1. The quote is from Genesis 6:4 (Cambridge: Cambridge University Press, n.d.: New Testament, p. 7). Floud, Wachter, and Gregory (1990) also make use of this quotation at the beginning of their volume but draw different implications from it. On the methodological importance of analogy and metaphor in the sciences and social sciences, see inter alia, Nelson, Megill, and McCloskey 1987; McCloskey 1994; Mirowski 1987; and the discussion in chapter 2 (''Long-Term Determinants of the Anthropometric Measures, 1901-1979," and appendix) of this volume. [BACK]

2. The relationship between the standard of living and what I call population quality is a thorny one. I distinguish between the standard of living and population quality on practical grounds. In common parlance indicators of potential physical strength and endurance and of intelligence and learning capacity reside within the organs and tissues of the individuals within a population and hence tell us something about the population's average quality. The standard of living is either a more subjective matter involving a psychological sense of self-worth and satisfaction or a matter of goods and services consumed and possessed by individuals within a population and hence external to physical bodies. As I argue in the text of this section and the next section of chapter 1, there is a relationship between Sen's definition of the standard of living, which revolves around capabilities conceived in terms of the physical and psychological capabilities to take advantage of economic and social opportunities, and my definition of population quality in the sense that both of us reject definitions based on per capita consumption of goods and services or on utility. However, my definition is narrower than his since I exclude a host of political and social factors like political freedom which Sen considers important.

Another reason I make a point of distinguishing between population quality and the standard of living defined in terms of per capita consumption is that I feel some scholars mistake the one for the other, or use the former as a proxy for the latter. For instance, in Fogel (1986) and in many of the fine papers in the volume edited by Komlos (1994) there is a tendency to confuse anthropometric measures with the standard of living. As an example, consider the title of Shay 1994. While I do think some aspects of what these scholars define as the standard of living, especially food consumption, are important to determining population quality as I define it, I do not believe that population quality is equivalent to these components of the standard of living defined in consumption terms. Moreover, I favor Sen's definition of the standard of living over that defined in terms of consumption of goods and services because I wish to emphasize the linkage between the standard of living and work capacity.

Steckel (1994a, 1994b) reviews some other proposed indexes that bear a resemblance to what I propose here: for instance, a physical quality of life index advanced by Morris and the UN development index and his own biological standard of living index. However, none of these is identical to population quality. To some extent, of course, it does not matter what we call the underlying variable we are measuring with the anthropometric proxy variables as long as we are clear about what we think statistical analysis is actually telling us. For instance, in this volume I utilize anthropometric measures as proxies for population quality, but persons who prefer to interpret results concerning the determinants of the anthropometric measures in terms of another conceptual framework in which the anthropometric measures are proxies for something else are free to do so. I do attempt, however, to show in Part II that an interpretation in terms of capabilities and work capacity is reasonable. [BACK]

3. The influence of the gene pool on anthropometric measures is one reason that it is difficult to make inferences about the standard of living of slaves in eighteenth- and nineteenth-century America on the basis of comparisons between figures on height and weight for slaves and nonslave populations like Irish immigrants to Boston. Of course, if we adopt Sen's concept of the standard of living, slavery, by constraining human freedom, automatically reduces the standard of living of the enslaved population. [BACK]

4. There is a literature in contemporary Japan known as the Nihonjinron which argues that the Japanese people constitute a racially distinct people. It is sometimes argued by advocates of this viewpoint that the digestive organs of Japanese differ from those of non-Japanese or the Japanese people learn the Japanese language on one side of the brain while other racial groups learn their languages on the other side of the brain, and so forth. Miller (1872) reviews many of these arguments, demonstrating that the experimental evidence adduced in their favor is shaky or nonexistent. He also shows that many of the arguments, such as that concerning language and the brain, are circular. The obvious problem with this is that since many individuals not of Japanese origin, like Koreans living in Japan whose mother tongue is Japanese, learn Japanese, and they either learn the language on the other side of the brain from the Japanese—in which case they are not really learning Japanese as a language—or they learn the language on the same side of the brain as the Japanese—in which case by

definition they are themselves Japanese even though they are not Japanese. It should be obvious from the text that I think the idea of a pure Japanese "race" is not a useful concept and I reject it throughout this book. [BACK]

