1This thesis aims to explain how the ruling elite of the Khitan Empire (916-1125) structured its web of matrimonial alliances. It explains how the identity of the mother played a fundamental role in the organisation of the family, which in turn influenced all marriage strategies, and that the Khitan “multiethnic” elite was divided into two intermarrying moieties.

Historical context: the Khitan Empire (Liao)

2The Khitan were nomadic people who occupied the region that spans the two sides of the modern frontier between Inner Mongolia and Liaoning. Their language belongs to the Mongolic (or Särbi-Mongolic languages) language family. Although this group existed from the 4^th to the 14^th century, we only have precise data from the 10^th to the first half of the 12^th century. It was during this period that the Khitan established an empire that, through Chinese historiography, became known as the Liao dynasty. The Khitan Empire lasted from 916 to 1125. It appeared when the Khitan qaghan Abaoji 阿保機 (*Abogin in Khitan) took the Chinese title of huangdi, therefore creating a status above that of qaghan. Under his reign and that of his son Taizong 太宗 (r. 927-947), the Khitan conquered a vast territory incorporating most of the present-day Dongbei, the Mongolian Plateau, and the northern edges of China, stretching as far south as modern-day Beijing. The subjects of the Liao empire were divided into four categories: the Khitan, the Qai (Xi 奚), the Han’er 漢兒 (i.e. the Chinese), and the Bohai 渤海.

3The elite mostly comprised of Khitan, with a significant Chinese minority, and several Qai and Bohai families. All emperors had the family name “Yelü” while all empresses (except one) were “Xiao”. Imperial marriages were set up to ensure that the empress was exclusively a woman from the Imperial Uncles. Jennifer Holmgren has previously explained the origins of this peculiar Yelü-Xiao alliance. When it became clear that Abaoji would not follow the tradition and limit his reign to three years, his brothers revolted against him in 911. To reinforce his power against his brothers, Abaoji gave extensive privileges to his in-laws. His wife Empress Yingtian seized power when he died in 926 and imposed her own will on the succession. The status of Yingtian’s family increased under the second reign and became equivalent to that of the imperial house. From this time on, the two families entered into an intermarriage relationship, which still left room for unions with other clans.

4Members of numerous clans started to use the Chinese surname “Xiao” during the second half of the 10^th century. Aisin-Gioro Ulhicun clarified that “Xiao” empresses all came from one clan named the “Imperial (Maternal) Uncles Tent,” a parallel to the “Horizontal Tent” of the Yelü emperors. The “Tents” (Kh. ger; Ch. zhang 帳) delimited the clan in Khitan society, but only the two mentioned above were not under the authority of a section (Kh. *niour; Ch. bu 部), which acted as the first administrative level for the Khitan and Qai people. They and other “tents” served as the main conceptual framework for the localisation of individuals within the elite.

Historiography and corpus

5Studies on the Khitan Empire are amid renewal. During the last century, hundreds of funerary inscriptions (muzhiming 墓誌銘) and other stone inscriptions dating from the Liao period have been unearthed or rediscovered. They often yield significant genealogical and biographical information which has permitted the reconstruction of genealogies and marriages of the most prominent clans. In parallel, the discovery of several tens of Khitan Small and Large Script inscriptions has permitted remarkable advancements in deciphering the two scripts. The use of both Chinese and Khitan epigraphy proved to be essential to completing the otherwise partial knowledge gathered from transmitted documents. Nonetheless, the 14^th-century Liaoshi 遼史, or “History of the Liao”, currently remains the backbone of our knowledge of the history of the Khitan Empire. This book, which was written more than two centuries after the fall of the Khitan empire, contains many valuable accounts, although several sections are fabricated or inaccurate. These results have mainly been achieved through reading older source materials, such as the Qidan guo zhi 契丹國志, or “A monograph on the Khitan state”, and various descriptions and reports written by Song emissaries and travellers. All these documents concur with our understanding of the Khitan elite’s internal organisation and the empire’s political situation, particularly in the 11^th century.

