What was life like for the people who lived along the Silk Road?' Absorbed by the movement of people, religions, and trade goods, we rarely pause to consider how the long-distance overland caravan trade affected the communities through which it passed. Traders must have frequented inns, religious sites, markets, and financial institutions, we suppose, but in fact we know surprisingly little about the day-to-day functioning of the oasis economies ringing the Taklamakan Desert. For those interested in understanding the Silk Road trade of the Tang dynasty, the Turfan oasis offers the best case study. The century from 640 (the Tang conquest of the independent Gaochang kingdom) to 755 (the Tang withdrawal from Central Asia following the An Lushan rebellion) marks the apogee of the Tang dynasty's involvement in Central Asia.' Nearly two thousand documents span the period before Tang rule and continue after 640, when Gaochang was renamed Xizhou .

Les Sogdiens en Chine
The residents of Turfan had an unusual custom: they outfitted their dead with paper boots, hats, belts, and shoes. And because paper in the Silk Road oasis was scarce, they recycled government documents, contracts, and other texts to make these funeral vestments. Nearly two thousand documents have been found in 205 tombs at the Astana graveyard. 4 After disassembling the paper shoes and hats, scholars painstakingly deciphered the handwritten documents. The fragmentary nature of these documents means that they can be frustrating: just as one begins to piece together the course of events, a crucial name is missing, and frequently large chunks of text have been cut away to make shoe soles. Even so, the surviving contracts, depositions given in legal disputes, and travel passes offer a fleeting but informative glimpse of the overland trade of the sixth through eighth centuries. These documents make it very clear that, among the various Central Asian merchants active in Turfan, the Sogdians were the most numerous.
To assess the impact of the Silk Road trade on Turfan, this paper will analyze the oasis's residents in concentric rings. At the center stand the small group of Sogdian merchants who worked full-time as traders. Never long in Turfan, they were constantly on the move with their goods and their households from one oasis to the next. Local officials referred to them as sojourning West Asian merchants (literally "Non-Chinese 'Hu' merchants conducting business" xingsheng hushang ", often shortened to xinghui? They must have coined this term in response to local conditions because it occurs neither in the official histories nor in The Tang Code. Surrounding this core of Silk Road merchants were the residents of Turfan who had frequent contact with them and whose livelihood depended on the Silk Road trade. This group included government officials who supervised and taxed the trade, their interpreters, the people who worked at the inns where the travelers stayed, religious practitioners, Many of the Sogdian traders were adherents of Zoroastrianism, named for the legendary prophet Zoroaster, (also spelled Zara-thustra), also called Mazdeism, a label derived from the name of the supreme deity, Ahura Mazda. Finally, in the outermost orbital are perhaps the most interesting, yet least documented, group. These people, who comprised the bulk of the cultivators of Turfan, had occasional contact with the Silk Road traders but earned their living independent of them. When they borrowed money or purchased goods (often animals or slaves) from Valerie Hansen The impact of the Silk Road trade the Silk Road traders, they usually drew up contracts.' Sixth-and seventh-century contracts for even the smallest amount show the influence of Silk Road commercial consciousness: many prices are recorded in silver coins minted by the Sasanian dynasty (reigned 224-651, in modem-day Iran), and contracts charge a ten percent penalty each month."

The Ethnic Composition of Turfan
By the sixth century Turfan's multi-cultural population was a mix of Chinese and non-Chinese peoples who had settled there during the preceding centuries. The Silk Road merchants, those who worked full-time with them, and those who did not, drew their members from both the Chinese and non-Chinese residents of Turfan.
The original inhabitants of Turfan were a semi-nomadic people who did not have an indigenous system of writing. The dynastic history of the Han reports that they "lived in felt-tents, kept moving in pursuit of water and grass for grazing, and had a fair knowledge of farming"." Artifacts found in early, non-Chinese style tombs at Jiaohe resemble those of the nomadic peoples to the west (whom archeologists conversant with Soviet terminology often refer to as Sarmatians)." We do not know what language they spoke or what they called themselves. The Chinese called them the Jushi l or Gushi peoples, and these peoples often adopted the surname Ju when they took Chinese names.
