The Native Chinese State

In general, laws and regulations relating to marriage in the larger society also prevailed at the imperial level of the native Chinese state. Some were already in place in preimperial times (see Thatcher's chapter in this volume) and may be summarized as follows: (1) surname exogamy (TL 14:262-63; SHT 14:218-19; ML 6:16; Feng 1948:33-46); (2) serial monogamy with concubinage (TL 13:255; SHT 13:214; TML 13:275; Ch'ü 1972:46); (3) the wife as titular or legal mother to all offspring (Tjan 1949:260; Watsons chapter in this volume); (4) a ban on demoting wives to the rank of concubine and on elevating concubines to the status of wife (SHT 13:214-15; ML 6:9-10; TML 13:275-76; Ebrey 1986); and (5) a ban on cross-generational alliance (TL 14:263; SHT 14:218-19; TML 14:287; Ch'ü 1961:94-95). At the imperial level, the emphasis on monogamy meant that the title of empress went to the woman who had held the position of wife before her husband's accession to the throne. It also meant that there could be only one empress at any one time (Chung 1981:147) and that she was the titular mother of the next ruler whether or not she herself was his biological mother. The rule of surname exogamy meant that no matter how distant the relationship, members of the ruling family could not marry imperial clansmen but had to marry out and downward. Finally, although cousins of a different surname might marry (MBD/FZD), unions between heirs to the throne and their aunts (MZ/FZ) or nieces (FDD/MDD) were forbidden (see n. 3). These restrictions were but one of many conditions preventing individual families from gaining an exclusive permanent hold on marriage relations with the throne.^[1]

There were other, more general customs and practices that conditioned marriage both in the larger society and at the imperial level. For example, married women continued ties to their natal kin (Ebrey 1981; the chapters by Thatcher and Watson in this volume); in the event of minority rule, widows managed their husbands' estates on behalf of a successor (Ch'ü 1961:104; Shiga 1978:120; Holmgren 1985:7,16); and widows maintained rights in heir and spouse selection (Ch'ü 1961:30-31, 104; Waltner 1981:142; Ebrey 1984a:234-37). Widows also demanded total obedience from the father's sons (Ch'ü 1961:120-2l; Ch'ü 1972:53). In its broader form, filial piety—obedience to parents, with the wife subordinate to the husband (see Mann's chapter in this volume)—underpinned the entire moral-political order from the highest levels of society down to the village.

Although members of the ruling family had to marry down in the native Chinese state, care was taken that alliances were contracted with families from the highest stratum of the elite. The intention was not so much to conform to any ideal of class endogamy but rather to control powerful interest groups. The usual pattern was for early supporters of the founder, often

military leaders, to be given preference in marriage relations until such time as they had ceased to be actively involved in the critical functions of government. There was thus a strong pull toward repetitive intermarriage with the same set of families. But the marriage circle always remained open, tending to evolve according to shifts in the ruling elite.^[2]

Where members of the ruling family were concerned, senior officials of the outer court (chosen, in theory, according to merit) acted as guardians of the moral-political order (see Fang 1952:106-7; 1965:384; Fitzgerald 1968:27-28; Chung 1981:47, 62-63). This is not to say that strong-minded rulers and those who controlled the throne, including senior bureaucrats, did not try to manipulate custom and law when it suited their purpose.^[3] Indeed, as a group, senior officials often openly agreed to bend or break some rules in order to square conditions with other, more important considerations. They would not usually object, for example, to an emperor's wish to promote his favorite concubine to the rank of empress if that position were vacant and if the concubine were the mother of the eldest son (see Chung 1981:47, 49). As we shall see, the practical need to override the established rule that a concubine should not be elevated to the status of wife derived from the particular form of succession that operated at the imperial level.

Rather than partible inheritance among all sons regardless of the mothers' status—the normal and preferred option among the scholar-elite and peasantry (Schurmann 1955-56:511-12; Shiga 1978:135, 141-43)—only one person could succeed to the throne. That person was the eldest son born of the empress (Tao 1978:173-75; Chang 1966:5-6).^[4] If she produced no son, the eldest, regardless of his mother's position, was usually selected as heir, with the empress/wife acting as titular mother (also discussed by Thatcher in this volume). The heir (and his eldest son) resided in the palace beside his parents. All other sons were given fiefs at some distance from the locus of power. Although well provided for, they spent their lives in political obscurity. Direct access even to the wealth of the fief was denied, centralized bureaucratic control over its management being absolute. In sum, a distinction was drawn between male offspring on the basis of seniority and the status of the mother, with inheritance portions for younger sons being limited to a share of the tax revenue. Such was the case in the mature systems of all major Chinese regimes.^[5]

Because younger sons of the emperor were without political, military, or fiscal authority, the only kinship bonds that could be effectively used by the throne for political purposes were its ties to wife-givers and wife-takers. The marriages of women—both the bestowal of the ruler's aunts (FZ), sisters, and daughters on families outside the palace and the entry of women into the imperial harem—therefore took on unparalleled political importance, with the ruler's maternal relatives becoming at once his chief means of support

and a most dangerous threat to his independence. The section below outlines the main features of the relations between the emperor, his mother, and his wife.