6The reconstruction of genealogies has led to a better understanding of the marriage alliances between clans. Jennifer Holmgren, Aisin-Gioro, Qi Wei, Shi Fengchun and others have studied political implications and the marriage system of the Khitan elite in great detail, but have tended to focus on one isolated clan. To date, a systematic study of the web of marriages that formed the Khitan aristocracy has yet to be attempted. This thesis is by no means an exhaustive account of the marriage alliances of the Khitan elite; however, it aims to provide an overall picture of the marriage relations, clan organisation, and matrimonial relationships between the Khitan and non-Khitan clans. As diplomatic marriages constituted a minor element in Liao foreign policy, they have not been included in this study.

7To better identify the different layers of family units, this thesis has used a diverse vocabulary. In Khitan terms, the clan is called a ger, which I translated following the Chinese zhang 帳 as “tent”. As for a ger within a ger, I called it a “house” following the Chinese fang 房. The main subunits are the branches, which then divide into lineages, then households. Modern research on Khitan noble clans often uses the units mentioned in the sources (these are the branches), which are only relevant for the early to mid-10^th century. As the branches grew, the number of individuals and lineages continuously increased, therefore, to maintain clarity this thesis considers a lineage as a family group of two to four generations with a coherent marriage strategy during a given period.

Structure of the thesis

8This thesis is divided into six chapters, with an introduction and a conclusion. A general history of the Khitan Empire and an overview of the sources are included in the introduction, with a discussion on the diverse problems a reader of the Liaoshi might encounter (pp. 39-50). This short presentation of the content of these chapters is herein followed by a general discussion of the contents of this thesis and its conclusions.

9The first chapter describes, through a comparison with the Mongol and Yuan cases (presented in pp. 72-87), the marriages of the Khitan elite and imperial family. The cases of the family of Yelü Shuji (pp. 88-97, 112) and of the matrimonial dynamics of the imperial family (pp. 97-129 passim) are given special attention. This chapter presents a general overview of the principles elite Khitan families abided with when contracting marriage alliances between themselves.

10The second chapter focuses on the early history of the Horizontal Tent (pp. 136-157) and of the Imperial Uncles Tent (pp. 157-173), more precisely on how they formed under the reigns of the first two emperors of the Liao. This is followed by a discussion of the origins and significance of the Chinese surname “Xiao” as it started to be applied to the “Imperial Uncles” shortly after the formation of the Imperial Uncles Tent in the second half of the 10^th century (pp. 173-189).

11The third chapter examines how these two “Tents” acted as more than just clans. It also demonstrates that the references to Tents in the Liaoshi are fabrications of the 14^th‑century chroniclers who misunderstood their sources (pp. 192-214 provide an in-depth critical reading of the sections relating to the Tents in the “Baiguan zhi” 百官志). Epigraphic and transmitted sources are then read concurrently to discern several aspects of the “Tents”, which indicate that they formed a noble unit within the aristocracy.

12The fourth chapter delves into the history of imperial marriages during the “short 10^th century” of the Khitan Empire. The death of Xiao Siwen in 970 brought to an end half a century of Yelü succession struggles intertwined with lineage rivalries within the Imperial Uncles Tent. It also examines the obscure period in the history of the avuncular clan under the reigns of Shizong 世宗 (r. 947-951) and Muzong 穆宗 (r. 951-969), both the political winter and the first period of embranchment of the Imperial Uncles into rival lineages (pp. 268-290, 297-301).

13The fifth chapter follows the fourth in its aim to rebuild the history of imperial marriages. It focuses on the “long 11^th century” that started in 970 and ended with the dynasty in 1125. The discussion revolves around how the marriages of the sons and daughters of the emperors were managed, and how they, in turn, established strong hierarchies between the imperial in-laws (for a study on the lineage of Xiao Xinning 蕭信寧, see pp. 306-332). It also attempts to explain why the marriage rules initially followed by the imperial family proved unsustainable, and how they ultimately evolved.

14The sixth chapter comes back to the problem of the Chinese surnames “Yelü” and “Xiao” and explains the rationale behind their attribution to diverse clans. It discusses how the Khitan aristocracy, centred on the Tents nobility and comprising a minority of Qai, Bohai and Han’er clans, acted as a dual society divided into a “Yelü” moiety and a “Xiao” moiety.

