Introduction

Soon after 209 BC, the unification of various tribes and kingdoms in the eastern Eurasian Steppe led to the formation of a large-scale, complex polity, often referred to as the “first nomadic empire” (Barfield 1981; Kradin 2001: 23; 2011: 77; Miller 2024). Known in the historical accounts of the Han Dynasty (202 BC–AD 220) as the Xiongnu 匈奴, they went on to dominate the Inner Asian steppes until AD 91. The growing power and status of the ruler and his clan in the Inner Asian political arena allowed them to take control of the main communication routes over a vast area and establish an extensive tributary system and diplomatic network, supported by regular marriage alliances (Di Cosmo 2004: 188–90; Chin 2010). All these changes affected the functioning and appearance of the centres of power in the steppe (Honeychurch and Amartuvshin 2006: 265–8).

The main Xiongnu political centre has been the subject of much academic debate, primarily focused on identifying its geographic location and establishing whether there was one centre or several, whether they were mobile or fixed places, and how the names recorded by other polities relate to archaeological remains (e.g. Wang 1983; Tan 1996; Yü 2008: 384; De Crespigny 2007; Bemmann 2011; Loginov 2012; Iderkhangai et al. 2020). Another theme has been to examine the political centre within the broader context of urbanism in the eastern steppe, since Han Dynasty chroniclers began referring to certain Xiongnu settlements as ‘cities’ (cheng 城) around the same time that urban sites appear in the archaeology. Studies on this theme have mainly linked materials on the Xiongnu to the emergence of urban centres among later polities (Danilov 2004; Rogers et al. 2005; Waugh 2010; Honeychurch 2015; Kradin 2019; Vasyutin 2022), whereas the functions of these sites and their connections to political authority in the Xiongnu Empire remain unclear.

Han written records represent the main sources on Xiongnu political centres; however, this information is yet to be the subject of focused analyses. Specifically, the Han sources use a variety of different names to refer to the political centre. Following the basic principles of the semiotic approach to analyses of historical narratives (Lotman 1990), including analyses from both synchronic and diachronic perspectives (Saussure 1959), we examine the terminology for rulers’ residences in the Han records and then compare this to the archaeological record attributed to the Xiongnu to verify the accuracy of the texts. The synchronic aspect involves compiling the list of names used by Han authors for Xiongnu settlements then analysing these comparatively and contextually to elucidate the structure, proper naming conventions, and connection to the sociocultural and linguistic environments, and to identify different names for the same settlements. The diachronic aspect involves determining the date of each name according to information provided in the texts or inference from general context and then comparing the chronologies and meanings of different names with known Xiongnu history, geography, and archaeology. This allows us to trace changes both in the names for rulers’ residences and in the settlements’ attributes over time and across regions.

1 Our Sources

The primary source of information is the historical records of the Han Empire: the Shiji 史記 (Records of the Grand Historian) compiled by Sima Qian 司馬遷 around 91 BC; the Han Shu 漢書 (Book of Han) compiled by Ban Gu 班固 and others circa AD 82; and the Hou Han Shu 後漢書 (Book of the Later Han) by Fan Ye 范曄 and others circa AD 445. In these texts, information about settlements is provided only tangentially when the authors describe the territorial and political structure of the nomadic state, its customs, and the most important military or political events, especially Han campaigns against the Xiongnu. The types of settlements are also limited to only those that are significant enough to Han politics, meaning they are typically the locations of rulers’ residences, with some infrequent mentions of residences of the nobility and border settlements. Thus, information about site location is usually incomplete, and the localization of most residences remains contested. Generally, none of these sources provide details about the settlement’s architectural appearance or the materials and technologies used in its construction, and sources sometimes contradict themselves. For example, the beginning of chapter 110 in the Shiji claims that the Xiongnu “move from place to place, following water and grass, [they] have no inner and outer city walls, fixed places to live, nor practices of cultivating fields” (逐水草遷徙，毋城郭常處耕田之業). However, the same chapter of the Shiji later mentions two cities and further confirms their existence in chapters 108, 111, and 112. Similar descriptions of the Xiongnu’s lack of fixed settlements are given in the Yan Tie Lun 鹽鐵論 (Discussions on salt and iron) by Huan Kuan 桓寬, which includes details of discussions at the Han court in 81 BC (ch. 52: 503–4): “the Xiongnu have no protection of inner and outer city walls, nor fortification of ditches and moats” (匈奴無城廓 之守，溝池之固), instead they “use mountains and valleys as inner and outer city walls” (因山谷為城郭); also, “the men do not carve ingenious items, and do not have skills of constructing palace buildings and inner and outer city walls” (男無刻鏤奇巧之事，宮室城郭之功). However, in the same way as the Shiji, Chapter 44 of the Yan Tie Lun contradicts these statements by mentioning the existence of a city. The situation is further confused by the simultaneous use of several terms, names, and transcriptions to denote settlements.

Archaeology presents another source of information. More than 20 structurally diverse, walled sites, some with palatial remains (which is an obvious indicator of a high-level political power), have been discovered in Mongolia, as well as Transbaikal, and Southern Siberia (Perlee 1957; Kyzlasov 2001; 2006; Turbat 2013; Danilov and Zhavoronkova 1995; Danilov et al. 2016; 2018; Eregzen 2017; Botalov and Battulga 2017; Ankhbayar 2017; Egiimaa 2019; Miller et al. 2019; Iderkhangai et al. 2020). Traditional mobile dwellings are also depicted in Xiongnu art (Miniaev and Sakharovskaia 2007). In contrast to the historical records, the archaeological data comprises settlements of all types. Not only do these have clear geographical locations, but they also provide evidence for construction technologies, materials, and architectural features.

Both forms of evidence offer information about Xiongnu rulers’ residences that the other lacks, and an interdisciplinary investigation of the two can provide a deeper understanding of architecture and settlement patterns, as well as the history and culture of the Xiongnu. Some settlements mentioned in the texts have even been identified with archaeological finds in recent years, such as one on the Orkhon River that produced roof tiles marked with the title of the Xiongnu ruler (Iderkhangai 2021). Similarly, a settlement with a Han-style palace in Southern Siberia—previously argued to be the residence of Xiongnu nobles Li Ling 李陵 (Evtyukhova 1946) or Xubu Dang 須卜當 (Zhou 1956), or the impostor emperor Lu Fang 盧芳 (Kovalev 2011)—has been demonstrated to be the northernmost residence of the Xiongnu ruler Zhizhi 郅支 for the period 48–43 BC (Esin 2022a; Miller 2024: 158–159).

