China Marches West: The Qing Conquest of Central Eurasia (Peter C. Perdue)

Формат документа: pdf
Размер документа: 4.98 Мб

Прямая ссылка будет доступна
примерно через: 45 сек.

Теги: Китай. Английский язык. Династия Цин
  • Сообщить о нарушении / Abuse
    Все документы на сайте взяты из открытых источников, которые размещаются пользователями. Приносим свои глубочайшие извинения, если Ваш документ был опубликован без Вашего на то согласия.


the qing conquest ofcentral eurasia
Peter C. Perdue
the belknap press of
harvard university press
Cambridge, Massachusetts London, England 2005

Copyright © 2005 by Peter C. PerdueAll rights reserved
Printed in the United States of America
Library of Congress Cataloging-in-Publication Data Perdue, Peter C.
China marches west : the Qing conquest of Central Eurasia / Peter C. Perdue. p. cm.
Includes bibliographical references and index. ISBN 0-674-01684-x (alk. paper)
1. China—History—Qing dynasty, 1644–1912. 2. Russia—History—1613–1917. 3. Mongols—History. I. Title: Qing conquest of Central Eurasia. II. Title.
ds754.p47 2005
951′.03—dc22 2004059472

To Linda, Kay, and Alex, and the memory of my parents

Preface xiii
Acknowledgments xvii
Note on Names, Dates, Weights and Measures,and Chinese Characters
Introduction 1
History, Time, and Memory 6
The Qing Conquests as a World Historical Event 9
part one The Formation of the Central Eurasian States
1 Environments, State Building, and National Identity 15
The Unboundedness of Central Eurasia 19
Trade, Transport, and Travel 33
The Frontier Zone 41
Isolation and Integration 44
2 The Ming, Muscovy, and Siberia, 1400–1600 51
The Ming and the Mongols 52
State Formation in Muscovy and Russian Expansion 74
Siberian and Chinese Frontiers 89
3 Central Eurasian Interactions and the Rise of the
Manchus, 1600–1670
Building the Zunghar State 101

The Rise of the Manchus109
Mongolian Influence on the Manchu State 122
Early Modern State Building Compared 127
part twoContending for Power
4 Manchus, Mongols, and Russians in Conflict, 1670–1690 133
Kangxi the Ruler 133
Galdan’s Intervention 144
Kangxi’s First Personal Expedition 152
The Treaty of Nerchinsk and the Excluded Middle 161
5 Eating Snow: The End of Galdan, 1690–1697 174
The Dolon Nor Assembly 175
The Battle of Jao Modo 180
The Emperor Rewrites History 190
The Final Campaigns and the Fate of Galdan 193
6 Imperial Overreach and Zunghar Survival, 1700–1731 209
The Rise of Tsewang Rabdan 210
Three Central Eurasian Travelers 213
The Penetration of Turkestan and Tibet 227
The New Emperor Changes Tack 240
7 The Final Blows, 1734–1771 256
Transforming the Barbarians through Trade 256
The Death Knell of the Zunghar State 270
The Conquest of Turkestan 289
The Return of the Torghuts 292
part three The Economic Basis of Empire
8 Cannons on Camelback: Ecological Structures and
Economic Conjunctures
Galdan the State Builder 304
Nian Gengyao and the Incorporation of Qinghai 310
Administering the Frontier 314
9 Land Settlement and Military Colonies 324
Deportation from Turfan 329
Settlement of Xinjiang 333
Colonization and Land Clearance 342
Economic Development 353

10 Harvests and Relief 358
Harvests and Yields 359
Granary Reserves 362
The Contribution Scandal 365
The Relief Campaign of 1756 370
11 Currency and Commerce 378
Money on the Frontier, from Song through Ming 379
Integration and Stabilization 383
Commerce as a Weapon of War 397
Tribute and Frontier Trade 402
part four Fixing Frontiers
12 Moving through the Land 409
Travel and Authority 413
Marking Space in Stone 429
Maps and Power 442
Expanding the Imperial Gaze 457
13 Marking Time: Writing Imperial History 462
Kangxi’s Campaign History 463
Yongzheng and the Dayi Juemilu 470
Qianlong’s Account of the Zunghar Mongols 476
A View from the Frontier 481
Nomadic Chronicles 486
part five Legacies and Implications
14 Writing the National History of Conquest 497
Statecraft Writers and Empire 497
Geopolitics and Emperor Worship 504
Chinese Historians and the Multicultural State 506
Soviet and Mongolian Attacks on Qing Aggression 511
Empires, Nations, and Peoples 513
15 State Building in Europe and Asia 518
The Political Ecology of Frontier Conquest 520
European, Chinese, and Inner Asian Models 524
Theories of Nomadic Empires 532
Rethinking the Qing in the World 536
contents ix

16 Frontier Expansion in the Rise and Fall of the Qing 547
The End of the Qing State 549
Northwest and Southern Frontiers 551
The Negotiated State 555
Commercialization and Regionalization 558
A Rulers and Reigns 569
B The Yongzheng Emperor Reels from the News of the Disaster, 1731 571
C Haggling at the Border 575
D Gansu Harvests and Yields 583
E Climate and Harvests in the Northwest 589
Abbreviations 591
Notes 593
Bibliography 671
Illustration Credits 707
Index 711

1 The Qing empire, ca. 1800 2
2 The Zunghar empire 3
3 Ecological zones of Eurasia 22
4 Tribal peoples and Russian settlements in the sixteenth andseventeenth centuries
5 The Sino-Russian frontier 168
6 The Kangxi emperor’s Zunghar campaigns, 1690–1697 182
7 The Qianlong emperor’s western campaigns, 1755–1760 273
8a Grain price integration in Gansu, 1739–1864 390
8b Grain price integration in Gansu with famine years omitted, 1739–1864
9 Grain price integration in Xinjiang, 1777–1860 395

I began the research for this book in more peaceful times, purely out of in-
tellectual curiosity about a little-known region of the world and a neglected
topic in Chinese imperial history. Today, the issues raised by this study have
come all too close to home, both for the Chinese and for us. The PRC gov-
ernment has now convinced the UN Security Council to classify the East
Turkestan Independence Movement as a terrorist organization. At the same
time, the PRC has launched a visionary project to “develop the West,” so as
to bring the Central Eurasian and interior regions of China firmly into the
economic web generated by the reform program of the last two decades.
Although I believe that China will maintain control over this region for the
foreseeable future, I make no predictions. But the imperial legacy of con-
quest still hangs heavy over the future of the Chinese nation-state. Specialists have, for disciplinary convenience, usually divided the region
in two. “Inner Asia,” conventionally defined as modern Mongolia (Inner
and Outer), Manchuria, Xinjiang, and Tibet, has historically been pri-
marily the province of those knowledgeable in Chinese, Mongolian, and
Manchu languages. “Central Asia” generally refers to the area of Turkic
peoples bounded by the former Soviet Union, and much of its scholarly lit-
erature is in Russian. But we cannot let a simple Russian–Chinese divide
define a cultural arena, especially for the early modern period, centuries be-
fore either nation existed. “Inner Asia” has its uses as a subdivision of Asia
setting it off from East, Southeast, and South Asia, but it is misleading to set
its boundary simply along the former Sino–Soviet border. “Asia,” a Euro-
pean concept not embraced by any indigenous peoples of the region until

the twentieth century, has fairly clear maritime borders to the east and
south (ignoring the problematic status of Australia, New Zealand, and
Polynesia) but no clear borders on the northwest.These geographic descriptions reflect a binary division which has now
vanished. I prefer the term “Central Eurasia,” now used by the U.S. State
Department and by the only Department of Central Eurasian Studies in the
United States, at Indiana University. It is a less familiar term, but it lacks the
historical baggage of other words. It also reminds us of the artificiality of
the distinction between “Europe” and “Asia,” and the awkward status of
Russia in this definition. Without being foolishly consistent, I have used
“Central Eurasia” most of the time in this book, and have used “Inner
Asia” only when the discussion clearly refers to the Chinese side of the
border. I chose the title of this book carefully. “China marches west,” because
this is the story of a conquest endorsed by the Chinese nation-state today.
But it is the “Qing Conquest,” not the “Chinese Conquest,” because many
of the major participants were not Han Chinese, and “Central Eurasia” in-
stead of more common terms like “Inner Asia,” “Central Asia,” “Mongo-
lia,” or “Xinjiang” in order to indicate the large scale and blurred bound-
aries of the territory the empire gained. Chinese historians would prefer
a title like “The Unification of the Mongolian and Uighur Peoples under
the Qing Multinationality State”; Russians and Mongolians might choose
“The Invasion of the Mongolian Peoples by the Aggressive Manchu Qing.”
Every choice of words has political implications; none is neutral, but some
can be more detached than others. Throughout, I try to stress the pre-na-
tional character of the processes described here. The warriors and traders
of the eighteenth century did not think in terms of nation-states; they pur-
sued their own interests as they saw them, and we should both recover and
respect their views. Nevertheless, they laid the necessary groundwork for
nationalism by their actions. This study both critiques nationalism and
demonstrates the continuity between empire and nation. Its dates are indefinite. We might begin with the birth of Nurhaci in
1559, the first Russian penetration of Siberia in 1582, the first declaration
of an empire in Manchuria in 1616, or the launching of the wars against
Galdan in 1690. Likewise, this story could end with the death of Amursana
in 1757, the final suppression of rebellion in Turkestan in 1765, the “re-
turn” of the Torghut Mongols to Qing territory in 1771, or extend through
the final Russian conquest of Central Eurasia in the mid-nineteenth century.
History being a continuous flow, all periodization is inevitably Procrustean.
I indicate significant turning points throughout. This is a long book, to be sure. If I had found an adequate study of this
xiv preface

topic, I would not have gone on at such length. And yet it falls far short of
being a comprehensive analysis of Qing China’s relations with Central Eur-
asia, or even of the Zunghar campaigns alone. To my surprise and horror,
what seemed to be a manageably focused study of a “peripheral” subject
soon revealed vast seas of archival and published documents and secondary
research. There still remain huge numbers of documents and questions
which I have not had the time or energy to study in depth. I have ended
with many more questions unanswered than resolved. I only hope that by
insisting on the importance of China’s northwest campaigns I can stimulate
others to follow the frontier trail.Although this is a large, sprawling book, it has a simple format. It alter-
nates between structural analysis and narrative. Part One sets the stage,
looking at the ecological conditions under which the contests took place
and the background to dynastic decision making up until the seventeenth
century. Part Two follows the basic story from the rise of the Qing in Man-
churia, the establishment of the Mongol state, and the arrival of the Rus-
sians in the early seventeenth century, through the consolidation of control
at the end of the eighteenth century. Part Three turns back to a structural
analysis of the economic and environmental constraints which the conquer-
ors and their rivals had to overcome. Part Four looks at the cultural perfor-
mances and symbolic representations that legitimated the conquest and left
a legacy to the future. And Part Five sums up the implications for modern
China and modern theorists about state building in China and elsewhere. I
have tried to give a balanced account of the motives of all the major parties
involved in the conflict, but since the bulk of the sources are in Chinese, bi-
ases are unavoidable, even if we read texts against the grain. This is primar-
ily a book about the Qing rulers and the world they made, but it tries to
rescue from “the enormous condescension of posterity” (to quote E. P.
Thompson) many others who played their parts in the drama. I take no po-
sition on whether the empires described here were good or evil: that kind of
judgment is not the historian’s role. But I do insist that their conquests were
not inevitable, that resistance and accommodation both deserve their story,
and that the course of this struggle left traces that persist today.
preface xv

T his study has taken far too long for me to be able to thank adequately
everyone who has contributed to it. Joseph Fletcher, with whom I studied
Manchu at Harvard, died in 1984 while I was in the archives in Beijing
completing my first book and exploring a new topic on northwest China. I
and many others have tried to carry on the work that he so heroically
pioneered. I received valuable help from the staff at the Number One Ar-
chives in Beijing, the Palace Archives in Taiwan, the Archive of Foreign Pol-
icy, and the State Archive of Ancient Acts in Moscow. Galina Khartulary
and Valerii Klokov were invaluable guides to the mysteries of Moscow. In
Beijing, I would like to single out Cheng Chongde, Director of the People’s
University Qing History Institute, and his colleagues, and Li Bozhong, an
old friend and colleague, among many others. Back home, I am especially
grateful to Dean Philip Khoury for his constant support, and to all my col-
leagues on the MIT History Faculty for comments and encouragement.
Anne McCants, Pauline Maier, Harriet Ritvo, and Elizabeth Wood offered
especially valuable reactions to an early chapter. John Dower has been a
constant source of sagely wisdom. I could make progress on this book even
while serving as department head only because of their goodwill and dedi-
cation. Many parts of this text have been presented at seminars at the
Fairbank Center for East Asian Studies at Harvard University, Columbia
University, the University of California, Irvine, the California Institute of
Technology, the Institute for Agrarian Studies at Yale University, the Uni-
versity of Kansas, and annual meetings of the Association of Asian Studies,
the American Historical Association, and seminars as far-flung as Istanbul,
Taibei, Beijing, and Japan. I greatly appreciate all the helpful reactions re-

ceived from those who attended, especially Tim Brook, Pamela Crossley,
Nicola Di Cosmo, Philip A. Kuhn, Robert Marks, Ken Pomeranz, James C.
Scott, Carl Strikwerda, Charles Tilly, Frederic Wakeman, Joanna Waley-
Cohen, R. Bin Wong, Alexander B. Woodside, Donald Worster, and many
others. My deepest regret is that Benjamin Schwartz, a model of scholarly
rigor and personal engagement, died before this manuscript was finished. I
would have loved to hear his reactions. I deeply appreciate the tireless labor
of the research assistants at MIT and Harvard who helped with the prepa-
ration of the book: Cherng Chao, He Wenkai, Jiangti Kong, Ellen McGill,
Nana Okura, Helen Tang, and Angela Xiang. I am also grateful to the
many scholars who have gone before me in this field. For centuries, in many
languages, often without knowing of one another’s existence, they accumu-
lated an invaluable fund of synthetic and documentary knowledge. If we
see a bit further, it is because we stand on their shoulders.I must also offer special thanks to the Chao family, whose endowment of
the T. T. and Wei Fong Chao Professorship helped substantially in provid-
ing the resources to prepare this manuscript for publication, especially the
beautiful artworks included in it. I thank Lin Man-houng of Academia
Sinica, Taiwan, for her invaluable help in obtaining artworks from Taiwan.
My editors, Kathleen McDermott and Elizabeth Gilbert, have been par-
ticularly generous, firm, and efficient in bringing this book into its pres-
ent form. Amanda Heller meticulously edited the final copy, and Phil
Schwartzberg drew the elegant maps. In mostly unconscious ways, many environments have influenced the
perspectives of this book. Travels to Beijing, Japan, Mongolia, Moscow,
and Xinjiang have helped to ground the abstractions of ancient texts with
contemporary experience and the natural world. I am grateful to all those
whom I met there and who helped me on the way. More locally, I reflect on
Jamaica Pond, the Arnold Arboretum, the Charles River, Dorchester, MIT,
and Harvard Yard, where many of the ideas in this book were generated. Some chapters draw on previously published material in revised form:
“Boundaries, Maps, and Movement: The Chinese, Russian, and Mongo-
lian Empires in Early Modern Eurasia,” International History Review20
(June 1998), pp. 263–286; “Empire and Nation in Comparative Perspec-
tive,” Journal of Early Modern History 5, no. 4 (2001), pp. 282–304; “The
Agrarian Basis of Qing Expansion into Central Asia,” in Papers from the
Third International Conference on Sinology: History Section (Zhongyang
Yanjiuyuan Disanzhou Guoji Hanxue Huiyi Lunwenji Lishizu) (Taibei: In-
stitute of History and Philology, Academia Sinica, 2002), pp. 181–223; and
“A Frontier View of Chineseness,” in The Resurgence of East Asia: 500-,
150-, and 50-Year Perspectives, ed. Giovanni Arrighi, Takeshi Hamashita,
and Mark Selden (London: Routledge, 2003), pp. 51–77.
xviii acknowledgments

Note on Names, Dates, Weights andMeasures, and Chinese Characters
Achieving consistency in the romanization of personal and place-names has
caused me constant headaches. I have tried wherever possible to write each
person’s name in the romanized version of the language of his primary eth-
nic group. Alternative romanizations are given in parentheses, preceded by
an indication of the language (Ma. for Manchu, Mo. for Mongolian, Ch.
for Chinese, and T. for Tibetan). For Chinese names, I use pinyinromani-
zation; for Manchu names, the spelling used in Jerry Norman, A Concise
Manchu-English Lexicon (Seattle: University of Washington Press, 1978).
Russian names generally follow the Library of Congress system. Mongo-
lian names come in many flavors, reflecting differences of dialects and dif-
ferent romanization systems. Foolish consistency being impossible, I have
used spellings that, I hope, will be at least pronounceable by those unfamil-
iar with the subtleties of Altaic languages, and at least recognizable by
those who care about philological accuracy. For Tibetan, I give a pro-
nounceable version following R. A. Stein, Tibetan Civilization(Stanford:
Stanford University Press, 1972), with the original spelling in parentheses.
Turkic names and places usually follow James A. Millward, Beyond the
Pass: Economy, Ethnicity, and Empire in Qing Central Asia, 1759–1864
(Stanford: Stanford University Press, 1998). See Appendix A for the names
and reign years of the major Central Eurasian rulers discussed here. I use different romanizations to distinguish Soochow, the great textile
city of the lower Yangzi, from Suzhou, the garrison town in western Gansu.

All dates in the text are given in the Julian solar calendar; Chinese refer-
ences in the notes follow the Chinese lunar calendar with reign name (KX:
Kangxi; YZ: Yongzheng; QL: Qianlong). An asterisk indicates the lunar in-
tercalary month following the numbered month.
Weights and Measures
1 jin (catty) =1.3 pounds or 0.6 kg
1 li= 0.36 miles or 0.576 km
1 mou =0.1647 acre or 0.0666 hectare
1 shi =103.5 liters or 2.9 bushels (23.5 gallons). One shiof milled rice
weighed approximately 175–195 pounds (80–89 kg).
1 patman (Ch.bateman) =a Turkestani land and grain measure, equiva-
lent to 4.5–5.3 shiof grain, or 26.5 mouof land (from Millward, Be-
yond the Pass, p. 54).
Chinese Characters
Because I wrote this book for readers who may not know Chinese, it does
not contain Chinese characters. A character list is available on the Web site; search for “Perdue.”
xx note on names, dates, weights and measures

The conquest of the earth, which mostly means the taking it
away from those who have a different complexion or slightly
flatter noses than ourselves, is not a pretty thing when you
look into it much. What redeems it is the idea only.
Joseph Conrad,Heart of Darkness
You have conquered the empire on horseback; but can you
rule it on horseback?
Lu Jia (ca. 228–140 bce) to Emperor Han Gaozu

F romthe seventeenth to the mid-eighteenth century, three great empires—
the Manchu Qing (1644–1911), the Muscovite–Russian (1613–1917), and
the Mongolian Zunghars (1671–1760)—contended for power in the heart
of Eurasia. The distances were vast, communications slow, military cam-
paigns extended and costly, and cultural alienation was huge. By the end of
this epic confrontation, an early version of the “Great Game,” only two
empires were left standing.
1The Qing and Russians faced each other along
an extended border. They had become two of the largest empires in world
2The Zunghars had vanished. Despite nineteenth-century upheav-
als, this binary division of Eurasia lasted until the collapse of the Soviet
Union in 1991.
3This book examines the division of the continent between
the two empires, emphasizing the Qing empire’s conquest of Mongolia and
Xinjiang. (See Maps 1 and 2.) No one could claim that the Qing campaigns in Central Eurasia have
been ignored. Every textbook mentions them. This is not the recovery of a
long-forgotten event. And yet, there has never been an adequate full-length
study of them in English.
4Studies in other languages also have their limita-
tions. This book, based on primary sources, archival and published, in sev-
eral languages, tells a dramatic story almost unknown to English-language
readers. It also critiques some of the dominant paradigms of modern Chinese his-
toriography, most prominent in the work of Chinese scholars but also im-
plicitly adopted by many in the West. In brief, most historians, supported
by the prevailing nationalist ideology that reigns on both sides of the Tai-

Map 1. The Qing empire, ca. 1800.

Map 2. The Zunghar empire.

wan Strait, take the current territorial and cultural boundaries of the Chi-
nese nation-state for granted. They view the conquest of Mongolia and
Xinjiang as the fulfillment of a development that culminates with the Re-
public of China in the twentieth century, or, in Chinese terms, the “unifica-
tion” of the peoples of Central Eurasia in the multinational Chinese nation-
state. These historians view the Qing expansion at its maximal extent,
around 1800, as the end of this process. Since then, in the nineteenth cen-
tury, the empire declined from its peak of glory until it recovered sover-
eignty over much of the region by 1949. Mongolia, of course, and much of
Manchuria north of the Ussuri River still remain outside the People’s Re-
public, as does Taiwan. In the ideals of Chinese nationalists, all these re-
gions should be part of the modern nation-state. Indeed, from time to time
the mainland Chinese leaders have indicated that they still see these regions
as part of the nation. They of course strongly insist that Taiwan belongs to
one China. Even though nationalists rejected the Manchus as an obstacle to
China’s modernization, the Chinese nation-state derived its concept of its
ideal boundaries from the maximal expansion of the Qing empire in the
eighteenth century. Like other nationalists, the Chinese built on the past
they rejected.From this perspective, the conquest of Xinjiang and Mongolia was im-
portant but not epochal. But what if we reject the view of the conquest as
merely one link in a chain that inevitably led to the nation-state? What if it
was a contingent product of the crosscutting wills of many players, only
some of whom were Han Chinese, and none of whom had modern nation-
alism in mind? The security threat from the northwest was indeed long-
lasting, in every imperial ruler’s mind. What emperor did not wish to rid
himself of the damned nomads who harassed the northwest frontier? All
previous efforts, however, with the brief exception of the Mongol-ruled
Yuan dynasty (1279–1368), were unsuccessful, until the Qing achieved the
final solution in the mid-eighteenth century. Why were the Qing able to
eliminate the Mongol threat once and for all, and to dominate this region
ever since, when no previous dynasty succeeded? That will be one of the
central questions of this work. Specialists in Central Asian history have generally ignored this period.
The entire era from the fall of the Mongol empire to the present day, re-
garded as the “Époque de la Décadence,” accounts for only 30 pages out
of over 350 in Denis Sinor’s introduction to the field. They have empha-
sized philology over ethnography (only 8 pages in Sinor) or other social sci-
entific perspectives.
5Luc Kwanten’s survey of steppe empires stops at the
year 1500, because he, like many others, regards the Mongol empire of
Chinggis Khan as the culmination of the steppe imperial tradition, followed
4 introduction

by centuries of decline. 6Many historians have used the paradigm of a clas-
sical age of florescence and expansion, followed by a period of “decline,”
to structure their accounts. Historians of the Ottoman and other empires
have grown increasingly critical of this perspective.
7The main defect of this
paradigm is that it takes the end for the beginning by reading history back-
ward from a time of weakness to its origins in an “era of decadence.” Like
nationalist historiography, the paradigm of decline forecloses consideration
of roads not taken, crucial decisions, and contingencies, and denies agency
to the actors. We need to reintroduce independent agency to the Central
Eurasians in this story, without taking for granted either the hostile per-
spectives of their Manchu rivals or nostalgia for the days of Chinggis Khan. A careful examination of these events raises many questions about the
predominant historiography both of China and of Central Eurasia. The
best overview in a western European language is still that of Maurice Cou-
rant, a French diplomat who in 1912 wrote L’Asie Centrale aux 17e et 18e
siècles: Empire Kalmouk ou Empire Mantchou? It relies almost entirely,
however, on a single Chinese source. I. Ia. Zlatkin’s Istoriia Dzhungarskogo
Khanstvo, 1635–1758 (History of the Zunghar Khanate, 1635–1758), the
best survey in Russian, uses Manchu and Russian but not Chinese sources.
Chiba Muneo provides an extended Japanese narrative, and a number of
Japanese scholars have written relevant works.
8Chinese scholars have pro-
duced an enormous secondary scholarly literature and a major synthesis.
But each of these scholarly traditions has deficiencies, which reveal particu-
lar cultural and national characteristics. In the final section of the book, I
briefly review the historiography of the conquest in order to indicate how
historians of this topic, like all historians, have reflected the concerns of
their time. The study of modern Chinese history has increasingly turned inward.
Early works focused on China’s international position and included study
of the Manchu nature of the regime, the military institutions, and the secu-
rity considerations of the center.
9The rise of the Fairbank school shifted in-
terest toward diplomatic relations with maritime Western powers, the cen-
tral civil institutions of the Qing, and intellectual history. The influence of
G. William Skinner’s regional systems model led scholars to turn toward in-
tensive socioeconomic studies of single provinces or sub-regions. Much was
gained from the turn inward, but something was lost, too. Particularities of
individual variation overrode wider contexts. Placing China in world his-
tory should be a priority today, and study of the frontier is a promising way
to integrate China with the wider Eurasian world. This study of Qing ex-
pansion offers points of comparison with studies of frontiers elsewhere in
the world. It also questions the rigidity of Skinner’s regional systems model.
introduction 5

Instead of fitting our regional analyses into a prescribed geophysical frame-
work, we must recognize that often political and social processes moved
fluidly across regional boundaries, driven by considerations of military se-
curity, economic gain, or cultural transmission.
History, Time, and Memory
As well as reshaping conceptions of space, this study aims to adjust our
views of Eurasian time. Historians have often viewed Chinese and Central
Eurasian history in terms of cycles. The dynastic cycle paradigm has pro-
vided a convenient explanation of the rise and fall of Chinese imperial re-
gimes ever since the Zhou conquered the Shang in 1045bce. The repeated
unification and collapse of steppe empires has also inspired cyclical expla-
nations. Although many scholars see repetitive patterns in the state forma-
tions of both China and the steppe, they find widely divergent causes for
them. We may roughly classify these causes as moral and material. Moral
explanations look for the essence of change in human psychology. China’s
dynastic cycle theorists attributed the rise and fall of empires to the moral
characteristics of the ruler. If he governed wisely, he made his people happy
and attracted others to support him. If he was debauched, the peasantry
suffered, the soldiers rebelled, and officials deserted his cause. Updated ver-
sions of the dynastic cycle paradigm, by contrast, focus on material bases of
imperial rule, such as maintenance of irrigation works, equitable tax collec-
tion, or famine relief. In this perspective, if bureaucratic management of
these key social institutions failed to meet the social challenges, mass rebel-
lion or external invasion brought the dynasty down. For steppe empires, the predominant material explanation has invoked
climatic determinism, assuming that cycles of desiccation inevitably drove
nomads out of their steppe homelands to invade the settled societies around
them. By contrast, Ibn-KhaldÄn, the great Islamic historian and social theo-
rist, outlined a theory based on social solidarity, or ‘asabiyya, arguing that
the collective unity generated by tribal societies created a form of power
which allowed them to conquer decadent urban civilizations. After the con-
quest, as the moral codes generated in the steppe-desert environment de-
cayed under the influence of urban ways, the tribal conquerors lost their
solidarity and became victims of new invaders. Ibn-KhaldÄn’s theory shows
striking similarities to the Chinese dynastic cycle model, even though he ap-
plies its moral definitions not to “civilized” Confucianized rulers but to no-
madic warriors.
The cyclical models have the advantage of highlighting common patterns
6 introduction

that transcend the narrative of particular events. They thus point toward
generalizations about broad processes of social change. Long before the
rise of Western sociology, both Chinese and Middle Eastern social theorists
had begun to analyze the long-term evolution of their societies in compara-
tive terms. The danger of cyclical theorizing, however, lies in its denial of
human agency and its neglect of linear change. Strict adherence to cyclical
explanation implies that people do not learn from experience. They are
doomed to follow in the same tracks as their ancestors. But we know well
that no social process ever perfectly repeats the past, for two reasons. First,
technological advances greatly altered the kinds of conflicts that tore socie-
ties apart and the pressures that held them together. New weapons, reduc-
tions in transport costs, or new forms of communication, for example, all
transformed the means of integration and disintegration of Chinese dynas-
ties. Second, both Chinese and nomadic peoples learned from history. They
knew the precedents set by their ancestors, and they constantly used their
knowledge to build on past experience or to learn from their mistakes.
Hence the writing and rewriting of history formed a key element in shaping
the strategies of empire builders, officials, and ordinary social interactions.The geographical and strategic environment of Chinese–nomadic inter-
action created recurrent situations, but those involved dealt with these situ-
ations in different ways, depending on their historical and technological re-
sources. I describe in the chapters that follow examples of frontier trade,
horse breeding, and the use of gunpowder weaponry to indicate how the
Qing–Zunghar conflict partially followed old paths but often broke new
ground. It is particularly important to stress innovation by steppe empire build-
ers, since settled observers have so often viewed nomads as doomed to
stay in the same tracks forever. The striking continuities in nomadic life
from the creation of pastoralism to the present often inspired images of
monotony, repetition, and stagnation. The settled observers who provide
nearly all of our sources usually saw the nomads as catastrophic eruptions
into their civilizations, comparable even, in Braudel’s words, to biblical
11 Mechanical theses of environmental determinism likewise de-
prived the Central Eurasians of any agency, tending to reduce them to mere
biological actors. We shall see biological, naturalistic analogies invoked re-
peatedly by the enemies of the Zunghars, so as to exclude them from the
human world and justify their extinction. During the age of high imperialism, in the nineteenth century, Westerners
likewise portrayed Chinese as victims of natural forces, or as liable to sud-
den inexplicable outbreaks of violence. They applied simple stereotypes of
determinism and Asian autocracy to explain China’s persistent resistance to
introduction 7

Western intrusion. Our perspectives on China have changed considerably
in recent years. Serious scholars have abandoned old stereotypes of “Orien-
tal Despotism.” The “Naito hypothesis” and the “early modern hypothe-
sis” both hold that decisive socioeconomic changes occurred in China dur-
ing the ninth to tenth or late sixteenth century, respectively. In the last
section of this book, I support the “Eurasian Similarity Thesis,” which ar-
gues that Chinese empires were just as economically dynamic as European
ones up to the end of the eighteenth century.
12By giving China an internal
dynamic of its own, comparable to but separable from that of the early
modern West, these analyses point toward the true incorporation of China
into world history. Only recently, scholars have also argued for substantial historical change
in pastoral societies. Thomas J. Barfield’s model of nomadic state building
recognizes linear evolution, but his theory makes nomadic state builders
dependent on the changes in Chinese empires and does not allow them
much agency of their own. Nicola Di Cosmo has made the most cogent ar-
gument for autonomous dynamics in the formation of steppe empires. He
argues for freeing our study of Inner Asian peoples from biological imagery
and mechanical causation, and develops a periodization of state formation
based on the means by which state builders obtained revenue from external
actors. From the second century bceto the eighteenth century ce, revenue
extraction developed from tribute extraction, to trade-tribute partnerships,
to dual administration of settled and nomadic peoples, to fully developed
procedures for taxing an agricultural base.
Our story finds the Zunghars making use of all of these methods as they
attempted to support their expanding state. They exacted tribute from their
neighbors, traded extensively with the Qing, Russians, and Central Eur-
asians, practiced a form of dual administration in Turkestan, and devel-
oped fiscal structures in the Ili valley. As Di Cosmo notes, however, they
lacked the “breathing space” to consolidate their state because of the si-
multaneous expansion of the Qing and the Russians against them. Hence, I
see the seventeenth and eighteenth centuries as the decisive turning point in
steppe-settled interactions, determined by the actions of both the agrarian
empire builders and the nomadic state makers. The closure of borders cre-
ated by the concurrent expansion of the Qing and Russian empires substan-
tially increased the constraints on the peoples in between. The Zunghar
leaders did, as Barfield argues, try to use whatever resources they could
get from their imperial neighbors, Russian and Chinese, to support their
state, but they also tried to develop internal resources from the oases of
Turkestan and the Eurasian caravan trade. Material resources mattered,
but simple desiccation had little to do with state formation. Attacks of
8 introduction

drought could just as easily drive Mongols to seek refuge under Qing pro-
tection as to attack the empire. How nomads reacted to drought was a
product of personal and diplomatic negotiation between the Qing and the
Khans.Throughout this discussion I stress the multiple opportunities available
to all the actors and the indeterminacy of the outcome. The Qing conquest
and elimination of the Zunghars was never inevitable. Some environmental
factors favored the Qing, but others favored the Zunghars. Personal deci-
sions, accidental deaths, misunderstandings, and deceit all played impor-
tant parts. This story would have no drama if Heavenly mandates, environ-
mental conditions, or teleologies of the nation predetermined the outcome.
Instead, I place this story in its broadest context while keeping in the fore-
ground the contingent results of human decisions.
The Qing Conquests as a World Historical Event
Two contradictory paradigms have shaped our general outlook on the sig-
nificance of Central Eurasia. Many theorists have described it as the region
least prepared for modernization. In their view, its isolation from global
trends, its political and cultural fragmentation, and its poverty of natural
resources gave it few of the prerequisites for rapid development.
14For these
theorists, universal processes of economic growth, political and social inte-
gration, and communication inevitably drag backward, traditional socie-
ties into agonizing transformations. The backward societies have little or
nothing to offer to the advanced societies. Sooner or later, in this linear,
railroad track view of history, they follow the leaders. An alternative perspective is one we may call the “world historical”
point of view. Rejecting the automatic classification of societies by nation-
state units so characteristic of modernization theory, it denies that national
units can develop in isolation from one another. Often major civilizational
units transgress contemporary nation-state boundaries, and all these re-
gions constantly interact. Central Eurasia, in this view, was not a remote,
isolated region but the crossroads of the Eurasian continent, and it had a
decisive impact on all the settled societies around it. A current of histori-
cal interpretation from the Russian Eurasian school, to the geopolitics of
Halford Mackinder and Owen Lattimore, to the contemporary world his-
torians and theorists of world systems has viewed Central Eurasia as a key
region of the Eurasian world system.
In the early modern period, Central Eurasia was indeed the crossroads of
the Eurasian continent. Every major religion reached it. Trade routes be-
introduction 9

tween China, the Middle East, Russia, India, and Europe crisscrossed the
region. Up until at least the sixteenth century, the caravan traders following
the old Silk Routes played a major part in global trade. Even in the seven-
teenth and eighteenth centuries, as the old Silk Route trade declined, the
Russo-Chinese tea and fur trade remained important. Religious diversity,
linguistic pluralism, and cosmopolitanism characterized the oasis cities.
More than anything else, it was the conquest of the region by the “modern”
empires of China and Russia that relegated it to backwardness in the nine-
teenth century, not any essential features of the region. Scholars critical of
the modernization paradigm have blamed imperial conquest for causing
backwardness in other parts of the world. Perhaps conquest had the same
effect here. At least we can say that the Qing and Russian conquest of the
region strongly affected the potential of Central Eurasian peoples for devel-
opment in the nineteenth and twentieth centuries.The Qing conquests, then, were a major world historical event in three
senses. First, for the empire’s rulers and subjects, these victories fundamen-
tally transformed the scale of their world. By vastly expanding the territo-
rial reach of the state, the conquests opened up new terrain for colonial set-
tlement, for trade, for administration, and for the literary imagination.
Second, the expansion of the Qing state formed part of a global process in
the seventeenth and eighteenth centuries. Nearly everywhere, newly cen-
tralized, integrated, militarized states pushed their borders outward by
military conquest, and settlers, missionaries, and traders followed behind.
Western European historians often characterize this period as that of the
“seventeenth-century crisis” of state formation followed by eighteenth-
century stabilization, and other historians have found parallel processes
around the globe.
16Treating China’s imperial expansion as part of a global
process helps us to view China in a broader perspective instead of seeing ev-
erything about its imperial experience as unique. Third, China’s expansion
marked a turning point in the history of Eurasia. Across the continent, the
great empires founded by Central Eurasian conquerors in the wake of the
disintegration of the Mongol empire had captured the heartlands of densely
settled regions, used the resources of these regions to supply military forces,
and pushed back from the heartlands into the core of the continent. When
their borders met, they negotiated treaties that drew fixed lines through the
steppes, deserts, and oases, leaving no refuge for the mobile peoples of the
frontier. The closing of this great frontier was more significant in world history
than the renowned closing of the North American frontier lamented by
Frederick Jackson Turner in 1893.
17 It eliminated permanently as a major
actor on the historical stage the nomadic pastoralists, who had been the
10 introduction

strongest alternative to settled agrarian society since the second millennium
bce. Many Inner Asianists regard the decline of the nomadic warrior as
having been completed in the sixteenth century because of the diffusion
of gunpowder or shifts in the Central Asian caravan trade.
18One last ma-
jor nomadic state, however, held out against the military forces closing
in on the steppe. These Mongols battled vigorously against the Manchu
Qing armies. They adopted gunpowder weaponry in response to the mili-
tary threats around them, and the caravan trade remained a significant
source of revenue for all the contending parties. Thus the true world histor-
ical transformation that tipped the balance against unfettered nomadism
happened from 1680 to 1760, and the Qing rulers were major forces in this
momentous change.
introduction 11

The Formation ofthe Central
Eurasian States

Environments, State Building,and National Identity
T hree theoretical perspectives inform this work: frontier environ-
ments, state building, and the construction of national and ethnic identities
through historical representation. Many historians, recognizing the critical
role of the environment in shaping human affairs, have focused on the in-
teraction between man and nature. Western writers have, however, inflicted
a debilitating environmental determinism on the analysis of Asian history
that reflected nineteenth-century European assumptions of racial superior-
ity. In the eighteenth century, Baron de Montesquieu contrasted the flat,
monotonous plains of Asia with the varied landscape of Europe to explain
the contrast between Asiatic despotism and European freedom. Karl Au-
gust Wittfogel’s Oriental Despotism carried on the same style of analysis by
connecting Chinese imperial autocracy and Soviet communism alike to
state control of hydraulic works.
1Both used grossly oversimplified con-
trasts between European and Asian environments to undergird ideological
arguments protecting Western liberties against threats from the East. Re-
sponsible environmental history has to shake off this tainted legacy and
look at much more nuanced relationships between humans and their natu-
ral surroundings without political preconceptions. The American geographer Ellsworth Huntington drew a direct connec-
tion between environmental change in Central Eurasia and the evolution of
civilization. As he stated in The Pulse of Asia,after travels through Chinese
Turkestan from 1903 to 1906:
In relatively dry regions increasing aridity is a dire calamity, giving rise
to famine and distress. These, in turn, are fruitful causes of wars and

migrations, which engender the fall of dynasties and empires, the rise
of new nations, and the growth of new civilizations. If, on the con-
trary, a country becomes steadily less arid, and the conditions of life
improve, prosperity and contentment are the rule. There is less tempta-
tion to war, and men’s attention is left more free for the gentler arts
and sciences which make for higher civilization.
He found the same “parallelism between climatic changes and history”
from Turkey to China, in Europe, and in the New World. For Huntington,
desiccation processes, most visible in Central Asia, repeatedly forced no-
mads out of the center of the continent, causing everything from the bar-
barian invasions of Rome to the rise of Islam, and even threatening a poten-
tial invasion of the United States by starving hordes of Chinese in the
twentieth century.
Arnold Toynbee used Huntington’s findings to support his claims that
nomads were trapped by climatic conditions in a stagnant way of life, leav-
ing them a “society without history.”
4Only environmental changes could
alter their inevitable submission to the cycle of seasons, when droughts
forced them to invade settled civilizations. Toynbee, like his predecessors,
based his definitions of civilization on a contrast between the creative mas-
tery of nature in the settled agrarian realms and the animalistic vulnerabil-
ity of nomads to cycles of desiccation. Despite Huntington’s blatant, naïve racism and his almost laughable cru-
dity, he did correctly single out aridity as an important defining characteris-
tic of Central Eurasia. In his travels he observed carefully the local geology
and the relationship of the local people to their environment, mixing sen-
sitive description with vast biased speculations on human character. His
analysis alerts us to the inevitable political contamination of all efforts to
link human and natural history. Owen Lattimore at first endorsed Hunting-
ton’s and Wittfogel’s views, but in his later work he argued for the greater
importance of social over geographical causes. He found “mechanical ex-
planations, in the form of climatic cycles or progressive desiccation [ac-
counting] for the ‘blind’ eruption of nomad hordes” too crude, because
they ignored the “dynamics of social groups.” More recent work has tried
to shake off its imperialist traces by placing European and American devel-
opments in a broader global context.
Most of this new work has concentrated on the impact of Europeans on
the North American frontier, and of Americans on the western United
States during the nineteenth century. We need to see, however, the Ameri-
can expansion westward across the continent as part of a worldwide ex-
pansion of agrarian frontiers in the eighteenth and nineteenth centuries.
16 formation of states

Both the Russian and Chinese empires pushed from opposite directions
across Eurasia beginning in the sixteenth and seventeenth centuries. The
natural environments of North America and northern Eurasia, in fact, have
many similarities. Both include temperate forests which are home to fur-
bearing animals, which are trapped by native peoples; both have grasslands
where nomadic peoples roam; and both have deserts and arid regions as
well as fertile agricultural zones. The Chinese reached no Pacific coast, of
course, and found no gold (although the Russians prospected for gold in
southern Xinjiang), but like their American counterparts, they moved west
in search of land. The Russians reached the Pacific and beyond to Alaska in
their inexorable quest for furs. In all three cases, the interlocking environ-
ments of forests, deserts, steppes, and agrarian settlement shaped their his-
tories. Natural forces beyond human control and organized efforts to tap
the environment for human livelihood drove the settlers onward.Environmental historians have closely examined settlement on new fron-
tiers. William Cronon’s work, for example, has progressed from a study
of the impact of English settlers on seventeenth-century Massachusetts,
to the takeover of the Great West by the forces of capitalism centered in
Chicago, to the exploration of the mining frontiers of Alaska.
7In each
case he has focused on the interplay between indigenous peoples, the natu-
ral environment, and the disturbance created by intrusive aggressive set-
tlers from western Europe. Alfred Crosby’s studies of the “Columbian ex-
change” likewise followed in detail the ecological disruption brought by
European settlers to the New World. Recent work on Oceania likewise
demonstrates the devastating impact of European biota on the ecosystems
of the Pacific Islands. There is thus a close link between environmental and
frontier history.
Students of America’s frontiers inevitably have to refer to Frederick Jack-
son Turner’s famous thesis about the formation of the American character
by pioneer settlement.
9Critics of Turner’s excessive Anglocentrism aim to
offset his biases by exploring the histories of the other peoples who inhab-
ited the American West before the arrival of English settlers, Native Ameri-
cans, Hispanics, and Chinese among them. Others promote a separate re-
gional history of the West, autonomous of the forces of eastern capitalism,
with its own special identity.
10 Most of the participants in this debate,
however, do not examine North American frontiers comparatively. On the
Eurasian frontiers, Chinese and Russians also viewed their settlement pro-
grams as the penetration of civilization into wilderness, ignoring or repress-
ing the independent histories of other inhabitants. The national histories
of both countries are intimately bound up with their expansion into the
Eurasian interior. Although they have maintained national states, unlike
environments and state building 17

the indigenous Americans, Mongols and other Central Eurasian peoples
likewise still find much of their history written in terms of penetration of
their territory by expanding empires. Neither China nor Russia produced
as prominent a figure as Turner to epitomize its national character, but each
country’s experience demonstrates useful analogies to the North American
process. The Eurasian frontiers, because of their remoteness from western
Europe, are much less well known than those of America, but they deserve
to be included in any comparative study of frontier settlement.Similarly, theorists of state building have aimed primarily at explaining
the rise of European states from the sixteenth century forward.
11Some the-
orists explicitly exclude “empires” like the Qing from their analysis, put-
ting them in a different category from “states.”
12Here, however, I argue for
resemblances among the three major competing agrarian empires in Eur-
asia—the Zunghar Mongols, the Russians, and the Chinese—and the Euro-
pean states. All three, driven by geopolitical competition, mobilized their
resources for war, trade, and diplomacy against one another. In order to
compete with its rivals, each regime strove to increase its “stateness”—its
“formal autonomy, differentiation from nongovernmental organizations,
centralization, and internal coordination”—by extracting resources from
its subject peoples and from its neighbors.
13These efforts led to substantial
social and institutional reforms. The self-interest of state rulers and their
officials led them to develop bureaucratic methods to tap agrarian and
commercial production all across Eurasia. Western Europe did not follow a
special path. The collapse of the Soviet Union and the resurgence of nationalism in
eastern Europe and elsewhere have nurtured a new crop of theories of na-
tionalism and national identity.
14 These new theories argue that nations,
as “imagined communities,” are created, not born. Nationalist ideologies
claim two contradictory positions at the same time: that their nation has
deep roots in land, language, and culture, and that it is a new creation, re-
jecting the stagnant legacy of the past. The contemporary study of national-
ism highlights this contradiction, stressing the artificial, constructed nature
of every nation, including China.
15Language and subjective identity count
for much more than the “objective,” nearly timeless features of land and
race in defining the boundaries of the modern nation-state. Ideologues mo-
bilize particular interpretations of past events as tools for constructing new
national identities. As Eric Hobsbawm has remarked, “History is the raw
material for nationalist or ethnic or fundamentalist ideologies, as poppies
are the raw material for heroin addiction.”
16 History becomes a weapon,
easily turned into myth, appropriated for the purposes of nation builders
and citizens. The real events of the past are viewed indirectly, mediated by
18 formation of states

memory. But Europe was not the only nationalist construction site, and
“geobody building” did not begin in the nineteenth century.
The historian’s task is to recognize both the falsity and the effectiveness
of myth.
18As soon as the battles for territory ended, the battles for control
over the historical meaning of conquest began. The conquest of Eurasia
played a critical role in the national conceptions of all three of the competi-
tors, but it was interpreted in strikingly different ways. Thus our story be-
gins with nature, continues with individual actors, and ends with the histo-
rians. After a long excursus through the events, we return, in a cycle of
recapitulation, to examine how myth making, which began under the Qing
empire, created the elements that composed the nationalist histories of the
twentieth century. The conclusion returns to the perspectives sketched here,
to address the implication of this story for general paradigms of Chinese
and world history.
The Unboundedness of Central Eurasia
Central Eurasia has never coincided neatly with national boundaries. Only
under the brief rule of the Mongols was the region united under one impe-
rial ruler. Until 1991, China, Mongolia, and Russia or the Soviet Union
controlled the bulk of the region, with other parts in Iran, Afghanistan, and
the Ottoman empire. Today, eight independent nations (the five former So-
viet Central Asian Republics, Russia, Mongolia, and China) divide up most
of the territory. Fragmentation has been by far the most common experi-
ence of Central Eurasians. The bipolar Soviet–Chinese split turns out to
have been a brief interlude. Most broadly defined, Central Eurasia extends from the Ukrainian
steppes in the west to the shores of the Pacific in the east, from the southern
edge of the Siberian forests to the Tibetan plateau. But all of its borders are
so ambiguous that endless disputes arise. If Central Eurasia includes all
grasslands and steppes, it extends through the Ukraine into the Hungarian
plains. By cultural and linguistic criteria, including the Ural-Altaic language
family, Turkic and Mongolian peoples are found as far afield as Finland,
Manchuria, and, arguably, Japan and Korea. Steppe nomadism alone is not
sufficient as a defining feature, because all across the region nomads coex-
isted with settled agriculturalists; and other nomads, of the Middle East, or
reindeer herders of Siberia, are omitted. Almost every scholar defines the boundaries of the region differently.
Cyril Black includes the five Central Asian Republics, Iran, Afghanistan, Ti-
bet, Xinjiang, and Outer Mongolia in “Inner Asia,” but not Manchuria or
environments and state building 19

Inner Mongolia, because Chinese dominated these areas in the twentieth
century. He estimates a total area for this region in 1989 of 4.9 million
square miles (12.7 million square kilometers), with a population of 135
19Joseph Fletcher divides the region into the “desert habitat charac-
teristic of Central Asia (a region that includes, roughly, present-day Af-
ghanistan, Soviet Central Asia—which does not include Kazakhstan—and
Sinkiang [Xinjiang] south of the Tianshan mountains)” and the “steppe
habitat” which “covers a wide zone running from Europe to Manchuria
along roughly the fiftieth parallel north latitude, its main regions being
the south Russian steppe, Kazakhstan, Zungharia, much of Amdo (present-
day Tsinghai [Qinghai] province), and Mongolia north and south of the
20He omits Tibet. Denis Sinor’s organization of the region, primarily
linguistic, includes the speakers of Finno-Ugric, Altaic, Manchu, Mongo-
lian, and Turkic languages.
21 Finns, Lapps, and Hungarians enter the pic-
ture, but Chinese, Russians, Tibetans, Arabs, and Iranians are omitted, ex-
cept as the writers of sources. Richard Frye confines himself to Indo-Iranian
language speakers.
Rather than choosing among these disparate definitions, I would stress
the great indeterminacy of the cultural characteristics and territorial bor-
ders of this zone, which resulted in constant competition by empires, reli-
gions, and cultural groups to define and control it. No simple myths of
well-bounded space could ever confine it: no imaginary hexagon of France,
no sceptered isle of England, no manifest destiny stretching between two
ocean coasts. The huge flatlands offered no decisive barriers to invaders,
and the mountain ranges left large gaps through which conquerors could
pour their troops. The bewildering fluctuations of a multitude of empires,
each ruled by different peoples with different boundaries and institutions,
demonstrated the great plasticity of the landscape. I shall try to avoid anachronistically imposing definitions from the age of
nationalism on the region, focusing on the critical importance of the con-
quests of the seventeenth and eighteenth centuries for the evolution of the
Chinese and Russian empires. Manchuria, Mongolia, and Tibet cannot be
excluded from this story, but Iran and Afghanistan played very small roles.
I view the conquest not as the “Chinese” conquest of “Xinjiang,” but as the
expansion of a Central Eurasian state that used the massive resources of the
Chinese bureaucracy and economy to bring as much as possible of Central
Eurasia and the Chinese core under its rule. The main actors in our drama, in fact, covered nearly all the major zones
of classical Central Eurasia, from the south Russian steppes in the west,
where the Volga Kalmyks settled, to Manchuria in the east. Siberian native
20 formation of states

peoples in the north provided the furs that drew Russia to the region, and
Tibetan lamas to the south provided religious, bureaucratic, and clerical ex-
pertise critical for the formation of the Zunghar state. Caravan traders of
many nationalities linked all these territories together. This story encom-
passes a vast scale, not neatly bounded by national territories. The end of
the nomadic empire builders meant the end of unbounded space and the
clear demarcation of the region into mutually exclusive agrarian empires,
soon to become even more exclusive nation-states.To free ourselves from national divisions, it is best to begin by looking at
ecological zones, starting with the largest scale and then narrowing down.
(See Map 3.) Central Eurasia contains forests and fields, oases and deserts,
grasslands, rivers, and mountains. Its inhabitants include pastoral nomads
roaming the grasslands, traders in the oasis towns, and the agriculturalists
surrounding them. The homogeneity and uniformity of the ecological zones
of the region make it quite distinct from other, more fragmented parts of
the world. Very high mountains surround its periphery, but the great ma-
jority of the territory is broad and flat. Climate and precipitation determine
the basic characteristics of vegetation and animals in four broad horizontal
strips, defining tundra, forest, steppe, and desert. It is not the absolute
amount of precipitation but the ratio of precipitation to evaporation that is
decisive. Very low rainfall in the cold tundra still produces wet soil, while
more rain in the southern deserts yields no moisture. South of the Arctic
tundra, where only reindeer herders can find subsistence, are the forests of
Siberia, which overlap with the steppe grasslands. The grasslands shade
into steppe desert and large expanses of pure desert in the south.
Change did occur in Central Eurasia, but outside observers often missed
its signs. On the one hand, the uniformity of the ecological zones allowed
easy movement east and west, which especially suited the herding and no-
madic peoples who flourished there. On the other hand, the lack of di-
versity within the zones limited possibilities for greater development. The
Chinese and other sedentary peoples incorporated diverse agricultural and
maritime production areas within their borders, and used the variegated
products of their different zones to build up complex civilizations and em-
pires. The Central Eurasian peoples often seem to have repeated the same
adaptations to the monotonous environment without change for millennia.
Grazing animals, the tent, the mounted warrior, and his mobile but precari-
ous existence recur in the descriptions of the many different peoples of Cen-
tral Eurasia from the Scythians to the last of the Mongols. The “civilized”
writers who provide our sources tended to regard the Central Eurasian peo-
ples as universally greedy, primitive, and poor, and identified them almost
environments and state building 21

exclusively as nomads. They ignored the different ways of life in the re-
gion and neglected its possibilities for technological development. They
“allow[ed] the assumed Nomad to obscure the living man.”
The substantial uniformity of climatic zones from east to west also made
it impossible to draw sharp boundaries across the region. Since no obvious
divisions marked the end of one culture and the beginning of another, all
22 formation of states
Map 3. Ecological zones of Eurasia.

sorts of travelers and warriors could move freely back and forth. The great-
est problem of empire builders was to define their stopping point, and, once
they had stopped, to make provisions for security along their frontier. Basic
natural features such as rivers, mountains, and settlements did not define
well-enclosed spaces but tended to create centrifugal patterns. As Justin
Rudelson writes, “The historical focus of the Xinjiang oases was not in-
ward, toward each other, but outward, across borders.”
25 Empires and
their borders shifted constantly across the region because no natural
boundaries existed.
For the most part, the rivers of Central Eurasia flow inward, or into frozen
Arctic seas. They do not link the region to the world around it. The Amu
Darya and Syr Darya flow from the high Pamir and Tianshan Mountains
into the Aral Sea, embracing a desert, and no fertile crescent, between
them. Far to the west, only the Volga connects via the Caspian Sea with the
Middle East, and to the east, the Liao River of Manchuria supported agrar-
ian settlement. The other main rivers are not very effective transport arter-
ies. The major Siberian rivers, like the Ob, Yenisei, and Lena, flow north
into the Arctic Sea. The Russians were able to cross Siberia by following
their tributaries as they branched out to the east and west, but at the cost of
many difficult land portages. The lakes, too, though very large, isolated
among deserts and high mountains, generally lead nowhere. But along
stretches of certain rivers, agrarian settlers could interact with nomads.
The location of these rivers had the greatest effect on the builders of the
Zunghar empire, who had to obtain agricultural resources from small river
valleys and oasis production. The Ferghana valley supported sizable cities,
such as Kokand and Andijan. Farther north, the Ili River valley connected
with Lake Balkash, which linked the Zungharian steppe with the Kirghiz-
Kazakh steppe. In Mongolia, from the region of modern Ulaan Baatar, the
Orkhon and Selengge rivers flowed north into Lake Baikal, and the Kerulen
River flowed east into Hulun Lake in modern Inner Mongolia. Each of the major steppe empires put its political and spiritual base in
one of these river valleys.
26 The Selengge and Orkhon rivers served as the
headquarters for the Eastern Turks, the Uighur empire, and the later Mon-
gol empires of the Oirats. Ögödei, son of Chinggis Khan, built the Mongol
empire’s central capital at Karakorum on the Orkhon River. After the cap-
ital collapsed, the site still remained sacred to the Mongols. The monas-
tery of Erdeni Zu, built in 1586, became the headquarters of the chief Bud-
dhist cleric of Mongolia. The Liao River valley was the core region of
environments and state building 23

the Manchurian-based empires of the Khitan Liao, the Jurchen Jin, and
the Manchu Qing. The northwest Gansu corridor (especially the irrigated
lands of the Yellow River around Yinchuan) served as a base for the Tangut
Xixia kingdom.Vast distances separated all of these small valleys, and they pointed in
different directions. Except for Chinggis Khan, no steppe ruler ever con-
trolled all of them simultaneously, nor did any single settled empire until
the Qing. For would-be empire builders, the Liao River was the most prom-
ising, both because of its fertile agricultural lands and because there were
few physical barriers to moving south into the North China plain. The
Selengge-Orkhon region was the farthest removed from any settled empire.
Empires here benefited from isolation, but they had few available resources
beyond the steppe to draw on nearby, except for the fur hunters of Siberia.
The Zunghars, when they were based here, had to expand in all directions
at once: east and south toward Manchuria, north to Siberia, and southwest
across the Mongolian steppe, aiming for Ferghana. Ferghana, or ancient Turkestan, had been the site of many oasis king-
doms, periodically overrun by steppe conquerors like the Qaraqanids and
Timur. Prosperous both in agrarian resources and in caravan trade, it
linked directly east to the cities around the Tarim basin and west to Iran
and the Silk Road. Naturally it was an attractive target, and the Zunghars
early on made attempts to extract wealth from it. The Qing conquerors,
however, after defeating the Zunghars, took the east Turkestan Tarim basin
cities of Kashgar and Yarkand, but did not immediately push over the
Tianshan-Pamir Mountains to the richer settlements of Kokand, Ferghana,
Bukhara, and Samarkand. Qing armies invaded Kokand in 1830, but then
withdrew, because in the eighteenth century the Qing court had made a de-
cision to stop at this point. The ancient links between these cities, and their
unified integration in previous empires, reminds us that there were no “nat-
ural” limits to Chinese expansion in this direction, despite the towering
peaks dividing east and west Turkestan. Manchu, and later Chinese, rulers
consistently tried to keep control of Xinjiang, despite huge costs, but lim-
ited China’s borders to those defined in the mid-eighteenth century. They
stopped their advance not because of inevitable geographic constraints or
the expression of essential Chinese territorial claims, but for contingent po-
litical and cultural reasons.
Richard Frye writes that “the history of [Central Asia] is primarily one of
oases, large and small.”
27He overstates the case, but oases undeniably pro-
24 formation of states

vided the primary resources for the settled populations of the region. Oasis
settlements are “perpendicular civilizations.”
28 They are settled societies,
but very different from those of great agrarian civilizations, where village
populations spread across a broad landscape. The Turkestan oasis towns
were independent, self-sufficient units. Their irrigation water came from
melting snow in the Tianshan Mountains, which also contained small pas-
tures. Agriculture in the valley depended on channeling the snowmelt down
to fields below, where high temperatures allowed a long growing season.
These prosperous farmers could support an urban population, but trade
and contact with the outside world were extremely difficult. Only the pass-
ing caravans of the Silk Road traders, a completely distinct society, linked
the oasis communities with one another and with the outside world. The oasis communities were stable but fragile. Prolonged drought dried
up the river sources, forcing valley people to move up into the mountains.
Drought that killed mountain pastures prompted mountain pastoralists to
raid the towns below. Such disturbance of this miniature pastoralist–settled
symbiosis quickly allowed the desert sands to move in. The abandoned cit-
ies of Turkestan and Zungharia that litter the desert testify to the vulnera-
bility of the oasis communities to political and ecological change. Nevertheless, the oasis dwellers accounted for the primary concentra-
tions of agricultural and commercial wealth in this vast landscape, focal
points of attention of all empires, nomadic and settled. When possible,
they sought protection and support from powerful neighbors. Turfan and
Qomul (Hami) in the east, the closest major oases to China, often entered
into tributary relations with the empire. The presence in Turfan of two
abandoned garrison towns from the Han and Tang dynasties reflects Chi-
nese imperial interest in controlling the region. As a gateway to the steppe,
far beyond the Great Walls, Turfan was a valuable security resource for ex-
panding Chinese dynasties. Other oases, farther away, could not count on China’s protection. They
entered into relations with nomadic conquerors, often serving as their ma-
jor tax base and administrative center. Khocho, near Turfan, and Besh-
balik, near modern Ürümchi (Ch. Wulumuqi, Urumqi), became Uighur
principalities after they lost Mongolia. Timur made Samarkand into one
of the most flourishing urban centers of the fifteenth-century world. The
symbiosis of nomadic warriors and oasis cultivators benefited both when
the warriors brought their wealth back home and attracted merchants to
their base. But when the nomadic empires contracted, the oases could easily
revert to isolation, or abandonment. We seldom learn the fates of the peo-
ple who left their oasis homes, but we know that nearly the entire Turfani
population sought refuge within Qing borders for several decades in the
environments and state building 25

eighteenth century. They were, fortunately, able to return, but other peo-
ples disappeared into the sand.
For Fernand Braudel’s Mediterranean, “mountains come first,” but the
mountains of Central Eurasia do not determine its borders or its overall cli-
29 The distance of the entire region from the ocean creates its conti-
nental climate, with extreme temperatures in summer and winter. Precipita-
tion and evaporation define the east-west climatic zones. Mountains, by
blocking moisture-bearing winds, create regional features, like deserts and
oases, but no single mountain barrier marks off the region. One great series of mountain chains runs northeast to southwest from
Lake Baikal and Manchuria to the Hindu Kush.
30 Although some of the
ranges are 3,000 to 4,000 meters high, with peaks over 5,000 meters, there
are large gaps between them, especially from east to west. Along the Ili
River valley, nomads and travelers moved freely back and forth, and farther
north the Zungharian gate opened into the Kazakh steppe. The mobility of
nomad armies for a long time frustrated the efforts of the Qing empire to
confine them or hunt them down. Mountains and forests offered them ref-
uge but did not block their movement. The Zunghars learned how to use
forests to blunt the impact of Qing artillery fire. Mountain ranges delimited the southern boundary of the region more
closely, dividing it into fragments. The great ranges of the Caucasus, the
Elburz of northern Iran, the Pamirs, and the Tianshan, marked off the
steppe to the south. Farther east, the Tianshan range divided Xinjiang into
two very distinct sections. The Kunlun Mountains sealed off Tibet from the
north, and the Himalayas sealed it off in the south. The Altai and Sayan
ranges divided Mongolia from Xinjiang to the south and Siberia to the
north. The Khingan range marked off Manchuria from Mongolia. The
Urals, perhaps the range best known to Europeans, were the least sig-
nificant physical barrier; only by cultural convention did they become the
line demarcating European from Asiatic Russia. Nevertheless, each of these
ranges contained passes and gaps. India, Afghanistan, and Iran maintained
constant cultural connections with the Central Eurasian peoples. A huge zone of deserts stretches from the Caspian Sea to the Gobi and
Ordos in Mongolia.
31 These deserts, too, break up the region. Their ex-
treme aridity eliminates inhabitants from the core, allowing settlements
only on the periphery. The Taklamakan desert makes life possible in south-
ern Xinjiang only on the oases on the rim of the Tarim basin, as the Gobi
divides Mongolia in two. The deserts did not, however, block movement
26 formation of states

entirely. They could all be crossed by caravans linking the oasis towns; so
they did not form impassable barriers, but they did prevent close integra-
tion of the settlements on their edges.These physical barriers broke up the unity of culturally similar peoples,
although they were not the sole determinant. The Mongols of Inner Mon-
golia and the Khalkhas of eastern Outer Mongolia were drawn ever closer
into the Chinese orbit, while the Oirats in the west, separated from them by
vast empty spaces, never succeeded in re-creating the Mongol empire of
Chinggis. Cultural unity and disunity were, however, a product of politi-
cal strategy, not geographic determinism. Manchu frontier officials espe-
cially knew how to exploit divisions among the Mongols. By selectively
tying the Eastern Mongols closer to the Chinese economy, they promoted
increasing development of trade and settlement in Mongolia from the sev-
enteenth through the nineteenth century. Geography yielded to commer-
cial penetration.
Forests extended for 6,000 miles across northern Eurasia, from Manchuria
through Siberia. The coniferous forests of the taiga “form the most exten-
sive tree cover in the world.” The climate is subarctic, and much of the soil
is permafrost. Although elk, deer, bear, lynx, and other large animals live in
the forest, the small fur-bearing animals—sable, fox, ermine, marten, and
squirrels—proved to be the most valuable forest product for the Russians
who moved there.
In the prehistoric past, large forest zones covered much of the North
China plain during warmer and wetter climatic periods; but since the defor-
estation of the plain in the Han dynasty, large extensive forests remained
for the most part confined to Siberia and Manchuria.
33 The peoples of
the forests practiced small-scale hunting and gathering or agriculture with
primitive tools, and remained “on the margins of world history,” primarily
because of the low productivity of farming in such a cold climate.
forests on mountain uplands provided shelter for pastoralists who moved
to higher altitudes with their herds in the summer. Like the oasis peoples,
the forest peoples remained confined to small horizons; they never built big
political structures on their own, and they were victims of pressure from
the expanding empires around them. Although the forest peoples do not figure as major actors in this story,
they provided the resources for the state builders who contended over the
Eurasian steppe. The native peoples of Siberia, who paid tribute in furs, at-
tracted all three greedy empires, who demanded iasak,or tribute, from
environments and state building 27

them. Without the fur-bearing animals of Siberia and Manchuria, the Rus-
sian adventurers, traders, and military colonists would never have been in-
terested in expanding to the east, and the Manchus themselves, expert
hunters in their own right, could never have built a state.
We may divide the steppe into three horizontal bands: wooded or forest
steppe in the north, pure grasslands in the middle, and desert steppe in
the south. The central grasslands extend from the Ukraine, the northern
Caucasus, southern Urals, and Kazakhstan through eastern Mongolia and
Manchuria, including the high grasslands of Zungharia and the valley of
the Ili River. Rich black soil under the continuous belt of grass supported its
annual regrowth, despite the cold and arid climate. The grass belt allowed
fluid movement by the nomads across the vast distances east and west.
Thus, in the early seventeenth century the Torghuts could break off from
the Western Mongols and move 3,000 kilometers west to the banks of the
Volga without ever leaving the steppe. They undertook an extremely dif-
ficult return to their homeland in the late eighteenth century for political,
not environmental, reasons.Over five thousand species of plants grow in the Central Eurasian steppe.
All of them must withstand aridity and violent climatic changes. They ripen
early and quickly in the spring, and usually go into dormancy in the sum-
mer and winter. Not all of them are suitable for grazing animals, of course,
and the short blooming season means that much of the grazing must be
done in the spring. “Livestock are always moved toward a perpetual
spring”: pastoralists drive them up to alpine meadows in the summer and
down to lower elevations in fall and winter.
35 The rigid restrictions of cli-
mate and pasture defined tight limits on nomadic mobility, despite the ap-
pearance of freedom. Access to pasture required close attention to weather,
geography, and the needs of animals, as well as enough organization to
ward off invaders of the tribal pastures. Pastures left dormant could easily
be raided by others. The boundaries between grasslands and forests shifted north and south
over prehistoric time in tune with the changing climatic parameters of tem-
perature and moisture. Tundra, desert, grassland, and forest-steppe domi-
nated northern and northwest China during the first half of the Wurm gla-
ciation (70,000 bp–40,000bp), but as the climate warmed from 40,000
to 25,000 bp, forests spread into Manchuria and the North China plain.
Colder, drier climate from 25,000 to 15,000 bpdrove the forests out, re-
placing them with treeless dry steppe. Warming followed again around
28 formation of states

12,000bp, and broadleaf forest spread once more through North China,
but a drier climate after 11,000 bpbrought back grassland to replace forest
in parts of northern and northeast China. Clearly the ability of agriculture
to flourish in the borderland forest-steppe zone depended heavily on the cli-
matic conditions, and the frontier was always much more vulnerable to
dryness than the North China plain. The earliest agricultural sites in eastern Inner Mongolia date from 5300
bce, during another warming period, following cold conditions from 8000
to 6000 bce. Forests spread through the Western Liao River area, and three
successive Neolithic agricultural regimes followed, each increasingly linked
to developments in North China. Agricultural tools, ox and sheep bones,
and finally the appearance of foxtail millet, the main grain of the northwest
environments and state building 29
Grasslands of Mongolia. One of the photos taken by Frederick Wulsin during his ex-
pedition of 1923, and colorized in Beijing.
[To view this image, refer to
the print version of this title.]

around 2000–1000bce, indicate the culmination of agriculture in this re-
gion during a time when mild winters and moist, temperate climate flour-
Around 1000 bcethe climate turned cooler and drier, agriculture de-
clined, and a pastoral nomadic culture displaced the settled farmers in the
Western Liao River. Pottery declined in quality, horses appeared, and pigs
disappeared. Cultural influences now came from the north and west instead
of from the south and east. Hunting and fishing cultures may also have
evolved toward mobile hunting and animal husbandry. The shift in cultures
in this region corresponds to the appearance of references to the Rong and
Di tribes in historical sources.
Asian nomadism first evolved in the thirteenth century bcein the Upper
Yenisei River, north of the Altai Mountains. From there it moved south-
east and southwest into China’s borderlands and across Central Eurasia.
Colder, drier climates disfavored agriculture in marginal areas but favored
mobile pastoralists living off grasslands and herds. Thus, over the long
term, ecological forces underlying two sharply distinct cultural regimes
pushed nomads and settlers back and forth. The steppe nomads originated at the margins of the steppe, not in the
heart of it. Owen Lattimore argued that some may have come from rein-
deer herders and seed gatherers in the Lake Baikal forest region who moved
south, shifting from dependence on reindeer to the steppe assemblage of
sheep, cattle, horses, and camels. Others might have originated in oasis
peoples who went hunting in nearby mountains and grasslands, eventually
domesticating animals at the outer edges of oases. A third source was Chi-
nese peasants at the edge of the loess soil region in Gansu-Shaanxi, who
eventually gave up on farming and turned to pastoralism.
Harold Peake and Herbert Fleure have noted the contrasts between the
northern steppe (the grasslands extending from Hungary to eastern Man-
churia) and the southern steppe (from western India through Persia and
Arabia to North Africa).
37 The inhabitants of the southern steppes, from
prehistoric times, had been in contact with irrigated agriculture and the
earliest urban civilizations. They raided the settled cultivators, but just as
often traded goods with the towns, where they picked up the central reli-
gious ideas of the region, especially the belief in a supreme deity. Their pri-
mary animal was the one-humped camel, not the horse, whose great de-
mand for water limited its spread through the desert. By contrast with the south, in the north nomads had less contact with
settled cultivators and traded less with cities. The northern steppes are
walled off from the ancient Middle East by the linked mountain ranges of
the Sayan, Altai, Tianshan, Elburz, Pamir, and Hindu Kush. Much of the
30 formation of states

land to the south is high plateaus, 500 to 1,000 meters in altitude. For the
people of the north, the horse was far more important than the camel be-
cause of its ability to endure cold. Its demand for water was not such a
problem in colder climates. The two-humped Bactrian camel became the
main pack animal of the Silk Road for the caravan traders but not the ani-
mal of the nomad.Peake’s division of northern and southern steppe resembles Joseph
Fletcher’s distinction of “desert” and “steppe” nomadic habitats. As
Fletcher writes, “The desert nomad understood agrarian cultivation and
urban society.” When the Turks moved across Eurasia, they followed paths
prepared previously for them by the Arabs. They came in gradually, in
small groups, and did little damage to the settled societies. They soon as-
similated and adopted Islam. The (northern) steppe nomads, the Mongols,
lived apart from settled peoples. Geography here separated the two worlds
instead of uniting them. On the Chinese end, “the steppe-sown dichotomy
was sharper than anywhere else in the Eurasian steppe,” and “Mongolia
and China confronted one another throughout much of history as worlds
38 The great ecological separation, along with the suddenness of
the Mongol conquest, explains the tremendous destruction inflicted by
environments and state building 31
Camel caravan crossing Mongolia. Photo taken by Frederick Wulsin, 1923.
[To view this image, refer to
the print version of this title.]

Chinggis compared to other steppe invasions. Eastern Eurasian nomads
and Chinese, for much of their history, regarded each other with mutual
contempt, literally as animals: the Chinese used “pig” radicals on the
characters for barbarians of the northwest; the Mongols, when they con-
quered the North China plain, regarded Chinese peasants as “herds” to be
driven out.Farther east, however, Manchuria linked the steppe and settled worlds.
Although the region was mainly populated by hunters and fishermen, no-
mads repeatedly conquered the region to use its resources for their state. In
the thirteenth century, the Jurchens extended their control south to the Chi-
nese settled population, combining both the settled and nomadic worlds
under a dual administration.
39The Qing rulers drew explicitly on the expe-
rience of these ancestors for administering both steppe and settled worlds. In sum, in the northern grasslands, and especially in eastern Eurasia, the
focal point of our story, the steppe and the sown diverged sharply, and the
nomads remained for a long time quite isolated from both settled agricul-
tural and urban civilizations. Their main contact with them was through
raiding. Conversely, the settled Chinese peasant and urban-centered official
alike often saw the nomad as utterly foreign and hostile, not a dynamic ele-
ment in the civilizational complex, but an alien threat to be bought off,
walled off, or driven out. The Manchus, like their predecessors the Jurchen,
occupied the key bridge between the two worlds.
What we now call “Xinjiang” is a recent creation. The Qing created the
name “Xinjiang” (New Frontier) during their eighteenth-century con-
quests. It did not become a province of the empire until the late nineteenth
century. It has no essential geographical unity; as noted earlier, its topogra-
phy fragments the region. It was a conglomerate of different cultures, ecol-
ogies, and peoples, mostly separate and oriented toward their local envi-
ronments. Yet this region is at the center of our story. Xinjiang lies directly across both northern and southern steppe zones.
North of the Tianshan, the Zungharian steppe acts as a gateway connecting
the plateaus of Mongolia with the lowlands to the west. The Altai Moun-
tains bound it on the north and northeast, but a corridor leads east into
Mongolia north of the Gobi desert. The center of Zungharia is desert, but
grasslands ring it on the North and South. The Ürümchi oasis, now the cap-
ital, but always an urban center under different names, lies at the southern
end of the basin, on the northern edge of the Tianshan. The northern
branch of the old Silk Road leads from here east via Hami to Jiuquan in
32 formation of states

Gansu province, and west through the Ili valley to Kuldja (modern Yining).
To the west, the Tarbaghatai Mountains and Zunghar-Alatau ranges run
east and west, leaving an opening into the Kazakh steppe. Major nomadic
empires always tried to occupy the Zungharian basin because of its large
expanse of grasslands and its productive oasis towns. The Zunghars, who
gave their name to it, were only the last in a series. Conversely, until Qing
armies could reach the Zungharian basin, they could not destroy the
Zunghar state.
Trade, Transport, and Travel
Humans learned to domesticate the horse around 4000bcein the southern
steppes of Russia.
42By 1950 bcethe first horse-drawn chariots appeared in
western Asia, and by 1200 bcethe Chinese were using chariots in warfare.
Riding on horseback took longer. The first mounted warriors, the Scythians
and Cimmerians, appeared in the ninth century bce, and soon invaded As-
syria but diffused only slowly to China and Mongolia. Large horses derived
from Arab stocks were bred in northern Iran, Kazakhstan, Turkmenistan,
Uzbekistan, Tajikistan, and Kirghizstan in the first millennium bce, but the
basic stock of the east Eurasian warriors, the small pony or “Przewalski’s
horse,” did not appear in Mongolia until the fifth century bce. Within a
century, these horsemen had become a powerful presence on China’s north-
west frontier.
43 Possession of the horse turned the pastoralist from a mere
herder into a powerful predator. He could raid sedentary societies for other
goods, or simply use the threat of raids to extort better terms of trade.
Horses demanded greater areas of pasture than the other grazing animals;
acquiring them was an investment in a capital good, which brought greater
returns by enabling conquest of other pasturelands and better terms of
trade from settled societies. When horse-riding nomads entered regions without sufficient pasture,
they dismounted. Their use of the horse demonstrates adaptation to dif-
ferent environments. The Huns, for example, once they crossed the Car-
pathian Mountains, found a terrain lacking the large pastures of Central
Eurasia. By the fourth and fifth centuries ce, their battles with the Romans
were more like pitched infantry battles than swift cavalry raids. The Hun-
garian Alföld, by Lindner’s estimate, contained 42,400 square kilometers,
which could support at least 320,000 horses, assuming that one horse re-
quires 25 acres (10.1 hectares) of pasture. But allowing for the presence of
other grazing animals, as well as the presence of forests and marshes mixed
with pasture, limits the number of horses to about 150,000, enough to sup-
environments and state building 33

port at most 15,000 mounted warriors—less than two Mongolian divi-
sions. As Rudi Lindner writes, “Hungary is not Mongolia. To expect the
Huns to have retained the domestic economy of the steppe once they had
reached the Danube is to reject the role of ecology in history.” Thus the
Huns turned sedentary when they entered Europe, and “the Carpathians
mark the far western frontier of the history of nomadism.”
The horse was both the mainstay of the nomadic economy and the one
essential element in warfare which the sedentary civilizations could not
breed in sufficient numbers for their own needs. Thus it was the key linch-
34 formation of states
The “precious and auspicious” horse Baoji. The Torghut Mongols presented this
horse to the Qianlong emperor after their return from Russia, and the Czech court
painter Ai Qimeng (Ignatius Sickeltart) painted it in 1773.
[To view this image, refer to
the print version of this title.]

pin linking steppe and sedentary peoples. Ever since Herodotus commented
on the Scythians, historians have realized that “in combat the [nomadic]
horse always puts to flight the horse of the enemy.”
46Although, in princi-
ple, the nomads could be a self-sufficient society, they rarely chose to live
solely off the steppe. They wanted to trade with settled peoples for luxury
goods, but they often traded in more common products such as linen, tea,
and grain.
47 Likewise, the settled peoples could do without steppe trade
only if they were willing to endure constant raiding. The trade of rare prod-
ucts for horses seemed advantageous to both sides. Sechin Jagchid argues
that peace was always possible with the nomads when the Chinese satisfied
their needs for trade, but the Chinese often “failed to discover that poverty
and famine caused the nomads to invade China to supply their needs by
48But several factors made the negotiation of trade precarious. Re-
peatedly the Chinese dynasties set up markets for trading tea for horses, but
all were problematic.
49First, not only in China but also in Europe and the
Byzantine empire, critics found it humiliating to pay extortionate prices to
the barbarian raiders.
50Second, on the nomadic side, stable trade relations
required a leader who could negotiate and enforce agreements to prevent
border raids while ensuring a reasonable supply of horses.
The Chinese learned very early the critical importance of the supply of
horses to military success. The first clear account of the use of mounted
men in China appears in the story of Zhao Wuling’s adoption of nomadic
cavalry warfare in 320 bce.
52Efforts to obtain horses by other means than
trade with the nearest enemy account for the first major Chinese expedition
into Central Eurasia. Han Wudi in 104 bcesent an army of thirty thousand
men to Ferghana to capture the famous “blood-sweating horses” of the
Ferghana valley. This first expedition failed, returning with only 10 to 20
percent of its men. The second expedition, with sixty thousand men, suc-
ceeded in getting the horses, but again the cost in men and resources was
high. Only ten thousand men returned home.
Later dynasties tried to breed horses within the empire but succeeded
only briefly. The Tang were able to build a supply of up to 700,000 horses
through an extensive breeding program but still had to rely on extensive
imports from Samarkand. They traded large amounts of silk with the Turks
to obtain them.
54The rebel military governor An Lushan had much greater
success than the Tang central government in supplying his forces, because
he could pick the best cavalry horses available on the northwest frontier.
After An Lushan’s rebellion in 755 blocked Tang access to the northwest,
the rising price of horses drove the government deep into debt. The finan-
cial crisis created by the closing of the Silk Road brought attacks on Bud-
dhism in the ninth century, including the melting of statues for coin. There-
environments and state building 35

after the Chinese always depended heavily on Mongolia for their horse
supply. The Song and Ming mainly used the officially supervised tea-for-
horse trading markets, even though these produced poor-quality animals at
exorbitant prices. Chapter 3 describes the difficulties with frontier trade in
the Song and Ming. This experience informed the very different approach
of the Qing.The Manchus were well aware of their predecessors’ dilemmas. They
took special care to develop close relations with the Eastern Mongols be-
cause of their desperate need for reliable horse supplies for the northwest-
ern campaigns. Inducements of trade, titles of nobility, and grain for famine
relief all served to bring over cooperative Mongol Khans to the Qing side.
In return, they owed levies of horses, men, and supplies when demanded by
the Qing. The 1757 rebellion of the Eastern Mongols is the clearest demon-
stration of the heavy burden placed by the Manchus on their Mongol sub-
ordinates and of the critical importance of allied Mongols in the campaign.
We know much less about the role of the other four essential domesticated
animals in nomadic society: sheep, goats, camels, and cattle. None of them
had the prestige of the horse, because none of them were suited for warfare,
but they were economically much more fundamental. Owen Lattimore in
the 1940s found that pastoralists in Xinjiang owned 11.7 million sheep and
goats, 1.55 million cattle, 870,000 horses, and 90,000 camels.
55Sheep and
goats were the essential animals for subsistence, and the nomads stripped
them of everything they produced. They put a heavy burden on the grass-
lands: one sheep needs 5 to 10 hectares (twelve to twenty-five acres) of pas-
ture, and consumes a harvest of 45 to 180 pounds of dry matter per acre.
Sheep and goats monopolized their pastures because they grazed the grass
too close to leave anything for horses. Horses had to graze separately.
Even if horse-riding warriors looked more dashing, any ruler aiming to
increase his influence had to pay attention to his sheep. Batur Hongtaiji, the
founder of the Zunghar state, became known as the “sheep-raising king”
because of his efforts to build up his herds.
58Since the Qing armies and set-
tlers also needed sheep, the Zunghars were able to sell them profitably at
regulated markets.
59 Sheep were the basic source of meat for frontier set-
tlers and nomads alike. The severe overgrazing which has now damaged
many of Inner Mongolia’s grasslands has long historical roots. The size
of herds began to increase when the Zunghar state and the Qing armies
fought each other in the eighteenth century.
The two-humped Bactrian camel, the chief transport animal of the desert
36 formation of states

caravans, also served the military and diplomatic missions. The huge cara-
vans crossing the Gobi desert were essential to the tea trade with Russia.
Camels brought coal to Beijing from the northwest.
61The emperors’ mili-
tary campaigns depended heavily on camels to bring supplies to distant
outposts. Camels were, however, cantankerous animals, difficult to control
and difficult to breed. Qing armies had to employ specialized personnel to
rear young animals and to drive them across the desert. The Qing con-
environments and state building 37
Goats in a village in Gansu province. Photo taken by Frederick Wulsin, 1923.
[To view this image, refer to
the print version of this title.]

quests, a multicultural venture, depended on Turkic camel drivers as much
as Mongol horsemen and Chinese infantry. Camels could even carry artil-
lery between their humps and, when they sat down, provide a platform for
firing the guns. This use of camels in warfare, depicted clearly in eigh-
teenth-century engravings of Qianlong’s expeditions, persisted through the
nineteenth century in Persia, Bukhara, and Turkestan.
Travel across this great land mass was arduous, dangerous, and slow.
In the 1920s, Lattimore obtained detailed figures on rates of travel from
Guihua (modern Hohhot) to Guchengzi, just southeast of Ürümchi, via the
Great Road. To travel this distance of 1,800 miles took 120 days for heavy
caravans, 90 days for express freight, and 70 days for travelers going by
forced march, a rate of 15 to 25 miles per day. Kangxi’s army in 1690
crossed the Gobi at a rate of fifteen miles per day. On the difficult desert
road from Guihua to Morhgujing, north of Baotou, it took 100 days to
cover 285 miles.
63 We might take these as the maximum and minimum
rates of travel across the Chinese steppes and deserts. Camels, the main
beasts of burden in desert territory, could travel only two to two and one-
half miles per hour. It was possible to travel rapidly on the Yellow River, as
the Kangxi emperor did on his return from one of his military campaigns.
The unique pifazi,or rafts supported with inflated hides, could transport up
to twenty-five tons of goods from Lanzhou to the upper bend in the river,
but this river passed only through the Ordos region. The other steppes and
deserts had little river transport.
Animals, of course, had to carry most of their fodder with them when
crossing deserts. This in turn limited the quantity of goods they could carry
and raised the cost of equipping a caravan. Nevertheless, the caravan trade
routes, some of the oldest long-distance trade routes in the world, endured
for millennia. Caravan traders could profit only from the exchange of light-
weight and valuable commodities, carried peripherally to the main bulk
of the caravan’s animal and human supplies. Chinese silk, of course, was
the preeminent commodity in value of the Old Silk Road, but China also
exported porcelain, metalware, and jade. Caravan trade was tightly con-
fined by the ecological parameters set by the desert, steppe, and oasis envi-
ronment. Three institutions were critical to it: garrisons and watchtowers
manned by soldiers to keep the peace; postal relay stations, originally es-
tablished by the Mongol empire, for rapid communications; and caravanse-
rai, to provide lodgings and trading places in the oases. Traders formed a
small group of sixty to one hundred men with a minimal number of ani-
mals and a small load of highly valuable goods. They could not afford to
travel with large defense forces. This made them perfect targets for bandits
and nomadic raiders. Only in times of relative peace could the trading cara-
38 formation of states

vans move unhindered. Trade waxed and waned, depending heavily on the
ability of settled and nomadic rulers to establish minimal order. This trade,
more than any other, was extremely vulnerable to changes in the surround-
ing political environment. Many scholars have argued that the caravan
trade declined in the sixteenth century because of new competition from
European overseas trade, but this inverts the relative importance of politics
and economics. It was rising political and military instability in Eurasia
during the sixteenth and seventeenth centuries, not European competition,
that killed the long-distance trade. In fact, the trade did not disappear but
moved north to a new secure channel, carrying furs and silk directly be-
tween Russia and China. Commerce was much more dependent on politi-
cal power than vice versa.
One of the most important transport routes led from Gansu province to
the northwest, through a narrow corridor between deserts, to the oases of
environments and state building 39
Raft on the Yellow River. Wulsin, like the Kangxi emperor and many others, used
this traditional form of raft to travel down the river.
[To view this image, refer to
the print version of this title.]

