293

SHORT-TERM TRENDS IN BASTARDY IN

Richard E. Barrett*

By international standards, bastardy was a In recent years, historical demographers relatively common occurrence in Taiwan have published a considerable amount on during the Japanese colonial period (1895- European patterns of illegitimacy during 1945). This may come as somewhat of a the last century (Shorter, Knodel, and van surprise to many of those familiar with an- de Walle, 1971), and on certain smaller thropological accounts of Chinese popula- areas for longer periods of time (e.g., Las- tions ; in most of these studies, illegitimacy lett, 1977:102-159). Yet for China, aside is presented as a rare phenomenon. 1 from Arthur Wolds s pioneering work with Taiwanese household registers from the *Richard E. Barrett was awarded the Ph.D. in Haishan district, we have almost no reli- sociology in 1978 by the University of Michigan. He is able data on illegitimacy (see Wolf, 1975; now Assistant Professor in the Department of Sociol- Wolf and Huang, 1980:251-260). ogy at the University of Illinois, Circle. Chicago There are a number of levels at which The author would like to thank Martin Whyte, of can be anal- Ronald Freedman, John Knodel, Arthur Wolf, Wil- the phenomenon bastardy liam Lavely, and John Casterline for their suggestions yzed. One can study its prevalence, its and criticisms; of course, final responsibility for the social and economic context, its meaning ideas and interpretations presented here rests with the for the parents or for the bastards them- author. In addition, the staffs of the Institute of Eth- selves, or its prevalence as perceived by the nology, , and the Taiwan Branch local and the which Library of the National Central Library (both are in people meanings they , Taiwan), and of the Orientalia Division, attach to it. Anthropologists of China have United States Library of Congress, deserve thanks; tended to concentrate on the latter two as- their policies of allowing researchers free access to their stacks had a direct effect on the discovery of some of the findings reported in this paper. Second, many of these studies were written at a time when "functionalism" was a dominant paradigm ’See, for example, Levy, 1968:337; or Yang, 1968: in British and American social science. While func- 118-121. Illegitimacy on was gen- tionalism provided many insights into Chinese insti- erally ascribed to the penetration of urban or tutions, this view of social relations can lead to an "modern" influences into the traditional family sys- overemphasis on the degree to which prevalent norms tem. influence actual behavior. There appear to be three major reasons for this Third, the field study of anthropologists may be portrayal of Chinese bastardy. First, many of the a poor method by which to study bastardy. During English-language anthropological studies of China in the course of a year’s stay in a village the anthropolo- the 1920-1949 period appear to be conscious attempts gist may witness few illegitimate births, or the nature to emphasize the more "normal" or mundane aspects of these births may be successfully concealed. The in- of village or family life as an antidote to missionaries’ vestigator who is dependent on the goodwill of his or horror stories or Western imperialist apologists’ pic- her hosts may consciously or unconsciously avoid ask- ture of China as a "decadent" culture which deserved ing potentially embarrassing questions about the the foreign exploitation to which it was subjected. legitimacy of births in the area. 294 pects ; while information on such percep- 5. Marital and non-marital age specific tions and norms is valuable, there is no fertility rates indicate that the of necessary relation between such percep- illegitimate children tend to be somewhat tions and the actual prevalence of bas- older than the mothers of legitimate chil- tardy. Hopefully, future anthropologists of dren. China will distinguish more clearly which Most of the salient facts about the pat- parts of their work are based on thorough terns of Taiwanese marriage, divorce, investigation and which are largely the re- widowhood, and marital fertility have sult of informed speculation or conjecture. already been discussed by George Barclay Of course, if one is to hold the anthro- (1954a:210-254). Most women married in pologist to such rarified standards of pro- their late teens, and husbands were usually fessional honesty, one should be willing to about four or five years older than their take a tentative step in this direction one- wives. While the mean age at first mar- self. Thus it should be stressed that the riage increased slightly for women, it de- data which will be presented below can clined for men during the period: in 1906, only answer certain questions about the the average age at first marriage was 18.1I prevalence and distribution of illegitimacy, years for women and 24.0 years for men, not about its social meaning for the Tai- and in 1942, it was 19.0 years for women wanese.2 and 22.3 years for men (Bureau of Ac- An analysis of the colonial vital statistics counting and Statistics, 1946:190). and census records shows that: As Barclay has shown, between 1905 1. While the vast majority of Taiwanese and 1935, 99 percent or more of all Tai- women married at least once, there were wanese women over thirty-five years of age frequent extramarital births to single, had married at least once. Yet relatively divorced, or widowed women. large numbers of them were widowed at 2. The illegitimacy ratio and non- early ages, and a sizeable number of them marital fertility rates appear to have been divorced or were divorced by their hus- on the increase from 1905 to 1935, but de- bands. However, because of the strong cline slightly after that period. propensity towards the remarriage of 3. Non-marital fertility rates in Taiwan- divorcees, and the tendency of younger ese cities were little different from those widows to remarry, widows, divorcees, and prevalent in the surrounding countryside, single women constituted only a small pro- and regional differences tended to be more portion of women of childbearing age at striking than urban-rural ones. any time (see Barclay, 1954a:216-237).’ 4. Illegitimacy ratios of Hokkiens and Throughout the 1906-1943 period, the Hakkas, the two major Chinese cultural Japanese colonial police administered an groups, exhibit no systematic differences. efficient and highly accurate system of household registration (see Barclay, 1954b:1 2The prevalence of illegitimacy, however, may well be related to its social meaning. The Chinese did have certain traditional mechanisms for the integra- 3Anthropologists and sociologists of China have tion of bastard children into the ritual life of the com- belatedly taken heed of Barclay’s convincing demon- munity and for providing them with some access to stration of the high rates of remarriage of widows and resources. In Taiwan, for example, illegitimate chil- divorcees in Taiwan (see, for example, Freedman, dren can participate in their father’s funeral ritual 1966:59-67). From Barclay’s work and my own in- (see Wolf in Freedman, 1970:194-195). Moreover, vestigation of the conditions of the marriage market the Ch’ing dynasty inheritance code used in Taiwan in colonial Taiwan (see Barrett, 1978b), arguments provided for the rights of bastard sons as well as that divorced or widowed women resorted to bastardy legitimate ones (see Okamatsu, 1971: appendix, as a means of old-age security because they were un- xxiii). able to find husbands are not supported by fact. 295