2 Terms for the Chanyu’s Residence

In the historical records of the Han Dynasty, several Chinese terms are used simultaneously to designate the place where the Xiongnu ruler, the chanyu 單于, resided for long or short periods of time. These include quluo 區落, ying 營, ting 庭, cheng 城, du 都, and ci 祠. Three of these—ting, ci, and cheng—are used in compound names in multiple variations (Table 1), which complicates the picture. We use ‘residence’ as a general term for the variants discussed below.

View Full Size

The frequencies of compound names for Xiongnu rulers’ residences in Han-Dynasty historical records

Citation: Journal of the Economic and Social History of the Orient 68, 1-2 (2025) ; 10.1163/15685209-12341640

• Download Figure
• Download figure as PowerPoint slide

2.1 Quluo

The residence of the founder of the steppe empire—Modu Chanyu 冒頓單于 (r. 209–174 BC)—is called a quluo 區落 in chapter 23 of the Hou Han Shu, meaning ‘an encampment, a settlement [of non-Chinese tribes]’ (Oshanin 1983–1984, 2: 1259). This implies traditional Xiongnu dwellings made of a light-weight willow frame covered with felt¹ that apparently lacked the folding lattices of later yurts (Vainshtein 1976; Kryukov and Kurylev 2000). The name was transcribed from the Xiongnu language into Chinese as qionglu 穹廬 and qionglü 穹閭² (Shiji, ch. 110: 2913; ch. 27: 1338). These dwellings can be identified in Xiongnu art which provides more details of the design (see section 3.2). The term quluo appears in the account in a figurative comparison of two first chanyus’ residences, with the later one dating to the end of the 1st century AD. This comparison was part of a commemorative text compiled by Ban Gu in AD 89 after a successful Han military campaign, during which the chanyu’s residence north of the Gobi was captured and burnt: “thus, [they] passed over Zhuoye [Mountain], crossed the Anhou [River], ascended Yanran [Mountain], trampled the quluo of Modu, burnt the Longting of Laoshang (r. 174–161 BC)” (遂踰涿邪，跨安侯，乘燕然，躡冒頓之區落，焚老上之龍庭). On the way back into Han territory, the commanders Dou Xian 竇憲 and Geng Bing 耿秉 performed a sacrifice on a mountain, engraving Ban Gu’s text on a rock surface to commemorate their achievements (Hou Han Shu, ch. 23: 815). The rock inscription was recently discovered by Mongolian researchers (Battulga et al. 2016) on the lower reaches of the Ongi River (Fig. 1.a). The text on the rock, however, deviates slightly from that recorded in the Hou Han Shu (Qimudedao’erji and Gao 2019). According to the Han Shu (ch. 94b: 3803), Modu Chanyu’s place of residence was in the Yin Mountains 陰山.

View Full Size

(1–3) Locations of the main Xiongnu rulers’ residences, (a) the rock inscription composed by Ban Gu, (b) the Tsaraam cemetery, and probable identification of several toponyms: Zhiju shui 郅居 水 for Selenga river; Anhou he 安侯河 for Orkhon river; Yuwu shui 余吾水 for Ongi river; Zhuoye shan 涿邪山 for Gurvan Saikhan mountains; Tianyan shan 窴顏山 / 闐顏山, Yanran shan 燕然山 for Khangai mountains. The seats of the Han commanderies, Yunzhong 雲中, Dai 代, and Shanggu 上谷, are also marked. Red – Shiji; Green – Han Shu; Blue – Hou Han Shu

Citation: Journal of the Economic and Social History of the Orient 68, 1-2 (2025) ; 10.1163/15685209-12341640

• Download Figure
• Download figure as PowerPoint slide

2.2 Ying

In chapter 110 of the Shiji, the place where the chanyu lived is designated with the word ying 營, meaning ‘camp’ (Oshanin 1983–1984, 2: 2361). The text states that “in the morning, the chanyu comes out of camp [and] pays obeisance to the rising sun, in the evening [he] pays obeisance to the moon” (而單于朝出營，拜日之始生，夕拜月; Shiji, ch. 110: 2892). The contexts in which this word is used in Han records indicate that the camp was most likely military in nature. A variant of such a camp is described during the Han campaign against the Xiongnu in 119 BC: “And then the General-in-chief ordered covered war chariots to be placed in a ring to make a camp” (於是大將軍令武剛車自環為營; Shiji, ch. 111: 2935). The same camp arrangement is described in the Han Shu, when the Han commander Li Ling 李陵 established a military camp in Xiongnu lands in the autumn of 99 BC by arranging large, wheeled carts to form a camp: “by means of large wheeled carts [he] made a camp” (以大車爲營; Han Shu, ch. 54: 2452).

The method of arranging carts to form a defensive structure was also utilised by the Xiongnu. A noble’s residence protected by carts in front of the qionglu is depicted on a vessel from an elite Xiongnu burial in the Tsaraam Valley, Transbaikal, dating to the end of the 1st century BC–1st century AD (Fig. 2). The excavators suggest that this was the grave of the chanyu or one of his close relatives (Miniaev and Sakharovskaia 2007: 50). Due to its geographical location in the lands of the Dingling (Esin 2022a: 34–5, fig. 9), it could be the grave of the King of Dingling (Dingling wang 丁靈王). The design on the vessel may show either his own camp or that of the chanyu.

View Full Size

A birch bark box decorated with a depiction of the mobile residence of a high-ranking Xiongnu noble from Tsaraam cemetery. The biggest qionglu of the highest status (d) sits at the centre of several smaller qionglu (c, e) and covered carts (a, b, f) with flags (1). After Miniaev and Sakharovskaia (2007: fig. 11)

Citation: Journal of the Economic and Social History of the Orient 68, 1-2 (2025) ; 10.1163/15685209-12341640

• Download Figure
• Download figure as PowerPoint slide

It is possible that this type of camp used by some Han military leaders was an adaptation of nomadic warfare tactics, as seen in other aspects of military operations against the Xiongnu (Di Cosmo 2002: 273). Unfortunately, the spatial and chronological usage of the word ying as a chanyu’s residence is not defined in the Shiji. The general context indicates that the term ying was used to denote the ruler’s residence in the early period of Xiongnu history. However, mobile types of residence undoubtedly existed throughout Xiongnu history to accommodate the ruler and aristocracy during military campaigns and other trips. The same practice was true for the Han.