5. There is a large literature that deals with the potential interactions between the environments phenotypes function in and genetic evolution. One line of argument is sociobiological and posits that maximizing survival of the gene pool may be the basis for altruistic behavior toward one's own children and kin. On this literature, see Goldsmith 1991 and Wallace 1972a, 1972b. [BACK]

6. Our knowledge of the specific impact that the nutrients in food (and their processing through cooking and preparation for eating) have on human growth and potential physical and mental exertion and to long-run coevolutionary change over time certainly leaves much to be desired. Nevertheless, progress has been made. See, for example, Cohen 1987; Durnin 1983; Ebrahim 1979; Eide and Steady 1980; Gibson 1990; Greksa, Pelletier, and Gage 1986; Harris and Ross 1987; Jelliffe and Jelliffe 1979; Pellett 1987; Pike and Brown 1967; Taylor 1982. [BACK]

1. Unless otherwise indicated, the term "height" refers to standing height. Later on in this chapter we shall encounter figures on sitting height, and I will specifically use the term "sitting height" when this is what I am discussing. [BACK]

2. Shay (1994) relies on figures for military recruits for his analysis. This makes a good deal of sense given the fact that he concentrates on regional analysis rather than time series analysis (the data for recruits are available for regions during the early twentieth century and later on for prefectures). In chapters 4 and 5 I provide a much less systematic analysis of regional differentials in height and the other anthropometric measures for military recruits than does Shay and therefore recommend that the reader interested in regional differentials consult his interesting analysis. [BACK]

3. I am grateful to Professor Osamu Saito of the Institute of Economic Research at Hitotsubashi University for pointing this out to me. [BACK]

4. For the weights and for comparisons between calorie intake on a per capita and on a per consumer unit weighted basis, see table 1. [BACK]

5. Kurosaki (1967) provides an excellent discussion of regional variation in food intake during the preindustrial period. He shows that in some areas of southern and northern Japan, for example, in southern Kyushu, potatoes, not rice, were the main staple foodstuff. Kito (1986) also gives figures on food consumption for regions of Japan during the nineteenth century. [BACK]

6. After the Rice Riots of 1918 Japan began to systematically import rice and other foodstuffs from its colonies, Korea and Taiwan. The figures on food consumption are based on consumption and not production and therefore take account of imports. For a discussion of imports and food policy before World War II, see Brandt 1993. For a discussion of Japanese food policy during World War II, see Johnston 1953. For a discussion of imports after World War II and estimates of income and price elasticities of demand for various foodstuffs, see Sanderson 1978. Uchino (1977) provides an excellent treatment, based on sur-

veys of households, of the changing nature of diet and of the role of urbanization in the shift toward more "Western" types of food consumption. [BACK]

1. The entitlement concept used in this way is from A. K. Sen. For his use of the notion, see Sen 1987 and Sen et al. 1987. For a discussion of political unrest and the struggles over entitlements in European history, see Tilly 1983. [BACK]

2. My treatment of the establishment and operation of the Tokugawa political system and the consequent expansion in rice cultivation is cursory. For much more detailed treatments, see Beasley 1972; Kikuchi 1986; Ramseyer 1989; Smith 1988; Totman 1967; and Vlastos 1986. [BACK]

3. I am grateful to Hiroshi Kawaguchi of Tezukayama University for explaining this point to me. For a monumental treatment of the development of shinden in Japan, see Kikuchi 1986. [BACK]

4. For a discussion of urbanization during the late Tokugawa period and the expansion of Nagoya in particular, see Mosk 1995b. [BACK]

