Aristocratic elites in the Xiongnu empire

NOMADEN UND SESSHAFTE

Sonderforschungsbereich Differenz und IntegrationWechselwirkungen zwischen nomadischen und sesshaften Lebensformen

in Zivilisationen der Alten Welt

Herausgegeben im Auftrag des SFBvon Jörg Gertel, Stefan Leder, Jürgen Paul und Bernhard Streck

BAND 17

WIESBADEN 2013DR. LUDWIG REICHERT VERLAG

Nomad Aristocrats in a World of Empires

Herausgegeben von Jürgen Paul

WIESBADEN 2013DR. LUDWIG REICHERT VERLAG

Bibliografische Information der Deutschen NationalbibliothekDie Deutsche Nationalbibliothek verzeichnet diese Publikation

in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http://dnb.dnb.de abrufbar.

© 2013 Dr. Ludwig Reichert Verlag WiesbadenISBN: 978-3-89500-975-4

www.reichert-verlag.deDas Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt.

Jede Verwertung außerhalb der engen Grenzen des Urhebergesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar.

Das gilt insbesondere für Vervielfältigungen, Übersetzungen,Mikroverfilmungen und die Speicherung

und Verarbeitung in elektronischen Systemen.Gedruckt auf säurefreiem Papier

(alterungsbeständig pH7 –, neutral)Printed in Germany

Gedruckt mit Unterstützung der Deutschen Forschungsgemeinschaft.

Cover:A court of biis. Turkestanskij Al’bom.

(Library of Congress, Prints & Photographs Division, LC-DIG-ppmsca-09951-00106)

Contents

Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Nomads in History. A View from the SFB. With Comments by Anatoly M. Khazanov . . . . . . . . . . . . . 17

Nicola Di CosmoAristocratic Elites in the Xiongnu Empire as Seen from Historical and Archeological Evidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

A. C. S. PeacockFrom the Balkhān-Kūhīyān to the Nāwakīya: Nomadic Politics and the Foundations of Seljūq Rule in Anatolia . . . . . . . . . . . . . . 55

Jürgen PaulSanjar and Atsız: Independence, Lordship, and Literature . . . . . . . . . . . . . . . . . . . 81

Rudi Paul LindnerThe Settlement of the Ottomans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

Shahin MustafayevBetween Nomadism and Centralization: The Ottoman Alternative in the History of the Aqqoyunlu State . . . . . . . . . . . . . . . 143

David SneathAyimag, uymaq and baylik: Re-examining Notions of the Nomadic Tribe and State . . . . . . . . . . . . . . . . . . . . . 161

Anatoly M. KhazanovThe Eurasian Steppe Nomads in World Military History . . . . . . . . . . . . . . . . . . . . 187

Aristocratic Elites in the Xiongnu Empire as Seen from Historical and Archeological Evidence*

Nicola Di Cosmo

Preliminary Remarks

The concept of “elite” is omnipresent in the study of ancient nomadic societies and his-tory. It is used in a generic sense to indicate members of elevated social and political rank, and is at times conflated with other terms, such as aristocracy, nobility, or ruling class. The plasticity of this concept is useful when speaking of elites because we need not clarify what level of social stratification, or composition, a given polity attained, or how an elite status was ascribed, attained, or transmitted. It is sufficient to know that there was some social dif-ferentiation and that a certain group of people had access to greater wealth and power than others, whereas in the case of terms such as aristocracy and nobility, which carry notions of hereditary status, ranked structure, and power relations with respect to a putative organizing center, a far more precise definition is required.

Ancient nomadic societies have been particularly difficult to corral into schematic rep-resentations of social and political relations. The notion of a “nomadic feudalism” proposed almost a century ago by Vladimirtsov lost much of its appeal with the crisis of the term “feudalism” itself. Other schemes, based on the assumption that ancient pastoral societies develop socially and politically only under the influence of external impulses, have been in-effective at defining phenomena such as elite formation or social differentiation. Moreover, an excessive (nearly exclusive) attention to dynamics of conquest or “trade or raid” has pre-vented any analysis of the emergence of elites within their societies as a necessary condition for certain types of interactions with other societies. This orientation, which has assigned to any pastoral society a subaltern role with respect to those agrarian societies by which they were supposedly influenced, has therefore tended to ignore internal dynamics of political development and formation. Such approaches are increasingly at odds with the archaeologi-cal evidence on Eurasian nomads.

The recent advances in archeological research in Mongolia, Tuva, the Altai region, Trans-baikalia, and other parts of Siberia and Kazakhstan show the nontenability of any theory that would divide the Eurasian world into opposing camps: a non-self-sufficient “nomadic” camp and a rich agrarian one, the first being permanently poised to assault the second. It is quite clear that the long-term dynamics of social development included inter-nomadic warfare, the transmission of knowledge (including political ideas, religious beliefs, and technology)

* I would like to thank Ursula Brosseder for valuable comments on an earlier version of this paper. I am of course responsible for any mistakes.

Nicola Di Cosmo24

from and through various societies – which could be agrarian, pastoral, or mixed – and that the rise of more complex nomadic societies and political formations is a process that can-not be reduced to single-factor explanations, such as economic dependency. Long trends of human settlements and movements, migration and conquest, about which we have but few traces, went together with the adoption of more sophisticated forms of political organiza-tion, technological development, economic specialization, and gradual class differentiation. The construction of large kurgans such as that in Tuva (Arzhan), dated around the end of the ninth century or early eighth century BCE, cannot be understood without taking stock of the long tradition of kurgan-building in the Eurasian world, and yet it represents a new stage in the development of ancient nomadic cultures.1 Likewise, later developments, both in the Altai region and to its east and west, can hardly be comprehended without taking into consideration centuries of significant cultural change among the nomadic elites.

It is therefore against this backdrop of long-term developments that the rise of nomadic empires has to be placed. Interactions among various nomadic groups and agrarian societies were surely important, but to what extent they can be regarded as generative of change is open to question, and in any case no change can be assessed without an understanding of the political infrastructure internal to nomadic societies, such as social order, access to resources, military mobilization, ritual practices, and the ways in which external relations – including trade and warfare – were organized. Today, the rapid growth of a vast archaeological litera-ture on Eurasian nomads allows us to ask more searching questions about the formation and development of “elites” in relation to the distribution of power across political networks, to their economic role in view of the presence of a variety of productive activities in areas previously regarded as purely pastoral, and to the cultural changes that the elites themselves underwent.

In archaeology, the term “elite” continues to have broad uses and applications, encom-passing the full spectrum of the material record from burials to material remains, and is in-terchangeable with terms such as royal or aristocratic, monumental, and ostentatious if refer-ring to burials and luxury or prestige if referring to goods. The research field that is Xiongnu archaeology has also made tremendous headway over the past several years, due especially to numerous excavation projects in Mongolia and Transbaikalia. This is a critical aspect of the study of nomadic aristocracies, because the Xiongnu played the special historical role of being the first empire formed to the north of China, and, as such, the precursor of many other nomadic empires (Turks, Uighur, Khitan, Jurchen, and especially Mongols), achieving distinction, if not always appreciation, in the historical records of Eurasian civilizations. Ar-chaeological research is transforming our knowledge of the Xiongnu phenomenon by reveal-ing its cultural complexity and at the same time stimulating a reexamination of the written sources with a new appreciation of the internal diversity of the ancient nomadic world.

The nature of the Xiongnu polity is a matter that has been debated for decades, and inevitably divergent opinions are reflected in the discordant terminology, as scholars define it as a state, a tribal confederation, or a “supercomplex chiefdom”. Glossing over such dif-

1 Grjaznov, Der Groβkurgan von Aržan in Tuva, Südsibiren, pp. 70–75. Hanks, “Archaeology of the Eurasian Steppes and Mongolia”, pp. 476–477. Parzinger, Die frühen Völker Eurasiens, p. 609.

Aristocratic Elites in the Xiongnu Empire 25

ferences, everyone agrees about the existence of a political “elite”, as documented in the Chinese sources, and in particular in the narrative accounts of the Xiongnu in the Shiji and Hanshu. Another point of agreement is that the Xiongnu, no matter what definition we use to refer to their polity, formed an empire. Like other empires, they projected their power and extended their territory far beyond their original homeland (whose actual location remains undetermined archaeologically),2 possessed a variety of ethnic and linguistic components, and had relations with China and other states in which the imperial status of its supreme head, the chanyu, was implicitly or explicitly recognized as having the same dignity as that of the Chinese emperor. As a contribution to the study of the Xiongnu empire, this essay attempts to address the question of the political nature, composition, status, and symbols that defined its elites in both the historical and the archaeological realms.

The first part of this essay consists of a reexamination of the historical sources on the composition and structure of the upper layers of the Xiongnu political establishment, and the types of “elite” that can be identified within it. The second part will address Xiongnu archaeology, focusing in particular on the question of elite burials and goods.

1. The Xiongnu Elite in the Historical Sources

The “Account of the Xiongnu” in the Shiji (chapter 110) provides a cursory explanation of the basic structure of the Xiongnu political elite, including a fairly detailed description of Xiongnu titles, ranks, and functions. While this structure has been mentioned many times before, it is useful to focus on specific passages that can be regarded as diagnostic to identify, even if only in a sketchy manner, the qualities and attributes that may clarify the social posi-tion and political role of “elite” persons.

A revealing passage is the one that describes the parricide committed by Modu (this is another reading for Maodun) against his father Touman:

He [i. e. Modu] followed his father, the chanyu Touman, on a hunt and when he shot at Touman with a whistling arrow, all his left and right attendants, following the whistling arrow, also shot and killed the chanyu Touman. Subsequently, [Modu] executed all his stepmothers and younger brothers as well as those among the great ministers who did not listen and follow. Modu installed himself as chanyu. [Shiji 2888; all emphases are mine]

This account of Modu’s violent coup d’état, through which he seized the supreme command of the Xiongnu and began his people’s imperial expansion, reveals in a single sentence at least three types of Xiongnu elite: the left and right attendants, the members of the royal clan, and the chief ministers. Who are these people? The left and right attendants must refer to Modu’s own bodyguards, whom he had previously recruited and trained to blind obedience. Here we notice the directional separation into left and right (that is, east and west) that was one of the organizing principles of the Xiongnu empire. The existence of this principle prior to Modu’s seizure of the imperium shows that the political establishment of the Xiongnu already functioned according to a principle of bilateral affiliation, which was used as the

2 Bemmann, “Was the Center of the Xiongnu Empire in the Orkhon Valley?”, pp. 455–461.

Nicola Di Cosmo26

empire grew to establish separate governments for the eastern and the western half. Since the bodyguard was essentially a military unit, one may speculate as to whether the principle of dividing the empire into two regions may have originated with a military organization whereby army units were arrayed into left and right formations. Stepmothers and younger brothers, all of them members of the royal clan, were executed as they presented a challenge to Modu’s rise to power, and in this we recognize a system of hereditary succession from fa-ther to son wherein the vertical principle prevailed rather than a lateral principle from older to younger brother, in which case a challenge could also have come from his father’s broth-ers. While not all instances of succession to the throne occurred according to the principle of vertical transmission, this was the most common system. The ministers loyal to his father may have also formed a hostile group. These were probably high dignitaries in charge of state affairs, including military, civil, or perhaps even religious affairs.

