1 / The Frontier Ground and Peoples of Northwest China

Well, I at least find myself reflecting on this point. A geographical area keeps a certain flavour, which manifests in all its happenings, its events. . . . I sometimes wonder if this thought may not be usefully taught to children at the start of their “geography lessons.” Or would one call it history?

Doris Lessing, Shikasta

Muslims live almost everywhere in China. A few small clusters are located in the south, more in the northeast, and hundreds of thousands on the north China plain, with the densest concentrations at Beijing and Tianjin, though they constitute only a tiny minority among the non-Muslim Chinese.¹ Tens of thousands of Muslims live in Yunnan’s cities and market towns, and smaller numbers may be found along the trade routes leading to Burma and Tibet from the southwest. But Muslims in contemporary China still live most densely along the ancient Silk Road, which connected Central Asia with north China. Eastern Turkestan (now the Xinjiang Uygur Autonomous Region) has had an indigenous population that is almost entirely Muslim and non-Chinese-speaking for centuries, joined only during the past forty years by millions of non-Muslim Chinese.

Between north China and Turkestan are two zones of dense Muslim habitation: the Ningxia region of the middle Yellow River valley (now the Ningxia Hui Autonomous Region) and a crescent of territory in southwestern Gansu and northeastern Qinghai Provinces. These zones, one on either side of the regional core at Lanzhou, are the ground of most of this book—a frontier of four cultures, a region known in China for little except poverty, marginality, archeological riches, and bellicose inhabitants. Though I must begin with a general history of Islam’s arrival and development in China, since no such narrative exists in English, the lion’s share of this work focuses on the northwest.

MAP 1. Contemporary provinces, autonomous regions, and municipalities of the People’s Republic of China.
(Map by Philip M. Mobley)

Therefore this book is not about China, but about a particular part of China, a frontier very distant from the core areas of Chinese culture and very strange to most Chinese. That does not, of course, reduce its importance in Chinese history. The distance of the northwestern frontier from “China proper” (which northwesterners call neidi, the interior) enhances its value as a lens on the range and diversity of Chinese life. Local history, not in the mode of case studies in search of the typical but as an understanding of the particular for its own sake, deepens and subtly diffuses our comprehension of what it is to be Chinese. The generalization and homogenization of Chinese society by scholars, both Chinese and foreign, distorts the real and compelling variety created by distinct environments and their particular histories.

The acts recorded by chroniclers of northwest China have tended to be violent and antisocial, construed as immoral by the Confucian judges of history. In a narration of this rowdy past, we must ask why people, often neighbors for years or for generations, take up weapons to kill one another at particular historical moments. What begins what Barbara Tuchman has called “the march of folly,” leading to bloodshed on a small or large scale? It will not do to say, as many have, that Muslims are naturally violent and fanatical people because of their doctrine. Nor can we aver that Chinese people do not care about human life because there are so many of them. Or that frontiers are just violent places. Historians of the particular cannot ignore or devalue the often peaceful Muslims, the often life-affirming Chinese, the often calm fronters. In order to discover patterns and clues in the specific time and place, we must patiently explore what people actually did.

In the civil, avowedly antimilitary and culturally homogeneous China of the dominant mythology, violence may properly serve only the holder of the mandate of Heaven.² Yet in folk tradition, from the early legends to the battles of Cao Cao to the heroes of the Shuihu zhuan and the Luding Bridge, warfare and martial heroism have vaulted men and women into exalted memory, and many parts of China have dramatic, explosive local traditions of violence going back to the Han dynasty (206 B.C.E–C.E. 220) or even before. A local historian of Jiangsu tells me that there are areas of central China where collective or large-scale violence plays almost no part in history. But to call such places “typically Chinese” ignores not only stereotypically violent frontiers such as the northwest but also counties in Zhejiang, Anhui, and even Jiangsu itself that have rich and troubled traditions of violence. To understand China in its particulars, we must assimilate that violence into our imagery.

Thus we can build on Hsiao Kung-ch’uan’s famous metaphor—China as a vast mosaic of environments and histories. In this book I focus on a small part to discover the colors and textures of its many fragments, their brightness and sharp corners, their diverse shapes. As local history becomes more popular and more viable, a clear vision of the parts must, in our history of scholarship, clarify the complex human effort that created the whole.

FRONTIER GROUND

October 1938. Gu Jiegang was tired and worried. His father was dying back home in Suzhou, but the forty-five year old professor of ancient history knew he could not travel east to do his filial duty. He had been on the road for over a year, having left Beiping just ahead of the invading Japanese, and had arrived in Sichuan only in September, after an exhausting fact-finding tour of the northwest. Not yet ready to report to the Sino-British Cultural and Educational Endowment Fund, which had sponsored his excursion in the hope of learning more about education among the northwestern Muslims, he nonetheless agreed when an old friend asked him to give a lecture. He had to speak to his audience—the faculty and students of the Mongolian-Tibetan School of the Central Political College—in a makeshift hall, for Chongqing’s universities, like its sewers and housing market, had been flooded by refugees fleeing the Japanese:

When Demchukdonggrub [Ch. De Wang] started the autonomy movement in Mongolia in 1933, I met with him and his associates at Bailingmiao. Only after that did I realize the gravity of frontier issues, so I changed my direction and began to study frontier problems. . . . [On my recent trip] most of the places I visited were inhabited by Hui and Fan [Tibetans]. . . . Banditry is a really serious problem there. . . . Though we met with some local desperadoes, fortunately they didn’t rob us.

