This is the first report from my current research on the Tokugawa economy and demography which was initiated under a grant from the East Asian Research Center, Harvard and is currently supported by a National Science Foundation grant. I am grateful to Professor T. C. Smith who made valuable criticisms which materially improved the final version of this article and to the Editor of this JOURNAL, whose incisive comments were extremely helpful. To my friend, E. Sydney Crawcour, I express my gratitude for introducing me to Tokugawa economic history.

¹ Most Japanese authors are content to join Nomura who said that: “During the Tokugawa period, the samurai class suffered increasing poverty, and this is evident.” Nomura, Kanetarō, Tokugawa hōkenshakai no kenkyū (A study on the Tokugawa feudal society) (Tokyo: Nikkō Shoin, 1931), p. 3.Google Scholar

² See fn. 33 below.

³ In Tokugawa Japan, the samurai class consisted of the retainers of the Shogun and the retainers of daimyō. The retainers of the Shogun were the bannermen and the housemen (gokenin). The latter were low income retainers who, unlike the bannermen, had no right to an audience with the Shogun. Shogun literally is a military ruler. Daimyō were feudal lords who administered fiefs granted to them by the Shogunate.

⁴ There were a dozen bannermen who were paid in cash. But these were exceptions and limited to those in the lowest income class. Yamaguchi, Tōru, “Bakumatsu ni okeru hatamoto zaisei” (The bannermen's finances during the late Tokugawa period), Shakai Keizai Shigaku, XXVIII, 2 (1962), 164–65.Google Scholar

⁵ Takayanagi, Kaneyoshi, Edo jidai gokenin no seikatsu (The life of the housemen during the Edo period) (Tokyo: Yūzankaku, 1966), pp. 275–85.Google Scholar

⁶ In any case, even had the Osaka price not been the dominant price, we were forced to use it since this was the only series available for a sufficiently long period and, in many cases, the only series available on a monthly basis. The dominance of Osaka as the nation's trading center in rice and other commodities is fully described in Furushima, Toshio, Kindai nihon nōgyō no tenkai (The development of modern Japanese agriculture) (Tokyo: Tokyo University Press, 1968), pp. 61–80.Google Scholar For evidence on the basic similarities of prices in Edo and Osaka before the 1830's, see Crawcour, E. S. and Yamamura, K., “The Tokugawa Monetary System: 1787–1868,” Economic Development and Cultural Change, XVIII (July 1970), 489–518.CrossRefGoogle Scholar

⁷ Takayanagi, The life of the housemen, p. 74.

⁸ The ratios were: 1.245 (1722), 1.183 (1723), and 1.269 (1779).

⁹ This source, Kansei chōshū shokafu, was compiled by the Bakufu officials and was issued in 1812. This 22 volume source was edited and revised by a group of Japanese historians and published by Taibōsha Publishing Company in Tokyo between 1963 and 1966. An original sample of 5,000 was chosen, but it was not possible to obtain any information on the stipends of 44 bannermen. A sample rather than the universe was used because of die extremely time-consuming process required in obtaining necessary information from the genealogies.

¹⁰ The requirement ranged from five (one lancer, one swordsman and three manservants) for a bannerman with 200 koku to 235 (20 riflemen, 10 archers, 30 lancers, 16 swordsmen, 10 cavalry men, 3 banner carriers and 149 manservants) for a bannerman with 10,000 koku. Though enforcement of this requirement was relaxed over time, it remained theoretically in effect until 1862. Shinmi, Yoshiharu, Hatamoto (The bannermen) (Tokyo: Yoshikawa Kōbunkan, 1967), p. 157.Google Scholar

¹¹ How long and how seriously the Bakufu enforced the decrees on the required military retainers are debated subjects among Japanese historians. See Shinmi, The bannermen, p. 124. We know that the requirement was formally changed to cash contributions in 1862 when the inflation of the period made the real burden of the requirement lighter. Kitajima noted that the requirement became “a dead letter” much earlier, but it is not clear exactly when enforcement was relaxed. Kitajima, Masamoto, Edo Bakuftt no kenryoku kōzō (The power structure of the Edo Bakufu) (Tokyo: Iwanami Shoten, 1964), p. 414.Google Scholar

¹² Discussions in this and later sections are based on a theoretical model constructed to evaluate interrelated effects of changes in the price of rice, the price of consumer goods and services, wage levels, the proportion of the stipend which the Bakufu gave out in cash, and the differential in the market and posted price. Basically, the model sees the problem at hand as one of constrained utility maximization over time: U_t = f(R_t, N_t) where R is rice and N is non-rice consumer goods and services. The model, which was not included in this paper for reasons of space, can be furnished to interested readers.

