The early states and empires of Inner Asia played a pivotal role in Eurasian history, with legacies still evident today. Yet, in spite of more than 100 years of scholarly contributions, the region remains a relatively unknown heartland (Di Cosmo 1994; Hanks 2010; Lattimore 1940; Mackinder 1904). As pivotal as the history of Inner Asia is in its own right, it also holds special significance for how we interpret complex societies on a global basis. Recognizing the broader relevance of Inner Asia requires articulation of some distinctive and challenging characteristics that are both familiar and nearly unique. These polities emerged, existed, and collapsed in ways that often defy conventional understandings of what constitutes early complex societies. Such hallmarks as sedentary populations, cities, complex bureaucracies, defined territorial boundaries, and agriculture, so typically associated with early states, play less conspicuous roles in Inner Asia. Instead, mobility, scale, extralocal interactions, nonfixed property, dispersed aristocratic control hierarchies, and the economics of multiresource pastoralism serve as alternative foundations for these complex social systems.

To evaluate the early states and empires of Inner Asia from a broad comparative vantage point, I focus first on their unique characteristics, followed by a discussion of major theories and debates that have long influenced interpretations. These major themes are considered briefly in relation to past and present archaeological and documentary research in the region. Theory and data sources are then contextualized in an analysis of the political, organizational, and economic processes in the historical trajectories of 15 steppe polities beginning with the Xiongnu (Khunnu), around 200 B.C., and concluding with the Zunghars in the mid-18th century. I follow with a focus on the emergence of states and empires in terms of how power relations are constructed across a variety of practices, social categories, and institutions of authority. From this analysis is derived a series of conclusions about Inner Asian polities in the context of how early states and empires everywhere rose, fell, and were sometimes sustained for long periods of time.

Although diverse, the environment of Inner Asia is dominated by grasslands that support the herding of several different animals. Rather than simply a stage for human action, the environment is an interactive context that contributes to and modifies human events on a continual basis. Inner Asia is a vast region of steppe grasslands, deserts, and mountains, situated between the boreal forests and taiga of Siberia to the north and the Central Plain of China to the south. In the east the Greater Khingan Mountain range, the forests of northeast China, and the Korean peninsula form a permeable margin; to the west the steppe lands of eastern Kazakhstan beyond the Altai Mountains form a western margin (Fig. 1). Farther west is Central Asia, extending from eastern Kazakhstan to the Black Sea. Figure 1 is a map showing the approximate extent of Inner Asia, including northern China (Inner Mongolia and portions of northeast China), all of Mongolia, the southern Siberian portion of Russia, and eastern Kazakhstan, including key geographical features and select archaeological sites mentioned in the text. In some instances Inner Asia is described as a much more expansive swath of central and northern Asia. In this sense, the region described here might be termed eastern Inner Asia. Throughout the region weather conditions can be extreme, with hot summers and long bitterly cold winters. The territory occupied by the modern nation of Mongolia is at the heart of Inner Asia and was the key setting for development of several of the polities discussed here. In this environment a diverse set of dynamic polities emerged that illustrate important differences and similarities over time and space.

Fig. 1
figure 1

Inner Asia, consisting of northern China (Inner Mongolia and portions of northeast China), all of Mongolia, the southern Siberian portion of Russia, and eastern Kazakhstan, showing the location of selected geographical features and archaeological sites discussed in the text

The commonly used chronology that encompasses the development of complex polities begins with the Bronze Age, which in Central Asia is defined as ca. 2500–1000 B.C. (Frachetti 2008). Hanks (2010, p. 471), who places the Bronze Age at ca. 3500–1200 B.C., recognizes a broad transitioning into the Iron Age between 1200 and 300 B.C. Although there are many alternative ranges given by different authors, based on emerging data that reflect regional differences and new theoretical perspectives, the date ranges provided by Hanks are a good compromise when considering all of Eurasia. In the eastern steppe of Inner Asia, especially in Mongolia, the Bronze Age is less well defined. Current research, however, is generating large numbers of dates between 3000–700 B.C. (e.g., Frohlich et al. 2010; Kovalev and Erdenebaatar 2009). The Bronze Age is followed by the Early Iron Age (700–400 B.C.) and the Late Iron Age (ca. 400 B.C.–A.D. 1200). The time from A.D. 1200 to 1500 is generally described as the medieval or Middle period, although there is growing recognition that the concepts associated with the medieval period and the feudal system in Europe do not apply in Inner Asia. After A.D. 1500, Inner Asia entered the Early Modern period. Within these periods, most literature on the early states and empires use the dating given in historical documents for the individual polities (Fig. 2).

Fig. 2
figure 2

The chronological sequence of polities discussed in the text, beginning with the Xiongnu and concluding with the Zunghar

The meanings of such terms as state, empire, polity, and “complex social system” vary across different fields of study (Cederman 1997, pp. 17–19; Grinin et al. 2008; Johnson and Earle 1987, p. 246; Kradin 2009a; Mann 1986, p. 37; Service 1975; Sinopoli 2001, pp. 444, 447; Turchin 2003; Weber 1946, p. 78). Although clear distinctions in terminology are important, my objective is not to create exclusive definitions and typologies but to examine the interplay of sources of authority, technology, and modes of adaptation. Less-specific definitions allow for more analysis of variation without the reification of a specific set of attributes. I define states generally as political entities with a recognized form of managerial control that extends over a particular population and territory. An empire has the characteristics of a state that has expanded, and is inclined to expand further, to include other states and sociopolitical entities, incorporating increasingly multiethnic populations over a diverse territory (Morrison 2001; Rogers 2007, p. 250). Both states and empires have recognized boundaries, although these may be defined in only a vague sense. Whereas states and empires share characteristics, an empire is not simply a large state. The scale of empires presents different managerial challenges and embodies far different forms of organization. I use the term polity in a generic sense to refer to any sociopolitical entity with internal rule and a prescribed population beyond the scale of lineage and clan. A polity may be a state, an empire, or a different type of grouping, such as a tribe or confederation.

Organizing ideas

There are several key ideas fundamental to the trajectory of early states and empires in the region. The foundations derive from mobility and economic adaptations to the steppe and adjacent regions. Within this context the political and social process of how steppe polities expanded their power opens the potential to interpret the transformation of social roles that could be integrated and stratified in a variety of functions to develop increasingly complex social systems. Closely linked to the dynamics of state formation is the nature of regional interactions and especially interaction with the states of China and their influence on steppe polities.

The characteristics of mobility

Within the diverse landscape of Inner Asia the forms of social systems and economic adaptations that were the foundation of early polities emerged after the domestication of the horse, especially after horses were used for riding (Jacobson-Tepfer 2008; Kradin 1992). By 3500 B.C. (Outram et al. 2009), the Botai culture of Kazakhstan consumed horse milk and meat and also used harnesses that probably facilitated riding. The domestication of the horse and subsequent riding, however, were not immediately followed by its widespread adoption or the transformation of local economies (Kohl 2007, p. 140). Across the Central Asian steppe—from north of the Black Sea to eastern Kazakhstan—there is substantial evidence for diverse mobile pastoralist economies, but primarily after 2500 B.C. (Benecke and von den Driesch 2003; Frachetti 2009). Once horseback riding became common in Central Asia, the herd animals previously domesticated in the Middle East were incorporated in mixed-herd pastoralism, including varying proportions of sheep, goats, cattle, yaks, and camels. The riding of horses also was a transportation breakthrough that promoted the spread of cultures, languages, and technologies such as bronze-working along various trade routes (Anthony 2007; Boyle et al. 2002; Frachetti 2008, p. 46; Kohl 2007). The use of horses, camels, and other pack animals greatly enhanced both east–west trade connections and north–south trade between Siberia and southerly regions. Importantly, the horse also was the foundation for techniques of warfare that later fueled mobile pastoralist successes in their conflicts with more sedentary societies.

With mobility, landscapes shrink and vision expands. Individuals are able to move across large distances and share communication with widely distributed destinations. The experience of connection with place is culturally constructed and embodied in the relationship that exists between meaning, function, and movement. Mobility is therefore culturally experienced and may or may not have a direct correlation to geographical distance. Mobility is important for all societies and can be the source of great advantage. For pastoralists, however, mobility becomes an essential component of the social world. Mobility links political and social practice on the landscape, through concepts such as communal land tenure, reciprocity, universal traveler hospitality (usually within political boundaries), and the individual nuclear family as the basic social and sometimes independent productive unit.

The mobility that developed as part of this pastoralist adaptation is perhaps the single most important underlying principle for the polities discussed here. This type of mobility is likewise closely tied to expansive environments and the concepts of nomadism and pastoralism. Like the definitions of states and empires, pastoralism is a term that has been typologized along a scale ranging from pure nomadism to pastoralists living in permanent locations. In some cases the distinctions made in these typologies, between categories such as pure nomadism and seminomadism, have been used to invoke an evolutionary progression towards sedentism. Such an evolutionary view, however, is not supported when compared against the variation that may exist at one time in a particular region (Pletneva 1982, p. 145). Salzman (2002), for instance, observed that pastoralists often incorporate a range of subsistence practices, including agriculture. For some time, anthropologists working with pastoral nomads have noted considerable differences among these societies and that there is no single social characteristic exclusive to this group (Dyson-Hudson and Dyson-Hudson 1980, p. 55). Archaeological findings have expanded even further the complexity of the forms of adaptations that come under the heading of steppe pastoralism. The pastoralists of Inner Asia certainly capture a range of variation, and it is not realistic to think of a single pastoralist tradition, whether in the recent or more distant past. For the purposes of this study, multiresource pastoralism—that also may include agriculture—is a more realistic way to describe the subsistence adaptations of households and communities within these early states and empires (Honeychurch and Amartuvshin 2007; Salzman 1972).

Mobility is a hallmark of Inner Asian societies, but, like pastoralism, it is not a unitary concept. Just because a pastoralist family may live in a tent does not mean that they move the tent very much at all. Numerous ethnographic studies from the 20th century have illustrated various forms of mobility, involving both long-distance and local sporadic movement under a variety of political restrictions (Fernández-Giménez and LeFebre 2006; Humphrey and Sneath 1999; Simukov 1934; Vainshtein 1980; Vreeland 1957).

Although mobility is clearly recognized and different types of seasonal and daily movement have been defined, it is less common to disassemble the different forms and circumstances of movement on the landscape. Simukov (1933, 1934, 1935, 1936) was perhaps the first in Inner Asia to develop a comprehensive typology of pastoralist subsistence-related movement through his ethnographic work in the 1930s. While Simukov’s typology of herder movement has resonance with other sources of information, there is significant variation and change over time hidden within his formulation of spatial relationships. Given emerging systematic archaeological information on monumentality, subsistence practices, and settlement patterning, it seems clear that variation in mobility has always been part of pastoralism in Inner Asia (Frachetti 2008, p. 17; Frohlich et al. 2010; Honeychurch et al. 2009).

The relationship between the landscape and the forms of social practice that led to complex polities is mediated by the types of mobility fundamental to pastoralists with access to effective transport, like horses. The potential offered by mobility greatly influenced how the social and political landscape was structured in terms ranging from localized sacred places to entire regions to where and how settlements and other aspects of the built environment were positioned (Ashmore and Knapp 1999; Charleux 2006; Honeychurch et al. 2009; Ucko and Layton 1999). The pastoralist polities often built, although sometimes reluctantly, places that qualify as urban centers. They were aware that permanent constructions actually diminished their greatest advantage. These constructions typically have a temporal dimension that mirrors social and political trajectories to an extent not seen in more sedentary societies. For instance, several polities built settlements at new locations as part of a planned project and likewise abandoned these locations as a single event. Often these urban centers were never resettled, although reuse did become more common in later times (Charleux 2006, p. 184; Rogers 2009; Rogers et al. 2005). This is different than in many sedentary polities based on agriculture, where urban centers themselves are tied to a particular location and therefore embody a complicated history of change—richly complex but difficult to interpret archaeologically. In theories of the development of complex social systems, urban centers are often the conceptual center of discussion (Trigger 2003). In quite a few cases the city is even coterminous with the state (Nichols and Charlton 1997). Beyond the city is a hinterland that supplies the needs of the urban residents. For the steppe polities it is necessary to turn the view around and look not from the city outward but from a decentered vantage point. From this perspective mobility is a central concept.