Whenever the throne was vacant or the emperor too young, ill, or feeble-minded to govern in his own right, the wife in the oldest surviving generation (empress, empress dowager, or grand empress dowager—hereafter referred to as the "widow" or "senior widow") acted as regent for state affairs (Chao 1937:155-58; Yang 1968:50-51; Bernard 1972:65-75). When the main line failed, the widow selected an heir from among her husband's collateral kin (de Crespigny 1975:5). She also had powers of spouse selection and dethronement. Even when she was little more than a child, with de facto power in the hands of senior members of the bureaucracy who were her natal relatives (as happened in the case of the fifteen-year-old empress dowager, née Huo, of Former Han [Tao 1978:179]), the legal power of dethronement still lay with her (Wallacker 1987). In all this, the empire was treated as family property, the chief difference being that the widow worked in concert, not with her husband's collateral kin, but with senior members of the bureaucracy who were either unrelated to the throne or were its wifely or maternal kin.

Whereas the role of the widow accorded with basic Chinese principles of family management and the all-important code of filial piety and was considerably strengthened by the political restrictions imposed on male agnates, filial piety and the idea of a female regent were at odds with the concept of sovereign power of the emperor (see Ch'ü 1972:59). The compromise was an uneasy one, for it involved voluntary withdrawal by the widow, keeping her powers of heir selection and dethronement in reserve for times of emergency. In practice, the widow usually withdrew from court in the formal way demanded by the bureaucracy but continued to maintain her hand in government affairs through (1) her psychological advantage as the emperor's mother or grandmother; (2) controlling the marriage process; and (3) placing her male kin in key civilian and military posts. These strategies were first put into place in imperial times by the empress dowager, née Lü (d. 180 B.C .), of the Former Han (Ward-Czynska 1978:2-6) and from there became the standard means by which ambitious mothers, wives, and favorite concubines tried to perpetuate their influence. In these cases, MBD/FMBSD alliances for heirs to the throne (figure 2.1) aimed to avoid a clash with incoming women, while the placement of relatives in strategic administrative posts ensured control over information reaching the throne.

In the native Chinese state it was thus common 'to find relatives of the senior widow controlling communication between the emperor and the outer court. No consort family managed, however, to maintain its political dominance for more than two or, at the most, three generations of rule. One difficulty lay in the legal restrictions imposed on intermarriage with close kin, and another in the role played by senior members of the bureaucuracy as

Fig. 2.1.
Marriage with MBD/FMBSD in the Chinese state

guardians of the moral-political order, which led to their consequent interest and participation in the spouse-selection process (see Fitzgerald 1968:27-28; Chung 1981:62-63). Senior ministers jealously guarded their right to independent access to the throne and, through that, their right to join the imperial marriage circle. There were thus no customs or laws limiting the choice of wife to women from one particular family or group of families. This meant, however, that the system suffered from chronic tension between the throne, consort families, and the larger bureaucratic elite.

Another factor that prevented the development of a closed marriage circle lay in the custom whereby the wife could not control the marriage process until after the death of her husband.^[6] Consequently, she often controlled only the marriage of a grandson and was thus constantly at odds with her son's wife chosen either by her husband or by his mother or grandmother. In this way, conflict between successive generations of imperial wives became a recurrent theme in the history of the native state. In the outer court, the conflict was represented by alliances between relatives and supporters of the ruler's grandmother and wife against the family of his mother.^[7] From the emperor's point of view, his mother's family was useful in combatting either the combined influence of his grandmother and wife or that of the bureaucrat-

ic elite in general whenever it obstructed his purpose. Thus, when the emperor was in control, he often gave his mother's relatives the choice posts for high office.^[8] In this way, the top echelons of the bureaucracy were rarely free from domination by consort kin, yet the position of any one family was highly unstable, with the death of an influential widow often heralding a violent shake-up of the administration.