¹³ About 25 percent of the bannermen in the middle to high income classes received their stipends in land. In 1705, there were 1,268 bannermen receiving at least 100 koku but less than 200 koku; 1,386 between 200 and 300 koku; 1,298 between 300 and 400 koku; 339 between 400 and 500 koku; and only two bannermen received as much as 9,000 koku. This skewed income distribution remained essentially unchanged throughout the Tokugawa period. Suzuki, Hisashi, “Tokugawa bakushindan no chigyō keitai” (The patterns of fiefs of the Tokugawa Shogun's retainers), Shigaku Zasshi, LXXI, 2 (Feb. 1962), 144–45.Google Scholar

¹⁴ The decision to use these commodities in constructing the price index was reached on the basis of the limited evidence on the bannermen's expenditure patterns and of the constraints imposed bv the data sources cited in Appendix B. Nomura's A Study on the Tokugawa Feudal Society, pp. 63–120, includes eight sample budgets that were recommended by the Tokugawa essayist, Keikan Hashimoto, in his Keizai zuihitsu (Economic essays) published in 1825. The budgets are for “frugal bannermen” in eight different income classes between 100 hyō and 500 hyō. Though there are slight differences by income class and though the amount spent on salt cannot be separated from expenditures on kindling wood, the importance of these five items and their relative weight are sufficiently apparent. Three actual budgets included also support the above observation: one for a bannerman with a stipend of 1,500 koku; one with 500 koku; and the last for a bannerman whose income could not be established. The years represented were 1778, 1781, 1861. Of course, the higher the income and the later the time period, the smaller was the proportion of total income spent on these five items. The budgets described above indicate that charcoal and candles were equally important. However, prices of these two commodities are not available.

¹⁵ These estimates of the coverage of the indices are based on sources cited in footnote 14, in the footnote of Table 2 and in other budget studies referred to in this article. For qualifications, see footnote 14.

¹⁶ A deflator constructed with a weight of seven for cotton cloth does not yield a real income index which is sufficiently different to necessitate revisions in our conclusions.

¹⁷ Honjō, Control of the rice price, and Nakazawa, A history of rice price fluctuations (both fully cited in Appendix A) have good descriptive accounts of price changes.

¹⁸ The source is: Kyoto Daigaku Kinsei Bukkashi Kenkyukai (The Kyoto University group to study prices in the 15th-17th centuries), Jūgō-juhichi seiki ni okeru bukka hendō no kenkyū (A study on the price fluctuations in the 15th-17th centuries) (Kyoto: Dokushikai, 1962).Google Scholar This source includes prices of nearly 150 commodities up to 1652. Most of the data were collected from records preserved in shrines and temples in Kyoto. These were retail prices. Though the prices fluctuated for bean paste, soy sauce, salt (pp. 95–7) and other commodities consumed daily, there were no discernible upward trends.

¹⁹ The source is Changes in major prices during the late modern period, cited in Appendix A. This source includes daily wages for. carpenters, masons, tatami [straw mat] artisans and tema [unskilled helpers]. None of the series is consecutive but all series have entries in sufficient frequency. Also see, Hayami, Akira, Nihon Keizai eno shikaku (A viewpoint on Japanese economic history) (Tokyo: Tōyōkeizai Shimpōsha, 1968), pp. 150–55Google Scholar, which discusses articles published on the wage levels of live-in servants in the peasant households of Kantō villages.

²⁰ The budgets cited in footnote 14 show this. Also the data on one budget discussed in detail later in this article show that wage expenditures constituted one of the most important items that did not rise later in the Tokugawa period. A bannerman of nearly 2,000 koku experienced the following changes in wage expenditures: If we consider the wage expenditures of 1795 to equal 100, they were as high as 300 in 1711 (wage expenditures could have been higher than normal, but we have no data for adjacent years), fluctuated between 74.5 and 134.0 between 1796 and 1805, and rose steadily during the last few decades of the Tokugawa period but at a rate below that of the inflation experienced during the period. Yamaguchi, “The Bannermen's finances during the late Tokugawa period,” pp. 164–65.