Turkestan. 66 This “Imperial Highroad,” as Lattimore called it, was much
more stable than a caravan route. All Chinese dynasties which could con-
trol the region maintained this road, but its final form was a product of
Qing security needs, a major highway built to provide supplies for the gar-
risons conducting campaigns on the frontier. Wheeled vehicles could travel
part of the way faster than camels and donkeys; military quartermasters
brought large numbers of carts here from interior China. Transport costs
still remained extremely high: shipping grain from Hexi, the agricultural re-
gions in Gansu west of the Yellow River, to Hami, the nearest oasis, raised
its price by a factor of ten.
67Road construction helped to reduce transport
costs, but efforts to increase agricultural production in the oases themselves
worked better. The difficulty and cost of land transport thus stimulated ef-
forts to develop the local agrarian economy. Military expeditions by settled regimes into the steppe could defend
themselves easily against nomadic raiding, but at heavy cost. Their struc-
ture was the inverse of the merchant caravans. They proceeded more
slowly, but with large numbers of men and animals, including replacement
transport. Often they brought sheep and cattle with them for meat supplies.
They valued neither profit nor speed but the presence of numbers. Usually
they followed the existing trade routes, but sometimes they had to leave the
main roads in pursuit of nomad enemies, leading them into difficult des-
ert crossings. Military supply demands overwhelmed the modest merchant
caravanserai, so they had to pitch separate camps. From the Han dynasty
up to the mid-Qing, no Chinese army could last more than ninety to one
hundred days in the steppe. This logistical barrier established a fundamen-
tal limit on the ability of Chinese expeditions to penetrate Central Eurasia.
Until the Qing, almost none of these aggressive campaigns succeeded. As
we shall see, nomad military strategy took advantage of the logistical limi-
tations of Chinese steppe incursions. Pilgrims, ambassadors, spies, diplomatic envoys, and marriage partners
(generally Chinese women sent as spouses to win over nomad rulers) also
traveled these arduous routes. Even though they were not the dominant po-
litical or economic actors, many of them left valuable accounts about Cen-
tral Eurasia. Chinese emperors are another special kind of steppe traveler.
The best known are the Yongle emperor (r. 1402–1424), who conducted
five campaigns in Mongolia; the hapless Zhengtong emperor, captured by
Mongols in 1449 near Datong after the failure of his military expedition;
and the much more successful Kangxi emperor. Kangxi’s letters to his son
while he was on campaign deserve to be included among the most striking
and vivid travel narratives in Chinese literature.
40 formation of states

The Frontier Zone
Fernand Braudel writes: “The question of boundaries is the first to be en-
countered; from it all others flow. To draw a boundary around anything is
to define, analyse, and reconstruct it, in this case select, indeed adopt, a phi-
losophy of history.” To which David A. Hollinger would add: “But not all
exclusions are bad, the conventional wisdom of our time will be quick to
remind us, and we are all left with the responsibility for deciding where to
try to draw circles, with whom, and around what.”
The borderlands between the core of China and the farthest nomadic
pastures were a zone of frontier interaction, a “middle ground” where peo-
ples following radically different ways of life adapted to one another and to
the environment.
69Because the steppe was filled with people constantly on
the move, all those moving through it had to adopt to some extent the cus-
toms of the nomads, those best suited to life in the steppe. Chinese armies
ate more meat than they would at home and got used to traveling with
herds of animals. They had to leave their forts and stay in tent encamp-
ments. They used horses and cavalry forces much more than in the interior
and had to deal with recalcitrant camels and mules instead of docile oxen
pulling carts. Merchants and other professional caravan men likewise
changed their ways on the frontier. As Lattimore noted during his trip
through Mongolia in 1926–27, 90 percent of the caravan men were Chi-
nese who had cut their links to settled fields, ancestral homelands, and
heartland customs. On the caravan routes, they made offerings to gods of
fire and water, not ancestral deities; for clothing, food, and drink, they re-
lied on sheep, not pigs and chickens.
70The trade frontier was a social space
in which core ethnic identities had to bend to fit rigorous geographical con-
ditions. Mongols on the caravan routes had to put up with more restricted
mobility than they could gain in the broad pasturelands; Chinese became
accustomed to a life of much greater wandering than their settled peasant
confrères in the interior. The frontier zone was a liminal space where cultural identities merged
and shifted, as peoples of different ethnic and linguistic roots interacted
for common economic purposes. Most Han Chinese officials found this
environment hostile, abhorrent, and alien. Their Manchu and Mongol col-
leagues did not find it so strange. The idea of Chinese turning native, aban-
doning the essential elements of civilization and preferring a mobile life,
shocked established powers but attracted others. The story of the eigh-
teenth-century Qing empire is of an effort to seal off this ambiguous,
threatening frontier experience once and for all by incorporating it within
environments and state building 41

the fixed boundaries of a distinctly defined space, and by drawing lines that
clearly demarcated separate cultures.North Americans and Europeans had similar experiences. The peoples
on either side of an ambiguous border often have more in common with
each other than with the heartland of the nations they belong to. Frontier
peoples have ambiguous loyalties; they share a resentment of the economic
and political advantage of the centers that rule them.
71 Frederick Jackson
Turner saw frontier settlement as the penetration of empty space by Anglo-
Saxon settlers, a process that reaffirmed the quintessential Americanness of
the pioneers. By contrast, new historians of the American frontier empha-
size the special regional characteristics of the West, the persistence of the in-
digenous inhabitants in the face of the Anglo-American invasion, and the
formation of new hybrid identities with ambiguous loyalties to the metro-
politan centers of the East. In both China and the New World, power-
ful empires based in eastern metropolises incorporated large, sparsely pop-
ulated arid western zones. Although the conquerors pretended that the
newly conquered regions were empty of civilized peoples, contact trans-
formed both the metropole and the frontier. The most conspicuous display of demarcation in China’s frontier zone,
of course, is the “Great Wall,” or more accurately translating the Chinese
word changcheng, the “Long Walls” that many Chinese dynasties built on
the northwestern frontier. Arthur Waldron has shown that, contrary to
Western and Chinese myth, the “Great Wall” is not thousands of years old.
Not until the sixteenth century did the Ming dynasty construct a single,
nearly continuous defensive barrier.
72 Yet for many centuries the goal of
cutting off the steppe from the settled zone inspired Chinese frontier policy.
In Lattimore’s interpretation, the Great Wall was built not to keep the no-
mads out but to keep the Chinese in. It “was an attempt to establish a per-
manent cultural demarcation between the lands of the nomad tribes and
the lands held by settled people.”
73No other settled empire on the periph-
ery of the steppe ever tried to create such a sharp demarcation between the
steppe and the sown, although there is a partial Russian analogue in the
seventeenth century.
74 For most of Chinese history these efforts, in fact,
failed militarily. Since Chinese efforts to demarcate the frontier never suc-
ceeded, the zone was never stabilized. It always contained transitional so-
cial groups—sinicized nomads, semi-barbarized Chinese, Tibetans, Mus-
lims, and other non-Han peoples—mixtures of merchants, nomads, oasis
settlers, and peasants. In Lattimore’s words, “China could never put an end
to the ebb and flow of frontier history and maintain the civilization of
China in the closed world that was its ideal.”
Only after the mid-Qing, when the wall became militarily irrelevant, did
42 formation of states

it acquire effective symbolic value as a cultural marker among both Chinese
and Westerners.
76The history of China’s Long Walls highlights the interac-
tion between cultural definition and geography on the frontier. The Qing
conquest likewise relied on both symbolic and ecological manipulation to
define its achievements. As Qing officials cleared land, settled peasants, and
drew maps, they moved both real sand and lines in the sand.
Even the Qing conquests did not settle things for long. Stabilization of
the frontier zone has never proved permanent. The Sino-Russian treaty ne-
gotiations of 1689 seemed to demarcate a clear boundary across the Eur-
asian continent. But in the mid-nineteenth and early twentieth centuries,
when China’s defenses weakened, the Russians transgressed the border line
in the Ili crisis and built the Far Eastern Railway through Manchuria. Con-
currently, the British pushed their interests in Tibet, motivated by fear of
Russian expansion. Strategic analysts dubbed this geopolitical competition
the “Great Game.” By the early twentieth century, however, the exagger-
ated British fears that Russia intended to threaten India seem to have
subsided, and the frontier stabilized again. Treaties included the Anglo-
Russian settlement of the Pamir boundary in 1895, the Anglo-Russian con-
vention of 1907, by which both parties agreed not to penetrate Tibet, and
Russian agreements with China over Outer Mongolia in 1911, in which,
like the British in Tibet, the Russians agreed to representation in Mongolia
but not occupation.
The “second thirty years’” war of the twentieth century, however, from
1914 to 1945, destroyed this and all other colonial agreements. When the
dust settled, Central Eurasia still was divided in two, now by the cold war.
Although for a short time Eurasia seemed to fall under a monolithic com-
munist bloc, by the 1960s the Sino-Soviet split established a line across the
continent not too different from the border of 1689, except that Outer
Mongolia, part of the Ili valley, and northern Manchuria now belonged to
the Soviet Russian sphere. This division, in turn, lasted only thirty years.
With the collapse of the Soviet Union, five independent states in addition to
Russia and Mongolia now occupy Central Eurasia. Although the borders
of China with Russia and Central Asia have not changed, unrest in Tibet
and Xinjiang calls into question any assumptions about permanence and
stability. Instability, indefiniteness, and physiographical unboundedness
still challenge the unceasing efforts of nation-states to draw lines and settle
their peoples in immobile, fixed territorial and psychological sites. Fixing people in place territorially requires material and organizational
resources: armies, border guards, passports, visas. Fixing people psycho-
logically requires intellectual and cultural resources: nationalist symbols,
rewritten history. Both strategies contest natural human urges to move, to
environments and state building 43

change, and to evolve, but they are supported by equally natural human
urges for security, fixity, and stability. Excessive stabilization can bring re-
pression and stagnation, but excessive fluidity means chaos and anarchy.
All states, including the Qing and modern China, struggle to find the ap-
propriate balance between stability and freedom.
Isolation and Integration
David Christian has outlined how these distinct ecological zones and their
frontiers shaped the states and societies of Russian Eurasia. The two geo-
graphic features of Central Eurasia that define it as a distinct unit in world
history are its low natural productivity and its “interiority,” or distance
from oceans. Northern latitude and continental climate patterns meant
cold weather, low rainfall, and poor agricultural productivity. Five distinct
ecological adaptations in succession responded to these ecological and geo-
graphic characteristics: hunting, pastoralism, pastoral nomadism, agrarian
autocracy, and command economy. Each was designed to achieve the maxi-
mum possible concentration of resources from a region where these agri-
cultural and human resources were few and highly dispersed. Each form, in
turn, marked off Central Eurasia as very distinct from the richer settled so-
cieties around it, and these societies often regarded the natives of Central
Eurasia as utterly alien. Each of the civilized societies had its alien other in
Central Eurasia, and each evolved as a reflection of the other.
Systematic, specialized hunting, beginning with the mammoth hunters of
the Upper Paleolithic, divided early humans from the specialization in gath-
ering on the periphery. The rise of agriculture in the Neolithic period had its
contrasting counterpart in the rise of pastoralism from 6000 bce. Pasto-
ralism based on grazing herds allowed population growth and required
military mobilization to defend the herds. The warrior chieftains of the
kurgan (burial mound) cultures from the fourth to the third millennium
bce developed into full-fledged nomadic warrior societies by the second
millennium bce. As the nomads mastered the horse, settled societies, un-
able to pasture horses, had to trade with nomads to obtain horses from the
steppes. Even the first agrarian empires of Mesopotamia and Egypt had to
contend with constant raids from 2600 bce.
79 After 1000bce, nomadic
military superiority, based on rapid mobility and universal use of the horse,
dominated the steppes until 1500 ce.
Agrarian autocracy was a product of the forest at the western periphery
of the steppe. The rise of Muscovy from a conquered subject of the Mongol
empire to a powerful independent state by the seventeenth century rested
44 formation of states

on its ability to exploit brutally but effectively the extensive low-productiv-
ity agricultural regime of the forest and steppe. The steppe environment,
and the constant interaction of Russians with nomadic states to their east,
accounts for many of the military and cultural similarities between Rus-
sians and the nomads they fought. They learned much from their enemies,
as we shall see. When nomadic warriors conquered China, they brought
Central Eurasian institutions with them, mixing with those of the settled
core. By contrast, when Han Chinese from the core ruled China, especially
under the Ming dynasty, they showed very little respect for or willingness to
learn the lessons of the steppe nomads, even when these lessons could have
brought them military success. This accounts for the general impression of
a gradual shading off of Russian into nomad, as opposed to the sharp de-
lineation of Han from non-Han that is characteristic of China. As Lord
Curzon put it, in typically British terms, “The Russian fraternizes in the
true sense of the word. He is guiltless of that air of conscious superiority
and gloomy hauteur, which does more to inflame animosity than cruelty
may have done to kindle it, and he does not shrink from entering into social
and domestic relations with alien and inferior races.”
As mentioned earlier, the modernization paradigm regards Central Eur-
asia as an especially isolated region, cut off from the major trends of the
modern world by physical and cultural barriers. By contrast, in the “classi-
cal” early modern perspective it is seen as the “crossroads” of Eurasia,
linked to all the sedentary societies around it through long-standing net-
works of trade, conquest, and religious and cultural exchange. We will face
this dual paradox repeatedly. Settled civilizations on the periphery regarded
Central Eurasia as remote, peculiar, hostile, and threatening; within the
region, Central Asians came in contact with outsiders from the entire Eur-
asian continent: travelers, pilgrims, missionaries of different religions, con-
querors, traders, and explorers. The same ambivalence characterizes de-
scriptions of Central Eurasian ecology: we can view it as an extremely
isolated region, or one of the most integrated with the rest of the continent,
depending on our perspective. The impact of smallpox in Central Eurasia illustrates its paradoxical eco-
logical position. According to John R. McNeill, the isolated biota of the
Pacific Islands were unusually labile: that is, subject to sudden and unpre-
dictable change when they came into contact with outside forces. Pacific Is-
land biota had developed few defenses against the common predators of
the Eurasian and North American continents; hence the newly arrived rats,
deer, snakes, cows, pigs, and sheep devastated the existing populations of
birds and plants. Dramatic changes occurred with the first arrival of hu-
mans in the islands; even more dramatic change followed the era of Cap-
environments and state building 45

tain Cook after his Pacific voyage of 1769. In many places, like New Zea-
land, European portmanteau biota nearly completely overwhelmed native
plants and animals and people; on other islands, native peoples and plants
recovered, but in constant competition with the aggressive European,
American, and Asian invaders.
By contrast, no ocean barriers cut off Central Eurasia from the outside
world. Humans, animals, and other hitchhiking biota have for millennia
crisscrossed its land routes. Still, the common analogy of the grasslands to a
great internal sea does hold up in the ecological sense to some degree.
Deserts and high mountains block passage of many organisms, and ex-
46 formation of states
Smallpox victim, one of numerous illustrations of the disease in
a Qing imperial medical encyclopedia.
[To view this image, refer to
the print version of this title.]

treme heat and cold kill off temperate zone organisms. Most notably, Cen-
tral Asians remained nearly isolated from the European and Asian disease
pools until the eighteenth century. Then smallpox, among other diseases,
decimated the Mongolian population when it came into contact with Chi-
nese settlers, just as Native American and Oceanic populations died off af-
ter the European conquests of the New World.Mongols knew in the mid-fifteenth century that they could catch small-
pox from the Chinese, and the Chinese in turn warned them not to settle
too close to the border to avoid spreading it. The Ming dynasty held only
sporadic horse fairs where Mongols and Chinese mingled; then Chinese
bans on frontier trade in retaliation for nomadic raids had as a by-product
the effect of protecting the Mongols from infection.
82 Chinese migrants
who went beyond the Great Wall, however, could also spread the disease,
and there were more than 100,000 of them in southern Mongolia by 1590.
Still, few Mongols caught the disease under the Ming. The Manchus, before the founding of the Qing, also rarely encountered
smallpox, but they knew of its danger. Mongols and Manchus who had not
been exposed to the disease were exempted from coming to Beijing to re-
ceive titles of succession. The main response of the Mongols and Manchus
to those who did fall ill was quarantine. Li Xinheng commented that if any-
one in a tribe caught smallpox, his relatives abandoned him in a cave or dis-
tant grassland.
83Seventy to 80 percent of those infected died. The German
traveler Peter Simon Pallas, who visited the Mongols three times from
1768 to 1772, commented that smallpox was the only disease they greatly
feared. It occurred very seldom, but spread rapidly when it struck: “If
someone catches it, they abandon him in his tent; they only approach from
the windward side to provide food. Children who catch it are sold to the
Russians very cheaply.”
84The Mongols whom Pallas visited lived far from
the Chinese border, but they knew well that smallpox was highly conta-
gious and nearly fatal. The Chinese discovery of variolation—a method of inoculation—was of
great aid in reducing the severity of attacks. The Kangxi emperor himself
was selected as heir in part because he had survived the disease in child-
hood; his father had died of it. In 1687 he inaugurated regular inoculation
of the royal family, and his successor extended mandatory inoculation to all
Manchu children.
85The Manchus adopted this Chinese medical practice in
order to protect themselves against the virulent strains that were absent
from the steppe. Only Manchus who had survived the disease were allowed
to be sent to the Mongolian steppe. Mongols close to the Manchu and Chi-
nese border gradually grew immune, but those farther away suffered great
losses in the nineteenth century when Chinese penetration increased.
environments and state building 47

Disease determined critical turning points in the conflict between the
Manchus and Zunghars. Ligdan Khan, the first major Mongol rival to
Manchu rule, died of smallpox.
86In 1745, when the Zunghar Khan Galdan
Tseren died, outbreaks of smallpox caused upheaval among the Zunghars;
one report stated that 30 percent of them died.
87Another epidemic struck
Zungharia in the 1750s, just as the Qianlong emperor launched his final
campaign. The last rebel against Manchu domination, the young prince
Amursana, died of smallpox at the age of thirty-five, opening the way to the
complete conquest of Xinjiang. Wei Yuan estimated, after the Zunghars
had vanished as a people, that 40 percent of them died of smallpox—more
than lost their lives in battle or fled to Russia. The Mongols for their part tried to avoid contact with Han Chinese as
much as possible. Apparently they never learned the variolation techniques,
so their only recourse was isolation. When negotiating licensed trade with
the Qing in the 1740s, Galdan Tseren feared that his envoys would catch
the disease when they passed through Chinese territory, so he asked for per-
mission to avoid the northwestern towns of Hami and Suzhou and instead
go direct to Dongkeer. Tibetans also tried to avoid traveling in the Chinese
interior: the Panchen Lama used his lack of immunity to smallpox as an ex-
cuse to avoid an audience with the Kangxi emperor in Beijing. The Man-
chus themselves often tried to accommodate the Mongols in order to spare
their lives. The Kangxi emperor noted that many surrendered Mongols liv-
ing in the capital were dying of diseases. He pitied them because “the cap-
ital’s food and drink are against their nature,” and they were “like caged
birds and animals,” so he provided them with tents and settled them be-
yond the wall in Zhangjiakou and Guihua. When Mongol children flocked
to Kangxi’s military camp in the Ordos while he was on campaign, he
called in a special doctor to inoculate them.
So in respect to certain disease vectors, the pathogens of Eurasian settled
regimes hit Central Eurasia with devastating impact by the eighteenth cen-
tury. The vulnerability of the Mongols to smallpox is eloquent testimony to
their isolation from the germ pools of dense populations. At the same time,
their constant contacts with the Manchus and Chinese made them aware of
the danger, even though they could not prevent its incidence. The Manchus,
by contrast, could take active measures against the disease, having closer
regular contact with the Chinese, greater medical knowledge, and greater
acquired immunity. They in turn used this knowledge to inoculate Mongols
who surrendered to them, leaving those who resisted them to face the rav-
ages of the disease. The disease environment itself significantly affected
the outcome of the conflict, but the disease vectors acted through human
48 formation of states

Another conspicuous ecological impact was the invasion of the grass-
lands by the agricultural regimes of Chinese and Russian settlers. Here eco-
logical integration brought more ambiguous results, in one sense following
the inverse cycle to that of disease, in another sense similar to it. From the
smallpox bacterium’s point of view, the eighteenth century brought success-
ful colonization of new human and animal territory, followed by eventual
adaptation and acquisition of immunity by the hosts and a slowing down
of new colonization—a typical logistic curve. From the human point of
view, severe population decline from the new diseases was followed by slow
recovery after several centuries. Similarly, the new crops introduced by ag-
ricultural settlers, especially wheat and millet, had great initial success,
pushing back the grassland frontiers and greatly expanding the cultivated
area. This benefited peasant cultivators, whose population grew, at the
cost of pastoral nomads and their grassland-dependent animals. But by the
twentieth century, at the end of the logistic curve, excess cultivation in arid
regions without sufficient irrigation brought the desert back. The remain-
ing grasslands are still under severe stress, but agriculture is not prospering
The difference between the spread of disease vectors and the spread of
agriculture is that agriculture always required heavy subsidies from outside
to succeed. In such a hostile host environment, only greater investments in
water supplies, seed, animals, and tools from the state could make it work.
A glance at Central Eurasian agriculture evokes strong analogies to the set-
tlement of the American West, where the arid region west of 100 degrees
longitude could never support settlement without heavy subsidies from the
government back east.
Thus we can view Central Eurasia in the same light as other formerly re-
mote, forbidding frontiers invaded by aggressive European biosystems. The
main incursion began in the eighteenth century, but the actors were primar-
ily Chinese, and secondarily Russian. Analyses that focus only on western
European and American examples of cultural, ecological, and political im-
perialism need to pay more attention to the other expanding empires which
concurrently marched their agrarian frontiers across the eighteenth-century
world. In short, David Christian’s analysis of Russia fits the Manchus too. The
parallels are striking. In the late sixteenth century two forest peoples—the
Russians and the Manchus—began to build strong states and to expand un-
til they divided the northern steppe. In both cases, organization and politi-
cal structure had to make up for demographic inferiority. The lands of
western Europe and south of Manchuria were far more favorable to dense
agrarian settlement than were the heartlands of Muscovy and the Man-
environments and state building 49

chus. Yet both states expanded. Kinship networks bound together the rul-
ing elites; serfdom, or bond servitude, tied the agrarian producers to the
militarized state. “Autocracy, then, was a response to the difficulties of cre-
ating a powerful agrarian state in Central Eurasia.”
91 Both empires had
greater resources than the nomadic peoples of the interior, but both were
on the periphery of the settled agrarian civilizations to their west and south.
Both extracted resources from agrarian producers and used them to domi-
nate both the steppe and the agricultural zones. Both used military aristoc-
racies effectively to concentrate the limited resources of the agrarian pe-
riphery. The “Manchurian solution,” in Barfield’s terms, had been tried
twice before by the Khitans and Liao in East Asia, with only partial suc-
92 Manchus and Russians found parallel solutions to the problem of
mobilizing the resources of the northern forests and fields to dominate the
Eurasian continent.
50 formation of states

The Ming, Muscovy, andSiberia, 1400–1600
E ach of the central players in our drama—the Manchu Qing empire,
the Zunghar Mongolian state, and the Muscovite–Russian empire—had
long experience with the steppe. Although they did not directly confront
one another until the seventeenth century, the previous century set the stage
and provided the ideological, material, and political resources with which
they conducted their geopolitical game. This chapter briefly summarizes the
story that led up to the seventeenth-century conflict and highlights some of
the major issues confronting these Central Eurasian regimes in their efforts
to dominate the steppe. All narratives are selective. I begin with a brief discussion of the Ming dy-
nasty background to the clash over the steppe, focusing on themes relevant
to later analysis: the logistics of military supply, especially the interrelation-
ship between frontier trade and horse purchases; the strategic decisions for
attack and defense; and the consequences of these decisions for the struc-
ture of the state and for relations between the Han and Mongols. The Ming rulers inherited problems with frontier defense faced by earlier
Han-ruled dynasties. They practiced two different strategies, neither of
which succeeded for long, and neither of which was imitated by the Qing.
For the first half of the fifteenth century, they led aggressive campaigns
against the Mongols, as far as the Orkhon, Onon, and Kerulen rivers. The
military campaigns ended with the embarrassing capture of the Zhengtong
emperor by Esen Khan in the Tumu incident of 1449. Parallel to the mili-
tary campaigns, the Ming rulers launched logistical efforts to obtain ade-

quate supplies of horses for the frontier garrisons. They promoted govern-
ment-supervised trade in the form of horse and tea markets. As in the Song
programs of Wang Anshi, their goal was to exchange products of the Chi-
nese interior for the one necessary product the Han Chinese could not
produce internally: militarily capable horses. This new institution also col-
lapsed by 1449, undermined by both military failure and the internal pres-
sures, ecological and commercial, on the tea-producing regions.The early Ming experience highlights the same problems and advantages
that would be faced by the Qing during their expansive period. Negative
factors for Chinese rulers included the limitations of supply in the steppe,
the unreliability of Mongol allies, and the constant instability of steppe pol-
itics. In the favor of the Chinese was the poverty of the steppe and the
Mongols’ dependence on Chinese goods, the rising importance of trade
relations for the Mongol chieftains, as well as their interest in gaining
Chinese legitimation of titles, useful in the constant succession struggles.
Divisions among the Mongols helped the Ming rulers most. As the split be-
tween Eastern and Western Mongols (Oirats) became the fundamental di-
viding line in the steppe during the fifteenth century, China’s ability to ward
off attacks on the frontier depended on careful manipulation of the fears of
each group of Mongols about the other.
The Ming and the Mongols
Our story begins around the year 1400 ce, shortly after the collapse of
the Great Mongol empire founded by Chinggis Khan. Peasant rebellions
in China against Mongol rule brought down the Yuan dynasty when the
Mongol invaders fled without putting up a fight. Zhu Yuanzhang, peasant
leader and former Buddhist monk, captured Beijing in 1368, but he estab-
lished his main capital at Nanjing, close to his hometown of Fengyang in
The early fifteenth century was a critical turning point in the history of
both China and Central Asia. In 1399 Prince Di of Yan rebelled against
his nephew, the reigning Jianwen emperor. His successful military coup
brought him to power as the Yongle emperor (r. 1403–1424) and initiated
the series of aggressive military campaigns, led personally by the emperor,
against the Mongols. In 1405 the death of Timur just before launching an
invasion with 200,000 men spared China the devastation he inflicted on the
Middle East. China no longer faced a major threat from a consolidated em-
pire led by a conqueror with the ambition of Chinggis Khan.
52 formation of states

China’s Mongol rivals for power over the steppe divided into eastern and
western groups bitterly at odds with each other.
The first reference to the western Mongolian tribe known as Oirats
(Oyirad) occurs in Rashid ad-Din’s description of the rise of Chinggis
Khan. In 1201, under their chief Khudukha-beki, they joined in a military
alliance with their neighbors, the Naiman and Merkid, to fight a losing bat-
tle with Chinggis. The Oirats were called the “forest people” (perhaps from
Mongolianoi,“forest”), living in the upper reaches of the Yenisei River
near Lake Baikal.
3Their main occupation was hunting and fishing, not pas-
toral nomadism. To their north and south lived the Turkic Kirghiz and
Naiman. To their east and west were the Mongolian Merkid and Tumad.
They formed a distinctive group speaking a different dialect from neighbor-
ing Mongol peoples. Among them the political power of shamans was espe-
cially high: the term beki(shaman) indicates that their ruler, Khudukha-
beki, was one of the leading priests.
4In the great battle of 1201, Khudukha-
beki tried to use his shamanic powers to stir up a storm against Temujin,
but he failed:
The Oyirad Khudukha Beki [and three others]...these four men [led]
Jamugha’s forces to battle...Just as the two armies began to charge
one another...Buyirugh Khan and Khudukha, who were great sha-
mans, began to conjure a storm of darkness. They began to raise winds
and darkness in order to blind us, when suddenly the storm turned. In-
stead of striking our army their storm blinded their own men. Their
soldiers fell into ravines on the mountainside unable to see, crying:
“Heaven’s turned against us!” and their army dispersed...Khudukha
Beki ran back to the forests as far as the Shisgis.
Khudukha fled back to his forests but soon after surrendered to Temujin
with his four thousand households of followers. In 1217 he aided Chinggis
Khan’s son Jochi in an expedition to subdue the other forest peoples of the
region. In return Khudukha was allowed to marry the women of his tribe to
Chinggis’s descendants.
6When Chinggis created his organizational system
of thousands and myriads (tümen)designed to undercut tribal ties, he of-
fered the Oirats along with the Ongguts the exceptional privilege of main-
taining their tribal affiliations within the tümenunits.
After Chinggis’s death, the Oirats occupied an important strategic posi-
ming, muscovy, and siberia 53

tion at the juncture of the four giant appanages(ulus)belonging to his
sons: Tolui to the southeast, Ögödei to the southwest, Chagatai to the west,
and Jochi to the northwest. This four-way division of Chinggis’s conquests
became the basis for the four great successor Khanships of Eurasia: the
Yuan, Il Khan, Kipchak, and Chaghatai Khanates. The Oirats supported
the 1260–1264 rebellion of Arika Buga against Khubilai, and their “Waila”
army helped in the attacks.
8After their defeat by Khubilai, the Oirats disap-
peared from the historical records for over a century, but they reappeared
in 1388 as opponents of the last Yuan emperor, Toghon Temur, the direct
descendant of Khubilai.
9The year 1399, when an Oirat commander killed
the successor to the Mongol Khan, marks the decline in independence of
the Eastern Mongol Khans and the rise of Oirat hegemony over the Mon-
golian steppe.
Zhu Yuanzhang had ordered, but not led personally, nine campaigns
against the remnants of the Mongol empire in China, and his fourth son,
Prince Di of Yan, the future Yongle emperor, had participated in two of
10 He was an experienced military commander, very familiar with
steppe conditions, and determined to eliminate Mongol power from the
northwest. He also knew well how to manipulate divisions among the
Mongols to China’s advantage. During the military campaigns within
China against the Jianwen emperor and his supporters, he made sure to
keep on friendly terms with the Mongols, avoiding a two-front war. He
gave several chieftains high ranks, and when Mongols in the Northeast,
suffering from famine, requested permission to barter horses for grain in
1407, he opened the first of the horse markets on the frontier that would
later become an important trading link between Chinese and Mongols. The Mongols had suffered continual anarchy for two decades, but in
1408, the Eastern Mongol chieftain Arughtai killed the Khan and recalled
Bunyasiri, a direct descendant of Chinggis, from Beshbalik to become a
new Khan of the Eastern Mongols. Arughtai, the real power, never claimed
the Khanship for himself, but worked behind the scenes as the Khan’s as-
sistant. Although Yongle at first tried to conciliate the new leaders, he
broke with the Mongols when the Khan killed his emissary in 1409. Yongle
then turned to the Oirats to offset the rising Eastern Mongol power. He
granted Chinese titles to the three Oirat chieftains, of whom the most im-
portant was Mahmud. Mahmud cooperated by launching an attack against
Bunyasiri while at the same time Yongle sent out 100,000 men under his
best general, Qiu Fu, against him. Qiu Fu, unwisely advancing too far,
54 formation of states

found himself surrounded by a Mongol army near the Kerulen River. Qiu
Fu was killed and the Chinese army destroyed. Vowing revenge, Yongle
personally planned a major campaign against Bunyasiri, the year after he
definitely decided to move the capital of China north to Beijing.In this first great campaign he assembled at least 100,000 men (the Ming-
shi figure of 500,000 must be exaggerated) and 30,000 carts for transport,
leaving Beijing on March 15, 1410. The army went from Xinghe, 50 kilo-
meters northwest of Kalgan, north to Mingluanshu, where the emperor
held a great military parade to impress the Oirat envoys, thus ensuring their
neutrality. Although Bunyasiri wanted to flee, Arughtai disagreed, so the
two leaders separated, each acting in effect as head of a separate tribe, fa-
tally weakening their strength. Yongle first attacked Bunyasiri to the east,
pursuing him to the Onon River, where he defeated Bunyasiri’s troops.
Bunyasiri in flight was killed by the Oirat Mahmud. The emperor then
turned east against Arughtai, who fled with remnants of his army. Yongle
returned in victory to Beijing on September 15 after six months on cam-
paign. The wily Arughtai then presented tribute and proposed to surrender, if
the emperor would make him overlord of all foreign barbarians—much the
same proposal that would be presented by the Mongol prince Amursana in
the mid-eighteenth century. Yongle, now needing the Eastern Mongols on
his side, granted Arughtai the title of Hening Wang and enrolled him as an
ally. The new threat came from the rising power of the Oirat Mahmud,
who had named Bunyasiri’s son Delbek (Ch. Daliba) the new Mongol Khan
and besieged Karakorum, the old Yuan capital. In 1413 Mahmud, fearing
the new alliance of China and the Eastern Mongols, began his attack on
China, sending thirty thousand troops to the Kerulen River. This attack in-
stigated Yongle’s second personal campaign. He left Beijing on April 6,
1414, marching to Xinghe, then through the steppe on the same route up to
the Kerulen and beyond to battle the Oirats at the upper Tula River. An in-
novation in this battle was the Chinese use of cannon, which the Mongols
could not resist. Although Yongle forced Mahmud to retreat, he could not
pursue him. Delbek Khan and Mahmud escaped, and the emperor returned
to Beijing on August 15. Arughtai, supposedly the emperor’s ally, begged
off participating in the campaign because of illness. Mahmud sought reconciliation after his defeat. Although the emperor
was suspicious, relations might have warmed up, but then Arughtai in
1416 attacked and killed Mahmud and Delbek Khan. The emperor en-
feoffed Mahmud’s son Toghon with his father’s title of Shunning Wang and
tried to establish peace between the two Mongol groups, but the incessant
seesaw of Mongol politics was to a large extent a result of deliberate Chi-
ming, muscovy, and siberia 55

nese policies of divide and rule. Each time one ruler grew stronger, China
supported the other, thus preventing the unity of the Mongols while ensur-
ing continual conflict. Still, Yongle had achieved temporary success. No
further attacks by the Oirats ensued for six years.During the last years of Yongle’s reign, however, the emperor cam-
paigned three more times against the same players as before. Now the
growing power of Arughtai led him to pillage caravans on the way to
Beijing, and in 1422 he overrran Xinghe, the outpost fortress north of Kal-
gan. Yongle now prepared his largest army to date, spurning the advice of
important officials, who urged him not to launch another campaign. Just as
in the Qing, a classic conflict between a remonstrant official and an auto-
cratic emperor focused on the issue of supplies for a northwest campaign.
These officials were well worth listening to. They included Xia Yuanji,
Minister of Revenue for sixteen years, a man who “was noted for his ability
to know exactly what funds and grain supplies were held anywhere in the
empire”; Wu Zhong, the Minister of Works; and Fang Bin, the Minister of
War. They argued that the empire could not afford to supply such a massive
campaign. When Fang Bin committed suicide to demonstrate his commit-
ment, the furious emperor imprisoned Xia and Wu and confiscated Xia’s
estate. Both men survived torture in prison until they were released by the
next emperor after Yongle’s death in 1424. Xia earned a legendary reputa-
tion as an upright, forceful Confucian official, although his main talent was
as an efficient if unoriginal administrator (somewhat like Chen Yun in the
PRC). Xia, notably, also criticized the extravagance of Zheng He’s mari-
time treasure fleets and was mainly responsible for their abolition.
Supplies for the third expedition included 340,000 donkeys, 117,573
carts, 235,146 men, and 370,000 shiof grain. In addition, Shandong,
Henan, and Shanxi provided 200,000 donkeys for officers and men, which
must have been a tremendous burden on these provinces. On April 12,
1422, the emperor left Beijing and headed for Kulun. When Arughtai fled,
the Chinese took booty from his camp in typical steppe fashion. The frus-
trated Yongle then took out his anger on three Urianghai Mongol tribes not
involved in Arughtai’s attack, plundering them mercilessly. After this unedi-
fying looting expedition, he returned to Beijing on September 23. The final
two campaigns generally repeated this frustrating pattern of futile pursuit
of Arughtai, followed by supply shortage and ruthless plunder of any avail-
able Mongols along the route. In the final campaign of his life, the emperor
was forced to turn back, fearing the approaching winter. He died on the re-
turn journey on August 12, 1424. Yongle was the last aggressive emperor of the Ming. His Qing succes-
sors uncannily repeated his experience, with similar campaigns over much
56 formation of states

greater distances. Each time he faced serious supply shortages; although he
won several decisive battles, his enemies escaped to fight again. Mongol al-
lies proved unreliable, just as likely to raid and plunder after submitting
tribute as before. By setting Eastern and Western Mongols against each
other, Ming rulers warded off the unlikely threat of a unified Mongol alli-
ance. The great campaigns were far more expensive than the intermittent
raiding they were supposed to prevent. They seem to reflect a thirst for per-
sonal vengeance more than rational calculation of strategic necessity. The
route to controlling the Mongols via trade had not yet been discovered.
Far-sighted officials who advised against the heavy expense of these cam-
paigns suffered imperial censure. Yongle’s aggressive campaigns did not
even provide a temporary solution to the problem of frontier defense.Nevertheless, Yongle left important lessons for his followers in the Qing.
He truly loved campaigning in the steppe. In his Beizhenglu,an account of
his campaign travels similar to Kangxi’s letters, he insisted on the bracing
effect of personal experience of the grasslands: “Scholars see only what’s on
paper; this cannot compare with what you see with your own eyes.”
failure to capture Arughtai demonstrated that a single army setting out
from China could not track down the swift Mongol formations. Kangxi, by
contrast, adopted the Mongol tactic of using multiple armies converging
on one objective. Most important, Yongle reoriented the empire institu-
tionally toward the north and northwest. From 1407 to 1421 he con-
structed China’s new capital at Beijing, where it has remained ever since.
His successors rebuilt the Grand Canal to supply Beijing and focused much
of their fiscal and agrarian policy on maintaining the large state presence in
North China, far from the most productive agriculture in the south. For the Mongols, this was an unsettled time. No leader emerged to rival
the stature of the Yuan emperors or of Timur. No Mongol really aspired to
rule the entire steppe. China was a target for raiding and plunder, but not a
genuine goal of conquest. The only Chinese territories seriously attacked
were border outposts. Unification of the Mongols required a convincing
leader of the Chinggisid line; this is why Arughtai supported Chinggis’s de-
scendants while working behind the scenes. But the Khans remained weak
and incapable of rallying all the Mongols behind them. After Yongle’s death, the Oirats became the aggressive party. Reunited
under their new leader, Toghon, who had killed two rival chieftains, they
named Toghto-buqa their Khan, defeated Arughtai in 1431, and killed him
in 1434. But the great new leader of the Mongols, who was to lead the
Oirats to dominate the greatest portion of the steppe since Timur, was
Esen, the son of Toghon. Peace had reigned on the northern frontier under
the Xuande emperor (r. 1426–1435), but with the rise of Esen, the new
ming, muscovy, and siberia 57

Zhengtong emperor (r. 1436–1449), enthroned at the age of eight, faced a
vigorous rival.
Esen in 1443 took over his father’s title of Tayisi, or military assistant to
the Grand Khan. He quickly overshadowed Toghto-buqa, the Chinggisid
Khan, who was far more inclined toward peaceful relations with China.
Esen first focused on the Prince of Hami, who was a Mongol normally loyal
to China. Repeated raids and threats by Esen and Chinese lack of support
forced him to submit in 1448. Esen then took Gansu, proclaiming his own
provincial government there. Mongols fleeing Gansu appealed to the Chi-
nese emperor for aid but received none. Once Esen had secured his rear,
he could now prepare for a major attack on China. He plundered the
Urianghai Mongols on China’s northeastern frontier, forcing their submis-
sion. At the same time, he used tributary missions to China to build up his
economic resources, expanding the number of envoys to over two thou-
sand in the 1440s, and three thousand in 1448, despite Chinese complaints
about the great cost of feeding such huge numbers. The refusal of the Chi-
nese to allow even greater tribute missions and complaints about plunder-
ing by Mongol envoys on their way to Beijing served as the pretext for
Esen’s invasion, but Esen could also raise legitimate complaints about being
cheated by Chinese merchants when conducting trade in the capital. Yet de-
spite warnings of Esen’s preparations for attack, the Ming court made few
preparations. The young emperor was completely dominated by his eunuch
tutor, Wang Zhen, a man with no military experience, interested more in
his private wealth than in the empire’s security. Finally, when Esen moved
against Datong, the key strongpoint of the Great Wall in Shanxi, Wang
Zhen persuaded the emperor to lead personally a huge army, said to be
500,000 men, against Esen. Hostile sources claim that the only reason
Wang Zhen insisted on the emperor’s personal leadership was to ensure
that the emperor would visit Wang’s hometown in Shanxi. Thus began the process leading up to the “ridiculous” Tumu incident of
1449, when the “foolish, incompetent” Zhengtong emperor succeeded in
getting himself captured by Esen.
15 Despite constant warnings by compe-
tent officials about the danger of being stranded in the steppe without ade-
quate provisions, the massive army began to run short of supplies almost as
soon as it left the walls of Beijing. Eunuch advisers to military officials pre-
vented adequate preparations along the march to Datong. The Mongols fell
back as the army approached the city, waiting for the chance for an am-
bush. Only after reaching Datong did Wang Zhen recognize the danger and
decide to turn back. On the return journey, following classic nomadic strat-
egy, Esen first attacked and destroyed the Chinese rear guard. The starving
58 formation of states

troops sought shelter at the border outpost of Tumu. Wang refused to ad-
vance to the nearby walled city of Huailai, not wanting to leave behind his
personal baggage train. So the emperor and his exhausted troops were sur-
rounded and captured at Tumu, and Wang Zhen and the eunuchs were
killed.Esen failed to follow up his amazing windfall victory with an immediate
attack on Beijing. Instead he first extorted payments of 20,000 taels from
the garrison commander at Datong and then returned with the emperor to
the steppe. This behavior should make us doubt whether Esen really in-
tended to reconstruct the empire of Chinggis Khan or to conquer China. By
the time he turned back to lay siege to Beijing two months later, Yu Qian,
the very capable finance minister ruling as regent in the emperor’s absence,
had placed the emperor’s younger brother on the throne and organized a
tough defense. Esen quickly gave up the siege, returned to the steppe, and a
year later sent the emperor back. Negotiations with two envoys sent by the
new Jingtai emperor demonstrate that the new emperor was not particu-
larly eager to have his brother back, since he gave no instructions for the re-
turn of the emperor to either mission.
16The captive emperor, desperate to
return, guaranteed that he would not seek to recover his throne. Esen him-
self was quite anxious to unload his illustrious captive, now useless to him.
The emperor upon his return was held under close house arrest until his
supporters staged a coup in 1457, killed the reigning emperor, and restored
him to the throne as the Tianshun emperor (r. 1457–1464). After the emperor’s return, Esen briefly reached his height of power, once
again subduing the Eastern Mongols and the Khan Toqtobuqa near Turfan
in 1451. Toqtobuqa was killed in 1452. At its peak, the Oirat empire ex-
tended from Urianghai and the Jurchens in the east to Hami in the west. But
his fatal step was to proclaim himself Khan in 1453. Even though the Chi-
nese emperor approved his designation, Mongols still expected leadership
only by legitimate descendants of Chinggis. Revolts by his military com-
manders forced him to flee, until he was killed in 1455. Esen’s conflict with the Ming can easily be misinterpreted. Frederick
Mote argues that the Mongols, remembering their unification under the
Yuan, had achieved new self-awareness of themselves as a “Mongol na-
tion” under Esen and that Esen effectively appealed to all Mongols to re-
construct the empire of Chinggis Khan.
17Little evidence supports this inter-
pretation, and much makes it implausible: Mongol chieftains in Hami and
Urianghai submitted to Esen only reluctantly, under threat of force, and re-
verted to Chinese alliances when possible. Esen could hardly have had un-
equivocal designs on conquering the Chinese empire, since he did not fol-
ming, muscovy, and siberia 59

low up his capture of the emperor with an immediate attack on Beijing.
Instead, he engaged in the typical raiding and extortion, and quickly gave
back his captive when he found him of no immediate use.The dialogue between Esen and Yang Shan, the second envoy sent to re-
trieve the captive emperor, is a very revealing account of a nomadic ruler’s
attitudes toward China, one of the few examples we have of direct dialogue
between a steppe ruler and Chinese representatives. It is, however, reported
only through Chinese sources. (Later, we shall examine a similar conversa-
tion of a Western Mongol leader with a Russian envoy in the eighteenth
century, in which the Russian sources give a different impression.) When
Yang Shan accused Esen of “black ingratitude,” Esen replied, “Why did
you lower the price of horses and why did you often deliver worthless and
spoiled silk? In addition, many of my envoys disappeared without a trace
and never returned home...”YangShan blamed Esen for supplying too
many horses, thus lowering the price, and blamed “government contrac-
tors” (tongshi) for supplying “worthless silk.” He claimed that many of
Esen’s three to four thousand envoys were “thieves and brigands” who
feared returning home. Threatening Esen with great losses if he conducted
further attacks, he offered him great profits if he resumed trade: “If you
now return the emperor, and make peace as before, the riches of China will
arrive daily, and both countries will be happy.”
18 Yang Shan had a weak
hand to play, since Ming forces were weak, and the court gave Yang no
valuable gifts to present; but in the end, Esen sent back the emperor in re-
turn for resumption of trade in silk and horses. The incompetence of Chinese policy in this period highlights the require-
ments for success. Arrogance toward “barbarian” nomads, failure to pay
attention to military logistics, clumsiness in managing diplomatic alliances
with Mongols, and mismanagement of tribute trade by eunuchs in search
of profit all greatly facilitated Esen’s rise to power. Yet the Mongols’ desire
for trade induced them to make peace in the end. The Qing combined
trade, diplomatic, and military pressures much more effectively.
After Tumu, the Ming rulers abandoned forward campaigns into the
steppe. They found themselves on the defensive on all fronts, constantly
warding off repeated attacks by many autonomous Mongol chieftains. The
period from 1450 to 1540 marked an inconclusive shift to a full-fledged
Great Wall strategy. Even though no single unifying leader arose among the
Mongols, the ability of Ming frontier commanders to protect the border
declined even further. Since Arthur Waldron’s excellent study of the Great
60 formation of states

Wall focuses on this period, I will only highlight the issues of frontier mili-
tary supply and nomadic diplomacy and trade, which are somewhat under-
played in Waldron’s account.
Waldron divides the strategic history of the Ming into three periods: the
first, from 1368 to 1449, a time of an open frontier, with no major wall
building; the second period, 1449 to 1540, a time of inconclusive shift from
offensive to defensive strategies, with great controversy and the beginning
of major wall based defenses; and the third, from 1540 to the end of the dy-
nasty, marked by reinforcement and consolidation of settled garrisons at
large fortifications and the completion of the full Great Wall complex. The Oirats withdrew to the Orkhon River after Esen’s death in 1455 but
continued raids on the northwest border. Now, however, the Eastern Mon-
gols gained strength under Bolai, who attacked and plundered Shaanxi and
Gansu. In 1461 General Li Wen, angry at being refused promotion, refused
to mobilize his troops, causing the Chinese to suffer a great defeat.
20 The
Chinese aimed to buy peace by establishing tribute trade with the Mongols,
allowing them to choose their own routes and plunder along the way. Cen-
sor Chen Xuan delivered a devastating report on the laziness of the bor-
der commanders, the lack of supplies, and the mistreatment of poor sol-
diers, but his insistence on wholesale replacement of commanders was not
More ominously, in the 1470s Mongols began to move into the fertile
steppes of the Ordos, inside the Yellow River bend, a strategic salient of the
steppe held by the Chinese. Debates over the defense of the Ordos raged for
the rest of the century.
22Wang Yue, the most capable military commander
of the time, realized that an army of 150,000 would be necessary to defend
the Ordos, but that it would be impossible to supply such a huge number in
the steppe. The Chinese were forced to withdraw their outposts to the
south and tacitly agree to the loss of the Ordos, while Mongols used the re-
gion as a base for raids to the south and west. Yu Zijun proposed the build-
ing of major defense fortifications in 1471 and 1472, and after gaining the
support of Wang Yue against considerable opposition, succeeded in 1474 in
carrying out his plan. The wall he built ran 600 miles from northeastern
Shaanxi to northwestern Ningxia, using a labor force of forty thousand
men at a cost of over 1 million taels. The wall also provided a shelter for
military farms which yielded sixty thousand bushels of grain per year. It
proved its effectiveness in 1482 when the soldiers at the wall held off a ma-
jor Mongol offensive. This project began the first serious Chinese effort to
link economic rehabilitation with a defensive strategy. Instead of an aggres-
sive “extermination campaign” (jiao),Wang supported “restoring the peo-
ple’s livelihood” (shao su min lao).
23 Wang Yue’s famous victory over the
ming, muscovy, and siberia 61

Mongols at Red Salt Lake in 1473, the first in half a century, did not initiate
further forward moves into the steppe; instead it bought time for Yu to
complete his wall. Wang Yue was the last Ming commander in the north-
west who could effectively manage both supply and mobilization. Many
believed, mistakenly, that serious problems with supply shortages and cor-
ruption of frontier commands began only after his death in 1499.China was still partly protected as long as the Mongols remained divided
into eastern and western groups, but in 1483 Batu Möngke (Dayan Khan)
again unified the Mongols. He conducted raids every year against Liao-
dong, Gansu, Datong, Xuanhua, and Yansui. Small detachments of swift
Mongols routed incompetent Chinese military officials, who reported false
victories to the court. Wang Ao reported at a court conference in 1501 on
the desperate situation of the frontier, where commanders quarreled with
one another, no one dared fight nomads openly, officials were lazy, and
the troops showed no courage.
24 In 1513 Batu began building fortified
camps in the Xuanhua and Datong areas, from which he staged increas-
62 formation of states
A Russian delegation passes
through the Great Wall.
[To view this image, refer to
the print version of this title.]

ingly dangerous incursions with up to fifteen thousand cavalry even closer
to Beijing.
25 At the same time, internal succession struggles were tearing
apart the principality of Hami, a useful buffer for China against the west-
ern steppe. Turfan conquered Hami in 1513. The high point of Mongol
power came in 1517, when Batu moved on Beijing itself. Although the Chi-
nese held him off in a major battle, he continued to threaten the capital
until 1526. The issue of the recovery of the Ordos still dominated Ming strategy dis-
cussions in the early sixteenth century. Hard-liners regarded trade or nego-
tiation with the irredeemably violent Mongols as impossible. They continu-
ally advocated unrealistic plans to drive the nomads out of the region.
Advocates of compromise had to consider soberly the relative costs and
benefits of major military expeditions compared to trade, but no one could
actively advocate trade relations with the Mongols in this period. Wall
building emerged as a second-best policy, approved by neither side but ac-
ceptable to both. At the same time, the links between economic develop-
ming, muscovy, and siberia 63
[To view this image, refer to
the print version of this title.]

ment and frontier defense grew more apparent. Qiu Jun’s highly regarded
manual of historical statecraft, theDaxue Yanyibu,published in 1487, con-
tained a chapter on the Ordos question that stressed the importance of con-
currently strengthening defense and the local economy. Zeng Xian, the
great proponent of both recovery of the Ordos and of wall building in the
1540s, relied on Qiu Jun’s work to argue for the connection between eco-
nomic development and military defense, first in building walls at Linqing
around the Grand Canal, and later in the northwest.
At the same time, the increasing unreality of court politics undercut fron-
tier defense. The Zhengde emperor (r. 1506–1521) purported to love mili-
tary affairs, and he frequently wore Mongol dress. He liked to tour the
frontier, built a palace there, and defeated Mongol raiders in 1517, but
failed to develop a coherent strategy.
27In some ways his behavior curiously
foreshadows that of the Qianlong emperor. He took a Uighur concubine,
he set out on a costly southern tour despite official protest, and he enjoyed
promoting the image of a vigorous martial emperor, but in fact he did no se-
rious strategic planning. Qianlong, however, surrounded by able Grand
Councilors and vigorous Manchu, Mongol, and Chinese commanders,
could afford his indulgences much better than could his Ming predecessor. In 1506 Yang Yiqing began to increase fortifications along the frontier,
but eunuch interference thwarted his plans after only thirteen miles were
28The Jiajing emperor’s reign (1522–1567) featured further factional-
ism at court and rising contempt for Mongols. Officials ordered, for exam-
ple, that in documents the character yifor “barbarian” be written as small
as possible. In 1551 the emperor prohibited all trade with the Mongols on
pain of death. But in the 1540s a second major debate over the Ordos
marked a significant turning point in frontier policy. Altan Khan (1507–1582), the grandson of Batu, had risen to power in
the mid-sixteenth century as the next great Mongol raider of the Chinese
empire. He never unified all the Mongols, but he led the twelve Tümed (ten
thousand–man units) under his control north of Shaanxi and Shanxi in
continuous attacks along the frontier, followed by requests for permission
to conduct tribute trade—requests which the Chinese nearly always re-
jected. This repeated cycle of “request, refusal, raid” continued for forty
years until 1570.
29 Once again debates raged about defense strategy at
court. Weng Wanda proposed major wall-building projects in Xuanfu and
Datong to block the most accessible points and the erection of permanent
watch posts. The emperor gave him 600,000 taels, with which he repaired
fortifications and helped to restore the local economy, but he rejected
Weng’s proposal to allow Altan to send tribute. Zeng Xian submitted de-
64 formation of states

tailed reports in 1547–48 proposing an aggressive campaign to drive the
Mongols out of the Ordos, but Weng pointed out that it would be impossi-
ble to support the huge baggage trains necessary in this desert-steppe envi-
ronment.Zeng proposed an attack by 300,000 men by land and water routes, in-
cluding the use of firearms, at a cost he estimated at 2 million shiof grain
and 3,000 taels of silver. Zeng estimated a total cost of over 300,000 taels,
but the real costs would be much larger, up to 1.3 million taels. Waldron es-
timates that including transport costs would drive up the total to 3.75 mil-
lion taels.
30The Ministry of War, stressing the difficulties of finance, urged
the need first to rebuild the border economy. Zeng nevertheless began his
preparations, stirring up the hostility of the local population by corvée im-
pressment, and by requisitioning cooking pots and farm tools to cast into
weapons. Both Weng and Zeng had close advisers of the emperor to support their
case, but the eunuchs lobbied against Zeng. In 1547 the emperor ordered
Zeng executed, permanently rejecting the idea of a recovery campaign.
Zeng Xian and Xia Yan later won reputations as patriots, despite the un-
realism of their recovery project. Further raids ensued, and cowardly bor-
der commanders concluded that it was better simply to surrender to the
Mongols and then buy them off than to resist. Now the Mongols gained the
support of many Chinese, including deserters from the army, criminals, ref-
ugees, and agricultural settlers in the Ordos region. Chinese military advis-
ers pointed out the weak points of the Ming defense lines and the best
places to attack. Altan began building a new city, Köke Khota, or Guihua
(present-day Hohhot), under their influence as he moved toward a more
settled way of life.
32The Chinese denounced these traitors in Mongol ser-
vice, and demanded their heads, but continued to refuse trade. One attempt
to open horse markets in 1551 was quickly shut down after Mongol raids,
and the emperor prohibited all trade with the Mongols on pain of death. Only under the next emperor’s reign (Longqing, 1567–1572) could the
Ming, in a brilliant stroke of frontier diplomacy, bring itself to negotiate
peace on the frontier. The genius behind the successful policy was Wang
Chonggu, Governor-General of Shaanxi, who realized that all the elements
for a settlement were in place. He was supported by the vigorous Grand
Secretary Zhang Juzheng, Ming China’s most illustrious high official, who
took a great interest in frontier affairs.
33Altan Khan wanted peaceful trade
relations; he raided only if tribute was refused. The Ming had reinforced its
walls and mainly needed horses from the Mongols for the mobility of the
garrisons. The Chinese wanted their deserters back, and the Mongols were
ming, muscovy, and siberia 65

willing to surrender them for a price. The opportunity for a settlement ap-
peared when, in 1570, one of Altan Khan’s grandsons, feuding with his
chief, surrendered to the Chinese. Wang Chonggu recommended accepting
his surrender and exchanging him for the Chinese deserters, promising
trade relations with Altan if he swore to cease his attacks.Wang’s report is a good outline of future Qing policy. He recommended
giving honorary rank and official degrees to selected Mongol chieftains,
regulating the time and size of tribute embassies, limiting them to 150 men
once per year, and limiting frontier horse markets to eight hundred men
from the nomads, supervised by five hundred troops.
34After stormy discus-
sions at court, the Ming emperor and Zhang Juzheng squelched cries for
Wang’s impeachment and accepted the new policy, opening frontier horse
markets and overturning the trade prohibition of 1551. Altan was given the
honorary rank of Shunyi Wang, sixty-three other chieftains received appro-
priate titles, and frontier trade was tightly regulated at fixed dates and re-
stricted, unlike other tributary missions, to the frontier. Merchants flocked
to the frontier to sell silk, fur, grain, and cooking pots to the Mongols; the
government collected taxes on the trade and used the income to buy poor
horses at high prices from the nomads.
The Lama Buddhist church began its rise to influence in Mongolia in the
late sixteenth century as well. The first Mongol contact with Tibetan Bud-
dhism since the thirteenth century occurred in 1566, when the Ordos Mon-
gol Khutukhtai Secen Hongtaiji (1540–1586) traveled to Tibet, offering to
protect the church if the Tibetans submitted to him. Secen converted to
Buddhism and brought with him over 100,000 Mongolian, Tibetan, and
Chinese converts. In 1576 he advised Altan Khan to use the religion to
strengthen his position among the Mongols.
After the peace settlement, Wang Chonggu encouraged the emperor to
promote Buddhism among the Mongols, and Mongol princes hired Chi-
nese craftsmen to build temples in Mongolia and Kokonor. The Mongols
agreed to donate horses and camels to monasteries instead of sacrificing
them to their ancestors. Altan invited Sodnam Gyamtsho (bSod-nams-rgya-
m-ts’o) of the Gelugpa (dGe-lugs-pa) sect in Lhasa to visit him at the first
major temple built in Kokonor. In 1578 Altan gave him the title of “Dalai
Lama” (Oceanic Teacher in Mongolian), and the Dalai Lama declared
Altan to be the reincarnation of Khubilai Khan. The grandson of Altan
Khan became the first Mongol to take the post of Dalai Lama in 1586.
66 formation of states

Many scholars argue that Altan Khan’s invitation marked the first intro-
duction of Tibetan Buddhism into Mongolia, but Henry Serruys has shown
that “all the facts point to an uninterrupted Lamaist tradition in Southern
Mongolia going all the way back to Mongol Lamaism of early Ming times
and in final analysis to the end of the Yuan period.” The close and continu-
ous association of the Mongols with the Tibetan Buddhist establishment
had begun centuries earlier, but the sixteenth-century situation allowed the
“expansion and consolidation of something that had never completely dis-
appeared from Mongolia and was only waiting for the right opportunity to
reassert itself.”
36Later the Qing would try to cut these ties to Tibet but keep
the Mongols focused on Buddhism centered in Beijing. The late sixteenth century was also the heyday of wall building along the
Ming border. The Nine Border Garrison system reached its highest elabora-
tion, existing walls were strengthened with brick and stone, and the walls
were linked together with watchtowers, all at heavy cost in silver. Most of
the present-day Great Wall dates from these sixteenth-century works. All
of these elements—controlled trade and tribute, patronage of Buddhism,
grants of titles, defensive strengthening, and investment in the local econ-
omy—became key points of Qing strategic policy. Conversely, the Mongols also exhibited debilitating signs of division
which plagued their efforts to unite against the Ming. Altan was not a
would-be unifier of all the Mongols. Unlike in the northwest, the tribes of
the northeast frontier refused to make peace with China, even after Altan’s
submission. The Mongols in the northwest only temporarily followed
Altan’s lead. As Dmitrii Pokotilov puts it, “they collaborated only as long
as there existed a common interest and the chance of easy and quick plun-
dering, but when it became necessary to intercede for the interests of some
individual prince with the risk of strong resistance, all solidarity immedi-
ately disappeared, even among near relatives.”
37Only in the late sixteenth
century did the Ming finally succeed in exploiting these divisions among
Mongol tribes. This divide-and-rule strategy, which, as mentioned earlier,
became the cornerstone of Qing policy, kept peace along the northwest
frontier until the end of the dynasty. The new threats to the Ming arose
from domestic social disturbances within northwest China and elsewhere,
and from the rising Manchu state in the Northeast. Ming strategy thus evolved from unsuccessful efforts at military penetra-
tion of the steppe to costly defensive walls and garrisons, supported by the
acceptance of limited trade with the tribes in the guise of tribute missions.
The trade lever proved to be the least costly and most effective in removing
the threat of raids. Sechin Jagchid has argued that it was only Chinese mis-
ming, muscovy, and siberia 67

understanding of the need of nomads for trade that prevented peaceful rela-
tions with the steppe. In his view, once emperors accepted regular trading
relations, they faced little military threat:For a period of two thousand years, trade was the chief determinant of
peace and war between the nomadic and the Chinese peoples along
China’s northern border. The nomadic peoples were dependent on a
few key products produced by the agriculturalist Chinese, particularly
grain and cloth. When they were able to obtain these goods peacefully
through the mechanisms of trade, bestowals, and court-to-court inter-
marriage arrangements, stability along China’s frontiers was possible,
but when they were denied ready access to these essential commodi-
ties, war was almost a certainty.
Trading relations, however, did not by themselves necessarily weaken
the strength of nomadic state builders. A far-sighted challenger to Chinese
power could use trading privileges to build up the resources of his state.
This was the strategy pursued by Nurhaci, the founder of the Manchu
state, in his early years. In order to reduce the threat, Chinese dynasties had
to combine peaceful trading relations with efforts to ensure that the steppe
tribes remained disunited. This meant gathering intelligence and gaining al-
lies among nomadic tribes to balance against rising powers. When Chinese
rulers could take advantage of the “fatal individualism” of the Mongols to
keep them weak, they reduced their defense costs by diverting nomads from
raids on the frontier toward attacking one another. The Ming court partly
succeeded in this strategy in the late sixteenth century in Mongolia but
failed to recognize the significance of the growing strength of the Manchus
in the Northeast.
I have described the evolution of Ming strategy from military offense to
trade and defense in terms of the recognition by the Ming of the value of
trading relations to the nomads of the frontier. Frontier trade, however, was
equally important for the Ming court itself. A Han-ruled dynasty like the
Ming, which did not control the steppe, faced the same problem as its pre-
decessors: an inability to produce within its borders sufficient numbers of
militarily capable horses to supply its army. Even for defensive warfare, do-
mestic Chinese horses were inadequate against the tough, swift ponies of
the Mongols. Matteo Ricci noted in the late sixteenth century that most
Chinese army horses were “so degenerate and lacking in martial spirit that
68 formation of states

they are put to rout even by the neighing of the Tartars’ steeds and so they
are practically useless in battle.”
39 Well aware of this problem, the first
Ming emperor established two organizations to breed horses for battle:
the Yuanmasi (Pasturage Office) and the Taipusi (Court of the Imperial
Stud) under the Ministry of War. But domestic pasture was hopelessly inad-
equate. “Horse households” in North China were supposed to provide
horses to the Taipusi in Beijing, but by the end of the fifteenth century, the
government had to allow them to buy replacements. Nearly all of the re-
placements were “western horses” (xima)obtained from private traders
who had bought them at frontier markets in the northwest. The traders
charged the horse households very high prices and gave them horses nearly
useless for warfare.
40 So efforts to provide horses from inside the Great
Wall once again ended in failure. The only option was to obtain horses by trade with the nomads. This re-
quired two basic elements: the Han Chinese had to produce a product the
nomads wanted, and they had to find nomads who would cooperate in pro-
viding horses to be used in military activity against other nomads. Diplo-
macy and trade had to work together. For about a half century, from 1393
to 1449, the “gold tablet system” (jinpai xinfu)seemed to provide the an-
swer. Chinese officials believed that the nomads had a desperate need for
tea. As one official put it, “All the barbarians need tea to survive. If they
cannot get tea, they become ill and die.”
41This delusion supported a policy
of large-scale shipments of tea to the frontiers, with the conviction that
China could use the supply of tea to control nomad attacks. Every three
years, 1 million jinof tea from Sichuan was to be traded with specially li-
censed nomads for fourteen thousand horses at three principal designated
frontier “tea–horse markets” (chamasi)in Hezhou, Daozhou, and Xining
(in present-day Gansu and Qinghai).
42Hezhou was the largest market. The
success of this system depended on Chinese government monopsonistic
control over the purchasing of tea and monopolistic control over its ex-
change at the markets, as well as the ability to designate acceptable no-
madic horse traders. Only a limited number of tribal chieftains could re-
ceive one of the forty-one coveted gold tablets that entitled them to trade
with government representatives. The Ming also attempted to fix the price
of horses, first at the level of 30 to 40 jinper horse, but later at prices rang-
ing from 50 to 120 jindepending on quality.
The gold tablet monopoly system at first dramatically raised tea produc-
tion and the horse supply, but Sichuan was too far away and transport costs
too high to make this effort successful for long. In 1397 the state obtained
13,518 horses in exchange for 500,000 jinof tea. Sichuan’s quota of tea
rose to 1 million jinevery three years, but by 1444 it was reduced to
ming, muscovy, and siberia 69

420,000jin,until abolition of the quota in the 1450s. Sichuan tea planta-
tions declined in importance as Hanzhong fu,in southern Shaanxi, became
a much more adequate source. Sichuan’s production used to be three times
that of Hanzhong, but the proximity of Hanzhong to the frontiers favored
it, and the government’s promotion of refugee settlement in the region en-
couraged land clearance for planting tea. While Yongle’s aggressive campaigns succeeded, the tea monopoly sys-
tem provided him the necessary logistical support, but his demands ex-
panded as the campaigns proceeded. The total number of horses possessed
by the army increased from 310,617 in 1415 to 1.2 million in 1422. With
the end of the campaigns, the gold tablets held by friendly nomad horse
traders were lost, scattered by hostile attacks. Famine struck Shaanxi in
1450, requiring local officials to sell their tea stocks for grain. Military of-
ficials were too busy fending off attacks from Esen Khan to transport the
tea, so they used 10,000 taels from the treasury to buy horses.
External attacks from Western Mongols were, however, only one cause
of the system’s demise. The internal challenge from private commerce was
just as serious. From the beginning, Ming rulers tried to stamp out private
trade in many commodities, driven by an ideology of agrarian self-suf-
ficiency. Private trade in tea was especially threatening, because merchants
could link up with unauthorized horse traders, driving up the official tea–
horse exchange ratio, which was set below the market price. But given the
inadequacy of the official trade, Yongle himself had to rely heavily on pri-
vate traders for his horses, paying high prices for them (80,000 jinfor sev-
enty horses). After 1450, when officials shifted their attention from Sichuan
to Shaanxi, they were forced to recognize the usefulness of private trade.
They also shifted from levying tea taxes in kind to collections in currency.
In the fifteenth century a new merchant contracting system, “Equitable Ex-
change” (of grain for salt), or kaizhongfa,had developed, at first in the salt
monopoly, to provide grain supplies to frontier garrisons. The state gave
salt merchants monopoly licenses to trade in salt in exchange for providing
designated amounts of grain to the northwest frontier.
43In 1505 the Ming
initiated a similar tea merchant contracting system (kaizhongcha)in re-
sponse both to repeated famines in Shaanxi and to the need for horses.
Merchants who contributed grain for famine relief received licenses to
trade in tea, which they could exchange for horses at the frontier. They
were obliged to provide the state with a fixed number of horses, and trade
outside these channels was prohibited. The effect was to stimulate greatly tea production in Hanzhong, and to
promote tea trade as far away as Hunan and Guangdong. Merchant profits
rose dramatically; by 1501 Shaanxi had obtained 9 million jinof tea. But
70 formation of states

the officials soon noticed that they received very few horses, and those they
did obtain were of the poorest quality. In 1503 the government’s share
of the trade was 33 percent; it obtained only 10,000 horses at a cost of
500,000 to 600,000jin.Calls rose to abolish private markets again.
Yang Yiqing, who directed the horse administration in the northwest, an-
alyzed the critical situation in great detail and recommended drastic re-
44 He proposed yet another system, called “Equal Cooperation of
State and Merchant” (guanshang duifen).He recognized that tea produc-
tion had grown rapidly, stimulated by private commerce, but the state re-
ceived only 10 to 20 percent of total production. Private tea traders were
driving up the fixed official prices for horses but provided only the worst-
quality horses as their quota to the state. In principle, Yang wanted to re-
vive the government monopoly gold tablet system, yet he knew that ban-
ning private trade would only leave the tea plantation workers without a
livelihood and dry up supplies to the frontier markets, while failing to
relieve the shortages of horses. Reviving the trade and ensuring profit for
the state required incentives for both officials and merchants. In 1505 he
proposed and implemented a policy of “Summoning Merchants to Pur-
chase and Transport” tea to the frontier (zhaoshang maiyun).Selected mer-
chants were invited to purchase up to 10,000 jinof tea each for a total of
500,000 to 600,000 jin,transport the tea to the frontier markets, and sell
33 percent of the tea to the garrisons at a price of 50 taels per jin(25 taels
for production costs, 25 taels for transport). They could keep half of the tea
for their own trade. Unlike the previous kaizhongsystem, this allowed the
government to keep control of the exchange of tea for horses with the no-
mads so it could profit by fixing prices, and illegal smuggling was strictly
prohibited. Anyone aware of the course of the economic reforms in China of the past
two decades will recognize remarkable parallels here. Sixteenth-century
China was a halfway house in the movement toward reform; the state
could not yield entirely its monopoly control of strategic trade but recog-
nized the importance of providing incentives to merchants and tea pro-
ducers. The state held on to what modern socialist governments call the
“towering heights” of the economy, closely tied to security needs, while en-
couraging private trade in less strategic commodities. This reform, how-
ever, worked no better than the earlier ones. Frontier officials accumulated
large stocks of tea but found no horses they could buy, because merchants
persisted in the private tea–horse trade, cleaning out all the good horses
from the market. In 1532 the three chamasiheld 870,000 jinof tea but
could not buy horses. By the mid-sixteenth century, Ming officials tried to
limit merchant-contracted tea to 500,000 to 600,000 jinand the number of
ming, muscovy, and siberia 71

merchants to 150. By 1586 the Hanzhong tea tax had been completely
commuted to silver. Merchants bought the entire local tea supply, and they
went down the Han River to Xiangyang in northern Hubei, which had be-
come a major tea entrepôt. The state could not stop illegal trade with the
nomads of the northwest, but by the end of the sixteenth century their stra-
tegic importance had declined.Perhaps the most basic factor underlying the increasingly defensive strat-
egy of the Ming in the northwest during the sixteenth century was the
shortage of horses and grain on the frontier. When the kaizhongfamer-
chant contracting policy failed to attract enough supplies to the frontier,
the Ming allowed merchants to establish their own merchant colonies
(shangtun) near the garrisons.
45 On these agricultural settlements, mer-
chants with salt licenses paid tenant farmers to produce the grain for ship-
ment to the garrisons. But by the late fifteenth century, excessive state de-
mands from the merchants caused them to abandon the colonies on the
border and return home. By 1530 there was not enough mercantile capital
in the frontier region to make colonies feasible. The Ming experience demonstrates the very precarious balance of com-
mercial and strategic interests that the court had to pursue on the frontier.
Three different systems of horse–tea trade evolved over two centuries, from
complete government monopoly of both products in the early fifteenth cen-
tury, to nearly complete private contracting for purchase, transport, and
sale in the late fifteenth century, to a mixed government-merchant contract-
ing and transport system in the sixteenth century. None worked for long.
None provided critical strategic goods in sufficient quantities to meet im-
mediate military needs. But the principle held: the only way Chinese rulers
could obtain the two crucial logistical components of preindustrial war-
fare—horses and grain—was to link state requisitions to the developing
commercial economy of the interior, then devise means of transporting sup-
plies to the northwest. The steppe barrier proved a major obstacle to Chinese expansion. Fron-
tier markets under government control had many advantages. On the one
hand, they kept out foreign merchants from the center of the empire; they
lowered transport costs for horses obtained from the steppe (but raised
them for tea produced in the interior); and they were small in number and
kept under close military control. On the other hand, the state could never
entirely prevent private trade. The frontier in the sixteenth century became
a classic “borderland,” only weakly patrolled, where powerful officials and
military officers diverted supplies of tea and horses for their own profit.
Yang Yiqing recognized that people on the border “could speak the barbar-
ian languages, and soldiers from many provinces collect there to trade
72 formation of states

horses with the barbarians. They hire local people as guides, and penetrate
far into barbarian territory.” Cultures were mingling, despite imperial ef-
forts to keep them apart, but Yang’s reforms could not impose authority
over these local frontier profiteers.
The sixteenth century marked a new high tide in the advance of commer-
cial relations throughout China.
47The influx of silver, first from Japan and
later from the New World, provided the medium for expanding long-dis-
tance monetary exchange. It made possible the nearly complete monetiza-
tion of the tax system, known as the Single Whip reforms, throughout the
century. Military defense needs played just as important a role as private
trade and taxation in spreading silver around the empire. In the late six-
teenth century, Beijing sent over 4 million taels of silver per year to the
northwest garrisons for purchasing goods from local peasants and frontier
48These garrisons formed a vast consumer belt demanding con-
stant replenishment of grain rations and textiles. Their demands fostered
the growth of a trading system that linked the northwest to the lower
Yangzi, through merchant contracting with the state and through private
networks. The military and civilian distribution systems were intertwined
through mechanisms of silver payment. The key challenge for Ming fron-
tier officials was to tap enough of this flow of goods to supply their garri-
sons, and to limit the profits scooped up by the aggressive Shanxi mer-
chants. Because they failed to ensure an adequate supply of grain and cloth
at reasonable prices, the troops remained poorly fed and clothed, inciting
them to desert their posts at the end of the dynasty. Ming rulers had shut
down the great Southeast Asian voyages in the early fifteenth century be-
cause they put greater priority on northwest defense, but they still had to
rely on the commercial resources of the southeast to supply the defensive
wall on the northwest. Tea, horses, grain, and silver remained vital components of Qing frontier
trade, too. The Qing also established frontier markets, mainly with the
Russians at Kiakhta, where tea exports were a valuable item of trade. But
they solved the horse problem differently. Unlike the Song and Ming, the
Qing did not try to establish either a monopoly or a subcontracted tea-for-
horses trade. It obtained its horses on private markets with government
funds, or it requisitioned horses from surrendered Mongols. What accounts for the Ming difficulties on the frontier? Ray Huang
traces Ming strategic failure to the fundamental weaknesses of the Ming
garrison, or weisuo,system.
49These hereditary military forces always suf-
fered from inadequate funding, even as expenses rose. Ming fiscal institu-
tions could not efficiently gather revenue from the interior and ship it to the
frontier. Revenue management was fragmented, and there was no central-
ming, muscovy, and siberia 73

ized auditing. The Ming founder designed his empire’s fiscal structure for a
small-scale village economy; it could not accommodate the expanded and
commercialized revenue needs of the sixteenth century.Although Huang blames Ming fiscal structures, three other factors had
greater weight. First, the Ming economy put limits on the empire’s defense.
Before the sixteenth century, Ming China lacked a sufficiently commercial-
ized economy to make possible fully monetized acquisition of strategic
goods. Even 4 million taels of silver was still inadequate to meet the annual
needs of the northwest garrisons. Second, the Han ethnicity of the rulers
and officialdom isolated them from the steppe environment. Ming Taizu,
his successors, and his officials came predominantly from southern China,
far from the frontier. The Manchu Qing rulers, by contrast, knew Mongols
well. Mongols and Manchus intermarried and cooperated on military cam-
paigns. Third, the Ming state lacked essential tools of communication and
administration that would be much more highly developed in the Qing.
Thus ecology, ethnicity, and state structure interacted to produce different
results. Still, the Ming did last over 250 years. For part of this period it estab-
lished a stable defensive relationship with the steppe. It, more than any
other dynasty, relied on the Great Wall for real defense. In this sense it was
the true culmination of a Han-centered strategic polity that had begun with
the Qin dynasty of the third century bce. Although it began as a severely
agrarian regime, hostile to commerce, and repeatedly shut down promising
moves toward commercial developments by calling off the Southeast Asian
naval expeditions and campaigning against the seaborne traders along the
southeast coast, it also attempted in two major institutions, the kaizhongfa
and shangtun, to mobilize mercantile incentives in the interests of strategic
defense. As Mark Elvin notes, this effort to combine commerce and de-
fense, even though unavailing, left a valuable repertory of experience for its
successors, who built on the Ming logistical framework.
State Formation in Muscovy and Russian Expansion
Let us now introduce the third major player in the Central Eurasian Great
Game: the expanding state of Muscovy, which became the Russian empire.
Russia’s entrance into steppe politics is conventionally dated from the mid-
sixteenth century, after the takeover of Kazan’ and the beginning of the Si-
berian expeditions; but Muscovy was an active player on the steppe a cen-
tury earlier, when it emerged from the breakup of Chinggis Khan’s empire.
74 formation of states

Mongolian rule had strongly shaped the Muscovite state, and its foreign
and military policies derived first from steppe conflicts and only later from
contact with western Europe.The Kipchak Khanate, better known by the anachronistic term “Golden
Horde,” was the portion of the Chinggisid empire that included the Rus-
sian steppe. It originated as the ulus,or personal territory, of Jochi, the el-
dest son of Chinggis Khan. By the late fourteenth century the Khans’ con-
trol of the ulushad severely weakened. Leadership struggles between rival
successors divided authority at the center, while several new states formed
in eastern European and Russian territory to challenge the Khans’ rule. The
Grand Duchy of Lithuania, Moscow, Tver’, and other states competed with
one another and with the successor Khans for local control. Moscow, lo-
cated at a strategic point in command of the major trade route of the Volga,
was in an excellent position to play off rival Khans against one another.
The rise of Timur (Tamerlane) to power in eastern Eurasia in the 1360s
further weakened the ability of the Khans to control Muscovy. Dmitri
Donskoi, declared the ruler of eastern Russia in 1375 after crushing ef-
forts of Tver’ and Lithuania to invade Moscow, then moved down the
Volga River. His most famous battle with the Mongol ruler Mamay, at
Kulikovo field on September 8, 1380, forced the Mongol ruler to flee. But
Tokhtamysh, who had taken over Jochi’s ulus,attacked and looted Mos-
cow in 1382. Only the clash between Tokhtamysh and Timur in the 1390s
saved Moscow from once again being subordinated to Mongol rule. Timur crushed Tokhtamysh’s independence, but he did not challenge
Muscovite resistance when he prepared to storm the city. By the time Timur
died in 1405, he had fatally undermined the trade of the Golden Horde by
disrupting the caravan routes and destroying its cities, and had made it im-
possible for later Mongol rulers to subordinate the newly autonomous Rus-
sian states. It was in the first half of the fifteenth century that “East Russia,
actually, if not yet formally, emancipated herself from Tatar domination.”
Muscovites made only token tribute payments to the Khans, and there was
no real interference by the Khan in Moscow’s affairs. When Moscow beat
back Tatar efforts to storm the city in 1451, many Tatars subordinated
themselves to Moscow’s ruler. Thus the rising Muscovite state included
a mixed population of Russians and Tatars in the service of the Grand
Russian historians have debated for centuries the influence of the
Mongols on Russia. The eighteenth-century historian N. M. Karamzin de-
clared that “Moscow owes its greatness to the Khans,” recognizing both
the heavy debt owed by Muscovite rulers to the Chinggisid empire along
ming, muscovy, and siberia 75

with the severe impact of Mongol oppression on political liberties and the
“deterioration of morals.” Nineteenth-century nationalist historians like
S. M. Solov’ev and V. O. Kliuchevksy tended to play down Mongolian in-
fluence, although others, like Prince Nicholas Trubetskoy, insisted that an
understanding of the Mongol empire was essential to understanding the
Muscovite state. The twentieth-century historian George Vernadsky argues
that the greatest influence of the Tatars on Moscow came only with the
breakup of the Horde and the incorporation of Mongols into the service of
the Grand Duke after 1480. Donald Ostrowski, however, has argued for a
much greater direct influence of Mongolian traditions on the formation of
the Muscovite state during the period of its rise to independence. He sug-
gests that the Muscovite princes created a “sharp rift in institutional conti-
nuity” in the early fourteenth century by introducing Mongol political and
military institutions on a wide scale.
The major institutions of fourteenth-century Muscovy paralleled very
closely those of the Kipchak Khanate. These included the dual administra-
tive structure, dividing authority between the chief military commander
(Kipchak bekalribek, Russiantysiatskii) and the chief controller of the trea-
sury (Kipchak vizier,Russian dvorskii); the use of Mongolian and Turkish
terminology for taxation and currency (tamga, “commercial tax”;kazna,
“treasury”; den’gi (tengge), “money”; etc.); the elaboration of a sophisti-
cated postal system for transmission of documents and information (Mon-
gol jam, Russian iam,Chinese zhan); the use of petitions to the prince
(chelom bit’e, derived via Turkish from the Chinese ketou); and the creation
of a “clan polity,” in which only one family could produce the ruling prince
or Tsar, and the heads of other leading clans were ranked in a strict hierar-
chy (the Mongol ungu bogol,Russianmestnichestvo system). Russian mili-
tary institutions, strategy, and tactics also were derived from the Mongol
system. Finally, the principle of “lateral succession,” dubbed “bloody tan-
istry” by Joseph Fletcher, allowed all the brothers and uncles of a ruler to
compete for succession, leading to violent internecine warfare during suc-
cession conflicts. This principle competed with the vertical succession by
the ruler’s son, introduced from Byzantium, until 1425.
Muscovy’s military formation also clearly reflects its steppe origins. 55 It
did well in resisting Tatar attacks because it recognized how Mongolian
military formations worked. Once Muscovy freed itself formally from the
Mongols in the 1460s, it became a contender for power over the other suc-
cession states of the steppe. This was the process that led to the dramatic
expansion in the mid-sixteenth century of Muscovy eastward against
Kazan’ and Astrakhan and into Siberia. Muscovy, like Kazan’, the Crimea,
76 formation of states

and others, was a succession state of the Golden Horde, and it raised itself
to power by deploying the familiar techniques of steppe politics: using sup-
port from the Khans to legitimate its rulers’ local authority while manipu-
lating the Khans against one another to create maximum autonomy.
Such striking similarities between two very disparate societies, nomadic
and settled, demonstrates the power of cultural borrowing and the great in-
fluence a conqueror can have on the subjugated society. Muscovy’s close
contact with the steppe led it to adopt steppe institutions in order to
strengthen the rulers’ power. Ming China, more closed off from the steppe
by the defensive barrier of the Great Wall and the ritualized tribute system,
did not adopt nearly so many Mongolian institutions, but its successor, the
Manchus, would use methods similar to those adopted by Muscovy two
centuries earlier. The general trend in recent historiography of this period is toward a
more polycultural analysis, one that does not try to make a rigid separation
between essentially “Russian” and “Mongolian” elements but recognizes
the creative role of cultural mingling in the transformation of states. It at-
tempts to avoid the Eurocentric and colonialist premises of nationalist his-
toriography that saw the Mongols as nothing more than cruel Asiatics or a
gang of bandits.
57 Certainly the initial impact of the Mongols on all Eur-
asian states was destructive, but after the conquest, the Mongols promoted
the revival of the caravan trade, from which Russian princes profited greatly. The Mongols ruled Russia indirectly, by staying in the steppe to pre-
serve the classic nomadic warrior’s way of life. In China and Iran, where
they occupied settled regions, the urban garrisons lost their dedication to
the military discipline of the grasslands. Furthermore, since pasturelands
were scarce in agrarian China, the Mongol military machine decayed from
within as urban warriors began to assimilate to Chinese ways. The indirect
rule over Russia meant that Mongol domination lasted a century longer
there than in China or Iran. It also meant greater Mongol influence on
Muscovite institutions than in China or Iran. This happened despite the
“larger social distance” between Mongols and Russians created by the sep-
aration of Mongolian steppe dwellers and Russian settled agriculturalists.
The Mongol solution to the dangers of assimilation as practiced in the
Kipchak Khanate required separation of the realm into two spheres, one
nomadic and one settled. Periodic raids and invasions of the settled zone,
with limited official representation and intervention through tribute collec-
tion, kept the settled area intermittently quiescent, while the nomads could
maintain their pastoral way of life relatively undisturbed. The Chinese-
Iranian solution brought greater administrative elaboration in the settled
ming, muscovy, and siberia 77

region, closer contact of conquerors and conquered, and in the end under-
mined the basis of Mongolian domination. In the centuries that followed,
these two basic choices—separation or assimilation—faced the Manchus
as well.
Muscovy began its great expansion eastward by conquering the steppe
state of Kazan’ in the mid-sixteenth century. The prevalent nationalist his-
toriography of this conflict depicts a rising Christian Muscovite state strug-
gling to overcome the united opposition of the “remnants of the Golden
Horde” composed of Turkish and Mongolian tribes together with Islamic
merchants. As Edward Keenan and Jaroslaw Pelenski argue, this ideology
is a later creation of Russian Orthodox priests, legitimating Muscovy’s con-
quests as a combined religious and military crusade.
59It does not represent
the actual relations between Muscovy and the steppe polities in the six-
teenth century. Rather than depict Muscovy as a separate state struggling
against “remnants of the Horde,” it is better to see Muscovy along with the
Tatar Khanates as a group of successor states to the Chinggisid empire,
each seeking a modus vivendi in new conditions. The Eurasian steppe has often been called a “land sea,” whose cities
are its ports and caravans its convoys.
60But the nomadic inhabitants of the
steppe are the third crucial element that distinguishes land from water. The
catastrophic decline in importance of the steppe began with Timur’s de-
struction of its economic bases in the 1390s. Until the 1460s centrifu-
gal forces directed power to peripheral states, which struggled to consoli-
date their rule. Then these rival states (Moscow, the Crimea, Kazan’, Siberia)
aimed to stabilize their relationship as they cut back the Great Horde’s cen-
tral power. Finally, from 1520 to the 1550s, the three states contended with
one another over the power vacuum left by the collapse of the Horde, and
Moscow emerged as the dominant force in the western steppe. Muscovy was far more closely linked to steppe politics than its contem-
porary, the Ming dynasty. The Muscovite rulers knew intimately the game
of nomadic politics, and they effectively exploited divisions within the
steppe polities for their own ends. But in general, they did not seek military
conquest of the steppe. Muscovy’s main strategic concerns lay to the west,
against the Polish-Lithuanian state. They tried to keep peace in the east in
a situation of divided and confusing allegiances, as they also tried to use
the commercial resources of the steppe to build up the state treasury in
preparation for war in the west. The major activities of Muscovite state
78 formation of states

builders in the sixteenth century foreshadowed those of Peter the Great and
his successors.
The Khanate of Kazan’ (1445–1552) was a confederation of agricultural
and nomadic peoples whose nucleus on the Volga River included Tatars,
Turks, and Muslims. Princes of dominant clans administered peasant com-
munes headed by elders, and these groupings were collectively called the
“Land” (Zemlia). Above them, the Khan, a descendant of the Chinggisid
Jochi, conducted diplomacy and war, collecting his revenues from personal
landholdings, levies on the local aristocracy, and taxes on trade. The oppo-
sition between the Khan and his court on the one side and the “Land” on
the other was common to all Turkic polities, the Muscovite state, and the
Manchu Qing empire. Succession practices, too, showed commonalities de-
riving from the Turkish tradition, following the “bloody tanistry” of pass-
ing the throne first through all the brothers, then to the eldest son of the el-
dest brother. This pattern ensured frequent instability and power struggles
upon the death of each Khan. The divisions within the polity created open-
ings for exploitation by rival states that could back contending parties to
ensure their dominant influence. The Muscovite play within Kazan’ again
mirrors the play of the Manchus within the Zunghar state on a much
grander scale in the eighteenth century.Kazan’ was not technically a state. It was formally a iurt,or territory
of the Golden Horde. In fact, however, as a major trade entrepôt on the
Volga, it collected significant revenues from both riverine and overland
trade with nomadic regions. Muslim merchants controlled the main Volga
River routes, but Russians gained rising influence. Merchants and the Land
generally favored peaceful trade, while the Khan and his court might seek
to expand the territory by war. Other actors in this steppe political game included the Nogai confedera-
tion of nomads, the Crimea Khanate, and the Ottoman empire. The Nogai
were a purely nomadic confederation, extending east from the Volga to the
Irtysh River in Siberia. They had no fixed capital, and were ruled by a coun-
cil led by a Grand Prince. They, not Muscovy, were the decisive forces in
Kazan’s foreign policy. For Muscovy, the Nogai were most important as the
main source of horses for riding, just as Muscovy was the Nogai’s main
source of income. The horse drives to Moscow in the sixteenth century
brought as many as thirty to forty thousand horses to the capital annu-
61This was three times as many as the Ming usually obtained. Here is
ming, muscovy, and siberia 79

another indication of Muscovy’s intimacy with and Beijing’s distance from
the steppe: Russians secured their main military mounts in the capital itself;
Chinese confined horse trade to the frontiers.Early encounters of Muscovite princes with the Khans of Kazan’ were
pragmatic and cooperative. Each of the principalities tried to outbid the
others for a iarlyk,or Khan’s charter.
62 Vasilii Vasilevich of Moscow was
granted a iarlykby the Khan Ulu-Magmet investing him as Prince of Mos-
cow. Even though Ulu-Magmet later attacked and defeated Vasilii, they
negotiated a peace which lasted until Vasilii’s death in 1462. Commerce be-
tween Moscow and Kazan’ flourished. Ivan III (r. 1462–1505), Vasilii’s suc-
cessor, focused his attention on his main enemy to the west, the Grand
Prince of Lithuania, and strove to keep stability to his south and east
through an alliance with the Khan of the Crimea. Stability on this frontier
left him free to attack Novgorod, the great northwestern trading city, and
force it into submission in 1478.
63Despite sporadic raids by Akhmed Khan
from Kazan’, Ivan kept good relations with him, and the necessary horse
drives by the Nogai continued. Akhmed’s desperate attack on Moscow and
its failure in 1487 “marked the entry of Muscovy as an important force in
the political dynamics of Volga and steppe politics.”
64 The collapse of the
Golden Horde had left a power vacuum which would eventually bring all
the successor states into conflict with one another. Nevertheless, stability lasted from 1480 to 1510 based on the Moscow
Crimean alliance balanced against Kazan’. The final remnants of the Great
Horde were mopped up when Mengli Girei, the Crimean Khan, captured
Astrakhan in 1502. After this, hostility between Moscow and the Crimea
grew. The Crimean Khan broke with Moscow and allied with Lithuania.
When Vasilii III (r. 1505–1533) succeeded Ivan, he increased Muscovite in-
tervention in Kazan’ against the wishes of the Crimean Khan. A contested
succession to the Khanate in Kazan’ led certain groups to invite Moscow’s
intervention on one side and others to invite Crimea to intervene on the
other. The Land invited the Crimean candidate to become Khan in 1521,
but aggressive moves by Kazan’ against Moscow were defeated when
princes in Kazan’ deposed their Khan with Moscow’s support. The main
point is that internal Kazan’ politics drove developments in this period, and
Kazan’ did not take direction from either Moscow or the Crimea. By the
1540s Moscow mobilized against Kazan’, but also aimed to reach accom-
modation with the princes of the Land in Kazan’. Notably, nationalism or
religious loyalty played very little role in these alliances: Tatars, Muslims,
and pagans fought on Moscow’s side, and Kazanians failed to get support
from their fellow Muslims. There never was a united Turkic-Muslim front
against Orthodox Moscow.
80 formation of states

Moscow’s siege of Kazan’ in 1545 brought the collapse of political insti-
tutions in Kazan’. Ivan IV came of age in 1547 and was crowned Tsar.
When he left on his campaign against Kazan’ in 1549, he gained considerable
support from Kazan’s people and was prepared to enter the city peacefully,
when certain groups within the city rebelled against the arranged interven-
tion. Ivan finally had to assault the city, which fell to him in October 1552.
In sum, Kazan’ fell primarily because of internal divisions. It was unable
to balance the differing interests of nomadic Nogai, Siberian, and Mongol
military aristocracy, Muslim merchant classes, and Muscovite agents. The
Muscovite agents knew well how to maneuver these factions in their own
interest, just as the Manchus used their close relations with the Mongols to
prevent a united front against them. Like that of the Manchus, “Muscovy’s
participation in steppe politics was not that of an outsider or intruder, but
that of friend and brother, i.e. a participant in a system to which it owed its
political origins and traditions.”
65Both Muscovy and the Qing grew out of
the traditions of Central Eurasian state competition but learned to adapt
their institutions so as to make maximum use of the settled societies they
ruled. At the same time, they used their knowledge of the steppe to ensure
that no future rivals would unite against them. Both succeeded brilliantly in
this double expansion into settled and nomadic realms. The conquest of Kazan’, however, marked the end of Muscovy’s active
participation in steppe politics. For the next century, it turned gradually
westward to nourish its growing empire, and it expanded across Siberia,
north of the steppe, seeking the wealth to support its territorial and admin-
istrative ambitions. Siberia was to Muscovy what Russia had been to the
Mongols: a peripheral region to be ruled indirectly at low cost, yielding as
much wealth as possible from the native population for the purposes of the
Khan or Tsar.
“There was from the beginning, the Russian land...always one Russian
land, where now the city of Kazan’ standeth.”
We continue the story of Muscovite expansion across Siberia, but let
us first examine how the Tsars justified their conquests. Accounts of Mos-
cow’s relations with Kazan’ before the conquest diverge considerably from
the justification of the conquest by chroniclers after the event. As we have
seen, diplomatic and archival sources provide a picture of Muscovy and
Kazan’ as common but rival successor states within the world of the Mon-
gol Horde. They conducted commercial and diplomatic relations with each
other on pragmatic and friendly terms. Divisions within the Khanate some-
ming, muscovy, and siberia 81

times led to Muscovite interventions on one side or the other, but for most
of the period from 1450 to 1550, the dominant rulers in Kazan’ acted inde-
pendently of Moscow. Religious or cultural differences were not the funda-
mental driving forces of the relationship. In the end, Moscow intervened at
the invitation of elements within Kazan’ to protect them against rivals sup-
ported by the Crimean Khan.Russian chroniclers, however, depicted a world of sharp, irreconcilable
struggle between two forces alien to each other: the barbarian pagans of
Kazan’ and the virtuous Christians of Russia. They separated radically the
two regimes, stressing the national, cultural, and religious differences. They
also provided spurious legitimations for the conquest based on claims of
continuous Muscovite rule from the antique past. The ideological history of Muscovite-Kazanian relations produced by
clerical chroniclers in the late sixteenth and early seventeenth centuries
demonstrates strategies of legitimation that would be used to justify other
conquests of Central Eurasian states by settled agrarian regimes. First was
the false legal claim that the Muscovite ruler had an established right to
grant investiture to the Kazan’ Khans ever since 1487, when Ivan III had in-
tervened on one side of a dynastic struggle in the Kazan’ Khanate.
the Russian rulers made a second legal claim, that Kazan’ was a patrimony
(votchina orotchina, i.e.,iurt) of the Muscovite Grand Princes. Muscovy
also invoked the “right of conquest” derived from its victory in war, and
the idea of continuity of the Russian princes, claiming that they had ruled
over the Tatar Khanate lands since antiquity. The Kazanskaia Istoriia of the 1590s, more historical fiction than chroni-
cle, incorporated these ideological arguments into a very widely read ver-
sion of the struggle that was taken as fact by many successive historians.
This retrospective justification rested on two central themes: the promi-
nence of territory (the land) and the claim to unity (one Russian land). Both
of these themes fit much better into the new territorial states built after
the conquest than into the fluid personal relations before it. Only careful
study of archival sources and diplomatic documents independent of the
Kazanskaia Istoriia has reconstructed the truly complex intercultural nego-
tiations that marked this and other encounters on the steppe. The remarkable similarity of arguments in the Russian chronicles to later
nationalistic justifications of conquest, and to nationalistic Chinese inter-
pretations of the Qing conquest of Central Eurasia, demonstrates two im-
portant conclusions:
• The ideological reconstruction of the history of the encounter of settled
and Central Eurasian regimes displays remarkable similarity at both
ends of the steppe.
82 formation of states

• In both cases ideological reinterpretation began soon after the comple-
tion of the conquest; it did not have to wait for nineteenth-century na-
Instead of depicting a complex process of pragmatic negotiation, in
which Moscow or Beijing often compromised and often acted by norms
well understood by parties in the steppe and in the settled regime, the
chroniclers substituted a much simpler opposition of “barbaric” Tatars, or
nomads, and “civilized” rulers. The victors claimed to have held continu-
ous control since antiquity over the territories they had just recently con-
quered. With this claim, those who resisted could be classified as “internal”
rebels, not autonomous state powers. The expansion of the state appeared
as a continuous, organic process of incorporation of peoples who naturally
belonged to its dominion. Both China and Russia built on this ideology of
incorporation to create their modern nationality policies. One element notably different in the Russian ideology from the Chinese
was the stress on religious opposition. Russian clerical chroniclers naturally
brought Providence into their accounts of battles, describing them as God-
inspired conflicts between pagan forces of darkness and Christian forces of
light. The sins and vices of Christians explained Russian defeats, and God’s
grace and intervention in a just cause explained Russian victories. Chinese
legitimation did not invoke sharp religious oppositions between victors and
vanquished, because the Qing incorporative religious policy accepted Bud-
dhism as part of the system. Nevertheless, Chinese official writers still saw
Heaven’s hand behind the emperor’s victories, casting out the opponents as
“bandits” who defied Heaven’s will. Chinese legitimation also invoked long-standing continuities with previ-
ous imperial regimes that claimed sovereignty over Central Eurasia. It too
treated the resisting autonomous Mongol states as “internal” rebels on
what was essentially Chinese territory. It drew a sharp cultural boundary
between the “raw” barbarian Zunghars, who deserved extermination be-
cause they did not belong to the civilized realms, and the “cooked” Mon-
gols, who had voluntarily submitted to Manchu rule. Although direct inter-
vention by an active Providence was not invoked, the Chinese emperors did
claim that their successful conquest reflected the natural order of the cos-
mos, and they inscribed this view on stelae all over the empire. This effort to turn the unpredictable, mixed, pragmatic interactions of
armies, states, and people into foreordained binary conflicts is of course
common to many other rewritings of history. In the settled–Central Eur-
asian encounter, however, the inequality of resources makes it especially
difficult to recover a genuinely polycultural perspective. The vast majority
of surviving documents come from the victorious Chinese and Russian
ming, muscovy, and siberia 83

states. These states created themselves and constructed their own histories
through their archives and through suppressing the archives and accounts
of alternative versions.
Thirty years after the taking of Kazan’ and Astrakhan, Russia began its sec-
ond great eastward expansion.
70The Cossack Ermak’s defeat of the Khan
of Kuchum in 1582 opened up the forests of Siberia to Russian settlement.
Fortress by fortress, Russians established strongholds on the major rivers,
moving steadily eastward until they reached the Pacific, and across the Be-
ring Strait to Alaska. The motives and process of this second Asiatic expansion differed greatly
from those of the first. No major nomadic state ruler blocked the way.
There was no need to play intricate diplomatic games in the style of steppe
politics. The Tsar himself was not directly involved. Merchant-entrepre-
neurs (promyshlenniki) and Cossacks, semi-independent representatives of
the state under the incomplete supervision of governors (voevody),pushed
the frontier forward and conducted the negotiations with local tribes. Ex-
traction of wealth, not security, was the main goal. Furs, “soft gold,” had
provided critical sources of income for the Muscovite state since medieval
71Exhaustion of nearer territories drove the Russians to pursue sable,
otter, mink, and other fur-bearing animals farther eastward. Like slash-
and-burn agriculturalists or Canadian and American pioneers, Russian
tribute collectors drained a region of its surpluses, then moved on to ex-
haust new regions. Expansion altered the identity of the emerging empire. The image of
Siberia as a land of abundance under colonial domination, a Russian equiv-
alent of El Dorado, grew firmly established in the seventeenth and eigh-
teenth centuries. Peter’s declaration in 1721 that Russia was now an im-
perii (empire), no longer a Tsardom, placed Russia in the European balance
of power as an expanding empire with its own distinct Asiatic domains.
The geographer Vasilii Tatishchev created the conceptual division of Rus-
sian Europe from Asia at the Ural Mountains, inspiring the cartographic
delimitation of the border by the Swedish officer Philipp Johann von Strah-
72 Often neglected in Russian scholarship, the Siberian expansion
into Asia shaped Russia’s fate just as much as its better-known moves west-
ward into Europe. In its early stages, the Siberian expansion was linked to the struggle with
Kazan’. The Principality of Novgorod had profited from trade in gray
squirrel furs with Siberia since the twelfth century. Muscovy crushed Nov-
84 formation of states

Tobolsk, the headquarters of the Russian administration of Siberia.
[To view this image, refer to
the print version of this title.]

gorod in 1471, taking over its fur-bearing tributaries, but it expanded far-
ther to the northeast to acquire the much more valuable sable furs. In 1483
and 1499 the Muscovites attacked to the northeast, crossing the Urals, past
Tiumen’ to the Ob River, and subordinating the Iugri and Voguly tribes.
Moscow traded the luxury furs—sable and ermine, especially—with Euro-
pean and later Ottoman merchants. It also used the furs as diplomatic gifts
for the Crimean Khan, who in turn allowed Muscovite caravans to cross
the steppe and trade at his Black Sea ports. Moscow’s expansion had two
goals: to acquire sables as tribute by directly subordinating the northeast-
ern tribes, and to control the trade route bringing commercial furs to mar-
ket at Ustiug. When Muscovy and Kazan’ were at peace, Moscow’s north-
east aggression slackened, because it could rely on the Kazan’ market for
furs; when hostility grew, Moscow pressed northeastward to secure its fur
supplies. The early sixteenth century saw reduced expansion to the north-
east, but after the conquest of Kazan’, Moscow resumed its Siberian at-
tacks. The Siberian expansion continued and followed on the long-standing
Muscovite drive to gain economic resources from the eastern steppe and
forest zones at the expense of the formerly dominant Khanates of the re-
gion, Kazan’ and Sibir’.
Until the mid-sixteenth century, a Central Eurasian Khanate blocked di-
rect Russian access to Siberia. It was founded in the late fifteenth century by
Nogai Tatars fleeing north from the Russian expansion south of the Urals.
Its capital was at the town of Sibir’, or Kashlyk, near the future Russian city
of Tobolsk. Under Ediger Taibugid, the Khanate paid tribute to Muscovy,
but Khan Kuchum conquered Sibir’ from Ediger in 1563, rejected tribute
payments, and gained control of a large expanse along the Tura, Tobol, and
Irtysh rivers. He then aimed to establish friendly relations with Ivan of
Moscow, sending tribute missions in 1570, but Russian settlers had already
begun moving into his territory. Gregory Stroganov had established a tax-
exempt settlement on the Kuma River in Perm in 1558. In 1579 a nephew
of Gregory’s hired the Cossack ataman(headman) Ermak Timofeevich to
protect his possessions from Kuchum’s raids. Ermak until this time had led
a freebooting pirate’s life plundering caravans and Tsarist officials along
the Don and Volga rivers. Ermak, leading 840 men, in 1582 defeated Khan
Kuchum’s forces at the Tura River and forced the local Ostiak and Vogul
tribes to pay tribute. He captured the Khan’s capital, and the Khan fled
south. He offered the newly conquered lands to Tsar Ivan IV and begged
the Tsar to forgive his past crimes. By 1583–84 he reached the mouth of the
Tobol River, but in 1584 or 1585, besieged by Tatars, he drowned in the
river, and his troops retreated.
74Part bandit, part folk hero, Ermak opened
up Siberia to Russian colonization.
86 formation of states

Consolidation of the Siberian holdings continued under the reign of
Boris Godunov (regent 1584–1598, Tsar 1598–1605), with the founding
of the first major fortresses (ostrogi)at Tiumen in 1586 and Tobolsk in
1587. Tara, founded in 1594, became the main base for operations against
Kuchum, driving him into the Nogai Horde in 1598, where he was put to
death. According to George Lantzeff and Richard Pierce, the founding of
Tara, designed to secure the safety of the trade route used by Bukharan
merchants, “might also be considered the first step of the Russians toward
central Asia.”
75Step by step the fortresses marched eastward—to Narym in
1596, Eniseisk in 1619, Yakutsk in 1632—reaching the Pacific coast in
1649. The Russian strategy of conquest was based on rivers, portages, and the
ostrog (fort).
76First, Cossacks explored the river valleys, then voevodyfol-
lowed with soldiers who built a fort as a base to annex further territory.
The dispersed tribal peoples of the north could not offer any systematic re-
sistance, much as they detested the inexorable tribute exactions. The Tsars
were mainly preoccupied with Russia’s frontiers to the west or south, so
they left the local voevodyalone.
Supplying even these small garrisons was difficult, since the Siberian for-
ests produced low agricultural yields. The native peoples could live off
meager agriculture because of their active hunting and fishing, but the Cos-
sacks, ignorant of local resources, could not manage to support themselves.
Desperately short of food, they plundered native villages to find caches of
grain, dried meat, and fish. Permanent garrisons could only survive on reg-
ular shipments of supplies from Moscow and the west. During the “Time of
Troubles” in the early seventeenth century, “hungry, without supplies and
reinforcements, the garrisons dwindled from death and desertion.”
Vasily Poyarkov’s expedition to the Amur in the 1640s excited great in-
terest when he claimed to have discovered broad river valleys with fertile
fields, a large population growing crops, and an abundance of sable and
fish. He boldly proclaimed, “The warriors of the Sovereign will not go hun-
gry in this land.”
79Before they discovered the Amur, his expedition spent a
horrible winter eating bark and roots, during which forty of his men died of
starvation. It was their own fault, because they had plundered the local
Daur people to relieve their grain shortages, causing the natives to flee from
the Russians and abandon their fields. The Amur promised great riches that
could feed the other Siberian garrison towns, obviating the need to import
supplies from faraway European Russia. This was the main motivation behind the fortification of Albazin by
Erofei Pavlovich Khabarov in 1650. When Khabarov discovered that the
local Daurs paid tribute to China, he drew up grandiose plans, supported
ming, muscovy, and siberia 87

by Moscow, to attack the Chinese empire and seize its large supplies of
gold and silver. Manchu troops attacked in 1652, but temporarily with-
drew, convincing the Russians that they could occupy the region success-
fully. Wild rumors spread about the wealth of the Amur, leading farmers,
promyshlenniki,and soldiers to desert their fields and garrisons and rush to
the region. But the Russians were soon disappointed. After further clashes with the
Chinese troops, the expedition leader Onufry Stepanov reported, “As for
grain, there is very little of it on the Amur because the Bogdoi Tsar [Chinese
emperor] has forbidden the natives to sow grain and has ordered them
to move into his territory.”
80 In 1658 the Manchus surrounded and de-
stroyed Stepanov and his men and killed him. From 1658 to 1672 the
Amur became a no-man’s-land, to whose defense Moscow was not com-
mitted. Caught in a freebooters’ camp beyond Moscow’s control, settlers
suffered food shortages, because the desperado Russians had driven away
the native peoples. By the time of the Nerchinsk negotiations in 1689, Moscow had will-
ingly given up the region, realizing that the Amur could not provide food
for the Siberian centers of the fur trade. In short, food shortages in Siberia
drew the Russian state to the Amur, but the primary importance of the fur
trade led them to give it back to the Chinese. Contrary to many accounts,
the Russians had no consistent commitment to territorial aggrandizement.
Their drive east was in pursuit of trade and food, not land. Regional ecol-
ogy determined the Siberian expansion, as the exhaustion of furs in one
area of the north drove the traders and Cossacks farther east. Once the fur
trade was exhausted on land, the fur drive continued to the Pacific, relying
on sea otters as the main source. By the nineteenth century, when profits
from fur declined, Siberia’s image had changed from a land of abundance to
a land of desolation. Once the seductive target of traders, adventurers, nat-
ural scientists, and runaway serfs, it became the land of exile, frost, and
In Alan Wood’s words, “It was, after all, the conquest and settlement of
Siberia in the late sixteenth and seventeenth centuries which more than
anything else . . . originally transformed the land-locked medieval Tsardom
of Muscovy into the mighty Russian Empire.”
82 The conquests of Kazan’
and Siberia played a crucial role in establishing the most distinctive feature
of the Russian state: Its vast territorial expanse. The autocratic power of
the Russian Tsar was the second feature most noted by European observ-
ers. But this aspect is easily misinterpreted. Although from the days of Ivan
IV the Tsars claimed total authority over their dominions, in practice their
control was much more limited. Likewise, it is misleading to interpret Rus-
88 formation of states

sian state building as solely a coercive process. 83The conquest and rule of
Siberia involved a motley collection of different groups, none of which was
completely under the thumb of the Russian state, and it was not part of a
“master plan” conceived in an ambitious Tsar’s head. Siberia contained a
“complex symbiosis” of many participants, including “military service-
men, hunters, merchants, officials, Orthodox clergymen, fugitive serfs, en-
trepreneurs and tradesmen (promyshlenniki),convicts, religious dissidents,
foreign prisoners of war, cossacks, artisans, adventurers and vagrants.” An
“intricate web of mutual dependency existed between state and private in-
dividuals, military, hunters, peasants, craftsmen, merchants.”
84 Even the
provincial governors (voevody)dispatched by Moscow had great indepen-
dent powers. They combined military and civil authority in their hands,
they enjoyed great personal wealth from compulsory “gifts” and bribes ex-
tracted from their subjects, and they acted with almost autocratic power.
But they too did not control entirely the influx of “wandering people”
(guliashchie liudi)—who could be runaway serfs, deserting soldiers, or
criminals fleeing the law—into these vast spaces. This great influx of volun-
tary fugitives in the eighteenth century increased the Russian population of
Siberia from 169,000 adult males in 1719 to 412,000 in 1792.
Siberian and Chinese Frontiers
The Chinese and Russian frontier settlers expanding into Central Eurasia
resembled each other in many ways. Both faced an alien presence, but not
an entirely incomprehensible one. In this respect, they differed from the
conquistadors of the New World. The Russian explorers of Siberia were
not like their western European counterparts, who returned from the New
World across the Atlantic Ocean with tales of wondrous discoveries and
“marvelous possessions.”
85 Seventeenth-century Muscovites seem to have
had no real interest in the curious customs of other peoples. They did not
try to incorporate them into elaborate schemes of classification, some of
which included foreign peoples in the same category as animals. Musco-
vites stood out by virtue of their single-minded devotion to profit and lack
of missionary zeal, quite different from the Spanish and Portuguese. Fur-
thermore, because there were no “wide seas, high mountains, or other sym-
bolically significant divides” between Russians and the Siberian peoples,
the Cossacks, again unlike the Spanish conquistadors, moved only gradu-
ally into the territory. They “never entered a new world becaus e...they
had not been sent to a new world and because they had no ‘public’ that
wanted to hear about new worlds.”
ming, muscovy, and siberia 89

For the Chinese on the northwest frontier, Mongolia was not a new
world either. Chinese rulers and soldiers had been familiar with nomads on
the steppe for centuries. They certainly regarded them as an alien race, and
often explicitly compared them to animals, but they did not regard en-
trance into the frontier in the seventeenth century as an exotic, radically
new experience. Neither the Ming soldiers camped along the Great Wall
nor the Manchu warriors on campaign felt the need to increase their store
of knowledge about the exotic peoples of the steppe, except for immediate
security needs. Whereas the Russians sought profit, and took interest only
in places that held fur and ivory, the Chinese sought security, and focused
only on immediate threats. To be sure, the Chinese had a much longer experience with their frontier
than the Russians did. Discussions of steppe peoples in China went back to
the formation of the first states in the first millennium bce. Frontier defense
and diplomacy deeply influenced Chinese political philosophy from then
on. Russian state creation came much later, in the ninth century ce. One
could argue, on the one hand, that the Chinese frontier experience was
deeper and more influential than the Russian one because of its prominence
over time and in political debate. On the other hand, the continual gaze of
Russians toward western Europe since the eighteenth century has led many
historians to neglect Russia’s equally important Central Eurasian roots. If
we compensate for the neglect of the subject in Russian historiography, one
could argue that frontier relations were just as significant for Russians as
for Chinese. Both Cossacks and Ming soldiers camped in static fortresses in remote
locations, but the Cossacks were much more isolated. The Russian govern-
ment mostly left the fortresses to their own devices, separated by hundreds
of miles of forest. Each fort created its own self-sufficient community, based
on extraction of tribute, small amounts of arable cultivation, and trade
with local peoples. The Chinese forts, by contrast, were in constant contact
by way of the sentry posts and signal fires along the Great Wall. Shipments
of salt came from the interior in exchange for tea. Paradoxically, greater
connection of the Chinese fortresses with the interior meant less integration
with the steppe, while the greater isolation of the Siberian fortresses forced
their occupants to adapt to their new environment. Trade with northern
peoples was a necessity for survival in Siberia. Thus, despite oppressive re-
lations of exploitation, the Russians did not view the Siberian peoples as
utterly incomprehensible aliens. The Ming Chinese, by contrast, shored up
their defenses to sharpen the differences between themselves and the alien,
threatening nomads. Both empires, expanding on land frontiers, faced
much more blurred boundaries between themselves and the natives than
90 formation of states

did the maritime empires, but their reactions differed. While the Chinese at-
tempted systematically to integrate their defensive frontier regions with the
core of the empire, the Russians left the garrisons to fend for themselves.The eighteenth century marked a shift in the perspectives of both em-
pires. Under Peter the Great, ideas of civilization defined by the Enlighten-
ment entered Russia. Peter promoted scientific investigation of the minerals
and birds and other curiosities of Siberia. Officials who described and clas-
sified its flora, fauna, and peoples began to realize that the Enlightenment
ideals of rationality, cleanliness, and social graces excluded the natives.
Now terms like “alien” and “wild men” (dikii)entered the Russian vocab-
ulary. As the fur tribute declined, the Russian state took a greater interest in
the region and concluded that the timid, helpless, iasak-paying people must
be separated from the Russian middlemen and iasakcollectors. The interest
in classification, the replacement of a focus solely on trade with a view of
native peoples as humans in their own right, and the effort to separate them
from the destructive impact of commerce by state controls all have parallels
to eighteenth-century Chinese frontier policies. The use of Siberia as an ex-
ile colony also parallels the Chinese use of Xinjiang.
In sum, Siberia offers many points of comparison with China’s colonial
domains in Central Asia, particularly Mongolia and Xinjiang. Both regions
mixed multiple ethnic, economic, and social categories. The conflict be-
tween native peoples and conquering settlers, the lure of trade, the promise
of freedom combined with the enforcement of exile, and the predominance
of military rule based on isolated garrison islands in the midst of a sea of
desert or forest form common background conditions. The native peoples
of Siberia were much smaller in number than those of Chinese Central
Asia, and they offered only sporadic resistance. The Mongolian and Turkic
peoples fought back much more stubbornly. Both in the end lost to the
pressure of the armies and settlers of the expanding empires. Disease, too, played an important role in weakening native resistance
on both frontiers. Smallpox epidemics appeared in western Siberia in the
1630s, reducing one native group’s population by half. Spreading east of
the Yenisei in the 1650s, the disease wiped out 80 percent of the Northern
Tungus and Yakuts. By the mid-eighteenth century, it reached the Buriat
Mongols. The timing of the spread westward from China to the Mongols is
very similar, and the effects on the Mongols were nearly as devastating. In
addition to their superiority in arms and wealth, the Russians and Chinese
had the great advantage of centuries of exposure to the germ pools of Eur-
asia, which had become increasingly united by the development of mari-
time trade. Like the Native Americans and Hawaiians, the isolated terres-
trial inhabitants of Central Eurasia received this massive biological shock
ming, muscovy, and siberia 91

at the same time that they encountered superior weaponry and the lures of
gold, tobacco, and alcohol.Conquest and settlement influenced the state structure of both empires
even before they came into contact with each other. The Russians faced
only minimal opposition as they moved east, so they could easily control
huge expanses with scattered fortresses and small garrisons up to the mid-
seventeenth century. Their first contact with the congealing Mongolian
state to the southwest led them to recognize that they had encountered a
much more formidable foe. The building of Kuznetsk in 1618, the south-
ernmost outpost of Siberian expansion, responded to Mongol raids and ri-
val claims for tribute from the Kirgiz Kazakhs of the steppe.
88 During the
eighteenth century, beginning with Peter the Great’s ambition to system-
atize Russian domination, the Russians began building a fortified defense
line against nomadic raids, stretching from Ust-Kamenogorsk in the west to
Kuznetsk in the east. Every 100 kilometers they set up two defense posi-
tions with two hundred soldiers. The aim was to fortify the watersheds of
all major rivers. Kuznetsk was rebuilt in stone in response to the threat of
the Zunghar state, but with the decline of the Zunghars after the mid-eigh-
teenth century, its significance diminished. This great wall-building cam-
paign of the eighteenth century repeated the goals of Muscovy’s seven-
teenth-century defense line in many aspects, as it also echoed the Ming
defense line strategy of the sixteenth century.
89In each case, the expanding
state attempted to defend against repeated raids from a consolidating no-
madic state by investing large amounts of resources in static fortified po-
sitions closely connected with one another. The Russian and Ming states
resembled each other in this respect more closely than the Qing, who un-
dertook an aggressive campaign to destroy the Mongol state at its source in
the steppe. Yet the curious combination of extreme autonomy with extreme autoc-
racy distinguishes Russian state building from that of both Ming and Qing
China. Although Tsar Ivan IV became notorious for the brutal terrorism of
his private security force, the oprichnina,he also designated the power-
ful mercantile Stroganov family as nearly independent landowners on the
eastern frontier, and implicitly permitted them to hire the bandit Cossack
Ermak to create their own autonomous army. This form of “contract colo-
nialism” has more in common with European maritime colonial empires
than with Chinese practices. Europeans often delegated the initial con-
quest, by default or explicitly, to missionaries, adventurers, and trading
companies. In one sense, the Stroganovs were the most businesslike of all
colonial enterprises, nearly completely uncontaminated by rival impulses
of religious conversion. Ming and Qing China, by comparison, mobilized
92 formation of states

merchant capital to develop its frontiers but always maintained official
control of trading licenses. When private traders threatened the tea–horse
trade, Ming officials shut down the system. Russian expansion into Siberia
thus combined features of absolutist western European regimes, Central
Eurasian political formations, and responses to nomadic state formation in
the steppe analogous to those faced by Chinese dynasties.
ming, muscovy, and siberia 93

Central Eurasian Interactionsand the Rise of the Manchus, 1600–1670
A sthe Russians moved east, approaching Lake Baikal, they came
into contact with groups of Mongolian nomads, who proved to be much
more formidable adversaries than the Arctic peoples of western Siberia.
The Mongols had last been united under Dayan Khan in the mid-sixteenth
century. By the early seventeenth century they had fragmented once again
into independent tribes, each under its own leader. In succession, from west
to east, the Russian Cossacks and voevodycame into contact with each of
these major tribes during the early seventeenth century: first the Oirats
(later known as Zunghars), then the Altyn Khans around the Altai, and
later in the 1640s the Eastern Mongol (Khalkha) leaders, the Chechen,
Tüsiyetü, and Jasaktu Khans.
1(See Map 4.)
The Western Mongols during the Ming had been known as the Der-
ben Oirat (Four Oirats), designated variously as including the Khoshot,
Zunghar, Derbet, Torghut, and later Khoit and Choros tribes. In fact the
term “Four Oirats” rarely indicated any formal confederation of these
tribes, who spent most of their efforts fighting one another. In the sixteenth
century some of the Oirats had moved west and subordinated themselves to
Kuchum Khan of Siberia. The collapse of Kuchum’s dominion under Rus-
sian attack set many Central Eurasian peoples in motion, some of whom
sought protection from the Russians, while others aimed to maintain inde-
pendent domains. The extension southward to the edge of the steppe of the
Russian fortresses, particularly Tara, Tobolsk, Tomsk, and Kuznetsk, in-
creased contacts between the two disparate ecological zones of the taiga
and the steppe. Tobolsk was the town most constantly in contact with the

Map 4. Tribal peoples and Russian settlements in the sixteenth and seventeeth centuries.

