Historical Demography Zhongwei Zhao

2.3. The Historical Investigation of Fertility Transition in Europe
At about the same time as the Cambridge Group began its investigations into English population and social history, another major effort made in historical demography by scholars in Europe and North America was also launched under the title, the European Fertility Project. This project, organized by the Office of Population Research at Princeton University, had two principal objectives: to provide detailed quantitative information about the fertility experience of over six hundred provinces of Europe during the period of their demographic transition; and to identify the social and economic conditions that prevailed at the time when the modern reduction in fertility started (Coale and Treadway 1986).
Researchers involved in this project collected large amounts of fertility data for more than six hundred European provinces, recorded over a period of more than one hundred years (data collected for European aristocracies covered a much longer period). A set of fertility indices (I_f, I_g, I_h and I_m), which will be discussed later, was developed and calculated at ten-year intervals for each of the provinces included in the study. Analysis of these data showed that, while noticeable fertility fluctuations (including reductions) were observed in pre-transition societies such as England in the mid-seventeenth century, sustained fertility decline in most European countries other than France only took place in the late nineteenth century. This decline was far more complicated than that described by classical demographic transition theory. It was first observed in Northwest Europe, followed by peripheral areas in Southern and Eastern Europe. If France was excluded, more than half of the six hundred European provinces began their long-term fertility decline between 1890 and 1920. While there were provincial leaders and laggards in this transition, the differences between the onsets of their fertility decline were not very large. By 1930, most provinces had already witnessed a more than 50 per cent fertility reduction from their pre-transition level (Watkins 1986).

The European Fertility Project found that in pre-transition societies fertility fluctuations were often related to changes in marriage patterns, whilst changes in fertility behaviour within marriage were a major factor explaining the secular fertility decline during the demographic transition. The project also showed that in most pre-transition populations, fertility levels of married, or of all, women were considerably lower than among the Hutterites, a religious group who lived in the United States and Canada and was widely credited with the highest recorded fertility. Differing from the stylised description offered by classical demographic transition theory, fertility levels and patterns varied greatly across regions in pre-transition Europe. Fertility decline tended to start earlier in populations who lived in urban areas, had better education or experienced a lower infant mortality than in populations living in rural areas, with less education or experiencing a relatively high infant mortality; but the relationship between fertility decline and industrialization (as measured by proportion of people engaged in agriculture), levels of literacy and infant mortality was not strong. The socio-economic conditions under which fertility began to decline in these European provinces were also remarkably different. Interestingly, while it is difficult to, on the basis of economic conditions, determine the threshold of sustained fertility decline, similar fertility changes were often recorded in populations sharing common cultural tradition, language or religion. Although the evidence uncovered by the European Fertility Project did not offer a strong support for the classical demographic transition theory, it significantly enriched our knowledge about the demographic transition in Europe by providing great insights about pre-transition fertility regimes, and their profound changes during the late nineteenth century and first half of the twentieth century.