3-13; Chen, 1975:391-416; and Barrett, had the world’s highest ratio of police to 1978a:68-88). With a few exceptions, field people-the accuracy of the system was investigations by anthropologists and probably due in large part to the cross- various statistical checks on reliability ap- checking of the quinquennial census re- plied by demographers have shown that sults against the registration of vital events the data give an accurate picture of the ac- in the household registers, a procedure tual demographic conditions of the time, which reduced errors in both sets of data. and that the data are far more reliable Happily, this time-consuming procedure than any extant data from the Chinese also permits later researchers to interfile mainland (c.f. Cohen, 1976:19). census and vital statistics data in com- Apart from the efficiency and ubiquity puting various demographic rates and of the colonial police-the island may have measures.

TABLE 1. NUMBER OF LEGITIMATE AND ILLEGITIMATE BIRTHS AND BIRTHS TO CONCUBINES, AND SEX RATIOS AT BIRTH, TAIWANESE WOMEN, SELECTED YEARS, 1906-1940. 296

The Prevalence of Bastardy mon perception of who was &dquo;married&dquo; in a Taiwanese community in this period did The annual distribution of all Taiwanese not vary greatly. While Taiwanese villagers births across the three categories of legiti- tend to differ among themselves about the macy used in the colonial records in vari- exact point in the marital negotiations the ous years is presented in Table 1. The marriage actually begins (some feel it only Japanese colonial authorities required that begins with the recording of the event in all births in an area be recorded in the the household register; see Gallin, 1966: household (kosekl) register maintained at 212), there is no question that a con- the local police station. Compliance was tractual relationship has been undertaken, enforced through a system of police super- and that the contract has a legal status (see vision and joint responsibility, an effective Okamatsu, 1971:appendi~, v-x). Accord- form of the traditional pao-chia system ing to Okamatsu, even &dquo;brief ceremonies&dquo; (see Chen, 1975). of the common people were recognized by Legitimate births included all those the nineteenth century Ch’ing dynasty pro- births to currently married women. While vincial government, and there is no reason Barclay points out that the legitimacy of to believe that the Japanese colonial gov- birth is thus largely determined by admin- ernment made their stipulations any more istrative fiat, and may have failed to in- stringent. clude some marriages which had not been The major question centers on whether registered with the colonial authorities, I or not a sizeable minority of people mar- think that this objection to the use of these ried but failed to register the marriage statistics is a relatively weak one (see Bar- with the Japanese authorities. While it is clay, 1954ae241).&dquo; The problem of marital true that registration of marriages was not definition occurs in almost all literate cul- required during the Ch’ing dynasty (Oka- tures, and tends to confound most studies matsu, 1971:appendix, vii), none of the of illegitimacy, particularly those which published investigations of household regi- purport to chart changes over time (see stration materials from the Japanese col- Shorter, Knodel, and van de Walle, 1971: onial period by anthropologists indicates 380-382). that underregistration was a serious prob- There are several reasons to believe that lem. the household register data and the com- For the Chinese, marriage is a religious ceremony, but it is a family-based one which requires neither church approval 4It is possible that because of a reliance on an nor a church ceremony. In fact, the partic- administrative rather than a cultural definition of of Taoist is ex- marriage the illegitimacy rates presented here are ipation &dquo;clergy&dquo; largely insofar as it to en- overstated. Nevertheless, Barclay presents no evi- traneous, except helps dence that this was a serious problem in the data; if sure supernatural cooperation in a key anything, the colonial statistics are remarkable for event in the family life cycle (see Gallin, the of women who were considered high percentage 1966:209). Religious groups have never "administratively married," usually at fairly young had to forbid so ages. any legal power marriage, One researcher with extensive fieldwork experi- that one of the major sources of defini- ence and use of the household registers has pointed tional ambiguity does not complicate the out that while underregistration of minor marriages Taiwanese case. was possible, the public festivities associated with The colonial authorities allowed three major marriage and the usual rural practice of in- of unions to be as full viting the colonial police to the wedding make it un- types registered likely that they were not recorded (A. Wolf, personal marriages: major marriage (where the communication). bride married into the groom’s house), 297 minor marriage (where the bride had well as the numbers of married, never- previously been adopted into the groom’s married, widowed, and divorced women house as a &dquo;little daughter-in-law&dquo;), and given in all the censuses. uxorilocal marriage (where the groom The Japanese also registered all births in married into the current household of the each year in one of three legitimacy cate- bride). The authorities did not permit the gories : fully legitimate births, illegitimate registration of unions with concubines as births, and births to concubines. Births to full marriages. concubines had a semi-legitimate status According to the Sasos, the Japanese because such births were acknowledged by colonial authorities did not want to en- a father, even though there was no legal courage concubinage. But they had marriage contract between the parents. already decided that the Ch’ing dynasty In the analysis below, primary emphasis common law code, which legally recog- will be placed on the measurement and in- nized concubinage and certain forms of terpretation of &dquo;true&dquo; illegitimate births to slavery, would be one legitimate form of women who were neither currently married common law in Taiwan for the Taiwanese nor concubines. The registered births of (Saso and Saso, 1975:62). As a result, con- children to concubines will be excluded cubinage was recognized as a &dquo;quasi- from the computation of all rates and marriage,&dquo; and the children resulting ratios of illegitimacy in order to avoid the from such a match would have a right to possibility of exaggerating the extent of il- share in the inheritance. But legitimacy by including births which may have seemed legitimate to some Taiwan- Since the quasi-spouse was not considered to even the authorities be a legitimate spouse, in civil and criminal ese, though disagreed. cases bigamy and adultery could not be con- Conceivably, illegitimate births include sidered as legal cases to be handled in court. those births to concubines whose consorts In a series of legal decisions, the Japanese refused to recognize paternity when the soon established the customary privileges of births were registered, even though the con- the concubine in her quasi-marriage, as re- ventional reason for a concubine flected in the Koseki family registers. If a con- taking a heir. cubine decided to marry another man as a was to produce male (Despite this legitimate wife, divorce proceedings were un- F.L.K. Hsu found considerable slippage necessary, nor could she be accused of adul- between the ideal and the reality in West tery (Taiwan Supreme Court, 1905) or bigamy Town, Yunnan see Hsu, 1971: (Taiwan Supreme Court, 1917) (Saso and province; Saso, 1975:63). 105.) If we examine the sex ratio at birth from Thus the consort of a concubine, like 1915 on (see Table 1), we find that the sex the father of an ordinary bastard, could ratio for legitimate births is either 105 or not compel an exclusive sexual relation- 106, which is close to what is thought to be ship with his partner. It can be argued that the normal distribution of births by sex for the minimum condition for a marital re- large human populations.5 However, lation, at least in a male-dominated society the sex ratio of illegitimate births is consis- like China’s, is an enforceable contract for tently below this figure, indicating a sexual exclusivity on the part of the . If so, then in Taiwan at least, con- cubinage was not a full form of marriage. 5It has been shown (Barclay, 1954a:242, fn.) that there was a towards the In some censuses, the listed tendency underregistra- Japanese tion of female stillbirths and neo-natal deaths, but concubines as a of married sub-category this problem had largely been corrected by 1915. As a women. Thus in those years we have the result, the sex ratios at birth for different legitimacy number of concubines of various ages, as categories in 1906 and 1910 may not be too reliable. 298