The marriages of the elite

15Khitan aristocratic families, like the Mongols, privileged durable alliances with a selected group of in-laws, mainly through marriage exchanges. This practice entailed reciprocal alliances where two families would each give daughters to each other. Cousin and cross-generational (often aunt-nephew or uncle-niece) marriages were common as long as spouses did not belong to the same patrilineal family – a tent (ger). Widows frequently, if not systematically, remarried into their original in-law family through levirate or sororate. However, differences between the Khitan and the Mongol traditions existed. Under the rule of the Yelü, a man’s first wife and her sons and daughters held a distinctly prominent status in the household. Concubines are barely mentioned in the texts, and their children rarely obtained the opportunities given to their half-brothers. Consequently, one man could only have one main in-law family and arrange the unions of his sons and daughters to both meet the requirements of exchange marriage and expand his network of alliances.

16Marriages in the imperial house, that is to say, of the closest kins of the emperor, although following the same pattern, serve obvious political purposes. The economy of marriages in the imperial house served to establish a clear hierarchy between the in-law clans. Sons and daughters of the empress were to marry into the main Xiao lineage only, and more precisely to the close kinsfolk of the empress. The several exceptions to this rule under the reign of Shengzong are explained in chapter 5 as being due to exceptional circumstances. As the ruling Empress Qitian failed to produce viable heirs, the emperor designated Yelü Zongzhen, a son born from the concubine of an Imperial Uncle, as the official heir in 1017. Consequently, his siblings, a brother and two sisters, came to be considered as children of empresses. Similarly, their mother Noujin was given the title of Yuanfei 元妃 and appears to have been treated as a quasi-empress. Another exception was that Shengzong’s kinsfolk far exceeded that of Qitian, making it impossible to marry all children of Jingzong to the same lineage. For this reason, it was decided that Shengzong’s two younger sisters were to marry a secondary lineage of the Imperial Uncles, that of Xiao Xinning. In doing so, the emperor confirmed the status of the Xinning lineage among other secondary in-laws, as they previously gave the third emperor his second empress. This privileged status given to former main lineages of the Xiao family is verified in chapter 3 through a study of all known Chancellors of the North (bei zaixiang 北宰相).

17The second and most important purpose of marriage in the imperial family  –  producing an heir for the empire  –  closely relates to the first. Particular attention to the question of succession under the Khitan has been given in this thesis. In chapter 1, a chronology of the transition from brother-to-brother succession to primogeniture has been explained. Contrary to what is commonly asserted, imperial younger brothers were always perceived as the most legitimate heirs. In chapter 5, for example, is explained that Shengzong chose Zongzhen as his heir the year his last true brother Yelü Longqing 耶律隆慶 died. Zongzhen, as Emperor Xingzong, deliberately chose to veer off from the tradition and designated his eldest son as the heir. This indirectly provoked the failed rebellion of his brother Yelü Zongyuan 耶律宗元 (Zhongyuan 重元 in the Liaoshi) in 1063. Zongzhen, who became Emperor Daozong 道宗 (r. 1055-1101), as the target of the rebellion, also chose to not give power to his brothers. This decision can be seen through the fact that Yelü Hongshi 耶律弘世 is the first younger brother of an emperor who did not marry a woman of the main Xiao lineage. As all brothers were potential heirs before this date, it was necessary for them to all marry into the same lineage to not endanger the privileged status of the main in-law lineage. These marriages, therefore, functioned in a close circle between the lineage of the emperors and the lineage of the empresses.

18The lineage of the empresses, which is the main “Xiao” lineage, changed several times throughout the dynasty. During the short 10^th century discussed in chapter 4, changes in the imperial lineages automatically shifted the identity of the main in-laws. The branch of Tuyu, Abaoji’s first son, ultimately imposed itself in 969 when Jingzong seized power following the assassination of the fourth emperor. Jingzong, still unmarried before the events, married the daughter of his close ally Xiao Siwen. Since Siwen belonged to another branch of the Imperial Uncles, the maternal kin of Jingzong revolted and assassinated the new imperial father-in-law in 970. In this case, political necessities presided over the tradition. This event, however, consecrated the Xiao Siwen descendants as the first stable main Xiao lineage of the dynasty. Their downfall came from a concurrence of events leading Shengzong (son of Jingzong) to designate a son born from a concubine as an heir. In doing so and arranging new marriages, he made the Xiao He 蕭和 family the second main Xiao lineage. This situation ended with the death of Shengzong in 1031 when the concubine Yuanfei and her supporters seized power and orchestrated the assassination of the childless empress and her close kins.