2.4. Historical Demography in East Asia
Historical demography has also made remarkable progress in East Asia in recent decades. In Japan, A. Hayami, also strongly influenced by Henry’s family reconstitution work, started his research in historical demography in the late 1960s. Since then, an increasing number of studies has been carried out by Japanese scholars, shedding new light on Japanese population history. According to available evidence, population registration had already existed in Japan in the eighth century. Records surviving from this period are often very fragmented, however, and do not provide enough information on which to base systematic demographic analysis. Most published studies are conducted using demographic data recorded since the seventeenth century. The primary data sources for Japanese historical demography are population registers (Ninbetsu Aratame Cho) and religious faith investigation registers (Shumon Aratame Cho), although other data sources such as death registers compiled by Buddhist temples, are also used.
On the basis of their examination of these historical data, Japanese scholars considerably revised previous population estimates for the early Tokugawa period. They have shown that, although the national population growth stagnated in the late Tokugawa period, there were marked regional variations in population growth. These variations were closely related to major disasters taking please during this time, migration to the cities, and the relatively high mortality recorded in urban areas. Their analysis of surviving population registers has uncovered much detail about marriage, fertility, mortality, migration, and residential patterns in pre-modern Japan (Hayami 2001; Saito 1990; Tsuya and Kurosu 2004). Hayami’s studies, for example, have shown that there were large age differences between husbands and wives in some historical Japanese populations. Marital fertility among women born after 1700, those between 1751 and 1800 in particular, was moderate or low in comparison with that observed in many historical European populations. The age structure of deaths differed considerably between urban and rural areas. The studies have also revealed that work-related migration was common in some regions, where many people worked away from home (dekasegi). These population movements exerted a notable impact on women’s age at marriage, levels of fertility, and age patterns of death in urban and rural areas (Hayami 2001).
China has had a sophisticated population registration system since ancient times and still possesses a large collection of officially produced demographic statistics. These records, along with those made privately (for example family genealogies), provide valuable data sources for historical demography. Although studies of population history have been confined almost entirely to the examination of summary population figures calculated at national or provincial levels in Mainland China, great progress has been made in the analysis of demographic data recorded at the level of individuals in recent decades by scholars working in Taiwan and western countries.
T. Liu (1992) systematically analysed 50 lineage or family genealogies collected from 12 provinces in China. These genealogies record nearly 300,000 people belonging to more than 20 generations who lived mostly between the fifteenth and nineteenth centuries. Liu and her research team examined marriage, fertility, and mortality patterns in the lineages, their population growth, demographic constraints on household formation, social status of lineage members, and major socio-economic functions of lineages and families. The study offers a good example of using genealogical materials for demographic investigations, and provides remarkably detailed information on population changes among these lineage populations during China’s Ming-Qing periods (1368-1911).
Another major effort in pursuing Chinese historical demography is the work of J. Lee and his collaborators. Their studies are based primarily on two types of historical records. The first is the Qing imperial lineage genealogies, which are of high quality, especially in the early period. These data provide unique opportunities for the investigation of demographic behaviour in the population and for the study of fertility and infant mortality in the past. The second is the population registers of people living in various parts in China during the eighteenth and nineteenth centuries, which are census-type records updated (or collected) every three years. Lee and his collaborators have examined the demographic history of these populations. On the basis of their findings and those reported by other scholars, they have challenged the Malthusian view that historical China was ‘the prime example of a society dominated by the positive check and virtually devoid of any preventive check’. According to Lee and his collaborators, ‘Chinese demographic behiviour not only provides an alternative demographic model’ to the one proposed by Malthus; ‘it also reveals that many differences in population behaviour between East and West are a product of regional and historical differences in social organization rather than of different population checks’ (Lee and Wang 1999: 5 and 7; Lee and Campbell 1997).
Studies conducted by T. Telford, S. Harrell and Z. Zhao have also made contributions to the development of Chinese historical demography. Aside from using genealogical data and computer micro-simulation to study marriage, fertility, mortality, and potential residential patterns in the past, these researchers have systematically examined major under-registration problems found in genealogical records and the potential biases of using such data in the study of demographic changes (Harrell 1987 and 1995; Telford 1986 and 1990; Zhao 1994 and 2001). Their studies provided useful references for those who use genealogical records in their research.
One of the major research findings in Chinese historical demography is that marital fertility was, as in Japan, relatively low in Chinese history compared with that in many historical European populations. Chinese women married at relatively young ages, but their birth intervals (both between marriage and first birth and between successive births) were relatively long. They also stopped having children at relatively young ages. Some scholars have suggested that the relatively low marital fertility was an indication of deliberate control of fertility or family size (Lee and Wang 1999; Zhao 1997). These findings have important theoretical implications, and have stimulated considerable debate and much further research into these issues in recent years.
2.5. Historical Demography in Other Areas and Its Recent Development
In addition to the development discussed above, a large number of historical demographic investigations have been undertaken in Canada, Germany, Italy, Spain, Sweden, the Netherlands, Austria, the United States of America and many other countries in the last few decades. Similar studies have also been conducted in Latin America, South and Southeast Asia, although they tend to be relatively small in scale and carried out by fewer researchers in comparison with those in other areas. These studies are not detailed here because they are, in many respects such as methods, data sources, and questions being examined, similar to those discussed in previous sections. Nonetheless, two recent developments are worth further commenting.
The first development is observed in some European, especially north European, countries. Many of these countries possess historical data covering a long period, with good quality and much detailed information. For example, Sweden began compulsory parish registration in the second half of the seventeenth century, and since the mid-eighteenth century, records of causes of death and continuous instrument-based measurements of surface temperature have also been collected in some areas. Detailed demographic and non-demographic data of this kind provide researchers with a unique opportunity to examine many questions that have never been examined before. For example, historical demographers have, using such data, examined the effects of childhood experiences such as stress in early life on old age mortality (Bengtsson and Lindstrom 2000). The richness of these historical data also generates great interest in inter-disciplinary research among scholars with different academic backgrounds such demography, epidemiology, history, genetic and environmental sciences. One such example is a recent investigation of the impact of ambient temperature on the sex ratio at birth and male longevity, which has been conducted using historical demographic and meteorologic data collected from Demark, Finland, Norway, and Sweden (Catalano, Bruckner and Smith 2008).
