The Xiongnu built the first empire of the eastern steppe, and the question of how they did it is the foundational question of the field: everything that came after them, the Göktürk khaganate included, inherited the shape of what they made. The answers on offer divide on a single point. One family of explanations locates the cause on the Chinese frontier, so that the steppe empire is a function of the agrarian one; the other locates it in the steppe, and holds that the frontier explains the timing at most. This article sets out both and what the evidence does to them.
Frontier models
Owen Lattimore’s account assumed a sharp divide — ecological, economic and political — between the pastoralists of the north and the farmers of China, and a specialisation on both sides that left the nomads dependent on goods they could not produce. From that dependency follows the pattern usually called trade or raid: the forceful appropriation of what the steppe economy lacked.[1]
Thomas Barfield built the best-known version on that foundation. In it the steppe polities of Mongolia are shadow empires, secondary formations that arise in answer to Chinese imperial expansion; their stability depends on extorting wealth from China through pillage, tribute, border trade and the re-export of luxuries rather than on taxing the steppe, so that when China is centralised and powerful the nomadic empire is too, and when China collapses into anarchy the unified steppe polity collapses with it.[1] The nomadic world unites, on this reading, when it needs enough force to extract goods from a neighbour that has grown both richer and stronger.
Two objections have been pressed. Michael Drompp, reviewing the formation and disintegration of the first and second Göktürk khaganates and the Uyghur empire, found the picture more complex than the model allows and the system of diplomacy and trade multipolar rather than bipolar.[1] Peter Turchin objected that there is no time lag between the unification of China and the rise of a steppe empire, and that steppe states came into existence in several cases when China was weak and divided.[1]
Turchin’s own model keeps the frontier but lengthens the clock. Over several centuries, he argues, polities on both sides of the frontier grew larger and more complex together: as the smaller Chinese states were swallowed by larger ones, the nomads scaled up in step, and the Xiongnu empire is the product of that coevolution rather than a reaction to a finished Chinese empire.[1] The weakness of the model is one of evidence. What happened on the Chinese side is documented; the parallel development in the steppe is a postulate, and Nicola Di Cosmo holds it impossible to demonstrate on either textual or archaeological grounds.[1]
Where the institutions came from
The decisive test is not the timing but the apparatus. The Xiongnu had an army organised on decimal lines, a government of paired offices, imperial rituals, a centralised conduct of trade and diplomacy, and a language of heaven-given sovereignty. If that apparatus is Chinese in origin, the frontier models have their mechanism; if it is not, they lose it. Di Cosmo takes the three elements that have been offered as evidence of Chinese derivation and argues that each of them fails.
The first is the title. The Han Shu gives the ruler’s style as Chengli Gutu Chanyu and glosses it son of Heaven, which invites comparison with the Chinese Tianzi.[2] But Chengli is tengri, reconstructed as such with little dissent and long before this argument was made, and a borrowing from China could not have produced a native word different from the Chinese one; if the non-Chinese word was already in place, there is no reason to think the concept attached to it was not.[1] Chanyu has been compared with a later Altaic dārγā, which belongs to a body of political terms outside the Chinese tradition altogether.[1]
The second is the division of every office into a left and a right. Chinese titles were paired that way from the Zhou down to the Han, with the left taking precedence, and the arrangement has been derived from the ruler’s facing south so that his left hand lay east, toward the rising sun. The Shiji, however, describes the Xiongnu ritual otherwise: when the Chanyu and his men sit down, the senior officials sit on the left and face north — all of them, not only those styled of the left.[3] The Xiongnu pairing is not ceremonial placement at all but a territorial order, two branches of government administering two halves of the empire, and its closest parallel is not in China. It is the division of the Göktürk realm into an eastern and a western half.[1]
The third is colour symbolism. The Shiji says that when the Han emperor Gaozu was surrounded, the Xiongnu cavalry was white on its western side, grey-spotted on the east, black on the north and bay on the south, which matches the Chinese system of the Five Elements.[3] The association is certainly there; whether it was taken from China is another matter, since colour-and-direction correlations belong to a shamanic substratum that early Chinese civilisation shared with the rest of inner Asia and may reach back to the Bronze Age or earlier.[1]
What the rest of the apparatus points to is the west. The decimal organisation of the army has its closest and perhaps its only parallel in the Achaemenid empire. The oath sworn over wine mixed with the blood of a sacrificed animal and stirred with a jinglu blade is a Scythian rite, as is the cup of wine given for an enemy’s head taken in battle, and the name of the sacrificial dagger itself has been reconstructed as an Iranian word.[1] That runs with what the vocabulary shows: an Eastern Iranian stratum in the Xiongnu titles and dairy terms, set out in the article on the language.[4]
Di Cosmo’s conclusion is that the roots of Xiongnu political culture are not in China, that a Chinese origin has still to be demonstrated, and that with a Chinese influence excluded the coevolutionary hypothesis cannot be sustained. What the evidence points to instead is a continuum between the nomadic world north of China and the Scythian cultures of Siberia and central Asia, in forms of political organisation as well as in material culture.[1]
State or chiefdom
A second argument concerns what to call the thing. Nikolay Kradin sets four criteria for an early state: access to managerial position by some selection outside clan and kin, regular taxation to pay officials, a judicial power separate from the political one, and a class of functionaries running an administration for the whole community. By that measure the Xiongnu polity meets the requirements at an embryonic level at best, and Kradin calls it a super-complex chiefdom.[1]
The case on the other side is that the military and civil apparatus stood apart from the kin hierarchies, that its commanders and functionaries were maintained from a centre headed by a charismatic ruler who conducted the rites of the whole community, and that the judicial power was in fact separate.[1] An empire of that extent must have held a great many middle-ranking posts — commanders of smaller units, local administrators, superintendents of the post, of supplies and of the court — and it cannot simply be assumed that all of them were filled from the three or four royal clans named in the Shiji.[1] The argument about lesser kings and regional leaders in the Xiongnu article bears directly on this point.