Having lived long enough to reach the status of senior widow able to control the marriage of her grandson, a woman found that any alliance with her family arranged for the heir was marred by the incoming spouse, who came with a set of ready-made female relatives (namely, a mother and grandmother) whose loyalties often lay with their natal kin. Thus, the senior widow always faced competition from relatives of incoming wives (see figure 2.1), and the throne might ally itself with any one of these families against that of the widow. But perhaps the most serious problem faced by imperial wives and their relatives lay in biology: the empress was often not the biological mother of the incoming ruler. Infertility and high infant mortality certainly were factors,^[9] but the emperor's deliberate sexual avoidance of the empress also played a role (see de Crespigny 1975:29). The dislike felt by many an emperor for his wife is easily explained: the woman was imposed on him when he was little more than a child. His favorite, then, was usually a concubine or palace attendant, and these women, rather than the wife, tended to conceive heirs to the throne.^[10] The emperor's dislike of his empress intensified when she was a close relative of an overly dominant senior widow. She was then seen to be an enemy or spy and indeed often acted as such (ibid., 8; Chung 1981:62, 75-76).

Such marriage dynamics explain many incidents related in the historical accounts of the Chinese court. They also explain the great variety of adoption and informal support strategies used by unrelated females in the harem and palace service sector—for example, the protection an empress or influential wet nurse gave to a young attendant, slave, or orphaned child. Here, the older woman hoped the young protegée might become the prince's favorite and keep rivals away from the throne. Such strategies were most effective when they cut across social barriers in the outer society and status hierarchies within the palace. Only in this way could the older, more senior woman be fairly sure of maintaining her authority over the younger woman in the harem (see PCS 9:128; Fitzgerald 1968:18-20; Chung 1981:60; MS 113: 3514).

Like wives in the larger society, an empress could be divorced for barrenness (see Tai 1978), and many a ruler tried to use this law to rid himself of his wife following the death of the senior widow (see Ward-Czynska 1978:14-17; Chung 1981:28, 47; MS 113:3513). Yet from the middle of the Former Han when wives came mostly from powerful lineages with secure footholds in the upper echelons of the administration (see Ward-Czynska 1978), members of

the bureaucracy became increasingly adamant that the empress should not be cast aside simply to make way for a current favorite. In fact, T'ang and Sung law expressly forbad divorce on the ground of barrenness (Tai 1978:86-87). Against this, the bureaucracy had to weigh the cost of infighting between the biological concubine-mother and the titular mother (and regent) when both women survived into the next generation. Then, a clash of wills was inevitable. The Chinese solution to this problem was to prevent wherever possible the separation of the wife's political functions from her biological one of producing an heir to the throne. To this end, officials invariably agreed to elevate a favorite concubine who had produced the eldest son to the rank of empress if and when the current empress died or was divorced. The possibility of such promotions, however, encouraged rulers and favorite concubines, and sometimes mothers and maternal relatives, to engineer the death or disgrace of the empress, a problem also recorded for preimperial times (see Thatcher's chapter in this volume).

Given the frequency with which concubines' sons came to the throne (see n. 10), a ruler might well be surrounded by, not two, but up to four distinct sets of competing relatives centered around his biological and titular mothers (dowager empresses) and his biological and titular grandmothers (grand dowager empresses), some or all of which might be competing with the family of his wife (the empress). This competition, it should be noted, was important in preventing usurpation of the throne by consort families. Wang Mang (r. A.D . 9-23), whose case provides one of the few examples of successful usurpation, had to remove three other consort families from power before he could properly begin to engineer his coup d'état (Dubs 1955:44-49; Loewe 1974:286-306; Ward-Czynska 1978:41-47, 56-61). In contrast to Wang Mang, most imperial relatives confined themselves to the more realistic goal of domination. Even so, the lack of support from male agnates and the permanent tension among consort families, the throne, and members of the bureaucratic elite meant that fear of maternal kin was a major destabilizing factor in the history of the native state.

We have seen that an imperial wife held an unstable position in the early years of her career, when she relied in large part on the senior widow and members of the bureaucratic elite (including her own kin) for protection. With her husband's death, however, she might become head of the ruling house and on many occasions de facto ruler of the empire. Here, the widow not only assumed leadership of the ruling house but also (and quite logically) the de facto headship of her natal lineage (see Ch'ü 1972:58-59). Her influence was most widely felt when the emperor was her biological son or grandson and when he was young, weak-willed, or uninterested in government affairs. Thus, she invariably chose an infant or child rather than an adult when in a position to exercise her authority in choosing an heir (see HHS 10A:423). But the accession of a series of young and ineffectual rulers,

as occurred during the Later Hah (see Ch'ü 1972:215-16), did little for the morale of the administration because direct access to the throne was effectively blocked for all but the woman's family and supporters. For this reason, the death of an influential senior widow often heralded a violent fall in status for her family, sometimes bringing in its wake the death or disgrace of the empress and/or the wife of the heir-apparent.

