History of Xinjiang: Ancient Encounters (Earliest Times-8th Century)

Eurasian Crossroads: A History of Xinjiang. Editor: James A Millward. New York, NY: Columbia University Press, 2007.

Writers have attached a number of metaphors to Xinjiang. It is the hub of the Silk Road, the crossroads of Asia, the heart of the continent. Owen Lattimore—caravaneer, Inner Asianist and, according to Senator Joseph McCarthy, ‘the top Soviet espionage agent in the United States’—called Xinjiang the ‘pivot of Asia’. Some might question Lattimore’s assessment of the region’s geopolitical significance in the mid-twentieth century. Nevertheless, when viewed broadly, the idea behind all these labels is clear enough: Xinjiang’s centrality and intermediate position in Eurasia. This book is devoted to examining the region’s ‘betweenness’ over a long chronological perspective. A good place to begin is from another sort of long perspective, from a satellite above the earth at the beginning of the twenty-first century.

A glance at a satellite image or topographical map of Eurasia reveals the dense knot of mountains at the continent’s core. These mountains define Xinjiang as we know it: the region is a rough triangle consisting of three basins, the Tarim to the south, Turfan (Turpan) to the south-east, and Zungharian to the north. The Kunlun range and its eastern offshoot, the Altyn Tagh, form the southern boundary of the Tarim Basin, and the Tianshan forms its northern edge, dividing the Tarim Basin from Zungharia. The Altai range divides Zungharia from Mongolia on the north-east. To the south-east, the Quruq Tagh, a spur of the Tianshan, separates the Tarim and Turfan Basins.

Xinjiang thus defined by mountain ranges is a relatively recent geological creation. It is not, for example, the Xinjiang known by Tienshanosaurus or other dinosaur species from the Jurassic and early Cretaceous whose remains have been found there. In fact, what most dramatically distinguishes Tienshanosaurus’ home some 160 million years ago from the Xinjiang we know are precisely those mountain ranges, the high ground at the centre of Asia. The Qinghai-Tibetan plateau, the world’s highest place, and the landform which, along with Antarctica and the Arctic, exercises the most climatic influence on earth, began to rise some 100 million years ago with the subduction of the Indian plate under Eurasia and the breaking away of the Indian continent from Gondwanaland. India sped north and collided at high-speed (up to 15 cm per year) some 60 million years ago with continental Asia, there joining South China and North China, which had themselves only relatively recently met and taken their place at the Eurasian table. The Indian collision lifted the Qiangtang plateau of northern Tibet and raised mountain ranges in a series of folds across the centre of Eurasia.

The uplift continued, by twenty-two to fifteen million years ago reaching a threshold height that created the Asian monsoon weather pattern. In this weather cycle, warm and moist air from the Indian and Pacific oceans is drawn in the spring and summer into the Indian sub-continent and East Asia by rising air masses over Tibet and Mongolia, bringing the monsoon rains. In the autumn and winter, cool high pressure zones centred on the high Inner Asian regions block warmer, moister air from the sea, causing relatively dry winters in Asia. The mountains and the monsoon climate they produced have profoundly affected the history of Asia, influencing everything from agriculture to navigation in the Indian Ocean and South China Sea to, arguably, social and political structures in eastern and southern Asia.

We may thus add another metaphor: if Tibet is the roof of the world, Xinjiang lies in its eaves. This tectonic and orographic history has also shaped Xinjiang’s history, albeit in a fashion diametrically opposed to that of areas lying closer to the sea: The mountains ringing Xinjiang and lower ranges further east create a barrier effect, shielding the Tarim Basin from south-easterly winds and the influence of the monsoons that determine China’s and India’s climate. Moreover, whereas long-term shifts to warmer temperatures have increased precipitation in the rest of China, such climatological changes have increased aridity and desertification in Xinjiang. Conversely, climatic cooling produces aridity in eastern China but increases precipitation and glaciation in Xinjiang’s highlands. The mountains, then, separate the Xinjiang region from the climatic regime of eastern and southern Asia.

In fact, Xinjiang lies at the midpoint of two climatic belts extending across Eurasia’s midriff. A climatic map of Eurasia reveals a series of horizontal bands. Furthest north lies the arctic tundra; below that, a strip of coniferous forests, or taiga, comprising most of Siberia. The next band, a belt of steppe grasslands reaching from northern Mongolia in the east to Hungary in the west, crosses northern Xinjiang. South of that lies the band of deserts running from the Gobi to the shores of the Caspian and contiguous with the Arabian peninsula, north Africa and the Sahara. This band, of course, runs through southern Xinjiang.

The populous societies that have formed the civilisational cores of world history lie around the rim of the continent. Xinjiang lies between them, astride the steppe and desert bands, and roughly equidistant from the population cores of China, India and the Mediterranean basin. I will make a good deal of these geographic facts underlying Xinjiang’s centrality in the narrative that follows, but won’t be the first to capitalise on them: entrepreneurial Chinese geographers have recently determined that the ‘exact’ geographical centre of Eurasia lies in a newly developed theme-park located a convenient distance from the tourist hotels of Urumchi (Ürümchi).

The Geographic Setting

With an area of 1,664,900 square kilometres, Xinjiang is the size of Great Britain, France, Germany and Spain—or Texas, California, Montana and Colorado—combined. If it were a country, it would be the world’s sixteenth largest, just smaller than Libya and just larger than Iran. Xinjiang comprises one sixth of the land area of the PRC (People’s Republic of China), yet in the year 2000 was home to only 1.5 per cent of China’s population. Although official PRC sources claim the region has been part of China since 60 BCE, its population has only in the past century become Chinese-speaking, and the region remains, by Chinese standards, sparsely settled.

The Xinjiang region was not an integrated political unit with its current boundaries until the eighteenth century. Before then control of the territory was usually divided among many local oasis rulers or warring empires; parts of Xinjiang were often ruled together with lands in what are now the Central Asian Republics, and parts by China or Tibet. Nevertheless, there is a geographic, cultural and geopolitical coherence to the region that makes writing its history over the very long term a reasonable thing to attempt.

With its snowy mountains and dry basins Xinjiang is home to both the second highest and the second lowest places on earth. To the northeast the Altai range rises to 4,000 metres, forming the border with Mongolia. The Tianshan range running east-west down the middle of the region has peaks of up to 7,000 metres at its higher western end; the landmark Boghda shan (5,445 metres) is visible on rare clear days from Urumchi. The Kunlun range, to the south, divides Xinjiang from the Tibetan plateau and includes the world’s second-highest peak, Mt Godwin Austen (Qiaogeli, 8,611 metres). In the south-west, routes over the high Pamir plateau skirt peaks of 7,500 metres en route to Kashmir, Pakistan and Afghanistan.