2.3 Ting

The place where the chanyu lived is most often designated with the character ting 庭 in Han accounts, and the most typical compound name with this term is Chanyu ting 單于庭 (Table 1). After the division of the Xiongnu into northern and southern factions in the middle of the 1st century AD, further specification was briefly given with Bei ting 北庭 or alternatively Shuo ting 朔庭 for the Northern Chanyu (Bei chanyu 北單于) and Nan ting 南庭 for the Southern Chanyu (Nan chanyu 南單于). Sometimes ting is combined with the name of the people themselves, Xiongnu ting 匈奴庭, or with the derogatory term lu 虜 ([northern] barbarian) to create Lu ting 虜庭.

According to the Shuowen jiezi 説文解字, a dictionary compiled in the Eastern Han period (AD 25–220), the word ting refers to the inner part of a palace (Shuowen jiezi zhu, 443). In the context of Han culture, ting is typically translated as ‘court’, as in the case of Han ting 漢庭 to mean the ‘Han court’ (Han Shu, ch. 94b: 3832; Taskin 1973: 66). Thus, the royal family and councillors, constituting a ruling power, were designated via their association with a palace by lexical semantic change. However, many publications translate ting as ‘headquarters’ or ‘encampment’ for Xiongnu contexts, which suggests that a nomadic ruler’s residence was more likely to be impermanent and consist of a yurt or tent rather than a fixed palatial structure (Bichurin 1950: 49; Kychanov 1997: 11–2). Even scholars who use ‘court’ assume that the structure was “presumably, a mobile encampment” (De Crespigny 1984: 507). A similar concept based on theories of the spatial politics of nomadic states in the eastern steppe understands the nomadic court as a group of elites that moved constantly from place to place (Honeychurch 2015: 311). Han writers evidently viewed the Chanyu ting quite differently from how modern researchers do—although they used the word ting to demonstrate a settlement’s political function, for them it was inextricably linked to a palace. But how does this fact compare with descriptions of the geographical location and other contexts of the word itself, as well as with archaeological data on Xiongnu settlements?

Across different periods of Xiongnu history, the geographical location of the ting varied greatly. Conversely, however, within these periods, the location was stable. In the Shiji, there is one definite description that relates to the 2nd century BC: “The court of the Chanyu is located opposite [i.e. to the north of] the Dai and Yunzhong [commanderies]” (單于之庭直代、雲中; Shiji, ch. 110: 2891). Thus, the court is described as lying northeast of the Ordos Loop in the Yellow River and south of the Gobi (Fig. 1.1). This place corresponds to the Yin Mountains, in which Modu Chanyu lived and no doubt his successors continued to live for several decades, but which were lost to the Xiongnu during the reign of Yizhixie Chanyu 伊稚斜單于 (r. 126–114 BC) due to increased Han military pressure (Han Shu, ch. 94b: 3803).

From the end of the 2nd century BC, the chanyus’ residence is described as lying to the north of the Gobi on the Orkhon River, which the Han Shu calls the ‘Anhou River’ (Anhou he 安侯河; Yu and Li 2014: 489). This relocation occurred around 121–120 BC, after Zici King 自次王, formerly Zhao Xin 趙信, Marquis of Xi 翕侯 in the Han, presented his plan to “reside north of the desert” (ju mu bei 居幕北) to Yizhixie, assuming the region to be beyond the reach of Han troops (Shiji, ch. 110: 2910), and Han records initially called the new residence Zhao Xin cheng 趙信城. Details on the routes to the Orkhon ting and surrounding toponyms have been preserved in several Han military accounts, especially those headed by Li Guangli 李廣利 (Han Shu, ch. 94a) and Dou Xian (Hou Han Shu, ch. 23). The remains of a chanyu’s residence were recently discovered on the Orkhon River at Kharganyn Dörvölzhin (Iderkhangai et al. 2020); its location corresponds with the description of the Chanyu ting north of the Gobi Desert (Fig. 1.2).

After the division of the Xiongnu into northern and southern states in the mid-1st century AD, the Orkhon ting remained in the hands of the northern faction but was destroyed in AD 89–91 during hostilities with the Han. Meanwhile, the Southern Chanyu established his ting near a bend of the Yellow River. According to the Hou Han Shu (ch. 89: 2943 and 2945), in AD 50 the Guangwu Emperor (r. AD 25–57) ordered the ting be established 80 li from the western part of the frontier (xibu sai 西部塞) in Wuyuan 五原 Commandery, potentially on the same place as where the earlier residence of Huhanye 呼韓邪 Chanyu (r. 58–31 BC) had existed in 52–42 BC. Soon after, the Emperor gave permission for the Southern Chanyu to live in Yunzhong Commandery, then, in AD 51, he ordered the ting be moved to Meiji 美稷 County in Xihe 西河 Commandery (Fig. 1.3). These residences of the Southern Chanyu, in addition to those of the first chanyus prior to the move to the Orkhon, still await discovery.

Though the Han sources suggest that the chanyus’ residences ting were more geospatially fixed and permanent than the modern theories argue, this does not preclude the use of traditional nomadic dwellings there. The Shiji (ch. 110: 2913) records that the Wuwei Chanyu 烏維單于 (r. 114–105 BC) hosted formal diplomatic receptions in a qionglu. This, however, does not mean there were no long-term constructions at the Wuwei Chanyu’s residence, as yurt-like structures could have been used alongside other types. An entry in the Hou Han Shu for the summer of AD 47, for example, mentions that an official meeting of Xiongnu chiefs took place in the Chanyu’s tent (Chanyu zhang 單于帳) during a traditional seasonal gathering at the ting (ch. 89: 2942).

Archaeological materials support the interpretation of ting as relating to a palatial structure, with Han-style palaces appearing in the steppe at a certain point. The Orkhon River residence is a prime example of this, with two main, south-facing palaces (Figs 3.a.1 and 3.a.2) and a large pond (Fig. 3.a.3) in the central part. These are enclosed by an earthen wall forming a rectangle with dimensions 250×210 m (Fig. 3.a.4) set with gates. The entire complex has an earthen outer wall with a perimeter of approx. 550×550 m (Fig. 3.a.5; Iderkhangai et al. 2020; Iderkhangai 2023). In addition to the extant, long-term structures, there is plentiful space to pitch yurts both within and without the central enclosure. The Han influence on the structure is clearly demonstrated by the construction materials used for the roof and inscriptions on it. The majority of finials of the roof tiles (wadang 瓦當) are decorated with moulding in small seal script auspicious characters: 天子單于千萬歲與天毋極 “To the Son of Heaven Chanyu—a thousand and ten thousand years, together with Heaven having no limit” (Fig. 3.b). Wadang of a later type (Fig. 3.c) without this elevated title (“To the ruler—longevity and loyalty of officials” 主壽臣忠) were discovered on the southern gate of the inner wall, which reflects the long-term use of the residence and its alteration.

