Somebody had to carry the goods, whoever was paying for them, and for roughly six hundred years the people who did it across the middle of Asia were the Sogdians.
The previous lesson argued that much of the silk moving into Central Asia entered circulation as Chinese state spending. That says who supplied the currency and says nothing about who did the work. This lesson is about the merchants themselves, and it is the one place in the subject where we can watch a commercial system from the inside, because a bag of their letters was abandoned in a watchtower and never delivered.
A people without a state
Sogdiana was the land between the Oxus and the Jaxartes, the Amu Darya and Syr Darya, centred on the Zarafshan valley in what is now Uzbekistan and Tajikistan. Its cities were Samarkand, Bukhara, Panjikent and Kesh, and it was never once, in its entire history, a unified independent state.
It was a province of the Achaemenid Persians, then taken by Alexander, then held in turn or in part by Seleucids, Kushans, Hephthalites, the Turkic khaganates, and finally the Arabs. Between and beneath those overlords, the Sogdian cities ran themselves as an aristocratic patchwork of walled towns and fortified estates, each with its own ruler, its own coinage, and no common authority.
That absence is not a footnote. A people with a small, fertile, densely irrigated homeland, a position exactly halfway between China, India, Persia and the steppe, and no empire to defend or feed, is a people with every incentive to become intermediaries and no capacity to become anything else. What Sogdiana had to sell was position and organisation.
How we know where they went
The single most useful piece of evidence about the diaspora is a naming convention. Sogdians settling in China took Chinese surnames drawn from their city of origin, in a fixed and well-attested scheme: Kang for Samarkand, An for Bukhara, Shi for Kesh or Tashkent, Mi for Maimargh, He for Kushaniya, Cao for Ishtikhan. Chinese sources call them the Nine Surnames of Zhaowu.
The consequence for a historian is remarkable. Chinese administrative records, tomb inscriptions and census fragments are full of ordinary Chinese names, and a subset of them announce that their bearer's family came from a specific Central Asian city. It becomes possible to map a diaspora through the bureaucracy of the country it settled in, which is not possible for any other trading people of the period.
Example. A Tang register from a garrison town lists a militia unit whose members include several men surnamed Kang, two surnamed An and one surnamed Shi. What can you conclude, and what can you not?
You can conclude that a Sogdian community was settled there rather than passing through, and that it had been there long enough to be enrolled in a Chinese institution and taxed like everyone else. Names in a register are evidence of residence and of legal integration, and militia service in particular means these men were treated as subjects rather than as foreign visitors. What you cannot conclude is that any of them was a merchant. The surname records an ancestral city, not an occupation, and by the eighth century many families of Sogdian origin in China were soldiers, farmers, officials, grooms, musicians or entertainers, several generations from anyone who had traded anything. An Lushan, whose rebellion nearly destroyed the Tang in 755, carried the surname of Bukhara and was a frontier general. The convention tells you about origin and settlement, and inferring a caravan from a surname is exactly the overreach it invites.
Now you. The Chinese bureaucracy created an official post, the sabao, from the Sogdian term for a caravan leader, and made it a ranked position responsible for the resident foreign merchant community. What does the existence of that office tell you?
Answer
It tells you the communities were permanent, large enough to need managing, and already self-organised under leaders the Chinese state found it easier to co-opt than to replace. States do not invent bureaucratic ranks for transient visitors. It also shows the direction of the borrowing: the title is Sogdian in origin, so the state was recognising an existing institution, the head of a caravan and by extension of a merchant community, and slotting it into its own hierarchy with a rank and a salary. That is a standard imperial technique, indirect rule over a diaspora through its own notables, and it has a double edge. It gave the Sogdians legal standing, protection and a channel to officialdom, and it made the whole community legible and therefore vulnerable, which is what it turned out to be after 755, when a rebellion led by a general of Sogdian descent was followed by a violent turn against Sogdians generally.