The Zungharian (Jungharian, Dzungarian) Basin, or Zungharia, making up the Xinjiang region’s northern half, is bounded by the Altai and Tianshan ranges. Although parts of this basin are desert, much is prime grassland, and some areas support agriculture, especially the fertile valley of the Yili river in the south-west. The Yili drains west into Lake Balkash in what is today Kazakstan. Northern Xinjiang’s other major river, the Irtysh, likewise flows west, filling Lake Zaysan and continuing on to the north-west to Semey (Semipalatinsk) and on past Omsk into Russian Siberia. In fact, along much of its western edge the Zungharian Basin opens without obstacle into Kazakstan’s Semirech’e (Yettisu, Seven Rivers) region. Zungharian grasslands have through history generally formed a single unit with the watersheds of the Talas, Chu and other Semirech’e rivers; with these rich pastures to keep their herds strong, nomad powers could dominate both southern Xinjiang and the heartland of western Turkestan, a region now divided among Uzbekistan, Tajikistan and Kyrgyzstan.

Several passes through the Tianshan link northern and southern Xinjiang, permitting the passage of merchants and warriors. Towards the east, where the city of Urumchi is located today, a large corridor opens southward into the Turfan Depression, which lies between the main chain of the Tianshan and a south-eastern spur, known as the Quruq (Kuruk) mountains, which divide Turfan and Hami from the rest of southern Xinjiang. Ayding Kol, a salt lake at the bottom of the depression, lies at 154 metres below sea level—only the Dead Sea is lower. Although the desert surrounding Turfan is unforgiving and summer temperatures in the region can approach 50 degrees Celsius, ample water made the Turfan and surrounding oases a rich agricultural area, one which nomadic powers to the north coveted for the produce and tribute they could exact. For centuries farmers in the Turfan area dug and maintained underground canals called karez to tap underground reservoirs filled by mountain run-off and to water their fields with a minimum of evaporation. From Turfan a long desert road leads to the Gansu corridor and thence to the interior provinces of China. Some Chinese dynasties established fortifications along this corridor: Jiayuguan, some 700 kilometres east of Turfan, marks the westernmost extent of the Ming dynasty great wall. Other dynasties, notably the Han and Tang, made the Turfan Depression itself a base of operations.

West of Turfan and the low Quruq Tagh range is the Tarim Basin, also known as Altishahr. This southern half of Xinjiang is dominated by the vast Taklamakan Desert (327,000 sq km), a sand-trap surrounded by mountain ranges that cast an almost total rain shadow from north, west and south. The shifting sand dunes of the Taklamakan can reach 100 metres in height. Encroaching sands have buried towns, and seasonal sandstorms, called qara buran, ‘black winds’, have asphyxiated travellers caught without shelter. But the mountains that starve the Tarim Basin of precipitation also provide the snow and glacial run-off that it lives on. The Khotan, Yarkand, Kashgar, Aksu and many smaller rivers feed into the Tarim River, which drains the west and north of the basin in a clockwise direction before emptying south-east into the desert. Around the Tarim Basin stretch a chain of oases, some spring-fed, others supported by mountain run-off. These oases have provided the grain and other products needed by nomads from the north, as well as tribute to military conquerors. They also facilitated trade: the caravan routes running through the ancient cities along the southern and northern rims of the Tarim Basin comprise the central stretches of the Silk Road. Passes through the Pamir and Karakoram ranges allowed communication between the Tarim Basin and Afghanistan, India and Pakistan as well as with what are now collectively referred to as the former Soviet Central Asian Republics, especially Kyrgyzstan, Kazakstan, Uzbekistan and Tajikistan. Historically, both classical Greco-Roman and Arab sources referred to the heartland of these ‘stans’ as ‘the land beyond the river’, referring to the Oxus or Amu river. For convenience below I use the English version of this term, Transoxiana, to refer to the region between the Amu Darya and Syr Darya (darya means ‘river’), an area that includes the Ferghana Valley as well as the cities of Tashkent, Samarkand and Bukhara.

With only 20-150 mm of annual precipitation, the Tarim Basin today supports agriculture only in the piedmont zone, where extensive irrigation can harness the glacier- and snow-melt from the Kunlun, Pamir and Tianshan mountains. Though run-off levels are currently stable, they have fluctuated in the past, with dramatic effects on human habitation in the southern Tarim. Remains of Niya, Dandan-oilik, Endere and other sites deep in the Taklamakan are evidence of times when greater volumes of water brought the rivers flowing out of the Kunlun well north of their current termini. Climatic changes in the late third and late eighth centuries led to the abandonment of these cities to the desert. The desiccated city of Loulan (Kroraina, Krorän) in the south-east of the Taklamakan, once capital of the Shanshan kingdom and a vigorous trade entrepôt, is further testimony to the vagaries of water in southern Xinjiang. When the Tarim and Kongque River system shifted course to the south-east around 330 CE, its terminal lake, Lop Nor, disappeared, leaving Loulan high and dry. Abandoned by its people, Loulan settled into the shifting sands until rediscovered at the turn of the twentieth century.

Since the eighteenth century Qing and subsequent Chinese authorities have expanded the agricultural capacity of the Tarim Basin through vigorous development of hydraulic works. This process accelerated after 1949. But as the Tarim is an ancient sea and an entirely inland drainage, salinisation and alkalisation plague up to a third of arable land—its telltale white residue looks like frost on the surface of many farms. In desert areas groundwater can be saltier than blood. As humankind exploits ever larger areas of land and volumes of river water, natural vegetation, too, suffers, leading to desertification in many areas abutting the desert. From the 1970s to 2000 upstream uses exhausted the Tarim River, leaving little or no water to flow into the desert. The wilderness of salt flats where the river once emptied are now home only to the last surviving wild camels, who have somehow adapted to drinking briny water and dodged the nuclear tests conducted in the Lop Nor region.

Xinjiang in Prehistory

Stone Age

Although archaeological exploration of one sort or another began in Xinjiang over a century ago, our knowledge of the lives of the region’s earliest human inhabitants remains sketchy. Siberia and other parts of Inner and Central Asia have all yielded evidence of Palaeolithic human cultures, but in Xinjiang only a few sites show possible Palaeolithic cores, flakes and evidence of fire use. There is more evidence of humans in Xinjiang dating from the time of the last glacial maximum some 20,000 to 15,000 years ago, when the southern Tarim Basin was less arid than it is today and more snow in the mountains provided more run-off in the deserts. Pebble choppers and other simply chipped tools found south of Khotan testify to the presence of hunter-gatherers exploiting the abundant animal and plant life in the piedmont region. Other, possibly Mesolithic, sites near Khotan, Hami, Musang and Shanshan show a full complement of arrowheads, blades, scrapers and other well-made stone tools from around 10,000 years ago. Many surface artefacts dating possibly to the Neolithic (roughly 10,000 to 4,000 years ago) have been found in sites scattered throughout northern and southern Xinjiang, where there is also evidence of fixed habitations, coloured pottery and a wide variety of microliths and larger chopped and ground stone implements, including mortars and pestles. In these Neolithic settlements, then, people cultivated food-crops and processed grains, probably to supplement continued hunting and gathering. However, in this period there is not yet any indication of animal husbandry, nor can we ascertain much about possible linkages with better-researched Neolithic cultures in western Central Asia, Mongolia or the Gansu-Qinghai area of China.