Another important development in historical demography is the increase in international research collaboration, as demonstrated by the Eurasia Population and Family History Project, which was initialized by A. Hayami in 1993. Researchers involved in this project assembled rich historical data from five countries: Belgium, China, Italy, Japan and Sweden, which provided detailed social and demographic information for recorded individuals and their households and communities. Through their examination of these materials, the researchers compared population dynamics in the five countries during the Eighteenth and Nineteenth centuries. Their research findings have been reported in a large number of articles and several books and conference proceedings (Bengtsson, Campbell and Lee 2004). This project has a number of features that distinguish it from others. First, historical demographic data gathered from these populations have been analysed using similar and sophisticated statistical methods, which has gone beyond many previous studies. Second, demographic behaviour of individuals have been analysed in the context of their household structure, socio-economic profiles of their community and their social status and positions within the family, which has brought about considerable insights on how people’s demographic behaviour and its variations were influenced by cultural values and micro social structure. The research project provides a good example of international collaboration in advancing historical demography.
3. Major Data Sources for Historical Demography
Many countries possess rich historical demographic data, and they have become increasingly accessible in recent years thanks to the progress made in historical demography and the advancement in computer technologies. This section briefly describes some of the most important data sources used in the investigation of population history.
3.1. Parish Registers
Parish registers are records of baptisms, marriages, and burials made by ecclesiastical authorities. They have been the most widely used and the most rewarding historical data source in the study of European population history.
Keeping parish records already existed in certain Italian city states in the late medieval period. The practice spread first to western France and England, then to most of western and central Europe. By the eighteenth century, registering baptisms, marriages and burials was widely observed in many European countries and their overseas colonies (Willigan and Lynch 1982). In England, for example, parish registration was established in 1538 by Thomas Cromwell (1485-1540) who ordered the clergy to record all baptisms, marriages, and burials performed in the Christian church. During the next three centuries continuous efforts were made to maintain and improve such registration. This resulted in a huge collection of parish registers, many of which are still available today. The importance of parish registers as major records of vital events and the primary data source for historical demography was only replaced by the civil registration and modern censuses during the nineteenth century.
The level of detail of parish records varies considerably across region and over time. Some entries are very simple and include only the name of the person who was baptised, buried, or married, and the date of the event. Others give a more elaborate description of the individuals and families affected by these events. Thus, a baptism record may also contain the date and the place of birth, the names of the parents, and the occupation or the rank of the father. A marriage record may provide information on the ages of the bride and groom, their past marriage history, their parents’ names, occupation or rank, and the occupation of the groom. A burial entry may include the age, occupation or rank, normal place of residence of the deceased, and even the cause of death. In general, the extensiveness of information contained in parish registers tends to increase over time, and many detailed records were found in the eighteenth and early nineteenth centuries (Willigan and Lynch 1982).
While these records may at first glance not seem to be very useful for demographic analysis, the technique of family reconstitution can transform them into a comprehensive data set consisting of detailed information on all, or most, of the demographic events experienced by the recorded individuals. This gives parish registers a privileged position as the key data source for understanding the demographic history of many European countries and their overseas colonies from the sixteenth to the early nineteenth centuries.
The parish registers as a major data source for demographic studies have some limitations, however. As a type of religious records compiled by the Christian church, parish registers might cover only part of the whole population living in an area. People with other religious faiths were usually excluded from such registration. Because baptism, marriage, and burial registration recorded only people involved in such events and did not have information on migration (except data about those married into the parish), they cannot provide accurate information on the total population in a parish at any particular point in time. Moreover, these records do not provide information on residential patterns or co-residing family members, although the relationship between family members may, in many cases, be determined on the basis of these data.
While parish registers do not exist in non-Christian populations, records sharing a similar nature were also found in some countries. In historical Japan, for example Kakocho or Books of the Past, were compiled by many Buddhist Temples. Kakocho also listed people who had died sometimes including their age at death, causes of death, and the date of death, providing an important data source that has been used for the study of demographic change in the past (Hayami 2001; Jannetta and Preston 1991).
3.2. Population Registers and Census-type Materials
While often differing from modern censuses, census-like operations have long existed in many parts of the world. Governments have collected population data in order to obtain reliable information for military drafts, corvees, and taxation purpose. Governments or religious authorities have also gathered the required demographic data through maintaining and updating various population registers. Thanks to these efforts, a formidable amount of historical population data has been accumulated in many countries. In China, for example, aggregated population statistics computed at national, provincial and prefectural levels for the year 2 AD are still available. Population registers made during the Tang Dynasty (618-907 AD) have been discovered in Dunhuang. Surviving records of this kind, especially those made in the last few centuries, have increasingly been found and used in historical demography in recent years (Lee and Campbell 1997; Liao 2001).
Population registers and census-type materials are particularly abundant in East Asia. Hukou Ce made in the Qing Dynasty (1616-1911) and used by J. Lee and his collaborators (Lee and Campbell 1997), Ninbetsu Aratame Cho produced in Japan during the Tokugawa era (1603-1868) and used by many Japanese scholars (Hayami 2001), and Hojǒk compiled in Korea over the seventeenth and eighteenth centuries and used by K. Kim (2005) all belong to this type of records. Historical data of similar nature also exist in other parts of the world. For example, population registers made in the nineteenth century have been found in some parts Italy, Belgium and Sweden, and used in recent investigations of population changes in the past (Campbell 2004; Bengtsson, Campbell and Lee 2004).
These population registers and census-type materials were generally maintained for governmental administrative purposes such as tax collection, but they might also have religious purposes. Shumon Aratami Cho, for example, literally meaning ‘religious faith investigation registers’. These registers were widely used during Japan’s Tokugawa period to control the spread of Christianity. They recorded religious beliefs and detailed demographic information on household members. Similarly, catechetical examination registers, which are essentially a kind of population register, were produced and used by clergymen in Sweden to identify eligible parishioners whose knowledge of the Bible was to be examined (Campbell 2004). Listings of inhabitants were also compiled by clergymen in some parts of England (Laslett 1966).
Population registers and census-type materials generally record the population living in a particular area with a clearly defined boundary. Residents are normally listed by households and the relationship between household members and household heads is usually indicated. Some registers also include the age, sex, marital status, and occupation of household members, as well as other useful information such as land holding of the household. All these properties make the population registers and census-type materials particularly useful for historical study of micro-social structures. However, records of this type do not usually include information on number of births, deaths, and marriages, and therefore do not allow computing birth, death, and marriage rates; although these rates may be estimated if successive population registers or census data are available.