Wendrich and Barnard (2008, p. 9) provide useful grounding for conceptualizing mobility. They identify four categories that describe the range of mobility considered here: moment (duration), movement (range and pattern), motivation, and population segment. Each category encompasses a diverse range of possibilities. Within a particular region and time, herders are motivated to construct a predictable landscape of resource access. At the household level, subsistence-related mobility is tied to the maintenance and growth of herds based on the vagaries of weather and the availability of grass and water. Recent studies by Fernández-Giménez (1997, p. 10) and Humphrey and Sneath (1999) highlight some of the risk aversion strategies that may either expand or restrict mobility, including maintenance of flexibility, dispersion, seasonal short- or long-distance movement, and herd splitting. Closely connected to subsistence-related movement are the limitations and opportunities tied to differential wealth and social boundaries. Mobility certainly functioned as a social distancing mechanism, to either lessen or expand communication, for such things as the maintenance of social networks and the flow of goods and services. Along with the advantages of mobility came communication challenges typically solved through specific traditions of social networks, polity-sponsored strategies, and cultural practices that included periodic social and political gatherings.

Elements of power and empire

As today, early states and empires existed as socially stratified societies under various political control hierarchies. Power was derived from economic sources, knowledge structures (networks, rituals, symbols, kinship, and materiality), and armed forces as the basis for the dynamics of social change (Earle 1997; Yoffee 2005, pp. 38, 40). These various sources of power are best thought of as overlapping networks of interaction and institutions that are always in the process of being created for particular goals. The nature of these institutions and even their goals often remain contested rather than forming a unitary concept of society in which internal variation is minimized and change moves along a single trajectory (Mann 1986). Typical archaeological reliance on stages, as in cultural evolution, needs to be disassembled and the particular characteristics considered in their own right, not so much as processes that produce stage change but as clusters of characteristics that change at different rates and for different reasons.

The nature of power and its relationship to political hierarchy is a very important aspect of any polity, especially in terms of how authority typically becomes centralized in the hands of a few individuals (Stein 1998, p. 26). The polities of Inner Asia present some unique characteristics, although they also are unmistakably part of the larger segment of political strategies applied at many places and times over the last several thousand years. In Inner Asia the construction of political authority has been pursued by historians, anthropologists, and political scientists by incorporating a close reading of the historical record with new information from archaeological projects. The Inner Asian polities were not always clearly centralized under a single ruler who held absolute authority; instead, traditions of hierarchy were embedded in the aristocratic lineages from which leaders came. It was under these lineages that the general population was stratified and controlled.

Hierarchy in social systems is closely connected to complexity. Hierarchy itself, however, has multiple definitions (Crumley 1995, p. 2; Pumain 2006). In considering the diverse uses of the hierarchy concept, Lane (2006, pp. 83–86) notes four different, yet partially overlapping, kinds: order, inclusion, control, and level. An order hierarchy refers to a simple ranking of a set of something based on a criterion such as size. An inclusion hierarchy is similar to an order hierarchy, except that it adds the extra dimension of each level encompassing all previous levels. A control hierarchy—the most commonly used—refers to the power to execute decisions that affect lower levels of the hierarchy. Typically, information flows up the hierarchy and decisions flow down. The administrative bureaucracies of the various polities described here are examples of control hierarchies. Level hierarchies may have a developmental quality, as in cultural evolutionary frameworks with inclusive progressions of sociopolitical entities in which lower levels give rise to higher levels in a form of upward causation. Level hierarchies, as used to describe complex systems, do not necessarily function in a unilineal trajectory, and there also may be downward causation. Entities in a level hierarchy have specific spatiotemporal scales and associated differences in complexity (cf. Heylighen 2000). The multidirectional nature of causation in a level hierarchy is important to consider, as it represents an alternative perspective to cultural evolutionary stages and other commonly used progression sequences.

The elements of control hierarchies, level hierarchies, and multidirectional causation are incorporated in a theory of collective action, called canonical theory (Cioffi-Revilla 2005; Rogers and Cioffi-Revilla 2009). In this theory, polities respond (or fail to respond) to situational change through collective action imbedded in a series of fast-branching processes. Actions taken may result in the loss or accrual of social complexity (upward or downward causation), eventually resulting in a longer-term slow process of change. In canonical theory there is no assumption of evolutionary direction resulting in increased complexity, although there is a tendency for polities to develop the knowledge to permit greater success under certain circumstances. The traditions of statecraft and control hierarchies that developed in Inner Asia are examples of accrued knowledge.

The maintenance and distribution of power occur through kinship, political office, ideological appropriation, differential expertise of the participants, existing social categories, informal power arrangements, and coercive force. Most, if not all, early states also might be described as having a royal or imperial form of leadership for the channeling of power that combined elements of kinship and political office (Trigger 2003, p. 264). Centralized control hierarchies are the typical mechanism through which power is exercised at the level of polities. The tangled character of how power is actually distributed, as discussed above, is echoed by Mann’s (1986, p. 1) observation: “Societies are constituted of multiple overlapping and intersecting sociospatial networks of power.” All social systems have these alternative sources of power, sometimes referred to as “heterarchy” (Crumley 1995, 2001; Stein 1998). These alternatives may be co-opted by official hierarchies, and sometimes they coalesce as rebellion or subaltern ways of contesting existing power structures. A common alternative power network in Inner Asia was a type of distributed power that contradicts existing ideas emphasizing the centralized and unitary nature of authority. On the steppe authority was often decentralized for long periods of time and remained so even in some of the most successful polities.

Different opinions on the nature of empires in Inner Asia also are closely tied to the idea of mobility and its broader impacts on social organization. Some have argued that political centralization was inherently difficult because of the low population densities of pastoralists and the tendency to disperse under unfavorable political circumstances (Allard and Erdenebaatar 2005; Irons 1979; Khazanov 1994a, pp. 152–164). For instance, it is generally assumed that pastoralist families could cancel their allegiance to a particular leader by simply moving away. In reality, it was a dangerous undertaking and generally required finding a new allegiance and protection, or suffer the consequences. Sometimes, perhaps even often, groups of families did find new allegiances, but this is more a refection of the corporate nature of the steppe “tribes” than a commentary on their lack of organization. Defection was certainly not a phenomenon restricted to pastoralists. In China farmers might flee an entire region to escape high taxes (Rossabi 1987). The steppe was not a vacant landscape; it was politically and culturally partitioned in complex ways. In actuality, there is strong evidence from a variety of sources that powerful hierarchies asserted control over the lives of individual families in very fundamental ways, at least by the beginning of the Bronze Age (during the middle to late 3rd millennium B.C.), and probably before. The Bronze Age on the steppe was a time that saw the formation of chiefdom-like hierarchies and other characteristics later associated with early states.

The expanding polity

In addition to mobility, political centralization, and general issues surrounding how power is built and maintained, there also is the dynamic process of expansion itself. Following successful conquest, expansion is largely about the management of ethnic, cultural, and organizational diversity, which represents a major challenge for expanding polities, whether in Inner Asia or elsewhere in the world. All polities must employ mechanisms to construct order and through which “others” are encountered and dealt with following the incorporation of new territories (Rogers 2005, p. 334, 2007, pp. 263–264). There are six widely used strategies. The first strategy, the elimination of diversity, was often implemented through the prohibition of certain practices, or even genocide. Attempts to eliminate deeply held cultural practices, for instance, often sow the seeds of later uprisings. A second strategy, the manipulation of continuity, employs the creation of new forms of continuity and the disruption of preexisting patterns, such as when the conquerors adopt the courtly style and bureaucracy of the conquered people. The expanding polity also may co-opt and redefine traditions as part of the long-term struggle for control of collective memory (Jonker 1995; Woolf 1994). A third strategy, organizational imposition, similar to Flannery’s (1972, p. 410) “linearization,” occurs when the entire indigenous administration is eliminated and replaced by a new structure. This strategy is often related to colonization. A fourth strategy that was employed several times in Inner Asia is dual administration. In this practice the conquerors maintain separate administrations for the indigenous population and resident invaders, who may either have colonized portions of the new territory or taken up positions as the new elite. In the fifth strategy, overlay organization, a small administrative presence is developed, primarily to exact tribute or to monitor compliance with other demands. Numerous empires, including several in Inner Asia, have followed a sixth strategy, marginal incorporation, where conquered regions are left to their own administrative controls as long as they comply with demands.

These strategies, often implemented in an ad hoc way, are the fundamental motivations that create new complex political formations that distinguish an empire from an otherwise similar but smaller state. These approaches are not presented as formal types of engagement as multiple strategies were employed simultaneously or sequentially by the same polity in different regions and for different lengths of time.

The pastoralist–agriculturalist dichotomy and dependency models

In the characterization of early complex polities, the above points concern the creation and maintenance of authority, the nature of continuity within the trajectory of Inner Asian polities, and strategies for expansion. The nature of continuity within the Inner Asian traditions raises a key theoretical question about whether these polities should be viewed as a by-product of the states that formed earlier in China and elsewhere. For Inner Asia the question of interaction with China is very important, but the scope of the relevant contacts through trade and other means extends in all directions and not just to the south (e.g., Amory et al. 2006).

The relationship with China has traditionally been viewed as a divide between pastoralism and agriculture, in which the pastoralists are in some way dependent on the agriculturalists. Perhaps the most innovative and contested perspective to emerge in the recent wave of research is the challenge to this commonly expressed dichotomy (Rogers 2007; Sneath 2006). In some cases dualities or paired sets of oppositions do express valid distinctions. The pastoralist/agriculturalist dialectic, however, has been folded into a much broader set of characterizations used extensively in social typologies of pastoralism (Khazanov 2003). The analytical meanings behind this dichotomy, as it pertains to early steppe polities, are typically linked by describing pastoralists as marginal and dependent on nearby sedentary states for both the ideas of statecraft and the products needed to sustain control hierarchies and the complex operations of large political systems (Barfield 2001; Khazanov 1994a; Kradin 2002). Coupled with this is the fact that almost all written sources come from China or other sedentary societies and not from the pastoralists themselves (Lindner 1982). This imbalance highlights the potential for interpretive bias and the further need to seek new sources of information from archaeology.

The pastoralist/agriculturalist duality for Inner Asia can be traced back to early Chinese sources that described the people living beyond the northern borders. The first of these northern peoples for which there is substantial information is the Xiongnu. While they are noted in earlier records, their first major appearance is in the Shiji (Records of the Historian) by Sima Qian, who lived between 145 and 90 B.C. (Watson 1961). These early Chinese references to northern peoples as barbarians have remained an influential strand in the characterization of pastoralists. Although there is no single analog in Chinese for the word barbarian, there are usages that imply a pejorative meaning (Kriukov 1978), and the use of the term as a typical translation in European languages has done much to influence thinking (Di Cosmo 2002, p. 95). In the original Chinese sources the use of several terms for those people beyond the “boundaries” also reflected an emerging us/them dichotomy that crystallized as a concept of Chinese identity, well articulated in the literature of the Warring States period (476–221 B.C.), but also in earlier sources. While the dichotomy of a Sino–foreigner world was routine, simple dualities were complicated by instances of cooperation, cultural exchanges, and acculturation in both directions (Pines 2005, p. 70). Geography also was conceived as a dimension of identity, with the notion of “natural boundaries” that existed between the northern pastoralists and the centers of agriculture in China (see also Drompp 1989, p. 141).

From a broader perspective, the anthropological literature often describes mobile pastoralists as people with an unstable economic base occupying marginal environments (Dahl 1979; Krader 1978; Lattimore 1940, p. 77; Weissleder 1978). In part, these perspectives derive from ethnographic observations in the 19th and 20th centuries when pastoralists were indeed often marginalized within the framework of emerging nation-states. The typically low population densities outside urban centers (usually less than 1.5 people per km2) and the volatile nature of wealth (i.e., herds) has led many scholars to discount mobile pastoralists as capable of independently originating more complex forms of social organization; rather, they have existed in a dependency relationship with sedentary polities, in this case, those of China and southern Central Asia (Barfield 1989; Dahl 1979; Golden 1992, 2001; Jagchid and Symons 1989, p. 165; Khazanov 1994a; Lattimore 1940). The most extensive recent example is Barfield’s work (1989, 1991, 2001, 2006), which characterizes the steppe polities as tribal organizations that either developed centralized control hierarchies or became decentralized in response to patterns of centralization in China. In this model, the “shadow” empires of the steppe required the wealth of China to fund their emergent aristocracies and the political aims of their fragile secondary states. Without this external wealth, it is argued that the steppe polities would revert to a simpler tribal formation (Barfield 1989, p. 8).