probable underregistration of male births. was a gradual rise in the illegitimacy ratio Conversely, the sex ratios at birth for con- until 1935, when illegitimate births com- cubines’ children are consistently higher prised 4.4 percent of all births. The concu- than 105, which shows that there was an binage ratio hovered close to one percent overregistration of male births. of all births for most of the period. I do not One possible explanation for these dif- have any ready answer for why there was a ferences is that the consorts of concubines rather sharp upsurge in the concubinage were more likely to acknowledge paternity ratio (1.7 percent) in 1940. of a male child than a female one. This This gradual increase in the illegitimacy would account for the overrepresentation ratio is over-shadowed by the magnitude of of boys among concubines’ births and their its areal variation. While figures for the underrepresentation among illegitimate lower-level units, the districts and town- births. As was suggested in footnote 5 ships, do not appear to be extant, an above, it seems unlikely that these differ- examination of the prefectural statistics on ences in sex ratios at birth were the result illegitimacy shows that there appear to of underregistration of early female infant have been different cultural patterns or deaths after 1915. adaptations of illegitimacy, and that these Demographers are generally interested patterns were relatively stable over time. in the proportion of illegitimate births The colonial government began record- among all births. This proportion gives an ing illegitimate births by prefecture in indication of the social importance of bas- 1920. A comparison of the illegitimacy tardy, as well as the contribution of illegiti- ratios for prefectures in 1920 and 1935 macy to the overall rates of fertility. shows the stability of these social patterns over the fifteen-year period (see Table 3). TABLE 2. DISTRIBUTION OF TAI- As in previous analyses of Taiwanese WANESE BIRTHS BY LEGITIMACY fertility (e.g., Barclay, 1954a:232), we will STATUS, TAIWAN, 1906-1940, SELECTED place primary emphasis on the patterns of YEARS, IN PERCENT. fertility in the western coastal plains where most of the Taiwanese lived. The two East Coast prefectures had large proportions of aborigines in their populations and much lower ratios of people to arable land, and the Japanese colonial authorities excluded the central mountain aborigine preserve from the statistics and census system. There are considerable differences in the illegitimacy ratios of the West Coast prefectures: Taipei’s ratio hovers at about Source of Data: Bureau of Accounting and Sta- 9 percent, while ’s rate is only tistics, 1946:216-217. about 2 percent in both 1920 and 1938. While economists and anthropologists The simplest of these measures is the il- have identified certain regional differences legitimacy ratio, which is illegitimate in economic and family organization and births expressed as a percentage of all dominant crop type, these variations do births. This ratio, together with its equiva- not appear to suggest any immediately ob- lent, the &dquo;concubinage ratio,&dquo; is given for vious explanations for differences in the various years in Table 2. Apparently there illegitimacy ratios. 299

TABLE 3. ILLEGITIMACY RATIOS, TAIWANESE PREFECTURES, 1920 AND 1935.

Note: While the present analysis centers on the west coast prefectures, Taitung and Hualien prefec- tures were included here to show that their illegitimacy ratios were relatively stable over time as well. The reasons for the high illegitimacy ratios among the populations of these two east coast prefectures are not clear, but they may be related to the fact that both were still &dquo;frontier&dquo; areas during the colonial period. *Included in Prefecture until 1926.

