I am much indebted to Moses Abramovitz, Paul David, Bruce Johnston, Lawrence Lau, Koji Taira, and Robert Ward for reading an early version of this article and making useful suggestions, but I am solely responsible for any errors of judgment or fact.

¹ Kaempfer, Engelbert, The History of Japan (Glasgow, 1906) II, 297Google Scholar.

² Thunberg, C. P., Travels (London, 1795) III, 106Google Scholar.

³ Yamazaki Ryūzō, “Edo kōki ni okeru nōson keizai no hatten to nōminsō bunkal,” Nihon rekishi. Kinsei (Tokyo, Iwanami) IV, 347.

⁴ Norifumi, Watanabe, “Jūshū enden,” Nihon sangyō shi taikei (Tokyo, 1960) VII, 26–32, 38, 55–58Google Scholar; Yoichi, Kodama, Kinsei enden no seiritsu (Tokyo, 1960), p. 562Google Scholar, for the development of the salt industry in a neighboring district that throws light on Kaminoseki.

⁵ Shinobu Seizaburō, Kindai Nihon sangyō shi josetsu (Tokyo, 1942), pp. 47–49Google Scholar; Jun'ya, Seki, Hansei kaikaku to Meiji ishin (Tokyo, 1956), pp. 25–34, 69Google Scholar.

⁶ Edited by the Yamaguchi Prefectural Archives; hereafter cited as BFC.

⁷ BFC, V, VI.

⁸ There were a number of other districts consisting of small off-shore islands which have /been excluded from consideration here by reason of their peculiar ecology.

⁹ The term machi seems to have been used in at least two senses in the Tokugawa period: first, to designate places with a greater density of population than villages or mura and with more conspicuously commercial functions; and, second, to designate places under the administration of a machi bugyō (or town magistrate) and to which “town” (machikata) rather than “village” (mwakata) law applied. Machi in the first sense were not necessarily machi in the second or legal sense, although machi in the legal sense were probably almost always machi in the social sense. The machi in Kaminoseki were clearly towns in the first sense, though very small ones, but it is not clear whether they were also legal towns. kyōgikai, Chihōshi kenkyū, ed., Nihon no machi (Tokyo, 1958), pp. 306–14Google Scholar.

¹⁰ BFC, V, 279.

¹¹ Ibid., p. 288.

¹² There is no description of the machi in Hirao, which, however, must have been very similar to the town part of Murotsu, also a port and described in detail. Ibid., VI, 243.

¹³ Ibid., V, 280.

¹⁴ Excluding a total of 135 eta or outcast families who were not included in output, expenditure, or food consumption figures; and also excluding a total of 350 samurai family heads who, like the eta, were listed as present in the districts but not included in output, expenditure, or food consumption figures. See appendix.

¹⁵ Districts with machi were Ono, Hirao, Murotsu, Befu, Shimotafuse, Hano, and Sone.

¹⁶ Assuming, that is, the productivity of all families in each district was the same, so that the proportion of nonfarm families and of nonagricultural income would be identical.

¹⁷ Agricultural output was listed down to the most minute quantities of fruits, nuts, and herbs, for example, while there were obvious omissions on the nonagricultural side, certain nonfarm families such as ishi (doctor) being listed for whom no income was recorded, and for some items of nonagricultural income only that part sold outside the district being recorded. See appendix for the treatment of these problems.

¹⁸ The weight of the land tax varied by district but ran about 45 percent of the value of output on average. Salt production was taxed at a roughly comparable rate, but textile production was untaxed. Of course, strictly speaking, the land tax, though paid in rice, was levied on the total economy of farm families and not on their farming only. Conscious efforts were made by officials elsewhere to take nonagricultural production into account in setting the land tax, and one may assume that this possibility was not overlooked by Choshu tax officials.

¹⁹ Figures are difficult to come by, but we have estimates of product per worker Agriculture/Nonagriculture for Japan in 1878–99 (0.345), and in the U.S. in 1839–49 (0.37) and in 1869–88 (0.47). Mimeographed “Summary” of an International Conference on Economic Growth—A Case Study of Japan's Experience, September 5–10, 1966, p. 2. For indirect indications of relative productivity in farming and other employments (depending on wage comparisons and the difficulty of attracting labor out of agriculture) in England during the eighteenth and early nineteenth centuries, Redford, A., Labour Migration in England (Manchester, 1926), pp. 54–69Google Scholar; Mantoux, Paul, The Industrial Revolution in the Eighteenth Century, rev. ed. (London, 1948), pp. 430–35Google Scholar; Landes, David, “Technological Change and Development in Western Europe, 1750–1914,” The Cambridge Economic History of Europe (Cambridge, 1965), VI, 344–47Google Scholar.