The most severe and most pressing problem in the northwest is transportation. . . . If you haven’t been there, you couldn’t imagine it, but between neighboring counties, even townships, people rarely communicate. . . . If the northwest’s transportation problem is not solved, there’s no sense even talking about the others.

Of the places I visited, many were districts with complex racial [Ch. zhongzu] and religious intermingling. . . . The Muslims have Old Teaching, New Teaching, and New–New Teaching, while the lamaists have Red, Black, Yellow, and Flowery sects. Among the races there are Han [Chinese], Manchus, Mongols, Hui Muslims, Qiang and Fan, Salar Muslims, and Turen [Monguors, a Mongolic-speaking people]. . . . Because northwesterners have all these factions, all these mental barriers, they know only that there are sectarian divisions, and they don’t know that they all are citizens of the Republic of China!

. . . people from the rest of China rarely go to the northwest, and those who do are all merchants, low-class salesmen, with small minds and love for high profits, who often jack up prices and cheat ignorant Tibetans and Mongols.

The places we went on this trip are actually in the middle of our country’s territory; when people say we got to the frontier, it makes us feel really ashamed. . . . So our responsibility in “frontier work” must be gradually to shrink the frontier, while enlarging the center, so that sooner or later the “frontier” will just be the border.³

We have no record of the audience’s reaction to Gu’s lecture, but he remained convinced that the northwest held one of the keys to China’s future, so he spent much of his career investigating its history. His father died that winter without ever seeing his son again.⁴

The Frontier Ground of Gansu

Gu Jiegang was right about the northwest’s physical position inside China’s border—the provinces he visited are only slightly north and west of the country’s geographical center, if the vast areas conquered by the Qing are included in “China.” But he was certainly ingenuous about culture. Despite its proximity to the “center” of China, what he called the northwest constitutes the meeting ground of four topographical and cultural worlds: the Tibetan highlands, the Mongolian steppe, the Central Asian desert, and the loess of agricultural north China.⁵ This frontier zone encompasses the perimeters of cultures that have been in evolving contact on this same ground for centuries. We might have a comparable area in the United States if large numbers of Navajo farmers, Sioux hunters, Anglophone ranchers, and Spanish-speaking herders all had been packed for a long period of time into a three-hundred-mile-square, ecologically diverse region of New Mexico. The comparison between northwest China and the American west did not escape a contemporary Chinese urbanite sent to work in Qinghai during the Cultural Revolution:

The road passed through hills and valleys that were at times bleak and empty, reminding me of what I had imagined the American wild west would look like, based on a movie I had once seen in Shanghai. “Where were the cowboys and the Indians?” I asked myself, and later of course it turned out that I would be one of the cowboys and the Tibetans would be the Indians.⁶

In physical space eastern Gansu forms the transitional zone between steppe/highlands and arable lowlands, its climate dry and severe, its topographical contrasts sharp and sudden.⁷ Both of the historically crucial roads in the region—between Central Asia and the cultural cores of China (west-east), and between Tibet (Xizang) and Mongolia (south-north)—pass through the Gansu corridor, the narrow gap between the northern Tibetan mountains and the Mongolian desert.

The complex topography has contributed to equally complex human geography. In historical time, most of Gansu has been numerically and politically dominated by Chinese people and the states ruling them, both the conquest dynasties and the domestic. As Gu noted, sharing or competing with them for the productive and strategic resources of Gansu have been diverse non-Chinese peoples, among them Tibetans sedentary and nomadic (including a few Muslims), Turkic-speakers (Muslim and non-Muslim), Mongolic-speakers (Muslim and non-Muslim), and mixtures among the four. New ethnic identities have evolved as peoples and states advanced, contracted, and mingled along this multicultural margin, strategically crucial to many states but central only to those who live there. Personal and collective identities, elusive and processual in all human societies, have proved particularly troublesome in frontier areas when modern states attempt to rigidify boundaries and classify people; Gansu is no exception.

In central Gansu, at Jiayuguan, lies the symbolic end of the Chinese agriculturalist’s domain, the terminus of the Great Wall. Even inside the wall, however, the Gansu topography dictates a diverse economy, with livestock breeding and trade in animal products playing crucial roles beside cultivation of food grains and artisanal production. The eastern half of the province, with its New England–like climate, produces good tobacco, fruit, millet, and medicinal herbs. Both eastern Gansu and the Gansu corridor have soils and conditions well suited to the opium poppy, planted in great abundance in the nineteenth and early twentieth centuries.

Much of Gansu, however, does not look as agrarian China should (see plate 1). The hillsides and grasslands, especially in the high country of the Tibetan mountains and Alashan, provide pasture for sheep, goats, local breeds of hardy cattle, horses, deer, and yaks. Between the river valleys, away from water sources, much of Gansu lies barren, the loess deeply eroded into steep hills without sufficient topsoil or water for agriculture. There the population spreads more thinly across the landscape, and even the best roads wash out in the rainy season or collapse when earthquakes strike. Gu Jiegang knew from firsthand experience how many parts of Gansu, remote from regional or even local centers, remain difficult to reach because of the infrastructural obstacles posed by topography and climate.