One possible explanation for these dif- 1935 shown in Table 4 indicate that the ferences is that they reflect the more devel- ratios are variable in both areas, but that oped urbanization in some prefectures. the urban sections of prefectures have con- Since urban areas often have surpluses of sistently higher ratios than the rural young, unmarried men or women, hypoth- regions. Thus, illegitimacy appears to be eses about the effects of urban living on linked to urban residence. the levels of illegitimate fertility often use Nevertheless, the illegitimacy ratio says variations in the sex ratio. As Knodel and nothing about the rate of bastardy among Hochstadt point out, there are two contra- currently unmarried women. It is possible dictory arguments about the influence of that the illegitimacy ratio is strongly af- the sex ratio on illegitimacy. One theory fected by the larger proportion of unmar- says that a surplus of women at marriage- ried women in cities, and that the actual able ages will give males an advantage in fertility of these unmarried urban women sexual bargaining, and the result will be may actually be below that of their rural more illegitimate births. The other argu- counterparts. Another demographic mea- ment holds that a scarcity of women will sure of the rate of illegitimate births must subject the few available unmarried be computed if we are to settle this ques- women to a constant hounding by men, tion. which will lead them to grant favors and These are a variety of ways of measuring inflate the illegitimate birth rate (see the rate of illegitimate fertility. I have Knodel and Hochstadt, 1976:28).66 chosen to compute Coale’s Ih, which is &dquo;an The illegitimacy ratios of the urban and illegitimate fertility rate (number of illegit- rural sections of West Coast prefectures in imate births/1,000 unmarried women of childbearing age) which has been ’in- directly standardized’ for the age distribu- 6It has also been that a severe short- suggested tion of the unmarried women (this is im- of women would Taiwanese to age encourage parents because of differences in fer- sell their daughters into prostitution, which would portant age and van de tend to inflate the illegitimacy ratio (N. Diamond, tility)&dquo; (Shorter, Knodel, personal communication). Walle, 1971:379). The Ih is the most 300

TABLE 4. ILLEGITIMACY RATIOS FOR TABLE 5. SEX RATIOS, ILLEGITIMACY URBAN AND RURAL SECTIONS OF PRE- RATIOS, AND THE INDEX OF ILLEGITI- FECTURES, TAIWAN, 1935. MACY (Ih) FOR URBAN AND RURAL SECTIONS OF PREFECTURES, TAIWAN, 1935.

Note: The urban areas of the prefectures inclu- ded : Taipei Prefecture: Taipei City, Chitung City; Prefecture: Hsinchu City; Taichung Prefec- ture : Taichung City, Changhua City; Prefec- ture : Tainan City, City; Kaohsiung Prefec- ture : Kaohsiung City, Pingtung City. All other sec- tions of each prefecture were included in the rural area.

sophisticated measure that can be com- puted from the available data, and it is a useful one because it measures non-mari- tal fertility against an objective standard. Since Coale’s Ih measure has been widely used in European historical demographic research, it also permits us to compare - -- .. --- , - - .-- - . - . - . - - . - . , . - aNote: the of and Taiwan with that and other areas of the For specification urban rural areas of these see Table 4. world.’7 prefectures, b’This sex ratio is the ratio of all currently un- The is to the Ih superior illegitimacy married men aged 15-34 years to all currently un- ratio for the determination of the actual married women aged 15-34 years in each area. probability of births to unmarried women between the ages of 15 and 49 years. In tive rural regions, the differences are much Table 5, the levels of the Ih, the illegiti- smaller than when the illegitimacy ratios macy ratio, and the sex ratio of currently were used. This indicates that the greater unmarried men 15 to 34 years to un- incidence of illegitimate births in cities is married women of the same ages are given. partially due to the greater number of un- First, it should be noted that while the married women in those areas. Moreover, values of Ih for urban areas of prefectures some rural regions of prefectures, such as are still higher than those for their respec- Taipei or Kaohsiung, have higher indices of illegitimacy (Ih) than some urban areas of such as those of ’The four measures of fertility developed by prefectures, Hsinchu, Ansley Coale (i.e., If, Im, Ig, and Ih) are useful tools Taichung, and Tainan prefectures. This in comparative historical demographic research fact shows that differential illegitimate fer- (see Coale, 1969, for definitions of these measures). tility is unrelated to the variations in the Estimates of these measures for Taiwan at various of urbanization between prefec- dates can be found in Barclay et al., 1976:616, and in degree tures. Barrett, 1978a:171. For what is probably the most accurate estimate of these measures for Taiwan, see The comparison of the sex ratios of un- Freedman and Casterline, 1979. married men to women between the ages 301

of 15 and 34 years and either measure of ethnicity on the illegitimacy ratio (see illegitimate fertility does not provide much Tables 6 and 7, and Figure 1). support for either of the theories men- While the Hokkiens and Hakkas did tioned above. Neither the scarcity nor the have minor difference in marriage patterns abundance of unmarried women appears in this period (see Barclay, 1954a:231-233), to have much relation to the current level the data show that ethnic differences had of bastardy. Of course, it is possible that little effect on illegitimacy ratios. Again, the kinds of social interactions implied in region seems to be a more important deter- these theories occurs within a much minant of illegitimacy than ethnicity; the smaller area than the prefecture. If this is ethnic groups show marked differences the case, then these data should not be only in areas where the total number of taken too seriously. However, since the births to women of one of the ethnic groups colonial authorities do not appear to have were less than one thousand. In such cases collected data on the legitimacy of birth we might well expect some instability in the for areas below the prefectural and city ratios, due to the small number of cases level, it may be necessary for future re- used in their computation. searchers to attack this question through It has been argued that Hakka women the reconstruction of household registers had a higher status relative to that of rather than through the use of census and women in other southern Chinese ethnic vital statistics data. groups. Hakka women did not bind their feet, and there may have been less occupa- IMegitimacy Ratios of tional segregation (especially in farming) in tiokkiens and Hakkas Hakka communities. However, the greater The Japanese divided the &dquo;Taiwanese&dquo; &dquo;freedom,&dquo; if that is the inference one population into three categories: those of wishes to put on such cultural differences, Fukien origin (Hokkiens), those of Kwang- did not appear to have a marked effect on tung origin (Hakkas), and the lowland and their illegitimacy ratios. highland aborigines.8I While there are Non-marital Age Specific Fertility Rates some ethnic or linguistic differences within the two major Chinese groups, the Hok- One of the drawbacks of working with data kiens and the Hakkas, one hypothesis on births in colonial Taiwan is the unavail- about differences in illegitimacy is that it ability of specific information on the ages of reflects differences in culture between mothers. Researchers have tried to get these two major groups. Since the 1925 around this problem by using age specific and 1935 census and vital statistics vol- fertility rates of countries such as Japan in umes provide data on marital status and computing plausible estimates of the Gross ages of women and legitimacy status of Reproduction Rate (see Barclay, 1954a: births by ethnic group in each prefecture, 246). But the drawbacks of this kind of it is possible to investigate the influence of solution are fairly obvious. The recent de- bate among historical demographers about the prevalence of &dquo;natural fertility&dquo; 8The most data on the ethnic back- complete among premodern European and Asian ground of the Taiwanese (and perhaps of any Chinese has made this lacuna even in a 1926 census of ethnic back- populations area) appear special 9 ground conducted by the colonial authorities. An more regrettable.9 analysis of the results of this census and estimates of the degree of spatial segregation of the twelve ethnic and dialect groups can be found in Barrett, 1978a: 9The recent collections of studies in the volumes 227-273. edited by Lee (1977) and Tilly (1978) provide many 302

