Final exam

Rethinking the imperial harem 177

8 Rethinking the imperial harem

Why were there so many palace women?*

Chinese emperors, from the First Emperor of Qin (r. 221–206 ) on, sur- rounded themselves with attractive young women. Many let the number of palace ladies increase unchecked until it reached several thousand. These women (in Chinese called shinü, yushi, gongren, gongnü, or the like) were potential mates, young and unmarried, kept on only as long as they were considered sexually desirable. They formed a huge pool of women available to the emperor, and any he was attracted to could be promoted into the lower ranks of consort if the emperor chose to favor them. Although these women served as palace attendants or servants, they were not recruited, promoted, or retained on the basis of their skill as seamstresses or cooks, but rather for their sexual allure. Despite Confucian criticisms, dating back to the Zhou period, of rulers who let themselves become befuddled by women, no dynastic founder ever proposed refraining from setting up a harem or keep- ing only two or three women. Even Hong Xiuquan, the nineteenth-century rebel Taiping emperor, who insisted that his followers practice monogamy, ended up with what looked very much like a harem.1

Why would an emperor want or need thousands of potential mates? A harem numbering in the thousands incurs all sorts of costs – financial, emotional, managerial, and political – that do not seem obviously balanced by greater benefits. To assure an ample supply of heirs, surely dozens of women would have been sufficient. To demonstrate that the emperor was the most powerful man in the country, a couple of hundred would surely have been plenty, since even the richest subjects did not keep more than a couple of dozen concubines. Another strategy historians often use in looking for explanations – the search for origins – is no more enlighten- ing. The first emperor to keep a really large harem, the First Emperor of Qin, was never held up as a paragon to be emulated by all later emperors who wanted the respect of posterity. To the contrary, those who allowed their harem to grow large in the centuries after Qin Shihuangdi did so

* Earlier versions of this essay were presented at Beijing University, Harvard University, Seattle University, and the University of British Columbia. Its present form has benefited from the questions and comments received on those occasions.

Women and the Family in Chinese History Patricia Buckley Ebrey

178 Women and the family in Chinese history

despite the general condemnation of his style of governing. To me this is evidence that there were cultural logics, just as “Chinese” as the Confucian protests, that led rulers to let their harems grow to considerable size. In this essay I pursue these logics, looking most closely at the case of the Northern Song (960–1127).

References to the numbers of palace women are not as precise or reliable as one would like. Most of what we have are round numbers used by those criticizing an emperor for extravagance. The First Emperor of Qin, we are told in an early commentary to the Shiji, not only built a huge palace complex, but also filled it with women from the harems of the rulers he had conquered, so that his “rear palace” had over 10,000 women.2 Although the first three Han emperors did not have over 1,000 palace ladies, by about 100 , under Wudi, their numbers reached several thousand, and the figure “several thousand” recurs in the Later Han as well.3 The number “10,000” recurs in the third century  with the last ruler of Wu in the Three Kingdoms and in the fourth century with Wudi of Jin.4 This record was surpassed by the early seventh-century Sui emperor, Yangdi, who was said to have had the truly implausible number of 100,000 palace ladies in the secondary palace in Yangzhou.5 For the Tang emperor best known as a ladies’ man, Xuanzong in the mid-eighth century, poets guessed 3,000 or 8,000 palace women but the official history stated 40,000.6

Traditionally, there were three main ways of looking on the size of the imperial harem: the voyeuristic, the romantic, and the moralistic. Voyeuristic literature revelled in evocations of the sensual indulgence of the palace. A good example here would be the novels about the second emperor of the Sui, Sui Yangdi, who, as mentioned above, was said to have had 100,000 palace ladies. Westerners have also been fascinated with the erotic possib- ilities of the imperial harem and written about it in voyeuristic ways. The Book of Marco Polo contains a description of the selection process used for Khubilai’s harem. Each year the khan had messengers choose 400 or 500 of the most beautiful of the “very handsome and fair-skinned” Ungrat tribe, giving each girl numerical ratings for “the hair, the face, and the eyebrows, the mouth, the lips, and the other limbs.” Those rated highest would then be brought before Khubilai. “When they are come to his presence he has them valued again by other judges, and of them all he has thirty or forty who are valued at most carats chosen for his own room.” Not only were great efforts invested in selecting the women most likely to please the khan, but efforts were taken to make sure he never grew tired of any of them:

It is true that every three days and three nights six of these girls are sent to wait on the lord when he goes to rest and when he gets up, both in the room and in the bed and for all that he needs; and the great Khan does with them what he pleases. And at the end of these three days and of three nights come the second six girls in exchange for these, and those depart. And so it goes all the year that every three days and three

Rethinking the imperial harem 179

Figure 8.1 Illustration from the seventeenth-century novel about Sui Yangdi (Sui Yangdi yanshi).