Gansu Geographical Regions

The Yellow River, its tributaries, and the deforestation of its watershed have carved out much of this harsh terrain. The rivers—very rapid as they flow down from the high mountains—obstruct transport, irrigate narrow flatlands along their banks, and erode the slopes from which the trees have been removed. They also define the regions into which the province has traditionally been divided. Lanzhou, the provincial capital and commercial and transportation center of the whole region, lies on the Yellow River’s south bank, directly between the two zones of dense Muslim habitation. Through the city runs the Silk Road, now a paved highway, the main land route from central China to Central Asia, as well as a north-south road connecting Gansu to Sichuan and distant Yunnan. Though nearby hills make extensive farming difficult, the Lanzhou region does produce grain and fruit, especially large (and very sweet) pears. Upriver lie the gorges of northeastern Tibet, and downriver dangerous rapids hinder irrigation of the poor farmland along the banks.

MAP 2. The five provinces and autonomous regions of northwest China. (Map by Philip M. Mobley).

Ningxia. The Yellow River finally straightens out northeast of Lanzhou, running sedately across the loess plain east of Alashan to the city of Ningxia (now Yinchuan), the “Lower Yangtze of the steppe.” There, at the edge of the Gobi Desert, an ancient irrigation system supports a rich and mixed agricultural economy, including paddy rice. The irrigated plain east of the river stretches toward the alkaline desert of the western Ordos, a source of wealth from salt extraction but a dangerous obstacle to commerce with north China. Southeast of Ningxia, the ill-favored loess country of Haicheng and Guyuan contains some of Gansu’s worst land and poorest people, ending in the rugged Liupan Mountains, where Chinggis Khan is supposed to have died in the thirteenth century.

Longdong.⁸ The eastern quarter of the province lies at the upper end of the Wei River valley. The Wei, a tributary of the Yellow River at the great bend (Tongguan) in southeastern Shaanxi, flows through one of the earliest centers of Chinese civilization. Shaanxi Province, also called Guanzhong, the “Area within the Passes,” saw the flourishing of the Zhou dynasty, the rise of the Qin, and the stable hegemony of the western Han, all before the beginning of the Christian era. The Sui dynasty’s canals brought the produce of the south to this yellow-earth country in the sixth century C.E., and a few decades later the Tang emperors built their supremely cosmopolitan Chang’an on the Wei’s southern bank, at present-day Xi’an, where many former capitals had stood.⁹ The upper Wei valley in Gansu has sufficient water, a temperate climate, and enough flat land for reasonably well-graded roads, with the dry, barren Liupan mountain range as its source and its northwestern limit. Longdong remains more closely connected to Shaanxi and China proper than does the rest of the province, its large towns almost all lying on the Wei River or on the Xi’an–Lanzhou Road.

Longnan. South of the Wei watershed, east of the Tao River, where Gansu and Sichuan meet, the southern part of the province lies partly inside the Yangzi watershed across a divide in the Qinling Mountains. Well watered but high, with a short growing season, southern Gansu farmland produces barley and oats as well as millet and wheat. Protected by its mountains, this part of the province attracted refugees from wars and disasters to the north and east.

The large Gansu tributaries of the Yellow River—the Xining (Huang), Daxia, and Tao Rivers—all rise in the Tibetan highlands. Flowing among chaotic mountain ranges, they subdivide southwestern Gansu into distinct regions: Huangzhong, Taoxi, and Longnan, all bordering northeastern Tibet (Tib. Amdo).¹⁰ Each of these rivers traverses a transitional landscape from its source in the high mountains to steep foothills surrounding narrow valleys, then to broad valleys marking the entrance from cultural Tibet into cultural China. Divided from one another by the rapidly flowing rivers, which could be crossed with ease only in winter, these regions also contain a variety of ecological zones, from high grasslands suitable for nomads to well-watered arable plains, and thus constitute meeting and mixing grounds for a variety of ethnic groups and cultures.

Huangzhong. Centered on the Xining River watershed between the Yellow River and its major city, Xining, cut by rapid rivers and battered by a harsh climate, the narrow valleys of northeastern Tibet nonetheless can yield good crops to their ethnically diverse cultivators.¹¹ A local proverb says, “Gansu is dry, but Qinghai is green” (Ch. Gansu gan, Qinghai qing), the green including both fields in the valley bottoms (which sometimes lie above 8,000 ft.) and dense forests on some of the slopes.¹² Bounded by Qinghai itself (lit., “the blue-green lake,” Mong. Kökenōr), the Tibetan massif, and the Qilian Mountains in the north, this area has often been of strategic importance as a meeting ground for Tibet, Mongolia, and Turkic-speaking Central Asia, as well as for China’s relations with all three. Included in Gansu by the Qing but made part of a separate province (Qinghai) in the late 1920s, Huangzhong also demands a pastoral or semipastoral life from many of its inhabitants, for the steep, high hill country does not offer large expanses of arable land (see plate 2).

Taoxi. The triangle west and north of the Tao River and east of the mountains, with the Dongxiang mountains at its apex, constitutes a core zone for this book. The Muslim-dominated city of Hezhou (now Linxia) and the Tibetan center at Labrang define an ethnic-cultural progression between China and Tibet. Taoxi marks the southwestern boundary of arable loess and the northeastern boundary of Tibetan pastoralism, and its ethnically mixed population reflects the topography. Though much of Gansu has adequate rainfall in good years, the rain often arrives suddenly in large quantities rather than spreading itself over the growing season. Heavy rainfall can be disastrous in loessial areas such as Taoxi, for that dusty soil must be watered evenly and slowly to produce good crops. Both the Yellow River and the Wei carry heavy loads of loess silt away from Gansu, leaving the northwest more leached and less arable every year. Deforestation and subsequent erosion cut sharp valleys through the loess plains, creating a jumble of tiny plots among the steep slopes.