TABLE 6. HOKKIEN ILLEGITIMACY RATIOS BY PREFECTURE, TAIWAN, 1925 AND 1935.

Note: The &dquo;All Taiwanese&dquo; category includes Hakkas, Hokkiens, other Chinese, lowland aborigines, and highland aborigines not under separate Japanese police administration in the mountain reserves. *Part of Kaohsiung Prefecture until 1926. TABLE 7. HAKKA ILLEGITIMACY RATIOS BY PREFECTURE, TAIWAN, 1925 AND 1935.

Note: The &dquo;All Taiwanese&dquo; category includes Hakkas, Hokkiens, other Chinese, lowland aborigines, and highland aborigines not under separate Japanese police administration in the mountain reserves. *These prefectures had fewer than 1,000 Hakka births per year; as a result, some of the illegitimacy ratios may be unstable. In 1935, there were no Hakka births in Prefecture. **Part of Kaohsiung Prefecture until 1926. In the course of working in the stacks of survey of the colonial demographic re- the Orientalia Division of the Library of sources (see Taeuber, 1961).’° This report Congress in Washington, D.C., I came across a Taiwanese demographic report which had apparently escaped the notice 10This slim volume (Government General of was and misfiled in the of either George Barclay or Irene Taeuber, Taiwan, 1943) uncatalogued Manchurian section of the library’s Japanese-lan- who had also conducted a fairly thorough guage holdings. As far as I know, it is unavailable at either the Taiwan Branch Library of the National comparative data on natural fertility in many areas. Central Library or at the Main Library of National In addition, Henry’s (1960) reworking of Chi-hsien Taiwan University (both are in Taipei, Taiwan), the Tuan’s data (1958) shows how the natural fertility two other main repositories of published Japanese re- hypothesis might be applied in a Taiwanese context. ports on colonial Taiwan. 303 304

provided the following information, by shrinkage in each single-year cohort over prefecture and urban and rural areas, on the eight-month period could be estimated all Taiwanese and Japanese births on the with a high degree of accuracy (see Govern- island in 1941: 1) age of and father ment General of Taiwan, 1942:54-66). for all legitimate births; 2) age of father for After the deaths for each single-year co- all births to concubines; and 3) age of hort were subtracted, the survivors were mother for all illegitimate births. In addi- apportioned among marital statuses ac- tion, the report provided the same infor- mation on ages of parents by marital status for all still-births. TABLE 8. AGE-SPECIFIC FERTILITY the went to such to Why Japanese lengths RATES FOR ALL WOMEN, ALL CUR- all of the classify approximately 240,000 RENTLY MARRIED WOMEN, AND ALL births to Taiwanese in 1941 of by age par- CURRENTLY UNMARRIED WOMEN, TAI- ents is not clear. never They published any WANESE, 1941. analysis of this study, and it has never been used by later researchers. What makes this study of considerable importance is the fact that when these data on births are com- bined with the census data of 1940, it is pos- sible to compute marital and non-marital age specific birth rates for single years of age.&dquo;I The 1940 census results provide the number of women at single years of age for each prefecture in October, 1940 (see Bur- eau of Accounting and Statistics, 1953 :40- 45). In addition, the 1940 census results give the marital status of women by five- year age categories for each prefecture; and a good estimate of the distribution of mari- tal status by single years, within these five- year age categories, can be obtained from the more extensive 1935 census (Govern- ment General of Taiwan, 1937). However, before these marital propor- tions could be applied to the number of women at each age, it was necessary to &dquo;age&dquo; the October, 1940 population about eight months so that all age specific rates would apply to the 1941 mid-year popula- tion of women (see Shryock and Siegel, 1973:472-485). Since single-year age spe- cific death data were available from the an- nual vital statistics reports, the actual

11The data on marital age specific fertility rates will be presented in a separate report. 305

cording to the estimates computed from the latter case, the bastard births were divided ~.935 and 1940 censuses. (The assumption by the number of unmarried women. It is was made that the death rates for women of important to note that because no age the same age but different marital statuses specific data on the distribution of concu- were the same.) The never-married, cur- bines’ births were available, these births rently divorced, and currently widowed were excluded from Table 8 and Figure 2.’z women in each age category were grouped together to form the currently unmarried 12As can be seen from Table 2, births to concu- category. bines comprised about one percent of all births in Table 8 shows the age specific fertility most years. If these births were added into the nu- rates for single years of age for the currently merators of the age specific fertility formulae, it is married and currently unmarried groups of logical to assume that marital age specific rates would increase little more than one While I women. In the former case, the rate was percent. suspect that births to concubines tended to be con- the total number of computed by dividing centrated at younger ages than births to all married legitimate births by the number of cur- women, these data are insufficient to test this rently married women at each age; in the hypothesis.

