A Longitudinal Historical Population Database in Asia. Taiwanese Historical Household Registers Database (1906-1945)

e-ISSN: 2352-6343 PID article: https://hdl.handle.net/10622/23526343-2020-0011 The article can be downloaded from here. © 2020, Lin, Chen, Chuang, Yang, Wilkerson, Hsieh, Yap, Huang This open-access work is licensed under a Creative Commons Attribution 4.0 International License, which permits use, reproduction & distribution in any medium for non-commercial purposes, provided the original author(s) and source are given credit. See http://creativecommons.org/licenses/.


A Longitudinal Historical Population Database in Asia
For the past 35 years, the Taiwan Historical Household Registers Database (THHRD) has been significant for historical demographic research on Asia. In recent years, researchers have continued adding new demographic information to the database. This allows for the expansion of research on the topic of historical households in the region. However, there are still many issues to address in the field of Asian historical demography. This paper provides a brief introduction on the uses of THHRD for future research.
Stanford anthropologist Arthur P. Wolf was the first scholar to recognize the academic value of the Taiwanese household registers. He utilized them to explore the marriage and adoption customs of the Han Chinese (Wolf, 1968;Wolf & Huang, 1980). Thanks to Professor Wolf we have some unpublished manuscripts about the Taiwan Historical Household Registers Database, 1906-1945, maintained by the Program for Historical Demography (PHD) at the Academia Sinica, Taipei, Taiwan. This paper is based on Professor Wolf's 2009 manuscript, as well as recent developments and achievements.
In 1985, Professor Arthur P. Wolf began to cooperate with Ying-chang Chuang, Research Fellow of the Institute of Ethnology, Academia Sinica. Together they collected the Taiwanese household registers 1 compiled during the Japanese colonial era. They then digitized the information in these registers in order to construct a historical demographic database. In 2003, the Program of Historical Demography 2 was officially launched at the Research Center for Humanities and Social Science (RCHSS), Academia Sinica, to continue the project of digitizing the data of the Taiwanese household registers.
Taiwan was ruled by Japan from 1895 to 1945 after China was defeated in the Sino-Japanese War of 1894. In order to tightly monitor and control this newly seized island, the Japanese colonial government soon implemented a carefully designed household registration system ( Figure 1). The language of the household registers is Japanese. It covered great details of the social life of individuals and families at that time and serves as a precious legacy left behind by the colonial authorities. 'One can recover a large part of the life history of every person alive in the period 1905-1945, and reconstruct the exact composition of each family he or she joined.' (Wolf, forthcoming). These detailed records allow for a variety of studies on demographic events from a longitudinal perspective and can make a contribution to many academic disciplines.

Figure 1
A household register of Colonial Taiwan Source: The Program for Historical Demography (2020).
Note: Due to privacy protection regulations, the names in Figure 1 are censored. This paper first briefly introduces the historical institutions and contents of the Taiwanese household registers from the Japanese colonial period. Then, it offers an overview of the THHRD and its limitations. Finally, it reviews studies that made use of the THHRD and discusses the potential value of the THHRD for research.
After taking over Taiwan in 1895, the Japanese soon issued an order pertaining to the implementation of a household registration system and its related regulations in the following year (Liao, 2010). The task of conducting a household survey and registering every household and its members were carried out by the military police. The registers compiled by the military police were no longer used, and were replaced by household registers of a new format after new registration regulations were promulgated on December 26, 1905. The new registers were officially established on January 15, 1906.
Instead of the military police, the new law made the municipal police and the so-called Baochia 保甲 in charge of the administration of household registration. The word Baochia denotes the head of Bao and the head of Chia. One Chia consisted of ten households; one Bao consisted of ten Chias (Lin, 2011;Wolf & Huang, 1980). Individuals had to report changes in their situation, including births, deaths, marriages and changes of address, to the BaoChia 保甲, which were the heads of the local HoKou 戶口 ('household') networks, within ten days, and then the police paid a home visit to verify the report (Engelen & Hsieh, 2007;Katz & Chiu, 2006). The Baochia functioned as a Chinese community policing system (Katz & Chiu, 2006).
To establish the initial set of the new household registers, the Japanese police reexamined the information from the defunct household registers. They interviewed family members and consulted private sources such as ancestral tablets and clan genealogies when the information appeared insufficient or missing in the records (Wolf, forthcoming). The household registration system required everyone to be registered in one household. The police also made regular door-to-door checks and irregular household surveys to make sure the accuracy of the information in the registers (Hong, 2013).