Bronze and Iron Ages

Archaeologists have not yet found evidence of indigenous development from Stone Age to Bronze Age cultures in Xinjiang. Nevertheless, work on the second and first millennia BCE in Xinjiang has yielded exciting finds. Excavation has begun on a good many sites in both southern and northern Xinjiang, especially tombs, and many artefacts and human remains have been recovered. Although associations between the findings from different Xinjiang sites or with developments in regions outside Xinjiang remain tentative, Xinjiang archaeologists have in recent years succeeded in separating the archaeology of the region from the periodisations and other assumptions of Chinese archaeology. Silk, lacquer and cowrie shells from China show up in some tombs; likewise jade from Khotan was common in ancient China. Nevertheless, analysis of the archaeological record in Xinjiang demonstrates that despite these early contacts, the main story of the Bronze Age in Zungharia, the Tarim Basin and the Turfan Basin involved migrations of peoples from western Central Asia and Siberia. Bronze metallurgy in Xinjiang, beginning in the early second millennium BCE, likely preceded that of Central China: cut-marks on logs found in an ancient cemetery site in Xinjiang from that time are too deep and well-defined to have been made by a stone tool.

The Bronze Age in Central Eurasia is characterised by the emergence of mobile peoples, who applied bronze metallurgy to all aspects of their lives, and from the third millennium BCE employed wheeled vehicles, first heavy carts and ultimately the light war chariot. Xinjiang’s Bronze Age is attested by sites in Tashkurgan (in the Pamirs), Loulan, Alagou (Turfan county) and in the Altai and Tianshan mountains. These sites, representing different cultures and different burial types, have yielded a variety of woven materials, ceramics, jewellery, ornaments and figurines, as well as farming tools, grains and animal bones. Interestingly, though there are many small bronze objects in these sites (not large bronzes as in China), agricultural tools from this epoch are still made of stone. Bones of sheep, oxen and horses, hides, woven woollens and felts in many sites demonstrate that animal husbandry had by the beginning of the second millennium BCE become an important component of people’s livelihoods, if not yet an independent economic strategy. Xinjiang peoples were in contact with the Andronovo culture which spread across Central Eurasia from the Urals to South Siberia during the Bronze Age. These contacts may have been the conduit by which the chariot was introduced to China by around the thirteenth century BCE.

Iron items begin appearing in Xinjiang by around 1200 BCE; this is earlier than anywhere else within Chinas present borders, but in line with the beginning of the Iron Age further west. Although there are mines and evidence of local smelting in Xinjiang, the advances in metallurgy may be connected with migrations of pastoral nomadic peoples across the Eurasian steppes. The Iron Age across Central Eurasia is associated with the rise of pastoral nomadism—as opposed to sedentary or semi-sedentary herding, and more militarised and socially stratified society on the steppe that was able to intrude upon agricultural areas. This ‘Scythian’ type culture is distinguished by dynamic representations of animals in metalwork and other media. Some of the ornaments found in Xinjiang display the ‘animal style’ found across the Eurasian steppe in this period, such as the bronze belt plaques found in the Xiangbaobao cemetery in Tashkurgan or the silver and gold plaques from Alagou, fashioned into energetically posed tigers and lions. Sites in Zungharia and the Turfan-Hami area in particular show signs of independent nomadic pastoralism, as opposed to stock-breeding supplementary to farming— again, in keeping with general Central Eurasian trends. However, the line between nomadic pastoral and sedentary agrarian societies should not be drawn too sharply, and the prehistorical evidence from Xinjiang shows a relationship characterised by complex interactions of herders and farmers and mixed agricultural and pastoral land use.

Consider the woman buried in Qäwrighul near Loulan around 1800 BCE (more conservative estimates date her to the first millennium BCE) whose well-preserved remains have gained her acclaim as the ‘Beauty of Loulan’. She was wrapped in a shawl of woven sheep’s wool, and the cemetery in which she and others are buried has no agricultural tools, suggesting that the inhabitants of this sandy region were not farmers, but rather herders, hunters and anglers. Yet she was interred with a winnowing basket and grains of wheat, a testimony to the symbolic importance of agriculture to these people and to possible close interactions with agriculturalists. Likewise, in major archaeological sites throughout western Gansu and southern Xinjiang dating from c. 1000 to 400 BCE, stone and metal agricultural implements have been found in association with bronze items in the style usually favoured by nomads.

Given the relatively early stages of archaeological work in Xinjiang, the region’s prehistory remains beset with puzzles. Data from material artefacts, human remains and even historical linguistics must be accounted for. One scholar who has attempted to synthesise this complex picture is Victor Mair, the impresario of Tarim mummy studies in the West—and a man who himself bears a striking resemblance to the best-looking male mummy, ‘Chärchän Man’. Working together with J.P. Mallory, a specialist on the ancient Indo-Europeans, Mair theorises that from the beginning of the Bronze age (c. 2000 BCE), waves of migrants entered the Xinjiang region. The earliest of these, Mair believes, were probably speakers of Tokharian, an Indo-European language attested only in two dialects in the northern Tarim Basin in materials dating from the first millennium CE, but which was probably spoken much earlier in the region (historical linguists believe that Tokharian was among the earliest Indo-European languages to diverge from the main Indo-European linguistic stock then probably located on the Russian steppes north of the Black Sea). Through the second millennium BCE, later migrations brought other peoples from the west and north-west into Xinjiang; they were probably speakers of various Iranian languages (at that time Iranian was common across Eurasia, and was only just becoming the language of Persia). These migrants were mobile peoples, mostly animal herders, and the last waves, corresponding to the Iron Age discussed above, included true nomadic pastoralists.

These or other migrants also brought agrarian technologies and aspects of the ritual culture of sedentarists to the Tarim Basin, possibly from the band of Central Asian agricultural lands from northern Afghanistan to the Aral sea (a region known as Bactria and Margiana). One indication of such an imported origin for the Bronze Age in Xinjiang lies in the fact that both the grain crops (barley and wheat) and domesticated sheep come from the west. Moreover, in some Xinjiang Bronze Age sites archaeologists have found ephedra, a medicinal herb also used in that period in ceremonies in Bactria, India and Iran. Furthermore, as seen above, there are indications that textile and metallurgical technologies attested in Xinjiang also entered from the west.