View Full Size

(a) Plan of the Kharganyn Dörvölzhin walled site on the Orkhon river, 1 and 2: main palaces; 3: pond; 4: inner palace wall; 5: outer wall; 6: southern gate of the inner palace wall. (b) The main type of a roof tile finial (diameter – 17.5 cm) with moulded inscription with the title Tianzi Chanyu 天 子單于 ‘Son of Heaven Chanyu’, and (c) less common type with word zhu 主 ‘ruler’. After Iderkhangai (2021; 2023)

Citation: Journal of the Economic and Social History of the Orient 68, 1-2 (2025) ; 10.1163/15685209-12341640

• Download Figure
• Download figure as PowerPoint slide

Complicating the concept of ting as exclusively the chanyu’s residence is the fact that the same historical records use the term for residences of the highest echelons of Xiongnu aristocracy. Chapter 110 of the Shiji includes references to the ting of the Xiongnu kings (wang ting 王庭), and chapter 94a of the Han Shu mentions the ting of the Luli King of the Right (You luli ting 右谷蠡庭). The import of these residences is attested by palatial ruins at several sites outside the Upper Orkhon, especially in Eastern Mongolia (Miller et al. 2019: fig. 1) in the territory of the Empire’s most powerful kings of the Left. Similarly, a heavily fortified residence (with palatial building, inner (55 m side length) and outer (119 m side length) earthen walls, wooden wall, ditch, gates) has been found in the Lower Selenge Basin—Bayan Under (Danilov 2004: 37–8; Bessonova et al. 2020)—which could have belonged to the King of Dingling.

The earliest, narrowly-dated example of ting being used in relation to the Xiongnu in the Shiji relates to an incident in 124 BC, when the court of the Worthy King of the Right (You xian wang ting 右賢王庭) was seized (ch. 111: 2926), and slightly later in 119 BC when the Xiongnu were defeated: “After that, the Xiongnu fled far away, and there were no kings’ courts left south of the desert” (是後匈奴遠遁，而幕南無王庭; ch. 110: 2911). The Hou Han Shu also uses ting for the Laoshang Chanyu’s residence (see section 3.1).

2.5 Du

When describing the Xiongnu civil war in the middle of the 1st century BC, the Han Shu refers to the new political centres several times as du 都. During the Han Dynasty, this designated the main, capital city where the residence of a state’s ruler was located. The ancestors’ temple where the ruler made sacrifices was also an important part of a du. If there was no such temple, then the city was called yi 邑 (Shuowen jiezi zhu, 283). The word ‘capital’ was used for the residences of two competing chanyus after the Xiongnu split into separate factions:

遂復都單于庭。

After that [Huhanye] restored [his] capital to Chanyu ting (Han Shu, ch. 94b: 3796).

呼韓邪破，其兵走，郅支都單于庭。

Huhanye was defeated, his warriors fled, Zhizhi relocated [his] capital to Chanyu ting (Han Shu, ch. 94b: 3796).

These examples show that Chanyu ting for the Han author denoted not one or another elite group but rather a certain place, which was considered the permanent, fixed political centre of the entire Xiongnu state. It can be identified with the walled palace site of Kharganyn Dörvölzhin (Fig. 3).

Around 49 BC, Zhizhi Chanyu left Chanyu ting and soon founded a new du in the Jiankun 堅昆 Kingdom (Han Shu, ch. 70: 3008; ch. 94b: 3800). As in previous cases, use of du for this new political centre helped the Han authors distinguish it from the residence on the Orkhon River, though in one case this new residence of Zhizhi is also referred to as a ting (Han Shu, ch. 70: 3009). The new du has now been identified with the remains of settlement and a Han-style palace (Fig. 4) excavated on the Abakan River in Southern Siberia (Esin 2022a).

View Full Size

(a) Plan of the Abakan Palace based on documentation of the excavated area in 1940–46 (see Evtyukhova 1947 and Kyzlasov 2001) and (b) reconstruction of the front view. Probable function of the rooms: C1 – entrance hall; C2 – hall for receptions, gatherings, and various ceremonies (X – place for the ruler); A1, B1, D1, E1 – rooms for heating system operation, for servants and guards; C4, D4 – rooms for cooking; D3 – private room of the ruler.

Citation: Journal of the Economic and Social History of the Orient 68, 1-2 (2025) ; 10.1163/15685209-12341640

• Download Figure
• Download figure as PowerPoint slide

3 Dailin and outuo

Besides Longcheng, another major gathering mentioned in the Shiji took place at Dailin 蹛林, which was apparently located near the Chanyus’ residence in the Yin Mountains. According to Fu Qian’s 服虔 (fl. late 2nd century AD) commentary on an account of the same event in the Han Shu (ch. 94a), it was held during the eighth moon. Dailin has been suggested to be another city (Psarras 2003: 125) and possibly even the winter capital of the Xiongnu (Kyzlasov 2006: 145). However, it is more likely that Dailin was the name of a geographical feature, as dai 蹛 means ‘to surround, around’ and lin 林 means ‘a forest’, making it highly likely that this was the name of a forest (Watson 1993: 137; Kychanov 1997: 33). This is supported by chapter 2 of the Liao Shi 遼史 (History of the Liao), which mentions that, many centuries after the disappearance of the Xiongnu Empire, Emperor Taizu 太祖 of Liao (r. 907–926) “paid obeisance to the Sun in Dailin” (拜日於蹛林) near the former residence of the Xiongnu chanyus on the Orkhon River (Dmitriev 2009: 85). This toponym perhaps appears on the Orkhon after the Chanyu’s residence was relocated north of the desert, which also caused a change in the locations of the three main state rituals.

