Introduction
Although many scholars consider the emergence of the Chinggisid power structure to have been a watershed in Eurasian political and social transformation because it appeared to have replaced the region's ‘tribal’ order with a highly centralised state, some still regard it as a ‘supercomplex chiefdom’.Footnote 2 Indeed, with the collapse of the Mongol Empire, Eurasia is believed by many to have reverted to a ‘tribal’-based social and political order.Footnote 3 Thus, according to this paradigm, the existence of centralised state remained transient in pre-modern Eurasia, if not entirely elusive, and the ‘tribal’ order remained resilient, if not always dominant.
However, David Sneath rejects this paradigm as “colonial-era misrepresentations, rooted in nineteenth century Eurocentric evolutionist theory” and argues that:
in Inner Asia many of the forms of power thought to be characteristic of states actually existed independently of the degree of overarching political centralization. . .The local power relations that since ancient times have made the Inner Asian state possible were reproduced with or without an overarching ruler or central ‘head’. . .It was not ‘kinship society’ but aristocratic power and state-like processes of administration that emerged as the more significant features of the wider organization of life on the steppe. . .The political relations of aristocrats determined the size, scale, and degree of centralization of political power. . .The centralized ‘state’, then, appears as one variant of aristocracy.Footnote 4
Furthermore, Sneath argues “the recognition that stratification and the state relation are not dependent upon a centralised bureaucratic structure makes it easier to discern the substrata of power, the aristocratic order that lay at the base”.Footnote 5 Thus, according to Sneath, pre-modern Inner Asian societies were not tribal but aristocratic and, their ‘states’, whether centralised or de-centralised or even ‘headless’, were aristocratic political structures. Thus, Sneath views what many scholars see as the state or the centralised state (such as a Chinggisid political structure) “as a variant of aristocracy”, that is, an aristocratic, probably centralised, state. Although Sneath elaborates very little on the origin, nature and transformation of his aristocratic order, his scheme leaves an impression that aristocracy was always present and it is the aristocracy that produces the centralised state rather than the other way round.
However, more conventional scholarship maintains that the “medieval Eurasian nomadic tribe was a political organism open to all who were willing to subordinate themselves to its chief and who shared interests with its tribesmen”.Footnote 6 Scholars of medieval Mongolia mostly subscribe to this view and Thomas Allsen for instance, writes that:
though defined in genealogical terms, the lineage and the tribe were essentially political entities composed of individuals whose ties of blood were more often fictive than real. In the steppe, common political interest was typically translated into the idiom of kinship. Thus, the genealogies of the medieval Mongols (and other tribal peoples) were ideological statements designed to enhance political unity, not authentic descriptions of biological relationships.Footnote 7
Similarly, Peter Golden emphatically confirms that “Much of the modern scholarship on the Eurasian nomads long ago recognized that ‘tribe’ and ‘clan’ were complex phenomena, involving the political integration, often unstable, of heterogeneous elements”Footnote 8 , while Thomas Barfield asserts that
contrary to his [Sneath's] assertion that anthropologists remain wedded to a genealogical conical clan model, the more common view is that such Inner Asian confederations were the products of reorganization enforced by division from the top down rather than alliance from the bottom up. It has long been accepted that ‘actual’ kinship relations (based on principles of descent, marriage, or adoption) were evident mostly within smaller units: nuclear families, extended household.Footnote 9
However, Golden still believes that “kinship. . .clan and tribe. . .must have been the building blocks on which later expanded and no longer stricto senso kinship-based political structures were constructed” and maintains “that Sneath's thesis may be feasible for the late Chinggisid Mongol world”.Footnote 10 Yet, he does “not find it a useful tool for assessing the pre-Chinggisid steppe polities”.Footnote 11 Similarly, Barfield not only finds Sneath's model ‘problematic’ for projection onto the pre-Qing period but also maintains that such “hierarchy was much weaker or non-existent among nomadic pastoral societies” outside of Inner Asia.Footnote 12 In Barfield's scheme, the tribal order is indigenous, intrinsic or essential to Eurasia while the state is an exception.Footnote 13 Thus, Barfield boldly claims that: “The tribal organization never disappeared at the local level”.Footnote 14
Indeed, scholars who subscribe to the ‘tribal paradigm’ seem to take ‘tribe’ as an essential or natural political organisation of pre-modern nomads, although many still see ‘tribe’ as an extension of kinship organisation. ‘Tribe’ for them is a higher level of kinship incorporation, albeit more complex and more abstract, involving many fictive elements.Footnote 15 Thus, there was no need to look at the origin or genesis of ‘tribes’. The ‘tribes’ were the essential socio-political units of nomadic Eurasia. Thus, we have two different interpretations of pre-modern Eurasian political structure: tribal non-state and aristocratic state.
In this article, however, I would like to look at the origin or genesis of the named categories, that is, the ‘tribes’ or the “aristocracy-led named groups”, as Golden puts it, in order to see whether pre-modern Eurasia's political order was tribal or aristocratic.Footnote 16 Thus, my questions are: where did these named categories come from, and how were they formed? Furthermore, in terms of political order, how principally different was the pre-Chinggisid Mongolia from that of the pre-Qing Mongolia? To be sure, not only is raising the question of the origin or genesis of ‘tribal’ and ‘aristocratic’ controversial but providing a well-grounded answer is extremely difficult if not virtually impossible. Furthermore, addressing this question by comparing the Liao, the Chinggisid and the Qing incorporations of Inner Asia would certainly raise a number of doubts.Footnote 17 Therefore, to introduce the necessary caveats, I should add that what follows is not only unconventional and controversial but also conjectural and speculative. In all probability, this analysis will not be useful for understanding pre-Xiongnu (or Khunnu) Inner Asia. However, it is, I think, applicable to post-Xiongnu Inner Asia, and is certainly appropriate for post-Chinggisid Eurasia. Moreover, although the scheme does not pretend to explain every single case, I believe it to be valid in many of the historical cases found in the study of post-Xiongnu Eurasia.