3.3. Family or Lineage Genealogies
Genealogy is the written record of family or lineage members, which recounts one’s descent from an ancestor or ancestors by enumeration of the intermediate persons. Constructing genealogies has a long history, and the practice is found in many societies, especially those in a new frontier where a large number of migrants settled and those where lineage organizations formed an important social institution and the worship of ancestors was widely encouraged. Genealogies vary greatly in their format and levels of detail. In their simplest form genealogies list lineage or family members and the relationship between them, but they can also record detailed social demographic information such as dates of birth, death, and marriage, and academic and political achievements of their members. These records provide a rich mine of information for historical research.
Genealogies distinguish themselves from other population records in a number of ways. People recorded in genealogies normally have common ancestry. The generation to which a recorded member belongs, and the relationship between recorded lineage members, are all clear. This makes genealogical records very useful in the study of social mobility of the family, population genetics, and heterogeneity. Differing from most other population records, which are usually obtained from clearly defined territories, or particular social or religious groups in certain areas, people recorded in genealogies are less, if at all, restricted by their place of residence. Because of that genealogies cannot provide information on the total population of a given region, but they can be very useful for the study of geographic movements of lineage or family members and migration patterns if such information is available. Another notable feature of genealogies is that some extend over many generations and contain much information, which provide rare opportunities for the study of early population history.
The way in which genealogies are compiled also makes them differ from other historical population data. Genealogies are often constructed by people of several generations, usually under different circumstances, and over a long period. The rules of compiling genealogies, or the implementation of such rules, may change from time to time. The interval between the time when a certain social demographic event took place and when the event was recorded in the genealogy could be very long. While genealogical records have many desirable qualities, in comparison with other historical population data they are more likely to have suffered from selective biases, which arise mainly from the following influences.
First, because genealogies were usually constructed for the purpose of recording and glorifying the history of the family or lineage other than demographic research, certain family members, for example, children who died young, female descendants who are often regarded as less important to the family in a patrilineal society, and persons who have brought disgrace to the family, are frequently excluded from genealogies. Genealogies also tend to be compiled by families or lineages with relatively high socio-economic status (Harrell 1987; Telford 1990; Zhao 1994).
Second, genealogies can be constructed in different ways. The compilation of a genealogy may be started from and by people who lived hundreds or even thousands years ago, their descendants being later added into the genealogy generation by generation. This results in a descendant genealogy. Alternatively, a genealogy can be compiled by a person or a group of people of later generations who trace their ancestors generation by generation back through time and add them into the genealogy. This produces an ascendant genealogy. Genealogies compiled in this backward fashion are most likely to obtain the records of ancestors from church books, vital registrations, or other data sources. The genealogies, therefore, can be viewed as secondary data sources. Since ascendant genealogies require that each generation to have at least one member surviving into adulthood and to produce at least one child, they tend to exhibit lower mortality and higher fertility than the general population.
Third, many available genealogies are records of lineages that survived to the time when the genealogies were updated last time or collected, which could be hundreds of years since the start of their compilation. One reason, among many, that these lineages could avoid extinction is that they have experienced favourable demographic conditions such as higher fertility, lower mortality, young ages of marriage, and higher proportion marrying, which all promote population growth. Among survival patrilineages that extend through the male line, sex ratios may also be higher than the population as a whole. Biases of this kind tend to be found in the first few generations recorded in a genealogy, and their impact becomes less observable thereafter (Zhao 2001).
3.4. Other Data Sources

