¹ kyōkai, Nihon shiseki, Hansei ichiran [Summary of han governments], 2 vols. (Tokyo, 1928–1929Google Scholar). The entries are arranged alphabetically (iroha order) by name of domain.

² Specific instances will be given below, as they become relevant to particular points of the analysis. There is one point which might be made here, however, since it refers to the question of comparability. Not all the details given in the returns are for the same date. For example, most of the revenue returns are based on five-year averages, but they are not always for precisely the same years (though the differences are small). Some domains gave ten-year averages, others gave no indication what practice they had followed. I have made no attempt to compensate for these differences, since it would be extremely difficult to do so, and there is in any case no indication that one period of years would necessarily be more typical than any other. I have not used returns which contained evidence that they had been seriously distorted by the occurrence of some special event like flood or typhoon.

³ There were more than 260 domains during most of the Tokugawa period, the number varying slightly from time to time. It is difficult to determine exactly how many existed at the time of the Meiji survey, since a good many changes were made as a result of the civil war early in 1868, but Hansei ichiran includes returns from 281.

⁴ In all tables—and generally throughout the article—the term fudai should be understood as including Tokugawa branch houses.

⁵ As a check on the reliability of the regional variations revealed by the figures used in this article, averages were also worked out comparing the domains by type (tozama or fudai) and size (over 10,000; over 50,000; and over 100,000 koku respectively). The only consistent variation to emerge was that the Kusadaka figures (see next section) were higher for tozama than fudai. For the rest, the results varied somewhat from the overall regional averages, but for each set of figures there was a regional pattern not so very different from diat given by analysis of the whole.

⁶ Eight of the domains in the sample are of this kind: those numbered 5, 11, 15, 18, 30, 55, 72 and 117 in the Appendix. It is worth noting that these domains do not appear to have any special distinguishing characteristics which would mark them off as a group from the others, at least in terms of the figures for yield, revenue and population with which we are concerned in this article.

⁷ See Smith's, Thomas C. article, based on the tax records of II villages in different parts of Japan, “Land Tax in die Tokugawa Period,” Journal of Asian Studies, XVIII (1958), 3–19.CrossRefGoogle Scholar On Tokugawa agriculture generally, widi special reference to increases in production, the reader is referred to die same author's recent book, The Agrarian Origins of Modern Japan, Stanford, 1959.

⁸ This conclusion is supported by the facts cited in Thomas C. Smidi's article, noted above: in almost all the villages he studied the assessed yield was constant or nearly so from the 17th to the 19th centuries, while at least one recorded increase in assessment was due to the addition of new arable to die tax rolls. See also Shikitaro, Ōyama, “Bakumatsu ni okeru denso oyobi jōnōkin,” in Bakumatsu keizaishi kenkyū (Tokyo, 1935), 298–374Google Scholar, for similar evidence concerning some Tokugawa estates.

⁹ Column (v) of die Appendix gives figures for individual domains. Local and regional averages are given in Table 2 and in graph form in Graph A. The averages given were actually prepared by aggregadng the relevant figures in koku, not by averaging the separate percentages for each domain. The results are therefore weighted in accordance with die size of domains. The same method has been followed in arriving at all the figures given as averages in Table 2.

¹⁰ In describing this distinction as being between a sort of land tax on the one hand and mostly commercial taxes on the other, I am generalising from an inspection of the returns and undoubtedly over simplifying the picture. See Smith's, Thomas C. note on komono-nari, “Land Tax,” p. 9.Google Scholar

¹¹ All figures used’ in preparing these notes were reduced to koku of rice (disregarding items which amounted to less than 1 koku). Where it was necessary to convert from cash to rice, the following equivalents were used: 80 kamme copper = 1 koku rice; 1 kamme silver = 1.5 koku rice; 8 ryō gold = 1 koku rice. The last, especially, differs sharply from the rate (1 ryō = 1 koku) which is usually accepted for earlier years; but it is the rate most commonly found in the returns and may reflect die sharp rise in prices after 1864. It seems probable in any case that the conversion rates used by the domains were approximate and conventional, adopted for bookkeeping purposes.

¹² It is possible, for example, that these domains have included as additional revenue the returns from special levies on samurai and farmers. Such levies were widespread in this period. They were euphemistically described as loans by domain governments, but constituted in fact a more or less permanent addition to normal dues. Since they are not specifically mentioned in the returns it is not clear whether they were treated as regular or as additional revenue. The former is more likely, but practice may well have differed from place to place.

¹³ Smith, “Land Tax,” 11–14. It is interesting to note that the villages he cites vary considerably in the rate of tax they bore. There were even two in the same domain, Kōriyama, of which one paid about 45% and the other less than 40%. According to the Hansei ichiran figures (No. 57 in the Appendix) the average for Kōriyama was 35.48%.

¹⁴ By this method of calculation, Yamaguchi (No. 30 in the Appendix), at just over 56%, comes very near its regional average, while Yonezawa (No. 110 in the Appendix), at approximately 36%, still remains low by comparison with other domains in its area.

¹⁵ Despite the unreliability of the figures, it is worth noting that a graph drawn to show feudal class as a percentage of total population proved to have a general shape not unlike that of Graph C. A few figures in south Kyushu were very high (25% or more), with an average of just under 10% elsewhere in that island and in Shikoku. In western Honshu figures varied about approximately 7%, in the central region about 6%, and the Tokaido and Kanto about 5% or a little under. North along the Japan Sea coast the average was something like 7% again, with a sharp rise in the extreme northeast to 15% or even more. A further graph, drawn to show total population in relation to kusadaka, showed wide individual variations (round a mean figure not far from the traditional “one person per koku,” the variations being greater in the northeast and southwest than in the centre), but there was no apparent regional pattern.

¹⁶ The matter is almost certainly more complex than this, however, for there is no direct correlation between fudai status and a low level of samurai population in the domain. On the other hand, fudai territories varied less widely in this matter than did tozama. Separate regional averages (starting from Kyushu on the left) are as follows: tozama: 100.9, 64.0, 56.0, 34.6, 65.3; fudai: 77.2, 79.8, 50.2, 49.7, 69.6.

¹⁷ This is still more marked if one takes the figures for ordinary revenue as a percentage of hondaka, since this eliminates the irregularity due to exceptionally high kusadaka and low tax rate in Chōshū.

¹⁸ Takao, Tsuchiya, Hōken shakai hōkai katei no kenkyū (Kyoto, 1927Google Scholar).

¹⁹ Reliable figures cannot be worked out for Kagoshima (Satsuma), but the following rough approximations may be of interest: kusadaka as percentage of hondaka, 112%; Samurai (excluding sotsu) per thousand koku 281; Ordinary revenue as percentage of kusadaka, 34%. However, this figure for ordinary revenue excludes, among other items, the returns for the sugar crop, which is normally regarded as the mainstay of Satsuma finances.

²⁰ This time the average for the northeast is not unduly distorted by the return for Kanazawa. Even if one excludes Kanazawa, the tozama average for the region is still 54.7%.