5. An interesting question whose full discussion would carry us far beyond the thematic concerns of this volume is whether the stem family system came into existence at the end of the early Tokugawa period because population had been growing rapidly over the preceding century. That is, after a century of substantial population growth Japan was reaching a point at which constraints on new land availability characterized by a highly inelastic supply curve for newly developed arable land was limiting the ability of families to divide land or find land for sons and daughters in new villages or to arrange for successful marriages through which sons or daughters leaving the family's estate could expect to enjoy a reasonable level of sustenance. Some scholars, like Smith (1959, 1988), argue that an inelastic supply of new arable land and technological changes led to the breakup of extended families into smaller stem family units, that is, that the stem family was basically a by-product of the new economic environment at the end of the early Tokugawa period, in particular, a by-product of the problem of inheritance in an environment in which land could not easily be divided any further. Other scholars argue that the statistically observed decline in family size whereby small families emerged in the Tokugawa period after a time when large family units were recorded in the temple registers is largely an artifact due to changes in registration practices (see Mosk 1995a for a brief discussion of this issue). In any event, regardless of which viewpoint one adopts, one cannot deny the independent role of culture in creating the stem family system in Japan. The reason is, I think, obvious: there are many societies where new arable land is inelastically supplied and yet they do not develop a stem family system.

For a discussion of the post-1990 diffusion of the samurai model of the stem family system into middle-class and farming populations and the role of the educational system in this diffusion process, see Smith 1983. [BACK]

6. For a theoretical discussion of this problem and empirical evidence concerning the relationship between infant mortality and ages for first marriage for women in Japan, see Mosk 1983. [BACK]

7. That families in Japan used in-adoption of fictive heirs in such instances raises an interesting question as to whether such families could be said to be maximizing reproductive success in terms of the gene pool, an idea put forward by sociobiologists. The issue is conceptually a thorny one, and I raise it here simply as one to consider. [BACK]

8. For recent articles on the practice of infanticide and mortality in Tokugawa Japan, see Cornell 1994; Janetta and Preston 1991; Morris and Smith 1985; Ohta and Sawayama 1994; Saito 1993; Smith 1988; Tomobe 1994. Kawaguchi (1993) gives graphic details from the diary of a household concerning infanticide. For a discussion of the quality of the shumon-aratame-cho * registers, see Cornell and Hayami 1986. [BACK]

9. That the fiefs provided this entitlement insurance is one reason in my opinion why peasants were encouraged to reduce their fertility. The problem peasants face in many agricultural settings is a highly variable probability of adequate harvest. In such a risky setting households are likely to pursue a "safety first" strategy, which means not taking excessive risks in crops planted and avoiding small family sizes, since infants are likely to die in times of dearth. But because of entitlement insurance, even households in regions where harvests were often poor did not tend to pursue a "safety first" strategy as far as fertility was concerned. [BACK]

1. For a treatment of labor markets during the Meiji period which discusses the wage floor issue, the existence of surplus labor in agriculture (i.e., the existence of a labor pool in which the marginal product of an hour's worth of work is positive but in which workers can be removed without loss of output because the remaining workers work harder), see Mosk (n.d.: chap. 2). Mosk (1995c) also discusses the issue of how nonwage payments (like subsidized housing and meals) affect measured wage levels and wage differentials between agriculture and nonagriculture. In principle, none of the wage figures given in table 16 includes the market value of nonwage remuneration. For a discussion of the development of training programs within companies, see Iwauchi and Sasaki 1987. [BACK]

2. Most recruits to the textile mills came from tenant households, and the contracts were typically signed by the head of the household on behalf of the girl involved, a fact demonstrated by Tsurumi (1990). Hence the relevant opportunity cost for the agricultural sector for a female textile worker was the marginal productivity of her labor on tenant farms. I deal with productivity and income differentials between owner, part-owner/part-tenant, and tenant farm households in chapter 5. On the importance of female labor inputs in rural Japan during the late feudal and early modern periods, see Saito 1991. [BACK]

3. For further and more detailed discussion of these issues, see Gordon 1985 and Mosk 1995c. [BACK]

4. My discussion concerning the importing of Western medicine and public health during the late Tokugawa and Meiji periods draws on the accounts in Fukutome 1986; Hirota 1957; Miura 1978, 1980a, 1986; Nihon Koseisho * 1955; Sugaya 1976, 1982; and Sugita 1969. [BACK]

5. A highly restrictive set of relief measures known as jutsukyu * (or jukkyu * ) kisoku (relief regulations) were carried over the bakufu domain and some of the fiefs. But this hardly constituted a program of entitlements since it was highly selective in who qualified for coverage. The law had special provisions for those over seventy and for those under thirteen years of age who did not have an adult taking care of them, and otherwise it emphasized the need for self-reliance. For a discussion of the slow development of entitlement policy during the Meiji period and later, see Adachi 1988; Fukutome 1986; Hotani 1994; and Onishi 1988. Yamanaka (1966) provides a useful history of the development of private and public initiatives in the life insurance industry. [BACK]