²⁰ The average wage remittance per worker for the whole county was taken as an index of nonagricultural productivity, and one-half the average agricultural income of farm families in the county (assuming an average of four members per family, composed equally of workers and dependents) as an index of agricultural productivity. The per worker wage remittance was then reduced, for purposes of comparison, by one-third on the assumption that the nonagricultural workers who earned them worked one-third more days a year than farmers. This gave a productivity ratio of 1.5:1 on a man-day basis. It does not take account of the value of the food and shelter earned by the remitting workers; but this factor was probably offset by a certain amount of unreported income among farmers (including the use-value of their housing).

²¹ See the case of the craftsmen-farmers described in the text. For a description of a country town in another county in which it is explicitly stated that many town merchants and artisans farmed part-time, see BFC, XVI, 123.

²² For some examples, see ibid., V, 38, 88, 136, 190.

²³ See particularly the clear statement in the Ogo returns, ibid., p. 38.

²⁴ Ibid., p. 296.

²⁵ A manuscript manual on salt making from Choshu, evidently written early in the eighteenth century, offers wage and other cost data to support the contention that a farm family could break even if it worked no more land than it could cultivate with family labor, but would suffer a net reduction of income by cultivating additional land, owing to the high cost of hired labor. Seki, Hansei kaikaku, p. 63.

²⁶ BFC, XVII, 145, 377.

²⁷ Agricultural income per farm family for the county as a whole was 562 momme.

²⁸ Yosan suchi, mss., National Diet Library, unpaginated.

²⁹ Eitarō, Tamura, Ōkura Nagatsune (Tokyo, 1944); p. 29Google Scholar.

³⁰ Noboru, Nomura and Kitarō, Yoshii, eds., Kinsei shomin shiryō (Osaka, 1955), p. 194Google Scholar. This volume consists of the headman-author's jottings over a lifetime, ranging from hints on agriculture and observations on government to brief sketches of the history of various families in the village—all of extraordinary interest.

³¹ For example: “The silk worm being a living organism requires constant attention. The slightest neglect results in immediate loss. Raising cocoons is the more demanding, too, because it overlaps with the spring wheat harvest, and the cocoons require the most attention just before the summer rice planting. After that, reeling must be immediately begun; one day's delay results in loss. Then, when it is time for weeding the rice, and the weather is unbearably hot, the summer cocoons mature and must be reeled. In these busy seasons, the sericulturalist faces one job after another and is so busy he scarcely knows whether it is night or day.” Yosan kensetsu published in 1813; in Sansō kōhen shūsei (Tokyo, 1930), p. 333Google Scholar. For other contemporary descriptions of the intensity of work entailed by combining farming and by-employments, Eitarō, Tamura, Ōkura Nagatsune (Tokyo, 1943), p. 29Google Scholar; BFC, V, 38, 88, 136, 190.

³² Smith, Thomas C., The Agrarian Origins of Modern Japan (Stanford: Stanford University Press, 1959), pp. 112 ffGoogle Scholar.

³³ Holdings were small throughout the county; the smallest average in a district was 1.8 tan of arable in Murotsu and the largest 5.9 tan in Usagi. (1 tan = 0.245 acre.) BFC, V, 38, 88, 136, 190, 227, 263, 296, 338, 382; VI, 75, 127, 170, 210, 245.

³⁴ Ibid., V, 228.

³⁵ ibid., V, 296.

³⁶ Ten of fifteen villages in this report listed income from wage remittances, in four cases accounting for 19, 17, 15, and 8 percent, respectively, of total income. See Table 5, column (6).

³⁷ Stereotyped as the ”sanyo” or “three intervals” of fanning free for other employments— especially for study, according to classical writers.

³⁸ For example, mountain villages clearly exported straw ropes and mats as well as charcoal and firewood to coastal villages which made salt, receiving in return the numerous services concentrated in the latter.

³⁹ Hano held markets in the machi that were curiously bunched on the tenth, fourteenth, and nineteenth of each month, when “straw mats, hats, bamboo, cotton cloth, and other industrial commodities are brought to the market from surrounding villages for sale.” BFC, V, 229.