These three categories can all be regarded as “elite”, but obviously their social and politi-cal positions cannot be easily compared. What emerges is a possible differentiation between an aristocracy by birth, consisting of clan members, and an aristocracy appointed by the chanyu for their individual merits, consisting of bodyguards and ministers. Another passage makes it clear that the positions included in the top hierarchy of state were all hereditary:

They establish Worthy [xian 賢] Kings [wang 王] of the Left and Right, Luli [谷蠡] Kings of the Left and Right, Grand Commanders [da jiang 大將] of the Left and Right, Grand Com-mandants [duwei 都尉] of the Left and Right, Grand Household Managers [da danghu 大當戶] of the Left and Right, and Gudu [骨都] Marquises [hou 侯] to the Left and Right. The Xiongnu call a worthy a tuqi [屠耆]. Therefore, they usually take the Heir-Apparent to be the Tuqi [i. e., the Worthy] King of the Left. From the likes of the Worthy Kings to the Left and Right down to the Household Managers, the great ones have ten thousand horsemen [and] the small ones have several thousand, all twenty-four chiefs [zhang 長] are appointed with the title of “[Commander of ] Ten Thousand Horsemen”. All great ministers 大臣 have hereditary positions [shiguan 世官, my emphasis]. The three surnames of the Huyan lineage [shi 氏], the Lan lineage, and later the Xubu lineage constitute their nobility [guizhong 貴種]. All Kings and Commanders [jiang 將] of the Left direction reside in the Eastern region. Those who reside right across from Shanggu and beyond, border in the east on the Weimo and Chaoxian. The Kings and Leaders of the Right direc-tion reside in the Western region. Those who reside right across from Shangjun and further west border on the Yuezhi, the Di and the Qiang. (Shiji 2890–91)

Here we have a ranked honor system that differentiates, at the upper level, between three types of titles: king, chief, and minister. The term da chen for “great ministers” should per-haps be understood generically as “grandee” or “high dignitary” to indicate an aristocrat with a political position. All twenty-four of them had hereditary positions, were also considered chiefs, and held concurrent military positions, being heads of a myriarchy (ten thousand warriors).

Since the text specifies that among these twenty-four chiefs some are more important than others, and that their relative significance is measured on the basis of the actual number of troops at their disposal, it may seem contradictory that they are referred to by the same generic title (grandee, or da chen) and regarded as holding equivalent military ranks (Heads


of Ten Thousand). However, if we may allow an analogy with some later instances of Inner Asian military and political organizations, we can see that in both the Mongol and Manchu military systems the highest noblemen were given military titles as Head of Ten Thousand (tumen) in the case of the Mongols or as Head of a Banner in the case of the Manchus, regardless of the actual number of soldiers in the Tumen or Banner, which depended upon the subordinate units that were effectively included under their command. In the Xiongnu hierarchy, those aristocrats given the highest ranks held a dual position: one was a title linked to a specific political and government post and the other was a generic military title that in-dicated the holder as the commander-in-chief of the army recruited from among the popula-tion under his authority (de facto his subjects). This organization implies the existence of a system of territorial fiefs that were run in a semi-independent and fully autonomous manner, as indicated in the following passage:

Each [leader] has an allocation of land and moves according to a cycle in search of water and grass. As to the Worthy Kings to the Left and Right and the Luli Kings to the Left and Right, these are the greatest. The Gudu Marquises of the Left and Right assist in the government. Each of these twenty-four chiefs also establishes on their own authority Chiefs of a Thousand, Chiefs of a Hun-dred, Chiefs of Ten, Supporting Lesser Kings 裨小王, Administrators of Fiefs, Commandants, Household Managers, Juqu 且渠, and others. (Shiji 2891).

The context makes it clear that “each” refers to one of the twenty-four leaders, and that they follow, with their people, a nomadic cycle of movement from pasture to pasture on a seasonal basis. We should note that some of the titles established at the local level are the same as the state titles, but were not prefixed with “Great” (da 大) and were not divided into a left and right position, as for instance the Household Managers (danghu) and the juqu (Shiji 2890; Hanshu 3751). Clearly the Xiongnu elite included a central and a local level. The members of the very top echelon were distributed across the empire, and controlled a portion of it as a private domain, where each was free to appoint his subordinate officials ac-cording entirely to his own preference, and without political interference from the Chanyu. Naturally, their unlimited authority over their domains limited the Chanyu’s sovereignty over the empire as a whole.

To delve further into the question of the formation of local elites, let us consider briefly this statement, which refers to Xiongnu military campaigns undertaken sometime between 206 and 200 BCE:

Later [Modu] in the north subjugated the states of the Hunyu 渾庾, Quyi 屈射, Dingling 丁零, Likun 鬲昆, and Xinli 薪黎. Hence, the nobles 貴人 and the great ministers 大臣 of the Xiongnu all submitted [to him] and regarded Modu as worthy. (Shiji 2893)

Since this passage comes after the description of the establishment of the Xiongnu state un-der Modu, we have to assume that not all Xiongnu chiefs and grandees had initially accepted his leadership, and that, therefore, the Xiongnu were not unified until sometime after Modu became chanyu. Modu’s successful military leadership and expansion of the Xiongnu terri-tory persuaded the Xiongnu leaders to finally join him, but this was a nobility, as well as a government class of high dignitaries, that already existed, even if it was not yet unified. The

Nicola Di Cosmo28

implication of this state of affairs is that the authority of the chanyu before Modu’s ascent to power did not extend to all Xiongnu leaders, and suggests that the Xiongnu expansion transformed the political order by creating a centralized hierarchy, but that this was based on an already existing aristocracy and government system, which was expanded and to a certain degree restructured to fit the new “imperial” growth.

The specific question raised by this passage, then, is how those nobles and high digni-taries who joined later were incorporated into the new state. They may have retained their authority and territory, but since these were presumably outside the “command structure” of the Xiongnu upper elite, one can only surmise that it was from their ranks that the lower, local elites were selected. This would fit the notion that the twenty-four “grandees” could appoint local chiefs to various positions in their own fiefs, such as Head of a Thousand or Fief Administrator. Further textual evidence of this can be found in a speech by several Xiongnu dignitaries to the chanyu Huhanye, in which it is stated: “Today brothers fight over the [Xiongnu] state, [yet] if [power] is not with the elder brother, then it is with the younger brother, and although they may die, they nevertheless have power and fame, and their sons and grandsons continue to be the commanders of many states” (Hanshu 3797). This pas-sage illustrates the principle that the Xiongnu upper nobility was in charge of ruling subject states, and the local elites were pressed into service under their command.

It is plausible, based on the texts we have examined, that the growth of the Xiongnu empire was characterized by the gradual cooption of local nomadic chiefs into the fief ’s bureaucracy and military apparatus, while the upper (state) ranks remained in the hands of the three royal clans and the nobility closest to Modu. Incidentally, whether the Xiongnu did or did not have surnames (xing 姓) aside from the three royal clans, as Sima Qian says (Shiji 2880), is of limited relevance here, as it probably means that only the upper nobility’s three lineages generated the members of the aristocracy with the highest political ranks. This system’s weakest point consisted of a perverse mechanism: the more the empire expanded, the greater the power of regional “satrapies”. Since the sovereignty of the chanyu was based on two elements, that is, first, his monopoly over foreign policy, with the attendant revenues derived from tribute and trade, and, second, his right to mobilize troops from the whole em-pire, both sources of authority could be undermined by powerful “fief-holders”, who could use their own armies to take independent political action, for instance by raiding China of their own accord or by expanding their power base without the chanyu’s approval.

There was also a hierarchy within the upper elite, and some positions were more power-ful than others and had greater military weight. The high dignitaries bearing the name of “kings” were mightier than the “ministers”, and the Worthy (or tuqi in Xiongnu language) King of the Left was the appointed Heir Apparent. Since the twenty-four grandees held specific positions within the central government, they must have gathered at the court of the chanyu from time to time to hold council concerning pressing political matters and military affairs and to attend religious ceremonies, but they probably resided in their fiefs most of the time.

The Shiji lists the three royal lineages (the Huyan 呼衍, Lan 蘭, and Xubu 須卜) imme-diately before the statement that the twenty-four grandees held hereditary positions. Wheth-er all of them were chosen from these three clans is not said explicitly, but it may be surmised


from the proximity of the two statements. In other words, the nobility of the Xiongnu consisted of the members of these three clans, from among which the top commanders (the “puissant chiefs”) as well as the Heir Apparent, were selected. The top positions were also arranged territorially into two halves, an eastern (left) and a western (right) one, with the left sector deemed superior to the right. The grandee, who was also a fief-holder, entrusted the administration of his own domain to a hierarchy of political and military leaders internal to the fief. This hierarchy was modeled after the higher state-wide structure, and the titles of the local (fief-based) elite by and large replicated the structure of the upper, imperial elite.

Such considerations force us to open a parenthesis and re-examine briefly an old but still influential model of empire-building by pastoral nomads, according to which – and the Xiongnu are thought to be no exception – a “supratribal” elite was established by the khan once a unifying political process had been completed and a tribal confederation, or even a nomadic state, had come into being. In this model, the supratribal elite is supposed to be qualitatively different from the pre-existing tribal elites, which, however, continued to exist in a relationship of both subordination to and partnership with the ruling clan. While the ruling clan was the sole legal social group that was entitled to assign and hold the supreme political position in the land (chanyu, khaghan, khan, etc.), allied, co-opted and subaltern clans and tribes could hypothetically exercise the option to secede from the imperial no-madic union. The essential features of this model (sketchy as they are) do not fit what we have so far said about the Xiongnu in at least two ways.

First of all, the Xiongnu do not appear to be divided into clans and tribes across the whole social spectrum. Certainly, they absorbed a number of other peoples and it is likely that some of these peoples retained their own elites as well as a degree of control over their territory, as we shall see below. However, the core twenty-four positions were not parceled out and handed over to conquered people, but were held by the Xiongnu nobility, and were hereditary. These noblemen had special posts which their descendants continued to occupy essentially in perpetuity, and which became, therefore, the personal property of a certain family or descent line, originally associated with one of the royal clans. These people in turn replicated on a smaller scale the same central government structure within their fiefs. The relative position of each grandee within the upper hierarchy also determined (and was possibly partially determined by) the size of the population under each one, which corre-sponded to the number of soldiers they could put in the field. Thus, the title Commander of a Myriarchy simply means commander of the army of that particular fief, located within the eastern or western half of the Xiongnu empire, and held by one of the twenty-four Grandees in addition to their nobility title.

Second, it would be a mistake to assume without any supporting evidence that this sys-tem was established by Modu, and that, therefore, it represented a qualitative leap from a previous – tribal – Xiongnu political system to an imperial and supratribal one. The Xiong-nu undoubtedly went through a political transformation to resolve a political crisis, caused by the Qin army’s massive invasion of nomadic lands after the unification of the empire, and the very parricide committed by Modu could be interpreted allegorically as a replacement of the old aristocracy by a new group of leaders whom Modu commanded with unchal-lenged authority. After his bloody rebellion, Modu proceeded to place his own men in key

Nicola Di Cosmo30

positions, ruthlessly eliminating all opposition. This scenario, as presented in Sima Qian’s account, is in my view entirely plausible. What is less believable is that Modu invented a completely new political and military system right after his accession to the throne. Nor does Sima Qian say that he did.

What is more likely to have happened is that the basic structure of the previous order was retained and adjusted to the needs of a new order. Later on, thanks to the extraordinary military successes achieved during the thirty-five years of his reign, Modu (r. 209–174 BCE) vastly expanded the territory and the population under Xiongnu dominium and presented himself as the emperor of a united polity in his confrontation with China. The growth of the empire was accompanied by a duplication of the key positions into right and left ones, and by the replication of the upper structure at the lower level. The conquered lands were appor-tioned as personal domains to the closest family members and collaborators of the chanyu. In turn, the fief-holders created bureaucratic and military elites internal to their domains, possibly by co-opting local elites. (The “supporting lesser kings” mentioned by Sima Qian could be chiefs of local peoples.) Therefore, if we can speak of a qualitative leap, this should be seen primarily in terms of the growth of the Xiongnu empire, and of the reordering of the internal hierarchies and power relations within the newly conquered lands. At the level of the chanyu court and state command structure, the political order changed in the sense that a new, highly centralized political elite came to power, but this new order was built upon a pre-existing system of government that used to function on a smaller and possibly more flexible scale, then remapped on a much larger scale to fit the needs of an expansive empire.