In Inner Asia there is no question that pastoralist interactions with the states of China and southern Central Asia were complex with deeply intertwined histories. Describing the relationship as one of dependency, however, discounts the growing evidence for independent traditions of authority building and the complex interactions that the steppe pastoralists maintained in all directions (Di Cosmo 1999; Honeychurch and Amartuvshin 2006; Rogers 2007; Sneath 2007; Vasjutin 2003, p. 52). There are further reasons to question dependency interpretations, based on the conceptual orientation of scholars working on these issues. The dominant interpretive framework, until recently, has fostered dichotomies of pastoralist versus agriculturalist as parallel representations of the dichotomy of tribe versus state. If pastoralists are viewed as nonhierarchically organized and occupying marginal lands, then they serve in the tribal role in these perceived dichotomies. Agriculturalists would then be associated with states. This is a misrepresentation of the evidence.

Pastoralism allows ample opportunity for the emergence of social inequality and the development of traditions of authority. A growing number of archaeologists, historians, and cultural anthropologists have not only questioned the dependency perspective but provide convincing evidence that the resource base, trade connections, and native social systems for the steppe pastoralists are diverse and self-sustaining (Anthony and Brown 2007; Frachetti 2008; Honeychurch et al. 2009; Houle 2009). The dependency model assumes that the steppe pastoralists were not normally organized in hierarchical ways beyond tribal kinship affiliations and simple hierarchies. It also presupposes a fundamental kinship basis that was itself already entangled with a long-standing tradition of complex societies that developed on the steppe in the Bronze Age. However, extensive archaeological evidence from throughout the steppe regions of Eurasia shows that traditions of hierarchical authority emerged before 2000 B.C. (e.g., Chang 2008, p. 338), given the broad evidence for pastoralism and agriculture in different regions of the steppe (e.g., Kohl 2009; Koryakova and Epimakhov 2007; Levine et al. 1999).

In some regions, especially the western steppe, emerging control hierarchies were based on relatively settled villages practicing a mix of agriculture and pastoralism (Anthony 2007; Lamberg-Karlovsky 2002). In other regions pastoralism was practiced, at least initially, without direct connections to agriculture, even where there is evidence for substantial dwellings, such as in the late Neolithic (4th millennium B.C.) Atbasar cultural complex of northern Kazakhstan (Kislenko and Tatarintseva 1999). In the same region, the later Botai cultural complex of the 3rd millennium B.C. incorporated settled villages with horse breeding and management of wild herds. Evidence for reliance on pastoralism is present by 2500–2000 B.C. in some regions (Frachetti 2008, p. 47; Kislenko and Tatarintseva 1999, p. 201; Zaibert 1993). Recent evidence from Begash in Kazakhstan, which dates the presence of millet and wheat in a ritual context to around 2200 B.C. (Frachetti et al. 2010), highlights the variable timing and role of agriculture.

A major study by Jean-Luc Houle (2010) develops direct archaeological evidence for emerging social complexity in the Late Bronze Age (1400–700 B.C.) of northern Mongolia. The study was designed to test the China-dependency hypothesis as a source for sociopolitical developments among the pastoralists. The results provide evidence for the relatively restricted movement of the pastoralists, stable economic self-sufficiency, the construction of monuments and burial facilities following a broad regional pattern, and archaeological evidence for limited status differentiation. Taken together, these findings further challenge the expectations associated with the dependency hypothesis.

Evidence from the Bronze Age throughout Central and Inner Asia presents a complex mosaic, both spatially and temporally, in the styles and timing of economic and social changes (Hanks and Linduff 2009, p. 3; Prušek 1971). Frachetti (in press) has synthesized the Central Eurasian evidence for the emergence of regionally diverse agropastoralist economies. He introduces the concept of “non-uniform institutional alignments” in the emergence of greater social complexity by the second millennium B.C. to account for regional and local diversity. The resulting forms of political social complexity do not fit well with common culture evolutionary typologies. Instead, the evidence documents unique steppe traditions with achieved and ascribed social hierarchies that could easily provide the conceptual foundations for the steppe-based aristocracy that played a central role in later empires.

Additional evidence of continuity in steppe cultural practices comes from the late Bronze Age/Early Iron Age transition in Central Asia in a broad cultural complex referred to as Scythian (approximately 800–400 B.C.), which has been researched extensively archaeologically (Alekseev et al. 2002). The term Scythian was known to the Greeks and Persians. Around 450 B.C. Herodotus described a group of people he termed Scythians as a hierarchical society of pastoralists who also practiced agriculture (Godley 1981, pp. 218–219). Although there continues to be disagreement over the link with the historical documents, there is substantial archaeological evidence for complex status hierarchies, including the royal tombs (kurgans) at Arzhan (von Zabern 2010) and related Pazyryk burials dating between the 6th and 3rd centuries B.C. from the Altai Mountains that are part of a broad Scythian cultural horizon. In the standard interpretations, the Scythian cultural complex was eventually eclipsed by the Sarmatian in the regions north of the Aral Sea, but also extending as far east as Tuva in Inner Asia. In the eastern steppe, including parts of Inner Asia, a similar archaeological culture known as the Saka, dated to roughly the same time, has likewise revealed very substantial evidence for royal lineages (Chang et al. 2003). Taken together, the accumulating evidence supports a far more dynamic interpretation of an emerging complex social system, one in which pastoralists did not depend on sedentary agriculturalists.

Past and present research

There is a long history of archaeological and documentary research in Inner Asia going back to the 17th century (Vladimirtsov 1934). The majority of the work in the 19th and early 20th centuries is relevant to the study of early states and empires, as questions about the legacy and origins of the Mongol empire of the 13th and 14th centuries was of considerable interest. The ill-informed image of the mounted barbarian warriors sweeping across the steppe is engrained in the popular imagination and remains a common stereotype in much of the world. The primary reason this image has remained so durable is that the Mongols were indeed incredibly successful and thoroughly devastating in their attacks, and the histories of the time were written by those conquered. Lattimore (1951, pp. 61–65) can be credited with partially opening the eyes of Western scholars, but the uncritical image of barbarian invasions has endured and is generally applied to all early polities of Inner Asia, going back nearly a thousand years before the Mongol empire.

During the early and middle 20th century, prominent Russian, Chinese, and Mongolian archaeologists and historians began the process of developing a deeper understanding that did not depend solely on early historical sources. Through the 20th century, projects centered on the early polities added new details at a slow but steady pace (Bichurin 1950; Pletneva 1982). Their primary objective was to develop a detailed culture history that could be linked to the Chinese and other historical sources. By mid-20th century, new interpretations were being developed independent of the documentary sources, especially through the excavation of tombs, such as the Xiongnu tombs at Noyon Uul (Rudenko 1962) and Gol Mod (Brosseder 2009). These objectives still represent the focus of most research in Inner Asia. With important exceptions (Lattimore 1940; Wittfogel and Feng 1949), relatively few European or American scholars were involved before the late 20th century.

In China, historians and archaeologists have focused on the origins of Chinese civilization for decades. The traditional view that the Central Plains of China were the cradle of Chinese civilization has been effectively criticized and expanded to a more multiregional concept reflecting complex regional histories informed by recent archaeological research (Keightley 2000; Liu and Chen 2003, 2006, p. 151). Although archaeological research in China has typically been viewed as a subfield of history, recent research on the later pastoralist states and empires of northern China have produced a variety of important insights (An 1989; Shelach 1999).

The pace of international research, especially in Mongolia, expanded dramatically after the breakup of the Soviet Union in 1991. While Russian research efforts have a long tradition in Mongolia, other countries became involved in the 1990s. A recent example of the diversity of researchers is the first international conference on Archaeological Research in Mongolia, which was held in Ulaanbaatar in 2007, with participants from 13 countries (Bemmann et al. 2009). This meeting encouraged a diversity of theoretical perspectives applied to field research as well as interpretation of the findings. During the 20th century, in much of Inner Asia archaeology operated predominantly under a cultural evolutionary theory that was closely tied in numerous studies to an evolutionary progression utilizing axiomatic interpretations based on Marxist theory and related definitions of pastoralism (Tolstov 1934; Vladimirtsov 1934). In practice, most research was directed to expanding the empirical base to build better culture histories using a relatively general idea of cultural evolution. Also very prevalent was the effort to link place names from the early Chinese, Islamic, and European accounts with actual locations (Kiselev 1965; Perlee 1961). The early texts are still the most important sources for details about all of the polities; therefore, archaeologists continue to pay close attention to how actual places relate to these texts (e.g., Ochir and Erdenebold 2009).

Major Inner Asian polities

The construction of leadership, the characteristics of pastoralism as practiced over a huge contiguous region, and the contingencies of history all contributed to the dynamics of Inner Asian polities. Below are brief descriptions of those political entities that played a substantial role in the complex history of Inner Asia and beyond. Each polity presented here has the characteristics of early states and empires. Most are poorly known archaeologically, which is not surprising considering that these are political entities, often with a relatively short existence and often composed of the same ethnic groups as in preceding polities. For these reasons a distinctive material culture may not exist to distinguish one from another. Even so, it is in the study of these polities that the efforts of field archaeology and documentary history meet to provide the most multifaceted picture possible.

The dates listed for each polity are commonly accepted dates; however, they may be controversial as there is often no consensus as to when exactly a polity begins or ends (Fig. 2). For instance, one author may consider a major military defeat as spelling the end, while another may consider the death of the leader as the final blow. In some instance a polity might be dislodged from its former territory but not defeated in a definitive event. Aristocratic lineages may reemerge with newly formed political alliances decades after having been relegated to the sidelines. These factors, and more, serve to complicate the political and spatial landscape and point to the need to recognize emerging power relationships that crosscut social and ethnic distinctions and blur chronological boundaries. Some of these aspects are highlighted in the brief descriptions offered below.


The Xiongnu polity is the first widely recognized steppe political entity to be identified as an empire (Brosseder and Miller 2011). Like other empires to follow, the Xiongnu polity was multiethnic and multilingual given the large geographical scale. The Xiongnu were the first to leverage the forms of multiresource pastoralism, developed in previous centuries, in combination with effective military strategies and technologies to not only conquer their steppe rivals but pose a threat to China (Barfield 1981). Not only were they a threat, but they soon became a serious rival for the newly emerging Han dynasty (202 B.C.–A.D. 9). It was during the initial period of Xiongnu expansion that the so-called Great Wall was constructed, forming a hard border between the Xiongnu and China. The Xiongnu expansion focused on two strategies in the core region—manipulation of continuity and organizational imposition. Both west and east of the core region the Xiongnu exacted tribute from polities through marginal incorporation.

Although there were pastoralist predecessors under the collective name Hu that were noted in the Chinese histories in the 4th century B.C., none had developed sufficient power to be of real concern. Archaeological research on the Bronze Age supports the existence of societies with complex social hierarchies throughout the steppe regions. Chinese historians do note in the decades prior to the consolidation of the Xiongnu empire that three major steppe confederations existed: the Yuezhi in the west, the Xiongnu in the center, and the Dunghu in the east. There also is brief mention of other groups such as the Wusun and many other smaller groups (Barfield 1989, p. 32; Christian 1998, p. 184; Prušek 1971, pp. 24, 97). The Xiongnu apparently were able to absorb these other peoples.

For the Chinese, Pan Ku (1997, p. 3843), who lived A.D. 32-92 and wrote the Han shu, sums up the attitude towards the Xiongnu: “The barbarians are covetous of grain…human faced but animal hearted.” Such was the often-repeated attitude towards the northern peoples. By the time Chinese authors were making their first detailed reports, the Xiongnu already had a fully developed tradition of aristocratic hierarchy and a military organization that would be repeated in modified form over the centuries by many of their steppe successors, including the Mongols.

It appears that multiple ethnic and tribal affiliations existed within the Xiongnu polity (Di Cosmo 2002, p. 165), although to some extent these represented fictive relationships, since allegiance of individual households was highly variable and changed over time with evidence for factional competition and the realignment of groups under one or another leader. The basic organization of all the steppe polities has been tied to a simple form of the tribal/kinship model (e.g., Jagchid and Hyer 1979, p. 245). However, this model is not well supported, considering the complex dynamics of shifting affiliations under different political hierarchies evident in the histories.

In 209 B.C. Maodun became the supreme Xiongnu leader under whom many of the early successes took place (Di Cosmo 2002, p. 186). The overall organization represented a melding of military–administrative systems with a local administrative component. There were three main geographical divisions consisting of the east and west, otherwise known as the “left and right” kingships, with an implied third division being the core central region administered directly by the paramount leader. Within this system there were 24 regional leaders, each with the title of “ten thousand horsemen.” The 24 regional leaders represent a dispersed control hierarchy, probably composed of the leaders of long-standing aristocratic lineages (e.g., Sneath 2007, p. 116).