19Emperor Daozong put an end to this situation. Instead of favouring one of the other possible households of the Xiao He descendants, he chose a girl from another branch of the Imperial Uncles Tent to marry his grandson and heir. Thus, the imperial marriages transitioned from the domination of only one family to that of concurring lineages. Nonetheless, this change did not jeopardise the prestigious status of the Imperial Uncles Tent within the Liao aristocracy.

The moieties of the Khitan elite and the Chinese surnames “Yelü” and “Xiao”

20The surnames “Yelü” and “Xiao” are systematically attributed to the Horizontal Tent and Imperial Uncles Tent members in Chinese sources. Advances in studies on Khitan inscriptions have shown that Khitan never used these surnames, instead identifying themselves with a “little name” and a “second name”, neither as a patronym. The attribution of surnames, therefore, was only relevant in a Chinese-speaking (or writing) context. Chapter 3 explains that the “Yelü” surname appeared early, as the transliteration of the original Khitan fief of the Horizontal Tent. The “Xiao” surname, however, came from the name “Xiao Han” 蕭翰 that Li Song 李崧 (d. 948) created for one important member of the Imperial Uncles Tent in 947. Although the name started to be given to all men and women of the Tent in the 980s, epigraphic inscriptions indicate that it was already attributed to them in the previous decades.

21Studies on Khitan scripts, especially those by Aisin-Gioro Ulhicun, have also shown that the “Xiao” surname applied to multiple clans and not only to the Imperial Uncles Tent. Khitan and Qai families, including descendants of former Qai qaghans, also appeared as “Xiao” in Chinese texts, which made minimal to no effort to discern the different patrilineal groups. Parallel to this phenomenon was of the integration of other families into the Horizontal Tent and within the “Yelü” surname group. From what is known, this concerned the “nine tents” of the Yaonian 遙輦 (Kh. *Yulrin), which comprised the descendants of the nine Khitan qaghans that preceded Abaoji, and the Chinese Han 韓 family, more specifically, all descendants of Han Kuangsi 韓匡嗣 (918-983). In sum, “Yelü” and “Xiao” did not correspond to only two patrilineal clans. More importantly, the Khitan court only recognised these two surnames and applied them to a variety of different families. This connects with the observation that no marriage between two “Yelü” or two “Xiao” appears in any of the sources. This means that two families coming from very different clans cannot intermarry if they both belong to the “Xiao group”. Considering that the corresponding “Yelü group” comprised only the imperial clan and a few associated lineages, this meant that the majority of the Khitan and Qai elite became grouped into one exogamous group that only married the “Yelü”. The Khitan elite, therefore, functioned as a dual society including two moieties “Yelü” and “Xiao”. Furthermore, this dualistic organisation extended to clans that were not identified with the two surnames mentioned above. The descendants of Geng Chongmei 耿崇美 (893-948) who became one of the prominent Chinese families in the imperial court only married Yelü, and never Xiao. Likewise, members of the Han family frequently allied with the Xiao and the Geng 耿 but never intermarried with the Yelü, even before their integration into the Horizontal Tent. This means that the Han and the Geng respectively entered the “Yelü” and the “Xiao” moiety from the time they started intermarrying with Khitan nobles. The “Xiao group” also comprised members of the Da 大 royal family of Bohai, the Liu 劉, and the Liang 梁 families.

22The rationale behind the attribution of the “Yelü” and “Xiao” surnames has long been attributed to a conscious desire to “Sinicize” the Khitan elite, which adopted the Chinese concept of surnames and the attached rule of “those of the same family name shall not intermarry” (tongxing buhun 同姓不婚). In reality, the stricter rules presiding over Khitan marriages already fit with the Chinese custom. The dualistic structure of the ruling elite, however, might not have been understood by the Chinese, as it was not, from what we understand, clearly explicated in the official discourse we see in the inscriptions. Therefore, the official distribution of only two surnames to the Khitan houses might have been motivated to make the two exogamic groups evident and understood in the eyes of the Chinese tradition.