The meaning of the columns in the household registers are shown in Figure 2. If a household consisted of more than five persons, there were more pages for their household. In the Chinese household formation system, it is possible to have more than one family in a household/a house address. There were three types of household registers in colonial Taiwan: the active register, the inactive register, and finally the sojourner register. A register was a set of sheets. At the beginning of a household, the household register belonged to the active register file, when the household closed, the household registers would be moved to the inactive register file. Sojourners were the transient residents in the household. They were not part of the family and sojourned in the household because of work or some other reason. The three types of household registers shared the same registration form (Lin, 2011).
To ensure that the data in the Taiwanese household registers were accurate, they were updated regularly. Those individuals that were distrusted (vagrants and criminals) had to report to the registry office every month. For people belonging to higher social classes, this obligation was extended to every three or six months (Wolf & Huang, 1980).
The household register should be read from right to left and from top to bottom. The first item, column I, is filled in for every household register and includes the address of the household and the date and the reason for its establishment (Lin, 2011). The other columns, II to VI in Figure 2, record personal events, including the name of the other persons involved, the date when the event took place, the related address and the type of event. Sometimes a member of the household was struck out from the record, in such circumstances the individual's departure would be recorded in the personal event column (Wolf & Huang, 1980). In the case of more information, small pieces of paper would be pasted on top of it to provide additional writing space to record the new event (Lin, 2011). Each of the columns II-VI is assigned to one member of the household, except the column II which is always reserved to the household head.
Beneath each event column are six square boxes (box 2-7) for other personal information. The boxes are classified as 'ethnicity', 'opium addiction', 'bound feet', 'classification' (for criminal records), 'disabilities', and 'vaccination' (Wolf & Huang, 1980). In the space of 'ethnicity', people were distinguished in the light of ancestral origins, native language as well as biological affiliation (Hong, 2013;Wolf, forthcoming). The space of 'classification' (Zhongbie 種別) notes the rating of a person given by the police on a three-point scale. The police classified people into three groups according to their social status and behavior. There were three categories: class 1(rich citizens and outstanding citizens); class 2 (general citizens); class 3 (pauper, criminals, entertainers and people with 'bad reputations') (Lin, 2011;Wolf, forthcoming).
The record in the opium addiction box was like a drug license. If an individual had 'A' 阿 (a contracted word of opium) written in this box, it means that he was allowed to smoke opium. Although opium abuse was harmful to an individual's health, it brought in substantial tax revenue for the Japanese government. The Japanese government therefore allowed the Taiwanese to smoke opium by buying it at official shops (Lin, 2011).
The vaccination box indicated whether an individual had received a smallpox vaccination. The box concerning 'disabilities' stated whether the person was blind, deaf or dumb. As for the record of bound feet, there were three types: 'not bound' for women who never had bound feet; 'bound' for those who had; and 'unbound' for women who had had bound feet but released them, probably because foot binding was forbidden by law in 1915. The practice of documenting criminal records (classification) was abolished in the late colonial period and the original records were covered with black ink (Lin, 2011).
As said, the horizontal rectangular box below the six square boxes indicates the relationship of each individual to the head of the household. The listing of the members was prioritized according to their relationship to the head of the household when the household register was first established. This rule was broken when somebody was married or moved into the household. In any circumstances, the first person in the record always was the head of the household (Lin, 2011).
The information collected in an individual member column (in the boxes 8-13) contains the name of a household member, the person's birthdate, his/her parents' names, his/her same-sex sibling order, and his/her relationship to the household head. The occupation of a household head and in some cases, other household members, when they had occupations which differed from the occupation of the household heads.
In 1935, the regulation of the Taiwanese household registration system was altered (Engelen & Hsieh, 2007). The information about ethnicity, police rating, physical deformities, smallpox immunization, opium smoking and foot-binding were no longer required in household registers according to the amendment (Hsu, 2014;Lin, 2011). With an exception for the occupation of a household head and in some cases other household members when they had occupations which differed from the occupation of the household heads. The sojourner register was set up only after 1935 (Chiu, 2003). Before 1935, there was no specific sojourner register book, and the sojourners were recorded together with household members in the active register. As a consequence, some sojourners have two similar records in the household registration system: one in the sojourner's original household, the other in the temporary household.