Amdo. West of the Daxia, south of the Yellow River, northeastern Tibet is largely pastoral and inhabited primarily by Tibetan and Turkic speakers. Unsuited to any but high-country agriculture in which barley is the dominant grain, the mountainous Amdo country rewards pastoralists more than farmers. Based in the market towns by the rivers, Muslims have for centuries moved between the pastoral and agricultural zones as middlemen, brokers, and translators. Connected to Lhasa only by long and dangerous roads, Amdo nonetheless constituted a crucial transportation link between Tibet and Mongolia, especially once the latter culture area came to be dominated by Tibetan Buddhism and its lamasery-based political culture. Three great lamaseries—Labrang (south of Hezhou), Choni or Zhuoni (on the upper Tao), and Kumbum (near Xining)—and a host of smaller ones controlled the wealth and military power of the region well into modern times.

Multiple Cores, Multiple Peripheries

In such a hodgepodge of environments, Gansu society could hardly have been a unified, geographically centered system. In G. William Skinner’s mapping of China’s regions, Gansu is the periphery of the periphery, the outer edge of the northwestern macroregion, which has its core at Xi’an. This book argues that such a single-centered view does not adequately describe Gansu. The Tibetans, Salars, Chinese-speaking Muslims, Mongols, and Mongolian-speaking Muslims all had cores and peripheries arranged very differently. Without understanding their systems, and the interactions of these conceptions of spatial relations and hierarchies, we will misconstrue both the geography and the history of the region, as Gu Jiegang did and many scholars still do.

For example, at this distant edge of their civilization, the Chinese of Gansu did not simply display the language, habits, and values of their core; they also adapted to the harsh environment and to the non-Chinese cultures of their neighbors.¹³ Frontiers are always in motion and always have two sides, so Gansu cannot be seen simply as an edge of cultural China. It also constitutes an outer edge of Tibet, Mongolia, and Muslim Central Asia. Beginning with their own unique histories, cultures, and cores, the non-Chinese of the frontiers also evolved and changed through their contact with one another and with China. These mutual adaptations, rather than any one-way transmission, encompass the social and cultural processes of this book, as the severe, diverse Gansu terrain forms its geographical setting.

The process of incorporation that Gu Jiegang recommended and so deeply desired has indeed changed the nature of China’s peripheries as China has changed. The Gansu frontier zones, which had been shifting areas of Chinese/non-Chinese interaction and syncretism, have gradually and sometimes painfully given way in modern times to the new and powerful concept of political boundary, a demarcation within which national sovereignty rather than imperial virtue holds supreme power and demands local loyalty. As the Chinese state changed, so did the pressures imposed on peripheral peoples, especially on leaders who chose to ally or identify themselves with the political center. The Qing tusi system, under which hereditary “native chieftains” (to use the most popular nineteenth-century English translation) were granted imperial titles to rule the frontier, had to be replaced by “national” rule or by local leaders thoroughly imbued with the values of the nation-state. Such leaders would generally be from the group who have come to call themselves Han, the “culturally Chinese,” or from the elites of non-Chinese groups who see their advantage, and possibly that of their communities, in representing state power locally.¹⁴ The great diversity of Gansu peoples made local leadership a particularly thorny problem, so the emergence of a new northwestern Muslim elite committed to the political center of China—first Qing, then Republican (1912–49)—beginning late in the nineteenth century forms an important part of this book’s story.

Gu Jiegang did not tell his audience in 1938 that China, newly self-conscious and struggling to constitute itself as a nation-state, had imposed on its peripheries with a heavy hand as it underwent its painful modern adaptations. National leaders and nationalist thinkers needed to form a new China both as an idea and as an institution, a modern nation-state utterly different from what they saw in the anachronistic Qing empire, yet covering the same territory and tied to the past with the strong bonds of language and historical consciousness. Gu spoke of expanding the center (Ch. zhongyuan) until it includes all of China, leaving the national border as the frontier; that is clearly the goal of national integration and nation building all over the world. But even now Gansu and Qinghai people refer to their provinces as “frontier” (Ch. bianjiang) and the rest of China as the “interior,” for they live among cultural Others and see themselves as living on China’s cultural edge.

The first part of that nation-building process, the conceptualization of a new world, took place almost entirely without reference to China’s peripheries. There were a few new-style schools (Ch. xuetang) in Gansu after 1900, and small groups of young Communists had to be eliminated by the Lanzhou authorities in the 1920s.¹⁵ But the modern world arrived in the northwest primarily as military technology, wielded by leaders who fought their most important battles back east. The reformers, revolutionaries, cultural essence (Ch. guocui) advocates, and New Culture intellectuals neither knew nor cared much about Muslims, whom they knew to be violent, or about wild and woolly Tibetans, except as potential threats to the unity of the nation.

