Xiongnu Confederation

The first confederation

Drafted from Cosmo, Ancient China and Its Enemies: The Rise of Nomadic Power in East Asian History (2002)formation

Modu unified the steppe tribes around 209 BC and immediately did what no steppe power had done: forced the Han empire into a treaty of equals, sealed with a princess, silk payments and a frontier settlement. The payments continued for sixty years.

Di Cosmo's argument is that this is not a story about nomads raiding for what they lacked. The Xiongnu state was built by a confederation that needed the tribute to hold itself together — the chanyu redistributed Chinese silk and grain to keep subordinate chiefs loyal — and the Han empire's expansion was as much a cause of steppe centralisation as a response to it. The two states made each other.

That relationship is the template, and every steppe power for the next seventeen centuries worked some version of it — the Türks, the Uyghurs, the Khitan, the Jurchen, the Mongols. So is the ending: the confederation split into northern and southern halves at a disputed succession, one half submitted, and the other was finished off by a subordinate people that had been waiting.

The whistling arrow

Drafted from Cosmo, Ancient China and Its Enemies: The Rise of Nomadic Power in East Asian History (2002)expansion

The Xiongnu confederation was assembled by a man who began by murdering his father, and the story Sima Qian tells about how he prepared for it is the most quoted paragraph in the history of the steppe.

The training

Modu was Touman's heir, and Touman wanted a younger son to succeed instead, so he sent Modu as a hostage to the Yuezhi and then attacked them, expecting them to kill him. Modu stole a horse and got home.

Given a command of ten thousand horsemen, he had whistling arrows made, and ordered that whatever he shot at with a whistling arrow, everyone was to shoot at too, and anyone who did not would be executed.

He shot a bird. Some men did not shoot and were killed. He shot his own best horse; some hesitated, and were killed. He shot his own wife; some were too appalled to shoot, and were killed. He shot his father's best horse, and everybody shot.

Then he went hunting with his father and shot him with a whistling arrow, and every man in his guard shot too. He killed his stepmother, his younger brother and the officials who would not obey him, and made himself chanyu.

Sima Qian is writing a century later and this is a formed literary anecdote, not a report. But what it is about is exact: the problem of converting a tribal following into an instrument that will do a thing its members find abhorrent, on a signal, without asking. Every steppe state that worked afterwards solved that problem somehow.

The two demands from the east

The story continues into the founding of the empire, and the second half is better than the first.

The Donghu, the dominant power to the east, tested the new chanyu. They asked for Touman's famous thousand-li horse. Modu's advisers said refuse; he said one does not begrudge a neighbour a horse, and sent it. They asked for one of his consorts. His advisers were furious; he said one does not begrudge a neighbour a woman, and sent her.

Then the Donghu asked for a strip of abandoned wasteland on the frontier that neither side used. This time his advisers said it was worthless, give it to them.

Modu said that land is the foundation of the state, executed everyone who had advised giving it away, mounted, ordered death for anyone who lagged behind, and destroyed the Donghu in a surprise attack.

What was assembled

He then broke the Yuezhi and drove them west — the beginning of a migration that eventually put them in Bactria — subdued the Dingling in the north and the Kirghiz country, and recovered the Ordos loop that Qin garrisons had taken.

By about 200 BC the confederation ran from Manchuria to the Tarim and from Lake Baikal to the Great Wall. Its structure, so far as the Chinese sources allow it to be reconstructed, is a permanent one: a chanyu at the centre, a Wise King of the Left and a Wise King of the Right holding the two wings, and a descending order of commands nominally reckoned in ten thousands, with subordinate peoples attached at every level.

That is a real state apparatus, and it is the first one on the steppe. What it was for is the subject of the next chapter, and the answer is not raiding.

The treaty of equals

Drafted from Cosmo, Ancient China and Its Enemies: The Rise of Nomadic Power in East Asian History (2002)peak

In 200 BC the founding emperor of the Han led an army north against the Xiongnu and was surrounded at Baideng for seven days before being let out. The settlement that followed is the thing this confederation is actually for.

The heqin terms

The agreement — heqin, harmonious kinship — had four parts. A Han princess of the imperial house went to the chanyu as a wife. A fixed annual consignment of silk, grain, wine and other goods went north. The two states were declared brothers of equal standing. And the frontier was fixed at the Great Wall, with markets opened along it.

The payments ran for about sixty years, and were increased repeatedly.

That is a settled empire treating a steppe confederation as a diplomatic equal and paying it an annuity — which is not how the Han described its relations with anybody else, and is why the arrangement was so unpopular at court that it had to be argued about for four generations.