At its height the Xiongnu empire controlled a region encompassing all of Mongolia and extending south to the Ordos region in the loop of the Yellow River in northern China (Fig. 1). In the north control extended to the boreal forests of Siberia, including Tuva and Buriatia in the Russian Federation (Kradin 2005a). Based on a variety of maps and geographical information on the location of different groups, at its height the Xiongnu empire encompassed a region on the order of 4,000,000 km2. Historical sources indicate that while the Xiongnu were primarily pastoralists, they practiced some agriculture (Hayashi 1984). Botanical analyses were conducted at the Ivolga site complex, an important example of a fortified settlement of 2,500–3,000 people specializing in agriculture and metal production in the Transbaikal region (Davydova 1995; Kradin 2005a). In the Egiin Gol region of north central Mongolia, archaeobotanical studies also have confirmed the presence of agricultural production. The first full-coverage survey at Egiin Gol has produced the most comprehensive local picture of Xiongnu use of a specific region (Honeychurch and Amartuvshin 2003, 2007). Projects focused on other regions and sites, such as Takhiltyn Khotgor (Miller et al. 2009), Gol Mod (DesRoches and André 2009), and Noyon Uul (Polos’mak et al. 2008), are now also providing more comprehensive regional pictures (Fig. 3).

Fig. 3
figure 3

Xiongnu elite mound 20 pit burial at Noyon Uul (Noin Ula). Burial associations included a Chinese chariot, numerous artifacts, organic materials, and a lacquered coffin (Polos’mak et al. 2008, p. 86, reproduced with permission from Elsevier)

Throughout Inner Asia more than 400 Xiongnu tombs have been excavated. Although Xiongnu cemeteries include tombs of varying scale, many show a level of labor and wealth investment that links them to elite individuals. Several tombs have recently been excavated by joint Mongolian–French, Mongolian–Korean, and Mongolian–Russian teams (Brosseder 2009; Chang et al. 2007; Guilhem 2002; Konovalov 2008; Miniaev and Elikhina 2009; Miniaev and Sakharovskaia 2007). The larger tombs were often terraced pits, some more than 20-m deep with a ramp leading down. The tombs were usually located in areas away from settlements. In the Ordos regions of North China, the cemeteries of Budonggou, Daodunzi, and Xigoupan, dated between the 2nd century B.C. and the 1st century A.D., include mortuary traits related to the Xiongnu (Wu 1990). A variety of settlements are known, including the walled sites of Gua Dov, Tereljiin Dorvoljiin, and the settlement of Boroo Gol, with semisubterranean houses reminiscent of the structures at the Ivolga site (Danilov 2009; Perlee 1961; Ramseyer et al. 2009).

The Xiongnu conducted many battles and border raids against the Chinese. The greater mobility of the Xiongnu cavalry proved to be very effective against Chinese infantry and chariots, and for a long period of time a treaty (198 B.C.) was in effect, bringing vast quantities of Chinese tribute into the Xiongnu court (Barfield 1981; Yü 1967). Before 50 B.C., the Xiongnu split into a northern and southern polity. Both remained well organized and expansionistic at first, but eventually the southern Xiongnu (estimated at 200,000 people) became a vassal state of the Han Chinese, and by A.D. 150 their political control was virtually nonexistent. Periods of famine and civil war within the Xiongnu polities may have weakened them further (Christian 1998, p. 202). It is generally acknowledged that by A.D. 155 the Xiongnu empire no longer existed, although some authors date the end of the empire as early as A.D. 93 (Honeychurch and Amartuvshin 2006, p. 262). Compared to their rise, the decline of the Xiongnu appears to have taken place relatively slowly. Although the dynamics of the decline are poorly known, the main causes appear to be internal succession struggles, famines resulting from herd loss due to weather events, and increasing military pressure from the later Han.


Unlike the Xiongnu, the Wuhuan maintained a dispersed set of control hierarchies, with only partial evidence for a single paramount leader. Their organization was far more egalitarian, and centralized leadership tended to be merit-based rather than hereditary (Fan 1937; Parker 1892–1893). While herding was the principal aspect of the Wuhuan economy, they also practiced extensive agriculture at different points in their history (Kessler 1993, p. 67). The Wuhuan polity originated in northeast China (Fig. 1), and there are several sites in the Mongol Autonomous Region that relate to both the Wuhuan and the neighboring Xianbei (Inner Mongolia Cultural Relics Team 1974; Kessler 1993, p. 68).

The Wuhuan were periodically in conflict with the Xiongnu, beginning by at least 78 B.C. (Barfield 1989, p. 59). In that year the Wuhuan attacked the Xiongnu and raided their royal tombs. The Xiongnu retaliated and won an easy victory, forcing the Wuhuan under their control. Following Xiongnu civil wars between 58 and 49 B.C., the Wuhuan reasserted their autonomy. Throughout their history, the Wuhuan were a major polity who engaged in a series of conflicts in China, often as an ally of one or another of the local warlords or as Chinese intermediaries against the Xiongnu (Gardiner and de Crespigny 1977). However, by A.D. 207 conflicts with China resulted in a definitive Wuhuan defeat. Although the Wuhuan continued to function for a decade as three separate chiefdoms, in A.D. 218 further Chinese (Wei) attacks thoroughly destroyed Wuhuan power.


The Xianbei were another polity from northeast China, with origins closely related ethnically and linguistically to the Wuhuan. By A.D. 155 the Xianbei had eclipsed the Wuhuan and were poised to fill the gap left by the fall of the Xiongnu polity. By approximately A.D. 170 the Xianbei controlled a vast region approximately 3,000 km east–west and 1,500 km north–south (Gardiner and de Crespigny 1977, pp. 29–30). Like the Wuhuan, their political organization was not highly centralized, with leadership reportedly based on merit, although given other evidence this could not have been the only criterion. This may have allowed lesser leaders to operate relatively independently, which served to limit Xianbei effectiveness against the later Han dynasty (A.D. 25–220) in China (Barfield 1989, p. 86; Gardner and de Crespigny 1977, p. 2). A report from around A.D. 1 notes that the Xianbei were composed of 36 tribes made up of 99 clans (Dien 1991, p. 41). Although Xianbei bureaucracy was minimal, there is evidence that they divided their territory into eastern, central, and western divisions, reminiscent of the more formal Xiongnu administrative structure, and utilized a form of organizational imposition and dual administration as the empire expanded. In their expansion they also absorbed several different ethnic groups and incorporated families (reportedly 100,000) that were formerly part of the Xiongnu polity, suggesting incorporation through manipulation of continuity.

Like the Wuhuan, Xianbei history has many intersections with the Xiongnu, either as vassals or enemies. The Xianbei were often allies of the different Chinese dynasties and war lords fighting against the Xiongnu, but at times the Xianbei also were at war with the Chinese (Fan 1937). Circa A.D. 60, the Chinese court reportedly paid annually the contemporary equivalent of 270,000,000 dollars in bounties and gifts to the Xianbei. In A.D. 87 the Xianbei destroyed the northern Xiongnu. Reportedly, 280,000 northern Xiongnu surrendered at the Chinese border and 100,000 families joined the Xianbei (Fan 1937). With the fall of the Xiongnu in A.D. 155, a new Xianbei leader, Tanshikhuai, formally declared the formation of a state (Gardiner and de Crespigny 1977, p. 15). They continued to raid China, sometimes with the help of the southern Xiongnu and the Wuhuan. Subsequent to the death of Tanshikhuai, his brother came to power, followed by a nephew, and then an unrelated leader (Kebineng), but unity was ephemeral and by A.D. 235 the Xianbei broke into a series of smaller polities, eventually reemerging as the Toba (northern) Wei polity.

The Xianbei practiced multiresource pastoralism, including hunting and some emphasis on growing wheat, barley, and millet. In their home region of northeast China several sites are associated with the Xianbei; however, because of the short existence of their expanded polity it is difficult to identify Xianbei sites farther west. The Xianbei reportedly practiced shamanism. Two possible Xianbei sites with graves are located at Wangong and Zhalainuoer, both in Hulunbeier League in northeast China. Burial offerings at these sites show trade connections with China and the Xiongnu. Ceramics and other objects are markedly different than other known assemblages, so it may be possible to define a distinctive Xianbei material culture. Zhalainuoer is the later of the two sites, with objects not earlier than the 1st century A.D. Other sites in northeast China and Inner Mongolia are connected to the Xianbei, such as sites with dwellings and burials in the Wuerjimulun River valley, east of Nanyangjianingzi, Balin Left Banner (Dien 1991, pp. 41–43; Su 1979). In A.D. 160 the Xianbei leader moved his headquarters to that of the former Xiongnu leader in the Khangai Mountains in Mongolia, although this location has not been conclusively identified. Among known sites is the walled settlement at Shengle, reportedly built in A.D. 258, just north of modern Holingol. The site was occupied by Chinese farmers under Xianbei control (Dien 1991, p. 45).

Jujan (Rouran)

The Jujan polity emerged in the central portion of Inner Asia under a powerful leader (Shelun) who forged a confederation of several local groups that lasted 175 years. At its height, Jujan geographical boundaries were similar to those of the Xiongnu (Kradin 2005b, p. 155). A typical challenge for the steppe polities was the succession process, which might involve lineal or lateral kin, or might be a matter of election by a group of leaders, or even assertion of leadership through military force. The Jujan managed, however, at least 16 royal successions while maintaining the polity. Eleven of these took place peacefully, while five others were by violence. Like others before them, they adopted the Xiongnu-style three-part regional organization, while also maintaining a simplified bureaucracy composed primarily of dispersed royal and noble lineages (Kradin 2000, 2005a). Unlike the Xianbei, they established a more centralized imperial confederation with systematic warrior registration and rules of behavior. However, there was no written legal system or functionaries to administer laws beyond the military.

The Jujan relied on multiresource pastoralism in addition to agriculture. Historical documents make reference to Jujan permanent settlements, including palaces and a capital (Mumocheng). Although some archaeologists have attributed certain locations to the Jujan, confirmation remains elusive (Kradin 2005b, p. 150; Perlee 1961, pp. 35–44; Taskin 1984, p. 290). Jujan history is marked by almost constant conflicts with the agropastoral Toba Wei state to the south in China and with major steppe neighbors. Jujan strategy focused on the long-standing steppe tradition of raiding south into China for plunder but not to gain territory. There was plenty of opportunity for this considering that in A.D. 400 there were at least 12 independent states in northern China. Although records exist of raids on China, almost nothing is known about conflicts oriented in other directions. At the height of Jujan expansion in A.D. 460, they had extended control west to the Turfan region (eastern Turkestan and Xinjiang).

Historical sources report that by A.D. 500 the Jujan were actively adopting a variety of Chinese influences, including the use of written Chinese for official records. In A.D. 522 the Jujan asked for 340,000 liters of millet as seed stock to expand agricultural production, and there is evidence for the adoption of Chinese customs. In A.D. 552 the Jujan suffered a crushing defeat at the hands of the Turks and all Jujan males over the age of 16 were put to death.

Toba Wei

The aristocratic lineages that descended from the Xianbei polity reemerged as the Toba Wei. The Toba were originally the most westerly of the Xianbei “tribes” that composed the ancestral polity (Barfield 1989, p. 118). As the polity took shape in the A.D. 380s, it first occupied eastern Mongolia but later expanded to include northeast China. By 396 the leader, Tuoba Gui, formally declared himself emperor of the state of Wei. Following an earlier Xianbei model, they implemented overlay organization and dual administration, with separate policies and roles for the pastoralists under Xianbei administration and for the agriculturalists under Chinese administration. The Toba Wei political structure was a unique hybrid system that departed from the traditional three-part (left, right, and center) pastoralist administrative system, while also incorporating significant elements of Chinese bureaucracy (Holmgren 1982; Wei 1937).

The Toba Wei practiced a combination of pastoralism and agriculture. Over their history they increasingly emphasized the agricultural and settled aspects of the economy. Initially, however, the Toba Wei built few, if any, permanent structures. For instance, the founder of the polity lived in a tent camp for at least half his reign (A.D. 386–409). Between roughly A.D. 400 and 450, the Toba Wei maintained six major garrisons between themselves and the Jujan to the north (Holmgren 1982). In approximately A.D. 410 the Toba Wei leader moved the capital south out of the steppe to Pingcheng, at modern Datong in China, where it remained for nearly 100 years (Dien 1991, p. 44). In A.D. 494 the capital was moved farther south to Loyang. There are several sites, including tombs, from the 6th century.