In contrast to relatives of the emperor's mother and wife (wife-givers), the husbands and offspring of his sisters and daughters (sororal, or wife-taking, kin) were in a weak position when it came to succession politics. Sisters and sisters' sons were ineligible for the throne, and sororal kin were excluded from regency powers both by custom and by jealous wifely/maternal kin. In this way, however, the marriages of imperial sisters and daughters provided the throne with a way of establishing nonthreatening relations with influential members of the bureaucratic elite.^[11]

As with incoming women, sororal ties were established by emphasizing customs and laws found in the wider society—property shares for daughters in the form of dowry (see Ch'ü 1972:273 n. 107, 283; Tai 1978:105-6; Ebrey 1981; Holmgren 1985) and educational and economic strategies centered on the mother (see Waltner 1981:144-52; Walton 1984:44-49). In addition, sororal ties benefited from the law of inheritance, that is, ranks and titles conferred on a woman without reference to her husband or son were to be treated as if granted to a man (TL 2:38; Johnson 1979:100-101). In the case of the princess, this condition was achieved by dispensing with the general law that made married women liable for punishments meted out to members of the husband's lineage (TCTC 76:2425). The princess's exemption from this law (see Ch'ü 1972:57; HTS 83:3647, 3650, 3653) meant that her status in no way depended on the position of her husband and his family. Rather, she retained her membership of the ruling line and was subject only to the throne (see Tang 1975:42; Wong 1979:136-37; DMB 1976.1:211). Because her status was conferred without reference to her husband, it could pass to her children. Thus, so long as they did not become wittingly involved in plots against the emperor, sororal cousins, nephews, and nieces were, like their mothers, exempt from severe punishment (see SKC 50:1198; 57:1340).

The material wealth a princess brought into marriage symbolized her condition. The lavish wedding gifts provided by the throne (see Chaffee's chapter in this volume) indicated the social and political superiority of her natal lineage over the husband and his family; the fief and its accompanying stipend (see Bielenstein 1967:21-22) symbolized that marriage had not altered her status. If the marriage lasted, most of the property eventually passed to the husband's family and out into the wider society through the woman's Offspring, the fief title going to her eldest son in accordance with the general law (see Dull 1978: 63 n. 188; Tang 1975:100; Shiga 1978:118). As befitting their elevated status as honorary members of the imperial line, the

woman's sons received imperial patronage in selection for high office. In this way, the sororal bond was transformed into an arm of the ruling line, reaching out into the wider community and establishing pockets of loyalty within other, potentially dangerous, lineages without the threat of domination or usurpation.^[12]

Although the recipient lineage derived some comfort from the knowledge that at least one of its branches was insulated from political disaster, the throne benefited more from the arrangement because the princess could be counted on to put the imperial interest above that of her husband's family (see n. 12). For the woman's part, her exemption from severe punishment gave her a personal freedom denied to other members of the society, including her brothers. Unlike male agnates, who were perceived as a threat to the throne, the princess could remain in the capital at the center of power. Moreover, being female, she was not barred from the inner recesses of the palace as were the male officials of the outer court. Nor was she confined to the palace like an imperial mother or wife. In every respect, then, she was ideally placed to act as a power broker between the throne and families of the wider elite.

As a permanent member of the ruling line, the social status of the princess was higher than that of the imperial mother or wife. Indeed, her social position closely approximated that of the emperor himself. During the Southern Dynasties (A.D . 317-589), one woman used her exalted position to argue that, like her brother, the emperor, she too should have a harem. She was given thirty male "concubines" (Wong 1979:10). In short, although the princess never reached the political heights of an imperial mother or wife who acted as regent and de facto ruler of the state, and although her influence had no strong legal basis, her freedom of movement was considerable and her position solid. Because her status was not conditioned by ties beyond the throne, her influence depended on the accession of a strong ruler able to control his maternal relatives and to override any objections from the bureaucracy about her behavior. The tension between sisters and wives (princesses and empresses) at the imperial level was thus acute. Moreover, because all attention was directed toward a single male, competition among the princesses themselves was also intense—both among sisters and among different generations of female offspring (FFZ, FZ, and Z; see n. 15).

The princess's considerable status, which she passed to her children, meant that those children were eminently suitable as spouses for the next generation of imperial offspring. In fact, rulers often gave their children in marriage to sororal nephews and nieces. Although it was relatively easy, however, for a favorite sister to persuade her brother to give one of his many daughters in marriage to her son, her ultimate aim was to marry one of her daughters to the heir-apparent (FZD). Such a union would enable her to become de facto ruler of the state through her daughter, who would be in line