When a household member died or exited because of adoption, marriage or household division, his name was deleted from the register. When a new member arrived by means of birth, marriage or adoption, he was added to the next open column in the register. When a household head died or retired, information concerning all household members was copied onto new forms, and the old register was assigned to a file containing all the registers retired that year. When a family 3 moved from one registration district to another, all information relating to its current members was copied onto fresh forms, which were sent to the registry office responsible for the household's new dwelling place. The old register was then cancelled and filed as closed registers. However, when a family moved to another village or neighborhood within the same registration district, the relevant sheets of the register were simply passed from one registry office to another without a note 'the household was quitting' in the records held by the registry office of the community (Wolf, forthcoming).
Use of the Japanese registers continued through most of 1947, despite the change of government at the end of World War II. However, the THHRD did not include any of the data recorded in 1946 and 1947 since the change of administration may have affected the reliability of such sources (Wolf, forthcoming). , 1906, -1945. These two databases are based on the same household registers, but include different locations and are based on different computing systems. The database of Stanford archive only consists of household registers from the Hai-san area where Arthur Wolf did his field research for many years. For more information about the Stanford Archive, see Wolf (forthcoming). In this article, we mainly introduce THHRD which is a longitudinal historical demographic database.

Arthur Wolf's book (forthcoming) introduces two kinds of databases of Taiwanese household registers: the Stanford Archive and the Taiwan Historical Household Registers Database
The THHRD database covers the period of 1906-1945. In 1906, the household registration system, as carefully designed in the household registration regulations of 1905, was put into practice. It lasted until the nationalist party took over Taiwan from the Japanese colonial government in 1945. The time span of 40 years contains up to 4 generations (Dong, Campbell, Kurosu, Yang, & Lee, 2015). By 2019 there were 369,373 individuals, and 55,810 households in THHRD. The database was constructed from various contributions. We are grateful to the collectors (see Table 1) who kindly provided their collections of household registers with the researchers. 3 In traditional Chinese society, a household could include more than one family. A family consists of at least two generations.  Table 1 and Figure 3). They include eight sites in Northern Taiwan, four in Central Taiwan and four in Southern Taiwan. Among these sites, some are in urban areas, and others are in rural areas. Therefore, the THHRD allows for an observation of urban and rural differences. The mentioned program for historical demography only provides open access to the data of three research sites: Chu-pei, Pei-p'u and E-mei. As for the rest of research sites, please first contact with contributors and the program for historical demography.

Figure 3
The distribution of research sites in THHRD (2019) In the THHRD, following the design of Professor Wolf and Professor Chuang, there are 9 different tables (see Table 2) to transcribe the enormous and complex details from the household registers. The history of a household could be recorded in more than one register if the household ever experienced the death or retirement of a household head, or if one of its members ever departed to create a new household. As for an individual, he/she could also be observed in different registers in any of the previously mentioned cases as well as if he/she ever departed from his/her original household through adoption, marriage or other means.
Through linkage, household and individual histories can be reconstructed and various research tasks can be conducted. For instance, with the help of the recorded relationship of each individual to his household head, it is possible to identify the relationship between any two individuals within one household and thus to reconstruct the exact structure of a household (Wolf, forthcoming). By linking the Personal Stat Table, the Occupation Table and so on, Chia-chi Lin (2011) has published a book on female-headed households in Eurasian societies. There is always a limitation in everything. As for using THHRD, users have to keep the following things in mind. First, events occurring before 1906 were usually recorded a long period after they actually happened and were mainly based on a reporter's recollection. In addition, as mentioned before, the dates of these events might have been reported according to the lunar calendar instead of the solar calendar used by the Japanese government. The lunar calendar was used in the pre-colonial farm society. This changed under the Japanese governance. So, one has to consider the accuracy of the dating of the earlier events (Hsu, 2014;Wolf, forthcoming).

CONTENTS
Second, the way by which 'persons in temporary stay' (Jiliouren 寄留人) are dealt with in the database deserves some attention. Because the types of 'person in temporary stay' are of great diversity, it is hard to handle them with a computer program. Therefore, they are excluded from the Personal Dynamic Table and the Personal Location Table, while they are still included in the Personal Stat Table. The reason to include them in the Personal Stat Table is that some people in temporary stay could have blood ties with members of the household in which they were registered.
Third, information about ethnicity, police rating, physical deformities, smallpox immunization, opium smoking and foot-binding was no longer recorded in the registers after 1935.