According to these theories, then, Xinjiang’s Bronze Age culture was not an indigenous development from Neolithic roots, but rather derived from an influx of Indo-European peoples and technologies from the northern steppes and over the Pamirs. By the late first millennium BCE some of these Indo-European-speaking peoples in Xinjiang’s archaeological record can be associated with particular peoples known from historical or numismatic evidence. However, as it is difficult to reconcile the names in Chinese histories with those in Greek and other Western sources, such associations are often speculative, even controversial.

One group of Indo-European speakers that makes an early appearance on the Xinjiang stage is the Saka (Ch. Sai). Saka is more a generic term than a name for a specific state or ethnic group; Saka tribes were part of a cultural continuum of early nomads across Siberia and the Central Eurasian steppe lands from Xinjiang to the Black Sea. Like the Scythians whom Herodotus describes in book four of his History (Saka is an Iranian word equivalent to the Greek Skythos, and many scholars refer to them together as Saka-Scythian), Sakas were Iranian-speaking horse nomads who deployed chariots in battle, sacrificed horses, and buried their dead in barrows or mound tombs called kurgans. Royal burials included rich inventories of metal objects, often decorated in ‘animal style’ motifs—one famous example is the trove that yielded the gold fittings known as ‘Golden Man’, of which a reproduction now stands on a pillar in downtown Almaty. There are Saka sites across Central Eurasia. In Xinjiang sites with Saka artefacts and Europoid remains have been identified dating from c. 650 BCE through the latter half of the first millennium BCE in Tashkurgan (west of Kashgar, in the Pamirs), in Yili, and even near Toqsun, south of the Tianshan. This latter site, dated to between the fifth and second centuries BCE, yielded a statuette of a kneeling warrior wearing a tall conical hat coming to a curved point—headwear associated in Persian sources with the Sakas, and well known in classical Greek and Roman writings as a ‘Phrygian cap’, after another group of steppe invaders who sported it. An identical hat, fashioned from felt, was found in a tomb near Cherchen in the southern Tarim, dating to c. 1000 BCE, and on the heads of three women buried in Subashi in the Turfan Basin in the fourth or third century BCE. Very similar hats would centuries later adorn the heads of Soghdians in Tang depictions of these Central Asian, Iranian-speaking merchants.

Another group of early Xinjiang inhabitants appears in Chinese sources of the second century BCE as the Yuezhi (Yueh-chih). The identity and movements of this group have posed one of the great questions of ancient Central Asian history: who were the Yuezhi originally, and what eventually became of them? Many scholars believe the Yuezhi and the Tokharians may be one and the same people. Mallory and Mair argue that the ancestors of the Yuezhi lived first in the region of the Altai mountains and Yenesei River basin, where they formed what is known as the Afansevo slab-grave culture, before migrating south to Gansu and Xinjiang; A.K. Narain, on the other hand, suggests that the Yuezhi may have been indigenous inhabitants of the region around Dunhuang and the Qilian (Ch’i-lien) Mountains of Gansu long before they enter historical accounts—there is a continuous cultural tradition in this area east of the Chinese central plains from as early as can be ascertained archaeologically. In any case, the earliest of the Tarim mummies—that is, the Bronze and Iron Age inhabitants of southern Xinjiang—may have been the ancestral or proto-Yuezhi.

We know from Chinese records of the Han period that in the second century BCE the Xiongnu (Hsiung-nu), then the ascendant nomadic power in Mongolia, attacked and dispersed the Yuezhi from their homelands. While some Yuezhi moved into Qinghai (Kokonor, Amdo) and others may have trickled into the Tarim, the main branch of their ruling clan migrated first to the upper Yili valley (where they encountered another people, the Wusun, who had themselves displaced or somehow merged with Sakas in the area). When attacked again, the Yuezhi moved to the Amu (Oxus) river, today the border of Afghanistan and Uzbekistan, where they took control of the Hellenic states of Bactria, the legacy of Alexander’s eastern conquests. The Yuezhi divided into five subgroups, each under its own chief (yabghu), but one of these divisions seized control of the others. Chinese sources called this new empire Guishang, though some still used ‘Yuezhi’. Greek sources, on the other hand, refer to this poeple as ‘Tokharoi’. This empire is now known as Kushan.

In the first century CE Kushan exercised military and political influence over the western oases of the Tarim Basin. The exact nature and extent of the Tokharian element in Xinjiang’s early linguistic, demographic and political history has not been determined, but Tokharian dialects were employed in the Tarim andTurfan Basins through the first millennium CE, either because a Tokharian-speaking population was continuously settled in southern Xinjiang from prehistoric times, or because Yuezhi took refuge there at the time of the Xiongnu attack, or because of the later influence of the Kushan empire—or due to some combination of the three possibilities. In any case, most important for our purposes here, the Yuezhi/Tokharians/Kushans demonstrate an early case of a phenomenon displayed over and over in Xinjiang’s history: a nomadic royal house and its followers forging a confederation and establishing imperial control over sedentary populations.

The Oldest Surviving Taklamakanians

Finally, a word about the mummies. In recent years popularisations of Xinjiang archaeology, as well as magazine and television specials, have made much of the ‘discovery’ of ‘European’ mummies in Taklamakan sites. Some, including an eminent geneticist, have gone so far as to call them blonde and blue-eyed.

Though understandable as a means of selling magazines and garnering research funding, the hype is misleading for two reasons. First, European and Chinese scholars have known about the Europoid mummies for a century, so the novelty of the ‘discovery’ is relative. Secondly, these people were not ‘Europeans’! The popular idea that the mummies are European derives mainly from two factors: their appearance (they are not Mongoloid), and the body of evidence that suggests the Yuezhi or some other early group in the region spoke Tokharian, an Indo-European language. But neither of these factors amounts to real evidence that any of the mummies or any of their ancestors had ever been anywhere near what is usually understood as ‘Europe’—that is, Eurasia west of the Bosporus. The original speakers of Indo-European languages probably lived on the steppes north of the Black Sea; from there branches migrated both east to Central Asia, India and Iran, south to Anatolia and west to Europe. Judging from calculations based on differences in different branches of the Indo-European family and rates of linguistic change, the Tokharian and Saka (Iranian) speakers were among the earliest to take their leave of the ur-Indo-European population. If they were indeed Tokharian speakers, therefore, the Yuezhi must have come east very early indeed—well before the inhabitants of Europe themselves spoke Indo-European. As speakers of an Indo-European tongue, ancient Tarimites may well share a common ancestor with later Europeans, and this would explain the linguistic relationship and even the technical similarities that Elizabeth Barber has identified between woven goods in the Taklamakan and some ancient textiles preserved in central European salt mines. But to call the denizens of the ancient Tarim’ Europeans’ is equivalent to calling Americans of British heritage ‘Australians’. Or, to put it another way, if we describe the ancient Tarimites as ‘European’, then we should call Iranians and Indians ‘European’ as well—but we generally don’t. Anyone familiar with the history of the distribution of the Indo-European peoples across Eurasia should not be overly surprised to find brown-haired, high-nosed mummies—or living Europoid people—in Xinjiang. But if one remains impressed by the apparent racial incongruity of finding these mummies in what is now part of China, then one should simply call them Europoid or Caucasoid and not confuse the picture by associating them with the western extreme of the Eurasian continent.