6. I classify the prefectures by levels of urbanization here (figures for the prefectures on labor force structure and per capita income are not available for this period), but I wish the reader to be aware that there is a high correlation between levels of urbanization, the proportion of persons employed outside of primary industry, and per capita income. On this point please refer to table 20 and to the discussion surrounding table 20 in chapter 5 below. [BACK]

7. Emi (1963) provides a useful account of national spending on social security and other programs. [BACK]

8. Miura (1980a, 1980b) gives details concerning the various governmental and nongovernmental research agencies set up to study conditions in the factories during the Meiji and Taisho (1915-1916) periods. [BACK]

9. Honda and Shay (1994) demonstrate that throughout the period 1899-1937 average heights of military recruits in agricultural districts (defined as those prefectures with 60 percent of its labor force in primary production in the year 1930) fell consistently below the average for the industrial prefectures (those with less than 40 percent of its labor force in primary production in 1930) and the absolute level of the difference in average heights was actually increasing somewhat throughout the 1899-1937 period. Shay (1994), using the same prefectural data and making comparisons in the ranks of the prefectures according to average height for military recruits for the years 1898, 1902, 1907, 1912, 1917, 1922, 1917, 1932, and 1937, shows that while prefectures producing on average short and tall recruits tended to remain the same, the ranks of the prefectures within the groups of prefectures producing tall and short recruits did vary over time. One problem with both of these analyses is the use of average male height figures: the absolute differences in averages between the prefectures at adulthood tends to be very small. In this context see table 20 in chapter 5.

For an illuminating study of the changing regional pattern of illiteracy in Meiji Japan, which bears on the question about continuity and discontinuity in regional patterns during the era of balanced growth, see Kurosaki 1983. [BACK]

1. For a general history of tenant unions and landlord-tenant disputes in Japan, see Smethurst 1986 and Waswo 1977, 1982. [BACK]

2. Smethurst (1986) argues that rebellion broke out in rural Japan because tenant farmers who had been enjoying an improvement in income for some time

wanted to improve even more their commercial market opportunities by doing away with the practice of paying rents in the form of rice, thereby allowing diversification into a wider range of crops, and because tenants paid excessively high rents (typically about 40 percent of the tenant's crop). At one time high rents might have been justified because of the entitlement insurance provided by landlords, but the demand for such "insurance" had been on the decline for a considerable time because of the secular improvement in productivity per acre. As appealing as the argument may seem, it has been harshly criticized on factual grounds. See in this regard the systematic critique by Nishida (1986). In my view the Smethurst thesis misses the point for two reasons: it focuses on absolute levels of income instead of relative levels of income; and it fails to take into account the fact that many farmers were seeking greater price and income stability through government intervention in the market aimed at guaranteeing minimum levels of income for marginal producers and guaranteeing tenant rights to continue to cultivate land they had been cultivating for many years. Stability of rights to continue to farm land became increasingly important as farm households, responding to the changing terms of trade and the rising relative cost of labor to land, increased their investments in machinery and other forms of capital equipment. For further discussion of Smethurst 1986, see Brandt 1989. [BACK]

3. In tabulations that are not displayed in table 21, I found that small farmers tended to work longer hours per worker than large farmers, and other things being equal, part-owner/pan-tenant households worked longer hours per worker than did owner households. And of the three groups I found that tenant households worked the longest hours per worker. From the last line of panel A of table 21 it is apparent that dependence on hired labor declined for all size and tenancy groups so that the demands of agricultural work were increasingly being met by family members regardless of ownership status. However, in tabulations not reported here, I found that owners used more hired labor than did part-owner/part-tenant households and that tenant households were the least likely to use hired labor. This is one reason tenant household members worked longer hours than did those in part-owner/part-tenant households or in owner households. [BACK]

4. For further details, see Gordon 1985 and Mosk 1995c. [BACK]

5. For a review of the various studies concerning industrial health conducted by governmental agencies and by private or semiprivate research institutes during the interwar period, see Kano 1974 and Miura 1980b, 1981. [BACK]