While parish registers, population registers, census-type materials, family and lineage genealogies are the most widely used data in historical demography, other data sources are available and can be used in the study of population history as well. In one of the earliest books devoted to historical demography, T. H. Hollingsworth listed nineteen data sources that, in addition to those already discussed above, include Bills of Mortality, vital registration data, fiscal documents, military records, inventories of property, wills, marriage settlements, eye-witness estimates, long-term price records, the number and extent of towns, archaeological remains, methods of agricultural economy, ecclesiastical and administrative geography, new buildings, colonization of new land, cemetery data from both skeletons and tombstone inscriptions (Hollingsworth 1969). The list can be extended further. For example, detailed land registers made in Chinese history, biographical information on ‘virtuous women’ and ‘faithful widows’ kept in historical Chinese county record books, convict indents which detailed the convicts sent to Australia, and other migration records such as ship lists could all provide useful demographic information and have been used in the study of past population changes. In comparison with the data sources discussed in the previous three sub-sections, these data tend to be less common, more selective and more difficult to handle. They are, therefore, less frequently used in demographic research.

Aside from historical data recorded at the level of individuals or households, many countries also accumulated considerable amounts of aggregated population statistics (for example population totals by regions) throughout their history. These data also shed light on population changes in the past and can be used in historical demography. Fluctuations in recorded population totals often reflect both actual demographic changes and the strength and influence of the government or the size of the area under its jurisdiction. Statistics of this kind can also be influenced by the purposes of, and procedures used in, collecting these data, and suffer from various types of under or over registrations.
A further data source that is of great importance for the international comparative study of population history consists of various demographic databases recently constructed around the world. One such example is IPUMS-International (Integrated Public Use Microdata Series – International). The project began in 1999 and is managed by the Population Centre at the University of Minnesota. Its objective is to inventory, preserve, harmonize, and disseminate census microdata collected around world. By the year 2002, a preliminary database comprising 48 million people in six countries was already released. Since then much larger amounts of census data obtained from many more countries have been added into the database and subsequently released worldwide. Although most of these census records were collected in the twentieth century, they will increasingly become an important data source for the study of recent population history. Another example is the HMD (Human Mortality Database) created by the Department of Demography at University of California, Berkeley and the Max Planck Institute for Demographic Research. This database, which contains detailed mortality data collected from more than 30 countries over the last three centuries, has already been used in more than 200 studies.