First, I will examine the Qing incorporation of Mongolia to show how the process transformed Mongolia as a whole, as well as Mongolia's former tümens and otogs, into categories of quasi or non-political and sub-state (or sub-imperial) importance by demolishing their political structures and incorporating them into Qing politico-administrative structures. I will also show how the Qing incorporation of Mongolia created the relevant vocabularies of quasi-political population categories and sub-state or ‘under-state’ administrative ones. Then I will briefly examine the language of the Chinggisid incorporation to show the similarities between the Chinggisid and the Qing incorporations. Next, I will examine pre-Chinggisid Inner Asia to reveal the origins and transformations of the relevant polities and to compare the pre-Chinggisid political order with that of the pre-Qing period. Finally, I will provide a brief analysis of the nature of what I call “a military-oriented numerical hereditary divisional system” and its significance to the origins of ‘tribes’ or ‘aristocratic orders’.
Imperial Incorporations in Comparison
In this section I will compare three successive imperial incorporations of Inner Asia, the Liao, the Chinggisid and the Qing, to show how these imperial projects consistently employed hereditary numerical divisions in incorporating Inner Asia into their realm. Starting with the Qing incorporation of Mongolia, I will move on to the Chinggisid and the Liao incorporations regressively to emphasize the transformations and continuities of hereditary divisional systems and the political order.
The Qing Incorporation of Mongolia: The Creation of Khoshuus and Ayimaqs
By the middle of the nineteenth century, the Mongols had come to be known in European scholarship as a ‘race’ or ‘nation’ consisting of numerous ‘tribes’ such as Chakhar, Khalkha, Khorchin and so on. Thus, for example, Robert Latham, in his 1859 Descriptive Ethnology,Footnote 18 provides a lengthy description of numerous Mongolian ‘tribes’ and assures that:
in no part of the world, Arabia itself not excepted, is the tribal system more developed than in Mongolia. The connection between the members of a tribe is that of blood, pedigree, or descent; the tribe itself being, in some cases, named after a real or supposed patriarch.Footnote 19 The tribe, by which term we translate the native name aimauk, or aimak, is a large division, falling into so many kokhum [khoshuu] or banners.Footnote 20
So, the belief that pre-modern Mongolia had always consisted of numerous ‘tribes’ established itself as the dominant paradigm for the understanding of pre-modern Mongolia and since than this paradigm has dominated the field. Even the most vociferous critic of tribalism Morton Fried himself conceded that “secondary tribalism existed as the prelude to the rise of the massive secondary state[s]” in Eurasia such as Xiongnu, Turk, Mongol and Manchu Empires.Footnote 21
Indeed, under the Qing, all of the Mongols were divided into some 200 khoshuus (or banners). While all of the khoshuus of the Qing ‘External Mongolia’ (gadagadu mongqol)Footnote 22 were the hereditary territorial domains of Mongol ruling princes, some of the Mongol khoshuus that were not part of this ‘External Mongolia’ were divisions governed by Qing-appointed officials, such as Chakhar. While most of the khoshuus of External Mongolia were grouped into chuulgan (aka league), many were not. Chuulgan was an assembly of the ruling princes of the several khoshuus grouped together by the Qing. They held an assembly once in three years at a fixed locality after which the chuulgan was named.
At the same time, Mongolia was divided into nearly 40 ayimaqs. Most of these ayimaqs had their own distinctive names, while six of them shared a common name.Footnote 23 While all the ayimaqs had clearly demarcated territories, four of them had nominal khans after whose titles they were named. While some of the ayimaqs were formed of but a single khoshuu and were grouped into chuulgan with several other ayimaqs, most were formed of two or more khoshuus, while some larger ayimaqs were divided into more than 20 khoshuus and were formed into chuulgans in themselves. Yet, if we take khoshuu and chuulgan as parts of the Qing governing structure, we can see that ayimaq was clearly not. It is these ayimaqs that became known as ‘tribes’. What, then, were they?
Most of the names of these ayimaqs do not appear in the sources of the thirteenth and fourteenth centuries. Instead, their names started to appear in Ming sources.Footnote 24 In the Mongolian sources of the late sixteenth and the early seventeenth centuries, they appear as tümens and otogs or khoshuus. According to these sources, pre-Qing Mongolia consisted of tümens and otogs (or khoshuus at least in the Khalkha and the Altan Khan's codes).Footnote 25 The tümen and otog were hierarchically nested divisions of Mongolia: that is, while Mongolia was divided into tümens, these tümens were, in turn, divided into otogs (or khoshuus). In fact, on the eve of the Qing incorporation, eastern Mongolia consisted of six tümens divided into 54 otogs while western Mongolia was consisting of four tümens and numerous otogs.Footnote 26 All the tümens and otogs had distinctive names, designated territories, governing princes and a multitude of officials, and they shared common governing institutions and legal codes and so on. However, because the Dayan Khanid princes apportioned these tümens and otogs as hereditary shares (qubi) or appanages among themselves, these divisions became autonomous princely domains in themselves.Footnote 27 Thus, by the early seventeenth century, Mongolia with its fragmented sovereignty and decentralised political power – where the sovereignty of the Great Khan was nominal, while the khans and khung taijis or the territorial princes were the effective rulers of their immediate domains – was a fragmented, decentralised realm similar to the Holy Roman Empire.Footnote 28
Consequently, by the early seventeenth century some of these tümens were referred to in Mongolian chronicles as ulus.Footnote 29 The term ulus was used for princely domains of various sizes and is best understood as meaning the political community established by the state or rulership project.Footnote 30 In the documents detailing Manchu–Mongolian relations and in early Qing documents all of the tümens and some of the otogs were called ulus in Mongolian and gurun in Manchu, indicating a certain level of political autonomy. Yet, with the declaration of the Qing Empire, all of them were uniformly designated ayimaq.Footnote 31 Indeed, until the Qing rule of Mongolia, as Christopher Atwood has demonstrated, ayimaq was never used to designate intra-Mongol divisions or categories.Footnote 32 Instead, it was mainly used to denote confessional/religious and as well as temporary task units. Then, how did the former tümens and otogs come to be known as ayimaqs? What made them ayimaqs? What was an ayimaq?