FIGURE 2. MARITAL AND NON-MARITAL AGE-SPECIFIC FERTILITY RATES, TAI- WANESE WOMEN, 1941. 306

FIGURE 3. NON-MARITAL AGE-SPECIFIC FERTILITY RATES, WEST COAST PRE- FECTURES AND WHOLE ISLAND, TAIWANESE POPULATION OF TAIWAN, 1941.

Table 8 and Figure 2 show that the age ages 30 to 34 years. I have no easy answer specific fertility curve for married women for why Hsinchu exhibits this deviant peaks at age 20 and declines steadily there- pattern. It is conceivable that it is related to after. While the non-marital fertility curve the large proportion of Hakkas in that pre- is at a much lower level throughout the fecture, but as evident in Tables 6 and 7, fecund years, it increases until about age the Hakkas do not appear to differ in their 24, declines slowly until age 32, and then overall levels of non-marital fertility. Since drops off more quickly. these 1941 data were not classified by The 1941 non-marital age specific fertil- ethnic background, it is not possible to pro- ity rate, like the 1935 Ih measure, shows a vide a conclusive answer to this question. fair amount of variation between prefec- While it is possible to specify the ages of tures (see Figure 3). While the level of il- mothers of illegitimate children and their legitimacy varies, there appears to be a gen- fertility rates, the available data do not al- eral similarity in the shape of the distribu- low us to determine whether these children tion across ages. The only major exception were born to never-married, divorced, or here appears to be Hsinchu Prefecture, widowed women. Any meaningful evalua- where non-marital fertility peaks late, at tion of the social significance of bastardy 107

in Taiwanese communities must be linked One possible argument about the preva- to the issue of the previous marital status lence of bastardy in Taiwan is its high inci- of these mothers.’3 dence among prostitutes, but its relative From his examination and reconstruc- infrequence among the general population tion of the household registers from Hai- of women. The Japanese colonial authori- shan District in the southern Taipei pre- ties collected information on the occupa- fecture, Arthur Wolf has concluded: tion of women as a part of the census in- vestigations, and also listed the occupa- an and a Despite early marrying age puritani- tional status of the mothers of cal attitude towards premarital sexual rela- illegitimate children from the 1920s until 1942. tions in our six districts, 12.3 percent of all early women raised there bore at least one illegiti- Yet neither set of statistics specifies prosti- mate child [before they marrieda. But this tutes. Further, Barclay says that the stan- does not mean that most young women experi- dards of classification used in assigning mented with sex before marriage. Since the births to occupational groups differed majority of those who bore illegitimate chil- from those used in the census. 14 dren were the daughters (or adopted daugh- ters) of people who were dependent on them If we are willing, however, to approach because they had no sons, most of these chil- this question of the marital status of dren were probably the fruits of prostitution mothers of bastards with another defini- (Wolf, 1975:102). tion, it may be possible to shed some light on the question. Suppose the Confucian 13Let me stress that what I am to here referring ethics did guide the behavior of most are the rates of marital or non-marital fertility (i.e. women, but that there was an &dquo;under- per 1,000 women in the category), not to the actual frequency of occurrence. The frequency distribution class&dquo; of servants and prostitutes who did of legitimate and illegitimate births by age of mother not follow these norms, even though today in 1941 was as follows: the use of earnings from prostitution to FREQUENCY DISTRIBUTION OF LEGITIMATE support one’s aged parents is still recog- AND ILLEGITIMATE BIRTHS BY AGE OF nized as a form of filial piety in Taiwan. MOTHER, 15-49 YEARS, IN FIVE-YEAR INTER- Further, let us suppose that none of VALS, TAIWAN, 1941. these ur:derclass women were either cur- rently married or widowed. This ieaves us with two models: 1) all widows and divor- cees are chaste, but never-married women are at risk of illegitimate births; and 2) all widows are chaste, but divorcees and never- married women are at risk of illegitimate

14See Barclay, 1954a:241. Throughout his work, Barclay rejects the use of the colonial data on the em- ployment of women as a reliable source of informa- tion 1954a:73-101). Some Note: As before, births to concubines, which (Barclay, anthropological evidence, however, suggests that the colonial data on totaled 4,360, were not included in this table. employment of women may be more accurate than Thus if we look at the actual frequencies, we find that that collected more recently (see Barrett, 1978a:390, illegitimate births are common to women in the 15-19 fn). The many available data on historical trends in year age group (23.43 percent of all illegitimate births employment in Taiwan have received only a very as opposed to 9.79 percent of all legitimate births). superficial treatment by labor economists and soci- Nevertheless, the large proportion of unmarried ologists, and the linkages between employment and women in that age group leads to low rates of non- marital or non-marital fertility have not been ex- marital fertility. plored in any systematic fashion. 308