Oirats. Tara, the farthest south of the Russianostrogy,reported the first
contacts with Oirats in 1606. Russian military expeditions against the
Mongols in 1607 pushed them back in order to secure Russian monopoly
of iasak payments.
The Russians soon realized that with so few military forces facing the
steppe warriors, they had to act carefully. The Mongols began to send em-
bassies to explore negotiated relationships. The first Oirat embassy to Mos-
cow in 1607–8 requested rights to pasture along the Irtysh and Ob rivers
and to trade in Siberian towns in return for an alliance against their Mon-
gol rival, the Altyn Khan. The first Altyn Khan (r. 1567–1627) had built a
powerful state just south of the Mongolian Altai Mountains, which exerted
pressure on the Oirats to the west.
3The Russians granted pasture rights but
refused to commit themselves to an alliance unless the Mongols swore
allegiance to the Tsar and paid iasak.This became the constant refrain of
96 formation of states
Mongol family in the eighteenth century. The woman is preparing kumyss,a fer-
mented drink, for a visiting lama.
[To view this image, refer to
the print version of this title.]

Russian–Mongol relations: Mongols aimed to use Russian backing against
their rivals, Mongolian, Manchu, or Chinese, and to gain the benefits of
trade at Russian fortress towns, while the Russians steadily pushed for the
subordination of individual tribes to the Tsar.In their dealings with the Oirats, Russian officials used the terms they in-
herited from their relations with the western steppe. They referred to the
oaths of submission as shert’,a term derived from the Turkish and Arabic
words shart,meaning a clause of a treaty.
4From Moscow’s point of view,
these were not equal relations but the submission of non-Christian peoples
to the Orthodox Tsar. Russians called offers of negotiation chelobit’e,or
“petitions,” literally “knocking of the head (on the ground),” derived from
the Chinese term ketoutransmitted through the time of Mongol rule. In
Russian accounts, the Mongols begged officials to “allow them to be under
our Tsar’s lofty hand.”
5Whatever words they used, the Mongols’ inten-
tions were quite different. Mongols generally saw the Russian demands for
oaths and tributes as temporary expedients, indicating alliances between
equal powers, while the Russians saw these signs of allegiance as perma-
nent recognition of subordination. Very few Mongolian leaders ever ac-
cepted permanent subordination to the Tsar, the one exception being the
rise of the manchus 97
Encampment of the Derbet Mongol ulusnear the Volga in the eighteenth century.
The memorial in front was built to honor a deceased lama in 1772.
[To view this image, refer to
the print version of this title.]

Torghuts of the Western Mongols, who moved thousands of miles west to
settle on the Volga River. Most maintained the essence of freedom in the
steppe by preserving their rights to mobility.In 1608 the Oirats defeated the Kazakhs and drove the Altyn Khan off to
the southeast, so they no longer needed Russian protection. They rejected
Russian demands for submission, but they increased trade activity at the
fortresses of Tomsk and Tara. Mongols brought 550 horses to Tara in 1607
in exchange for money, writing paper, and cloth. Thus was established the
essential basis of commercial exchange between Siberia and the steppe. The
Siberians, like the Chinese, suffered most acutely from a shortage of horses,
very difficult to raise in the forests. The Mongols supplied these essential
transport animals from the grasslands in exchange for the manufactured
products of settled civilization. Later there appeared the other essential
source of Mongol revenue, furs, which they controlled as intermediaries in
the trade between Russia and China. The Russians nearly succeeded in obtaining the genuine submission of an
98 formation of states
An obo, or sacred Mongolian site, usually marked by a pile of rocks and prayer
flags. This monument honors a deceased Khoshot prince.
[To view this image, refer to
the print version of this title.]

important Mongolian leader in 1614, when the Oirats suffered heavily
from a cold winter that killed their herds and from losses to Altyn Khan
which obliged them to pay him tribute. The governor of Tobolsk offered
peace to the suffering Oirats if they would submit to the Tsar, and sent the
embassy of Ivan Petrov and Ivan Kunitsyn to stay for two months with
leading Mongol princes. Dalai Taisha, the strongest leader of the Derben
rise of the manchus 99
The Tibetan Buddhist deity Yamantaka embraces his consort. Mongolians and Ti-
betans respected the terrifying power of this manifestation of Buddha.
[To view this image, refer to
the print version of this title.]

Oirats, claimed to be willing to submit to the Tsar, and offered ten thou-
sand warriors to serve in attacks on the Nogai. When Dalai reneged on his
offer, the governor prepared military expeditions to enforce his subordina-
tion, but he was rebuked by Moscow, which ordered him to maintain peace
with the Mongols. Russia in the early seventeenth century, weakened by
wars with the Poles and Swedes, wanted to avoid military action in the east.During this mission, the Russian envoys tried to learn as much as they
could about the fabled Chinese empire. They asked about its size and popu-
lation, its allies, and the beliefs of its people. They obtained some valuable
though misleading information from their Mongol hosts, who told them
that the Chinese lived in brick cities, on large rivers, whose names they did
not know, and their ruler (the Wanli emperor) was called “Taibykankan”
(Da Bogda Khan?). According to the Oirats, the Altyn Khan and the Chi-
nese shared the same language and beliefs. Clearly, all they knew of China
were the Mongols settled near the Ming dynasty frontier.
Khara Khula (d. 1634/35?), the Zunghar prince, also sought Russian aid.
In 1620 he sent an embassy to Moscow and received in return the first edict
of a Russian Tsar to a Mongol leader, stating, “You, Khara Khula, will sub-
mit yourself to Russia and gain our protection from your enemies.”
7But re-
lations between Russia and the Oirats were broken off for fifteen years as
Khara Khula began his wars with the Altyn Khan Ombo Erdeni to his east. Meanwhile, to escape civil war, the Torghut Mongols (known to the Rus-
sians as Kalmyks) began moving west, toward the Russian fortresses of Ufa
and Tiumen. The Torghuts also refused genuine submission to the Tsar, but
they wanted to trade horses at Siberian towns for furs, rifles, and iron.
Moscow rejected their demands. When, after 1623, they began raiding
Russian territories and iasakpeoples, Moscow ordered renewal of regular
trade if the Kalmyks would maintain their distance from the Russian towns
and cease their raids. In the 1630s the Torghuts began their great migration
across the Kazakh steppe to their ultimate settlement on the banks of the
8Other groups continued to migrate there through the end of the
1640s, and they kept up ties with their tribesmen back in Mongolia. Thus
one part of the Oirat confederation became a genuine dependent of the
Russian Tsar as the remaining tribes unified under Zunghar leadership in
western Mongolia. Next to the Torghuts, the early Altyn Khans were the most active of the
Mongols in seeking Russian aid. Russian relations with the Altyn Khan,
who ruled northwest Khalkha Mongolia, differed from those with the
Oirats, because of the Khan’s proximity to China. In 1616 the Khan enticed
a Russian embassy with the offer of access to the China trade, one month
away through his territory, in return for supplies of rifles and furs. The em-
100 formation of states

bassy of Ivashko Petlin, Russia’s first major embassy to China, did pass
through Altyn Khan’s territory unobstructed, but when the Khan asked for
military aid against his Oirat rivals, the Tsar, angered, cut off relations
through the 1620s. Here too the Russians followed a consistent approach,
aiming to stay out of internecine conflicts while securing access through
Mongolian territory to China. Likewise, the Mongols, for their part, aimed
to draw the Russians to their side through tactical declarations of submis-
sion to the Tsar. The new Altyn Khan in 1631 renewed relations with Rus-
sia, offering to swear an oath of allegiance, in order to gain aid against his
new rivals to the south. He also sent the first tea caravan to Russia from
China in 1638. By the end of the 1640s, however, the Altyn Khan had allied
with the rising Manchu power, and his successors refused to recognize pre-
vious oaths sworn to the Tsar. Interestingly, in the 1660s, the last Altyn
Khan, Luvsan, asked for Russian aid to help him build a fortress on the
Khemchik River at Tuba, but he was refused. His submission to the Qing in
1681 and his death in 1682 ended the autonomous power of the northwest
Khalkhas.The Russians entered the Mongolian steppes just as the fragmented tribal
heads were beginning to unify themselves under Oirat leadership, and as
the Ming dynasty was in decline. China had little political influence in the
region, but the lure of its market for furs drew the Russians steadily east-
ward. Even in this early period it became clear that Russo-Mongol relations
would be tense, because Mongols refused to submit unconditionally to the
Tsar, and Russians mainly considered Mongolian territory a way station to
China. Chinese nationalist historiography to the contrary, the Russian Tsar
never allied with the Oirat–Zunghars against the Chinese state, but despite
claims of Russian and Mongolian historians, their relations were not uni-
formly harmonious.
Building the Zunghar State
While the Ming fell and the Russians encroached, from the late sixteenth to
late seventeenth century the Western Mongols tried to create a unified state,
against great odds. The Oirats in the late sixteenth century were extremely
fragmented. After individual chieftains lost battles with Eastern Mongolian
chiefs, they were driven west to the Altai Mountains. At the same time they
were attacked from Kazakhstan and from Turfan. By the 1590s some Oirat
princes had been forced to submit to the Kazakhs in order to keep their pas-
tures. Altyn Khan began a major drive against the Oirats in the 1580s. When
rise of the manchus 101

he sent an army searching for their leader, he found no one unified confed-
eration but instead autonomous princes scattered along the Irtysh River. In
1616 Russian envoys who contacted the Oirats learned that they were di-
vided into four tribes: the Derbet, led by Dalai Taiji; the Zunghar under
Khara Khula; the Khoshot under Baibagas; and the Torghut under Kho
Urluk. Dalai Taiji was the strongest among the four, but he was not the
Khan. Baibagas of the Khoshot was called Khan, since he was descended
from the Khans who founded the Oirats in the thirteenth century.
though Baibagas Khan (d. ca. 1630) served as the leader of the assembly of
princes (chulgan dargi, Ch.qiuergan), he had no real independent power.
Each of the Khans controlled his own people autonomously. In response to
Altyn Khan’s threat, however, Baibagas Khan was able to collect a joint
army of thirty thousand Khoshot, eight thousand Derbet, six thousand
Choros, four thousand Khoit, and two thousand Torghuts.
11Although this
was a first step toward united action, the Oirats remained divided by con-
stant rivalry between the princes, especially between Baibagas Khan and
the Zunghar chief Khara Khula. Internal struggles over pasturelands coun-
tered by efforts to unify in response to outside threats would sharpen con-
siderably in the early seventeenth century. At a general assembly in 1616–17 the Oirats agreed to establish internal
peace and not to help those who attacked fellow Oirats, but the rivalry of
the princes continued. At the same time certain tribes began to establish re-
lations with the Russians moving across Siberia. Kho Urluk of the Torghuts
had begun negotiations for submission to the Tsar with the governor of
Tara in 1607.
12 When the great war with Altyn Khan broke out in 1608,
most Oirats lost interest in Russian ties, as they briefly achieved unity
against the threat from the east. But in the 1620s both Khara Khula and
the Altyn Khan sent embassies to Moscow seeking aid against each other.
Khara Khula, notably, expressed his need for iron weapons. Moscow ac-
cepted Khara Khula’s submission, as noted in the edict quoted earlier, but
refused to intervene on either side. By 1622 the dissatisfied Khara Khula
and other Oirats were raiding Kuznetsk for weapons.
During the 1620s the Oirats succeeded in putting together a coalition of
the four tribal leaders to march against the Altyn Khan. In the unified Oirat
army, Baibagas Khan commanded sixteen thousand troops, Khara Khula
led six thousand troops, and three other leaders commanded a total of
fourteen thousand men.
14In 1623 the combined forces attacked the Altyn
Khan, with indecisive results. They took many captives, but Khara Khula
lost many of his own men. Further battles led to victory over the Altyn
Khan in 1628–29. Oirats who had fled for protection to the Siberian for-
tresses were now able to move back into their homelands in Zungharia and
eastern Turkestan.
102 formation of states

Khara Khula tried to increase his influence by mediating disputes over in-
heritance, but he remained only one of the prominent leaders within the
Oirat confederation. A major civil war then tore the Oirats apart in 1625.
When one Khoshot prince died, his brothers Baibagas Khan and Chokur
quarreled over the division of his inheritance. Khara Khula and Dalai Taiji
of the Derbet both tried to mediate, but the dispute was resolved only after
Chokur mobilized a ten thousand–man army against his brother and de-
feated him.
15Even when the Oirats faced a serious threat from Altyn Khan,
they could not restrain themselves from internecine conflict. Both Khara
Khula and Dalai Taiji tried to strengthen their unity, but the tribal structure
remained unstable. At the same time, Kho Urluk began the negotiations that would lead his
Torghuts across the steppe to the Volga.
16Most of the Torghuts had left for
the Volga with about fifty thousand families (200,000 to 250,000 people),
removing the major source of division within Zungharia. The Torghuts’
emigration reduced the Oirats’ total strength but aided their unity by re-
moving the most disgruntled tribes from Khara Khula’s control. Tibetan Buddhist influence in western Mongolia proceeded more slowly
than in the east but became a powerful force in the late seventeenth century.
Baibagas was dissuaded from entering a monastery, but many Mongol aris-
tocrats offered their sons to the monasteries. Zaya Pandita (1599–1662),
the adopted son of Baibagas, strengthened the Western Mongols’ unity
through a close alliance with Tibet.
17In 1616 he was sent to a monastery in
Tibet, where he studied Buddhism and tantric arts for twenty-two years.
When he attended the installation of the seventeen-year-old Dalai Lama in
1635, the Panchen Lama gave him the mission to spread Buddhist teach-
ings among his Mongol people through translations of Tibetan texts. On
his return home in 1639, he first went to Baibagas’s son Ochirtu Taiji, who
became the Chechen Khan, but soon received invitations from the other
top leaders of the Mongols, the Jasaktu and Tüsiyetü Khans, and visited
the pastures of the four chief Oirat leaders, including the leader of the
Torghuts. He spent nearly every year of his life traveling from one tribe to
another, in each place conducting funerary and matrimonial rituals, found-
ing temples, preaching, fasting, and translating Tibetan texts. In 1648 he
devised the Todo script, a variant of the Mongolian script designed spe-
cially for the Western Mongolian dialect. During his life, he translated over
177 valuable Tibetan Buddhist texts into Mongolian. In 1650, on one trip
to Tibet, he brought 110,000 taels of silver, which he donated for the pro-
duction of statues and gave in support of the Tashilumpo monastery. He
also asked the Dalai Lama his opinion of the new Qing emperor. Impressed by his great knowledge of Tibetan Buddhism, the Khans gave
him disciples, servants, and large herds of horses and cattle. In return, he
rise of the manchus 103

was able to exert considerable political influence. Zaya Pandita was the key
mediator between the left and right wings of the Oirats when they went to
war against each other after 1657, and he attended the large assembly
(chulgan)held in 1660 to reconcile the opposing factions. When he died on
his return from Tibet in 1662, his disciples brought his ashes back to Lhasa,
where he was honored by both the Red Hat and Yellow Hat lamas, and the
Dalai Lama had a large silver statue made of him. Zaya Pandita founded
the Oirat literary tradition, kept the Mongolian Khans linked to one an-
other with his constant traveling and preaching, and ensured the domi-
nance of Buddhism in western Mongolia. Although he failed to prevent
the Mongols from fighting one another, he kept the rival Khans in contact,
constantly emphasizing their common bonds as patrons of the Buddhist
church. The nineteenth-century Russian explorer Nikolai Przewalski found Bud-
dhism to be a “religion that sapped vitality and hindered progress” in Mon-
golia, as did Owen Lattimore, who thought that Buddhism withdrew the
most productive Mongols from pastoralism and paralyzed the nobility by
fixing their territories around monastic settlements.
18 But Buddhism did
not have the same effects in the seventeenth century, when the Oirats were
building their state. Zaya Pandita offered the Khans their strongest legiti-
mating force beyond the claim to military superiority and descent from
Chinggis Khan. Lamas offered important resources to the Mongolian chiefs,
giving them legitimacy as reincarnated Khans, providing them with a writ-
ing system, and linking them to a cultural system distinct from that of
either the steppe or China. Although Ming officials found Buddhism useful
among the Mongols, they could not control what the Mongols made of
it. Only the Manchus, in the eighteenth century, devised methods to en-
sure that Mongols and Tibetan lamas subordinated themselves to the Qing
state. Most scholars of this period, from Peter Simon Pallas in the eighteenth
century to I. Ia. Zlatkin and Hiroshi Wakamatsu more recently, have ar-
gued that when Khara Khula died in 1635, he passed on a legacy of prog-
ress toward unification that allowed his son Batur Hongtaiji (r. 1635–1653)
to declare himself sole leader of the Oirats, and the founder of the Zunghar
Khanate. Miyawaki Junko, however, has severely criticized this interpreta-
tion. In her view Batur never became Khan, because only patrilineal descen-
dants of Chinggis Khan could assume this title.
19The true unification of the
Oirats as a “Khanate” occurred only in 1678, when Galdan received the ti-
tle of Boshoktu Khan from the Dalai Lama after killing his father-in-law
and rival, Ochirtu Chechen Khan. In her view, the erroneous interpretation
of the Zunghars as having created a “nation-state” under Batur as Khan
104 formation of states

derives from reading sources created in the eighteenth century instead of
looking at the seventeenth century itself.Miyawaki’s interpretation fits well with our project of reconstructing
the contingencies of state formation in this period, and her critique of
retrospective application of inappropriate concepts is well founded. Even
though Batur’s power has been exaggerated, however, we may still at least
credit him with an incomplete state-building project,one that aimed at
further centralization and stabilization of the Oirat community under
Zunghar rule. The title of Khan was held by the Khoshot leader Gush Khan, who took
it on the death of Baibagas in 1630. Gush Khan married his own daugh-
ter to Batur Hongtaiji, so their family relations were close. But Batur, as
Miyawaki notes, could not be the undisputed leader of the Oirats because
he had no direct Chinggisid ancestors and could only name himself “Hong-
taiji,” second in command to the Khan. Gush Khan, Batur, and Ochirtu
Taiji marched together in a military alliance against the Kazakhs in 1634–
35. Gush Khan also led an expedition to Lhasa in 1636 and Batur accom-
panied him in expeditions to Kokonor in 1637.
Batur invites comparison to his Manchu contemporary Hong Taiji. Like
the Manchu leader, he tried to centralize power among the tribal chiefs, al-
though, unlike the Manchus, he could not claim the supreme title of Khan.
(Among the Manchus, the title did not require descent from Chinggis.)
Circumstances favored him, but his legitimacy was never secure. Threats
from the divided Kazakhs in the west and from the Altyn Khan in the east
declined. The departure of the Torghuts had reduced population pressure
on the grasslands. In 1635 the Khoshots, 100,000 of them, had moved to
Kokonor to create their own state. Those who did not want to subordi-
nate themselves to Batur chose exit over resistance. Batur aimed to induce
the breakaway groups to return to a united Zunghar leadership, along
with the Eastern Mongols, but he could not conduct military campaigns
against them. Instead, he strengthened ties with the Russians, renewing diplomatic re-
lations after a hiatus of fifteen years. He sent thirty-three embassies to Mos-
cow during his reign, and received nineteen in return from Siberia.
21By ar-
ranging for the return of Cossack war captives and the settlement of border
conflicts, he competed with the Altyn Khan for the favor of the Tsar. Two
major sources of conflict were at least temporarily resolved: the access to
salt sources and the allegiance of the Kirgiz. Rivalry over the Kirgiz tribes
living on the Yenisei River grew in 1641 when Batur claimed the right to
collect tribute from them, rejecting the claim of the voevodaof Tobolsk
likewise to collect tribute. War was avoided by an ingenious device, the rec-
rise of the manchus 105

ognition of dual sovereignty, obliging the unfortunate Kirgiz to payiasakto
both the Russians and the Zunghars.
22At the same time, Batur was allowed
to expand duty-free trade in Tobolsk, Tara, Tiumen, and especially Tomsk.
A separate quarter of Tobolsk grew up, called the Tatar Settlement. Trade
was mostly conducted under the name of diplomatic gift giving, in which
the Zunghars exchanged their horses, cattle, sheepskins, and furs for handi-
crafts made of cloth, leather, silk, silver, walrus ivory, and metal.
23 “Bu-
kharan” merchants from Turkestan prospered as the intermediaries in this
new Central Asian trade. The Russians, by now well aware that the trade
route to China passed through Zunghar territory, favored the growth of
peaceful trade relations with the Mongols.
Batur demonstrated a clear goal of accumulating resources of trade, agri-
culture, and population in order to build his state’s power. In 1638 he asked
the Siberian voevodato provide pigs and chickens to be raised on Zunghar
farms in newly annexed areas of Turkestan. In 1639 he asked for guns,
armor, and bullets. Told that it was impossible to obtain these items in
Tobolsk, he sent a mission to Moscow for them. In 1650 he demanded
greater “gifts” from the Tobolsk governor in order to ensure the continua-
tion of peaceful trade relations. At the same time he asked for Russian arti-
sans, including stonemasons, carpenters, blacksmiths, and weapon makers,
to be sent to Zungharia.
24Russia agreed to send pigs and chickens but re-
fused to supply weapons makers, consistently avoiding the promotion of
arms flows out of Russia to Mongolia. From 1636 to 1638 Batur did, how-
ever, begin building his capital, a stone fortress and monastery at Kubak
Zar between Lake Yamysh and the Irtysh River.
25 At first it was a small
city with only around three hundred people. Russian emissaries reported
that Chinese and Mongol artisans built it of stone, with walls 100 meters
around and 6 meters high. Chinese, Mongols, Bukharans, and lamas lived
in separate districts. Batur had Turkestani peasants brought in to culti-
vate the fields. Fortresses surrounded the capital, containing four cannon
brought from China. Batur Hongtaiji himself pastured about seven days
away and visited the town from time to time. He asked the Russians to send
him more artisans, and he built several other towns in the next four years.
Like Altan Khan before him, Batur had begun to lay the foundations of a
settled regime, but his city did not last long. After his death, it decayed, and
now no traces are left. Salt became another key resource for the growing state. It was perhaps
the scarcest essential nutrient on the northwest frontier, and its supply had
been a major focus of Ming frontier policy. Siberian settlers also needed salt
in their diet, and there were no sources of it in the forest zone. The Russians
and Mongols had clashed over access to salt since the early seventeenth
106 formation of states

century. In 1611 the Mongols had captured some sources near Tara, forc-
ing the Russians to mount a military expedition to drive them off. In 1613
Siberian settlers arrived at the great salt lake of Yamysh. This lake, how-
ever, lay in a region under Zunghar domination. Until 1620 the Russians
developed the lake and extracted major supplies of salt with no opposition.
Then the Mongols arrived and blocked further production. As armed con-
flicts broke out, the Russians planned to build a fortress at the site, but they
gave up the idea, realizing that the difficulty of access and the lack of pas-
tures and fields prevented support for resident troops. Instead they sent pe-
riodic military detachments to collect salt as needed. But in 1634, two
thousand Mongols attacked the Russian military expedition. Batur offered
to replace armed conflict with peaceful trade. In 1635 the Tsar agreed to
allow Mongol salt mining at the lake, while Batur approved regular salt
caravans traveling from Lake Yamysh to Russia in exchange for increased
trade. Around Lake Yamysh grew a mercantile settlement, which became
the largest trading center in Siberia until the designation of Kiakhta as the
China trade center in 1689. Fairs lasting two to three weeks attracted mer-
chants from all over Central Eurasia, who sold horses, Chinese goods, and
slaves (despite prohibitions) to Russia in return for metals, textiles, and
glass. The Russians promoted the growth of civilian trade but strictly for-
bade the export of weapons and gunpowder.
The great assembly(khuriltai)of 1640 represented the high point of efforts
to gather the Mongols together in a loosely united confederation.
27It gath-
ered the Khans of the Khalkha Mongols, Kokonor, the Volga Kalmyks, and
representatives of the Tibetan church in Tarbaghatai to discuss ways to
avoid internecine conflict. Only the Chahars, already subordinate to the
Manchu state, were excluded. The assembly agreed on a code, which aimed
to regulate disputes, unite the Mongols against outside threats, and in-
crease the powers of Khans and princes.
28Attacks by one tribe against an-
other would be punished by fines, tribes would aid one another against out-
side attack, and each prince would protect his own pastures and return
fugitives from other tribes who fled to him for protection. Tibetan Bud-
dhism was declared the official church of the Mongols.
Most historians credit Batur Hongtaiji and Zaya Pandita with bringing
together this great alliance of Khans and princes, but Miyawaki demon-
strates that the leading force was the Jasaktu Khan of the Khalkhas, and
that the assembly was held in Khalkha, not Oirat, territory. The Jasaktu
Khan faced the most imminent threat from the rising power of the Man-
rise of the manchus 107

chus and had the greatest incentive to form an alliance against them. The
Mongol-Oirat code, however, did not create a founding law of a new Mon-
gol or Zunghar state; like earlier Mongol laws, it applied only to relations
between tribes and left each Khan autonomous in his own domains.
In 1640 the Manchus were not yet a serious enough threat to most of
the Mongols to provide a strong stimulus for a united front. Although some
of the southern Mongols had submitted to them, the Manchus were still
building their state in Manchuria and concentrating on the battle with
the Ming. The code’s provisions were also internally directed. The code
strengthened the powers of individual Khans to settle disputes within their
domains but left them completely sovereign and took no steps toward a
larger confederation. Just at the time when the Manchus were building a
powerful centralized state based on a banner system that crosscut tribal al-
legiances, and Japan’s unifiers were beginning to combine a strengthened
feudalism with a centralized military regime, the Mongols achieved only
partial internal consolidation without creating an even loosely united con-
federation. In the 1640s serious internal wars broke out among the Oirats,
and a decade later the Khalkhas and the Khoshots of Kokonor were coop-
erating with the new Qing state. Although the khuriltaiof 1640 could have been an important step to-
ward the institutionalization of the Mongol state, it lagged far behind its ri-
val states in formation in Russia, Manchuria, and Japan. Early modern
states were built in two stages: first, by a local prince who attained suprem-
acy over his military followers, and second, by a federation of the princes
under a recognized military leader. In Japan, for example, Tokugawa Ieyasu
consolidated his shogunate by combining strong local daimyÃpower
within the domains with a careful balancing of daimyÃforces against
one another. The Manchu leaders created new organizations, the banners,
that incorporated some of the tribal leadership in non-tribal organizations.
They borrowed the idea from Chinggis Khan’s ten thousand–man units
(tümen). The Mongol successors to Chinggis, however, in the much vaster
spaces of Central Eurasia could not join under one recognized leader. In-
stead they remained divided into several rival groups that could at best join
in temporary alliances with one another, but could just as easily ally with
non-Mongol neighbors on all sides. Batur could not be Khan, but he could build a city, increase his people’s
prosperity through trade, and participate in efforts to construct a coalition
against the Qing. On his death in 1653, however, all his efforts fell apart.
One of his nine sons, Sengge, succeeded him, but Sengge’s brothers dis-
puted his control over half of his father’s people. They killed him in 1670.
One brother, Galdan, who had been sent to a Tibetan lamasery, returned to
108 formation of states

take charge during this succession crisis. Galdan became the great unifying
leader the Mongols needed, as he expanded the Zunghar state into the larg-
est power in Central Eurasia.
The Rise of the Manchus
Many historians have described the astonishing conquest of the Chinese
empire by the Manchus, a tribal people dispersed through the forests and
fields of China’s northeast frontier. Here I concentrate on three themes that
will recur throughout our story: the struggle between clan-based organiza-
tion and bureaucracy, the exploitation of an agrarian base, and the role of
Manchu–Mongol relations in forming the early Manchu state. Each of
these themes highlights the interactions between the growing Manchu state
and its neighbors in Central Eurasia while emphasizing similarities of the
challenges faced by all three.The predominant theme in most studies of the rise of the Manchus has
been the transformation of the clan society of the Manchu tribes into a cen-
tralized bureaucratic state. Franz Michael attributed this change from “feu-
dal” to bureaucratic relations to the influence of Chinese advisers and clas-
sical texts. In his view the Manchus, like previous conquerors, needed to
adopt Chinese ways before the conquest in order to succeed: “There was no
essential difference between conquest from outside and from inside the
country...[A]s a general system (the Manchus) had to accept the Chinese
way of life and Chinese civilization.”
Unlike later scholars, Michael argued that “the Manchus never became
completely absorbed into the Chinese culture. They remained the privileged
group of conquerors; the group which retained part of the military and feu-
dal past all through its history.” Yet he still believed that “it was the Chi-
nese civilization which had to be recognized as [the Manchus’] political and
ideological standard.” For Michael, the Manchuness of the Qing derived
only from the early years of conquest; it was not embedded in the perma-
nent institutions of the state.
Michael’s main source was the Kaiguo Fanglue,published in 1789 under
the Qianlong emperor’s aegis, but this text was selectively edited in order to
produce a spurious impression of a Confucianized Qing regime that was
the legitimate successor to the Ming. It fails to reveal evidence of conflicts
between the state founder, Nurhaci, and his nobles and family, and it pro-
vides little evidence of the economic crises that plagued the early Manchu
state. More recent studies, relying on the valuable Manchu language re-
cords in the Old Manchu archive, give a very different picture. They carry
rise of the manchus 109

on the tradition of “Altaic” interpretations of the Qing established by Japa-
nese scholars in the early twentieth century.
33 The Manchu state builders
did not smoothly acculturate themselves to Chinese ways as they built their
state; there were major internecine conflicts within the elite over this
wrenching transformation. Chinese influence in the form of printed texts
was negligible in the early years, because very few texts were translated into
or published in Manchu, and most of them were not philosophical clas-
sics. Instead, the bureaucratization of the state must be seen as a prag-
matic, contingent reaction to the conquest process itself, influenced by
advice from surrendered Chinese of the frontier regions, but also from
Mongolian allies, Manchu nobles, and bicultural “transfrontiersmen” (in
Frederic Wakeman Jr.’s term) who spanned the cultural gaps among the
three peoples. The state lurched from brilliant military victories to severe
subsistence crises; much of the discontented Chinese population under its
rule revolted, but other Chinese officials supported the Manchus against
the Ming. Succession crises threatened to dismantle it, but centralization
overcame factionalism in the end. Bloody tanistry plagued the Manchu state from its origins. Under Ming
influence, the Manchus could also invoke the principle of primogeniture,
but not as a regular practice. Nurhaci (1559–1626), the founder of the
Manchu state, began his career of military conquest in 1583 in alliance
with a few hundred men of his tribe, a confederation with his eldest son,
Cuyeng, and his younger brother Shurhaci. By 1609 the three men had
fallen out; Nurhaci had Shurhaci killed in 1611. Nurhaci at first felt obliged
to appoint Cuyeng his successor, but he later denied the succession to
Cuyeng and had him killed in 1615. Nurhaci named no successor on his
death in 1626. Like the Mongols and their ancestors the Jurchens, the
Manchus found it difficult to agree on the uncontested rule of a single
Military conquest, however, reduced internecine wrangling as it brought
more people under the military state’s control and provided the human re-
sources for further expansion. As Nurhaci and his army subjugated the ri-
val tribes of the Northeast, they incorporated them as companies (niru)of
three hundred men each within the army-state.
35The Hada tribes, defeated
in 1599, became some of the first companies that went into the formation
of the banners, the most vital institution of the Manchu state. The ban-
ner system began to form as early as 1601, derived primarily from the
Manchu practice of organizing military maneuvers as hunting expeditions.
The Eight Banners (Ch. Qi, Ma. gûsa), created in 1615, laid the basis for
the multiethnic coalition that Nurhaci proclaimed as the Latter Jin empire
in 1616. As Mark Elliott points out, the banner companies (niru)main-
110 formation of states

tained ethnic homogeneity, even though the banners(gûsa)mixed Mongol,
Chinese, and Manchu companies together. Thus the banners flexibly in-
corporated members of several different ethnic groups while maintaining
their distinctiveness. The banners cut across kinship connections within the
Manchus, tying them together in a new military and civil organization that
owed total loyalty to its commander.
The banners fixed new ties of solidarity at the base of the new state, over-
riding the kinship loyalties of the early tribes of Manchuria and Mongolia.
Yet each banner commander remained nearly autonomous, a powerful
warlord with loyal troops under him. To create a centralized state, Nurhaci
still needed to unify it at the top. In 1615, along with the banners, he estab-
lished the sunja amban (Five State Councilors), vesting leadership in the
hands of Nurhaci and five of his sons-in-law. He named his four remaining
sons Hosoi Beile (Senior Chieftains). Although Nurhaci appointed the rul-
ers of each banner, these men had the power to choose the leader’s succes-
sor, and commanders “held [the banners] virtually as private property.”
As the Manchus conquered Liaodong and incorporated Chinese into
their state, Nurhaci used his new subjects to limit the power of the Manchu
nobles, or beile.Thebeile were either relatives of Nurhaci or rulers of inde-
pendent tribes before the conquest. They controlled their own people and
captives, but the Chinese were directly subordinated to Nurhaci as Khan,
not to the beile.Nurhaci kept moving his capital farther westward and
southward toward the Chinese border and away from the beile’s territory,
against their resistance. The beileAdun was killed in 1621 for defying the
“order of the state” (gurun i doro)by sowing discord, probably for refusing
to accept the move of the capital. Nurhaci also selected eight scholars,
called baksi, as his principal advisers.
38 These men, like Dahai (d. 1632)
and Erdeni Baksi (d. 1623), were fluent in Manchu, Mongolian, and Chi-
nese. Some may have been originally of Han origin, but their ancestors
went native and joined Manchu clans on the frontier. They wrote procla-
mations in Chinese and translated Chinese classical texts. As “transfron-
tiersmen,” they were multicultural in their background, not simple agents
of sinicization. By offsetting the power of the beile,they served as crucial
aids to reinforcing the Khan’s central power.
Nurhaci had recommended collective rule by the senior beilein his last
testament: “The one who succeeds me should not be one who relies on
force. If this type of man becomes ruler, I fear that by relying on the power
of his position he would offend Heaven. Furthermore, one man’s wisdom is
no match for the wisdom of many. You eight men are the leaders of the
Eight Banners, and if you can rule the nation with a common purpose, then
nothing will be lost.”
rise of the manchus 111

[To view this image, refer to
the print version of this title.]

His eighth son, Hong Taiji (1592–1643), however, as commander of
both the Bordered Yellow and White Banners, had a decisive advantage
in the succession struggle. He probably usurped power by force after
Nurhaci’s death and soon took immediate steps toward autocratic rule,
contravening his father’s intentions.
41 He eliminated Amin, Daisan, and
Manggultai, three of the senior beilewho had ruled with his father. Amin,
for example, who attacked and looted Korea against his orders and mas-
sacred Chinese in Yongping, was indicted for crimes and imprisoned in
42By 1636 Hong Taiji had enlarged the Chancellery, which originated
in his father’s eight baksi,set up the Six Boards and the Three Courts, re-
cruited more Chinese scholars, and dismissed the beilefrom the boards.
The Deliberative Council of Ministers (Yizheng Wang Dachen Huiyi), cre-
ated in 1622 and expanded in 1637, derived from Nurhaci’s collective ad-
visory bodies. But Nurhaci’s councils included the eight highest-ranking
beile, while Hong Taiji excluded Manchu princes from his council. He in-
cluded only the direct administrators of the banners.
Hong Taiji’s Chinese advisers urged him to invade China quickly, and
they supported the concentration of power in his hands. Newly acquired
wealth from conquest, they argued, should go directly to the Khan, to be
distributed by him to the beile.They rejected the collective rule of the beile
and the principle of each beilepassing out his loot autonomously to his fol-
lowers. As one memorialist put it, “Ten sheep with nine shepherds will cre-
ate disorder and disunity.”
44 This move toward authoritarianism clearly
benefited the Chinese advisers, who expected large material rewards from
conquest of the rich lands to the south in return for their loyalty to the
Khan, at the expense of beilewho wanted to maintain autonomy.
In short, Hong Taiji, at his death in 1643, left a seemingly permanent leg-
acy of a strong authoritarian state. He had named the Manchus as a single
people, elevated himself to supreme leader above the nobility, created new
governmental institutions, and announced the Manchu “historical mis-
sion,” the creation of a new dynasty, named Qing, to replace the Ming. Re-
jecting the earlier designation of the state as the Latter Jin, Hong Taiji delib-
erately signaled a break in continuity both from Ming to Qing and from his
Jurchen ancestors, stating openly: “The Ming is not a descendant of Song
and I am not descendant of Jin; times are different.”
The new state was, however, still fragile. Quarrels over the succession
under the regency that followed Hong Taiji’s death could easily have led to
its collapse. From the death of Hong Taiji in 1643 to the Kangxi emperor’s
rise of the manchus 113
Hong Taiji (r. 1627–1643), the second ruler of the Manchu state.

assertion of his personal rule in 1669, six factions ruled successively as re-
gents over the fledgling state.
46Except for the Shunzhi emperor’s personal
rule from 1653 to 1661, these groups were dominated by Manchu noble
banner commanders. These nobles had varying policies toward the adop-
tion of Chinese ways or the incorporation of Chinese into high-level posi-
tions. Dorgon (r. 1643–1651) was the first to introduce Chinese officials
into high administrative posts, but the regency led by Oboi (r. 1661–1669)
promoted a return to old Manchu ways, downgrading Chinese institutions
and officials. The Manchu reaction, however, did not imply the reassertion of beile
control. In fact, the regents also promoted the ascendancy of the central
ruler over his Manchu nobles, and they helped to institutionalize emergent
bureaucratic organizations whether derived from Ming precedents or new
creations of the rising state. Bureaucratic and autocratic supremacy over
collective noble rule was a continuous trend during this period, but it was
not a smooth course toward “sinicization.” The drive to autocracy was
powered by the military and administrative demands of the expanding
state, and not by any growing acceptance of the superiority of Chinese civi-
lization. Attitudes toward the role of Chinese institutions in the state could
change, but the progressive ascendancy of the Khan and his advisers over
the beile could not easily be reversed.
Dorgon, as regent for the five-year-old Shunzhi emperor, completed the
conquest of Beijing in 1644.
47He began the “dyarchy” policy of having for-
mer Ming and Manchu officials serve together in high administrative posts.
(“Dyarchy” is in fact a misnomer, since Chinese bannermen were a third
significant group of officials who fit neither category.) He also abolished the
assignment of Manchu princes to the Six Boards. Upon Dorgon’s death in
1650, factional rivalry between the Manchu regents, led by Jirgalang, en-
sued for two years, until the Shunzhi emperor took personal control from
1653 to 1661. Shunzhi relied heavily on non-Manchu groups to offset
Jirgalang, encouraging contacts with the Jesuits, Buddhists, and eunuchs.
He supported the right of Chinese to memorialize with scathing attacks on
the corruption of the Manchu noble elite. The state seemed to be heading in a strongly Chinese direction under the
young emperor, but on his death in 1661, the Empress Dowager in alliance
with the Manchu nobles supported a reactive move. They put the new em-
peror, Kangxi, at age eight sui(seven Western years), on the throne under
the supervision of four powerful Manchu regents: Oboi, Soni, Suksaha,
and Ebilun.
48 Stressing the need for a new reign of virtue, they forged the
will of the Shunzhi emperor, putting in his mouth severe criticism of his at-
tachment to corrupt non-Manchu practices. The regents purged the eu-
114 formation of states

nuchs and Buddhists, severely interrogated the Jesuit astronomer Adam
Schall, and executed many other Christian followers. They claimed to be
returning to the idealized, militarist, pure Manchu state of Nurhaci and
Hong Taiji, although in fact both these leaders had skillfully blended multi-
ple cultural elements. The Council of Ministers was cut back in size but
expanded in its functions, while the influence of Ming institutions—the
Neisanyuan, Six Boards, and censorate—declined.
The regents became notorious for their harsh policies toward local Chi-
nese officials. Their support of a crackdown on tax deficits in Jiangnan
touched off a revolt by students and officials, who were subjected to mass
execution in 1661. They introduced the systematic evaluation (kaocheng)
of local officials in order to root out corruption and ensure prompt tax col-
lection. The regents did not, however, restore the principles of collective
rule or local beileautonomy, even in Manchuria. Southern Manchuria be-
came a new province, Fengtian, under the rule of a military governor, and a
military governor took over control of northern Manchuria when it was
threatened by Russian incursions. Only in one respect did the regents allow
excessive autonomy: they permitted Wu Sangui and the other military com-
manders of feudatories in southwest and South China to expand their mili-
tary power and administrative independence. In foreign policy, they were
cautious and not expansive when facing the Mongols, the Russians, or
Zheng Chenggong’s regime on Taiwan. Overall, the regency period did not undermine the stability of the state,
despite its apparent reassertion of Manchu traditional values. Further cen-
tralization and military authoritarian rule increased, but we cannot attri-
bute this centralization very much to Chinese influence. In seventeenth-
century Russia, during the Time of Troubles (1602–1613), the Muscovite
state had collapsed, and local barons strongly attacked the claims of the
centralizing Tsars. Succession struggles, foreign invasion, peasant revolt,
economic decline, and assertion of baronial autonomy all nearly destroyed
the state, until the advent of the new Romanov dynasty in 1613. The Man-
chus faced all of these challenges as well. The two founders of the state had
created it as pragmatic military administrators using the most promising of
the diverse cultural elements of their frontier environment. The Manchu re-
action failed because it aimed to “purify” the state of contamination from
Chinese practices, without understanding that many of these elements were
integral parts of the state, and without producing a successful alternative
model. The regents, notably, were never able to produce a “Sacred Edict” an-
nouncing the fundamental principles of their emperor’s rule, as the Kangxi
emperor did in 1670. The Sacred Edict spelled out the “bare bones of Con-
rise of the manchus 115

fucian orthodoxy as it pertained to the average citizen.” 50 Later versions
fleshed out the emperor’s maxims and delivered elaborations of them orally
to villagers across the empire. Unlike the Kangxi emperor, the regents did
not have enough confidence in a coherent program to address the Han pop-
ulation so directly. Nevertheless, the regents did promote the fundamental
objectives of the founders, as they too were driven by their environment to-
ward greater centralization and authoritarian control. Only when the Kangxi emperor finally asserted his personal power in
1670 were his clan members permanently excluded from all administrative
posts. Thus the seesaw struggle between clan and bureaucracy followed no
simple line toward growing sinicization, but moved back and forth depend-
ing on contingencies of personality and the progress of the military and
economic resources of the state. By comparison with the Russian and Mongolian states, the Manchus
faced a simpler geopolitical situation. They maintained a clear focus on a
single enemy: the Ming dynasty to the south. On their other borders they
faced mainly sympathetic Mongols to the west, inoffensive tribal peoples to
the north, and the relatively weak Korean king to the southeast. They had
defeated one rebellion by the Sunid Mongols allied with the Khalkhas in
51Russian incursions into northern Manchuria had not yet become a
serious threat in the 1660s. The focus on a single front gave a clear struc-
ture and purpose to the state leader’s goals, and the demands of military ex-
pansion drove consistent projects of administrative and cultural reform.
After they moved south of Beijing, from the 1640s to 1660s, the rulers con-
tinued to drive the Ming holdouts south, as they established local institu-
tions for governing the huge Chinese subject population. Even after the
strong leadership of the two founders, the regents, who practiced a cau-
tious foreign policy, followed the momentum established in the early years.
They put down local rebellions and evacuated the southeast coast in order
to isolate Zheng Chenggong and his Ming supporters on Taiwan. They ele-
vated the status of the Lifanyuan, which became the core institution for
dealing with northwestern affairs, separated off from the Libu, which dealt
with tributaries to the east and southeast.
52This decision institutionalized
the bifurcated foreign policy of the Qing and ensured that the northwest
border received a distinctive kind of attention. But the Manchus deferred
the subjugation of most of the Mongols until they had secured the south.
The Oirat Mongols, by contrast, constantly alternated between different
fronts, confronting Mongols to their east, Russians and Kazakhs to the
west, and Tibet to the south. By 1670 the leaders were embroiled in fratri-
cidal strife, and no one could claim the legitimate title of Khan. The Man-
chus’ steady centralization under both emperors and regents laid the foun-
116 formation of states

dation for the expansive ambitions of the Kangxi emperor after he took
personal power.Centralized power, however, could not by itself obtain enough resources
to sustain the growing state. Manchuria has a deceptive ecology. Viewed
from the forests of Siberia, or from the deserts and steppes of Central Eur-
asia, its rivers appeared to be filled with fish and bordered with rich agricul-
tural lands, and its forests abounded in game, forest products, and miner-
als. So the Russian settlers thought as they pushed east across Siberia,
hearing tales of the riches of the region. But they were soon disappointed.
The local tribes looked poor, the fields provided very little grain for the
armed garrisons, and the fortresses soon found themselves short of food.
As noted earlier, the Russians ran into food supply problems even before
they knew that they were trespassing on the territory of the new Manchu
state. When the Manchus were a dispersed tribal population, they could sup-
port themselves at a low level of subsistence without too much difficulty by
practicing a mixed economy that combined hunting, fishing, and small
amounts of agricultural production. They also prospered from trade in for-
est products, particularly jinseng root, sold to Koreans and Chinese, and
furs. Ming officials had effectively used this Manchu desire for trade to di-
vide the tribes by granting each of them separate patents that allowed par-
ticipation in tribute missions. They could thus use the rivalry over patents
to play the tribes against one another. Nurhaci, in order to centralize his
power, had to take control of all the patents himself. This meant depriving
the other tribes of their access to tribute resources. Only after Nurhaci began building his military machine designed for
state unification and ultimately for conquest of China did the logistical
bottlenecks become severe. As he defeated rival clan leaders and incorpo-
rated them into banners, he incurred responsibilities for provisioning these
troops. The best available sources for new provisions were the neighboring
peoples of Korea and Liaodong, so his first moves were to plunder them in
time-honored Central Eurasian fashion. Nurhaci also promoted economic
development from the earliest years, as the Manchus used Chinese artisans
to smelt iron, mine gold and silver, and develop sericulture and cotton pro-
53 But the Ming responded by cutting back tribute income after
1609 and closing down the profitable border markets. By 1615, before he
declared his new dynasty, Nurhaci was already admitting that his state
lacked enough food to feed its own people, and it could not fight neighbor-
ing Mongolian tribes. He urged the clearance of more land to fill the grana-
54From this time until the conquest of Beijing, the urgent need for grain
supplies became a major factor in the expansion of the state. In 1619
rise of the manchus 117

Nurhaci conquered the Yehe clan in order to get more food. At the same
time that he was encouraging the Mongols to join him in the conquest, he
told them that they had to bring their own provisions when they marched
against the Ming, and they could not steal food from the local population.
After the declaration of the Latter Jin dynasty in 1616, Nurhaci faced the
dual problem of feeding his own population and winning over the Chinese
of Liaodong to the south. He promised to “nourish the begging poor” and
to overcome the disparities of wealth created by the corruption of Ming
56 Newcomers to Manchu rule were guaranteed grain rations in
1623, and the allocations were made equal for Chinese and Manchus. Con-
quered Chinese households could be incorporated in a range of statuses,
from free to slave. In 1624 Chinese households who had 5 to 7 Manchu sin
of grain (equal to 9 to 13 shi,or about 800 to 1,000 kg) were given land
and houses, while those with less were made into slaves.
57But the status of
boo-i niyalma, oraha, did not correspond exactly to the Chinese category
of “bondservant-slave” (nupu)or to that of “free man” (zhuangding).It
was a relationship of personal dependency on a master which in theory
guaranteed close personal relationships and equal treatment. As the Khan
stated, “The Master should love the slave and eat the same food as him.”
By insisting on the paternalism of the relationship, notably through guaran-
teeing food supplies, the Manchu ruler aimed to convince the Chinese of
Liaodong that they could gain security and overcome unjust disparities of
wealth under the new state. But the conquest of Liaodong in 1621 led in practice to much sharper
conflicts between Manchus and Chinese and much more severe subsistence
crises. Rebellions by the Chinese against unequal treatment also focused on
food supplies: Manchus had to fear poisoning of wells by their new neigh-
bors and beware of eating pigs, salt, chickens, eggplant, and other prod-
59Nurhaci promoted co-occupancy of households, whereby Manchus
and Chinese lived and worked together, in order to overcome antagonism,
but food shortages as well as cultural differences made this experiment a
failure. The Manchus had no grain reserves of their own in the winter of
1621–22, so they had to extract food from the Chinese, who tried to con-
ceal it, and institute general grain rationing as an emergency measure.
They ordered equal distribution of land the next year, at 5 cimari(30mou)
per adult male, but it is not clear if this was ever carried out. The expanding
state tried to resolve the logistical crisis with new corvée and grain taxes
levied on the larger Chinese population, but free land was scarce in Liao-
dong, and grew even scarcer as Chinese fled the fighting in Liaoxi to the
west. The refugees found that they had “not enough grain or salt to eat.”
Manchus under economic pressure exploited the Chinese in their house-
118 formation of states

holds, treating them as slaves and stealing their possessions. The Chinese
of Liaodong openly rebelled in 1623, setting fire to buildings, poisoning
Manchu associates, and stealing grain from the granaries. Although the re-
bellion was easily suppressed, the Khan heard repeated reports that his peo-
ple ignored edicts commanding fair treatment of the Chinese. An even more
serious uprising in 1625 exposed further abuses by local Manchu officials
and commanders.
Hong Taiji listened to newly enrolled Chinese advisers against the wishes
of some of the Manchu nobility. The Chinese urged further centralization
of power and greater leniency toward the subject Chinese population, con-
vincing him that social unrest was undermining the Khan’s authority and
favored the interests of his unruly clan nobility, the beise.Those who re-
sisted undue favoritism toward Chinese advisers in the state, like Amin,
wanted to return to the good old days of plunder and raiding. Amin in-
vaded Korea in 1627 and pillaged the territory, defying the Khan’s orders,
and massacred and looted the Chinese town of Yongping in 1629. Hong
Taiji, declaring him an enemy of the state, had him thrown in prison in
1633, where he died in 1640. The social and economic crises of the state led Hong Taiji to abandon
Nurhaci’s fruitless attempts at Manchu–Chinese integration. He instituted
a strict segregation policy that attempted to create separate but equal do-
mains. The co-occupancy of households was abolished, and a new organi-
zation, the tokso,enrolled Chinese into units of thirteen households under
a single headman, who served under a Manchu banner commander. Chi-
nese were to hand in all weapons, while all Manchus had to be armed.
Stricter controls over mobility, including rewards for returning fugitive
bondservants, curiously parallel the contemporaneous effort of the Russian
state to tighten up the bonds of serfdom. But the Manchu state was “on the brink of economic disaster” in 1627.
The first major defeat of the Manchu army by the Ming in 1626 exposed
the state’s great vulnerability. The marginal economy of Manchuria could
only barely support its growing population, but to provide in addition for a
large army on a military campaign, the army had to gather booty from its
victories. This was exactly the dynamic of the typical steppe nomadic feder-
ation, which expanded explosively when it succeeded in gathering enough
loot from victory to reward its supporters but collapsed implosively at the
first signs of defeat. The food crisis of 1627, the most severe yet, drove up
grain prices eightfold over 1623, to 8 taels per sin(1 Manchu sin=1.8 shi),
and brought tales of cannibalism and robbery. There was no grain for
newly surrendered subjects, and the granaries were empty. Nor was there
land for newly arrived Chinese.
65 Other food crises struck in 1635 and
rise of the manchus 119

1637. The lack of provisions for the army weakened the Manchus’ military
power considerably. Their horses could not pursue the enemy because they
were too tired and weak.
66Efforts to raise agricultural production in Liaoxi
failed; exhortations to rich landowners to provide for the poor fell, as
usual, on deaf ears, and the Manchus could not afford to alienate them by
forcing them to sell at low prices.
67Korea again offered a tempting target.
The Manchus forced Korea to provide grain under threat of military in-
vasion, and cut off its profitable tribute trade with China so that they
could monopolize it for themselves. Relations with the Mongols, crucial to
Manchu military strength, also threatened to turn sour if the Manchus
could not provide enough grain in exchange for Mongol horses. They had
to prevent the Mongols from selling horses to China and repressed the
secret Mongol trade in grain with Chinese. Thus, despite conventional impressions, the Manchu state and army
were in a weakened condition in the 1630s. Although the Manchu archival
documents do not provide detailed information on this period, the repeated
pattern from 1636 to 1644 of Manchu attacks on Chinese cities followed
by harvesting of the fields and withdrawal indicates that the Manchus were
not strong enough to pursue a plan of direct conquest.
68They were in dan-
ger of reverting to the old raid and extortion policy instead of creating a vi-
able new state, let alone conquering a Chinese dynasty. Hong Taiji was firmly on the throne by 1635. He had overcome his
Manchu rivals and declared the new Qing dynasty in 1636. Having cap-
tured a Yuan dynasty seal from the Chahar Mongols, he could style himself
the rightful successor to Chinggis Khan as well. But his weakening army
was opposed by the Chinese commander of the pass, Wu Sangui, whom he
could not defeat. He was saved only by the fortuitous events of 1644:
the conquest of Beijing by the rebel Li Zicheng, the death of the Ming
emperor and the flight of his court, and Wu Sangui’s decision to leave his
defensive position to attack Li Zicheng, allowing the Manchus to pour
through the pass.
Severe logistical constraints on the new Manchu state, then, nearly
caused it to collapse. In fact, the Manchus were not alone. All major agrar-
ian states in the sixteenth and seventeenth centuries stretched their state-
building ambitions far beyond their means. The multiple crises that struck
the globe in the seventeenth century in England, France, Russia, Germany,
China, and the Ottoman empire, among others, have been traced to the ex-
cessive burdens placed on agrarian economies with limited productive po-
tential by burgeoning state and military apparatuses. Population growth
also increased demands for subsistence on limited agrarian production sys-
tems in western Europe and China. Fiscal exactions ran up against the bar-
120 formation of states

riers of agrarian production and local elite resistance. Ming surtaxes to
support military defense touched off the rebellion in the northwest that
brought down the dynasty, and French noble resistance to royal taxation
likewise tore the country apart in the Fronde civil wars of the seventeenth
In this context, the Central Eurasian states occupied a special niche. On
the one hand, unlike the western European states, they were not burdened
by growing populations within a limited territorial area. There was still
plenty of room for peasants to move across the steppes and open lands of
Russia. The main problem for Russian state builders was to tie the peasants
to the land in order to extract labor dues and taxes from them. On the other
hand, the rate of agrarian production was so low that only small amounts
over subsistence could be extracted. Extensive agriculture was accompa-
nied by extensive state extraction. Russians also discovered a second lucra-
tive source of state revenue in the Siberian fur trade. Exploitation of the
northern forest peoples, which had begun as early as the twelfth century by
Novgorod, brought in valuable fur-bearing animals at relatively low cost.
The Manchus likewise could rely on the forest resources of their region as
the main base for the early stages of state building. Nurhaci was, in effect, a
forest tribal leader who conquered his fellow tribespeoples and used their
production as the basis of his military regime. But both Manchus and Russians faced severe strains when they moved
into more densely settled regions. Conquering these regions required nego-
tiations with local elites much more securely based than the forest peoples
of Siberia or Manchuria. In Poland–Lithuania, the Ukraine, or Liaodong,
these frontiers required major military siege campaigns and adroit negotia-
tions before they could be incorporated into the new state. The conquer-
ors had to support the subject peoples and ward off subsistence crises to
avert social disorder. Both states nearly collapsed in the mid-seventeenth
century. The Manchus avoided failure only when the Ming court aban-
doned Beijing, while Muscovy survived only after the citizens of Moscow
rallied an army to drive off Polish invaders, put down peasant revolts, and
elect a new Tsar.
72A breakthrough to new sources of economic wealth and
technology by the end of the century provided new expansive powers for
both regimes: Peter the Great turning decisively toward the West but build-
ing on the Muscovite legacy; the Qing rulers taking over China’s abundant
agricultural wealth in the south but also not abandoning their strategic
goals in the northwest. In the early to mid-seventeenth century, by con-
trast, the survival of both states seemed precarious. In the stability of lead-
ership and logistics, the Russian, Manchu, and Mongol states were close
rise of the manchus 121

Mongolian Influence on the Manchu State
The Manchus’ relations with the neighboring Mongol tribes were crucial to
the strengthening of the state from its earliest years. Kinship and ideologi-
cal intimacy tied them closely together. Nurhaci had exchanged wives and
concubines with the Khalkha Mongols since 1594, and he received the title
Sure kündulen Hanfrom them in 1607, replacing his clan title of sure beile
(wise prince), thus raising his claims from that of a local clan leader to the
level of a Central Eurasian Khan. In 1616 he gave himself the title Geren
gurun be ujire genggiyen han (the radiant emperor, whom Heaven has des-
ignated to nourish the many countries), consolidating his claim to the Mon-
golian traditions of leadership.
73The ideal of a multinational ruler, or uni-
versal Khan (Mo. gür qan, dalai-yin qaghan), derived from the Buddhist
“wheel-turning king” (chakravartin),had been transmitted through Tibet
to Mongolia in Yuan times. Tibetan lamas gave sacral character to Ching-
gis and Khubilai Khan in their writings, regarding them as incarnations of
powerful boddhisattvas and tutelary deities.
74 The last claimant to this
ideal rulership in the seventeenth century was the descendant of Chinggis
Khan, Ligdan (Linden) Khan, of the Chahars. Any Inner Asian claimant to
imperial power had to link himself to this ideological tradition. Other concepts of imperial rule, according to David Farquhar, derived
not from Chinese but from Mongolian sources. The Manchus had not yet
begun, in this early period, to translate basic Chinese classical texts. As
mentioned previously, the earliest literate advisers to Nurhaci were men
like Dahai and Erdeni Baksi, who had grown up with Mongolian connec-
tions. Dahai and Erdeni were probably Chinese in origin, “transfrontiers-
men” in Liaodong whose ancestors had changed their cultural identity,
adopting Manchu names and cultural traditions.
75Dahai (1592–1632) was
in charge of all official communication with the Koreans, the Mongols, and
the Ming court. He conducted peace negotiations with Ming commanders
during the Manchu military campaigns, and he read out the emperor’s
proclamations in Chinese in 1629–30. He also translated military books
and the Ming penal code into Manchu. Erdeni Baksi (d. 1623) knew Mon-
golian and Chinese texts from his early youth and could speak with both
Chinese and Mongols in their native tongue. Erdeni told Nurhaci that
omens in the sky (the northern lights), so conspicuous in the years 1612,
1614, and 1615, indicated that the Mandate of Heaven was due to be
changed soon. This interpretation, combining Central Eurasian reverence
for Heaven with Chinese mandate theory, induced Nurhaci to proclaim the
Latter Jin dynasty in 1616.
These tricultural men were the key brokers who synthesized the essential
122 formation of states

elements of the Manchu state from diverse roots. They were later given bi-
ographies in theBaqi Tongzhi(Eight banner gazetteer) as the first “Confu-
cians” (ruren), even though their backgrounds were more Central Eurasian
than Chinese. The later generations of literati given biographies in this
text have very little Mongolian background; their chief contributions came
from the translation of Chinese texts into Manchu. They performed a dif-
ficult and dangerous cultural straddling act, one which made them vulnera-
ble to attacks from rivals. Dahai was dismissed from office in 1623, and
Erdeni was executed in 1623 for unspecified crimes. They may have been
the victims of false rumors spread by beilewho resented their intimacy with
the rising Khan.
The concept of “Heavenly destiny” (Ch. tianming,Ma.abka-i fulingga:
the first reign title of the Manchus, declared in 1616), if it ultimately had
Chinese origins, had been transmitted to the Manchus from Mongolia.
Khubilai Khan had claimed this mandate for his empire in 1272.
78 The
Manchu beile,who combined the roles of hereditary princes and officials,
corresponded exactly to their Mongolian counterparts, the noyan.The idea
of a dual nature state (Mo. qoyar yosun) divided between secular and reli-
gious spheres derived from sixteenth- and seventeenth-century Mongolian
theories of the “nature of the state” (Mo. törö sajin,Ma.doro sajin). Thus
the “higher” concepts of a state that transcended clan leadership, even
though they looked similar to Chinese classical concepts, came to Manchu-
ria “through a Mongolian strainer.”
Even the Manchus’ most distinctive innovation, the banner organization,
had just as strong Mongolian roots as Chinese. Franz Michael argued that
the banners were modeled directly on the Ming weisuo,or hereditary mili-
tary garrisons, but the original banner model had nothing in common with
the decimal organization of the Ming weisuo.
80It was based on a division
into two geren(hordes), each gerendivided into four tatan(camps) of sev-
enty-five men each, but each banner also contained cavalry (uksin)and foot
soldiers (yafahan). Only later did the banner assume a more decimal form
of organization. In any case, the Ming weisuothemselves originated from
Yuan dynasty decimal military formations.
81The other distinctive features
of the banners also had common Mongolian origins: the combination of
military and civil administration, the use of sacred flags, and the appoint-
ment of hereditary officers, for example. By its final form in 1623, the banners were based on the niru(arrow,
company) which was originally a hunting and fighting unit of the Manchus,
very similar to the qosighunof the Mongols in the thirteenth and four-
teenth centuries. The gûsa(banner), like the qosighun,was based on “the
able-bodied men of a tribe or other autonomous political unit when consid-
rise of the manchus 123

ered as a military force.” 82Its head was a hereditary prince, and its official
titles were modeled on the Mongolian ones. Thus the central institution
that provided the backbone of the Manchu conquest could be traced to
Mongolian roots.
Just as in the Russian case, it is less important to prolong the argument
over the relative importance of Chinese or Mongolian sources of the ban-
ners than to recognize the creative synthesis produced by the Manchus out
of their frontier location. The fluidity of institutional and personal iden-
tities characteristic of the region, the mixing of peoples and ecologies and
social structures, provided the cauldron in which innovative social forma-
tions could be brewed, which led to a military and social structure of irre-
sistible force. For Roberto Unger, “the Mongols’ experience in the steppe
and the borderlands and the problems they had to face as a conquest elite
prepared them...fortherole of teachers of statecraft. They taught the les-
son of continued mobilization, of hostility to vested rights and local prerog-
atives, and of insistence on insulating at least parts of governmental power
from the influence of local landowners and notables.” The Mongolian pat-
terns, adopted by the Qing, prevented China from falling back again into
“reversion cycles,” in which local landowners took over control of agrarian
resources, starving the central state and shutting down commercial ex-
change networks. The Qing banner system “kept the conquest elite ready
for service and prevented it from sinking into a class of landowners anxious
about their local interests.”
The banners and other Manchu institutions are examples of productive
hybridity, combining “pure” Mongolian elements (the script, for example)
with Chinese elements transmitted through Mongolian filters, and, later
on, purely Chinese elements that entered the state as it expanded to the
south. Intermarriage with Mongolian noble families further cemented the alli-
ance between the two peoples. In 1612 Nurhaci himself married the daugh-
ter of the chief of the Khorcin Mongols. From 1612 to 1615 Nurhaci and
his sons together married six Mongolian women. After 1617 other Man-
chus married into the Mongolian nobility, and surrendering Mongol tribes
enrolled in Manchu banners provided significant numbers of women to
marry the Manchu elite.
Hong Taiji expanded the marriage alliance policy, marrying twelve of his
daughters to Mongolian chieftains. He used marriage ties to draw in more
of the twenty-one southern Mongolian tribes that joined the Manchu alli-
ance. The Manchu–Mongolian alliance became even more systematized
and regulated after the Qing conquered the rest of China. Special annual
subsidies of silver and silk, elaborate funerals, and the granting of high
124 formation of states

rank to sons all encouraged intermarriage. At the same time, restrictions on
visits to the capital kept the Mongolian nobles under control. Manchu
women married to Mongolians could come to the capital, their home-
town, only once every ten years, and they could stay for only sixty days.
But young Mongolian boys were selected as consorts for the princesses,
brought to the capital for education in Manchu and Chinese culture, and
sent back to their tribes suitably acculturated. The Qianlong and Jiaqing
emperors expanded the system to encompass thousands of marriages be-
tween all levels of the Manchu and Mongolian noble classes. At the same
time, commoners of both sides were kept rigidly segregated, and intermar-
riage was banned.This highly regulated, large-scale kinship creation program differed sub-
stantially from previous dynasties’ uses of occasional marriage alliances to
ensure peace with northwestern frontier peoples, the so-called heqinpolicy,
which went back to the Han dynasty.
86Han and Tang use of heqinalliances
was intended only to ward off frontier threats, but the Manchus aimed at
intimate linkages with their fellow Central Eurasians—linkages that did
not, however, obliterate boundaries between the two peoples. Marriage al-
liances were used alongside other techniques—gifts, stipends, tax exemp-
tion, education, access to official posts—to provide strong incentives for
Mongols to cooperate with the rising Manchu power. In 1689 the Kangxi
emperor boasted, rather prematurely, that because of his intimate ties with
the Mongols beyond the wall, he had brought peace to the northwestern
frontier, eliminating the constant menace of nomadic raids that plagued the
Ming dynasty.
Despite the growing intimacy of Manchu–Mongol ties, one major Mon-
gol leader, Ligdan (Linden) Khan of the Chahars, resolutely opposed the
growing Manchu power. As the last descendant of Chinggis Khan, he held
an official Yuan seal and viewed himself as the legitimate representative of
the Mongolian imperial tradition. But after his losses in battle to the Man-
chus in 1628 and 1632, the Manchus took over the Yuan seal and enrolled
the Eastern Mongols as a whole in the banner system. Ligdan Khan’s son
married a Manchu princess after Ligdan died of smallpox in Qinghai. The
Chahar and Khalkha Mongols comprised 384 niruwith 19,580 families,
the Khorcin 448 niruwith 22,308 families.
88Only after this definitive vic-
tory could Hong Taiji proclaim the genuine three-nationality empire, which
he named Da Qing in 1636. No longer did the Manchus need to hark back
to their Jurchen predecessors of the twelfth century, the regional state of the
Jin (1115–1234 ce); now they could legitimately claim to be building a uni-
versal empire. Ligdan Khan played another important function in the Manchu state. He
rise of the manchus 125

had translated the large Tibetan Buddhist canon, the Kanjur, into Mongo-
lian during his rule. The Manchus were introduced to Tibetan Buddhism
through this channel, preparing the way for the visit of the Dalai Lama to
Beijing in 1652.
The successful incorporation of the Eastern and Southern Mongols dur-
ing the early conquest period provided a substantial fund of experience that
the Manchus could use when they confronted the more isolated, hostile,
and autonomous Mongolian tribes farther west. Over the century and a
half following the conquest, in their extension of the empire, they returned
to the techniques of seduction, threat, personal ties, and authoritarian bu-
reaucratic regulation that they had used successfully in the relations be-
tween Mongols and Manchus during the first fifty years of building the
The greatest gift of the Mongols to the Manchus, of course, was the
Mongolian script.
90In 1599 Nurhaci ordered Erdeni Baksi and G’ag’ai to
create a script for the Manchu “national language” (guoyu).They objected
that the Manchus had long used the Mongolian script and language, and
they could not create a new one. Nurhaci then said, “When the Chinese
read out their writing, people understand it, whether or not they can read
Chinese; likewise for Mongols; but our words must first be translated into
Mongolian; then [the Manchus] don’t understand it.”
91 He then ordered
them to create a new alphabetic script, using the Mongolian script as a
Taizu [Nurhaci] asked, “Why is it difficult to write down our lan-
guage, but easy to learn the languages of other countries?” G’ag’ai and
Erdeni replied: “It would be best to create a script for our country’s
language, but we do not know how to transcribe the sounds.” Taizu
said: “If you put a letter for ‘ma’ after a letter for ‘a,’ is this not ‘ama’
[father]? If you put a letter ‘me’ after a letter for ‘e,’ is this not ‘eme’
[mother]? My mind is made up; you just try it out.” Thereupon they
took the Mongolian script and wrote the Manchu language. The cre-
ation of the Manchu script began with Taizu.
So Erdeni and G’ag’ai, following Nurhaci’s orders, created the new writing
system, and soon began to translate Chinese texts into Manchu, as well as
using Manchu in imperial proclamations. Dahai, in 1632, added the dia-
critical marks to distinguish different Manchu vowels, along with extra
symbols for particular Chinese consonants; this “pointed” script became
the standard Manchu writing system for the rest of the dynasty. Nurhaci was, of course, wrong to assume that classical literary Chinese
126 formation of states

could be understood when read out loud. His advisers, well acquainted
with Mongolian imperial language, resisted the introduction of Manchu
writing probably in order to maintain ties to the Mongolian institutional
tradition. To judge from his discussion, Nurhaci had in mind a syllabic
script (like Japanesehiraganaandkatakana), not the actual Mongolian or
Manchu scripts, which were alphabetic. Nurhaci’s motives were political,
not linguistic. What he stressed was oral communication of written com-
mands by the ruler to the entire Manchu population, literate and nonlit-
erate. He needed a scriptural apparatus to bolster his new state because he,
like all previous Central Eurasian rulers, needed to communicate his per-
sonal will beyond the boundaries of person-to-person contact. His edicts
could now be read out in their own language to all his Manchu subjects,
and texts could be translated into their native language for their own edu-
cation. In effect, by creating a distinctive script, Nurhaci broadened the cul-
tural horizons of his people, allowing them to adapt non-Manchu ideas but
maintain their distinct identity. The new technology of writing made possi-
ble the expansion of the state to cover all the Manchu people. But it also al-
lowed the introduction of large quantities of Chinese classical literature
through translation into the Manchu literate world, which had formerly
been much more closely tied to Mongolia and the Buddhist world of Cen-
tral Eurasia. In short, the Mongols contributed a great deal to the early Manchu state.
They provided military allies, horses, and a tradition of legitimation reach-
ing back to Chinggis Khan. Along with the Yuan seal came the concept of a
universal empire encompassing many peoples, an ideal of rulership that
vastly transcended either the state of the Manchus’ ancestors, the Jurchen
Jin, or that of the Ming. Personal connections through kinship and literary
connections through the script bound the two peoples together. Nurhaci of-
ten invoked the common heritage of the Mongols and Manchus when pro-
moting alliances, although at other times he stressed their differences. Not
all the Mongols accepted the supremacy of the young Manchu state; only
decisive victory in war persuaded them to submit. From its origins, the rul-
ers of the Manchu state learned how to devise interrelated strategies of war,
diplomacy, and economic inducement to ensure the loyalty of their critical
allies in the northwest.
Early Modern State Building Compared
The Chinese and Russian empires displayed many common patterns as they
expanded over Eurasia. They both grew explosively in size, in little over
rise of the manchus 127

fifty years spreading from small tribal kingships to domination of vast for-
ests, fields, steppes, and deserts. At about the same time, the three Middle
Eastern “gunpowder empires,” the Mughals, Safavids, and Ottomans, also
expanded. Even though gunpowder was not the primary factor, new mili-
tary organizations partially based on gunpowder weaponry vitally aided
their conquests. The Manchus, Russians, and Mongols also relied on new
military organizations to bind their states together. They shared a basis of
common interactions with peoples of the steppe, from whom they learned
the importance of rapid mobility and careful attention to military logistics.
They sought the easiest sources of wealth from the settled peoples nearby,
in oases, paddy fields, or fur trapping villages, but they went beyond raid-
ing and rapid extraction by using these resources to build new states.In this chapter, the Central Eurasian steppe has been the element linking
the three states together in the early period of their formation. After their
success, however, the conquerors attempted to disguise the complex pro-
cesses that generated their new states. Russian clerical chroniclers created a
simple dualism of struggle between Christianity and paganism to legitimate
the new Tsars, while Chinese historians inserted the Manchus into the or-
thodox doctrine of inheritance of the Mandate of Heaven and sinicization.
In fact, the conquerors succeeded not because they followed rigid reli-
gious ideologies and cosmological precepts, but because they pragmatically
mixed together multiple traditions. Russian Tsars relied on astute steppe di-
plomacy, horsemanship, and many Mongolian heritages as well as Ortho-
dox Christianity and subordination of Slavic peasantry. Manchus mingled
Mongolian rulership, horsemanship, and language with Chinese classical
texts and Manchu kinship connections. The Zunghar Mongols adapted Ti-
betan Buddhist practices to Mongolian clan connections and oasis cultures. The Ottoman empire provides another example of such creative min-
gling. Historians for a long time interpreted the rise of the Ottomans as the
product of a tribe of righteous warriors for Islam (ghazi)on the fringes of
the Byzantine and Turkish empires. But Cemal Kafadar has recently shown
that the early Ottomans grew out of a fluid frontier environment where
identities and loyalties constantly shifted. Rudi Lindner has also pointed to
the important role of nomads in the early Ottoman state. Later, in the sev-
enteenth century, the Ottomans were challenged by armed groups, called
celali, or “bandits.” Many historians viewed the celalias outright oppo-
nents of the Ottoman state, but Karen Barkey has shown that they were in
fact products of the Ottoman regime, mainly demobilized soldiers who
sought higher positions within the state. The Ottomans once again success-
fully negotiated with multiple armed groups in ceaselessly competitive
frontier conditions to keep their state alive.
93 The Ottoman state endured
128 formation of states

for so long because its rulers knew how to negotiate with diverse groups
within the polity.If we examine each of these empires in isolation, we tend to see only the
apparently distinctive features that mark them as most innovative. Con-
sidering the rise of three or four empires together reveals instead substantial
resemblances in the role of nomads, the relations between central state
structures and local nobilities, and negotiations with rival state builders.
Some of these techniques were borrowed from neighbors; others evolved as
common responses to a similar frontier environment. But the Central Eur-
asian connections of the state founders marked all of them and left a per-
manent legacy.
rise of the manchus 129

Contendingfor Power

Manchus, Mongols, and Russiansin Conflict, 1670–1690
Inthe late seventeenth century, the vigorous young ruler of China, known
as the Kangxi emperor, took decisive action to expand and strengthen his
new empire. In an astonishingly short period of time, a mere thirty years
from taking personal power, he decisively imposed his will on the regents—
his own uncles, the generals ruling the southwest, Taiwanese aborigines,
and, most impressively, the free nomadic military leaders of Mongolia. By
1700 it seemed that no one could resist the imperial command. None of
these achievements was determined in advance, and victory was never cer-
tain. The emperor had to overcome substantial opposition from his clos-
est advisers, and his troops battled incessantly against immense human
and natural obstacles. Although the structure of the emerging Qing state,
including its Central Asian elements, supported Kangxi’s vigorous cam-
paigns, we cannot ignore the sheer force of personal will that so strongly
marked these critical years. Kangxi’s dynamic intervention transformed
the Qing from a promising but limited enterprise into an unprecedented
project of expansion. The Mongol campaigns signaled most definitively
this transformation of the Qing into a Central Eurasian empire with world
Kangxi the Ruler
Aisin Gioro Xuanye (1654–1722) was the third son of Fulin, the Shunzhi
emperor. His grandmother and the four Manchu regents selected him to be-

[To view this image, refer to
the print version of this title.]

come emperor, with the reign name Kangxi, when his father was dying of
1Fear of smallpox, a virulent disease among both Manchus and
Mongols, played an important role in the choice of Xuanye, who had sur-
vived exposure to it as a boy. By the 1630s, Manchu generals, believing that
Mongolian territory was an especially dangerous site of the disease, tried to
ensure that only officers already exposed to smallpox could lead expedi-
tions through this region.
The new emperor combined the blood of the three ruling peoples in his
veins. His ancestry was in fact less than half Manchu, since his father and
paternal grandfather were sons of Mongol princesses. He was one-fourth
Chinese on his mother’s side, although she belonged to the Manchu im-
perial clan.
3Eight years old at his succession, the boy emperor grew up
under the supervision of his uncles. When he reached the age of fourteen sui
in 1667, a Chinese censor urged him to take personal power.
he formally assumed personal rule in that year, he did not move actively
against the constraints of the regency until June 1669, when he arrested
Oboi, the chief regent, and had him prosecuted for a detailed list of crimes.
Oboi soon died in prison. The new emperor then proceeded to purge all the
Six Board presidents and major banner officials to assert his undisputed
power. Kangxi’s first task was to eliminate the factionalism at the top levels of
government, which, he believed, had brought about the downfall of the
Ming dynasty and also plagued the regency. He needed to alleviate the se-
vere tensions between Manchus and Chinese caused by the harsh Manchu
nativism of Oboi, and to convince the Chinese literate class that only his
rule would bring peace. He also needed to complete the unfinished tasks of
gaining military control of Taiwan and southwest China. By suppressing
the Three Feudatories revolt (1673–1678), the new emperor demonstrated
dramatically that he would reject the cautious, vacillating policies of his re-
gents. Kangxi’s personal will injected a new dynamism into Qing expan-
sionism that brought back the spirit of the early days of the growth of the
state. The Chinese generals Kong Yude, Shang Kexi, Geng Zhongming, and,
most important, Wu Sangui had crucially aided the Manchu conquest. The
defection of Chinese officials and generals to the Manchu side was essen-
tial in winning the battle against the holdout forces of the Southern Ming
dynasty. As a reward for their services, these generals had been given nearly
total autonomous power over southern and southwest China. In 1660
manchus, mongols, and russians 135
The Kangxi emperor (r. 1662–1722), in a scholarly pose.

[To view this image, refer to
the print version of this title.]

Shang Kexi ruled Guangdong, Geng Jimao controlled Fujian, and Wu
Sangui controlled Yunnan.
5Wu, the strongest of them, commanded 65,000
troops and held high noble rank. Central government support of his army
cost 9 million taels, two-fifths of state revenue in the Shunzhi period.
regents did not challenge the growing independence of the feudatory gener-
als, nor did they expand their own banner forces to balance them. Kangxi
would expand the banners by 179 more companies from 1667 to 1675, to
a total of 799 companies. Once the Southern Ming forces abandoned South China, the military ser-
vices of the feudatories were no longer needed, but they aimed to preserve
their power base. Wu Sangui offered his retirement to Kangxi in 1667 to
test Kangxi’s will, but Kangxi, not ready to challenge him, kept him on.
When Shang Kexi offered to retire in 1673, the time for the true test had
come. The key issue was whether or not the central government would
allow these territories to become hereditary family possessions of the
founding generals. The Deliberative Council agreed to accept his retire-
ment, but they would not allow his son to inherit his command. When Wu
and Geng offered their retirements, the majority of Kangxi’s councilors ad-
vised against accepting them, because the court was not prepared to face a
military challenge from these strong regional powers. Kangxi, however,
overruling his advisers, accepted Wu’s retirement, thus touching off the re-
volt by Wu Sangui in December 1673. Shang Kexi and Geng Jingzhong,
Geng Jimao’s son, joined Wu’s revolt, thus throwing all of southern China
into rebellion. Kangxi won this difficult campaign for several reasons: the
feudatories could not cooperate effectively with one another; Wu Sangui
was discredited as a defender of Chinese interests because he had collabo-
rated with the Manchus; and Kangxi’s expanded banner troops were still
the most effective fighting force in the empire. But the fortuitous death of
Wu Sangui in 1678 put a quick end to what could have been a very drawn-
out conflict. Kangxi’s successful suppression of the southwest secessionists had conse-
quences for Qing attitudes toward control of other parts of the empire. No
subsequent ruler would ever allow any region the degree of autonomy that
the feudatories had attained. Even though the other frontier regions, after
their incorporation by conquest, maintained distinctive and somewhat au-
tonomous local administrative systems, the Manchus, by controlling ti-
tles, succession to power, lands, and taxation, ensured that breaking away
would be difficult.
manchus, mongols, and russians 137
The Kangxi emperor wearing armor.

Kangxi’s first military campaign set the tone for the later northwest cam-
paigns. Kangxi developed a long-standing suspicion of the Dalai Lama dur-
ing this period. The emperor wanted the Dalai Lama’s support for crushing
the revolt, but he knew that Wu Sangui, whose territory bordered Tibet,
had also been making overtures to the lama. Even though the Dalai Lama
denied supporting Wu, the emperor remained suspicious. He received ex-
cellent support from the Chahar Mongols during the campaigns, which jus-
tified further efforts to mobilize surrendered Mongol forces in military ex-
peditions. Overall, however, the emperor was very dissatisfied with the
performance of his field generals, whom he regarded as excessively cautious
and untruthful.
By 1678 Kangxi already had reason to suspect that the Western Mongols
under Galdan, unlike the eastern tribes, would not docilely accept subordi-
nation to Manchu rule. But in the 1680s his main goal in the northwest was
to ward off the increasing Russian penetration of northern Manchuria.
Kangxi’s forward moves culminated in the attacks on the Albazin fortress
in 1684 and 1686, and the negotiations leading to the Nerchinsk treaty of
1689. Preventing the Russians from supporting the rising power in western
Mongolia was one of the main motivations for the treaty. By 1684 or 1689 Kangxi could regard his consolidation of Manchu
power as complete.
7He had won over the Chinese scholar elite to his rule
and successfully incorporated nearly all the Ming elite with the boxuespe-
cial doctoral exam of 1679. He had secured the boundaries of the state on
all its frontiers and defeated major opposition from Ming loyalists, the
maritime fortress base of Zheng Chenggong on Taiwan, the fortress com-
manders in Siberia, and the powerful regional generals controlling the
southwest. He had ensured the supremacy of the emperor over rival Man-
chu magnates and established stable institutions of bureaucratic rule. Why,
then, ten years later, did this secure ruler launch another series of extremely
risky campaigns of expansion, again defying his cautious advisers, and even
participating in the campaigns personally? There was no strategic impera-
tive for the Qing empire to expand further; it already exceeded the size of
the Ming empire by 1678, and the Zunghar Mongols arguably did not con-
stitute a serious threat to the core of Chinese rule. We will explore later
how the campaigns against Galdan developed out of the contingent interac-
tion of Russians, Mongols, Manchus, Tibetans, and other actors in the
fluctuating environment of the steppe. Relations between Galdan and Kangxi began amicably, following prece-
dents set by Galdan’s ancestors. In 1653 Gush Khan had been allowed
to send tribute missions to Beijing, and received an imperial seal and title
from the Shunzhi emperor. Ochirtu Khan, the Chechen and Tüsiyetü (Ch.
138 contending for power

Tuxietu) Khans, and other Mongols stationed near the Gansu border had
also been accepted as tributaries. The Qing accepted Galdan’s request in
1677 to offer tribute. The court knew something of Galdan’s character:
that he was “rough and crafty and likes fighting,” that he had made himself
a leader by executing his vow of vengeance against those who had killed his
brother Sengge, and that he had quickly achieved dominance over many of
the Western Mongols.
8His rapid rise from small chief to a great prince enti-
tled him to present tribute on an equal level with the other Khans. The Qing
aimed to ensure peace among the quarreling Mongol tribes by appealing to
them as tributaries, equally subject to the emperor’s benevolence or wrath.
In 1655 eight of the Khalkha leaders were granted the title of “Jasak” and
provided with fixed tribute allowances. They had to present annually eight
white horses and one white camel—the “Nine Whites” (jiubai).In return
they received silver, tea, brocades, and other textiles—all profitable items of
9These Mongols from beyond the pass could enter the empire only on
strictly regulated tributary missions. The expansion of Galdan’s power, however, disrupted Qing goals of
maintaining peace among the tributaries and of keeping them confined to
regions beyond the passes. By the end of 1677, Ölöd (Western Mongol)
leaders defeated by Galdan had begun to cross the border illegally, stealing
horses from border guards and local people.
10 The most powerful among
them were Erdeni Qosuuci (Ch. Erdeni Heshuoqi) with over ten thousand
tents and Morgen Alana Dorji with several thousand. Threatened by winter
famine, these invaders crossed the border, killing and plundering for food
and animals. The emperor, realizing that they were driven by desperation,
ordered border troops not to kill them but to attempt to drive them out.
Above all, he preferred a defensive policy: strengthening the alertness of
border guards to prevent the Mongols from crossing the border in the first
place. In 1676 or 1677 Galdan had defeated and killed his father-in-law, the
Khoshot leader Ochirtu Khan. He tried to recruit Ochirtu’s men to join
him, but many fled.
11 Major attacks were reported on the Mongols in the
Ordos region by the remnants of Ochirtu’s defeated bands in early 1678. At
the same time, panicked Mongol loyalists reported rumors that Galdan was
planning to attack the Qing borders himself. His own headquarters were
at Jinshan, two months’ march from Jiayuguan, but he had called for an
assembly of the Mongols in the Ordos, intending, they said, to plan an in-
12 Meanwhile, through 1679, Erdeni Qosuuci continued his plun-
dering, while the emperor tried to get Erdeni’s lord Dalai Batur Taiji and
Morgen Taiji to control him. Kangxi, who needed Galdan’s aid in resolving
the border troubles, since he was preoccupied with the Three Feudatories
manchus, mongols, and russians 139

war, asked Galdan to capture Erdeni Qosuuci’s band. Galdan replied that
Erdeni now “roamed with the wild animals” and could not be found. The
emperor responded that Galdan must capture Erdeni and return the plun-
dered goods, or else the Qing would take their own measures.
13By fall of
1679 Galdan had ten thousand troops prepared to invade Turfan, and he
was sending scouts farther east to Hami, only 525 kilometers, or ten days’
march, from the Suzhou garrison in Gansu. Still, General Zhang Yong con-
cluded that the news about Galdan’s intentions was not reliable and sent
men to Jiayuguan to investigate.
Galdan did take Hami and Turfan, and he had already conquered the
rich oasis towns of eastern Turkestan. In October 1679 Zhang Yong re-
ceived a message from Galdan, along with gifts of horses and furs for the
emperor. Galdan stated, “The northwest area I have now entirely taken.
Only Xihai [Qinghai], which was divided between my ancestors and your
ancestors, now you control alone. I want to take it back.” Zhang Yong’s
Mongol spies had learned a few more details about Galdan: he was born in
1644, so was now age thirty-six sui;he was “violent and evil, and addicted
to wine and sex.” The previous year he had mobilized troops to invade
Xihai, marched toward it for eleven days, and then dispersed his army. He
had sent troops to Turkestan (the “turban-headed Hui territory”). He did
not yet dare to challenge the imperial troops on the border.
We have only the (hostile) Chinese sources’ version of Galdan’s message,
but even in this version, Galdan claimed only the right to partition the
steppe along the lines of the Ming dynasty. The Ming rulers had never con-
trolled Mongolia or Qinghai, and they extended only a thin strip into
Gansu along the Great Wall. He himself did not appear to threaten the
Qing’s borders; the main disturbance came from the other defeated Mon-
gols seeking refuge within the passes. Zhang Yong concluded that Galdan
had no intentions of invading, and there was no need to worry about
Gansu. Meanwhile, Galdan took advantage of Kangxi’s involvement in the
southwest to increase his scope of action. With the Dalai Lama’s approval,
he took the title of Bushuktu Khan. Bushuktu,derived from the Mongolian
“Boshugh” (decree of Heaven, fate, destiny, command) has very similar
connotations to the Chinese concept of the “Mandate of Heaven.”
16In Oc-
tober 1679 he sent an envoy with tribute gifts to Beijing, announcing his
new title. The Lifanyuan (Court of Colonial Affairs) noted that never be-
fore had a Mongol Khan simply announced a new title to the Chinese em-
peror without first obtaining imperial approval and an official seal. In this
case, however, the court recommended that Galdan’s embassy be accepted,
because he seemed to be submitting to the Qing. Galdan, however, had
140 contending for power

clearly marked himself as determined to act more autonomously than the
other Khans, who first requested seals from the Qing emperor before they
assumed titles.Kangxi’s Mongol policy was to limit Qing intervention on the frontier by
encouraging his tributaries to police themselves. He rejected a request by
the Khalkhas to increase Qing guards at border posts, instead urging the
Khalkhas to maintain control of their followers to prevent them from pil-
laging the frontier.
17After the suppression of the Three Feudatories revolt
in 1681, he summoned his Mongolian envoys to a banquet and instructed
them in the general rules of Mongolian policy. Taking with them imperial
orders and gifts, they would urge the Mongols to make peace among them-
selves. All officials who did not speak Mongolian should have their discus-
sions translated by interpreters, and all discussions should be recorded.
They should respect Mongolian customs and not be excessively concerned
with following Chinese ritual. They must keep tight control of their embas-
sies to prevent causing disputes with the Mongols. This was a comprehen-
sive effort by the Qing ruler to establish contacts with all the Mongolian
Khans on a basis of mutual understanding. The emperor sent an embassy to Galdan in response to his request to of-
fer tribute. It reported back to Beijing in the fall of 1683. Galdan, who had
never before received imperial envoys, had been delighted to hear of their
arrival. He insisted on selecting an auspicious day to receive the imperial
orders, and he was concerned to follow the correct rituals. As he said:
“Chinese customs are complicated. Our country’s customs are simple and
easy. When Chinese board officials receive an imperial order, the banquet is
held only a month later. Although our country does not have boards, we
have zaisang [ministers]. If a zaisangreceives the order, this follows Chinese
custom. If we follow our custom, the Khan himself receives the order and
gifts directly.” He decided to receive the imperial gifts and orders directly.
Interrogating the envoys, he said, “I’ve heard you recently had a rebellion,
which has now been suppressed.” The envoys replied, “There were rebels
recently, but our emperor did not want to disturb the people with arms, so
he gradually induced them to surrender. Some surrendered, some were ex-
terminated, and now all is settled.” Galdan was also impressed that the
Qing had sent Tibetan-speaking envoys to Tibet. He demonstrated his Ti-
betan connections by making the embassy watch lamas doing ritual dances
and chanting scriptures. He offered them numerous banquets and sent
them off with lavish gifts, including four hundred horses, sixty camels,
three hundred sable and five hundred ermine pelts, and, interestingly, “four
Ölöd fowling pieces [niaoqiang].” The envoys agreed that Galdan would
attempt to capture Batur and Erdeni if possible, and both sides would ob-
manchus, mongols, and russians 141

serve the two chiefs’ activities until the fourth month of 1685. Even if he
could not capture and punish them, Galdan promised to prevent the two
from attacking the Qing frontier.
This first embassy to Galdan indicates the flexibility of both sides and
their mutual interest in good relations. Each side was aware of the cultural
differences between them, and neither one wanted to let ritual details inter-
fere with communication. Galdan was seen as less threatening than the
Russians, and a useful resource for the Qing in maintaining control over
the less powerful Mongolian tribes who raided the border. As so many pre-
vious dynasties had done, the Chinese emperor helped one steppe leader to
secure preeminence in the hopes of winning him over to establish stable
tribute relations.
Tribute presentations served as a flexible means of using economic incen-
tives to secure border control. The Qing envoys pressed Galdan to assert
tighter authority over his tributary envoys. Many envoys had come pur-
porting to be Galdan’s men, but without carrying any official credentials.
Galdan noted that his direct envoys all carried his official seal, but the allied
tribes—Derbets, Torghuts, Khoshot, and others—lived too far away to be
provided with official credentials. Also, Galdan’s own kinsmen carried no
papers when they crossed the border. The looser, more casual controls of
the Mongolian Khans clashed with the demands of Chinese bureaucratic
paperwork. When the Qing envoys insisted, “If you send sons and nephews
without credentials, we will not allow them to cross the border. Do you un-
derstand?” Galdan replied, “Whether or not they cross the border is your
emperor’s decision.”
Only two months later the emperor cracked down more severely on
abuses by Galdan’s missions. The embassies were getting too large—up to
several thousand men—and the members were committing crimes along
the route to the capital: “They loot and plunder the horses of Mongols be-
yond the pass, and pasture them at will after they enter, trampling fields,
and plundering people’s goods.”
21 Kangxi had previously set no limits on
the size of Galdan’s missions, but now he restricted them to two hundred
men. Others would have to trade at the border towns. Reliable headmen
(toumu) were to watch carefully over the actions of all who entered the
pass. Although he had pardoned the crimes of the previous embassies, in
the future all who committed crimes in the interior would be punished by
Chinese laws. Extraterritoriality, the right of foreigners to be tried by their
own legal codes, was not part of the emperor’s thinking for these missions. Nevertheless, the next year Galdan sent an embassy of three thousand
men to test Kangxi’s order. All but two hundred were turned back at the
22Galdan disputed the emperor’s restrictions, arguing that since an-
142 contending for power

cient times, trade with the Ölöd had followed fixed regulations, which did
not limit the size of missions; but the Lifanyuan responded that the rules
had changed, and Galdan’s “claim from tradition” was rejected. Galdan,
like so many nomadic state builders, clearly needed the tribute trade to ob-
tain resources, and the emperor, while supporting Galdan’s claim to preem-
inence over the Western Mongols, used the tribute lever to restrain his de-
mands. This was one of the issues that lay behind Galdan’s more aggressive
moves eastward three years later. But the Qing did not cut off all access to
the capital by Galdan, and in fact backed up his exclusive rights in 1686,
declaring that only Galdan and the four great (Khalkha) princes had rights
to trade in the capital; all others must trade at the border.
Trade restrictions did put pressure on Galdan’s authority. His kinsmen
blamed Galdan, not the emperor, for their loss of opportunities to trade in-
24Indirectly, making Galdan the sole authority entitled to send
missions to the capital and requiring all Western Mongol traders to obtain
seals from him created tensions within the Mongolian tribes—tensions that
could be used by the Qing to pit them against one another. As Galdan expanded, he forced other Western Mongolian tribal lead-
ers to attack the Chinese frontier to survive. When Galdan killed Ochirtu
Khan, he took over his possessions and territory, but Ochirtu Khan’s son
Lobzang Gunbu Labdan and his nephew Batur Erke Jinong fled to seek aid
from the Dalai Lama. The Dalai Lama found a place for them to settle at
Alake Shan. In 1683 reports came from the Ordos chief that Batur was pas-
turing along the shores of the Yellow River. He asked that Batur be sent
back to his tribal territory. The emperor sent a letter to Galdan saying,
“This [Batur] is one of your Ölöd people. If you want him back then take
him; otherwise we will take measures.” Galdan replied that he would deal
with Batur the year after next. Kangxi declared that the Qing could abso-
lutely not accept Batur, although he knew that Galdan’s people were said to
be so poor that “anyone who owned one horse is considered very wealthy
there.” Expecting internal divisions to break out soon in Galdan’s reign, he
decided to wait before taking action.
Prince Gandu had also been roving along the frontier ever since Galdan
killed his grandfather. In 1684 Batur, Erdeni, and Gandu requested the em-
peror’s pardon in return for swearing loyalty to the Qing.
26 The emperor
made it clear that he could have sent troops to exterminate (jiaomie)them
as criminals, but he pardoned them because of the long-standing presenta-
tion of tribute to the Qing by Ochirtu Khan. He decided to settle Batur’s
people together with those of Lobzang Gunbu in a convenient place. Stress-
ing that Galdan had refused to accept Batur’s people, he proclaimed his
merciful acts to an assembly of Jasaks, lamas, and representatives of the
manchus, mongols, and russians 143