4. Methodological Development
Because historical demographers are often required to work with surviving population data that were not collected for the purpose of demographic research and thus difficult to use, they have frequently turned to non-conventional and sometimes sophisticated methods to accomplish their analysis. Partly for this reason, historical demography has made several major contributions to the advancement of demographic techniques over the past half century. Some of these developments are summarized in the following sub-sections.
4.1. Family Reconstitution
Family reconstitution is a technique of record linkage. It brings together scattered demographic information from baptism, marriage and burial records, and reconstructs the demographic history of individuals and their families. The technique is widely used by historical demographers in Europe and North America where large numbers of parish registers are available (Levine 2003).
High quality parish records are the foundation for family reconstitution. Detailed baptism, marriage and burial records give information about a person’s birth and childbearing (date of birth or baptism, the name of the new-born, the names of the parents and maybe the occupation of the father), marriage (date of marriage, the names and ages of the bride and groom and sometimes their marriage history) and death (date of death or burial, the name, sex and age of the deceased). When linked together, these records could provide a detailed account of the demographic history for an individual and important information for his or her immediate family members.
Family reconstitution can be carried out either manually or by computers. To assemble available demographic and sometimes non-demographic information relating to a marriage, Henry developed the Family Reconstitution Form (FRF). The FRF or similar forms were used widely by historical demographers involved in family reconstitution studies before computer software was developed to complete the task. Although not exactly the same, these forms are generally used to record the date of the marriage, dates of birth (or baptism) and death (or burial) of the husband and wife, and dates of birth (or baptism) and death (or burial) of their children. Occasionally, related socio-economic information is also recorded on these forms. The linked data are then analysed using conventional demographic methods to generate required population statistics (Fleury and Henry 1956; Wrigley 1966b). When performed manually, family reconstitution is very time consuming. Partly for this reason, many early studies involved only a small number of parishes. Rapid advances in computer technology have prompted the development of computer-based and more efficient methods of record linkage. Thanks to this improvement, large scale family reconstitution studies have been conducted successfully in several countries in the world.
Family reconstitution is a very effective way of deriving useful demographic information from baptism, marriage, and burial records. Demographic history reconstructed using this method sometimes provides more detailed and reliable information than that obtained from retrospective surveys, which may be biased by respondent’s recall errors. Moreover, because parish registration was maintained for hundreds of years in many countries, demographic records generated through linking these registers have become one of the most significant data sources for the study of long-term population changes.
Because of the constraint imposed by parish registers that allow only births, marriages, and deaths to be inferred from recorded religious activities such as baptism, marriage ceremonies, and burials, family reconstitution alone can generate data about population flows (such as total number of births, deaths and marriages in a specified period), but not population stocks. Furthermore, information about migration such as moves into (except through marriage) and out of the parish cannot be obtained from parish registers or family reconstitution. Demographic results revealed by family reconstitution could, therefore, be affected by selection biases if people who moved away from the parish behaved differently from those who did not. This possibility should be examined whenever such inquiries could be conducted.
4.2. Inverse Projection, Back Projection, and Generalised Inverse Projection
In addition to providing opportunities for reconstructing demographic history for millions of recorded individuals through family reconstitution, parish registers generate large amounts of aggregated demographic data such as total numbers of recorded births (or baptisms) and deaths (or burials) in a population or an area. How these summary data might be used in the study of population history of a region or country was another major challenge that faced researchers in the early years of the development of historical demography.
A breakthrough was achieved when R. Lee developed the method of inverse projection in the early 1970s and used it to estimate detailed fertility and mortality rates for historical English populations. Lee’s method is based on assumptions about age-specific mortality and fertility schedules and migration, and requires information about population size and age distribution at a given time in the study period. It can then use a sequence of births and deaths recorded over time to derive sequences of vital rates and age distributions of the population at any time or time interval over this period. The method uses a technique that is similar to population projection but in an inverse form. Here, the inversion is not temporal but logical, in the sense that the method differs from the conventional population projection where sequences of births, deaths and age distributions are derived from a given sequence of vital rates and a starting population. Theoretically, the method of inverse projection ‘can be used either forward or backward in time’ (Lee 1974: 495).
On the basis of the inverse projection, J. Oeppen developed the method of back projection. This method uses totals of births and deaths recorded in a period and information about population size and age composition at the point in time as its input parameters. On the basis of ‘assumptions (or direct information) about certain other characteristics of the population, such as the age patterns of mortality and net migration’, it projects the population backward in time. The method of back projection can provide detailed demographic estimates such as population size and age distribution, crude birth and death rates, gross and net reproduction rates, expectation of life at birth, and other useful information at any time interval considered appropriate, or specified, by the researcher (Wrigley et al. 1997: 7). The method was used by Wrigley and Schofield in their study, The Population History of England 1541-1871, to produce detailed demographic data for historical English population over a period of more than three centuries.
While back projection was regarded as an important technical breakthrough in historical demography, it was open to a number of criticisms, however (Lee 1985 and 2004). Partly as a response to these criticisms, J. Oeppen improved his early method and developed it to a wider class of models that he termed ‘Generalized Inverse Projection’ (GIP), within which both back projection and inverse projection can be defined. According to Oeppen, General Inverse Projection ‘provides a flexible framework for the consistent integration of data and a priori information to produce constrained demographic projections. The basic structure can be applied to problems in population reconstruction, interpolation, and data correction’. Oeppen’s refined method allows researchers to estimate a series of mortality and migration parameters ‘that correspond to the period for which totals of births and deaths are available, and, simultaneously, a series of population age structures that are consistent with the data and the parameters’ (Oeppen 1993: 245-246). In addition to Generalized Inverse Projection, other related techniques such as Differentiated Inverse Projection and Stochastic Inverse Projection have also been developed on the basis of the Inverse Projection pioneered by Lee. These methods have been used to study long-term demographic patterns, their changes over time, and the relationship between those changes and economic development. Detailed information about these developments can be found in D. Reher and R. S. Schofield (1993) and E. Barbi, S. Bertino and E. Sonnino (2004).
4.3. Computer Simulation
Computer simulation is a technique in which researchers use computers to mimic ‘reality’, conduct ‘experiments’, and examine the impact of given events in which they are interested. The technique is increasingly used to study questions that are too difficult, or not possible, for conventional methods to handle. Computer simulation is neither new nor does it belong to population studies alone, but it has played an important part in advancing historical demography, as evident in the development and use of two computer micro-simulation systems: SOCSIM and CAMSIM.
As mentioned earlier, when, in the early 1960s, P. Laslett made his important discoveries about household composition in historical England, it became necessary to test the validity of these research findings and to establish that the high proportion of simple family households recorded in English history was indeed a result of people’s residential propensity. To do this, K. Wachter, E. Hammel, P. Laslett and other researchers developed a sophisticated computer simulation system SOCSIM. Their simulation investigations of household formation in past times led to the publication of a landmark work, Statistical Studies of Historical Social Structure (Wachter, Hammel and Laslett 1978). Following this initial success, J. Smith and J. Oeppen developed another computer simulation system CAMSIM (Smith 1987; Smith and Oeppen 1993).