Eurocentric sensationalism aside, the mummies do provide valuable evidence about the history of the Xinjiang region. Preliminary mitochondrial DNA data has supported claims of a possible western Eurasian origin for some of the ancient Tarimites. But craniometrical comparisons point to additional complexities, with some earlier mummies displaying similarities to peoples in the Indus valley and later ones showing more affinity to populations in the Oxus (Amu) River valley. Earlier and later mummy groups, moreover, were quite different from one another. Though this complicates the story, it shows that Xinjiang was a Eurasian crossroads with a diverse population already in the second millennium BCE.

The Classical Period

The Han and the Xiongnu

The relationship between nomadic pastoralist and sedentary agrarian societies in Xinjiang took its classic form after a confederation of tribes known as the Xiongnu (Hsiung-nu, possibly related indirectly to the peoples known as Huns in Europe) formed an empire encompassing Mongolia, north-west China and Zungharia. Unlike the Indo-European groups discussed above, most Xiongnu probably spoke an early form of Turkic. The Records of the Grand Historian (Shiji) by Sima Qian (Ssuma Ch’ien), China’s ‘father of history’, contains a long ethnographic description of the Xiongnu as pure nomads ‘chasing grass and water’, a phrase that became the stock description of northern nomads in later Chinese sources. Sima Qian tells us of a people all but born in the saddle: children first learned to ride on sheep, shooting bows and arrows at rats and birds before graduating to larger mounts and game. But archaeologists have discovered Xiongnu houses and fortifications and evidence of agriculture along river valleys from Manchuria to Zungharia. The picture of pure nomads living entirely on the hoof is much overdrawn.

From c. 500 BC, as we know from both Han Dynasty historical sources and frozen tombs in the Altai mountain site of Pazyryk, the small city-states of the Turfan and Taklamakan areas had traded with and paid tribute to nomad overlords based north of the Tianshan—Xiongnu or their predecessors. Dominance by Xiongnu over south-eastern Xinjiang was politically institutionalised in the first half of the second century BCE, when Xiongnu defeated the Yuezhi and drove them from the Gansu corridor (175 BCE), and then attacked them in what is now northern Xinjiang (162 BCE). By 130 BCE the Yuezhi had made their migration south-west to the Amu river, yielding the rich valleys of the Yili and Chu rivers back to the Wusun. The Xiongnu established a commandery south of the Tianshan, in the region of Lake Baghrash (Bohu or Bositeng), to control the Tarim Basin. From this office (referred to in the Chinese sources as ‘Commandant of Slaves’) the Xiongnu levied taxes and conscripted labour from Loulan and other Tarim Basin cities.

Meanwhile, further east, the Han dynasty had its own problems with the Xiongnu. From 198 BCE the Han court entered into a pragmatic if somewhat degrading relationship with the powerful northern nomads: known as heqin, it entailed recognising the Xiongnu rulers, or chanyus (sometimes spelled shanyu) as equals to Chinese emperors, letting them marry Han emperors’ daughters, and paying tribute to the nomads to keep them at bay. However, the cost of the tributes kept rising, and both Han generals who had defected to the Xiongnu and Xiongnu generals continued to raid Han territory anyway After some sixty years of dispatching silk, wine, grain and princesses off to the barbarians without securing peace, the Han turned to a more aggressive approach under a new emperor, Wudi. Wudi first attempted to eliminate the Xiongnu ruler and his core force with a trick. The ambush failed dismally and open war soon broke out, in which one Han strategic objective was to ‘cut off the right arm of the Xiongnu’ by driving them from the Turfan and Tarim Basins.

Xiongnu prisoners reported that the Xiongnu chanyu, upon driving the Yuezhi from Gansu, had killed the Yuezhi monarch and fashioned his skull into a drinking cup. Wudi’s court concluded, not unreasonably, that the Yuezhi might be amenable to an alliance with Han against the Xiongnu. Wudi thus decided to dispatch an envoy to discuss the idea with the Yuezhi. Zhang Qian, a former palace attendant, volunteered for the job and left Han in 139 BCE with an escort of over a hundred men, heading across the heart of Xiongnu territory toward the Yuezhi in the Yili valley. The party was promptly captured by the Xiongnu, who held Zhang Qian prisoner at the chanyu’s court north of the Gobi, but gave him a Xiongnu wife with whom he had a son. A decade later Zhang Qian escaped with his wife and some of his men and made his way west. By this time the Yuezhi had already decamped from the Yili and Chu valleys, so Zhang Qian journeyed on to Ferghana (Dayuan), Soghdiana (Kangju) and Bactria (Daxia), where he finally found the Yuezhi on the north bank of the Amu River. By then, of course, the Yuezhi had left the Xiongnu and dreams of vengeance far behind them. Zhang Qian thus returned east via the southern Tarim, and after another year-long Xiongnu detention returned to the Han court, where he became Han’s pre-eminent ‘Western Regions’ specialist. Though he had not secured a military alliance with the Yuezhi (and would not with the Wusun either during a later mission in 116 BCE), the intelligence Zhang Qian gathered on his journeys formed the core of Han geographic, strategic and ethnographic knowledge of lands to the west.

A series of victories against the Xiongnu allowed the Han to penetrate the Gansu corridor as far as Lop Nor by around 120 BCE. For the next sixty years Xiongnu and Han forces engaged in a tug of war over the Tarim Basin. The region’s many small city-states did what they could to weather the geopolitical storm, with much pragmatic shifting of allegiance. For example, the important trade city of Loulan, near Lop Nor, sent hostages to both Han and Xiongnu courts. Jushi, straddling a pass through the easternmost spur of the Tianshan (between modern Turfan and Jimsar jimsa]) was the scene of many Xiongnu-Han battles, until the Han relocated the city’s population wholesale.

The Tarim and Turfan Basin oases, especially Jushi, were essential to the Xiongnu as sources of agricultural products, man-power and tribute revenue. The Han, on the other hand, was fighting for strategic position, not economic gain. This Han campaign is best characterised by the scholars Hulsewé and Loewe:

[From c. 115] … up to 60 BCE the governments of Western Han were ready to take drastic and violent action to secure or promote their interests. We know of two military expeditions designed to force the submission of other peoples and their acceptance of kings favoured by Han (Dayuan and Qiuzi); of five occasions in which Han officials staged or were implicated in plots to murder a local king (Suoju, Yucheng, Wusun, Jibin and Loulan); of one case when the local inhabitants were all put to the sword as a reprisal for resistance (Luntai); of one instance in which a puppet king was set up and the inhabitants displaced from their lands, which were then made over to the Xiongnu (Jushi); and one case in which the authority of a state and control of its population was divided between two local leaders (Wusun).