Source: After Hegel 1981: 66.

180 Women and the family in Chinese history

nights they are changed from six to six girls until the number of those hundred is completed, and then they begin again another turn.7

This sort of description can be classed as voyeuristic because it offers male readers the pleasure of observing from afar the sexual lives of the emperors who could have all the women they could possibly want.

The romantic approach to the thousands of palace ladies looks with sym- pathy at the plight of the women who have to compete with so many other women to win the affections of a single man. The lonely, neglected palace lady became a romantically appealing image. This is seen in the large genre of “palace style” poetry. Poets described anonymous palace ladies as pas- sionate but passive women, women with strong feelings they cannot act on because the object of their passions is a man unavailable to them. Confined spatially in elegant quarters, they spend their time at home with their feel- ings, often unable to sleep, or silently weeping.8 There are also stories and plays about particular palace women, such as the famous Wang Zhaojun, the palace lady in the Han who ended up sent to marry a Xiongnu chief because she did not bribe the artist to make her look beautiful so the emperor would not be willing to part with her.9

The moralistic approach is found throughout the histories. Confucian moral critics regularly pointed a finger of warning at the examples from history of rulers who came to bad ends because they were too given to indulging their passions for women.10 Confucian scholars did not worry solely about dissipation, about rulers wasting their time and energy on excessive numbers of women. In their view, even emperors who confined their atten- tions to a few favorites had to be wary, because the imperial harem was a place that had a pernicious effect on women. Given the level of competition in the harem, and the disparity between those favored and those not, women there were seen as particularly likely to become manipulative and scheming, and emperors therefore had to be on their guard, lest they fall under their influence.

The main reason to keep in mind these traditional ways of viewing palace women is that they create biases in the sources. All of them tend to exagger- ate the numbers of women and frame the institution in terms of a sexual relationship between one man and many women, in the process obscuring other aspects of what was going on.

Modern scholarly studies of the “rear palace” have usually concentrated on aspects related to imperial succession and court politics, such as the role of empresses in deciding succession and ruling when the heir was a child, or the selection and promotion of empresses, consorts, and female officials.11

Many of these studies reflect an early-stage feminist impulse to give women their due as power holders, to show that women, like men, could hold polit- ical power and occupy places in complex bureaucratic organizations. Generally they adopt an implicit functionalist framework, which explains the organization of the harem in terms of the importance to the imperial

Rethinking the imperial harem 181

Table 8.1 Northern Song emperors, their consorts and childrena

Emperor

Taizu 960 –975 34 16 2 10 .63 40 Taizong 976–997 38 22 11 3 13 .59 38 Zhenzong 998–1022 30 25 11 3 7 .28 57 Renzong 1023–1063 13 41 16 6 16 .44 25 Yingzong 1064 –1067 32 4 4 1 0 0 0 Shenzong 1068–1085 20 18 14 9 26 1.44 8 Zhezong 1086 –1100 10 15 9 1 5 .71 80 Huizong 1101–1126 19 25 19 13 65 2.6 17

a There are several discrepancies in the number of sons listed for each emperor in the SS and the HSSCGY. Here I have taken the larger number. Also, in the case of the first two emperors, I had to estimate how many of their children were born after taking the throne, and tried to err on the side of counting as many as possible as born after accession.

b This is the cumulative number of empresses and consorts who held high consort titles, as listed in HSSCGY.

A ge

i n

su i

on a

cc es

si on

Y ea

rs o

n th

ro ne

H ig

h- ra

nk in

g co

ns or

ts w

hi le

o n

th ro

ne b

H ig

h- ra

nk in

g co

ns or

ts w

ho h

ad c

hi ld

re n

C hi

ld re

n bo

rn w

hi le

em pe

ro r

C hi

ld re

n /y

ea r

on th

ro ne

a ft

er a

ge 1

8

% o

f ch

il dr

en b

y lo

w er

-r an

k w

om en

institution of an ample supply of heirs. However, from their work it is evident that the need to staff a female, inner court bureaucracy cannot account for a harem in the thousands. As Priscilla Ching Chung showed, in the Northern Song there were six bureaus in the palace staffed by 279 female officials, including everyone down to the level of clerk.12 It was not to keep these offices staffed that the numbers of palace women reaches well into the thousands.