Why the chanyu needed it

The usual explanation for steppe pressure on China is scarcity: nomads lack grain and cloth, so they raid.

Di Cosmo's argument is that the Xiongnu state was built on the tribute rather than driven to it by hunger. The confederation was an assembly of peoples with their own chiefs, and what held it together was distribution. The chanyu who could hand out Chinese silk, grain and manufactured goods to subordinate chiefs, in front of their own followers, was a chanyu who was obeyed; the one who could not was a chief with a title.

That inverts the usual reading. The raiding is not an alternative to the tribute — it is what happens when the tribute is interrupted, because the centre has to find something to distribute or come apart. And a settled empire that pays is not being extorted so much as underwriting a neighbour's political system, which is precisely what the court critics who wanted to stop paying kept saying.

It also means the two states made each other. Han expansion under Wudi was as much a cause of steppe centralisation as a response to it: a large, aggressive, unified neighbour is exactly the condition under which scattered pastoral peoples benefit from combining, and the confederation grew in response to pressure as much as it applied it.

What the archaeology adds

The Chinese portrait is of a people with nothing, taking. The ground shows something else.

Excavation in Mongolia and southern Siberia has produced substantial settlements with rammed-earth walls, agriculture including millet cultivation, iron and bronze working on a large scale, and elite burials at Noin-Ula and elsewhere containing imported Chinese lacquer and silk alongside local goldwork of very high quality and textiles from as far away as Bactria and the Greek world.

That is a stratified society with its own production and its own long-distance trade, not a horde living off a frontier. The silk in the burials came from China; the carpets did not.

What was actually being sold

Read together, the treaty and the graves describe a transaction rather than a tribute.

The Han was buying a frontier. What it got for its silk was sixty years in which the northern border was a market rather than a war, at a cost the court could calculate and complained about but kept paying, because the alternative — which Wudi eventually chose — was campaigns into the Gobi that emptied the treasury, lost tens of thousands of horses, and did not settle anything either.

The split

Drafted from Cosmo, Ancient China and Its Enemies: The Rise of Nomadic Power in East Asian History (2002)end

The confederation was not destroyed. It divided, and then each half was absorbed by something else, which is how most steppe empires end and is much harder to date than a battle.

What Wudi's wars did

From about 133 BC the Han abandoned the payments and went over to campaigning. The wars ran on and off for decades and they were enormously expensive: armies of a hundred thousand and more sent into the Gobi, horse losses that ruined the imperial studs, and a fiscal crisis that produced state monopolies on salt and iron to pay for it.

They did not conquer the steppe and were not designed to. What they did achieve was strategic: the Han took the Hexi corridor, which separated the Xiongnu from the Qiang to their south-west and opened the road to the Tarim, and then spent a century contesting the Tarim oases — because the oases were the Xiongnu's other source of the goods a chanyu had to distribute.

That is the mechanism by which a steppe confederation is actually broken. Not by defeating its army, which can ride away, but by cutting off the things it needs to pay its own chiefs with.

60 to 36 BC

The strain came out where such strain always comes out, at the succession.

In the 50s BC a disputed succession produced rival chanyus — at one point five at once — and the confederation split. The southern claimant, Huhanye, submitted to the Han in 51 BC, went to court, and was settled on the frontier as a subordinate; his brother Zhizhi went west and was killed by a Han expedition in 36 BC at his fortified base on the Talas.

Huhanye's submission is worth being precise about. He did not become a Han subject population. He received a title, a subsidy and a frontier position, and his people remained a distinct polity — which is to say the Han had finally found a way of paying that came with a hierarchy attached, and the price of the subsidy was now a formal acknowledgement of inferiority.

AD 48

The final division is a repeat, a century later, of the same fracture.

Another succession dispute split the confederation into a Southern Xiongnu, which submitted and was settled inside the frontier as a federate people, and a Northern Xiongnu, which held the Mongolian steppe.

The northern half was then broken between about 89 and 91 by a coalition of Han armies and the Xianbei, a rising power on their eastern flank — the ordinary fate of a weakened steppe centre, finished off by a subordinate people that had been waiting.

Where the northern remnant went is the great open question of the field. A population moving west out of Mongolia in the second century and the appearance of the Huns on the Volga in the fourth is a tempting line to draw, and the evidence for drawing it is thin and much argued over. It should be treated as unresolved.

What became of the southern half

They stayed inside the frontier for two hundred and fifty years as a federate military population, supplying cavalry, and gradually became one of the several peoples of the northern Chinese borderland.