Later on, as Modu was replaced by less successful chanyus and the Xiongnu political crisis deepened, two parallel phenomena developed. On the one hand, there were intrigues, plots, rebellions, and internecine warfare among the members of the Xiongnu upper nobility, all of whom had access to resources of their own in their personal power bases. On the other, the subject peoples, such as the Wuhuan and the Hujie, who had been previously defeated and subjected by the Xiongnu, rose up in arms to attack the local Xiongnu leaders. There was, in the end, no simple way for tribes to “walk away” in search of a better leader, but rather there were bitter fratricidal wars – ruthless wars of revenge by formerly subject groups, which included the defilement of Xiongnu aristocratic tombs.

The scenario I have proposed to account for the basic features of the Xiongnu elite, then, includes a two-tier aristocracy with government functions with clearly marked hierarchies in both the higher (central, imperial) and lower (local, fief-based) nobilities, and an east-west distribution of fiefs, which was also asymmetric in terms of political priority and hierarchy. This picture is complicated by two aspects that warrant further consideration. The first is the incorporation of foreign people into the Xiongnu elite, such as nobles from conquered states or surrendering Han aristocrats and generals. The second is whether we can see changes in the composition of the Xiongnu elite, in particular the appearance (or disappearance) of elite titles.

The Xiongnu elite undoubtedly included foreigners, who were incorporated within the upper ranks of Xiongnu nobility, such as surrendering Chinese generals and local non-Xiongnu leaders. Let us consider the following passage, which refers to the case of the Mar-


quis of Xi, a figure emblematic of the subversive role played by the Xiongnu with regard to the Han dynasty, who defected to the Xiongnu :

After the Chanyu had obtained [the surrender of ] the Marquis of Xi 翕 [i. e., Zhao Xin], he made him a king subordinate [only] to himself, had his [own] elder sister marry him, and had him join in strategy planning regarding the Han. (Shiji 2908)

This statement shows that the Marquis of Xi was not appointed to one of the twenty-four highest positions but was nevertheless made a noble and placed directly under the authority of the chanyu. This reveals the existence of a group of noblemen and aristocrats who formed a royal council directly under the chanyu’s authority. Such a council would make perfect sense, as many Han defectors could be of invaluable assistance in diplomatic and military affairs. Another case in point is the eunuch and Han envoy, Zhonghang Yue. According to Sima Qian’s narrative, after Zhonghang Yue surrendered to the chanyu, the chanyu became “very close” (shen qin 甚親) to him and “favored” (xing 幸) him (Shiji 2898). As is well-known, Zhong-hang Yue played a key role in negotiating with Han envoys and in handling Xiongnu diplomacy. Finally, we may mention the case of the Han general Li Ling. After this general, who fought valiantly against the Xiongnu, was forced to surrender, the chanyu ennobled him (gui 貴) and gave him his daughter as a wife (Shiji 2918). Interestingly, the Hanshu reports that, when a Han envoy met Li Ling (d. 74 BCE), the Han defector was wearing nomadic clothing and his hair was braided in the manner of the Xiongnu. When asked by the Han envoy whether he would return to China, “he went silent and made no reply, and after contemplating the length of his hair, he answered, ‘I am now dressed like a nomad!’” (Hanshu 2458). All three people, who were among the most famous Han “traitors” who defected to the Xiongnu, had a personal relationship with the chanyu and were raised to the rank of Xiongnu nobility in virtue of their marriage to women of the royal clan.

These cases beg the question of whether a council – an inner circle of personal advisors to the chanyu – existed separately from the twenty-four high dignitaries or other nobles. More clues point in that direction. For instance, in the description of one of the main religious rituals performed at court – namely the daily worship of the sun and moon – we are told that the chanyu was accompanied by his chiefs, or commanders, who sat on his left and faced north. The word used in the Chinese text is zhang 長 and such usage may be meaningful, in that the people who flanked the chanyu in these daily activities may not have been the da chen – that is, the fief-holders appointed to the highest state positions – but rather the trusted members of the comitatus, who lived permanently at the chanyu’s court. On the other hand, the term zhang is sufficiently generic to mean simply some chiefs, and therefore the hypothetical distinction between zhang (as members of the chanyu’s council) and da chen (as fief-holders and army commanders with state responsibilities) must remain speculative.

This scenario brings to light an important feature of the Xiongnu elite at the upper level – namely, the political separation between the twenty-four grandees, who represented the state apparatus, and the chanyu’s personal council of trusted advisors. Within this inner group, and possibly with quite different functions and responsibilities (which may have been political, military, or administrative), there were people who had pledged their support for the chanyu, with whom they had established a direct and intimate relationship sealed by kin-

Nicola Di Cosmo32

ship ties ensured through marriage with women from the royal clan, including the chanyu’s own sisters and daughters. We may also recall that, at the time of the heqin policy, Chinese princesses were sent to the Xiongnu court with their retinues and established themselves in his entourage. The heart of the Xiongnu political establishment may be imagined as a com-posite group formed of various elites, who also represented themselves according to specific cultural choices. Some of the foreign members may have acculturated to the nomadic life-style, as in the case of Li Ling, while members of the Xiongnu aristocracy may have adopted new symbols of elite status, for instance by dressing in imported Chinese silk.3 This Xiongnu court was, in other words, open to various people whose talents and devotion to the chanyu probably trumped ethnic affiliations or cultural preferences and taste, but whose political role required, in order to retain legitimacy and cohesion, being fully incorporated within the hereditary structure of kinship-based nobility.

A point that cannot be exhaustively researched in this essay, but that should be con-sidered with regard to possible historical changes in the composition of Xiongnu nobility, is that we find in the Hanshu a number of Xiongnu titles that in the Shiji are either absent or are mentioned in passing without a clear context. For instance, the term Rizhu King, for the Xiongnu high-ranking dignitary who was entrusted with the government of the western-most part of the empire (a title also divided into Left and Right), and presumably a position at the level of the twenty-four da chen, is found in the chapter on the Xiongnu in the Hanshu (3795), but is mentioned in the Shiji only in the Tables (Shiji 20, 1068), not in the chapter on the Xiongnu or in any other narrative part of the Shiji. Likewise, the title Yizhizi (伊秩訾) King, also divided into a Left and Right positions, is found in the Hanshu (3797; 3806; 3823) but is completely absent in the Shiji.

A different title that also appears in the Hanshu (e. g., 3788, 3790) but not in the Shiji is that of Yujian (奧鞬) King. While this title, like most Xiongnu titles, was divided into Left and Right positions, it was not a Xiongnu title, but rather came from the name of the king of the Yujian city in Kangju (Transoxiana) mentioned in Chapter 96A of the Hanshu.4 An intriguing passage states that, after the Left Yujian King passed away the chanyu of his own accord established his minor son as Yujian king, and made him reside at court. However, the noblemen of Yujian (奧鞬貴人 Yujian guiren) together established the original son of the Yujian King as the new king (Hanshu 3790). Clearly, the Yujian king was a foreign vassal of the Xiongnu, and the chanyu’s attempt to replace him with his own son was not accepted by the aristocrats of the city of Yujian, who instead appointed the dead king’s son. But it is still peculiar, and difficult to explain, that the position was divided, following Xiongnu custom, into a Right and a Left one. One possible explanation is that some of these provincial kings’ designated heirs were given the Left rank as an additional honorific when they were required to reside at the Xiongnu court. The Worthy King of the Left was the designated heir to the chanyu’s throne, and the custom of having the sons of local leaders sent to reside at another people’s (sometimes their enemies’) court was common among both the Han and the no-

3 The replacement of nomadic clothing by the Xiongnu is the object of a scorching reprimand by the Chinese eunuch, Zhonghang Yue, whom we can imagine to have also traded Chinese clothes for nomadic attire (Shiji 2899).

4 Hulsewé, China in Central Asia, p. 131.


madic peoples. Hence, the passage above may indicate that a king of Yujian who was the original heir apparent and (therefore legitimate) king died, and when the Xiongnu wanted to make his own son king of Yujian and have him reside at court (presumably until he at-tained his majority), the Yujian nobles rebelled. This is only speculative, but may explain a further use of the prefix “Left” as an indicator of legitimate succession.

One more title that comes from a foreign people found in the Hanshu is the Hujie 呼揭 (glossed as Huqie) King, who was also based in the western region of the empire. The Huqie were a population that had been conquered in 174 BCE by Modu, and “had become Xiong-nu” (Shiji 2896). According to the Hanshu, this king was located in the western part (xi fang) rebelled and declared himself chanyu (Hanshu 3796). Moreover, in the Shiji and Hanshu we find that the kings (wang 王) of the Loufan and Baiyang peoples, who were located “south of the river” and had been integrated into the Xiongnu empire, were attacked by the Han general Wei Qing (Hanshu 3766). These examples show how the aristocracy of conquered and surrendered peoples continued to retain their positions as they were incorporated into the Xiongnu empire, and formed a local elite, presumably under the overall authority not only of the chanyu but also of the upper Xiongnu elite, that is, the “fief-holders” who ruled over the region in which these local “kings” were based.

Another interesting case concerns a potentially non-aristocratic level of the elites, which is exemplified in a position that belongs in the Xiongnu bureaucracy, but again is found in the Hanshu but not in the Shiji. The term husului 呼遫累 (Hanshu 3796; for the pronun-ciation “lui” see note on p. 3797) is glossed as the name of an official. We cannot say what this position represented or whether it represented a change or addition to the Xiongnu bureaucracy from the time of Sima Qian. Another bureaucratic title, given only in Chinese, is the Xiongnu Master of Guests (Xiongnu zhuke 匈奴主客, Shiji 2912), which appears to be a court position, held by one serving directly under the chanyu, who may have been either a court bureaucrat or a member of the aforementioned “inner circle”. These are just a few examples, reported here simply to show that the composition of the Xiongnu elite as provided in the Shiji and in the Hanshu may not represent the full spectrum of the Xiongnu elite, which also probably changed as the nature of the empire was transformed by its expan-sion first and later by its shrinkage and fragmentation.

Tentative as this reconstruction of the Xiongnu elites may be, it provides us with a plat-form that allows for some conclusions regarding the composition of the hierarchy of the Xiongnu political establishment, in particular in relation to the identification of a distinc-tion between upper and lower elites, the presence of foreign elements within the elite, the separation between hereditary and non-hereditary positions, and the role of the court and clan system in the production, as well as reproduction, of elites. As we have argued, the Xiongnu elite was structured at several levels, of which we can perhaps identify two broad classes, each including several groups. The upper elite was formed by the original Xiongnu aristocracy, which included the three royal clans, the chanyu and his queen (her Xiongnu title was yanzhi 閼氏), and the top ranking officials, who also came from the royal clans, consisting of the twenty-four fief-hold “grandees”, divided into the Right and Left divisions. To the same upper level, but separate from the original Xiongnu aristocracy in terms of political role and patterns of appointment, recruitment and cooption, belonged the council

Nicola Di Cosmo34

and retinue of the chanyu, and other positions attached to the court, which we have defined as the chanyu’s council. All of these positions appear to be hereditary. The lower elites can be divided into local Xiongnu elites and the elites of surrendered peoples. The fief-holders appointed their own military and civil officials within their domains, and also controlled the aristocracy of the subordinate peoples, who probably retained a degree of autonomy as heads of their own communities and were entitled to rule over their peoples. These positions may have been hereditary, but could also be altered or abolished, as the aforementioned case of the Yujian King shows.

This general survey also indicates that the Xiongnu replicated at the local level the same political and military titles and hierarchical structures established at the central level, and in so doing also incorporated the elites of conquered peoples, who thereby “became Xiongnu”. Each of the twenty-four high dignitaries had his own power base, military force, and full au-tonomy to make appointments in the lands he ruled, not just as “governor” but as a virtually independent king. The empire grew in size (in terms of both territory and people) thanks to a system of cooption of local elites, both Xiongnu and especially non-Xiongnu, while at the same time the ability of the political center to penetrate and control politically the “provin-cial” (“fief”, “satrapy”) level grew weaker.

2. The Xiongnu Elite in the Archaeological Contexts

One may be tempted to seek confirmation of the hypothetical reconstruction of the Xiong-nu elites presented above by investigating the vast archaeological literature available today. However, this is a temptation that should be resisted. Trying to superimpose a theoretical scheme drawn from historical sources onto an altogether different set of evidence and ques-tions, born out of archaeological research, would be methodologically unsound. A com-parison between different forms of identification, distribution, and hierarchical structures of the aristocracy in the Xiongnu empire, if possible at all, must be preceded by a thorough examination of elite status based exclusively on an analysis of the material record.