As the sources show, the Manchu rulers readily recognised most of the Mongolian tümens as well as many otogs as uluses before Ligdan's demise. This recognition of Mongolian tümens as uluses was aimed at making them independent from the Mongol Great Khan.Footnote 33 In the process, however, the Mongolian Great Khan was removed and Mongolia was dismantled as a realm. By claiming the imperial throne left vacant by Ligdan's demise, Hong Taiji successfully persuaded his Mongol noble allies to unite under his sway. In adopting and embracing the rhetoric of Mongol unification, Hong Taiji and the Mongol nobles had to put the emergent Mongol uluses (as potentially autonomous princely domains) back into common Mongol fold.Footnote 34 In doing so, Hong Taiji and the Mongol nobles, wary of reviving Mongolian political hierarchy and integration, adopted the term ayimaq from a Manchu precedence and cognate (aiman) to embrace the emergent uluses as equal divisions or subdivisions of Mongolia, thus transcending their former tümen and otog hierarchies. Furthermore, Mongol tümen–ulus–ayimaqs were incorporated into the Qing Empire as ‘external vassal domains’ (gadagadu muji (literally external province) or gadagadu ayimaq (literally external divisions) or tulergi golo (literally external province) or wai-fan (literally external fiefdom), and just as ‘External Mongolia’ (gadagadu mongqol) was established as the dominant designation for the Manchu concept of tulergi golo, so ayimaq was established as the term for its divisions.Footnote 35 Thus, the dismantlement of Mongolia as an independent polity and its incorporation by the Qing as discrete vassal princely domains transformed the tümens and otogs into ayimaqs.
What, then, was an ayimaq? Certainly, ayimaqs were divisions or subdivisions of External Mongolia, a peripheral territory of the Qing Empire.Footnote 36 At the same time, as vassal princely domains of the Qing Empire, ayimaqs were subdivisions of the Qing Empire and as such the ayimaqs were subordinate or inferior to the imperial state units. Yet, as ‘external vassal domains’, ayimaqs enjoyed a degree of status and autonomy, including a degree of self-rule, a distinctive name, ‘territorial integrity’ and, the ability to form chuulgan assemblies and the privilege of special entries in imperial registers. However, to win the allegiance of the Mongol nobles and to break the power of the larger tümen–ulus–ayimaqs, Hong Taiji and subsequent Qing emperors continued to recognise the hereditary ruling claims of the minor Mongol nobles over their immediate subjects and territories by recognising them as hereditary ruling princes (jasaq) and institutionalising their domains as khoshuus. All of the larger ayimaqs were divided into a number of khoshuus and no larger ayimaq was left with a single overarching ruler. Instead, each khoshuu was an autonomous domain independent from all others. The power and authority of former rulers of the ayimaqs (or tümen and otog) was effectively reduced to their own immediate khoshuus, though all of them enjoyed higher titles, ranks and annuities and some even retained their titles of khan. Thus, there was no real integration and rulership above the khoshuu level in the larger ayimaqs. This effectively resulted in the dismantling of the tümen–ulus–ayimags. The former tümens were, to be sure, the largest of these units. However, small ayimaqs formed by a single khoshuus survived as a single politico-administrative unit in the form of the new Qing khoshuu. The larger ayimaqs, however, became more or less headless realms.
At the same time, in place of phrases such as mongqol ulus or mongqol irgen, the Qing deployed the single, simple category ‘Mongols’ (mongqolchud or mongqolud) to designate Mongolia's population.Footnote 37 Thus, the Qing uniformly categorised the population of Mongolia simply as ‘the Mongols.’ Since the ayimaqs (as uluses) had been emergent kingdoms and princedoms before their incorporation into the Qing, we would expect their populations to have developed distinct categorical identities, bearing their ayimaqs’ names. However, this does not appear to be the case. We do not find any mention of Khorchins, Kharachins, Tümeds or Ordoses among the population of the Qing External Mongolia (contemporary Inner Mongolia) in the sources, referred to in the way that the ‘Mongols’ (mongqolchud or mongqolud) are. Instead, we find only references to “the Mongols of. . .” Khorchin, Kharachin, Tümed and Ordos.Footnote 38 We do, however, find a few instances where the populations of Khalkha, Oyirad and Chakhar are referred to as Khalkhas, Oyirads, and Chakhars.Footnote 39 The Chakhar banners were not part of Qing External Mongolia, and Khalkha existed as an independent power until the end of the seventeenth century; indeed, even thereafter Khalkha remained relatively separate and its khans retained their titles too. Certainly, the Oyirads were able to maintain their independence for the longest period of time. Yet, the majority of the sources still talk of the ‘Mongols of Khalkha’ and so on. We also do not find any origin myth or any cultural description of the ayimaqs comparable to what we find with regard to the Mongols.Footnote 40 All of the princes and princely lineages were, however, invariably identified by their ayimaqs. Indeed, the Iletgel Shastir actually equates ayimaq to noble houses. As this work is a genealogy of the Mongol princes that extensively integrates information from the Da Qing Yi tong zhi 大 清 一 統 志, it fuses ayimaqs and khoshuus with their ruling lineages. Thus we read statements such as “Khorchin's ancestor (övög) is Khavt Khasar, younger brother of Yuan Taizu” or “Tsitsig. . . is the ancestor of the two khoshuus of Tüsheet khan Uuba and Zasagt jün wang Budach. . . Namsrai. . .is the ancestor of the three khoshuus of Darkhan chin wang Manzushir, Bint jün wang Khongor and beile Dongor” or “Among the external Mongols, there are two who are named Dörwöd. Though the name is identical, descent/lineage is different. One has Choros ovog, and the offspring of Buhan, a noble of Oirad. . . The other has Borjigid ovog, and the offspring of Khavt Khasar”.Footnote 41 However, a closer reading reveals that Khasar is not the ancestor of the Khorchin ayimaq as a domain with its population but the ancestor of the Khorchin ruling princes as it says: “Present six zasags (or ruling princes) of Khorchin, and also Jalayid, Dörwöd, Gorlos, Ar Khorchin, aimag of Dörwön khüükhed, Muumyangan, Urad, Alashan, Khökh nuur [and] Khoshuud all are his [Khasar's] offspring”.Footnote 42
Thus, we see how the Qing incorporation dismantled Mongolia as a political entity and transformed her tümens and otogs – hierarchically organised administrative divisions – into ayimaqs – discrete subdivisions of External Mongolia within the Qing Empire. The important insight is that the Qing ayimaqs were actually Mongolia's administrative divisions – tümens and otogs – before their incorporation into the Qing. Furthermore, Mongolia's tümens and otogs, because of the Dayan Khanid apportionment of them as princely shares, were increasingly becoming hereditary princely domains even before the Qing incorporation. And, as hereditary princely domains they were becoming more like autonomous lordships than the hierarchically subordinated divisions of a state. If this was the case at the time of the Qing incorporation and pre-Qing Mongolia, then what was the case during the Chinggisid incorporation and in pre-Chinggisid Inner Asia?