In AD 304 one of their leaders, Liu Yuan — descended from the chanyus, educated in the Chinese classics, and claiming the Han imperial surname on the strength of the marriage alliances his ancestors had made — declared himself king of Han and began the destruction of the Jin dynasty.

That is the last act of the Xiongnu name in China, and it is a precise inversion of Modu: not a steppe confederation extracting tribute from an empire, but a man of Xiongnu descent claiming the empire itself, in its own vocabulary, on the strength of a princess sent north four hundred years earlier.

Everything is Sima Qian's

Drafted from Cosmo, Ancient China and Its Enemies: The Rise of Nomadic Power in East Asian History (2002)aside?What aside means

The Xiongnu left no writing. Everything narrative about them comes from Chinese histories, principally Sima Qian's Shiji, composed by a court historian of the empire they were extracting tribute from, roughly a century after Modu.

Di Cosmo's argument is that the resulting portrait — nomads as a natural, unchanging predatory force, driven by scarcity to raid the settled world — is a Chinese political construction rather than an observation, and that it was serving an argument at court about whether to appease the steppe or campaign against it. The archaeology gives something less convenient: substantial settlements, agriculture, metalworking, and burials indicating a stratified society with considerable wealth that was not all taken from China.

This matters beyond the Xiongnu, because the portrait was applied to every steppe people afterwards by the literate societies they bordered — in the Chinese histories of the Türks and the Khitan, in the Persian and Arabic writing on the Turks, and in the European accounts of the Mongols, which reach for the same vocabulary with no knowledge of the Chinese original. The steppe is almost entirely documented by its enemies, and the enemies were mostly arguing with each other about frontier policy.

The same caution applies to the figures. The reach commonly given for this confederation — nine million square kilometres — is one of the largest anybody attaches to a pre-modern state, and it describes a polity whose internal structure is inferred rather than recorded. It is a published estimate rather than an invention, and it should be read as an area over which tribute was plausibly collected rather than a territory anyone administered.

The pattern

Drafted from Cosmo, Ancient China and Its Enemies: The Rise of Nomadic Power in East Asian History (2002)afterlife

The Xiongnu are the first steppe empire that can be described in detail, and what they left is not a state, a script or a religion. They left a shape that recurred for seventeen centuries.

The shape

A pastoral confederation forms on the Mongolian steppe under a single supreme title. Its cohesion depends on the ruler's ability to distribute goods he does not produce, so it needs access to a settled economy — by treaty, by trade, by tribute or by force, and usually by all four in rotation. The settled empire alternates between paying and campaigning, and finds that both are expensive. The confederation eventually splits at a succession, one part submits and is settled on the frontier, and a subordinate people that had been waiting takes the centre.

Every element of that is visible in the Xiongnu between 209 BC and AD 91.

Then: the Xianbei, the Rouran, the Türks, the Uyghurs, the Khitan, the Jurchen, the Mongols. Not one of them is a repetition — the Uyghurs built a city, the Khitan ran two administrations at once, the Mongols took the settled world instead of taxing it — but every one of them is recognisably working the same problem, and several of them said so.

The Orkhon inscriptions, carved eight centuries later, are a Türk qaghan telling his people that Chinese generosity is a weapon and that a steppe empire that settles near the market dies of it. That is the Xiongnu problem, stated from the inside, by a man whose own state was living on the same silk.

What it cost the Xiongnu to be first

Being the template has a price, and the Xiongnu paid it: they became the definition rather than a case.

Because Sima Qian wrote them up first and wrote them up well, his portrait — nomads as a permanent, unchanging predatory force, driven by scarcity toward the settled world — became the frame through which every later steppe people was described by everyone who bordered them. It is in the Chinese histories of the Türks and the Khitan. It is in the Persian and Arabic writing on the Turks. It is in the European accounts of the Mongols, which reach for the same vocabulary without any knowledge of the Chinese original.

The frame was a political argument at the Han court about frontier policy before it was an ethnography, and it has been in service ever since.

The other name

There is one more afterlife and it is a philological argument rather than a fact.

Xiongnu in the Chinese sources, Hun in the Latin ones, and the two are similar enough that they have been identified with each other since the eighteenth century. Sogdian letters found near Dunhuang, written about 313, use a word for the people who had just sacked Luoyang that is the same word later used for the Huns, which is the strongest single piece of evidence for a connection.

Whether that is descent, a migration, or a prestigious name picked up by unrelated people on the western steppe is not settled and may not be settleable.

It is a fitting last item. A confederation that left no writing of its own is known through its enemies, has given its portrait to every steppe people since, and may or may not have given its name to the one that finished the Roman frontier — and the best evidence either way is a merchant's letter, in a third language, complaining about a sacked city.