Another part of the modern process, the creation of a new type of state, forcibly included the periphery in the recasting of an imperial (and imperialized) anachronism into a unified polity capable of survival in the hostile, aggressive twentieth-century world system of nation-states. New China engaged its frontiers with power rather than persuasion, so the ideologies and discussions of the 1920s touched Gansu very lightly.¹⁶ The provincial administrators, bureaucrats, and generals of Republican China used the tools and techniques of the modern world in the northwest only to enhance their control over territory. The Christian general Feng Yuxiang arrived in Gansu not with programs against opium but rather with campaigns against local militarists and factions of unruly Muslims.

The late Qing and Republican periods did see the transformation of the northwest, not by the polemics or ideals of modernization or development but by the process of military incorporation. The “new” local elite rose not primarily from the ranks of students or cadets, not from new political activists participating in boycotts or strikes, not even primarily from the more traditional elites who joined the provincial assembly. Among the non-Muslim Chinese elite, local officials of the defunct Qing dynasty retained their high status and gained Republican legitimacy, while a Gansu Muslim elite evolved from the Qing military, sometimes supported by the Muslim politico-religious groups (menhuan) that had enough cash to recruit and equip armies.

With elites such as these, Gansu (in contrast to much of China proper) had a period of warlord domination virtually without a New Culture movement of urban intellectuals. The northwestern periphery was a symbol to be secured, a bastion to be defended, a resource to be tapped, but not a place congenial to the self-conscious designers of New China’s identity. Gu Jiegang’s intellectual contemporaries tended to call the northwesterners outlandish, violent, and hostile to a civilized new order. Gu Jiegang and his companion Wang Shumin, however, trekking between Lanzhou and Xining, visiting schools and soldiers and officials, saw China’s Muslims and the whole northwest with very different eyes, describing them as potential wellsprings of virile energy for the building of a new nation. In that fateful year of 1937, Gu wrote: “In the revival of the Chinese race, the Moslems would have a large responsibility.”¹⁷ He knew, if many of his compatriots did not, that the Muslims of northwest China were Chinese.

FRONTIER PEOPLES

The Salars have their own independent language, naturally different in basic vocabulary and grammatical structure from the languages of the surrounding Han and Tibetan peoples. It is close to the languages of the Turkmen and the Uzbeks living in Samarkand. . . . They all belong to the Yugus branch of the western Xiongnu group of the Turkic languages, which are part of the Altaic language family. The Salars’ relatively tall and large physique, abundant facial hair, high-bridged noses, and deep-set eyes and other external features clearly mark their differences from the neighboring Han and Tibetan peoples and prove that they have a close racial resemblance to the peoples of Central Asia.¹⁸

Many a traveler has tried to describe anthropologically the Salar type—tall, aquiline noses, etc. Such an undertaking has very little value. As we have already seen, the Salars, at least in the past two hundred years, are a great mish-mash of Turks, Mongols, Tibetans, Chinese, and Huihui [Chinese-speaking Muslims].¹⁹

People of the Frontiers

In Gansu, agriculture, sedentary pastoralism, and various degrees of nomadism could flourish in close proximity. Frontier peoples, including the Chinese, thus evolved their individual and collective identities in extended contact with many cultural Others. In Gansu live a wide variety of mixed peoples, all on the frontiers of their linguistic cultures: Turkic-speaking Muslims, Mongolic-speaking Muslims, Chinese-speaking Tibetans, Tibetan-speaking Muslims, Monguor-speaking Muslims and non-Muslims, and more. These groups, most of them now carefully but inconsistently classified as “minority nationalities” by the state, have in earlier times been more fluid, flexible entities. Individual fronter folks, by conscious choice or unconscious adaptation, defended and altered their identities among their neighbors.²⁰

In Gansu a non-Muslim Chinese male might convert to Islam to improve his chances in business, his female cousin in order to marry a Muslim man who had accumulated substantial wealth. A Muslim merchant might eat pork in order to enhance his working relationship with a non-Muslim or to avoid offending an important host. A Tibetan nomad might make a permanent home in a village in order to plant crops and avoid the seasonal disasters of herding. A Turkic-speaker might encourage his children to learn the local dialect of Chinese to better their prospects in an area dominated by the Chinese, or discourage them in order to preserve the purity of his Islamic heritage. A Mongolic-speaking Muslim might be able to earn a living only by migrant labor in non-Muslim areas, so he and his family would travel and learn Chinese by necessity. To take a concrete example, in Tangwangchuan, on the west bank of the Tao River south of Lanzhou, half of the Wang clan were Muslims while the other half were not, and the same happened within the Tang lineage (see plate 3). While kinship ties united Muslims and non-Muslims, members of each clan also could be enemies of kinsmen and allies of coreligionists, when religion became a divisive valence of identity.²¹ In some Tibetan families that did business with Muslims, one son might convert to Islam, sharing the family home with a brother who had become a lama. Such decisions and changes, one by one, have created the complexity of Gansu ethnic identities.

According to an ideological tenet held by many Chinese in premodern times, non-Chinese must either remain in the peripheral darkness of barbarism or they must laihua, come and be transformed by the civilizing power of Chinese culture. In this mythic construction, China exerts a stable, immobile magnetism that draws the susceptible (that is, the proximate) into sinification by its moral power. This characterization must be carefully glossed by historical reality.²² Many people have existed along China’s borders for millennia, altered by China’s presence but never especially attracted to or absorbed into Chinese culture and society, as the case of the Tibetans most vividly shows. Though some Chinese might argue that this demonstrates their reactionary tendencies or sheer barbaric stupidity, as observers we may conclude that not everyone who knows China wants to be Chinese. Some may want to trade, others to raid, but they certainly desire to remain themselves, however syncretically their frontier culture may have grown.