Starting in the third and fourth centuries, waves of Chinese migrants moved to Turfan and absorbed, displaced, or conquered the indigenous peoples. The Chinese historical record contains almost no mention of the oasis's original residents. When the non-Chinese Qu Wi family established the Gaochang Kingdom around 500,11 Turfau's population was largely Chinese. The Gaochang rulers followed Chinese models for almost every measure they imposed on their kingdom. Serving a government whose political structure mirrored the Chinese state, Gaochang officials used Chinese as the language of administration. The capital city at Gaochang was surrounded by walls with named gates on four sides, just like other Chinese cities. Many of the city's residents spoke Chinese at home, their children studied Confucian texts at school (possibly with glosses in local languages ),12 and their king was a devout Buddhist. When the Tang armies conquered Turfan in 640, they took over a kingdom that was so culturally Chinese that its name in the Sogdian language (a language spoken in the Les Sogdiens en Chine region of Samarkand that was distinct from, but closely related to, Middle Persian) was "Chinatown" or Town of the Chinese." After the conquest, the Tang government introduced the equal-field system then in effect throughout the rest of the empire and renamed Turfan Xizhou . Before redistributing the land, Tang officials compiled household registers. Their census listed the population of the kingdom as 8,046 households with 37,738 residents and 4,300 horses in three prefectures, five sub-prefectures, and twenty-two cities." This figure recorded the population of the entire Gaochang kingdom, not just the capital at Gaochang city. Some ten years earlier, the Gaochang king had boasted to the pilgrim Xuanzang that several thousand clerics lived in his kingdom.IS (Unlike the Sui and the Tang dynasties, the Gaochang rulers did not exempt monks and nuns from taxes, a further indication that clerics formed a large sector of the population.) With a population of nearly forty thousand people, Gaochang city and its satellite settlements would have been one of the largest, if not the largest, trading centers on the Silk Road. Interestingly, the Tang census did not record the ethnic identities or native lan-guages of Turfan's mixed population. Nor do the various household registers that survive distinguish between the Chinese and non-Chinese residents of the oasis, an indication that a black-and-white distinction to us was more variegated to the people of the time. 16 Historians of Turfan have devoted considerable energy to identifying the different ethnic groups resident in Turfan in spite of the limited social and cultural information and in spite of the very few pictorial representations of non-Chinese. Of the various tomb figurines from Turfan that have been published, three male figurines and two detached heads are clearly non-Chinese. Two of these grooms , whose hands have a hole for a rope, led ceramic Valerie Hansen The impact ofthe Silk Road trade or mud figurines of camels or horses laden with goods for the next world. 18 Standing 56 em. high, they wear boots, colorful robes, and distinctive headgear, whether a tall pointed hat with a design or a rounded felt cap with the brim rolled up. Their facial features are exaggerated: dark eyebrows, big noses, and heavy beards mark them as non-Chinese. Both were buried in tomb 206 at Astana and date to either 633, when the husband was buried, or more likely to 689, when his wife was. Several of the female dancer figurines from this tomb have arms made of pawn tickets from Changan, a clear indication that they -and possibly these non-Chinese grooms -were also manufactured there. 19 The other Sogdian figures were probably made locally ]. A tomb guardian provides the most extreme example of stereotyping . The craftsman who made him topped his panther's body and cloven hoofs with a noticeably hairy non-Chinese face. We should not assume that these figurines accurately depict the non-Chinese residents of Turfan. Not portraits of real-life Sogdian residents of Turfan, they are rather the creations of craftsmen deploying familiar stereotypes. With the exception of these figurines, the only clue to the ethnic identity of the Turfan population is their Chinese-language names. Tang-dynasty sources refer to people from Sogdiana (the region around Samarkand that straddles modem-day Tajikistan and Uzbekistan) as the nine jeweled clans (zhaowu jiuxing ), even though most lists do not include exactly nine different family names." The most common Sogdian surnames and their place of origin were:"
An Bukhara (modem Bukhara)
Cao Kabudhan, Gubdan (north of the Zerafshan River)
He Kushaniyah (between Samarkand and Bukhara)
Kang Samarkand (modem Samarkand)
Mi Maimurgh (either southeast of the Zerafshan River or
Panjikent)"
Shi Kesh (modem Shahrisabz)
Shi Chach (modem Tashkent)

Les Sogdiens en Chine are Central Asian; the slaves, names look as though they were transliterated into Chinese from another language, quite possibly Sogdian. The list is intriguing: what were all these people doing in a single household? Wu Zhen offers a daring reading: "of course the male and female slaves could have undergone training of various types -like listening and speaking basic Chinese, becoming familiar with Chinese manners and customs, and even learning some types of tasks -at the hands of the musicians and personal retainers. The goal was to increase the sale price of these slaves". 66 It is also possible that the members of the inferior classes were also for sale (contrary to the provisions of The Tang Code) and not there simply to train the slaves. Still the large size of the household strongly supports Wu Zhen's hunch that this was a slave-producing establishment. The few documented pairings of Chinese male owners with young Sogdian girls raise the question how often Sogdian and Chinese families intermarried. The historical record is largely silent on this topic, but Rong Xinjiang has found throughout Tang-dynasty China a total of twenty-one recorded marriages in the seventh century in which one partner was Sogdian, and in eighteen cases, the spouse is also Sogdian. The only exceptions are very high-ranking Sogdian officials who married Chinese wives." He concludes that most Sogdian men took Sogdian wives, and we may surmise that the pairings between Chinese men and Sogdian women were usually between a Chinese male master and a Sogdian female slave. The Outermost Orbital: Those Least Affected by the Silk Road Trade Of the 212 Turfan contracts listed by Yamamoto Tatsuro and Ikeda On, only a handful can be linked to the long-distance exchanges of the Silk Road. Three of these were included with travel passes because Tang law stipulated that a caravan owner had to have documentary proof that he owned the slaves and animals traveling with him. These are not the actual market certificates required by law because they bear no official seals. These contracts expressly identify one of the parties to the contract as a non-Chinese merchant (xingsheng hu, see above discussion). Let us look at the three examples in chronological order.