The same toponym is, however, also seen in other areas (Malyavkin 1981: 145), indicating that this ritual was not exclusively the prerogative of the ruler but also performed at the local level in different parts of the nomads’ lands. The name Dailin in this case may originate from the ritual known as dai. In chapter 94a of the Han Shu, Yan Shigu comments that the Xianbei 鮮卑 continued this autumn ritual, during which they rode horses rapidly three times around a forest or a ritual object constructed from upright branches (Kyuner 1961: 310; De Crespigny 1984: 508). A similar ritual is documented among the High Carts (Gao che 高車), a people descended from the Dingling. In the autumn, when the horses became fat, the High Carts rode their horses several times as a group around an altar on which many willow branches were placed vertically (Bei Shi, ch. 98: 3271).⁸

Researchers have typically viewed the autumn gatherings as an event related to taxation and therefore significant to the Xiongnu economy (Kychanov 1997: 33; Kyzlasov 2006: 145). However, well-fed, healthy horses and skilled riders will have been key to successful military campaigns, so such an event was a way for the chanyu, as well as other, lower-ranking chiefs of the polity, to assess available forces, as suggested by the Southern Chanyu’s letter of AD 88 quoted above in which he orders horses and warriors from all parts of his domain to attend the autumn sacrifice in preparation for a military campaign (Hou Han Shu, ch. 89: 2952).⁹

Another word encountered throughout the Shiji and Han Shu deserves attention—outuo 甌脫 (also written 區脫 in the Han Shu).¹⁰ Similar to long, it is a Chinese transcription of a word from the Xiongnu language. Early post- Han commentators considered it as a term and suggested it means an ‘earthen room’, ‘earth pit’, or ‘place of the border guard’ (Taskin 1968: 131; Kychanov 1997: 21; Chen 2016: 5–7). Some modern researchers have linked and continue to link it to the Turkic word ordu, suggesting it was a military camp and headquarters of the chanyu (De Groot 1921: 52; Taskin 1968: 132; Kradin et al. 2020: 44; Vasyutin 2022: 59). However, the Han Shu (ch. 94a: 3783) states that outuo was governed by a special king, the Outuo Wang 甌脫王, and contexts in the records show that outuo designated military camps on the borders of Xiongnu lands. The Shiji (ch. 110: 2889) mentions such camps on the border with Donghu 東胡 in the late 3rd century BC indicating that both nations had their own outuo (“each [of them] lived on their own edge [of abandoned land between Xiongnu and Donghu, where nobody had settled] making outuo” 各居其邊為甌脫), while the Han Shu—on the border with Han in the 1st century BC. Apparently, these military camps were located on traditional ways through the desert and served for defense from enemy attacks, control over the movement of tribes and livestock across the border (Han Shu, ch. 94a: 3788) and sometimes for carrying out frontier raids (Han Shu, ch. 54: 2465).

In 80 BC, during a raid, Outuo Wang was captured. Fearing that he would show the Han army the way through the Gobi, the Xiongnu retreated from the southern border and “sent people to encamp at outuo.¹¹ The next year, [they] again sent 9,000 horsemen to encamp at Shouxiangcheng¹² for defence against the Han” (發人民屯甌脫。明年，復遣九千騎屯受降城以備漢; Han Shu, ch. 94a: 3783). The use of the Chinese word tun 屯, which means to set up a military camp, reinforces the meaning of outuo as a military camp. In this context, tun outuo evidently meant to strengthen existing border camps with additional warriors. As with the camp near Shouxiangcheng, the records contain references to similar Xiongnu border camps but do not use the word outuo. For example, in 68 BC, “to protect against the Han, the Xiongnu again dispatched two encampments of 10,000 horsemen each” (又發兩屯各萬騎以備漢; Han Shu, ch. 94a: 3788). Both encampments were led not by the Outuo Wang but by other high-ranking military leaders.

The last use of outuo in the records dates to 43 BC (Han Shu, ch. 94b: 3801), but in Han terminology associated with the Xiongnu other similar transcriptions can be found. In 7 BC one such word appears slightly modified as part of a noble title Wen Outu Wang 溫偶駼王 (Han Shu, ch. 94b: 3810). This latter title designated a noble living on the Xiongnu’s southern border north of Zhangye commandery 张掖郡. Even later, descriptions of military events in AD 73–76 mention the title Wen (Yu/Ou)du Wang 溫禺犢王 (Hou Han Shu, ch. 89: 2949),¹³ which is a clear equivalent to the previous title. This title appears in relation to the southern border, with Wen (Yu/Ou)du Wang described as repelling enemy attacks with an army of several thousand north of Zhangye near the Zhuoye Mountains 涿邪山 (now Gurvan Saikhan Mountains). Wen (Yu/Ou)du Wang is further recorded at the same area as fighting the southern Xiongnu in AD 85 (Hou Han Shu, ch. 89: 2950) and the army of Dou Xian and Geng Bing in AD 89 (Hou Han Shu, ch. 23: 815). These later titles are all similar in structure and phonetic form, but most importantly they are all associated with the same border area and function. A similar title with a different third character for two military leaders of the Left and Right (which matches well with the meaning of ou 偶/禺) existed around the same time among the southern Xiongnu, namely Wen (Yu/Ou)di 溫禺鞮 (Hou Han Shu, ch. 89: 2944; Miller 2015: 161).

Analysis in synchronic perspective suggests that the title wang 王 was not part of the original Xiongnu title but was added for the Han people to clarify the individual’s high status (Psarras 2003: 127). The original title will have consisted of two components: (1) wen 溫 + (2) outuo/outu/oudu. Since we know that the second component referred to military units on the border, we can assume that wen indicated the role of the title bearer in their managing. His authority was similar to a Han wang, which caused wen to be dropped in some passages of the Han records to make Outuo Wang. Analysis in diachronic perspective shows changes in the transcription of the Xiongnu term for military camps: 甌脫/區脫 (late 3rd–middle 1st centuries BC) > 偶駼 (late 1st century BC) > 禺犢 and 禺鞮 (late 1st century AD).

The number of riders mentioned in the records gives some indication as to the maximum size of these camps. They could have been structured into discrete groups of traditional mobile dwellings. If post-Han commentators are correct, then such camps had earthen structures, feasibly earthworks and ditches, around their perimeters. From the end of the 2nd century BC, when the Xiongnu resettled north of the desert and before the collapse of the Northern Xiongnu, border camps were permanently located south of the Ongi River near the Gurvan Saikhan Mountains on a key route from the Ordos Loop of the Yellow River through the Gobi to Khangai. This area was adjacent to the Han fortresses outside the fortified line, which were used from 104–68 BC and the remains of which have already been discovered (Shiji, ch. 110: 2915; Han Shu, ch. 94a: 3787; Amartüvshin et al. 2011; Kovalev et al. 2011).