Through the decades of A.D. 400–439, the Toba Wei expanded through all of northern China and into the northern steppe. In 439 they defeated the last remaining Chinese border state, the northern Liang (Eberhard 1949). By around 410 the Toba Wei leadership began a conscious effort to move away from the old Xianbei pastoralist traditions towards Chinese traditions, with Chinese-style schools, language, and dress. Speaking the Toba Wei language was prohibited at court. In 524 and 525 the primarily pastoralist troops at the northern border revolted, although unsuccessfully. Around 530 the Toba Wei divided into western and eastern states (Dien 1991, p. 49) and many of the old traditions reemerged. In 581 the Sui dynasty successfully conquered the Toba Wei.

Turk I and II

The Turks were first reported in the Chinese chronicles as a member group of the Jujan confederation, with their ancestral origins in the Altai Mountains. Unlike some earlier groups, there are a variety of sources available for the Turkic polities (Drompp 2005; Golden 1992; Li 1937; Ling-hu 1937; Liu 1937; Ou-yang Hsiu 1937; Sinor 1990). Although the two Turk empires are distinct, they are combined here because of similar organization and their spatial and temporal proximity. For both, there were at least four recognized levels in the administrative hierarchy, almost all of whose members came from the ruling Ashina clan. Succession was lateral to brothers of the khaghan. If there were no available brothers, a high council was appointed to select a leader, but this did not prevent military confrontations and civil war. According to the Chinese chroniclers, there were 28 hereditary ranks or titles in the Turk political system, suggesting a formal bureaucracy but not an entirely centralized administration. Both Turk empires utilized the three-part geographical divisions first seen with the Xiongnu (Golden 2006, p. 50). Evidence suggests that the mechanisms of empire expansion included both organizational imposition and overlay organization. The first Turk polity was apparently divided into four relatively autonomous divisions (Sinor 1990, p. 298). At its maximum, the first Turk empire extended from the Caspian Sea to northeast China and from the boreal forests in the north to the Gobi Desert in the south.

The Turk polities practiced pastoralism with at least a small amount of agriculture. They also participated strongly in trade networks extending throughout the empire and beyond to Persia and Byzantium. There were no major urban centers; in fact, the Turkic general and counselor, Tonyukhukh, is credited with the quote, “If we build castles and give up our old customs, we shall be vanquished” (Tkachev 1987, p. 114). The Turkic leaders took this advice, although there is a report of a settlement built at a place called Dalee (Perlee 1961, p. 47; Rogers et al. 2005, pp. 812–813). Archaeological work in Inner Asia relating to the Turkic empires has concentrated on the Khoshoo Tsaidam temple complex in the Orkhon River Valley in central Mongolia (Amartuvshin and Gerelbadrakh 2009). This region was important as early as the Xiongnu empires, but especially during the Turkic, Uighur, and Mongol empires (Allsen 1996; Kiselev 1965). There are several major inscriptions in the Turkic runic script from Khoshoo Tsaidam but also from the Tuul, Ongi, and Selenge River basins. Throughout Inner Asia there are several tombs and stone statues associated with the Turkic polities.

The defeat of the Jujan in A.D. 551 led to a rapid expansion of the first Turkic polity. However, only 30 years later civil war over a succession dispute split the polity into mutually hostile eastern and western sections. Although separate sections often were embroiled in internal disputes, they managed to successfully attack China and thwart counterattacks. At other times the Turkic polities were closely allied with either the Sui (A.D. 581–618) or the Tang (A.D. 618–907) dynasty (Sinor 1990). By 627 internal rebellions and a Tang invasion resulted in the dissolution of the first Turkic polity. Beginning in the 680s a new series of Turk successes resulted in the formation and rapid expansion of the second expansive polity. For decades the second Turkic polity raided Tang China to exact tribute. In 734 the famous khaghan Bilgee was assassinated and a variety of infighting among factions continued for a decade. By 744 an internal coalition emerged and defeated the last imperial elite and their troops.


The Uighur polity began from an initial coalition of nine smaller groups. Together this coalition was responsible for the fall of the second Turkic empire. The Uighurs adopted many of the formal administrative structures that had become traditional in the Turkic empire (Liu 1937). Royal succession was lineal, which reduced the potential for civil war. Uighur political organization was relatively centralized, with several levels of administration, including a system of tax collection. Still, a leader often served dual civil and military functions. Some evidence indicates that local leaders were relatively autonomous and that royal edicts were not always the law of the land (Mackerras 1990, p. 328). Associated with the centralization of authority were the expansion strategies of manipulation of continuity and organizational imposition.

The Uighurs practiced a form of agropastoralism that varied significantly by region, depending on local environmental conditions. Through their early adoption of the Manichaeanism religion and reliance on Sogdian advisors, agriculture became even more important and their links with the Iranian world expanded. There are several sites that have been associated with the Uighur empire. The most famous is the huge capital city of Ordu Balik (Khar Balgas), currently being excavated by a Mongolian–German project. It is estimated that the city encompassed a walled area of 25 km2 (Kiselev 1957; Radloff 1892; Rogers et al. 2005). In A.D. 821 the traveler Tamīn ibn Bahr visited the city and described it as agriculturally rich with many outlying villages (Minorsky 1947, p. 283). Another important Uighur site, Baibalyk on the Tsagaan River (Bayar 1999, p. 176), consists of three large square enclosures, one still with very substantial walls. Recently, a Uighur site known as Khedun in the Chinese chronicles has been identified as the site currently referred to as Chintolgoi in Bulgan province, Mongolia (Ochir and Erdenebold 2009).

Uighur history is not unlike that of the other steppe polities. They conducted many military campaigns against other pastoralists and against China. In A.D. 758 they reportedly achieved a victory over the pastoralist Kirghiz, to the north. In 762 the Uighurs successfully supported the Tang emperor against a rebellion. The Uighurs were frequently Chinese allies, which involved several marriage alliances between the royal courts. Campaigns against their neighbors and internal conflicts continued to be the norm. In 839 there was a report of heavy snowfalls, a great famine and pestilence, and the loss of many sheep and horses. These events, along with the constant warfare with the Kirghiz to the north, may have triggered the rapid downfall of the Uighurs, although a gap in the Chinese sources gives no information on the previous decade (Mackerras 1990, p. 342). The Kirghiz destroyed the Uighur capital in 840 but did not take the opportunity to form an empire or a unified state.

Khitan Liao

The Khitan first appear in documentary sources in the 4th century as a people of the south-central portion of northeast China, from a region of mountains and open grasslands. Under the leadership of Abaoji, the Khitan rapidly adopted a centralized royal form of organization, with clear similarities to Chinese traditions but also incorporating some of the familiar steppe pastoralist strategies (Wittfogel and Feng 1949, pp. 59–65). Although Buddhism was a central feature of the polity, as the empire expanded into Central Asia the majority of the population was actually Muslim (Biran 2006, p. 66; Dunnell 1996, p. 4). The incorporation of Chinese craft and administrative specialists is closely tied to the transition from a confederation to a state. Prior to this phase of empire-building, the various “tribes” within the Khitan polity had collectively elected a central leader for three-year terms. As with other steppe polities, the term “tribe” was applied to groups that were part of the Khitan confederation but not necessarily ethnically Khitan. In some cases certain “tribes” had only recently come into existence and were linked to other ethnicities, such as the Uighurs (Franke 1990, p. 404). With the expansion of the empire, the Khitans organized around three geographical divisions, a north, a central, and a southern region. Inside these core regions organizational imposition was practiced. Beyond the core, overlay organization and marginal incorporation were utilized to administer the diverse populations. Dual administration also was used, with Khitans in the northern division governed by traditional law and Chinese subjects in the south governed by Chinese administrators (Biran 2006, p. 66).

As the empire expanded farther into the steppe, local populations were displaced by agriculturalists brought in from the south (Enkhtur Altangerel, personal communication 2010; Wittfogel and Feng 1949, p. 123). Traditionally, ethnic Khitans lived in tents and the royal court partially maintained a pattern of seasonal movements common in pastoralist societies. As the empire grew, central, western, and southern capitals were established, along with many other major settlements and border outposts (Jagchid 1981; Mullie 1922; Perlee 1962; Scott 1975; Steinhardt 1997). All of their capitals were in the region of the Great Wall, and the southern capital was located at present-day Beijing.

There are several sites in central and western Mongolia that probably served as border outposts inhabited by relocated Jurchen and Chinese (Ou-yang Hsüan 1937a). One of these is the site of Khar Bukhyn Balgas. Like several other sites, it consists of a large square defensive wall constructed of rammed earth enclosing an area of nearly 1 km2 (Rogers et al. 2005, p. 807). Additional sites in Mongolia that have large defensive walls are the Kherlen Bars 1, Kherlen Bars 3, Sumt, East Wall, and West Wall sites (Dashnyam et al. 1999). At Kherlen Bars 1 there is a standing brick stupa (Fig. 4). Currently under study by a Mongolian–Russian team is another walled border settlement, Chintolgoi, situated 250 km northwest of Ulaanbaatar. This site was originally built by the Uighur polity and later reconstructed by the Khitan. It was apparently inhabited by Bohai people relocated from the east (Kradin 2005c; Kradin and Ivliev 2009). Uglugchiin Kherem is a fortified site with the unusual feature of massive stone walls positioned on the side of a steep hill in Khenti Province, Mongolia. Most of the enclosed space at these sites shows no signs of habitation. It is likely that open areas within the walls were for tent neighborhoods. Other sites are located in Inner Mongolia, such as Shangjing, the supreme capital near modern Chifeng in northeast China, and the tomb of Yelü Dashi, a noted Khitan official in the same region (Kradin and Ivliev 2009, p. 462; Ta 2006, p. 54). There are more than 50 chamber tombs and other notable constructions associated with high-ranking Khitan or Han Chinese who served in the Khitan administration. Most of these tombs are located in Inner Mongolia and Liaoning Province in China (Kuhn 2006, p. 28). The extensive site survey by the Chifeng International Collaborative Archaeological Research Project (CICARP) provides extensive documentation for numerous sedentary agricultural communities in the Chifeng region for the Khitan, other polities, and earlier periods (Chifeng 2003; Linduff et al. 2002).

Fig. 4
figure 4

Multistoried brick stupa at the Khitan Liao settlement of Kherlem Bars 1 (photograph by Amelia Yonan)

Over the centuries prior to the formation of their expansive state, the Khitans alternated alliance and war with virtually all their neighbors, but especially different Chinese states. In A.D. 947 the Later Jin dynasty (A.D. 1115–1234) surrendered and the Khitan state took on the name Great Liao (for the Liao River). Successful Khitan attacks farther south against the Song dynasty (A.D. 960–1279) resulted in substantial tribute payments and a trade agreement that continued for 100 years. After initial expansion the Khitans did not continue to press against the Xia steppe empire or against Korea; diplomatic agreements were put in place instead. The principal challenge to the empire came from internal dissention by several of the original groups of the confederation. By 1114, a Jurchin alliance instigated a civil war in cooperation with the Sung Chinese. The capital was captured and the emperor killed in 1125 (Franke 1990).

Tangut Xia

The Tangut state of Xia (Xi Xia) was located west of the Khitan Liao in the Ordos region of modern north-central China. The beginning date used here refers to the creation of the autonomous Ordos state. The imperial period formally began with the first emperor in A.D. 1038 and the emergence of a more formal centralized government. Even with a royal court, imperial power was far from absolute and continued to rely on compliance from “tribal” leaders who were the foundation of the confederated state (Dunnell 1996, pp. 27–28). From pastoralist foundations the Tanguts built a highly stratified empire based on Chinese traditions of administration. The leaders of the multiethnic Xia polity adopted and promoted Buddhism, (Allsen 1996, p. 119; Dunnell 1996, 2009). Even with extensive Buddhist influences, it is not possible to characterize the Xia state in any sort of monolithic way since they also practiced traditional shamanistic beliefs and adopted Confucian teachings (Kychanov 1987). Dunnell (1996) provides a strong argument for viewing initial Xia state formation as a means of organizing resistance to Song Chinese pressure.

The economy of the polity was based on multiresource pastoralism plus irrigation agriculture and a focus on trade. The latter element of the economy was particularly important as they were able to serve as intermediaries in the long-distance trade along an important strand of the Silk Road connecting Europe with Asia.