The Chinggisid Incorporation of Eurasia: the Creation of Minqans and Irgens
The political world of the medieval Mongol scribes was a world full of uluses and irgens. The sources abound with these two terms. For example, we find ulus 106 times and irgen nearly 200 times in the Secret History of the Mongols (SHM) alone.Footnote 43 However, only the word ‘Mongol’ is designated by both words; being described as both an ulus and an irgen, while all the other contemporary polities or population categories such as Kitat, Tang'ut and so on were only ever designated irgen. None of these contemporary categories is described using the term ulus except the Mongol, and an analysis of the available sources confirms that this apparent pattern (of the usage of ulus and irgen) was a reflection of a well-established convention.Footnote 44
According to this usage convention, the medieval Mongol ulus was a category of government conceptualised as a community of a realm, while the medieval Mongol irgen was a community of a realm conceptualised as community of a descent. In other words, the Mongol ulus was a political entity and community, while the Mongol (or any other) irgen was thought of as a community of language, custom and descent; an ‘ethno-cultural’ category in modern academic framing, although, in fact, it was also a community of a given realm. Thus, Mongol ulus and Mongol irgen were two different political and ‘ethno-cultural’ conceptualisations of one and the same entity. Accordingly, all the categories other than Mongol were reduced to a category of irgen-hood, subject-peoples: stateless, thus quasi-political and, consequently, ‘ethno-cultural’ or tribal in modern academic parlance.Footnote 45
The usage convention for the terms ulus and irgen and their conceptual division and framework mirror the political make-up of the Mongol imperial world. Indeed, since all the sources belong to the post-1225-1226 period, their perspective is that of the imperial Mongols viewing their own imperial world. Therefore, what we read in the sources, including the SHM, is the imperial reconstruction or ex post facto imposition of the Mongol imperial framework onto its historical past. Indeed, the author of the SHM, looking through the lens afforded by the imperial framework and speaking through the contemporary language of the day, unwittingly made a systematic revision of the pre-1206 past. Thus, all the categories other than Mongol, including those of pre-1206, were reduced to irgen because they were irgens/subject-peoplesFootnote 46 of the empire at the time of his writing.Footnote 47
Indeed, the conquered and subjugated polities no longer existed as independent political entities. In most cases, their sovereigns and political structures were destroyed and their territories were incorporated into the Chinggisid territory. Their populations, stripped of their own political structures, were integrated into the Chinggisid minqans and their decimal structures. Many of the subject populations such as Kereyid, Naiman and Mergid no longer constituted organised communities because they were dispersed into different Chinggisid decimal divisions. Those loyal or submissive to Chinggisid polities were made into minqans and tümens. Indeed, all of the pre-Chinggisid polities were reduced to categories of ‘quasi-political’ in that they did not imply independent or sovereign political existence. Some of them, as minqans, were reduced to the status of sub-state-administrative categories. In short, what was left of the once hegemonic Kereyid kingdom were surviving individual persons, families and lineages known by the identity category of Kereyid, although dispersed in various Chinggisid decimal divisions.
On the other hand, in dividing Mongolia's population into military-oriented minqans, decimally organised hierarchical administrative divisions, the Chinggisid power structure produced numerous military-oriented population divisions with hereditary rulership and membership. At the same time, the Chinggisid state, as a new ulus or regnum or imperium, necessarily embraced all these divisions and imposed a new common ulus/regnum/imperium identity on its population. This not only necessarily re-appropriated previous political identities and placed them into the category of ‘quasi-political’ population categories, but also reduced them to a sub-regnum or sub-state categories.Footnote 48
Thus, the Chinggisid power structure produced, on the one hand, a multitude of quasi-political identity categories and, on the other, numerous military-oriented minqans. Furthermore, as a new ulus or imperium it also imposed a common political identity on its population. Meanwhile, the memory of Kereyid and other polities was enshrined in imperial historiography as irgens – quasi or non-political (and consequently ethnic or cultural), sub or intra-state categories.
The Chinggisid and the Qing incorporations display similar languages and similar patterns. The Chinggisid incorporation transformed pre-existing polities into irgens – stateless, thus, ‘quasi-political’, and consequently, ‘ethno-cultural’ categories of subject-peoples.Footnote 49 Similarly, the Qing incorporation transformed Mongolia as a realm into ‘the Mongols’ as a population; and Mongolian tümens and otogs were transformed into ayimaqs – headless, thus, non-hierarchical and quasi-political subdivisions of External Mongolia, a vassal peripheral territory. If one were to take these ayimaqs as population groups, which is justifiable and, indeed, many did, they would appear as sub-groups or sub-categories of stateless people – tribes.Footnote 50
What, then, can we say about the multitude of intra-Mongol irgens before their integration into the Chinggisid power structure? Were they tribes and clans as later writers, such as in Rashid al-Din's elaborate terminology,Footnote 51 describe them? Were they aristocratic houses and their domains? Or were they instead hierarchical politico-administrative divisions similar to pre-Qing Mongolian tümens and otogs, perhaps in the process of transforming into autonomous lordships? What were the origins of these categories and what were they called before their integration into the Chinggisid state?