Fig. 2.2.
Marriage with FZD/FMDD in the Chinese state

for the position of empress. Here, however, the princess faced fierce opposition from the ruler's grandmother, mother, and wife, as well as from other princesses with the same ambition. The senior widow, in conjunction with members of the outer court, controlled the marriage of the heir. So in practice a princess might achieve her ambition only (1) if she were the favorite sister of a strong-minded ruler whose mother and grandmother had passed away; (2) if she were the biological daughter of the senior widow; or (3) if an accident brought to the throne a younger half brother of the appointed heir—one who had previously married into her family (figure 2.2). Some women attempted to induce the latter condition by forming an alliance with an imperial concubine who had a son in order to bring about the disgrace and demotion of the appointed heir and his mother. But few women were able to bring such a complex strategy to fruition.^[13] In practice, then, the empress in the native Chinese state was often a maternal cousin of the emperor (MBD or MBD/FMBSD), but she was rarely his paternal cousin (FZD).^[14] In this way, sororal relatives (or wife-takers) tended to remain somewhat separate from wifely/maternal kin (or wife-givers), again helping to prevent the emergence of a truly aristocratic closed marriage circle. In short, the open-ended marriage system described in the chapter by Thatcher for the Spring and Autumn period continued on into late imperial times.

The political importance of the sororal bond for the throne meant that unmarried aunts (FZ), sisters, and daughters were always in short supply. Thus, those who were widowed (with or without children) usually remarried. If a recipient family fell from grace, the woman might well be summarily

divorced and given to another lineage, young sons following her into the new alliance. Because a favorite sister was also well placed to obtain a divorce on grounds of incompatibility and because marital discord might itself bring political catastrophe to the lineage, it was not unusual for an imperial princess to become the de facto head of the husband's family, controlling its finances, organizing its marriages, and determining its political strategy. For this reason, some families tried to avoid a sororal relationship with the throne (see Ch'ü 1972:295-96; Wong 1979:206 n. 66). The problem was not confined, however, to imperial relationships: any family interested in furthering its economic, political, or social condition through the upward marriage of a son chanced subordination to the wife and her kin (on this point see Chaffee's chapter).

Although this inversion of the normal husband-wife relationship brought complaints (see HHS 62:2052-53; Wong 1979:95-96, 99; and below), and although praise might be heaped on the princess who refused to remarry after widowhood or forced divorce (see Sui shu 80:1798), greater exception was taken to women who flaunted their superior status before the husband's parents, thus violating the code of filial piety. In such cases, senior officials pressured the throne into trying to get the princess to moderate her behavior. As we have seen, however, the issue of filial piety posed an even greater problem for the emperor's personal relationships. Divorce was kept within reasonable bounds by the knowledge that a crucial political alliance was at stake and by the idea that the severing of an established bond between a man and his wife for political gain was objectionable to senior members of the bureaucracy because it undercut family harmony, the foundation stone of the moral-political order, and because they themselves were the potential victims of policies embracing arbitrary divorce (see Chaffee's chapter in this volume). In the main, then, an imperial princess was not summarily divorced unless her husband suffered a drastic punishment like exile or execution. At the same time, princesses, unlike other women, were not expected to undergo any hardship or punishment that might accrue to the husband's lineage, and a man who abused his princess-wife (whatever the provocation) could be sentenced to death (see Ch'ü 1972:58).

In summary, we can say that the political impotence of the emperor's male agnates, who were greatly distrusted by both the throne and the bureaucracy, heightened the political significance of the marriages of women and the position of the emperor's mother, wife, aunt (FZ), sister, and daughter. The mother, or senior widow, and her relatives provided support both against the ambitions of male collateral kin and against the larger bureaucratic elite when either attempted to obstruct the will of the throne. For this reason (and because the power of maternal kin was balanced only by sororal bonds established through the marriages of female offspring), the political authority of the imperial wife was greatly feared. Because attention was

focused on a single male (the emperor or his heir), however, competition · between different generations of imperial wives and between them and the ruler's sisters and daughters, as well as among sisters themselves, was fierce. Such rivalries helped to prevent the development of a monopoly by any one family, greatly reducing the threat of usurpation by maternal kin. Yet the marriage circle tended to follow shifts within the political elite, so that competition also destabilized government. The following discussion focuses on the development of this system and how the founders of some regimes addressed the problem posed by the power of maternal kin.

Historical Development and Evolution

Perhaps the nearest a native Chinese regime came to instituting an exclusive system of marriage exchange with one particular family was at the beginning of the Former Han, when the empress dowager, née Lü (d. 180 B.C .) arranged a series of strategic marriages in order to perpetuate control of the government by members of her family (Ward-Czynska 1978:2-5). With the struggle to oust the Lüs after her death, however, potential heirs to the throne who were connected with her family were killed. Moreover, the next emperor was chosen for the lowly status and meager numbers of his maternal and wifely kin (Ward-Czynska 1978:1-11; Tao 1978:178; Kamada 1962:80). The next few reigns saw a relatively free system of marriage that regarded the selection of a spouse as a purely private or family matter. No definitive marriage group emerged because little formal connection existed between consort families and the established elite. In this period, then, wifely and maternal kin were even more vulnerable to changes in fortune than in later times, when they had an independent foothold in the bureaucracy.