Modern linguistic analysis has shown that it is phonetically impossible to reduce outuo to ordu—the word is more comparable to the Proto-Turkic *кoλ-ut (singular *кoλ) meaning ‘encampments’ (Dybo 2007: 91 and 114). What is clear is that outuo was distinct in meaning compared to long, giving some idea of the system for designating settlement hierarchy in the Xiongnu language.

4 Discussion

The terms characterizing Xiongnu settlements can be divided into three groups based on the features they highlight. Those in the first group describe architectural features or components of a settlement (Table 2.A). Among them, qionglu is probably a transcription of a non-Chinese word¹⁴ for the traditional, domed dwellings made of willow branches and felt of the nomads. In the Hou Han Shu, however, this Xiongnu word was replaced by the Chinese words zhang (tent) and jimu 罽幕 (woollen tent), which were used as synonyms. The Xiongnu ruler also used long-term palatial buildings—danei—alongside the traditional nomadic dwellings. The existence of such buildings, as well as earthen and wooden walls, gates and ditches has been confirmed by archaeological excavations of a number of Xiongnu settlements. Wells and storage pits are not yet among the studied constructions but should be the subject of a search in the archaeological record at walled sites. Table 2 provides the most complete list of features according to the written sources, but of course not every Xiongnu walled site necessarily had all these elements. In addition, some features present in certain sites may not have been mentioned by the Han dynastic records, such as the pond (perhaps accompanied by a small park) at Kharganyn Dörvölzhin and Gua Dov sites.

View Full Size

Summary and classification of terms describing Xiongnu settlements in Han-Dynasty historical records. Black: Chinese words; Green: Chinese transcriptions of Xiongnu words

Citation: Journal of the Economic and Social History of the Orient 68, 1-2 (2025) ; 10.1163/15685209-12341640

• Download Figure
• Download figure as PowerPoint slide

Terms for designating settlements form two groups. One includes names associated with the form and structural elements of settlements (Table 2.B). Among them are cheng, indicating the presence of long-term defensive walls, quluo—as a name for a traditional Xiongnu settlement, and ying—camp. All of them evidently included dwellings in the form of felt tents, but cheng could also have long-term buildings such as palaces. Ying undoubtedly had some kind of protective fence, for which, for example, carts placed in a circle might be used. It is difficult to say what was meant by quluo. One possibility is that it was a settlement similar to ying, but without an external fence. For Xiongnu military camps, the records also used a term from their own language—outuo. In the Shiji this latter is undoubtedly a term associated with the form of settlement. In the Han Shu, however, it seems that this name is rather applied not to any military camps, but to camps in one specific location; that is to say, it sometimes gets the value of a place name for the Han authors. If the characters for this transcription were selected considering the correspondence of their meaning to certain aspects of the designated object (as in many other cases), then the earliest ou 甌 (bowl) most likely indicates the round shape of such a camp, which, therefore, is similar to ying.

Another group is formed by terms indicating the special status/function of settlements (Table 2.C). Among them is the Xiongnu’s own name for the ruler’s settlement—long. Han authors perceived this latter more as a place name, so they constantly combined it with some Chinese terms into compound names. It has two Chinese equivalents: (a) ting ‘court’, indicating the location of the ruling family along with close associates; (b) du ‘capital’, denoting the political, economic, religious centre of the country. Another term used was ci, which specifically referred to the ritual function of this political centre.

The most common words for the ruler’s residence—ting, cheng, and ci—are usually part of a compound name. At some point, all three words are combined with the transcribed Xiongnu word long, forming Longting, Longcheng, and Longci. The repeated use of long suggests that these three names may designate the same settlement. In general, compound names for the rulers’ residences have two parts: 1) a term for the settlement form or function (cheng, ting, ci); and 2) an individual qualifier to specify the particular settlement from the Han point of view. This individual qualifier came from four potential themes: 1) the Xiongnu term for a ruler’s residence; 2) the title of the Xiongnu ruler; 3) geographic or ethnic characteristics; and 4) the name of the founder or owner (Fig. 5). As a functional term, long for the Xiongnu was as applicable to traditional mobile settlements with felt dwellings as to sites with earthen walls and palaces. For the Han, however, there were some restrictions in the use of terms. In particular, for the Han cultural milieu, the concept of ting was inseparable from a palace. Similarly, du and ci also had associations with long- term structures.

View Full Size

Schema of compound name construction for Xiongnu rulers’ residences in Han-Dynasty historical records: the central triangle shows common terms for the settlement form or function and the circles show words used to specify and contextualise specific places. Grey: Chinese words; Green: Chinese transcriptions of Xiongnu words; Bold: typical combinations. The frequencies are indicated in red (Shiji), green (Han Shu), and blue (Hou Han Shu), see also Table 1

Citation: Journal of the Economic and Social History of the Orient 68, 1-2 (2025) ; 10.1163/15685209-12341640

• Download Figure
• Download figure as PowerPoint slide

All six Chinese terms represent an interpretation of Xiongnu settlements from the perspective of Han language and culture. There are several possible reasons that different words were used, including the existence of different settlement types, changes to settlement structure over time, and the individual observations or preferences of different compilers and different authors in different periods, whose primary reports were later used to compile the Han Dynasty records.

Differences in the meanings of the terms over time (Fig. 6) could reflect chronological changes to the appearance of the settlements. Thus, in his memorial text on the military campaign against the Xiongnu, Ban Gu calls the residence of Modu a quluo, but refers to the residence of his successor Laoshang as a ting, creating a chronological distinction between the residences (Hou Han Shu, ch. 23). These may reflect information available to Ban Gu from Western Han period documents, and, if so, reflect real changes in the arrangement of the chanyu’s residence, with palatial structures appearing. A few decades after the reign of Laoshang, by 124 BC, the upper echelons of Xiongnu aristocracy also had palace structures that were referred to in the Han records as ting.

View Full Size

Chronological usage of the Chinese terms for Xiongnu rulers’ residences in Han-Dynasty historical records.

Citation: Journal of the Economic and Social History of the Orient 68, 1-2 (2025) ; 10.1163/15685209-12341640

• Download Figure
• Download figure as PowerPoint slide

In reference to narrowly dated events, the word cheng appears later than ting, being used for the first time for the proper name Longcheng around 134 BC when plans for a military campaign against the Xiongnu were being discussed at the Han court. However, since Longcheng was perceived to be the Xiongnu’s permanent political centre, the settlement had presumably already existed for more than one decade.