Nor can the rate of increase of palace women be explained by linking it to the amorous interests of individual emperors. Northern Song evidence shows that the numbers could increase even when an emperor was a child and an empress dowager ran the palace, or when it is evident that the emperor limited his attentions to no more than a dozen of the women in the palace.

Table 8.1 provides a basic overview of the Northern Song emperors and their consorts and children.13 Neither of the first two Song emperors, Taizu and Taizong, had large harems. Both came to the throne as married adults (ages 34 and 38), and their habits were set before they lived in the palace. Taizu’s first wife had died before he became emperor, and he had his next wife made empress, and when she died, took another empress, but besides these three women he had no other high-ranking consorts. He did take palace women from the harems of the rulers he defeated, and four of his ten

182 Women and the family in Chinese history

children were born to unnamed low-ranking consorts, so it is not as though he was monogamous. Still, the total number of palace women never exceeded 300, and he once dismissed a quarter of the women in his palace, letting anyone go who wanted to go.14

His brother Taizong was on the throne longer (twenty-two years) and took a larger number of high-ranking consorts (eleven). Only three of these women, however, bore him children. In addition to the ten children they bore, Taizong had five other children, born to lower-ranking women whose names have not been preserved.

The first emperor reared in the palace, Zhenzong, was on the throne for twenty-five years, from age 30 to 55, but had only seven children. Only three of his eleven high-ranking consorts bore children by him, and they bore only one each; the other four children were by lower-ranking women. In 1008 and 1015 he dismissed 120 and 184 women respectively, but the total nevertheless gradually increased, reaching around 500.

It was under the next emperor, Renzong, that the numbers of palace women soared. Renzong came to the throne at age 13 and lived another 41 years, making his reign the longest in the Northern Song. Although Renzong was generally praised by Confucian scholars and historians, it was not for his handling of the “rear palace.” In his early years he was dominated by the Empress Dowager, and once she died in 1032, he promptly ousted the empress she had picked for him when he first took the throne, though to placate the officials, he also had to agree to oust her rival as well.15

In the 1040s and until her death in 1054, Renzong was particularly enamored of one consort, named Zhang, who bore him three girls. Never- theless, during this period, the number of palace women steadily increased. By the Jiayou period (1056 –1063), memorials from officials troubled by the expense reported that the number of palace women had reached nearly 10,000.16 Here the functional explanation of the harem seems to have some validity: one reason for the increase in palace women may have been that despite his many years on the throne Renzong did not beget many sons. Thirteen of his sixteen children were girls, and none of the three boys lived to maturity. As an economy measure, in 1060 the emperor agreed to cut the number of palace women to the level of 1023, but it is unclear how effectively this was carried out.17 No massive dismissals for 1060 or 1061 are recorded (though 450 had been dismissed in 1059).18

Yingzong, Renzong’s adopted son, came to the throne as an adult (age 32) and was on it only four years, so his case is too anomalous to be worth considering. All of his eight children were by the woman who was his wife before he took the throne.

Yingzong’s son Shenzong came to the throne as a young man of 20. He had been largely raised outside the palace and was already married to a woman who became his empress and would outlive him. In addition to her he had thirteen high-ranking consorts. He had twenty-three children by nine of these fourteen women. Thus, of the emperors discussed so far, he was the

Rethinking the imperial harem 183

Figure 8.2 Song statues of serving women attending the Jade Emperor, at the Shrine of the Jade Emperor, Jincheng, Shanxi province.

Source: After ZGMSQJ diaosu bian 5.88.

184 Women and the family in Chinese history

one to take the greatest advantage of the opportunities of the harem: he had children by the largest number of women as well as the most children. However, there are no complaints that the number of palace women increased during his reign.