There are several types of sites known, including large fortresses, communities (e.g., Khara Khoto), palaces, kilns, and a royal cemetery containing the tombs of several of the emperors (Dong 1978; Dunnell 2009). The fortress community of Khara Khoto, later known as Edzina, was the first place attacked in the Mongol invasion of A.D. 1226. Khara Khoto was studied by Russian and other archaeologists in the early part of the 20th century (Guo and Li 1987). Other Xia sites are known in the Ordos region of North China, such as the capital of Zhongxing at present-day Yinchuan and the fortified site of Gaoyoufang, with walls forming a square enclosure of nearly 1 km2 (Chen and Long 1987). Buddhist temples, monasteries, and shrines also are well represented (Lei et al. 1995; Xu and Du 1995).

For much of their history, the Xia were part of a three-way power struggle with the Song dynasty to the south and the Khitan Liao state to the east. Although there were periodic agreements and periods of peace, there also were seemingly endless border conflicts. By A.D. 1205 the Xia polity was under pressure from the Mongols to the east. In 1209 Genghis Khan captured the border fort at Wulaihai and then the capital. The Xia capitulated to the Mongols and began paying tribute. The accumulation of tribute obligations was the initial Mongol strategy, but in 1226 the Mongol armies invaded and incorporated the Xia directly into the expanding Mongol empire.


Jurchen history overlaps with that of the Xia in the century before the Mongol expansion. The Jurchen derived from southern northeast China and are described as having had local elected leadership, a strong military, and little in the way of overall cohesive political organization (Ou-yang Hsüan 1937b). The military, organized in the common decimal system (units of 10, 100, 1,000, and 10,000), not only conducted campaigns but also was essentially the organizational structure beyond the home territory (Franke 1990, p. 417). Eventually they adopted a Chinese model of administration, along with many cultural practices. Newly conquered groups, even Chinese, were incorporated within the polity by receiving traditional Jurchen titles, a practice suggesting the complex affiliations and origins of internal groups that have often been referred to simply as tribes in the literature. Although the polity took on a very strong Chinese style, there also were significant differences. They initially maintained the dual administration used by the Khitan Liao, in which Jurchen and Chinese were under separate and unequal authority and laws. Buddhism became the state religion.

Derived from small villages based on hunting, animal husbandry, and some agriculture, the Jurchen were not initially a steppe pastoralist group, but they adopted many steppe practices as their polity expanded. Agriculture also became a more significant component of the economy as they colonized appropriate areas (Franke 1990, p. 418). There are several sites that were either built or reoccupied by the Jurchen. One is the city ruins of Fengzhou, constructed by the Khitan Liao, repaired by the Jurchen, and reused again during the later Yuan dynasty (Li 1977).

Between A.D. 1114 and 1120, the Jurchen successfully conquered the Khitan and defeated a huge imperial army sent by the Song dynasty (Ou-yang Hsüan 1937a; Wittfogel and Feng 1949, p. 596). By 1126 the Jurchen took control of all of North China (Barfield 1989, p. 179). To a large extent the Jurchen were able to take over the Khitan Liao empire, including maintenance of its administrative structure. In the process of adopting a more centralized Chinese-style administration, the emperor eliminated the traditional council. In 1142 a peace treaty was signed with the Song dynasty in which the Jurchen received an annual tribute. In 1150 the dual administration system was ended and in 1153 the capital was moved farther south (Barfield 1989, p. 181). Between 1161 and 1189 there were rebellions and efforts to revive traditional Jurchen customs. In 1211 the Mongols attacked and eventually captured the southern capital at present-day Beijing. By 1234 the Jurchen state had fallen to combined attacks from the Mongols and the Song (Franke 1990, p. 420).

Kara Khitai (western Liao)

The Kara Khitai polity represents a western reemergence of the Khitan Liao state that had previously encompassed northern China and much of Mongolia. The Kara Khitai consolidated a region extending from western Mongolia and the Tarim Basin to the Aral Sea (Biran 2005a, map 4). They incorporated many of the organizational strategies that had been used by the Khitan Liao, including the use of coins, the Chinese calendar, a well-defined Confucian administration, and centralized control (Biran 2005a, pp. 94–102). Even so, there was a strong tendency to maintain Khitan ethnic identity and pastoralist traditions in a dual administration in which Khitan were governed separately from local groups. The military was organized in the traditional steppe decimal system, although they drew salaries, unlike typical steppe practice (Biran 2006). The Kara Khitai polity is an interesting example of a steppe empire that did not adopt the local religion of the lands they conquered (Biran 2005b). Buddhism was already well established, at least among the Kara Khitai leadership.

The Kara Khitai empire had a strong connection to pastoralism; however, it also certainly incorporated the irrigation agriculture of the settled populations that became part of the polity. Their location on major east–west trade routes established them as part of an important trading network. There are many known communities, fortresses, and other sites, including cities such as Samarkand, that still exist today. Initially, the royal Kara Khitai court utilized an encampment of tents, but later a capital was established at Balāsāghūn in contemporary Kyrgyzstan. A variety of ruins from the old city are still present, including the restored Burana tower.

The origins of the Kara Khitai polity are associated with the leader Yelü Dashi, who abandoned the failing Khitan Liao empire and departed to the west with his followers, including Khitan, Mongols, and Chinese (Ou-yang Hsüan 1937a). Over the next 13 years the emerging polity expanded rapidly and consolidated control over several groups, including the Uighurs, Karluks, Kanglis, Kirghizs, and Seljuks, and the cities of Besh Balik, Kashgar, Khotan, and Samarkand, among others. Although the Kara Khitai attempted to dislodge the Jurchen from their former homelands, they were unsuccessful. The western Khitan population continued to grow and population pressure may account for their efforts at conquest, although most sources indicate that the Khitan were invited to assist in local conflicts, which gave them allies and opportunities (Biran 2005a, pp. 41–42).

Although the period following the death of Yelü Dashi in A.D. 1143 was relatively stable, new conflicts emerged with pastoralist groups. In 1216 the Mongols entered the Kara Khitai territory with the mission of capturing the emperor. Although the Mongols engaged in at least one major battle, the fall of the Kara Khitai did not involve destruction of the cities or wholesale slaughter of the population. In some instances the local sedentary populations were already inclined to join the Mongol cause against what had been an oppressive regime (Biran 2005a, pp. 85–86).


The Mongol empire is rightly the most famous of the steppe polities. During the 13th and 14th centuries the Mongols developed the largest contiguous land empire in history. The imperial period of expansion lasted about 150 years, although Mongol khans continued to rule Inner Asia and portions of Central Asia for another 300 years and into the era of modern nation-states (Atwood 2004a, p. 365). The early history of the empire and the personal rise of Genghis Khan (born Temüjin) are documented in The Secret History of the Mongols, the only significant historical document of Mongol authorship (Cleaves 1982). Because of the extent and impact of the empire, the relevant literature is extensive, ranging from Marco Polo to studies of DNA linking the ancestry of many people to the Mongol expansion (Buell 2003; Dawson 1955; Fitzhugh et al. 2009; Prawdin 1953; Zerjal et al. 2003).

As was the case with the ancient Xiongnu polity, the Mongols also adopted a military decimal organization. At least initially, the Mongols also employed the common east, west, and center regional military divisions. Genghis Khan also is credited with several innovations that made the Mongol polity effective, especially in their early and rapid expansion. Three that stand out are the implementation of rigorous military rules, the placement of individuals in positions of authority based on loyalty and ability rather than kinship, and the implementation of a rapid horse relay communication system. Under Genghis Khan the army numbered about 95,000 soldiers, almost entirely cavalry (Barfield 1989, p. 192). The literature on the core of the Mongol empire speaks extensively of clans (approximately 35) and tribes that made up the foundation of the empire; however, it also is clear that these units were composed of people from many ancestries and do not conform to kinship-based definitions of clan and tribe (Atwood 2006, p. 210; Cleaves 1982; Sneath 2007, p. 167).

As the empire expanded, new layers of administration were added, especially to manage and exact taxes from the diverse agriculture-based societies of the southern regions (Allsen 2006). Typically, the Mongols employed local administrators who either operated relatively independently or under the supervision of an imperial representative, as both marginal incorporation and overlay organization. In China this amounted to a dual administration in which the local people were under native administration and the Mongols were under military authority.

The economy of the Mongol empire was eventually as diverse as the regions they conquered (Allsen 1991; Atwood 2004a; Morgan 2009). The original Mongol polity under Genghis Khan was based on pastoralism with a variety of resource inputs including some agriculture, as was the case with almost all of the other steppe polities discussed here. Added to this was an early control of trade networks across the steppe, which further fueled the growing empire (Allsen 1997).

In several parts of Asia, settlements and other places associated with the Mongol empire have been studied by archaeologists, including burial sites associated with the different divisions of the empire, such as the Golden Horde in Russia (Narozhny 2007). In the Inner Asian heartland of the empire, the capital city of Kharakhorum and other urban centers, palace sites, inscriptions, and burial sites have been studied (Perlee 1961; Rogers 2009; Rogers et al. 2005; Tseveendorj 1999; Ulambayar 2009). There are 27 sites alone attributed to the Yuan dynasty period in the Inner Mongolia portion of northern China (Gai 1999, pp. 2–74).

Kharakhorum served as the capital of the empire primarily from A.D. 1235 to approximately 1260 when Kublai Khan moved the capital to Shangdu (Xanadu) and later to Daidu (Beijing) (Steinhardt 1988). Major recent excavations by a Mongol–German expedition have supplied a wealth of new information about the architecture and history of the settlement. As the capital of the largest empire in the world, it was not particularly impressive to visitors at the time. For instance, William of Rubruck, an emissary on a religious mission from St. Louis at Acre, France, described Kharakhorum as a very active place but only about the size of the French village of St. Denis. He also noted that the monastery at St. Denis was ten times more impressive than the palace of the khan (Dawson 1955, pp. 183–184). Kharakhorum had an exterior wall with four gates and a moat enclosing an area of roughly 1 km2. Adjacent was the khan’s palace, also enclosed by a major wall. Recently, new interpretations have linked the location of the palace with the current Erdene Zuu Buddhist Monastery founded in 1586 (Erdenebat and Pohl 2009; Hüttel 2009; Roth et al. 2002).

Beyond the capital there are several sites and burials associated with the period of the Mongol empire, including a series of palace sites. It was common practice for the khans to shift their residence periodically, resembling a seasonal pastoralist migratory pattern (Boyle 1972; Shiraishi 2004). In some cases they used royal tents, but there were also sites with more permanent constructions from the 14th century. It has been hypothesized that Avraga was the place Genghis Khan was proclaimed Great Khan (Kato and Shiraishi 2005; Perlee 1961; Shiraishi 2009). Also potentially associated with Genghis Khan are the palace sites of Saari Keer and Khara Tün (Rogers 2009; Shiraishi 2002). There are at least three other sites that may be palaces associated with Ögödei Khan (A.D. 1229–1241), the Melikhen-Tolgoi Hill Palace, Doityn Balgas, site of the Giegen Chagan Palace (Fig. 5), and Bayan Gol (Shiraishi 2002). There are other palace sites associated with Möngke Khan (r. 1251–1259). Shaazan Khot, on the north edge of the Gobi Desert, is likely the winter palace mentioned in early texts and is located along a possible trade route between Kharakhorum and China (Moriyasu and Ochir 1999, plate 20a; Rogers et al. 2005, p. 810; Shiraishi 2004, pp. 113–114). The site of Erchuu Khot, northwest of Kharakhorum, is probably another palace (Perlee 1961, pp. 104–105). In the heartland of the largest empire on earth, none of the palace sites have defensive walls.

Fig. 5
figure 5

Plan view of individual buildings and central platform mound at the Doityn Balgas site, hypothesized to be the seasonal palace called Giegen Chagan, built by Islamic workers around 1237 for the Mongol khan Ögödei

In addition to the capital and various palaces, there were a variety of other constructions. Khar Khul Khaani Balgas, a probable administrative center on the Khanui River in north-central Mongolia, consists of ten enclosures, the largest of which had defensive walls higher than 3 m (Enkhbat 1986; Rogers et al. 2005, p. 809). In the Tuva region of southern Russia, Russian archaeologists have studied six urban sites relating to the Mongol empire. Sites like Den-Terek may have served as frontier outposts but also as agricultural, trade, and production centers (Kiselev 1965, p. 118; Kyzlasov 1969, p. 159). The large number of urban sites in the core of the Mongol empire, compared to that of earlier steppe polities, is indicative of some of the new perspectives and investments that paralleled the expansion of a world empire.