Pre-Chinggisid Inner Asia: the Liao Incorporation and the Zubu-Kereyid Hegemony
Ethnographically, the demography of pre-Chinggisid Inner Asia might have been much more heterogeneous than that of the pre-Qing Mongolia. However, their political structures were not that divergent. This is to be expected since both periods were dominated by orders described as either tribal or aristocratic. However, what I would like to argue here is that the pre-Chinggisid Inner Asian political structure was similar to the pre-Qing structure in the sense that the pre-Qing Mongolia's tümens and otogs were hierarchically subordinated politico-administrative divisions and pre-Qing Mongolia constituted a single realm, albeit a decentralised and fragmented one of what was, probably, a reviving power. And while the Kereyid was the seat of the legitimate ruler of the realm, just as Chakhar was for the Northern Yuan, the Naiman, Merkid, Tatar, Mongol, Oyirad and so on were (former) divisions of this realm, perhaps relatively autonomous ones.Footnote 52
An idea of ulus congruent with the Chinggisid state of 1206 was well established in pre-Chinggisid Inner Asia and, indeed, the Chinggisid state of 1206 was built upon the Kereyid state and her sphere of hegemony.Footnote 53 Certainly, scholars have noticed the centrality of the Kereyid polity in pre-Chinggisid Mongolia.Footnote 54 Paul Rachnevsky even maintained that Temüjin mounted the Kereyid throne by destroying the ‘Kerait Empire’.Footnote 55
Although the Chinggisid sources distort the past, their careful examination not only clearly shows that the idea of a single realm of the ‘felt-tent ulus’ was well established in pre-Chinggisid Inner Asia but also allows us to argue that pre-Chinggisid Mongolia formed a single, yet decentralised and fragmented realm under the Kereyid throne. According to the SHM, in 1201, on the bank of the Ergüne River, an assembly of the ruling lords of almost all of the major polities of the Mongolian plateau enthroned Jamuqa as the Universal Khan (gür-qa[n]). Buyuruq Khan of Naiman, Quduqa-beki of Oyirad, Jali-buqa of Tatar, Qutu, the son of Togto'a-beki of Merkid and Tarqutai-Qiriltuq of Taichiud were among many others who enthroned the Universal Khan.Footnote 56 The Universal Khan declared war on Ong Khan and Temüjin and immediately set out to bring them under his rule. However, his campaign was ill-fated. Not only did the Universal Khan lose the battle, but his forces were scattered, effectively ending his reign.
Ill-fated as the Universal Khan's reign was, his undertaking reveals one very important political idea universally shared by the aristocracies of the ‘felt-tent ulus’. There was to be a single universal ruler over all the ‘felt-tent ulus’, meaning that all of the ‘felt-tent ulus’ were to form a single realm and the parties involved were to form its subdivisions. Both Ong Khan and Temüjin shared this idea, but Jamuqa was not the universal khan for them. Instead, their universal khan, or the legitimate ruler of the realm, was Ong Khan, while Jamuqa appears to have been enthroned to displace him.
Other than the ill-fated Gür Khan, no contemporary ruler of the ‘felt-tent ulus’ held a more senior title than the Ong Khan (Wang Khan), and while Ong Khan's deposed predecessor held the title of Gür Khan, his heir apparent was called ‘Sangun,Footnote 57 which means prince in the Cathaian language’.Footnote 58 The highest title that Tayang Khan held was Tai Wang meaning ‘khan's son’ or prince.Footnote 59 The title of Wang or King, or ‘ruler of kingdom or country’ and ja'ut quri [centurion] that the Kereyid ruler and Temüjin received respectively from the Jin dynasty, indicates their relative status within the realm.Footnote 60 Furthermore, while Tayang Khan's exclamation ‘How can there be two khans on Earth?’Footnote 61 presupposes a single, overarching ruler, his charge that Temüjin usurped Ong Khan's throne ‘to become the khan’ assumes that Ong Khan was that overarching ruler.Footnote 62 This clearly shows that Ong Khan was not just a ruler who happened to hold a very senior title but was the commonly recognised overarching ruler of the realm.
Moreover, while the SHM narrows down the field to focus on Jamuqa as the principal pretender to the throne, with Teb-Tenggeri and Qasar as minor obstacles in Temüjin's bid to rule the ulus, the Compendium of Chronicles depicts Jamuqa, Qasar, Sacha Beki and Alag-Udur of Merkid as all being contenders for the rule of the ulus, along with Temüjin.Footnote 63 However, none of these sources mention Ong Khan as a contender. This telling evidence suggests that Ong Khan was the de jure sovereign of ‘Mongolia’, for whose throne these junior lords vied. Also, the Compendium of Chronicles does not include Nilqa-senggüm, the heir apparent, as a contender in the above list, although it reports Nilqa-senggüm as declaring to Temüjin that “We will do battle, and whichever of us emerges will be khan”.Footnote 64 Certainly, Nilqa-senggüm as the heir apparent was not a contender for the throne but its legitimate heir. Indeed, if Ong Khan and Nilqa-senggüm were neither the legitimate ruler nor his heir apparent, they must have headed the list of the aspirants. Thus, the Kereyid throne was the ultimate prize for all these men.
In fact, Ong Khan's ancestor (probably his great grandfather or grandfather) was one of the earliest known rulers of ‘autonomous’ – i.e. autonomous of the Liao – ‘Mongolia’– that is the realm described as Tsu-pu or Zubu.Footnote 65 According to Rashid al-Din, Ong Khan's grandfather Marghuz (Marcus or Markos) Buyuruq Khan, the ruler of the Kereyid, was captured by Na'ur Buyuruq Khan of Tatar, a vassal “to the monarchs of Cathay and the Jurchids”, and was “sent to the King of the Jurchids, who had him killed by being nailed to a wooden donkey”.Footnote 66 According to Liao-shi, in 1089 Liao appointed (or rather, recognised) Mo-ku-ssü, a Zubu ruler, as “the chieftain of various tribes” or, more properly, as the Great King of the Zubu, a ‘subordinate state’ to the Liao since Zubu was categorised as a ‘subordinate state’ and its ruler was addressed by the Liao as the ‘Great King’.Footnote 67 However, Mo-ku-ssü soon invaded Liao and destroyed the greater part of the Liao army in 1093 and continued to war with the Liao until he was captured by Yeh-lü Wo-t’ê-la in 1100 and sent to the Liao court to be “hacked to pieces in the market place”. This Mo-ku-ssü is none other than Marghuz Buyuruq Khan, who obviously was betrayed by the Tatar ruler.Footnote 68 Thus, the kingdom of Kereyid and its hegemony over the others must have been established at least by the end of the eleventh century (or even before the Khitan conquest of Mongolia), and, thus Ong Khan, according to the SHM, really was an ‘ancient old great khan’ (erten-ü ötögü yeke qan).Footnote 69
Indeed, the Chinggisid state was directly built upon the Kereyid state and many of its institutions were adopted from it. After toppling Ong Khan, Chinggis Khan adopted the decimal organisation and created institutions such as the office of six cherbi [stewards] and kesikten [guards] consisting of 80 kebte’ül [night guards] and 70 turqa'ut [day guards], and a vanguard force of 1000 ba'atut [heroes].Footnote 70 The turqa'ut and ba'atut were clearly Kereyid institutions.Footnote 71 In fact, Ong Khan had much larger force of turqa'ut (1000 turga'ud) and ‘presumably a corresponding number of sentries, or night guards’.Footnote 72 Probably, his force of ba'atut was larger too.Footnote 73 Furthermore, Ong Khan's forces were organised in decimal divisions such as minqan turqa'ud and tümen-tübegen (10,000 centre or 10,000 assault troops?) Kereyid. Thus, the Chinggisid state was built upon the Kereyid state, and as such, was its successor. Indeed, according to the SHM, Chinggis Khan himself admitted that he, “by the favor of eternal heaven, achieved the lofty throne deposing the Kereyid people”.Footnote 74
While many scholars, such as Peter Golden and Nicola di Cosmo, emphasise the continuity of the Inner Asian imperial political traditions including the decimal system, Hansgerd Cöckenjan persuasively documents the continuous use of the decimal system by Xiongnu, Hsien-pei, T’ü-chüeh, Khitan Liao, Jin and Mongol and Timurid empires.Footnote 75 In fact, if many of these imperial powers rose and ruled Mongolia directly, the greater part of Mongolia was under the Liao before it was subjected to the Jin policy of ‘divide and rule’.
