EXPLORATIONS IN ECONOMIC HISTORY 15,269-289 (1978) Fecundity, Infanticide, and Food Consumption in Japan* CARL MOSK Department of Economics, University of California, Berkeley INTRODUCTION Of the many apparently unique facets of Japanese history, none appears more perplexing than Japan’s demographic history. On the face of it, the Japanese experience would seem to violate the basic tenets of demographic transition theory. According to the demographic transition rubric, prior to modernization, fertility and mortality are both high; then mortality falls and after a lag, natality also drops, leading to a state of low death and birth rates. However, according to official Japanese statistics for the Meiji (1868- 1912) period and the recent findings of village family reconstitution studies for the Tokugawa era, prior to modem economic development low natal&y and mortality prevailed in Japan. With the opening of Japan to the West, dramatic increases in per capita income, birth, and death rates seem to occur, and only after the first census conducted in 1920 do the trends in the vital rates turn decisively downward. Figure 1 illustrates the trends in vital rates suggested by the official sta- tistics. From a low of 25.3 in 1875 the CBR advances to 31.6 in 1919. The CDR rises from 19.1 to 22.8 over the same period. It should be stressed that these data are not entirely flawless. Soon after the Meiji Restoration the new government issued the koseki-hc?, the law requiring registration of all families and of vital events occurring to family members. The house- hold registration book, the koseki, became the basic source for computa- tion of vital statistics and, prior to 1920, for all population counts as well. Japanese demographers tend to suspect that, initially, much of the popula- tion failed to register and establish koseki, and only gradually did the cover- age of the system approach completeness. It is usually assumed that coverage was almost complete by 1920. As a result, a number of Japanese demographers have utilized various ingenious techniques and assump- tions to project backward the 1920 census and to estimate population totals and birth and death rates for the period between the 1870’s and 1920. Re- * I am especially grateful to Simon Pak for his research assistance on this project. The appendix was prepared jointly by Mr. Pak and myself. Conversations with Richard Sutch, Pranab Bardhan, and Robert Eng were of great assistance to me in formulating the approach developed in this paper. 269 0014-4983/78/0153-0269$02.00/0 Copyright 0 1978 by Academic Rear, Inc. All right.3 of reproduction in any form reserved.