SOCSIM and CAMSIM share a number of similarities. Both of them are computer micro-simulation systems in which all demographic events are simulated at the level of individuals rather than at the level of population or sub-population groups as in a macro-simulation. The two systems are capable of addressing both the central tendency in people’s demographic behavior and outcomes, and their intra-population variability. Both SOCSIM and CAMSIM are ‘stochastic’ and execute all demographic events randomly according to the probabilities that govern the occurrence of those events. This allows researchers to examine not only the deterministic aspect of the demographic process, but also the impact of chance. Both simulation systems were initially designed for the historical investigation of kinship structure and household composition, and have been used widely in historical demography, studies of changing kinship structure in contemporary and future populations, and investigations of other theoretical and methodological issues. In spite of their similarities, however, SOCSIM and CAMSIM differ notably in their system designs, major assumptions and simulated outcomes. These technical details, which are crucial for using the systems and interpreting their results, are discussed in a number of publications (Hammel, Mason and Wachter 1990; Smith 1987; Smith and Oeppen 1993).
Computer demographic simulation differs from empirical research by nature. In simulation studies, a model population is generated by the computer. Changes in the population and their implications are simulated under the demographic conditions determined by input demographic parameters. These conditions can be manipulated according to the research design. Computer simulation can handle complicated demographic processes, and the simulation can be conducted repeatedly under the same or different conditions. This is very useful in measuring the impact of chance and in identifying the influence of major factors being studied. Because of these characteristics and its effectiveness in examining complex issues, computer simulation has become an important aid for historical demographers. The two systems mentioned above have been used in a large number of studies (Zhao 2006). In addition to SOCSIM and CAMSIM, other demographic simulation systems (for example MOMSIM) have also been developed and applied in historical demography (Ruggles 1987). Their use has made important contributions to our understanding of population and social history in recent decades. It is important, however, to note that in the real world observed population changes or other social phenomena are affected by many interrelated factors. In contrast, computer demographic simulation takes into account only the factors that have been specified in the simulation system. Accordingly, simulation results tell us about the implications of the theory and assumptions that are embedded, both explicitly and implicitly, in the process of simulation, not about what has actually happened in a real world. Simulation results differ from, and should not be mistaken for, reality.
4.4. Coale’s Fertility Indices and Other Methodological Developments
The European Fertility Project discussed in section two also made considerable methodological contributions to the development of historical demography and population studies. This project was designed to exploit the wealth of information contained in summary demographic data gathered from some six hundred European provinces in the nineteenth and twentieth centuries and to compare fertility patterns across regions and over time. To achieve this, Ansley Coale (1917-2002) developed a set of fertility indices (I_f, I_g, I_h, and I_m). I_f, I_g and I_h indicate relative levels of general fertility, marital fertility and non-marital fertility; I_m indicates the proportion of married. Using these indices, researchers involved in the project successfully showed how observed general, marital and non-marital fertility levels differed from ‘a clearly defined maximum fertility’ as indicated by fertility rates among the Hutterites, and the extent to which the general fertility was affected by the levels of marital and non-marital fertility, and by patterns of marriage in all European populations included in the study (Coale 1986: 162).
Methodological developments have also been made in other research areas such as record linkage, the analysis of household composition, the evaluation and correction of historical data, and the use of genealogical records and other less common historical materials. Some of these techniques and methods are relatively easy to understand and have been widely used (for example, methods used in studying household composition), while others are still in the early stages of development, or tend to be data source specific and have only limited use. Therefore, discussion about the development of these techniques and methods is not presented here.
5. Concluding Remarks: Historical Demography at the Beginning of the Twenty-first Century
This chapter provides an introduction to the history of historical demography, its principal data sources, and major methodological developments made in recent decades. The discussion has concentrated on conventionally defined historical demography – the application of demographic techniques to historical data. Some topics have not been discussed though they are also related to population history. One such topic is paleodemography, the study of demographic characters of ancient populations. The other is qualitative historical demography promoted by D. Kertzer and others (Kertzer 1997). While these research areas are important for our understanding of people’s demographic behaviours in past times, their underlying causes, and implications, they have not been discussed in this chapter because the methods, evidence and information used in paleodemography and qualitative historical demography differ considerably from those used in conventional historical demography.
Historical demography has made impressive progress since it was established as a research discipline half a century ago. Scholars have assembled a vast quantity of surviving population records, from which they have derived valuable demographic information through the use of family reconstitution or other ingeniously developed methods. These efforts have generated considerable insights into demographic behaviour, population changes, and their interaction with socio-economic development in the last few centuries, which have provided crucial empirical evidence for major demographic theories. Methodological developments made in historical demography have also greatly advanced population studies. Whilst remarkable achievements have been made in historical demography in Europe, North America and East Asia, much remains to be done in other parts of the world.
In the light of these achievements, it is clear that more effort should be made to promote historical demography as a means of advancing understanding of modern demographic problems, especially in the areas where such studies are still rare. One of the most important findings of historical demography is the great diversity in demographic behaviours and in demographic responses to emerging changes in social, economic or environmental conditions. Although detailed historical demographic investigations have been conducted in limited numbers of populations, they reveal significant variations in patterns of marriage, levels of fertility, and ways of controlling family size. Knowledge of pre-transition societies has had significant implications on the development and interpretation of population theory. Many current demographic theories, models and methods have been developed based on demographic experiences of limited countries, especially developed countries. While these theories play an important role in guiding demographic research and these models and methods proved to be useful in population studies undertaken in other areas, there is a risk that some of these theories, models or methods may not be applicable to demographic reality of these populations. This is a problem that has to be addressed if we want to gain a better understanding of world population issues. If for no other reason, this concern alone justifies the importance of further promoting historical demography in the world.
Further effort should be made to improve the use of historical data and to promote research collaboration. More attention should be given to the linkage and use of records from different sources, for example, linking parish registers and vital registration data to census records, jointly using household registers and genealogical data, and jointly using demographic and non-demographic data. Operations of this kind provide opportunities for researchers to identify registration problems in different data sets, to overcome the limitations of the data of a particular type, and to examine the questions that could not be studied using records from a single source. Another important step in advancing historical demography is to further promote research collaboration between historical demographers from different countries and collaboration among scholars from disciplines. This is particularly helpful in countries where historical demography is in the early stages of development and major investigations into population history often call for multi-disciplinary effort.
Conditions for advancing historical demography have been greatly improved over the last few decades. We have gained much experience in using historical data for demographic research. Increasing amounts of surviving historical data have been collected, catalogued and made in computer readable forms. Researchers no longer need to spend extensive resources on gathering and inputting records. Many studies can now be conducted on a scale that could hardly be imagined a couple of decades ago. Increasing computing power and the development of many techniques and analytical tools make historical demography a less challenging task than ever before. In this favourable environment, the early twenty-first century is expected to see a further boom in historical demography in the world.
Glossary
Historical demography – the application of demographic techniques to historical population data.