(1.) In 673 a company commander (duizheng ) bought a camel for fourteen bolts of silk from Kang Wupoyan a non-resident merchant from Samar-kand (Kangzhou )


In many cases people with zhaowu surnames also have first names that clearly have been transliterated from another language. Yoshida Yutaka has begun to work in this promising, but difficult field, and has identified some of the most common Sogdian names in Chinese and their possible Iranian reconstructions. As is evident from Professor Yoshida's chart, Sogdian children were often named for the guardian deity of the day of the month on which they were born, and many of these deities were associated with Mazdean, or Zoroastrian, beliefs. (Scholars writing about the Sogdians in China tend to prefer the term "Mazdean", derived from the name of the supreme deity Ahura Mazda, over the term "Zoroastrian", which they reserve for the less polytheistic form of the religion.) The Sogdian names in this list provide a snapshot of the deities thought most important by the Mazdeans resident in China. They are an eclectic group drawn from Iranian (Mithra and Ram), Mesopotamian (Nanai), and Indian (Buddha) traditions. (The name Jesus probably reflects Manichean -not Christian ~ influence because the prophet Mani acknowledged Jesus as one of his predecessors.Of those people whose identities we can surmise on the basis of their Chinese names, the vast majority hailed from Sogdiana. In this respect Turfan was no different than any other Chinese town, whether on the overland trade routes or in the interior. Every trading town on the Silk Road and in the interior of China had its own resident community of Sogdians. The Core: the Silk Road Traders The first detailed information that survives about the merchants moving along the Silk Road is a group of thirty-seven separate tax receipts recording payments made over the course of a year, probably around 600, at a single tax station outside Turfan." Cut out from shoe soles, the receipts contain gaps and are not continuous. Local officials tallied the taxes they collected every fifteen days, and recorded the number of silver coins they had collected. On nine separate fifteen-day periods, spread out between the first and twelfth months, they collected no coins at all, an indication that the traffic at this particular tax station fluctuated. As Jonathan Karam Skaff has brilliantly explained, 26 the Gaochang kingdom collected a scale-fee tax every time someone bought a good that had to be weighed. For each transaction, tax officials recorded the type of good purchased, its weight, the amount of the tax levied, the date of the transaction, and the names of the buyers and sellers. The merchants paid the tax with silver coins minted by the Sasanian empire. Famed for their purity, these silver coins enjoyed wide use in Turfan. Forty-one of the forty-nine people have Sogdian surnames, clear evidence of Sog-dian dominance of the Silk Road trade in this early period. Of the eight merchants who did not have characteristic Sogdian names, two (both named Bai B) were from the oasis of Kucha, which lay to the west of Turfan on the northern Silk Route, three (two named Di ~,27 one named Ju *) were descendants of the indigenous peoples of the region, one was Turkish (Gongqin Daguan [=Tarqan] , one was Chinese (Ning ), and one (A fIDJ) cannot be identified." The scale-fee receipts reveal that most of the transactions involved five key commodities: spices (9 times), gold (6 times), silver (6 times), silk thread (5 times), and ammonium chloride (6 times). Ammonium chloride was used as a medicine or as a flux, both in the melting of metals and in dyes. The merchants traded other commodities only once that year: brass (toushi ~E), medicine, copper, the spice turmeric, and sugar. Quantities ranged from quite large (800 jin , equal to perhaps 500 kilograms of spice) to equally small (4 liang m, equal to 160 grams of gold), with roughly one-third over 100 jin. These documents offer one major surprise. Not one of the merchants bought silk cloth! Because silk was sold by length, and not by weight, it was not subject to the 'scale fee' tax. The omission of silk provides a useful reminder. The scale-fee tax receipts do not cover sales of animals or slaves, two of the most frequently traded goods on the Silk Road. We should not leap to the conclusion that all Silk Road traders were Sogdian. Other accounts, to be discussed below, reveal the participation of Chinese merchants in the Silk Road trade, and, by the eighth century, more and more people with distinctly Turkic names appear, evidence of increasing Turkic influence in Central Asia in the seventh and eighth centuries. The Silk Road merchants readily formed partnerships with one another, we learn from a fascinating series of affidavits about a trade dispute between a Chinese merchant and his Sogdian partner that occurred around 670.29 Had the Chinese merchant Li Les Sogdiens en Chine Shaojin borrowed 275 bolts of silk and failed to repay his Sogdian partner? Or, was Merchant Li telling the truth when he denied borrowing the silk? The Sogdian partner's death made it even more difficult to determine what had happened.
The case involved five Silk Road merchants, none of them resident in Turfan:
1.
Li Shaojin ~~f1~i (also called Li San *-=) a Chinese merchant resident in Chang'an ijingshi Han /~grp~);

2.
Cao LushaniRokhshan ~t~LLr, a thirty-year old Sogdian merchant, also resident in Chang'an;

3.