Dalai Lama: “You lamas constantly plan for the common people with mer-
ciful hearts. All the Ölöd nobles present gifts to the lamas and follow re-
spectfully their laws. I know that Lobzang Gunbu Labdan and Batur Erke
Jinong are descendants of Ochirtu Khan, and Ochirtu Khan long protected
the laws of the lamas. How can you bear to watch silently as his descen-
dants are driven into poverty!”
Kangxi thus launched an effective propaganda campaign to persuade the
lamas to settle Batur’s and Lobzang’s people under the emperor’s protec-
tion. These were the first Western Mongols to seek Qing protection. Nego-
tiations ensued to establish clear boundaries for their settlement. The em-
peror held the power to eliminate the tribes, as he made clear, but gained a
reputation for mercy which would suit him well in the future, and his ap-
peal to shared values between the Buddhist lamas and the Qing regime was
a step toward drawing the Tibetan Buddhist church away from support of
the Mongols and toward the Manchus.
Galdan’s Intervention
A major split between the two “wings” of the Eastern Mongols drew both
Galdan and Kangxi into conflict. On the “right” (western) wing were
the Jasaktu Khan and Altyn Khan; on the “left” (eastern) wing were the
Tüsiyetü and Chechen Khans.
28 Their dispute went back to the “Lobzang
incident” of 1662, the first year of Kangxi’s reign, when Lobzang, of the
Altyn Khan lineage, attacked and killed the Jasaktu Khan, Wanshuke.
The Kangxi emperor reprimanded Lobzang for violating one of the Qing’s
tributaries. The left-wing Tüsiyetü Khan also came to the aid of the Jasaktu
Khan and crushed Lobzang, who fled across the Russian border. The
Jasaktu Khan’s rule was saved, but during the chaos many of his people had
fled to the Tüsiyetü Khan’s lands and remained there. After the Tüsiyetü
Khan installed Chengun, Wanshuke’s brother, in the Jasaktu Khan’s posi-
tion in 1670, peace returned to his lands, but the Tüsiyetü Khan refused to
send back his people who had fled to him for protection. Galdan had him-
self been involved in the dispute as a mediator, trying to draw in the Dalai
Lama to support the Jasaktu Khan. By 1684 the Jasaktu Khan was still seeking the return of many of his peo-
ple from the Tüsiyetü Khan, with no result, and he appealed to Kangxi and
the Dalai Lama for help. The emperor was inclined to have the Dalai Lama
settle the matter:
You, Dalai Lama, are compassionate...allrespect you, Khalkha lords
worship you, respect your teachings and laws, and you are sincerely
144 contending for power

Portrait of the Lcan-skya (Ch. Zhangjia) Hutukhtu, dated 1693. The high lama is
flanked by Tsongkhapa, founder of the Yellow Sect, on the left, and Shakyamuni, the
historical Buddha, on his right. Below are the demon Demchog with his partner,
White Tara, and the four-headed Mahakala.
[To view this image, refer to
the print version of this title.]

obedient to this dynasty. I am distressed that the Jasaktu Khan’s people
cannot gather together; this will create mutual killing and war, then
how can the two Khans coexist? You must have the Jasaktu Khan’s
people return and make peace between them, following my impartial
benevolence. You can send a high lama to meet my high official on the
Khalkha border, and set a time to meet.
But the Dalai Lama’s representatives were unable to get the two sides to
make any agreements that year. On the death of the Jasaktu Khan, Kangxi
confirmed his successor. By 1686 he heard that battles had broken out
between the two wings, with “brothers and kinsmen attacking one an-
other,” and men were fleeing from the left wing to the right wing and vice
Arni, the emperor’s envoy, met at Kuleng Barqir (Mo. Küriyen belciger)
with the collected Khans and nobles of the two wings and the Dalai Lama’s
envoys to urge the battling Khalkhas to make peace. Also present was
146 contending for power
Engraving of the Potala, the Dalai Lama’s central temple in Lhasa.
[To view this image, refer to
the print version of this title.]

the brother of the Tüsiyetü Khan, titled the Jebzongdanba Khutukhtu, the
leading Buddhist cleric of Mongolia, who claimed equal and independent
authority with the Dalai Lama in Tibet. Arni passed on the emperor’s
words: “You are all descendants of the same ancestor. If you continue fight-
ing, you will all be eliminated. There is no profit in peace for me; there is
gain in war for me.” The emperor implied that although he and the Qing
would actually profit from the division of the Khalkhas, their real interest
was in a unified, peaceful Mongolian people. The Mongolian princes ex-
pressed their appreciation for the efforts of the emperor and Dalai Lama to
bring them together, but noted that “whether we make peace is up to us
two Khans.” On October 10, 1686, before the Jebzongdanba Khutukhtu
and the Dalai Lama’s envoys and Buddha’s image, the Khans, princes, and
over sixty ministers swore oaths of eternal peace, and each promised to re-
turn to his own lands.
The compilers of the official history of Kangxi’s campaigns commented
on the emperor’s activities here, demonstrating the difference between the
contemporaneous Qing view and the retrospective view developed at the
time of compilation in 1708. They noted that “state builders” (liguozhe)
are threatened primarily by internal dissension, and secondarily by external
threats. Because the Ölöd and Khalkhas were close neighbors, any weak-
ness among the Khalkhas would lead “the strong to gobble up the weak.”
When brother fought brother among the Khalkhas, with “the roots torn up
the leaves would wither,” and the Khalkhas were threatened with extinc-
32In the compilers’ view, Kangxi’s intervention thus persuaded them to
make peace before they were destroyed by Galdan’s invasion. But the em-
peror at the time made no reference to Galdan, nor did he indicate that
Galdan was a major threat to either the Khalkhas or the Qing. He was still
setting Galdan up as the official tributary trader of the Western Mongols.
Kangxi’s border policy continued to be defensive, aiming to ward off refu-
gees from the Qing borders by trying to establish peace among the Khal-
khas. Only after the defeat of Galdan was history rewritten to suggest that
he had always been an implacable enemy of the Qing. The next year the emperor confirmed the succession of the Chechen and
Tüsiyetü Khans with imperial seals and congratulated them for making
peace. In the second month, however, he received a communication from
Galdan complaining about a violation of ritual propriety at the 1686 meet-
ing. The Lifanyuan, reporting that the rituals of the Yellow Hat sect, the
Dalai Lama’s sect, had been followed properly, regarded the matter as
closed. But Galdan could not let it rest. He sent letters to Arni, the Jebzong-
danba Khutukhtu, and the Tüsiyetü Khan spelling out his complaint: at the
Kuleng Barqir meeting, the Jebzongdanba Khutukhtu, a subordinate of the
manchus, mongols, and russians 147

Dalai Lama, had sat directly across from the Dalai Lama’s representative,
violating the clerical hierarchy.By July 1687 Galdan had moved his headquarters eastward from the
Altai and summoned the followers of Jasaktu Khan to meet with him. The
Tüsiyetü Khan feared an immediate attack, but Kangxi did not respond
quickly. By October the Tüsiyetü Khan was reporting definite evidence that
Galdan planned to attack by two routes, and his followers were urging him
to meet Galdan in battle. The emperor only continued to urge the two sides
to make peace, and issued a remarkably conciliatory edict to Galdan. He
once again repeated his intention for all his tributaries to live in peace, al-
lowing that he might have heard false rumors that Galdan was preparing
an attack: “You, Galdan, have been obedient and presented tribute. I may
not have reliable information because the distance is so far, or some follow-
ers may be stirring up trouble.”
33He denied permission to Batur Erke, the
refugee Mongol leader settled in Qing territory, to take his men to support
the Tüsiyetü Khan by the shortest possible route. At the end of 1687 the Tüsiyetü Khan invaded the Jasaktu Khan’s terri-
tory, killed the Khan, and scattered his people. Galdan’s younger brother,
who had led four hundred troops against the Tüsiyetü Khan, was killed in
the attack. In revenge for the loss of his brother, Galdan’s troops crushed
the Tüsiyetü Khan and invaded Erdeni Zu, the great monastery built dur-
ing the Ming dynasty at the old Yuan capital of Karakorum.
34 Galdan re-
mained in the rear while six thousand men plundered the Erdeni Zu region,
destroying temples and scriptures of the Jebzongdanba Khutukhtu. Thus
they took their revenge for his insolence toward the Dalai Lama. Galdan’s attack shattered the Khalkha confederation. As the Chinese
chronicle puts it, everywhere the Khalkhas “abandoned their tents, posses-
sions, horses, sheep and fled south, day and night without stopping.”
Other Khalkha tribes fled north. They besieged the new Russian envoy,
Fedor Alekseevich Golovin, at Selenginsk, until he drove them off. Several
chieftains offered to swear oaths of submission to Russian rule in return
for protection from Galdan. Golovin, however, demanded “eternal loy-
alty” to the Tsar, while the Mongols sought only temporary alliances. Some
of them rejected the Russian demands and turned to join the rising power
of Galdan, while others maintained Russian connections until Galdan’s de-
feat. The Khalkha tribes split in three directions—Russian, Chinese, and
Zunghar—finding no unity while they were under attack.
What were Galdan’s motivations for these attacks? Most Chinese histori-
ans treat his bringing up the issue of ritual propriety toward the Dalai
Lama as a mere pretext for his plan to destroy the Tüsiyetü Khan. But
Galdan, raised from youth under the shelter of the Tibetan lamas, had
148 contending for power

strong feelings about proper deference toward the Dalai Lama. The restric-
tion of tribute trade opportunities by the Qing also forced him to make
a decisive move. A true peace among the Khalkhas would have blocked
Galdan’s expansion to the east and thwarted his ambitions to dominate all
the Mongolian peoples—if he had such ambitions. And finally, the desire
for personal revenge for the death of his brother led him to order an orgy of
destruction of the temples belonging to his enemy. We need not assume any
single consistent plan behind the tempestuous Galdan’s actions, but only an
accumulation of economic, cultural, and personal slights.The pro-Manchu Eastern Mongol and banner commander Lomi, author
of the History of the Borjigid Clan [of Chinggis Khan](Mongghol borjigid
obogh-un teüke), written from 1732 to 1735 based on contemporary
sources, believed that Galdan’s motive for attacking the Khalkhas was sim-
ply to avenge the death of his brother. The Jesuit Jean-François Gerbillon
(1654–1707), who accompanied the Kangxi emperor on his campaigns,
agreed. These views differ from later accounts, which ascribe to Galdan
much more ambitious aims. If Lomi’s account reflects “the general under-
standing of the class of his time,” it probably best represents Galdan’s im-
mediate views when he invaded.
At first, Kangxi again reacted defensively to Galdan’s invasion. He sent
2,500 troops to the border, who soon had to hold off defeated Khalkhas
in flight. The Tüsiyetü Khan and the Khutukhtu first fled to the Chechen
Khan’s territory, but even though the Khan offered the refugees support,
their people continued to flee south, crossing the Qing border at the Sunite
guard post (kalun).Investigators found six hundred lamas and two thou-
sand families, a total of twenty thousand individuals, desperately seeking
Chinese protection.
38The emperor ordered a full discussion by the Deliber-
ative Council on whether to drive the Khalkhas away from the border or
protect them from Galdan’s pursuit. The council advised that it would be
impossible to drive the desperate refugees away, but if they stayed long
in the area, they would destroy the local grazing supply. The emperor
still temporized, deciding to wait another month to see how the situation
developed. Meanwhile, Galdan sent a letter explaining that he was punishing the
Khan and the Khutukhtu for violating the respect due the Dalai Lama,
saying he had “fought for the Dalai Lama’s soul by destroying their dwell-
39He told the Qing ruler to refuse to accept the Khutukhtu’s surren-
der and instead arrest him. Again the emperor was remarkably noncommit-
tal, refusing to blame either Galdan or the Khan and Khutukhtu. But by the
next month, as further flights by defeated Khalkhas increased pressure on
the border region, the Qing had to send troops to protect the Khutukhtu.
manchus, mongols, and russians 149

Galdan had now reached Hulunbeir (Mo. Külün bayir), seven to eight
days’ march from the border. The emperor prepared ten banner compa-
nies, a total of ten thousand men, for defensive action on the border under
the command of the Khorchin Tüsiyetü prince Shajin.
40Emissaries sent to
Galdan heard him say, “If I make peace with the Tüsiyetü Khan, who will
recompense me for [the loss of] my brother Dorjizhabu? If I put out all my
effort, in five to six years, I can destroy the Khalkhas, and capture the
Jebzongdanba Khutukhtu.”
41When Galdan withdrew from the Qing bor-
der to attack the Tüsiyetü Khan, however, the emperor declined to pursue
him. Then, when the Tüsiyetü Khan desperately appealed for military aid
from the Qing to attack Galdan, Kangxi laid down his conditions: the
Khan could retain his title, but he must surrender to the Qing and agree to
be settled under Qing supervision. On August 28, 1688, the Tüsiyetü Khan
and Galdan fought a pitched battle for three days. Galdan crushed the
Khan, who fled, isolated and weak, across the desert to the Khutukhtu’s
territory. Kangxi took further defensive measures, sending two thousand
troops to guard the Ordos. Two months later the emperor ordered a partial withdrawal. The border
troops lacked clothing for the oncoming winter, their horses were worn
out, and they were running out of supplies of rations and fodder. Much
of the available grain at the border had been distributed to the refugee
Tüsiyetü Khan’s people.
42 By the spring of 1689, over twenty thousand
starving Khalkhas had come to request relief, and supplies were low. The
emperor was now inclined to agree with Galdan’s version of events: it was
the Tüsiyetü Khan who had violated the oath of peace, killed the Jasaktu
Khan and Galdan’s brother, and first launched the invasion of Mongolian
territory. The Tüsiyetü Khan had “brought on his own destruction” (ziqu
miewang), and Galdan was not to blame.
43Gerbillon also believed that the
Jebzongdanba Khutukhtu, the younger brother of the Tüsiyetü Khan, had
“ruined his family by his arrogance, because he put himself on a par with
the Grand Lama of Tibet.”
44The emperor did not even blame Galdan very
much for destroying the Buddhist temples and images in Khalkha territory.
Nevertheless, the emperor pardoned the Tüsiyetü Khan’s crimes and re-
fused to surrender him and the Khutukhtu to Galdan for punishment. Once
again he urged the Dalai Lama to intervene to make peace between the con-
tending parties, proclaiming his disinterested benevolence: “I do not rejoice
at the defeat of others as my own profit. But if they surrender to me, I am
Lord of Heaven, and must take care of them, else who will? I will give to all
who seek help; just as you lamas will not let people die, you will take care
of them too. My goal is to dissolve the contention between Khalkha and
Ölöd, and create peace. If you can send lamas to Galdan, urging him to
150 contending for power

make peace, he will follow your laws.” 45At the same time, he threatened to
cut off all of Galdan’s trading privileges if he did not abandon his vows of
revenge. While Galdan was moving east to meddle in Khalkha affairs, conflict was
brewing back home in Zungharia. Tsewang Rabdan, the son of the assassi-
nated Sengge, had grown to maturity and become a threat to Galdan, who
tried to have him killed in 1688. Tsewang Rabdan escaped, but his brother
was killed, and when Galdan left to march against the Khalkhas, he at-
tacked Hami with his own men. This forced Galdan to move back to meet
Tsewang Rabdan and relieved the pressure on the Eastern Mongols, allow-
ing the Qing to reduce their troop allotments by one-half. Arni’s interview
with Galdan in the fall of 1689 indicated that the Mongol leader was still
determined to avenge his brother’s death, even though his people were con-
fronting famine, “eating human flesh,” and facing a hard battle.
also indicated his suspicions that the Qing were about to forge an alliance
with the Russians and the Khalkhas against him, since Arni had just re-
turned from Selenginsk with two thousand troops. Arni assured him that
he had only negotiated border issues with the Russians, and the troops he
brought with him were not to be used to support the Khalkhas. Kangxi’s
edict to the Dalai Lama at the end of the year reflected his understand-
ing that Galdan had been defeated by Tsewang Rabdan and was in desper-
ate straits. Asking the Dalai Lama for confirmation, he offered to accept
Galdan’s surrender to the Qing. Within less than a year, in the emperor’s
eyes, Galdan’s forces had been reduced from a serious military threat to a
starving remnant band deserving of pity. Qing officials proceeded to organize the Khalkhas into banners and settle
them temporarily on the border. The emperor told the Khalkhas that they
had no discipline (fadu);they needed a strong hand of supervision to pre-
vent them from dispersing. Governors (Jasak) were appointed to collect the
remnant Mongols, assign them to banners, and allot them fixed pastures.
But the emperor also promised that they could return to their old homes
when the Ölöd conflicts were settled. Through the winter of 1689–90 Galdan remained in Khobdo to resist the
rebellion by Tsewang Rabdan, at the same time collecting troops for a
move east against the Khalkhas. Although he had a few thousand follow-
ers, many of them were tempted to desert him for Tsewang, unless there
was promise of significant booty from attacking the Khalkhas.
47 Kangxi
continued to send small detachments to the border for defensive measures:
500 to Guihua, 1,500 to the Ordos. At the same time, he established con-
tact with Tsewang Rabdan to find out the reason for his rebellion. The
Chechen Khan promised to attack Galdan with 10,000 troops if the em-
manchus, mongols, and russians 151

peror gave his backing. For the first time, the emperor now allowed the
sale of Qing military supplies to the Khalkhas. He doubled the number of
troops heading to the Tula River and specified that they should bring can-
non with them.
Meanwhile, the Dalai Lama appeared to diverge from the Qing em-
peror’s goals. The regent (T. sDe-pa,Ch. Diba), the highest-ranking official
in Tibet, told the emperor’s emissary that he must arrest the Tüsiyetü Khan
and Khutukhtu and deliver them to Galdan. This new support for Galdan
aroused suspicions about the Dalai Lama. What Kangxi did not know was
that in fact the fifth Dalai Lama had died in 1682 and that real secular
power was then assumed by the sDe-pa,who refused to announce the death
of the Dalai Lama or to install a new one. The sDe-pawas an active sup-
porter of Galdan’s campaigns and an enemy of the influential Khoshot
princes in Tibet and of the Khalkhas. Kangxi, however, continued to appeal
to the Dalai Lama as his ally in making peace among the Mongols.
On June 9, 1690, Galdan led thirty thousand troops across the Urja
River to attack the Kundulun, Chechen, and Tüsiyetü Khans.
50Reports in-
dicated that Galdan intended to meet up with supporting Russian troops.
Songgotu (Ch. Suoetu), the Manchu envoy to the Russians at Nerchinsk,
quickly delivered them the message that Galdan was threatened with inter-
nal revolt, his troops lacked food, and the Russians should not intervene.
Qing generals sent more troops and cannon to the frontier. The Russians
denied Galdan’s request for twenty thousand troops.
Kangxi’s First Personal Expedition
On July 27, 1690, the emperor announced that he would personally lead
the expedition against Galdan. Kangxi’s first personal campaign (qinzheng)
was now under way. Three armies would converge on Galdan’s camps: one
led by Prince Fuquan, the emperor’s brother, leaving from Gubeikou, one
led by Prince Changning from Xifengkou, and one led by the emperor him-
self. He invited the Jesuits Gerbillon and Pereira to accompany him. Proba-
bly sixty thousand troops participated in the campaign.
52This sudden shift
from defensive measures to a personal campaign remained without jus-
tification at the time by the emperor. Clearly it cannot be explained, as later
historians did, as a response to a growing threat from Galdan’s rising
power, for Kangxi knew that Galdan was in desperate straits. Galdan’s
march to the east was primarily a plundering expedition to support his
starving troops. Although he had a large army, many of them would
quickly desert him if they did not get their loot, and Tsewang Rabdan still
152 contending for power

threatened to his rear. By the end of the fifth month Galdan was camping
on the lower Kerulen River with his supporters, who had dwindled to ten
thousand men. Their food was nearly gone, and they were consuming their
horses to survive. Yet Kangxi proceeded cautiously. He did not expect a
long campaign. He issued four months’ rations to the troops, half of which
they would carry with them, the rest to be bought on the way.
53But he or-
dered the advance troops to wait until the main army had arrived with suf-
ficient supplies of cannon and men and to avoid battle at all costs. Other re-
ports from fleeing Khalkhas put Galdan’s strength at (an exaggerated) forty
thousand men, but no one knew exactly where he was. Kangxi told his
Mongolian allies to mobilize ten thousand men from the Khorchin banners.
Galdan, by contrast, continued to believe in making peace with the Qing,
and tried to confine his conflict to internal Mongolian matters. Meeting
several of the Mongolian bannermen, he told them, “We are of the same
tribe [buluo]; why do you fight us? The Khalkhas are our enemy,” not the
Mongolian bannermen of the Qing. Galdan’s younger brother, who com-
manded the troops, assured the Qing that his men would not cross the bor-
der. But meanwhile, Kangxi was preparing his full-scale expedition. He set
the date of departure of the main army from Beijing for August 8, 1690,
followed by the emperor himself two days later. He paid special attention
to providing cannon for the army. In his edict to Galdan, he accused him of
violating the empire’s frontiers and refusing to return imperial envoys, and
he claimed that Qing troops were acting purely for defensive purposes. En-
voys to Galdan were not to inform him that the emperor and his princes
were personally leading the campaign. In fact, Kangxi made clear his inten-
tion to destroy Galdan only in secret instructions to the armies on the fron-
tier: “Have one man take a clerk to the armies, and secretly tell them: ‘The
emperor has sent us to tell you to advance slowly, because the great army
will soon arrive. Prepare your troops, but do not engage in battle. Every
night send out patrols, and wait for reinforcements. If Galdan receives or-
ders and retreats, then stop him and tell him not to retreat. If he does not
stop, have the armies attack immediately.’” In other words, the goal was to
keep Galdan from fleeing, and to avoid battle until the main body of the
army could arrive.
Military operations seldom proceed according to plan. As Galdan con-
tinued to pursue the Chechen and Tüsiyetü Khan’s men along the Khalkha
River, Arni, the Qing commander, instructed to observe but not to fight, got
into battle unintentionally when he ran across Galdan at the Urhui River.
Arni’s two hundred crack Mongolian troops and five hundred Khalkha
raiders could not withstand Galdan’s twenty thousand men drawn up in
battle formation. Galdan had both fowling pieces and cannon, but Arni’s
manchus, mongols, and russians 153

cannon had not yet arrived. The Zunghar attack drove the Qing troops into
ignominious retreat.Now Kangxi had to salvage the situation from Arni’s blunder. First he
stripped Arni of his position and degraded him by four ranks. Then he tried
to persuade Galdan that it had never been his intention to attack the Mon-
golian leader, putting all the blame on Arni’s illegal action. His thoroughly
disingenuous edict stressed the common interests of Zunghars and Qing
and the purely defensive nature of the Qing mobilization, and evaded Gal-
dan’s main demands: the extradition of the Tüsiyetü Khan and Khutukhtu.
Once again, the envoy to Galdan indicated vaguely that Qing troops were
moving to the frontier but concealed the presence of the emperor and the
high princes.
By the seventh month there were already problems with supplying the
troops beyond the passes. Reports indicated that the horses were tired, the
sheep worn out, and rations scarce, requiring more grain shipments. The
Qing strategy was still to advance slowly and wait until the Great Army
had arrived. Galdan, it was feared, would retreat in the face of such large
forces, making it impossible to crush him in battle. The first army was ex-
pected to arrive at Galdan’s camp in ten days, so it was necessary to stall
him with the false offer of peace talks. Thus the Qing would “use these dip-
lomatic messengers going back and forth to delay him and wait for the
Great Army.” Galdan indicated genuine interest in peace negotiations; he
claimed that he never intended to violate the Chinese frontier, but was in-
terested only in taking revenge on those responsible for his brother’s death.
Kangxi, in reply, reproached Galdan severely for raiding the Qing border
posts, but offered to discuss peace under the Dalai Lama’s auspices. He
again claimed that the again huge army he was assembling was “not to
punish you, but to establish discussions.” The emperor invoked his recent
experience with the Russians, in which troops were sent to the frontier
to establish peace negotiations and withdrawn when the treaty was con-
cluded. He neglected to mention the siege and destruction of the fortress at
Even though Galdan had now moved away to the north, the emperor
was determined to persevere with his personal expedition. It seems that the
emperor was looking for pretexts to continue his personal campaigns in the
face of criticism. Instead of justifying his campaign as a personal drive
against Galdan, he claimed that he needed to accompany the troops in or-
der to put down disturbances among the Khalkhas.
57The Khorchin prince,
whose supplies were running out, had already given up the pursuit of Gal-
dan, so he returned to his pastures, despite the emperor’s severe criticism.
As Galdan continued to indicate interest in peace talks, the emperor asked,
154 contending for power

“How can we rein in [jimi] Galdan so as to wait for the Shengjing army to
58He sent Galdan sheep and set a time and place to meet for talks.
At this critical juncture, in late August, the emperor was forced to return
to the capital at the urging of his ministers. “Hot weather,” probably the
cause of an illness lasting through the month of September, prevented him
from spending a long period beyond the passes. His ministers also had
feared unrest when the Chinese in the south learned that the emperor had
left the capital.
59Galdan, at the same time, supported by the Dalai Lama’s
envoy, requested a meeting at Ulan Butong (350 kilometers north of Bei-
jing) to discuss the return of the Tüsiyetü Khan and Jebzongdanba Khu-
tukhtu. He was only 40 li(23 km) away from the Qing army.
Fuquan, the Prince of Yu, then reported a crushing defeat of Galdan.
Having advanced to Galdan’s camp at Ulan Butong, he saw the enemy at
noon on September 3, 1690. The Qing fired their “deerhorn cannon” and
ordered an attack. The Zunghar troops protected themselves in the forest
at the foot of the mountains and used their camels to shield themselves
from artillery fire. As Maska, an officer in Fuquan’s army, described it, the
enemy “bound the feet of ten thousand camels, covered them with felt, and
lay down behind them, spread out like a wall. We called it a ‘camel wall’
[tuocheng]. They shot arrows from slits in the wall.”
60The left wing of the
Qing army was able to surround the Mongol troops in the mountains and
inflict great casualties on them, but the right wing, blocked by a great
marsh, had to return to its original position by nightfall. Although Fuquan
reported that Galdan had suffered heavy casualties, in fact he had not
driven him from his position, and he knew that Galdan was prepared to re-
sist strongly the next day. Gerbillon did not witness the battle but heard re-
ports of it immediately. Although he described the battle as a “defeat” for
Galdan, he knew that the Qing troops had retreated to their camp after the
first day’s fighting, and neither side was prepared to fight further. The Rus-
sian envoy, Kibirev, who was with Galdan’s forces, did not see the battle as
a Qing victory.
61The emperor, however, was highly optimistic, and for the
first time publicly announced his goal of a “final solution”: “We must now
consider how to pull up the roots entirely [qiong qi genzhu], wipe out the
remaining followers, and clean up everything permanently with one blow
[yiju yongqing].”
His optimism was premature. Both Fuquan and Galdan knew that the
Qing might win a second battle, but the outcome was uncertain. Qing
troops suffered severe supply difficulties; Galdan had lost many men and
horses but remained encamped securely in the forest. Qing artillery could
not ensure victory, as the Mongols were protected by both camels and trees.
When Galdan sent emissaries to talk peace on September 7, the field com-
manchus, mongols, and russians 155

[To view this image, refer to
the print version of this title.]

The battlefield of Ulan Butong, where Qing forces fought the Mongol leader Galdan
in 1690. Contrary to later accounts, this was not an overwhelming victory for the
Qing. Galdan escaped, and Commander Tong Guogang, the Kangxi emperor’s uncle,
was killed in the battle. According to legend, Tong’s cannon sank into the marshy
ground, and spring water gushed out to form a lake, now called the General’s Lake.
[To view this image, refer to
the print version of this title.]

manders considered it best to hold off the attack. Galdan now conceded the
main issue: he did not demand the immediate return of the Khan and
Khutukhtu, but asked only that the Khutukhtu be sent to his superior, the
Dalai Lama, for investigation. The generals told Galdan that he must move
far away from the Qing borders and promise never to raid the frontier
again. Their troops were prepared to advance and exterminate(jiao)Gal-
dan if he did not agree. In their thinking, they should not further weary
the troops with an unnecessary battle; either Galdan would withdraw to
a remote region, or, if he remained nearby, they could wait for the planned
meeting in four to five days with the advancing troops to strike the deci-
sive blow.
The emperor, highly suspicious of the crafty Galdan, urged the generals
into rapid pursuit, but he implicitly had to accept the logistical limitations
of the frontier. He allowed his son Yinti to return to the capital, under the
assumption that there would be no further attacks. Galdan, meanwhile,
having successfully stalled for time, fled as fast as he could to the north un-
der cover of darkness. The generals wanted to pursue him, but “the horses’
strength did not allow [them] to advance.” On the ninth, as planned, they
met with the Shengjing Ula and Khorchin troops; the Great Army was still
100 li(58 km) away. Galdan’s envoys, who had promised to obtain an oath
of allegiance from Galdan, were sent after him to get his written commit-
ment. Galdan, in the course of flight, plundered twenty thousand sheep and
over one thousand horses to replenish his army, but the generals continued
to wait for his answer. Although the emperor urged a quick advance, he re-
alized that “the princes and generals are with the army, and they personally
see the situation.”
By the middle of September, troops had begun to withdraw to defen-
sive positions. Frontier garrisons provided for returning troops who lacked
food and horses. Galdan’s official oath arrived on September 20. He had set
up a Buddhist statue in his camp, bowed low before it, and asked forgive-
ness for his crimes. He promised to withdraw far from the Qing frontier in
search of a place with “good water and grass, and no people.” After discus-
sion with the Deliberative Council, the emperor, still distrustful, accepted
Galdan’s oath. Once again he proclaimed his common interest with the
Dalai Lama in peace between the Zunghars and Khalkhas. Galdan had vio-
lated the Dalai Lama’s teaching by entering Qing territory and plundering
the frontier Mongol tribes, and the proper Qing policy would be to pursue
him “across steep mountains, to remote frontiers” and to execute him as a
rebel. The emperor made the extraordinary claim that his princes and min-
isters all urged a second battle but he alone had stopped them. (On the con-
trary, the emperor had urged battle against his generals’ resistance, but by
158 contending for power

distorting the facts, the emperor could pose as a peacemaker.) In return,
Galdan must not only withdraw from the frontier but cut off all contact
with other Mongolian tribes as well. The emperor promised to destroy him
mercilessly if he violated his oath.
Even allowing for normal diplomatic duplicity, Kangxi’s craftiness cer-
tainly exceeded that of Galdan. The emperor was in fact furious at Galdan’s
escape, and he took his revenge on his generals. But the brute facts of dis-
tance, food, and horses made his most desired goal, the elimination of an
autonomous Mongol leader, impossible.
At the end of the year, trials determined punishments for the generals
who had failed to destroy Galdan. Fuquan, the emperor’s elder brother,
was pardoned because of inexperience in combat. Although other high-
ranking Manchus accused Fuquan of negligence, the emperor’s son would
not criticize his uncle’s conduct on the campaign. Fuquan was stripped of
his rank and forced to remain on the frontier for several months. Other of-
ficers received whippings and imprisonment, notably the artillery captains
who had abandoned a heavy cannon on the battlefield.
The aftermath of the abortive campaign revealed clearly the severe sup-
ply problems that had already begun to plague the Qing troops from the
first days of mobilization. Until the Qing army overcame fundamental sup-
ply constraints in the mid-eighteenth century, this pattern repeated itself.
Apparent smashing victories were followed by rapid abandonment of the
frontier, allowing the nomadic rival to revive. As soon as Galdan fled be-
yond reach, the Qing generals began their withdrawal. The first troops to
withdraw did not even have horses; camels loaded with food supplies had
to be sent to meet these starving soldiers on their return.
66Violating regula-
tions, Duke Zhuoketuo and Vice Minister Alami of the Board of Revenue
had illegally advanced to the Shengjing army two months’ extra rations and
five months’ extra salary because they knew that official rations could not
meet the urgent needs of these soldiers about to go out on campaign.
The emperor pardoned these officials because they had been moved by the
troops’ poverty and hunger. Two months after Galdan’s flight, troops were
still waiting on the frontier for word of Galdan’s oath of submission, but
this army needed an immediate shipment of four hundred camels with sup-
plies to sustain itself. Instead, the emperor ordered a full withdrawal back
to the border posts, including cannon, leaving only four hundred men.
Even before securing Galdan’s official submission, the Qing armies could
not support their expensive campaign beyond the wall. But supply costs had dogged the expedition from its earliest days.
Maska, a Manchu general who wrote a diary of his experiences with the
army he accompanied from Beijing, described vividly the climatic extremes
manchus, mongols, and russians 159

beyond the wall. 68 After crossing the Great Wall at Zhangjiakou, the sol-
diers first had to traverse the Dabaghan mountain range in cold, drenching
rain on narrow paths under steep cliffs. Beyond the mountains they entered
the steppe, which was so dry that they had to dig wells to find water, even
though they were pelted with “hailstones big as peaches.” Once they en-
tered the Gobi desert, the only water they found “stank of rotting flesh”
and made them vomit because of the infestation of tamarisk roots, and they
found no animals or birds except for small marmots.
69Good drinking wa-
ter lay five feet deep under the sand. Having crossed the desert, they once
again were struck by heavy rains, which nearly drowned both men and
horses, and by now their grain supplies had run out, leaving them unable to
advance. The exhausting struggle with the elements made it difficult for
men and horses to move at more than a crawl. It took twelve days to cross
the Gobi, a distance of 300 kilometers. Only when they met up with other
troops could they proceed into battle against Galdan. Horses were costly to purchase and to feed. Before the campaign, the em-
peror had caused a riot in Beijing when he set a price of 12 to 20 taels
per horse, probably because the excessively low price induced officials to
seize horses from local people, including high-ranking literati.
70During the
preparations for the campaign, each captain took charge of pasturing ten
horses, which he would turn over to the Ministry of War when they were
strong and fat. Nevertheless, inspectors constantly reported thin, starving
horses at frontier garrisons.
71No unit could remain long in a fixed location
without quickly running low on supplies for both horses and men. The em-
peror had expected a short campaign of only two months, but as soon as he
announced his personal expedition, he started to hear of exhausted troops
and horses. Tired men could only travel slowly, and they were forced to de-
pend on the local population to “help” them with supplies. He had to pro-
hibit soldiers from selling weapons to the Mongolians in exchange for food
and horses. Troops carried both grain and silver with them, but silver pur-
chases drove up prices on local markets, and many soldiers were reluctant,
or unable, to carry two months’ worth of rations on their backs. Even
though the furious emperor wanted to punish his generals who retreated,
he recognized that they had suffered severe shortages, stressing that “China
cannot rest in levying troops and planning supplies.”
Even after the campaign, supply costs took their toll in the form of heavy
debts. As noted earlier, the emperor spared from punishment two officials
who had illegally lent official funds to their deeply indebted soldiers. After
indebted soldiers threatened to riot in the capital, and several even tried to
force their way into the palace, the emperor announced that he would take
responsibility for their debts, at a total cost of 16 million taels.
Qing officials, despite their intentions, had to supply not only their own
160 contending for power

troops but also those of their Mongolian allies. For a while they refused the
military aid offered them by the Khorchin prince, until he promised to pro-
vide his own grain and horses. Even after accepting the alliance, they at first
refused to allow him to purchase supplies on local markets, but finally re-
lented. The next month the prince had to return home when his supplies
were exhausted. Even though he “deserved execution” for his betrayal, the
emperor had no choice but to pardon him.
74 When the Khalkha Mongol
refugees fled across the border, however, the Qing had to provision over
twenty thousand starving people immediately. Commander Fiyanggû sold
his stocks of tea and cloth and used silver to buy them animals and several
months’ worth of grain.
75Refugee demands for subsistence pushed the em-
peror to attempt to finish Galdan once and for all. Even though the compilers of the campaign history fulsomely praised the
emperor for sparing the common people the burdens of supplying his cam-
paigns, imperial demands strained the edges of the local economy and cen-
tral resources. Chen Feng estimates the total cost of the first Zunghar cam-
paign at 3 million taels, or 6 percent of treasury holdings in one year.
Qing armies had proved that they could confront and drive off a Mongo-
lian force in the steppe, but they learned that the logistical costs of such ex-
peditions were high and the economic burdens heavy. Galdan lost many
men and horses but survived to fight another day. Exterminating the rival
state required more than ambitious military plans; the Qing rulers needed
allies, both Mongolian and Russian, to achieve their goal, and they needed
economic development to support military mobilization.
The Treaty of Nerchinsk and the Excluded Middle
The Chinese emperor’s successful campaigns against Galdan would have
been impossible without Russian acquiescence. China and Russia signed
the Treaty of Nerchinsk in 1689 and the Kiakhta trade treaty in 1727.
These treaties had decisive consequences for Central Eurasian power rela-
tions. Their most important effect was to reduce the ambiguity of the fron-
tier by eliminating unmapped zones. Peoples in between the two expanding
agrarian empires took advantage of the fluidity of this zone to protect their
identities through shifting allegiances. After 1689, refugees, deserters, and
tribespeople had to be fixed as subjects of either Russia or China. Maps,
surveyors, border guards, and ethnographers began to determine their
identities and their movements. The treaties served both empires internally
and externally by stabilizing movements across borders and enabling the
suppression of groups who did not fit into imperial definitions of space. Many scholars have examined the diplomatic negotiations that led to
manchus, mongols, and russians 161

Nerchinsk, site of the treaty negotiations between Russia and China in 1689.
[To view this image, refer to
the print version of this title.]

these treaties, but most of them concentrate on the bilateral Russo-Chinese
relationship. The success of these frontier negotiations, however, depended
on the relationships among four parties—Russians, Mongols, Manchus,
and Jesuits—each of whom had separate interests. In this section I focus on
the importance of the Mongolian connection for both the Russians and the
Manchus. The two empires had gradually groped toward each other across
vast underpopulated spaces in the seventeenth century. Both sides came to
realize initially that they had a common interest in fixing the allegiances of
the tribal peoples on their borders, and, later, that they needed to prevent
each other from allying with the rival Mongol state to their west. The lure
manchus, mongols, and russians 163
Qing map of the region from Urga (Ulaan Baatar) to the Sino-Russian border at
Kiakhta. North is at the top, with Kiakhta at the extreme top center. The large tem-
ple on the center right is the residence of the Khutukhtu, high lama of Mongolia.
Walls, guardhouses, and a willow palisade guard the city on the west; the circles
with yellow labels are watchtowers(karun).
[To view this image, refer to
the print version of this title.]

of trade with the rich Chinese market drew the Russians to the Qing side,
while the Russians’ control of the northeast border attracted their Qing
counterparts. Mongol efforts to find allies proved unavailing.
As the Russian settlers moved east across Siberia, they soon became
aware of the riches of China. Siberia contained “an inexhaustible supply of
what had constituted Russia’s chief marketable commodity—her gold in
short—namely the fur-bearing animals, such as ermines, black and silver
foxes, beavers, otters, mink, but best of all, sables.”
78Furs provided a sig-
nificant amount of revenue to the Muscovite state, and Beijing offered the
best market for them. The Russians knew very little in detail about the Chi-
nese empire but saw it as a profitable trading opportunity from the begin-
ning. After an unsuccessful attempt to reach Beijing by caravan in 1608, for
example, they had learned that “[the Chinese] use firearms, and people
come from many lands to trade with them. And they wear golden robes,
and to him [the emperor] they bring all kinds of precious stones and other
things out of many countries.”
The diplomatic mission of Ivashko Petlin and Ondrushka Mundoff in
1618–19 was the “first European mission to [China] in modern times that
succeeded in reaching Peking and returning in safety,” but the Tsar learned
little from it. Because they had no tributary gifts to present, the envoys
could not get an audience with the emperor. (They had received a letter
from the Wanli emperor inviting them to present tribute, but no one was
able to translate it until 1675.)
80 The Russians knew the Great Wall well,
because they had traveled along it for ten days, but they had no access to
the court in Beijing. Russians had much more contact in this region with
Mongols, especially the Altyn Khan, than with either Manchus or Chinese,
and they had no real knowledge of the Manchus’ powerful state until they
ran directly into them on the Amur River in 1644. Russian Cossacks had begun to move into the Amur watershed in 1632,
when they founded Yakutsk, but Vasily Poyarkov’s expedition of 1643–
1646 was the first to provide detailed information about the resources of
the region. Poyarkov’s arrival alerted the Manchus to Russian expansion,
stimulating them to take defensive measures. When Erofei Pavlovich Kha-
barov returned on his second expedition in 1650, he defeated tribes who
were vassals of the Manchus, but he was surprised by the arrival of a
Manchu force in 1652. The Manchus, unaware that Khabarov represented
a serious attempt at colonization by the Russians, did not try to destroy the
Cossack force, but instead retreated. In 1654 and 1658, however, Manchu
troops decisively defeated the Cossacks and drove them out of the Amur
back to Nerchinsk. The Qing did not follow up this action, however, so
outlaws and Cossack bands filtered back eastward. In the 1660s a Polish
164 contending for power

exile and some Cossack criminals built a fortress at Albazin, which grew to
hold three hundred people.
Meanwhile, Russian embassies continued to pursue the prospects of
trade in Beijing. The embassy of Fedor Isakovich Baykov in 1653 had a di-
rectly commercial purpose, since the Russians were now desperate to de-
velop the fur trade. Because he refused to perform the ketou,Baykov’s gifts
were rejected, but he gathered important information. In 1658 and 1668
the embassies of Setkul Ablin, a Bukharan dispatched by Baykov to trade in
furs, demonstrated that selling furs to Beijing could be very profitable. In
1672 Ablin returned to Russia with an 18,700 ruble return on an invest-
ment of 4,500 rubles.
82Since Ablin was designated only a “messenger” of
Baykov and not an official envoy, he could avoid the restrictions of tribu-
tary ritual and concentrate on trade. By now fur had become the dominant
element in the Russo-Chinese trade, exchanged in Beijing mainly for silk
and other textiles. Russian hopes for profit rose. Conflicts over the allegiance of the tribes of the Amur valley, however,
blocked trade relations for the next twenty years. Gantimur, for example, a
Tungusic chieftain who left Manchu control to become a Russian subject,
exemplifies the opportunities available before the closing of the frontier.
The defection of Gantimur caused intense diplomatic conflict between the
two empires but ultimately led both to realize that they shared an interest in
establishing a well-defined border. Gantimur had fled from his homeland in
Siberia to avoid Russian demands for tribute in 1653. The Russian com-
mander Khabarov built the fortress of Nerchinsk in Gantimur’s territory in
1654. Gantimur received a mandarin rank and led troops against the Rus-
sians but then came back under Russian control in 1666–67. The Russians
aimed to attract local tribes to Albazin by offering military protection and
food supplies, while the Manchus ordered Tungus tribal leaders to move
away from the border, closer to Qing control. Gantimur and his powerful warriors were an attractive resource for both
sides. The Kangxi emperor demanded Gantimur’s return, fearing that oth-
ers would follow his example, but the Russian governor, in turn, demanded
that the Qing emperor declare himself a vassal of the Tsar. Such an outra-
geous demand was too shocking even to be translated from Russian. Fortu-
nately, in 1675, the Tsar in Moscow, determined to promote the fur trade,
sent a major embassy to Beijing, led by Nikolai Milescu Spathary, to negoti-
ate diplomatic relations and trade. Spathary’s embassy failed, primarily be-
cause he refused to send Gantimur back. The Qing feared that Gantimur
would aid Siberian Cossacks who were raiding their frontiers, while the
Russians feared that returning him would encourage other tribes to refuse
iasak tribute payments on which they relied to support their garrisons. Yet
manchus, mongols, and russians 165

both sides gained valuable information from this encounter. Kangxi real-
ized that raids on the border could be stopped by offering trade prospects
to the distant Tsar, and the Russians confirmed the power of the new dy-
nasty and the attractiveness of the China market.Both sides now agreed that clarifying the ambiguous border was crucial
to trade and security on the frontier. By the 1680s they were ready to nego-
tiate. Kangxi had repressed the Three Feudatories uprising in 1678 and
taken Taiwan in 1683. He sent two letters in May 1683 offering negotia-
tions if Albazin were evacuated. By the time the letters arrived in Moscow
in November 1685, Qing troops had destroyed the fortress, and would
besiege it again after Russian reoccupation in 1686. In response, the Tsar
dispatched Golovin from Moscow in January 1686 as plenipotentiary am-
bassador to negotiate a demarcated frontier and to establish commercial re-
lations. He arrived in Selenginsk in October 1687. Golovin first intended to
invite the Khutukhtu of Mongolia to act as mediator between Russia and
China, but then he received an invitation to write to the Kangxi emperor di-
rectly. Kangxi agreed with the Russian proposal to meet in Selenginsk in
1688, but the Qing officials called off the meeting when Galdan began his
attack on the Khalkhas in the same year. The primary Qing concern became preventing the Russians from sup-
porting Galdan, whose power was growing. The Russians likewise realized
that turmoil in Mongolia threatened a resolution of border conflicts with
China. Galdan had driven the Khalkhas into flight both north and south.
Some sought shelter on the Qing border; others moved north against Selen-
ginsk, first besieging Golovin in the fortress and then forcing him to flee to
Udinsk. By March 1689 he had defeated these Mongols and recovered
Selenginsk. The Qing officials now agreed to his suggestion to meet at
Nerchinsk farther east, freer from Galdan’s disruption. Envoys from both
sides arrived there in July 1689. Golovin, with his entourage of about one
thousand, met with seven ambassadors, led by Songgotu, and a supporting
cast including the two Jesuits Gerbillon and Pereira, military regiments,
and Buddhist clergy numbering at least ten thousand.
84The Qing represen-
tatives were all high Manchu officials, including Tong Guogang, the Chi-
nese uncle of the Kangxi emperor, and Sabsu and Langdan, who had led the
troops that demolished Albazin. No Mongol princes attended the meeting at Nerchinsk, but their hidden
presence affected two critical issues of the negotiations: the means of com-
munication and the delineation of the border. The delegates first had to
choose a language of discussion. Neither side could use its native language,
because preserving the illusion of equality was essential to success at the ne-
gotiating table. Both parties had open tents side by side, with equal num-
166 contending for power

bers of men. The Qing officials did have Russian translators available but
did not use them. As high-ranking Manchus, they excluded participation
by any Chinese in border negotiations. Tong Guogang had argued for the
presence of Chinese in 1688, but his proposal was rejected. The Russians
had tried to obtain Manchu translators from their Daur tributaries but
could not find any competent ones. Both the Russians and the Manchus,
however, were quite familiar with Mongolian, which had been the most
common language of communication among different peoples in this fron-
tier region. And yet the primary language of the Treaty of Nerchinsk be-
came not Mongolian but Latin, a language known to only one or two Rus-
sian representatives and the two Jesuits serving the Qing.Because the Jesuits had inserted themselves as crucial mediators, they
could decide the terms—and the language of communication. On the first
day of the meeting, August 22, the envoys agreed in principle to communi-
cate in Latin.
85According to Golovin’s report, they believed that there were
not enough Mongolian translators, and these were not reliable, so both
sides agreed that it would be more “objective” to rely on the Jesuits’ Latin.
The Pole Andrei Belobotskii acted as the Russians’ Latin translator. Discus-
sions, however, soon became hung up on the question of where to draw
the border. (See Map 5.) At first, the Manchus claimed all the territory up
to Lake Baikal, basing their claim on the fact that all the Mongolian tribes
of this region had paid tribute to the Yuan empire. The Russians held
out for preserving Albazin and Nerchinsk, suggesting that the border be
drawn along the Amur River. They heard the Manchus threaten them with
military strikes if they did not concede immediately. When they realized
that the Jesuits were “inserting words” in their translations, they asked
to communicate with the Qing envoys in Mongolian. After a long discus-
sion among themselves in Manchu, the Qing envoys said that they had
“only directed the Jesuits to speak of the border issue, and not of military
Each time discussions deadlocked, the Russians tried to communicate
directly with the Manchus, using Mongolian translators, but the Jesuits op-
posed them on the pretext that the translators were incompetent. The Jesu-
its also told the Russians not to speak to the Manchus in between negotia-
tion sessions, and they told their own interpreters and Manchu official
assistants (jargochi) never to speak to the Russians alone in Mongolian.
Mongolian could certainly have served as a bridging language just as easily
as Latin; by excluding it, the Jesuits put themselves in the position of get-
ting better terms for the propagation of their religion from both sides. They
enticed the Russians into promising favorable treatment from the Tsar by
pretending to be able to dissuade the Kangxi emperor from war, and they
manchus, mongols, and russians 167

obtained an Edict of Toleration from Kangxi in 1692 by taking the credit
for successful treaty negotiations. In the time-honored tradition of power-
ful mediators up to Henry Kissinger in our own day, they were determined
to exclude any communication channels outside themselves.
88 Monopo-
lizing the language and the access of each side to the other, they successfully
kept any Mongolian interests out of the negotiations. The conflict over where to draw the border line was ultimately settled by
the threat of force and by the ambiguous loyalty of the Mongols between
the two empires. On the second day, Golovin agreed to allow the border to
run along the Bystry, or Burei, River, a small tributary of the Argun, if the
168 contending for power
Map 5. The Sino-Russian frontier.

Qing would pay compensation for the destruction of Albazin. Songgotu
then proposed that the border should run along the Shilka River, following
its course into the Amur. This line would leave the Russian fortress of
Argunsk in Qing territory, along with a valuable salt lake and mines. Golo-
vin rejected Songgotu’s proposal to adjourn the conference until both sides
had submitted letters outlining their demands to their emperors; he sus-
pected that the Manchus would only use the time to move troops into the
frontier region. He also tried to win over the Jesuits to his side by promising
them favorable treatment for proselytizing in Siberia.Golovin then learned that the Qing were inducing at least two thousand
Buriat and Onggut Mongols near Nerchinsk, who had paid iasakto the
Russians, to desert to the Qing. Songgotu offered to set the border at the
Gorbitsa River, south of the Shilka, if the Russians would agree to de-
termine a border with the Khalkhas along the Selengge River. Since the
Khalkhas had not yet submitted to the Qing, Golovin rejected the authority
of the Qing to determine the borders. Songgotu then mobilized an army of
12,000 men, plus the Buriat and Onggut deserters, to surround Nerchinsk,
while Golovin with his 1,500 men prepared a last-ditch defense. Two days
later, knowing that his position was hopeless, Golovin gave in to most of
the Qing demands. Reasoning that there were few settlements and few furs
between the Gorbitsa River and Albazin, and his Mongolian tributaries
were deserting in increasing numbers, he abandoned his claims to Albazin
but retained access to the salt and mines north of the Argun. Under the final
terms of the treaty, the fortress of Argunsk would be moved to the north
side of the river, into Russian territory; the Qing would pay no compensa-
tion for Albazin but would allow traders access to the regions; and the bor-
der would be drawn north of the Amur River along the nearest mountain
range, determined by stone markers. A stele at the mouth of the Argun gave
the text of the treaty inscribed in Russian, Chinese, Manchu, Mongol, and
Latin. Other parts of the border would be delimited later. Golovin thus succumbed to an adroit use of military threats and entice-
ments by the Manchus. Losing his Mongolian tributaries would have cost
him nearly all control of the Transbaikal region. At the price of giving
up Albazin, he preserved access to lands north of the Argun, and he kept
control of the tributaries currently under Russian control. The Qing gave
up claims to land which it never controlled in the first place, and by offer-
ing trading access ensured that the Russians would not support Galdan.
Today, nationalist historians on either side argue that the other side got
the better deal. They reduce the treaty to a bipolar Sino-Russian confronta-
tion, neglecting the significance of the other two parties affected by the ne-
gotiations. The Russian and Qing officials calculated their positions not just in terms
manchus, mongols, and russians 169

of the effect on the other party but in light of the consequences for all par-
ticipants in the steppe power struggle. It was the Jesuits who clearly got the
most out of Nerchinsk, gaining great credit with Kangxi, who allowed
them free access for missionary activity. The Jesuits had already gained the
trust of the emperor by providing him with weapons, teaching him geome-
try, and informing him about the wider world. Now they showed their dip-
lomatic skills as well. In the end, however, they never succeeded in convert-
ing the Manchus or the Chinese to Christianity, and they lost influence
170 contending for power
Boundary markers between Russia and China determined by the Nerchinsk treaty.
A Russian Orthodox cross is atop the left marker; the Manchu writing on the right
says “border”(jecen i ba).
[To view this image, refer to
the print version of this title.]

when Kangxi left the throne. The Qing rulers had used them for their own
purposes and then cast them aside. The Mongols, both Galdan and the
frontier tribes, lost the most because the settlement locked them on one side
of the border and deprived them of the ability to find allies.Galdan discovered the effects of the Nerchinsk treaty immediately. In
early 1690 he sent an envoy to Golovin in Irkutsk seeking Russian military
support for his planned attack on the Khalkhas.
89Since the Russians them-
selves had been attacked by the Tüsiyetü Khan, Galdan expected an alli-
ance against their common enemy. Golovin replied that he had tried to con-
tact Galdan in 1688 to discuss an alliance, but his messengers had been
unable to get through. Now, however, he was unable to join forces with
manchus, mongols, and russians 171
A detail from the Jilin Jiuhetu(Map of Nine Rivers of Jilin), a large map of the Man-
churian–Russian border region used by Qing negotiators at Nerchinsk. All rivers
and villages are labeled in Manchu.
[To view this image, refer to
the print version of this title.]

Galdan. He did, however, send an envoy to inquire about Galdan’s troop
strength and to find out how many Russian traders were in Galdan’s terri-
tory. After signing the treaty with the Manchus, Golovin no longer had an
interest in a Zunghar alliance. His main concern was to protect Russians in
Galdan’s territory and to prevent the defection of Mongols who were pay-
ing tribute to the Tsar. Golovin’s envoy told Galdan to allow free trade at
Irkutsk and to send back the brother of the Altyn Khan, who had rejected
his tribute obligations to the Tsar and sought protection from Galdan.As soon as he heard of Galdan’s approach to the Russians, the Kangxi
emperor reminded the Russians that aiding an attack on the Khalkhas,
who were subjects of the Manchus, would violate the Nerchinsk treaty.
Nerchinsk now served as the model for defining obligations on the frontier,
and the Qing had defined the terms. Galdan was too late. He continued to
send embassies to the Russians in Siberia throughout the rest of the decade,
but they refused to allow him to appeal to Moscow. In the end the lure of
the China trade was much stronger than any prospects for gold exploration
in Zunghar territory. The Qing’s greatest gain from Nerchinsk was thus to
deprive the Zunghars of a potential ally. At Nerchinsk and Kiakhta the Qing signed their first treaties with a West-
ern power, the only ones for over two centuries negotiated on a basis of rel-
ative equality. Yet neither empire’s rulers believed in equal-status negotia-
tions between sovereign states. Both acted from hierarchical assumptions
of tribute, vassalage, and deference. How could these treaties be negotiated
under such contradictory understandings? Negotiations succeeded only be-
cause two other parties intervened as vital cultural intermediaries. The hid-
den presence of the Zunghar Mongol state induced both empires to adjust
conventional diplomatic rituals. The Jesuits and their Latin language were
the last survivals of the fluid intercultural communications of the vanishing
frontier, before formal state-to-state contact took over. In order to secure
their monopoly, however, the Jesuits had had to exclude the ability of the ri-
vals to communicate in Mongolian. Unlike McCartney’s experience in the late eighteenth century, the issue of
kowtowing to the emperor did not ultimately derail these negotiations. As
in the later negotiations with other European powers, the two sides had
very different goals. The Russians wanted trade, while the Chinese wanted
security. But unlike in the nineteenth-century “unequal treaties,” each side
gained what it needed without inflicting unacceptable costs on the other. Each side found the other both culturally familiar and alien. Both Chi-
nese and Russians had been accustomed for centuries to diplomacy with
Central Eurasians, but for the most part they had dealt with nomadic tribal
confederations that were militarily powerful but institutionally unstable.
172 contending for power

Now each confronted a large, established imperial rival. Norms of protocol
had to change. The Russians took a pragmatic view of frontier expansion:
they had moved into Siberia not for glory but for profit. Territory and im-
perial honor took second place to promotion of the fur trade. The Qing, for
its part, compromised on ritual propriety in order to secure Russian neu-
trality in the coming conflict with the Zunghars.Nationalism and contemporary politics have strongly influenced inter-
pretations of Nerchinsk and Kiakhta.
91 Until the Sino-Soviet split of the
1960s, Russian and Chinese historians interpreted the treaties as the roots
of the “fraternal alliance” of the 1950s, the only successful equal treaties
between China and the West. Since the 1960s, however, Russians have
viewed them as “unequal” treaties forced on a weakened Russian empire
by aggressive Manchu expansion. Chinese historians regard the Russians as
treacherous imperialists who signed the treaties but continued to give sig-
nificant aid to the Mongol state. Most recently in China, nationalists have
claimed that the treaties themselves were “unequal” to China’s disadvan-
tage, because China gave up claims to large parts of eastern Siberia pur-
portedly occupied by “Chinese” (i.e., Tungusic) peoples. Both sides con-
struct the other as ineradicably “aggressive” in order to shore up their own
insecure national communities. Like the military campaigns, diplomatic
history too has become a tool of nationalist ideology.
manchus, mongols, and russians 173

Eating Snow: The End
of Galdan, 1690–1697
D uring the six years between his first and second campaigns, the
emperor shored up his defenses and planned to isolate Galdan from poten-
tial allies. Galdan, for his part, aimed to win back the Khalkhas who had
surrendered to Kangxi, recover his strength far from the border, and en-
sure support from the Dalai Lama. Tibet became a primary focus of ri-
valry. The emperor tried to cut off communication routes between Galdan
and Tibet running through Hami and Xining. Galdan, in turn, tried to cut
contact between Kangxi and Tsewang Rabdan, because the emperor aimed
to use Galdan’s nephew against him at his rear. Both rulers publicly pro-
fessed their dedication to peaceful relations while each plotted against the
other. Kangxi began to recognize the severe supply problems that prevented
him from pursuing Galdan to his lair. He had vowed to “exterminate Gal-
dan root and branch,” but he could not reach him in distant Khobdo. He
could only act defensively, until either his logistical foundations were more
secure or Galdan could be lured into a closer attack. The first steps to-
ward building the giant logistical network that culminated in Qianlong’s
campaigns of the 1750s began during this period. Likewise, the “fatal in-
dividualism,” or divisions between the Zunghar and Khalkha Mongols,
and within the Zunghars themselves, prevented any joint action by them
against the Qing regime.
1These divisions, arising in the 1690s, ultimately
caused the destruction of the Zunghar people in the mid-eighteenth

The Dolon Nor Assembly
Although angry at Galdan’s escape, the emperor realized that Galdan’s ab-
sence offered him an opportunity to extend his influence over all of the
Khalkha tribes. Soon after his victory at Ulan Butong, he made plans to es-
tablish “order and discipline” among the Khalkhas.
2He convened a great
meeting of the Khalkha Khans at which he would organize them into ban-
ners like their fellow Chechen Mongols in the Forty-nine Banners, and set-
tle them permanently in designated territories. The emperor once again
faced opposition to a dangerous trip beyond the wall.
3Censor Shen Kai-
zheng urged him to delay the meeting because of bad weather and risks to
his health, but the emperor insisted that these major issues could be settled
only with his personal presence. Leaving Beijing on May 9, 1691, he first
conducted a great hunt, laid out as a military campaign, then set out for
Dolon Nor. Kangxi knew that the primary cause of disorder in Khalkha Mongo-
lia was the internecine warfare among the Khans, stemming from deeply
rooted personal feuds. These rival parties had drawn Galdan and the Qing
into intervening in Khalkha affairs, but the Qing’s decisive military vic-
tory gave it the authority to settle these disputes permanently according to
Manchu rules. Kangxi established a clear order of precedence among the Khans before
the meeting, placing the Jebzongdanba Khutukhtu, the Tüsiyetü Khan, the
Jasaktu Khan’s younger brother Tsewang Jabu, and the Chechen Khan in
the first rank, and ordering the other nobility in seven declining ranks. Over
550 members of the Mongolian nobility were assigned to specific ranks.
Each of the Khans performed specified rituals, including three kneelings
and nine prostrations, and each was assigned a seat at the great banquet. The meeting was held from May 29 to June 3, 1691, at Dolon Nor, a
small settlement on the edge of the steppe 250 kilometers north of Beijing.
The emperor received the Khalkha Khans at a great banquet, followed by
military parades to impress them with the empire’s might. The firing of can-
non and the display of firearms caused them to “tremble with fear and ad-
4A total of sixty-four small cannon, eight large cannon, and
eight mortars were placed in the visiting Mongol camps. The emperor him-
self, armed, on horseback, led the demonstration of seventy pieces of ar-
tillery. He asked Gerbillon if European kings also made great voyages,
and complained that the Manchus were receiving inferior guns from Jesuit
At the meeting, the emperor openly declared that the Tüsiyetü Khan and
eating snow 175

Khutukhtu had committed crimes. Galdan’s invasion had been brought on
by the Tüsiyetü Khan himself, leading to destruction of his state and the
loss of his family. Out of his own benevolence, the emperor had rescued the
Khan’s people. He pardoned them, as they begged his forgiveness. The
Jasaktu Khan, by contrast, deserved pity, as he had been harmed for no rea-
son. Both the Tüsiyetü Khan and the Jebzongdanba Khutukhtu were given
fiefs and Manchu noble titles, and all swore to maintain peace. Refugees
would be returned to their homes.In the emperor’s view, the Khalkhas were a “disorderly” people in need
of “discipline” (fadu).Enrolling them in banners kept their territories dis-
tinctly separated, avoiding pastureland conflicts. Each Khan kept his title;
the younger brother of the murdered Jasaktu Khan succeeded to his posi-
tion, with a title approved by the Qing. Qing officials took over the final
authority for granting titles of leadership among the Khalkhas.
The Khalkhas gained materially from their submission, as the Qing pro-
vided them with food and animals to relieve their suffering. They also
gained new titles, confirming their authority. But in return, they gave up
their right to move at will. Enrolling in banner organizations meant that
Manchu officials strictly supervised their movements between pasture-
lands. Not all Khalkha leaders accepted the terms of this arrangement.
Batur Erke Jinong, though he had joined the Qing at Dolon Nor, rejected
such controls, insisting on his right to change pastures. The Dolon Nor assembly also provided the Qing with greater authority
over the Mongols in its rivalry with the Dalai Lama. An envoy was sent to
Tibet to inform the Dalai Lama of the emperor’s success in bringing peace
to the Khalkhas.
7At the same time, he warned that Galdan, whose follow-
ers were starving, might turn to the Dalai Lama for aid. The emperor
vowed to exterminate (jiaomie)Galdan if he violated his oath not to attack
the Khalkhas. Isolating Galdan from Tibetan support, however, required
both threats and material incentives. Trade restrictions had been imposed
at the border town of Dajianlu (T. Dar-rTse-mDo), now in Sichuan, which
divided Qing and Tibetan realms, during the time of conflict. These were
lifted now that peace was restored. Although both armies faced severe shortages, Galdan’s forces, mean-
while, were in a state of much greater exhaustion, and, unlike the Man-
chus, Galdan had no secure refuge. First he fled north, intending to replen-
ish his dying herds by capturing animals from Khalkha tribes. When these
efforts failed, he had to continue on foot.
8Galdan then headed west toward
the Ordos region. Qing troops were not able to block his movements be-
yond the Yellow River because they too lacked supplies, but they could es-
tablish defensive positions to keep him out of the Ordos. Galdan retreated
176 contending for power

to Khobdo, far away in Mongolia beyond the reach of Qing troops. Pro-
posals for a major expedition there against Galdan had to be rejected be-
cause of his inaccessibility.
9His forces could threaten Hami, an important
supply point with a sympathetic Muslim population, which was too far out
the Gansu corridor for Qing forces to garrison it in strength, but as long as
the Qing kept troops in Ganzhou, Galdan did not dare to attack. The Qing’s greatest fear was that Galdan would be able to rebuild his
strength from the resources of the pasturelands of Mongolia combined
with grain production of the Central Asian oasis populations. After his loss
in battle, with nearly all his cattle and sheep gone, his men were forced to
cultivate fields and even catch fish to survive, but if the Zunghars could es-
tablish independent sources of grain, they could revive their formidable
strength. Qing officials therefore proposed confiscating grain stores from
the Muslims in Hami and holding them at the military garrison in Jiayuguan. Despite their anxieties, there was little they could do to mount another
major campaign immediately. Most of their concerns during this period
were about reducing the size of frontier garrisons in order to economize on
10Repeatedly, efforts to move small units beyond interior forts ran
up against limited supplies of grain and horses. In addition, the new Mon-
gol allies, many of them desperate refugees from the battlefields, required
relief grain.
11 Food was a useful weapon in the contest for control of the
Mongols, but even the huge Chinese agrarian economy could deliver only
limited amounts to the frontiers. The Qing made further efforts to isolate and encircle Galdan by contact-
ing his kinsman and enemy Tsewang Rabdan, whose rebellion in 1690 had
almost forced Galdan to give up his intervention in Khalkha Mongolia.
Qing envoys, hoping to exploit the division between them, sent gifts to
Tsewang Rabdan in 1691. Telling Galdan, “Your animals are all gone, you
have nothing to eat. Men are dying in extreme want,” the emperor offered
to make peace between him and Tsewang Rabdan if he would submit. This
strategy aimed to reproduce the success of the Qing as peacemaker among
the Khalkhas, but Galdan rejected it. In Hami, one of Galdan’s subordi-
nates killed the Qing envoy Madi, sent to Tsewang Rabdan.
12 Although
Galdan denied responsibility, the emperor blamed him directly, combining
threats and incentives to induce his surrender. Tsewang Rabdan had sent a
secret memorial to the emperor, along with tribute gifts, and maintained
contact after Madi’s death. The Qing policy had apparently succeeded in
detaching the Khalkhas, gaining the Dalai Lama’s and Russian neutrality,
and splitting the Zunghar homeland. Yet Qing officials had to give up the
idea of an all-out advance to Khobdo in search of Madi’s killers. Galdan
was still out of reach.
eating snow 177

Galdan himself, aiming to win over allies wherever he could find them,
approached the Russians, and tried to enroll the Khorchin prince Biliketu
on his side.
13But the Qing captured Galdan’s letter to the prince, only lift-
ing the cloud of suspicion from him after an investigation, and won him
back to Qing submission. He presented gifts to the emperor on his birth-
14 The Russians, as noted earlier, rejected Galdan’s feelers in order to
preserve the terms of the Treaty of Nerchinsk. Relations with Tibet became the most critical contest. Until the battle of
Ulan Butong, the emperor had treated the Dalai Lama and his envoys with
great respect, constantly stressing their shared interest in peace among the
Khalkhas. After the submission of the Khalkhas at Dolon Nor, the edicts
from Beijing sounded the same tone but added a note of menace: the Dalai
Lama must not respond to Galdan’s appeals for material aid, or else the
Qing would cut off trade relations with Tibet. The emperor’s suspicions of
collaboration between Galdan and Tibet increased; in November 1691,
after receiving an appeal for peace from the Dalai Lama, he responded
forcefully: “You lie when you claim to be advising peace. You are now plot-
ting with Jilong Khutukhtu, who is part of Galdan’s camp. He is not advis-
ing peace. Galdan is now sneaking into our borders, plundering the
Uzhumuqin area. You lamas are not passing on my edicts. You are greedy
for profit, deceitful, and are concealing Galdan’s activities.” Kangxi as-
sumed collusion between Galdan and the Jilong Khutukhtu, the envoy of
the Dalai Lama whose proposal for negotiations after the battle had given
Galdan time to escape.
By 1692 Galdan had reduced his demands. Instead of three requests,
he had only one, supported by the Dalai Lama: the return of the seven
Khalkhas to their original lands. This, of course, would have meant releas-
ing them from the bonds of the banner system and making them vulnerable
to Galdan’s pressure. The emperor had to reject this request, but he also
strongly warned the Dalai Lama not to persist with any further support of
Galdan’s independence. He had not revealed to the Tibetan lamas his goal
of exterminating Galdan, and he continued to stress their common interest
in benevolence, as demonstrated by his feeding of the starving Khalkhas.
But the message to isolate Galdan became sterner. In December 1693 the sDe-parevealed that he, and not the Dalai Lama,
had been managing affairs in Tibet. He agreed with Kangxi’s refusal of
Galdan’s three requests, and at the same time asked for an official seal from
the Qing, which was granted. At the same time, he asked the emperor not
to deprive Galdan and Tsewang Rabdan of their titles as Khan. This re-
quest was refused, with the response that an “outer barbarian” (waifan)
had no right to determine decisions of the Chinese emperor.
178 contending for power

Strengthened by these successful rebuffs of the Dalai Lama’s mediation,
the emperor insisted on Galdan’s surrender. Galdan tenaciously offered to
apologize for his rude language but asked for an imperial grant of 50,000
to 60,000 taels. The emperor insisted that only a personal audience would
make it possible for Galdan to submit and receive imperial favor. He had
no expectations that Galdan would accept such an audience, but this was
part of his strategy to lure Galdan closer to Beijing so that he could do bat-
tle with him. At the same time, the emperor grew increasingly suspicious that Galdan
was gaining allies among the Muslims of Xinjiang. Because Galdan used
Muslim messengers whenever he sent envoys to the Qing, the emperor ac-
cused Galdan of sending Muslim spies into China. Of course, the Muslim
oasis inhabitants of Hami and Turfan, long familiar with both nomadic
and Chinese trade, made natural intermediaries, but the Kangxi emperor
wanted to remove middlemen between the Qing and Mongols. By 1695 he
had even apparently become convinced that Galdan himself had converted
to Islam.
17 Perhaps his curious delusion is explicable only by the convic-
eating snow 179
Cannon cast in 1690, with Manchu and Chinese inscriptions, for use in the Galdan
campaigns. The cannon was named “General who exerts power over long dis-
tances.” Because of its small size (about 1 meter long) and wheeled carriage, it could
be transported on long campaigns.
[To view this image, refer to
the print version of this title.]

tion of the Qing emperor that he had coopted all the other peoples in Cen-
tral Asia, winning them over to submission or at least non-intervention. If
Galdan appeared to reject the Dalai Lama, the Mongols, and the Qing, he
must belong to the only group as yet beyond Qing control: the Muslims of
Central Eurasia.Reports that Galdan was moving east out of Khobdo stimulated a new
Qing mobilization. One hundred camels with special food supplies were
bought to carry heavy cannon.
18The emperor expected Galdan to attempt
to move south to Tibet via Kokonor. To block this move he ordered prepa-
rations for a scorched earth campaign to burn all the grasslands along the
Ejina River, north of Kokonor.
19But the campaign to destroy Galdan once
and for all depended on drawing him into another battle in Mongolia. In September 1695 the emperor outlined a definitive plan to entice Gal-
dan into battle. The Khorchin prince Biliketu, formerly suspected of collud-
ing with Galdan, had captured some of Galdan’s abandoned documents,
including an invitation to a meeting. He would send an envoy to Galdan,
telling him that ten Khorchin banners wanted to submit to him and inviting
him to advance east. Kangxi vowed to “personally lead a large army thun-
dering after him, so that he cannot escape. We will definitely exterminate
20He knew by now that Galdan was not preparing any new attacks
on the empire, but he had determined to eliminate his tenacious rival. He
prepared to set out on his second personal expedition on April 1, 1696.
The Battle of Jao Modo
Discussions of strategy for the second campaign focused first on determin-
ing Galdan’s intentions. The crucial issue was logistics: supplies of horses,
cannon, and grain. Unless Galdan could be lured in closer, they all had to
be hauled thousands of miles from northern and northwest China to Mon-
golia. The emperor’s greatest fear was that Galdan would once again es-
cape, as he had at the battle of Ulan Butong. In August 1695 Galdan seemed to be moving out of Khobdo toward the
Qing frontier, but by the next month it became clear that Galdan would not
march closer. He remained in the Kerulen–Tula River area to wait out the
winter snows. Efforts to entice him nearer would not succeed.
21 Spies re-
ported large concentrations of troops along the Kerulen River, where “the
ground was all trampled,” estimating Galdan’s force at five thousand to
six thousand troops.
22 Although he had few sheep, he now had abundant
horses and camels. Even though Galdan’s lack of aggressive intent would
make it harder for the Qing to reach him, it allowed time for the Qing to
supply a huge army.
180 contending for power

Galdan himself, in his memorials to the emperor, indicated no awareness
of the plans to exterminate him. He still offered to resolve outstanding dis-
putes, with a view to establishing a clear boundary line between the two
Cautious, “cowardly” officials in the court, however, were not enthusias-
tic about a plan to march long distances through the desert. Many advised
waiting for Galdan to approach more closely. Heading all the way out to
the Kerulen River immediately, and marching through the winter, as the
emperor wanted, risked arriving before the spring grasses had sprouted.
They knew well that the lack of grass for horses would be a major limit on
the mobility of the army.
24But the emperor’s intense resentment of Galdan’s
escape at Ulan Butong led him to insist on immediate preparations. Even
after mobilization began, there was continual opposition to the emperor’s
personal participation in the campaign. Councilors urged the emperor not
to risk the health of his “jade body” but instead to rely on the valiant ef-
forts of his troops. They were, of course, concerned about the great costs of
the campaign and the risk of instability at home while the emperor was
away. At this early point, the emperor did not yet become angry at his
Han officials’ “reluctance to tire themselves with military affairs.” He only
noted that because he had been too ill to be present at the battle of Ulan
Butong, the crafty Galdan had escaped, so he insisted that his personal
presence was required. His eldest son could handle affairs at home.
Fiyanggû, commander of the West Route Army, was the only enthusiastic
supporter of the campaign, but even he persuaded the emperor that it
would be best not to march in winter but rather to leave the following
spring. The Qing forces would have time to fatten up their horses during
the winter, move rapidly, and catch Galdan in early spring, just before his
horses could be nourished on new spring grass.
26Three armies set out: the
West Route Army, led by Fiyanggû, with 30,000 men; the East Route Army,
led by Sabsu, with 10,000 men; and the Capital (Central) Army, led by
the emperor, with 32,970 men. (See Map 6.) Sabsu’s army stopped at the
Khalkha River to block movement of the Mongols to the east, so only
two armies took part in direct pursuit. Another army, led by Sunsike, with
10,000 men, would set out from Ningxia to join Fiyanggû. Fiyanggû’s role
was to block Galdan’s escape route, so he had to arrive at the Tula River
before the Central Army reached Galdan’s camp at the Kerulen River. It
was 2,000 li(1,160 km) from Guihua, where the West Army set out, to
Galdan’s camp at Bayan Ulan, and an additional 1,000 li(580 km) from
the capital to Guihua. The West Route Army would leave around March
22, carrying eighty days’ supplies, with fifty days’ supplies sent to follow.
Zhili, Shandong, and Henan would together provide 1,333 carts, each car-
rying 6 shiof grain. Governors of each province would provide carters
eating snow 181

and escort troops. All of North China’s provinces bore the burden of the
campaign, but the man chiefly responsible for grain supplies was Yu
Chenglong, Governor-General of Zhili. He had six thousand carts built
to carry supplies to the frontier. Extra allocations from the center were
given to each provincial governor to spare some of the burden on the local
182 contending for power
Map 6. The Kangxi emperor’s Zunghar campaigns, 1690–1697.

Grain purchases were vital to the campaign’s success. Men who worked
in grain supply would receive equal military honors with soldiers who
fought in battle.
29 But the poor northwestern provinces would have great
difficulty supporting the army’s needs. They grew only one crop per year,
and they suffered frequent droughts. Distributing rations to troops and of-
ficials in the capital before they left would relieve the strain on local grain
markets, but it was too much of a burden on the men to carry all their sup-
plies with them. The Shaanxi army needed 22,400 shifor five months,
much more than it could carry. Substituting cattle for one month’s rations,
and having soldiers themselves carry 0.05 shiwith them, reduced the to-
tal somewhat, but they would still need to carry silver to buy supplies in
Gansu markets. Every man was allotted silver taels, ranging from 20 taels
per month for generals to 2 taels for soldiers.
30 On the march, merchants
followed along behind the army, trading in separate encampments. Since
many of the troops, who came from the northwest, knew local markets
well, General Yin Huaxing gave them money to buy their own grain.
Horses, the second vital element, could not be obtained from the interior.
For them, the Qing depended almost entirely on its Mongol allies. As early
as August 1695, officials went to the Khalkhas to purchase horses. They
bought one thousand at the major horse markets in Guihua and two thou-
sand from each of the six Ordos and ten Khorchin banners; other tribes
supplied smaller numbers. Even feeding the horses in the capital over the
winter proved to be difficult. Fengtian prefecture, north of Beijing, made its
contribution to the campaign by providing 300,000 bundles of grass for
eight thousand horses.
But the Mongols made for unreliable allies. Severe discipline had to be
enforced against both Mongols and Chinese troops who sold horses to
them. The Khalkhas were “expert at stealing horses.” Two Khalkhas dis-
covered trying to steal horses were sentenced to have their arms and legs
broken and their ears and neck cut.
33Yet Mongol troops were essential to
the campaign, especially crack troops for rapid pursuit. Each chief had to
supply troops, but they had to be the strongest, most experienced men, par-
ticularly from the prosperous elite families.
34 It was equally important to
avoid unrest among the Mongols, especially those who might fear that the
army passing through their territory intended to punish them as well as
Galdan. Here Tsewang Rabdan, the Zunghar leader who had split with
Galdan, became a useful Qing ally. He could reassure other Mongols that
Galdan was the only target, and thereby undercut Galdan’s efforts to rally
all of the Mongols against the Manchus and Chinese.
35 Tsewang Rabdan
was allowed to increase the size of his tribute missions from two hundred
to three hundred people, thus increasing his profits from trade, along with
eating snow 183

gifts of brocades, silver, tea, and furs; and the emperor promised to send
to him any Muslim merchants captured in Galdan’s territory during the
Gunpowder weaponry was the third crucial element. Addressing his
troops, the emperor allocated cannon to each banner to ensure total Qing
victory, saying, “Nothing is fiercer [meng] than [gunpowder weapons],
they are vital weapons for the army. Gunpowder is the key to exterminating
37 Except in the Three Feudatories rebellion, where the enemy
used gunpowder but did not prevail, imperial armies had won nearly all
their battles without using cannon. This time, by contrast, the army would
be well provided. At the great meeting at Dolon Nor, where “the firing of
cannon thundered through the mountains and valleys,” cannon fire had im-
pressed the Mongols extremely. The army carried at least 235 large cannon
weighing from 8,000 to 10,000 jin(5,000 kg), and 104 lighter cannon
weighing 100 to 800 jin.There were many varieties, including “Western
[Xiyang] bronze cannon” and “Taiwan light cannon.” In a military review,
when all the banners’ cannon fired in succession, deserters from Galdan’s
army viewing the display were “greatly astonished” and predicted that
“Galdan would be destroyed in a day.”
38 Beyond Guihua, the cannon
would have to be transported on camelback. The cannon taken to Ulan
Butong were too heavy, so lighter cannon were used, but transporting this
artillery over such long distances considerably slowed down the army. It
nearly caused Fiyanggû to miss his scheduled rendezvous to prevent Gal-
dan’s flight. The emperor set the departure date of the Central Army from Beijing at
March 26, 1696, between 3 and 5 am. The West Army left from Guihua on
March 20 between 9 and 11 am. Sun Sike gathered his ten thousand troops
at Ningxia and left from there on March 24.
39 The Central Army could
march in four stages to Dushikou 150 kilometers northwest of Beijing
without extra horses; beyond Dushikou, they could average up to 60 liper
stage, a total of approximately sixty stages. For each stage, a relay station
provided forty relief horses. Along the march, strict discipline was imposed. Troops had to rise be-
tween 3 and 5 am; they were not allowed to light fires to cook breakfast in
order that they might break camp early. Negligent officers who failed to get
the soldiers moving early were punished. Water, grain, grass, and horses
were the primary considerations. This early in the spring, most ponds and
springs were frozen over. Wells had to be dug by chopping through ice at
each stage of the march. At one spot soldiers dug seventy-five wells and cre-
ated forty-five ponds. It was seen as a very good omen when water sud-
denly bubbled up from a frozen spring.
40 Horses could easily die from
184 contending for power

overexertion. Sweating heavily on the march and suddenly stopping for rest
could exhaust them; then they had to be galloped to warm them up. They
could not be given water until their sweat had dried. The troops had to be
careful to avoid poisoned wells and grasses.
Thirteen hundred carts followed behind the army, the supply carts
bunched close together to prevent attacks. The emperor, his generals, and
troops, who marched ahead of the grain carts, were very relieved when the
supplies arrived. Yu Chenglong had continual problems with his supply
train and soon fell behind schedule. Rain and mud slowed progress, and
many cattle died. Soldiers had to build roads out of willow sticks and mud
in order to cross giant sand dunes. As the carts moved out into the desert,
they left behind part of their supplies at magazine posts to provide for the
return journey.
Meanwhile, the emperor urged his army on. Accompanied by the Jesuits
Gerbillon and Pereira, he took regular sightings of the polestar to deter-
mine his latitude.
43He prayed to the spirits of wind and rain to provide a
smooth journey. Six of Kangxi’s sons accompanied him on the campaign,
but his eldest son, Prince Yinreng, remained in the capital to take charge of
affairs in the emperor’s absence. Kangxi’s letters to his son describe the
landscape, the progress of the campaign, and his personal health. These
Manchu letters give fascinating insights into Kangxi’s character. They form
one of the most remarkable travel documents in Chinese literature, provid-
ing an almost daily record of the emperor’s changing moods. As new infor-
mation or supplies arrive, or fail to arrive, the emperor’s confidence waxes
and wanes. He also describes in detail the terrain through which the troops
pass, paying attention to different types of grasses, the abundance of mar-
mot burrows, water sources, and the different kinds of deserts. He sends
back plants for the prince to grow. When stuck in a place with “nothing but
sand and rocks,” he still collects colored stones in a chest to be sent as a gift
to his family in Beijing.
44These letters, which vividly depict the hardships
and delights of the campaign, are analyzed further in a later chapter. A severe snowstorm struck the expedition soon after it set out, but for
the most part the emperor found the environment he passed through quite
favorable. As the weather warmed, supplies of grass and water became
more easily available, and his horses stayed healthy. At one point, however,
when no water or grass was found within a radius of 40 to 50 li,the em-
peror wondered if Heaven had abandoned him. Then, late at night, a spring
was found on the top of a mountain, just enough for a one-night stopover.
On May 14 he passed a stone marker left by the Yongle emperor during his
expeditions in the early fifteenth century.
46Kangxi worried about constant
rain and snow; but the Mongols rejoiced. They praised him for bringing
eating snow 185

good grass in the fourth month, the early spring, when sheep and horses
were critically weak after the long winter.
Not everyone shared Kangxi’s confidence. When it was reported that
Galdan had sixty thousand Russian troops supporting him, councilors
Songgotu, Yisanga, and Tong Guowei urged the emperor to return to the
48This time it was not civilian Han officials but two of the highest-
ranking Manchu officers, his close advisers, who foresaw calamity. Furi-
ous, the emperor said: “I have made careful plans, and made offerings to
Heaven, Earth, and the ancestors...Among the soldiers there is no one,
even down to the stable boy, who does not want to exterminate Galdan.
But you high officials are base womanly cowards who fear putting out
effort...Iwill certainly kill anyone who hesitates, or withdraws from
this campaign.” After a dramatic confrontation, they knelt before him,
begged his forgiveness, and agreed on the correctness of his plan. Galdan’s
clever disinformation had nearly halted the expedition. As the officials later
learned, Galdan had in fact sought Russian aid and discussed with twenty
Russian envoys obtaining at least one thousand troops and cannon, but the
Russians had made no definite commitment.
The Manchu advisers’ primary concern, however, was not the Central
Army but Fiyanggû’s West Army. Galdan did not yet know that the Qing
were leading a huge army against him, but he was expected to flee west as
soon as he learned of it. If Fiyanggû could not reach his designated battle
station before then, Galdan would have escaped again, and the campaign
would fail. Early reports from Fiyanggû were encouraging. He reached the
Qing frontier posts on April 14, and expected to reach Wengjin (Onggin)
on May 3, Tula on May 24, and Bayan Ulan, Galdan’s camp, on May 27.
To keep up his speed, he did not wait for his supply carts to catch up.
then, after a month with no word, Fiyanggû reported that snow and mud
had bogged him down. Unable to move his cannon, he had left most of
them behind at the frontier, although he had been able to carry fifty-nine
cannon farther on by camel. He could bring only fifteen days’ worth of ra-
tions to Wengjin, and twenty days’ to Tula; but by advancing quickly, he
would meet with Sunsike on May 30. He now hoped to reach Tula on June
2 and Bayan Ulan on June 6.
Hearing this, the Central Army faced a great dilemma: to advance or to
wait? Waiting would use up precious rations; advancing too quickly would
drive Galdan into flight before Fiyanggû could block him. A military coun-
cil discussed the options.
52 Those who felt that Galdan would stand and
fight urged a quick advance, but many feared that he would escape. It
would still take nine days to reach Galdan’s camp, giving him plenty of time
to move far away. The army waited for several days, until May 23, risking
186 contending for power

the chance that Galdan would soon hear of the army’s presence. Then the
grain carts arrived, 300 camels and 173 wagons, with 1,000shiof grain
(300 tons): supplies were “heaped up like mountains,” astonishing the
Khalkha troops with their abundance. But the army had already consumed
sixty days’ worth of its total rations and needed to send for more. Yu
Chenglong still could not bring up all his carts. The troops were now
camped in an area with “nothing but sand and rocks.”
Reports, which later appeared to have been exaggerated, claimed that
Galdan had ten thousand soldiers, ten thousand armed servants, and seven
thousand vassals, with plenty of provisions and animals.
54 Fiyanggû was
advancing toward the Kerulen with ten thousand men, but the exhaustion
of his horses forced him to leave many men behind. The army of Sunsike
was reduced to two thousand Chinese troops, most too exhausted to
march. General Yin reported: “We rushed here without time to nourish our
horses, so many died crossing the Gobi. Our water is all gone, and we con-
stantly faced strong winds and rain for days at a time. The troops were cold
all night, and starving men and horses collapsed.”
55 By the time they ar-
rived at Onggin, they had lost nearly all their horses in the cold and had
supplies for only one month. Sending back all but the strongest men in or-
der to conserve supplies, Sunsike marched on with Fiyanggû. On May 26 the emperor decided to follow a third plan: to send envoys to
Galdan offering negotiations. The envoys took with them several Zunghar
captives, who would be released to join Galdan’s forces. By stalling Galdan
for a while with communications, the Qing army would gain time for
Fiyanggû to block his retreat.
56Once again the emperor used disingenuous
language, asserting that his sole interest was peace on the border, that
he had no intention of exterminating Galdan. He had saved from starva-
tion the Khalkhas who had fled Galdan’s attacks; now he wanted to meet
Galdan personally to form an alliance. “Let us meet and determine our
boundaries [dijie] and resume our former tributary relations ...Iamnot
trying to entice you into destruction.”
Imperial envoys told Galdan’s nephew Danjila that the emperor was
coming in person to negotiate with Galdan. When they informed him
(falsely) that Fiyanggû had reached the Tula River and that escape was im-
possible, Danjila gave a “cry of anguish.” The next day, all the imperial
troops were drawn up in ranks, “filling the hills and fields without end,”
their “weapons glistening in the sunlight.”
58The emperor was certain that
the sight of this massive force would break Galdan’s spirit. By now, he
expected Fiyanggû to have arrived at the Tula River, so he did not fear
Galdan’s escape, even though he hoped for a direct confrontation. On June
7 he reached the Kerulen.
59The river, flowing through steep hills, had very
eating snow 187

little water. Zunghar patrols watched the troops advance but did not resist.
There was no sign of Galdan’s troops, although traces of his encampment
were found. Clearly, he had left only a short time before. Denouncing
Galdan as a coward, the emperor vowed to continue his pursuit. Leaving
behind all but sixteen cannon, he sent envoys ahead to urge Galdan to sur-
render. Captured soldiers reported that Galdan had fled into the forests at
Bayan Ulan. Had Fiyanggû got there before him?At the Kerulen, the Qing troops found exhausted horses, abandoned
by Galdan, and severe drought conditions. The grass had not sprouted.
Old people left behind described the Zunghar flight as a “panic.” Parts of
Galdan’s army were attacking one another; women and children were com-
mitting suicide. And word came that Fiyanggû had reached the Tula River
as planned on June 6. These were auspicious signs, but the Qing troops
now faced a severe grain and fodder shortage. They had used up the rations
they had brought with them, and Yu Chenglong’s reserve supplies had not
yet arrived. The emperor himself was eating nothing but mutton. Reluc-
tantly, he realized that he would have to turn back with the main army to
find his grain carts. Sending Maska ahead with a small detachment of cav-
alry and light artillery to continue pursuit, he began his “victory march”
home on June 12. Although he expected success, he knew that the West
Route Army also lacked grain supplies. He revealed his depressed mood in
a letter to the prince, describing a landscape “with no good places for thou-
sands of miles,” asking for clothes to be sent from home, and expressing his
longing for his son.
Fiyanggû’s long-awaited report arrived the next day. Fiyanggû had
known that Galdan was at the Kerulen, but his troops were too weak to ad-
vance rapidly. Their rations would run out between June 3 and June 10. Yu
Chenglong could drag his carts over the sandy hills at a rate of only 20
to 30 liper day, and they could not move against a strong wind. On May
31 Galdan was ten days ahead of them, and he had burned all the grass
for miles around. Galdan had gone to the Kerulen expecting to find Rus-
sian musketeers and large cannon. By June 2 Fiyanggû was able to block
Galdan’s escape route with fourteen thousand men. On June 12 the two
armies met in battle at a “terrible place” in the midst of the desert, called
Jao Modo, from the Mongolian “Jaghun Modu,” meaning “one hundred
trees.” It was a small valley with a river at the bottom, surrounded by
61Fiyanggû’s men, unable to carry provisions, had marched for eleven
days, living off horseflesh and camel meat just like nomads. Galdan had
only five thousand men armed with two thousand fowling pieces. General
Yin argued strongly for occupying the hills, even though the sun was set-
ting. To climb the hills, his men had to fight fiercely against Mongol sharp-
188 contending for power

shooters. When the army finally camped above Galdan on the hill, they had
gained the strategic advantage. The Manchu troops fired their great cannon
and advanced behind a wooden barricade, protecting their bodies with
padded cotton armor. When they were ten steps away from the enemy,
“arrows fell like rain.” Even the news that the emperor was approaching
had frightened many Mongol soldiers into abandoning their weapons and
fleeing. Galdan was unable to control his troops, who broke ranks and fled.
His kinsman Arabdan attempted to resist, but then the Manchu cavalry at-
tacked, killing thousands of men and capturing over twenty thousand cattle
and forty thousand sheep. Galdan and Danjila escaped with only forty to
fifty men.
The emperor received Fiyanggû’s victory report before he crossed the
Great Wall on July 3. He entered Beijing four days later. Kangxi had gained
a great victory, justifying his iron determination to advance into the steppe.
In an expedition of ninety-eight days, traveling over two thousand kilome-
ters to the Kerulen and back, he had “extinguished” Galdan’s flame. He re-
turned to hold a great victory celebration. Even though the emperor, and
possibly Galdan himself, viewed the victory as ordained by Heaven, it had
been a close call. The emperor’s thanks to Heaven reflected tremendous re-
lief at a miraculous victory. The surrender of Galdan’s greatest general, Qasiqa, revealed Galdan’s
strategic plan, one which fit perfectly with classic nomadic strategy.
dan had thought he could attain his “Great Enterprise” (amba baita),the
unification of the Mongols, by staying in the area of the Kerulen and Tula.
He regretted his penetration to Ulan Butong. He had planned to retreat if a
large Manchu army advanced until the troops exhausted their grain and
silver supplies. But on hearing that the emperor was with the army, the
Mongols lost their courage. Galdan wanted to fight the Manchus in the for-
est but could not stop his troops from fleeing. He was prepared to fight the
West Route Army alone, but the military feat of sending three armies to the
Kerulen had terrified his troops so much that they were beyond his control. Both sides were at the extreme limit of their supply lines, but the Qing
troops depended crucially on grain from the interior, while the Mongols
still had plentiful sheep and cattle. Without grass, they could not last much
longer in one place, but they were free to move. If Fiyanggû had not been in
exactly the right place to block them, Galdan could have escaped. Fiyanggû
had had to travel very far west to find water, and the journey had exhausted
nearly all of his animals. He had abandoned many supplies because he
lacked animals to carry them. When he arrived at Tula, his men were “à la
dernière extremité,” in Gerbillon’s words. They would have died of hunger
if Galdan had not come to meet them in battle. Capturing Galdan’s abun-
eating snow 189

dant provisions saved their lives. 64Ironically, Galdan could have retreated
back to the Kerulen, saved himself, and left Fiyanggû’s army to starve; but
he overestimated his own strength in facing the weakened Chinese. The
Manchus and Chinese fought desperately, knowing that they had nowhere
to go and nothing to lose. Galdan was also betrayed by his fellow Zunghars. When he sent a mes-
sage to his kinsman Arabdan at Bayan Ulan, telling him that Kangxi was
coming, Arabdan replied, “You have land with no women, children, or cat-
tle, I have land with women, children, and cattle. Didn’t you know what
the Manchus are like? I will not fight with the Manchus.”
65Arabdan aban-
doned Galdan, although he later appeared at the battle of Jao Modo. Choruses of praise for Kangxi’s stunning victory came from Chinese of-
ficials, Manchu chiefs, and Mongolian Khans and Jasaks. They expressed
their gratitude to the emperor for relieving them of Galdan’s plundering at-
tacks. The Mongols viewed the emperor as a Khan with magical pow-
ers, who could bring water and fresh grass to the areas through which he
passed (though, in fact, all the digging and searching for wells was done by
the Mongols themselves). The emperor responded by incorporating a wider
circle of Mongols within the Qing kinship realm: “I formerly saw all within
the passes as one family; now all within the Kerulen–Tula are one family.”
One Mongol prince reported (erroneously) that Galdan had killed his wife
and children when he heard that the great army was coming.
The Emperor Rewrites History
Historical reinterpretation was already under way. The emperor and his
ministers quickly placed the victory within a long historical context. They
compared the campaign to those of Yin Gaozong and Zhou Xuanwang, an-
cient rulers whose campaigns took three years and traveled thousands of li,
but our emperor went much farther into the steppe, over three thou-
sand li, and only took eighty days...[T]he Mongols collect like birds
and disperse like animals, lacking any fixed abode. That is why it is ex-
tremely difficult to exterminate them. This time we surrounded them
with a pincer movement of troops, thus we completely eliminated
them [jian]. This was done by Heaven; no human force could have
done it. Now the deserts are permanently cleared, and the border is se-
cure. This is an achievement rarely seen in history books. The Han
could not do this to the Xiongnu, Tang could not do this to the
190 contending for power

The Qing victory had already been inscribed in the history books as one
surpassing the achievements of the greatest emperors, despite the fact that
Galdan still remained alive and the frontier threat had not ceased. A con-
sciousness of the Qing as completing and transcending the achievements of
its predecessors had nevertheless begun to form, placing the dynasty in a
progression of increasingly expansive territorial conquests. The editors of
the military chronicle Record of the Emperor’s Personal Expeditions to
Pacify the Northern Frontiers (Qinzheng Pingding Shuomo Fanglue) noted
that “since ancient times expeditions against Mongolia have been a vain
waste of food and exhaustion of men and horses.”
68 This time, however,
Heaven had decreed the total extermination of the Mongolian menace. By
attributing favorable outcomes to Heaven, the Manchu rulers incorporated
the fortuitous accidents of history into a broader historical perspective. Op-
ponents of the campaigns had failed to recognize the auspicious signs.
Many Manchu and Han officials had urged the emperor to abandon a per-
sonal expedition, or to delay its start, and to give up pursuing Galdan after
he had fled. But the emperor had rejected their advice, and as a result “the
borders are firm, the inner and outer realms are peaceful.” Such retrospec-
tive historiography attributed superior insight into Heavenly forces to the
emperor, who incorporated sagely wisdom and Heavenly will in his strate-
gic thinking. Not only did the emperor follow the ritual prescriptions of his
predecessors, as when he performed the same sacrifices to Heaven and
Earth that the Yongle emperor had performed after his frontier expedi-
tions, but also he went beyond them by turning the “propensities” (shi)of
Heaven to more effective use.
69His “sacred military might” (shengwu)al-
lowed him to “defeat the fierce enemy like breaking a rotten branch.” In
1842 Wei Yuan’s famous book A Record of Sacred Military Campaigns
(Shengwuji) would synthesize the concepts of unity of Heaven, military vic-
tory, and sage rulership that originated with these conquests.
Inclusion and exclusion were the twin mythistorical strategies deployed
after the victory.
71As the newly subordinate Mongols were welcomed into
the Qing family, Galdan and his followers were forced out. Many other
peoples with undetermined allegiances, however, remained on the frontiers.
It was still necessary to drive a wedge between them and Galdan. The em-
peror printed hundreds of leaflets for the Mongolian princes of Kokonor
urging them to capture all remaining members of Galdan’s family. He
stressed that Galdan had violated the ways of the Dalai Lama by invading
the Qing frontiers and had claimed support from the princes of Kokonor,
the Russians, and “China’s Muslims” (Zhongguo Huizi) in order to plot to
conquer China and set up a Muslim as its ruler.
72Kangxi once believed that
Galdan had converted to Islam, and he feared Galdan’s use of Muslim spies
eating snow 191

in the Hami and Turfan oases. He also realized that it would be difficult to
use Muslim troops against Turkic peoples in Kokonor. Clearly, the Muslim
peoples of the region were not reliable, even though the chieftain of Hami
had offered his services in capturing Galdan. The emperor’s primary strat-
egy at this point was to win over Mongols by convincing them that Galdan
had violated his oaths to the Dalai Lama, thus excluding himself from the
Way of the Buddha, Tsongkhaba. Turning his attention to the princes of
Kokonor meant deeper involvement in the links between the Dalai Lama in
Tibet and Mongolians.Tibet, however, remained another ambiguous subject, of which the Qing
rulers had little knowledge. Kangxi’s greatest shock came after the victory
at Jao Modo, when Zunghars who surrendered told him that the Dalai
Lama had died nine years earlier. In fact, the fifth Dalai Lama had died in
1682, and the sDe-pa,or regent, had assumed power. Kangxi only learned
the true date of the Dalai Lama’s death later from Tsewang Rabdan. Events
in Tibet were certainly murky. Kangxi presumed that the sDe-pahad
usurped power and concealed the Dalai Lama’s death in order to forge a
stronger anti-Qing alliance with the Zunghars. Zahiruddin Ahmad argues,
using Tibetan sources, that the Qing misunderstood the nature of power re-
lations in Tibet, and later deliberately propagated the story of usurpation
and concealment in order to highlight the treachery of Galdan and the sDe-
73Contrary to the Qing view, the sDe-pawas no “minor official” who
illegitimately usurped power after the Dalai Lama’s death; he was a senior
adviser to the Dalai Lama, entrusted with secular duties in 1679. Soon af-
ter, the Dalai Lama withdrew into meditation. It was not uncommon to
postpone notices of a lama’s death for astrological reasons. Ahmad argues
that there was no deliberate attempt by the sDe-pato deceive the emperor,
and that the Chinese could not understand the Tibetan view of reincarna-
tion. The sDe-pamay have believed that the fifth Dalai Lama had left
his body to go into meditation, and would return later in his sixth reincar-
nation. Nevertheless, a clear shift in Tibetan policy was apparent after the fifth
Dalai Lama’s death. The sDe-pahad actively promoted a more sympathetic
view of Galdan by trying to mediate between Kangxi and the Khan, and by
trying to ward off military action. Now, in the emperor’s eyes, the sDe-pa
stood exposed as an active confederate of Galdan, and the Kokonor Mon-
gols’ obedience to Tibetan lamas threatened to draw them away from the
Qing. The Kokonor princes themselves proclaimed dual and equal alle-
giance, saying, “In the East is the emperor, in the West is the Dalai Lama.”
At the risk of alienating them, the emperor moved strongly against the sDe-
pa. Earlier emissaries from the Dalai Lama were kept in separate residences
192 contending for power

outside the city, but the latest emissary from thesDe-pawas to be arrested
for collaboration with Galdan.
Two months later, as the fruitless pursuit of Galdan continued, the em-
peror severely rebuked the sDe-pafor instigating Galdan’s revolt: “You
were originally the Dalai Lama’s subordinate, and I granted you the title of
king of Tibet. Now I know that you publicly revered Tsongkhaba, but se-
cretly allied with Galdan, and deceived the Dalai Lama. You pretended that
the long-dead Dalai Lama was still alive. You sent Jilong Khutukhtu to
Galdan’s camp to recite scriptures, prolong negotiations, and allow Galdan
to escape. You promoted the marriage of the Kokonor princess Boshokhtu
Jinong with Galdan. Galdan believed your inciting words.”
According to a captive’s report, Galdan had told his followers after the
defeat: “It was not me who wanted to penetrate deeply into China; it was
the Dalai Lama’s order. He said that a southern expedition would be auspi-
cious. The Dalai Lama has killed me, and I have killed you people.” From
this report, the emperor concluded that the sDe-pabore the ultimate re-
sponsibility. Clearly the Dalai Lama would not have approved a military in-
vasion if he had been alive. The emperor vowed to send another large army
“to Yunnan, Sichuan, or Shaanxi. Following the precedent of Galdan, I will
personally lead an army to punish you.”
Meanwhile, Galdan’s whereabouts remained unknown. One report had
it that Danjila and Arabdan, Galdan’s top lieutenants, had met at the
Bortala River, then separated to look for Galdan. No one knew where he
had gone. He could not go to Hami, held by his enemy Tsewang Rabdan.
Ayuki Khan of the Torghuts was too far away, and not friendly to Galdan
in any event. The Russians were interested only in trading relations, not in
sheltering a defeated Khan. So the sympathetic sDe-pain Tibet seemed to
be Galdan’s only option, although he might instead try to attack Tsewang
Rabdan at Hami, or head for the Altai Mountains. The goal of the Qing
was to prevent Galdan from crossing through Kokonor, picking up Muslim
followers, and reaching the protection of the sDe-pain Lhasa. Kangxi’s
third expedition against Galdan aimed to wipe out this “lone wolf” once
and for all.
The Final Campaigns and the Fate of Galdan
Conditions were unpropitious for a third expedition in 1696. The West
Army still desperately lacked food. Its horses were exhausted, its carts were
all broken, and relief grain from the Central Army had not yet arrived.
Galdan had gathered a total of over five thousand troops, although he had
eating snow 193

very few animals. He had headed first for Wengjin, from which he would ei-
ther attack Hami or head for Tibet. Fiyanggû was ordered to head in per-
son for the border to block Galdan, to burn all surplus grain supplies so
that they could not be captured, and to send his cannon back to the capital.
Time was of the essence.Once again the Qing profited from divisions among their enemies.
Galdan met with Danjila and Arabdan, but the three could not agree on
strategy. Galdan wanted to seize grain at Wengjin and march on Hami.
Danjila favored heading for the Altai. Arabdan wanted to loot the Russian
area. Because the majority disagreed with Galdan, no decision was taken.
Arabdan broke with Galdan, leaving with two thousand men. Galdan and
Danjila at most one thousand troops, no tents, no clothes, and no food. His
men were “only following him to death, searching for a place with land
[you guotu zhi di].”
82 They headed for Tamir, plundered the region, then
moved farther into the desert. By now Galdan had just one or two horses
and no cattle or sheep. His followers abandoned him, surrendering to
Tsewang Rabdan or the emperor rather than face starvation. As Galdan’s followers dwindled, the emperor gained new servants, in-
cluding the “Huihui king Abdulishite” of Yarkand, who visited the capital.
Galdan had seized his father at Ili in 1682, and only now was he able to re-
turn. The king vowed to use his twenty thousand troops at Yarkand to seize
Galdan, or to send troops from Turfan to capture Arabdan. Tsewang
Rabdan was praised for his loyalty and ordered to capture and execute
Galdan if he came to Hami: “We cannot let Galdan remain in the human
world. If he wanders into your territory, or flees to Hami, if you capture
him, execute him and send his head to us. This will display your obedi-
Further instructions to the Dalai Lama aimed to win him over and sepa-
rate him from the sDe-pa’s pro-Galdan policy. The emperor claimed that
sDe-pa had facilitated Galdan’s escape at the battle of Ulan Butong. Clearly
the Dalai Lama would not do this; it was the sDe-pa,he believed, assuming
the Dalai Lama’s name.
84 The sDe-pa was ordered to hand over Jilong
Khutukhtu to the emperor, to let the Panchen Lama rule the Tibetan faith,
and also to send him to China along with Galdan’s daughter.
85 All corre-
spondence in Tibetan or Mongolian between the Dalai Lama, the sDe-pa,
and Galdan was to be intercepted or confiscated. Qing officials luckily did
intercept envoys from Galdan to Tibet passing through Xining, obtaining
letters that revealed Galdan’s hopes for Tibetan support.
86 In these letters
Galdan reported on his loss in battle and asked for aid from the Dalai
Lama. He hoped that recitation of sutras in temples in Tibet would help
him out of his plight.
194 contending for power