By A.D. 1260 a four-year civil war and the difficult logistics of maintaining central control resulted in the subdivision of the empire into four states, the Golden Horde in Central Asia, the future Ilkhanate in Iran and Iraq, the Chaghadai khanate in Inner Asia, and the Yuan dynasty in China. Each was connected but essentially independent with nominal central authority. Each was led by a member of the Genghis Khan lineage. The Yuan dynasty was the division of the empire under Kublai Khan (Sung 1937). The changes that the Mongols underwent in China are especially instructive for understanding the administrative issues common to many empires. The Yuan dynasty continued to expand into Southeast Asia and the Korean Peninsula. To effectively lead an empire that was now primarily Chinese and composed of settled agricultural villages and large cities, Kublai Khan learned to speak Chinese and adopted the trappings of a Chinese imperial court, even though inside the walls of the imperial palace Mongolian felt tents (gers) also were in use and shamans from the steppe continued to perform traditional rituals. The Mongols remained acutely aware that they were a small minority of foreigners and always vulnerable to overthrow by revolt (Rossabi 1987). In the census of 1290, the difficulty of maintaining control in Yuan China because of the sheer demographics was apparent. At most there were 2,000,000 Mongols and others from Central and Inner Asia, including those in their homelands and in China, whereas there were 70,000,000 Chinese, Manchurians, and Koreans (Dardess 1963).

In A.D. 1311 approximately one third of the Yuan dynasty revenue was being used to defend the Mongolian homeland and provide relief for hundreds of thousands of poverty-stricken families on the steppe (Barfield 1989, pp. 222–223). Historical sources suggest that for the average Mongolian family the benefits of building a vast empire had become more of a burden than a blessing. Archaeological skeletal research at Egiin Gol and Baga Gazaryn Chuluu provide some evidence for nutritional stress that may relate to this period (Nelson et al. 2009). For the next 50 years the Yuan dynasty was embroiled in suppressing rebellions and numerous internal challenges. In 1368 consolidated rival forces, as the Ming dynasty, seized the capital at Daidu and the Yuan court fled to Mongolia and constructed a fortress, believed to be the site of Kherlen Bars 3. This effectively ended the consolidated Mongol empire, although the other divisions of the empire continued with their own trajectories. In the steppe heartland of Inner Asia, the fall of the Yuan dynasty ushered in a long period of unstable conditions in which no one group was able to achieve control. Following the collapse of Yuan forces on the steppe, two competing groups eventually gained prominence—the western Mongols (Oyirad) and the eastern Mongols.

Oyirads–western Mongols

The western Mongols (Oyirads) were active participants in the western expansion of the Mongol empire. It was, however, only after the rise of their leader, Esen, that they became a consolidated expansionistic force 71 years after the fall of the Yuan dynasty. From the Altai Mountain region and Xinjiang, they expanded east and briefly held much of Mongolia and northern China. In the tradition of steppe politics, the western Mongols were a composite of four smaller groups that periodically exacted tribute from China while engaged in a constantly rotating set of alliances. Political control was dominated by a group of aristocratic lineages that oversaw distinctive classes within the population. Atwood (2006, p. 209) describes the political organization of the Oyirads as a “quad (marriage-partner)-keshig (imperial guard) dynasty linked to puppet Chinggisid khans.” The economy of the western Mongols was based on the core pastoralist traditions with a minor amount of agriculture. Almost nothing is known about the archaeology of this polity.

The western Mongols were often at war with the eastern Mongols. It was between A.D. 1439 and 1449 that Esen was able to consolidate much of the steppe, and in 1449 the western Mongols successfully defeated the Ming forces at the Great Wall. They captured the Ming emperor and held him for ransom. In 1453 Esen declared himself khan, but in 1455 he was assassinated and the empire disintegrated (Barfield 1989, p. 242). The western Mongols returned to the west where they maintained control over a region of Turkestan and surrounding areas. With the fall of the western Mongol polity, the eastern Mongols attempted to unify the region but were largely unsuccessful. There were, however, some powerful leaders who rose to power over the next century, such as Dayan Khan and Altan Khan of Chingissid descent (Crossley 2009).


The Zunghar polity is the last independent major steppe power (Courant 1912; Zlatkin 1964). At its height it encompassed a region that included portions of Central Asia, Kazakhstan, western Mongolia and surrounding areas of southern Siberia, and Xinjiang. The leaders of the polity incorporated western Mongol Oyirad history and descent from the leader, Esen, more than 125 years earlier, as part of their legitimacy. The leaders of three of the four groups within the Zunghar confederation were direct descendants of Esen (Courant 1912; Zlatkin 1964). Atwood (2006, p. 209) describes the Zunghar political structure as “a confederated pattern of several ruling lineages competing for domination, and linked by marriage alliance.” Such a system contrasts sharply with the single-lineage type of power structure found in the earlier Mongol empire. As with almost all of the other Inner Asian polities, succession of leadership was a great point of contention. In the Zunghar case, the leader Galdan favored the Inner Asian tradition of lateral succession, from elder to younger brother, but the Zunghars also had a tradition of direct succession from father to son. A lack of consensus on these alternatives meant that there were multiple claimants to the throne. The contest for control resulted in a very damaging civil war. Like all the other polities described here, the Zunghars were multiethnic and Oyirads were only a portion of the population. The state political structure was closely tied to the traditions of Yellow Hat Buddhism (Atwood 2006, pp. 232–235).

The basic economy of the Zunghar polity was certainly multifaceted. Pastoralism was the core; however, agriculture was very successful and widely practiced. There also was attention given to trade and to the manufacture of iron and cloth (Courant 1912, p. 67). There are several sites associated with the Zunghar polity, often representing settlements or fortresses reused from earlier times.

By A.D. 1625 the Zunghar polity had resolved internal conflicts and began a period of expansion. Some of these expansions were legitimated as supporting the growth of Buddhism. In 1640 the Zunghar leaders were instrumental in convening a great conference of steppe leaders. The outcome was a confederation with no single leader but a shared code of laws. Sneath (2007, pp. 182–183) describes the union as a “joint project of rulership by powerful aristocrats.” This confederation incorporated many typical aspects of states, such as courts, conscription, and centralized leadership (Atwood 2006). The Zunghars participated in the defeat of the last Tibetan king (1642) and the installation of the Dalai Lama as the leader of both Tibetan Buddhism and the state. In 1671 Galdan became the supreme khan, and he successfully expanded the Zunghar polity against the Central Asian Muslim oasis cities. By 1688 conflict between the east and west led to the breakup of the confederation. The Zunghars in the east still remained powerful and successfully annexed Mongol territories farther east, thus bringing most of Mongolia under their control. After initial successes, the Zunghar forces returned home to deal with internal conflicts. It was not until 1694 that the Zunghar armies advanced again on Mongolia, this time apparently because of a devastating famine. The Zunghars were finally defeated by Qing forces, apparently as a result of internal conflicts and a smallpox epidemic (Barfield 1989, p. 294). With the fall of the Zunghar polity, the political struggles of Inner Asia moved towards the power struggle between Russia and China and the formation of modern nation-states.

The state of the past

Like early states and empires elsewhere, the steppe polities provide a glimpse into the patterns of complexity and power relations that remain visible today as the social forces that structure much of the world around us. Although seemingly remote in time and place, Inner Asia provides important new dimensions to our understanding of how increasingly expansive polities faced organizational challenges. The principal arenas through which forces of organization are implemented can be summarized as including the technology of production (e.g., plant and animal domestication, tools, and weaponry), organizational strategies (e.g., the structure of social groups, specialization, the uses of ideology as motivations for collective action, legitimization of leadership, and control of materiality), social interaction (e.g., trade, information transfers, and exploitation of others), opportunities and limitations of the environment, and the contingencies of a particular history. Rather than considering reified social categories to identify an idealized type, like tribe, chiefdom, or state, it is valuable to consider each element independently. Select aspects of these forces of organization are evident in the trajectories of steppe traditions, but especially in how early polities brokered, managed, and built power emerging from social practices and institutions of authority.

The steppe polities shared core commonalities, but they also existed at differing scales and used a variety of organizational strategies. Their principal similarities are tied to the physical geography of the Inner Asian steppe, deserts, and mixed forests of adjacent regions, subsistence strategies founded on pastoralism, and a complex social geography of ethnic groups and political entities of varying scale, all adjacent to sedentary agricultural states with different cultural and organizational traditions. Primary states existed in China more than 1,500 years before the Xiongnu (Chang 1986, pp. 305–307; Liu and Chen 2003). The evidence, however, from the Bronze Age and the histories of the individual polities illustrates a unique orientation to empire-building that took inspiration from steppe antecedents (Houle 2010; Rogers and Cioffi-Revilla 2009). The evidence raises the question, long debated, about the relationship between primary and secondary states and the nature of expansion (Price 1978). It is often argued that pristine states are responsible for the formation of secondary states at their margins (e.g., Morris 2009, pp. 154–155). This may be true initially, but throughout the world primary states are rare and secondary states are the norm. History is dominated by situations in which traditions of statecraft already existed. In this sense, there is no theoretical reason to conclude that steppe polities must have formed as a result of Chinese influences.

Based on the histories of the individual polities and emerging archaeological evidence, the political and organizational structures of the Inner Asian polities developed primarily from steppe traditions with very deep roots, even those polities originating in northeast China at the eastern margins of the steppe. As these polities expanded and transformed their political structure, they also frequently adopted a range of administrative practices originating with the people they conquered or with whom they had direct contact. This ability to adopt new techniques and strategies was itself a major adaptive advantage for empire-building. Archaeological evidence from the Late Bronze and Early Iron Ages (1400–400 B.C.) confirms the existence of control hierarchies associated with wealth differences and sociopolitical power well before the emergence of expansive political entities. Much of this evidence comes from a variety of burial mounds that exhibit status differentials consistent with the types of hierarchies seen in chiefdoms (Erdenebaatar 2002; Fitzhugh 2009; Frohlich et al. 2010). By the beginning of the Xiongnu polity at 200 B.C., the leadership hierarchies noted in the early Chinese sources already had a very long history. The Xiongnu may have adopted some Chinese practices, but their overall political structure was of steppe origin.

For a polity, expansion is a time-honored, problem-solving approach that reduces infighting by identifying an external enemy or objective with the potential of offering collective gain. For the steppe polities, expansion was the opportunity to acquire more wealth in the form of loot or tribute and new territories with the potential for sustained wealth extraction. There is little evidence, however, that expansions took place because of population pressure or climate change, although epidemics (smallpox) or extreme weather events like droughts or snowstorms sometimes may have created local conditions that produced conflict with neighboring groups. Local events were sometimes identified in the writings of the Chinese historians as a proximal cause. The vast majority of the evidence, however, supports explanations in which social factors predominate.

The brief overview of each polity also confirms that conceiving of the basic social organization as composed of clans and tribes is an insufficient model for interpreting social and political structures. The people who developed the major polities had a history both before and after the nexus of empire. Through exploration of longer histories, a greater sense of continuity emerges along with the realization that the foundations of control, in the form of social hierarchies, are consistently present and are not the by-product of pressures or opportunities originating in China. Recent research by Sneath (2007) has crystallized an emerging set of challenges to how complex polities were formed and functioned on the steppe. Although Sneath’s book has engendered debate (Golden 2010; Kradin 2009b; Sneath 2010), he argues effectively that the concept of a kinship society, commonly used to describe steppe pastoralists, does not adequately consider the distribution of authority as “decentralized aristocratic power.” Rather than deriving from a fundamental kinship basis, those groupings translated as “clans” or “tribes” in early Chinese historical sources were often poorly understood composite corporate–political organizations with shared allegiance but not necessarily kinship (Atwood 2006, p. 218; Ecsdey 1972; Honeychurch, in press). In addition, terms like “clan” and “tribe” have specific definitions in the anthropological literature that do not match their usage in Inner Asia. Additional confusion is introduced since these terms and others, such as chiefdom, have figured prominently in the study of the evolution of culture and political economy (Bondarenko et al. 2003, p. 8; Johnson and Earle 1987; Service 1962, pp. 12–13; White 1959, p. 122). The organization of individual families into named groups that had the appearance of a kinship structure was a major function of statecraft (Sneath 2007, p. 3). There are benefits to maintaining the semblance of a kinship structure in that it provides an additional mechanism of social affiliation that tends to produce stronger ties of responsibility. This is part of the well-known process of generating legitimacy by inventing or rebuilding a history to match new objectives, described above as manipulation of continuity.

With the steppe tradition of political hierarchies, the polities also innovated new forms of administration, though they also sometimes adopted ideas originating from multiple external sources. Allsen (2006) notes three sources that played a strong role on the steppe: administrators from the conquered lands, merchants–former pastoralists who were familiar with steppe practices but had become agriculturalists, and city dwellers. Expanding polities everywhere used diverse sources of expertise to solve issues such as managing multiethnic populations, diplomatic complexities, tax/tribute collection, information gathering, and technological expertise. In some cases experts were recruited from neighboring sedentary states, but more generally they came from populations newly incorporated into the polity. In other cases expertise was imported in a massive way, such as the possible movement of agriculturalists to the Ivolga site complex in Buriatia to generate agricultural products for the Xiongnu pastoralists.