Ci is the latest term to be used for a chanyu’s residence, appearing in Han sources only during the Eastern Han. Use of ci emphasises the residence’s ceremonial value and may even reflect the emergence of special temples at the ruler’s residence. Such temples, probably constructed in the late years BC– early years AD, have recently been identified at the site of Talyn Gurvan Kherem near the chanyu’s residence at Kharganyn Dörvölzhin (Ochir et al. 2020). However, it is most likely that the use of ci arose through increased exposure of Han culture to activities and ceremonies held at the Southern Chanyu’s residence in Xihe Commandery. Unfortunately, the remains of this residence have not yet been discovered.

The question of how сheng-type settlements appeared in the Xiongnu realm deserves special attention. In particular, a new residence built for Yizhixie Chanyu is connected with the defector Zhao Xin, who became the Chanyu’s closest adviser, brother-in-law, and a noble of high rank after he defected back to the Xiongnu from the Han Dynasty in 123 BC. On his return, he brought a detachment of 800 people, who potentially participated in the construction of what the Han initially referred to as Zhao Xin cheng (Shiji, ch. 110: 2910), distinguishable from the previous residence south of the Gobi, which circa 140–120 BC was known as Longcheng at the Han court. The name Zhao Xin cheng also appears in the Han Shu in connection with military events in 90 BC (Han Shu, ch. 94a: 3778), though Zhao Xin had already died around 108 BC. It is clear from the description that the city was located on the west bank of the Orkhon River and was at that time the residence of Hulugu Chanyu 狐鹿姑單于 (r. 96–85 BC).

At the end of the 80s BC, under the leadership of the former Han ambassador Wei Lü and in collaboration with ‘Qin people’ (Qin ren 秦人)¹⁵ to defend against Han troops, the construction of several cheng began¹⁶ (Han Shu, ch. 94a: 3782). Apparently, the same ‘Qin people’ who would defend these settlements were also involved in their construction. It is possible that in the 90s or 80s BC Wei Lü reconstructed as well the former Zhao Xin cheng, decorating it anew with roof tiles stamped with the title ‘Son of Heaven Chanyu’ (Fig. 3). This title is never used in the Shiji, appearing for the first time in the Han Shu (ch. 94a: 3751). In one of his letters to the Emperor, Hulugu uses a similar but alternative structure, referring to himself as the ‘Beloved Son of Heaven’ (Tian zhi jiao zi 天之驕子; Han Shu, ch. 94a: 3780). This suggests that the title was in use during Hulugu’s reign and went out of use before 31 BC, when the chanyus started to style themselves as ‘Divine Chanyu’ (Ruodi chanyu 若鞮單于) (for the translation, see Pulleybank 1966: 19; Dybo 2007: 112). For now, however, it cannot be completely ruled out that the title ‘Son of Heaven Chanyu’ may have already appeared on the wadang during the construction of the new residence in the reign of Yizhixie, but was not mentioned in the Shiji for political reasons. Wadang from the inner wall’s southern gate with a neutral designation apparently appeared after Huhanye Chanyu returned to the Orkhon Valley at the very end of the 40s BC. Since he recognised his lesser status relative to the Han Emperor, it was impossible for him to use the same title. Also, after a long political crisis among the Xiongnu, he was in dire need of the loyalty of chiefs from the northern and western parts of his empire previously supported his rival Zhizhi which well corresponds to this wadang’s last characters.

Since people from the Han Empire played a key role in building later cities with walls in Xiongnu lands, they were probably also involved in the appearance of the first Xiongnu city south of the Gobi, Longcheng. One possible scenario is that the courtiers who accompanied Han princesses dispatched to the chanyu’s residence as part of the heqin 和親 marriage strategy were potentially involved, since a similar situation is known for a herding society close to the Xiongnu in Wusun 烏孫 Kingdom. After the arrival of the first Han princess and her several hundred-strong entourage circa 105 BC, the first known, Han-style palace was built there: “The princess, on reaching their [Wusun] kingdom, herself constructed palace buildings to reside, once or twice a year together with the Kunmo¹⁷ held a gathering, set out a banquet, and bestowed the king and noblemen of the Left and Right with valuables and silk” (公主至其國，自治宮室居，歲時一再與昆莫會，置酒飲食，以幣帛賜王左右貴人; Han Shu, ch. 96b: 3903). In fact, her palace became the location for at least one seasonal gathering of the Kunmo’s court per year, similar to those held by the chanyu. As we know from the same source, in her deep longing for her homeland, the princess composed a song in which she mentions her life in a qionglu,¹⁸ but this, undoubtedly, was not her permanent dwelling. Several decades after the palace appeared, the same residence came to be called Chigu cheng 赤谷城 ‘City of the Red Valley’ (Han Shu, ch. 96b: 3906–7).

The first Han princess sent to the Xiongnu in 198 BC had higher status than the one sent later to the Wusun: she was introduced to the Xiongnu as the eldest daughter of the emperor’s main wife. The emperor and his court had great hopes for her marriage and long-term plans to subjugate the Xiongnu through family ties and generous gifts (Shiji, ch. 99: 2719). Her retinue had to constructed a palace for her at the chanyu’s residence too, though in the Shiji there is no such detailed information about princess for the Xiongnu as in the case of the Wusun in the Han Shu. During the reign of Laoshang, this palace may have become the summer residence of the Chanyu himself and was given the name Longting in Han records. Soon after, the settlement became bigger and was referred to as Longcheng. A Chanyu advisor from the Han court, Zhonghang Yue 中行說, could also have played a certain role in the arrangement of the residence at that time.