So far, a discussion of elites in Xiongnu archaeology has, by and large, been in the context of the study of specific sites identified as “elite”, but has not produced a distinctive theory of social stratification. A critical aspect of Xiongnu archaeology is its evolution into two sepa-rate areas of investigation, which have been developing along separate trajectories – namely, the archeological research conducted in the “Northern Zone” (beifang diqu) of China, south of the Gobi desert, and the excavations of Xiongnu cemeteries in Mongolia, Buriatia, Trans-baikalia, and generally to the north of the Gobi. Archaeologists have offered different in-terpretations of the distribution, chronology, and characteristics of what is supposed to be a “Xiongnu culture”. It would be more accurate, in view of the sprawling development of Xiongnu archaeology and of the complex questions that it poses, to speak of a “Xiongnu pe-riod” that embraces the duration of the Xiongnu empire but must also include other groups coeval with the Xiongnu, and thus avoid the common fallacy of attaching a single ethnic label to a set of cultural markers shared by a whole cultural complex of horse-riding pastoral peoples. As far as I can tell, there is no clear periodization of the Xiongnu, either as a “cul-ture” or as a political phenomenon, that can be inclusive of all the areas in which Xiongnu


remains have been identified. Archaeological sites regarded as “Xiongnu” that are located to the south of the Gobi, in Inner Mongolia and adjacent areas included in the “northern zone” of Chinese archaeology, show distinctly different features from the Xiongnu archaeology of the north of the Gobi, which are characterized by large necropolises, monumental burials and distinct ritual ceremonies.

Constructing a synthesis of a “Xiongnu period”, therefore, would be premature at this stage of the research, but we can approach the issue of identification of “elites” from two common starting points, namely by looking at “elite burials” and “elite goods” separately in each of these two areas. Elite burials in the northern part of the of the Xiongnu “phenom-enon” are characterized by monumental burials, not found in the southern area, which also contain elite goods, whereas elite status in the Xiongnu tombs of southern Mongolia and northern China is represented only by elite goods. The following summary analysis is by no means exhaustive of the debates, but may serve to highlight the complex issues that arise from the identification of “elites” in Xiongnu archaeology.

2.1 Xiongnu elites in the “Northern Zone” of China

Historians have primarily cited archaeological material related to the Xiongnu in northern China in support of claims that the evolution of the Xiongnu “state” towards a higher level of political integration was “stimulated” by technological progress – the development of “productive forces”, such as the greater sophistication of metallurgy and the incremental use of iron.5 In an archaeological context strongly influenced by evolutionary approaches and theories, the search for the origins of the Xiongnu has been rooted in the belief that they represented the most advanced manifestation of the discrete pastoral cultures that evolved in the steppe areas, reaching the peak of the developmental curve in the late Warring States and Han periods.6 The search for elites in the archaeological record, thus, intersects several layers of interpretation, from the internal evolution of pastoral communities, to dynamics of contact and exchange between the Northern Zone and the Central Asian cultural complex to its north and further west, and to contacts with China.

However, the extensive debates about the origin, direction, and frequency of material exchanges and influences have not satisfactorily resolved the question of how social and economic developments within the Northern Zone, such as the increase in pastoral produc-tion, decrease in agricultural production and progressive emergence of a warlike mounted aristocracy, came about. The wealth of sites investigated to date that bear evidence of their affiliation either to non-Chinese northern cultures or to Sino-northern cultural admixtures pose questions that cannot be answered by solely considering problems related to dating, classification, and typological analysis of the material record. Were these cultures converted to nomadism through a process of long-distance, regional or local development? What was

5 Lin Gan, Xiongnu shiliao huibian, pp. 1–3.6 One of the best descriptions of the Xiongnu sites, and of the material culture associated with them is provided

by Akiyama Shingo (“Nei Menggu gaoyuan de Xiongnu muzang”, pp. 375–392). The attribution of the ethnonym Xiongnu to the ancient culture of the southern Mongolian steppes is discussed very briefly on pp. 389–390.

Nicola Di Cosmo36

the nature of the nomads’ interaction with neighboring communities? How mobile were these societies, and how did migrations contribute to the spread of a pastoral economy? What was their social structure, and how did it change over time? And finally, how did they become so powerful? All of these questions, of course, bear on the issue of the historical rel-evance that these people eventually acquired, as the frontier between China and the nomads became not only a place of cultural transmission but also a battlefield of opposing armies and a boundary between actively or potentially hostile powers.

Given its fragmentary nature, the data available – by now substantial but by no means uncontested – supports not only different theories but also different directions of investi-gation. So far, a degree of consensus has been reached with regard to certain aspects. For instance, the Northern Zone metallurgical tradition displays original features that have led researchers to regard it as one of the important centers of ancient metallurgical cultures. Around the second half of the second millennium, and especially during the first half of the first millennium BCE, pastoralism gradually expanded in the area, replacing earlier ag-ricultural and mixed agro-pastoral cultures. Animal breeding probably expanded over time – albeit the timeline, territorial centers and modalities of this expansion are documented vaguely at best – while farming did not disappear completely. Pastoral peoples’ remains, largely limited to burials, shared some common features, including above all metal objects, such as weapons, buckles, ornamental plaques, and horse and chariot fittings, often featur-ing a distinctive “animal style” and geometric decor, in bronze, but also gold, silver, and iron. These forms show continuity with earlier types and wide trans-regional distribution. In other respects, such as burial customs, pottery tradition, and bone artifacts, northern cultural sites differ profoundly, and local characteristics predominate.

The archaeological picture shows, around the middle of the first millennium BCE, a world engaged in pastoral economy and extensive military activity, rich in weaponry and sophisticated personal ornaments, and one that also made extensive use of horses. This is a world in which a martial lifestyle became more pronounced, possibly among both men and women, and it would seem reasonable to posit, as a preliminary and admittedly elemen-tary conjecture, that the more highly organized, militarily stronger, and possibly but not necessarily more technologically advanced pastoral groups imposed their rule over weaker, smaller, or less organized communities. Wider adoption of the technology associated with nomadic cultures, including the manufacture of bronze weapons and horse fittings, and an extensive knowledge of animal domestication, might in turn have been responsible for the growth of the political power, social status, and economic relevance of a mounted nomadic aristocracy.

Because the affirmation of the power of the Xiongnu vis-à-vis China is assumed to re-present the zenith of the political and military might of pastoral nomads, the study of the Xiongnu archeological culture has likewise been influenced by an evolutionary approach.7 While Chinese archaeology has often resorted to historical records in its classification of Northern cultures (names derived from the written sources such as Di, Rong and Hu are of-

7 For an example of the evolutionary approach to the definition of the Xiongnu culture, see Tian Guangjin, “Jinnianlai Nei Menggu diqu de xiongnu kaogu”, pp. 7–24.


ten attributed to archaeological sites), the Xiongnu culture appears as something altogether more significant and, so to speak, in a different league with respect to those other peoples. The conflation of the historical reading of the emergence of the Xiongnu as a new political force in East Asia with the archaeological record of the development of pastoral cultures in northern China means that, once evidence can be produced to show that the Xiongnu were a particularly advanced and “leading” pastoral nomadic culture, the premier role that their elites played in the political and military unification of the steppe nomads would ipso facto be explained.

However, the assumption that military or political superiority has to coincide with a higher level of development of “productive forces” and better technology lacks an empiri-cal basis in the history of Inner Asian nomads. Indeed, the written sources do not refer to any technological superiority of the Xiongnu as compared with other nomads, all of whom shared the military advantage that the extensive use of cavalry gave them. This is not to say that economic forces did not play a role; economic development by itself, however, does not provide a sufficient or adequate explanation for the rise of the Xiongnu empire. What is, in my view, a more promising avenue to identify possible transformations within nomadic so-cieties, rather than looking at mechanically determined changes in economic relations, is to focus on the ways in which the social, political and economic role of the aristocracy changed, as reflected in the material record. What archaeology can contribute to our understanding of nomadic aristocracies is not specific knowledge of the development of specific peoples, but rather insights into the wide-ranging cultural changes that may in turn shed light on historical processes.

Keeping this general methodological orientation in mind, it may be useful to refer to the distinction between “corporate” and “network” political economies within which elites emerge, which has been proposed by Blanton, Feinman, and others, and effectively adopted by B. A. Shepard in a recent essay.8 We can identify the following features of the network/corporate theory, according to Shepard’s abridged description. The “corporate” mode is based on local interactions, and the rise to power of elites occurs in close association with “communal identities”. This type of political economy is characterized by a more egalitarian distribution of wealth, shared power, emphasis on food production, power embedded in a group, monumental ritual places and other features. Individuals in higher positions of power and prestige are connected to the community through their “mastery of esoteric knowledge” linked to various cosmological themes. Most importantly, there is little emphasis on per-sonal accumulation of wealth “in the form of prestige goods obtained through long-distance exchange”. The “network” mode, on the other hand, is one in which it is individuals (as opposed to the community) who “confer power” and where this power “derives from long-distance interactions and the production, control and display of the wealth objects that these interactions generate”. Hence, such political economies are characterized by concen-trated wealth, prestige goods, princely burials, personal glorification and ostentatious elite

8 See in particular Blanton et al., “A Dual-Processual Theory for the Evolution of Mesoamerican Civilization”, pp. 1–14; Feiman, “Corporate/Network: New Perspectives on Models of Political Action and the Puebloan Southwest”, pp. 31–51. This paragraph paraphrases, abridges, or reproduces the relevant passages in Shepard, “Political Economic Reorganization among Non-state Societies”, pp. 366–367.

Nicola Di Cosmo38

adornment. The ostentatious aspects of the “network” elites are embedded in non-local or rare materials, those that require extremely expensive labor investments or that cannot be reproduced. Such differences in the political economies of archaic societies could be useful to identify different modes of self-representation of nomadic elites.

The funerary assemblage of pastoral nomadic sites in the Northern Zone of China can be examined with this particular aim in mind, that is, to identify the “corporate” or the “network” aspects of a given community’s elite. The objects that are usually regarded as sig-nificant to determine status consist of metal objects (in bronze, silver, gold, and iron) and precious artifacts (jewelry, precious stones, and objects of fine craftsmanship), while pottery, as important as it is in determining the cultural continuity and other features specific to a given culture or site, cannot provide sufficient information about the status system. Animal remains have also not been taken into account, because their presence is extremely wide-spread and, while the nature of the animals sacrificed and their position and number in the burials may indeed give an indication of status, specific correlations have to my knowledge not been systematically investigated.

We can divide the mortuary assemblages into two broad categories. The first category is characterized by martial symbols, in particular weapons and horse fittings, together with a variety of ornaments that are not intrinsically very rare or precious but nevertheless must have carried indications of status. The second category, on the other hand, is dominated by ostentatious, prestige objects, ranging from ordinary types to extremely rich ones, sometimes found in large quantities in a single grave. These two “scenarios” may be indicative of the different ways in which leadership status (rather than simply social status) was represented in the burials following the corporate/network classification mentioned above. That is to say, a “martial” burial may be indicative of a leadership that emphasizes a warrior’s status, charisma, and personal prowess, but not his accumulation of personal wealth, and finds its raison d’être in a system of values internal to the community, such as the protection of the group and the rituals associated to it. On the other hand, the prevalence of objects that ap-pear to display personal riches may point to a type of leadership that is more connected with hereditary prestige, luxury exchange, ostentatious wealth, and perhaps commercial opportu-nities, thus being closer to a “network” mode of political economy. These two scenarios only very rarely appear in their “pure” form, but consistent variations do point to differences in the ways in which what we might call a “nomadic aristocracy” constituted itself in relation to its own community as well as to foreign elites. Relations with other political entities pro-vided discrete opportunities and necessities, whereby defense, military expansion, trade, or the government of heterogeneous ethnic groups might have intervened to shape the nature of a given group’s leadership.