The benefits of bringing together diverse sources of expertise also had an important cultural dimension. Much has already been written about the Mongols as a foreign dynasty in China that incorporated many practices of the people they conquered. When considering the Mongol empire from the Chinese perspective, a foreign dynasty was not all that uncommon (Mair 2005). Especially in North China and into the Central Plain, there had been Xiongnu, Wuhuan, and Xianbei from northeast China, Tibetan, and other polities who had asserted control and established Chinese-style dynasties with large percentages of Chinese subjects. Allsen (2006) points out that accommodation to foreign overlords also was not a new phenomenon in regions farther west, such as Turkestan and northern Iran. In these regions, as in China, it was acceptable for local officials to participate honorably in the administration of the conquerors, thereby supplying a level of expertise that was typically not part of the infrastructure of the emerging steppe polities. Essentially, throughout the Mongol empire the administration was multiethnic and had strong ties to the merchant class and the local bureaucracy.

Spiritual expertise was not as widely utilized. Steppe polities rarely employed an ideology of expansion or consolidation based on the spread of religion or cultural practices, although such policies were common elsewhere in the world, as in the Aztec, Inca, Roman, and Spanish empires (Conrad and Demarest 1984; Goldstone and Haldon 2009). Most of the polities described here were instead tolerant of cultural and religious diversity. The Mongol empire was famous for religious tolerance (Atwood 2004b). There were, however, religions promoted within a particular polity, usually first adopted by the ruling elite. The Uighur leadership, for instance, adopted Manichaeanism as the state religion in the 8th and 9th centuries, but it is not clear how widely it was practiced by the multiethnic population. With the emergence of Tibetan Buddhism, however, the picture began to change (Khazanov 1994b). The Xi Xia polity (A.D. 990–1227) engaged in the promotion of Buddhism in the two centuries before the Mongol expansion (Dunnell 1996). In the successor states of the Mongol empire, Atwood (2006, p. 211) points out that the promotion of Buddhism in the east and Islam in the west essentially catalyzed an enduring division between western and eastern portions of Central and Inner Asia. Uzbeg, the khan of the Golden Horde, for instance, was initially rebuked when he suggested general conversion to Islam. Archaeological evidence from burials of the Golden Horde suggests that a process of Islamization occurred in the 13th and 14th centuries (Narozhny 2007). In Inner Asia the Buddhist traditions that emerged at the end of the Mongol empire were carried forward with expanding influence. The Zunghar polity (1625–1757) promoted Buddhism (specifically Lamaism), and in later times under Manchu rule the Buddhist monastery system became even more prominent (Bold 2001; Purevjav 1978). Even with the extensive adoption of Islam and Buddhism on the steppe, there still remained a widespread continuation of more ancient shamanistic practices involving worship of the gods of the sky and earth.

As religion was treated variably, so too was the actual structure of political power and administration. Generally, organization was expansive and not highly bureaucratized, existing with different degrees of centralization and forms of leadership. In most early states there is at least nominal power-sharing through councils of nobles or advisors. Often these councils had minimal political impact. Several of the steppe polities, however, utilized a form of collective leadership rather than a single central authority. The social control hierarchies that did exist were not a product of centralization but could exist without it.

As among the Mongol, authority in the Turk I and Turk II empires was indeed highly centralized and restricted to succession within the ruling lineage. In almost all Inner Asian polities, leadership succession might be lineal or lateral—the source of many internal conflicts. In other cases a council elected a central leader for a term of years, such as among the Khitan Liao (Wittfogel and Feng 1949, p. 571) and the Kazakh khanate (Sneath 2007, p. 187). In the Zunghar polity (1625–1757) and among the smaller Kalmyk and Upper Mongol confederations of that period, competing lineages linked by marriage alliances formed a kind of shared leadership (Atwood 2006, p. 209). Sneath’s (2007, pp. 181, 186–187) description of the “headless” state of 1640 as a political union involving the Zunghar polity, the Khalkha Mongols, and other related groups represents yet another form of distributed leadership. Within the steppe polities, the three common aspects of the political systems that fueled internal challenges to leadership were direct versus lateral succession, council-based election of leaders, and the distributed power base itself, which gave local leaders the opportunity to challenge or simply ignore central leadership.

Inner Asian aristocratic lineages were everywhere the source of rising leaders with larger political ambitions. The collective action described in canonical theory (Cioffi-Revilla 2005; Rogers and Cioffi-Revilla 2009), to either seize an opportunity or mitigate a potential disaster, is usually attributed in early historical writings to individual actors with special talents. Like the inscriptions and stelae raised to celebrate the accomplishments of rules everywhere, the early Chinese sources did not fail to record events and deeds attributed to charismatic leaders. These leaders were engaged in projects of emergent structure that would accommodate a control hierarchy capable of adding new groups, resources, and territory to a polity. However, the changes that accounted for the accrual or loss of social complexity are far too often attributed uncritically to the actions of a single individual. Leaders operated in a broader realm of motivation and opportunity, which produced a variety of outcomes that illustrate the interplay of process and agency (Flannery 1999). It is precisely in this mix of fast processes of individual decision-making and annual cycles and slow processes that occur over the course of decades and centuries, that the idea of a control hierarchy takes on a decidedly tangled character (Lane 2006, p. 117), existing across social levels in complementary and contradictory ways.


States are always an unfinished project in the process of becoming something different. As Sinopoli (1994, p. 162) points out, the projects of expansion and consolidation that constitute the emergence of complex polities usually happen in response to crisis and opportunity. The study of these processes is particularly challenging for archaeology. Polities are by definition political and economic entities that may not be directly connected to a distinct materiality—whether imperial monuments or the ephemeral traces of a herder’s camp. Also, they are entities that often expand and contract more rapidly than can be traced by the best available absolute dating techniques. For these and other reasons, the study of Inner Asian polities will always be an interdisciplinary venture that uses all available historical documents in combination with a rapidly expanding set of information and new ideas contributed by archaeologists and scientists working in a variety of allied fields. With these multiple sources of information in mind, this overview has sought the comparative grounding needed to interpret a complex 2,000-year history of unique events, trends, and broader patterns of change. Of necessity, emphasis has been placed selectively on competing theories of social dynamics, the sources of social power, the nature of tradition as a source of organizational knowledge, and the mechanisms of expansion.

As developmental typologies and standard interpretive dichotomies have dominated attempts at explanation, there is a growing recognition of the necessity of moving to more dynamic perspectives. Kradin, for instance, notes that the Jujan and other steppe polities display characteristics of both complex chiefdoms and simple states (Kradin 2005b, p. 166): this is a reflection of the “inapplicability of the terminology of settled societies to the history of pastoral nomads.” It also is a reflection of the overreliance on historical texts and the fact that these polities were organized in several different ways.

Most of what we know about the early states and empires of Inner Asia comes from historical documents, primarily from Chinese sources. With these histories, archaeology assumed its first role—to verify, locate, and fill in gaps. Regional archaeology is now moving rapidly beyond this first set of objectives toward development of novel interpretations and challenges to documentary interpretations. To facilitate this new level of innovation, there is a need to step beyond the confines of existing disciplinary boundaries at every level—from new techniques in information recovery to new theories of state origin.

Some of the most valuable work on how to understand emerging empires comes from studies aimed at in-depth interpretations of earlier Bronze Age phenomena (e.g., Chang et al. 2002; Frachetti 2008; Parzinger 2006; Shelach 2009). Our historical vision of the centuries before the Xiongnu is limited by the fact that Chinese chroniclers took little notice of the earlier political and cultural developments on the steppe. However, it was during the Bronze Age that aristocratic traditions emerged and the first significant efforts at creating collective power were implemented. Archaeology must take center stage to identify these origins and innovations.

Another long-standing set of issues revolves around the nature of the economic basis of the steppe polities, especially the role of agriculture. The evidence is now clear that the steppe polities practiced varied combinations of multiresource pastoralism. The role of agriculture is particularly at question here because of the perceived link between agriculture and the emergence of states and empires, at least in other parts of the world. Although always recognized as a part of the steppe economy, it now appears that at different times and places agriculture was even more important than generally acknowledged. Archaeology now needs to produce the hard evidence through studies of macro (seeds) and micro (pollen, phytoliths, and starch grains) paleobotanical materials. Such studies also will support paleoenvironmental reconstructions of vegetation and climate change. New detailed studies are needed at key sites to extend paleoclimatology research through lake cores, tree-ring dating, and sedimentation studies (e.g., Namkhaijantsan 2006; Panyushkina et al. 2010; Terasov et al. 1998).

Early polities amassed power and implemented strategies of incorporation as they expanded and sought to remain stable. A question still remains about how political solutions were implemented at the local level, something that is rarely discussed in the documentary record. In a landscape in which settlement was dispersed and left few traces, the emphasis has been on excavation of specific locations. Newer forms of landscape archaeology, however, are producing integrated regional settlement histories that allow new questions to be addressed. Some of the recent projects using full regional coverage include the Khovsgol Mound Project (Frohlich et al. 2010), Egin Gol Survey (Honeychurch and Amartuvshin 2007; Honeychurch et al. 2007), Khanuy Valley Project (Houle 2010), the Baga Gazaryn Chuluu Survey (Amartuvshin and Honeychurch 2010), work in Kazakhstan by Frachetti (2008) and Chang et al. (2003), and the Chifeng project in Inner Mongolia (Chifeng 2003).

Finally, where do the steppe polities fit within a larger comparative view of empires? Several answers to this question have been described in these pages. For instance, some have argued that the polities were ephemeral in their duration because of the volatile nature of steppe pastoralism; yet many ranged in duration from 18 to 354 years, with a mean of 155 years. This variability describes those political formations that were “successful” at achieving some level of stability, although many others never made it to that level. If the successful polities are compared to a worldwide sample, it becomes clear that Inner Asian states and empires were no more ephemeral than similar sociopolitical formations elsewhere in the world (Cioffi-Revilla et al. 2008; Sinopoli 2006; Taagepera 1997). Empire dynamics studied within a world-systems perspective consider scale and duration (Cederman 1997; Chase-Dunn and Hall 2000; Stein 1999), yet the steppe polities have been relegated to the periphery in these studies, partly due to the lack of easily accessible information for researchers from outside the region.

There is no single causal force that accounts for the emergence, change, and variation seen in the structure and function of early states and empires. Instead, explanations must be multifaceted. The fundamental conclusions I offer challenge theories of steppe pastoralist dependency on agricultural states by providing evidence for sustainable economies, diverse trade connections, and traditions of political power. Drawing a rigid analytical dichotomy between sedentary agricultural societies and pastoral societies is not productive, as histories show the intertwined nature of social change and empire development. The steppe polities were fundamentally influenced by mobility offered by the horse and the potential of movement associated with animal herding. Derived from mobility were forms of authority based on dispersed aristocratic lineages and deep traditions unique to the steppe regions—not originating in China or south central Asia. The strategies employed by most of the steppe polities were organizationally adaptive, utilized diversity, were militarily efficient, extracted resources, and were not highly bureaucratic. The degree of administrative centralization was variable and poorly correlated with long-term political stability. Not all of the techniques of administration, nor the idea of distributed hierarchies, are unique to the steppe polities; however, most theories of state formation argue that centralized authority is a key element of the successful polity. In Inner Asia leadership often took the form of hereditary kingship, but there was significant variation as to how authority was actually constructed and maintained. As elsewhere in the world, the steppe polities utilized a variety of strategies to expand and solidify their control, with an emphasis on those techniques that did not require direct or extensive intervention.

The techniques of statecraft described here offer a constellation of characteristics that expands our vision of how early socially complex systems operated. In recent studies of states and empires in other parts of world, there is still little recognition of how the pastoralist empires may have innovated new approaches to state formation (Yoffee 2005). Of the many characteristics discussed, the construction and maintenance of power is foundational for the existence of any polity. The Inner Asian polities utilized distributed power that cut across social practices and the traditionally recognized institutions of authority, such as centralized imperial authority. Coupled with this viewpoint is the recognition that the city as a central place or the productivity of agriculture does not always define an empire or a civilization. Instead, for Inner Asia, mobility, scale, extralocal interactions, nonfixed property, dispersed aristocratic control hierarchies, and the economics of multiresource pastoralism were an alternative foundation for complex societies.