Shenzong’s oldest son Zhezong was only 10 when he came to the throne, and like Renzong was under the domination of an empress dowager, in his case his grandmother. In preparation for Zhezong’s eventual marriage, Empress Dowager Gao had over 100 girls from suitable families brought into the palace and eventually selected one named Meng to become the new empress. After Empress Dowager Gao died in 1093 Zhezong seems to have taken out his resentment of his grandmother’s domination on the bride she had chosen for him. Not only did Zhezong encourage or at least tolerate his favorite, Consort Liu, to act rudely to his young empress, but on the basis of rumors ordered one of the chief eunuchs to conduct a judicial investiga- tion into alleged conspiracies in the harem. Thirty of the palace women were tortured, some ending up with broken bones or their tongues cut out. After nothing was discovered, Zhezong had Empress Meng deposed and made a Daoist nun. He had her rival Liu installed as empress.19

Zhezong was only 25 when he died, but by that time he had five children, only two of whom were by one of his nine high-ranking consorts, the other three by lower-ranking women. His only son died in infancy, so the succes- sion went to his younger brother, then 19 sui.

Huizong had spent most of his life in the palace, but had moved out a couple of years before he ascended the throne, and was already married. In his twenty-six years on the throne, he had two empresses and seventeen high-ranking consorts. He copied his father Shenzong in making use of the harem, having children by twelve of his high-ranking consorts (about one for every two years on the throne, just like Shenzong). But he exceeded Shenzong in producing children, averaging 2.6 per year to his father’s 1.4, possible because six of his consorts had four or more children.

Most of Huizong’s consorts, again like his father’s, had entered the palace as ordinary palace ladies. Historical sources describe Huizong as a man who could get very attached to particular consorts. His first favorite, Consort Liu (Mingda), had joined the palace service at age 14 in 1100 and bore her first child three years later at 17. By the time she died at age 27, she had borne six children altogether. Huizong was so distraught that he not only wrote poems in her memory, but also broke precedent to have her posthum- ously made empress. Even three or four years later, Huizong still missed her a great deal. He even asked Daoist priests to help him communicate with her.20

The girl the first Consort Liu had “adopted” then caught his fancy, and the histories report that he let her tend to him from morning to night. This second Miss Liu, we are told, was exceptionally good looking, skilled in applying make-up, and dressed stylishly. Once when Huizong was feasting some of his top officials in the rear garden, he invited them into a hall that

Rethinking the imperial harem 185

had a painted portrait of her, and later let them meet her in person, a rare privilege.21 After this second Consort Liu died in 1121, Huizong fell into deep grief and would weep with whoever came to console him. One of his consorts, Consort Cui, did not seem to him to be in grief, which annoyed him so much that he had her demoted to commoner, even though she had borne him six children.22 This seems to be the only time he treated one of the women close to him harshly.

A variety of evidence indicates that by the end of Huizong’s reign the number of palace ladies had again reached about 10,000. We know that Huizong periodically released women; from 1108 to 1118 he released 1,819 women, sometimes releasing as few as 68, sometimes as many as 600.23

Probably these were women who had already served more than five years with- out attracting any attention or otherwise distinguishing themselves. There was not, however, a simple rotation process of admitting 200 girls a year and after eight years beginning to release 200 or so. For after Huizong abdicated in 1125, 6,000–7,000 palace women were released, and yet there were still several thousand in the palace.24 This suggests something more like taking in 500 a year and later letting around 200 a year go, so that each year the total increased.

We have further detail on the numbers of women in Huizong’s palace because the Jurchens, after seizing the Song capital, Kaifeng, treated these women like booty and made lists of those they seized. When the Jurchen were placing demands on Huizong and his successor Qinzong in the first month of 1127, they were able to demand 2,500 palace women and 1,500 female musicians to distribute among their soldiers.25 Later, when the author- ities made an inventory of the members of the imperial family and palace taken north, they listed 504 palace women, for whom no details were listed, and 143 consorts of Huizong’s, who are listed by name, title, and age. The five of highest rank (also referred to as his wives, with ranks of huanghou, guifei, shufei, defei, and xianfei), ranged in age from 34 to 42, and all were mothers of children. The next 31 (also referred to as concubines) held several different titles and ranged in age from 19 to 39. Several of them were also mothers of princes and princesses. The remaining 107, none of whom had borne children, held the lowest titles and ranged in age from 16 to 24.26

From these figures we can discern a system in which women rise in rank both by bearing children and by getting older, but not in a simple lock-step fashion, as there was a lot of overlap.