As we examine more closely the “martial” assemblage, a common theme is the presence of three types of objects: 1. weapons and tools in bronze and iron; 2. personal ornaments in bronze, and rarely in gold and silver (generally in tiny numbers); 3. horse and chariots fittings, attesting to the importance of the horse and to a probably greater mobility of the


group. Sites with these characteristics have been located not only in Inner Mongolia and the Ordos region, but also in Qinghai, Ningxia, and Gansu.9

At Maoqinggou (Liangcheng County, Inner Mongolia), one of the most prominent, larger, and better known sites in which burials of the upper stratum have been attributed by archaeologists to the “Xiongnu” culture, we find a metal inventory in which weapons and ornaments predominate.10 With eighty-one excavated tombs and a settlement, this site has been divided into four phases based on stratigraphic analysis, spanning from c. 700 to c. 300 BCE. It shows a remarkable continuity in the composition of the funerary assemblage, even though the style of the objects and their workmanship changed. Among the metal objects, weapons, belt ornaments, and ornamental plaques predominate. Horse-related findings, such as a bronze bit, are very few, and appear in only two tombs. Iron objects increase in the later period, while no gold remains have been found. The military nature of these elites is evident, I believe, in their being buried with their weapons, while wealth was stored mainly in relatively common bronze ornamental objects.

Taohongbala, another important Ordos site, dated to the Warring States period, is re-garded as one of the most representative of the Xiongnu archaeological culture.11 Here too we find several weapons, such as pick-axes, an axe, a club-head, and bronze and iron knives. Personal ornaments also abound; all are bronze with the exception of a pair of gold ear-rings. However, unlike Maoqinggou and Fanjiayaozi (see below), the burials of Taohongbala include several horse fittings – such as bits and chanfrons – and equestrian ornaments. In terms of precious objects, we find here only one gold pendant, similar to the gold coils found at Nanshangen (Liaoning) and Beixingbao (Hebei province). This may indicate the develop-ment of a “prestige exchange” network in which gold objects may have played an important role, but given the isolated nature of this find, its relevance to the definition of social status is unclear.

A second example of this type of assemblage is the “Xiongnu” tomb excavated at Yulong-tai, in Inner Mongolia. The rich material assemblage shows a mixture of weapons and tools, in both bronze and iron, and a variety of bronze objects, which constitute the greater por-tion of the assemblage, in which horse fittings such as bits, ornaments, and cheek-pieces, are widely represented. The ornamental plates include animal and geometric style plates, and finials shaped in the form of a horse head or crouching deer. The only “luxury item” is a silver necklace similar to one found at Wa’ertugou.12

The contrast between Ordos sites with and without equestrian remains can be seen when comparing the sites of Fanjiayaozi and Hulusitai. The first includes the usual bronze inven-tory of weapons and “animal-style” ornamental plaques but lacks horse-related items.13The

9 For a recent survey of Xiongnu sites in northern China, see Pan Ling, “A Summary of Xiongnu Sites within the Northern Periphery of China”, pp. 463–474.

10 Höllmann and Kossack, Maoqinggou: Ein eisenzeitliches Gräberfeld in der Ordos Region (Innere Mongolei); Nei Menggu wenwu gongzuodui, “Maoqinggou Mudi”, pp. 252–287.

11 Tian Guangjin, “Taohongbalade Xiongnumu”, pp. 131–142.12 Neimenggu Bowuguan and Neimenggu Wenwu Gongzuodui, “Neimenggu Zhunge'erqi Yulongtaide Xiong-

numu”, pp. 111–114.13 Li Yiyou, “Neimenggu Helinge'erxian Chutude Tongqi”, p. 79.

Nicola Di Cosmo40

metal inventory of the second set of burials (three tombs) also presents a large number of horse fittings (chanfrons) and in one tomb the remains of a large number of sacrificed horses (twenty-seven heads).14 Both sites have been attributed to the Xiongnu culture.15

A somewhat later Xiongnu burial site, also located in Inner Mongolia, is that of Bu-donggou, which comprises nine tombs.16 This has been dated to the Han dynasty on the basis of a TLV mirror with late Western or early Eastern Han characteristics, and presents several iron items including horse bits and equestrian ornaments. The presence of bronze cauldrons found at this site allows for the attribution of this site to the “Xiongnu” nomadic complex, which by this time encompassed the whole northern region of China, Mongolia, and parts of Siberia and Central Asia. The continuity of this type of assemblage with early ones is clear from one personal weapon (a sword), tools, ornaments, and especially horse fittings.

As mentioned above, broadly the same type of objects are present in a number of sites that are not normally attributed to the Xiongnu, probably because the geographical coordi-nates do not correspond to the region that they are presumed to have occupied. Neverthe-less, the cultural similarities are striking. In the northwestern sites of the Yanglang “culture”, in Guyuan county (Ningxia), the martial nature of the elite is even more prominent, and this may possibly be taken as the region where a martial mounted aristocracy developed more precociously and more fully than elsewhere. At Yanglang, forty-nine graves have been dated to the pre-Han period (c. 700–200 BCE).17 Here, military paraphernalia include one full iron sword, and the remains of several others, in addition to bronze weapons such as ge halberds (also found in several Ordos sites), spearheads, daggers, knives, arrowheads, pick-axes, and axes. Bronze and sometimes iron ornamental objects, such as plaques in the animal style, belt hooks, earrings, and belt ornaments, are present in large quantities. The chariot and horse ornaments, including bits, cheek pieces, chanfrons, and chariot caps, are especially numerous. A small percentage of findings consists of precious objects, namely, nine gold and silver earrings and some silver beads. The precious objects, the ge 戈 halberds and other products of Chinese origin attest to the existence of trade and exchange, but the abundance of military objects suggests that the local nomadic elite did not define themselves primarily in terms of their commercially derived wealth.

At the site of Shilacun, in the same county of Guyuan, we find the same type of assem-blage dominated by bronze horse fittings and weapons, with metalwork closely reminiscent of the remains found at “Xiongnu” sites such as Taohongbala and Yulongtai.18 Moreover, a general survey of the metal inventory of several graves excavated in the same area again confirms the basic nature of this type of inventory, in which we have weapons, horse and

14 Ta La and Liang Jinming, “Hulusitai Xiongnumu”, pp. 11–12.15 Tian Guanjin and Guo Suxin, E’erduosi shi qingtong qi, pp. 222–226.16 Yikezhaomeng wenwu gongzuozhan, “Yikezhaomeng Budonggou Xiongnumu Qingli Jianbao”, pp. 27–33.17 Ningxia Wenwu Kaogu Yanjiusuo and Ningxia Guyuan Bowuguan, “Ningxia Guyuan Yanglang Qingtong

Wenhua Mudi”, pp. 13–56.18 Luo Feng, “Ningxia Guyuan Shilacun Faxian Yizuo Zhanguomu”, pp. 130–131, 142.


chariot fittings, and largely bronze ornaments (although some gold pieces are present, but in very small quantities).19

Another significant site is Yujiazhuang, near Pengpu, in Guyuan county (Ningxia).20 Here twenty-eight tombs have been excavated, with nearly seven hundred bronze objects recovered, and a few items in iron and in gold. This site has been attributed to the Western Rong people, but in fact the metal assemblage is similar to (although not identical with) the type of bronzes found in the Ordos region. Both artifacts and sacrificial remains indicate that the horse was important in their culture. On the other hand, we also find numerous traces of local and independent cultural developments. The dating is confused, and the presence of two different types of burial customs renders the interpretation of the cultural attribution of the site tentative at best, but the general picture is that of a florid northern nomadic culture, whose valuables (weapons, personal ornaments, and horses) were similar to those found in other northern regions and belonged to an elite of horse-riding warriors.

This type of inventory is also broadly similar to what we find at various sites in Qingyang county (Gansu province),21 where horse and chariot ornaments and fittings are even more prominent and some tools show a close typological connection with the above mentioned Taohongbala and with the site of Xichagou in Liaoning (see below). Several findings from sites in Qin’an county (also in Gansu) have also been attributed to the Xiongnu, and include in particular weapons and bronze ornaments similar to findings from the sites of Taohong-bala, Fanjiayaozi, Xigoupan, and Beixinbao.22

Martial symbols, the main markers of status in some nomadic communities, also appear in some later sites. On the eastern side of the Northern Zone, such features can be seen in the pre-Han site of Wudaohezi (Lingyuan, Liaoning province).23 Bronze weapons, orna-ments, and horse fittings predominate. No iron objects have been retrieved, but two golden ornaments may indicate this site’s limited participation in a broader luxury trade, which does not appear to be significant in the definition of elite status. The Xiongnu site of Xicha-gou, a broadly distributed necropolis also located in Liaoning of about five hundred graves dated to the Western Han period (206 BCE-9 CE), features objects of Chinese origin, such as mirrors and coins, but the dominant elements of the funerary inventory are still chariot and horse fittings, and weapons, including a large number of iron swords and spears. No gold or silver objects have been found.24

The burials of the second category display not just a predominance of prestige and wealth symbols, including gold and silver items, but an absence of weapons and horse-related ob-jects. At Guoxianyaozi (Liangcheng, Inner Mongolia) – a burial site of circa twenty tombs dated around 500–400 BCE – we find a large number of bronze plaques (forty-four, in

19 Luo Feng and Han Kongle, “Ningxia Guyuan Jinnian Faxiande Beifangxi Qingtongqi”, pp. 403–418.20 Ningxia Wenwu Kaogu Yanjiusuo, “Ningxia Pengbao yujiazhuang mudi”, pp. 79–107; Zhong Kan, “Guyu-

anxian Pengpu Chunqiu Zhanguo Muzang”, pp. 255–256.21 Liu Dezhen and Xu Junchen, “Gansu Qingyang Chunqiu Zhanguo muzangde qingli”.22 Qin’an xian wenhua guan, “Qin'an xian Linian Chutude Beifangxi Qingtongqi”, pp. 40–43.23 Liaoning sheng Wenwu Kaogu Yanjiusuo, “Liaoning Lingyuanxian Wudaohezi Zhanguomu Fajue Jianbao”,

pp. 52–61.24 Sun Shoudao, “Xiongnu Xichagou Wenhua’ gumujun de faxian”, pp. 25–35.

Nicola Di Cosmo42

animal or geometric style), belt ornaments, buttons, bells, earrings, and various clothing or-naments.25 Beside these, the only other bronze objects are two knives and one pick-axe. The absence of the classic martial and equestrian markers of nomadic elite may indicate that the people here were defining status in terms of wealth represented by rich or rare ornamental objects.

Even more striking examples of ostentatious wealth are the sites that present large num-bers of precious objects, such as Aluchaideng (Ordos), tombs number 2 and 4 at Xigoupan (Ordos), Nalin’gaotu (Shaanxi province) and Shihuigou (Ordos). At Aluchaideng, two graves attributed to the Xiongnu yielded a very large number of luxury items, totaling 218 gold and five silver objects. These include a gold headdress, ornamental gold plaques, includ-ing some set with gems in a design of a tiger and birds, fifty-five animal style plates, forty-five rectangular gold buckle ornaments, and various other ornaments such as buttons, beads, one chain, and a pair of earrings.26

In tomb number 2 at the burial site of Xigoupan, also dated to the third century BCE, we find several gold items, such as two ornamental plaques, a necklace, a pair of earrings, scabbard ornaments, and bird-shaped ornaments.27 Silver ornaments have also been found. At the same site, several luxury items were found in tomb number 4, dated to a later period, possibly early Han (second-first century BCE). Here too jewelry predominates, including a gold headdress, earrings, ornamental plates, strings of beads with gold thread, large gold-rimmed open-work jade pendants, gold belt ornaments, and a buckle. Although identified in the literature as Xiongnu sites, their metal inventory is strikingly different from that of other Xiongnu tombs.