Both traditional and modern Chinese historiographic propaganda have played down the military character of Chinese penetration of Central Asia. In a curious hybrid of the two, the Nationalist Chinese leader Chiang Kai-shek (Jiang Jieshi) described the above events as ‘the Han Dynasty’s direct contact with the “Western Region”’. The Various stocks’ of China, he asserted, including that ‘east of the Pamirs’ came to be ‘amalgamated’ because of cultural attraction (wangdao —the way of the prince) and not force (badao —the way of the hegemon):’… they have now become integral parts of one nation. In this process, culture and not military might has been the actuating force; the method of assimilation has been by a stretching forth of a helping hand, and not by conquest.’

After 60 BCE the Han stretched its hand further into the Tarim Basin, replacing the Xiongnu ‘Commandant of Slaves’ with its own ‘Protector General’ (duhu). By splitting the nomad confederation into northern and southern factions, moreover, the Han temporarily relieved pressure from the Xiongnu. The Han also began to establish military agricultural colonies, or tuntian, an institution that Chinese regimes right up to the present PRC government would use, with minor variations, to resolve logistical problems and enhance frontier security, especially in Xinjiang.

Tuntian colonies were state farms worked by Han soldiers. They allowed the Han to station loyal troops in places too far to supply from the Chinese interior, without imposing upon local populations for food. Even before 60 BCE, the Han founded tuntian colonies in Luntai and Quli (between modern Kucha and Korla); later, groups of a few hundred soldiers irrigated and farmed lands in the Turfan-Hami region, near Lop Nor, and in the eastern part of the southern Tarim Basin. At the height of Han Central Asian power, in the mid-first century BCE, there was even a tuntian near Issyk Kul in Kyrgyzstan, but during the latter half of the Han period (Eastern Han, 25-220 CE) the dynasty maintained a military farm only in Yiwu, near present-day Hami.

The Han was forced to leave the cities of the Tarim pretty much to themselves during a civil war in the Han heartland (Wang Mang’s usurpation of the imperial throne, 8-25 CE). When the Han dynasty was restored (as the Later or Eastern Han, 25-220) it initially had little to do with the Western Regions, and the oases’ city-states took to warring among themselves for supremacy. The Northern Xiongnu exploited the power vacuum and re-established nomad overlordship over the lands south of the Tianshan. Finally, in the 70s-90s CE renewed Han military offensives and the emergence of other nomad powers in Mongolia drove the Northern Xiongnu to Zungharia, while the Han general Ban Chao re-established military colonies and bullied the Tarim city-states into renewing their vows of allegiance to Han and sending token tributes east to the Han capital. Although his biography notes thousands of Western Regions heads lopped off, Ban Chao is famous for achieving the reconquest of the Tarim more by boldness and guile than by superior force. In one celebrated incident the rebellious King Zhong of Kashgar (Shule) was promising to renew his loyalty to the Han while in fact conspiring with Kucha (Qiuci) and to kill Ban Chao in a surprise attack. Ban Chao pretended to be fooled and invited Zhong and his men to a banquet. When wine had been served, Ban Chao shouted an order and Zhong was bound and beheaded. Han troops then killed 700 of Zhong’s men, and the southern route through the Tarim was pacified.

However, after Ban Chao’s return east in 102 CE the Han again retrenched, once more ceding the Tarim to the Northern Xiongnu from 107 to 125. Ban Chao’s son, Ban Yong, then embarked on a new series of campaigns and the Han re-established a measure of control over the Tarim from 127 to c. 150.

After that the Yuezhi—through their Kushan descendents—reclaimed the stage. From the latter half of the first century CE to the late third century Kushan seems to have enjoyed political sway in the southern Tarim Basin equivalent to that once enjoyed by Han. Administrative documents in the Indic dialect and script used by the Kushan empire (Kharoshthi) have been found in Loulan, Khotan and Niya; there are also bilingual Kharoshthi-Chinese coins.

Han imperial policing of the ‘Western Regions’ had encouraged the exchange of local goods and luxuries from west and east along what would later be named ‘the Silk Road.’ After the Han retreat Kushan, from its position in Bactria astride trade routes linking China, India and Rome, likewise maintained and was enriched by this trade. Kushan also played a part in the introduction of Buddhism to southern Xinjiang and China, as well as in the translation of Buddhist texts into Chinese and other languages.

Assessing the Han and Xiongnu Record in Xinjiang

At this point, since the Han period is officially cited as the beginning of Chinese rule in Xinjiang, it is worth pausing to take stock. Between 162 BCE (when the Xiongnu established their commandery south of the Tianshan) and 150 AD (after which neither Han nor Xiongnu enjoyed any influence in the south) the Xiongnu managed some seventy years of definitive control over the Turfan and Tarim Basin oases together, and the Han some 125 years. The rest of this 310-year-period of engagement with the region may be characterised as partial control and tug-of-war, the waxing and waning of one power or another. There were, to be sure, spectacular temporary Han military successes, such as Li Guangli’s punch through to Ferghana to obtain ‘blood-sweating horses’ (102 BCE) and Ban Chao’s oasis-hopping juggernaut (73-102 CE). The Han partly settled or at least garrisoned the Tarim Basin with military colonies; but the Xiongnu had maintained an administrative centre in southern Xinjiang continuously for a century (longer than the Han Protectorate General that displaced it). Moreover, the Han never had a foothold in Zungharia (northern Xinjiang), which the Xiongnu and the Wusun dominated for this whole period. The impression that all Xinjiang was Chinese territory throughout the Han Dynasty is a distortion arising from later historians’ emphasis on certain aspects of this mixed record. In this case, the historians have proven more powerful than armies.

More important for us, however, are the two dynamics underlined by the Han-Xiongnu struggle: first, a nomad power in Mongolia and Zungharia exploited the Tarim and Turfan Basin oases for foodstuffs and revenue. Second, in its war against the northerners, a China-based power campaigned west to Xinjiang in order to undermine its nomad adversary’s resource base and thus reduce its ability to harass north China. Han expansion into the ‘Western Regions’, in other words, arose from its long rivalry with the Xiongnu and was motivated by security concerns, not desire to secure trade routes or new land. This is a pattern we will see again.

Measured Autonomy and Continued Connections, 3rd-6th Centuries

The situation in Xinjiang during the three hundred years following the decline of the Xiongnu and fall of the Han is poorly documented historically. Due to political and military chaos in China, north-China-based states managed only limited and sporadic involvement in the Tarim Basin. But although this period is often treated as a kind of dark age in China, trade, diplomatic and religious communications remained vibrant in the Tarim Basin and even continued to link north-western China with India and Soghdiana. Indeed, during these centuries Soghdian commercial networks expanded across the Tarim Basin and Gansu Corridor, and Buddhism developed, nourished largely by these same lines of communication.