Cao Yanyan f!l 3k}Jf, his deceased older brother, a non-resident Sogdian merchant ([J1!]~i!i}1);

4 & 5. Cao Guoyi V*~, Cao Bisuo V~~ (also called Cao Er W=?), two Sogdian, merchants, temporarily resident in Chang'an, where their families lived (hu, kejingshi,youjiakouzai )The geographic range of these merchants operations is impressive. Based in Chang'an, Li Shaojin and Cao Yanyan formed a partnership in Gongyuecheng C3Y3~ (modem-day Almaligh), which lies some 2500 kilometers to the west in the Yili River basin close to modem China's border with Kazakstan." In Gongyuecheng, Merchant Li borrowed 275 bolts of silk from Cao Yanyan, Rokhshan testified, and the two men, who had no common language, agreed to meet in Kucha, some 300 kilometers to the south. At the time of their parting, the Sogdian was leading two camels, four cattle, and one donkey who carried his wares: silk, bowls, saddles, bows and arrows. The Sogdian merchant never arrived at his destination. One witness speculated that he might have died at the hands of Turkish bandits who wanted his cargo of weapons and saddles. Not surprisingly, the Chinese merchant never paid back the 275 bolts of silk he had borrowed from the dead man. Then, sometime between 665 and 673, and probably before the Tibetan incursions of 670, CaoRokhshan brought a complaint beforethe authorities inTurfan on behalf of his deceased brother. His name marks him clearly as a Sogdian; Cao was a surname Chinese assigned to Sogdians who lived north of the Zerafshan River (which runs through Samarkand), and Lushan was the Chinese transcription of the Sogdian name Rokhshan "bright", the masculine equivalent of the English "Roxanne". Of course, this was An Lushan's name, toO.31 .. In his affidavit the Merchant Li denied borrowing anything from his Sogdian partner. But then the court officials confronted him with the testimony of two Sogdian merchants who had witnessed the original loan of the 275 bolts of silk at Gongyuecheng. Although the copy of the contract belonging to the deceased . Sogdian partner had disappeared, and . although the Chinese merchant denied it, the two Sogdian witnesses vouched that the Chinese merchant had indeed borrowed the silk. According to Tang law, their testimony had the same legal standing as a copy of the contract. . Ruling in Rokhshan's favor, the court ordered the Merchant Li to pay back 275 -bolts of silk in addition to interest. We have no way of knowing whether the Chinese partner ever paid the younger brother back, because the documents come to an abrupt stop. Did other Silk Road merchants travel distances as great as those mentioned in this case? Absolutely.
Ruins of Gaochang city. with a Each time a merchant caravan crossed detailed plan of the market zone. through a pass, local officials were required to check the caravan's travel pass, called a guosuo , in order to certify that each member of the caravan, whether human or animal, belonged to person whose name appeared on the pass. Family members could travel together, and several of the documents give the reason for travel as bringing family members to Chang'an." Some of the travelers are classed as zuoren {IFA, a dependent laborer whose status was higher than that of a slave because zuoren could not be bought and sold." Slaves, both male or female, required a market certificate showing that they had been bought legally -unless they were born to a slave already belonging to the family -as did all draft animals like cattle and horses that could be used by the military. In one case the travel document listed the age and color of eight different horses." . Twelve travel passes found at Turfan confirm that the deceased Sogdian merchant's party was indeed typicaL Merchants often traveled with a dozen or so human companions and as many draft animals. The travel passes, unfortunately, do not list their caravans cargo, but Cao Yanyan's load of silk, bowls, saddles, bows and arrows gives some indication of the items a caravan might have carried. The distance from
Les Sogdiens en Chine Almaligh to Chang'an, great as it was, was equalled by other merchant's itineraries: one man with a Chinese name (Tang Yilian ) wanted to travel all the way to Fuzhou, Fujian, from Turfan in 733 in the company of one male slave and two female slaves 35 One guosuo document, also dated 733, records the reasons given by Wang Fengxian ê 0 0Þ ( (0Þ ( (Þ ( ( J0 0 0  for why he twice visited one place on his permitted itinerary: to pursue a debtor. The authorities drew up a new guosuo document for him since merchants were not permitted to diverge from their prearranged itineraries unless they obtained official permission. How big were the caravans traveling on the Silk Road? As with so many other basic questions about the Silk Road trade, the fragmentary information is subject to interpretation. Buddhist legends translated into Chinese in the fourth and fifth centuries often speak of groups of 500 merchants, clearly a stereotype, but possibly an informed one. One or two caravans of several hundred people are mentioned in preTang sources, but caravans going to and from Turfan during the Tang seem to have been smaller, often a dozen people or so. The guosuo documents discussed above indicate that these small parties sometimes traveled together, forming groups of fifty people. Larger groups may have formed when crossing greater obstacles such as the Pamir Mountains or long stretches of desert. Smaller caravans may indicate greater security: because the Tang exercised greater control over the major routes than had its predecessors, merchants felt safe even in smaller convoys." The First Orbital: Those Working Full-time with the Silk Road Merchants As the travel passes show, government officials, usually working for the Section of the People (Hu Cao ) or market supervisors (shi yi m1~) watched over the movement of these caravans, drew up the market certificates required by The Tang Code each time an animal or slave was sold, and heard complaints like that brought by Cao Lushan. These officials employed interpreters when communication was im... possible. And they sent merchants to inns and doctors when they fell ill and even disposed of their corpses in the event of their deaths. Few documents describe the marketplace at Turfan where the Silk Road traders gathered (Xuanzang, who spent most of his time in the king's palace and a monastery, says nothing at all about a market.) The most likely location for the market was in the southwest comer of Gaochang city, near the ruins of the monastery that still stand today . Archeologists who have surveyed the ruins of Gaochang city have tentatively identified two open areas near to the ruins of the large monastery in the southwest comer of the city as plausible candidates. Rows of houses flank: the marketplaces, with the remains of stalls (hang iT) and the walls dividing the fang :tjj sub-districts of the Tang dynasty still visible (no fang gates have been found). Some of the walls are quite close together, suggesting these were originally workshops, which then supplied the merchants who sold similar goods in the same rows in the markets." The sole Sogdian-Ianguage contract found in the Astana graveyard indirectly suggests that a central market existed and that it was supervised by an official of the Gao-chang Kingdom. but the term may simply indicate some kind of Buddhist adherent") named Zhang for 120 silver coins in 639, just one year before the Tang conquest."