Fourth, the household registers did not record everyone's occupations. Only 10% of the recorded occupations were from other persons in the household other than that of the head of the household. However, these records did not include the dates when the occupations were valid, but they did note changes of occupation.
Fifth, the THHRD database is based on regions. Once individuals moved out of the selected regions, there was no information in the database anymore. If individuals arrived from other regions, only the reason of moving and the address of the former household from the foregoing period were recorded.
The THHRD has been applied to various research topics intending to understand the family composition and social structure of Taiwan in the early 20th century. The topics include adoption, ethnicity, fertility, gender differences, marriage, mortality and the related life course researches. Chuang and Wolf (1995) claim that different marriage forms have different issues of concern. Moreover, in 1996, on the cooperation of Professors Arthur P. Wolf, Ying-chang Chuang and Theo Engelen of Radboud University, the project of the 'Population and Society in Taiwan and the Netherlands' was launched. The THHRD, together with the database of the Historical Sample of the Netherlands (HSN), was utilized to make Eurasian comparative studies, and a series, Life at Extremes, produced four books from the project. The first book, Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis, reviewed the contribution of Hajnal's hypothesis to historical demography (Engelen & Wolf, 2005).
The second book, Positive or Preventive? Reproduction in Taiwan and the Netherlands, 1850-1940(Chuang, Engelen, & Wolf, 2006, discussed the Malthus hypothesis proved by the results of the Eurasia historical population database. Moreover, Two cities, One Life. Marriage and Fertility in Lugang and Nijmegen provided a case study furthering discussion on Eurasia difference (Engelen & Hsieh, 2007   Taiwan and the Netherlands 1850-1945 gave information on mortality topics. Another important element in Malthus' hypothesis (Engelen, Shephard, & Yang, 2011 Professor Ko-hua Yap published a paper about foot-binding studies. In general, the Hoklo people preferred bound feet while the Hakka people preferred natural feet. He found that Hoklo female with cross-ethnic family background were more likely to get rid of foot binding, according to the household registers. Furthermore, the Hoklo women without cross-ethnic family background but close to the Hakka areas also had unbound feet as compared to their counterparts in other Hoklo areas. Professor Ko-hua Yap explained the above phenomenon as 'arms race'. That is, the more popular foot binding was in the surroundings, the more likely parents bound their daughters' feet, to avoid having a disadvantage in the marriage market. Conversely, the existence of people who did not desire bound feet in the surroundings, drastically reduced the pressure of pursuing foot binding (Yap, 2017).
Professor James Wilkerson (2010) published a chapter in a Ruizhi Lian and Ying-chang Chuang edited volume on late Qing dynasty era literati marriage. The literati status in Hsinchu at that time was a legal status for men who successfully passed one or more of the various levels of the Qing imperial exams. Both before and after marriage, women in literati families were under constant surveillance through standard Chinese style sub-bureaucracies (BaoChia) to impose a variety of prescriptive standards for female 'domestic decorum' (Fudao). When these literati are compared with non-literati, the literati had extremely low rates of violations of female domestic decorum in comparison with the high rates of violations of female domestic decorum as reported and discussed by Arthur P. Wolf and others for Hsinchu and elsewhere in Taiwan. This chapter is a series of publications discussing marriage and subbureaucracies in Taiwan and China.
In 2015, Professor Wen-shan Yang further joined the East Asian Population and Family History Project (EAP 2), therefore the THHRD was included in EAP 2. This project focuses on neighboring populations in East Asia that are more similar in terms of background and context. Instead of comparing European and Asian populations, it aims to point out the similarities and differences in East Asian population behavior through comparative analysis of 5 population register databases from East Asia (Dong, Campbell, Kurosu, Yang, & Lee, 2015).
The Taiwan Historical Household Registers Database has been known as a treasure for historical demography of Asian study. It has been established for more than 30 years and keeps growing. It has provided fruitful and successful research results. With the advancement of statistical methods and computer science, we believe there are still many issues that researchers might study by using the THHRD. This paper is a brief introduction for those who are interested in the Taiwan historical household registers database, and it is also an invitation to join us!
The source data from the Taiwan household registers used in this paper are collected by Arthur Wolf, Hill Gates, Ying-chang Chuang, Wen-shan Yang, Guang-hong Yu, Ing-hai Pan, Hsiang-shui Chen, Paul Katz, James Wilkerson, and Chia-chi Lin. Those registers were digitized as 'Taiwan Historical Household Registers Database, 1906-1945' by the Program of Historical Demography,