The mailbag
The five Sogdian Ancient Letters, found in a watchtower west of Dunhuang and dated to 313 or 314, are the earliest substantial evidence of how this trade was organised, and their value comes from the fact that they never arrived.
Letter II is from a merchant named Nanai-vandak, writing east to west to his employer or patron in Samarkand. It is a business report in the middle of a catastrophe. He describes the destruction in China, the emperor's flight from Luoyang because of famine, the burning of the palace, and the deaths by starvation of Indians and Sogdians in the city. Then, without pause, he turns to accounts: what is owed, who holds what, what he has sent, what should be done with a sum of money, and a request for instructions.
That mixture is the point. This is a firm with a principal at home and agents distributed across thousands of kilometres, sending consignments on credit, keeping accounts across years, and expecting written instructions to arrive from a city three months away. It is not adventurers with a camel. It is a business organisation, and one recognisably like the family firms of the medieval Mediterranean.
Letter III is not a business letter, and it is the reason the collection is famous outside the field. Miwnay, a Sogdian woman stranded in Dunhuang with her daughter Shayn, writes to her husband Nanai-dhat, who has left them without support. She has appealed to the local Sogdian community for help and been refused. She would rather be a dog's or a pig's wife, she writes, than his. A separate letter from the daughter to her mother's relatives asks them to intervene.
Both letters were in the same undelivered bag, and both stayed there for sixteen centuries. The system that moved silk across a continent also stranded a woman in a desert town on the far side of it, and the same community network that enforced commercial obligations declined to enforce a domestic one.
The problem a diaspora solves
Why should a trade of this kind be run by one ethnic group at all, rather than by whoever has capital? The answer is a problem in enforcement, and it can be made precise.
A merchant in Samarkand who wants goods sold in Chang'an must put them in the hands of an agent 3,500 km away, in a jurisdiction whose courts he cannot use, whose language he may not speak, and where his agent could simply sell the cargo and disappear. No contract helps, because no authority spans both ends.
Example. An agent holds a cargo worth 100 and earns a commission of 10 a year for continuing to work for the firm. Treat theft as taking the 100 now and losing the commission stream forever. At what discount rate does honesty stop paying, and what does the answer imply?
The value of the honest stream is for a discount rate , so honesty beats theft when , that is when . At a 5 per cent discount rate the stream is worth 200 and the agent stays honest; at 10 per cent it is worth exactly 100 and he is indifferent; at 20 per cent it is worth 50 and stealing is clearly better. Since a merchant in a violent, plague-ridden, politically unstable world discounts the future heavily, and 10 or 20 per cent is if anything optimistic, the pure business relationship does not deter theft. Repeat dealing alone is not enough, which is the whole difficulty of long-distance trade before the modern state, and it means the mechanism that actually held the Sogdian network together must have added something to the agent's payoff calculation.
Now you. What could a diaspora community add that raises the cost of theft enough to close that gap?
Answer
Three things, all of which the Sogdian evidence shows. It can make the punishment collective and portable: a man who steals from a Samarkandi firm is known in every Sogdian colony from Semirechye to Chang'an, cannot trade with any of them, and cannot go home, so the loss is not one stream of commission but every stream available in his profession for life. It can attach the family: agents were commonly kin, and property, marriage prospects and standing at home are hostages the agent cannot take with him. And it can hold real assets in the homeland, so that theft abroad is answered by seizure in Sogdiana, which converts the problem from one no court can touch into one the local court can. Add those and the payoff from stealing 100 falls well below 100 while the honest stream is unaffected, which restores honesty at realistic discount rates. This is the same mechanism Avner Greif reconstructed for the eleventh century Maghribi traders of the Cairo Geniza, and the fact that two unconnected trading diasporas converged on it is evidence that it is the solution the problem forces, not a cultural quirk. Notice the cost of the solution, which is that it only works inside the group, so the network cannot recruit talent or capital from outside and is destroyed by anything that scatters the community.