The Gansu Muslims: Patchwork Society

The Muslims of Gansu did not function as a unified community in any sense of the word but were divided along many lines, including geographic ones.²³ A wide variety of Gansu urban neighborhoods, towns, and villages may be described as “Muslim,” but they differed in location, size, relations with non-Muslim peoples, economic activities, religious affiliations, and political allegiances (see plates 4–7).

Close to the Shaanxi border, in a loess valley, lay the entirely Muslim market town of Zhangjiachuan. No non-Muslims lived within miles of the place, except a few artisans who were careful to obey the strictures of Muslim dietary purity in order to continue living there. The tax officers from the county seat could visit and collect, but otherwise non-Muslims came only for the periodic hide and leather market and for the livestock fair of nearby Maluzhen. The town was, in the late nineteenth and twentieth centuries, dominated by a religious leader of the Jahrīya suborder of Naqshabandī Sufis, and his justice was supreme in the valley.²⁴

In sharp contrast, Lanzhou, the provincial capital, contained several Muslim neighborhoods, but they constituted no more than 10 percent of the city’s population. Though the Muslims did prosper in trade, they had no independent sphere of political action. Their communities built some large mosques, but they never functioned as religious centers for Muslims from other parts of the province. Sometimes besieged by Muslims from Hezhou or Xunhua, the Lanzhou Muslims could not or would not rebel in their own right, living as they did under the eyes of the governor and his garrison.²⁵

Muslim political hegemony in Gansu, at least in Qing and Republican times, was associated with the community of Hezhou (Linxia), now the central city of Linxia Hui Autonomous Prefecture, set in the valley of the Daxia River southwest of Lanzhou. Muslims and non-Muslims lived there in equal numbers, perhaps forty-five thousand of each in the early twentieth century.²⁶ The non-Muslims lived in the walled city to the north, the Muslims in the suburbs and rural hinterland (see plate 8). In economic and social power, the Muslims dominated the entire region. Even in the realm of coercive power, they controlled elements of the garrison and had powerful local militias of their own for much of the modern period. The Muslim community also supported many religious institutions, including the tombs of several important Sufi saints. Underscoring the city’s centrality in Gansu Muslim life, most of the Muslim generals who dominated the northwest during the Republican period claimed Hezhou as their “official” home district, though many lived elsewhere. As a market for both Chinese goods and the hill country produce of Tibet, Hezhou prospered in peacetime but suffered horribly from communal conflict in wartime. The populous but vulnerable Muslim suburb of Bafang was sacked and burned a number of times as frontier violence escalated after the Qianlong period (1736–96).

Tianshui, now an important town in the upper Wei valley, had a Muslim community of about five thousand, with eight mosques, among a population of fifty thousand. But the entire hinterland of this county seat was non-Muslim, so the town’s small Muslim community had to maintain a conciliatory attitude, especially since a garrison was usually stationed there. Though the Muslims there belonged to a supposedly militant revivalist organization, the Jahrīya (Ch. Zheherenye or Zhehelinye), headquartered nearby at Zhangjiachuan, they never rose up against their neighbors.²⁷

Near Tianshui and Zhangjiachuan the town of Fujiang (now Gangu), a market town and county seat, had only a small Muslim community. To a European anthropologist working there in the early twentieth century, the place strongly resembled entirely Chinese towns much further east and seemed culturally very far removed from the geographically close Muslim centers. Indeed, in a lengthy ethnography Joseph Dols did not mention any Muslims at all.²⁸ But the Muslims were there and did belong to the Jahrīya. Through their leaders, they participated in the anxiety and planning attendant upon that Sufi order’s tribulations and uprisings.²⁹

With the usual, substantial warnings about the inaccuracy of the late Qing and Republican population statistics, we may plot the modern distribution of Muslims in Gansu and find it very uneven. They gathered in Hezhou, Xunhua, Zhangjiachuan, Xining, Taozhou, Ningxia, and Guyuan/Haicheng. Large sections of the province—between Lanzhou and Ningxia, in the east, and in the south—held only small Muslim communities. The Gansu sections of the Wei valley were completely dominated by non-Muslims, though Muslims lived at Tianshui, Fujiang, Pingliang, and Zhangjiachuan. As Gu Jiegang found to his pain, Gansu roads were notoriously bad, often closed by natural disaster or bandit menace, placing yet another obstacle in the path of Muslim networks of communication.

In sum, the condition of Muslim communities in Gansu and their relations with non-Muslim neighbors depended to a great extent on local factors. For example, after 1873 Muslims were excluded from living within the city walls of most towns in the province, for fear of a repetition of the violence of 1862–73. In Hezhou, Xining, and Ningxia, all of which had large Muslim communities, they continued to live only in the suburbs into the mid-twentieth century. But in Tianshui and Lanzhou Muslims lived in the city with everyone else, so nonthreatening were their small communities. In some districts Muslims and non-Muslims lived in the same village,³⁰ whereas around Hezhou the separation between them seemed quite strict:

On one occasion we made a trip with Muslim muleteers. They were friendly and suggested that we stay with them in their village, so for that night we stopped in a little Muslim world. The children were called I-ssu-mer (Ismael), Fa-ti-mai (Fatima), Er-pu-tu (Abdul). . . . Before darkness fell, pious bearded men said their evening prayers in public places.