In listing those who were bound to recognize the transaction -non-resident merchants, householders, the king, and high officials -the text offers a snapshot of Turfan society . The contract closes with the names of the witnesses as well as the recording official, an administrator of the Gaochang kingdom who oversaw the Sogdian community. The contract does not say so explicitly, but it makes perfect sense that such an official would have an office in or near the market where the Sogdian traders were most active. The officials of the Gaochang kingdom were responsible not just for monitoring private traders, like those described above, but also for hosting envoys from the different regional kingdoms of Central Asia as well as from Central China. One document from Turfan lists the commodities sold to the Gaochang kingdom government officials, which included a form of brass, felt rugs, Persian brocades, and gold. This list does not specify the sellers, names, but it seems most likely that they were the envoys themselves," for these envoys often engaged in private trade at the same time as they conducted official business. As a result, the distinctions among private trade and official tribute missions were blurred. After the Chinese recovered Turpan of 640, Turfan renamed Xizhou, one of three hundred prefectures in Tang China. As such, it could no longer host envoys from different Central Asian states, and the Turfan merchants became subject to the provisions of The Tang Code. All sales of animals and slaves had to be registered on a market certificate, and only market officials had the legal authority to issue such certificates. The market supervisors inspected all market stalls every ten days to make sure that accurate weights were in use and that merchants charged prices within the range stipulated by the government.
Chinese officials depended interpreters to translate for them, but, because they did not always record their names or even mention them, we do not know whether or not interpreters worked full-time for the government." Unusually, one interpreter's name appears on a travel pass, dated 685, issued to five men traveling east together. At the start of their trip, they had been unable to obtain a travel pass because no officials had been available." The beginning of the document is torn, but the appearance of the date, the interpreter's name, and the name of the reporting official (who signs only one character of his name [Heng -7], as was the practice in the Tang bureaucracy) suggest that only a few lines, if that, are missing. The interpreter's name Di Nanipan l1m{~51 is clearly non-Chinese: Di is a standard last name for the descendants of the indigenous Gaoju peoples, and three syllables suggests a non-Chinese first name as well.46 Each of the travelers draws three finger lines below his name (the equivalent of signing one's name with an X in Europe") to show that he was present and was subject to the legal jurisdiction of the officials issuing the pass. And so too does the interpreter, probably after checking the written form of the document to make sure that it matches the oral testimony he translated. At the end of the document, the travelers give their full names and the names of those traveling with them: two are named Kang, one He, and two identify themselves as Tokharians (Tuhuoluo), or residents of Tokharistan, or Bactria in northwest Afgha-nistan, which had been conquered by the Turks in the early seventh century. Tang law required them to give the names of five guarantors, whose places of residence cover a large area: Tingzhou M1\[,r (Beshbalik), Yizhou fjt1\['[ (Rami), Yanqi ~~ (Karashahr), and Xizhou (Turfan itself). Surely these men traded in all of these places. Three of the five guarantors are identified as commoners (baixing B ~), meaning that their names were entered on the household registers of the towns listed and had the same legal status as other free people in the empire. Each man listed the slaves, the dependent laborers (zuoren) , women, and animals in his party. The interpreter Di Nanipan does not appear on these lists because he was not traveling with the caravan but was, instead, working for the Xizhou authorities in Turfan. Like interpreters, innkeepers provided a service crucial to the smooth functioning of the overland trade. Merchants often stored goods they could not carry with them in inns, which served as warehouses. Market officials had close ties to local inn-keepers, as we learn from an inquiry into the unexpected death of a visitor -maybe a merchant -named Laifeng *1( (his surname is missing) in 643. 48 A Sogdian named He Shementuo testified that local officials had ordered him to provide Laifeng's meals and medical care, suggesting that He Shementuo was probably an innkeeper (The family name He denotes those Sogdians from Kushaniyah, north of the Zerafshan River.) Even though the Sogdian He had called a doctor, Laifeng had died while under his care. The case was complicated by Shementuo's failure to fill in the appropriate forms. At the end of his deposition, He gave the name of someone who could corroborate his testimony: a Mr. Kang (the most common Sogdian surname). The case fell under the jurisdiction of the official in charge of the Jieyi W~~ sub-district, apparently the name of the fang where the market was located. The residents of Turfan definitely maintained inns, we learned from a series of depositions, dated 762, about a cart accident." Two eight-years-old children -a girl from the Cao WJ family and a boy from the Shi 9::. family -were playing in front of an inn owned by Zhang Youhe ~):fi1S, when a driver Kang Shifen m~* lost control of his ox-cart. In his deposition Kang called himself a commoner of the Chumi tribe (Chumi buluo), which lived in the northern Tianshan mountains and Southern Jungarian Basin east and west of Beshbalik. His use of the surname Kang is intriguing; it may indicated that some Sogdians had joined the Chumi tribe, or, more likely, it shows that some non-Sogdians used the surname Kang." Further, he explained that he had been hired by a temporary resident (xingke ), Jin Chennu (whose name does not look Chinese either). Kang explained that the cart was not his own, that he had little experience in driving a cart, and that he had unintentionally wounded the two children when he lost control of the cart. The local authorities followed the provisions of The Tang Code to the letter: Kang was ordered to take care of the children for fifty days. If, after the stipulated time, the children recovered, Kang could go free. If not, then he would receive the punishment appropriate for a murderer."