Chinese links with the Tarim Basin were attenuated but not cut off after the fall of the Han. The Chinese annals register occasional diplomatic visits by emissaries both from Xinjiang city-states and further west. Twice in this period (324 and 382 CE), rulers in the Gansu region dispatched armies to subdue Karashahr (Qarashahr, Yanqi) and Kucha (Qiuci, Qucha) and awe the other petty principalities of the Turfan and Tarim Basins into pledging allegiance and sending tribute. The general in charge of the second invasion, Lü Guang, needed 20,000 camels to bring his plunder back when he returned east two years later. Gaochang (Qocho, Karakhoja), near Turfan, while enjoying political autonomy, retained Chinese language for official documentary use. In parts of the Tarim Basin further west and south, the archaeological record suggests the rising influence of the Kushan empire, which was based in Bactria (present-day northern Afghanistan and southern Uzbekistan), but there remained some continued Chinese presence or at least Chinese-style agricultural settlements at Niya, and Chinese official documents found in the Lop Nor region dating to the Wei (220-65) and especially Western Jin (265-316) testify to the presence of Chinese outposts from these courts in the Shanshan kingdom, of which Loulan (Kroraina) was the capital. Chinese merchants dealt in silk and precious stones throughout the Shanshan kingdom’s five main provinces in the south-eastern Tarim and Lop Nor area.

When the travelling monk Faxian (Fa-hsien) passed through Shanshan in 399, however, he noted no Chinese there. Moreover, he described the landscape he had crossed on his way from Dunhuang in the following famous terms:

Amidst these rivers of sand are evil spirits and hot winds—all who encounter them perish. No birds fly above; no beasts walk below. One gazes all around as far as the eye can see, hoping to find a place to cross, but can chose none. The bleached bones of the dead provide the only sign.

Yet Loulan could hardly have become a thriving entrepôt between Dunhuang and Khotan and a Silk Road hub if the route had always been so treacherous. In fact, conditions had worsened since the first century. From roughly 50 CE the climate grew increasingly warm and dry, resulting in less precipitation in the mountains and therefore less run-off. Several cities on the southern rim of the Tarim Basin, including Karadong and ancient Niya, were abandoned around the end of the third century, probably because the rivers flowing northward out of the Kunlun mountains, on which these cities relied, penetrated less far into the desert. As noted above, the Tarim and Kongque system changed course in the early fourth century, and the ancient Lop Nor, which these rivers fed and which supported the city of Loulan,‘moved’ with them. During the fourth or early fifth century Loulan’s inhabitants abandoned the site entirely. While other cities in the Shanshan area survived and new ones were founded, from this time on the main east-west trade route began to shift to Kucha and Aqsu along the northern rim of the Tarim Basin. The Chinese presence in the south-eastern Tarim Basin in the post-Han period apparently did not survive climatic and hydrological changes and the geographic realignment of trade routes which resulted.

Even before these developments, and despite the Chinese garrison in Loulan and occasional Chinese intervention on the northern Tarim route, the larger principalities of the Tarim and the Shanshan area were in the third and fourth centuries mostly governed independently by local rulers. The Prakrit-language documents written in Kharoshthi script found in Loulan and Cadh’ota, a site in the desert north of Minfeng, indicate a feudal, agrarian kingdom carrying on its business with little reference to either Kushan or China except in matters of luxury trade. An élite of officials presided, while serfs and slaves raised sheep and goats and worked the land, growing primarily grain (wheat, millet and barley) and grapes (for wine). Women could own property, but were also bought and sold, both as slaves and adoptees, in higher numbers than men. Provincial governors and county officials adjudicated disputes, applying a sophisticated legal code, and collected taxes. Among the most burdensome tax levies was the corvée, which consisted mainly of watching camel herds. Buddhism, both Hinayana and Mahayana, thrived in the kingdom, but the monks lived not in monasteries but dispersed among the community, and could own private property, marry and have children.

Further west the kingdom of Khotan displayed its intense devotion to Mahayana Buddhism with pagodas in front of each house, fourteen large monasteries and countless smaller ones. Faxian describes a great procession of the image of the Buddha, with attendant bodhisattvas and devas, in a carriage decorated with precious metals, gemstones and silk pennants and canopies.

When the images are one hundred paces from the city gate, the king takes off his cap of State and puts on new clothes; walking barefoot and holding flowers and incense in his hands, with attendants on each side, he proceeds out of the gate. On meeting the images, he bows his head down to the ground, scatters the flowers and burns the incense. When the images enter the city, the queen and Court ladies who are on top of the gate scatter far and wide all kinds of flowers which flutter down and thus the splendour of decorations is offered up complete.

Kucha was another great Buddhist city in the third and fourth centuries. We know that the Kucheans

[had] a walled city and suburbs. The walls are threefold. Within are Buddhist temples and stupas numbering a thousand. The people are engaged in agriculture and husbandry. The men and women cut their hair and wear it at the neck. The prince’s palace is grand and imposing, glittering like an abode of the gods.

Kucha was one of the most important Buddhist centres in Central Asia, and the birthplace of Kumarajiva (Ch. Jiumoluoshi; 343P-413 CE), a celebrated Buddhist monk born of an Indian father and Kuchean mother. When his mother decided to become a Buddhist nun, the nine-year-old Kumarajiva accompanied her to Kashmir, where he studied several schools of Hinayana Buddhism. A few years later mother and son left Kashmir and returned to Kucha by way of other Central Asian Buddhist monastic centres, including Yarkand, where Kumarajiva encountered and embraced Mahayana Buddhism. By the time they returned home Kumarajiva enjoyed an international reputation as Buddhist scholar and polyglot. He was famous even in north China, and General Lü Guang brought him back east in 384 after conquering Kucha. Kumarajiva then set to work training disciples and translating Buddhist sutras into Chinese. His knowledge of Chinese, Sanskrit and various Buddhist schools made him an apt interpreter of Buddhism into a Chinese idiom, and his work left a deep impact upon Buddhism in China. In the proud words of a modern Indian scholar, ‘Kumarajiva symbolises the spirit of cultural collaboration between Central Asia and India and the joint effort made by the Buddhist scholars of these countries for the dissemination of Indian culture in China.’

Near Kucha were many Buddhist monasteries and sanctuaries, including the rock-cut caves at Qizil (Kizil). Qizil represents the extension to Central Asia of the Indian tradition of excavating and painting caves as sanctuaries; other examples of this temple form include caves at Bamiyan (Afghanistan), Bezeklik (near Turfan), Dunhuang (Gansu), Binglingsi (near Lanzhou) and Longmen (in Henan province). From at least the third century CE until around the time of the Tang conquest in the seventh century, local aristocracy and wealthy merchants sponsored the decoration of caves at Qizil and other sites in the Kucha area with three-dimensional Buddha images, frescos of the life of the Buddha and other scriptural scenes, drawing on diverse influences from Soghdia, Gandhara, India and Iran. The striking green and blue palate of the Qizil frescos, however, is uniquely Kuchean.