By October the emperor was ready. Fiyanggû’s resistance to active pur-
suit of Galdan was making him impatient. He rebuked Fiyanggû for having
failed to arrive at the rendezvous point in the previous expedition, allowing
Galdan to escape, and he believed that they still could have pursued and
crushed Galdan at the time. Now that Galdan was starving, this was a
“great opportunity sent from Heaven” which they could not pass up. Yu
Chenglong returned to the capital, having transported 27,000 shiof grain,
distributing 18,000 shiand storing the remainder in granaries. Supplies
seemed sufficient. The emperor announced that he would set out on Octo-
ber 14, 1696, from the capital on a “hunting expedition” to the Ordos.
This expedition could not capture Galdan, whose whereabouts were un-
known. Its goal was to demonstrate Qing wealth to the Mongols of the re-
gion, with the hope of enticing all of Galdan’s supporters to surrender. Un-
like in the previous expedition, the emperor now had substantial numbers
of Western Mongol allies. Hosted by leading lamas in the region, he trav-
eled at a leisurely pace. One hundred thousand sheep traveled with him.
Passing through Guihua, he reached the shores of the Yellow River on No-
vember 22, the site of a large warehouse for rice, containing 70,000 shi.
Twenty days’ supplies were distributed. On the next day the emperor mea-
sured the breadth of the river with his Jesuit surveying instruments, report-
ing that the grass there was “so high you can’t see the horses.” A week later
he crossed the frozen Yellow River with his supply train and entered the
Ordos region.
Meanwhile, Fiyanggû reported a serious defeat of an effort by Danjila to
raid the grain stores at Wengjin (on the Onggin River).
89The Qing troops
had already burned most of the grain stores in the region, leaving them lit-
tle to live on. In writing to his son, the emperor said that Danjila’s original
intention had not been to steal grain at Onggin, but his followers claimed
that it was better to attack than starve to death. With the failure of this raid, Galdan’s only way to get food was to attack
Hami, but this could offer an opportunity for the Qing armies to crush him
again. Or he could hope to survive the winter in his camp, in the Jasaktu
Khan’s territory, forty days’ march from the Qing frontier. The emperor
wavered. Why waste men and horses trying to crush Galdan now, espe-
cially since it would be impossible to reach his camp during the winter?
He told Fiyanggû to turn back. They could fatten up their horses, wait un-
til spring, and then pursue Galdan. Meanwhile, they would attract more
Zunghar followers to surrender and disperse them to scattered grazing
The Ordos was a rich region with fine grass for the horses and excellent
prospects for hunting. In his letters home the emperor praised the invigo-
eating snow 195

rating fresh air, the wonderful taste of the mutton, and the fine workman-
ship of the Mongolian saddles.
91The Mongols in turn were extremely im-
pressed with the huge herds of cattle that the Manchus had brought with
them. Observing herds of sixteen thousand cattle and seventy thousand
sheep for one banner, they said, “Since our ancestors’ times, we have con-
sidered one to two thousand head as making a man rich. We have never
heard of ten thousand head.”
When Fiyanggû arrived with his exhausted West Route troops, the em-
peror praised their endurance and held a great celebratory banquet.
Fiyanggû now had sufficient supplies, but his troops needed rest. The sim-
plest strategy would have been to wait comfortably in the Ordos while
Galdan died of starvation in his isolated camp. Surrounded on all sides by
enemies, losing supporters daily, he was like an animal in a cage.
93Still, the
emperor was anxious to march to Hami. Fiyanggû objected that the dis-
tance was too great and his troops were famished. Jiayuguan was a thou-
sand kilometers away, and it was twenty days’ march from Jiayuguan, at
the end of the Great Wall, to Hami, a distance of five hundred kilometers.
Sunsike could not go beyond Suzhou, near Jiayuguan, because he lacked ra-
tions. The emperor was still anxious to proceed at least as far as Ningxia.
But on December 19 he met the famished troops returning from the distant
steppes and discovered how truly exhausted his army was. Fiyanggû then
reported the arrival of Galdan’s envoy, Geleiguying, to discuss surrender
It could have been a trick. Fiyanggû recommended seizing Geleiguying
and not sending him back. But Geleiguying kowtowed to the emperor
and vowed that the Zunghars sincerely recognized their crimes and wished
to surrender: “We Zunghars were ignorant. We coveted the wealth and
women of the Khalkhas and did not realize that Heaven’s will is without
partiality. This was our crime. Now many Zunghars want to submit. Our
lord cannot be ranked with the Khalkhas.” The emperor smiled and said,
“These words are just. Even though you are foreigners [waiguo zhi ren],
you understand reason [li]. Galdan is ignorant; he has failed to accept my
benevolence. He has chosen his own death.”
95He offered generous terms.
Zunghar nobles would be given official rank; others would be assigned
to banner service. Captured women and families would be returned. He
claimed that unlike previous rulers, who loved military victory, he prided
himself on putting peace foremost. The traditional Chinese conception of
“coerced followers” (xiecong)also provided grounds for pardoning even
the most treacherous rebels. Offers of wealth and honors were expected to
bring around even the stubborn Galdan.
196 contending for power

Galdan’s letter to Kangxi, brought by Geleiguying, clearly affirmed his
Buddhist beliefs and the justice of his campaigns:
Buddha teaches that humans cannot clearly foretell events, but all
great Khans who unify the world worship the three treasures of the
Dalai Lama, as we have. That is why since Altan Khan the seven ban-
ners of the Khalkhas have been the patrons of the Dalai Lama. Since
Gush Nomun Khan, we four Ölöd have also been patrons of the Dalai
Lama. We have each lived peacefully and separately in our lands. We
have not waged war against the Jebzongdanba Khutukhtu or Jasaktu
Khan. They failed to respect the Dalai Lama’s representative, causing
the great turmoil. My cause was just, but I will submit to the emperor’s
Galdan faced great pressure from his ministers and followers to sub-
mit to the emperor. Even though his own letter scarcely admits any guilt,
Geleiguying’s report to the emperor reveals considerable distress in the
Zunghar camp. Most likely, from Galdan’s perspective, the surrender talks
were only a tactic to win time, pacify his people, and give them the strength
to last the winter. He still hoped for rescue from the Dalai Lama and the
Khans of Kokonor. On December 21 the emperor decided to accept Galdan’s surrender. He
sent Geleiguying back to Galdan, setting a time limit of seventy days for a
return report. He would remain in the Ordos hunting to await a reply. If no
reply came, he would advance his troops. Just at this time the bondservant Dadaduhu memorialized that the army
had nearly used up its supplies and had to turn back. The furious emperor,
accusing Dadaduhu of stirring up doubts among the people, ordered his ex-
ecution. Once again he vowed, “If grain supplies are exhausted, we will go
to riverbanks and marshes to get grain. I will eat snow to pursue Galdan;
we absolutely cannot turn back.”
97Dadaduhu’s crime was to reveal the lo-
gistical weaknesses of the army in front of Galdan’s envoy. The Qing gener-
als had Geleiguying followed until he was far from the camp. Only then did
Kangxi announce that the army would return to the capital. All the troops
were delighted. Both Galdan and the emperor were in extreme situations, although Gal-
dan was much worse off. Both faced increasing dissension from their high
generals and troops, but neither would give in. Weary fighters slugging out
the last rounds, they had to call a temporary truce to rebuild their strength
and regain support. Although the Ordos was rich, it would not support a
eating snow 197

large army over the winter. The weather was turning cold, the emperor
missed his sons, and he would get no reply from Galdan anyway for the
next seventy days. So, despite his vows to Geleiguying, he turned back and
arrived at the capital on January 12, 1697. This third expedition had lasted
ninety-one days. Several thousand Zunghars had surrendered, even though
there had been no decisive battle. It had succeeded as a demonstration of
Qing wealth and power to the already subordinated Mongols; it had
blocked Galdan from escaping to Kokonor or Tibet; but it had not yet elim-
inated Kangxi’s indomitable foe.Kangxi had little expectation that Galdan would surrender. Yet he also
knew that another expedition would require even more time and planning.
Once again, opposition surfaced. Censor Zhou Zihuang objected to an-
other personal expedition, arguing that subordinate officers could easily
subdue such a petty bandit, but the emperor replied that he must pursue
Galdan to the end. He invoked the great consequences of Wu Sangui’s up-
rising twenty years earlier, which had spread unrest even to the northwest
98 Still, he wavered for some time until the begof Hami reported
the capture of Galdan’s fourteen-year-old son Sebteng Baljur.
99 This was
“Heaven’s gift” to the Qing. It left Galdan abandoned and alone, without
any heirs. This was a strong incentive to advance once again to the north-
west and confront the isolated Galdan. He had only five hundred to six
hundred men left, many of whom would desert him when faced with a
strong military presence.
Galdan was now camped in the Altai Mountains, over 1,600 kilometers
northwest of Ningxia and twenty-nine days’ march north from Jiayuguan.
Because Galdan’s forces were so weak, the Qing armies could travel light
and fast. Two forces of three thousand troops each would set out, one from
Ningxia and one from Jiayuguan. The start of the expedition was set for
February 27, 1697. Because of their previous experience, the supply officers
developed much more careful and cautious plans than in the earlier cam-
paigns. Only Mongol allies with many horses could join the army. Less was
more. The route to Ningxia was well known, and the emperor was anx-
ious to go. He collected information from the Khalkha Mongols about the
route. On the previous expedition he had crossed the wall and headed
straight for Köke Khota, but on this trip, the emperor went along the inside
of the Great Wall to Yulin, taking the opportunity to inspect the Shaanxi
landscape. His small force of four hundred men struggled over rocky
mountains, ravines, and deep sand. The Ming Chengde emperor (r. 1506–
1521) had visited here during his northwest expedition, but Kangxi wrote
to his son that his travels would equal or surpass that hapless emperor’s. Al-
though Kangxi did not resemble the abusive Ming ruler, a poor warrior in-
198 contending for power

terested mainly in military pomp, this tour had more symbolic than mili-
tary significance. Historical precedents echoed in the emperor’s mind.
On March 26, at Shenmu, between the Yellow River and the Great Wall
in Shaanxi, the emperor interviewed Sebteng Baljur, Galdan’s son, trying to
find out if his father would surrender. The terrified boy could not give a
definite answer, but hoped that his father would give in to the imperial
might. Kangxi regarded the boy as a “small, inferior person” lacking his fa-
ther’s courage. Gerbillon, however, found him to be well built, “triste et
étonné,” and thought he held up well under questioning. The boy was sent
on to Beijing.
102 Kangxi planned to have the boy executed in the capital
along with his father. The Hami begwho had arrested and delivered Sebteng feared that
Tsewang Rabdan would take revenge on them, and, as expected, Tsewang
Rabdan indeed soon demanded custody of the boy, his greatest weapon
against Galdan. When Hami asked for Qing protection, it came under for-
mal Qing authority. The first Turkic oasis now joined the Qing sphere.
Soon after, the Princes of Kokonor submitted. Except during the Yuan dy-
nasty, Kokonor had never been under imperial control. This was an unprec-
edented submission to imperial rule of “outer barbarians,” as described in
former histories.
104 In a letter to his son, the emperor expressed his “great
joy” at gaining the submission of these powerful local rulers without a bat-
tle. Stretching the truth a good deal, he even claimed Tsewang Rabdan as
his subject. In reality, Tsewang Rabdan was only an ally of convenience,
willing to cooperate in the defeat of Galdan, but not a permanent subject.
Tibet remained the one major inaccessible region beyond Qing control.
But Galdan’s losses and the defection of other allies had shaken the regent’s
106 The “terrified” sDe-panow denied that he had ever supported
the “rebel” Galdan; he identified the emperor with Manjusri Buddha, and
vowed great gratitude to the emperor for granting him the title “King of Ti-
bet.” He did try to prevent the Panchen Lama and Galdan’s daughter, now
in Tibet, from being summoned to the capital, but the emperor refused his
request. The regent confirmed, at last, that the fifth Dalai Lama’s death had
occurred in 1682. His story was that the Dalai Lama before his death had
assured the lamas that he would be reincarnated the next year, but he told
them to keep his death secret until his successor was sixteen years old.
The sDe-pa promised that the present Dalai Lama, now fifteen, would be
revealed (chuding) by the end of the year. He asked the emperor to conceal
this fact until the tenth month, but it turned out that the new Dalai Lama
had already been revealed to Tsewang Rabdan. Refusing to cooperate in
this deception, the outraged emperor proclaimed it to all the Inner Mongo-
lian banners.
108 The revelation of the new Dalai Lama was actually advan-
eating snow 199

tageous to Kangxi, because all along Galdan had claimed to be acting on
the legitimate orders of the Dalai Lama; but now he stood exposed as rely-
ing only on the usurping regent.
109 The regent, likewise, expecting Galdan’s
defeat, withdrew his support in order to curry favor with the rising Qing
power. Now on the brink of victory, the emperor was magnanimous. He would
pardon the regent’s crimes and seek peace among all the Mongols:
I note that Mongols beyond the borders have always resisted China.
From the Han, Tang, Song and Ming dynasties they have harmed us
...Nodynasty until ours has exerted authority over Mongolia, and
made them submit in their hearts [guixin]. Using troops is cruel; the
ruler does it only if necessary. It is like using needles on a person with
illness; we do not inflict pain for no reason on the skin. Ruling is like
this: in times of disorder [luan] we need force, in times of peace and or-
der we use pacification [fusui]. Since ancient times those who loved
distant expeditions caused losses to the country’s spirit. So I value
most not creating trouble.
Having ensured the temporary subordination of Tibet, Kokonor, and
Tsewang Rabdan, the emperor continued his pursuit of Galdan. From
Yulin to Ningxia, the route led outside the wall and along it through the
southern Ordos desert. Unlike the previous short venture into the rich fields
near the Yellow River, this crossing of the Ordos led through deep salt and
sandy wastes with very little water. During the Ming dynasty this region,
where the Great Wall lay within the Yellow River bend, was the chief head-
ache of strategic planners, and a major subject of debate.
111 Mongol tribes
constantly attacked the wall and raided within it, while Ming officials dis-
agreed hopelessly over launching expensive, useless campaigns or hunker-
ing down for humiliating defensive tactics. Kangxi agreed that it was a wise
policy to extend the wall through the Ordos because it was not defensible
otherwise. He enjoyed having his aides debate different calculations of the
distance to Ningxia. He now realized that the route outside the wall to
Köke Khota offered much more water and grass than this more direct
desert crossing. Local peasants had dug breaks through the wall in order reach their cul-
tivated fields outside. It was made only of tamped-down earth and piled-up
rocks, fifteen feet high and six to seven feet wide at the top. Towers spaced
along it accommodated three to four guards each, and men to light watch-
fires. Defenses here clearly could not stop a serious cavalry attack, but
Galdan and other Mongols, unlike in the Ming, were in no shape to mount
200 contending for power

any challenge. 112 The emperor declared that he would give up the “trivial”
hunt for animals in order to focus on human prey. The weaker the Mon-
gol enemies became, the closer they came to the animal order in Manchu
On April 17 the Qing troops arrived at Ningxia, 1,400 kilometers from
Beijing, after fifty-one days on the march.
114 The Ningxia region was a
beautiful, rich landscape with cheap food supplies and abundant irrigation.
The emperor’s main concern was to prevent excessive extractions from the
area’s population. Summoning the local elite, he encouraged them to aid
the army without burdening the peasantry. Yu Chenglong planned grain
shipments to reach the Altai Mountains. He could carry 3,000 shiof grain
up the Yellow River to its northwestern bend in boats, then transfer sup-
plies to camel and baggage carts. It would take many days to reach Godoli
Balaghasun, a staging point 1,200 liaway in the Southern Altai Mountains,
but by June, Yu had transported huge quantities of grain there—forty-five
days’ supply for each man. His troops built walls six feet high and pits nine
feet deep to guard the enormous stores. Yu’s efforts made the critical differ-
ence in the victory over Galdan. According to Wei Yuan, after Galdan’s
death a Qing general told Danjila, Galdan’s chief general, “‘The man who
destroyed your state is the Grain Transport Commissioner Yu Chenglong.’
Danjila hung his head in shame.”
The emperor spent nineteen days in Ningxia making preparations for
this great logistical operation. His transition from frontline commander to
supply sergeant proved very frustrating. He complained to his son about
his constant preoccupation with arranging silver and grain supplies for the
116 By May 19 he had completed all his preparations and boarded a
boat to travel downstream to Baita on the bend of the river. The townspeo-
ple of Ningxia surrounded the troops, begging them to stay; local mer-
chants had made great profits from supplying the army.
117 Before leav-
ing, he sent out further calls for Galdan’s surrender, encouraged Tsewang
Rabdan to help in crushing his Zunghar rival, and set a date for troops to
start for Galdan’s camp.
118 Then he watched the troops set out from Baita.
Abandoning his earlier vows to march for twenty days across the desert
in pursuit of Galdan, the emperor continued downstream from Baita and
returned to the capital, this time taking the shorter route across the steppe.
Despite the merchants’ delight, the poor region of the northwest had been
heavily burdened by the imperial presence, and he had been away from the
capital for over seventy days. One letter indicates growing distrust of his
son’s “coldness,” foreshadowing the later severe loss of confidence in his
heir that would lead to the great succession crisis of the end of his reign.
The emperor returned home on July 4, 1697, after a 129-day journey, his
eating snow 201

longest yet. Most of his time, however, was spent not in the steppe but in
the towns within the wall or on the Yellow River.Kangxi could not know that all of his frenetic preparations had been ren-
dered moot by Galdan’s death even before he arrived at Ningxia. Reports
from the envoys sent to discuss surrender terms arrived in Ningxia shortly
after the emperor got there. They indicated severe dissension within
Galdan’s camp. Drinking with Galdan in his tent, Danjila and another
Zunghar leader, Urjanjab, denounced Galdan for bringing about the de-
struction of their state and failing to defend the “Way of Buddha”: “We
have followed you until the end...butnowwecannot bear it anymore,”
they said. “The road divides in two,” either surrender or death. Others re-
proached Urjanjab for praising the Khan in times of prosperity and desert-
ing him in adversity. Galdan had only three hundred men left, many of
whom had only one skinny horse. Others were starving to death. But Gal-
dan refused to surrender. Danjila invited the envoy to his tent, offering to
surrender to the emperor himself if he would be beneficent. Another noble
follower of Galdan, Noyan Gelong, praised the emperor as a living Bud-
120 The emperor’s order pointed out that Galdan was surrounded on all
sides by enemies; if he relied on the emperor, he would live, he would keep
his title of Khan, and he and his followers would become wealthy. Thus
Galdan’s men learned that Kangxi was camped in the Ordos with abundant
grain and cattle supplies. Danjila was clearly impressed enough to want to
surrender, but he could not move Galdan, so the envoys left. Geleiguying,
Galdan’s own envoy, himself deserted the camp and joined them. Soon after, refugees reported hearing cannon fire near Galdan’s camp.
They learned that Galdan was moving farther northwest, into the Altai
Mountains. He had split with Danjila and moved off with only one hun-
dred men. Then, on April 4, 1697, at Aca Amtatai, between Kara Usu Lake
and Khobdo, Galdan suddenly died under mysterious circumstances. The
emperor did not hear of Galdan’s death until the night of June 2, fifty-nine
days later, while resting at Baotou on his return journey. Fiyanggû met en-
voys from Danjila who reported that Galdan had caught an illness one
morning and died that same evening. They did not know what the illness
was. When asked why they had not reported this earlier, they claimed that
Danjila’s men and horses were so weak they could not move. Immediately
before his death Galdan had said, “I believed that the Zunghars were a
good people; I did not expect them to be so faithless.”
The emperor’s first reaction was to suspect poison. He told his son that
either Galdan had committed suicide or he had been poisoned by one of
his followers. The truth would not be known until they had interrogated
Cembu Sangbu, Galdan’s trusted doctor, who had prevented even Danjila
202 contending for power

from offering meat to Galdan. 122 But in his meeting with Fiyanggû the next
day, and in all subsequent pronouncements, he settled on the suicide story.
The emperor and his ministers had often predicted that Heaven’s great re-
wards to the Qing and Galdan’s despair would lead him to kill himself.
Once again, these predictions had come true. In fact, no evidence indicates that the astonishingly stubborn Galdan had
any intention of committing suicide, and his Buddhist commandments for-
bade him, a living Buddha and reincarnation of a high monk, to do so. It is
possible, as some scholars believe, that Galdan died of a sudden natural ill-
ness (such as a brain seizure).
123 His illness could certainly have not been a
slow degenerative one like smallpox, the most common and most feared
cause of death among Mongols exposed to outside influences. Yet he died
very shortly after the stormy meeting with Danjila and other top aides re-
ported by Kangxi’s envoy. Given the severe disagreements over surrender
within Galdan’s camp and his evident distrust of even his most loyal aides, I
think the emperor’s first suspicions were correct: Galdan was poisoned by
one of his followers, most likely Danjila, who gained credit with the Qing
for delivering Galdan’s body and saved himself from complete ruin. But this outcome did not fit imperial mythistory. Compilers soon altered
the story to make it accord more neatly with the thesis that the Qing em-
peror embodied Heaven’s will. The date of Galdan’s death was moved later
in time by a month, to the lunar month 3*/13 (May 3) instead of 3/13
(April 4), so that the entire expedition would not be revealed as useless.
In subsequent letters to his son and in proclamations to the army, the em-
peror insisted that his interrogation of Danjila’s messenger confirmed the
suicide story. Later commentators noted that “the emperor predicted that
Galdan would kill himself [zijin], and he has done so. The emperor is so
farsighted that he understands the enemy like a god.”
The Chinese edition of the Qingshilu(but not the Manchu one) endorsed
Kangxi’s claim, and so have nearly all subsequent historians. Yet the re-
ports contained in the Qinzheng Pingding Shuomo Fanglue, and Kangxi’s
original letters to his son, preserve the recalcitrant facts that contradict the
imperial mythmakers. On July 4 a great military review celebrated the emperor’s return to
Beijing. Bannermen, merchants, elders, and women lined the streets of the
capital, holding incense and prostrating themselves before the procession.
The Grand Councilors listed the emperor’s great feats—defeating the rebel-
lion of the Chahar Mongols, suppressing the Three Feudatories, conquer-
ing Taiwan and “entering it on the registers” (ru bantu)as an overseas terri-
tory (haiwai), collecting tribute from Russia, which never before had had
contact with China—now crowned with the “final elimination” of the
eating snow 203

Mongol menace. They boasted, “No emperor from the past can compare
with this.” At the Taihedian a great ceremony celebrated Galdan’s extermi-
nation. Officials reported the victory to the Temples of Heaven and Earth,
the Taimiao and other shrines, and the ancestors at the imperial tombs.
Qing officials paid tribute to Galdan’s great military skill, reckoning that
he had subdued 1,200 towns in the western regions, including the Muslim
oases, Bukhara, Turfan, and the Kazakhs, and defeated several hundred
thousand men in the Khalkha banners. This tribute to Galdan undermined
the notion of him as a mere “rebel,” but building him up as a great military
leader was necessary to justify the costly personal expeditions against him. Rewards and punishments followed the celebrations. Nearly all officials
and criminals under death sentences (except those convicted of the “ten
great evils”) received pardons. Several thousand troops received bonuses of
5 to 6 taels annually for three years. Ilagukesan, the high lama adviser to
Galdan, who recited sutras for him before battle, was condemned to death
by slicing, but over two hundred other lamas in contact with him were par-
doned. Even the Jilong Khutukhtu, the peacemaker sent by the Dalai Lama
who urged the Zunghars into battle, was pardoned, after pleas for his life
from the Panchen Lama. Surrendered Zunghar troops were incorporated
into the Chahar Mongol banners.
127 Muslims who had joined with Galdan
were also spared execution. Chinese officials, like Yu Chenglong, scheduled
to be punished for delays in grain transport, were now pardoned and re-
warded with promotions. Fiyanggû became first-rank prince. General
Maska, who had failed to capture Ilagukesan and been stripped of his posi-
tion, was now restored to a lower rank. The “probationary ethic” in action
had kept these officials in suspense until the victory, but now it was time for
It was still necessary to display military might before the Mongol Khans.
At a great banquet for the Khans of Kokonor, troops passed by and thun-
derous cannon roared, “shaking the mountains and valleys.” These Khans,
a crucial link between the Zunghars and Tibet, now presented tribute and
seemed to submit to the overpowering might and “godlike mystery” of the
emperor. Gunpowder in the imperial presence had a powerful symbolic ef-
fect. Soon, as the Yongzheng emperor discovered, these Khans would be
much less obedient.
Leniency did not apply, however, to the body of Galdan himself. It was
not enough for the Qing to remove Galdan from the human world. Even
his corpse had to be eliminated. The goal was to treat him the same as Wu
Sangui: to burn the body, display it at the execution ground in the capital,
then pulverize the bones and scatter them through the streets. The head
would be hung on the city wall, displayed to the people, and later sent
204 contending for power

around to all the Khalkha Forty-nine Banners who had surrendered to the
Qing. Thus the polluting presence of Galdan’s spirit could be eradicated.
His son would suffer the same treatment.
But the Mongols had other ideas. Danjila had already burned the body
but kept the head and ashes. Danjila was determined to take Galdan’s re-
mains to Tibet to present them to the Dalai Lama, but he was captured by
Tsewang Rabdan, who took custody of Galdan’s remains. The emperor’s
envoys demanded that he return them to Beijing. They also demanded that
he hand over Galdan’s wife, son, and daughter in his possession. After
the arrival of the ashes, they would be executed along with Galdan’s son
Sebteng Baljur, already in Qing custody. Tsewang Rabdan replied, however,
that by Mongol custom “we do not take revenge on useless people,” and
that he harbored no enmity toward Galdan’s family. Galdan’s children and
his ashes were not objects of revenge. It was also taboo to entrust a corpse
to other people.
131 He would throw the ashes in the water or scatter them in
the fields. He did, however, hand over for execution the lama Ilagukesan. As the Qing envoys made clear, however, by Chinese custom “all the
family of rebels are to be wiped out, corpses captured, all must be entirely
swept away [qiongjiu saochu].”
132 In the Qing view, if Tsewang Rabdan re-
fused to deliver the remains, his tributary relationship with the Qing, and
his earlier vow to help eliminate Galdan, would be exposed as a lie, and the
Qing would cut off all access to their territory for Tsewang Rabdan’s trad-
ers. Tsewang Rabdan agreed with the Qing that Galdan had been a threat
to both of them, and he rejected the usurpation of power in Tibet by the
sDe-pa, but he was already staking out an independent position on Gal-
dan’s legacy. After a severe imperial rebuke, he agreed to deliver the re-
mains but asked for mercy to be shown to Galdan’s family. He also asked to
keep Galdan’s remaining followers, because he needed their strength to
launch attacks on the Kazakhs to his west. The Lifanyuan insisted that
these men, and Galdan’s family, could not be allowed to stay in Mongolia,
but the emperor relented partially. He allowed Galdan’s men to remain, but
insisted on the return of his wife and children. Finally, in the fall of 1698, Manchus, Mongols, and Chinese assembled
on the military training ground in Beijing to watch Galdan’s bones be
crushed and scattered to the winds. Sebteng Baljur was pardoned and given
a wife and first rank in the Imperial Bodyguard. In 1701 Galdan’s daughter
arrived in Beijing to live with her brother.
What does this ritual dispute reveal about the Qing–Mongol conflict? As
Evelyn Rawski has shown, Qing legitimation relied heavily on the constant
performance of prescribed rituals.
134 These rituals were both public and pri-
vate, and they combined Confucian, shamanistic, and Tibetan Buddhist
eating snow 205

practices. The court frequently visited temples in the city to ensure auspi-
cious harvests and moral order. But there is one ritual she does not discuss:
executions. These, too, were highly conspicuous performances by which
the dynasty displayed its power and authority. As Lu Xun’s Ah Q noticed,
proper behavior by both the victim and the executioner was expected.Capital punishment of common criminals followed a hierarchy defined
by the penal code ranging from hanging to execution by slicing. Major re-
bels, however, because of the sheer scale of their crimes, went well beyond
ordinary violations of law. In fact, if Galdan were treated as a rival power,
his actions would not amount to a crime at all, since he would have been
playing by different rules. The Qing would have to recognize his interest in
the security and autonomy of his state instead of merely seeing a compul-
sion to destroy social order. Although such a geopolitical vision suited the
diplomatic interests of Qing officials before victory, it would not fit the im-
perial ideology of the Qing after its defeat of the Mongol armies. So Galdan
had to be reduced to a common criminal but treated to a punishment even
more extreme than for a common rebel. Although death by slicing was in-
tended to ensure that the victim’s soul would not survive, crushing bones to
powder and scattering them over a parade ground went even further. It was
a degree of obliteration not found in the penal code: it represented the ulti-
mate in imperial erasure of an enemy from both the human and cosmic
realms. The determination to eradicate totally Galdan’s alien presence ex-
plains the emperor’s tenacious insistence on recovering his body. In the
mid-eighteenth century the Qianlong emperor would attempt the same
against Amursana, the final representative of Zunghar resistance. Tsewang Rabdan, for his part, also asserted cultural values recognizable
to the Qing: we do not make enemies of our kinsmen. To him, Galdan was
family. If Galdan persisted, even symbolically, the Zunghar state could
rise again. The emperor had to cast Galdan out of the familial role and
place him in the criminal pigeonhole. Only after his bones were retrieved,
crushed, and ritually expelled from the human and even physical realm
could his specter be eliminated permanently. But the family principle won out for the rest of Galdan’s followers. The
principle of “coerced followers” (xiecong)allowed for leniency for nearly
all except the top leader. Regarded as innocent dupes of a clever manipula-
tor, followers of peasant rebels and nomadic raiders usually received par-
dons. Even Galdan’s top lieutenants were pardoned, and some were given
official positions. His wife and children were also pardoned. Their actions
were partitioned from the evil of Galdan himself. Incorporation and ex-
treme exclusion each required the other. Once the body and soul of Galdan
himself was physically annihilated, removing any potential pollution, all
206 contending for power

One of the French etchings commissioned by the Qianlong emperor to celebrate his military victories, depicting the presentation and exe-
cution at Beijing’s Meridian Gate of captives taken in the Jinchuan campaigns. The group of men kneeling on the extreme left are present-
ing the head of Khoja Jihân to the emperor, who sits on the high dais on the right.
[To view this image, refer to
the print version of this title.]

others personally connected with him were purified and could be reincor-
porated into the human realm.With the successful conclusion of the great Galdan campaigns began the
Qing project to establish a definitive demarcation of the dynasty’s position
in imperial space and time. Until Galdan’s defeat, the character and even
prolonged survival of the dynasty remained uncertain. Would it last for
centuries as one of China’s great eras, or would it be a relatively short-lived
conquest regime, overthrown by another more powerful Mongolian re-
gime, like the Jurchen Jin (r. 1115–1234 ce)? And would the extent of Qing
rule include permanent dominion over the vast areas beyond the wall, or
would it end up like the Ming, confined to defensive positions after brief
forays into the steppe? The elimination of the Zunghar leader seemed to in-
dicate relative stabilization and permanent control. So the emperor began a
great project to map the entire extent of imperial domains shortly after his
victory. In a later chapter I discuss the significance of this vast effort to in-
scribe legibility onto Qing domains, known as the Chart of the Compre-
hensive Gaze over Imperial Domains (Huangyu Quanlan Tu), or “Jesuit
Atlas.” It was published in several editions from 1717 to 1721, but the use
of Jesuits in surveys closely connected to strategic concerns began in 1700.
By the end of the seventeenth century, with their greatest adversary appar-
ently eliminated, Qing rulers could put down definitive boundaries of their
territorial claims. Similarly, the emperor laid the basis for placing his conquests in histori-
cal time by commissioning an official history of his campaigns, Outline
History of the Personal Expeditions to Pacify the Northwest Frontier (Qin-
zheng Pingding Shuomo Fanglue). Begun in 1699 under the supervision of
top officials and Hanlin compilers, it was published in 1708. Later I will
analyze in detail this complementary effort of the Qing to fix its narrative
alongside its mapping of imperial terrain. In both cases, the emperor and
the scholar-officials he patronized moved quickly to set in print a selective,
limited spatial and temporal perspective, excluding alternative versions and
designed to last for all time. The late-seventeenth-century conquests, however, did not conclude the
northwest story. Kangxi’s later years, and the curious reign of his son, the
Yongzheng emperor, indicated that serious challenges to his authority per-
sisted, and major underlying constraints on Qing mobilization had not yet
been overcome. As Chapters 6 and 7 demonstrate, the Zunghar resistance
to undisputed control of the northwest lasted for more than a half-century
beyond the Kangxi emperor’s victories. His successors built on his achieve-
ments, but in retrospect they were only temporary successes.
208 contending for power

Imperial Overreach and
Zunghar Survival, 1700–1731
T he death of Galdan by no means ended the power of the Zunghar
state. Under Galdan’s nephew and successor Tsewang Rabdan (r. 1697–
1727), the Zunghars reached the peak of their power. Galdan Tseren
(r. 1727–1745), Tsewang Rabdan’s son, not only held the state together but
also ambushed a large Qing army and drove it back in humiliation. Yet
within fifteen years after Galdan Tseren died, the Zunghar state and peo-
ple had vanished. When internecine struggles for leadership destroyed the
unity of the Zunghars, the young Qianlong emperor seized his chance to
eliminate them. Although Qianlong’s victory was sudden, it was not fore-
ordained. For the first half of the eighteenth century, the three empires
maintained an uneasy coexistence, interacting more through trade than
through war. From 1700 to the death of the Kangxi emperor in 1722, the frontier
remained relatively stable in Turkestan, but Qing intervention in Tibet
opened a new arena of competition. The 1720s were a period of transition
for all three states, marked by the death of Kangxi in 1722, the death of Pe-
ter in 1725, and the death of Tsewang Rabdan in 1727. New rulers aimed
to carry on the policies of their predecessors, but with considerably less
vigor. Of the three successors, it was Galdan Tseren who ruled the longest
and gained the most success. The Yongzheng emperor, at first cautious and
austere, launched a reckless attack in Mongolia that ended in a resounding
defeat. He reluctantly settled for a truce, and put economic pressure on the
Central Eurasians. The Russians gained their long-awaited access to the
China market, opening the frontier town of Kiakhta in 1727, but the trade

with China proved to be a disappointment. Galdan Tseren held his state to-
gether, gathered commercial resources, and maintained his position be-
tween the two giant empires.Modern Chinese historians reveal ironic ambiguities in their discussion
of this period. Although they must portray the Zunghars as “splittists”
who undermined the unity of the peoples of China, they also praise Tse-
wang Rabdan for resisting Russian “aggression” against “our people,” the
Mongols included. Russian historians, conversely, view Tsewang Rabdan
as holding off Manchu expansionism. In fact, it is difficult to convict any of
the regimes of unilateral expansionism. Nationalistic focus on unity and
the zero-sum binary game obscure the multiple interactions among the con-
tending peoples. This period is full of surprises.
The Rise of Tsewang Rabdan
Tsewang Rabdan was the son of Sengge, the brother whose assassination
brought Galdan rushing back from Tibet. Already in his twenties, he had
hostile relations with his uncle, who kidnapped the princess to whom he
was betrothed and sent assassins to kill him. Tsewang Rabdan fled to the
Bortala valley to escape Galdan’s pursuit, and held him off in a pitched bat-
1When Galdan moved east in 1690 to intervene in the Khalkhas’ inter-
necine struggles, Tsewang Rabdan took advantage of his uncle’s absence to
expand his strength into Khobdo in Galdan’s rear. After the battle of Ulan
Butong, which severely damaged Galdan’s strength, Tsewang Rabdan grew
even stronger, but his proposal to partition the Zunghar empire was re-
2As Galdan prepared for his next battle with the Qing in 1696,
Tsewang Rabdan opened secret contacts with the Qing, promising to cap-
ture and deliver Galdan to them. In return, he asked the Qing to send back
to him any Muslim merchants who fled from Galdan into the Qing interior.
He was allowed to send tribute embassies of up to three hundred men to
Beijing, which would provide him with valuable trade goods. With the death of Galdan in 1697, Tsewang Rabdan formally took un-
disputed control of his state. Because he lacked a Chinggisid lineage, he
could not call himself legitimately “Khan” of the Zunghars. He had ob-
tained from the Dalai Lama the title of Erdeni Zoriqtu Hongtaiji in 1694.
The Russians referred to him as “Kontaisha.”
3Instead of trying to confront
the powerful Manchus, he focused on the Kazakhs to his west. In 1698–99
he launched a series of attacks on the Kazakhs on the Irtysh and lower
Xier rivers.
4Although he used the pretext that the Kazakhs had blocked a
Zunghar trade embassy, the causes lay deeper. Galdan’s losses had deprived
210 contending for power

the Zunghars of a great expanse of pastureland, which Kangxi refused to
return. Meanwhile, the Kazakhs blocked their access to the west, and Rus-
sian settlements moving up the Siberian rivers threatened to confine them
even further.
Tsewang Rabdan, more than his uncle, had to focus much of his atten-
tion on preserving access to territory and trade routes instead of winning
over the allegiance of other Mongol tribes. He began to behave more like a
territorial sovereign hemmed in by other states and less like a free-ranging
nomadic conqueror. Greater stability in the definition of Zunghar borders
aided the growth of settlement. He developed agriculture, crafts, and indus-
trial production with the use of captives from Turkestan, Russia, and even
Sweden. He maintained control of Turkestan by granting lands to the Mus-
lim lords in return for payments in kind for his state. He also developed
closer ties with the faraway Torghuts (Kalmyks) on the Volga River. The
Qing themselves sent Tulisen on his great expedition to gain support from
their Khan, Ayuki, for an alliance against Tsewang Rabdan. When Sanjib,
the son of Ayuki, left the Volga for Tsewang Rabdan’s territory with ten
thousand to fifteen thousand households, Tsewang Rabdan gained a pow-
erful contingent of men and cattle. Sanjib had broken away from his father
and returned to Mongolia with the original intention of seizing Tsewang
Rabdan’s throne. Tsewang Rabdan defeated him, however, and sent Sanjib
back home, but kept his fighting men for himself.
Up to 1715 the Zunghars clashed most severely with Russia, not the
Qing. Russians began moving south of Krasnoyarsk in the early eighteenth
century, provoking attacks by Zunghar troops. Both sides competed for the
exaction of fur tribute (iasak)from native Siberian tribes. In 1710 the
Zunghars destroyed the Russian fortress of Bakan between the Biya and
Katun rivers, attacked Baraba and Kuznetsk, and exacted tribute from the
inhabitants of Barabinsk.
7Peter I in 1713 ordered Matvei Gagarin, gover-
nor of Siberia, to negotiate an end to these conflicts. Tsewang Rabdan,
however, consistently refused to recognize the Russians’ rights to this terri-
tory, and ordered them to destroy their fortresses at Krasnoyarsk, Kuz-
netsk, and Tomsk. Dual payments of iasakcontinued to be an issue be-
tween the Russians and Zunghars until the end of the Zunghar state. The
Russians could never exact unquestioned and exclusive revenue collection
rights in Siberia until their Mongol neighbor had been eliminated by the
Qing. But tensions were greatly heightened when reports came to the Tsar in
1713 of the discovery of gold in Zungharia.
8Governor Gagarin had pur-
chased samples from “Bukharans,” or Turkic traders, who told of huge de-
posits in “Eskel” on the Amu Darya River. (These reports probably referred
imperial overreach and zunghar survival 211

to Yarkand, which is not on the Amu Darya.) Tsar Peter, whose motto was
“Gold is the heart of the state,” was desperate to fund his Great Northern
War with Sweden. He ordered an expedition to travel up the Irtysh River
and to build fortresses along it and at lakes Balkash, Yamysh, and Zaisang,
in the heart of Zunghar territory. The expedition would then attempt to
capture Yarkand and investigate the sources of gold in the region. Led by
Lieutenant Colonel Ivan Buchholz, a former Captain of the Guard, the
army of two thousand men departed Tobolsk in July 1715, arriving in Oc-
tober at Lake Yamysh, where they built their first fortress. They took with
them a contingent of Swedish artillery and mining engineers, including
Lieutenant J. G. Renat, who would contribute substantially to the Zunghar
economy after his capture.
9Gagarin assured Tsewang Rabdan that the
Russians had only friendly intentions, and that they would support Tse-
wang Rabdan if he did not interfere. But Tsewang Rabdan’s brother and
best general, Tsering Dondub (Ch. Celing Dunduobu), besieged the Rus-
sians with ten thousand men; they stormed and destroyed the fortress on
February 9, 1716. Weakened by starvation and pestilence, the hapless Rus-
sians had to withdraw up the Irtysh River, where they built the more defen-
sible fortress of Omsk. Another Russian envoy had meanwhile returned to St. Petersburg with
samples of gold collected around the great “Blue Lake” (Kokonor), fur-
ther exciting Russian greed. The Russians persisted in exploration and for-
tress construction, rebuilding the fortress at Yamysh and building Semi-
palatinsk in 1718. Although Tsewang Rabdan objected, the Russians knew
that he was concerned with fending off attacks by the Kirghiz–Kazakhs and
holding the oases of Hami and Turfan against Chinese pressure, so they
paid him little attention. In 1719 Major Ivan Mikhailovich Likharev again
penetrated Zunghar lands, building a fortress on Lake Zaisang. At first the
Zunghar army feared a Russian–Chinese alliance and refused to attack, but
when they learned that there was no alliance, the Zunghars attacked in
force with twenty thousand men. Using heavy cannon, Likharev held them
off with his small force until a truce was agreed upon. Tsewang Rabdan’s
losses at Hami and Turfan, and his debacle in Tibet, meant that he could
not risk war with Russia. Fearing a Qing invasion, he had to allow the con-
struction of the fortress at Ust-Kamenogorsk in return for Russian with-
drawal and promises of aid against the Manchus.
10 The gold expeditions
ended without any success: the lure of El Dorado in Turkestan proved to be
a will-o’-the-wisp (until the oil discoveries of the twentieth century). But
the haunting possibility of gold discoveries gave the Zunghars one more
chance to win the Russians to their side. Tsewang Rabdan had accommodated Russian fortifications in the hopes
212 contending for power

of gaining an ally against the Qing, but he had no obvious mineral wealth
to offer that could compare with the Beijing fur trade. The Manchu envoy
Tulisen had succeeded in persuading the Russians not to support Tsewang
Rabdan; and Tsewang Rabdan would not submit as a vassal either to
China or to Russia. He stubbornly and successfully maintained his people’s
autonomy throughout his reign.
Three Central Eurasian Travelers
With most military action suspended, diplomatic activity flourished. Qing
resources had been exhausted by the difficult military campaigns against
Galdan, and the emperors were not inclined to launch new ones immedi-
ately. Tsewang Rabdan aimed to shore up his defenses during this lull by in-
vestigating opportunities for military aid from Russia in return for gold ex-
ploration. After Russia stabilized its border in 1689, the fur caravans sent
from Siberia to Beijing in the next two decades were extremely profitable
for the Russian state. Russia expected even more lucrative prospects from
the Kiakhta treaty of 1727.
11 All sides engaged in mutual spying, testing,
and exploration. Three travelers, each of different nationalities, left fascinating accounts
that illuminate this period of cross-cultural engagement. Best known to
Chinese historians is the Manchu envoy Tulisen (1667–1741), who was
sent on a mission to Ayuki Khan of the Torghuts from 1712 to 1715. He
wrote his report on the mission in Manchu, and a somewhat different
Chinese version appeared later. Shortly thereafter, the Scotsman John Bell
crossed Central Eurasia in the opposite direction on a mission from the
Tsar; and the Russian Ivan Unkovskii undertook a mission to Tsewang
Rabdan’s headquarters from 1722 to 1724. These accounts, in addition to
providing valuable geographical information, show how three Central Eur-
asians—a Russian, a Mongol, and a foreign servant of the Tsar—viewed
the Qing state. Each of the travelers acted in both a personal and an official capacity.
Bell traveled as a member of a Russian embassy led by Leon Vasilievich
Izmailov, whose main purpose was to obtain information about China’s
military capabilities and prospects for commercial exchange. Tulisen’s os-
tensible mission was to facilitate the return of Ayuki’s nephew Arabjur
from the Chinese frontier to the new Torghut homeland on the Volga River.
He expected Ayuki Khan to propose a joint expedition against the Zunghar
state, but he was instructed to decline any such alliance. Ayuki, dissatisfied
with the burdens of Russian servitude, told Tulisen that he felt closer to the
imperial overreach and zunghar survival 213

Manchu rulers than to the Russians, and asked for Qing intervention to im-
prove relations with Russia and the Mongols. Tulisen also took the oppor-
tunity while he journeyed through Russian territory to check out the em-
pire’s commercial and strategic resources, and to prepare for a possible
invitation to meet the Tsar. Tulisen and John Bell crossed paths at Selen-
ginsk, and Tulisen later escorted Bell from the frontier to Beijing. Coming
from opposite ends of the continent, the two envoys had much in common.
Each was a representative of a people on the edge of a great new empire in
formation, and each made a successful career out of serving the empire’s in-
By contrast, Unkovskii represented an effort by the Russians to explore
opportunities for alliance with the Zunghar Mongols against the Qing em-
pire. Tsewang Rabdan considered becoming a vassal of the Tsar like Ayuki,
and he offered the Russians access to suspected gold sands in the deserts of
Turkestan in return for their support in the form of a twenty thousand–man
army and weapons to defend him against Qing attacks. Each of the three texts represents a particular combination of state inter-
est and individual responses to the growing Qing power in Central Eurasia.
Each report mixes the genres of travel, espionage, and diplomatic litera-
ture. They combine exploration, geographical surveys, intelligence gather-
ing, diplomatic interviews, and personal reactions to the landscape. In four
different languages (Manchu, Chinese, Russian, and English), they provide
four different perspectives on the nature of China’s expansion on its north-
western frontier. Tulisen’s and Unkovskii’s writings were in the guise of of-
ficial reports to be presented to their sovereigns, but the texts themselves
have the character of personal journals. Bell’s Travelswas a private ac-
count, not published until nearly fifty years after his return.
During the early seventeenth century the Torghuts had migrated thousands
of miles across the steppe to the lower Volga, where they were accepted into
the service of the Russian Tsar.
13 They maintained kinship ties with the
Zunghars but never consented to be put under their dominion. Their rela-
tions with the Russians were also burdensome. The Tsars exempted them
from taxes but compelled them to join in military campaigns. Of all the
Mongols, they had migrated the longest distances to find open living space,
but even they could not escape the reach of the agrarian empires surround-
ing them. They also maintained contacts with Tibet, preserving their Bud-
dhist faith. Known as the Kalmyks, they remain today as an autonomous
republic in the Russian federation. They represent the westernmost exten-
sion of Buddhism in Eurasia.
214 contending for power

The pretext for Tulisen’s mission was closely tied to the growing links be-
tween the Qing, the Zunghars, Russia, and Tibet.
14Arabjur, the nephew of
Ayuki, Khan of the Torghuts on the Volga (r. 1673–1724), had traveled to
Tibet on a pilgrimage sometime between 1698 and 1703, but found himself
unable to return to the Volga because of wars between Ayuki and Tsewang
Rabdan. The Kangxi emperor gave him refuge, allowing him pastures in
the northwest. Finally, in 1712, Ayuki demanded the return of Arabjur just
at the time when a Russian caravan was setting out from Beijing. This was
an ideal opportunity for the Qing emperor to use his Russian connections
to establish a link with the distant Torghuts, so he asked the Russians to es-
cort five imperial envoys with a message to Ayuki to discuss Arabjur’s re-
turn. Tulisen was not the head of the mission, but he was the only one to
write an account of it. As one of the most extended pieces of Manchu writ-
ing about Central Eurasia, it reveals critical aspects of Qing policy toward
the region. Arabjur, however, did not join the embassy, and ultimately he
was never sent back. Qing objectives were grander than the return of one
Mongol to his tribe. The imperial instructions to Tulisen are a masterpiece of Qing ideologi-
cal dissimulation. Kangxi directed a special edict to Ayuki, responding to
his sincere presentation of tribute, expressing the emperor’s wish to have
Arabjur reunited with his uncle. Knowing that Ayuki and Tsewang Rabdan
were on bad terms, however, he expected Ayuki to propose a joint alliance
against Tsewang Rabdan. He gave Tulisen specific instructions on how to
If Ayuki says, “Let us, having joined together, strive to attack Tse-
wang Rabdan in a pincer movement from both sides,” you should
definitely not say anything. Say only, “Tsewang Rabdan is [on very]
good [terms with] the Great Khan [Kangxi]. He has been sending
countless emissaries to ask after the Khan’s health. The Great Khan
likewise has been moved to bestow grace in the same manner. Al-
though one might think that his strength is weak, that he is in need, or
that he is exhausted, there are certainly no circumstances under which
our Sage Lord will attack him. But this matter is great. If we envoys en-
tertain this suggestion it would not be right. Although you have re-
quested us to speak to our Sage Lord, we realize that our Lord says,
‘May all living beings under Heaven live in peace and well-being.’ We
can strongly affirm that the emperor has no intention of disturbing
Tsewang Rabdan.”
In other words, the envoys were to assert that the emperor had no intention
of destroying Tsewang Rabdan, and to reject an explicit alliance against
imperial overreach and zunghar survival 215

him. Even though Kangxi had repeatedly called for the extermination of
Galdan and his Zunghar state, he now claimed to have renounced ambi-
tious efforts to intervene in the distant steppes. The Torghuts themselves
saw this embassy quite differently.At the same time, the envoys were instructed to respond positively to any
invitations to meet with the Russian Tsar. They were also told to be very
flexible in adapting to Russian customs:
[If the Tsar sends you an envoy] it is all right to meet him following the
laws of his country. Say to the man that he sends, “Previously when
Nikolai of your country came to our country, he behaved badly. We
are not like that”...After you have met the Tsar, if he should ask,
“What does your country esteem?” answer only: “Our way of life
takes loyalty, filial piety, benevolence, righteousness, and faith as its
chief principles. We follow them seriously, both in ruling the country
and in our personal conduct. In matters pertaining to life and death,
we have nothing to fear. We simply say, ‘If we die we die’...Because
our country puts first such principles as loyalty, [etc.], we have no wars
and no heavy punishments. For many years, we have lived in well-be-
ing and peace.”
This embassy provided the emperor with an opportunity to spread an
idealized version of his rule to new peoples with whom the expanding em-
pire had come in contact. For external consumption, imperial rule repre-
sented peace, loyalty, and benevolence. The emperor did not boast of mili-
tary conquests to the Russians, as he did to his own subjects, nor did he
speak openly of diplomatic alliances or trade. But invoking his desire for
peace also included a promise to respond to threats: “If Russia dispatched
troops to its borders, we might become suspicious and send our troops
there. Our two countries have lived in harmony for a long time. We intend
to preserve this relationship. Yet ‘once troops have been sent to the frontier
for use, they will be used. Do not have any doubts.’” His emissaries, how-
ever, should stress the difficulty of any major armed confrontation:
I believe the Russians will certainly ask about what types of cannon we
have. If they do ask, say: “The distance is very far; it would be very dif-
ficult to bring them to the border. Along the way, there are very many
mountains, cliffs, woods, dense forests, precipices, and narrow passes.
There is certainly no way for us to bring them there. We do not per-
ceive a need for it, and to achieve it would be very difficult. Thus, by
our country’s laws, these kinds of things absolutely could not occur at
216 contending for power

the frontier. Our prohibitions are very strict. And even if our Sage Lord
ordered it, he certainly could not attain it.”
The envoys were also told that Russian customs were degenerate and
corrupt, and that they should avoid any excessive drinking or disorderly
behavior. Most of the imperial instructions were intended to present a
peaceful, disciplined image to these unsophisticated people. At the same
time, the emissaries were to gather extensive information about the country
through which they traveled. Tulisen, on his return, prepared a detailed
map of Siberia for the emperor which was at least equal if not superior to
the maps prepared by Russian and Western surveyors.
Tulisen provided his own intriguing insights into Russian attitudes. In
his account, Siberian Governor Gagarin praises Kangxi’s benevolence and
draws invidious comparisons with his own Tsar:
[Gagarin] then said—“Your emperor of China is indeed a most excel-
lent and most divine personage; while he is thus occupied in promoting
the prosperity and riches of his empire, and in preserving on all sides
the blessings of peace, your Excellencies may no doubt happily and un-
interruptedly follow your respective pursuits...Inthis empire also,
while the late Chahan Khan [Peter I] lived, we were free from labor
and care. In his reign, all men, whether of high or low estate, rested in
peace . . . But latterly, for these twenty years past, our empire has been
engaged in incessant wars; and to this day, we are still fighting and
contending without any respite...China is at present the only empire
which enjoys any peace or tranquility. Our present Chahan Khan, even
when he was yet a child, was always fighting and contending with the
children who were his playmates. Those children are now become gen-
erals in his armies. We should have still been at rest at this time, as
heretofore, if he had only followed the steps of his father.”
Gagarin is made to say that Peter the Great, who did indeed practice mil-
itary drills as a child, compares unfavorably with Kangxi, whose farsighted
benevolence achieves peace without aggressive military action. Is this the
genuine view of a Siberian governor far from Moscow, or is Tulisen putting
words in his mouth? Gagarin may well have viewed Peter’s engagement in
European wars as a distraction from his task of defending Siberia and nego-
tiating favorable terms with China. China, after all, produced considerably
more wealth for Russia’s treasury by its purchases of Siberian furs than did
the nations of Europe. It is not impossible that the views of Tulisen and
the Siberian governor converged, even though Tulisen expresses Gagarin’s
imperial overreach and zunghar survival 217

views in purely orthodox Qing rhetorical style. But it seems unlikely that
Gagarin would criticize his Tsar openly in Tulisen’s presence. In any event,
Tulisen, like John Bell, draws on Russian testimony to support the image of
the kindly, peaceful Qing emperor.After Tulisen returned to Selenginsk, he dispatched a letter to Gagarin
with a very different message. Tsewang Rabdan had by then begun his at-
tacks on Hami. The letter explained that Tsewang Rabdan used to be a
loyal tributary, who was treated “with great kindness and compassion,”
but “the disposition of Tsewang Rabdan is by nature deceitful and false
[Ma. banitai koimali holo]. His essence being thus, he definitely cannot
change. Inquiring into Tsewang Rabdan’s crimes, we have sent [an army] to
make war against him.”
20 Tulisen urged the Russians to seize Tsewang
Rabdan and his followers if they crossed the border in an attempt to gain
refuge from the Qing army.
Thus in the Qing view, Tsewang Rabdan, by rejecting imperial benevo-
lence, had removed himself from any reach of pardon and deserved to
be eliminated. The Chinese text referring to Qing action uses the term
zhengjiao (a righteous campaign of extermination), whereas the Manchu
text uses only dailambi(to make war on).
22“Extermination” was the term
repeatedly used by Qing rulers facing Zunghar leaders who insisted on
their autonomy. Kangxi used metaphors of “root and branch” elimination
against Galdan, and Qianlong incessantly harped on the need to extermi-
nate the last vestiges of the Zunghar state. The carefully constructed image
of Chinese benevolence was used to persuade the Russians to refuse shelter
to refugee Mongols. Tulisen was also prepared to answer questions from Ayuki Khan. The
Khan asked many questions about the Qing, especially the Manchu rulers
and their activities. He knew that the emperor often went hunting north of
the capital, and that he had a separate residence in Rehe, as well as an ad-
ministrative center in Mukden. He also knew that the Manchu script was
derived from the Mongol script. Concluding that Manchus and Mongols
were so similar that they must have had a “common origin” (Ma. emu
adali; Ch.tongyuan), he wondered how they had come to separate from
each other. Tulisen promised to obtain more information from the emperor
on his return. In Tulisen’s report Ayuki stated that although the Torghuts
were a “people of an outer state” (Ma. tulergi gurun-i niyalma;Ch.waidi),
in form and clothing they were extremely similar to the people of the Cen-
tral Kingdom (Ma. dulimba-i gurun;Ch.zhongguo) yet completely differ-
ent from the Russians.
23 Ayuki clearly had in mind the Manchus, not the
Han Chinese, when he stressed the similarities between the two peoples.
The Qing rulers also could offer Ayuki closer ties with Tibet, because they
218 contending for power

had frequent envoys from the Dalai Lama. Ayuki was especially interested
in obtaining medicines from Tibet, access to which was now blocked by the
Russians. Thus Tulisen could effectively use the multicultural reach of the
Qing state, embracing Tibetans, Manchus, and Mongols, to draw another
Mongol Khan over to his side.Tulisen’s report, originally written in Manchu, with the title Lakcaha
Jecende takûraha ejehe bithe (Jottings on the places where one sent me in
the cut-off frontiers [outside the empire]), was published in Chinese in
1723 as Yiyulu(Record of strange regions). It was so fascinating that Euro-
peans soon translated it into several languages. A French translation ap-
peared in 1726, a German and two Russian translations in the late eigh-
teenth century, and a (rather inaccurate) English translation by Sir George
Staunton in 1821. European observers, then and now, describe Tulisen’s
mission as an extraordinary event, for China, in their view, almost never
sent its official representatives out of the empire. Their parochial focus
blinded them to the empire’s many foreign contacts preceding the European
encounter. Seen in a Central Eurasian context, Tulisen’s embassy was only
one in a long lineage of exploratory expeditions, including Han Wudi’s ex-
peditions to obtain the blood-sweating horses of Ferghana, Zhang Qian’s
mission to the nomads in the second century bce, or the Tang emperors’
dispatches of monks to India. Obtaining information about the peoples of
Central Eurasia, especially new tribes who appeared on the borders, had al-
ways been a significant goal of those empires which, like the Han and Tang,
looked outward to the northwest. The Qing embassy set out at a time when
other world empires, such as the British and the Russians, were also send-
ing their envoys, merchants, and travelers on missions of exploration. Like
their contemporaries, the Qing emissaries combined strategic, geograph-
ical, and commercial objectives. The exact objective of the embassy is still unclear. Did the emperor hope
to conclude a military alliance with Ayuki against Tsewang Rabdan? Since
the Qing had previously allied with Tsewang Rabdan against Galdan, it is
certainly plausible that they would once again try to play off one barbarian
against another. Even though imperial instructions directed Tulisen to re-
ject initially any such proposal by Ayuki, this was only the opening negoti-
ating position. If Ayuki really had military strength to offer, the emperor
could have changed his point of view. The alliance with Tsewang Rabdan
itself was kept very well concealed. The Torghuts themselves saw a Qing
alliance as the purpose of the mission, and Tulisen’s envoy to Beijing may
in fact have proposed such an alliance when he requested the return of
24 The view of the Russian envoy Glazunov in 1730 was that
Kangxi’s goal was to induce the Torghuts to return to Qing territory. In
imperial overreach and zunghar survival 219

1712, however, Tsewang Rabdan had not yet attacked Qing territory, and
Kangxi had no intention of launching an army against him. The Torghuts
did come back to the Qing frontiers in the 1770s, but no such promises
were made at the time of Tulisen’s visit; and, as we shall see, whether or
not to receive the long-departed Mongolians was a contentious issue in
Qianlong’s reign.More important than establishing an immediate alliance was the gather-
ing of information. Within Tulisen’s report, Russia and eastern Mongolia
feature more prominently than Ayuki Khan himself. The court was willing
to make many concessions to Russian customs in order to obtain a meeting
with the Tsar. Tulisen provided the Qing court with much valuable infor-
mation about Siberia. His report, one of the great pieces of travel literature
of the time, offers astute observations on geography, natural environment,
and social customs. It is both an imperial ethnography and an intelligence
document, serving to guide foreign policy and to inform the Qing rulers
about their international environment. Strategic concerns about Russian
and Zunghar intentions motivated this distant expedition, and the report
itself broadened the horizons of the Manchu elite. John Bell performed a
similar task for his Russian masters with his travels eastward across Cen-
tral Eurasia from 1716 to 1720.
John Bell (1691–1780) left home in 1714 to serve the Russian Tsar, like
so many of his fellow Scots. In 1719 he set out from St. Petersburg for
Beijing, passing through Siberia and the territory of the Buriat and Khalkha
Mongols. Many years later, in 1763, he published his Travels from St. Pe-
tersburg in Russia to Diverse Parts of Asia, described by a reviewer in 1817
as “the best model perhaps for travel-writing in the English language.”
One part of Bell’s Travels,the “Journey from St. Petersburg to Pekin,” be-
gins with his departure from the Russian capital and ends with his arrival at
the Qing court. Along the way he describes in great detail the topography,
flora and fauna, and peoples across Central Eurasia, mingling a personal
diary with discussions of politics and natural history. In addition to his curiosity about the customs and local histories of the
places through which he passed, Bell had very perceptive insights into the
crucial role of logistics for mastery of the steppe. His primary sources of in-
formation were Russian. He learned most about steppe peoples “from an
ingenious and penetrating gentleman, who fills a public office in this place
[Tobolsk], and was employed in several messages to him from the late gov-
ernor of Siberia.”
26 His comments about relations between the Qing and
the Mongols, however, also closely reflect the Qing official point of view.
220 contending for power

After his arrival at Selenginsk on May 29, 1720, Bell noted:
The Mongalls are a numerous people, and occupy a large extent of
country, from this place to the Kallgan, which signifies the Everlasting
Wall, or the great Wall of China...Onemayeasily imagine, from the
vast track of land which the Mongalls occupy, that they must be very
numerous; especially, when it is considered, that they live in a healthy
climate, and have been engaged in no wars, since they were conquered,
partly by the Russians on the west, and partly by the Chinese on the
east; to whom all these people are now tributaries. In former times the
Mongalls were troublesome neighbors to the Chinese, against whose
incursions the great wall was built.Kamhi [Kangxi], the present emperor of China, was the first who
subdued these hardy Tartars; which he effected more by kind usage
and humanity than by his sword; for these people are great lovers of
liberty. The same gentle treatment hath been observed by the Russians,
towards those of them who are their subjects. And they themselves
confess, that, under the protection of these two mighty Emperors, they
enjoy more liberty, and live more at ease, than they formerly did under
their own princes.
Bell’s “Mongalls” are the Khalkhas, or Eastern Mongols, particularly
those under the leadership of the Tüsiyetü Khan. The gratitude of the Khal-
khas toward Kangxi stemmed from his intervention in the dispute between
the Jasaktu Khan and Tüsiyetü Khan, which had provoked Galdan’s inva-
sion. Kangxi’s “kind usage” at first meant sheltering refugees from the in-
ternecine Mongolian battles within Chinese borders and offering famine re-
lief to starving Mongols. In return, however, the Khalkhas were enrolled
into banners, their pasturelands were carefully delimited, the succession to
the Khanship and other ranks was put under Qing supervision, and they
had to provide levies of horses to support Qing campaigns. Russia’s rule of
its subject Mongols was arguably less systematic but more onerous. Trib-
ute exactions by Siberian governors and Cossacks were more capricious,
but Mongol “liberties” were not so circumscribed. The new “liberty” of
the Mongols really meant freedom from the dangers of war, while they
lost their customary freedoms to migrate and pasture where they pleased.
The border treaty between Russia and China prevented them from fleeing
across the vaguely demarcated frontier to escape excess exactions by one
side or the other. Some, like Gantimur and the Torghuts, escaped these con-
straints, but most were bound by them. Bell’s comment reflects a view of Mongol–Chinese and Mongol–Russian
relations that suited both the settled empires and certain elements of the
imperial overreach and zunghar survival 221

Mongol elite, but it tells less than the full story. In 1719–20, when he wrote
his account, the Kangxi emperor in his last years had indeed seemed to
achieve his goals of settling Mongolian disputes through “kind usage.” The
memory of his effort to eliminate the Zunghar state was now twenty years
old. Bell had absorbed both the Russian desire for peaceful commercial re-
lations and the self-image of the Qing as a benevolent, impartial arbiter in-
terested in peace among peoples.Bell also commented on the Qing wars with Tsewang Rabdan. After not-
ing his arrival at Tobolsk, on December 16, 1719, Bell wrote:
The Emperor of China was some time ago engaged in a war with the
Kontaysha about some frontier towns, of which the latter took posses-
sion, and maintained his claim with a strong army. The Emperor sent
against him an army of three hundred thousand men, under the com-
mand of his fourteenth son, who is reckoned the best general of all
his children. Notwithstanding their superiority in numbers, the Kon-
taysha defeated the Chinese in several actions. The Emperor thought it
best to accommodate the difference, and a peace was concluded to the
satisfaction of both parties.
This passage refers to Tsewang Rabdan’s attempted invasion of Hami
in 1715.
29 Yinti, the Kangxi’s emperor’s fourteenth son, was designated
Fuyuan Jiangjun and sent with a large force (but not 300,000 men) to
Gansu and Mongolia to deter Tsewang Rabdan. There were no major bat-
tles, but the Qing did recover Hami and Turfan, and a truce was later ne-
gotiated. The source of Bell’s version is rather unclear, but “Kontaysha,”
derived from “Hongtaiji,” is the regular Russian title for the Zunghar rul-
ers. Presumably, this was an account given to Bell by Russian officials in
Tobolsk, and reflects their best understanding of the relations between the
Qing and Zunghars. As texts discussed later also indicate, the Russians
tended to look on the Qing military forces as being of poor quality, but
they also gave the Qing emperor credit for being farsighted and benevolent.
This emerging image of a rather weak but benevolent regime in Beijing
suited the Russian aims of establishing diplomatic relations based primarily
on trade. Bell further notes:
It must be observed, that the Chinese, being obliged to undertake a
long and difficult march, through a desert and barren country, lying
westward of the long wall; being also encumbered with artillery, and
heavy carriages containing provisions for the whole army during their
222 contending for power

march; had their force greatly diminished before they reached the en-
emy. The Kontaysha, on the other hand, having intelligence of the
great army coming against him, waited patiently on his own frontiers,
till the enemy was within a few days march of his camp, when he sent
out detachments of light horse to set fire to the grass, and lay waste the
country. He also distracted them, day and night, with repeated alarms,
which together with want of provisions, obliged them to retire with
considerable loss.
This comment must refer to the battle of the Manchu commanders
Seleng and Elunte with Tsewang Rabdan’s forces at the Kara-Usu River on
October 5, 1718, in which the Qing forces, consisting of two thousand
Green Standard troops, ten thousand local “chieftain troops” (tusi),and a
small number of Manchu troops, were surrounded and crushed.
31 It was
only after this stunning defeat that Kangxi designated his fourteenth son to
lead the army. It is a very perceptive observation about the nature of war-
fare between the Qing and its nomadic rivals, or in fact about warfare be-
tween settled agrarian regimes and nomads in general. Similar withdrawal
and scorched earth tactics were used by the Crimean Tartars against the
Russian empire’s expansion to the south, and by the Parthians against the
Roman empire. Bell comments: “This method of carrying on war, by wast-
ing the country, is very ancient among the Tartars, and practiced by all of
them from the Danube eastward. This circumstance renders them a dread-
ful enemy to regular troops, who must thereby be deprived of all subsis-
tence, while the Tartars, having always many spare horses to kill and eat,
are at no loss for provisions.”
32As the discussion of Kangxi’s Galdan cam-
paigns has indicated, Russians and Chinese both had to solve this funda-
mental logistical problem of the steppe. Russian armies in the Crimea cre-
ated tabory, or “moving fortresses,” grain wagons that were surrounded by
defensive troops, which trundled forward as the Tartars retreated.
33 Only
by the mid-eighteenth century were the Chinese, over a much vaster terrain,
able to create a chain of magazine posts stretching from the Gansu border
out into Xinjiang. Bell perceived both the general problem of logistics in the
steppe and the way nomads deployed logistical attacks in battle.
Artillery Captain Ivan Unkovskii’s mission to Zungharia in 1722–1724
was the last futile Russian attempt to induce the Zunghar Khan into sub-
mission. Despite the failure of the Buchholz and Likharev expeditions, Tsar
Peter maintained his drive to find gold in Zungharia. Tsewang Rabdan had
imperial overreach and zunghar survival 223

driven off two Russian armies, but he still expressed interest in an alliance
with Russia against China. Unkovskii was sent to continue negotiations.
He left Moscow in February 1722 but only arrived at Tsewang Rabdan’s
camp in November, just after the death of the Kangxi emperor. Tsewang
Rabdan now took a harder position against Russia, launching further at-
tacks against the Kazakhs and reiterating demands for the destruction of
Russian fortresses. Unkovskii left in September 1723 with his mission
unfulfilled. He did, however, collect a great deal of valuable information
about Zungharia. His journal entries are extremely revealing; they provide
an almost unique account of personal dialogue with a Central Eurasian
On December 11, 1722, a high-ranking noble came from Tsewang
Rabdan’s camp to meet Unkovskii with a list of questions from the Khan.
The journal lists the questions submitted by Tsewang Rabdan’s ministers
(zaisang) side by side with Unkovskii’s replies. This exchange offers inter-
esting indications of the different perspectives of both sides about China:
Zaisang: Cheredov told [the Khan] that fortress towns [gorody] were built
by Russia along the Irtysh in case the Tsar wanted to make war on China.
Unkovskii: Cheredov was not authorized to say that; the towns were built
not for war but to search for ore...
Zaisang: Cheredov spoke of a search for golden ore.
Unkovskii: And I was ordered to ask permission of the Kontaisha [for this
search]. And if the search for gold and silver succeeded, the gains for
you would be large, as I explained in detail to the Kontaisha [Tsewang
Zaisang: The Kontaisha asks that the Mongols submit themselves to the
beneficent protection of your Imperial Highness, as Ayuki Khan submit-
ted; and we would rejoice at this, and request an army of twenty thou-
sand men [to be used] against the Chinese Khan, and nothing else.
Unkovskii: On this subject I explained to the Kontaisha in detail that when
he committed himself to written negotiations, as were done with Ayuki
Khan, then His Imperial Highness would defend you, as his subjects,
against your enemies; but he would first attempt to persuade the Chi-
nese Khan by his orders to commit no injuries against you, and if the
Chinese Khan did not listen, then he would find ways of bringing support
to you.
This dialogue has a refreshing directness seldom found in the official dip-
lomatic exchanges. Clearly the Zunghars expected Russian military aid
against China in return for their submission to the Tsar. They even asked
224 contending for power

for the Russians to help them subdue the Khalkha Mongols, just as the
Russians had helped Ayuki gain dominance over his neighbors. By invoking
the oral discussions with the previous envoy, Ivan Cheredov, they tried to
tie down the Russians to definite commitments of aid. Unkovskii, however,
avoided promising direct support, deferring the question to negotiations.
Tsewang Rabdan did not know of Ayuki’s dissatisfaction with Russian sub-
jection, which he had revealed to Tulisen, nor did he know of Ayuki’s pro-
posal to ally with the Qing. Ultimately, Tsewang Rabdan refused to become
a Russian subject when he learned of the death of the Kangxi emperor,
and when he realized that it would weaken his position to accept Russian
troops on Zunghar soil. He told Unkovskii that he had received envoys
from the Yongzheng emperor offering peaceful relations, and he had also
received emissaries from the Khalkha Mongols and the Khoshots. He felt
that China was now in a weakened position. Perhaps Tsewang’s victory
over Qing armies in 1718 and the truce settlement in 1722 had led him to
overestimate his strength and misunderstand the degree of his isolation.
Tsewang Rabdan, however, displayed curiosity about geopolitical rela-
tions across the Eurasian continent. He asked Unkovskii repeatedly about
Peter the Great’s fleet, his wars with the Turks and Swedes, the nature of
Russian religious beliefs, and whether the Russians drank tea. In a second
private dialogue with Unkovksii, he tried to learn about the power of
Kontaisha: The Chinese boast that no one is stronger and braver than them,
and all peoples bring them tribute [dan’].
Unkovskii: I hope you will not take it amiss if I say this, but His Imperial
Highness ordered me to bring various things for you: What do you con-
sider them to be: tribute or something else? He replied that the Tsar
sent gifts in gratitude for his [Tsewang Rabdan’s] beneficence, not as trib-
ute. I said it is just the same with the Chinese Khan, people send gifts
[podarki], not tribute [dan’].
Kontaisha: Whom do you consider to be stronger, the Turkish sultan or the
Chinese Khan?
Unkovskii: We consider the Turks to be braver than the Chinese, and the
Chinese behave poorly in military actions. After that I told him that
among all peoples there are bad characters [who bring unreliable reports
that one should not believe].
This discussion of the nature of gifts presented by one sovereign to an-
other is reminiscent of the issue faced by George Macartney at the end of
the century. Should goods offered to a sovereign be regarded as “tribute,”
imperial overreach and zunghar survival 225

indicating the willing subjection of the presenter to the recipient, or as
“gifts,” simply indicating an expression of friendship? It is a particularly
complex question in this Central Eurasian example because of the three-
way ties among Russians, Mongols, and Chinese. Izmailov’s embassy to
Beijing, of which John Bell was a part, did perform theketouaccording to
Qing ritual, and did not openly contest the Qing description of Sino-Rus-
sian trade as “tribute.” Even though the Russians had signed a treaty with
China providing for exchanges of goods under the guise of tribute, they did
not want to indicate to the Zunghar Mongols that they regarded them-
selves in any way as subordinate to the Chinese. Too close a relationship
with China would endanger their access to the golden sands they hoped to
find in Zungharia. So they portrayed their relationship, and that of all other
countries, with China under the guise of an amicable exchange of “gifts”
and not the hierarchical relationship of tribute. The Russian envoy took care to denigrate Chinese military power as
well. Since Tsewang Rabdan clearly knew of Chinese military might from
Galdan’s experience, and he probably also knew of China’s destruction of
the Russian fortress at Albazin, the Russians did not want to arouse suspi-
cions that they lacked either the capacity or the willingness to confront
China if necessary. On the one hand, Tsewang Rabdan may have been
gratified to hear that Russia did not consider China to be very strong. On
the other hand, he may have been misled into overconfidence about his
own chances of survival if he faced China on his own. He had recently won
victories against Qing armies, and the death of the Kangxi emperor might
lead to a less aggressive Qing policy. As a result, he refused to become a
Russian subject, and this decision effectively ended any prospects for Rus-
sian military aid. But the Russians made the right economic choice in the
end. The fur trade with Beijing proved more profitable than the illusory
golden sands of Turkestan. During his ten-month residence in Zungharia, Unkovskii enjoyed fre-
quent access to the Khan and his ministers, and he was invited to observe
religious and ritual processions and horse-riding competitions. Despite the
conflicting interests of the Tsar and the Zunghar Khan, the envoy could
meet Tsewang Rabdan in a “middle ground,” in Richard White’s term, a
“world of common contact and common meaning,” where mutual curios-
ity and geopolitical interest inspired the two cultures to learn more about
each other.
38Like Tulisen and Bell, Unkovskii reported factual information
to his sovereign that served state interests, but he gave detailed personal ac-
counts in his journal that go well beyond the restrictions of diplomatic pro-
tocol. For this brief period, Russia faced east as much as it faced west, while
the Qing relaxed its efforts to dominate the entire middle ground. By pro-
226 contending for power

moting the exchange of information across these vast spaces, these three
envoys helped to construct images of a common Eurasian world.
The Penetration of Turkestan and Tibet
During the late years of Kangxi’s reign, the Qing extended their control far-
ther west with major military campaigns against both the Zunghars and Ti-
bet. This expansion culminated in the expedition of 1720 to Lhasa. Qing
intervention in Tibet is often treated as a separate campaign, connected
with the internal politics of Tibet, but it was intimately linked to the goal of
exterminating the Zunghar state. It grew out of an unsuccessful effort by
Kangxi to eliminate the power of Tsewang Rabdan, Galdan’s successor.
Kangxi failed because Tsewang Rabdan was too far away, but he succeeded
in Tibet because of internal rivalries and because the Qing could gain access
to Tibet via Kokonor.Tsewang Rabdan avoided open conflict with the Qing until a schism oc-
curred in Tibet, instigated by Lazang Khan (1656?–1717) of the Khoshot
Mongols of Kokonor (Ch. Qinghai). Lazang (T. Lha bzang), the grandson
of Gush Khan, the great sponsor of the Tibetan lamas, killed his elder
brother and seized the Khanship in 1700. Supported by the Chinese, he
aimed to restore Khoshot supremacy over Tibet after Galdan’s failure. Ti-
bet became a vast ground of contention between these rival Mongol lead-
ers, but ultimately the rivalry played into the hands of the Qing.
Power in the Tibetan state had long been divided among different sects
of monastic Buddhists, each of whom depended heavily on support from
Mongol patrons. At the top of the state there was a three-way division of
power among the Dalai Lama, his Mongol patrons, and the temporal ad-
ministrator (T. sDe-pa,Ch.diba), who also served as regent when the Dalai
Lamas were in their infancy. In the mid-seventeenth century, Gusri Khan of
the Khoshot Mongols had installed the sect he patronized, the Gelupa (T.
dGe-lugs-pa), or “Yellow Hats,” in Lhasa, winning a victory over the rival
Karmapa, or Red Hat, sect.
40Gusri Khan supported the Dalai Lama, sup-
pressing rebellions against his power, allowing him to enforce discipline
in the monasteries and carry out a census of the tax-paying population. Af-
ter Gusri Khan’s death, the fifth Dalai Lama, Nag-dban-blo-bzan, became
stronger, as Gusri Khan’s successors constituted only weak rivals. The
Dalai Lama retired from secular authority, yielding power to his natural
son, the regent Sanggya Gyatso (T. Sangs-rgyas-rgya-mts’o). As discussed earlier, the sDe-paconcealed the death of the Dalai Lama in
1682 from the Qing and ruled by himself until the defeat of Galdan. He
imperial overreach and zunghar survival 227

aimed to establish a “really absolute government” that would centralize Ti-
bet and open it to the outside world. Many foreign visitors, including Indi-
ans, Chinese, Mongols, and Muslims, brought in gold, silver, cloth, and
other trade goods, stimulating economic growth. He used this revenue to
build roads and bridges and to support scholarship.
41As the regent turned
outward, Kangxi denounced him for his alliance with Galdan and his de-
fiance of the Dalai Lama’s instructions. In 1697 the sDe-pawas forced to
conciliate the furious Chinese emperor by officially subordinating himself
to the sixth Dalai Lama, Tshangs-dbyangs-rgya-mtsho.
42In fact, the regent
kept his power while allowing the young lama, described by the Jesuit
Ippolito Desideri as a “dissolute youth, addicted to every vice, thoroughly
depraved,” to indulge in wine, women, and the writing of love poetry.
The sDe-pa also resisted efforts to control Lhasa by the Khoshot and Khal-
kha Mongols, who were Galdan’s principal rivals. In 1705, however, Lazang Khan, with the support of Kangxi, invaded
Lhasa, killed the regent, and established himself as the primary ruler of
central Tibet, while uniting several kingdoms under the central power in
Lhasa. Under Lazang Khan, Tibet seemed to be regaining its unity and au-
tonomy. Lazang Khan, like the Kangxi emperor, welcomed foreign mission-
aries to his court, intimating that he might convert to Christianity, and
enormously impressing the Jesuit Desideri with his sophistication, generos-
ity, and curiosity.
44 Kangxi, equally impressed, granted Lazang the title of
Yifa Gongshun Khan, “The Law-Abiding Obedient Khan.” Blaming all the
disorder in Tibet on the usurping sDe-pa,the emperor feared that the sixth
Dalai Lama would succeed in reaching Tsewang Rabdan and inspiring a re-
newal of the Zunghar–Tibetan alliance. Supporting Lazang’s efforts to overthrow Zunghar influence by expelling
the sixth Dalai Lama, Kangxi told Lazang to seize the “false” (sixth) Dalai
Lama and present him to the court, but while under escort passing through
Kokonor, the Dalai Lama died. Although he may have died of illness, per-
sistent strong rumors suggested that he was murdered by Lazang’s men.
Lazang Khan placed his own candidate for Dalai Lama on the throne, but
the Tibetans were strongly loyal to their former spiritual ruler, who had
predicted that he would be reborn in Kokonor after his death. Soon reports
arrived that a young child had indeed been born, in Litang, on the eastern
border of Kham, in 1706, and that he was a Khubilghan, a living Buddha
incarnating the spirit of the former Dalai Lama. Lazang and his deputy
monks, denying that the child showed signs of being a true reincarnation,
forbade him to come to Lhasa, and the Chinese cooperated by keeping the
boy under arrest “for his own protection” in a fortress in Xining. (Note the
uncanny parallels to the conflict between rival Panchen Lamas in Tibet
228 contending for power

Meanwhile Tsewang Rabdan had begun to take an interest in Tibetan af-
fairs, responding to the new level of Chinese support for Lazang Khan and
to appeals from the Yellow Hat lamas, who hated Lazang Khan’s rule. His
attack on Hami in 1715 was intended to put pressure on the Chinese and
the Khoshot Mongols to keep them out of Tibet, but it also alerted Kangxi
to the encroaching dangers on China’s northwestern borders. As the Qing
began to prepare a large army to repel Tsewang Rabdan, in a surprising vic-
tory his army of two thousand men was driven away from Hami by the
Muslim begand some two hundred Qing troops. The begwas rewarded
with 15,000 taels from the Qing treasury. Qing rule now began to extend to
the oasis Muslims as well as the Eastern Mongols.
46The emperor immedi-
ately began to prepare a military expedition to “exterminate” (jiaomie)
Tsewang Rabdan just like Galdan, to “wipe out the evil so as to have eter-
nal peace.”
47 Expenses this time were estimated at about 3 to 4 million
taels to support three armies of ten thousand men each that would march
to Tsewang Rabdan’s headquarters in far western Mongolia. Tsewang Rab-
dan was estimated to have an army of forty thousand, plus the ten thou-
sand Torghuts he had captured from Ayuki Khan’s son. The Qing, for their
part, drew on fifteen thousand troops from their Khalkha Mongol allies, in
addition to the Manchu, Mongol, and Chinese banners. These expeditions attempted to repeat the pattern of the Galdan cam-
paigns, but conditions had changed substantially. The emperor was now
too old and ill to participate personally. Ever since he deposed his son
Yinreng for immoral behavior, in 1708 and again in 1712 he had suffered
repeated attacks of mortal illness; his hands shook too hard to allow him to
write, and he suffered fainting spells that impeded his decision making. Un-
able to decide on a successor, he repeatedly instructed his sons not to in-
trigue against one another, to no avail. At the same time, the quality of civil
and military administration declined. Corrupt superior officials extorted
funds from lower officials, who pressed on the people, inciting armed resis-
tance. During these military campaigns, Finance Ministry Vice President
Seertu, in charge of providing rations for the troops, embezzled official
funds, shortchanging Manchu, Mongol, and Han soldiers alike.
As before, Kangxi tried to lure Tsewang Rabdan closer to Beijing with
peace offers but threatened punishment if he withdrew. Now over sixty
years old, the emperor constantly recalled the bold decisions and great
successes of his prime twenty years earlier.
49 This time, however, the dis-
tances and supply problems were much greater than during the Galdan
campaigns. Hami lay far out the Silk Road, nearly 1,200 kilometers west of
Ulan Butong and over 500 kilometers northwest of Suzhou, the closest ma-
jor military supply base, and the garrison at Jiayuguan at the end of the
Great Wall. And Hami, ruled by a begloyal to the Qing, was only the east-
imperial overreach and zunghar survival 229

ern border of Tsewang Rabdan’s realm. The real target, in the most grandi-
ose plans, was Ürümchi, 500 kilometers further west, or even Ili and the
Irtysh River, far away in the Altai Mountains. The expansive ambitions of
the Qing generals, urged on by the emperor, led them to plan for stupen-
dous logistical and military feats. As it happened, they fell far short of their
dreams, but they did substantially strengthen supply routes, commerce, and
access to the region. These campaigns initiated the Qing penetration of the
great Turko-Mongolian region which would later come to be known as
Xinjiang (New Frontiers).Their first effort was to build a supply base at Barköl (Ch. Balikun), 100
kilometers north of Hami, to serve as a base for advances farther west. First
20,000 shiof grain were to be transported by horse cart, using three thou-
sand carts to carry supplies in twelve stages from Jiayuguan to Hami. But
by March 1716 it was clear that these supplies could not support the garri-
son, and five thousand troops had to return to Gansu. General Xizhu was
fired for failing to pay sufficient attention to supplies, even though he pro-
tested that “I only lead the troops; I have nothing to do with military sup-
Previous campaigns had divided responsibilities ethnically—Han gener-
als took charge of logistical support, and Manchu and Mongol generals
led troops—but now the roles blurred. Manchu generals had to learn mun-
dane details of grain transport, and several Han Chinese demonstrated
significant military talents. Nian Gengyao, in particular, of the Chinese
Bordered Yellow Banner, did an excellent job as Governor of Sichuan in su-
pervising supplies, leading him to be promoted to Governor-General in
charge of military affairs at Chengdu. Funingga (Ch. Funing’an), the Man-
chu general in charge of the Xinjiang campaigns, profited from his experi-
ence in his previous post as granary superintendent.
51 At the same time,
Qing troops as they advanced burned the grasslands to deprive Tsewang
Rabdan’s horsemen of their supplies. As John Bell had noted, this ruthless
tactic had been employed by nomadic warriors from Mongolia to the Cri-
mea, and both the Qing and the Russians now learned how to respond in
52In a few years, the Vice President of the Ministry of War, Li Xianfu,
could boast that his men were thoroughly familiar with grain transport to
the far-flung garrisons. Oases formerly isolated in the desert could now be
reached regularly by troops; the road from Hami to Barköl had become a
53 Li was praised as a “Hanren” who had learned not to be
afraid of the difficulties of supplying troops on the frontiers. In this case,
military cooperation promoted the unity of Han and Manchu. Other officials continued to criticize the expense and risks of these cam-
paigns, but the emperor would allow no objections. Liu Yinqu repeatedly
230 contending for power

warned against the suffering that transport to the frontier inflicted on both
men and horses. He claimed that three to four feet of snow blocked the
route from Ganzhou to Barköl, but he was fired for incompetence, and the
emperor claimed that he had lied about the difficulties.
54 After being re-
buked by the emperor several more times and condemned to death, he was
eventually pardoned and ordered to spend his time cultivating fields on the
frontier under military supervision. Shi Yide, the Gansu provincial Com-
mander-in-Chief (tidu),also faced execution, until he was pardoned, for de-
claring that it was a waste of money to spend 250,000 taels on supplying
these garrisons. The emperor replied, “He doesn’t understand that I will
spend several million taels if necessary.”
Despite all the planning, the limitations of the harsh environment re-
quired repeated postponements of a major effort. The campaign for 1715
was called off as the generals focused on building up supplies at Barköl.
Fearing renewed attacks by Tsewang Rabdan on Hami, they strengthened
their defenses. The next year they again put off any advances, as many ob-
jected to the difficulties of transport.
56Even the increasingly frustrated em-
peror had to admit that logistical obstacles could not yet be overcome. The Qing’s conception of its natural borders kept expanding as the ar-
mies marched west. Hami was now regarded as “no different from the inte-
rior” since it had repelled Tsewang Rabdan’s attacks and been fortified. By
April 1717, 8,500 troops at Barköl were prepared to advance to capture
Turfan. Then this vital oasis on the eastern edge of Turkestan would once
and for all be “entered on the country’s registers” (ru guojia bantu).Even
so, success was uncertain, and there were disturbing reports of “narrow-
minded people [xiaoren] who spread rumors and undermine morale.”
more modest objective, considered as a fallback, was not to attempt to hold
Turfan, but to advance so as to alarm Tsewang Rabdan’s supporters with
the army’s might in the hope of inducing the surrender of his Mongol allies.
A second army would head for Ürümchi with a similar objective. This plan
echoed Kangxi’s expedition to the Ordos, which achieved no direct military
objectives but greatly impressed the local Mongols. Turfan and Ürümchi,
however, unlike the Ordos, were oases in the desert 350 to 500 kilometers
west of Hami, and there were no pastoralists nearby to be impressed.
Both armies did advance far enough to run into Zunghar patrols, which
they defeated in two battles just east of the oases of Mulei and Pizhan. The
delighted emperor at first urged Funingga to proceed farther and track
down all the Zunghars he could find. Commander Furdan even planned for
a full-scale advance to Tsewang Rabdan’s headquarters as far as Ili or the
Irtysh River. But east of Turfan, three hundred Zunghars fought back, firing
their fowling pieces and killing an important Mongol prince before they
imperial overreach and zunghar survival 231

fled. At this point, word came that Tsewang Rabdan was preparing to in-
vade Tibet, so the Xinjiang expedition was called off. As with Galdan, it
was more important to defeat the Mongol leader personally than to capture
several remote oases.
Although the generals drew up plans to continue the “righteous extermi-
nation” the next year, the central strategic concerns shifted to Kokonor
and Tibet, making it difficult to carry them out. Finally, two years later,
Funingga defeated the Zunghars in two skirmishes, occupied Pizhan, and
marched his army to Turfan, where the Muslim begsurrendered. Further
campaigns deeper into Xinjiang were called off. Qing armies had now ex-
tended the empire’s reach into the Turkic Muslim areas beyond the Great
Wall. For the first time in one thousand years, armies from northeast and
North China had conquered one of the major oasis cities of Turkestan.
Outside Turfan, the abandoned Han and Tang garrison cities of Gaochang
and Jiaohe still stood, as they do today, to remind them of their predeces-
sors’ achievements.
Aggressive plans for frontier settlement evolved out of the demands of
supplying the campaigns. Military agrarian colonies (tuntian)had been a
time-honored practice since the Han dynasty.
61 Troops who tilled the soil
could make their garrisons self-sufficient while also establishing a perma-
nent presence on the frontier. Plans were made shortly after the capture of
Hami to develop settlements at Suzhou, Barköl, and Hami under military
supervision. Exiled criminals and soldiers were to be the first settlers from
the interior.
62Top officials received rewards for promoting land clearance.
In 1717 General Furdan reported that he had sown seed and dug canals,
and all the new crops were flourishing. After Qianlong’s conquests of Tur-
kestan in the eighteenth century, this program of land clearance and settle-
ment would be developed on a much larger scale. Qing generals also planned to build walled fortresses in the steppe at
Khobdo, Ulan Gumu, and other places in Khalkha territory. Here, too, ex-
iled criminals were set to work clearing land. Postal relay stations linked
these military settlements to the interior.
63Permanent settlements in Mon-
golia had begun with the conversion of the Mongols to Tibetan Buddhism
in the sixteenth century. The first towns grew up around the monastic es-
tablishment. These included Hohhot (Ch. Huhehaote), whose major con-
struction began in 1555, and Urga (modern Ulaan Baatar), the headquar-
ters of the leading Buddhist cleric of Mongolia since the early seventeenth
64 Qing military garrisons, containing from two hundred to one
232 contending for power

thousand troops, now became the second locus of settlement in pastoral
lands. The commanders searched for strategic sites in broad plains near
ample supplies of water, grass, and wood. The settlements guarded the
routes to the Altai and Outer Mongolia, and they served as base camps for
grain and weapon storage. These were large wooden enclosures, containing
spaces equivalent to two thousand rooms(jian).Construction workers sent
from the interior stayed for at least one year to build them. Gradually these
camps evolved from military and penal colonies into sites of civilian set-
tlement. In Kokonor as well, Qing troops began to construct walled fortresses
near Mongol pastoralists. After selecting a site to clear land, they left the re-
maining pastures for the Mongol tribes. They “drew a sharp border that
may not be crossed” to separate the settlers from the indigenous pastoral-
65The Kokonor Mongols, loyal to the Qing emperor, did not begrudge
small amounts of land to the new settlers, but an inexorable process of pen-
etration had begun. The construction of Qing fortresses in the steppe mirrored the earlier
Russian advance into Siberia, but the process of penetration differed
greatly. In both cases, garrisons built small fortified units dispersed across
alien territories, and military occupation paved the way for larger waves of
civilian agrarian and mercantile settlement. The Russians, however, had
spread first across the forest zone, avoiding the grasslands and desert, and
did not encounter substantial resistance until they ran into Manchu ter-
ritory. They extracted fur tribute from the tribal peoples without much
armed conflict. The Qing, by contrast, first had to confront the Mongols in
the grasslands with large military forces before they could secure the region
with a network of forts and settlers. When they began to settle, the sur-
rounding populations seemed to be both subdued and loyal. But the subse-
quent history of rebellion and conflict indicates that Qing penetration of
the pasturelands was still fragile. The influence of the Qing rulers in Tibet now depended on their protégé,
Lazang Khan, who had installed his own Dalai Lama in Lhasa. The Qing
also kept in reserve the new Khubilghan in Xining but held back from im-
posing a solution to the divided spiritual authority, awaiting Tsewang Rab-
dan’s actions. In 1716, having repelled Tsewang Rabdan from Hami, the
Qing rulers began shifting troops to Xining.
66 Although Tsewang Rabdan
was soon expected to enter Tibet, it remained unclear whether he would act
as Lazang Khan’s ally or his enemy. Lazang’s son had married Tsewang
Rabdan’s daughter, and Tsewang Rabdan promised to support Lazang mil-
itarily in exchange for gold to help him defend the Zunghar state against
the Russians and Kazakhs.
67But Lazang Khan had also accepted titles and
imperial overreach and zunghar survival 233

support from the Qing. Would he join with Tsewang Rabdan to attack
Kokonor or defend it on behalf of the Qing? Kangxi warned Lazang Khan
about Tsewang Rabdan’s treachery, but in either case Qing troops were
much too far from Lhasa to offer support. Many ministers advised a wait-
and-see attitude.In the summer of 1717, Tsewang Rabdan sent his best general, Tsering
Dondub, with ten thousand men against Lazang Khan. Tsering Dondub led
his troops into Lhasa on an extremely difficult march over “the highest
route in the world” through “absolutely barren regions,” forcing Lazang
Khan to barricade himself in the Potala Palace.
68 The Jesuit Ippolito
Desideri, who was in Lhasa at the time of the invasion, described Tsering
Dondub as “passionate, expert, and proud, daring, intrepid and warlike,”
even going so far as to compare him to Alexander the Great.
69 Shi Yide
noted that Tsewang Rabdan’s general had led three thousand men from
Yarkand and Kashgar into Tibet, crossing “three great snowy mountains
. . . marching 10,000 lifor a year. The men ate dogs’ flesh, and they had no
support troops or provisions. With only one horse each they marched into
west Tibet and attacked Lazang Khan.” The Zunghar army’s feat intimi-
dated Shi Yide, who reported the dedication of the soldiers with “won-
derment” (Ma. ferguweme giyangname). The emperor refused to back
down. The ultimate goal of exterminating Tsewang Rabdan justified any
Yet the nearly 100,000 Qing troops in Sichuan and Xining did not move.
They expected that Tsering Dondub’s men, exhausted by the heavy snows,
would have little strength when they arrived in Lhasa, even though they
could inflict a good deal of damage on the way.
71Only Lazang Khan’s des-
perate appeals for imperial support instigated Qing mobilization. Lazang
claimed that if he lost Lhasa, the entire Yellow Teaching would be elimi-
nated. Kangxi ordered the Manchu general Erentei (Ch. Elunte) to march
on Lhasa with seven thousand men, mainly Chinese and Muslims, via the
northern route from Xining through the desert, and Namujar would march
with ten thousand Tangut troops by the longer but more secure route
via Dajianlu in western Sichuan. But before they set out, word arrived of
dreadful looting and massacres perpetrated by the Zunghar army. On No-
vember 30, 1717, Tsering Dondub had ordered the city to be sacked and
the monasteries stripped of their treasures. Tsering Dondub, formerly a
monk from the rival Shigatze monastery, could now take revenge on the
privileged lamas of Lhasa, backed by a ferocious army. Three days later, he
attacked the Potala, killed Lazang Khan, and seized the Dalai Lama and
Panchen Lama. The northern relief army advanced to Kela Usu, on the banks of the
234 contending for power