The Liao directly controlled most of the eastern part of Mongolia for over two centuries while the rest of Mongolia was under its hegemony. The Liao maintained military garrisons in Mongolia, such as Zhenzhou and Hedong or Kedun, established in 1004, with a force of 20,000 troops recruited locally.Footnote 76 Zubu, the power that checked the Liao's expansion westwards, was a ‘subordinate state’ headed by a ‘Great King’ or Khan. The rest of Mongolia was organised into ‘tribes’ and those that were part of the 54 ‘imperial tribes’ were under the Liao ‘Northern Administration’.Footnote 77 Nearly 10 of the 54 ‘imperial tribes’ and some 24 of the ‘outer’ and ‘various tribes’ are identified either as Mongol or residing in Mongolia.Footnote 78 “Tribes were administered by a Khitan senggüm (xiangwen, from Chinese xianggong, lord chancellor), assisted by a lingqu (lingwen, from Chinese linggong, lord director) and staff”.Footnote 79 And, indeed, we find senggüm and lingqu in pre-Chinggisid Mongolia.Footnote 80 In addition to Nilqa-senggüm, we find Caraqai-lingqu and Senggüm-Bilge in the SHM.Footnote 81 Revealingly, these two were father and son. Obviously, Caraqai's son Bilge was promoted to the commander of a division, while his father served as an assistant to the commander. What is more telling is that Bilge's son Ambaqai was elevated as the khan of ‘Qamuq Mongqol’ after Qabul-qahan.Footnote 82 Furthermore, while ‘barbarians’ formed a significant proportion of the ordo (palace) troops of the Liao emperor, “Mu-tsung's ordo garrison is known to have included Tsu-pu (Zubu) tribesmen”.Footnote 83
Yet, the Liao ‘tribes’ were top-down creations. According to Karl Wittfogel, the Liao rulers literally built the imperial 54 ‘tribes’ (bu 部). They divided “relatively large tribal groups” into a number of smaller ‘tribes’, abolished ‘tribal statuses’ and “gathered together small fragments or even individuals and formed them into new groups” and granted them ‘tribal status’.Footnote 84 For example, in one case, “Each of the old eight Ch'i-tan tribes contributed twenty households to a special garrison which, after expanding considerably, developed into the T'eli-t'e-mien tribe”.Footnote 85 Furthermore, the Liao rulers mixed subjugated non-Khitans into Khitan ‘tribes’ and, indeed, seven of the eighteen T'ai-tsu's tribes ‘were all composites’.Footnote 86 Thus Wittfogel uses terms such as ‘synthetic tribes’ or ‘composites’. Indeed, Wittfogel elaborates that “Tsu [族 kin] and pu are Chinese terms that are nothing more than approximate equivalents for specific and sometimes unfamiliar aspects of Ch'i-tan [Kitan] society. The term pu is used as an equivalent for pu-lo 部落, tribe; it may also be used to mean a local group or an administrative unit”.Footnote 87 In fact, Wittfogel defines tribe as “aggregates of a number of local groups, held together by the need for joint military action, by the authority of a head chief”.Footnote 88 The Liao ‘tribes’ were subdivided internally and the smallest division seems to have been a chao or ‘hundred’.Footnote 89 For obvious reasons, the population of Mongolia that was under the Liao must have been subjected to these top-down administrative enterprises.
Indeed, the Liao presence in Mongolia seems to have been enduring. Even the powerful Zubu, who checked the Liao's expansion westwards by constantly warring with it, also suffered from the Liao's ‘divide and rule’ policy. After two decades of turbulence following the Liao emperor Shengzong's campaign against Zubu in 983, the Liao achieved the surrender of the Zubu Khan in 1003 and divided the realm “into several divisions, each under a Liao military governor”.Footnote 90 Later on, fighting against Markhuz, the Liao mobilised ‘tribal soldiers’ against him in 1092 and eventually won over many of Markhuz's former allies against him.Footnote 91 Even Yelü Dashi, the founder of Qara-Khitai, made Hedong, a Liao military garrison, his power base during 1124–1130 and seems to have organised those under his control into ‘seven prefectures’.Footnote 92 Interestingly, scholars were able to identify very few of ‘the eighteen tribes’ that Yelü Dashi reportedly addressed in 1124.Footnote 93 Obviously, many of these tribes (or rather divisions or dependencies), if they ever existed, were top-down ‘synthetic’ or ‘composite’ divisions rather than enduring bottom–up human groups with distinctive identities. Thus, the Liao's ‘divide and rule’ policy seems to have been far-reaching in Mongolia.