²¹ Though there are many sources both in English and Japanese, for an English source that surveys many recent findings, see Hanley, S. B. and Yamamura, K., “A Quiet Transformation in Tokugawa Economic History,” Journal of Asian Studies, XXX (Feb. 1971), pp. 373–384.CrossRefGoogle Scholar

²² Kimura, Motoi, Kinsei no shindenmura (Tokugawa villages on newly reclaimed land) (Tokyo: Yoshikawa Kobunkan, 1964), p. 66.Google Scholar

²³ Any mention of shinden is rare in The Kansei revised samurai genealogies cited in fn. 9.

²⁴ These observations are well established in Kimura, Tokugawa villages, pp. 25–73.

²⁵ The bannermen lived in Edo and for military reasons the Bakufu gave them fiefs near Edo. Only a few bannermen in high income classes had fiefs distant enough that they had to be given permission to leave Edo at least once a year to visit them. The large majority of the bannermen holding fiefs had their fiefs in the Kantō region. The above, which is clear from an examination of The Kansei revised samurai genealogies, is supported by Fujino, Tamotsu, Bakuhan taiseishi no kenkyū (A study on the history of the Bakufu-han system) (Tokyo: Yoshikawa Kōbunkan, 1961)Google Scholar, which notes the concentration of bannermen's fiefs in “the eight kuni [provinces] of the Kantō region,” p. 259.

²⁶ Though the data are not wholly reliable, there is no doubt that the Kantō region lagged in increases in agricultural productivity vis-à-vis the western parts of Honshū, and parts of Kyūshū and Shikoku islands. This observation is supported fully in Hanley, S. B. and Yamamura, K., “Population Trends and Economic Growth in Preindustrial Japan,” in D. V. Glass, editor, Historical Population Studies (1971)Google Scholar, forthcoming.

²⁷ Though daimyō were given considerable freedom in setting the tax rates on rice yields within their daimates, the bannermen were not. See fn. 31.

²⁸ This is a large subject on which literally hundreds of books and articles have been written. Interested readers should examine: Hanley and Yamamura articles cited in fns. 21 and 26; Kitajima, The power structure, pp. 533–4; Oishi, Shinzaburō, Kyōhō kaikaku no keizai seisaku (The economic policies of the Kyōhō reform) (Tokyo: Ochamomizu Shobō, 1961), pp. 121–67Google Scholar; Shinmi, The bannermen, pp. 98–99; and, Smith, T. C., “The Land Tax in the Tokugawa Period,” in Hall, J. W.and Jansen, M. B., editors, Studies in the Institutional History of Early Modern Japan (Princeton: Princeton University Press, 1968), pp. 283–99.Google Scholar

²⁹ A recent article includes data on cash income received from the sale of rice received by a bannerman. The data available are for 1711, for 1795–1805, and for the post-1829 years. The data show that after adjustments were made for changes in the rice price and the advance payment (borrowing by the bannerman), the amount of rice received was remarkably stable. An index of rice received (1795 = 100) fluctuated between highs of 125 (1801) and 120 (1711) to a low of 69 (1805). The remainder were within the range of 89 and 110. The same source also shows that 8 other bannermen, who shared 7 villages in the Kantō region in which the bannerman in question, a Master of Court Ceremonies, had his fief, experienced the same stability in real income from their tax rice. Yamaguchi, “The bannermen's finances during the late Tokugawa period,” pp. 158–9.

³⁰ See fn. 31. Also, Yamaguchi, Ibid., pp. 168–72, cited two interesting records of exchanges of agreements between bannermen and the peasants in their fiefs. These bannermen agreed to strongly worded peasants’ “requests” to reduce their household expenses in exchange for advance payments of tax rice and for loans. The bannermen also agreed to “an examination” of their expenses by the peasants. One of the documents, signed by the leaders of the peasants, stated: “All these measures are requested in view of your outstanding debts, and we hope that you will give due consideration to the matters stated in this agreement,” p. 172.

³¹ As early as 1602, the Bakufu decreed that all magistrates (daikan) who administered Bakufu fiefs and the bannermen were to obey all the directives pertaining to the rice tax (nengu). A clear statement authorizing direct appeals by peasants was also issued at the same time. By the Kan'ei period (1624–73), the peasants' appeal procedures were well established. For a discussion of the Order of 1602, the appeal procedures, actual cases of appeals, and further discussions on the related questions, see: Sasaki, The basic structure, pp. 232–35, and pp. 353–57.