We also have a record, something like what Marco Polo reported, of the regular introduction of new girls into Huizong’s bedchamber. A man who served as an interpreter for the Jurchen on the journey bringing the palace ladies north was asked about the practices in the palace. He reported that every five to seven days a virgin would be brought to Huizong. The next day she would be given a title, and if he sent for her again, she would get a pro- motion.27 This would imply that the 143 women with titles were all ones who had had sexual relations with Huizong. While it is certainly not inconceivable

186 Women and the family in Chinese history

that Huizong took 143 women over 26 years, which would be an average of a new woman every other month, it does seem surprising that he had children by no more than 20. And it is totally implausible that he took a new woman every week for 26 years, which would have meant 1,300 women altogether, and that no more than 20 of them bore children. The Jurchen liked to view their victory as the victory of the vigorous soldiers over decadent courtiers, so the translator may well have told them what he thought they wanted to hear.

Taking all of the Northern Song emperors together, over the course of a century and a half, the size of the palace harem increased many fold, but it did not do it steadily. It grew the most under two emperors, Renzong and Huizong; in each case reaching well into the thousands. These two emperors were both largely reared in the palace and may have been comfortable in this female-dominated world. They also had the two longest reigns. But that is about all they had in common. Renzong had trouble producing male heirs; Huizong had an abundance within a few years of taking the throne. Renzong let Confucian officials like Ouyang Xiu (1007–1072) and Sima Guang (1019– 1086) set the tone at court; Huizong went his own way and was more attracted to Daoism than Confucianism.28

Evidence that the number of palace women was only loosely linked to individual emperors’ love lives is given in Tables 8.2– 8.4.29 Two of these emperors, Renzong and Huizong, had huge numbers of palace women, and even during Shenzong’s reign there must have been several thousand, but the differences in the numbers of palace women, or the rate of their increase, seems to have had little to do with their love lives. All three had children by at least seven women, almost all of whom had entered the palace as palace women (not as empresses or as high-ranking consorts). None of the three was a “serial monogamist,” interested in only one woman at a time; in at

Table 8.2 Children born to Renzong by two-year period and consort

Yu Miao Zhang1 Zhang2 Dong Zhou Other

1037 1 2 1039 1 1041 1 1 1043 2 1 1 1045 1047 1049 1051 1053 died 1055 1057 1059 2 1 1 1061 died 1 1

Rethinking the imperial harem 187

Table 8.3 Children born to Shenzong by two-year period and consort

Xiang Song Zhang Xing Zhu Guo Chen Wu Lin Other

1067 1 1069 1 1 1071 1 1073 1 1 1 1075 2 1077 2 1079 1081 2 1 1 1 2 1 1083 1 1 3 1 1 1085

least some period of their lives all were begetting children by several women in a single two-year period. And all of them could lose interest in a particular woman after she had borne him children, although Huizong retained his interest the longest, in several cases for well over a decade; two of the women he remained interested in the longest had their last child at 32. Although Huizong had many more children than his father or great-grandfather, that was due not so much to having sexual relations with many more women, but by retaining his interest in particular women for more years.

On the basis of their sexual habits, then, none of these emperors needed more than a dozen or two palace women at a time, since new young girls were regularly brought into the system and older ones released, ensuring that there would always be some who would be new to the emperor, should his interest flag.

If emperors’ sexual appetites do not do much to explain the huge num- bers of palace women, is there anything else that does? Two features of Song palace institutions may have played a part. The first is the interests of the consorts of prior emperors. In Song times the consorts of a previous emperor stayed in the palace under his successors. When Yingzong first took the throne in 1063, Sima Guang urged him to release all the palace ladies from his father’s time who had never been “favored” or borne a child.30 All those with higher titles were automatically kept on, even if they had not borne children. When Zhezong came to the throne in 1085, for instance, among the consorts of prior emperors given promotions were six of Renzong’s, four of Yingzong’s, and nine of Shenzong’s consorts. These women were still very much a part of the “rear palace.” For instance, in 1190 when Shenzong’s widowed consort Lin died, Renzong’s widowed consort Feng took to raising Lin’s two sons. To put this another way, Zhezong’s step great-grandmother took to rearing his half brothers after his step-mother died. Previous emperors’ consorts needed palace ladies to attend them, add- ing to the size of the total establishment. The romantic view of the palace ladies sees them as competing with each other for the emperor’s attention,