Similar sites have been excavated in Shenmu county, in Shaanxi.28 Only one tomb found in Nalin’gaotu, however, provided a large body of precious goods: gold, silver, and bronze or-naments in open-work, relief, or other forms. These include artifacts typical of the “North-ern” style, such as one gold deer-shape fantastic anima, a silver tiger, and five silver kneeling deer, all in three-dimensional forms; one gilt silver dagger handle; a pair of gold plaques in the shape of a tiger, and various silver plaques, buckles, rings and bronze ornaments. In con-trast, other findings in the same area have yielded more average bronze items.

Finally, the late Warring States site of Shihuigou, has yielded, in a single grave, numerous silver plaques, and gilded bronze ornaments that may have come from the Central Plains. 29 The silver plaques display typical northern motifs, such as a tiger biting a deer and tigers in combat; ornamental buttons, also in silver, include animal combat and geometric designs, and various animal representations. The gilt bronze ornaments, set in iron, represent turtles, kneeling deer, a crane’s head, and a ram’s head. In addition, gold and silver inlaid iron re-mains have been found. Here we also find elements of an equestrian culture, such as chariot

25 Wei Jian, “Liangcheng Guoxianyaozi Mudi”, pp. 57–81.26 Tian Guangjin and Guo Suxin, “Neimenggu Aluchaideng faxiande Xiongnu yiwu”, pp. 333–338, 364, 368.27 Yikezhaomeng wenwu gongzuozhan and Neimenggu wenwu gongzuodui, “Xigoupan Xiongnu mu”, pp.

1–10; Yikezhaomeng wenwu gongzuozhan, and Neimenggu wenwu gongzuodui, “Xigoupan Handai Xiong-nu mudi diaochaji”, pp. 15–26.

28 Dai Yingxin and Sun Jiaxiang, “Shaanxi Shenmuxian Chutu Xiongnu Wenwu”, pp. 23–30.29 Yikezhaomeng wenwu gongzuozhan, “Yijinhuoluoqi Shihuigou Faxiande E'erduosishi Wenwu”, pp. 91–96.


ornaments, gilded saddlecloth ornaments, and two horse bits, but the context is overwhelm-ingly one that privileges the accumulation of precious goods, and the iron-set gilded bronze, as well as the gold and silver inlaid iron objects, may point to Chinese production, or to a technology imported from or influenced by China. Aside from the question of the range of contacts and the stylistic or cultural attribution of the various objects, what is relevant for the purpose of this cursory survey is the concentration of wealth in a single grave, where the symbols of a nomadic warrior’s prowess (his horse and weapons) either disappear or are rendered according to a material hierarchy that provides the object with intrinsic value, apart from its relationship to the life and values of its owner.

What I would like to suggest is that this type of inventory may indicate that the accumu-lation of wealth in the form of luxury objects had become a more important sign of aristo-cratic distinction than the weapons and horse fittings traditionally associated with nomadic steppe leadership. This may in turn point to a deep transformation among some nomadic elites towards a political and economic role quite different from the traditional one, which we assume was centered on military leadership and religious functions, as the weapons, horse fittings and sacrificial cauldrons suggest.

In contrast, the emphatic accumulation of precious objects reflects a “network mode” of elite representation. Nomadic elites became increasingly involved in long-distance contacts, and drew legitimacy and power from their connections with other elites. Exchange of pres-tige items, as well as trade and tribute, became the source of stored wealth that demonstrated and consolidated a lineage’s enduring power. Foreign connections and representations of one’s elite status in terms that would be readily recognized outside one’s community marked a transition, among certain groups, to a symbolic system resembling the “network” rather than the “corporate” mode. This transition, however, should not be placed in a linear evo-lutional pattern. Interaction among nomadic communities and long-distance contacts with the northern and western cultural areas, as well as with the Chinese states to the south, may have stimulated these changes in patterns that are so far unclear. Political events may have played an important role as traditional elites may have been challenged and replaced by others, and the development of commercial routes (as demonstrated, for instance, by the presence of Chinese silk in the later Pazyryk burials) may have contributed to changes in the elites’ value system. However, the diversity in the funerary assemblages provides a strong indication that a transition was taking place among local elites in the northern borderlands of the Zhou community of states, and that this transition may be related to both long-term and short-term political and economic changes.

2.2 Northern Mongolia: Xiongnu elite burials and material culture

Moving to the Xiongnu archaeology north of the Gobi, in today’s Mongolia, implies not just a spatial transition but also a temporal one, as the main sites are dated from the first century BCE, which is later than most Xiongnu sites in today’s China. Arguably the most visible and important display of elite culture in this region can be seen in the burials that emerge around the end of the first century BCE. These are very large and imposing tombs, notable especially for the complexity or “monumentality” of their construction, which are

Nicola Di Cosmo44

significantly different from the circular tombs with elite goods and other features of el-evated status that can be found in the earlier period. Several studies have been produced that give us a fairly reliable overview of the chronology and typology of these burials. The large ramped, rectangular or square burials that have been found in numerous necropolises in central and northern Mongolia and Buriatia, at sites such as Gol Mod, Noyon Uul, and Tsaraam, have been attributed to the Xiongnu and dated between the late first century BCE and first century CE, but especially towards the end of this relatively narrow chronological spectrum. According to archaeologists, the fundamental features of these burials, the overall architecture and the mode of inhumation, reflect a synthesis of two different traditions.30 The construction of the wooden burial chamber is derived from the large “Scythian” tombs in the tradition of the Altai and Pazyryk barrows. The overall architecture, with ramp, ter-raced construction and deep location of the burial chamber, on the other hand, is attributed to influences from China and was possibly even built by Chinese slaves and prisoners of war. The overall composition of the monument includes peripheral or ancillary burial pits that are much smaller and whose function is unclear, but that appear to belong to a cohort of people who accompanied in death the person buried in the main tumulus. These burials are clearly extraordinary monuments, and are meant to represent the very highest members of the nomadic elites. They are quite uniform in style and in terms of the composition of the funerary objects, many of which indicate a relationship with workshops and artisanal centers in China whose products were of a grade suitable for the Chinese nobility.

Archaeologists and historians have been baffled by the sudden emergence of these monu-ments, located within necropolises made of hundreds of large and small graves. The time period coincides neither with the emergence of new political structures (such as the forma-tion of a new empire) nor with the appearance of a new people. The written records do not shed much useful light on the political composition of the Xiongnu empire in the areas where these monuments appear, but two phenomena must be taken into consideration. The first is that the labor-intensive construction, the satellite round burial pits, and the richness and exclusivity of the objects accompanying the dead, point to a self-representation of the Xiongnu elite very different from the southern one. The second is that the location of many such tombs in various parts of Mongolia indicates the absence of a clear territorial hierarchy – these areas were not subordinated to a putative imperial center. That is, various aristocratic or “elite” existed simultaneously and defined themselves, as reflected in the funerary monu-ments, through very similar attributes. A reasonable conclusion would be that, short of find-ing a site that could be identified as a “court” dominating the whole territory, the Xiongnu aristocracy had become fragmented and divided into independent “principalities”.

We do not know whether this change represents the old Xiongnu imperial aristocracy taking on a new political role, or rather the emergence of an altogether new “local” aristoc-racy. In terms of political change, several scenarios should be considered. For instance, just as the crisis that led to Modu’s seizure of power produced a new aristocracy in the Xiongnu empire, the crisis of the civil war among the Xiongnu around the middle of the first century BCE may have stimulated the rise of a new aristocracy among the independent northern

30 Eregzen, “A Comparative Analysis of Xiongnu Noble Tombs and Burials in Adjacent Regions”, pp. 275–284.


Xiongnu. This new aristocracy was more fragmented and autonomous, had a well-defined territorial base, and probably entertained separate relations with the Han court. Alterna-tively, a change in elite culture occurred as the old Xiongnu aristocracy retreated to the North and established power bases that defined themselves in competition with China by displays of wealth and majesty that could compete with the Han dynasty and thus retain authority and prestige in the face of mass Xiongnu desertions and migrations to China. The presence of prestige objects of Chinese origin can be explained only in terms of gifts directly sent (perhaps together with brides, artisans, and engineers) from China to various Xiongnu aristocrats. This type of relationship can be easily explained in terms of the “divide and rule” frontier strategy adopted by the Han dynasty and by Wang Mang during the short interreg-num of the Xin dynasty. What this means is that the Xiongnu ceased to work as an “empire” and devolved into multiple aristocratic centers whose political dynamics are not exactly known but which profited both from the dissolution of a centralized state that absorbed resources now accessible to regional aristocrats, and from the turn of Han politics towards a more effective diplomacy of maintaining a balance of power in the steppe in order to avoid the emergence of centralized “imperial” centers.

At the current state of research, it is difficult to say for how long exactly these elites were able to continue building monumental tombs, but it is likely that they disappeared in conjunction with a decline of the Han dynasty. Contrary to theories claiming that nomadic power rises and becomes more centralized in tandem with the increasing wealth of China, the presence of monumental tombs proves the opposite, namely, that the emergence of re-gional Xiongnu elites reflects not a unified but a divided political space. This development (which is supported by ample documentary evidence) should be attributed to a Chinese policy of investing large amounts of wealth in order to maintain various power centers. The richest and most ostentatious elites are, then, proof of the decline rather than the rise of Xiongnu imperial power, and defy any simplistic association between wealth and power.

A very insightful hypothesis in this regard has been proposed by Ursula Brosseder in her study of Xiongnu elite burials.31 Following Kossack’s theory of ostentatious tombs, she attributes their appearance to a cultural change within the elite due to a particular crisis, identified by her as the split between Northern and Southern Xiongnu of 49 CE. In my view, this is correct, but the phenomenon of regionalization and segmentation of political power among the Xiongnu could be dated even earlier, from around the first century BCE, when the central power of the chanyu began to wane and the crisis deepened, provoking be-tween c. 56 and 36 BCE the fateful split between Huhanye and Zhizhi chanyu. Under Huy-andi chanyu, the Xiongnu became vulnerable to attacks by Wuhuan, Wusun, and Dingling groups, and even the tombs of the chanyus were desecrated by these enemies. The Han also sent various expeditionary forces against the nomads. In other words, the sources present a picture of extreme instability, military weakness, and political fragmentation. Given this sce-nario, the emergence of regional power centers, leading to the redefinition of the Xiongnu upper strata after they lost their imperial and unified dimension, the rise of the nomadic enemies that threatened their supremacy, and the projection of Chinese power in the region

31 Brosseder, “Xiongnu Terrace Tombs and their Interpretation as Elite Burials”, pp. 247–280.

Nicola Di Cosmo46

(political, symbolic and military) may have been responsible for the appearance of new ways in which the Xiongnu elite defined itself from around the turn of the millennium. What we do not know and shall never know is what kind of elite this was – whether it represented the reemergence of tribal groups previously conquered by the Xiongnu and then incorporated within the Xiongnu empire, or the “exploded” members of the Xiongnu elite that had recon-stituted themselves as regional power-holders.

While this historical reconstruction remains at best a preliminary attempt to explain the profound shift in the material record, it may offer insights into the political evolution of the Xiongnu “empire”. Whereas the earlier (southern) elites’ burials are less ostentatious and their prestige goods did not constitute a standardized body of accoutrements, and while they had access to fewer human and material resources, the northern elites display a very dif-ferent type of status consciousness. The precious objects and large burials indicate identical elite taste and attributes, and, possibly, sumptuary regulations as to rank and title. The large number of Chinese prestige items (chariots, silk, mirrors, and exclusive lacquers) may even indicate an important role played by the Han court in determining or confirming the rank and status of these local elites. The emergence of “ostentatious” terraced tombs in Mongolia, therefore, needs to be seen in relation to political events that brought a profound and signifi-cant transformation of social and cultural values in which the presence of Chinese symbols of prestige and impressive landmarks in the steppes played a key role.