Among features of the Xiongnu cities in the historical records, the palace, rammed earth walls, and silk flags indicate significant Han influence on their architecture, materials, and construction methods. This is substantiated by technologies identified at archaeologically excavated palaces and settlements, where the mass-produced roof tiles, a process developed in dynastic China, are especially indicative. Simultaneously, adaptation of Han traditions and technologies due to a different cultural environment, as well as natural conditions, took place. For example, Abakan palace lacks the ubiquitous earth platform and had specific, underground heating system (Kyzlasov 2001: 61–87). Roof tiles at Terelzhiin Durvulzhin and other places sometimes have the wave-line design specific to Xiongnu pottery decoration and squared impressions in the central part of the wadang similar to the technological features on the bottom of their pottery (Danilov and Tsydenova 2011). These indicate that the craftspeople who made Xiongnu pottery were potentially involved in the production of roof tiles. The application of the ruler’s title and noble clan-emblems on the roof tile finials also reflects the specific cultural traditions of the Xiongnu. In addition, on the tiles from the Abakan Palace and from Kharganyn Dörvölzhin, clan-emblems and images were sometimes drawn on the raw clay before firing by those who made them or controlled the manufacturing process; both the drawings and emblems related to the traditions of the nomads (Kyzlasov 2001: fig. 40–41; Iderkhangai et al. 2022; Iderkhangai 2023).

5 Conclusion

The existence of palaces and cities with fortifications, demonstrated by both written and archaeological sources, contradicts the Shiji’s and Yan Tie Lun’s claim that the Xiongnu have no such objects. As has been discussed in this article, the Han historical accounts generally show that only the early Xiongnu, before the reign of Laoshang, can be considered to have had “no cities”. This corresponds with analyses of the Shiji’s textual construction, which substantiates the fact that the information in this passage was based on a pre-Han text (Di Cosmo 2010: 312; Goldin 2011: 233 and 240). At the same time, such a view represents the sedentary authors’ general impression of ways of life among the main body of the Xiongnu. It was characteristic of many members of the Han elite, and would persist for many centuries, even in relation to mediaeval nomads.

The analysed terms include six Chinese and two Xiongnu words that were used to refer to Xiongnu settlements in Han historical records. They can be divided into two groups: (1) those that characterise the visible form and structure; and (2) those that characterise the function.

The first group includes: (а) quluo, indicating a special, non-Chinese type of nomadic settlement—encampments with traditional mobile structures; (b) ying, denoting a settlement in the form of a mobile military camp; (c) outuo, meaning military camp in the Xiongnu language; and (d) cheng, which highlights the presence of permanent walls in the settlement.

The second group includes: (a) ting, denoting the inner space of a palace where the ruling family and their entourage resided and hosted major sociopolitical engagements; it was also possible for high-ranking aristocrats of the wang status related to the chanyu to have their own ting; (b) du, denoting the city with the state ruler’s residence and the centre of state power in all its aspects—political, economic, religious; (c) ci, specifically highlighting the religious significance of a settlement as a place of ritual veneration for noble lineages and ancestors (including the Han Emperor during the Southern Xiongnu), as well as deities and spirits that legitimised Xiongnu rule, its rulers and bequeathed prosperity to the state; and (e) long, denoting the residence of the ruler in the language of the Xiongnu themselves, which is partly equivalent to the Chinese terms ting and du.

The six Chinese terms reflect the authors’ interpretations of the residence of the Xiongnu ruler from a Han cultural perspective. This made it possible for different terms to be used in parallel to refer to the same settlement, highlighting its different aspects. At the same time, this research has demonstrated chronological trends. The earliest residences of the Xiongnu rulers are designated by the terms quluo and ying. The term ci, on the contrary, was used only in the period of Eastern Han. The term cheng began to be used in the second half of the 2nd century BC but from the end of the 1st century BC it was completely replaced by the term ting, the first dated mention of which is associated with Laoshang Chanyu. The frequency of use of each of these terms in the records shows that Han authors considered the key feature of Xiongnu political centres to be the presence of a court with palatial buildings as its attribute rather than a city wall or the collective sacrifices conducted there. In many cases, the authors emphasise the political status of the settlement further by merging ting and cheng with the title of a ruler, and geographical (north, south) or ethnic (Xiongnu, lu) characteristics.

Unlike the Chinese terms ting and du, the concept of long arose in a nomadic pastoral society centred on the seasonal movement of settlements, and thus it does not show any connection to the built environment. The main method of governance in such a society was the regular gathering of the nobility for the most significant rituals of the annual cycle, organised by the ruler. The seasonal movements of the ruler’s headquarters (Chinese terms in the records for this type of residence—quluo and ying) were limited to the central part of the country with regular places for different seasons. With the emergence of palaces and permanent walled settlements among the Xiongnu, this tradition did not disappear, at least in the early stages. The holding of three annual rituals in different places and the opposition of Chanyu ting and Longcheng in one passage of the Shiji reflects precisely this practice. At the same time, from the standpoint of the Han cultural environment, the use of the words ting and cheng in the names means that both of these seasonal settlements had earth/clay architecture, but the second, summer one was larger. Consequently, the use of a new type of constructions was adapted to the seasonal rhythm of life and other traditions of the steppe society: each settlement was used by the ruler only for part of the year, and the rest of the time its population was reduced. The summer settlement was apparently more populated, where some of the ruler’s wives, children, other relatives, and probably also some officials, craftspeople, high status foreigners, and envoys could live permanently. In later sources, however, the contrast between the names disappeared: in the 1st century BC the same settlement on the Orkhon River is called both ting and cheng, and then only ting. There is no such opposition among the southern Xiongnu either. Perhaps by this time the rulers began to conduct rituals of the 1st and 5th months in the same residence, and the autumnal ritual in a location close by.

Although traditional yurt-type dwellings remained in use throughout the Xiongnu Empire, the main ruler’s residence at certain point came to be modelled on Han palaces. Complex technologies for the production of all necessary materials and the construction of both palace-type buildings and other structures could not have appeared in the steppe without the participation of people who were familiar with them and had mastered aspects of their construction. This is even more evident than in the case of the construction of another type of complex architectural object—the terraced tombs of the Xiongnu nobility (Polos’mak, Bogdanov 2015: 118). The first palace was most likely built by a Han princess’ retinue when she married the Xiongnu ruler in the early 2nd century BC, which, perhaps, later became the first Longcheng. Over the next century, Han immigrants, like Zhao Xin and Wei Lü, continued to play key roles in the construction of palaces and long-term fortifications relying on the experience and labour of other people from Han territory, of whom there were quite a few among the Xiongnu. Local people were also involved at different stages of the construction processes which together with other factors opened the way for the adaptation of new technologies. However, the construction of such objects corresponded to the interests, resources and wills of chanyus themselves, otherwise they could not have appeared. Along with exotic trade items like silk textiles, rare jewellery, and Chinese chariots found in the tombs of the Xiongnu elite, Han-style architecture began to be used by the cosmopolitan steppe rulers for their own cultural politics as a symbol of status and power.