⁴⁰ BFC, V, 172, 218; VI, 243. Markets (ichi) were few and infrequent; there were only three in the entire county as compared to at least six machi and perhaps more; and some markets met as rarely as twice a month.

⁴¹ Exports could not be calculated for agricultural products in the way explained in the text owing to insufficient disaggregation on the consumption side. Therefore a very imperfect method was adopted—counting as exports all items of agricultural output, the value of which was given in money rather than kind only—a procedure undoubtedly overstating exports since, though such commodities were probably sold rather than consumed at home, there is no way of knowing where they were sold.

⁴² Every district reported its food needs calculated at a certain average intake of rice and rice equivalents per person, excluding persons working outside the district, samurai and eta; and each district also translated its own food production, in its many forms, into rice equivalents. BFC, V, 275, 157, 310, 353, 94, 398, 240, 207, 436; VI, 112, 61, 216, 132, 182, 253.

⁴³ For example, Hirao produced 285,000 momme worth of salt and imported firewood worth 14,040 and coal worth 102,525 momme. Ibid., V, 296–97.

⁴⁴ Ibid., VI, 185.

⁴⁵ Ibid., V, 67–68, 97–98, 160–61, 210–11, 243–44, 277–78, 315–16, 357–58, 400–1, 438–39; VI, 115–16, 134–35, 186–87, 219–20, 264–65.

⁴⁶ Ibid.

⁴⁷ Ibid., V, 88, 136.

⁴⁸ For example, Shimotafuse, a district with no salt industry, produced approximately 29,000 straw mats, nearly 14,000 bundles of straw ropes and 1,593 loads of firewood—all items used in making salt in places like Hirao, where they were imported in large quantities. Ibid., V, 136, 207–8.

⁴⁹ In some districts income from dyeing was divided between dyeing done for people inside and outside the district; in other districts only that done for people outside was listed. BFC, VI, 114, 184; V, 64.

⁵⁰ Kidota Shirōgyō kōzō (Tokyo, 1960), p. 53Google Scholar.

⁵¹ Ashikaga orimono (Tokyo, 1960), I, 22Google Scholar.

⁵² Yao shishi. Shiryōhen (Yao, 1960), p. 317Google Scholar.

⁵³ Akita kenshi. Shiryo kinsei (Akita, 1963), pp. 852–53Google Scholar; Tokio, Mihashi, “Edo jidai ni okeru nogyō keiei no hensan,” Mataji, Miyamoto, ed., Nōson kōsō no shiteki bunseki (Tokyo, 1955), p. 16Google Scholar; Yūkoku, Fujita (1774–1826), “Kannō wakumon,” Nihon keizai taiten, XXXII, 228Google Scholar.

⁵⁴ Ryōsuke, Ishii, ed., Hampōshū: Tottori han and Tokushima han (Tokyo, 1961–1962), II, 155–237Google Scholar; III, 774–927.

⁵⁵ Seiichi, Andō, Kinsei zaikata shōgyō no kenkyū (Tokyo, 1958), pp. 248–335Google Scholar.

⁵⁶ Kentarō, Nomura, Mura meisaichō no kenkyū (Tokyo, 1949), pp. 106–51Google Scholar.

⁵⁷ Geihan tsūshi (Hiroshima, 1907–1915), II, 741–49Google Scholar.

⁵⁸ Smith, Thomas C., “Landlords and Rural Capitalists,” Journal of Economic History (June 1956), 165–68Google Scholar.

⁵⁹ Sangyō shi taikei, VII, 85.

⁶⁰ Siebold, Phillip Franz von, Nippon Archiv zur Beschreibung von Japan (2d ed.; Leipzig, 1897), I, 182Google Scholar. I am indebted to David C. Evans for this reference.

⁶¹ Ibid., VII, 83; Ashikaga orimono, I, 226; Seizaburō, Shinobu, Kindai Nihon sangyō shi josetsu (Tokyo, 1942), pp. 15–16, 27, 64–75, 185–207Google Scholar; Hideichi, Horie, Meiji ishin no shakai kōzō (Tokyo, 1954), pp. 194–209Google Scholar; Norifumi, Watanabe, Hiroshima ken engyō shi (Hiroshima, 1960) 75–99Google Scholar; Shiso, Hattori, Nihon manufuakucha shiron (Tokyo, 1937), pp. 3–39Google Scholar; Hiroshi, Higuchi, Nihon tōgyō shi (Tokyo, 1961), pp. 160–62Google Scholar; Nihon sangyō shi taikei, I 182–88; Sekai rekishi jiten (Tokyo, 1960), XXII, 386–92Google Scholar.