Emperor Wu (r. 140-87 B.C .) changed this marriage pattern. Coming to power through a palace intrigue manipulated by his paternal aunt (FZ), Emperor Wu began his reign under two warring consort factions (Ch'ü 1972:169). His first wife was his father's sister's daughter, chosen for him by his grandmother, the grand empress dowager, née Tou (FZD/FMDD; figure 2.3). He degraded his wife after the death of his grandmother, and replaced her with a singing girl introduced by one of his sisters (Ward-Czynska 1978:12-17; Loewe 1974:51). Then, after having lost his first heir through political intrigue, and with a very young son to succeed him, he tried to put an end to consort power by forcing the mother of that son to commit suicide. He chose three ministers of the outer court to act as regents after his death (Ward-Czynska 1978:20-25; Loewe 1974:35-70). But this strategy failed because the appointed regents, all of whom were connected by marriage, simply selected a relative (a child of six) to be wife and empress for the young ruler. This arrangement initiated a twenty-year period (87-66 B.C .) of complete subordination for both the bureaucracy and the throne to consort families (see Loewe 1974:73-81; Wallacker 1987:58-59). In the short term, then, Emperor Wu's strategy produced a return to conditions seen at the

Fig. 2.3.
Family ties of Emperor Wu of the Former Han
* Empress dowager Po and Empress Po were kinswomen.

beginning of the dynasty; that is, to domination of the government by a particular consort group. In the long term, the strategy instituted a system that drew wives from families of the political elite, with members of that elite deciding marriages in consultation with the senior widow (see Ch'ü 1972: 173-74, 210-29).

In Emperor Wu's time, the last vestiges of regional authority given to male paternal kin were stripped away (see Kamada 1962; Tao 1974). Some three centuries later, however, under the state of Wu (A.D. 222-80) in southeastern China, princes of the house were once again given regional military powers and made coregents and advisers to the throne (see Fang 1952; 1965). This condition continued on through Western Chin (A.D. 265-316) and the South-

ern dynasties (A.D. 317-589) into the early T'ang period (see Grafflin 1980:183-93; Tang 1975:50-54). For later Chinese the dramatic events of Western Chin (see Straughair 1973:2-5; Grafflin 1980:26-29, 94-95) reinforced two separate messages: first, that authority allotted to the ruler's male agnates would probably result in succession disputes that might well bring down the regime, and, second, that constant vigilance was needed against the ambitions of imperial wives, who—like the empress, née Chia, of Western Chin—might be tempted to depose the legitimate heir in favor of one of their own relatives (CS 31:965; CS 53:1459-62).

By comparing the history of Western Chin with that of other eras that gave male agnates a share of political power (see n. 5), we can see that influential male kin complicated the alliance patterns described above. An imperial wife could now appeal not only to her natal kin but also to individual princes of the house for support against other women or against the throne itself. Conversely, a princess was also able to support, or gain the support of, uncles and brothers whenever the throne appeared to be overly dominated by imperial wives. At such times, the political activities of female offspring were no longer masked by the machinations of the throne or maternal kin. As a consequence, the normal patterns of conflict between women not only spilled over into the outer court, disrupting the functions of government, but also set one prince against another, creating instability in the succession. Such disruptions were common in early and middle Former Han, Wu, and early T'ang and during the Western Chin when civil war erupted.

In the state of Wu, the roles played by princesses of the ruling house in succession problems (see Fang 1965) prefigure those of the early T'ang era (see Tang 1975). One can discern these same outlines in the history of early and middle Former Han (see Ward-Czynska 1978). In these periods, imperial princesses were not only divided among themselves and set in opposition to particular wives and heirs to the throne but also on occasion found themselves opposed to their own husbands. As members of the bureaucratic elite, the husbands of princesses sometimes sided with the heir or with another prince of the house to oppose female influence in general (that of mothers, wives, and sisters). Such alliances account in part for the high rate of divorce and widow-remarriage among imperial princesses in these periods. In early T'ang the underlying gender conflict between the all-male bureaucracy and females who provided the throne with its personal base of support was intensified by the presence of Empress Wu acting as de jure ruler of the state (A.D. 684—707).^[15]

In sum, conditions in early T'ang represent one of several major realignments of the elite marriage system. The first of these shifts occurred during the Later Han when the throne, the bureaucracy, and even individual mothers and wives became subject to males of the consort family (see Young 1986 for details); the last was to occur during the Ming dynasty when the

throne, mothers and wives, and their male relatives became subject to the bureaucracy, a development we shall discuss below.