The two labels are in any case not exclusive of the third. Whether the polity was an early state or a super-complex chiefdom, it was an empire: it reached far beyond its original territory and its original people and held a variety of lands and peoples under an imperial centre.[1]
What the ground says
Both families of model were built when Xiongnu archaeology meant tombs. The survey work of the last decades has changed the evidence under them. A full-coverage survey of 246 square kilometres in the lower Egiin Gol valley recorded 572 sites, among which those of the Xiongnu period are the most common, and found a settlement hierarchy, long-term occupation and grain production in what the Shiji describes as a land without walled cities, fixed dwellings or agriculture of any kind.[5] In the Khanui valley the domestic sites yield the four herd animals in the proportions of a working pastoral economy, and iron slag at six of the excavated occupation areas — which means that smelting was not confined to the workshops of fortified places like Ivolga but went on at open seasonal camps.[6] The people of Egiin Gol were agropastoralists living in settlements, and that alone is enough to dispose of any model that rests on the Xiongnu economy being exclusively one of animal husbandry.[1]
The radiocarbon dates carry a further point, and it is the one that bears hardest on every model of sudden imperial birth. In Egiin Gol the people using Xiongnu-style pottery lived in the valley for generations before the Xiongnu appear in the Han histories, and went on living there after they had ceased to be worth recording and other names had risen to prominence in the north-east.[5] The name entered and left the Chinese record faster than the population entered or left the valley.
Why it matters here
The Göktürk are the test case on which the frontier models were checked and found wanting, and they are also the heirs of the arrangement whose origin is in dispute. The left and right of the Xiongnu became the eastern and western halves of the first khaganate, the decimal army persisted, and the heaven-given fortune of the ruler, Kut, is the same claim in the same words.[1] If that inheritance came from China, the Göktürk state is a Chinese derivative at one remove. If it came out of the steppe, it is what it appears to be in the inscriptions: an order the Göktürk took over from those who held the Ötüken before them.
References
Sources
Literature
- Nicola Di Cosmo, Ethnogenesis, Coevolution and Political Morphology of the Earliest Steppe Empire: The Xiongnu Question Revisited (2011)
- Bryan K. Miller, Xiongnu “Kings” and the Political Order of the Steppe Empire (2014)
- Joshua Wright; William Honeychurch; Chunag Amartuvshin, The Xiongnu Settlements of Egiin Gol, Mongolia (2009)
- Jean-Luc Houle; Lee G. Broderick, Settlement Patterns and Domestic Economy of the Xiongnu in Khanui Valley, Mongolia (2011)
- Alexander Savelyev; Choongwon Jeong, Early Nomads of the Eastern Steppe and Their Tentative Connections in the West (2020)
- Peter B. Golden, An Introduction to the History of the Turkic Peoples (1992)
- Emre-E. Yavuz, The Gokturk Empire: 2nd Edition (2025)
What links here3
- Institutions and Titles
- Khan (Khagan)
- Historiography
- The language of the Xiongnu · The Turkic ethnogenesis debate
Cite this article
Yavuz, Emre-E. “Making of the first steppe empire.” Turkic Atlas, revised 13 September 2026. https://turkicatlas.business-5d4.workers.dev/historiography/steppe-empire-formation-debate/
Last revised: 13 September 2026