What kind of burial would two non-Chinese children living in Turfan have received if they had died after the cart accident? What happened to the body of the merchant who died while in the innkeeper's care? Given the recent spectacular finds of non-Chinese tombs in Taiyuan and Changan, one cannot help wondering how the non-Chinese residents of Turfan disposed of their dead. The appearance of the corpses, the style of the tomb architecture, and the presence of Chinese-language documentation suggests that, with few exceptions, almost everyone buried in the Astana graveyards was Chinese." Several wooden slips have been found at Astana that have the Chinese characters dairen ("substitute person" "in place of a person") as well as something difficult to read in Sogdian script." These slips suggest that the relatives of the deceased hoped to provide the dead with servants in the next world. Only bilingual people would have labelled them in both Chinese and Sogdian. Those commissioning the burial were probably Sogdians who had adopted many Chinese customs including Chinese-style burials. In the Sogdian heartland, in the centuries before the Islamic conquests of the eighth century, Mazdean believers feared that decaying human flesh would contaminate the earth so they buried only clean bones. The traditional Mazdean means of disposing of the dead was to expose corpses, to allow wild animals to eat the flesh from the bones, and then to place the bones in a container (an ossuary) for burial. Kageyama Etsuko has identified four ossuaries found in Xinjiang, two from Turfan." They were found at Toyok (Tuyugou) and their style suggests they date to the late seventh or early eighth centuries, the peak period of Sogdian-Chinese interaction. The strongest textual evidence for Mazdeism is also linked to Toyok. One of the most important Mazdean deities worshipped at Turfan was called the Heaven of Toyok (Dinggu tian ), which Zhang Guangda thinks may be an alternate name for the god of victory, Verethraghna. The names of several Mazdean deities appear in a series of documents from the mid-sixth century listing the dates on which animal sacrifices were made to them.55 The deities worshipped include the supreme deity Ahura Mazda (Dawu Amo *~~rsoJ~), Weshparkar, the god of wind (Fengbo } and tree, rock, and mountain gods. The regularity of the sacrifices suggests that a group of Mazdean priests -working full-time? -lived in Turfan and conducted religious services for the Sogdian residents. The first mention of a Mazdean temple at Turfan dates to a Buddhist colophon from the year 430 or 490.57 The Sogdian experience in other cities indicates that as soon as the Sogdian community reached a certain size -perhaps just a hundred households -it named someone as sabao, who served as both political and religious leader. This flexibility allowed Sogdians to sustain their patterns of worship and to maintain their identity as a group even in the first years they moved to a new site. The existence of a full-time Manichean priesthood at Turfan during the seventh and eighth centuries seems much less likely. The Chinese-language texts from Astana say little about Manichean beliefs among the non-Chinese population, but the four German expeditions to Turfan found several Manichean libraries at the beginning of the twentieth century. The many manuscripts are not dated, but some use archaic liturgical languages like Parthian and Middle Persian." If these manuscripts were created and then stored in Turfan, a Manichean community existed there as early as the seventh and eighth centuries. But in recent years the scholarly consensus has shifted to suggest that Manicheism became firmly established at Turfan only after 803, when the Uighur kingdom took control of the oasis. Carbon-14 dating has been done on the paper of some of the illustrated manuscripts, and produced dates consonant with early tenth to mid-eleventh centuries. The few cave paintings with undeniably Manichean subject matter also date to this late period." In sum, the people who lived in Turfan but worked full-time with Silk Road mer-chants included the officials who regulated the trade, either by issuing market certificates or travel passes, their interpreters, the inn-keepers, and Mazdean priests. " . One of the archeologists who exca-vated the Astana site, Wu Zhen, contends that, although many households along the Silk Road bought individual slaves, as we can see in the earlier documents from Niya, the Turfan documents point to a massive escalation in the volume of the slave trade."
A name register from Emperor Wu's reign (690-705) lists seventy-nine people who had been omitted from earlier registers for two households." The seventy-nine names include 1 musician of inferior status iyueshi ~$), 9 male and female personal retainers, also of inferior status (buqu, kenii fffiHtJ ' ~:y':), and 68 female and male slaves (nubi ~;Z~). Musicians and personal retainers belonged to the lowest legal cate-gory of The Tang Code, the inferior classes (jianmin ), and unlike slaves, could not be bought and sold." One of the two households contained 61 people, with twenty individuals under the age of fifteen. The slaves have no surnames, but the personal retainers do, and several
are Central Asian; the slaves, names look as though they were transliterated into Chinese from another language, quite possibly Sogdian. The list is intriguing: what were all these people doing in a single household? Wu Zhen offers a daring reading: "of course the male and female slaves could have undergone training of various types -like listening and speaking basic Chinese, becoming familiar with Chinese manners and customs, and even learning some types of tasks -at the hands of the musicians and personal retainers. The goal was to increase the sale price of these slaves". 66 It is also possible that the members of the inferior classes were also for sale (contrary to the provisions of The Tang Code) and not there simply to train the slaves. Still the large size of the household strongly supports Wu Zhen's hunch that this was a slave-producing establishment. The few documented pairings of Chinese male owners with young Sogdian girls raise the question how often Sogdian and Chinese families intermarried. The historical record is largely silent on this topic, but Rong Xinjiang has found throughout Tang-dynasty China a total of twenty-one recorded marriages in the seventh century in which one partner was Sogdian, and in eighteen cases, the spouse is also Sogdian. The only exceptions are very high-ranking Sogdian officials who married Chinese wives." He concludes that most Sogdian men took Sogdian wives, and we may surmise that the pairings between Chinese men and Sogdian women were usually between a Chinese male master and a Sogdian slave.