It is commonly thought that with the establishment of the examination system in the T'ang-Sung era, late imperial China experienced a social and political mobility that put an end to the early aristocratic system of government and produced a concomitant shift away from class endogamy toward a marriage system in which rank could be exchanged for material wealth, scholastic repute, and so forth (see Ebrey's chapter, this volume). The status of emperor is said to have changed from that of "first among equals" (many of whom provided spouses for the imperial house) to autocrat (Sung) and, finally, to tyrant (Ming). The despotic tradition of the Mongol-Yuan era (c. A.D. 1200-1360) is believed to have influenced the development of tyranny during the Ming dynasty. In a recent study of imperial marriage strategy in the mid T'ang, Chang Pang-wei (1986) linked the idea of a shift away from class endogamy to a demonstrated decrease in the proportion of imperial wives and concubines coming from the upper ranks of the elite. But other scholars have begun to question the accuracy of the underlying hypothesis: they suggest that depictions of pre-T'ang society as particularly static and of Sung society as highly mobile may be exaggerated. They also challenge the idea of a Mongol tradition of despotism arid the view of the late Ming ruler as tyrant.^[16] The discussion below investigates some of these issues as they relate to imperial marriage under the Sung and Ming dynasties.

During Sung, certain founding families of the dynasty continued to provide spouses for, and to receive women from, the royal house long after their political fortunes had begun to wane.^[17] This is one immediate explanation of Chang Pang-wei's data. By late Northern Sung these and other sororal kin were, furthermore, being specifically denied top-level posts in the central administration (SS 223:13579; SS 248:8777; HTCTC 153:4094; and Chaffee's chapter in this volume). In other words, royal spouses were no longer automatically chosen from the bureaucratic elite, and members of consort families were no longer given key administrative positions to the extent that they were able to enter and dominate the top ranks of the bureaucratic elite. The change had little to do with shifting attitudes toward marriage in the wider society. Rather, the initial impetus came from the throne and was an unintended side-effect of efforts to limit the power of the military. To this end, the first ruler of Sung came to an agreement with the founding generals of his regime that the latter would forgo real power in return for guaranteed social status through marriage ties with the royal house (Chung 1981:25). In fact, successful demilitarization of the political structure was a difficult and complex matter achieved through strategies that had nothing to do with marriage relations. Once demilitarization was under way, however, newly emerging bureaucratic groups saw the advantages of a policy that paired members of the ruling line with spouses from elite families on the decline.

The weakened relationship between political power at the top of the bureaucratic hierarchy and marriage relations with the throne also suited imperial wives. They could now dominate political affairs as dowagers and regents without hindrance from their male relatives or the emperor's sororal kin. For the consort families themselves, elevated social status coupled with political advantage at the lower levels of the hierarchy helped to sustain respectability at the local level (see Chung 1981:64-68, 79; Chaffee's chapter in this volume).

The growing disparity between the social and educational attributes of Sung, princesses and their spouses heightened the former's haughtiness toward their husbands and in-laws—a phenomenon that hardened attitudes about the husband's superiority over his wife and the wife's subordination to her husband's parents. This development was independent but supportive of trends in the larger society. The declining status of imperial marriage partners in general led to discussion of the qualities required of an empress: here "pedigree," meaning membership in the founding elite and/or respectable middle-ranking military background, as well as individual social skills—education, intelligence, and reputation for ethical behavior (Chung 1981:24-25)—became the chief selection criteria.

If action against the power of consort families was particularly gentle in the Sung, this was not so in the Ming, the first native regime to maintain a deliberate policy of choosing imperial spouses from families with no bureaucratic connections (Soullière 1988). The strategy was not entirely new, however. It had been tried (with little success) as early as the first century B.C. by Emperor Hsuan (d. 49 B.C. ) of the Former Han (see Ward-Czynska 1978:32-34, 40-41). But then the strategy had ignored the fundamental issue—the widow's role as final arbiter of the political process. The Ming addressed this issue by placing an additional ban on the delegation of imperial authority to harem women (Greiner 1979/80:6-7, 47; Soullière 1984:138; de Heer 1986:10). This too had been tried in earlier times—by the founder of Wei (A.D. 220-65), who had tried to dispense simultaneously with both the authority of the widow and support from close male agnates (TCTC 69:2206). Relying largely on sororal kin and a few select fraternal bonds to establish ties with the political elite of the former regime, the Wei throne had quickly fallen prey to usurpation by outsiders (see nn. 5, 8, and 12). The Ming dynasty did not suffer this fate, but a parallel nevertheless exists in the weakening position of the Ming throne vis-à-vis the bureaucracy in the absence of key matrilateral support.

Like the state of Wei, Ming governments found they still faced the problem of delegating authority during minority rule. Ironically, although no empress ever became regent (Soullière 1984:138), senior widows retained the authority to appoint a regent and to choose heirs and spouses just as they had in third-century Wei (see Fang 1965:165-66, 337, 352; de Heer 1986:25-27).