Even after the Liao, both the Jin and the Qara Khitai continued to have a divisive influence on Mongolia. Understandably, both of these powers exercised greater influence on their immediate neighbours – the Jin on the Tatar and Qara Khitai on the Naiman – and, in many cases, used them to frustrate the growing power of the Kereyid. Thus we have the Tatar warring with the Kereyid and its Mongol subordinates, and we see the Naiman forcing Ong Khan to flee. In response, the Kereyid ruler seems to have been actively engaging with all the surrounding powers, such as Qara Khitai, the Jin and Xi Xia, obviously to counter-balance them against each other. In fact, Ong Khan seems to have been a very able ruler: not only was he able to establish Kereyid hegemony but he also demolished the Liao divisive legacies, even in the face of the Qara Khitai and the Jin divisive interventions. If we critically approach the Chinggisid sources, we see that it was actually Toghril who was incessantly campaigning to subdue the Merkid, Tatar, Gür Khan, Merkid, Naiman, and Mongol.Footnote 94 As a result, in this uneasy climate, Toghril managed to revive the hegemony of Kereyid power over Mongolia, and won the title of Wang, that is, the king of the realm, a title very similar to that of his grandfather Markhuz, from the Jin court. However, one of his estranged vassals was to claim his throne and realm, surprising him at night while he was feasting unsuspectingly.Footnote 95
Thus, the pre-Chinggisid Mongolia was a good deal like pre-Qing Mongolia; a decentralised, fragmented realm under the hegemony of the Kereyid kingdom. While the Jin and Qara Khitai continued to have divisive influences on the region, the Liao seems to have left many ‘synthetic’ or ‘composite’ military-oriented divisions, since Liao ‘tribes’ were themselves military-oriented divisions. Indeed, that the heads of the pre-Chinggisid Mongolian polities held military or military-oriented titles such as senggüm, lingqu, buyruq, beki, bahadur or ja'ut quri Footnote 96 seems to suggest not only the origin of their divisions but also their subordinate status in relation to a higher authority. However, by the time of Chinggis Khan's rise these divisions were, in all probability, becoming kingdoms and principalities or, perhaps, lordships of various sizes.Footnote 97 Certainly, some of their commanders must have claimed the title of khan as Ambaqai, a son of a senggüm, and a grandson of a lingqu, did. Apparently, as Susan Reynolds observed in the case of medieval Europe, administrative divisions of one generation were becoming lordships in later generations.Footnote 98 However, because these lords believed that they should form a single realm long before Chinggis Khan actually completed the task, those who claimed the title of khan must have sought to build such a realm to answer to these expectations.
Thus, pre-Chinggisid Mongolia, with its various territorial domains under the hegemony of the Kereyd Ong Khan, appears uniquely comparable to pre-Qing Mongolia that was also divided into a number of princely domains under the hegemony of the “Chakhar” Ligdan Khan, the legitimate sovereign of the waning Northern Yuan. In short, if what was left of the Chinggisid decimal hierarchical divisions were in the process of transforming into autonomous princely domains in the pre-Qing era, and then, what was left of the Liao ‘synthetic or composite tribes’ were struggling to maintain their autonomy from Kereyid incorporation in pre-Chinggisid Mongolia.
Thus the comparison of the Qing and Chinggisid incorporations, as well as that of pre-Qing and pre-Chinggisid Mongolia, shows that the ‘tribes’ or ‘aristocratic houses and their domains’ were actually created by imperial powers as their top-down military oriented subdivisions. And, with the collapse of the imperial centres, these divisions began to emerge as autonomous political entities in their own right. However, with the next imperial incorporation they were transformed into ‘quasi-political’ categories while their populations were divided or reshuffled into new numerical divisions of the empire. This leads us to examine the nature of the hereditary divisional system that all the Inner Asian empires employed to incorporate and administer the nomadic populations of Eurasia.
A Hereditary Divisional System and the Pre-Modern Eurasian Political Order
The tribal paradigm seems to have precluded scholarly attention from examining the nature of the imperial incorporation, its institutions of local government and their structural encompassment of the population. In particular, the political significance of incorporating or dividing a nomadic population into divisions with hereditary membership and rulership, and its implications and consequences, has not sufficiently been addressed.
To illustrate the nature of the hereditary divisional systemFootnote 99 in general, and to trace the origin of the ‘tribes’ or ‘aristocratic orders’ in particular, I will briefly discuss the Chinggisid decimal structure of minqan or thousand, the most archetypal form of the hereditary divisional system. Minqan was mainly associated with the Mongol army and many scholars suggest that the system demolished the tribal order.Footnote 100 However, Sneath, citing Xiongnu and Jürchin decimal structures, decisively rejects this interpretation, arguing that a decimal system that existed well before the time of Chinggis Khan's rise was already a well-established administrative system and that “the process described by the SHM seems largely to be the official recognition of existing lordly domains and their integration into a uniform minqan administrative system”.Footnote 101 Thus there was no ‘‘‘Chinggisid revolution” that had destroyed the clan structure and completely reorganised steppe society along decimal lines’.Footnote 102 Moreover, criticising a ‘tribalist’ reading of Inner-Asian history, Atwood has recently advanced a richly documented, excellent argument that Mongolian minqan, otog and khoshuu were basic local administration units of Mongolia for over 500 years in succession and constituted politico-territorial units and functioned as ‘appanage communities’.Footnote 103
Upon his enthronement, Chinggis Khan incorporated the Mongqol ulus into 95 divisions called minqans and appointed their respective noyons (lords or commanders).Footnote 104 Minqans in turn were subdivided into za'un (hundred) and har'ban (ten).Footnote 105 Some but not all of the minqans were grouped into tümens (10,000).Footnote 106 The system created a legible, efficient and responsive governance of the population since it had numerically fixed units and an efficient chain of command, with each superior responsible for ten subordinates. Clearly, the system was to allow swift mobilisation of troops in time of war.Footnote 107 Obviously, the system was devised to concentrate power in the hands of the monarchFootnote 108 because it not only created a top-down chain of command structure but also, by creating divisions with equal size and status, precluded unwelcome expansion of power of the noyans. Thus, the system was not only efficient in mobilising the population but also effective in concentrating power in the hands of the monarch.Footnote 109 The system was flexible to incorporate the pre-existing larger polities by demolishing them into smaller divisions. Even loyal or submissive Onggud, Uru'ud, Onggirat, Ba'arin and Ikires were each divided into a number of minqans.Footnote 110
Evidently, minqan was the backbone of the Chinggisid administrative system and the most important unit of local government, with a uniform nested hierarchy of decimal structure and a specific territory (nuntuq) supervised by imperially appointed officials named nuntu'ucin.Footnote 111 While all the offices of the noyans of each decimal unit were created imperially, all the noyans of minqans were appointed by the emperor and all were subject to uniform legal statutes.Footnote 112 Apart from military service, all minqans were responsible for collecting tax and corvee for the imperial service: SHM indicates legislation where ‘minqans of all places’ were to provide a postal relay, mare's milk and sheep for imperial use.Footnote 113 Thus, a minqan was a unit of government and a unit of local administration.