³² See the sources cited in fns. 29 through 31.

³³ The degree of explicitness differs, of course, by author. But more recent books tend to be more explicit and less ambiguous than the prewar writers. A prewar example is Nomura who was cited at the beginning of this article, and an example of the most recent literature is Kitajima who observed, on the basis of descriptions of contemporary writers, that: “One can think of many reasons why the impoverishment of the bannermen became manifest about the Kan'ei period (1624–43). But, the most important reason could be an increased consumption prompted by the fact that the bannermen (by this time) had gathered in Edo to live. Kitajima, The power structure, p. 370.

³⁴ Kitajima, Ibid., p. 436.

³⁵ Ibid., p. 436.

³⁶ Ibid., p. 370.

³⁷ Sasaki, Junnosuke, Daimyō to hyakushō [Daimyō and peasants], Vol. XV of Nihon no rekishi [A history of Japan] (Tokyo: Chūō Kōronsha, 1966), p. 169.Google Scholar

³⁸ Ibid., p. 162.

³⁹ See fn. 10 above.

⁴⁰ A brief but excellent description of the period is found in Kitajima, Masamoto, Edo jidai [The Edo period] (Tokyo: Iwanami Shoten, 1958), pp. 125–40.Google Scholar

⁴¹ See Appendix A.

⁴² Naramoto, Tetsuya, Chōnin no jitsuryoku [The real power merchants], Vol. XVII of Nihon no rekishi [A history of Japan] (Tokyo: Chūō Kōronsha, 1966), p. 176.Google Scholar

⁴⁸ Nomura, A study on the Tokugawa feudal society, p. 64. The grants began at 30 ryō for the bannermen in the 500 koku class and decreased almost proportionally according to the income level of the bannerman.

⁴⁴ Kitajima, The power structure, p. 570.

⁴⁵ Ibid., p. 570.

⁴⁶ Kada, Tetsuji, Bushi no konkyū to chōnin no bokkō [Impoverishment of the samurai and the rise of merchants] (Tokyo: Ogawa Shoten, 1961), p. 264.Google Scholar

⁴⁷ Kuranami, Shōji, Edo jidai no shihai to seikatsu [Authority and life in the Edo period] (Tokyo: Sanwa Shobō, 1967), p. 123.Google Scholar

⁴⁸ Yamaguchi, “The bannermen's finances,” p. 158.

⁴⁹ Ibid., p. 158.

⁵⁰ Ibid., pp. 164–5.

⁵¹ Takimoto, Seiichi, Nihon hōken keizaishi [An Economic history of feudal Japan] (Tokyo: Maruzen, K. K., 1930), p. 350.Google Scholar

⁵² Ibid., p. 531.

⁵³ Yamaguchi, “The bannermen's finances,” pp. 164–5.

⁵⁴ Yasuoka, Shigeaki, “Kinsei no kinyū soshiki” [Financial institutions in the Tokugawa period], in Toyota, T. and Kodarna, K., editors, Ryūtsūshi [A history of trading] (Tokyo: Yamakawa Shuppansha, 1969), p. 332.Google Scholar During the latter half of the eighteenth century, the Bakufu issued further decrees to alleviate the worsening finances of its retainers. Following the decree of 1744, described earlier, another Mutual Settlement decree was issued in 1746 and it was followed by the Abrogation Decree (kienrei) of 1789. The 1789 decree absolved all the loans made by bannermen and housemen from rice-jobbers during the preceding six years. Again in 1796 another Mutual Settlement decree was issued. Several more such decrees were issued during the first half of the nineteenth century. Kitajima, The power structure, p. 570.

More evidence of the increased dependence of the bannermen on loans is indicated by the profitable and flourishing loan business enjoyed by the rice-jobbers after the Kyōhō period. Kuranami in his Authority and life, p. 570, wrote: The rice jobbers' fee for going to the Bakufu warehouse and receiving rice on behalf of a bannerman or a houseman was about one bu per 100 hyō. Their fee for marketing the rice was about 2 bu for the same amount of rice. However, what enabled them to live, during the last half of the Tokugawa period, a luxurious and splendid life known as the kuramaefū (literally, the “in front of the warehouse mode ) and what made a share in the rice jobbers' guild worth 1,000 ryō were not these fees. Their wealth was due to the interest earned from the loans they made to bannermen. That is, their prosperity rested on money-lending. As samurai finances deteriorated after the Genroku and the Kyōhō periods, rice-jobbers fully exploited their close relationship with the housemen and bannermen. The rice-jobbers were these retainers' financiers, and the former controlled the economic life of the latter.