⁶² Nakamura Masao, Amakusa no mura meisaichō, Kyushu bunka shi kenkyūjo, ed., Kiyō, no. 12, p. 279.

⁶³ Mitsuo, Oka, Hōken sonraku no kenkyū (Tokyo, 1962), pp. 152–53Google Scholar.

⁶⁴ Geihan tsūshi, III, 893.

⁶⁵ Hoi, Aicki ken chihō shi hensan iinkai, ed., Mikawa no kuni Hoi chihō shumon fimbetsu aratamechō (Toyohashi, 1961), pp. 557–64Google Scholar; Hayami Akira, “Shinshū Yokouchi mura chōki jinkō tōkei,” Keizai gaku nempō, no. 10, Bessatsu, p. 70.

⁶⁶ Furushima, “Shosangyō hatten no chiikisei,” Nihon sangyō shi taikei, I, 366–76.

⁶⁷ For contemporary impressions of the region, “Chūkoku Kyūshū kikō,” Satō Nobuhiro zenshū (Tokyo, 1927), III, 699–702Google Scholar.

⁶⁸ Oshima, Kumage, Mitajiri and Ogōri. “County” is used here to refer to administrative districts parallel to Kaminoseki.

⁶⁹ Some entire prefectures such as Osaka, Kyoto, Tokyo, Aichi, and Hyogo were probably economically well in advance of Kaminoseki.

⁷⁰ Akita kenshi, Shiryo kinsei, pp. 491–92.

⁷¹ Another factor was that by-employments, unlike the extension of arable, though not the expansion of urban population, did not necessarily give rise to an increase in food supply so that famine continued as a population check. One might object that this is inconsistent with a rise in per capita income, which should have resulted in a relative rise in food prices, channeling additional resources into food production. If this occurred, however, it was not on a sufficient scale to prevent the large-scale loss of life through food shortages for the following reasons: (1) much of the added income went to high income families with no desire to consume more food in normal years; (2) their added income was used in part to bid up the prices of nonfood commodities (silk, cotton, tobacco, paper, services, etc.), which competed with food crops for land or labor; and (3) large annual variations in the harvest made famine inevitable intermittently even though the food supply was normally adequate.

⁷² There is evidence of control over population growth in Yokouchi village in Shinshu, the population registers of which from 1671 to 1871 have been carefully analyzed by Professor Hayami Akira. Professor Hayami found that in the latter half of that period, when village population was stable, as compared to the earlier half (1671–1771), when it was growing rapidly, (1) the marriage rate for both men and women was slightly lower, (2) the average age of women at first marriage was somewhat higher, and (3) the number of children-ever of married women who lived in a married state until age 45 declined significantly. Hayami, Yokouchi, pp. 59–105. For similar evidence from England, see Wrigley, E. A., “Family Limitation in Pre-Industrial England,” Economic History Review, sec. ser., XIX, no. 1 (1966), 82–109CrossRefGoogle Scholar.

⁷³ Naotarō, Sekiyama, Kinsei Nihon no jinkō kōzō (Tokyo, 1957), pp. 137–41Google Scholar.

⁷⁴ Ohkawa, Kazushi and Rosovsky, Henry, “A Century of Japanese Growth,” Lockwood, William, ed., The State and Economic Enterprise in Japan (Princeton: Princeton University Press, 1965), pp. 66–67Google Scholar.

⁷⁵ Smith, Agrarian Origins, pp. 157–79.

⁷⁶ Of the presidents of a sample of 154 of the largest companies in the country in 1956, nearly 40 percent identified their father's major occupation as “farmer”—including resident but excluding absentee landlords. Most of these men began in business in the first two decades of this century. By comparison, among American business leaders who began in business when the United States was comparably rural (about 1870), farmers’ sons accounted for between 10 and 20 percent. Smith, Thomas C., “Landlord's Sons in the Business Elite,” Economic Development and Cultural Change, IX, no. 1, pt. 2, Oct. 1960, pp. 93–96CrossRefGoogle Scholar.

⁷⁷ Kuznets, Simon, “Underdeveloped Countries and the Pre-industrial Phase in the Advanced Countries,” Agarwala, A. N. and Singh, S. P., eds., The Economics of Underdevelopment (New York, 1963), pp. 143–44Google Scholar.

⁷⁸ The Capital of the Tycoon (New York, 1868), I, 387Google Scholar.