188 Women and the family in Chinese history

T ab

le 8

.4 C

h il

d re

n b

o rn

t o

H u

iz o

n g

b y

tw o

-y ea

r p

er io

d a

n d

c o

n so

rt

W an

g Z

he ng

W an

g2 L

iu 1

Y an

g Q

ia o1

W an

g3 W

ei C

ui W

an g4

Q ia

o2 H

an L

iu 2

O th

er

11 00

2 2

11 02

1 1

1 1

11 04

1 2

1 1

1 11

06 1

1 2

2 1

1 11

08 D

ie d

1 1

1 1

1 1

11 10

1 1

1 1

1 1

1 1

11 12

1 D

ie d

1 2

1 1

11 14

1 1

1 1

1 1

1 11

16 D

ie d

1 11

18 1

1 3

11 20

1 2

11 22

2 11

24 1

4

Rethinking the imperial harem 189

but many of the women in the palace were outside that competition, includ- ing not only the empress dowager, but also the lower-ranking consorts of previous emperors. There would have been no great reason for these women to oppose the tendency for the numbers of palace ladies to grow, as it made their world larger. Indeed, the palace was something of a women’s city. Emperors who grew up in this environment do not seem to have disliked it.

A second institutional feature that could have contributed to the prolifera- tion of palace ladies is the success of the Song at limiting the numbers of eunuchs in palace service to no more than a couple hundred.31 Thus senior women were largely served by junior women, not eunuchs.

Contributing as well would have been the natural tendency of institutions to grow and the effort it takes to shrink them. In that light, it is not surprising that the only emperor who seems to have managed to curtail growth of the harem was Shenzong. It may be of some importance that he did not grow up in the palace and was not dominated by an empress dowager. But prob- ably more important was his strong desire to establish the New Policies, get the budget under control, and build up funds for military purposes.32

Although these institutional factors must have contributed to the growth of the imperial harem, I do not see them as the sole explanation for the growth in the numbers of palace ladies, if for no other reason than they have more to do with the specific case of the Northern Song than the whole period from the Qin dynasty on. This brings me back to “cultural logics,” widely held ideas and powerful symbols that fostered the practice.

Consider, first, the relationship between palaces and palace ladies. In China, as elsewhere, rulers tried to manifest the splendor of their rule by building opulent palaces. In Europe, where Christianity required that a king have only one consort, an opulent palace like Versailles could please the senses through gardens, elegantly appointed rooms with paintings, tapestries, and sculptures, as well as by the presence of fashionable dressed courtiers. Certainly servants, primarily male servants, were numerous and well-dressed, but they were not a key component of the aesthetic appeal of the palace.

In China, however, a large part of what made palaces beautiful and splen- did was the palace ladies who, in a sense, furnished them. For instance, writers sometimes talk of completing a palace by filling it with “beautiful women.” Sima Qian reported that after Qin Shihuangdi built his palace in Xianyang, he filled it with the beautiful women and musical instruments he had taken from the feudal lords he had defeated.33 In a similar vein, the Han Wu gushi told the story that when Wudi built the Mingguang palace, he recruited 3,000 beauties from Yan and Zhao to fill it.34 The rhapsody that Ban Gu wrote about the Western Palace gave the names of the halls in the rear palace, and described their lavish decoration, with “gilt thresholds, jade stairs, vermilion courtyards” but also the “Red-gauzed beauties, sleeves dangling, with silk-braided ribbons, tangled and twisted, their pure radiance gaily glittering, bobbed up and down like goddesses.” He also mentioned that these “modest and retiring” “burgeoning blossoms” numbered in the

190 Women and the family in Chinese history

hundreds.35 Unlike European palaces, however, this form of splendor was a private, concealed splendor, for the emperor to enjoy and other men merely to imagine.

This image of palaces made beautiful by the presence of palace ladies was later absorbed into Daoist visions of paradise. By the end of the Six Dynast- ies Daoist paradises were pictured as palaces where elegantly dressed “jade maidens” wandered among brightly-painted halls and flower-filled gardens.36

Daoist images of paradise in turn shaped notions of the perfect palace, and palace ladies could be likened to immortals. To give a Song example, Hong Mai, in his Yijianzhi, records an anecdote that begins by describing the splendors of the Genyue garden that Huizong had constructed outside the palace, with all of its halls and palaces as well as stands of bamboo and other natural scenery. The sight of the palace ladies going in and out among them “like immortals” was enough to get Huizong to climb one of the taller buildings so that he could gaze down at them.37