Salween River over 1,000 kilometers from Xining. There the Zunghars sur-
rounded them, killed Erentei, and besieged them with their food supplies
running out. They destroyed the entire force in September 1718. Funingga
himself received no reports about this crushing defeat; he had to ask, in
vain, an envoy from Tsewang Rabdan to provide him with information.
Kangxi then designated his fourteenth son, Yinti, the Fuyuan Dajiangjun,
to lead three armies into Tibet from the northwest.
73 With this appoint-
ment, the aging emperor hinted that Yinti could be the new heir if he distin-
guished himself in battle.
As Yinti assembled his massive army of 300,000 troops in Xining,
Sichuan Governor Nian Gengyao, well aware of the difficult terrain, ar-
gued that Tsering Dondub might in fact surrender to the Qing. According
to Qing intelligence, Tsering Dondub was not on good terms with his co-
general, and he feared retaliation by the jealous Tsewang Rabdan. Nian’s
proposal to begin secret negotiations with Tsering Dondub was, however,
74The new living Buddha, or his agents, also were concerned about
the disruption Qing troops would create in his country, even though he re-
lied on the army to escort him back to Lhasa. The Mongol princes of
Kokonor still feared Tsewang Rabdan’s retaliation and were concerned
about even further dependence on the Qing. The majority of the top minis-
ters, Manchus and Han alike, argued for merely guarding the Sichuan and
Kokonor borders. But the emperor brushed off all such concerns. Distur-
bances in Tibet would clearly affect the border peoples in Sichuan, who
were “of the same kind” (yilei).He insisted on a comprehensive campaign
that would simultaneously strike at Tsewang Rabdan in Turkestan and at
his generals in Tibet. The Khubilghan, under the new title “Sixth Dalai
Lama, Diffuser of the Teaching, Awakener of Sentient Beings,” would be
escorted by the army and placed on the throne in Lhasa, with the support
of the Kokonor princes. Once again, as at Ulan Butong, timidity and indeci-
siveness threatened to undermine his bold strokes. In his view, in times of
crisis, only autocratic decision making (duduan)could achieve success.
Yanxin, the emperor’s nephew, and Galbi (Ch. Garbi) led the army out of
Xining in February 1720, while Yinti stayed in Xining to rally Khoshot
support and guard the Dalai Lama. The Dalai Lama followed in May. On
September 24, 1720, the army took Lhasa, as the Zunghars had already
fled. Five chief lamas who supported the Zunghar intervention were exe-
cuted, and the Dalai Lama was placed on his seat in the Potala. Tibetans
surrounded the soldiers, playing music and prostrating themselves, declar-
ing, “Since the Zunghars plundered and raped us we thought we would
never see the sun shine again.”
76 Eventually the classic lama–patron rela-
tionship between Tibetan clergy and Central Eurasian rulers would be
imperial overreach and zunghar survival 235

transformed, for the Qing had become “the dominant military power in the
But Qing troops could not stay long. Nian Gengyao led the victory
march home, by the shortest possible route, as supplies were nearly gone
and the horses exhausted. The close association of Nian Gengyao and Yinti
established during this Tibetan campaign would play an important role in
the shadowy politics of succession after the Kangxi emperor’s death and
earn them both undying suspicion from the Yongzheng emperor. Yinti was
recalled to the capital upon Yongzheng’s accession and put under house
arrest, from which he was not released until the death of Yongzheng in
The official sources give a picture of a uniform, cooperative alliance of
the Khoshot Mongols and the Qing army, who entered Tibet with over-
whelming force, driving out the destructive, barbarian Zunghars, to the de-
light of the Tibetan people, who welcomed their liberation by external in-
tervention. Other sources, such as Yinti’s suppressed memorials, give a
different picture. First, they show that the Khoshot Mongols, who had had
good relations with the Zunghars in the past, joined the expedition only
reluctantly. For several years the Khoshots failed to follow Qing orders
to cooperate with the army. Nian Gengyao reported divisions among the
Khoshots, which were denounced by the Kangxi emperor. When, on July 1,
1719, the emperor asked the Khoshots for ten thousand men, Lobzang
Danjin, though eager to support the Qing, was ashamed to confess that his
brothers would not send troops.
Yinti’s memorials also reveal the active role of Tibetan guerrilla warriors
mobilized by Sonam Gyapo, chief of Khangchen, known as Kancennas
(Ch. Kangjinai). Kancennas aroused bands of thousands of fighters, who
ambushed Zunghar patrols and succeeded in blocking the routes of retreat
to Zungharia. At a banquet Kancennas killed sixty-five Zunghar envoys,
and was praised for this by the emperor. Again, contrary to the impression given by the Fanglue,the Zunghar
army had already begun its retreat from Lhasa before the Qing army ar-
rived. Kancennas’s activities help to explain why. The troops had already
suffered damaging guerrilla attacks, indicating the hostility of the local
population. Many suffered as well from disease (possibly altitude sickness).
Once they heard that a Qing–Khoshot army of anywhere from 100,000 to
1 million men was coming, bringing the Dalai Lama from Kokonor, and
that the road home was blocked, the troops found themselves in a desper-
ate situation. “A Zunghar soldier was drinking wine and crying in the
house of the Tibetan Danjin. Danjin asked him, ‘Are you crying because
you are homesick?’ and he replied: ‘You should be able to understand my
236 contending for power

misery. Now Kancennas at Gari [and others] with three thousand troops
have blocked our way home. We cannot retreat. Furthermore, the great
general, Kangxi’s son, has brought several hundred thousand troops, and
the Qinghai troops are bringing the Dalai Lama back to his throne. How
can our few troops defend against such a force?’”
80Only one-quarter of the
Zunghar army ever returned home. Thus the success of the Qing Tibetan campaigns was the outcome of
complex motives of multiple actors, including the divided Tibetans, the
quarrelsome Khoshots, the desperate Zunghars, and the disciplined Man-
chu, Mongol, and Han troops led by the emperor’s son. The stele set up in Lhasa the next year to commemorate the conquest
told a simpler story. It described the history of events in Lhasa since the
death of the fifth Dalai Lama, including the usurpation of power by the re-
gent and his failure to report the Dalai Lama’s death to the Qing for sixteen
years. Lazang Khan, with the support of the Khoshots, then eliminated
the regent, “revived the Teaching [of Buddha; xingfa].” Tsewang Rabdan
merely plundered and destroyed temples and massacred lamas. Although
Tsewang Rabdan also claimed to “revive the Teaching,” he actually de-
stroyed it. (All of the rival contenders for power over Tibet justified their
military activities as designed to “revive the Teachings” of Buddha.) The
emperor then sent thousands of Manchu, Mongol, and Han troops, who
braved malarial jungles to reach Tibet, where they killed the enemy, “paci-
fied” [pingding] Tibet, and (truly) revived the Teaching (zhenxing fajiao).
According to the inscription, these unprecedented (conggu wei you)
achievements exceeded those of all previous generations, and all Mongols
and Tibetans praised the boldness of the emperor and his divine might
Tsewang Rabdan himself, in letters written to the Qing emperor, claimed
to share many of the same ideals. He too favored peace in Tibet, and he
had intervened in order to prevent heretical and immoral activities by the
monks who were rivals to the Yellow Sect and to stop the oppressive ac-
tions of Lazang Khan. He had “destroyed the Red Sect, which deviated
from the Way,” and seized Lazang Khan’s wife and children.
82 Tsewang
Rabdan offered to cooperate with the Qing emperor, just as he had done
when Galdan was alive, if the emperor would refrain from active interven-
tion in Mongolia or Tibet. But now the emperor had grander designs. The stele concisely summarized the essential elements of the imperial
message to the newly conquered regions, stressing that it was Qing power
that had restored the Buddhist order revered by Tibetans, testifying to
the emperor’s divinely inspired mission. Kangxi had taken a step beyond
his predecessors, embracing the Tibetan religious hierarchy in addition to
imperial overreach and zunghar survival 237

the Mongols, Manchus, and Han. Prince Yinti’s skill in coordinating the
diverse members of the alliance got little credit; now it was solely the em-
peror’s own boldness that deserved recognition. Each imperial victory
quickly subsumed the conflicts and contingencies of conquest under the all-
encompassing gaze of the universal sovereign.The success in Tibet inspired even more ambitious plans. Funingga pro-
posed to advance immediately with seventeen thousand men from Barköl
on a three-month campaign to eliminate the Zungharian base permanently,
at a cost of over 350,000 taels. Yinti and the top generals Furdan and
Funingga gathered in the capital to plan the great extermination of the
Zunghar state. Yinti enthusiastically supported an immediate invasion. At
first the emperor gave his approval, but then he postponed the assault for a
year, concluding that it would be impossible to capture Tsewang Rabdan in
far-off Ili.
83Rumors arrived of Tsewang Rabdan’s death, or of serious dis-
sension in Zungharia. A captured Zunghar soldier reported under inter-
rogation that Tsewang Rabdan had imprisoned his son-in-law Galdan
Danjin. The Zunghars had suffered losses from attacks by the Burut and
Kazakhs to their west, and many of their cattle had died of disease. They
were saying, “Surely the Chinese have put a curse on them.” He confessed
that his people “great and small, sadly said to one another: ‘Certainly the
Han have sent envoys to the Russians, Kazakhs, and Burut, and now they
will make an alliance to attack us simultaneously. If they do this, how can
we survive? What will happen to our pastures?’”
The emperor decided to send the Jebzongdanba Khutukhtu, the leading
Buddhist cleric of Mongolia, to Tsewang Rabdan to invite him to submit
to the Qing. He would have to yield up Lazang Khan, whom he was shel-
tering, and promise never to disturb the Qing borders again. If Tsewang
Rabdan rejected negotiations, the military campaign would be launched. This time the generals were confident that they could defeat the enemy
because the Zunghars had limited supplies of gunpowder and cannon and
lacked discipline. Qing generals were gaining increasing confidence in the
decisive power of gunpowder weaponry. They also felt that they could
overcome the nomadic military advantage by sending several cavalry divi-
sions in a pincer attack.
85 Yet they had serious concerns about providing
supplies for the army based in Barköl. The grain harvests in this oasis
region, centered in Turfan, were seriously limited. Building up military set-
tlements in Turkestan became a more important focus the longer the army
remained there. The troops and support staff at Barköl and Turfan, num-
bering over 33,000, needed 6,690 shiof grain monthly, and the 25,000
troops in the Altai needed 5,000 shimonthly, for a total of 140,280 shiper
year. Barköl had produced a harvest of only 20,000 shi,and Turfan was
238 contending for power

even less successful. 86The tight supply situation meant that the troops must
either advance quickly or else put their efforts into substantial restocking of
provisions before a campaign. So no definite decision had been made to ad-
vance as Yinti headed back to the frontier. On December 21, 1722, Yinti was suddenly called back to the capital
with the news that his father had died and his brother Yinzhen had as-
sumed the throne as the Yongzheng emperor (r. 1723–1735). The absence
of Yinti from the capital during the crucial last days of Kangxi’s life allowed
Yinzhen, with the support of the General Commandant of Gendarmerie of
Beijing, Lungkodo, to engineer his own succession. Lungkodo produced a
testament, claimed to be in Kangxi’s hand, declaring Yinzhen the heir to the
throne, but historians have debated the legitimacy of his succession ever
The Kangxi emperor had twice deposed his second son, Yinreng, who
had committed many improper acts, but he had never been able to desig-
nate an heir. As many of his sons formed factions to plot for the succession,
frontier campaigns played a vital role in determining whom the aging em-
peror would favor, for Kangxi regarded military merit as a key element of
fitness to rule. Yinreng himself, even after his second deposition in 1712,
tried to get himself appointed Generalissimo of the armies on the northwest
frontier in order to restore himself to favor.
88 Yinshi, the eldest son, who
had accompanied his father on three campaigns, gained great favor until he
was accused of engaging a sorceress to place a curse on Yinreng. Yinsi, the
emperor’s eighth son, also rose in stature because he had accompanied his
father on the Galdan campaign of 1696. When Yinsi was stripped of his
rank for engaging in intrigue, Yinti, as commander in chief of the armies of
the northwest campaigns, “was regarded by many as the Emperor’s real
choice for the throne.”
Yinti’s brother Yinzhen, however, had never led an army to battle but
only drilled troops in Beijing. The emperor also regarded ritual deportment
as another skill of a successful ruler, and Yinzhen had demonstrated his
abilities best in conducting imperial rituals on Kangxi’s behalf. He partici-
pated in twenty-two rituals, more than any other son, and also joined ac-
tively in policy debates in the capital. Feng Erkang argues that Yinzhen
was not a low-ranking usurper but just as plausible a candidate as Yinti.
Yinzhen, in fact, was a prince of the first rank (qinwang),while Yinti was at
first only a beizi,and later a prince of the second degree (junwang).Xu
Zengzhong, however, stresses that since military merit had always been the
primary qualification for rule since the Manchu conquest, Yinti’s successes
in his campaigns clearly made him the favorite choice as the next emperor.
In the final political battle for the succession, an active military man who
imperial overreach and zunghar survival 239

was often away on campaign lost to a civilian ritual and policy expert who
had never left the capital.
Some have suggested that Yinti was deliberately sent away from the cap-
ital as part of the succession plot. Others argue that the emperor recalled
Yinti to the capital in order to prevent him from achieving success on the
frontier that would bolster his claim to the throne. Despite the contra-
dictory rumors of the contending candidates, I believe that Yinti had le-
gitimate tasks to perform on the frontier to help the aging Kangxi em-
peror plan once again his culminating extermination campaign against the
Zunghars. Only by chance was he absent at the emperor’s death. Yong-
zheng later claimed that Yinti was incompetent and had little military expe-
rience, and, as noted earlier, tried to suppress evidence of his achievements
in the Tibetan campaign. Exactly whom the emperor named as heir in his
last days, or if he named anyone, will probably remain, in Huang Pei’s
words, an “insolvable puzzle,” but clearly, frontier expansion changed the
succession to the throne and the writing of Qing history at the same time.
The New Emperor Changes Tack
The Yongzheng emperor’s succession inaugurated a new phase in Qing ex-
pansion, one with mixed and contradictory goals. In Turkestan, Yongzheng
continued the gestures his father had begun toward Tsewang Rabdan to en-
courage him to stabilize the border and send tribute missions to Beijing. As
the paramount leader of the Yellow Teaching in Mongolia, the Jebzong-
danba Khutukhtu, who died in 1723 at the age of ninety, had been a loyal
supporter of the Qing, who led the Khalkha tribes to submit to the Qing
when Galdan rebelled. He had also been a valuable link between the Qing
and the Western Mongols. After his death the emperor dealt directly with
Tsewang Rabdan’s envoys, promising them that the Qing had “no inten-
tion of destroying your state.” All he required was the return of the rebel
Lazang Khan; but, they were told, “it depends on you whether we follow
the emperor’s mercy [ren’en] or military might [wulie].”
Although Yongzheng continued to threaten retribution if his generous
offers were refused, he indicated that he was much more interested in cut-
ting back on military adventures. He began a sustained program of troop
withdrawals from the distant frontiers. At first he declared that no troops
at all were needed in Kokonor; he planned to concentrate them instead in
Barköl. The outbreak of Lobzang Danjin’s rebellion in 1723 negated this
goal, but the program of consolidating troops in fewer posts closer to
the interior continued. Several considerations favored this program. The
240 contending for power

troops were exhausted from staying for long periods of time at their posts;
therefore, there should be regular rotations to the interior. Local popula-
tions, too, were excessively burdened with supply demands for the garri-
sons. Underlying this retrenchment was Yongzheng’s program of domestic
austerity and institutional reform.
94He pulled troops out of Tibet in April
1723, concentrating them in Xining and western Sichuan. Only after Nian
Gengyao protested did he agree to build small guarded fortresses on the
roads west of Dajianlu.
Funingga, who had led the most aggressive campaigns against the
Zunghars in the last years of Kangxi’s reign, was clearly frustrated when
the new emperor decided to negotiate for peace. The blunt general, first
praised for his outstanding military capabilities, must have offended Yong-
zheng on his return to the capital, because he was soon sent to a demeaning
post in Shaanxi and stripped of his noble ranks, for unclear causes, in
Meanwhile, in Turkestan, Nian Gengyao began to implement the new
defensive strategy, relying on static defenses, reduced forces, and the pro-
motion of military colonies (tuntian).Only 2,000 men remained at Barköl,
1,500 at Turfan, and 2,000 at Hami. At Bulongjir (500 lifrom Jiayuguan)
he built a new fortress with 5,000 troops. This large outpost was estab-
lished as Anxi zhenin 1725.
97 In Tibet, likewise, after the suppression of
Lobzang Danjin’s rebellion, the emperor rejected Nian Gengyao’s request
for 6,000 additional troops west of Dajianlu. Control in Tibet relied on
only a very small garrison in Lhasa and support from the leading Tibetan
aristocrats. The Muslims of Turfan became the objects of a tug-of-war between the
two rival states, as Tsewang Rabdan tried to move them north while Yong-
zheng tried to move them south. Turfan was more of a strategic burden
than an asset: its grain supplies were too meager to support both a large
garrison and the local population. Barköl, an entirely military town, was
more suitable. When Zunghar raids began again in 1731, General Yue
Zhongqi demanded troop reinforcements, but the emperor rebuked him for
failing to organize his resources and told him to focus his defenses on
Gansu. The Turfan question was a troublesome thorn in the side of the em-
peror, who wanted to pull his military forces back and reduce costs. He pre-
ferred moving people in to sending troops out.
New internal conflict quickly drew the emperor, despite his intentions,
into condoning a second substantial military intervention in Tibet. In 1720
the Qing invasion force had initially installed a military government in
Lhasa, which was welcomed by Tibetans happy to see the brutal Zunghar
forces driven out. The great Potala Palace of the Dalai lama, looted by the
imperial overreach and zunghar survival 241

[To view this image, refer to
the print version of this title.]

Zunghars, was restored and even improved with imperial support. The of-
fice of regent was abolished, while the new seventh Dalai Lama, a twelve-
year-old boy, served as the figurehead for rule by leading Tibetan nobles.
The two most powerful were Sonam Stöbgyal, the chief of Polha in western
Tibet, known as Polhanas (Ch. Polonai), and Kancennas.
99Both had orga-
nized popular resistance to the Zunghars.
100 Three Manchu officials, the
Asahan Amba, supervised the administration with a garrison of three thou-
sand men. But the local government remained unstable in the hands of re-
gional rulers who could not create a functioning central council. The Chinese occupation army was a heavy burden for the Tibetans. The
price of grain was rising on local markets, even as the Qing was spending a
large amount to transport grain thousands of kilometers from the interior.
Generals Nian Gengyao and Yanxin had agreed with Kangxi that troops in
Lhasa should be reduced as soon as possible.
101 But Yongzheng ordered a
rapid and complete withdrawal to support his retrenchment drive, based
on maintaining peace with the Zunghars and relieving burdens on the civil-
ian population. Kancennas urged the emperor to reconsider as the troops
marched away. The Khoshot princes had been accustomed since the days of Gusri Khan
to exerting substantial influence in Tibet, but the Qing intervention fol-
lowed by sudden withdrawal left them no role. The Tibetans themselves
were to govern Tibet, without a regent or a Khoshot Mongol protector. The
old tripartite balance of Mongols as military protectors, the Dalai Lama as
spiritual authority, and the regent as temporal administrator had lost two
of its props. Instability and confusion resulted. A number of the Khoshots
resolved to invite Tsewang Rabdan to intervene in order to restore their in-
102 They believed that the Qing had betrayed a promise from Kangxi
that allowed them to appoint a Khan of Tibet from among themselves.
Prince Lobzang Danjin (T. Blo-bzan-bstan-dsin) was a grandson of Gusri
Khan, a powerful Kokonor chief who had supported the Chinese interven-
tion in Lhasa.
103 Chaghan Danjin, another grandson of Gush Khan from a
different lineage, was his major rival. Chaghan, backed by Yongzheng, ap-
peared to be expanding his territory at Lobzang’s expense. In July 1723
Lobzang attacked three Khoshot princes, also grandsons of Gusri Khan, in-
cluding Erdeni Erke Toghtonai, who fled in defeat, asking for Qing protec-
imperial overreach and zunghar survival 243
The Yongzheng emperor (r. 1723–1735) in court dress.

tion. Nian Gengyao advised non-intervention, saying, “If the [Kokonor
princes] should now forget their great indebtedness to our country and kill
their own flesh and blood, it is of no concern to us.” He knew well that, as
so often in nomadic warfare, an advance by Qing troops “would only
waste the strength of our own forces while the Mongols fled far off into the
distance on their well-fed horses.”
104 But the new emperor overruled Nian
244 contending for power
The Yongzheng emperor in daily life.
[To view this image, refer to
the print version of this title.]

Gengyao, ordering him to protect Erdeni Erke if he fled to Qing lands. Qing
support of refugees from a rising Mongolian power paralleled the course
that had drawn Kangxi into internecine disputes among the Khalkhas and
Zunghars. Lobzang Danjin would be offered the opportunity to repent and
submit to the Qing, but if he refused, a righteous army of extermination
(zhengjiao)would be sent against him. Lobzang, as expected, rejected Qing
offers to mediate a peace agreement. Changshou, the Qing envoy, reported
that Lobzang intended to defeat Chaghan Danjin and assemble all the
princes of Kokonor to proclaim him Khan. Rumors spread that the
Zunghars would soon enter Kokonor.
Nian and the emperor then agreed on a secret plan to “secure” (ding)
106 What they feared most was a potential alliance of the discon-
tented princes against Qing rule and a positive response to their appeals
to Tsewang Rabdan. They granted Chaghan Danjin the title of qinwang,
equivalent to Lobzang’s rank, so as to encourage him to split from Lob-
zang. As expected, the Qing shift of support induced Lobzang to attack
Chaghan Danjin on September 16, 1723. Lobzang assumed to himself the
title of Dalai Hongtaiji, aiming to unite all the Khoshot Mongols as they
had been under Gusri Khan. This provided the pretext for Qing military in-
tervention. Nian was again cautious, believing that Chaghan Danjin could
hold out against Lobzang, but the emperor ordered him to advance im-
mediately. On November 16, 1723, the Qing army battled with Lobzang
outside the Taersi, or Kumbum monastery (20 kilometers from modern
Xining), driving him back. Lobzang besieged the Xingcheng fortress on
November 27 and attacked the garrisons in Ganzhou and Liangzhou along
the Gansu corridor, but he lost quickly, and in less than a month the war
was over. Zunghar support for Lobzang never arrived because Tsewang Rabdan
was busy holding off the Russians and Kazakhs. He had had no intention
of supporting the princes in Kokonor, but he did shelter Lobzang Khan
when he fled there for refuge. Before the conquest, the emperor urged the troops not to mistreat the lo-
cal population. They should not rape women, violate graves, loot the prop-
erty of those who surrendered, or destroy houses, temples, and monaster-
107 After the conquest, however, the Qing troops took revenge on the
lamas and villagers who had supported the princes’ resistance. Their ulti-
mate goal was to destroy Lobzang Danjin and all his followers. They killed
hundreds of noncombatants, burned down 150 villages, and launched a
terrifying assault on the Gonlun monastery (Ch. Kuolongsi), headquarters
of the lCang-skya Khutukhtu (Ch. Zhangjia Hutukhtu), an ally of Lob-
zang, killing its six thousand monks and burning it to the ground.
imperial overreach and zunghar survival 245

Rituals in the capital celebrated this campaign as equivalent to the re-
pression of the Three Feudatories and the destruction of Galdan.
109 Of-
ficials formally reported the victory to eleven temples around the city,
including the temples of Heaven and Earth, the altars of Grain and the Im-
perial Lineage, and the tombs of the imperial ancestors. A stele inscribed in
the Imperial Academy depicted Lobzang Danjin as an evil rebel with a
“wolf’s heart and owl’s nature,” and praised Nian Gengyao for killing his
“crazed” followers and obtaining the surrender of hundreds of thousands
of others in such a short time.
110 Even though this campaign was much
shorter and easier than the Galdan campaigns or the repression of the
Three Feudatories revolt, Yongzheng could rightly claim to have brought
another vast territory under permanent Qing control. No longer could the
princes of Kokonor act with real autonomy from the Qing. After the repression, officials imposed strictly the banner and jasakad-
ministration on Kokonor that had been enacted in Mongolia. Nian Geng-
yao oversaw detailed plans for incorporation and reconstruction of the
territory. The princes (taiji)now became jasaks,or banner commanders,
under the supervision of Qing military personnel. They were allowed three
tribute missions per year. Khalkha Mongols in Kokonor were organized
246 contending for power
The Kumbum monastery in Kokonor. Photo taken by Frederick Wulsin, 1923.
[To view this image, refer to
the print version of this title.]

into separate banners, which freed them from subordination to the
Khoshots and provided the Qing with a balance against Khoshot indepen-
dence. As Nian recommended, the entire territory now became part of the
“interior”(neidi).Local Tibetans (Fan) would be governed by tusi(tribal
headmen) watched over by Qing garrisons. Their ties to the Dalai Lama
would be cut. He could no longer collect taxes from Tibetans in Kokonor;
instead, he would be granted 5,000 jinof tea annually as compensation.
The temples, which Nian described as “filthy hovels,” had been autono-
mous governing entities, collecting tribute from their followers and used
for weapon and food storage as well as religious activities. Now they would
be limited in size to two hundred rooms (jian),with a maximum of three
hundred lamas, and inspected twice a year.
The causes of the war with Lobzang Danjin have been much debated, but
the simple assumption of the emperor and his generals that it was a “rebel-
lion” against Qing rule is problematic. The term “rebellion” (panluan),
used by both Qing and modern Chinese historians, implies a previously
“loyal” population who then rejected imperial grace and directly chal-
lenged the emperor’s authority.
112 Until 1723, however, the Khoshot Mon-
gols had been quite autonomous of Qing authority, although they had en-
gaged in temporary alliances with the empire. Many of them had joined in
the expedition to Lhasa, though some with great reluctance. Katà Naoto
argues that there were pro-Qing and anti-Qing factions among the Ko-
konor princes, and that Lobzang Danjin’s attack on Chaghan Khan repre-
sented a struggle between these two factions; but Ishihama Yumiko ar-
gues more convincingly that there were no clear lines between the feuding
princes, and at one point there was a serious potential for an alliance of the
princes against the Qing, with Zunghar support.
113 The Yongzheng em-
peror, seeing the division among the princes, took advantage of their weak-
ness to launch a military campaign with the intention of bringing all of
Kokonor under Qing control. The suppression of Lobzang’s power elimi-
nated permanently the autonomy of the Kokonor princes and excluded
them from any influence over Tibet. The Qing acted primarily out of fear
that these princes and their followers could forge an alliance with the
Zunghars or move to Zunghar territory. Lobzang himself never intended to
draw in a major Qing army, and his rival, Chaghan Danjin, who had sup-
ported the Zunghar invasion of Tibet, was also no friend of the Qing. But
because of their internal strife, the Yongzheng emperor could seize the op-
portunity to gain a quick victory relatively close to the Qing military base
at Xining. Yongzheng in this case pushed his aggressive policy against the resistance
of the cautious generals Nian Gengyao and Yue Zhongqi. Once entrusted
imperial overreach and zunghar survival 247

with the campaign, however, Nian and Yue acted with great decisiveness,
brutality, and success. The emperor at first praised Nian highly in his re-
scripts. Any expansive military policy would have to depend heavily on this
vigorous Chinese-martial general. But Nian fell out of favor very quickly. In
1726 he was condemned to execution for ninety-two crimes of maladmini-
stration, corruption, and treasonous intrigue, and was allowed to commit
suicide. Yue Zhongqi helped provide evidence of his “crimes.” Nian’s rapid
downfall was closely connected to his intimate knowledge of the shadowy
circumstances of Yongzheng’s succession; his great military success made
him both indispensable and dangerous. Without Nian, the emperor would
find himself unable to expand further, and he would overreach his grasp,
suffering an embarrassing defeat.
Control of Kokonor still remained precarious. Disturbances in Tibet in
1727 threatened to spread to the Kokonor Mongols, but no open resistance
broke out. Lobzang, however, remained alive in Ili and could still exert an
attractive power from there. As Galdan Tseren, Tsewang Rabdan’s succes-
sor, continued to refuse to send him back, the court feared that some of the
Kokonor Mongols would abandon their responsibilities to the Qing and try
to move to Zungharia. The Kokonor banners now owed horses to the Qing
armies; these were supposed to be purchased at market prices, but local of-
ficials could exact extra burdens. When the Khoshot Prince Rajab (Ch.
Lachabu) attempted to move away from his pastures to avoid the burden-
some horse levies, he was commanded to return, while the emperor ordered
an investigation by local officials. In 1731 Prince Norbu, who had been
sent to guard against Zunghar invasions of Kokonor, rebelled against the
burdens of serving the Qing, but after ten days of plundering, he fled west
and was soon captured.
115 Most of the Kokonor princes actively supported
the Qing in putting down the rebellion. Still, the concerns about the loyalty
of the Kokonor princes led the emperor to deliver them a long edict, re-
minding them that they were descendants of Gusri Khan, who had received
many favors from the Qing, and that the Qing had supported the Yellow
Teaching, while the Zunghars were violators of the faith and came only to
stir up trouble. He told them that if they tried to flee to join Lobzang, “it is
too far, and your herds will die; how can you support yourselves? . . . Also
the Zunghars are surrounded by enemies and are always fighting. In every
battle they will put you on the front lines. Isn’t it better for you to stay
in your ancestors’ pastures and enjoy eternal peace?”
116 Even though the
Khoshot princes had originally moved to Kokonor from Zungharia, the
emperor encouraged them to regard Kokonor as their ancestral home. He
offered them peace and stability under Qing rule, provided they gave up the
freedom to move.
248 contending for power

The program of troop withdrawals resumed after the end of the rebel-
lion. Contributions for degrees, allowed as a temporary measure to raise
funds for the campaign, were now eliminated.
117 Tibet soon became an
arena of conflict again, however, because of increasing divisions among the
Tibetan nobles.
118 The capable Kancennas, the anti-Zunghar governor on
whom the Qing relied, had aroused increasing hostility among other minis-
ters, which culminated in his murder on August 5, 1727.
119 The Manchu
brigadier general Oci requested immediate reinforcements to put down re-
sistance. Meanwhile, Polhanas quickly mobilized his troops to wipe out his
rival Napodpa (Ch. Arbuba). Jalangga, President of the Censorate, and
Governor-General Mailu organized fifteen thousand troops from Shaanxi,
Sichuan, and Yunnan to enter Tibet, leaving Xining on June 13, 1728. On
July 3 Polhanas seized Lhasa before the Qing troops arrived and arrested
Napodpa and the leaders of the opposing forces.
120 He had kept the Qing
emperor informed of his activities, quickly winning support for his side in
the civil war. The Qing armies marched quickly through Kokonor, taking
care of logistical demands by having the soldiers carry two months’ grain
with them, plus 4 taels each to purchase more.
121 When Jalangga arrived
in Lhasa in September 1728, he and Polhanas put Napodpa and the others
on trial for the murder of Kancennas. Tibetan sources report that the rebel
ministers defended themselves by accusing Kancennas of contacts with the
122 But the Manchu judges, unimpressed, decided in favor of
Polhanas and condemned the ministers to death by slicing. Polhanas was
appointed governor of both eastern and western Tibet and richly rewarded
by the emperor. Jalangga, when he withdrew his forces, removed the inef-
fective Dalai Lama from Lhasa. He left two high Manchu officials (amban)
in Lhasa with a garrison of only two thousand men, as the emperor realized
that a large force would severely burden the supplies of the poor Tibetans. Both of the contesting sides in Tibet tried to link the civil war to the
Zunghars. The Qing official account, reflected in Wei Yuan’s Shengwuji,ac-
cuses the Zunghars of colluding with the Tibetan rebels.
123 But as Luciano
Petech concludes, confirmed by contemporary documents in the Fanglue,
the revolt was an internal Tibetan affair, not a rejection of the Qing protec-
torate; charges of Zunghar involvement came after the fact to justify Qing
intervention. Only in 1729, when the Yongzheng emperor launched a more
aggressive anti-Zunghar policy, did he try to suggest any Zunghar involve-
ment in the Tibetan war. After his second successful military intervention,
Yongzheng congratulated himself on “finding a policy to settle Tibetan af-
fairs once and for all” (yongyuan zhi dao).
124 When it came to Mongolia,
however, he would not be so lucky. When Tsewang Rabdan died in 1727, his son Galdan Tseren took power.
imperial overreach and zunghar survival 249

Relatively peaceful relations with the Qing lasted for only a year, as both
sides tried to feel their way to a new relationship. The Qing secured the bor-
der with Russia with the Kiakhta treaty of 1727, finally resolving the dis-
puted questions addressed in the Nerchinsk treaty of 1689. The Russians,
induced by trading concessions, could be relied on not to support the
Zunghars, and they would exert control over the mobile peoples of Siberia
and Manchuria, fixing them in place. The negotiations determined the bor-
der between Siberia and the Khalkha Mongols, who were now under Qing
control, over a distance of 2,600 miles. No refugees or criminals would be
sheltered by either side, and boundary stones would mark the frontier. Sava
Vladislavich, the Russian envoy, called the border an “everlasting demarca-
tion line between the two empires.”
Galdan Tseren, likewise, was offered regular tribute trading missions if
he would agree to determine borders with the Qing. A Zunghar embassy
visited the capital in 1728. But two strict conditions limited the prospects
for peace: Qing determination to cut off the Zunghars from any formal re-
lations with Tibet, and the surrender of Lobzang Danjin to Qing custody.
When Galdan Tseren requested permission to send men to conduct ceremo-
nial tea offerings for the monks in Lhasa, called man-ja(Ch.jiancha or
aocha), in order to aid the propagation of the Buddhist faith, the emperor
contemptuously refused: “The Zunghars are a small tribe in the northwest-
ern corner; what relations can there be between the propagation of Bud-
dhism and your offering of a man-ja?”
126 These ceremonial missions, with
their accompanying trade caravans and lamas, were an important spiritual
and commercial link to Tibet for all the Mongols. Earlier, during Tsewang
Rabdan’s reign, the Qing stripped the rank from a Mongolian jasakwho
had allowed Tsewang Rabdan’s envoys to cross his territory on a tea mis-
sion to Tibet.
127 Qing policy now aimed at stabilizing but also isolating the
Zunghar state from both its closest cultural neighbors, Mongolians and Ti-
betans, and from potential aid from Russia. The surrender of Lobzang
Danjin would complete this process by removing all concerns about di-
vided loyalties of the Kokonor Mongols. For this reason, Galdan Tseren
could not cut his fragile links to Kokonor and Tibet. In 1729 the emperor abandoned his peace policy and called for an ag-
gressive campaign to exterminate the Zunghar state. He engaged in a
lengthy justification of his decision, reviewing all the outrages committed
by the Zunghars since the days of Galdan. All the other Mongols had now
submitted to the Qing; only the Zunghars stubbornly refused to surren-
der. According to Yongzheng, after his victories over Galdan, the Kangxi
emperor should have proceeded to eliminate the Zunghars, but he feared
criticism for “exhausting the army with excessive campaigns” (qiongbing
250 contending for power

duwu),and so he did not. Later, Kangxi vowed to exterminate Tsewang
Rabdan, after first offering him a pardon. Yongzheng had also offered to
pardon Tsewang Rabdan, but the latter’s raids on Hami and intervention in
Tibet indicated his hostile intent. Galdan Tseren now said he wanted peace
with the Qing, but the emperor accused him of following his father’s path
and introduced the new charge of colluding with rebels in Tibet. Even
though sagely emperors tried to avoid qiongbing duwu,using troops only
as a last resort, Yongzheng vowed to finish the task begun by Kangxi. There
would be no military glory to be gained by wiping out these remote tribes,
but if they were allowed to survive, they would seriously endanger the na-
tion (guojia).
Yongzheng cited many of the same causes as Kangxi to justify his cam-
paign, but he struck a more defensive tone, one of exasperation rather
than confidence. From the imperial point of view, merciful acts intended to
arouse gratitude and promote prosperity among these obstreperous Mon-
gols had been repaid only with violence. A repeated pattern of Zunghar
broken promises justified reluctant endorsement of extreme acts of vio-
lence. The Qing, by contrast with the Ming, viewed the Mongols as human
actors who could take responsibility for their decisions. Ming rulers could
see them only as equivalent to beasts, constantly driven by greed and vio-
lence, over which the empire could exert no control. Paradoxically, alien-
ation from the Mongols and belief in their inhumanity had led the Ming to
practice a more defensive policy. The rulers made no grand efforts to eradi-
cate these nomads: since they were more like a natural force than a human
society, it would be as futile as trying to eliminate wolves or floods. By con-
trast, the Qing goal of universal peace among humans led the Qing to
endorse elimination of those humans who obstinately refused to knuckle
under to the imperial view. Humans who chose to resist the Qing terms
remained human, but they had to pay the costs of their choice: “righ-
teous extermination” (zhengjiao),designed to return the world to a rational
Concepts of universal peace and benevolence, however, fit awkwardly
with endorsement of violent repression. These tensions are more prominent
in Yongzheng than in Kangxi. His repeated references to the classical
phrase qiongbing duwu indicate that he knew that the proper role of a wise
ruler, in Confucian terms, was to exert moral suasion, not material force.
Kangxi, by contrast, was closer to a Central Eurasian tradition that valued
individual valor on the battlefield along with efficient administration and
moral authority. The three values were more coherently united during his
reign, as the expansion of his rule brought with it increasing peace and ac-
tive support of the Manchu regime. Yongzheng faced greater tensions in an
imperial overreach and zunghar survival 251

empire now straining at the limit of its administrative and logistical re-
sources. His most common response was retrenchment, careful husbanding
of resources, and lengthy, cautious preparation. His primary goal had been
withdrawal and consolidation of troops, sparing burdens on the treasury
and on the local people. The intervention in Kokonor against Lobzang
Danjin was exceptional, brief, and apparently successful. But the persis-
tence of the Zunghar state ate away at the imperial sense of unity like an
open sore. As long as one group recognizably part of the traditional Central
Eurasian world of Mongols, Tibetans, Muslims, and Manchus insisted on
rejecting the universal sovereign, he could not be content. Yongzheng’s atti-
tudes toward the Zunghars exhibit glaring contradictions between express-
ing contempt for them as a “small, remote tribe in the northwest corner”
and viewing them as a major threat to the security of the entire national
community. The new aggression of 1729 was not based on strategic logic.
The emperor’s overweening ambition was generated by the burgeoning
ideology of the empire, which had to enforce universal peace over all its
component parts. The Qing, of course, was not the only empire that has el-
evated minor powers into major security threats.
130 Yet Yongzheng vacil-
lated over launching a new campaign, since he lacked a legitimate cause,
except for the demand for the return of Lobzang Danjin.
In April 1729 Furdan was made Generalissimo for Suppressing Rebels
(Jingni Dajiangjun), in charge of the North Route Army, while Yue
Zhongqi, with 324 officers and 26,500 troops, was appointed Generalis-
simo for Pacifying the Distant Frontier (Ningyuan Dajiangjun) of the West
Route Army. Both would set out in June and July, converging on Galdan
Tseren’s base in Ili.
132 A great review of the troops was held in Beijing in
June 1729. Suddenly, the emperor and his brother both fell ill with influ-
enza, and his brother passed away on June 18, 1730. With the death of
his brother, Yongzheng lost a key adviser.
133 Then an envoy arrived from
Galdan Tseren, promising to hand over Lobzang Danjin.
134 Seizing this op-
portunity to negotiate peace, the emperor recalled the generals to the cap-
ital and delayed the army’s advance for a year. Furdan and Yue Zhongqi re-
turned to the capital in January 1731. Just as they arrived, Zunghar detachments began raiding frontier posts in
Kokonor and Barköl, taking away large numbers of horses. Driven off
from Turfan, they besieged the strategic pass of Gas in Kokonor. Yong-
zheng took no aggressive action, even though Yue Zhongqi insisted on
strengthening the frontier garrisons with at least five thousand more
troops. The emperor insisted on keeping major forces in the Gansu corridor
and at Xining, and not sending large forces out beyond the pass. He regret-
fully turned down Yue’s sixteen proposals for bolstering the frontier gar-
risons, admitting that he respected Yue’s passion to wipe out the enemy
252 contending for power

who had stolen so many horses, but confessing that “now is not the right
time.” He found Yue’s efforts to support Turfan, however, to be “non-
sense” and “every proposal ill considered.”
135 Luckily, the garrison at
Turfan succeeded in driving off the Zunghars and recovering the stolen
horses, while the Mongolian jasakssuccessfully defended Kokonor. Here
was a complete contrast to Kangxi’s relationship with his commanders:
now it was the cautious emperor who held back an aggressive frontier gen-
eral. Yongzheng said, “I really cannot make up my mind...[A]t this mo-
ment according to Heaven’s will and the general state of the situation we
can do nothing but wait quietly.”
The approach of spring renewed the confidence of the emperor and his
generals. By March 1731 all the ice had melted on the rivers and roads,
making transportation more convenient. Hearing reports of Galdan Tseren
moving to attack Halashar, a small Muslim settlement 200 kilometers
southwest of Turfan, Yue Zhongqi again asked for authority to launch an
attack, but the emperor told him to stay put and prepare his defenses in
At the same time, he grew very sensitive to criticism that army pro-
visioning policies were driving up local market prices in Shaanxi. Revealing
his paranoid fantasies, he blamed these complaints on his old enemies,
Yinti, Nian Gengyao, and Yanxin, who had “spread underground rumors
to incite people.” Announcing openly that “in the future, our demands for
military supplies will be greater, and prices will inevitably rise,” he de-
nounced the “ignorant people with selfish goals” who failed to understand
the larger benefits that the campaign would bring. Once again, Yongzheng’s
elaborate self-justification and fear of criticism contrasted notably with
Kangxi’s ferocious repression of dissent. He did not cashier any dissenters,
but sent officials to Shaanxi to “persuade” the people (huadao)into accept-
ing the burden of supplying the army, while at the same time warning that
he would punish any extortion by local officials from the people.
Faced with continuing harassment from Zunghar raids at Turfan, the
emperor finally agreed to allow Yue to advance against Ürümchi, 170 kilo-
meters northwest of Turfan, where he could build a fortress and wipe out
the enemy “nests.”
139 Seizing Ürümchi, a much larger oasis, might also re-
solve the severe supply problems at Turfan, which could not support a large
force. The emperor now deferred to Yue’s knowledge of the local situation,
stating, “I am thousands of liaway; you must decide on the spot.” Unlike
Kangxi, he expressed no interest in personally leading an expedition. He
also indicated skepticism of Yue’s abilities to resolve the difficulties of both
attacking Ürümchi and defending Turfan, saying, “I am beginning to lose
faith in you.”
Meanwhile, Furdan with his northern army had advanced to Khobdo
imperial overreach and zunghar survival 253

and begun to build a fortress there. This was farther west than any Qing
forces had ever penetrated into Mongolia. Galdan Tseren was said to have
10,000 troops guarding his border with the Kazakhs, while two loyal gen-
erals, the elder and younger Tsering Dondub (Ch. Celing Dunduobu), had
set out with 30,000 men to attack the Qing army. Another general, Lob-
zang Tsering, had split with Galdan Tseren and led his 3,000 households
south, heading for the border with Kokonor. Furdan thought that this divi-
sion of forces offered the Qing a good opportunity to attack. Ten thousand
men in three divisions advanced to meet the Zunghars, leaving 7,300 be-
hind to guard Khobdo. According to reports, the younger Tsering was three
days’ march away with only a small force. On July 20, 1731, he found
2,000 of the enemy and attacked them with 3,000 troops, forcing them to
flee.Furdan was actually marching into a trap. Small Zunghar forces lured
him forward with harassing raids while the main force stayed hidden in the
mountains. On July 23 the Zunghars poured out of the mountains and sur-
rounded the Qing army with a full force of 20,000 at Hoton Nor, a small
lake 210 kilometers west of Khobdo.
141 After fierce fighting, on July 27
Furdan was able to break the siege and retreat with his men back to
Khobdo. The first reports coming to the capital indicated that Furdan had
lost his entire army; later it appeared that he had sustained heavy damage.
Only 2,000 men made it back to Khobdo.
142 Eighty percent of the Qing
army was lost, and nearly all of his Mongolian allies were wiped out. Furdan, confessing that he had advanced too rashly, desperately pleaded
with the emperor to sentence him to execution. The emperor rebuked but
pardoned Furdan, praised him for leading a successful retreat, and insisted
that he focus his efforts on the defense of Khobdo. Furdan launched into an
active construction campaign to fortify the walls at Khobdo, planning for a
large fortress, 7 kilometers in circumference with walls nearly 5 meters
high, and a garrison of 16,000 troops. This garrison would be 1,500 li(870
km) west of the nearest Qing base at Chahan Sor (near Uliyasutai), very dif-
ficult to supply, and too far away for immediate support. The Grand Coun-
cil decided to abandon the fortress and withdraw troops back to the more
easily defensible line at Chahan Sor.
Yongzheng could take some small consolation from Yue Zhongqi’s suc-
cessful raid on Ürümchi. When reports of Furdan’s encounter with a large
Zunghar force first came in, Yue immediately set out for Ürümchi, hop-
ing that this gesture would force the Zunghars to divert their forces there.
But Furdan was routed too quickly for this feint to have any effect. Yue
Zhongqi could not hold Ürümchi; he soon withdrew to Barköl. The military reverses of 1731 drove the emperor to despair. In an ex-
traordinary rescript to Yue Zhongqi, he poured out his anguish in one of
254 contending for power

the most emotional and revealing documents produced by any ruler of the
Qing. Nothing had worked as expected. Because his armies had violated
basic rules of military strategy, they had met disaster. Ultimately, he had to
take the blame. As he said, “Painfully I reflect on my responsibility, and I
find that we, ruler and minister, have brought all the blame on ourselves.”
(See Appendix B.) The supreme Lord of Heaven found himself baffled by
inscrutable foes and abandoned by cruel Heaven. The world seemed out of
joint. For now, only Yue and the emperor knew the full scope of the disas-
ter, but they needed to make long-range plans. The emperor considered
abandoning the Zunghar campaigns entirely, but that would mean giving
up the great enterprise begun by his grandfather: to end once and for all the
nomadic menace. He resolved to press on, but much more cautiously, wait-
ing until more propitious times arrived. Turfan could be abandoned after
putting up a fight; endorsing body-count logic, the emperor had no other
strategy than to kill as many Zunghars as possible.Galdan Tseren, encouraged by his victory, sent his armies to plunder
the region southeast of Khobdo, hoping to weaken Qing and Khalkha re-
sistance. He also appealed to the Khalkhas to join with him, as common
descendants of Chinggis Khan, against the Manchus. His letter to the Khal-
khas declared, “You are descendants of Chinggis Khan, who are subordi-
nate to no one. Why not move your pastures to the Altai, and live together
with us in peace? We can unite to resist any military threat.”
144 But the
Khalkhas refused and instead joined battle with the Zunghars near Erdeni
Zhao (or Erdeni Zu), the seat of the highest-ranking living Buddha of Mon-
golia. The Khalkha leader was Tsereng, a descendant of Chinggis Khan
who had been appointed head of a new Khanate, the Sayin Noyan, in 1725
as a reward for his services to the Qing. In the battle, over ten thousand
Zunghars were killed, but a remnant escaped west of the Altai.
145 The
grateful emperor appointed Tsereng military governor of Uliayusatai, built
him a city with a palace, and on his death made him the first of only two
Mongols who had memorial tablets installed in the Imperial Ancestral
Temple. Yongzheng could claim no credit for this victory, since his Khalkha
allies had done the fighting, but now both sides, exhausted by war, were
ready to negotiate. Yongzheng’s aggressive ambitions had led to a military disaster, the first
major defeat of a Qing army by the Zunghars. Driven by the goal of sur-
passing his father’s glories, Yongzheng violated his own instincts for cau-
tious retrenchment. Once again, tried-and-true nomadic tactics had lured
an army from China beyond its supply lines and destroyed it. Retrench-
ment and stabilization of the border became the main watchwords of Qing
policy for the next twenty years.
imperial overreach and zunghar survival 255

The Final Blows,1734–1771
T emporary peace set in for over a decade, as new leaders took power.
Qing tactics shifted to trade and tribute, using the classic “five baits” of the
days of the Han and Xiongnu to transform the Zunghars, or to soften them
up for conquest. The new emperor never gave up the goal of eliminating
the rival state, but he waited for more propitious times. Meanwhile, the
Zunghar leaders desperately searched for allies and resources, from Beijing,
Lhasa, Central Eurasian traders, and Russia, to hold their fragile state to-
gether. But when a succession crisis broke out in 1745, one faction invited
Qing forces into Zungharia, leading to the extermination of the Zunghar
state and people. Fatal individualism was indeed fatal.
Transforming the Barbarians through Trade
In September 1734 Yongzheng sent his top ministers to Zungharia to nego-
tiate a peace that would divide the Khalkha and Zunghar domains. The ne-
gotiations did not produce a settlement because Galdan Tseren preferred a
boundary at the Khanggai Mountains, while the Qing envoys argued for a
line running along the Altai Mountains and Irtysh River.
1Yongzheng or-
dered maps to be made of the border, but no peace treaty was signed.
Yongzheng immediately began reducing the numbers of forces stationed on
the frontier. Galdan Tseren sent his first tribute embassy to Beijing in 1735,
but then the Yongzheng emperor died. Yongzheng’s successor, Hongli, the
Qianlong emperor (r. 1736–1795), rejected the first tribute mission, but he

ultimately decided to continue the truce policy. Under Yongzheng, the em-
pire had spent over 50 to 60 million taels fighting wars with the Zunghars,
but it had failed to eliminate them.
2War had damaged the northwest econ-
omy severely; two-thirds of the Zunghars owned no animals.
3Peace, stabi-
lization of the border, and trading relations offered considerable attractions
to both sides. Commercial exchange, however, could never be sharply sepa-
rated from security interests. In the seventeenth century, Galdan had also pushed for closer trading re-
lations. He sent his first trading mission in 1671, and trade volume rose
with successive missions up to 1688.
4But the growing conflict with the
Khalkhas led the emperor to restrict most trade to the border, and after
Galdan’s invasion of Khalkha territory, all trade was cut off. Tsewang
Rabdan likewise requested commercial access to Beijing, but his invasion of
Tibet ended these efforts. Qianlong used the strong desire of the Zunghars for trade as a lever to
obtain a final delimitation of the boundary. In 1739 a truce was agreed on
and regular trade relations were established.
5For the next fifteen years, the
Qing and Zunghars closely joined their economies together. This officially
regulated “tribute” trade, from the Qing perspective, allowed three types
of missions: embassies to the capital, border trade at Suzhou in western
final blows 257
Trading post, or caravanserai, in Inner Mongolia. Photo taken by Frederick Wulsin,
[To view this image, refer to
the print version of this title.]

Equestrian portrait of the Qianlong emperor (r. 1736–1795), by the Italian court
painter Lang Shining (Giuseppe Castiglione), ca. 1790.
[To view this image, refer to
the print version of this title.]

Gansu, and “presentation of boiled tea”(aocha)to lamas in Tibet, follow-
ing a route through Xining. Substantial archival sources allow us to investi-
gate these trade routes in detail. Although it lasted less than two decades,
this trade reveals a great deal about the nature of Qing trading relations
with all “barbarians” on the frontiers, northwest and elsewhere.
As Governor-General Qingfu put it, the goal of trade was to “transform”
(xianghua) the barbarian peoples by offering them goods from the interior
in exchange for peaceful relations.
7The Zunghars were allowed to send
missions to the capital every four years, in 1738, 1742, 1746, 1750, and
so on, and to trade at the border in 1740, 1744, 1748, 1752, and so on.
The basic regulations followed the same principles as the Russian caravan
final blows 259
Table 7.1 The Qing-Zunghar trade: Tribute trade at Beijing and Hami
Year No. of
traders Goods Silver
1735 22 Various hides 14,197
1735–36 26 Sheep (344); horses (237); camels (113) N/A
1737–38 24 Animals; grapes; sal ammoniac; antelope horns 17,111
1738–39 42 Horses (428); camels (145) 40,000+
1739–40 65 Sheep (3,000); horses (701); camels (388); grapes (1,700jin); sal ammoniac
(10,000+ jin); antelope horns (5,000+) 53,000+
1742/2–7 months 42 Sheep (5,000+); horses (484); camels (715);
grapes (174 packs); sal ammoniac (86
packs); antelope horns; hides N/A
1742/9 26 Sheep (5,629); horses (146); camels (114) N/A
1743–44 Sheep (545); horses (84); camels (42) N/A
1744–45 38 Cattle (378); sheep (7,669); horses (543); camels (191); hides; grapes; antelope horns N/A
1745–46 28 Cattle (28); sheep (945); horses (290); camels (95) N/A
1746–47 46 Cattle (690); sheep (13,700+); horses (913); camels (217); hides N/A
1748 28 Sheep (1,267); horses (407); camels (87);
hides N/A
1749–50 47 Cattle (129); sheep (2,585); horses (678); camels (181) 10,200+
1750–51 52 Cattle (156); sheep (3,600–3,700); horses (957); camels (346) 10,500+
1751–52 9,000+
1754 33 Animals; hides 8,175
Source:Zhungar Shilue Bianxiezu, Zhungar Shilue(Beijing: Renmin Chubanshe, 1985),
pp. 134–137.

trade, but the missions to the capital avoided arriving during the same year
as the Russians. The main goods offered for sale by the Zunghars were ani-
mals (horses, sheep, cattle, and camels), furs, certain medicinal products
(sal ammoniac and antelope horn), and dried grapes from Turkestan. (See
Table 7.1.) In exchange, the Qing provided brocades, tea, rhubarb, and, if
necessary, silver. The size of embassies was limited to one hundred men for
border trade and two hundred to three hundred men for trade at the cap-
ital; traders could stay at the border no more than eighty days to conclude
business, and the numbers and quantities of goods were specified in ad-
vance. Exports of gunpowder, metals, and weaponry were prohibited.
Following the emperor’s orders, officials made great efforts to ensure
that trade proceeded smoothly and that the embassies were well treated.
They offered medical care to Mongols who became ill while in the capital—
an important service, as nomads were known to be vulnerable to smallpox
and other urban diseases.
9Even though the two societies had just con-
cluded nearly fifty years of warfare, there was no wrangling over the ketou
and ritual precedent. Mongols and Manchus understood each other, and
260 contending for power
Table 7.2 Trade at Hami and Suzhou
Year No. of
traders Goods Hami
(taels) Suzhou
(taels) Total
1743–44 122 Cattle (260) Sheep (26,800)
Horses (545)
Camels (726)9,790 41,000 50,790
1746 213 Cattle (2,642) Sheep (40,615)
Horses (1,628)
Camels (726)13,130 95,923 109,053
1748 136 Cattle (402) Sheep (71,505)
Horses (984)
Camels (585)12,744 74,000 86,744
1750 301 Cattle (2,200) Sheep (156,900)
Horses (1,900+)
Camels (1,000+)7,868 186,200 194,067
1752 200 Cattle (1,200) Sheep (77,000)
Horses (1,279+)
Camels (588)N/A N/A N/A
Zhungar Shilue Bianxiezu, Zhungar Shilue,pp. 134–137.

the emperor viewed the Zunghars as part of his realm, deserving his protec-
tion and generosity(fusui).
The Suzhou trade, however, quickly grew well beyond expectations,
straining at official limits. (See Table 7.2.) From a value of 10,000 taels in
1738, it leapt to 105,000 taels in 1741. Furthermore, these figures omit a
substantial amount of private trade carried on alongside the officially re-
ported trade. The Zunghar traders constantly pressed for relaxation of
trade restrictions, asking for annual missions, and that missions heading
for the capital be allowed to sell non-tribute goods at the border. Border of-
ficials were caught by surprise when huge herds of animals and large num-
bers of men turned up at their posts, straining available local pastures.
Rather than be held responsible for keeping these people at the border, they
agreed to allow trade even in off years. As the tables show, despite the of-
ficial restrictions, trade occurred every year, with one exception, from 1738
to 1754, and sometimes traders arrived twice in one year. Volumes were
higher in official years, and highest during capital mission years, but sig-
nificant trade occurred every year. When Zunghars asked that trade be per-
mitted at Hami, 875 kilometers northwest of Suzhou, to relieve the pres-
sure, officials reluctantly agreed.
10 The traders asked very high prices for
their animals, causing officials to engage in weeks of haggling, and they ex-
tended their time at the border waiting for more goods to arrive.
The second tribute mission to Beijing illustrates vividly the pressures that
final blows 261
Table 7.3 Boiled tea (aocha)missions to Tibet
Year No. of
traders Goods Silver
1741 300Cattle (400)
Sheep (7,392)
Horses (1,716)
Camels (2,080)
Sal ammoniac (19,000+ jin)
Antelope horns (82,700)
Grapes; hides 105,476
1743 312Sheep (2,800+)
Horses (2,300+)
Camels (1,700+)
Hides; grapes; antelope horns 78,000
1747 300Sheep (3,000)
Horses (3,000+)
Camels (2,000+)
Hides 164,350
Zhungar Shilue Bianxiezu, Zhungar Shilue,pp. 134–137.

the traders put on frontier officials. 12 Chuinamuke, a relative of Galdan
Tseren, headed the caravan of forty-two men, 634 pack loads, and five
thousand sheep which arrived at the border early in 1742. Chuinamuke im-
mediately asked for permission to sell the animals at Hami or Suzhou, in vi-
olation of the tribute regulations. Yong Chang, the military commander at
Anxi, decided to be lenient, following the emperor’s policy of “cherishing
men from afar” (huairou yuanren). Most of the men, animals, and over five
hundred packs stayed in Suzhou, lodged at the Qing’s expense, while a
small group of fifteen men presented the tribute in Beijing. Yong Chang
supplied them with grain, tea, and tobacco and sent them on their way on
March 3. The emperor agreed to recompense the traders for any animals
that died en route. A month later, however, Governor-General Yinjishan took a much
harder line. He regarded the traders as “crafty and greedy” barbarians who
“traded useless products for the goods of the Central Kingdom” and plot-
ted to spy out conditions in the interior. In 1739 Shaanxi and Gansu local
officials had bought 17,000 jinof dried grapes, for which there was no de-
mand, but eventually unloaded them at one-third the original purchase
price, a loss of 10,000 taels to the treasury. Yinjishan agreed that officials
should be “flexible in accordance with circumstances” (quanheng zhong-
qing) but also firm, so as to control the traders’ “insatiable desires” (wuyan
zhi qiu). The previous year, on the pretext of presenting boiled tea to Tibet,
the traders had brought large piles of goods to Xining, then turned around
and gone home without ever entering Tibet. Clearly they had “inconstant
natures” and could not be trusted. When the delegation returned from
Beijing, however, after further haggling, the Governor-General ultimately
agreed to buy up the unsold goods, calling in merchants to negotiate sales
at reduced prices. Obviously relieved when the traders returned to their
“nests,” he once more vented his exasperation: “They have hundreds of
tricks,” he complained; the only way to treat them was with “a principled
balance between generosity and restraint.”
The arrival of the 1748 caravan at the border led to more amusing nego-
tiations between exasperated officials and disingenuous traders, reported
in vernacular Chinese in the documents. (See Appendix C.) Once again,
the wily caravan men brought along extra people, described as “doctors,
cooks, and accountants,” attempted to unload excess goods at outrageous
prices, pleaded for help in disposing of sick animals, and expressed fulsome
appreciation of the emperor’s benevolent grace. Grumpy officials eventu-
ally gave in to most requests, and the traders promised not to commit of-
fenses again.
One other incident of this year indicates that trade relations did have
262 contending for power

wider effects in the Zunghar territory. 15 Ajibardi and Niyasi, two young
Turkestani men from Turfan, galloped up to the frontier on stolen horses to
plead for refuge. Captured by the Zunghars at a young age, they had been
forced to perform slave labor for the masters, and had been beaten severely.
They vowed that it would be “better to die in Chinese territory than under
the Zunghars.” Ajibardi, whose parents were discovered to be alive in
Guazhou, was handed over to Emin Khoja, the begof Turfan, while Niyasi
was sent to the capital. A few months later a thirty-year-old Zunghar es-
caped from servitude, having heard that he could “live a better life” under
Qing rule. Though rather suspicious of the motives of these deserters, the
officials found them to be useful sources of information, and giving them
refuge helped reinforce the emperor’s reputation for generosity. Peaceful
trade relations did appear to be softening up the Zunghar empire’s support.
Soldiers and merchants were engaging in private trading on the border, de-
spite official prohibitions.
16 Border trade, though exasperating, promised
larger political gains if the wealth of the Chinese interior lured away the
Zunghars’ subject peoples. The border trade not only altered Zunghar internal relations but also be-
gan to change relations with the frontier merchants. Border officials, realiz-
ing that merchants knew prices better than the government, decided to co-
operate with them. They created a system of “merchant management under
overall official supervision” (shangban er guan wei zongshe zhaokan).
Nineteenth-century advocates of self-strengthening programs would later
call this arrangement “official supervision and merchant management”
(guandu shangban).The quantities of goods which the Zunghars brought
to the border exceeded what local markets could bear. Dried grapes and
rare medicinal products like sal ammoniac and antelope horn, obtained
from mines in Turkestan and pastures in Mongolia, piled up in warehouses
when no one could arrange distribution. Cattle and sheep served local
interests better because they could be used to support military garrisons,
but even these herds exceeded local demand. Furthermore, Zunghars con-
stantly insisted on being paid in silver, thus threatening to cause a substan-
tial bullion outflow. Border officials did not have much tea and cloth on
hand, and they would exhaust their treasuries if they bought everything
in silver. The Qianlong emperor resolved to contain silver within the em-
pire so as to maintain the stability of the currency. Turning to merchants
would solve the problems of limited demand, bullion outflow, and insistent
traders. In fact, the “Zunghar” missions were dominated by experienced Central
Asian merchants who moved bulk goods and currency along the tracks of
the ancient Silk Roads. In 1748, for example, of a total of 136 men, 46
final blows 263

were Mongols and 90 were Turkic Muslims (Chantou Hui). Three of the
four headmen of the caravan were Turkic.
18 The caravans mixed together
diplomatic envoys, Zunghar officials, herders, merchants, and certainly,
as the Qing suspected, spies. They pursued joint goals of generating reve-
nue for the Zunghar state, reviving the devastated pastoral economy, and
gathering information about their huge neighbor. Qing policy aimed to re-
duce intercultural contact as much as possible, especially to prevent the
Zunghars from confronting any Khalkha Mongols, by limiting their stay
and keeping them constantly under close military escort. But the growing
size of the trade forced officials to bring in other participants who had pri-
vate interests separate from those of the state. Very few merchants in the northwest, however, had sufficient capital to
handle the large quantities of goods. A merchant needed at least 100,000
taels to engage in this trade, but demobilization of the military had under-
cut the primary support of Suzhou’s economy. In Gansu there were almost
no wealthy merchants, and in Shaanxi most merchants ran small pawn-
shops scattered all over the province. In 1744, however, Li Yongxiang ar-
rived from Xi’an to take responsibility for the border trade. The Governor-
General loaned him 28,741 taels to supplement his capital, expecting re-
payment the next year. Thereafter, the government had to continue giving
inducements to draw traders from the interior. It paid many of their trans-
port costs and supplied them with carts formerly used for military supplies.
When the merchants complained that the amount of goods to be brought
by the Zunghars was unpredictable, officials agreed to loan them funds to
make up the difference.
19 When the “useless” high-value medicinal goods
could only be disposed of at low prices, causing merchants to lose money,
the officials made up their losses. Gansu Governor Huang Tinggui argued that it was vital to expand mar-
ket demand, because this trade was a “national security” affair (guojia
gongshi) in which all parts of the empire were involved. He allotted respon-
sibility for disposing of goods primarily to the northwest provinces—30
percent to Gansu and 70 percent to Shaanxi—but he urged Zhili, Henan,
Shandong, and Shanxi to mobilize their markets as well. These provinces
had dense populations and abundant merchant capital, unlike the north-
western regions “at the back of beyond” (tianmo zhi qiongbian).
20 It was
their responsibility to relieve the northwest of its glut of unsold goods.
Whatever could not be sold in the North China region should be sold at
Chongwenmen, the main gate of the imperial capital. Over time, the trade evolved more systematically to suit the needs of
the Zunghar traders. When they specified the types of cloth and tea they
needed, border officials would recruit merchants and loan them capital
264 contending for power

(ziben)to proceed to Jiangnan to obtain the necessary goods. 21These inter-
est-free loans ensured that Zunghars would dispose of their goods quickly,
while merchants had sufficient capital to pay them. Often, however, the
treasury lost funds after the trading season ended.
Concern about the outflow of silver drove the state itself to engage in
substantial commercial transactions. Silver constituted only 6 percent of
total exports in 1738, but its share rose over time. In 1743 Zunghar trad-
ers accepted half of their payment for furs in silver, but Governor Huang
Tinggui thought this excessive. The Zunghars were supposed to “supply
what they lack, not increase what they own” (ji qi suo wu, bing fei li qi suo
you), that is, trade sheep for daily use goods, not grapes for silver.
22In 1746
the Zunghars demanded 40 percent payment in silver, but after much hag-
gling accepted 20 percent.
23Huang Tinggui and other provincial governors
tried as much as possible to make the trade a barter trade. In short, by ad-
vancing money to merchants the Qing state injected silver into its interior
economy so as to shore up a barter trade on its frontiers, a version of mer-
cantilism, or bullionism, that many European statesmen would have en-
dorsed. In 1750 the Zunghars brought goods worth 186,000 taels, the largest
amount ever, which they exchanged for 167,300 taels’ worth of cloth and
tea, with the balance in silver. Governor-General Yinjishan, concerned
about the loss of wealth, again urged strict limits to the trade. The Qian-
long emperor, reversing his earlier views, now regarded the loss of silver as
of slight consequence compared to the political benefits derived from the
24Still, the court resolved to cut it back to the level of 1748, or 80,000
taels, but internal upheaval among the Zunghars after the death of Galdan
Tseren in 1745 had already begun to hurt commerce. The high volume of
trade in 1750, after the assassination of Galdan Tseren’s successor, may in-
dicate strong efforts to accumulate revenue by rival contenders for leader-
ship of the Zunghar state. Traders at the border told officials that Tsewang
Dorji Namjal was “crazy,” and his elder brother Lama Darja had been put
in charge until he got better.
25 When Dawaci made his bid for power, the
Qing shut down all trade in 1753 to induce his submission. They viewed his
requests for trade missions as motivated by intelligence gathering. Seven-
teen fifty-four was the last trading year. By the next year the Qing army had
occupied Ili, ending the unified Zunghar state.
The “boiled tea” (Ch. aochaorjiancha, T.manja) trade with Tibet (see Ta-
ble 7.3) was even more sensitive than the border trade, for it allowed direct
final blows 265

access by the Zunghars to Tibetan monasteries. 26Broadly speaking,manja
meant a pilgrimage carrying religious donations from Mongols to Tibetan
lamas, accompanied by trade along the route.
27During the truce period, the
Qing allowed three major missions to Tibet to pass through their territory,
but each produced angry disputes since the Zunghar and Qing goals dif-
fered. For the Zunghars, trade with Tibet was an opportunity to reestablish
alliances with Buddhist clerics and with the Khoshot Mongols of Kokonor,
as well as to make commercial profits. For the Qing, this trade in the guise
of ritual presentation could unify Mongols under the Tibetan Buddhist
church, which in turn was subjected to Qing patronage. This trade clearly
illustrates how Buddhist patronage and commerce were strongly influenced
by security concerns. The manja trade route via Xining was established as early as 1642–43,
when the Khoshot Mongols invaded Lhasa to support the Dalai Lama’s
Yellow Teaching sect, giving Mongols much more convenient access to Ti-
bet than the dangerous route over the Kunlun Mountains.
28Once the Qing
established control of the Khoshot Mongols in Kokonor in the 1720s, all
trade missions had to pass through their jurisdiction. Soon after Galdan
Tseren took power in Zungharia, in 1728, he requested permission from
the Yongzheng emperor to send boiled tea missions to Tibet to express his
support for the Dalai Lama, but as noted earlier, the emperor contemptu-
ously refused him because Lobzang Danjin was still in Zunghar hands.
Lhasa was in the midst of civil war, and the emperor suspected that the anti-
Qing faction had Zunghar support. Under Qianlong, shortly after the truce negotiations were completed in
1739, Galdan Tseren once again requested permission to send missions to
Tibet, and this time the emperor agreed, under certain conditions. Missions
would be limited to three hundred men. (At first the emperor tried to limit
their size to one hundred men but relented when the Zunghars claimed this
number was insufficient.) They first had to pass through Hami and Suzhou
before reaching Dongkeer temple, outside Xining, where they would be al-
lowed to exchange goods. They would be under strict military escort, and
they would avoid all contact with the Khoshot Mongols, who would be
moved away from the route through which the traders passed.
29Five hun-
dred troops, provided with 30,000 taels for transport and food costs,
would accompany the mission. Claiming that his men, who were not im-
munized against smallpox, would be infected in the garrison towns, Galdan
Tseren tried to send the mission directly to Xining, avoiding Hami and
Suzhou, but the emperor angrily refused. Finally, in 1741, the mission ar-
rived at Dongkeer, bringing animals, furs, and dried grapes. There the trad-
ers sold their goods for a total value of 105,476 taels, of which 80,000 taels
266 contending for power

were in silver. 30They were, however, denied permission to present offerings
at the nearby Labrang and Taersi (Kumbum) monasteries. These monaster-
ies were key assembly points for the Khoshots, and the Taersi monastery
had been a stronghold of Lobzang Danjin’s rebellion. The Kokonor Mon-
gols were still restive under Qing control, as Norbu’s rebellion of 1731 in-
dicated, so the Qing wanted no Zunghar involvement in the region. By August 1741, having finished its business, the Zunghar mission asked
to go home without traveling to Tibet, claiming that the climate in Tibet
was too hot, or too cold. When the mission arrived at the border in 1742,
General Yong Chang berated the merchants for betraying the emperor’s
trust. He strongly suspected that the aochamission had turned back only
because they expected no profits from trading in Tibet. Furious, the em-
peror denounced them for violating imperial orders, swearing that he
would never allow any future missions.
31 But if the goal of their mission
was to establish contact with lamaseries and the Khoshot tribes in
Kokonor, they realized that severe Qing restrictions were designed to pre-
vent this. They achieved their commercial goals, but saw no future in their
diplomatic endeavor. Nevertheless, in 1743, under strict escort, a second mission traded up to
100,000 taels’ worth of goods.
32 The caravan sent in 1743 did arrive in
Lhasa, and presented rich gifts to the Dalai Lama. Polhanas, the pro-Qing
Tibetan ruler, had no love for the Zunghars, who had killed Lazang Khan,
but he accepted the mission, while surrounding the capital with troops.
The Qing’s greatest windfall from this mission was the capture of the elu-
sive Lobzang Danjin. He had grasped the opportunity to accompany the
mission to return home from Zungharia, but he was seized and sent to
Beijing. The third mission was sent in 1747, after the death of Galdan Tseren, by
his successor Tsewang Dorji Namgyal, to offer prayers on behalf of the de-
ceased leader. Its merchants traded goods worth 164,350 taels, while the
Qing officials allotted 160,000 taels for the costs of the escort.
34 In early
1748 they celebrated a great feast in Lhasa and presented a large lump of
gold to the Dalai Lama. The officials were concerned that the Zunghars,
whom they suspected of spying, might establish a connection between Tibet
and the Jinchuan rebels in western Sichuan, and the Tibetans complained of
the high cost of these missions. The emperor declared that he would allow
no further contacts between the Zunghars and Tibet. Chaos in both Tibet
and Zungharia in the 1750s put an end to these connections. Despite their mutual suspicions, both sides still found reasons to con-
tinue the trade. If Galdan Tseren called back his first mission because of the
tight restrictions, he clearly had more than commercial goals in mind. Reg-
final blows 267

ular contact with the lamaseries in Kokonor and Lhasa was clearly a cru-
cial support for his state. He may also have feared that Qing troops were
moving back into the region of Khobdo, so he needed to consolidate his
35Galdan Tseren astutely did not count on peaceful relations lasting
forever. He would take advantage of the trading loophole to extend his
links to the Tibetans, while the Qing made strong efforts to limit contacts
with unreliable subjects. The Qing relied on the hostility to the Zunghars of
Polhanas, their ally in Tibet, plus the large military escort to prevent any
trouble. If the Zunghars used the missions to spy out conditions in Tibet,
the Qing escorts, too, could demonstrate their support of the Tibetan rulers
while encouraging economic links with the interior under strict control. Ma
Lin has argued that the aochamissions “contributed to developing the
unity of our multinational nation,” but they were too limited to have much
of an economic effect.
36They did, however, serve Qing strategic goals by ty-
ing Tibet and Kokonor together under the aegis of the Tibetan Buddhist
church, demonstrating the power of patronage of the universal emperor. Aocha missions continued through the eighteenth and nineteenth centu-
ries on a smaller scale after the destruction of the Zunghar state.
37 Small
groups of fewer than ten people did not require licenses, but the central
government still approved larger expeditions by Khoshot and Torghut
Mongols who went regularly to Tibet to pay homage to the lamas. Most of
these expeditions were organized by lamas, not by secular nobles. Trade
and religious pilgrimage were still linked, but in dispersed fashion, as no ri-
val centralized state could use them for its purposes. Increasing communi-
cation with Tibet did begin to tie the frontier regions of the empire together,
while the emperors strengthened their control by supporting the ecclesiasti-
cal nobility to balance against the secular lords. The principle of divide and
rule operated effectively to fragment any potential opposition. Peace with the Zunghars did not genuinely soften Qing attitudes. The
Qing regarded these barbarians as greedy, violent, and untrustworthy. The
Qing believed, however, that the emperor’s grace would soften them so that
they would accommodate to imperial dominion. Barbarians by nature had
“insatiable desires” (wu yan zhi qing)and “shameless greed” (tanli wuchi),
but by controlling their actions and “cherishing” (huairou)them, the Qing
could tame them.
38 Tying the Zunghar elites to the interior with trading
links would make them less inclined to attack the frontier.
39 This type of
“loose reign” (jimi)policy was a well-established method of dealing with
northwestern nomads. Frontier trade was thus a “national security” affair,
to which all parts of the empire should contribute, even if the economic
gains were slight. Commercial concerns, however, were not irrelevant:
frontier officials were supposed to limit the outflow of silver and cooperate
268 contending for power

with interior merchants to produce goods that the Zunghars would buy, as
well as ensure distribution of Zunghar goods in interior markets.Officials constantly described this trade as “tribute” (gong),but this term
covered a great variety of meanings.
40Using the term “tributary system” in
the sense employed by John K. Fairbank to characterize all of Qing trading
relationships is in one sense trivially true but in another misleading. As
James Millward and others have pointed out, the word “tribute” encom-
passed many different kinds of trading and power relationships, in the
Ming as well as the Qing.
The Zunghars occupied an ambiguous, borderline role in this relation-
ship. On the one hand, they were clearly not alien, new peoples, like the
Europeans or Russians. The Manchu rulers knew well the habits of Mon-
golian nomads, and part of the ruling coalition included them. Nor were
they like the loyal tributary states of Japan and Vietnam, which partici-
pated in trading relations as part of a broad East Asian network.
in the Ming, when the rulers in Beijing had no control over the steppe,
the Manchus had already won over a substantial portion of the Mongolian
tribes. They were not making large protection payments to nomadic in-
vaders to stave off damaging raids; the treasury lost some money on the
trade, plus the expense of escorting missions, but the net economic cost
was low. On the other hand, the Zunghars could not be considered loyal Mongo-
lian subjects, even though they were referred to as being part of the “inte-
rior” (neidi). The frontier trade was a kind of experiment designed to test
whether or not the Zunghars deserved imperial protection and whether
they could coexist with the great Manchu empire. As Thomas Barfield has
argued, large nomadic confederations frequently rose in tandem with cen-
tralized imperial regimes, using them as sources of revenue through either
raiding or trading.
43 Rulers in Beijing had an interest in keeping nomadic
confederations united, because it was easier to channel goods through a
single ruler, so long as his power did not become too great. Nicola Di
Cosmo has argued that Xiongnu raids on the Han dynasty were a result
of the inability of the Xiongnu chieftain to control his nobility.
44 Ming
rulers likewise suffered economic damage on the northwest frontier both
because of the disunity of the Mongolian tribes and because their defen-
sive bulwarks could not protect the frontier against raids. The Qing solu-
tion, to draw a sharp boundary line through the steppe and induce quies-
cence through trade, seemed to stabilize relations for a short period. The
Zunghars might have become a dependent but partially autonomous state,
linked to the Qing and other Mongols through trade and Buddhist ritual
patronage. But at the first sign of division among the Zunghars, the rulers
final blows 269

were ready to embrace the military option and return to their primary goal
of eliminating the state and its people.
The Death Knell of the Zunghar State
The death of Galdan Tseren in 1745 stimulated the internal divisions that
brought about the Zunghar state’s destruction. Within five years of his
death, internecine conflict had torn apart the state, and defeated members
were seeking Qing support. Galdan Tseren had three sons and one daugh-
ter. His second son, Tsewang Dorji Namjal, succeeded him in 1746, taking
the title of Hongtaiji. He was a violent, perverse, paranoid man, said to be
interested only in drinking and killing dogs.
45When his sister, Ulan Bayar,
tried to restrain him, he had her locked up, fearing that others were trying
to take power from him. In this case he was correct. His more popular elder
brother Lama Darja (or Lama Dorji) plotted with Sayin Bolek, the husband
of Ulan Bayar, to kill Tsewang Dorji Namjal in 1750 while he was on a
hunting trip. When he discovered the plot, Tsewang Dorji Namjal marched
against the conspirators, but he was defeated. His eyes were put out and he
270 contending for power
The Manchu general Machang, famous for victories in the Turkestan campaigns.
Portrait by the court painter Castiglione.
[To view this image, refer to
the print version of this title.]

was imprisoned in Aksu along with his other brother, Dashi Dawa. Fol-
lowers of Dashi Dawa then surrendered and were settled in Chahar.Qing officials watched developments in Zungharia closely. With misgiv-
ings, they allowed Tsewang Dorji Namjal to send envoys to Tibet to hold
ceremonies in memory of his father. The death of Polhanas in Tibet in
1747 threatened to send Tibet into turmoil again, but fortunately for the
Zunghar envoys, they were able to return without difficulty. But Tibet then
broke out in rebellion, and General Bandi had to lead his army to Lhasa to
final blows 271
The Mongol prince Dawaci, supporter of the Qing campaign against Amursana. His
portrait was one of a series commissioned by the emperor to honor military allies of
the Qing.
[To view this image, refer to
the print version of this title.]

install a reliable successor. Lama Darja was refused permission to contact
Tibet. Zungharia was thus cut off from its spiritual roots.Meanwhile, Lama Darja fell into conflict with two leading warriors:
Dawaci, a Zunghar noble, the grandson of the elder Tsering Dondub, the
most famous of the Zunghar generals, and Amursana, a prince of the Khoit
tribe, the grandson of Lazang Khan and a descendant of Tsewang Rabdan.
Dawaci, centered in Tarbaghatai, refused Lama Darja’s orders to pursue
Dashi Dawa’s men into Qing territory. Instead, he attempted to submit.
When Lama Darja tried to arrest him, Dawaci fled to Kazakh territory
along with Amursana, then returned to confront Lama Darja in Tarba-
ghatai. In December 1752, when Lama Darja’s own troops revolted and
killed him, Dawaci was made “Great Chieftain” (Da Taiji) of the Zunghars
and killed all of Darja’s men. But Dawaci, once in power, became a drunk-
ard and a wild man, who developed a hatred for Amursana. Amursana did
not belong to the Zunghar nobility, so he depended on Dawaci’s position to
maintain his influence among the Zunghars. Yet after he gradually gained
influence among the Khoshots, Derbets, and Khoits through marriage alli-
ances and negotiations, he proposed to Dawaci to divide the Zunghar lands
between them. Dawaci refused and attacked Amursana, forcing him to flee
to the east. In September or October, Amursana reached Khobdo and made
the fateful Faustian bargain. With his five thousand men, he would submit
to Qing authority if the Qing would place him at the head of the Zunghar
state. Other Zunghar leaders had also been fleeing the internecine struggle
to find refuge with the Qing. They received lavish welcomes, including
princely titles and many banquets at the emperor’s summer residence in
Chengde. But all previous Mongol refugees had been settled within existing
Qing territory, under close control. Amursana was the first to return to
Zungharia with the backing of Qing armies. The emperor grasped this op-
portunity, at the invitation of one of the leading contenders for power,
to intervene directly in Zunghar affairs beyond the borders. By the end
of 1754, Qianlong had decided to send two armies of 25,000 men each
against Dawaci: one by the northern route through Uliyasutai, and one by
the western route leaving from Barköl.
46 Bandi would lead the northern
army, with Amursana as his second in command; Yong Chang and Salar led
the western army. In March 1755 the troops set out, with two months’ ra-
tions per man, planning to meet in Bortala. (See Map 7.) Qianlong gave elaborate justifications for his campaign, invoking
Kangxi’s campaigns against Galdan, but praising Galdan Tseren’s obser-
vance of the boundary and his presentation of tribute. Dawaci, he claimed,
was only a usurper, whose violence had caused many Zunghars to flee his
272 contending for power

Map 7. The Qianlong emperor’s western campaigns, 1755–1760.

lands, forcing the Qing to provide for them. He also betrayed concern that
the long period of peace had undermined Manchu commitment to war:
“Our old Manchu customs respect righteousness and revere justice. Young
and old, none are ashamed to fight for them. But after enjoying such a long
period of peace, inevitably, people want to avoid putting on armor and
joining the ranks of war.”
47But now the Qing had to seize this opportunity
to protect the Khalkhas under its embrace and establish security on the bor-
ders. Qianlong saw this campaign as a means of solidifying control over the
Mongolian frontier and keeping up the fighting morale of the increasingly
sedentarized, pacific banners. All the players in this final act of the eighteenth-century Great Game now
mobilized their supporters. The Qing put pressure on its Khalkha allies to
provide support for the coming campaign, while the Zunghars looked to
Russia for aid. The Kazakhs, new entrants on the scene, received appeals
from both Qing and Zunghar envoys. The Russians knew that with only a
few soldiers scattered among distant fortresses, they were in a weak posi-
tion to face major Qing forces. They refused to accept Zunghar refugees
and tried to make peace among the Zunghars. They accepted dual pay-
ments of iasakto both Russian and Zunghar collectors but refused requests
by Lama Darja to destroy fortresses on the Irtysh and Ob rivers.
Ablai, leader of the Middle Horde of the Kazakhs, had sheltered Dawaci
and Amursana against attacks from Lama Darja. When Amursana split
from Dawaci, however, he, like the Qing, seized the opportunity to capture
herds and territory from the fractured Zunghar state. The campaign itself was short. Deprived of allies, and under attack from
both east and west, Dawaci had no effective response. His men were di-
vided into small, uncoordinated groups, while he spent most of his time
49By the middle of 1755, many of his zaisanghad surrendered. The
west army passed through Ürümchi, joined with Amursana and the north-
ern army, and reached Bortala in June 1755. Hearing of the Qing ap-
proach, Dawaci fled to Gedengshan, 180 lisouthwest of Ili, where he made
his last stand with ten thousand men. A night attack on July 2, 1755, by a
small patrol scattered Dawaci’s men, forcing him to flee south across the
Tianshan, where he was captured by the Hakim Beg Hojis of Ush. Sent in
captivity to Beijing, he was made a prince (qinwang)and married into the
imperial clan. The Qing success was rapid, but the armies could not be supported for
long. The generals quickly withdrew their troops, leaving only five hundred
men in Ili. Qianlong claimed to have been reluctant to use force at first, but
now he boasted of achieving a permanent settlement of the nomadic secu-
rity threat, implying that he had succeeded where Kangxi and Yongzheng
274 contending for power

had failed. He praised Fu Heng for his support, in contrast to the timid op-
ponents of the plan. He stressed the need for a long-term settlement of the
Zunghar problem, but also noted the low expenses of this campaign com-
pared to the costs of the previous ones.Yet once again the elimination of one Zunghar leader only created an-
other problem. Amursana congratulated the emperor on his victory. As a
Qing supporter (though he had actually done no fighting), he had been
given the title jiangjun(general), but now he expected to receive the only ti-
tle that really counted among the Zunghars: that of Khan. Amursana had
raised this demand before the expedition was finished, but Qianlong had
deferred an answer, for his real intention was to divide and rule. He would
make Amursana Khan only of the Khoit, one among four equal Khans.
Amursana rejected this offer, arguing that the Zunghars needed one single
ruler. He refused to use the official seal or clothes given him by the Qing. In-
stead, he used the seal of Galdan Tseren, the last officially recognized leader
of all the Zunghars. Once again the specter of a unified Mongolian state
loomed over the steppe.
As Amursana continued to gather more troops while delaying his formal
submission, other disturbing rumors spread. Some said, falsely, that he
had gathered the support of fifty thousand Kazakhs; but it was true that
Amursana had offered money to the Tibetan lamas to support his claim to
lead all the Zunghars. Reports arrived of his efforts to knit alliances with
the Buruts, Yarkand, and Kashgar. Despite growing suspicions, Qianlong
hesitated to take action. He invited Amursana to visit him at Chengde to
be invested with his title, along with the three other Khans, but Amursana
declined. Perhaps fearing excessive reliance on his allies, Qianlong at first
declined the proposal of the Khalkha Mongol Erinchindorj to arrest
Amursana and bring him forcibly to the emperor, but then agreed to have
Erinchindorj escort Amursana to Chengde. On August 20, 1755, he or-
dered Bandi to capture Amursana, but Amursana had anticipated him. He
escaped Erinchindorj’s custody and fled to the Irtysh River before Bandi
could arrive. The scattered armies in Ili were unprepared for a major revolt. Bandi, be-
sieged in Ili after trying to capture the city, committed suicide. Yong Chang
quickly retreated from Ili and asked for reinforcements. Failing to hold
Ürümchi, he retreated to Barköl. Qianlong fired Yong Chang, replacing
him with Tseleng. Declaring that he would capture and wipe out Amur-
sana’s forces, the emperor offered rewards to Mongol chieftains who aided
the pursuit, and insisted that Ablai of the Kazakhs capture Amursana if
he fled there. He announced a scorched earth policy of devastating the
pasturelands of all nomads who rebelled.
final blows 275

In the middle of Amursana’s rebellion, the Khalkha prince Chingünjav
(Ch. Qinggunzhabu), ruler of the Khotogoits in far northwest Mongolia
and an important ally, rejected his orders and deserted his posts.
51From the
summer of 1756 through January 1757, Chingünjav mounted what would
be the most serious rebellion by Khalkha Mongols against Qing rule until
the end of the dynasty. Chingünjav’s revolt exposed the tensions created by
the vise-like grip of the Manchus and demonstrated the cost inflicted on the
Outer Mongols by Qing demands to support the Zunghar campaigns. Be-
fore they could finish off Amursana, Qing generals had to divert most of
their forces to ensure stability in Khalkha, where their control was fragile
and many Mongols suffered poverty and official oppression. At the same
time, the revolt showed that, like their Zunghar brethren, the Khalkha
Mongols could not unite effectively against the Qing juggernaut. After submitting to the Qing at the Dolon Nor assembly of 1691, the
Khalkha Mongols had been incorporated into a “banner” system, designed
to maximize administrative control while reinforcing their military organi-
zation, and to fragment their power while emphasizing their loyalty to the
dynasty. These banners, however, performed quite a different function than
the Mongol, Manchu, and Han banners that had created the Qing conquest
52Each basic banner unit, or khoshun(Mo.khushuu, Ch.qi), had at
its head a prince, or jasak,whose position required confirmation by the em-
peror. The three traditional Khalkha aimaq,or clan confederations claim-
ing a common ancestor, under the Sechen, Tüshiyetü, and Jasaktu Khans,
were now divided into multiple khoshuninstead of individual clans. At first
there were seven khoshun,but the number of jasakssoon expanded rapidly.
By 1755 the four Khalkha aimaqcontained over eighty jasaks.(The Qing
had created the new Sayin Noyan aimaqin 1725.) These new jasaksowed
their power to Qing rule, and they undermined the hereditary claims of the
Khans of each aimaq.Theaimaq had become territorial units, not confed-
erations personally subordinate to the Khan. With stipends, seals, ceremo-
nial duties, and required attendance at court, much as in the Japanese
sankin kÃtai system, the court kept these nominally independent rulers un-
der close watch while confirming their authority over their clans. No longer
could Mongols move freely from one tribal unit to another; they were con-
fined to fixed territories, carefully policed and mapped. A local bureaucracy in Urga increased the burdens on the Mongols,
obliging them to provide corvée duties at guard posts and postal relay sta-
tions and to sell their livestock to officials for military campaign use. Local
officials often paid below market price for animals, forcing the Mongols to
go into debt to Chinese moneylenders to support themselves. As they had
done earlier in Inner Mongolia, Chinese merchants penetrated the Khalkha
276 contending for power

territory, taking advantage of the Mongols’ needs for cloth, tea, and other
goods to bind them with debt contracts. The fixing of boundaries, corvée
obligations, and other requirements, such as pasturing imperial horses,
along with the rise of static monastic establishments, increased pressure on
the grasslands while reducing the manpower available to raise herds.The wars with the Zunghars in the late Kangxi and through the Yong-
zheng reigns further devastated the economy. By one estimate, the Qing
requisitioned 4 million animals from the Khalkhas from 1715 to 1735,
paying as little as 7 silver cash per sheep.
53The cost of supporting Zunghar
refugees further stressed the carrying capacity of the grasslands. In 1751
Qianlong prohibited trade between Khalkhas and Zunghars on the border.
In order to clear their debts, the Khalkhas had to go further into debt to the
Chinese merchants, and to repay Chinese debts, they had to sell their ani-
mals at low prices to get silver. Thus, over the first half of the eighteenth
century, the formerly mobile Khalkha nomads had seen their economy forc-
ibly monetized with Chinese capital, stressed by increasing demands for
livestock for Qing military campaigns, and confined within rigid adminis-
trative boundaries under close bureaucratic control. Chingünjav fit very poorly with this new ideal of the obedient, settled no-
final blows 277
The valiant warrior Ayuxi, a Zunghar who fought for the Qing. Condemned to
death after being captured, Ayuxi pleaded for his life, swearing loyalty to the Qing.
He displayed excellent riding skills and was unsurpassed in the use of the spear, so he
was put in command of 24 Turkestanis, who routed a force of 6,500 Zunghars. The
emperor had Castiglione commemorate him with this portrait.
[To view this image, refer to
the print version of this title.]

mad. A legitimate descendant of Chinggis Khan, he inherited control of the
Khotogoit tribe from his father, with Qing confirmation. Although he rose
to high posts in the local administration, he repeatedly clashed with of-
ficials, who accused him of neglecting his duties. Impeached for “laziness”
asJasak beile of the Jasaktu Khan’s aimaq,he was later attacked by the em-
peror and demoted for defying the prohibition on border trade with the
Zunghars. Chingünjav’s obstreperousness exasperated his Manchu superi-
ors, but it makes sense as the behavior of a Mongol official who envied the
freedom of the “true nomad,” in Owen Lattimore’s term.
54 Coming from
the remotest region of the Khalkhas, he prized his independence more than
bureaucratic regulation. Chingünjav grudgingly bore his burdens and joined Amursana to march
against Dawaci, but severe frost in the winter of 1755 left his Khalkhas
with little support to provide. The final straw was the execution of
Erinchindorj, elder brother of the Jebzongdanba Khutukhtu, and the escort
of Amursana, who had suspiciously allowed him to escape Qing custody.
At first, Chingünjav himself came under suspicion. He was stripped of his
rank and arrested three times but ultimately cleared. The Qing needed his
loyalty, but it also needed a scapegoat. The court then fastened on
Erinchindorj. According to some accounts, Chingünjav and the Jebzong-
danba Khutukhtu were forced to witness Erinchindorj’s public execution in
the capital in May 1756. (Later accounts claim that Erinchindorj was al-
lowed to commit suicide.) In July 1756 Chingünjav rejected Qing orders to pursue Amursana and
returned to his pastures. Soon after, Khalkhas all along the border fled their
guard posts, and the court declared Chingünjav a rebel while urging the
Khalkha nobility to remain loyal. Chingünjav spelled out his grievances in
detail: the Qing had oppressed the Khalkhas by seizing their horses and cat-
tle, and had executed Erinchindorj, a descendant of Chinggis Khan, the an-
cestor of all the Khalkhas.
55Qianlong, in his response, denied that his cam-
paigns had oppressed the Khalkhas: his men had paid for animals with
silver; and he insisted that no one, however noble his lineage, was immune
from punishment. But he knew that Chingünjav’s resistance threatened to
draw in all the disgruntled Khalkha nobility. The emperor, after some hesi-
tation, decided on severe repression of the Khalkhas before finishing with
Amursana. Diverting military forces from the pursuit of Amursana, he
made the Khalkha Mongol Tsengünjav (Ch. Chenggunzhabu) responsible
for putting down unrest. Meanwhile, Mongol mobs looted Chinese shops
at guard posts and relay stations, while Manchu local officials and jasaks
pleaded for the court to restore order. Chingünjav, however, failed to take advantage of the discontent. Even
278 contending for power

though he appealed to the Mongols as descendants of Chinggis Khan to
throw off Qing rule, very few of the nobility joined him. He also failed to
coordinate activities with Amursana. Although he approached the Russian
governor at Selenginsk for aid, the Russians took a cautious wait-and-see
attitude. They found it more beneficial to obey the terms of the 1727 treaty
with the Qing than to back a hopeless uprising. The decisive actor in the re-
volt was the Jebzongdanba Khutukhtu, the leading Buddhist cleric of Mon-
golia, who, more than anyone else, enjoyed popular support from all the
Mongols. At first he vacillated between calling on Russian aid and acting as
intermediary between Chingünjav and the Qing court, but in the end he
was overawed by Qing power. Telling the Khalkha leaders, “In my opinion,
the Emperor took pity upon and succored us Khalkhas when we were cast
into confusion by Galdan Boshukhtu, and has from age to age conferred
various weighty favors,” he urged them to return to their guard posts.
During the critical months of the winter of 1756, Chingünjav failed to gar-
ner more than two thousand men to support him. In January 1757 a fero-
cious assault by Qing troops crushed his army, captured Chingünjav, and
executed the ringleaders and their families. In a key concession, at the same
time the emperor announced that no Khalkha troops would be used in the
suppression of Amursana. Combining small carrots with heavy sticks, the
Qing withstood this “ragged affair,” which in practice had become an un-
coordinated series of desertions and looting rather than any coherent resis-
The most important long-term outcome of the revolt was the Qing decla-
ration that further incarnations of the Jebzongdanba Khutukhtu must be
discovered in Tibet, not in Mongolia. As he would later do with Tibet, the
emperor took control of the Buddhist hierarchy by removing local lamas
from participating in the selection of their chief cleric. Modern Mongolian historians have held up Chingünjav’s revolt as a he-
roic national resistance movement of the Mongolian people against Qing
oppression, while Chinese historians, by contrast, hardly mention it, or else
view it as nothing more than an outbreak of banditry. Western and Japa-
nese evaluations fall somewhere in the middle of the spectrum, finding evi-
dence of general resentment of Qing rule created by economic and adminis-
trative abuse, but seeing the Mongols as once again compromised by their
“fatal individualism,” with their loyalties focused far more on tribal units
and personalities than on national unity.
58As we shall see, the contrasting
interpretations of this revolt dictated by the anachronistic requirements of
modern nationalism mirror the later distortions of the historiography of
the entire Zunghar campaign, just as the fragmentation of the Mongols
which sabotaged their resistance and opened their society to foreign pene-
final blows 279

tration by Chinese merchants ironically foreshadows China’s own fate in
the nineteenth century.During the pursuit of Amursana, Qianlong was repeatedly disappointed
with his field commanders. One by one they let him down. Either they
failed to attack vigorously, letting their elusive enemies escape, or they
sought excuses to avoid dangerous encounters, or they failed to cooperate
with one another, allowing the enemy to slip through their uncoordinated
forces. Qianlong the desk general, confined to his palace in Beijing, showed
little understanding of or compassion for the constraints that faced his field
generals. Amursana’s successful flight from Ili brought down heavy recrimi-
nations and severe punishment from Beijing. The first victim was Yubao, who was blamed for permitting Amursana to
59Yubao claimed that his two thousand troops were too few. He had
also judged that with only four to five days’ rations left in the garrison, and
few horses, he lacked the resources for rapid pursuit, so he returned to Ili.
Suspicious of a cover-up, the emperor dismissed Yubao from his post but
allowed him to remain in the army to prove himself, demonstrating the
“ethic of conditionality” in action. Yubao was pardoned the next month
and demoted. One month later, however, he was arrested and sent to the
capital for punishment. The emperor’s wrath, however, soon turned on Yubao’s colleague
Tseleng, who was General in Charge of Pacifying the West (Dingxi Jiang-
jun) when Amursana escaped from Ili. Tseleng resisted orders to advance
rapidly against the fleeing Amursana, claiming that his troops lacked sup-
plies, but the emperor blamed him personally: “Leaders of troops should
suffer the same hardships as the troops, not indulge in extravagance and
waste.” Dardanga, by contrast, was promoted for his enthusiasm. Dar-
danga had advanced deep into Kazakh territory, while Tseleng sat at his
camp expecting new supplies. The emperor, however, had all fresh supplies
sent to the more vigorous Dardanga. He rebuked Tseleng for requesting
more troops, when he knew that Amursana had fled with very few men.
Tseleng wanted to leave five hundred men in Ili before advancing and to
withdraw detachments he had already sent out. Tseleng may have rightly
been cautious in the face of genuine supply problems, but the emperor
found his “stick in the mud” behavior outrageous. When Tseleng then told
Dardanga to withdraw his troops, after pulling back Yubao’s men to pas-
ture their horses, he drew an impeachment for military negligence. “The
way to use troops is to advance and not retreat,” lectured the emperor.
“The pursuit of the rebels is in the hands of timid, ignorant men.” How
could Tseleng claim that lack of rations prevented his advance, or worse,
required the withdrawal of already advanced troops? Would it not use
280 contending for power

more rations to send horses and men forward and back? 60Tseleng had ar-
rived at his camp by the western route twenty days earlier than Dardanga,
yet Dardanga had the surplus, while Tseleng was exhausted. Tseleng had
also paused to buy animals from nearby Mongolian tribes instead of merely
using the animals sent to him from Barköl. Tseleng and Yubao were ar-
rested and sent to the capital for punishment. The two unfortunate generals
were attacked by Zunghars while under escort back to Beijing.
Dardanga, by contrast, earned great credit for defying his fellow gen-
eral’s orders and engaging in aggressive, perhaps reckless, initiatives. The
emperor supported both his insubordination and his aggressive behavior,
rewarding him with the top general’s post, formerly held by Tseleng. Al-
though Dardanga won praise for killing and capturing many Kazakhs,
soon he too would fall victim to the emperor’s unrealistic expectations
about steppe warfare when he failed to capture Amursana. The approach
of winter forced the emperor to agree reluctantly to withdraw troops.
Ablai, the Kazakh chief, while denying that he knew Amursana’s location,
also suspiciously refused to provide evidence about him. With thunderous
threats, the emperor told Ablai that if he did not turn over Amursana,
the Great Army would return the next spring and completely exterminate
his tribe. Dardanga nevertheless had no choice but to withdraw without
success. Dardanga was showing signs of negligence in the emperor’s eyes. He had
sent an envoy to negotiate with Ablai, keeping two Kazakhs behind as hos-
tages, but the Kazakhs escaped because they were not guarded. In Novem-
ber 1756 Dardanga and commander Hadaha gathered their armies to-
gether for the winter but failed to coordinate their responsibilities. They
lost their double-eye peacock feathers for this error. The emperor had
shrewdly assessed another motive for Tseleng’s failure to advance: if he was
unable to pursue Amursana, he was more likely to be pardoned than if
he vigorously pursued him and failed to capture him. This was exactly
Dardanga’s fate. Blaming the cowardice of Yubao and Tseleng for Amursana’s escape al-
lowed the court to avoid confronting the constraints of the steppe. In the
court’s view, sheer human will could overcome the most difficult condi-
tions, if only the right man were in charge. Failure was due to “barriers in
the mind,” not material constraints.
61Even the yearly winter withdrawal of
troops, demanded by the severe weather, was accepted with great reluc-
tance. Each year the emperor, hoping to finish off the enemy, was tempted
to extend the campaign dangerously far into winter. In the end, Dardanga and Hadaha proved no more successful than their
predecessors. Both failed to capture Amursana, and they let the treacher-
final blows 281

ous Ablai escape. The enemy, the emperor charged, had “made fools of
them.” Dardanga’s army, it turned out, had been only one or twoliaway
from Amursana, a “cart ride away,” but unaware of Amursana’s location,
Dardanga had instead waited for the Kazakh Ablai to bring Amursana for-
ward. Hadaha had in fact met Ablai but failed to arrest him for his treach-
ery. Amursana’s nephew, now captured and under interrogation, revealed
that the two generals had narrowly missed an opportunity to capture
Amursana himself. But Amursana, knowing of the planned withdrawal of
the armies for the winter, had simply followed them on their retreat back to
his old haunts. He, in fact, was able to move much more quickly about the
steppe than the heavily armed Qing troops. Ablai had submitted when
faced with a force of only thirty men. With a little effort he could have been
caught a year before. And, like their predecessors, the two commanders
had failed to coordinate their actions. Dardanga, denounced as “hopelessly
incompetent” (hutu wuneng), and Hadaha, who was criticized for avoiding
all decisions, were ordered to face a military court-martial and blasted as
the “shame of the Manchus.” They were sent to Rehe to “wear armor” and
redeem themselves.
For the next year’s campaign, the emperor was determined to advance
troops much more deeply into more “remote, precipitous” parts of Kazakh
territory than Tseleng had been willing to enter.
63The failure of Tseleng and
Yubao to take responsibility must not be repeated by the new commanders,
Jaohûi (Ch. Zhao Hui) and Fude. They must cooperate with each other. The year 1756 had opened with two concurrent rebellions against Qing
authority by two of the most influential Mongol leaders: Amursana and
Chingünjav. Both threatened to mobilize more supporters. Amursana at
first claimed that he only wanted to return to the peace established under
Galdan Tseren, denying any intention to rebel. His disingenuous appeal to
precedent aimed to win over more Mongol allies. A smallpox epidemic in
Bortala, Amursana’s headquarters, forced him to flee to Kazakh territory
and seek allies there. Qianlong ordered a full-scale search, while inflicting
a scorched earth policy on Amursana’s abandoned pastures. His greatest
concern was the potential alliance of Amursana with the Kazakhs, the most
remote of the nomadic tribes known to the emperor, and the most difficult
to intimidate. The Qing ruler preferred to win them over as allies, but he of-
fered them both the carrot and the stick. If the Kazakhs captured and re-
turned Amursana, he would give them large rewards, but if they did not, he
would send a great army to wipe them out. By the winter of 1756–57 the emperor was a frustrated man. His armies,
backed by a huge supply train reaching from the plains of North China to
the steppes of Xinjiang, had harried Amursana’s men into the mountains of
282 contending for power

the Kazakh steppe, broken them up into small bands, and captured or ac-
cepted the surrender of thousands of his followers, yet the defiant man who
would be Khan still survived. Faced with two rebellions in critical regions,
and furious at Amursana’s betrayal, Qianlong rejected all leniency. He now
ordered the massacre of all Zungharian captives: “Show no mercy at all to
these rebels. Only the old and weak should be saved. Our previous military
campaigns were too lenient. If we act as before, our troops will withdraw,
and further trouble will occur.” In another edict he declared: “If a rebel is
captured and his followers wish to surrender, he must personally come to
the garrison, prostrate himself before the commander, and request surren-
der. If he only sends someone to request submission, it is undoubtedly a
trick. Tell Tsengünjav to massacre these crafty Zunghars. Do not believe
what they say.”
He clearly had to overcome resistance from local military commanders,
since he repeated his order several times, using the term jiao(extermina-
tion) over and over again. General Jaohûi was praised and awarded high
rank for reporting massacres, as was Tangkelu, who captured the Khoit
Chebudeng Dorji and “exterminated his followers.” Tangkelu was allowed
to incorporate his enemy’s families and cattle into his own tribe. Other
commanders, like Hadaha and Agui, however, were punished for merely
occupying Zunghar pastures while allowing the people to escape. The rem-
nants of Amursana’s shattered bands were to be tracked down even into
Russian territory and eliminated.
The emperor deliberately targeted young and able men in order to de-
stroy the Zunghars as a people. When Chebudengzhabu captured a group
of Khoits, whom he was going to award to the loyal Khalkhas, the emperor
instructed him to “take the young and strong and massacre them,” and
award only the women as booty.
66Even some Zunghar youths who surren-
dered after the defeat of their elders could not be spared, since “their ances-
tors had been chieftains.” They had to be executed or made bondservants
of the conquering soldiers.
67In 1756 the court had recommended the use of
food relief to win over the Zunghar people by giving them grain, tea, and
animals if they surrendered. Now the emperor implicitly recommended the
use of starvation tactics, commenting that it would be “easy to exterminate
rebels because they had run out of provisions.”
68 Old men, children, and
women were to be spared and sent as bondservants to other Mongol tribes
and Manchu bannermen, but they would lose their tribal identity; they
could not preserve their tribal (otoq)names or their titles, such as zaisang
(minister or clan leader). Reliable Mongols designated to supervise these
remnants took instead the Chinese official titles zongguanandfuzong-
final blows 283