The language that outlived the traders
The colonies ran in a chain: Sogdiana itself, then Semirechye and the Chu valley, then the Tarim oases, Turfan, Dunhuang, and on into China proper, with substantial communities at Chang'an and Luoyang. Étienne de la Vaissière's Sogdian Traders (2002, English translation 2005) assembled the chain from Chinese, Sogdian, Arabic and archaeological sources, and it is the standard reconstruction.
Along that chain Sogdian became the working language of the road, and this had a consequence no merchant intended. Sogdian was written in an alphabet descended from Aramaic. The Uighurs, whose khaganate dominated the eastern steppe from the eighth century and whose contact with Sogdian merchants and Manichaean missionaries was constant, adopted that alphabet to write Turkic. In 1204 Chinggis Khan had the Uighur script adopted to write Mongol, and in the sixteenth century the Manchus took the Mongol script to write Manchu, in which form it was the writing of the Qing empire into the twentieth century. The chancery script of the largest land empire in history is a direct descendant of the alphabet used by Samarkandi merchants to keep their accounts.
Example. The Ancient Letters were never delivered. What does an undelivered mailbag give a historian that a carefully preserved archive does not?
Freedom from selection. An archive exists because somebody decided its contents were worth keeping, and that somebody was usually an institution keeping records that served it: title deeds, tax registers, official correspondence. Everything else was thrown away, which is why so much of the surviving documentation of the ancient world is administrative and so little of it is ordinary. A mailbag is the opposite: it contains whatever happened to be travelling on one day, chosen by nobody, mixing a merchant's account of a national catastrophe with a woman's letter about her husband abandoning her. That is as close to a random sample of what people actually wrote as this subject ever gets. The catch, and it is a real one, is that a single day's sample from a single route is tiny, so the letters are unusually unbiased and unusually few, and any generalisation from five documents needs the caution the second lesson recommended.
Now you. Almost all the surviving Sogdian documents were found outside Sogdiana. Why, and what does that do to the picture we have of the Sogdians?
Answer
The reason is the second lesson's reason: Sogdiana is an irrigated, fertile, continuously occupied region where organic material rots and later building destroys earlier layers, while the places the Sogdians travelled to include some of the driest ground on earth. The exception proves it, since the one substantial archive from the homeland, the papers of Divashtich, survived because it was carried up a mountain to a fortress at Mugh and left there. The effect on the picture is systematic and worth naming: we see the Sogdians almost entirely as expatriates, as merchants, as a diaspora, because the evidence comes from where they went rather than from where they lived. Their agriculture, their landholding aristocracy, their internal politics and the majority of Sogdians who never left the Zarafshan valley are correspondingly faint, known mainly from the wall paintings of Panjikent and from what archaeology can recover from mud brick. A people who were mostly farmers have come down to us as a nation of traders, which is a distortion produced by climate rather than by history.
The end
The Arab conquest of Sogdiana, campaigned by Qutayba ibn Muslim from 705 and completed over the following decades, ended Sogdian political independence. One of the sharpest documentary windows on it is the archive of Divashtich, the last ruler of Panjikent, found on Mount Mugh in Tajikistan in the 1930s: letters, legal documents and accounts from a man losing a war, ending with his capture and execution around 722.
The commercial network outlasted the homeland by decades rather than centuries. The An Lushan rebellion of 755 made Sogdian identity dangerous in China, and many families concealed their origins. In Central Asia, conversion to Islam and Turkicisation absorbed the Sogdians into the populations that succeeded them, and their commercial role passed to Muslim merchants who used Persian and Arabic. The language itself did not quite die: Yaghnobi, spoken today by a few thousand people in a high valley in Tajikistan, is its direct descendant, the last living remnant of the tongue that once ran a continent's trade.
The Sogdians left one more thing behind, and it was not commercial. Their caravans carried missionaries and monks alongside the cargo, their colonies supported temples, and Sogdians appear in the record as translators of religious texts into Chinese in numbers far out of proportion to their population. The next lesson takes up the largest of those transmissions, Buddhism moving from India to China along the same roads, and asks the question the freight model cannot answer: why should a religion travel with a trade at all?