The Outermost Orbital: Those Least Affected by the Silk Road Trade
Of the 212 Turfan contracts listed by Yamamoto Tatsuro and Ikeda On, only a handful can be linked to the long-distance exchanges of the Silk Road. Three of these were included with travel passes because Tang law stipulated that a caravan owner had to have documentary proof that he owned the slaves and animals traveling with him. These are not the actual market certificates required by law because they bear no official seals. These contracts expressly identify one of the parties to the contract as a non-Chinese merchant (xingsheng hu, see above discussion). Let us look at the three examples in chronological order.
(1.) In 673 a company commander (duizheng ~.IE) bought a camel for fourteen bolts of silk from Kang Wupoyan mtJ~~g,68 a non-resident merchant from Samar-kand (Kangzhou ~jH).69
(2.) In 731 the Sogdian merchant Mi Lushan sold an eleven-year-old girl to a
Valerie Hansen The impact ofthe Silk Road trade
resident of Changan, Tang Rong f@f~, for forty bolts of silk." Five men served as guarantors, vouching that she was not a free person who been enslaved (The Tang Code banned the enslavement of commoners.) Of the four with household registration in Xizhou (Tang-dynasty Turfan), three had Sogdian last names (Shi, Cao, and Kang) and the one with the surname Luo mwas probably a member of the Tuhuoluo tribe, whose primary home was Tokharistan. The fifth guarantor, also with the surname Kang, was designated a temporary resident tjizhu tf1j:) of Xizhou, an indication that he had not yet become a commoner and that his name was not yet entered on the household registers.
(3.) In 733 a Sogdian commoner resident in Xizhou, Shi Randian 1:~$! bought a horse for eighteen bolts of silk from a Sogdian named Kang. Skaff has pieced together information from different documents to follow Shi Randian's route from Hami to Dunhuang, and concluded that he may have traveled the entire route specified by his travel permit from Guazhou, Gansu, to Kucha." When Shi Randian purchased the horse, three guarantors vouched that the horse was not stolen: one Tuhuoluo non-resident merchant, one non-resident merchant from Bukhara (surnamed An), and one commoner resident in Xizhou, a Sogdian named Shi.72 Although no document says so explicitly, it seems likely that the seller paid the guarantors a small fee to vouch for the legality of the goods being sold, since the guarantors were financially liable should the slave or animal in question turn out to be stolen. The presence of resident and non-resident Sogdian guarantors indicates that Sogdian trade networks included both non-resident merchants (xingsheng hu) as well as those entered on the household registers of Xizhou and other localities as commoners (baixing).
If it were not for the explicit labeling of the seller or guarantors as non-resident merchants, we would have no reason to class these three transactions as part of the long-distance overland trade. After all, they simply record the sale of a single camel, slave, or horse. Evidence of the Silk Road trade is equally difficult to detect in the twenty-one labor contracts surviving from Turfan." Most specify the terms for hiring someone to perform someone else's corvee labor obligations to perform a watch on the beacon towers of the Tang, but two -unfortunately quite fragmentary -seventh-century contracts hire someone to transport lian ~ silk to an unnamed destination." This type of silk, sometimes called "cooked silk" or "degummed silk" was ready to be dyed." Like service on the beacon towers, the delivery of silk was probably a form of corvee labor -not a task performed for the benefit of private merchants active on the Silk Road.
Two interest-free loans, made in 665, from a moneylender who usually exacted high interest hints -perhaps! -at a long-distance transaction." The contract was buried

in the tomb of the moneylender, Zuo Chongxi ii:tj~ (d. 673), whose tomb contains fourteen other intact contracts, one of our best sources for understanding the Turfan economy." One loan, to a military guard named Zhang Haihuan 5&51fi1X, was for forty-eight silver coins; the other, to Bai Huailuo E3 ){f{5~, was for twenty-four silver coins. If the two men failed to pay the money back within ten days, the contract authorizes the moneylender to confiscate Zhang Haihuan's house, possessions, or land as compensation. It seems likely that the three men formed a partnership with Zhang having twice as many shares as Bai (the moneylender's share is not revealed). If the deal was success-fully completed, then the moneylender earned nothing from his partners. But if it was not, they had to pay him back at the prevailing interest rate in Turfan of 10% per month, four points higher than the maximum of six percent each month stipulated by The Tang Code." Since Moneylender Zuo retained his copy of the contract, we can conclude that Zhang Haihuan and Bai Huailuo never paid back the money they had borrowed. The contract includes an extra line holding Zhang Haihuan's mother, a female household head, responsible for his debt. It calls her a 'big woman' idanii *-jJ:J, a term that does not appear in the official histories. Big women appear as the household head on government household registers because they bore responsibility for paying their family's taxes when the male household heads were absent. A survey of seventy census documents from the period of Chinese occupation showed that between 16 and 30 percent of all Turfan households were headed by such women. Several scholars have argued that households concealed the presence of men and gave the name of a woman as head because women paid lower taxes than men. It is equally likely that women managed these households because their husbands were genuinely away from home either to perform military service or to go on business trips." Even though they occurred at the peak of the Silk Road trade, the overwhelming majority of Turfan contracts document the day-to-day transactions of an agricultural community in which people buy, sell, and rent individual animals, slaves, or small plots of agricultural land or orchards. Many loans are for a small amount. Perhaps we should not be surprised. Even in today's age of high finance, most contracts are for the purchase of individual houses or cars. 80 Even so, the Turfan contracts reflect the high degree of commercialization of the Turfan economy in the seventh and eighth centuries. I use the term commercialization to indicate that these transactions all involved money -actually Sasanian silver coins-and were not barter. In addition, the penalty for failure to fulfill the contract was 10% for each month, the same interest rate charged on commercial loans.