After these edicts were issued, reports came in of Qing detachments pur-
suing rebel bands and slaughtering them by the thousands. Imperial orders
insisted that the captives must not be treated according to the usual rules
for dealing with bandits: “These are not ordinary cattle rustlers. They must
all be captured and executed. Why should we have to distinguish leader
and follower? These tribes have many bandits; if we do not completely ex-
terminate them [jiaomie jingjin], it will be no good for the Mongols and
merchants.” The goal was not merely to put down a rebellion but to “cut
off the roots” of the Zunghar resistance.
70 Russian governors in Siberia
heard that Manchu troops had massacred men, women, and children, spar-
ing no one.
Some signs of a more lenient policy appeared after mid-1757. By this
time, Amursana’s effort to unify the Zunghars had clearly failed, and no
longer posed a major threat. The imminent rebellion by the Khojas of
Turkestan meant that the Qing had to avoid driving other Zunghars over to
the side of the Turkestanis.
72Large numbers of Zunghars were submitting
to the Qing, who regarded them with great suspicion. Those who appeared
“completely trustworthy” were allowed to move to interior pastures after
September 1757, but the slightest indication of disloyalty justified their ex-
termination. The emperor clarified that he “did not formerly have the in-
tention [of eliminating the Zunghars]. It was only because they repeatedly
submitted and then rebelled that we had to wipe them out.”
Generals Jaohûi and Shuhede in the same month received criticism for fail-
ing to show sufficient zeal in exterminating the rebels. They had apparently
shrunk back from wholesale slaughter, despite continual prodding. By contrast to his policy toward the Zunghars, the emperor hesitated to
eliminate completely the Tibetan lamas of the Yellow Teaching. He was an-
gry with the lamas, whom he regarded as inciting the Mongols to resis-
tance, and he had ordered Tsengünjav to exterminate those he found upon
his arrival at Ili. But by July 1757, four Zunghar otoqhad surrendered, and
the court feared that a massacre of the lamas would unnecessarily alarm the
Zunghars. The rebel Nima tried to stir up the rest of the Mongols by saying
that the Qing “was wiping out the Yellow Teaching.” Soon, declared the
emperor, all lamas connected with his revolt could be killed, but for the
time being it was better to temporize.
74 Such tactical retreats from all-out
massacre in the interests of winning over possible Mongolian support did
not fundamentally alter the Qing’s basic aim, which was to eliminate all
prospects for autonomous Mongolian resistance. This deliberate use of massacre has been almost completely ignored by
modern scholars.
75 The massacre policy was a clear break from previous
Qing methods of managing relations with the Mongols. Until this time, the
284 contending for power

Qing rulers had primarily adopted the time-honored diplomacy of “using
barbarians to fight barbarians” by alternately supporting different factions
of nomads against one another, or else they executed individual ringlead-
ers of rebellions. But they had never attempted ethnic genocide before.
With this policy, the Qing succeeded in imposing a “final solution” to
China’s northwest frontier problems, which lasted for about a century. The
Zunghars disappeared as a state and as a people, and the Zungharian
steppe was almost completely depopulated. In his history of the Qing mili-
tary campaigns, theShengwuji,Wei Yuan, who estimated the total popu-
lation of the Zunghars at 600,000 people, stated, “Of several hundred
thousand households, 40 percent died of smallpox, 20 percent fled to the
Russians and Kazakhs, and 30 percent were killed by the Great Army. [The
remaining] women and children were given as [servants] to others...For
several thousand lithere was not one single Zungharian tent.”
was left as a blank social space, to be refilled by a state-sponsored settle-
ment movement of millions of Han Chinese peasants, Manchu bannermen,
Turkestani oasis settlers, Hui, and others. The use of massacre for ethnic extermination was also atypical of Qing
policy either before or after this time. The term “extermination” (jiao)had
been used before to describe the appropriate action toward non-Han bar-
barians. Manchus, of course, did use terror tactics against Chinese cities in
the lower Yangzi “pour encourager les autres.” The ten-day massacre at
Yangzhou in 1645 is the most notorious example.
78 Later suppressions of
domestic rebellions produced many capital sentences, but only after inter-
rogation and judicial proceedings. The theory of distinguishing between
ringleaders and “coerced followers” aimed to minimize the number of exe-
cutions. Many were pardoned after their sentences were reviewed at the au-
tumn assizes. Wholesale massacre of an enemy tribe was more typical of Central Eur-
asian practice. Even so, most of the killing was directed only at the young
male warriors who fought back. If a tribe surrendered, its men were usually
incorporated into the victor’s tribe, because nomadic conquerors needed
warriors. Many served as slaves or bondservants, but they might keep their
tribal identity or intermarry with the victors. Seldom did one tribe aim
to exterminate all the productive labor of its enemy. In the sparsely popu-
lated steppe, fighting men were scarce—and valuable. Women, children,
and older men were not eliminated after conquest. The Zunghars, however, could not be treated like other domestic rebels
(especially Han rebels). The Qing army did not need more Central Asian
warriors; it already contained Manchus and docile Khalkha Mongols. The
Zunghars were no longer out of reach, protected by logistical limits.
final blows 285

Kangxi had treated Zunghar recalcitrance as the product of his personal
struggle with the Zunghar leader, Galdan, and focused all his efforts on
eliminating him. He personally relished the idea of man-to-man combat (at
a distance) in the steppe. Qianlong, by contrast, stayed in the capital direct-
ing his field generals from afar. And he came up with a classically neat bu-
reaucratic solution to the unruly steppe: eliminate everything that moved,
and create a blank slate.The Zunghars were the one Central Eurasian people who, unlike the
Eastern Mongols, had ferociously insisted on preserving their autonomy
from Chinese rule. The Eastern Mongols had given the Qing the right to al-
locate their pasturelands, distribute relief grain, settle their disputes, and
levy troops and animals from them for its wars. They paid the cost of al-
lowing penetration of their territory by Chinese settlers, corruption of the
aristocracy by Han merchants and decimation by smallpox and other dis-
eases introduced by Han settlers, but they survived as a people to become
an independent state in the twentieth century. The Zunghars, by contrast,
completely disappeared. Only folk memory kept their struggle alive. In the final massacre, Qianlong bared his teeth. He had called himself a
ruler who showed equal favor to all, aiming to encompass a variety of dif-
ferent peoples under one harmonious realm. But those who resisted the im-
perial embrace faced extermination. The emperor’s edicts from this period
expose the tension within the mid-Qing between the ideal of benevolence
and the reality of repression. Taking a remarkably defensive tone, he de-
fended himself against unnamed critics who regarded the campaigns as
wasteful and cruel. Asserting that “the constant nature of the Zunghars is
greed, cruelty, craftiness, and the love of disorder,” he nevertheless insisted
that “the Zunghar masses are not basically strong willed,” and that it
would take only a few days to quell them:
Yet ignorant outside observers have raised irresponsible criticism.
They say this campaign should not have been begun, since, once it
started, it only caused more turmoil; wouldn’t it have been better not
to start at all? They really don’t realize that the Zunghars have caused
trouble on the northwestern border for over four hundred years. My
grandfather and father solidified the frontier against Galdan and Gal-
dan Tseren. Their repeated military expeditions made a long-lasting
peace on the frontier, and I have inherited this project from Heaven.
Now these barbarian tribes have become disaffected and their minds
have gone astray, offering an opportunity for revolt. How can I sit
back and let them be lost? Would I not be laughed at by later genera-
tions? How could I face my grandfather and father in the afterlife?
286 contending for power

This is why we made careful plans and toiled to do the unavoidable; it
was not out of a temporary love of glory, but to prevent border unrest
and diligently make long-term plans.
The emperor showed concern about his personal dignity and later repu-
tation, trying to ensure that others would give him credit for long-term
strategic planning and not merely a love of glory. He also defended his pol-
icy of extermination as a defense of “our Khalkha tribes” against violent
attack, arguing that he had intended to use only the classic “loose reign”
(jimi) policy that allowed border tribes a great deal of autonomy, but the
“ignorant” followers of Amursana had brought extermination on them-
selves by refusing to accept peace, contrary to Qianlong’s “basic senti-
ments” of generosity (benhuai).Finally, he argued that the great victories
on the frontier could not be thrown away by retreating and allowing the
nomad threat to revive. In addition, he noted, the total cost had been only
17 million taels, much less than the 50 to 60 million taels spent by Yong-
zheng. He concluded by proclaiming that he had dedicated himself to the
campaign “not because I abandon ease and love toil, but because the situa-
tion demands it,” because “the superior man daily concerns himself with
the people’s happiness and sufferings.”
80In this extended defense of his mil-
itary project, the emperor pleaded for sympathy from those who saw no
use in expensive campaigns on remote frontiers. By placing himself in the
lineage of conquerors since Kangxi, and emphasizing their ever-expanding
achievements, he promised to bring an end to a security issue that had
lasted for centuries. And he justified the complete elimination of those who
stood in the way. By October 1756 Qing troops had fought and defeated the Kazakh sup-
porters of Amursana and captured one of his zaisang.Amursana fled far-
ther west, still protected by the Kazakh Ablai. By November, winter cold
forced the troops to withdraw and wait until the next year. The Kazakh at-
titude toward Amursana had been ambiguous. On the one hand, Ablai of-
fered tribute and promised to capture Amursana, but, on the other, cap-
tured zaisang of Amursana reported that Ablai was giving support to the
81Ablai sheltered Amursana briefly in the summer of 1756, and of-
fered him a small number of men to resist the Qing, but soon expelled him
from Kazakh territory, realizing that he had little hope of victory. When
Ablai submitted to Jaohûi, he claimed that he had tried to capture Amur-
sana, but Amursana had stolen his horses and escaped. He expressed grati-
tude for Qing aid and promised the submission of all the Kazakhs. Less
than a week later, the Kazakh Nima was captured, and other Kazakhs sur-
rendered once they heard about Ablai’s submission.
final blows 287

Deprived of all allies, Amursana now offered to submit to Russia if the
Tsar would build a fortress for him between the Irtysh River and Lake
Zaisang to serve as a base for attacks on the Qing. The Tsar refused to take
any action that might lead to war and offered to protect Amursana only if
he agreed to resettle on the Volga along with the other Kalmyks. In early 1757 Jaohûi reported that Amursana had turned back from the
Russian border. Not knowing of Chingünjav’s fate, he turned east in a des-
perate effort to join forces. The emperor thought that now was the perfect
opportunity to capture Amursana, despite the winter cold, and that both
armies should make every effort to arrest him. He assumed that Amursana
could not escape to Russia because of the treaty agreements. The Qing ar-
mies divided into four units to track Amursana down. In July, Amursana
turned up at the Russian fortress of Semipalatinsk and surrendered to the
Russian commander. Discovering that he was fatally ill with smallpox, the
Russians quickly sent him to Tobolsk, where he was kept quarantined out-
side the city. He died there on September 21, 1757, at the age of thirty-five. The Russians concealed the facts of Amursana’s flight and death, hoping
to derive leverage from possession of his body. They told Qing envoys that
Amursana had drowned while crossing the Irtysh River.
82 Qing officials
dredged the river for a month but found nothing. Their suspicions grew. By
October 1757 Jaohûi was certain that Amursana had fled to Russian terri-
tory and that the Russians had lied about his death. They had, in fact, al-
ready buried him in Selenginsk. Repeated insistence by the Chinese on
Amursana’s return brought no results until February 1758, when the Rus-
sian representative admitted that Amursana had died of smallpox and in-
vited the Chinese to view the body at the Selenginsk fortress or at the
Kiakhta border. Even though Amursana’s body might have rotted by this time and be-
come unidentifiable, and the Russians made no mention of other followers
of Amursana captured with him, Qianlong decided to send two representa-
tives to the border not just to view the body but to reclaim it. He was con-
fident enough to announce Amursana’s death two days after receiving the
Russian report. But the Russians refused to return the body. The emperor,
however, insisted on obtaining Amursana’s corpse, saying: “The state only
needs to capture Amursana. When he has died, and his body is retrieved,
the entire Zunghar affair can be called a success.”
Wrangling with the Russians over the return of Amursana’s body caused
a major diplomatic conflict for many years. The Qing repeatedly sent letters
to St. Petersburg demanding the corpse, while the Russians replied that
there was no need to disturb their friendly relations over the issue of a
few rotten bones. The Qing put pressure on the Russians by placing their
288 contending for power

Orthodox monks in Beijing under house arrest, forbidding them to have
any contact with Han people, and threatening to cut off trade. The Qing
claimed that under the terms of the Kiakhta treaty of 1727, Russia was
obliged to return any refugees who fled across the border; but from the
Russian point of view, since the border with Zungharia had not been deter-
mined by 1727, Amursana and his followers did not qualify as “refugees.”
Russian intransigence also derived from a concern to protect the imperial
image in the eyes of other Central Asian peoples, who would be watching
carefully the relative balance of power between the two empires. Appar-
ently, the emperor never got Amursana’s body back.
84Unlike his grandfa-
ther, Kangxi, who could triumphantly display Galdan’s head and crush his
ashes on the military parade ground in Beijing, Qianlong had to settle for
an inconclusive victory. Chinese historians now claim that the Russians aimed to use Amursana,
dead or alive, to expand their control over Zungharia. They blame the
Amursana rebellion mainly on the Russians, who had tried to split apart
the “naturally united” Chinese Mongolian nationality. In their view, as a
divisive influence on the Mongolians, Amursana deserves a “totally nega-
tive historical significance.” Russian historians, by contrast, view Amur-
sana’s actions as a valiant effort of the Mongolian people to resist the ag-
gressive expansionism of the Qing in Central Asia.
85 Once again, the two
nationalisms mirror each other. Each ascribes essential unity to the Mongo-
lian people and blames conflict on intrigue by outsiders. Chinese charges of
Russian aggression are countered by Russian charges of Qing expansion-
ism. Just as the fate of the last Zunghar opponent of the Qing remained un-
clear even after his death, so do historical interpretations of the last days of
his struggle continue without resolution.
The Conquest of Turkestan
Qing wars did not end with the destruction of the Zunghar state. One more
set of campaigns penetrated even further into Central Eurasia, to the oases
south of the Tianshan range. I can describe these campaigns only briefly,
but they put the capstone on the empire’s relentless expansion, bringing
a large proportion of the Turkic oasis-dwelling population under Qing
The oases of southern Turkestan had been under the leadership of the
MakhdÄmzÀda Khojas, saintly families claiming descent from famous
shaykhs of a Sufi brotherhood of the fifteenth and sixteenth centuries. The
Zunghar rulers confirmed their power over Turkestan and relied on them
final blows 289

heavily for food supplies but kept them closely watched. Tsewang Rabdan
had declared Khoja Mahmut the leader of all Turkestan, but Galdan Tseren
kept him imprisoned in Ili. After Mahmut’s death and the first invasion of
Ili in 1755, Qing troops released his sons, BurhÀn ad-DÂn (Ch. Bulanidun)
and Khoja JihÀn (Ch. Huojizhan), sending BurhÀn ad-DÂn, known as the
Elder Khoja, back to Yarkand but keeping the Younger Khoja, JihÀn, in Ili.
Khoja JihÀn then joined Amursana’s revolt, but fled from the Qing army to
Yarkand, where he urged his elder brother to rebel, warning:If we follow the court’s orders to submit, we will be imprisoned in
Beijing, just like the Zunghars. My ancestors for generations were un-
der others’ control. Now, by chance, the powerful [Zunghar] state has
collapsed, and no one is pressing on us. If we do not seize this opportu-
nity to create an independent state, we will be slaves forever. That
would be disaster. The Central Kingdom has now taken Zungharia,
but has not yet decided its policy [toward Turkestan]. Its troops cannot
come here, and if they do come, we will resist them until their supplies
are exhausted. We can break them without fighting.
Khoja JihÀn evoked the same drive for autonomy and reliance on remote-
ness and difficulty of obtaining supplies as did the Zunghar state builders.
His strategic calculations were almost correct. His brother agreed and be-
gan a revolt by local begsandahunds (Islamic notables), who collected a
large army and fortified themselves in Kucha, the key town giving access to
the region from the east. Qianlong appointed Yarhashan General in Chief
of Repressing Rebellion (Jingni Jiangjun), giving him ten thousand men to
besiege Kucha. Expecting an easy victory, Yarhashan made no preparations
against a counterattack, preferring to spend his time playing chess. The
Turki army then burst out of Kucha on the evening of July 28, 1758, and
fled west. The emperor had Yarhashan called back and executed. The Khojas, however, found no support in Aksu, where the local beghad
already helped the Qing to capture Dawaci, so they moved farther west,
occupying Yarkand and Kashgar. Building walls and clearing the fields,
they devastated the surrounding land, driving the local population into the
city, so as to leave nothing for an advancing army. Now Qianlong sent his
best frontier general, Jaohûi, who had completed the suppression of the
Zunghars, south to the edges of east Turkestan to besiege the rebels. But at
the battle of Blackwater Camp, just outside Yarkand, the Younger Khoja’s
army, outnumbering Jaohûi’s forces by four to one, surrounded the Qing
forces and placed them under a desperate siege that lasted for three months.
Luckily for the Qing troops, they could dig for water, and they discovered
290 contending for power

underground grain storage pits, so they were able to endure steady attacks
until a relief column arrived from Ürümchi.Meanwhile, the Kirghiz attacked Kashgar, the stronghold of the Elder
Khoja. Khwush Kipäk, the former hakim begof Kashgar, advised Jaohûi on
the best invasion routes.
88 Both Khojas fled to Badakhshan, where they
were captured and killed. Once again, the Qing generals wrangled long and
hard with Sultan Shah (Ch. Suoledansha), the ruler of Badakhshan, to re-
trieve the two rebels’ corpses.
89After further threats, they finally obtained
the head of Khoja JihÀn and the corpse of BurhÀn ad-DÂn for transport
back to Beijing. On December 13, 1759, the Qianlong emperor proclaimed
the completion of the Zunghar campaigns, notifying “all subjects of the
Center and Peripheries” that with the “entry onto the registers” (ru bantu)
of Zungharia and the retrieval of the heads of the rebel Khojas, he had
achieved “eternal peace and security on the borders.”
The campaigns in Turkestan mimicked in brief the century-long
Zungharian adventure. The Khojas had sought to create an autonomous,
united state, protected from Qing power by its remoteness, but they were
undone by division between the oasis communities and the astounding lo-
gistical achievements of the Qing quartermasters. Once again, contrary to
modern nationalists, there was no unified “Uighur” nationality either fight-
ing against the Qing state or yearning to be incorporated within it. Despite
the Qing’s rapid success, some generals, like Yarhashan, demonstrated ex-
treme incompetence; Jaohûi’s travails, however, resulted from bad luck
rather than bad leadership. The armies’ composition was truly multiethnic,
including Mongol and Manchu generals, Han supply commanders, and
even some surrendered Zunghar troops. Although Qing forces now had
long experience with distant frontier campaigns, the emperor again dis-
played his “anxiousness to declare victory and retreat.”
91 He still tried to
minimize costs on the frontier, even at the risk of allowing small forces to be
nearly wiped out. Perhaps the easy victory over Amursana allowed the em-
peror and his commanders to think that they would have just as easy a time
in the very different conditions of the region south of the Tianshan. One final uprising indicated that the Central Eurasian borders were still
not entirely secure, and foreshadowed the greater troubles to come in the
nineteenth century.
92 In 1765 the town of Ush, a city of nearly 1 million
people in the Southern March of Altishahr, rebelled. The previous beghad
captured Dawaci and refused aid to the two Khojas, but the emperor, still
suspicious of his loyalty, had replaced his administration with carpetbag-
gers from Hami. The new begextorted grain and violated the wives of the
former beg,while the drunken military commander did nothing to stop
him. Two hundred and forty Turkestani porters began a revolt to reject op-
final blows 291

pressive corvée duties. They gathered local support. Two to three thousand
rebels did battle with local forces, defeating reinforcements sent from Aksu,
100 kilometers to the east. Fearing that thebegof Kokand might intervene,
the emperor ordered larger detachments from Ili and Kashgar. Exiled con-
victs were drafted to scale the walls of the besieged city. The Ili and Kashgar
commanders failed to cooperate with each other, dragging out the siege and
allowing many rebels to escape, but the city finally surrendered in Septem-
ber 1765. The Qing commanders killed over three hundred rebels and de-
ported tens of thousands to Ili. They then rebuilt the city from scratch, im-
posing strict controls on abuses by hakim begsand attempting to enforce
tight segregation of the Han and Muslim populations. The Ush rebels, like Chingünjav’s Mongols, resented both the Qing labor
levies and the presence of abusive Han officials and exploiting merchants.
Both attempted to garner wider support, but neighboring towns or tribes
generally refused to cooperate. Harsh repression brought tighter controls in
both Mongolia and Xinjiang, but unrest simmered in the latter. Drunken
criminal exiles touched off a riot in Changji in 1768, and in the early nine-
teenth century, Jahangir brought in a new rebel army from the west. Opium
smuggling from Kokand gave local officials the same problems with sup-
pression campaigns that plagued their fellow officials on the south coast.
Turkestan joined the empire as a by-product of the Zunghar campaigns,
but the elimination of the Zunghars did not by itself ensure firm control of
the region.
The Return of the Torghuts
In the early nineteenth century Thomas De Quincey observed:
There is no great event in modern history, or perhaps it may be said
more broadly, none in all history, from its earliest records, less gen-
erally known, or more striking to the imagination, than the flight
eastwards of a principal Tartar nation across the boundless steppes of
Asia in the latter half of the last century...Intheabruptness of its
commencement, and the fierce velocity of its execution, we read the
wild barbaric character of those who conducted the movement. In the
unity of purpose connecting this myriad of wills, and in the blind but
unerring aim at a mark so remote, there is something which recalls to
the mind those almighty instincts that propel the migrations of the
swallow and the leeming [sic], or the life-withering marches of the
292 contending for power

Out of the epic Long March of the Torghut Mongols from the banks of
the Volga back to their Mongolian homeland, De Quincey created a minor
classic of English literature. As historical fiction, his account rivals the
works of Sir Walter Scott in dramatic scenes, vivid characters, and melodra-
matic conflict. In fact, not all of De Quincey’s account is made from whole
cloth; the text he relied on, Benjamin Bergmann’s Nomadische Streifereien
unter den Kalmuken, relied in turn on oral Russian and Mongolian sources
which are not found in the official documents. The complete history of the
Torghuts, based on full use of Mongolian, Manchu, Russian, Chinese, and
Western documents, has not yet been told, nor is there room to tell it here.
I can only highlight its significance for the evolving Qing imperial vision
and the end of the last autonomous Mongolian tribe in Eurasia. Modern Chinese historians describe the Torghut migration as the inevita-
ble “return” of one of the empire’s lost peoples to the benevolent embrace
of the emperor, driven by Russian oppression. It is true that the Torghuts on
the Volga suffered from Russian military demands and from encroachment
by other settlers, but their incorporation into the Qing after their harrow-
ing journey was by no means inevitable. In fact, some of the Torghuts,
known as “Kalmyks,” stayed behind in the area where they still live today,
in an autonomous republic of Russia. And those who arrived on the Qing
border in 1771 did not get a uniformly enthusiastic welcome. Michael
Khodarkovsky has described in detail the experience of the Torghuts under
Russian dominion. Here I will recount only the debates about their admis-
sion to the empire that took place when they first arrived on the Qing fron-
tier. These debates reveal the initial uncertainty about the status of these
migrant nomads, and the legacy of Qing, Russian, and Zunghar interac-
tions in determining the decision to absorb these remnants of the “last
known exodus of a nomadic people in the history of Asia.”
Kho Urlük (?–1644) had led his people out of western Mongolia in order
to escape the turmoil of the early-seventeenth-century steppe and to find
pastures where they could live freely. Although Tsar Ivan had conquered
Astrakhan and Kazan’, the Russian state had not yet established a strong
grip on the lower Volga. Like the Don Cossacks, the Torghuts—or Kal-
myks, as the Russians knew them—formed nearly autonomous communi-
ties under their own chiefs. Although they swore oaths of allegiance to the
Tsar, they had more of the status of independent allies than of vassals. Un-
der Ayuki Khan (1669–1724), their greatest leader, they expanded their
control over a huge swath of territory of 300,000 to 400,000 square miles
in the steppes north of the Caspian Sea. They maintained contact with Mongolia and the Qing during this pe-
riod. They sent tribute missions and were allowed to join “boiled tea” mis-
final blows 293

sions to Kokonor. In 1698 Arabjur, the son of Ayuki Khan (1669–1724),
took five thousand men to Lhasa with Qing permission. Tulisen’s embassy
to Ayuki in 1712, ostensibly designed to facilitate the return of Arabjur, ap-
peared to the Torghuts as a Qing effort to recruit support against Tsewang
Rabdan, but the Torghuts refused to commit to an alliance. From time to
time they explored the possibility of returning to their homeland, but as
long as the Zunghars remained powerful, and at war with the Qing and the
Kazakhs, conditions in the steppe were too unsettled to allow them to
move. Over the seventeenth and eighteenth centuries, Russian administra-
tion gradually encroached on Torghut autonomy, with increasing demands
for military service coupled with efforts to intervene in local administra-
tion. Russian penetration of the steppe was similar to the Qing move into
Mongolia: in each case, Mongolian chiefs found themselves under growing
constraints while Russian or Han Chinese settlers moved into their terri-
tory. The Russians did not have the rigidly bounded banner units of the
Qing, leaving the Torghuts freer to move about, but they exacted substan-
tial military service from their vassals. In this respect, the Torghuts were no
different from the Cossacks and other mobile peoples settled on this steppe
frontier.The decision to leave the Volga in 1770 resulted from pressures building
up on the community and a fortuitous opportunity. Khodarkovsky finds
that “Russia’s growing control of Kalmyk administrative affairs, its exces-
sive demands for Kalmyk cavalry, the loss of prime pasture lands to ex-
panding military and agricultural colonies, and fear of coercive settlement
and coercion were among the critical factors contributing to the Kalmyks’
decision to depart for Jungaria.”
97 In 1762 Empress Catherine declared
that the Khan could not freely select his advisory council but had to obtain
the Tsar’s approval. New demands for Kalmyk cavalry to serve in another
war with the Ottomans in 1768 were the final straw. Catherine II arro-
gantly demanded 20,000 horsemen from the Kalmyk population of 41,523
tents, ignoring the Kalmyks’ need to defend themselves from raids by
neighboring Kazakhs and Kuban.
98Only 10,000 Kalmyks joined the cam-
paign in 1769, and they returned home in September, defying Russian com-
mands, in order to tend their herds over the winter. Internal rivalries for power also played a role. Tsebek Dorji Tayishi, the
leading Kalmyk noble, thought he deserved the Khanate, but under Russian
pressure the Kalmyks had chosen the young and irresolute Ubashi. To take
revenge on the hated Russians, and in the hope of establishing his own
power outside Russian territory, Tsebek Dorji planned the audacious es-
cape, drawing Ubashi along with him. Tsebek Dorji addressed a crucial
meeting of Kalmyk nobles, declaring:
294 contending for power

In all respects, your rights are limited by the Russians. Not only do
their officials mistreat the Kalmyks, but also the government itself
seems to have the intention of turning these independent steppe people
into settled peasants. The banks of the Yaik are covered with Cossack
fortresses; the northern borders are settled by immigrant Germans,
and soon you will be torn away from the Don and Terek, the Kuma
and Volga, by other settlements; your nomadic life will be limited to
waterless regions, and your herds destroyed. You have no other choice
in the future except to bend your neck under the yoke of slavery, or to
quickly leave the Russian empire to escape your destruction. The deci-
sion you make will determine your fate.
Ubashi Khan planned the mass exodus for early 1771, taking advantage
of the fact that not all the Russian soldiers had returned yet from the Otto-
man campaign. He wanted to wait for the Volga to freeze so that the
Kalmyks on the west bank could join the main body, but was forced to
leave early when the Russian governor discovered his plan. On January
5, 1771, over 30,000 tents, or 150,000 to 170,000 people, set out for
Zungharia. Ubashi brought about three-quarters of the nomadic popula-
tion with him. Although Ubashi told the Qing officials that the failure
of the river to freeze had forced him to leave 13,000 tents behind, those
who stayed back were mainly Derbet or Khoit tribesmen who disapproved
of the mass migration.
100 By June 1771 the emigrants had reached Lake
Balkash, eluding Russian pursuit but suffering great losses from winter
frosts, hunger, disease, and Kazakh raids. When they stopped there to re-
group, the Kazakhs surrounded them and inflicted devastating injuries on
their exhausted warriors and herds. By the time he contacted officials in Ili,
Ubashi had only 15,000 tents, or 70,000 people, left.
In this epic Long March, Ubashi had lost at least 100,000 people, and his
dreams of free pastures had been crushed. As C. D. Barkman notes, “It is
quite clear that the Torghuts had not intended to surrender to the Chinese,
but had hoped to lead an independent existence in Dzungaria.”
102 Only af-
ter their arrival in the old Ölöd pastures around Lake Balkash did they real-
ize that the steppes had no room for them. Qing armies had eliminated the
Zunghar state and induced the Kazakhs to pay tribute, while Russian for-
tresses marked the border across Siberia. “The Kalmyks had escaped Rus-
sian tentacles only to be ensnared in Chinese ones.”
Yet the frontier commanders were most reluctant to allow these long-lost
nomads to cross the border. Manchu documents in the special archive dis-
cussing the Torghuts reveal considerable suspicion and resistance.
104 Iletu,
the Manchu commander at Khobdo, first recorded receiving a Russian re-
final blows 295

port of Torghut movements on June 14, 1771, indicating that the Kazakhs
were blocking the Torghuts’ efforts to reach the border. The Russians had
demanded that the Qing not accept these “rogues and traitors,” Russian
subjects who had deserted military service.
105 Under the Kiakhta treaty, ref-
ugees, deserters, or criminals who crossed the border were to be returned.
Iletu was in a quandary. The Torghuts could not be forcibly returned to
Russia without causing a new war, but he saw no way to settle them near
Ili. He proposed to send them much farther east, into Chahar territory. The
emperor quickly denounced Iletu for his ignorance of frontier conditions:
he found no problem in mixing the Torghuts with other Ölöd tribes in
western Mongolia. He dispatched Shuhede, who had been suppressing the
rebellion in Ush, to Ili to take charge, limiting Iletu’s role to managing local
The Kazakhs, in the midst of war with the Torghuts, also called for aid.
They exaggerated Torghut strength, putting it at eighty to ninety thousand
tents, and as tributaries of the Qing, they expected military support, since
the Torghuts had plundered tens of thousands of their animals. Cleverly in-
voking the recent war with the Khojas, Ablai urged quick military action so
that the rebels could not flee into the distant steppe, where a large force
would be needed to track them down.
A further issue was the presence of Prince Shereng (Ch. Sheleng) among
the Torghut migrants. Shereng had joined Amursana’s revolt then fled into
exile. Now he was returning as one of the leaders of the Torghut ulusalong-
side Ubashi Khan. He claimed to want to submit, but could he be trusted?
Iletu was skeptical; he thought that Shereng was refusing to admit to his
crimes. Qianlong brushed aside such fears, concluding that Shereng’s sur-
render was sincere. The emperor had concluded remarkably quickly that the Torghuts, as the
final remnant of the Zunghars, who were now included as “our Mongols,”
deserved refuge within the empire. He rejected Russian demands on the
grounds that the Russians had refused to return Shereng and other support-
ers of Amursana.
108 The Torghuts were not refugees but subjects of the
Qing, even though they had left over a century earlier. Qianlong projected
the militarily established boundaries of the Qing backward, retrospectively
including within their embrace all the Mongols of times past. He was aware
that “if Ili were an empty land, the Torghuts might want to recover their
old pastures,” but now that it was filled with troops and walled cities, they
knew they had no choice but to settle in districts determined by the Qing.
He rejected Kazakh requests for aid, advising Ablai not to “fish for profit”
in turbulent waters. Although he regarded 90 percent of the Torghuts as
loyal, he admitted the need to guard against disruption by the disloyal 10
296 contending for power

percent. When Commander Batu Jirgelang wanted to levy twenty thousand
troops to guard the borders, however, the emperor rejected this request as
excessive. He thought that fewer than ten thousand would be needed. Re-
membering Chingünjav’s rebellion, the frontier commanders were espe-
cially worried about demanding fighters from the Khalkhas, so the emperor
stopped all levies of troops from them.
By July 29, 1771, Qianlong had overridden the objections and deter-
mined to settle the Torghuts within the empire. He ordered officials to
schedule an audience for Ubashi at Chengde, and to search for lands to
settle them. He had 200,000 taels of silver sent out from Anxi, Barköl, and
Ürümchi for relief supplies and travel costs. Further arguments in the
Torghuts’ favor derived from their feeling of alienation in Orthodox Chris-
tian Russia, where, the emperor claimed, they had not been allowed to send
missions to Tibet. By contrast, the Qing would permit the Torghuts to send
aocha missions to Tibet whenever they chose, after notifying the Lifanyuan.
When the Zunghars had had an independent state, control of aochamis-
sions had been a key tool for limiting their access to Lhasa; now, it served to
lure the Torghuts closer to the interior and away from the uncertain border.
Now that Tibet had “entered the registers” (ru bantu),in the Qing view, the
Yellow Teaching could bind the Mongols closer to the empire instead of
alienating them from it.
Although the frontier officials followed the emperor’s orders, not all of
them were convinced. Qishiyi’s account of frontier affairs, written in 1777,
displays a deeply skeptical view of the loyalty of the Torghuts. His perspec-
tive preserved a potent alternative to the benevolent imperial gaze, one
based more firmly on local experience of the turbulent frontier than on the
soothing banalities uttered in the capital. I shall discuss his views later. At least three commemorative stelae celebrated the return of the
Torghuts: one in Ili and two at the entrance to the miniature Potala Palace
in Chengde, the court’s summer villa north of Beijing.
112 The two at Potala
were written in four languages—Manchu, Chinese, Tibetan, and Mongo-
lian—although Manchu appears to have been the primary language of
the text. In the sycophantic but accurate words of Yu Minzhong, Hanlin
scholar and intimate adviser of the emperor and editor of his poems, “His
Majesty has discovered the difficult art of including many facts in very few
words, and of giving to these few words an extremely broad and deep
113 The inscription, titled “Record of the Return to Obedience of
the Entire Torghut Tribe” (Turhute Quanbu Guishunji), indeed summarizes
concisely the story of the arrival of the Torghuts on the border and their ac-
ceptance by the emperor in the face of opposition. According to Yu, the em-
peror wrote the text in little more than fifteen minutes, but he brilliantly
final blows 297

captured the essence of the event. The inscription frankly discusses the sus-
picions about the rebel Shereng, the frontier commanders’ demands for
more troops, and the emperor’s confidence that the Torghuts could do no
harm if they were given supplies and carefully watched. The text justifies
the unexpected and completely voluntary “return of the Torghuts to sub-
mission” (Ch.guishun,Ma.jihe bahaha) as a clear sign of Heaven’s will fa-
voring the benevolent emperor. Later I discuss in detail what these and
other stelae reveal about the empire’s sense of its historical position. They
put the public seal of completion on the entire great project that had begun
with Kangxi’s Galdan wars. Kangxi’s grandson could now claim convinc-
ingly that the “Western Regions have been fixed and will flourish.” He ex-
pected perpetual peace on the frontiers, because “of all the Mongolian
tribes there are none who are not the Great Qing’s subjects.”
The return of the Torghuts was, as the Russians would put it, the final
“gathering of peoples” under the all-encompassing Qing embrace. It was a
second ending to the millennial struggle of the settled empire with the
steppe. One people had been eliminated; another had been resuscitated and
returned to the fold. After extermination came revival, a satisfying conclu-
sion for the imperial project. Nationalists in the future would constantly in-
voke the trope of “return” (gui)as the metaphor describing the unification
of peoples under the new nation-state. The concept of history projected as a
cycle of exodus and return underpinned the imperial and nationalist states’
claims to legitimate rule over the multiple peoples under their domination. Unlike the imperial image, the Torghuts themselves did not prosper in the
long run. Ubashi enjoyed a sumptuous banquet at his audience at Chengde,
received many gifts from the emperor, and kept his title of Khan, but he was
no longer in charge of his people. They were split up into ten banners in
four leagues dispersed across hundreds of kilometers in northern Xinjiang,
in order to make sure that these Mongols would never unite to challenge
the empire.
115 Two banners under Shereng were settled near Khobdo, and
four others south of Karashahr. Qing officials forced many of them to take
up farming so as to prevent them from increasing their numbers. Qishiyi in
1777 found that many of the men had turned to banditry and the women to
prostitution. In hindsight, which group made the best choice: those who left the Volga,
or those who stayed behind? Did the Torghuts really gain greater “freedom
to lead their nomadic life, to preserve their customs, and to practice the
Buddhist religion than their brethren on the Volga?”
116 By 1947 the reset-
tled Torghuts in China numbered about 57,000, but under the People’s Re-
public they disappeared as an ethnic group into the general Mongol nation-
298 contending for power

ality. 117 By contrast, the 13,000 Kalmyks who remained now have their
own autonomous republic within Russia, with a population of 300,000 in
118 We could debate the wisdom of their choice, but it is clear that
when these last free nomads came under domination from the great agrar-
ian empires that surrounded them, the steppe ended, and a great chapter in
world history closed.
final blows 299

The Economic Basisof Empire

Cannons on Camelback:
Ecological Structures and Economic Conjunctures
A s we have seen, logistical limits repeatedly tied the hands of the
Qing generals. At Ulan Butong, for example, if Fiyanggû had not fortu-
itously run into Galdan’s retreating army, his men would certainly have
starved. The vast distances, barren deserts, and low-yielding lands of the
frontier protected the nomads while blocking Chinese dynasties from pro-
jecting power into the region.
Yet nomads also faced supply constraints. Although they, unlike Chinese
armies, could support themselves on the grasslands, their herds died when
severe winter storms covered the grasses with snow. The grasslands also
limited the size and mobility of large nomad armies, which depended on
quick forays and retreats, avoiding lengthy, static defense. As Galdan
learned at Ulan Butong, battles near the Qing capital brought disaster, but
he could still escape westward. Kangxi showed in his later campaigns, how-
ever, that he could lead troops all the way to the Selengge and Tula rivers,
near the heartland of Mongol power. The nomads were no longer protected
by distance. The millennia-long structural limits on Chinese power had be-
gun to break down. Strategic considerations drove both sides to alter the natural environ-
ment. Just as the Qing began to penetrate the northwest, the Zunghars, too,
began to exploit their own territory. Most accounts of the development of
Xinjiang ignore this crucial shift from a pastoral to a settled economy, but
like so many of their predecessors, the Zunghars built their state by com-
bining pastoral, agrarian, commercial, and mineral resources. Batur Hongtaiji had founded a capital near Kibaksarai (Kobuk Saur),

surrounded by cultivated fields. 2In 1644 he built three brick cities, includ-
ing monasteries, and asked the Tsar for chickens, pigs, hogs, carpenters,
stonemasons, and merchants. Cut off from China and eastern Mongolia, he
depended on Turkestan and Russia for grain and labor. “Bukharans” (war
captives from Turkestan) served as agricultural laborers and as the key
Central Asian trading links.
3Caravans heading for Moscow profited from
trade until the Russians stopped them in 1647. In the 1670s Sengge ex-
panded the use of Russians for trade and of captive Turkestanis to cultivate
fields. Zunghar troops forced the Russians to trade with them before allow-
ing them access to salt lakes.
4Galdan added trading missions to Beijing af-
ter restoring diplomatic ties in 1677. Twice a year he sent tribute missions
to exchange furs and horses for silk, cotton textiles, and money. In 1683 the
Qing cut them back in size from thousands of men to only two hundred,
forcing the rest to trade at the border. Beijing’s limitation of trading mis-
sions was one of the grievances that incited Galdan to attack the Khalkhas
in 1688. Galdan believed, with reason, that Beijing aimed to win over the
Khalkhas by favoring them with closer commercial ties.
Galdan the State Builder
The Ming loyalist statecraft scholar Liang Fen investigated Galdan’s early
efforts to build his economic base.
5Based on nearly a decade of travel in the
northwest, his book, Qinbian Jilue(An account of the northwest frontier),
written about 1687, describes how Galdan obtained minerals and gunpow-
der for his army: “[Galdan’s] wealth and population dominated the West-
ern Regions...[He] extracted oil from sand, baked earth to make sulfur,
and used sulfuric acid to make saltpeter, white as snow. Copper, lead, and
fine steel came from the ground. Rocks by the water’s edge produced gold
and pearls: [there were so many that] they put them aside and did not use
them. No one could surpass them in swift horses and numbers of barbarian
riders.” (As this passage indicates, he seems to have learned the technology
of Persian steel refining from his contacts with east Turkestan.) Liang Fen also notes:
[Galdan] did not obtain military supplies from distant places, because
he was very clever at making high-quality weapons himself. He made
armor with small links of chain mail, as light as cloth. If an arrow
could pierce it, he had the armourer killed. He also sent emissaries to
the Huihui [including Russians and Turkestanis], to teach him the use
304 economic basis

of gunpowder weaponry and battle tactics, using first fowling pieces,
then archery and swordsmanship. He ordered his soldiers to carry
fowling pieces, short swords, bows and arrows, and knives around
their waists. He loaded his cannon on camels. People who heard their
thunderous roar near and far submitted.
The cannons on camelback show that Galdan had transcended raiding,
supplying his army with heavy weaponry designed for permanent conquest.
He conquered Turkestan and Russian territories by adding overpowering
military might to the mobility of a nomadic army. According to Liang Fen,
Galdan also established a rudimentary taxation system by delegating a
traveling inspector to exact payments of horses, oxen, and sheep from fron-
tier tribes, and to keep careful track of income and expenses. This man,
who “represented Galdan’s eyes and ears...gathered up people and goods
in his net.”
To ensure regular income, Galdan demanded corvée labor from one-
cannons on camelback 305
Cannons carried on camelback. Although cannon were difficult to carry and of lim-
ited value in battle, both the Qing and the Zunghar armies highly valued these weap-
ons and employed foreign experts to cast them.
[To view this image, refer to
the print version of this title.]

third of the settled population each year in rotation and restricted the mo-
bility of the Mongolian tribes subordinate to him. In southern Turkestan he
collected tribute from the autonomous Khoja chieftains but prohibited the
sale of slaves, which would deprive him of manpower.
Tsewang Rabdan carried on Galdan’s projects of strengthening the state,
enhancing the productive resources of the local economy, and furthering
trade, but his dependence on the Qing increased. As noted in a previous
chapter, he returned Galdan’s ashes to Beijing only after the Kangxi em-
peror threatened to cut off the caravan trade. Although disputes continued
to break out with Russia over levies of tribute, refugees, and Russian mili-
tary expansion southward, Zunghar caravans traveled frequently to Semi-
palatinsk, Tobolsk, and Yamyshev and became a significant presence in Si-
berian markets.
9(See Table 8.1.)
Tsewang Rabdan and Galdan Tseren also developed agricultural produc-
tion at Ili, the Irtysh River, and Ürümchi by bringing in Turkic oasis dwell-
ers, called Taranchi, who knew the special skills of high-yielding irrigated
agriculture. A Qing soldier captured by Tsewang in 1731 reported seeing
wide fields and gardens, and even some Zunghars themselves began to
take up agriculture, in the form of military colonies, imitating Qing prac-
Finally, no discussion of the Zunghars’ construction of their economic
and technological base can ignore the extraordinary odyssey of the artillery
officer Lieutenant J. G. Renat and his fellow Swedes.
11By 1711 some eight
hundred Swedes captured by the Russians at the battle of Poltava two years
earlier had scattered across Siberia. They provided the Tsar with useful in-
formation and were especially skilled in mapmaking.
12 Renat then joined
Buchholz’s gold-hunting expedition to Lake Yamysh, where he and other
306 economic basis
Table 8.1 Trade between Zungharia and Siberia (via Yamysh and Semipalatinsk to
Tobolsk) (rubles)
Year Exports from
Zungharia Imports to Zungharia
17243,633 4,446
1725 3,621 3,691
1726 16,203 4,837
1727 11,679 4,041
1728 12,233 18,413
Total 47,371 35,430
Source: I. Ia. Zlatkin, Istoriia Dzhungarskogo Khanstvo, 1635–1758 (Moscow: Nauka,
1964), p. 381.

Swedes were captured by the Zunghars in 1716. He did not return to Swe-
den until seventeen years later.The Zungharian captives represented many nationalities. They com-
prised Russians, Swedes, Manchus, and Chinese, possibly including some
surveyors sent out by the Jesuits at Kangxi’s behest.
13Tsewang Rabdan had
them build factories to produce velvet, cloth, and paper; Renat was set to
work at military production. Supervising local corvée laborers, he pro-
duced at least fifteen cannon and twenty mortars during his stay. He also
taught the Mongols how to smelt iron for bullets and mine gold and silver,
as well as how to print books. Renat also aided the Zunghar force of five thousand men that attacked
Lukchun, near Turfan, and he fought against Qing troops in the Altai, help-
ing the Zunghars in their victory over Furdan. Major L. Ugrimov, a Russian
emissary who visited Galdan Tseren, mentioned that Renat had planted his
own garden on the model of the Taranchi Turkis, containing fruit trees sur-
rounded by a brick wall.
Even more crucially, Renat helped Galdan Tseren to produce the first
maps of Zungharia. (Galdan Tseren was no primitive nomadic chieftain; he
was said to carry with him a hundred camel loads of books.)
15 On his re-
turn to Sweden, he took with him two maps, with Mongolian place-names,
which are now held in the Uppsala University library.
16 One of them ap-
pears to be an original Mongolian map, drawn by Galdan Tseren himself.
The second one is a copy of a map captured from the Chinese at Barköl or
Turfan. These were the first Mongolian maps of Central Eurasia, and they
surpass in accuracy and detail anything done by the Russians or Chinese
during the eighteenth century. Although scholars debate the degree of Chi-
nese and Russian influence on them, they were unique hybrid cultural
objects combining Swedish cartographic techniques, Chinese forms, and
Mongolian local knowledge. They demonstrate the importance of cartog-
raphy to the military planners and the extremely accurate information they
could obtain from interrogating local people.
In sum, competition with the Qing state drove the Zunghars to un-
dertake significant steps toward “self-strengthening.” Like many earlier
nomadic empires, they established cities, developed agriculture, fostered
trade, and generated tax revenues, but the primary motivation was not “as-
similation” to settled societies’ customs but mobilization of resources for
defense. Internal upheaval after the death of Galdan Tseren in 1745, how-
ever, curtailed these investments. By 1748 the metal factories had been
abandoned as unprofitable, as the Zunghars reverted to roving raids and
internecine strife.
18 Within a decade Qing troops had taken advantage of
this internal disorder to eradicate the Zunghar state.
cannons on camelback 307

[To view this image, refer to
the print version of this title.]

Map of Zungharia drawn by the Swedish ar-
tillery officer Johan Gustaf Renat or by the
Zunghar ruler Galdan Tseren (r. 1727–1745).
Note the extensive labeling of rivers in Mon-
golian. South is at the top, following Chinese
tradition. Lake Balkash appears at the right.
(Numbers added by John Baddeley, a later
British commentator.)
[To view this image, refer to
the print version of this title.]

Nian Gengyao and the Incorporation of Qinghai
While the Zunghar leaders were reaching in all directions to gather re-
sources for their fragile state, their Qing rivals were also strengthening their
grip over newly acquired territory. After Mongolia, Kokonor (Ch. Qinghai)
became next on the list of incorporated regions. Unlike the piecemeal incor-
poration of Mongolia, the systematic planning for Kokonor exposes the es-
sential Qing methods for bringing the frontier into the empire.As described earlier, when the Khoshot Mongols of Kokonor threatened
to unite themselves under Lobzang Danjin and restore their control over
Tibet, the Yongzheng emperor designated Nian Gengyao as Fuyuan
Dajiangjun (Generalissimo in Charge of Pacification of Remote Regions) to
suppress this “rebellion.” After brutally defeating Lobzang Danjin’s army
in 1724, Nian outlined a plan for the reconstruction and incorporation
of Kokonor into the empire. Its provisions for military security, economic
development, and administrative reform were designed to ensure that
Kokonor, formerly an autonomous territory ruled by Mongolian tribes,
would become a permanent part of the Qing realm.
19Nian’s proposals epit-
omize the process of Qing imperial formation so well that they are worth
discussing in some detail. He first emphasized all the benefits that the
Khoshots had received from the Qing emperors, claiming all of Lobzang’s
tribesmen (tonglei)as subjects of the Qing. The initial task was to de-
termine rewards and punishments. Lobzang Danjin himself had escaped.
Three princes who aided the Qing suppression received titles. “Coerced fol-
lowers” (xiecong) could be pardoned, but it was decided that the active
leaders of the rebellion deserved execution. Nian explained the Qing deci-
sion carefully to an assembly of the Khoshot princes, outlining in detail the
crimes of eight men who could not be pardoned. Then he had the men
dragged before the assembly and had their heads cut off, in order to “rec-
tify the laws of the nation [guojia].”
Next, he fixed the territories of the Mongolian tribes. In his view, because
the Ming did not control the region, the hereditary lineages of Mongols and
Tibetans had maintained their autonomy, leading to continual plundering
and conflict. Now that the rebels had been rooted out, the Mongols would
be organized into banner companies, modeled on those introduced into In-
ner Mongolia. The pasturelands’ boundaries would be fixed, and the Mon-
gol leaders would be named jasaks:banner commanders subject to confir-
mation by the Qing court. Each tribe would be allocated to a separate
grazing ground, and no tribe could interfere with another tribe’s pasture-
lands. Following tradition, every year an assembly leader (mengchang)su-
pervised the annual meeting of the Khoshots. Lobzang had used his status
310 economic basis

as hereditarymengchangto consolidate his support, but Nian was de-
termined to prevent the emergence of another powerful unifier. Now the
mengchang would be selected by Qing officials.
Classification was the third effective control technique. Nian exploited
divisions among the Mongols to serve imperial unification. Some of the
Khalkha Mongols, instead of submitting to the Qing at the great assem-
bly of 1690, had fled south and sought protection from the Khoshots of
Kokonor. Now that the Qing had taken over this region, these Khalkhas
had the opportunity to free themselves from Khoshot domination by ac-
cepting Qing rule. Their chiefs would become jasaks,and they could take
over territory confiscated from Lobzang Danjin’s rebels. Nian explicitly
stated, “Thus we will divide the strength of the Kokonor princes, and the
Khalkha princes will no longer suffer the shame of being slaves; they will
become their own tribes.”
21 These Khalkhas were explicitly defined as a
new tribe (buluo)and settled separately from the Khoshots in Kokonor.
Thus, internally, the Qing constructed new ethnic communities in order to
divide and rule, just as its foreign policy of yiyi zhiyiplayed off one “bar-
barian” against another. Similar distinctions were applied to the Tibetans (Xifan) in Kokonor,
whom Nian regarded as the original inhabitants of the region. They had
joined Lobzang “in swarms,” saying “they only knew of the existence of
Mongols, and knew nothing of Chinese civil administration or garrisons.”
But now, Nian proclaimed, they were “our common people [baixing]...
their lands are our lands; how can they serve the Kokonor princes?”
These native Tibetans had to cut their ties to lamaseries and Mongolian
lords in order to become loyal subjects owing duties directly to the imperial
administration. They became another defined population within the terri-
tory. Nian thus aimed to sort out multiple populations who owed obliga-
tions to different superiors into defined groups subordinate to only one su-
perior authority. Enforcing control over the local Tibetans also meant challenging the
Dalai Lama’s suzerainty by splitting up the territorial boundaries of Tibet.
Once again, Nian justified this division historically. Of the four tribes of the
Tangut, two, Zang and Wei in western Tibet, belonged under the Dalai
Lama’s jurisdiction, but Kokonor and Kham in eastern Tibet, because they
were dominated by Gusri Khan, deserved a separate classification.
23 After
Lobzang’s defeat, the Kham region was to be attached to the neighboring
interior provinces of Sichuan and Yunnan. Nian did not regard this reorga-
nization as taking territory away from the Dalai Lama; his “lands of in-
cense and fire [xianghuo]” in the west remained under his control, but the
Qing would “save several hundred thousand Tibetans from water and fire”
cannons on camelback 311

by detaching them from the Dalai Lama’s jurisdiction. 24 In return for the
loss of revenue, the Qing would make annual payments to the Dalai Lama
and allow him to conduct trade at Dajianlu on the Sichuan border. This di-
vision of the Tibetan cultural region by the Qing corresponds closely to its
current administrative boundaries under the PRC. Further discipline of Kokonor required severe limitations on the power
of the lamaseries. These were some of the largest establishments in the en-
tire Tibetan region, containing up to three thousand monks and many in-
terconnected buildings.
25They were significant centers of power, supported
by dues from local Tibetans. From Nian’s point of view, the lamaseries
could not be authentic religious institutions, because they gave refuge to
“criminals,” stored weapons, and supported the rebellion. It was perfectly
proper to burn them down and massacre their residents. Afterwards, the la-
maseries would be limited in size to three hundred monks, all of whom had
to register with local officials and undergo twice-yearly investigations. Rent
payments could not go directly to the lamaseries but had to be submitted to
the government for distribution to them. In sum, after eliminating a minority of “rebels,” the Qing administrators
organized the remaining “coerced followers” into fixed “tribes,” defined
as territorially and administratively circumscribed units, with leaders ap-
pointed by the state. They intermingled different groups so as to balance
them against one another, distinguished “natives” from later arrivals, and
replaced multiple allegiances with one direct line of authority. On these frontiers, the Qing rulers faced the same challenges as other
Eurasian empires, and used the same tools. The Russian and Ottoman em-
pires also defined stable borders and sharpened their categorizations of
peoples in the eighteenth century, and the British did the same in India in
the nineteenth century.
26All of them conducted ethnographic surveys of the
new regions, fixing identities in place for fiscal convenience and preser-
vation of local order.
27 They replaced the fluctuating alliances of the au-
tonomous nomadic society with fixed, hierarchically determined posts of
authority. This drive toward clarifying the state’s vision of its peoples, how-
ever, did not erase all difference. The bureaucratic structure had to adapt to
the frontier; it did not replicate the uniformity of the interior. Jasakswere
not merely district magistrates in Mongol clothing, nor were they Manchu
bannermen. (Later on I discuss further examples of Qing administrative di-
versity in Xinjiang.) At the same time, Nian promoted immigration in order to forge links to
the interior and create a more peaceful, settled society. He proposed to send
ten thousand Manchu and Han settler households into Kokonor to “di-
lute” the strength of the Mongols and turn them toward stable cultivation.
312 economic basis

These policies partially contradicted the state’s urge to simplify. With one
hand local officials untangled tribal affiliations by creating separate territo-
rial and kinship entities under an orderly administrative structure, while
with the other they introduced more diversity into the region by adding
new peoples who could clash with the indigenous inhabitants. Nian’s poli-
cies contained in the microcosm of Kokonor the tensions that strained the
entire empire after its great push outward.Nian likewise fostered trade to bind the frontier to the center. Before the
conquest, in Nian’s view, the Mongols had traded as they pleased, exchang-
ing “useless hides and furs for our useful tea and cloth.” Han traders in
search of profit headed for the territory, creating a “spirit of wickedness”
28Free trade relations also had allowed Lobzang to spy out condi-
tions in the interior before he rebelled. Now trade at the frontiers would be
regulated. Mongolians would be divided into three groups; each group
could come to the capital on a licensed trade mission once every three
years, in rotation. Over nine years, all would have a chance to trade. Regu-
lar trade at border markets was allowed twice a year. Troops would patrol
the markets to ensure that no one crossed the border without permission. These Central Eurasian precedents set the pattern for other treaties.
Nian’s proposals of 1724 anticipated the similar trade regulations that the
Qing negotiated with the Russians in 1727, in which border trade was con-
fined to the town of Kiakhta, and tribute missions to the capital were al-
lowed only with official permission. The Canton trade system of handling
Western merchants at the end of the eighteenth century embodied the same
principles. As Joseph Fletcher has shown, China’s first “unequal treaty”
settlement was negotiated with the Khan of Kokand in 1835. Its provisions
for extraterritoriality, merchant autonomy, and permanent resident politi-
cal representatives were applied equally to the British after the Opium War
in 1842.
29Nian’s proposals demonstrate that the same was true for an ear-
lier period: regulated trading arrangements in Kokonor set the framework
for the Russian and British treaty arrangements. Finally, Nian proposed to construct a huge fortified border along
Kokonor’s northern frontier. He envisioned a connected series of fortresses
that would in effect extend the Great Wall line of defense far out the Gansu
corridor to Ganzhou and Anxi, cutting off Kokonor from contact with the
Zunghars to the north. He would clear out all Mongols from this area and
bring in large numbers of settlers to populate the garrison towns. As Nian
noted, and as the Qing found later in Xinjiang, criminals sentenced to mili-
tary exile would make ideal cultivators of the soil. Nian did not realize all of his grand vision. The unnecessary fortress line
was never built, and migration to Kokonor was slow. Yet his proposals un-
cannons on camelback 313

cannily forecast the primary measures of frontier control of the Qing, Na-
tionalist, and PRC governments. Repression, settlement, state simplifica-
tion, migration, and commercial integration sum up the policies of all three
regimes. Yet the empire was less thorough, and thus more benign. The Qing
repression of revolt in Kokonor was brutal but not wanton. The large la-
maseries were dismantled and reestablished on a limited scale, with re-
duced influence, but they were not eliminated. Unlike its Nationalist and
Communist successors, imperial ideology was not hostile to Tibetan Bud-
dhism per se; it was concerned only with the potential for the institution to
create centers of resistance. Under appropriate restraints, the monks could
facilitate imperial control, so long as the emperor put himself at the top of
their hierarchy.We may note some surprising parallels between Kokonor’s situation and
current events in Tibet.
30Both feature the appearance of rival incarnate la-
mas, one backed by the orthodox hierarchy, one by the Chinese; the use of
forcible intervention to impose a solution on Lhasa; and the concern to pre-
vent foci of resistance to state control. But the Qing never conducted all-out
ideological campaigns against Tibet’s religion; Mao’s notorious (apocry-
phal?) statement to the Dalai Lama that “religion is poison” would never
have been uttered by a Qing emperor. During the Cultural Revolution,
thousands of monastic establishments were destroyed by Red Guards, and
many monks were humiliated and killed. The Manchus inflicted only a
nasty, sharp military shock on the resisting lamas, after which they allowed
the institution to continue. Except when they had their own militarily dom-
inant empire in the seventh to ninth centuries ce, Tibetans have always
been in the unfortunate position of choosing between unsatisfactory, vio-
lent, or indifferent protectors. Mongolian patronage did help the Dalai
Lama centralize his state from the sixteenth to the seventeenth century; but
the eighteenth-century Zungharian intervention in internal conflicts left the
Tibetans with the choice of only one patron: the Qing emperor in Beijing.
Yet after the 1720s, Tibet, Mongolia, and Kokonor remained at peace un-
der loose Qing suzerainty. Not until the twentieth century did nationalist
upheaval bring new threats to Tibetans and Mongolians under Qing ad-
Administering the Frontier
By eliminating the rival Mongolian state, the Qing rulers added a huge ter-
ritory to the northwest region. It took an entire century to digest the new
lands and peoples. Many of these lands were added on to the existing pro-
314 economic basis

vincial administration of Gansu province. Within the northwest macro-
region, defined by G. William Skinner as including Gansu, Shaanxi, west-
ern Shanxi, and a piece of Sichuan, Gansu was the most peripheral part. It
formed the hinge between the regular administrative structure of the core
and the new requirements of the frontier regions. Imperial policies on grain
provisioning, taxation, and military support had to be adapted to the spe-
cial circumstances of this frontier, and they were stretched further when the
even more alien regions of the far northwest were added on. But the prac-
tices established as variations within a bureaucratic norm in Gansu served
as models for greater experiments in Xinjiang. In this section I examine the
position of the northwest region, Gansu especially, within the empire as a
whole, and the processes of administrative expansion and economic change
that drew Xinjiang closer to the interior.
As the empire expanded, officials created new administrative units and
reshuffled old ones to adapt to new needs. They redirected bureaucratic re-
sources not only through material flows but also through alterations in the
structure of field administration. G. William Skinner and Gilbert Rozman
have analyzed the institutional structure of the Qing empire by showing the
distribution of administrative levels in relation to the urban hierarchy.
Skinner, in particular, has demonstrated that many aspects of the formal
designation of bureaucratic posts were allocated in accordance with the
two crucial functions of revenue collection and military defense. Skinner’s
model, however, is a static analysis, reflecting generally the structure of the
empire in 1893, though some of his data are estimated for 1843. We also
need to examine how bureaucratic posts changed in response to the in-
crease in territory and economic activity. At the top of the administrative hierarchy of China proper stood the
Governor-General (zongdu),which first became a regular position under
the Qing.
33 He usually supervised two or more provinces and served con-
currently as governor of one of them. From the founding of the dynasty un-
til 1760, the numbers and jurisdictions of the Governors-General shifted
constantly, until settling down to a total of nine (eight in China proper and
one in Manchuria). As Figure 8.1 shows, the northwestern regions were
particularly unsettled. In the early Qing, Gansu and Shaanxi were grouped
together as the Shaanxi Sanbian, while the Sichuan Governor-Generalship
included Huguang. In 1653 Sichuan was added to Shaanxi and Gansu and
separated from Huguang, but in 1661 it was split off again under its own
Governor-General. Shanxi was added to Shaanxi and Gansu for a short
time, and Huguang was again joined with Sichuan, but by 1674 Shanxi had
no Governor-General, Shaanxi and Gansu were joined, and Sichuan was
separate from Huguang. This would ultimately be the stable configura-
cannons on camelback 315

tion, but the jurisdictions shifted back and forth for the next eighty-six
years, with Sichuan sometimes joined with Shaanxi and Gansu, and some-
times separate. Gansu briefly gained its own Governor-General during the
final years of the conquest of Xinjiang, but in 1760 the partition of respon-
sibilities was finally resolved with three separate Governors-General for
Shaanxi–Gansu, Sichuan, and Huguang. This highly volatile division of the
northwest indicates the changing responsibilities of the Governors-General
of the region.The Governors-General here, because of their military responsibilities,
dominated over much weaker provincial governors. For the same reason,
only Manchus could serve as provincial governors of Shanxi, Shaanxi, and
Gansu until the mid-eighteenth century.
34 The shifting boundaries clearly
reflected changes in the military demands on the northwest, such as the
war against the Jinchuan rebels in Sichuan in 1749 and the invasions of
316 economic basis
Provinces Total G-G
Year Shanxi Gansu Shaanxi Sichuan Hubei Hunan
1645 X 4
1649 X
1659 X
1661 X
X 15
1674 X
X 9
1684 O
O 6
1724 O
1734 O
X 10
1738 O
1748 O
X 8
1759 O X
1760– 1906 O X 8
ShaanGan Governor-General
ChuanShaan Governor-General
Huguang Governor-General
X Governor-General rules a single province
O No Governor-General for this province
Figure 8.1 Qing northwest and central Governor-Generalships (major changes
only). Only six of the eighteen provinces of Qing China are shown. Totalindicates
total Governor-Generalships for all provinces. Source:Qian Shifu, ed., Qingdai
Zhiguan Nianbiao, 4 vols. (Beijing: Zhonghua Shuju, 1980), 2:1510–11.

Xinjiang in the 1750s. After this period, Gansu had no provincial governor.
The Governor-General moved to Lanzhou and took charge of the main mil-
itary and economic functions: funneling supplies along the Gansu corridor
to the Central Eurasian frontier. After a century of turmoil, the mid-eigh-
teenth-century decisions fixed the definition of most boundaries of adminis-
tration.The distribution of the Governor-Generalships indicates that Qing ad-
ministration at the highest level responded to physiographic constraints.
After 1760 the boundaries of most of the eight Governor-Generalships of
China proper (excluding Manchuria) corresponded quite closely to the
eight physiographic macroregions described by Skinner. (See Table 8.2.)
To be sure, as Skinner insists, administrative boundaries never fit exactly
with physiographic regions at any level of the marketing hierarchy. The two
with the worst fit were the Governor-Generalships of Zhili and Liangjiang.
The Zhili Governor-General did not encompass the entire North China
plain, since Shanxi, Henan, and Shandong had no Governor-General at
all. The Liangjiang Governor-Generalship included Jiangsu, Anhui, and
Jiangxi, even though the northern parts of Jiangsu and Anhui belonged to
the North China plain, and Jiangxi belonged to the middle Yangzi macro-
region (or was a separate macroregion). Broadly speaking, however, the
other Governor-Generalships after 1760 accommodated themselves to the
existing physiographic regions, and those which fit best were on the edges
of the empire. Since the primary functions of the Governor-General were
those of military and logistical coordination, these men were especially
valuable in allocating resources efficiently for imperial expansion on the
frontiers. In the early Qing, however, when the constant shifting of positions indi-
cannons on camelback 317
Table 8.2 Governor-Generalships after 1760 and macroregion overlap
Governor-Generalship MacroregionDegree of overlap
YunGui YunGuiHigh
Liangguang LingnanHigh
Minzhe Southeast coastModerate
Huguang Middle YangziHigh
Sichuan Upper YangziHigh
ShaanGan NorthwestHigh
Liangjiang Lower YangziLow
Zhili North ChinaLow
Source: G. William Skinner, ed., The City in Late Imperial China (Stanford: Stanford
University Press, 1977), pp. 214–215; G. William Skinner, “Presidential Address: The Structure
of Chinese History,” Journal of Asian Studies 44, no. 2 (1985): 271–292.

cated the fluctuating pressures generated by territorial expansion, these
units did not fit closely with physiographic macroregions. It took over a
century of boundary juggling to bring economic structure and administra-
tive hierarchy into rough correspondence. From 1661 to 1760, the number
of Governor-Generalships grew as large as fifteen and as small as six. (See
Figure 8.1.) It is noteworthy that each emperor in the early years of his
reign experimented with different configurations: Kangxi briefly increased
the number of Governors-General to fifteen, and Yongzheng decreased
them to nine. Qianlong kept the total at eight, but shifted Sichuan back and
forth, and for a short time split Yunnan and Guizhou.A debate over the northwestern Governor-Generalship indicates the con-
flicting logics determining its boundaries. In 1759 ShaanGan Governor-
General Yang Yingju memorialized to request an adjustment of military re-
sponsibilities in the northwest.
36He argued that the conquest of the western
regions had made his responsibilities too great, so he proposed to make
himself the Governor-General of Gansu alone, and put Shaanxi under the
Sichuan Governor-General, changing the jurisdictions of many of the mil-
itary units. Yang wanted to ensure closer coordination and greater cen-
tralization of control over interprovincial transfers. He insisted that com-
munications between provinces should go through the Governor-General,
eliminating all lateral communications from one governor to another,
and all military financing must be approved by the Governor-General in
The Deliberative Council rejected his proposal; military campaigns were
still continuing, and the councilors did not want to cause disruptions.
Guyuan military commander Zhang Jietian proposed to station the new
Gansu Governor-General in Suzhou, in the far west of the province, so as to
ensure closer supervision of the western regions. But the emperor called this
a “muddled” idea, and argued in favor of keeping the official headquarters
in a more populated region in the east: “Liangzhou is better because mer-
chants can flock there, and it is convenient. Suzhou is far away, and it has
been neglected. These military people don’t understand this. For control-
ling the newly subjected tribes, Suzhou is closer. But we should not decide
according to private interests.”
38In effect, the emperor rejected arguments
based on military security alone in favor of greater revenue collection from
the more densely populated regions in the east. He and the top officials
knew that revenue and security were the prime determinants of administra-
tive boundaries, but their imperatives often conflicted. At the highest and
lowest levels of the hierarchy, the Qing constantly tinkered with bureau-
cratic boundaries and posts in response to these two demands. Turning to provincial administration, I examine changes in Gansu to il-
318 economic basis

lustrate how the Qing created a frontier provincial structure during the
eighteenth century, and how its experience provided a model for the much
larger task of incorporating Xinjiang at the end of the century. Gansu prov-
ince was created when Shaanxi province, inherited from the Ming, was
split into two Provincial Administration Commissions, Left and Right; the
Right Commission became Gansu province in 1667.
39 It contained four
prefectures and inherited the Ming structure of military units (weisuo),
which were autonomous garrison commands stretched along the north-
western frontier. The Qing gradually incorporated these units into the civil
administration. There were forty-nine weisuowithin the western section of
Ming Shaanxi. Twelve of these had been abolished in the Shunzhi reign and
four more during the early Kangxi period. But it was the Yongzheng reign
which carried out the most drastic reduction, abolishing thirty-one weisuo
and putting them under the civil administration. (See Table 8.3.) The reduc-
tion of weisuo was in accord with Yongzheng’s policy of reducing military
costs by putting more territory under civilian control.
Two key elements of the Qing field administration were the “span of con-
trol” and strategic designation of posts. Span of control refers to the num-
ber of subordinate units under a higher-level one. Skinner uses this term to
refer only to the number of county-level units under a prefecture, but it
can be applied at any level.
41In the Qing, the span of control at the center
(the number of Governor-Generalships, or the number of provinces) varied
from nine to fifteen for Governor-Generalships, and from fifteen to twenty-
two for provinces. At each level of the hierarchy there were several different
kinds of offices, each of which had a particular function. (See Figure 8.2.)
Creating new units at any level increased the span of control of the higher
units, but also added additional officials to the bureaucracy and reduced
the span of control at the next lower level. Thus, creating a new province
cannons on camelback 319
Table 8.3 Incorporation of military units into civil administration in Gansu
Reign Number of military
units abolished
Shunzhi (1644–1661) 12
Kangxi (1662–1722) 4
Yongzheng (1723–1735) 32
Qianlong (1736–1795) 1
Total 49
Source:Niu Pinghan, Qingdai Zhengqu Yange Zongbiao (Beijing: Zhongquo Ditu
Chubanshe, 1990), pp. 473–477. Note: An additional six weisuowere created from 1718 to 1724 and abolished in 1759.

(such as Xinjiang in 1884, or Taiwan in 1885) increased the span of control
of the capital by adding new provincial governors, but reduced the span of
control of provinces over prefectures. Creating new prefectures would in-
crease the provincial span of control but reduce the number of county-level
units per prefecture. Increasing the number of salaried officials was some-
thing that Chinese dynasties were reluctant to do, because it added to their
fiscal burdens, but reducing the span of control increased security by mak-
ing each official responsible for fewer subordinate units. Once again, fiscal
and security considerations stood in an uneasy balance against each other.The administrative units of the imperial bureaucracy were remarkably
stable at the lowest level but quite variable at higher levels. The number of
county-level units has changed little: the Qin empire of the third century
bce had about 1,000 xian,and China today, with a much larger territory
320 economic basis
[1b, 8–9]
[2b, 18–22]
[4a, 68–79]
[4b, 214]
[7ab, 1381]
[jasaks, banners,
begs, etc.]
[5a, 75]
[5a, 54]
[5b, 139]
[5a–6a, 571]
Figure 8.2 The Qing administrative hierarchy. Numbers in brackets indicate rank
in the bureaucratic hierarchy, followed by the total number of units, which varied
over the course of the dynasty. The figure follows Brunnert, which reflects the situa-
tion around 1907; Hucker and Skinner give slightly different numbers. For xian,the
common English term is “district” or “county.” Zhouandting are often called “in-
dependent subprefectures.” Sources:I. S. Brunnert and V. V. Hagelstrom, Present
Day Political Organization of China (Taibei: Book World Co., 1911), pp. 395–438;
Charles O. Hucker, A Dictionary of Official Titles in Imperial China (Stanford:
Stanford University Press, 1985); G. William Skinner, ed., The City in Late Imperial
China (Stanford: Stanford University Press, 1977), pp. 301–307.

and population, has 2,143 county-level units, many with nearly the same
42An expanding dynasty like the Qing, however, did have to in-
clude more territory under its civil administration, so it converted military-
ruled regions into new civil administrative units. The culmination of this
process at the highest level was the designation of Xinjiang and Taiwan as
provinces in 1884 and 1885, but analogous changes had been made at
lower levels since the eighteenth century.
Shifting units between levels was a more common, and more attractive,
administrative change than creating new units because it did not increase
the number of officials, although it could change their ranks and salaries.
Often the motivation was to increase supervision over strategically impor-
tant areas. When counties (xian)became zhilizhou orzhiliting, they were
placed directly under the provincial governor instead of having a prefect in
between. The governor appointed especially experienced men to fill these
posts. According to Skinner, the designation of posts and the span of control
varied by region in relation to the two prime concerns of the empire: reve-
nue collection and defense. The span of control was to be high in regional
cores and low in the peripheries so as to maximize the tax collection from
the most prosperous areas and maximize supervision of authorities in stra-
tegically important peripheries. Zhilizhouandting were to be concentrated
in the peripheries because of their special status as units directly adminis-
tered by the provincial governor. Gansu, however, when it was created in 1667 out of Shaanxi province,
did not fit this pattern at all. (See Table 8.4.) It was the most peripheral
province of the empire, but with four prefectures and thirty-seven counties,
its span of control, at 9.25, was very high.
43It contained no zhouorting at
all. In effect, the administration was split into two parallel hierarchies: the
weisuo military units and the civilian structure. Skinner’s generalizations
apply better to the “normalized” empire of the late nineteenth century, af-
ter it had digested its new territorial gains, thanks to the early Qing. In the
seventeenth century, new provinces like Gansu were quite anomalous, but
over time the administrative units were adjusted to bring them closer to the
normal structure. These shifts were especially common in northwest China. The short Yongzheng reign probably marked the most intensive reor-
dering of administrative units in the Qing dynasty until its very last years.
In Gansu, Yongzheng created four new prefectures out of the garrisons
and raised four subordinate zhoutozhilizhou. In Shaanxi also, Yongzheng
raised nine zhoutozhilizhou and created two new prefectures.
As with the lower levels of the hierarchy, creating new prefectural level
units could have one of two motivations: security or revenue collection.
cannons on camelback 321

Gansu’s new units fell into both categories. Even though Gansu as a whole
was only a periphery of the northwest macroregion, the province itself was
stratified into zones of different marketing density.The three eastern zhilizhou—Jingzhou, Qinzhou, and Jiezhou—were
clearly located in more productive regions. Jiezhou, in fact, the most pro-
ductive prefecture with the best rainfall and the farthest south, belonged
not to the northwest macroregion at all but to the upper Yangzi. Creating
these three units increased the span of control of the provincial governor,
thus increasing his access to revenue from these productive regions. They
controlled an average of 4.3 county-level units. The five western prefectures, however, were formed out of military units
for strategic purposes. They controlled an average of 3.6 county-level units,
and they contained three of the province’s five ting.Ningxia prefecture, in-
cluding four county-level units, fell in between, combining both functions.
The branching of the Yellow River here created considerable irrigation wa-
ter for the fields, but the prefecture ran directly along the Great Wall, bor-
dering Mongolia. It constantly provided grain supplies for other counties,
but it also supported substantial military forces. In the early to mid-eighteenth century, then, even before the final elimina-
322 economic basis
Table 8.4 Gansu field administration (after 1777)
Prefectural unit Creation Location No. of
zhou/ting No. of
Lanzhou Fu 1667East2/14
Pingliang Fu 1667East23
Qingyang Fu 1667East14
Gongchang Fu 1667East1/17
Jingzhou Zhilizhou 1777, from PingliangEast
Qinzhou Zhilizhou 1729, from GongchangEast
Jiezhou Zhilizhou 1729, from GongchangEast
Ningxia Fu 1724East1/04
Xining Fu 1724West 0/13
Liangzhou Fu 1724West 0/15
Ganzhou Fu 1724West 0/12
Suzhou Zhilizhou 1729West 1/01
Anxi Zhilizhou 1773West 1/02
Total 12/545
Source:Niu Pinghan, Qingdai Zhengqu Yange Zongbiao (Beijing: Zhongquo Ditu
Chubanshe, 1990), pp. 462–477.

tion of the Zunghar state, the Qing administrative structure, like a boa con-
strictor, had begun to digest its new territorial acquisitions. Most of the
new territory remained outside the regular civil administration, but parts
had begun to move from military districts into county-level units, and new
prefectures balanced the revenue and security demands of the new north-
west province. Many other administrative and social changes ensued on the
local level, such as the building of walls, the growth of cities, and the prolif-
eration of official and sub-official posts. The experience with Gansu in the
early period of Qing expansion in the northwest set the stage for the huge
task of swallowing Xinjiang at the end of the century.
cannons on camelback 323

Land Settlement andMilitary Colonies
N ian Gengyao’s proposal to introduce military and civilian settlers
into Kokonor carried on a long-standing imperial tradition. In principle,
military colonies (tuntian)combined several advantages. Economically self-
sufficient military units spared costs when soldiers practiced cultivation
and defense in rotation. Once they brought their families to join them, the
soldier-settlers formed the nucleus of permanent settlements. Merchants
followed, linking the garrisons with interior trading networks. Then peas-
ants arrived, relieving population pressure in poor interior provinces, di-
minishing the prospects of famine or revolt, and mixing non-Han people
with more loyal settlers from the interior. In Chinese terms, military colo-
nies were a policy of yiju liangde(killing two birds with one stone).
In fact, permanent military colonies faced great difficulties. The Han dy-
nasty, in the second century bce, had set up military colonies beyond the
Great Wall, as did the Tang and others, but these colonies were expensive
to maintain. During the great Debates on Salt and Iron, in 81 bce, literati
attacked the policy of establishing salt and iron monopolies in order to pay
the expenses of military colonies on the Han northwestern frontier.
The abandoned garrison towns of Jiaohe and Gaochang in the Turfan
oasis testify to the haunting presence of these ephemeral imperial efforts at
2Jiaohe reached a peak population of five thousand under
the Tang dynasty. Its ruins now stretch for 1,700 meters north-south and
300 meters east-west on top of high cliffs west of Turfan. Gaochang, even
larger, also began as a Han garrison town and expanded by the seventh cen-
tury to become a large Buddhist community and the center of the Uighur

kingdom of KaraKhoja in the ninth century. The Mongol invasion de-
stroyed both cities in the fourteenth century. Rediscovered by German
archaeologists Albert Grünwedel and Albert von Le Coq in the early twen-
tieth century, Gaochang yielded large numbers of manuscripts, mosaics,
frescoes, and statuary to these modern-day adventurers and plunderers.
The fate of the famous cave paintings of Bezeklik nearby, taken to Berlin by
von le Coq and partly destroyed by American bombs during World War II,
epitomizes the vulnerability of these lost Central Eurasian cities.Ming emperors and officials made valiant efforts to develop military
colonies on a large scale. They quickly failed.
3Zhu Yuanzhang, the first
Ming emperor, expanded them on an unprecedented scale, especially on the
northwest frontier and in Liaodong. By the end of his reign, 33,500 soldiers
had opened over 16,000 qing(400 square miles) of land in Gansu province.
Zhu boasted of his ability to support over 1 million men in an army with-
out using any civilian grain production at all. Two-thirds of the soldiers in
Shaanxi were ordered to cultivate fields, while the rest stood watch. But the
troops made poor peasants. By 1392 two-thirds of them had run away. The
low yields of frontier lands, hardships of agricultural labor, and embezzle-
ment of funds by officers made soldiering very unattractive, and rendered
it impossible to produce enough grain to feed the troops. In the fifteenth
and sixteenth centuries, reformers repeatedly proposed changes to restore
production and reduce corruption, to no avail. As Arthur Waldron notes,
“Military farming simply did not work.”
The failure of military colonies to support themselves required inno-
vative commercial policies. Merchant-supported colonies brought civilian
settlers and flows of silver to the frontier to pay the troops to buy grain. De-
spite their founder’s intentions, Ming officials created an expensive merce-
nary standing army, not a self-sufficient low-budget force of hardy yeomen.
They withdrew garrisons from advanced positions in the steppe, built the
Long Walls to shelter the colonists, and supported them with commercial
supply lines from the interior. This was just the opposite of the Han and
Tang system of far-flung isolated garrisons. Zhang Juzheng, the great re-
formist Prime Minister, tried to revive the tuntianin the mid-sixteenth cen-
tury. By 1582 he had increased their nominal area to 600,000 qing,or
15,000 square miles.
5Yet the system completely fell apart by the early sev-
enteenth century, after nomad raids drove off cultivators, much govern-
ment land and water conservancy was diverted to private uses, and unequal
tax and corvée burdens forced soldiers to flee. Private cultivators prospered
more than soldier-settlers, especially when the commanders themselves
leased out their garrison plots to land speculators. Ming defense strategy,
eroded from within by profit incentives and harassed from without by con-
settlement and military colonies 325

[To view this image, refer to
the print version of this title.]

