Patching the Holes in Chinese Genealogies: Mortality in the Lineage Populations of Tongcheng County, 1300-1880

PATCHING THE HOLES IN CHINESE GENEALOGIES: MORTALITY IN THE LINEAGE POPULATIONS OF TONGCHENG COUNTY, 1300-1880* Ted A. Telford Since the initial work using lineage genealogies for Chinese demographic history began, researchers have become accustomed to using very fragmentary data from which to reconstruct historical populations. Many estimates of population characteristics and dynamics have been teased out of various sources through the application of powerful "formal" demographic techniques designed for use with incomplete data. Fitting a set of model life table probablities , for example, to scattered bits of information in order to flesh out the otherwise bony framework of a specific mortality regime often yields perfectly accurate results. However, these techniques, like all power tools, must be used with caution. It is no secret that formal demographers are often criticized for their lack of attention to the troublesome detail of source materials. Many have also pointed out that without the historian's sensitivity to the vagaries ofthe specific documentary sources being manipulated, grossly distorted findings can result. This paper deals with a formal demographic technique that has been used to patch some of the informational "holes" described by Stevan Harrell in his comparison of a few lineage genealogies with a more complete and accurate source (Harrell 1987). This paper will also show how paying attention to the peculiarities of a particular historical demographic source can help to patch holes that heretofore have not been considered in need of repair. Finally, I will describe some findings on the mortality of one lineage population resulting from the use of these techniques. Mortality and Model Life Tables One method used to supply incomplete mortality data is the fitting of known probabilities of dying to one or another set of model tables that will 'Funding from the ACLS/SSRC Mellon Postgraduate Fellowship for Chinese Studies enabled the creating of the historical demographic data set analyzed here. Without such generous support the time-consuming reconstitution and editing necessary to generate such a large data set would not have been possible. Late Imperial China. Vol. 11, No. 2 (December 1990): 116-137© by the Society for Qing Studies 116 Patching the Holes in Chinese Genealogies117 then provide the unknown probabilities.1 Without going into great detail as to why this is such a powerful technique, it works well basically because the age pattern of human mortality is fundamentally fixed by human biology— high in the first year of life, lower in the childhood years, lower still through the teens and early twenties, followed by gradual increases to the highest levels in the very old. The fact that in no known human population is the death rate for infants under one year of age lower than the rate for children age one to four years, means that if the rate for those age one to four years is known, an estimate for those dying before one year is possible. Common variations in the level of the "normal" U-shaped (or J-shaped) age pattern of mortality are encompassed by calculating separate life tables for males and females and by creating composite or averaged "model" tables that reflect different age patterns and levels of mortality due to unique disease patterns or social conditions. One then need only determine which model table most closely fits the empirical data, say, from age twenty through fifty (a 1 For those completely unfamiliar with the construction and use of model life tables, the Coale-Demeny (1966) tables themselves are the standard introduction. Other sets of model life tables have been devised by such organizations as the United Nations, mostly from and for application to the populations of contemporary less-developed countries, whose demographic data sources are often woefully incomplete. Because the Coale-Demeny tables included a large number of the best documented historical populations, including Taiwan in the first half of the twentieth century, they are usually more appropriate for historical work. On the other hand, if one is working with a poorly documented contemporary population , the U. N. model life tables might be a more satisfactory choice. This will usually be the case since contemporary mortality patterns, even in the world's least developed countries, are dramatically different from those of...