In practice, an imperial wife during the Ming dynasty played much the same role, albeit less prominently, as she had in the past.

Ming marriage strategy originated with the founder of the regime and reached maturity over the course of several reigns. Spouses were selected from low-ranking military families in the capital area on the basis of temperament and physical appearance (Soullière 1988:1, 20, 23-24, 30). Without a solid political base, families of the bureaucratic elite that supplied women for the imperial house could not supply more than one empress (ibid., 39). Consequently, this period witnessed a degree of upward and downward social mobility among imperial in-laws not seen in the native state since early Former Han. It would seem, then, that the phenomenon observed by Chang (1986) wherein imperial wives were increasingly drawn from the lower ranks of the elite and even from commoner families is best explained, not by changing attitudes in the wider society, but by the increasingly deliberate denial of power to consort families. The rationale and strategies underpinning this development during the Ming dynasty were firmly rooted in the Chinese tradition of pre-T'ang times.

Studies of Ming imperial marriage relations by Soullière (1984, 1987, 1988) support the current view that, far from being a despot, the Ming ruler had gradually been reduced to titular ruler and was made a virtual prisoner in his own palace. Without a powerful consort family as his ally, he was unable to assert his independence against the bureaucracy and sometimes sought to withdraw from the system entirely, forcing senior bureaucrats to turn to eunuchs and harem women for help in maintaining the link with the palace (Soullière 1984:132-34). In this period there was a swing away from educating imperial princesses in the virtues of wifely submission (as in the Sung) toward that of teaching their husbands and in-laws the art of being subordinate (see Soullière 1988:21). The development accorded with the new relationship between the bureaucratic elite and eunuch factions in the palace in cutting the emperor off from his traditional base of support. In short, the policy developed by the founder of the Ming to ensure the integrity of the throne against imperial wives and consort families was gradually appropriated by the bureaucratic elite so that all imperial women—mothers, sisters, and wives—now owed obedience, not to the husband or to male natal kin as in the past, but to the outer court.

In summary, power at the Chinese court oscillated between three officially sanctioned forces: the emperor, the senior widow, and members of the bureaucracy. Four other forces lay on the periphery (eunuchs, not dealt with here, and the emperor's paternal, sororal, and wifely/maternal kin). Any of these could be called into action by one of the major parties. Whether in Han times or Ming, however, the key to major shifts in the balance of power lay with close male relatives of the widow (wifely/maternal kin). When they were in a position to bridge the gap between the harem and the outer court by

occupying senior positions in the bureaucracy, power tended to swing away from the throne and unrelated members of the bureaucratic elite toward the widow and her family. Conversely, the widow was at her weakest whenever her natal kin had no established foothold within the bureaucratic elite. When this happened—as in the Wei (Three Kingdoms) and the Ming—power initially resided with the throne but tended to gravitate quickly toward the bureaucratic elite. Caught between the other two forces, the throne attempted to play off the widow's party against the bureaucratic elite, or to invoke the aid of one of the lesser peripheral bodies named above.

Accounts of the relationship between the emperor and these other parties were mostly composed, or at least approved, by men whose families had well-established relationships with sectors of the bureaucracy mainly outside the imperial marriage system. Thus, any choice of words in these accounts suggesting despotism on the part of the emperor or usurpation of power by imperial wives, relatives, or eunuchs should be seen as a signal of a significant shift in power away from the bureaucratic elite. Conversely, literature on the Sung indicates the stable and equitable balance of power between the major parties. A similar absence for mid- to late Ming, however, would reflect the complacency of a dominant bureaucracy. In other words, the mature Sung and Ming dynasties were seen to have largely avoided not only the "faults" of the Later Han, when the bureaucracy (as well as the throne and individual wives) had been subject to the consort clan, but also the "problems" of Wu, Western Chin, and early T'ang in dealing with the emperor's paternal kin as well as the various evils of late T'ang, when eunuchs had been the dominant policymakers.

We have seen how shifts in the balance of power during the Sung and Ming dynasties were in fact brought on by one of the constants of early Chinese history—namely, the acknowledgment by all parties that the position of the senior widow was such that the emperor and the bureaucratic elite could best (or perhaps only) assert their power by breaking the bond between political service and intermarriage with the throne. In this respect, the institution and the development of a public examination system (T'ang-Sung) possibly helped to weaken that bond. Yet the examination system was only a coincidental factor, for, as we have seen, conditions under the Ming were quite unlike, and in no way contingent upon, developments in either the T'ang or Sung dynasties. Moreover, and most important, each of those developments, as well as those that took place during the Ming, fell within parameters of change (both structural and ideological) already set in place by the end of the third century A.D. In short, despite major social changes in the wider society (see Ebrey and others in this volume), structure and strategy at court remained much the same in the native Chinese state from Han through to the end of Ming.