The ancient abandoned garrison town of Jiaohe, near Turfan.
[To view this image, refer to
the print version of this title.]

stant raiding, never found a genuine solution in the precedents of the impe-
rial tradition.Qing efforts, however, were much more extensive and permanent. As
early as 1692 the Kangxi emperor had insisted that troops beyond the
Great Wall cultivate land to support themselves, so as not to burden the lo-
cal people.
6He first discussed military colonies in 1715, after the Qing had
subdued the Khalkha Mongols. With the approval of the Khalkha Khans,
Qing troops investigated the most suitable places in eastern Mongolia for
agricultural cultivation. These colonies, he declared, would “not only save
on transport costs, and allow one to exhaust the earth’s resources [jin dili],
but they will also make the enemy barbarians disperse to poor areas, while
our generals enjoy a peaceful sleep on their pillows and mats.” Once again
the settlements combined the goals of economic development and security.
They could even provide grain to Mongol allies in times of need. In places
like Ulan Gumu and Khobdo, small garrisons of a thousand troops would
be initially supplied with large grain shipments from the interior until they
could support themselves from their fields.
Criminals sentenced to military exile, of both high and low status, made
excellent subjects for involuntary migration to the frontier.
8Guizhou gov-
ernor Liu Yinqu, for example, who had opposed the emperor’s military ex-
peditions into Turkestan, was spared execution but sent to Mongolia as an
exile to cultivate the fields.
By the 1720s General Furdan could report steady progress in digging irri-
gation canals, clearing fields, and planting crops. The land was very fertile,
yielding up to twenty times the initial seed, but he urgently needed iron
plows to clear the land.
10 At the best locations, like Ulan Gumu, Chahan
Sor, and the area just south of Khobdo, where there were ample supplies of
water and wood, the Qing generals could pursue a large-scale program of
building fortresses across Mongolia. These fortresses held thousands of
troops, who cultivated the fields and stood guard.
As noted earlier, the Qing fortress-building project resembled in some re-
spects the Russian penetration of Siberia: by linking the garrisons together
with sentry posts, the military forces could dominate a large area with rela-
tively few men; and by cultivating the land, they could relieve the burden of
sending supplies from the interior. The Russians, however, faced easier eco-
nomic conditions and a less powerful native population. They profited
from exacting tribute from the native Siberian tribes in furs, which they
could sell in the markets in Beijing. The Qing presence in Mongolia gained
no economic profits; the Khalkhas were at least temporarily subordinate al-
lies, but a sense of insecurity always persisted, and the resources of the
328 economic basis

Mongolian grasslands and scattered forests and fields were much less abun-
dant than those of the Siberian forests.Under Yongzheng, the other principal region targeted for settlement be-
came the oases stretching along the Gansu corridor into eastern Turkestan,
from Jiayuguan as far as Hami (Qomul), Turfan, and Barköl. Here the pro-
motion of military colonies supported the larger shift in imperial security
policy toward defensive concentrations of troops at key points, fortress
building, consolidation of forces, and reduction of expenses. These were
highly vulnerable outposts constantly raided by the forces of Tsewang Rab-
dan, and they had insufficient supplies to support major troop concentra-
tion. At the main Qing military base at Barköl, 300 kilometers east of
Turfan, there were sufficient supplies and prospects for good yields from
newly cleared land, but troops could not necessarily reach Turfan in time
to hold off damaging attacks.
12 One proposed solution was to build up
Turfan’s agricultural resources through extensive land clearance. Within the Gansu corridor, at Suzhou, Shazhou, Guazhou, Dunhuang,
and Anxi, it was easier to find cultivable land; the main task was to pro-
mote settlement. Three to four thousand troops were sent to each of these
towns (five thousand in Anxi) from 1722 to 1724 to revive the abandoned
military settlements. Peasant settlers from the northwest provinces of
Gansu and Shaanxi were preferred, however, because they had experience
with dry field farming, and they came voluntarily.
For the most part, these early colonization efforts did not attract private
traders, so the state maintained strict control over the grain markets. But
there were already signs of commercial interest. General Jalangga feared
that at Anxi “wicked people from the interior would seek profit” by buying
up surplus grain supplies, so he ordered tuntiancultivators to sell only to
state officials. The emperor approved the policy, but urged officials to buy
only at market prices and to store surplus grain in granaries. Even this dis-
tant oasis had a grain market, which attracted traders from the interior
who responded to price signals.
Deportation from Turfan
The effort to create a defense line in east Turkestan based on military settle-
ments ran into great difficulties. The debate over Turfan, the largest oasis,
illustrates Turkestan’s severe constraints on military supply and defense.
As noted in Chapter 6, Tsewang Rabdan’s attacks on Hami and Turfan
in 1715 had initiated a new period of conflict. Qing conquest of the oases
settlement and military colonies 329

in 1720 represented the first expansion of Chinese imperial control into
Turkestan since the Tang dynasty. But holding these territories proved dif-
ficult. Yongzheng’s troop withdrawals left Turfan especially vulnerable to
raids should Tsewang Rabdan turn hostile. The emperor’s policy relied on
the assumption that the Zunghars would remain cooperative if they were
offered the opportunity to send trade and tribute missions through Khal-
kha lands. But Zunghar raids on Turfan continued, as it became clear that
Turfan could not produce enough grain to support both the local popula-
330 economic basis
An elderly Muslim gentleman from Turfan, in traditional garb.
[To view this image, refer to
the print version of this title.]

tion and a substantial garrison. In 1722 the troops needed to borrow sup-
plies from the natives when their own grain supplies ran out.Tsewang Rabdan, meanwhile, was attempting to carry off Turfanis to
Halashar, because these agriculturalists could be valuable producers of
grain for his state.
16 Many Turfanis instead fled to the east, appealing for
protection from the Qing. When Tsewang Rabdan retreated the next year,
after some discussion the Qing generals decided not to put a large garrison
in Turfan. Instead they would maintain the garrison at Barköl, six to seven
days’ march away. But supplies were tight at Barköl also, and these troops
could not arrive in time to protect the Turfanis from raids on their herds.
These considerations led to proposals to induce the Muslims of Turfan to
move closer to the Chinese borders. Four to five thousand people, or half
the population, were offered lands to settle around Anxi and Suzhou; Anxi
was still 225 kilometers beyond the end of the Great Wall but 665 kilo-
meters closer to the border. Kangxi had embraced them as “our people”
(womin) and vowed to protect them against Zunghar raids, but the new
emperor would not expend large military resources in the distant oasis.
fact, very few Turfanis accepted the offer. Only 650 people departed the
oasis for a new home, even though Qing troop withdrawals left them more
vulnerable. Those who did settle in Suzhou suffered from poor harvests and
mistreatment by local officials, which drove them into debt. In 1731, how-
ever, when Zunghar raids resumed, there was further unrest in Turfan. The
garrison now numbered three thousand, but it could not support itself, and
officials had to distribute relief to the local people as well.
Yue Zhongqi outlined an extremely ambitious sixteen-point proposal for
a massive military occupation of east Turkestan.
20First he urged an all-out
attack on Ürümchi with a large army, leaving only a small garrison in
Turfan. If he succeeded, he could move the defense frontier 175 kilometers
westward, relieve supply burdens in Turfan, and destroy the Zunghar army.
The emperor, however, was skeptical that he could succeed. In the end,
Yongzheng delegated the decision to Yue Zhongqi. Yue then asked for substantial increases in troop strength, expansion of
military colonies, and shipments from the interior. The thirty thousand
troops in Barköl prepared for the military expedition would move to Tur-
fan and be replaced by eighteen thousand more men from Ningxia and the
Ordos; twenty thousand men would attack Ürümchi from Turfan, and ten
thousand would advance from Barköl. In case of victory, eighteen thousand
would be stationed in Ürümchi to hold the town. Yue made detailed esti-
mates of the supply demands for such a large force. He expected that lands
cleared around Turfan city could support ten thousand troops, and smaller
towns in the region could support at least five thousand more. The total
settlement and military colonies 331

current harvest at Turfan, Barköl, and Tarnaqin combined was 50,000shi
of barley, to be ground into grain for making noodles, but barley was hard
for the troops to digest. Additional supplies of 30,000 shiof millet shipped
annually from Suzhou to be mixed with the barley would be necessary. The
expanded army also needed a total of 60,000 horses, including cavalry
mounts and pack animals. Forty thousand were available at Barköl, 8,000
could be purchased in Zhili, Henan, and Shanxi, and the allied Mongols
would be asked to provide the rest. The attacking army also needed 34,000
camels to carry over 60,000 shiof grain, and 200,000 sheep, while each
soldier himself carried two months’ rations on his back.
Yue’s careful estimates indicated the vast scale of preparations necessary
to launch a truly decisive campaign. He realized that a serious military ef-
fort would take at least three to four years, and would have to root out the
Mongols from their faraway nests, at the same time leaving enough troops
in the oases to defend against raids. The emperor regretfully refused Yue’s
requests. He understood how shameful it was for Yue’s garrisons to stay in
defensive positions holding off nomad raids, but now was not the right
time for a decisive campaign.
22Yue did launch a sally against Ürümchi, but
he could not hold the town. At this juncture, Imin Kwaja (Emin Khoja), the chieftain of Turfan, be-
sieged by the Zunghars and desperate for Qing support, began to organize
a mass emigration from Turfan. After repelling the Zunghar army, in 1733–
34 he led nearly ten thousand people on a long march inland across 700 ki-
lometers of desert, settling them in Xin Guazhou, just west of the Anxi gar-
rison. For his efforts he was granted the title of jasak fuguogong,and his
people were organized into a banner, with Imin as the banner chief, the first
Turki to receive this honor. For two decades the Turfanis lived there in
poverty, as in a refugee camp, while their homelands were devastated by
warfare. They were allowed to return in 1754, as the Qing prepared its
final blows against the fragmented Zunghar state. As they moved back to
Turfan, they left behind 20,000 mouof cultivated land and 4,800 small
houses which they had built. On their return, further conflict broke out, as
Mangalik Bek, another Qing banner jasak,who had remained behind and
supported the Zunghars until their defeat, resisted Qing imposition of Imin
Kwaja as an equal power in the oasis. When Mangalik Bek’s resistance was
suppressed, Imin Khwaja became the undisputed ruler of Turfan as a ban-
ner jasak prince. Turfan retained this peculiar structure, the only Turke-
stani oasis organized as a banner, because of its close involvement with the
Zunghar wars. The Turfani migration toward the border was “voluntary” in the sense
that Imin’s people chose to protect themselves from Zunghar attacks by
332 economic basis

abandoning their home. But an important factor in their decision was the
Qing refusal to guarantee the oasis against attack. The limitations on sup-
plies for large garrisons in the oases of Turkestan meant that armies could
not stay in one place very long; they had to attack vigorously or retreat. The
long-term solution was to build up the productive resources of Turkestan
so as to support both an expanded population and a military apparatus.
This development came after the mid-eighteenth century, when the Qing
began the aggressive promotion of the settlement of Xinjiang.
23Most stud-
ies of the Qing settlement of Xinjiang only begin with this later period, but
the earlier experience with Turfan is revealing. It shows, once again, how
precarious were the natural resources of the region, how many constraints
they placed on active expansion, and what difficulties these created for se-
curity and economic livelihood there. This semi-voluntary population movement was only one of many mi-
gration projects promoted by the Qing. Through coercive and material in-
centives, the Qing rulers shifted not just military forces but thousands of
agrarian settlers around their empire as they incorporated progressively
larger territories into the state. From its origins as a “booty state” in early-
seventeenth-century Manchuria, the Qing had used deportations and mass
kidnappings to build a human resource base. As Hong Taiji commented in
1643, “One is not happy enough when merely taking goods, one is only
satisfied when one takes people.”
24 Now the Qing induced migration to-
ward the empire’s borders in order to incorporate more peoples into the ex-
panding state. These early eighteenth-century negotiations set the stage for
the aggressive penetration of Turkestan thereafter.
Settlement of Xinjiang
After eliminating the Zunghar state, the Qing began to promote much
more actively the colonization of Turkestan.
25 This colonization program
has been studied extensively by Chinese scholars, and several shorter ac-
counts in English have appeared, especially the excellent study by James
Millward. Here I focus on how the Qing colonization program contributed
to the cultural diversity of the empire and changed the ecology of the
Modern studies of the settlement of Xinjiang carry on a tradition that ex-
tends from Wei Yuan’s Shengwujito other nineteenth-century works. By
stressing the contribution of Qing policies to the integration of the empire
and the positive effects on economic development of the region, they assert
that the formation of modern China’s identity as a “multinationality na-
settlement and military colonies 333

tion-state”(duo minzu guojia) grew smoothly out of these projects of “uni-
fication.” Scholars of the Nationalist period and in the People’s Republic
take very similar approaches. Chen Tsu-yuen opened his Ph.D. dissertation,
written in France in 1932, by noting that Sun Yatsen envisaged sending 10
million soldiers and civilian colonists to Xinjiang and Mongolia: “This is
why the land clearance of this province under the Qing is of great interest:
it can provide useful information to the Chinese government for future
27 He discusses the organization of local government, the
sources of labor for land settlement, the location and number of settle-
334 economic basis
Map of settlements in Xinjiang. These settlements lie just north of the Ili river. South
is at the top. The map features canals and, in red, dams built along the rivers. On the
upper right is the fortified town of Huiyuan, on the lower left is Bayandai. The rivers
flow at the base of the large mountain range on the bottom (north) of the map.
[To view this image, refer to
the print version of this title.]

ments, the development of communication and transport routes, taxation,
and the economic and political consequences of the incorporation of Xin-
jiang as a province. Many subsequent studies follow the same model.Chinese historians of this project view it much as Frederick Jackson
Turner viewed the settlement of the American West, as the “opening of the
frontiers” to settlement (bianjiang kaifa).The Han immigrants, and the
state subsidies they brought with them, raised the productivity of lands
which had been vacant or used only for low-yielding activities such as graz-
ing. Like the North American settlers of the New World, the modern histo-
rians and their Qing ancestors viewed the replacement of grazing with in-
tensive agriculture as an advance in social and economic development.
Fixed settlements, dense populations, and high agricultural yields attracted
merchants. Denser trade routes tied the newly colonized regions more
closely to the interior. From the Chinese nationalist perspective, coloni-
zation contributed to the integration of China’s minority peoples into the
nation. These historians describe much of the process of settlement accurately,
but the assumptions behind their accounts are problematic. They take for
granted that Xinjiang “naturally” belonged under the rule of the Chinese
state, and that its recovery from the “rebel” Zunghars fulfilled a preexisting
definition of national territory. For them, imperial expansion completed the
construction of modern Xinjiang as a “multi-nationality” autonomous re-
gion with clearly defined borders. They portray the Qing as a developmen-
tal state dedicated to improving the economic livelihood of all its subjects
and uniting them under the leadership of the Han nationality. Our post-nationalist, postcolonialist age brings new perspectives to this
fascinating subject, introducing themes of ethnic and political diversity,
ecological constraints, and the social tensions of colonialism. The Qing was
only one of several expanding, colonizing empires in eighteenth-century
29 Furthermore, because the Xinjiang frontier was only one of
many frontiers on which the Qing expanded, we need a holistic view of
frontier policy. The emperor, the Grand Council, and a specialized group of
officials who spent most of their careers in the peripheries defined its under-
lying objectives. The lives of each of these men, such as Agui, Nayancheng,
Chen Hongmou, Nian Gengyao, and Song Yun, would repay closer analy-
30 Since their common interests and talents set them off from those of-
ficials who served in the interior, they provided unity for imperial planning.
Although no single policy was applied all over the empire, the issues were
the same everywhere. This special Qing vision of a unified empire needs
careful examination. Unlike modern historians, the Qing ruling elite did not conceive of a uni-
settlement and military colonies 335

fied people(minzu)defined by nationalist ideology, nor did they claim that
Xinjiang had always belonged to China. They knew well that these con-
quests were unprecedented, and that colonization required new policies.
The real Qing motivations were simpler: security and self-sufficiency. The
main goal was to make the region pay its own way, including support for
permanent large military garrisons. Some scholars claim that yields from
the newly cleared lands were enough, but as James Millward has shown,
grain supplies alone came nowhere near meeting total military costs. Large
shipments of silver from the interior paid for salaries, equipment, clothing,
and construction costs. Despite the development of gold and jade mines,
Xinjiang never justified itself economically, and economic development was
never the primary goal. Because Xinjiang depended so heavily on substantial support from the
interior, its security costs were always controversial. Jiangnan literati who
criticized spending on the new territories always met with stern imperial
rebukes. More surprising than the emperor’s hostility is the fact that the
critics gained such an open and influential voice, indicating that they had
substantial support. The debate over the value of Xinjiang was in fact a dis-
cussion of the proper boundaries and identity of empire. For the Qing dis-
putants, the conquest was not an obvious outcome of a continuous expan-
sion to fixed boundaries. The boundaries had to be constructed in the
process of conquest and constantly justified. History provided one means of legitimizing the conquest. Then as today,
historians drew a picture of a continuous Chinese presence in the oases
from the Han and Tang garrison towns to the present. Modern maps, like
Tan Qixiang’s historical atlas, perpetuate the myth by drawing clear lines of
imperial territorial control. The historical atlas, another key component of
nation formation, supports the convenient fiction of the nation-state as a
geographically fixed entity with clear, continuous, long-lasting boundaries.
By painting one space in one color, these maps conceal the limited state
control over frontier territories, the varied types of administration, and the
shifts of empires and trade routes that caused cities to flourish and then col-
lapse into ruins.
31 Precarious, disconnected settlement was normal in this
vast region, as Sven Hedin and others discovered, and the Qing establish-
ment of a permanent military and civilian presence, which lasts to this day,
was in fact unprecedented, not a linear outgrowth of earlier expansions. Instead of placing Xinjiang’s colonization in a smooth sequence from the
Han to the modern nation, let us ask what allowed the Qing to break
through the limits that frustrated earlier imperial efforts. Military, institu-
tional, diplomatic, and cultural changes allowed this breakthrough, but the
eighteenth-century economy was the base of imperial control. For the interior, the terms of the early Qing fiscal settlement had made it
336 economic basis

impossible for the state to know clearly local agrarian conditions. In the
early seventeenth century, in order to win over the Han elite, the Manchu
conquerors had agreed to leave most of the Ming fiscal system in place,
abolishing its most burdensome surcharges, and they had given up their
initial project of conducting a new empire-wide land survey.
32 The state
collected a fixed sum of revenue in silver, based on land quotas set in the
late sixteenth century. Newly cleared land was supposed to be reported for
taxation, but elites in collusion with local officials successfully concealed
nearly all of it. China’s registered land area hardly increased at all from
1753 to 1910, even though cultivated land area grew by an estimated 33
33Despite repeated denunciations, top-level officials rarely learned
of concealed land clearance. Only in flagrant cases, where, for example, a
sandbar threatened to flood a major city, could Beijing take notice.
34 This
fiscal arrangement, so frustrating for economic historians, was convenient
for imperial rule. Guaranteed large, stable revenue, enough to support a
small salaried bureaucracy, a large standing army, and a luxurious imperial
household, the rulers could rely on the cooperation of the much larger local
elites to enforce local order. But the frontier was different. Here, on the one hand, there were no en-
trenched literati to block state consolidation. On the other hand, the estab-
lished elites had little sympathy for imperial administrators, and they were
often culturally alien. No longer held in check by embedded social forma-
tions, military and civilian administrators had the freedom to try out new
methods of control. Zungharia after the conquest was a tabula rasa for pio-
neering frontier officials.
35Its experience indicates the aspirations of the bu-
reaucracy once it was freed from social constraints. We can look at the col-
onization program in a new light, then, not just as economic support for
soldiers, or as the culmination of a nation-building project, but as the im-
plementation of colonialists’ dreams on an open frontier. In short, it paral-
leled the projects of empire builders around the globe. Scholars have begun
to examine the “tensions of empire” in Asia and Africa in comparative per-
spective, focusing on the British, French, and Indonesian experience. The
Qing empire should be added to their list.
As its name indicates, Xinjiang, or “New Frontier,” was the product of
an imperial vision. Xinjiang, however, was only one of many new regions
that came under Qing control in the eighteenth century. From 1683 to
1760, Xinjiang, Taiwan, the southwest provinces, Mongolia, Kokonor, and
Tibet all became permanent territorial acquisitions. This short, explosive
period of expansion generated new thinking about the character of the em-
pire. We cannot consider any one region in isolation from the others, since
issues of frontier rule were part of a single discourse. What ideals did the Qing rulers have for their new territories? Some
settlement and military colonies 337

scholars have described the Qing project as a “civilizing mission,” by anal-
ogy with French imperial aims.
37 In this view, imperial officials followed
traditional Confucian imperatives of transforming barbarian peoples into
civilized humans in order to create a uniform, orderly hierarchy. But Qing
goals were by no means so simple. Such “transformation” (hua)was only
one aim, and it was offset by equally strong imperatives to preserve primi-
tive peoples from corrupting contact with civilization.
38 Furthermore, the
ideals generated from the center, or from high-ranking officials, always ran
up against barriers to implementation from highly varied local situations. A
second major tension of the empire was the gap between these inconsistent
ideals and their implementation in local administrative practice. Difference was inherent in the imperial project, both because the vision
did not endorse uniformity and because localities resisted rationalization.
At the same time, the imperatives of bureaucratic efficiency pushed for
standardization of administrative practice. The special position of the
Manchus, as themselves minority rulers of a Han-dominated empire, made
them especially sensitive to the eradication of difference. A completely ra-
tionalized, civilized empire that treated all its subjects the same would have
removed all marks of Manchu identity. Scholars who endorse the “siniciza-
tion thesis” about the Qing make just this claim: that the Manchus became
completely assimilated to Han culture and rested their legitimacy solely on
claims to create order, following the Mandate of Heaven.
39But this thesis
ignores prominent aspects of Qing rule. It neglects the continual concern
of the Manchu elite to maintain its separateness from the Han mass, ex-
pressed in its marriage policies, separate residence, and religious rituals,
and especially in the banner institutions that were its basis of control.
addition, the sinicization thesis, like the civilizing mission thesis, ignores
how the Qing continually reinscribed difference alongside uniformity in its
subject populations. The tension between difference and uniformity applies
as much to the Han as to the non-Han populations, but it was on the fron-
tiers that it became most evident. Here cultural diversity was not just inher-
ited but constructed.
The Qing set up all of Xinjiang as a military camp in 1760 under the au-
thority of the commander (Ili Jiangjun) in Ningyuan (later moved to Hui-
yuan) and his deputies in Pizhan-Turfan, Kur Kara Usu (Qingsui cheng),
Tarbaghatai (Suijing cheng), Ush, and Kashgar.
41This fact alone set it apart
from the rest of the empire. As the one region which had been gained in its
entirety through a rapid military conquest, it stood out as the one where the
338 economic basis

army dominated. Mongolia had joined the empire in a gradual process of
negotiated surrender and military campaigns extending from the early rise
of the Manchu state until 1760. In Tibet and Kokonor, Qing armies had
made brief forays and left small garrisons behind but relied on local Ti-
betan and Mongolian elites. In Taiwan and the southwest, substantial Han
immigration in the Ming had preceded the arrival of Qing troops. In Xin-
jiang, except for Hami and the failed colony at Turfan, very few Han set-
tlers entered the region until after 1760. The garrisons of 10,000 to 23,000
men stationed around Xinjiang introduced considerable diversity them-
selves. They included Manchu, Chinese, and Mongolian bannermen, Cha-
hars, surviving Zunghars, Torghuts returned from the Volga, tribal Man-
chus such as the Xibo and Daghurs, and Green Standard Chinese troops
from Gansu and Shaanxi.Besides the military commander, there were multiple civilian administra-
tive structures, frequently reshuffled from 1759 to 1773. In 1759 Anxi pre-
fecture in Gansu governed the new zhilitingof Hami and Barköl, and in-
cluded Pizhan (Turfan) and Qitai in 1771–72. Ürümchi and Zhenxi became
zhiliting in 1759–60, followed by Ili in 1764. They were all formally under
the authority of the Gansu provincial treasurer until 1773, when Barköl
was elevated to become Zhenxi prefecture, governing Hami, Pizhan, and
Qitai, and Anxi was demoted to a zhilizhou.Dihuazhilizhou was created
to govern Ürümchi and the new counties of Changji, Fukang, and Suilai.
Until 1882 the regular civil administration included only one prefecture
and one zhilizhou in eastern Xinjiang and the zhilitingof Ili. Most of the
vast territory lay beyond it. Even in eastern Xinjiang there were multiple jurisdictions. The Muslims
of Hami and Turfan were organized, unusually, as banners, with their
chieftains designated as jasaks,who had autonomous control over their
people. Other Mongolian tribes in the region were also ruled by jasaks.In
addition, the military colonists followed their banner or Green Standard
commanders, while the civilian colonists (Han and Taranchi) and criminal
exiles supported the garrisons and were governed by their respective civil-
ian authorities. In Altishahr, the oasis towns ringing the Tarim basin, begsgoverned inde-
pendently, under supervision by military residents. Here the Qing kept their
forces separate from the native population in small cantonments and ro-
tated the forces frequently to prevent permanent settlement. They also kept
out civilian colonists and, as much as possible, Han merchants. Local ad-
ministration was run by Turkestanis, and justice followed Islamic law. Qing
influence, however, was still apparent. The begsthemselves were no longer
hereditary nobles; subject to the Qing rule of avoidance, they wore the
settlement and military colonies 339

queue and Han clothing. They could abuse their power much like district
magistrates through bribery, manipulation of prices, and ties of indebted-
ness, but alongside them the religious establishment flourished, and the
Qing did not intervene. It is hard to imagine a greater contrast with the situ-
ation in Kokonor, for example, where the lamaseries were deliberately de-
stroyed.In summary, there was a gradient of administrative structures across the
region from east to west, becoming progressively less similar to the interior
civil administration and more dominated by Central Eurasian customs as
the Han presence faded out. There were three distinct civil administrative
structures—junxian, jasaks, andbegs—each deriving from a different cul-
tural tradition (Chinese, Mongolian, and Turkic). On top of this civil struc-
ture was the military presence, which was concentrated in Ili in the north,
where nearly 100,000 personnel and their dependents resided, significant
in Hami–Turfan–Barköl, and minimal in the south, where there were no
permanent forces. The military structures were divided into banners and
Green Standards, and the banners themselves accommodated a dizzying va-
riety of ethnic groups. The differing structures of native, civilian, and mili-
tary administration balanced one another as the Qing rulers deployed their
limited forces across this huge space. This did not reflect a single civilizing
project or nationalist incorporation but rather multiple negotiations be-
tween bureaucratic, coercive, and local environments.
Tax obligations and property rights also reflected the diversity of the re-
gion. Han military colonists carefully surveyed their lands, estimated har-
vest yields, and handed over their crops to the state in return for stipends.
Bannermen, by contrast, acted like self-cultivating peasants. Their banner
units owned the land collectively in principle, but this changed to private
ownership in the nineteenth century. They supposedly supported them-
selves from their own production, although in fact they often contracted
out cultivation rights to immigrant Han peasants or merchants in exchange
for fixed rents. The exile colonies had no rights to their land; all their settle-
ments were attached to the military colonies they served. They received less
land than the soldiers, generally of poorer yield, and officials took the en-
tire crop. Han civilians (hutun)received an average of 30 mou(5 acres) per
household, and were organized under the same lijiatax collection system as
in the interior. They were supposed to repay the loans of seed, tools, and
animals which got them started, but even though they paid no taxes for the
first six years, the low prices of grain put them in arrears quickly. Even-
tually the state reduced or forgave much of their loan obligations. Usually
their tax rates were the same as in Gansu, where most of them originated.
The Muslim colonists, by contrast, reckoned their harvest not by yields per
340 economic basis

land area but by the ratio of crop to seed planted. This method of reckon-
ing harvests was characteristic of traditional Muslim agriculture, adapted
to a low-yield, extensive cultivation region.
Officials discussing these tax systems persistently paid attention to incen-
tives for improving agricultural output. Many local officials tried to devise
tax policies appropriate to each group that would ensure incentives for the
cultivator to increase production as well as increasing the surplus paid to
the state. Originally, on Muslim lands, the Qing state took 40 percent of the
crop in kind, considering these lands as state land. Imperial Commissioner
Agui argued for switching taxes to fixed rents in order to make collection
easier and improve incentives. On Han military colony land (bingtun),the
entire surplus went to the state, and settlers were bound to the land. Har-
vest yields were carefully measured, and rewards and salaries were given to
officers and soldiers for improving yields. Agui argued in 1766 for increas-
ing rewards for soldiers who could improve yields above the average. Bannermen, by contrast, owned their land collectively as part of their
garrison. But in the early 1800s the Qing implemented a land reform pro-
gram which granted banner soldiers hereditary land rights. As Song Yun
noted in his report of 1804, collective ownership of property led to “lazi-
ness,” so he recommended dividing up the land and expecting each man to
provide for himself. He believed that this would create incentives for each
cultivator to grow a surplus beyond his family needs, and that granaries to
feed the poor could be provided from this surplus. In fact, Manchu banner-
men, never interested in agricultural cultivation in the first place, soon
leased out their holdings to Chinese and Taranchi tenants. Private, com-
mercialized land relations developed, including loans at interest and mort-
gages. Han and Taranchi settlers also owed labor dues for construction work in
cities and on road building. The Turkestani settlers were state peasants, ow-
ing rents and labor dues to the state, with no freedom to leave the land. The
Han settlers, by contrast, ranged from criminal exiles, who had no free-
dom, to completely independent landowners, and even merchant landown-
ers. In Zungharia the Qing created no single unified system of land rights.
The diverse character of settlement was reflected in separate arrangements
for each group. In east Turkestan and southern Xinjiang a well-established land system
had already been in place long before the conquest. Local begswere large
hereditary landowners. Those who rebelled against the Qing lost their
lands, but the loyalists were given increased holdings, including attached
peasants, from confiscated property. Begswere incorporated into the ad-
ministrative system. Like other Qing officials, they collected “nourishing
settlement and military colonies 341

virtue” salary supplements, but in the form of land and people, not in cash.
Peasants here were in a nearly slave-like condition. The new rulers inserted
themselves at the top of an old hierarchy, confirmed most of the local rul-
ers, and left agrarian relations as they were.
Colonization and Land Clearance
We now turn to the material details of agrarian production. The Qing pro-
moted military colonization on a much more extended scale than their pre-
decessors. The decision to encourage colonization grew directly out of of-
ficials’ awareness of the difficulties of supplying armies in the field. The
colonization of Central Eurasia in fact began half a century earlier than the
final conquest of the region, in the early eighteenth century. The two pri-
mary impulses were to increase grain production in the oasis settlements so
as to provide reliable supplies for the troops, and to relieve population
pressure in the poorest regions of China’s northwest so as to give relief to
drought-stricken peasants and ward off social unrest. A third motive, often
not so openly expressed, was to ensure permanent imperial control of the
region by constructing a new settler society there, composed of a mixture of
peoples immigrating from the interior.Colonization began under purely military auspices, and settled soldiers
were the first new agriculturalists to clear land, but civilians soon followed
them. Over time, the civilian population expanded, and soldiers began to
lease their land illegally to civilian investors, creating a de facto trend to-
ward civilianization and privatization. Qing officials encouraged this move,
believing that giving peasants land rights would encourage more intensive
cultivation. But the ultimate result of the decline of military control was to
undermine imperial authority over the region. Social tensions created by
the settlement push in the eighteenth century had been held in check when
the military administration dominated, but in the nineteenth century, with
the decline of military forces, many of the new structures established in the
previous century broke apart, making Xinjiang a constant source of up-
heaval. The Qing also embraced many diverse types of frontier settlement, mili-
tary and civilian. It brought to Xinjiang different kinds of military units, for
different purposes, and later sent criminal exiles there under military su-
pervision. The civilian settlers, mostly Han peasants from the northwest,
worked under different forms of agrarian institutions, spanning the range
from near serfdom to complete independent ownership. Furthermore,
Turkic settlers moved from southern Xinjiang to the north brought with
342 economic basis

them their own distinctive structures and local leadership. Thus the Qing
constructed a new society in northern Xinjiang, much more diverse than
ever before. The evolving colonization project overlaid new peoples and
structures on earlier ones, creating a palimpsest composed of multiple lay-
ers of institutions and social groups.The high cost of grain transport during the Galdan campaigns stimulated
the Kangxi emperor’s interest in the history of military settlement. In 1700
he told his subordinates to look into the experience of the Han general
Zhao Chongguo, who had first implemented tuntianon the frontier.
1715 he began active efforts to establish colonies at Turfan, Hami, Anxi,
and Barköl in the east, and at Khobdo and other places near the Irtysh,
Orkhon, and Tula rivers in the north. The northern colonies aimed to pro-
tect the Qing’s Khalkha allies against retaliation by Tsewang Rabdan. In
the east, each of the attempts to use soldiers to clear fields lasted only a
short time. Barköl, the main concentration point for troops campaigning
against Tsewang Rabdan, established a small colony of five hundred men in
1716, but it was abandoned in 1726. After reestablishing it in 1729, Yue
Zhongqi then proposed to clear up to 100,000 mouusing five thousand
men, but the emperor rejected his proposal, and most of the troops were
withdrawn by 1734. Qing armies occupied Turfan in 1715 and sent five
thousand colonists there in 1722, but in 1725 all but one thousand men
were pulled back to Barköl. They returned briefly in 1729, but Qing forces
could not hold the oasis securely. Instead, as we have seen, the Turfanis as a
people moved into the interior. Hami and Anxi, which were closer to the
border, lasted somewhat longer, but by 1742 they too were abandoned be-
cause of poor harvests, even though Nian Gengyao had recommended in-
vesting in larger settlements. In this early period, soldiers cleared land only temporarily to support the
main garrisons while they prepared for a campaign. During the Zunghar
wars, the military’s primary aim was to use the maximum possible force in
battle. Diverting soldiers to agricultural work came second. Despite the
economic gains of production in the region, which would cut transport
costs substantially, both the Kangxi and Yongzheng emperors chose not to
make the necessary long-term investments. Immediate strategic consider-
ations outweighed the farsighted views of Yue Zhongqi and Nian Gengyao. Beginning in 1758, Qing officials launched a serious, continuous cam-
paign to promote extensive agricultural settlement. Each form of settlement
had particular fiscal and agrarian features. Five different types of settlers entered Xinjiang during and after the con-
quest. First on the scene were two kinds of military colonists, both Manchu
bannermen and Han Green Standard troops. The first major military colo-
settlement and military colonies 343

nies were established in Ürümchi; they expanded westward to reach a total
of 13,400 cultivator-soldiers.
45 Farther west, in Ili, rebellions forced the
Qing to give up on early efforts, but from 1761 to 1772 it established in-
creasing numbers of colonies. Again, military colonization is by no means
unique to China, or to the Qing dynasty. The notoriously brutal colonial
schemes of Tsar Alexander I and his general Alexis Arakcheev in early-
nineteenth-century Russia are the most conspicuous European example.
Still, the Qing colonies in Xinjiang were more successful than those of pre-
vious Chinese dynasties and of other imperial states: they lasted longer,
fewer soldiers deserted, and they grew into permanent settlements. Al-
though they were never self-sufficient, they did not bankrupt the treasury. Third came exiled criminals. After 1758 the Qing began sending convicts
regularly to the newly conquered region. Ordinary criminals became bond-
servants of Chinese garrisons. Cashiered officials were exempt from penal
labor and lived separately from commoners, but they often served in local
administrative posts during the period of their temporary exile. Joanna
Waley-Cohen has described the legal, political, and symbolic functions of
the exile system.
47Exile colonies had only slight direct impact on land set-
tlement because their numbers were small. Nevertheless, they did mix to-
gether people of very different social strata from all over China. Tax resist-
ers, secret society members, military deserters, corrupt administrators, and
talented officials who were victims of political intrigue all shared the rigors
of the frontier, at least temporarily. Among the literati it was a common
enough experience to forge bonds of loyalty that may have influenced lite-
rati activity after their return. In this way, exile created networks spanning
regional, cultural, and class boundaries. Fourth, Han civilian settlement began after 1761. The main focus of this
state-sponsored program was the poor peasantry of Gansu. This province
suffered frequent drought; it had the lowest agricultural yields in interior
China; its tax base was very low; and its peasants constantly faced threats
of starvation.
48The settlement program was designed to offer a safety valve
for this region by encouraging the poorest peasants to migrate to Ürümchi.
By 1781 nearly twenty thousand households had emigrated. They obtained
full official support for transport costs, animals, agricultural tools, seed,
and housing. Organized official settlement ceased after 1781, but other
Han peasants continued to head for the frontier on their own. Fifth were Muslim settlers from southern Xinjiang. Realizing that the
difficult agrarian conditions of the region required special expertise, Qing
administrators recruited Muslim Turkic-speaking dwellers of the oasis cit-
ies of southern Xinjiang (Altishahr) to move to the north to clear new
land. Their expertise with irrigation of arid lands was particularly valuable.
344 economic basis

These Taranchis, as they were called, set up separate Muslim colonies, espe-
cially in Ili. They brought with them not only their agricultural skills but
their local political system as well. Their hereditary Muslim leaders(begs)
were recognized by the Qing as official community representatives with bu-
reaucratic ranks. Although they came from southern Xinjiang, they too
were immigrants, just like the Manchu military and Han civilians. The only
“natives,” the Zunghars, had been nearly exterminated. The Muslims’ reli-
gion and language, local leaders, tax obligations, and property rights set
them off sharply from the Han and Manchu settlers. Such administrative and cultural diversity posed severe challenges to im-
perial governance. The Qing’s problems were those of empires everywhere:
how to persuade extremely variegated groups of people to remain under a
single authority. It was imperial expansion that had created the problem in
the first place. Military force and competent civil administrators could con-
tain tensions, but they also needed other techniques. Clearance by military colonies (bingtun)began in 1757 with the restora-
tion of lands in Gansu, Hami, and Barköl, followed by new clearances
in Ürümchi and the Ili valley through the 1770s.
49The soldiers came from
the Green Standard forces of Han troops, mostly from garrisons in the
northwest. The tunwas the basic settlement unit, marked by a walled-earth fort,
holding from 15 to 250 men. The fort included room for soldiers and their
families, agricultural tools, granaries, and local offices. It was not merely
a defense post but a complete administrative and production “unit,” like
the modern danwei.Everytuntian area was governed by a major official
(tuntian dachen) who reported to the military governor in Ürümchi. On av-
erage, each soldier received 20 mouof land, plus animals, tools, and seed.
He did not have ownership rights in the land, nor could he choose his
crops. The state granted the colonists usage rights only, and prescribed
which crops would be grown. The predominant crop was wheat, with
smaller amounts of barley, millet, and sesame. The surplus was distributed
in two ways. In Anxi only, the crop was split evenly between the colonists
and the state. In the other colonies, cultivators paid a fixed quota to the
state and kept the rest for their families. The latter system generally pro-
vided greater incentives to increase yields. In Ürümchi, the standard quota
was 12 shi,but rewards were given to cultivators and officials if they gave
15 shi or more to the state. In 1784 Fukangan rejected a proposal to raise
the required quota in Ürümchi. He argued that the labor supply was lim-
ited, and requiring more grain would divert men from their military duties;
it would also cause them to expand military landholdings, interfering with
private production. He thought that the existing land provided sufficient
settlement and military colonies 345

grain for the garrison. The troops needed 30,000shi;the previous year’s
harvest had been 90,000 shi,and there were 800,000 shistored in the
granaries. At this point it appeared that the colonies were working success-
The cities of Ürümchi and Ili also had special granaries designated for
military use. As the Sanzhou Jiluenoted, “The interior granaries are for re-
lieving the people, but Xinjiang’s granaries are mainly for military sup-
51The only nonmilitary use was the allocation of seed to new tuntian
settlers. Surplus grain could be sold on the market to raise revenue; but un-
like the civilian granaries of the interior, these granaries were not intended
to level prices or provide famine relief.
52 We have abundant information
about the functioning of the civilian granary system, but the scale and uses
of the military granaries still remain obscure. Study of Ürümchi’s military
granaries would help to illuminate this parallel system. During the nineteenth century, the effectiveness of the military colonies
and their granaries declined. Troops were withdrawn from the region in the
1820s to meet pressing military needs in the interior, and much of the mili-
tary land was converted to civilian use.
53 Although granary stores were
plentiful in the 1780s, much of the grain rotted in the succeeding decades.
As the scale of cultivation declined, troops depended more on market sales
of grain, but the influx of civilian settlers had driven down market prices.
By the early nineteenth century, 200,000 civilians in Ili had cleared 1.8 mil-
lion mou of land.
54Wu Dashan had calculated in 1766 that a garrison of six
hundred men, without their families, rotated every five years for other
troops, could support themselves and store a surplus, but his estimate relied
on selling surplus grain at the high price of 1.6 taels per shi.This was the
market price in 1762, but by 1770 it had dropped to 0.5–0.7 tael per shiof
wheat. At these prices, individual troops could not support themselves. The
shift to family farming by civilians, or resettled soldiers, was driven by
awareness of this deficit. The “decline” of the military colonies was, in an-
other sense, testimony to the success of the Qing in promoting economic
development of the region as it moved from a military garrison zone to-
ward a more civilianized economy dominated by private farmers. Other soldiers came to Xinjiang for different reasons. Banner colonies,
or qitun, were established to restore the morale and military strength of the
banner troops. Composed predominantly of Manchus and Mongols, they
had been the core of the conquest army.
55After the conquest, banner forces
in the northwest neglected military drill and seemed to become more “cor-
rupted” by the civilian life around them. They also became more impover-
ished, as their stipends could not support them, and they were forbidden to
346 economic basis

engage in commerce or agriculture. This general phenomenon had con-
cerned the court throughout the eighteenth century, but now it extended
even to those closest to the northwestern frontier. Xinjiang was seen as a
relatively unspoiled region that could restore their morale. Some 11,500
troops were sent there between 1764 and 1774. At first they did not till the
soil; they were supported with stipends in kind. The revenue came from
state-run money-changing shops established in Ürümchi. The Qing officials
thus used the resources of the flourishing commercial economy to shore up
the archaic virtues of the pure, if impoverished, bannermen. By the Jiaqing
reign, however, rising costs of living made these revenues insufficient, and
the bannermen were ordered to support themselves on the land. They had
cleared up to 120,000mouby the end of the century in Huiyuan and
Huining. Turkic cultivators provided them with seed, and the state invested
heavily in digging irrigation canals to make these lands productive. Unlike the bingtun,the bannermen themselves never cultivated the land.
This was to be done by “surplus men” (yuding),the sons of banner families
who had not inherited their father’s post. Rights to the produce also dif-
fered. At first, the lands were worked collectively. All the grain was to be
given to the state and then shared out among the banner commands. On
some of the lands near the main garrison, cultivators worked vigorously,
seeing themselves as sharing in the garrison’s welfare; but on other lands
farther away, they neglected the fields, seeing them merely as state land. In
1804 Governor Song Yun turned collective holdings of 80,000 mouin
Huiyuan and 40,000 mouin Huining into private property, eliminating the
soldier-cultivators’ stipends while establishing special state granaries to buy
up their surplus production. He argued that making landholding a private
property right would increase incentives to work: “At first, the land was
state property, cultivated collectively so as to instruct the men in culti-
vation. But this led to laziness, so the land was divided. Now each is not
paid allowances from the treasury but provides for himself. Formerly they
feared that if they worked hard on the land, their salaries would be cut;
now that this [new system] has been explained, they work hard.”
With this anti-collectivist agrarian reform, reminiscent of China after
1979, frontier officials shifted away from the original goal of maintaining a
collective banner identity toward ensuring that each soldier supported him-
self as a private landowner. In fact, the banner cultivators themselves began
calling in tenants, illegally, to till the soil for them, and by 1830 prohibi-
tions on renting out land were ended.
57By the mid-nineteenth century the
Qing had in effect created a new landed elite in the region, a Manchu and
Mongol stratum relying on tenant labor by immigrant Han peasants.
settlement and military colonies 347

Another distinct group of soldier-cultivators entered the region: elements
of the Suolun, Chahar, Xibo, and Oirat garrisons from Zhangjiakou and
Manchuria. Like the Manchu and Mongol banner colonists, they received
stipends, though at a somewhat lower level, and some of them progres-
sively established themselves as independent landowners. The Xibo gar-
rison members were the most successful. They progressively gained eco-
nomic power, expanded their landholdings, and maintained their distinct
prosperous colonies through the end of the Qing dynasty. They remain the
only native Manchu-speaking minority in China today.
Exiled criminals formed another class of involuntary settlers. Earlier dy-
nasties again provided precedents, but the Qing had much larger territories
available. Exiles came to Xinjiang from all over the empire, and from many
social classes. Wang Xilong gives sixty-six examples of criminals who faced
exile, including counterfeiters of copper coins, armed robbers, parents who
killed their children for having illicit sexual relations, drug peddlers, feud-
ing lineages, kidnappers, arsonists, those who brought false charges at
court, and deserters from military service, to name only a few.
59Many con-
victed of crimes deserving capital punishment had their sentences com-
muted to exile at the autumn assizes. Impeached and convicted officials
were another distinct group. Exiles generally first served as bondservants to
the garrisons, not as agricultural laborers, but beginning in 1716 they could
till the land if they chose. Labor scarcity provided strong motivations to
use the exiles productively. By one estimate, about 160,000 exiles went to
Xinjiang from 1758 to 1911.
60Like the soldiers and civilian colonists, they
received agricultural implements, seed, and up to 30 mouof land to estab-
lish themselves, but since they had little chance to own their lands, their
yields were notably lower.
From the beginning the exiles fought back: they staged the first major
revolt in Xinjiang in 1768. A tuntianofficial held a celebratory banquet
for some of his exile friends; he provided large quantities of liquor and
gathered men and women together. When the drunken official forced the
women to sing, their outraged husbands went on a violent rampage, killing
the official, raiding the military warehouses for weapons, and occupying
the city. With a force of one thousand, they marched out of the city to con-
front a relief force of only 150 men sent from Ürümchi, but the exiles’ un-
trained horses fled when the two sides exchanged gunfire. All of the rebels
were massacred.
Exiled criminals sometimes bribed soldiers to let them escape. After the
riot had revealed lax administration, officials tightened up controls. Tat-
tooing on the face the name of the crime and the place of exile made detec-
348 economic basis

tion easy, and those who fled faced immediate execution. 63After five years
of field labor, however, followed by eight to ten years of mining labor, crim-
inals could join the civilian population. Few managed to achieve this ardu-
ous goal, but the state did provide funds for exiles’ families to join them so
as to encourage them to settle down. Some were able to join the army and,
if they achieved extraordinary merit during a campaign, gain their release
to return home. Nearly all the exiles, however, remained in the region for
their entire lives. A second kind of exile had not committed such egregious crimes but
ended up in Xinjiang for being implicated in local disturbances. Officials
deported one hundred members of a powerful lineage in Hubei that had
abused its local influence and a group of one thousand miners on the south-
west frontier who had attempted to flee to Vietnam.
64These settlers arrived
in groups, which were dispersed around the region; they worked under civil
instead of military jurisdiction and acted like civilian settlers. They were yet
another of the many disparate groups that cleared the land. Many disgraced officials were also exiled to Xinjiang, but they did not
cultivate the land. Over fifty officials implicated in a corruption scandal in
Gansu in 1781 went to Xinjiang, returning thirteen years later.
65 Hong
Liangji, the famous scholar who narrowly escaped decapitation after he di-
rectly criticized the emperor in 1799, was first sent to Ili but was pardoned
three months later. Failure to prevent deficits, the inability to manage for-
eigners, or the commission of crimes by subordinates could all get officials
in trouble. More than 10 percent of the Governors-General serving from
1758 to 1820 suffered banishment to Xinjiang.
66Civil and military officials
of all ranks from the lowest to the highest could end up in the region.
Xinjiang thus collected a microcosmic sample of Chinese society from the
interior, added to the native Turkestanis, military administrators, and no-
mads already there. Civilian settlers (mintun)became the majority of new migrants to Xin-
jiang. Before taking Xinjiang, officials had begun to encourage settlement
in western Gansu. Here, too, the Qing experimented with several different
types of land allocation before settling on one that provided the maximum
private incentive to increase production. In western Gansu, for a time, of-
ficials established a sharecropping system, which divided the crop 50–50 or
40–60 between the new settlers and the state.
67 In the Yongzheng reign,
the Gansu settlers supplied much grain to the armies still in the field in
Xinjiang. By 1736, however, because of disappointing harvests, the share-
cropping system ended, and the settlers’ holdings became private land on
which they paid fixed taxes. In western Gansu, now securely part of the in-
settlement and military colonies 349

terior, the new peasants assimilated to the resident population. State farms
worked with hired labor—a second experiment—cleared about 50,000
mouin Anxi, but were terminated after four years.
The most successful projects relied on a form of indentured servitude
much like the kind used to settle the New World. Settlers received travel
funds, clothes, food, seeds, tools, animals, start-up loans, and 30 mouof
land. For five years they had to work the land, but they paid no taxes. After
five years, once they began paying taxes, they became independent land-
owners. In Ili, after three years, they paid taxes of 0.1 tael per mouuntil
their debts were repaid, and 0.05 taels per mouthereafter.
68This system es-
pecially attracted the poor peasantry of the northwest, who lacked the cap-
ital to clear their own land. In 1726 Yue Zhongqi claimed to have settled
3,300 poor households in Shazhou, Gansu. From the 1760s, significant
numbers began to move into Xinjiang out of western Gansu.
69The first set-
tlers were single men or landless laborers, but soon most migrants brought
their families along, especially after officials broadcast reports to encourage
emigration. By 1803 over 155,000 civilians had cleared 1,014,879 mouin
Barköl and Ürümchi. By 1820, the total land cleared, including Ili, equaled
more than 1,080,000 mou.
This state-supported settlement program attracted many others from
the interior, who rushed to the new frontier eager to improve their lives.
ShaanGan Governor-General Wen Shou, touring the region in 1773, enthu-
siastically recounted the sight of shops clustered “as closely as the teeth of a
comb” along market streets, peddlers thronging the roads, and peasants
busily working on fertile lands. In his view, Xinjiang was a “paradise”
(letu) that offered great opportunities to all newcomers, and he urged the
state to promote further clearance.
71Hired laborers who came to the region
enjoyed high wages—up to 1 to 2 taels per month—and paid low food
prices, so they could save their money and before long buy land. Merchants who brought large sums of capital could invest in land imme-
diately and hire laborers to clear it, with official support. These “merchant
colonists” (shanghu)formed another important group of civilian migrants.
Some controlled very large estates, like the thirty-two merchants in Ili who
owned a total of 39,600 mou.In Ürümchi in 1778, 1,136 merchant house-
holds obtained land and seed loans on the same basis as the peasant set-
tlers, even though they did not work the land themselves. Lazar Duman ar-
gues that the presence of these merchant capitalists made Xinjiang a highly
stratified society, dominated by a landlord-official elite, but most of the
merchant land clearances were not as large as those in Ili; 50 mouper per-
son appears to have been an average amount.
72 Governor-General Wen
Shou did not fear the influx of merchants from the interior; instead, he of-
350 economic basis

fered them investment opportunities so that they would become a settled,
property-owning class.Except for the merchants, who paid in silver, nearly all settlers paid taxes
in kind. Wen Shou set a rate of 0.08 shiper mou, or 2.4 shifor an average
plot. Given five years to establish themselves, this was a reasonable amount
for most cultivators. Once they became taxpayers, they enjoyed the full
property rights of all the settled peasantry. Turkic agriculturalists were the last important group of contributors to
support the Qing conquest.
73 They were the only cultivators who had ex-
tensive experience with the special form of irrigated oasis agriculture prac-
ticed in Xinjiang. They were known as “Taranchi,” originally a Turkish
word, which the Qing took from the Mongolian tariyaci(or Oirat tarän)
for “farmer.”
74As noted earlier, the Zunghars, when they occupied the Ili
valley, brought many Taranchi up from the southern oases to promote their
land clearance program. The Taranchi made up a substantial proportion
of the twenty to thirty thousand people whom the Zunghars brought to
northern Xinjiang.
75When they captured Yarkand in 1680, they deported
much of the local population to the north as bondservants producing for
the Zunghar state. At the same time, other Taranchi helped the Qing. The begof Hami pro-
moted land clearance after Qing troops drove off Zunghar raiders in 1718.
Qing officials helped him to dig irrigation canals. In 1719 he sent 608 shito
the garrison in Barköl. In 1730 he reported harvests of 3,000 to 4,000 shi
from four to five hundred new settlers, and between 1730 and 1736 he sent
a total of 27,500 shito the Qing garrison. The Qing paid him 1 tael of silver
for each shiprovided to the military. A second settlement was begun in
1739 at Caibash Lake, but the Qing stopped making payments in 1742,
and the lands were converted into civilian colonies, paying 40 percent of
their product in kind to the state. By 1753, because of poor harvests and
the silting of irrigation canals, both colonies were abolished. In Turfan, Qing officials and the local begattempted to establish agricul-
tural colonies on the Hami model until the entire population moved to the
interior in 1731. Over eight thousand Turfanis arrived at Guazhou, Gansu,
in 1733. Qing officials built five forts for them and gave them 8,000 shiof
seed to plant 40,000 mou,as well as exemptions from tax and corvée. Poor
harvests left them unable to repay their loans, from which they were finally
exempted in 1738. In 1755 they returned to Turfan, and the Qing called in
other peasants to take over these cleared lands. Ili became the major site of Turkic agricultural settlement under Qing
sponsorship after the conquest. In Ili the Taranchi were the largest group of
cultivators. (See Table 9.1.) War had nearly depopulated the region, but
settlement and military colonies 351

General Jaohûi established the first Taranchi settlement of one thousand
households there to support his initial garrison of four to five thousand
men. Immigrants arrived from Aksu, Kashgar, Ush, Shayar, Yarkand,
Khotan, and Sailimu in the south, and from Turfan and Hami to the east.
By 1768, 6,383 Taranchi households provided a substantial amount of the
resident Qing garrison’s grain. Qing officials recognized the need to give the
cultivators incentives to exploit the fertile land exhaustively. Agui, for ex-
ample, proposed to change from share rents to fixed rents, stating:[Under share rents] during a good harvest, [the settlers] will not gain
benefits, and will neglect the land; we will need to supervise them, but
the people are too many, and we cannot supervise them all. We fear
that during planting time they will embezzle seed and store grain illic-
itly for themselves after the harvest. If we set a fixed amount to be sup-
plied each year, and let them keep the surplus, not only will there be no
disturbances, but they will know that diligent effort will improve their
Because fixed rent collection risked creating deficits in poor harvest
years, the hakim beg established charitable granaries to loan grain to culti-
vators in times of need. A loan taken in the early spring, before the crops
had begun to sprout, could be repaid from the fall harvest with an interest
payment of 10 percent. Officials also supported experimental plots where they tried to improve
yields with different planting methods. In Ili, General Ming Rui brought in
special high-quality seed from Pizhan for testing. He found that more ex-
tensive sowing gave higher yields: 0.025 shiper mou sown on 43 mou
yielded a harvest of 55.3 shi,but 0.015 shiper mou sown on 66 mougave a
harvest of 101 shi.Recognizing that agrarian conditions in Xinjiang were
quite different from those in the interior, he recommended that these plant-
ing methods be extended to other fields.
352 economic basis
Table 9.1 Types of cultivators in Ili
Taranchi Military Civilian Banner Criminal Total
Population (indiv.) 30,415 2,500 1,085 5,073 117 39,190
Land(mou) 180,000 a 50,000 66,211 40,584 1,611 338,406
Source:Wang Xilong, Qingdai Xibei Tuntian Yanjiu (Lanzhou: Lanzhou Daxue Chubanshe, 1990), p. 214;
Wu Yuanfeng, “Qing Qianlong nianjian Yili tuntian shulue,” Minzu Yanjiu(1987): 96.
Note: Figures for Civilian and Criminal may be for households, not individuals.
a. Wu gives 90,000; this figure is taken from Wang.

Ili’s production was more than enough to meet demand. Each Ili cultiva-
tor received 30 mouof land and 1.5 shiof seed, in millet and wheat, and
was expected to deliver 16 shito the garrison.
78In a good year, 1.5 shiof
seed yielded 40 shiof grain, so the settlers were giving 43 percent of their
crop to the military—slightly more than the 40 percent paid in Hami.
Twenty thousand troops and officials in Huiyuan and Huining needed a to-
tal of 166,600 shi,of which the Taranchi provided 103,000 shi,or 62 per-
cent of the total grain supply. The Chahar, Eleut, Suolun, and Xibo garri-
sons supplied their own grain, and the Green Standard troops supplied half
of their grain, but the Manchu troops relied entirely on the huitunsettle-
ments. By 1782 the granaries held 500,000 shi,but their holdings declined
to 282,000 in the Jiaqing reign (1796–1820). Although the Taranchi were not permitted to leave, and some were pun-
ished for fleeing, many other tenants from the south flocked to Ili to till its
fertile lands. They preferred the Qing fixed rent system to the repeated ex-
actions of smaller but more arbitrary amounts by the begsruling in the
south. The Ush rebellion of 1765, discussed earlier, resulted from abuses by
the local beg,who could not be restrained by Qing officials.
In Ili, Qing officials reinforced the local social hierarchy to ensure sta-
bility and production, leaving in place the existing system of rule by lo-
cal begs. Unlike in Turfan, the begsin Ili were not guaranteed hereditary
posts; each succession had to be approved by the ranking Qing official. The
hakim beg at the top of the hierarchy supervised thirteen other ranks of
subordinates, a total of eighty-seven officials. Each beg,like the civil of-
ficials of the interior, received a “nourishing virtue stipend,” beginning with
500 taels for the hakim begand proportionately less for lower ranks. The
hakim beg and his subordinates also received large land grants: 200
patman of land with one hundred men to till the land.
Indirectly, Qing policies to fill up Xinjiang also developed Central Eur-
asia beyond the empire’s borders. By the mid-nineteenth century, control
over the settlements had declined. The hakim begsand their subordinates
demanded more rents from their people; they extended their lands to en-
croach on state and Han settlers’ land; and they neglected irrigation works.
Many settlers fled. The uprising in Ili during the Taiping rebellion caused
further instability. Russians occupied the Ili valley in 1871, remained for a
decade, and took 100,000 Taranchis and Chinese Muslims with them when
they withdrew. Chinese historians claim that the Russians forcibly de-
ported them, but in fact they were fleeing the savage repression inflicted by
Zuo Zongtang’s armies. Under the name of “Dongans,” 70,000 of these
Chinese-speaking Muslims now contribute to the ethnic mix in Kazakhstan
and Kyrgyzstan.
settlement and military colonies 353

Economic Development
Transcending the narrow goals of military support, the conquerors of
Xinjiang soon promoted a full-scale program of economic development.
Alongside the military colonies, civilians expanded their settlements with
state support. The increased agricultural population in turn attracted mer-
chants, who stimulated the growth of towns and commercial links with the
interior. Other state programs beyond land clearance supported these de-
velopments. Horse markets, mines, and towns completed the picture.Armies may march on their bellies, but their supplies move on horse-
back. During military campaigns, all generals had to concern themselves
first and foremost with the supply of horses. Because of limits on pasture-
land within China proper, every dynasty needed to trade with nomads for
horses. When it came to judging horseflesh, nomads always had great ad-
vantages over the typical Chinese civil official. Dynasties with stronger
Central Eurasian connections, like the Tang, could get better deals, and
larger supplies of horses, than those, like the Ming, who had sealed them-
selves off from the steppe. But horse supplies were always crucial to the sur-
vival of an empire and accounted for a very significant part of the budget.
The Ming rulers established the most systematic method of trading tea and
cloth for horses at fixed frontier markets. But their defensive policy was
very expensive and left them highly dependent on the Mongols. The Manchus had established an Imperial Stud (Taipusi) to supply ani-
mals for court use, and they had other pasturelands available in Manchu-
ria. But when the military campaigns of Kangxi penetrated Mongolia, the
cost of providing horses at these distances became prohibitive. Shaanxi and
Gansu could not supply enough to meet the demand, and sending horses
from Guihua was very expensive. Mongols and Tibetans in Kokonor pro-
vided mares for 8 taels each and stallions at 12 taels each, an extremely
high price.
81 As the foregoing narrative indicates, the high death rate of
horses on campaigns meant that replacements on a large scale were needed,
and the campaigns could last only a limited time before supplies ran out. In 1736 the first horse farms were established in Ganzhou, Liangzhou,
Xining, and Suzhou in Gansu, with a total of 1,200 stud horses. By the
early nineteenth century these had increased to 20,000. But the wars of the
1750s called for far larger supplies. A fifty-thousand-man army, at three
horses per soldier, required a minimum of 150,000 horses at the outset, and
more than 200,000 to provide for replacements over a four-year campaign.
Qing generals now had Mongolian allies from whom they could levy ani-
mals, but as we have seen, excessive exactions could drive the Mongols
354 economic basis

into revolt. Only in 1760 were large stud farms opened in Xinjiang at Ili,
Barköl, Tarbaghatai, and Ürümchi. Ili, the largest center, had to meet a
quota of 9,524 horses every three years. By 1826 it had 50,000 breeding
horses. Subsequently, Qing officials established additional farms for sheep,
cattle, and camels. By 1826 Ili had 10,000 cows, several thousand camels,
and about 42,000 sheep.
In addition to purchases from Mongols, Xinjiang officials could now
draw on a new source: the large pastures controlled by the Kazakhs, who
had become tributaries of the Qing. At Ürümchi, Chinese offered brocades,
cloth, tea, metal utensils, medicine, and porcelain—the typical trade goods
of the Silk Road—in exchange for cattle and horses. Ili soon surpassed
Ürümchi, and the Kazakh pastures drove prices down considerably, to 2.47
taels per stallion, mare, or gelding, and 1.5 taels per head of cattle. As one
official noted, the value of animals on the frontier was quite different from
in the interior: at these Kazakh prices, one cow in the interior was worth
four frontier horses, and one donkey was worth two horses.
83Kazakhs also
paid a tax of 1 percent of their herds when they crossed the border, and
Kazakh tribute missions presented special gift horses to the emperor. The stud farms were under tight military control, with a top official in
charge supervising soldiers, each of whom tended twenty-four or more ani-
mals. Careful inspections counted and culled the herds: the Ili farms had to
breed a new stock of horses equivalent to one-third of their total herd every
three years, and cattle 80 percent of their herd every four years. The heavy
demands on the farms were a result of the multiple needs for horses on the
frontier. The most intensive demand came from the military colonies and
guard posts, where 30 percent of the horses and 15 percent of the cattle
needed replacement every year. Routine military service was not as inten-
sive for livestock as field labor, as only 17 percent were worn out annually.
Horses were also needed in the mines, and they were given as relief supplies
to nomads struck by disaster. A major military campaign, especially the
campaign against Jahangir in 1826, could mean a sudden demand for fifty
thousand horses from the northwest pastures. For the most part, the stud
farms could meet routine demands, and sometimes there was a surplus, as
in 1782, when excess horses were sold at 3.3 taels each. Until the 1850s the stud farms seem to have solved one of the most per-
sistent supply problems that constrained all imperial armies. After 1850,
however, the stud farms shared in the general disorder of the region. Like
all the other new institutions there, they quickly degenerated when official
oversight slackened. Without the triennial inspections, herds rapidly de-
clined. By 1853, 40,000 of the 100,000 horses in Ili had died of disease.
settlement and military colonies 355

Pasturelands were turned into cultivated fields as soon as troops with-
drew. Efforts to restore the farms at the end of the century had only lim-
ited success; by the 1890s there were only 10,000 horses in Barköl and
Ili combined.
The intensive agriculture introduced by the Qing generated new demands
for agricultural implements, so officials invested in iron mines to produce
both tools and weaponry. Original stocks imported from the interior
quickly wore out. As with horses, it proved cheaper to produce replace-
ments locally. Beginning in 1773, Ili officials developed iron mines, while
other mines produced lead and copper for currency, bullets, and weapons.
The military ran them all.For the most part, Turkic cultivators used only wooden tools, but Han
settlers, who dug the soil deeper and cultivated it more intensively, needed
iron plows, hoes, sickles, and reapers. Their demands were constant, as 30
percent of the tools wore out each year. Green Standard soldiers did the
first hard digging. Metal artisans brought from the interior received a gen-
erous wage of 0.2 taelsper day plus 1 jinof food for their scarce skills.
After 1773, exiled criminals became the predominant labor force. New reg-
ulations provided that after five years criminals could become ordinary
civilians, and after eight years in the mines they could return home. The
minimum time period was later extended to ten to twelve years, indicating
that their labor was badly needed. Agricultural development in Xinjiang,
although it relied on independent farmers to till the fields, also required a
bonded labor force of convicts under military supervision who provided
the basic tools of production. (See Table 9.2.)
356 economic basis
Table 9.2 Iron production in Ürümchi
Year Production(jin) Surplus above quota
1765 61,440 8,660
1766 76,000 31,178
1767 61,440 3,016
1768 56,480 11,540
1769 56,800 11,200
1770 61,280 12,000
1771 56,800 22,500
Source: Ulumuqi Zhenglue, tiechang;cited in Wang Xilong,Qingdai Xibei Tuntian Yanjiu
(Lanzhou: Lanzhou Daxue Chubanshe, 1990), p. 251.

By incorporating and developing Xinjiang, the Qing empire brought on
itself opportunities and dangers. It secured the region from control by other
powers and from a disturbing autonomy, and it drew fixed borders with
Russia that, for a time, prevented incursions from the Tsarist empire. Colo-
nization and integration policies intensified the exploitation of natural re-
sources and encouraged substantial immigration from the interior. To some
extent, emigration from the northwest eased pressures on resources there.
By compensation, the officials had to invest heavily in irrigation works,
tools, seed, and animals in Xinjiang to keep its agricultural settlements via-
ble. They raised the productivity of the soil, expanded the land area at the
expense of pasture, and generated important commercial links with the in-
terior. Xinjiang become closely bound to the Han core in a way it had never
been before. Yet development generated tensions that could be contained only by
force. The first wave of settlers, the soldiers, cleared fields as part of their
mission but later became civilian landowners. They were the most reliable
population. The exiled criminals came unwillingly, and could easily riot or
flee when control became lax. Civilian settlers arrived in droves, provided
they received heavy subsidies from the state. As these groups mixed with
the local Turkic population, however, and as Chinese Muslims and other
traders joined the ethnic macédoine, tensions rose. The vast expanses and
variegated populations could not be encompassed by one administrative re-
gime; therefore the Qing employed many systems to embrace its multi-
tudes. Since the region never paid for itself, continual subsidies from the inte-
rior were required simply to hold the existing ecological and social balance
in place. During the nineteenth century, however, as resources were di-
verted toward the interior, Xinjiang began to unravel. The repeated rebel-
lions, culminating in wholesale occupation for twenty years by Yakub Beg
in the mid-nineteenth century, were an unintended consequence of the high
Qing’s developmental policy.
settlement and military colonies 357

Harvests and Relief
M anagingthe frontier economy required a vastly expanded infor-
mation-gathering apparatus. The Qianlong emperor implemented a system
of empire-wide reporting on prices, harvests, and rainfall which provided
voluminous data on agrarian conditions. The purpose of these reports was
to allow officials to intervene as needed with relief measures. Officials who
ran the “evernormal” granaries established in every county relied on mar-
ket reports for the timing of their interventions to stabilize prices. They
could sell when supplies were short and restock when they were abundant. The collection of standardized statistical data from localities is consid-
ered to be the hallmark of a modern state. The creation of the concept of
“society” as an entity has been ascribed to the normalization disciplines en-
acted by nineteenth-century European states, among which statistics was
1In this sense, Qing practices of the eighteenth century look pre-
cociously “modern.” The most detailed harvest reports in the entire empire came from Xin-
jiang. No other lands in the empire were watched so closely. Only where all
the cleared land was under military control could officials measure pre-
cisely the yields in relation to the seed. Regular reports on agricultural
yields indicated that garrison supplies were being carefully monitored so
that troops could be allocated accordingly. How successful was the imperial program to settle peasants, clear land,
and raise agrarian productivity in the most arid region of the empire? Were
the rural producers able to provide sufficient grain for both themselves and
the troops? How variable were yields and prices in relation to climate fluc-

tuations? What were the effects of grain markets and official granaries in
leveling prices and linking regions together? To what extent did agriculture
depend on extensive investment in production by the state? The Qing infor-
mation-collection apparatus on the frontier provided the data to judge the
sustainability of agriculture there.
Harvests and Yields
Collecting data ontuntianyields formed part of a larger systematic pro-
gram of information gathering on the rural economy carried out through-
out the empire in the eighteenth century. After the emperor ordered provin-
cial governors to build up storage in granaries to be used for leveling price
fluctuations, he also called for monthly reports on grain prices and regular
reports on climate, grain storage, and harvests. This comprehensive effort
at imperial mapping of the terrain made visible to state officials the local
details of agricultural production.
2Two essential elements of this program
were accurate measurement and standardization. For harvest reports, local
officials established a target figure that represented the ideal harvest under
perfect conditions. They measured actual harvest as a percentage of this fig-
ure, on a scale from 1 to 10 (or in some cases 1 to 100).
3The harvest figures
themselves represent not a uniform level of output per land area but a com-
parison to the best possible yield. They are a valuable source of data for
measuring the variability of harvests by region and over time, but they do
not in general indicate the absolute output. The Collected Statutes of the Qing Dynasty specified that harvests of 8
and above were “plentiful” (feng);6 and above were “average” (ping);and
5 and below were “deficient” (que).In practice, however, any harvest re-
port of 7 or less was accompanied by considerable evidence of drought,
poor yields, and threats of famine in some local areas. This was just as true
in the paddy fields of Guangdong as in the arid northwest. Officials avoided
reporting harvest ratings as low as 5 because that would have required
them to give relief to the entire population. Instead, they indicated the pres-
ence of poor harvests with reports of 6 or 7, which allowed them to distrib-
ute relief to selected districts and populations. In practice, then, a figure of
8 or more indicated an abundant harvest, 7 or less indicated a poor harvest,
and 6 indicated the presence of severe disaster conditions which qualified
the region for extensive relief operations. The emperor carefully scrutinized
low harvest reports. In 1772, for example, he insisted that the Gansu gover-
nor check the method by which he averaged local county reports to make
sure that the figure for the provincial harvest was genuinely 6.5 and not 7.
harvests and relief 359

Thetuntian lands, however, did report absolute yields per mou,or yields
in relation to the seed planted. Thus they show us not only how harvests
varied but also whether there were increases over time. Because military of-
ficials depended on these lands for their soldiers’ rations, they paid careful
attention to the yields delivered by the peasants under their control. On ci-
vilian lands beyond military control, civil officials, who collected only a
small fixed tax, were much less concerned with absolute amounts collected
than with whether the peasants dropped below a minimum level of subsis-
tence. Both of these harvest measures can be used to gauge the sustain-
ability of agriculture in the Qing northwest. We can examine these ques-
tions for selected years and regions, concentrating primarily on Gansu
province and eastern Xinjiang. In an earlier essay I discussed the important role of granaries run by civil
officials in ensuring constant food supplies in Gansu.
5Here I concentrate
on the equally important role of the military colonies in Gansu. The mili-
tary colonies represent the productive possibilities for agriculture under the
best possible conditions. On them, officials carefully supervised the cultiva-
tors and provided all the needed supplies of grain, tools, and irrigation.
Comparing the tuntianyields with the harvests on Gansu’s civilian lands
reveals the gap between the best practice attainable with state subsidies
and the normal situation faced by the peasant farmer. Since it was mainly
Gansu and other northwestern peasants who brought their cultivation
methods to Xinjiang, an examination of Gansu indicates the kinds of prac-
tices that developed in the more distant frontier. Gansu’s crops were predominantly millet and wheat, as they are today.
Some fields were planted with winter wheat, to be harvested in the summer,
and other fields with spring crops, to be harvested in the fall. The main
summer crops were wheat, beans, and barley; the main fall crops were
millet, buckwheat, and oats. Along the Yellow River, especially around
Ningxia, a relatively high percentage of the land could be cultivated, but
farther away from water sources, much of the land could be used only for
grazing. On the Gansu corridor extending westward to Xinjiang, in Gan-
zhou, Liangzhou, and Suzhou, intensive cultivation was possible only in
oasis towns. As one memorialist put it: “[These regions] are near the
Tianshan Mountains, and people rely on melted snow from the mountains
for water; areas without melted snow from the mountains are gobi[hard
Gansu’s climate was more similar to that of Xinjiang than to that of the
North China plain. Both regions experienced hot, dry summers, very cold,
dry winters, and low summer rainfall: the extremes of continental climates.
Traveling officials noted a definite shift in the climate as they proceeded
360 economic basis

west from Shaanxi to Gansu, and a further shift as they crossed the Yellow
River. The farther west they went, the colder and dryer the climate became,
and the later the crops sprouted in the fields. Annual precipitation of 25 to
50 centimeters was just barely adequate, but, as in North China, much of
it fell in the winter months. The absorbent loess soil, however, soaked up
the melting snow and retained it in the ground for crops sprouting in the
spring. Officials and farmers kept very close track of the vital snowfall in
the early spring and reported it annually to Beijing.
Gansu was the third-largest province in area in the empire, next to
Yunnan and Sichuan, but the most sparsely populated. (In the Qing, Gansu
included the modern Ningxia autonomous region and parts of Qinghai
province.) Because its cultivated land area was also so low, however, less
than 3 percent of the total, its ratio of cultivated land to population was
near average.
9In its cultivable regions Gansu supported a dense rural popu-
lation, nearly equal in density to that of much of the North China plain or
the uplands of Shaanxi. Its population totaled nearly 15 million in 1787. The tables in Appendix D list the harvest reports for selected years in the
eighteenth century. With an average of 7.6, most of the time Gansu’s har-
vests were sufficient to support its population, but harvests were highly
variable. During at least nine out of thirty harvest seasons for which we
have data, the figure for the average provincial harvest was below 7, indi-
cating that significant regions faced disaster. Even in the most abundant
years, certain counties always needed relief supplies. The correlation of harvests between successive years is low, an average of
0.31, where 1.00 indicates perfect correlation. Even between summer and
fall harvests of the same year the correlation is only about 0.60. The high
variability of harvests made it difficult for officials to plan for relief. They
could not even expect the same districts to be hit by disaster every year. Di-
saster could come in the form of drought, sudden floods, or hailstorms, and
could strike almost anywhere. Officials had to be prepared to shift grain
supplies quickly to meet harvest shortfalls. As noted earlier, on military farms in the northwest, the state tracked
agrarian yields very closely. Each year the officials loaned the cultivators
sufficient grain from the granaries for their subsistence and for planting
their fields. The cultivators themselves contributed small amounts of seed
capital (ziben) in the spring. After the fall harvest the officials first deducted
the original seed loans, then took half (or, in Anxi, 40 percent) of the har-
vest to store in the granaries to support the troops. Nearly every year the
government recovered its loans. From the available archival reports, we can
calculate the total yield on the three tuntianlands in Liangzhou, Anxi, and
Suzhou in Gansu, Muslim colonies in Hami and Anxi, and both military
harvests and relief 361

and Han civilian farms in Ürümchi. These reports provide some of the most
accurate and systematic data on farm yields anywhere in the empire. (See
Appendix D.)Average yields were quite high and stable. The average yield per mouin
Gansu was 2.35 shiper mou, compared to 1.5 to 3.0 shiper mou on rice
paddies in the lower Yangzi and 2.0 to 3.0 shiper mou on wheat fields in
Jiangnan in especially good years.
10Wheat yields in the lower Yangzi gener-
ally averaged only 1.0 to 1.5 shiper mou; Philip Huang estimates wheat
and millet yields in several North China villages in the 1930s ranging from
0.5 to 1.6 shiper mou, and the PRC government set a target of 2.5 shiper
mou (400 jin,or 200 kg) for lands north of the Yellow River in 1957.
Since annual rainfall in most of the northwest was less than the average for
North China, this makes the Qing achievement even more extraordinary.
The correlation between seed yields on military farms and the provincial
harvest was also quite low, only 0.39, indicating that harvests on these
lands were relatively protected from the droughts that hit the rest of the
The Muslim cultivators in Guazhou and Anxi planted seed over a wider
area than the Gansu peasants. They had extraordinarily high seed yields,
averaging 6.12, and higher yields per acre than in Gansu. In Hami, also,
where yields were measured in proportion to seed planted, seed yields were
quite high, ranging from 9 to over 15. In Barköl, the central military base,
however, where the weather was quite cold and only barley could be
grown, yields were much lower, averaging only 0.84 shiper mou. All of its
millet was imported from Hami. Its granaries frequently suffered deficits,
which were also filled from Hami.
As these data indicate, Gansu as a whole suffered from frequent drought
and uncertain harvests, but the heavy investment in military colonies by the
Qing officials succeeded in maintaining very high, stable yields for the priv-
ileged settlers under their direct control. The Qing also, however, mobilized
its state grain holdings to serve the civilian population in times of famine
through an impressive system of granaries.
Granary Reserves
By the mid-eighteenth century, Gansu had an extensive network of “ever-
normal” granaries (changpingcang),which could sell grain on the market
to reduce price fluctuations. A collective study of the operations of this im-
perial civilian granary system found that Shaanxi and Gansu had the high-
est per capita reserves in the empire, along with Guangxi and Guizhou in
362 economic basis

the southwest. In both of these landlocked frontier regions, with a heavy
military presence, imperial policy required high levels of official storage be-
cause “they must keep military provisions in readiness.”
These reserves grew over the course of the eighteenth century from 1.0
million to a peak of 4.8 million shi.
14From 1741 to 1792, when the level of
reserves empire-wide increased by 54 percent, Gansu more than tripled its
holdings. Its strategic location, its relative lack of merchant wealth, and its
distance from networks of water transportation made Gansu a prime target
of the officially run granary relief system. To be sure that grain reserves were used effectively for relief, we need to
know not only the per capita level but also the rate of distribution. Unlike
in some provinces, officials in Gansu did not report the actual turnover
rates of the granaries. The normal expectation was that 30 percent of the
stores would be sold each year, so that nearly the entire stock would be
turned over in three years. Since there was usually some harvest shortfall
somewhere in the province in any given year, and since officials constantly
discussed where and how to distribute grain, it seems that most of the civil-
ian grain supplies were actively employed in relief distribution and price-
leveling sales. Gansu’s problem was not too little use of its reserves but too
much. The actual holdings in the granaries often fell far short of the official
targets because so much had been disbursed for relief supplies and loans to
farmers. In 1763, for example, its true reserves were only 1.832 million shi,
far below the target of 3.2 million.
15Gansu officials had great trouble in re-
stocking their granaries. The collective study focused almost exclusively on the civilian granary
system, but the huge military granaries operated alongside it. (See Table
10.1.) Although imperial administrators tried to keep military and civilian
granaries distinct, in practice they interacted with each other constantly.
During major campaigns, especially, the abundant memorials on military
grain supply reveal that all the granaries faced pressures to divert their
holdings from civilian relief to military rations. Earlier efforts at empire-wide civilian granary systems, like that of the
Tang dynasty, had failed because they could not meet military needs.
great rebellions of the late eighteenth century caused “decisive aggravation
of granary problems” because endemic structural weaknesses made it im-
possible for the granaries to meet the combined demands of civilians and
soldiers. By the mid-nineteenth century, rampant diversion of civilian sup-
plies completely undermined the system.
17Gansu was one of the first prov-
inces to face these pressures. By 1762, 250,000 shihad already been allot-
ted to the military, reducing total reserves to less than 900,000 shi.
In addition to direct diversion of supplies, market operations also linked
harvests and relief 363

the civilian and military granaries. Prices on local markets soared when
military purchases coincided with harvest failure and the local population’s
need for relief. By loaning grain to soldiers in advance of the harvest, of-
ficials could spread out the impact over the year, but they could not reduce
the aggregate demand.
The frontier provinces of Yunnan and Guizhou had the greatest military
presence and also very high per capita granary reserves. James Lee esti-
mates that military grain allocations amounted to 500,000 shiof unhusked
grain per year, roughly the same as that for the civilian population.
20 To -
gether, these could support 5 to 15 percent of the registered population for
a year. The southwestern provinces also were hit with heavy military de-
mands during the wars with Burma from 1765 to 1770, and later during
the invasion of Vietnam in 1788. Military demands during wartime rose to
as much as one-third to one-half of the civilian granary stocks.
The southwest had, however, very different problems with its grain
stocks from those affecting Gansu. During the rapid buildup of reserves in
the mid-eighteenth century, grain stores might grow faster than the needs of
the local population, and the older reserves would rot. In the arid north-
west, spoilage was generally not a problem: only about 1 percent of the
grain was lost annually. In the damp southwest, by contrast, much of the
grain was lost if it was not sold: Guizhou lost 70 percent of its holdings by
1776 because officials had been unable to turn them over quickly enough.
During the late eighteenth century, officials had less trouble keeping grain
from rotting since wars and harvest failures generated increased demand.
They seem to have been relatively successful at keeping prices stable.
Gansu, by contrast, seldom had problems with civilian oversupply. There
was no lack of peasants and soldiers needing relief. To be sure, in certain
districts, like those bordering Sichuan, moisture was a definite threat. Gov-
ernor Wu Dashan, when he arrived from Sichuan in 1755, noticed that mil-
364 economic basis
Table 10.1 Military granary reserves in Gansu
Place Year Amount (shi)Source Comment
Hexi 1754 1,653,200 GZDQL 8.836
Hexi 1756 780,000 GZDQL 14.235
Anxi 1753 80,000–100,000 GZDQL 7.188
Anxi 1754 68,000 GZDQL 9.842 Annual purchases
of 5,000–6,000shi
Suzhou 1756 170,000 GZDQL 13.447
Source:GZDQL–Guoli Gugong Bowuyuan, ed., Gongzhongdang Qianlongchao Zouzhe
(Materials from the Gongzhong Archives, Qianlong Reign).

itary grain stores had increased to 62,000shiin Jiezhou prefecture, much
more than the troops there needed, and he feared that the grain would soon
rot. Earlier requests to shift to collection in cash had been refused, because
the Ministry of Finance felt that it was vital to maintain large grain stores in
the remote peripheries of the empire. Governor Wu now thought that the
commercial economy had developed enough so that grain could be pur-
chased on the market, so he asked that levies from the peasants producing
for the military be made in cash instead of in kind. This was further evi-
dence of the penetration of the commercial economy into the province.
The Contribution Scandal
In a number of ways, the eighteenth-century experience of Gansu as a fron-
tier province forecast problems that would face the entire empire in the
nineteenth century. Deficits in its holdings, the constant pressure of mili-
tary support, and annual harvest failures challenged the capabilities of its
officials to the utmost. Again, foreshadowing later developments, they
quickly realized the need to turn to the private grain market, and especially
to promoting relief in cash. The shift to distribution in cash instead of grain
meant not a “failure” of the imperial effort to store grain for relief but a
change in the methods by which the government would relieve harvest fail-
ures. Instead of relying primarily on distributions in kind and sales to lower
prices, the officials worked through the market by giving money directly to
the stricken farmers. Gansu, like Shaanxi, even though it was on the pe-
riphery, pioneered these methods.
Yet Gansu had to support a very large military population until the end
of the Zunghar wars, and this made its situation quite distinct from that of
other provinces. The military colonies could support the established garri-
sons, but the troops marching through the province on the way to Xinjiang
also drove up grain prices and strained local supplies. After the conquest of
Xinjiang, Gansu still had to provide substantial support for the new garri-
sons stationed there, which numbered at least 125,000 people, including
dependents. Gansu’s tax accounts, as well as its granaries, were almost al-
ways in arrears because of these multiple demands. Periodic exemption
from annual tax collection was one of the most significant ways of provid-
ing relief for the local population. There were, then, a number of good reasons to shift from distribution in
kind to distribution of money: the official granaries would put less pressure
on local markets because they would not have to purchase so much to re-
stock; granaries could meet more realistic targets; less grain would spoil;
harvests and relief 365

and grain traders would be attracted to remote regions. Officials, however,
constantly debated the relative advantages of cash and grain relief, and
many were skeptical.
25 Cash was harder to track within the bureaucracy
because it could easily be shifted between accounts. This practice, called
nuoyi (illegal diversion), was forbidden but extremely common. Peculation,
or diversion of public stores to private supplies, was also easier with cash
than with grain. Giving money to needy people required them to go to the
markets themselves, and this was not easy for the elderly, children, or the
sick. When taxes were collected in money, as they had been since the Ming
Single Whip reforms, the government made great efforts to ensure that tax-
payers paid individually in person. It still could not stop the practice of
“proxy remittance” (baolan),by which small farmers entrusted payments
to landlords or brokers. The same thing in reverse could happen with relief
payments: “proxy recipients” could collect the payments and deliver only
small amounts to the truly needy. But the Qing was moving in the direction
of contracting out many of its basic functions to non-official groups. This
was inevitable so long as the formal bureaucracy remained small while the
population and its social activities increased. “Pettifoggers,” or legal spe-
cialists, took over plaintiffs’ cases before the magistrate for a fee; merchants
in large cities, like Hankow, assumed substantial local government func-
In the early eighteenth century, Gansu also suffered from “bumper crop
famines” (shuhuang), characteristic of a region that was inadequately mon-
etized, where grain yields fluctuated widely, and the population was too
poor to store its own grain.
27 When farmers had a good harvest, they
rushed to the market to sell all their grain, thus driving prices to unprofit-
ably low levels. If a poor harvest struck the next year, they could be wiped
out, since they had neither cash nor grain on hand. Officials had a strong
interest in making sure that grain producers had sufficient cash to pay their
taxes. Here the state’s most important role was to prop up prices during
the harvest season by purchasing grain with official funds and storing it
in government granaries. Later, as more merchants came to trade in the
northwest, the bumper crop famines would become less common. Injecting
money into the frontier economy both kept up prices in the short term and
fostered commercialization in the long run. The increased pressure to provide grain induced officials to devise new
ways to encourage building up granary stores. In an ingenious but ulti-
mately fatal innovation, they turned to the merchants. Shaanxi–Gansu
Governor-General Yong Chang and Gansu Governor Eleshun, reporting on
the military colonies in Anxi in 1754, noted that its local granaries, holding
147,000 shi,used to be sufficient to support the garrisons and the small ci-
366 economic basis

vilian population. 28 But now more people had immigrated from the inte-
rior, and merchants had arrived to meet their needs. Furthermore, 80 to 90
percent of the soldiers now had brought their families with them, and the
garrison had increased to ten thousand men. Finally, Hami, the oasis far-
ther west, depended on Anxi for grain shipments up to 100,000 shiin bad
years, an amount that exceeded the local supply and was expensive to
transport. The officials proposed to take advantage of the presence of
wealthy merchants by encouraging them to contribute grain in return for
examination degrees (juanjian).By registering in Anxi and arranging to de-
liver grain to the granaries, their sons would get jianshengdegrees. Each
contributor would have to provide 80 shiof unhusked or 40 shiof husked
rice, or the equivalent in wheat, plus “labor costs” of 4 taels and “granary
fees” of 3.2 taels per degree. Costs of transport to the garrisons ran 1.0 to
1.5 taels per shiin Suzhou, and higher in Anxi. This meant that the cost of a
jiansheng degree could be as high as 300 taels per person, although the av-
erage was 130 to 200 taels. This was much higher than the standard rate of
108 taels set in 1736.
29 When rising prices threatened to discourage mer-
chant contributions, the amounts required were lowered by up to 30 per-
cent, depending on the local market price. These proposals had strong precedents, since Ming officials had also sold
degrees as a result of a military crisis.
30 The court initiated the sale of
jiansheng studentships for grain or horses after the emperor was captured
by the Mongols in 1449. Between 1678 and 1682 the Manchus briefly sold
degrees on a large scale during the repression of the Three Feudatories re-
bellion. Northwestern governors had promoted contributions (juanna)many
times before for brief periods, in order to increase grain stores, in 1691,
1703, 1714, 1715, 1717, 1720, 1724, and 1734, but these contributions
were mostly collected from serving officials and those who already held de-
31In 1691, for example, as the first Galdan campaigns began, a con-
tribution of 1,000 shiwould buy a recommendation for promotion from
district magistrate to prefect.
32 The southwestern provinces also sporadi-
cally used contributions to raise grain supplies. Yunnan collected over
100,000 shiin this way by 1681 and 428,638 shiin 1732, until the practice
was stopped in 1768.
The intense competition for degrees in interior China was getting tighter
as the rising population competed for a fixed quota. Gansu itself produced
very few natives who could get examination degrees. Governor Chang Jun
revealed in 1763 that 70 to 80 percent of those taking exams in Gansu
came from the lower Yangzi and Zhejiang.
34Immigrants had flocked to the
province to take advantage of its more lenient quotas, but as the number of
harvests and relief 367

immigrants increased, purchasing degrees became a more attractive way
to use their wealth, and one appealing to local officials eager to fill the gra-
naries.The prospects of getting degrees at higher levels made the purchase of a
jiansheng even more attractive. The northwest had very favorable quotas at
the two highest levels, the jurenandjinshi degrees. Gansu had failed to pro-
duce a single degree holder between 1644 and 1702, but then a reform of
the provincial quota system favored it so much that there were 255 jinshi
degree holders from Gansu by the end of the century. The jianshengwas the
first step on the ladder of success, but it made higher rungs possible.
Governor-General Yong Chang set a target of 150,000 shiof unhusked
grain from these contributions, equivalent to 1,875 contributors. Expecting
higher grain stores to result from contributions, he asked that the provin-
cial quota be raised from 3.43 million to 3.6 million shi.Even at such high
prices, he expected merchants to transport grain to the frontier “enthusias-
tically”, and he was right.
36From 1741 to 1745, Governor Huang Tinggui
had reported collecting 1 million shiof contributed grain, or an average of
200,000 shiper year, so this figure may have seemed reasonable. By 1761
newly constructed granaries held 700,000 shiof contributed grain.
Shaanxi governor Lu Zhuo also had great problems with stocking his
granaries. He likewise proposed to allow merchants from outside the prov-
ince to contribute grain in exchange for degrees in seven counties which
bordered the Ordos desert, but his proposal had been rejected. When he
tried again in 1756, the Board of Revenue agreed to allow a one-year ex-
38The court, however, dismissed Lu Zhuo for inadequate supervi-
sion of military grain supplies and replaced him with Chen Hongmou, from
Gansu. Chen discovered that Lu indeed faced a problem, since the grana-
ries in these strategic border counties were nearly all empty. Merchants,
however, traveled beyond the Great Wall to purchase grain (the Ordos was
one of the few northwest regions where cultivable acreage lay outside the
wall). Since their grain purchases did not have a harmful effect on local
markets within the passes, Chen recommended that merchants be allowed
to contribute this grain in exchange for degrees to fill the border granaries.
One of the main considerations, then, in discussions of this question was to
ensure that merchants purchased grain beyond the local markets. If they
purchased only locally, they would simply drive up prices, bringing no
benefit to the people, but if they could be induced to transport grain from
outside, the local people would benefit from increased trade. The contribu-
tion program, then, not only provided social mobility for the merchant
classes but also fostered increased trading linkages. By 1758 all but two counties in western Gansu were collecting contribu-
368 economic basis

tions. Prospects for increasing grain stores looked quite favorable. Ten
counties had collected over 8,000shifrom 114 people in only three or four
months, and the governor expected “merchants and people near and far”
to “rush to contribute” to the granaries. During 1758, 463 people contrib-
uted 6,967 shi.
In 1766, however, all collection of contributions in Gansu and elsewhere
was stopped.
40The argument given was that too much was being collected
in money and not enough in grain. The emperor feared that large purchases
on the market would drive up prices, or that coercive purchases (lepai)
from the people would deprive them of their own reserves. Instead, he al-
lotted 3 million taels to the province, to be used to accumulate grain re-
serves slowly, purchasing only when market prices were low. In this way of-
ficials could prevent the impact of bumper crop famines but at the same
time inject money into the economy when needed. But in 1774 the emperor
once again allowed Gansu and Shaanxi to continue collecting grain contri-
butions in return for degrees. Gansu was a high-priced early experiment in which the Qing state of-
fered social advancement to mobilize merchant capital to serve strategic
needs and to relieve the local civilian population. In some respects it fol-
lowed the Ming dynasty’s use of merchants to supply the frontier garrisons
in return for monopoly salt licenses, but with characteristic Qing differ-
ences. These frontier merchants obtained no monopoly privileges, and de-
grees were available to anyone with the money to pay for them. As the com-
mercial economy expanded on the frontier, the Qing sought to tap the new
flow of resources for the benefit of local stability. Shifting away from their
primary dependence on the land tax, officials looked for new sources of
support from trade. But relying on mercantile wealth had dangerous implications. Unlike sol-
diers or settled peasants, the merchants were mobile and not under official
supervision. They could just as easily use their wealth to bribe poorly paid
local officials, and they could leave the province as quickly as they had ar-
rived. An investigation in 1810 exposed a major corruption scandal that re-
vealed how deeply involved Gansu had become with mercantile interests,
and how these connections had unsettled the local administration. In 1781
a provincial treasurer and his cronies in the magistrate’s offices of Gansu
had taken advantage of the new famine relief policy to line their own pock-
41The provincial officials involved in this scheme diverted the silver con-
tribution funds into their own pockets and took huge amounts of wealth
with them when they left the province for other posts. They were discov-
ered only by accident, when the outbreak of a rebellion in Gansu forced the
new governor to investigate the provincial accounts closely. In this case a
harvests and relief 369

policy designed to have positive developmental effects by directing mercan-
tile capital to a poor frontier region only ended up recycling the silver back
into the hands of greedy southern officials.In addition to indicating the limits to official control over grain collec-
tion, the 1781 scandal also showed how monetized Gansu’s economy had
become. Only the wide availability of cash, and of merchants willing to
pay, could have made this scheme possible in the first place. Besides provid-
ing subsistence, a second key element of the Qing incorporation of the fron-
tier was integrating the border economy with that of the interior. This
meant both standardizing coinage and promoting closer coordination of
prices. Strategic concerns drove these policies of commercial integration as
well. Qing policies had caused considerable integration of markets within
Gansu by the end of the eighteenth century. The same policies were pro-
moted to draw Xinjiang closer to the interior, but on an even larger scale.
The Relief Campaign of 1756
We can learn more about the impact of military logistics on northwest agri-
culture by looking closely at the effects of Qianlong’s armies on grain prices
in Gansu in the 1750s. Because his ambitious campaigns coincided closely
with critical pressures on the local food supply, officials had to wrestle si-
multaneously with supporting the armies on the march and relieving the lo-
cal peasant population.
Grain harvests in Gansu were fairly abundant in 1754, but they began di-
minishing in successive years. (See Appendix D.) The harvests of 1775
ranged from 7.5 to 8.0, and 1756 harvests averaged 7.5, but 1757 and
1758 harvests dropped to the critical level of 6.5. In 1759, the worst year,
the average of 5.5 indicates nearly total drought conditions over the entire
province. (See map, page 371, and Appendix E.) Fortunately, harvests re-
bounded for the next three years, but 1763 was another time of dearth. By
then, however, military operations had been concluded. Droughts struck in
later years, but they did not have the same impact. Prices rocketed to peak
levels during 1759 and 1760. By my calculations, the average effect of
drought and military pressure in 1759 and 1760 was to increase high grain
prices by 2.10 taels and low grain prices by 1.19 taels over their average
level of 1.09 taels.
Although we do not have complete documentary information, we can
trace the outlines of the large-scale relief campaign conducted by the impe-
rial officials in the northwest during these years. They drew on all the
means at their disposal to ensure that peasants received relief and the
troops received their rations.
370 economic basis

The year 1756 was a busy time for northwestern officials. The emperor
had sent a great army after the Mongol prince Amursana, but in order to
pursue him, the generals had to build up stocks of horses, grain, and sol-
diers at the central headquarters in Barköl, and nearly all of these supplies
had to come through Gansu. Fortunately, this year Gansu harvests were on
the whole abundant, although certain counties had been struck by disaster.
In the middle of the year, when it seemed that the rebels had nearly been
captured, troop withdrawals were ordered, but Amursana escaped, to the
great shame of the generals.
43With Amursana out of reach for the winter,
the laborious process of accumulating supplies for the next year had to
begin all over again. By the end of the year, Huang Tinggui could report
that sufficient supplies were stored at Barköl, and there was no need for
further transport from Gansu.
44Grain stocks then amounted to 26,000 shi
at Barköl and 81,000 shiat Hami.
At the same time, twenty-six counties received disaster relief in Gansu
and thirteen in Shaanxi, even though the overall harvest figures averaged
7.5. Recognizing the heavy military demands on the province, the emperor
harvests and relief 371
The northwestern drought of 1759. The blue circles enclosing the numeral “5” indi-
cate the most severely struck regions.
[To view this image, refer to
the print version of this title.]

exempted Ganzhou, Suzhou, and Liangzhou from current taxes and can-
celed their back taxes. Shaanxi was able to refill its granaries after the har-
vest at a cost of 100,000 taels, and Gansu also had an abundant harvest.In 1757 Amursana was still at large. It was feared at first that he would
flee to the Kazakhs, but Ablai of the Kazakhs then submitted to the Qing
and promised to aid in Amursana’s capture. Even though the Kazakhs
would later become an important source of military horses, the campaigns
still had to rely primarily on accumulating supplies from China’s interior. In
preparation for the next year’s campaign, Gansu received 2 million taels for
military expenses and further tax exemptions.
45Governor-General Huang
Tinggui’s request for an extra allocation of 4 to 5 million shiof grain for
military rations had been rejected, but the capital officials realized that
Gansu could not support a large number of troops. Their strategy for 1757
was to assemble a fairly small force in Barköl that could move quickly to
secure a victory before its supply demands were too great. By keeping troop
numbers down, the emperor recognized, he could avoid complaints from
the interior provinces.
46 Grain supplies for these four to five thousand
troops would be met by adding stores from Shaanxi to Gansu’s holdings. The year 1757, however, was worse for harvests than 1756, and in 1758
harvests remained just as bad. (See table in Appendix D.) Rising prices for
grain, especially in Gansu’s western prefectures, not only threatened the
livelihood of the local population but also raised the cost of supplying the
garrisons in Barköl and Ili. These supplies included many other materials
besides grain. The army required thirty thousand horses, to be collected at
Barköl, but Gansu could provide very few of them.
47 Most of the horses
came from Shaanxi, and were sent through Gansu to Xinjiang. The official
price of 8 taels per horse was too low to meet market rates and had to be
raised to 10 taels.
48Oxen were even more critical to peasants, so that rais-
ing official prices from 4.4 to 6.0 taels was still not sufficient. Officials had
to pay 8 taels per animal, but they were ordered not to purchase too many
so as to avoid burdening the local population. Transport costs also rose when harvests were poor. The normal price for
military transport of grain in Hexi was 0.2 taels per shiper 100 li,but in
this year the cost rose by 50 percent to 0.3 taels per shiper 100 li.
49In east-
ern Gansu, over the steep mountain routes from Jingzhou to Lanzhou,
costs were lower than in the west, at 0.16 taels per shiper 100 li,but still
higher than normal.
50 Transport costs from Sichuan ranged from 0.11 to
0.16 taels per shiper 100 li.
The increasing harvest disaster affected both the soldiers and the civil-
52 Garrisons at Anxi were usually given cash to purchase grain in
Suzhou, but this year, because of high prices, they had to loan four months’
372 economic basis

grain rations to the troops, a total of 1.6shiper soldier, in advance, for the
winter months, to be repurchased in the spring. (Average yearly grain ra-
tions for soldiers were about 5 shiper year.)
53The military rationing system
was operating like the “evernormal” civilian granaries, holding off de-
mands on the market in times of scarcity in the hope of restocking after
better harvests arrived. Faced with increasingly serious signs of famine in 1758 and 1759, pro-
vincial officials launched a full-scale relief campaign in Gansu, exempting
all current taxes, accumulated arrears, and unpaid loans in the most heavily
hit areas.
54 These exemptions included diding(land tax), haoxianand
yanglian (nourishing virtue salary supplements), grain and hay supplies,
and loans of seed for planting. The haoxiansurtaxes were normally not ex-
empted, but Gansu was so severely hit that they were included, amounting
to a total of 33,400 taels and 158,640 shiof grain. The western districts,
Ganzhou, Liangzhou, and Suzhou, which faced the heaviest military de-
mand, received the greatest exemptions. Tax exemptions, of course, only had an impact on the future; they did
not provide immediate relief. Direct relief distribution had begun in late
1757 for twenty-two zhouandxian that were struck by frost and hail and
flooding from mountain torrents, but disastrous drought conditions
(hanzai) appeared to be confined to parts of Anxi, including 113,900 mou
of military land.
55In 1758 extensive distribution of relief began throughout
the province. At first it was still possible to purchase grain in low-price ar-
eas, like Ningxia and Gongchang, for transport to other areas that lacked
irrigated fields. Governor Wu Dashan estimated relief needs in mid-1758 at
500,000 shiof grain and 300,000 taels of silver. He also noted that Gansu
could not provide its share of military supplies that year.
56Transport costs,
however, were so high even within the province that it might be preferable
to buy locally even at higher prices. It cost 4 taels per shito ship grain from
Gongchang to Suzhou, for example. The normal practice in Gansu was to give relief half in kind and half in
money. The usual official conversion rate was set at 1 tael of silver per shi,
but now, because of high grain prices, rates were raised to 1.2 taels per shi
in eastern Gansu and 1.3 taels in the west. Because of the additional burden
of military supply, however, the emperor raised each rate an additional 0.10
tael to 1.3 and 1.4 taels, respectively. These rates indicate that the empire
expected military purchases to raise grain prices in the province by less than
10 percent.
In 1758, however, Governor-General Huang Tinggui pushed to have all
relief distributed in money for areas where grain was still available. He ex-
pected further grain shipments from neighboring provinces to make up for
harvests and relief 373

deficits. In fact, it was not possible to give out all relief in cash, but Huang’s
proposal indicates a move toward further monetization of the relief distri-
bution system and an effort to apply inherited rules flexibly.
58During 1759
he gave cash loans for planting seed in the spring in areas where the soil was
too dry to be planted immediately.
By mid-1758 loans and relief were being distributed to twenty-three or
twenty-four of Gansu’s sixty-three county-level units.
60This was clearly in-
adequate, and officials feared unrest. The emperor warned his officials to
be lenient: “Whenever there is a disaster, crowds form to plunder goods...
[and] local officials use crowd violence as a pretext not to give relief. Later
they punish the people as thieves. If crowds are truly famine-stricken peo-
ple who plunder for food, and if they do not have weapons and their num-
bers are small, governors should be lenient according to the situation.”
The emperor was more concerned that local officials would deny relief than
he was about mass unrest. As far as we can tell, there were no major upris-
ings in Gansu during these famine years, but the potential remained. In
1749, for example, crowds of over one thousand people had gathered at
some county offices to demand tax remissions and granary distributions.
They were led by “evil gentry” (liesheng),probably lower-degree holders
from the local population. The crowds dispersed quickly, but troops were
put on alert to prevent major conflict.
As the spring of 1759 approached, peasants faced the critical period
when crops planted the previous winter had not yet sprouted and fields had
to be prepared for summer planting. They called it the “time when the
brown and green do not meet” (qinghuang bujie);French peasants knew it
as the soudure. Every year at this time, grain stores reached bottom and
prices soared. The four-month relief campaign that carried the population
through the winter now had to be extended by three more months, as there
was little sign of rain. At the same time, government granaries sold their
stores to reduce prices: millet at 2.4 taels and wheat for 2.2 taels per shi.
Just at this crucial moment, the emperor’s long-trusted northwest Gover-
nor-General, Huang Tinggui, died. Because he had managed the logistics
of the campaigns so superbly, his death was a great blow. Governor Wu
Dashan, now promoted to Governor-General, was ordered to follow
Huang’s precedents strictly.
64Prices, however, continued to rise ominously
in Lanzhou. To avoid straining local markets, military quartermasters were
told to stop buying supplies on local markets and instead distribute sup-
plies in kind.
As prices peaked at over 4 taels per shi,officials desperately attempted to
induce merchants to bring supplies into the province.
66 Sichuan was the
most productive neighboring province, and grain could be moved by water
374 economic basis

routes into southern Gansu, but the cost of land porterage to the rest of
the province was extremely high. From Lueyang in Shaanxi, on the upper
Jialing River near the border, to Gansu by land, a distance of 1,200 to
1,700li,would cost 2 taels per shiin transport costs alone, in addition to
the water transport costs to reach Lueyang. Since Shaanxi held 1.2 million
shi in granaries on the border with Gansu, it seemed more logical to obtain
the bulk of grain supplies from Shaanxi and to use Sichuan’s surplus to re-
lieve the strain on Shaanxi. Shaanxi was facing its own harvest crisis, for
which it received tax exemptions, but it could ship up to 400,000 shiof
grain to Gansu. Ultimately, over 200,000 shiof Shaanxi’s grain were allo-
cated to granaries in Gansu.
This complex interprovincial transfer scheme relied not only on govern-
ment movements but also on expectations of the responses of merchants
and consumers to grain markets. As one official noted: “Even if transport
costs [to Gansu] are high, as prices rise, small peddlers and porters will not
sit still. Households who hold grain will hear of the [high prices] and offer
it for sale, thus helping the public by leveling prices.”
68Shaanxi’s shipments
could serve as a pump-priming mechanism, creating expectations that
prices would drop, thus inciting households with grain to unload their
hoards. At the same time, Sichuan’s supplies would dampen resentment in
Shaanxi at losing its own vital grain stores. The fear that “foolish people
will resent” having grain shipped out of their locality, and would create
grain blockages, was on the minds of officials and the emperor. Throughout
the crisis, concern surfaced that “people of the interior would resent” the
diversion of their local supplies.
69In many other parts of the empire, grain
blockages occurred frequently in response to the arrival of outside mer-
chants at local markets.
70 By announcing in advance that Sichuan grain
supplies were arriving, the officials could manipulate expectations in
Shaanxi and hope to keep grain prices low.
At the same time, officials insisted that the shortage must be relieved by
“all possible means” (duofang).Other, more favored provinces were urged
to economize on their consumption for the sake of the stricken region. Tex-
tile producers were urged to reduce consumption and send their surplus to
Gansu so that its people could spend their money on grain.
When material supplies were not enough, officials turned to other means,
including public works and ritual activities. They hired famine-struck peas-
ants to rebuild city walls, enacting the principle of “using labor to provide
relief” (yigong daizhen).
73 Recalling the great North China grain crisis of
1744, in which the government had relieved over 1.6 million people in
Zhili and Shandong, the emperor fasted, conducted prayers for rain, and
ordered the release of those imprisoned for petty crimes, so as to “gain the
harvests and relief 375

grace of Heaven.” 74As rain began to fall in the fourth to seventh months of
1759, relieved officials and farmers prepared to plant summer crops, look-
ing forward to a plentiful fall harvest.
75 But they were disappointed. The
rains came too late, and the fall of 1759 continued at disaster levels. Many
of the fields replanted in the summer turned out to be barren. Much of the rest of the empire was suffering, too. Zhili and Shanxi faced
insufficient rainfall for most of 1759, but the rains finally arrived, and
they produced adequate harvests, although swarms of locusts moved into
Shanxi from Henan and Zhili. Shandong required relief in sixteen counties.
Shaanxi, in addition to supporting Gansu, had to divert supplies from its ci-
vilian evernormal granaries to support military units.
76 Zhejiang also had
shortfalls, but purchasing relief grain in Jiangsu and Guangdong threatened
to drive up prices there, so 150,000 shiwere shipped down the river from
Hunan. Without the great grain production boom of Hunan, fed by the
clearance of lake and river borders by enthusiastic settlers from down-
stream, much of the lower Yangzi would have suffered severe shortages.
The overriding goal of the authorities during this crisis was that “mili-
tary supplies and agriculture should both not be obstructed [junxu yu
nongye liang wu yiwu].” Part of the evernormal granary stores, usually in-
tended only for civilian use, had to be diverted to the military. By the fifth
month of 1759, Gansu had already used half of the 2 million shiin its gra-
nary for both military and civilian relief.
Military grain requirements were much higher per capita than for reliev-
ing civilian famine. Daily rations for soldiers were 8.3 he(weighing 8 to 10
jin) of raw millet (boiled into gruel) or 1 jinof noodles and bread per day.
Relief for civilians was at most 5 heper day. The average monthly military
grain ration was 0.4 shiper month, or 4.8 shiper year, while adult civilians
received 1.8 shiper year.
80Usually, 80 percent of military rations was given
in cash and 20 percent in kind, but since purchases on local markets drove
up prices, soldiers were given advances in grain to last them through the
winter months. Officials also sold grain to level prices for the benefit of the
military. In Jingyuan garrison, for example, they released 1,000 shionto lo-
cal markets from county granaries so that troops could buy grain. Wheat
and millet prices had soared to 4 to 5 taels per shi;sales were made at 2.2 to
2.4 taels per shi.
81 In 1757 the government met military demands from
within the province by shipping 50,000 shifrom eastern to western Gansu,
but in 1759 extra supplies had to be brought in from Shaanxi.
Gansu officials staved off famine only by nearly totally depleting the
province’s granaries. Reserves of 2.35 million shihad dropped to 1.2 mil-
lion shiby the first month of 1759.
83 A later report, from 1763, revealed
how low the granaries’ holdings had fallen. After the two bad years of 1758
376 economic basis

and 1759, total grain stores in Gansu dropped to 490,000shi.By 1763 they
had still not recovered to their 1756 level of 3.3 million shi.Even the nomi-
nal holdings of 2.9 million shiincluded unrecovered loans of 1.2 million shi
and grain sold but not yet repurchased of 527,000 shi,so the actual hold-
ings were only 1.169 million shi.This report reveals the dual character of
Gansu’s granaries: on the one hand, they were never able to meet their tar-
get figure of 4.5 million shi,but on the other, their reserves were very ac-
tively used to relieve harvest failures. In good years, reserves could be built
back up to more than 3 million shi,giving Gansu province one of the high-
est per capita holdings in the empire. In sum, out of severely limited resources, northwestern officials fash-
ioned a highly interventionist famine relief regime that just barely managed
to save the local population. Military mobilization and drought strained
the empire’s food supply to extreme limits in its poorest, most vulnerable
interior region. Relief officials used every trick in the book to move grain
there: granaries, merchant inducements, interprovincial shipments, rain
prayers, public works, tax relief, loans, and gruel stations. Pulling out all
the stops worked in this case because the underlying concern for security
drove military and civilian officials to work together to relieve peasants and
soldiers together. In the future, such smooth cooperation would be rare.
harvests and relief 377

Currency and Commerce
Q ingmarkets, like all others, were made, not born. The system did
not work perfectly, and it did not grow naturally. Like the nationalist tele-
ology, the myth of laissez-faire views markets as growing organically and
sequentially. This myth likewise ignores the contingent character of market
systems and their dependence on the institutional context. The imperial
state fostered their growth with specific measures, which have been studied
in detail.
1What needs emphasis here is the link between these market-fos-
tering initiatives and the security needs of the state. From the sixteenth century, the flood of New World silver to China
helped the integration of local and regional markets. Studies of the correla-
tions of grain prices between regions in the eighteenth century indicate how
closely attuned peasants and merchants were to market conditions at a dis-
tance, and how grain flows directed by price differentials compensated
for harvest shortfalls.
2My analysis of Gansu found that these integrative
trends extended even to the northwest. The Yongzheng emperor, much
more concerned with the internal economy than with frontier supply, in-
sisted on the need to allow resources to “flow without obstruction” (tong).
He maintained that provincial officials should not have a “border con-
sciousness” (ci jiang ci bian) when it came to grain and silver trade. Qian-
long could build on the structure of the early eighteenth century, so that
when his officers purchased grain for military needs, markets responded,
and local shortages were compensated by imports. Since markets required money, Chinese officials had long experience
in discussing the relationship between currency supplies and commerce.

Rulers could not simply command money to flow; they had to devise incen-
tives to induce traders to direct currency toward border regions. Currency
policy and security aims, inextricable from each other, shaped the frontier
discourse on war and money. Here, I will examine three aspects: (1) the mil-
itary motives of monetary policy; (2) the coexistence of highly localized
markets and interregional trade displayed in currency circulation; and (3)
the impact of state policies on monetary integration of the frontier.
Money on the Frontier, from Song through Ming
Classical writers and dynastic rulers recognized that control of the currency
supply was vital to the maintenance of social and economic order.
the Han dynasty, the chief form of money under state control was the
“cash” (qian),a round coin with a square hole in the middle, made from an
alloy of copper and other metals (tin, zinc, or lead). Strings of cash of up to
a thousand coins served as the primary medium for all daily transactions.
By increasing the output of the mints, or changing the metallic composition
of the coin, officials could adjust the number and quality of coins in circula-
tion. The state had two goals in regulating the currency supply: obtaining
revenue, and stabilizing markets so as to ensure the welfare of the people. It
could obtain revenue directly from seigniorage charges, the difference be-
tween the costs of production at the mint and the value of the coins in circu-
lation. Indirectly, when taxes were collected in money, the treasury gained
from the extension of commerce through the country. Greater access to
markets for their crops made it easier for peasant farmers to pay taxes. No dynasty, however, could completely dictate the use of money. Mer-
chants, consumers, and farmers used only monetary media which they
trusted. State backing was not always enough to ensure that a coin would
be used. The paradoxical effects of Gresham’s law, in which bad money
drives out good, meant that a coin of high value would be hoarded, while
counterfeit and debased coins circulated. Coins with high copper content
could be melted down when the value of the metal itself exceeded that of
the coin. Mints in search of profit were tempted to produce debased coins