In the third month of 668, a season when many who worked the land were short of money, the cultivator Zhang Shanxi ~~l! signed a contract to borrow twenty silver silver coins. This was the same moneylender who lent Zhang Haihuan money for his ten-day business trip." Both men -one a long-distance trader, the other a fanner -borrowed silver coins at the same interest rate of 10% per month. Moneylenders like Zuo Chongxi served as a bridge between the long-distance trade economy and the local agricultural economy, and they charged the same high rate of interest to everyone, effectively pulling the local cultivators into the larger economy. If fanners like Zhang Haihuan wanted to borrow money, they had to do so at the same rate as Silk Road traders did. The Turfan contracts clearly document the shifts in the medium of exchange at Turfan." The earliest Turfan contract testify to the existence of a barter economy: in 273 a female household head (danu) bought a coffin for twenty bolts of degummed lian silk." Similar exchanges continue in the fourth and fifth centuries. The first mention of Sasanian silver coins in a real-life rental contract occurs in 584, when someone rented one sixth-acre (mu) for five silver coins." The use of Sasanian coins peaks in the century from 550 to 650, but people continue to use both grain and silk along with silver coins throughout the seventh century, in the years leading up to and following the Tang conquest of 640. Suddenly -just at the tum of the eighth century -Turfan residents stop using Sasanian silver coins and adopt the use of Chinese bronze coins (that the contracts call copper coins and whose main constituent was indeed copper)." Skaff suggests that several factors -the Tibetan occupation of the Tarim basin bet-ween 686 and 692, the Chinese infusion of monetary silk and bronze coins, and the growing Chinese preference for silver in ingot form -may account for the change. 86 Although the reasons for the sudden shift continue to be debated, no one debates the immediacy of the change. All the residents of Turfan -both rich and poor -switched from silver to bronze coins over night, sure evidence of how embedded they are in the larger economy. Here, too, we can see the undocumented role of moneylenders as instrumental in introducing these changes. In the 670s people borrowed silver coins, ten or twenty at a time, but in 703 two different people borrow 320 bronze coins each, the equivalent of ten silver coins." (A tax receipt gives the exchange rate as 32 bronze coins to 1 silver.") And where cultivators had paid their rent in silver coins in the 670s, they paid bronze coins in 703.89

Conclusion
The stunning archeological finds of beautiful silks from Niya and the man with the gold mask from Yingpan (west of Lop Nor) in recent years have reinforced the conventional view that many rich merchants plied their wares along the Silk Road of the first millennium of the Christian Era. But the excavated documents give an entirely different impression. The Kharosthl contracts from Niya'" and the loan contracts from Dunhuang (studied so thoroughly by Eric Trombert") very clearly document the exis-tence of a large subsistence economy in which cultivators bartered for simple goods. Like the Turfan documents, the Niya and Dunhuang documents provide hardly any evidence of the fabled long-distance Silk Road trade in silk, gold, silver, jewels, and pearls. Turfan's economy of the sixth to eighth centuries differed in important ways from Niya's in the third and fourth century and Dunhuang's in the ninth and tenth centuries. It was much more commercialized -even in the subsistence transactions documented in surviving contracts. Eric Trombert has argued that the central government of the Tang played an enormous role in the rise of the Silk Road trade." Rather than see many low-level entrepreneurs crisscrossing the Tarim Basin, he suggests instead that we focus on government finance. When the Tang state decided to send large quantities of silk to central Asia to pay its troops and occupying officials, the Silk Road trade boomed. And when the state withdrew from Central Asia, as it did so suddenly after 755, it ended its massive subsidies (usually in the form of silk) to the region. While the Chinese government continued to buy horses in the northwest at a high price," the Silk Road trade dwindled to a small trickle.
The Silk Road trade in the century of Tang rule had clearly visible spill-over effects on the Turfan economy, which was highly monetized and in which all transactions were subject to high interest rates. But even between 640 and 755, the Golden Age of the Tang on the Silk Road, more people earned their livelihood working the land than did trading on the Silk Road. These cultivators had little to do with the Silk Road trade except when they borrowed silver coins from moneylenders or purchased animals and slaves from long-distance traders. Could it be that the Silk Road trade played a small role in Turfan's overall economy? That is certainly what the limited numbers of surviving documents from Turfan suggest.