During these centuries, and indeed through at least the eighth century, Silk Road commerce was mostly in the hands of Soghdian merchants. There were Soghdian communities in most major cities along a belt running from Soghdia (Transoxiana) across Xinjiang and the Gansu corridor, and, by Tang times, across northern China. The term ‘Silk Road’ is in fact almost a misnomer: silk was only one of many products exchanged, of which the Western imports to China were as important as Chinese exports. And there was not a single but rather multiple routes. ‘The Soghdian network’ would perhaps be a better, if less romantic, term, for these Iranian speaking merchants dominated east-west trade from communities scattered across Semirech’e, Bactria and the Upper Indus Valley as well as Xinjiang and western and northern China. Ultimately they even opened trade links with Byzantium. Soghdian became the Silk Road lingua franca, and not just in commercial settings. In larger communities like Gaochang there were communities of Soghdian farmers and artisans as well as merchants by the early seventh century. Of the thirty-five commercial transactions listed in a tax document from Gaochang in the 620s, twenty-nine involved a Soghdian. After climate changes forced the abandonment of Loulan, it was élites with Soghdian names who established new cities in the Shanshan region. Besides engaging in commerce, Soghdians served as diplomatic envoys, and their caravans often provided passage for pilgrims and monks. Though most Soghdians were Zoroastrians, some adopted Buddhism and joined the cadre of translators rendering Buddhist scripture and commentaries into Chinese and other languages.

The Soghdian network was already in place across the Tarim Basin and Gansu corridor in the early fourth century. We know this because of a remarkable set of letters dating to 313 CE found in a watchtower near the Jade Gate, 90 km west of Dunhuang. The letters, written in Soghdian, discuss chaotic events in China, including the sacking of Luoyang by the Xwn (‘Hun’, i.e. Xiongnu), and also commercial matters. One letter is a business communication from a merchant based in Gansu to his ‘home office’ in Samarkand. Among the trade goods mentioned in the letters are silver, gold, wine, pepper, camphor, Tibetan musk and a white powder, possibly lead used for cosmetics. Two other letters are private correspondence written by a woman to her mother and her husband. All the letters are addressed and wrapped in a standardised fashion, indicating that the Soghdians in Samarkand, Khotan, Loulan and Dunhuang and elsewhere could communicate by means of an established postal system. Soghdians would become still more important players in Xinjiang and north China between the fifth and the eighth centuries.

Assessing the Tang, Turks, and Tibetans in Xinjiang

Once again, because of the influence of Chinese sources on the subsequent telling of history, it is worth reconsidering the Tang record in Xinjiang as a whole. From its first active efforts in the region in the 630s to the An Lushan rebellion Tang enjoyed, all in all, some hundred years of relatively firm sovereignty over the Tarim city states, divided into two episodes by a spell of Tibetan rule, and frequently disturbed by Türk and Tibetan attacks. Over this period the Tang also controlled Zungharia for about twenty years—something the Han Dynasty had never achieved. To be sure, for some of the time the Tang could claim ‘submission’ of Western Turk tribal leaders based north of the Tianshan, and when the chiefs were not actively attacking Tang interests this alliance can be said to have had some military and political substance. But it must be added that except for its easternmost garrisons in the Turfan region, nowhere were large numbers of Tang Chinese settled; to the contrary, Tang rule was indirect, with local elites left in place, and Tang garrisons largely staffed by Turkic soldiers and led by non-Chinese commanders like Arshila She-er and Ko Sonji. The merchants carrying out the trade between China, Xinjiang and parts west were mostly Soghdians, who dominated commerce not only in Xinjiang, but across north China up to the Korean border. Chinese merchants, on the other hand, are notable for their ‘stunning absence’ in Tang sources up to 755.

As noted above, here again the pattern of political control of the Tarim Basin by outside powers is evident. This time Tibet often dominated the southern Tarim Basin, Turkic nomads controlled Zungharia, and China-based powers held sway over the Turfan region and Gansu corridor. As during the Han-Xiongnu conflict, the political and military fate of Xinjiang was in this middle period again linked to the quasi-cyclical pattern by which great powers on the plains of north China and in Mongolia have risen almost simultaneously. Many scholars, including Ellsworth Huntington (1919), Owen Lattimore (1940), Joseph Fletcher (1986), Sechin Jagchid (1989), Thomas Barfield (1989) and Nicola Di Cosmo (1999b) have offered explanations for the often parallel emergence and interaction of powerful sedentary and nomadic states across the ecological divide between Mongolia and north China. Environment, strategy, trade, technological change, internal socio-political dynamics, external pressure and historical memory all have played a role in this process. The fourteenth-century Arab ‘sociologist’ Ibn Khaldun famously analysed the relationship between nomadic and settled people in terms of a waxing and waning cycle of ‘group solidarity’ (‘asabiyya) engendered by the hard conditions of the desert. Peter Perdue has recently provided a useful comparison and analysis of all these theories, and reminded us, on the basis of his study of the Zunghars and Manchus, that strong leaders have also played a large part. For our purposes, it is sufficient to note again that the political history of the Xinjiang region has been repeatedly implicated in this relationship across the Sino-Mongolian borderlands: rivalries with Mongolia-based states led powers in China to expand westward in order to cut off the nomad powers’ other sources of agrarian supply, commercial revenue and tribute.

The Xinjiang region also fell within other circuits of interstate power relations and economic interchange. Though the peaks of the Pamir and Karakoram are awe-inspiring, below them lie plateaus along which traders and armies moved with relative ease. Kushan, and the Hephthalites after them, projected power into the Tarim Basin. The Arabs came close to doing so. Tibet and Tang warred over control of these very passes because of the importance of the routes to Bactria and Transoxiana running through them. Turkic powers based in Zungharia, the Talas River or Issyk Kul area vied to hold both the western Tarim Basin and Ferghana. To the east, Qinghai-based powers including the Tuyuhun, Tibet and, as we will see in Chapter 2, the Tanguts (Xixia), struggled with China over the Gansu corridor, Dunhuang (where the silk route forked north and south) and eastern parts of what is now Xinjiang. The political and military fate of Xinjiang, particularly the Tarim and Turfan Basins, then, was closely linked in this period to struggles along three enduring political and cultural frontiers: the Sino-Mongolian, the Sino-Tibetan and the Pamir/Tianshan belt dividing western from eastern parts of Central Asia. Meanwhile, the Soghdians, who worked with whoever seized power in the Central Asian oases and north China cities, managed commerce and other cultural communication from Ferghana to Manchuria.