births. Either model could be a reasonable What does this exercise prove? An ines- approximation of the conditions of an capable conclusion is that at least some of idealized Confucian society. the illegitimate fertility of unmarried From the 1935 and 1940 census data women was due to non-marital births to described above, good estimates of the widows, particularly at age thirty and over. 1941 mid-year populations of women at There are simply too few never-married or different marital statuses and five-year age currently divorced women at those ages to categories can be obtained. Since we know account for the relatively large number of the number of births to mothers of various illegitimate births. ages in 1941, it is possible to specify the This finding, however, does not exclude age specific fertility rates which would be the possibility that most of the illegitimate implied by each of these models (see Table births were to prostitutes. It is possible 9 ). that some widows drifted into prostitution. One way of evaluating the plausibility of But the usual route of recruitment into each of these models is to compare the age prostitution at this time was the appren- specific fertility rates which each would ticing or outright sale of unmarried generate with the actual marital and non- by their parents or others having legal marital age specific fertility rates pre- authority over them. While few data exist sented earlier (non-marital age specific on this topic, I would guess that the pro- rates would include widows as well as portion of older widows engaged in prosti- divorcees and never-married women). As a tution was probably quite small. general rule, one would not expect that It is doubtful if future research can either prostitutes or those involved in squeeze out much more information on the casual non-marital liaisons would have a marital background of the parents of bas- higher birth rate than married women in a tards from the vital statistics and census traditional society like Taiwan. However, data. The evidence does suggest sizeable one can see that in both models the age numbers of divorcees and widows bearing specific fertility rates of older unmarried illegitimate children; the age distribution women actually exceed those of married of mothers suggests that these post-marital women. The birth rates at older ages im- or inter-marital births were not necessarily plied by both models are particularly un- the fruits of prostitutes. Fresh insights believable. await researchers willing to devote the time

TABLE 9. TWO HYPOTHETICAL NON-MARITAL AGE-SPECIFIC FERTILITY RATES, AND THE ACTUAL NON-MARITAL AND MARITAL AGE-SPECIFIC FERTILITY RATES, TAIWAN, 1941. 309

FIGURE 4. COALE’S INDEX OF NON-MARITAL FERTILITY IN TAIWAN AND IN SE- LECTED EUROPEAN COUNTRIES, 1840-1960.

Note: Reliable data on illegitimacy in Taiwan are not available prior to 1906 or after 1942. This figure is adapted from the similar figure in Shorter, Knodel and van de Walle, 1971:391, and the European data are reprinted with the permission of the authors.

and effort to reconstruct the original Taiwanese Bastardy in International household registers from various localities. Perspective Of course, researchers who wish to gener- alize from the results of local area studies Figure 4 shows the levels of illegitimacy should be cautioned by the considerable measured by Coale’s Index of Non-Marital variation in the rates and ratios of illegiti- Fertility, the Ih, in a variety of European macy between regions. countries between 1840 and 1960 and in 310

Taiwan between 1906 and 1940. Even at mate births in Taiwan; these data may the first date for which reliable data are prove to be a useful complement to the ex- available, Taiwan’s Ih was only exceeded tensive published information on total by those for Sweden, Denmark, Germany, and marital fertility. ’5 Austria, Hungary and Portugal. By 1940, Why would the rate of illegitimate births Taiwan’s Ih was higher than that of any of have declined from 1940 to 1978? While these European nations, a trend partially Taiwanese tend to marry at older ages explained by the general decline in illegiti- today, the wide availability of contracep- macy in Europe which accompanied the tives and relatively cheap abortion makes it demographic transition. But the relatively much easier for mothers to prevent or sharp rise in illegitimacy measured by the terminate unwanted pregnancies. Both Ih in Taiwan also contributed to this con- practices were illegal and unobtainable for trast. the vast majority of women during the It would be advisable to have further Japanese colonial period.’6I data on Ch’ing Dynasty (before 1895) or It is possible that at least a small part of modern (after 1945) trends in illegitimacy the rapid decline in fertility over the past before making broad generalizations twenty-odd years is due to a decline in non- about the significance of the Japanese marital fertility, as well as a decline in colonial data. Yet on the basis of the avail- marital rates. In this sense modernization able historical evidence, there is little rea- may actually have led to a decline in il- son to believe that illegitimacy was signifi- legitimacy, a pattern also known in Eur- cantly lower during the late Ch’ing Dy- opean countries during the twentieth cen- nasty period than it was during the early tury. years of the Japanese household registra- Of course, here I am assuming that fer- tion system. Further, the rise in illegiti- tility among unmarried women in Taiwan macy during the colonial period does not really has declined, and this conclusion is appear to have been directly associated based on much less extensive evidence than with any &dquo;modernization&dquo; of social condi- is available for colonial Taiwan. The tions during that time. As Barclay and modern patterns of bastardy in Taiwan numerous other observers have stressed, may well show as much variation as those the economic development of Taiwan-or examined here. more accurately, its agricultural develop- ment-was not accompanied by much sig- nificant &dquo;social development&dquo; in marriage 15Their publication may also allow the Ministry of the Interior to correct one facet of their or family patterns (see Barclay, 1954a; misleading annual Taiwan Factbook. to Kerr, and Chen, Demographic Up now, 1974; 1955a, 1955b). the "Total Fertility Rate for Currently Married The lack of on any readily available data Women" has used all births on the island as the nu- the modern period precluded the analysis merator, rather than restricting it to births to cur- of current trends in illegitimacy in this rently married women. The use of the latter category in the of this rate is recommended article. However, some scattered data computation by most authorities. from different leads me to speculate that at present about 1.5 per- cent of all births are illegitimate, com- 16The data on the widespread availability and pared with about 4 percent during much of use of contraception and abortion in modern Taiwan are available in a number of for a useful sum- the Japanese I have been studies; period. recently mary, see Freedman et al., 1974. For data on the informed of that the Ministry the Interior legal status of abortion in the Japanese and post-war may soon publish information on illegiti- periods, see Cernada, 1975. 311

