¹ Hajnal, John, “European Marital Patterns in Historical Perspective,” Population in History, ed. by Glass, David V. and Eversley, David E. C. (London, 1965), pp. 101–43.Google Scholar

² This research, currently being done at the Office of Population Research, Princeton University, has not yet been published but has been briefly described by Ansley J. Coale (“The Demographic Transition,” paper presented at the Kyoto Conference on Fertility Transition, Dec. 1–5, 1975, Japan). A problem of time horizon should be noted, however. Fertility may rise in areas of Central Asia for a century or so and then fall afterwards. If we look at the data over a long enough time span, they may well agree with the generalization that the birth rate falls with long-run modernization. I am indebted to Brinley Thomas for this point.

³ See esp. Taeuber, Irene, The Population of Japan (Princeton, 1958).Google Scholar

⁴ Akira Hayami and his associates at Keio University have pionered in these reconstitution studies for Japan. A number of their studies have been published in English in the journal, Keio Economic Studies. For detailed citations to Hayami's articles, as well as to related findings, see Eng, R. Y. and Smith, Thomas C., “Peasant Families and Population Control in Eighteenth-Century Japan,” Journal of Interdisciplinary History, 6 (Winter 1976), 417–45CrossRefGoogle ScholarPubMed. In addition, see the references cited in n. 5 below. The question of what is high or low is to a certain extent a matter of opinion. My own impression is that the rates found by Eng and Smith are high.

⁵ See esp. Hanley, Susan B. and Yamamura, Kozo, “Population Trends and Economic Growth in Pre-industrial Japan,” Population and Social Change, ed. Glass, David V. and Revelle, Roger (London, 1972), pp. 451–99Google Scholar. See also Hanley, Susan B., “Fertility, Mortality and Life Expectancy in Premodern Japan,” Population Studies, 28 (March 1974), 127–42.CrossRefGoogle Scholar

⁶ Hanley, Susan B., “The Japanese Fertility Decline in Historical Perspective,” p. 25,paper presented at the Kyoto Conference on Fertility Transition,Dec. 1–5, 1975,Kyoto, Japan.Google Scholar

⁷ Hanley, “Fertility, Mortality and Life Expectancy,” p. 133.

⁸ Of course one might argue that birth control in the sense of prevention of births is nothing more than a modern substitute for infanticide in Japan. In this line of reasoning infanticide, a method of population (but not birth) control, was the principal means of controlling family size in Tokugawa Japan. Family size was deliberately regulated by the practice of winnowing out of babies, or mabiki as the Japanese called it. Such a hypothesis could be stretched to account for the rising trend of fertility in the Meiji period by arguing that the national legislation prohibiting mabiki became increasingly effective during Meiji. Then the population had to develop alternative techniques (birth control) for the same ends as were obtained with mabiki; thus the modern fertility decline in Japan boils down to a change in family limitation methods. This view does not seem reasonable; in any case, as I shall argue below, there is considerable support for the inference that the population of premodern Japan (i.e., farmers) wanted large families.

⁹ Taeuber, Population of Japan, p. 41.

¹⁰ Estimates have been prepared by Yuzo Morita, Yoichi Okazaki, and Masaaki Yasukawa, among others. See Nihon, Kōseishō Jinko Mondai Kenkyōjo [Japan, Ministry of Health and Welfare, Office of Population Problems], Kenkyū shiryo dai 145. Meiji shonen ikō Taishō ku-nen ni itaru danjo nenreibetsu jinkō suikei ni tsuite. [Research Series, 145. Concerning estimates of the population from 1868 to 1920 according to sex and age] (Tokyo, 1962). Also see Morita, Yuzo, “Estimated Birth and Death Rates in the Early Meiji Period of Japan,” Population Studies, 17 (July 1963), 33–56CrossRefGoogle Scholar; and Yasukawa, Masaaki and Hirooka, Keijiro, “Estimates of the Population Size and of Birth and Death-Rates in Japan, 1865–1920,” Keio Economic Studies, 11 (1974), 41–66Google Scholar. A discussion of the various estimates of mortality and fertility during the Meiji period is contained in Ohbuchi, Hiroshi, “The Demographic Transition in the Process of Japanese Industrialization,” in Patrick, Hugh, ed. Japanese Industrialization and its Social Consequences (Berkeley, 1976), pp. 329–61.Google Scholar

¹¹ I used the data on infant deaths and on children aged 0 to reestimate the figures of births, a technique which was first applied to the Japanese data by Irene Taeuber. In computing the marital fertility rates, I adjusted for illegitimacy wherever I could.

¹² This quotation from a translation of the legislation appears in Irene Taeuber, Population of Japan, pp. 269–70.

¹³ Leibenstein, Harvey, “The Economic Theory of Fertility Decline,” Quarterly Journal of Economics, 89 (Feb. 1974), 1–31.CrossRefGoogle Scholar

¹⁴ Easterlin, Richard, “The Economics and Sociology of Fertility: A Synthesis,”rev. version of a paper prepared for the Seminar on Early Industrialization, Shifts in Fertility, and Changes in Family Structure, Institute for Advanced Study,Princeton, N.J.,June 18-July 9, 1972.Google Scholar

¹⁵ For a general discussion of the extent of rural-urban fertility differentials relative to the international spreads of fertility between the more developed and less developed countries, see Simon Kuznets, “Rural-Urban Differences in Fertility: An International Comparison,” Economic Growth Center, Yale University, Center Discussion, no. 166.