However, a minqan was endowed with a particular attribute that not only sets it apart from being a mere bureaucratic administrative unit but also gives it a unique constitution: a minqan had not only hereditary membershipFootnote 114 but also hereditary rulership.Footnote 115 Hereditary membership and rulership not only gave a minqan a unique constitution but also was consequential to the imperial incorporation. Indeed, the hereditary membership and rulership provides a minqan with the twofold identity that Atwood attributes to it: an appanage, “a defined territory and people assigned to the hereditary jurisdiction of a particular nobleman and his descendants, as a unit of local government” and “a closed corporate community, that is a group of people who maintain a perpetuity of rights and membership in appanage resources, who limit these privileges to insiders; and who are discouraged from active contact with outsiders”.Footnote 116 Indeed, if the hereditary rulership effectively separated minqans from one another, the hereditary membership virtually sealed them off. Thus, the system as a whole had a weak horizontal integration with cross-knitting bonds between individual minqans. Hence, beneath the imperial power, the system was bound to produce discrete corporate population groups or local communities with their own distinct ruling lineages.Footnote 117 Thus, in effect, the hereditary divisional system was harbouring numerous potential lordships beneath the imperial sway. If the imperial as well as regional princely power were to go, minqans or tümens would emerge as independent powers in themselves.Footnote 118
Meanwhile, the population that was assigned to minqans and tümens, as a result of their hereditary rulership and membership, were to develop and maintain their own distinctive categorical or group identities. Most of them were simply identified by the names of their noyans or names of their noyans’ choosing or affiliations, and these names were established as the names of the populations of respective minqans. For this reason, as Atwood perceptibly notices, we have many so-called ethnonyms or named population groups in the post-imperial period.Footnote 119 Certainly, their identity mythomoteur necessarily centred on the ruling noble houses.Footnote 120 Because of the importance of succession and the inheritance of the office of noyan, the lineages of the ruling noble houses must have been the best remembered and, over time, in many cases the putative ancestors or founders of the ruling houses of the noyans tended to become the eponymous ancestors of all the entire minqans. For example, Yuan shi’s biography of Jürchedei tells us that Jalayir, Uru'ud, Mangghud, Qinggirad and Ikires were descended from Nachin Ba'atur, an ancestor of Jürchedei, a distinguished noyan of minqan.Footnote 121 However, the SHM tells us that Nachin Ba'atur was an ancestor of Uru'ut and Manqgut.Footnote 122 While the hereditary divisional system left the apportioned population no alternative but to loyally serve the houses of their hereditary lords, the genealogical mythomoteur of their lordly houses embraced and encompassed the subject population around the lineage myths of the lordly houses. Thus, the hereditary divisional system was destined to make a division's population not only into a corporate community but also a distinct, durable identity group with a myth of common origin. Finally, when the empire and the regional khanates collapsed, these divisions emerged not only as autonomous minqan-uluses or tümen-uluses but also as discrete identity groups.
This process is unmistakably noticeable throughout the Mongol Empire in the fourteenth and fifteenth centuries. Many tümens and minqans transformed into autonomous lordships. The result was structurally similar throughout the entire empire and a fragmented, decentralised political order – Sneath's aristocratic order or conventional scholarship's multitude of tribes – emerged. We find in the eastern half of the Mongol Empire about a dozen tümens or uluses and numerous otogs or khoshuus, while its western half was divided into many tümens or ulusesFootnote 123 and numerous qoshuns or hazaras in the declining years of the Chinggisid successor states. However, these were by no means Fried's ‘secondary tribes’ because they were integral parts of the imperial government. Instead, to paraphrase Fried's own words, they were veritable emirates or kingdoms, but only after the collapse of the Chinggisid regional khanates.Footnote 124
Thus military-oriented hereditary divisional system was the engine that generated lordships (or aristocratic houses or tribes) with distinct identities.Footnote 125 Yet, as long as they recognised the centrality or superiority of the central ruler they remained subordinate divisions of a larger polity, a unit of local governance. Grouping or dividing population into military-oriented numerically identical divisions was a superb scheme for the concentration of power and mobilisation of troops in times of war. However, if the system as a whole created numerous separate identical divisions beneath the imperial power, making their rulership and membership hereditary transformed them not only into potential lordships but also into discrete human groups with distinctive identities. Indeed, in a situation where local rulers were hereditary, the local units tended to become autonomous lordships in themselves unless central governments took actions necessary to impede them.Footnote 126 This must have been especially true when central authority was very weak and “the entire ulus was in turmoil” as it was appropriately described in the SHM.Footnote 127
Conclusion
We see empire builders, in an effort to devise an efficient system to centralise power and mobilise troops, create sub-state politico-territorial communities with distinct identities by dividing their manpower into military-oriented hereditary divisions. Thus, my inquiry into the origins of ‘tribes’ or “aristocracy-led named groups” shows that they were the top-down creations of what scholars call a ‘centralised state’ or ‘imperial state’. They were created by the state as means to administer and mobilise its manpower. As such, they were the parts of the imperial structure, but not pre-state kinship-based, real or imagined, groups. Similarly, Sneath's aristocracies had their origins in the hereditary offices of the commanders of these divisions, an imperial appointment to reward and secure their service and allegiance.
However, with the waning of the central power, we see these hereditary commanders emerge as independent aristocratic lords contending for the imperial throne with their own autonomous lordships as power bases. The latter happened to be the former imperial divisions, the hereditary appointments issued as a means to secure their allegiance. Both the founders of the Mongol and Manchu empires started their ascendancy from this humble origin and both usurped the thrones of the established powers to legitimise their rule. Both rulers incorporated the pre-existing polities by dividing them into smaller numerical divisions – minqans and khoshuus; both granted heredity rulerships – noyan and jasag – to those who were loyal and submissive. Thus, we see state or empire-building efforts creating both the hereditary divisions as administrative forms as well as ‘quasi-political’ population categories at the same time. Hence the hereditary divisional system was the engine that generated numerous potential lordships (or ‘aristocratic houses’ or ‘tribes’) with distinct identities. All in all, we might say, projects of imperial incorporation created both aristocracies and ‘tribes’.