To conclude this section on monumental burials, it may be appropriate to clarify a point in the historical sources that has been brought to the attention of archaeologists. In the Shiji, it is stated that the Xiongnu did not build tumuli or large tombs, but this statement was glossed in the Jijie commentary with a note by Zhang Hua 張華 (232–300), the influ-ential scholar who served under the Jin dynasty in the third century CE, who said that the Xiongnu had burial mounds called douluo 逗落 in their language (Shiji 2893). This can be reconstructed as *dow-lak, with a first syllable based on Proto-Mongolian *dobu/*döbe, and pronounced *dow, and the second as the suffix *-lak. This word may therefore be related to Classical Mongolian dobuilγa, meaning a raised earthen mound, or protuberance in the land. If this reconstruction can be accepted, this word allows us to associate a Proto-Mongo-lian word with a Xiongnu artifact that did not exist at the time of Sima Qian, but did exist at the time of Zhang Hua. Leaving aside the question of who exactly were the Xiongnu to whom Zhang Hua was referring, and the exact reconstruction of the Xiongnu word, this tex-tual note proves, in my view conclusively, that the cultural change attested by the creation of large monumental tombs occurred after the time of Sima Qian, when the Xiongnu empire was already undergoing a process of dissolution and decentralization, and possibly before the formal separation between Southern and Northern Xiongnu, in the Eastern Han dynasty.

In addition to the burials, we need to take into consideration other aspects of “elite” attribution in a Xiongnu context to the north of the Gobi. First of all, we should mention that the sudden appearance of monumental graves and the typical Xiongnu round burial pits (regarded as non-elite burials) are preceded in the pre-Xiongnu Mongolian archaeology by “stone-slab” graves (eleventh-third century), that is, rectangular or square pits with sides lined with stone slabs. These are dated to the late Bronze or Early Iron Age periods, and are found prevalently in Western Mongolia; some are also found in Transbaikalia, very few in


other parts of Mongolia, and none in Tuva or the Minusinsk basin. The inventory consists typically of a few bronze or iron weapons, some pottery and parts of sacrificed animals, but here too we can trace some changes.

The site of Egiin Gol, in northwestern Mongolia, has been regarded as an ordinary Xiongnu burial ground. The grave inventory is poor, the ornaments are few, and there are no traces of concentration of power and wealth. Rather, both assemblage and grave distribu-tion seem to indicate a diffused social body. The presence of non-locally produced items, however, indicates access to foreign goods, such as cowries, turquoise and tinned bronze items. The main source of these items is the south: Inner Mongolia and the Northern Zone of China. Horse-riding can be attested perhaps from the end of the second millennium BCE, but there are few weapons and no evidence of extensive warfare. The limited number of precious ornaments and the generally low level of social differentiation seems to indicate a “corporate” rather than a “network” type of local economy.

As mentioned above, the situation changes dramatically from the first century BCE. When first excavated, the Noyon Uul tombs were regarded as kurgans and associated with the nomadic barrows of Central Asia. The funerary assemblage in the most sumptuous and large elite graves includes what can be seen as highly standardized sets of Chinese goods of the Han dynasty, such as chariots, silk, mirrors (always fragmented), and lacquer objects, to-gether with elite goods more typical of a nomadic culture, such as golden artifacts, weapons, and animal-style decorations. A special feature is the custom of satellite single burials, typi-cally in relatively shallow pits (one to two-meters deep), which may be burials of sacrificial victims.

The large burial excavated by Sergey Minyaev at Tsaraam,32 although severely disturbed and looted, contained both Chinese imported goods (lacquer, chariot and mirrors), and iron buckles in the so-called “animal style” covered with gold foil or other precious metals. The monumentality of the burial complex, which includes several satellite burials, indicates, according to Minyaev, an extremely high-ranking figure (perhaps a “khan”), and the goods, which are only prestige and luxury items, are therefore indicative of high political status. Likewise, the large tomb excavated at Takhiltyn Khotgor shows an assemblage of some pre-cious objects in gold, vessels, silk, turquoise insets and other objects, but no weapons. We should note that the gold objects, representing a sun and a crescent, probably have ritual significance, and may even be associated with the aforementioned ritual performed by the chanyu to worship the rising sun and moon.33 Moreover, the general inventory of these large “elite” tombs does not include any weaponry, with the exception of a single arrowhead found in the Andreev Kurgan at Noyon Uul.34

Elite goods can be also found in the smaller “satellite” burials that have a spatial relation-ship to the larger rectangular (terraced) tomb. At Gol Mod 2, the funerary inventory of a larger satellite burial included imported goods, such as lacquer and mirrors from China, a

32 Minyaev and Sakharovskaia, “An Elite Complex of Xiongnu Burials in the Tsaraam Valley”, pp. 71–84.33 Miller, “Permutations on Peripheries in the Xiongnu Empire”, p. 571.34 Brosseder, “Xiongnu Terrace Tombs and their Interpretation as Elite Burials”, p. 264.

Nicola Di Cosmo48

Roman glass bowl, and semiprecious beads.35 Foreign luxury goods are indicative of “elite” status, but exactly what they tell us about the actual social rank is not clear. At Duurlig Nars, in north-eastern Mongolia, the excavation of a large terraced tomb revealed a similarly rich inventory, with goods of high prestige, such as chariot, gold and gilded objects, Chinese mir-rors and ritual bronze vessels (an incense burner with a tray and a cauldron). Once again, no symbols of a military status have been found.

Other Xiongnu undisturbed burials are sometimes also designated as “elite” even though they appear to be much more modest constructions than the terraced tombs – while also larger and deeper than “ordinary” Xiongnu burial pits – such as the barrows at Khökh Üzüürin Dugui-II in Western Mongolia. Here, Chinese bronze vessels dated to the Han dynasty were found together with a bronze cauldron that shows possible connections with Kazakhstan.36 These objects may be related to ritual functions, and may thus belong to a different type of elite, not political but perhaps religious. A combination of remains that we may find to be more typical of a warlike nomad comes from two wooden coffin burials at Shombuuziin-Belchir,37 which include iron horse bits, an iron spearhead, arrowheads, and bows. These burials had been looted and in some there are traces of Chinese lacquer, and while these burials should also be inscribed under the rubric of “elite” on account of the luxury goods, their shape and construction indicate a different type of elite attribution from the larger monumental burials.

As pointed out by Nelson et al., the systematic looting and disruption of Xiongnu cem-eteries in antiquity may be related to the desecration of Xiongnu elite tombs (specifically the tombs of the chanyu) by the Wuhuang, which we have read of in the sources.38 But whether the archeological record can actually clarify these issues is doubtful. At present we can only register the development of a culturally homogeneous upper elite, distinguished by access to Chinese prestigious goods exclusively associated with monumental tombs. Other forms of elite status may be seen in some burials displaying objects associated with a military or ritual function.

Finally, I would like to address the issue raised by recent genetic studies of the Xiongnu elite as a biologically and potentially ethnically mixed social stratum. Korean and Mongolian scientists have analyzed the ancient DNA of an adult male buried in what is regarded as an elite tomb in north-eastern Mongolia (Duurlig Nars). 39 The results show that the person in question belonged to a “European” haplogroup (R1a1) common among the nomads of western Eurasia. The authors hypothesize that this particular specimen may be related to

35 Erdenebaatar et al., “Excavations of Satellite Burial 30, Tomb 1 Complex, Gol Mod 2 Necropolis”, pp. 311–313.

36 Kovalev, Erdenebaatar and Iderkhangai, “An Unlooted Xiongnu Barrrow at Khökh Üzüürin Dugui-II, Bul-gan Sum, Khovd Aimag, Mongolia”, pp. 291–302.

37 Miller et al, “Xiongnu Constituents of the High Mountains: Results of the Mongol-American Khovd Archae-ology Project”, pp. 9–13; Reisinger, “New Evidence About Composite Bows and their Arrows in Inner Asia”, pp. 45–58.

38 Nelson, Honeychurch and Amartüvshin, “Caught in the Act: Understanding Xiongnu Site Formation Pro-cesses at Baga Gazaryn Chuluu, Mongolia”, p. 227.

39 Kijeong Kim et al., “A Western Eurasian Male Is Found in 2000-Year-Old Elite Xiongnu Cemetery in North-east Mongolia”, pp. 429–440.


migrations from the west across Central Asia and Siberia of the Andronovo, Karasuk, and Tagar periods, eventually infiltrating Transbaikalia, and they conclude that the Xiongnu empire included a biologically diverse population. However, of the particular set of tombs examined, one is a large Xiongnu terraced tomb, which is extremely rich, but the other two, including the burial with the west Eurasian male, are distinctly smaller rectangular ones. The funerary goods buried with the “west Eurasian male” (Tomb no. 3) are two round jars, a lamp, a Chinese bronze mirror, gold foil, gold belt ornaments, an antler, a bronze ring, and an iron arrowhead. The time period is also different, as the smaller tombs are regarded as dat-ing back to 300–100 BCE while the Xiongnu tomb is dated to 100 BCE-100 CE, and most likely to the latter end of this range. Dating, shape of the burial, and material inventory do not belong, in my view, to a “Xiongnu” archeological type. Rather, the site seems to include tombs from different periods, and the earlier ones may belong to a pre-Xiongnu “Scytho-Siberian” population, akin to the population of south Siberia, and the Altai region. Thus, the conclusion that this particular person belonged to the Xiongnu elite should be revised, and the question of the possible incorporation of local “Saka” elites within the Xiongnu upper class therefore remains a moot point on the archaeological front.

There are, however, other sites in which preliminary research indicates that the Xiongnu, as they expanded into areas inhabited by other nomadic cultures, and in particular in the northwestern extension of their empire (Altai, Tuva, and western Mongolia), did not com-pletely replace the local population, and a degree of cultural fusion may have occurred.40 Other studies about the Xiongnu population have yielded results that are not useful, for the time being, to assess whether the Xiongnu elite was ethnically or biologically diverse.41 These findings nonetheless raise further questions concerning the nature of the Xiongnu expansion into diverse cultural areas and ultimately about the shape of their empire and its legacy in Inner Asian history.

3. Conclusion

We cannot assume that we can speak of “elites” in history and archeology as if they were interchangeable concepts. An archeological “elite” object, such as a burial, cannot be trans-ferred automatically onto a historical context by claiming a specific status for the person associated with that particular structure or artifact. Likewise, the various “elites” whose exist-ence is documented in the sources may never show up in the material record in a way that al-lows us to identify specific political positions, not to mention specific individuals. Moreover, changes in the elites’ status ascription, symbolic representation and cultural change cannot be traced through the historical record at the present state of research. One of the critical limitations of the textual sources, which is evident in Han historiography, is the difficulty in discerning whether the historian, in his description of a nomadic society, is following clichés and normative modes of representation, introducing fresh information derived from contemporary observational material of his own, or even importing narratives derived from

40 Leus, “New Finds from the Xiongnu Period in central Tuva. Preliminary Communication”, pp. 514–536.41 Lee and Zhang Linhu. “Xiongnu Population History in Relation to China, Manchuria and the Western

Regions”, pp. 192–200.

Nicola Di Cosmo50

the foreign society’s culture, such as legends and myths. On the other hand, “elite status” in the archaeological record is, at best, a relative concept inferred according to the parameters specific to each context. Moreover, the nature of what could be perhaps defined as a Xiongnu “phenomenon” prevents any absolute identification of a specific Xiongnu elite, especially in the southern region of Xiongnu archaeology. In the north, a clearer elite status is closely con-nected with the appearance of terraced tombs, while the relationship between these Xiongnu aristocrats and other coeval and later nomadic elites will probably remain unclear for some time.

In this essay two separate notions of “elite” have been investigated, on the basis of docu-mentary sources and material culture, respectively. The study of text-based descriptions and features of the Xiongnu aristocracy has led to an attempt to reconstruct, in broad terms, the political organization of the Xiongnu empire, its territorial administration and the incorpo-ration of local and foreign members into its hierarchies. The study of the symbolic repre-sentation of elite status and cultural change through the material record, on the other hand, has led to some considerations about cultural change among these elites. Deep differences in heuristic methods and interpretive strategies, and of course in the quality of the information gathered from textual or archaeological sources, have not produced mutually supporting or converging hypotheses except to a very limited extent. In the end, an investigation of Xiongnu elites highlights the “bifurcated” nature of the Xiongnu phenomenon, whereby it acquires a different ontology depending on its historical or material dimension.