BIBLIOGRAPHY 1955b "Population Change in Taiwan." Bulletin of the Department of Archaeology and An- thropology, National Taiwan University 6:86-119. Cohen, Myron L. 1976 House United, House Divided. New York: Barclay, George W. Columbia University Press. 1954a Colonial Development and Population in Coale, Ansley J. Taiwan. Princeton: Princeton University 1969 "The Decline in Fertility in Europe from the Press. French Revolution to World War II." In S. 1954b A Report on Taiwan’s Population. Prince- J. Behrman, Leslie Corsa, Jr., and Ronald ton : Office of Population Research, Prince- Freedman, eds. Fertility and Family Plan- ton University. ning : A World View. Ann Arbor: University Barclay, George W., Ansley J. Coale, Michael A. of Michigan Press. Stoto, and James Trussell Freedman, Maurice 1976 "A Reassessment of the Demography of 1966 Chinese Lineage and Society: Fukien and Traditional Rural China." Population In- Kwangtung. London: Athlone Press. dex 42:606-635. Freedman, Maurice, ed. Barrett, Richard E. 1970 Family and Kinship in Chinese Society. 1978a "Differential Fertility in Rural Taiwan, Stanford: Stanford University Press. 1905-1940." Unpublished Doctoral Disser- Freedman, Ronald, Lolagene C. Coombs, Ming- tation, Department of Sociology, University cheng Chang, and Te-hsiung Sun of Michigan. 1974 "Trends in Fertility, Family Size Prefer- 1978b "Divorce, Widowhood and Remarriage in ences and Practice of Family Planning: Tai- Colonial Taiwan: Some New Evidence on wan, 1965-1973." Studies in Family Plan- the Status of Women in a ’Traditional’ Chi- ning 5, 9:270-288. nese Society." Mimeographed. Freedman, Ronald, and John Casterline Behrman, S. J., Leslie Corsa, Jr., and Ronald Freed- 1979 "The Effect of Nuptiality on Fertility in Tai- man, eds. wan." Paper Presented at the IUSSP Semi- 1969 Fertility and Family Planning: A World nar on Nuptiality and Fertility, Bruges, Bel- View. Ann Arbor: University of Michigan gium, Jan. 8-11, 1979. Press. Gallin, Bernard Bureau of Accounting and Statistics (Provincial Gov- 1966 Hsin Hsing, Taiwan: A Chinese Village in ernment of Taiwan) Change. Berkeley: University of California 1946 Taiwan Province: Statistical Summary of Press. the Past 51 Years. Taipei: Bureau of Ac- Government General of Taiwan counting and Statistics, Provincial Govern- 1937 Census of 1935: Statistical Tables. Taihoku ment of Taiwan. (in Chinese) [Taipei]: Government General of Taiwan. 1953 Results of the Seventh Population Census of (in Japanese) Taiwan (1940). Taipei: Bureau of Account- 1942 Vital Statistics of Taiwan. Taihoku [Tai- ing and Statistics, Provincial Government pei] : Government General of Taiwan. (in of Taiwan. Japanese) Cernada, George P. 1943 Statistics on the Ages of Fathers and 1975 "Basic Beliefs About Human Life Relating Mothers, for Live Births and Stillbirths, to Ethical Judgements Family Planning 1941. Taihoku [Taipei]: Government Gen- Workers Make About Induced Abortion: eral of Taiwan. (in Japanese) Taiwan, 1973." Unpublished Doctoral Dis- Henry, Louis sertation, Population Planning Depart- 1960 "Fécondité légitime de paysans chinois de ment, University of California at Berkeley. Formose." Population 15:551-554. Chen, Ching-chih Hsu, Francis L. K. 1975 "The Japanese Adaptation of the Pao-Chia 1971 Under the Ancestor’s Shadow. Stanford: System in Taiwan, 1895-1945." Journal of Stanford University Press. Asian Studies 32:391-417. Kerr, George H. Chen, Shao-hsing 1974 Formosa: Licensed Revolution and the 1955a "Population Growth and Social Change in Home Rule Movement. Honolulu: Univer- Taiwan." Bulletin of the Department of sity of Hawaii Press. Archaeology and Anthropology, National Knodel, John, and Steven Hochstadt Taiwan University 5:76-103. 1976 Illegitimacy in Imperial Germany. Ann Ar- 312

bor, Michigan: Center for Western Euro- Shryock, Henry S., and Jacob S. Siegel pean Studies, University of Michigan. 1973 The Methods and Materials of Demogra- Laslett, Peter phy. Washington, D.C.: U.S. Government 1977 Family Life and Illicit Love in Earlier Gen- Printing Office. erations. Cambridge: Cambridge University Taeuber, Irene Press. 1961 "Population Growth in a Chinese Micro- Lee, Ronald D., ed. cosm : Taiwan." Population Index 27:101- 1977 Population Patterns in the Past. New York: 126. Academic Press. Tilly, Charles, ed. Levy, Marion J., Jr. 1978 Historical Studies of Changing Fertility. 1968 The Family Revolution in Modern China. Princeton, New Jersey: Princeton University New York: Atheneum. Press. Okamatsu, Santaro Tuan, Chi-hsien 1971 Provisional Report on Investigation of Laws 1958 "Reproductive Histories of Chinese Women and Customs in the Island of Formosa. Tai- in Rural Taiwan." Population Studies 12: pei : Cheng Wen Publishing Co. [Reprint of 40-50. 1902 English-language edition published in Wolf, Arthur P. Kobe, Japan at the "Kobe Herald" Office] 1975 "The Women of Hai-shan: A Demographic Saso, Nariko, and Michael Saso Portrait." In Margery Wolf and Roxane 1975 "A Study of Hsinchu Koseki Registers." Witke, eds. Women in Chinese Society, pp. National Taiwan University Journal of Soci- 89-110. Stanford: Stanford University Press. ology 11:61-75. Wolf, Arthur P. and Chieh-shan Huang Shorter, Edward, John Knodel, and Etienne van de 1980 Marriage and Adoption in China, 1845- Walle 1945. Stanford: Stanford University Press. 1971 "The Decline of Non-Marital Fertility in Yang, Martin Europe." Population Studies 25:375-393. 1968 A Chinese Village. New York: Columbia University Press.