The association of palace ladies with immortality is also worth closer examination. From Qin and Han times on, the popular conception of ruler- ship in China owed a great deal to the elaboration of ideas about the Yellow Emperor.38 In this tradition, the Yellow Emperor is militarily powerful, extremely yang; he is a Daoist Sage of vast powers; he is associated with immortality, knowing the secrets to it, presiding over various paradises. In medical literature, he is the fountainhead of medical knowledge. In the esoteric sexual treatises, he is instructed in the sexual arts by the Plain Maiden, and mastery of these sexual arts can lead to immortality. But full employ- ment of these sexual arts required an unlimited supply of young virgins – the more women whose yin essence is absorbed, the more progress is made toward immortality. Here it is worth noting that the Song imperial house, ever since Zhenzong, traced its descent back to the Yellow Emperor.39

It is unlikely that Renzong or Huizong or any of the other Song emperors ever consciously thought that they could either enhance their yang power, their military might, or their chances for immortality by gathering thousands of virgins in the palace. Nevertheless, ideas of kingly majesty and power were so entangled with ideas of the superhuman Yellow Emperor that they did not need to consciously think about such connections. Even if Confucian moralists condemned large harems as wasteful and dangerous, they were fighting against an equally strong tradition that associated the possession of thousands of young beauties with the Yellow Emperor, immortality, pleasure, and imperial majesty.

***

Does rethinking the imperial harem help us understand any aspects of life beyond the palace walls? Although only a tiny portion of the Chinese population ever had any direct experience of the imperial harem, the cultural significance of the harem should not be underestimated. Not only were

Rethinking the imperial harem 191

Figure 8.3 Illustration from chapter 21 of the novel Jin Ping Mei, showing the protagonist with several of his concubines and their maids.

192 Women and the family in Chinese history

stories of emperors and their favorites staples of Chinese drama and fiction, but even stories of the amorous adventures of other men often borrow imperial imagery. These stories helped form Chinese conceptions of gender and male–female relations; ideas about the attractiveness of the lonely, longing woman, and the impressiveness of the man attended by several beauties owe a great deal to the power of these images of the most powerful man in the country and his essentially unlimited supply of attractive young handmaidens.

I would like to go further here, however, and suggest that the imperial case offers insight into one of the major questions that has engaged the attention of historians of Chinese women in recent years, the question of change over time. Lately, authors of books on women in Chinese history, myself among them, have made every effort to show that women were not outside of history, but just as much enmeshed in it as men were.40 We all, for instance, tried to point to aspects of economic change that would have shaped women’s lives, ranging from changes in the way textiles were pro- duced and marketed, to the impact of increasing commercialization on the market for women. Moreover, we all bring in intellectual and cultural change, particularly developments in Confucian thought and the steady expansion of print culture. If anything, these books probably all err on the side of exaggerating the impact of change, since if you read them together you cannot help but wonder why women in the eighteenth century had so much in common with women in the eleventh, given the pace of change we all claim. For instance, Susan Mann’s description of Qing women making cloth sounds a lot like my description of Song women making cloth, even though Franscesca Bray claims that there was a dramatic change in the social organ- ization of textile production between Song and Qing times. Similarly, despite all the developments intellectual historians talk about, Mann’s description of Qing women’s Buddhist piety does not sound all that different than what I found in Song times. In other words, no matter how hard we look for change, we keep coming up with a lot of recurrent patterns as well.

It is not difficult to think of possible explanations for recurrent forma- tions, from long-acknowledged ones like the persistent hold of basic Confu- cian family ethics and the basic patrilineal family system, to more nuanced ones such as Franscesca Bray’s emphasis on the ways the basic structure of Chinese houses shaped people’s experience of the family and sexual segrega- tion. But the persistent tendency for the imperial harem to grow huge sug- gests that it is important not to dismiss the power of deep cultural codes underlying conceptions of gender. Notions of imperial majesty seem to have been conveyed outside of the voluminous Confucian literature on emperor- ship, and indeed in opposition to it. The image of the Yellow Emperor as the alpha male undercut the Confucian literature on the responsible em- peror, but at a level that no Confucian adviser could directly address since it was never made explicit. I suspect that notions of manliness and femininity in the larger society similarly drew from ideas conveyed through popular

Rethinking the imperial harem 193

religion and popular literature, including ones that undercut or counter- acted ideas and images conveyed at the explicit intellectual level in Confucian writings on family virtues, theoretical analyses of yin and yang, as well as the laws governing descent and transmission of property. This untheorized level of culture is much harder to dislodge, since it is hard to recognize and hard to attack. When features of gender relations persist despite changing economic and cultural opportunities, we need to think about how the per- sistence of unarticulated assumptions might be implicated.