We made another trip over the same general route but on this occasion had Chinese [non-Muslim] muleteers. . . . Somehow the landscape seemed to have a much more typically Chinese complexion than before. Again we stayed for the night in the native village of our muleteers. . . . The smell of incense was in the air. . . . pigs grunted in the courtyards, and in the streets, after dark, accompanied by the beating of gongs and the clashing of cymbals, effigies and masks were carried around in celebration.³¹

We must thus pay close attention to the local in our study of Muslims in northwest China, not relying on categorization schemes devised by national institutions, nor on ideologically informed systems of classification ordinarily used by Chinese and Western scholars. Gansu identities were no more rigid or monovalent than others. Locale, family, religious affiliation, occupation, education, and a host of other loyalties affected the choices that Muslims and non-Muslims made along the frontiers of their cultures.

1. A map including only the Muslims classified as Hui by the People’s Republic of China may be found on the flyleaf of Gladney, Muslim Chinese (table of distribution, p. 28). The Hui Muslims are largely Chinese-speaking, which sets them off from Muslims who speak Turkic, Mongolic, or Persian languages, though the government has included some non-Chinese-speaking Muslims among the Hui.

2. For a more comprehensive treatment of the problem of violence in the study of China, see Harrell, “Introduction.”

3. Gu Jiegang, “Kaocha xibei,” 12–16.

4. For Gu Jiegang’s interest in China’s Muslims, especially their internecine conflicts, see Nakada, Kaikai minzoku, 75–77.

5. Fletcher, “A Brief History.”

6. Frolic, Mao’s People, 146.

7. “Northwest China” usually refers to the contemporary provinces of Shaanxi, Gansu, and Qinghai, plus the “autonomous regions” of Ningxia and Xinjiang. Before the late 1920s, however, northeastern Qinghai and the Muslim areas of Ningxia were included in Gansu Province. Since most of the region of concern here lies within pre-1928 Gansu, I shall use that province’s name except when referring specifically to places that lay in Shaanxi or Xinjiang under the Qing.

8. The traditional name for Gansu, Long, is often found in geographical names such as this one, meaning “eastern Long.”

9. The Chinese word for loess, huangtu, literally means “yellow earth,” and its intimate relation to Chinese people’s conceptions of their own history and culture held a central place in the recent, controversial television documentary “River Elegy” (He shang), which compared China’s ancient, yellow, riverine culture unfavorably to the new, blue, expansive, ocean cultures of Euro-America.

10. Current administrative divisions of the People’s Republic of China make Amdo the northeastern part of Qinghai Province and the northwestern part of Sichuan Province, quite distinct from the Xizang Tibetan Autonomous Region. I use the term “Tibet” here in a cultural, linguistic sense, which would demand the inclusion of most of Qinghai, western Sichuan, and western Yunnan (as well as some pieces of other countries) in Tibet. No political claim is intended, for the eastern and northeastern parts of Tibet have not generally been governed from Lhasa but rather have often maintained semi-independence under Manchu, Chinese, or other nominal hegemony.

11. The Xining River is usually called the Huang River, a homophone of the Yellow (Huang ) River. I shall use the less common “Xining” name to avoid confusion.

12. Along the Yellow River the mountains east of Lantaishan still have some of their forest cover, but to the west, past Gaizi and Gandu, they stand bare and eroding, their trees lost to forest fires or, more commonly, to human beings hungry for fuel, housing, and wood for tools.

13. However the elites of China may have envisioned (or still envision) their civilizing project along the frontiers, the daily exigencies of frontier life made mutual adaptation necessary. In realms as close to the heart as language itself, the frontier peoples influenced one another—the Hezhou (Linxia) dialect of Chinese utilizes Turco-Mongolic case endings, possibly from the Mongolic language spoken by some Muslims in Taoxi, just as Salar has absorbed large numbers of Chinese and Tibetan loan words while retaining Turkic grammar and Arabic-Persian religious vocabulary (Dwyer, “Altaic Elements”). A Japanese anthropologist specializing in Chinese frontier studies proposed (just after World War II) that especially rich social investigation of mutual adaptation would be possible where Muslims and non-Muslim Chinese lived in close proximity. To that end, he recommended parallel study of Sino-Muslims living in China proper and non-Muslim Chinese living in Xinjiang (Iwamura Shinobu, “Chūgoku Isuramu shakai,” 12–15).

14. As noted in the Introduction, I have tried to avoid terms such as Han and Hui, which bear the heavy weight of contemporary ethnonymic practice in the People’s Republic of China. For the Sino-Muslims, the ethnonym Tongan or Donggan referred exclusively to the Sinophone Muslims of the northwest, but it has fallen into disuse in written texts, though it is still used by Turkic speakers in Xinjiang. I prefer to use “Muslim” and “non-Muslim,” plus other appropriate ethnic markers, such as language, rather than the minzu terminology that dominates Chinese ethnographic discourse.

15. Mu Shouqi, Gan Ning Qing, 31.28a, describes the secret activism, arrest, and imprisonment of eight young Communist women by the warlord government in 1927. Mu Shouqi’s vast compendium of documentary extracts, quotations, and opinions will be referred to hereafter as GNQSL.