The average annual income of the rice-jobbers was, according to Naramoto, The real power of merchants, p. 389, no less than 1,000 ryō. This was approximately equal to the income of the handful of bannermen receiving a stipend of at least 5,000 koku. Also see for similar assessments, Sasaki, Daimyō and peasants, p. 168. Nomura, A study of the Tokugawa feudal society, pp. 121–30 provides further evidence and discussion.

⁶⁵ Yamaguchi, “The bannermen's finances,” pp. 164–5.

⁶⁶ Evidence demonstrating that wants increased can be gathered from a wide variety of sources ranging from the budget data cited in fn. 14 to references in contemporary literature such as in the stories of Saikaku. For direct and indirect evidence, see the sources cited in fns. 11, 31, 33, 37, 40, 42, 43, and 47.

⁵⁷ Only about 200 out of 17–18,000 housemen received a stipend as high as 200 koku. Their modal income was within the range of 30 to 70 hyō Takayanagi, The life of the housemen, pp. 177–80.

⁵⁸ Good studies are: Sasaki, Junnosuke, “Han kashindan no tenkai katei; Suwa han o sozai ni” [“The Process of the establishment of han-retainer groups; using the case of Suwa-han”], Shakai Keizai Shigaku, XXVIII, 1 (1962);Google Scholar and Fujino, A study of the history of the Bakufu-han system, pp. 499–690. There are about 20 case studies covering nearly 55 han.

⁵⁹ For a good quantitative analysis, see Fujino, A study of the history of the Bakufu-han system, pp. 666–72.

⁶⁰ A typical example is Hirato han with a fief of 65,000 koku which is slightly above the average size of a han. For most of the eighteenth century and the nineteenth century before the Restoration, about 130 out of the total retainers of 390 received stipends within the range of 10 to 30 koku and all but 30 to 35 retainers received stipends of less than 200 koku. Fujino, A study of the history of the Bakufuhan system, p. 532.

⁶¹ The sankin kōtai system of the Bakufu required daimyo to reside in Edo every other year.

⁶² There are many sources that deal with peasant resistance to higher taxes and other restrictions imposed on them. But, the most eloquent among them is a long list of peasant revolts and protests between 1640 and 1866 appended to Gakkai, Osaka Rekishi [The Osaka Historical Study Group], editors, Hōken shakai no mura to machi [Villages and towns in the feudal society] (Tokyo: Yoshikawa Kōbunkan, 1958).Google Scholar

⁸³ For a good description and examples, see Naramoto, The Real power of merchants, pp. 294–99.

⁶⁴ By present day standards, the growth of agricultural productivity was limited even in the most rapidly developing regions of Japan. The annual average rate of growth of grain output, for the period between 1645 and 1873, measured in terms of the rice equivalent koku unit, was about .3 percent for the western part of Japan. In the slower growing regions, the northeastern parts of Japan, it was somewhere around .12 percent. These, of course, are very crude estimates based on the data gathered by nan. See also the source cited in fn. 26.

⁶⁵ Takayanagi, The life of the. housemen, p. 100.

⁶⁶ Noting a rapid increase in by-employments by the samurai, Ryūen Warashina wrote circa 1770: “[Samurai] sold vegetables, made pottery for sale—and all for gain. All suffered from poverty.” Seifū Murata (1746–1811) wrote: “For years now, the samurai have suffered from poverty and their minds have been occupied by making a living. ‘Buy this, sell that’ and ‘pawn this to pay for that’ has become all of their lives. Even for those dedicated to their duties, it was inevitable to debase themselves and to engage in unsavory conduct (i.e., engage in trading)… . Even the wives of those who were earning as much as 200 koku busied themselves in trading and in shops.” Warashina is quoted in Naramoto, The real power of merchants, p. 395, and Murata is quoted in Taldmoto, An economic history of feudal Japan, p. 351.

⁶⁷ Andō, Selichi, Edo jidai no nōmin [Peasants in the Edo period] (Tokyo: Shibundō, 1967), pp. 194–5.Google Scholar