However, awareness of the limitations inherent in the knowledge gleaned from various types of evidence should not make our study of ancient nomadic elites frustratingly con-strained and forever dichotomized. A better understanding of both historical and material processes, informed by methods of analysis specific to each field, may provide new insights and thus generate questions leading to a general advancement of our knowledge of the his-tory of the steppe region at the time of the Xiongnu empire.

ReferencesAkiyama Shingo, “Nei Menggu gaoyuan de Xiongnu muzang”, in Wei Jian ed., Nei Menggu wenwu kaogu

wenji, vol. 2 (Beijing: Zhongguo dabaike, 1997), pp. 375–92. Bemmann, Jan, “Was the Center of the Xiongnu Empire in the Orkhon Valley?”, in Ursula Brosseder and

Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspectives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 441–61.

Blanton, Richard E., Gary M. Feinman, Stephen A. Kowalewski, and Peter N. Peregrine et al., “A Dual-Processual Theory for the Evolution of Mesoamerican Civilization”, Current Anthropology 37(1) (1996), pp. 1–14.

Brosseder, Ursula, “Xiongnu Terrace Tombs and their Interpretation as Elite Burials,” in Jan Bemmann et al., eds, Current Archaeological Research in Mongolia. Papers of the First International Conference on “Archaeological Research in Mongolia”, held in Ulanbaatar, August 19th-23rd 2007 (Bonn: Vor-und Früh-geschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2009), pp. 247–280.

Dai Yingxin and Sun Jiaxiang, “Shaanxi Shenmuxian Chutu Xiongnu Wenwu” (Xiongnu artefacts exca-vated in Shenmu County, Shaanxi), Wenwu 12 (1983), pp. 23–30.


Erdenebaatar, Diimaazhav et al., “Excavations of Satellite Burial 30, Tomb 1 Complex, Gol Mod 2 Ne-cropolis”, in Ursula Brosseder and Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspec-tives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 303–314.

Eregzen, Gelegdorzh, “A Comparative Analysis of Xiongnu Noble Tombs and Burials in Adjacent Regions”, in Ursula Brosseder and Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspectives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 275–284.

Feiman, Gary M., “Corporate/Network: New Perspectives on Models of Political Action and the Puebloan Southwest”, in M. B. Schiffer, ed., Social Theory in Archaeology (Salt Lake City: University of Utah Press, 2000), pp. 31–51.

Grjaznov, Michail P., Der Groβkurgan von Aržan in Tuva, Südsibirien (München: Verlag C. H. Beck, 1984).Hanks, Bryan, “Archaeology of the Eurasian Steppes and Mongolia”, Annual Revue of Anthropology 39

(2010), pp. 469–486.Hanshu (History of the Former Han Dynasty) by Ban Gu (Beijing: Zhonghua shuju, 1962 [1990]).Höllmann, Thomas O., and Georg W. Kossack, Maoqinggou: Ein eisenzeitliches Gräberfeld in der Ordos Re-

gion (Innere Mongolei) (Mainz : Materialien zur Allgemeinen und Vergleichenden Archäologie, 1992).Hulsewé, Anthony F. P., with an Introduction by M. A. N. Loewe, China in Central Asia: The Early stage

125 B. C./ - 23 A. D: An Annotated Translation of Chapters 61 and 96 of the History of the Former Han Dynasty (Leiden: Brill, 1979).

Kijeong Kim, et al., “A Western Eurasian Male Is Found in 2000-Year-Old Elite Xiongnu Cemetery in Northeast Mongolia”, American Journal of Physical Anthropology 142 (2010), pp. 429–440.

Kovalev Alexei, Diimazhaav Erdenebaatar, and Tömör-Ochir Iderkhangai, “An Unlooted Xiongnu Barrrow at Khökh Üzüürin Dugui-II, Bulgan Sum, Khovd Aimag, Mongolia: Relative Chronological Dating and Its Significance for the Study of Xiongnu Burial Rites. Preliminary Report”, in Ursula Brosseder and Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspectives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 291–302.

Lee Christine and Zhang Linhu “Xiongnu Population History in Relation to China, Manchuria and the Western Regions”, in Ursula Brosseder and Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspectives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 192–200.

Leus, Pavel M., “New Finds from the Xiongnu Period in Central Tuva: Preliminary Communication”, in Ursula Brosseder and Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspectives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 514–36.

Li Yiyou, “Neimenggu Helinge’er xian chutude tongqi” (Bronze objects excavated n Heling’er county, Inner Mongolia), Wenwu 6 (1959), p. 79.

Liaoning sheng Wenwu Kaogu Yanjiusuo, “Liaoning Lingyuanxian Wudaohezi Zhanguomu Fajue Jianbao” (Excavation of the Tombs from the Warring States at Wudaohezi, Linyuan County, Liaoning), Wenwu 2 (1989), pp. 52–61.

Lin Gan, Xiongnu shiliao huibian (Historical sources on the Xiongnu) (Beijing: Zhonghua shuju, 1988).Liu Dezhen, and Xu Junchen, “Gansu Qingyang Chunqiu Zhanguo muzangde qingli” (Explanation of

burials of the Spring and Autumn and Warring States Periods at Qingyang, Gansu), Kaogu 5 (1988), pp. 413–424.

Luo Feng, “Ningxia Guyuan Shilacun Faxian Yizuo Zhanguo mu” (A Warring States Burial Excavated at Shilacun, Guyuan, Ningxia), Kaoguxue Jikan 3 (1983), pp. 130–131, 142.

Luo Feng, and Han Kongle, “Ningxia Guyuan Jinnian Faxiande Beifangxi Qingtongqi” (Recently Found Bronze Objects from Guyuan, Ningxia), Kaogu 5 (1990), pp. 403–418.

Nicola Di Cosmo52

Miller, Bryan K. et al., “Xiongnu Constituents of the High Mountains: Results of the Mongol-American Khovd Archaeology Project”, The Silk Road 7 (2009) pp. 8–20.

Miller, Bryan K. “Permutations on Peripheries in the Xiongnu Empire”, in Ursula Brosseder and Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspectives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 559–578.

Minyaev, S. S., and L. M. Sakharovskaia, “An Elite Complex of Xiongnu Burials in the Tsaraam Valley”, Anthropology and Archaeology of Eurasia 46(4) (2008), pp. 71–84.

Nei Menggu wenwu gongzuodui, “Maoqinggou Mudi” (The Cemetery at Maoqinggou), in Tian Guangjin and Guo Suxin, eds, E’erduosi shi qingtong qi (Ordos Bronze Wares) (Beijing: Wenwu chubanshe 1986), pp. 227–315.

Neimenggu Bowuguan, and Neimenggu Wenwu Gongzuodui, “Neimenggu Zhunge’erqi Yulongtaide Xiongnumu” (Xiongnu Burials at Yulongtai, Zhunggar Banner, Inner Mongolia), Kaogu 2 (1977), pp. 111–114.

Nelson, Albert Russell, William Honeychurch, and Chunag Amartüvshin, “Caught in the Act: Understand-ing Xiongnu Site Formation Processes at Baga Gazaryn Chuluu, Mongolia”, in Ursula Brosseder and Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspectives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 213–227.

Ningxia Wenwu Kaogu Yanjiusuo, “Ningxia Pengbao Yujiazhuang mudi” (The cemetery at Pengbao, Yuji-azhuang, Ningxia), Kaogu xuebao 1 (1995), pp. 79–107.

Ningxia Wenwu Kaogu Yanjiusuo, and Ningxia Guyuan Bowuguan, “Ningxia Guyuan Yanglang Qingtong Wenhua Mudi” (A bronze culture cemetery at Yanglang, Guyuan, Ningxia), Kaogu xuebao 1 (1993), pp. 13–56.

Pan Ling, “A Summary of Xiongnu sites within the northern periphery of China”, in Ursula Brosseder and Bryan K. Miller, eds, Xiongnu Archaeology: Multidisciplinary Perspectives on the First Steppe Empire in Inner Asia (Bonn: Vor-und Frühgeschichtliche Archäologie Rheinische Friedrich-Wilhelms-Universität Bonn, 2011), pp. 463–474.

Parzinger, Hermann, Die frühen Völker Eurasiens. Vom Neolithikum bis zum Mittelalter (München: C. H. Beck, 2006).

Reisinger, Michaela R., “New Evidence About Composite Bows and their Arrows in Inner Asia”, The Silk Road 8 (2010), pp. 42–62.

Qin’an xian wenhua guan, “Qin’an xian Linian Chutude Beifangxi Qingtongqi” (Northern-style bronze objects excavated at Linian. Qin’an county), Wenwu 2 (1986), pp. 40–43.

Shepard, B. A., “Political Economic Reorganization among Non-state Societies: A Case Study Using Mid-dle Holocene Mortuary Data from the Cis-Baikal, Russia”, Journal of Anthropological Archaeology 31 (2012), pp. 365–380.

Shiji (Records of the Grand Historian) by Sima Qian (Beijing: Zhonghua shuju, 1982 [1992]).Sun Shoudao, “Xiongnu Xichagou Wenhua gumuqun de faxian” (Excavation of a Group of Ancient Tombs

of the Xichagou Culture of the Xiongnu), Wenwu 8–9 (1960), pp. 25–35.Ta La, and Liang Jinming, “Hulusitai Xiongnu mu” (Xiongnu burials at Hulusitai), Wenwu 7 (1980), pp.

11–12.Tian Guangjin and Guo Suxin, eds., E’erduosi shi qingtong qi (Ordos Bronze Wares) Beijing: Wenwu chu-

banshe 1986. Tian Guangjin, “Jinnianlai Nei Menggu diqu de xiongnu kaogu” (Xiongnu archaeology in the Inner Mongolian region in recent years) Kaogu xuebao 1 (1983), pp. 7–24.

Tian Guangjin, “Taohongbalade Xiongnumu” (The cemetery at Taohongbala), Kaogu Xuebao 1 (1976), pp. 131–142.


Tian Guangjin, and Guo Suxin, “Neimenggu Aluchaideng faxiande Xiongnu yiwu” (Xiongnu remains discovered at Aluchaideng, Inner Mongolia), Kaogu 4 (1980), pp. 333–338, 364, 368 [Rpt: O’erduosi Qingtongqi, pp. 342–350].

Wei Jian, “Liangcheng Guoxianyaozi Mudi” (The Guoxianyaozi Cemetery in Liangcheng County), Kaogu xuebao 1 (1989), pp. 57–81.

Yikezhaomeng wenwu gongzuozhan, “Yikezhaomeng Budonggou Xiongnumu Qingli Jianbao” (Report on the Xiongnu tomb at Budonggou, Ih Ju League), Neimenggu Wenwu Kaogu 1 (1981), pp. 27–33.

Yikezhaomeng wenwu gongzuozhan, “Yijinhuoluoqi Shihuigou Faxiande E’erduosishi Wenwu” (Ordos-style Artifacts discovered at Shihuigou, Eijin Horo Banner), Neimenggu Wenwu Kaogu 1–2 (1992), pp. 91–96.

Yikezhaomeng wenwu gongzuozhan, and Neimenggu wenwu gongzuodui, “Xigoupan Xiongnu mu” (Xiongnu burials at Xigoupan), Wenwu 7 (1980), pp. 1–10 [Rpt. as “Xigoupan Zhanguomu”, O’erduosi Qingtongqi, pp. 351–365].

Yikezhaomeng wenwu gongzuozhan, and Neimenggu wenwu gongzuodui, “Xigoupan Handai Xiongnu mudi diaochaji” (Investigation of the Han dynasty burial ground of the Xiongnu at Xigoupan), Nei-menggu wenwu kaogu 1 (1981), pp. 15–26.

Zhong Kan, “Guyuanxian Pengbao Chunqiu Zhanguo Muzang” (The Spring and Autumn and Warring States burials at Pengbao, Guyuan County), in Zhongguo Kaoguxue Nianjian 1988 (Beijing, 1989), pp. 255–56.

Aristocratic elites in the Xiongnu empire

Documents