16. Yu Yao, “Wusi yundong,” attempts to demonstrate the significance of the eastern intellectual movement in Gansu but presents no convincing evidence.

17. Schneider, Ku Chieh-kang, 279, 286–93. Gu Jiegang was not alone in this observation. Several years earlier the Japanese observer Sakuma Teijirō (“Shina kaikyōto”), though interested more in Muslim separatism than Chinese unity, concluded that the majority of Sino-Muslims thought of themselves as Chinese and could be dissuaded from that position only by a wide-ranging Pan-Islamic movement emanating from Xinjiang or further west.

18. Salazu jianshi, 9–10.

19. Trippner, “Die Salaren,” 261.

20. As noted in the Introduction, the People’s Republic of China divides its Muslim citizens into ten minzu. According to this category system, the Muslim minzu living in contemporary Gansu, Ningxia, and Qinghai may be described as follows: (1) Hui, living throughout the region, are Muslims who predominantly use Chinese as their native language (though some in Huangzhong speak Tibetan); (2) Dongxiang, concentrated in Hezhou and its eastern district, spoke (and some still speak) a dialect of Mongolian, though many now use Chinese, and call themselves Santa; (3) Salar, living mostly along the two banks of the Yellow River above and below the town of Xunhua, now in Qinghai, speak both Chinese and Salar, a Turkic language, related most closely to Turkmen but now containing a vast number of Chinese and Tibetan loan words; (4) Baoan (or Bonan), a tiny group living in and south of the Salar country, speak a Mongolic dialect similar to that of the Monguor people; and (5) Kazak, a nomadic people dispersed west of Kökenōr and in the Gansu corridor, speak the same Turkic language as the Kazaks of Xinjiang and Kazakstan. The non-Muslim minzu of the region are classified as Han (Chinese), Tibetans, Yugu (or Qara Yugur), Mongols, Manchus, and Monguors. For a conventional discussion of the current conditions of these peoples, as the People’s Republic officially analyzes them, see Gansu Sheng Minzu Shiwu Weiyuanhui and Gansu Sheng Minzu Yanjiusuo (eds.), Gansu shaoshu minzu.

21. Ma Hetian, Gan Qing Zang, 3. The language spoken in this little town and its hinterland represents a vivid syncretism of local cultures. With Islamic religious loan-words from Arabic and Persian, some Turkic as well, and its Chinese-cum-Mongolic grammar, Tangwanghua symbolizes the entire ethnic, cultural, and religious frontier region in which it was created, although it does not appear to contain much Tibetan material (A. Yibulaheimai [Ibrahim], “Gansu zhangnei,” 33–47). The geographer Li Xudan, investigating the natural resources of the Taoxi region, also noted that Tangwangchuan was one of the few religiously and ethnically mixed towns of the region (“Xibei kexue,” 25).

22. Crossley, “Thinking about Ethnicity,” 1–3.

23. An earlier version of this argument may be found in Lipman, “Patchwork Society.”

24. Iwamura Shinobu, “Kanshuku Kaimin,” 144–47.

25. See the population statistics in Botham, “Islam in Kansu,” 377–90. Chongxiu Gaolan xian zhi (1892), 13.12a–12b, gives the 1887 figures as 478,294 “Han” and 24,863 Muslims, in contrast to Botham’s 50,000 Muslims. Though taken from the baojia (mutual security system) records, the gazetteer figures cannot be regarded as any more accurate than the missionary’s. The relative percentage, 5–10 percent Muslim, is more trustworthy than any specific numbers.

26. Iwamura, “Kanshuku Kaimin,” 145, cites the missionary journal Friends of Moslems 8:2, 31, for a 60 : 40 ratio of Muslims to non-Muslims. The Muslim population of Hezhou has been placed as low as 41,000 (Zhang Qiyun, “Taoxi quyu,” 3) and as high as 112,000 (Ma Zikuo, “Musilin zai Linxia,” 15–18). Both Ma Hetian and Nishi Masao record a 50 : 50 ratio of Muslims to non-Muslims, and this is confirmed without any scientific basis by most of the travel accounts. This figure does not, however, include the rural hinterlands, especially the solidly Muslim eastern subprefecture, whose inhabitants may have numbered as many as one-hundred thousand in the early twentieth century (Ma Hetian, Gan Qing Zang, 18; Nishi Masao, “Kanshuku Seikai,” 8–21 and 41).

27. “Gansu Tianshui xian gaikuang” gives no indication that the local Muslims were other than peaceful citizens, though it maligns the Hezhou Muslims as tufei (bandits) for attacking the city in the late 1920s.

28. Dols, “La vie chinoise.” This lengthy article emphasizes repeatedly the “Chinese” quality of the Fujiang community.

29. The interrogation of a Fujiang Muslim in 1781 revealed that the community had raised four-hundred taels of silver to aid Ma Mingxin in fighting the lawsuits brought against his Jahrīya Sufi suborder and during his imprisonment in Lanzhou (Agui and Li Shiyao [Qianlong 46.5.29, June 20, 1781] memorial, in Salazu dang’an shiliao, 85–88).

30. Li Xudan, “Xibei kexue,” 25, notes zones of exclusively Muslim and exclusively non-Muslim settlement, with mixed villages and towns between.

31. Ekvall, Cultural Relations, 19–20. The author explicitly refers to the feelings of “we” and “they” that he heard expressed in both villages and emphasizes the strict segregation between them.