¹⁶ In 1920 the linear correlation between the two variables was + 645 (46 prefectures); in 1930 it was + 734 (46 prefectures). Both correlations are significant at the 1 percent level.

¹⁷ For the 121 agricultural counties, the linear correlation between percentage females single 15 to 19 in 1925 and the CDR (average tor the period 1925–1930) is –.69, significant at the 1 percent level.

¹⁸ In 1953 we have cross-sectional data on the infant mortality rate (IMR) defined as deaths to children under one year of age divided by live births and the mean age of first marriage for females (MAFM) for the rural component of each of the 46 prefectures. The correlation between the two variables is – 49, significant at the 1 percent level. I define an agricultural prefecture as one in which at least 60 percent of the male labor force is in the primary sector. There are nine of these prefectures according to the 1930 census. For these nine prefectures Kendall's rank correlation between trends in the IMR over the period 1919/21 to 1959/61 (averages over the years 1919 to 1921 and 1959 to 1961 were used as benchmarks) is – 722, significant at the 1 percent level.

¹⁹ For example, Kendall's rank correlation between trends in life expectancy of males at age 42 and the MAFM over the trend period 1921/25 to 1954/56 is zero.

²⁰ The linear correlation between d_c and the MGFR in line 1 is — 561 (36 counties), significant at the 1 percent level; in line 2, – 227 (65 counties), insignificant at the 5 percent level; in line 3, + 332 (20 counties), insignificant at the 5 percent level. A number of objections can be raised to my “child labor” interpretation of the results in Table 4. For instance, it can be argued that the productivity of labor (and per capita income) is higher the lower the dc ratio, so that what we see may be a relationship between income per capita and fertility. For the prefectures we have data on income for 1930. I tested the hypothesis that income per capita and the MGFR are related among the nine agricultural prefectures, and found the rank correlation was + 056, insignificant at the 5 percent level. Another objection concerns age and marital structure. The MGFR is computed as a ratio of births per married woman 15 to 49, but the d_e ratio relates land areas to all adults 15 and over regardless of marital status and age. Could the relationship implied by the computations reflect varying family structure (i.e., the proportions of unmarried and aged in families) rather than a causal connection? A statistical test casts doubt on this objection: I computed the ratio of all married persons 15 to 49 to adults 15 and older; the correlation of this variable with the MGFR for the 121 agricultural countries was + 04, insignificant at the 1 percent level.

²¹ The linear correlation between E_f and the MGFR for the 121 agricultural counties is – 08, insignificant at the 1 percent level. It must be noted, however, that the E_f ratio has at least two possible interpretations, making it difficult to segregate the two potential effects.

²² Of the 46 prefectures of Japan, 36 had at least one county with the percentage of male labor force in the primary sector (P_m%) greater than or equal to 70 percent. For these 36 prefectures, I chose the county with the highest and that with the lowest P_m%. The correlation between the spreads in these corresponding P_m% values (for 1930) and the spread in their MGFRs is – 842, significant at the 1 percent level. The correlation between the spreads in the MGFR and spreads in the differences between the high and low P_m%values is – 464 for these prefectures, significant at the 1 percent level.

²³ See Mosk, Carl, “Demographic Transition in Japan” (Ph.D. diss., Harvard University, 1976), pp. 145–46.Google Scholar

²⁴ In 1920 there were 81 cities. The linear correlation between E_f and the percentage of women 24 to 24 single is + 445.

²⁵ About all that I could tease out statistically was that, for the 43 prefectures with at least one city in 1920, the linear correlation between trends in the percentages of women 20 to 24 single in the urban areas and the corresponding rural percentages was positive and significant at the 5 percent level.

²⁶ For the six censal years 1920–1955 and for the 46 prefectures (there were only 43 relevant prefectures in 1920) there were only five instances in which the MGFR for the rural sector of a prefecture fell short of the corresponding MGFR for the urban sector.

²⁷ For instance, the linear correlation over the prefectures between the MGFR in the rural sector of a prefecture and the MGFR for the corresponding urban sector is + 795 in 1920 (43 prefectures), significant at the 1 percent level; in 1930 the corresponding correlation is + 858 (46 prefectures), significant at the 1 percent level.

²⁸ On this see Leibenstein, “Fertility Decline,” p. 10. Ronald Dore discusses the contributions the dependence of the aged on the support of their children made to the benefits of children in rural Japan in his article, “Japanese Rural Fertility: Some Social and Economic Factors,” Population Studies, 7 (July 1953), 77.Google Scholar

²⁹ See Taeuber, Population of Japan, and Ohbuchi, “Demographic Transition.”

³⁰ See Hashimoto, Masanori, “Economics of Postwar Fertility in Japan: Differentials and Trends,” Journal of Political Economy, 82, supp. (March/April 1974), S. 170–94.CrossRefGoogle Scholar

³¹ Yamamura, Kozo and Hanley, Susan, “Ichi hime, ni Tarō: Educational Aspirations and the Decline in Fertility in Postwar Japan,” Journal of Japanese Studies, 2 (Autumn 1975), 83–125.CrossRefGoogle Scholar

³² For the decade of the fifties statistics on abortions are given in the Nihon, Kōseishō Daijinkanbō Tōkei Chōsubu [Japan, Ministry of Health and Welfare, Bureau of Statistical Survey of the Office of the Minister], Eisei nenpō [Annual health report]. (Tokyo, various years)

³³ On the use of the condom see Matsumoto, Y. Scott, Koizumi, Akira, and Nohara, Todahiro, “Condom Use in Japan,” Studies in Family Planning, 3 (Oct. 1972), 251–55.CrossRefGoogle Scholar