Demographic transition – the historical shift of fertility and mortality from both high to both low. The term also refers to the model describing this change.

Natural fertility – the level of fertility that would prevail if couples do not alter their reproductive behaviour according to the number of children already born.

Parity progression ratio – the progression of women who already have n children go on to have at least another child.

Parish registers – registers of baptisms, marriages and burials made by Christian churches in many European countries.

Genealogies – records recounting one’s descent from an ancestor or ancestors by enumeration of the intermediate persons.

Family reconstitution – a method linking together separately-recorded vital events in order to reconstitute the demographic history of individuals and their families.

Computer demographic simulation – the technique of using a computer program to model and study demographic processes and their outcomes.
Acknowledgement
The author would like to thank Richard Smith, Tommy Bengtsson, and Mac Boot for their comments on earlier versions of this chapter.
Biblography
Barbi E. Bertino S. Sonnino E. (2004). Inverse Projection Technique: Old and New Approaches, 139 pp. New York: Springer. [A collection of papers on inverse projection related techniques]
Bengtsson T., Campbell C. and Lee J. (2004). Life under Pressure Mortality and Living Standards in Europe and Asia, 1700-1900, 531 pp. Cambridge, Massachusetts: The MIT Press. [A publication concentrating on the study of mortality by the Eurasia Population and Family History Project]
Bengtsson T. and Lindstrom M. (2000). Childhood Misery and Disease in Later Life: The Effects on Mortality in Old Age of Hazards Experienced in Early Life, South Sweden, 1760-1894. Population Studies 50: 263-277. [A historical investigation of the impact of childhood experience on old age mortality]
Campbell C. (2004). Sources and Measures. Life under Pressure Mortality and Living Standards in Europe and Asia, 1700-1900 (T. Bengtsson, C. Campbell and J. Lee), 441-476. Cambridge, Massachusetts: The MIT Press. [A discussion of data sources and measures used in the Eurasia Population and Family History Project]
Catalano R., Bruckner T. and Smith K. R. (2008). Ambient Temperature Predicts Sex Ratios and Male Longevity. PNAS 105: 2244-2247. [An interesting study of the impact of temperature change on sex ratio at birth and mortality]
Coale A. J. and Treadway R. (1986). A Summary of the Changing Distribution of Overall Fertility, Marital Fertility, and the Proportion Married in the Provinces of Europe. The Decline of Fertility in Europe (eds. A. J. Coale and S. C. Watkins) 31-181. Princeton, New Jersey: Princeton University Press. [A paper summarising the major findings of the European Fertility Project]
Coale A. J. and Watkins S. C. (eds). (1986). The Decline of Fertility in Europe, 484 pp. Princeton, New Jersey: Princeton University Press. [An important publication by the European Fertility Project]

Fleury M. and Henry L. (1956). Des registres paroissiaux à l’historie de la population. Manuel de lépouillement et d’exploitation de létat civil ancien, Paris: INED. [An early work discussing parish registers and the method of family reconstitution]

Gautier É. And Henry L. (1958). La population de Crulai: Paroisse normande. 272 pp. Paris: Presses Universitaries de France. [An influential work reporting the family reconstitution study conducted in Crulai]
Goubert P. (1970). Historical Demography and the Reinterpretation of Early Modern French History: A Research Review. The Journal of Interdisciplinary History, 1: 37-48. [A review article on the early development of historical demography and its impact in France]
Hajnal J. (1965). European Marriage Patterns in Perspective. Population in History: Essays in Historical Demography (eds. D. V. Glass and D. E. C. Eversley) 101–143. London: Edward Arnold. [A classical study of marriage patterns in historical Europe]
Hajnal J. (1983). Two Kinds of Pre-industrial Household Formation System. Family forms in historic Europe (eds. R. Wall et al.) 65-104. Cambridge: Cambridge University Press. [An important study examining household formation systems in northwest Europe and other areas]
Hammel, E. A., Mason C. and Wachter K. W. (1990). SOCSIM II: A Sociodemographic Microsimulation Program. Berkeley: Graduate Group in Demography, University of California. [A discussion of technique details of the SOCSIM computer micro-simulation system]
Harrell S. (1987). On the Holes in Chinese Genealogies. Later Imperial China 8(1): 52-87. [A paper examining under-registration problems in Chinese genealogies]
Harrell S. (1995). Introduction: Microdemography and Modelling of Population Process in Late Imperial China. Chinese Historical Microdemography (ed. Harrel), 1-20. Berkeley: University of California Press. [A useful discussion on historical demography in China]
Hayami A. (2001). The historical Demography of Pre-modern Japan, 191 pp. Tokyo: University of Tokyo Press. [A major English publication on Japanese historical demography]
Henry L. (1953). Fécondité des mariages. Nouvelle méthode de mesure. Travaux et Documents, Cahier n° 16, INED-PUF. [An early discussion of parity progression ratio]
Henry L. (1956). Anciennes familles genevoises, 234 pp. Études Démographique: XVI^e-XX^e siècle. Cahier n° 26, Paris: Presses Universitaries de France. [A study examining the demographic history of the Geneva’s ruling class]
Henry L. (1961). Some Data on Natural Fertility. Eugenics Quarterly 18: 81–91. [An early discussion on natural fertility]
Hollingsworth T. H. (1969). Historical Demography, 448 pp. London: The Sources of History Limited in Association with Hodder and Stoughton Limited. [A major book on historical demography published in the early years of its development]
Jannetta A. B. and Preston A. H. (1991). Two Centuries of Mortality Change in Central Japan: The Evidence from a Temple Death Register. Population Studies 45: 417-436. [A study examining mortality changes in historical Japan]
Kertzer D. I. (1997). Qualitative and Quantitative Approaches to Historical Demography. Population and Development Review 23: 839-846. [A detailed discussion on qualitative historical demography]
Kim K. (2005). Eighteenth-century Korea Marriage Customs: The Tangsǒng Census Registers. Continuity and Change 20: 193-209. [A study examining marriage patterns in historical Korea]
Laslett P. (1966). The World We Have Lost, 380 pp. London: Methuen. [An influential book on English social history]