Sichuan's Population in the Nineteenth Century: Lessons from Disaggregated Data

Vol. 8, No. 1 Late Imperial ChinaJune 1987 NOTICE: Due to a series of errors in the editorial process, the version of this article published in the previous issue (vol. 7, no. 2) did not include the author's corrections and, moreover, contained faulty reproductions of the maps. The editors offer their apologies and present the correct version here. The author wishes to stress that this is the only version of the article that has been authorized for publication. Bibliographers and indexers should therefore cite only this version.___________________________________________ SICHUAN'S POPULATION IN THE NINETEENTH CENTURY: LESSONS FROM DISAGGREGATED DATA* G. William Skinner It is well known that, in form at least, official provincial population figures from 1741 to 1898 were aggregations oí baojia population registers. Initial compilations were made at county-level yamens; county-level data were then cumulated to give prefectural-level totals, and these were summed in turn to yield the provincial figure. Province totals produced in this fashion have been long been available for certain years of the Qing period, but only rarely have we had access to the disaggregated data on which they were based. This paper analyzes disaggregated annual data for China's most populous province for nine years: 1822, 1828, 1829, 1845, 1847, 1848, 1850, 1852, and 1887. (I shall also make use of a countylevel data set published in the Jiaqing edition of the Sichuan provincial gazetteer.)1 The series of which these nine annual registers form a part is the major source of our knowledge about Sichuan's early modern population history. The several nineteenth-century population figures for Sichuan that have become familiar to us are, in fact, cumulated totals of the data being analyzed here. For instance, the widely cited figure of 44.164 million for 1850 agrees with the total in the present data set, as does the official figure * I gratefully acknowledge support from the Center for East Asian Studies and the Academic Computing Fund at Stanford University. 1 am indebted to Philip L. Ritter for assistance with computation, to Zumou Yue for assistance with the analysis, and to Susan Mann for assistance with translations of imperial edicts. An early version of this paper was presented to the Workshop on Qing Population History, sponsored by the ACLS-SSRC Joint Committee on Chinese Studies, convened in August 1985 at the California Institute of Technology, Pasadena . I am grateful to workshop participants for helpful comments. 1 Sichuan tongzhi, Jiaqing ed., juan 65, Shihuo zhi 4, Hukou 2. Reprinted as Appendix Table 10 in Lu Zijian 1984. 2 G. William Skinner for 1887 (73.179 million).2 Moreover, the population data presented in local gazetteers, at both county and prefectural levels, appear in almost all cases to have been taken from the same annual series of cumulated baojia registers. Thus, the present data set provides an opportunity to evaluate the source of most of what we think we know about Sichuan's population during the nineteenth century. Errors in local data may, of course, be simply canceled out through aggregation, so that cumulations of bad data are rendered plausible. To what extent this has occurred in Sichuan can be assessed only through systematic evaluation of county-level data and their cumulations. More importantly, if even some of these data can be shown to have reasonable validity, they open up exciting prospects for the analysis of variation through historical time and regional space. Indeed, it was the possibility of pursuing a temporal-cum-spatial analysis that induced me to take up the challenge of this particular data set. Regional-systems theory leads us to expect that demographic processes will vary systematically according to place in the structure of regional systems — an expectation borne out by preliminary studies of both France and Japan in the nineteenth century. Such an analysis of Qing China has been thwarted by the lack of a longitudinal , disaggregated data set covering the greater part of at least one macroregion. Since Sichuan's territory accounts for nearly nine-tenths of the Upper Yangzi macroregion (see Maps 1 and 2), the present data set would seem to be eminently suited to such an analytical venture. These data can also...