Seven
Farming, Sericulture, and Peasant Rationality in Wuxi County in the Early Twentieth Century

Lynda S. Bell

Peasant decision making is a subject of interest to a wide range of social science theorists as well as policy planners. A compelling reason for this interest is that in the third world, improving the productive capacity of the rural sector is an important component of development. Not only is a larger food supply necessary to support a growing population of industrial workers, but peasants must have improved standards of living so that they can transform themselves into investors and consumers to support fledgling industry. Therefore, understanding how and why peasants decide production and resource-allocation issues has become an important theme in economic decision theory.

Within this context, there has been a debate about whether or not peasants behave rationally. For example, Samuel Popkin has argued vehemently in favor of the idea, stating that peasants constantly take risks to improve their profit-making potential. This is a purist approach to the issue of rationality, and makes no concessions to other circumstances that may affect the peasant's decision-making processes.^[1] However, many economists who study peasant economies have come to accept a slightly modified view of peasant rationality. Rather than seeing peasants as pure profit maximizers, they observe that even when markets are well developed and functioning

[1] Samuel Popkin, The Rational Peasant: The Political Economy of Rural Society in Vietnam (Berkeley and Los Angeles, 1979). Popkin argues specifically against James C. Scott, The Moral Economy of the Peasant: Rebellion and Subsistence in Southeast Asia (New Haven, 1976). Scott claimed that peasants operate according to the principle of safety first and that patron-client ties figure prominently in peasants' efforts to avoid risk. Scott's work was influenced by the earlier social theorist Karl Polanyi, who made the important point that social institutions other than the marketplace greatly affect peasant decision making in precapitalist societies. See Scott; Polanyi, The Great Transformation (New York, 1944).

smoothly, peasants often lack the education and information necessary to take full advantage of them. Moreover, peasants also face the constraints of environment, low land-labor ratios, adverse land tenure relationships, and low levels of technological development.^[2]

Under these circumstances, the concerns of most peasants involve not only profit maximization but also the satisfaction of the immediate consumption demands of all household members. To accomplish this second goal, peasants often work at many sorts of supplementary tasks in addition to their principal crop-producing activity, a strategy that spreads risk and provides additional household income. Perhaps the best way to describe this behavior in the language of economics is to say that peasant families act simultaneously as both profit-maximizing producers, or firms, and income-maximizing consumers. Consequently, individual peasant households sometimes display behavior that in their role as producers could be considered irrational, since their supplementary activities often bring far fewer returns per unit of labor than prevailing market wages in agriculture. However, if there are a number of institutional and environmental constraints at work and satisfaction of basic consumption needs can be fulfilled in no other way, adopting such behavior helps to maximize the total real and future income of the family as a whole and become a consumer-specific form of economic rationality.

As described here, my analysis builds on the work of A. V. Chayanov and of Philip Huang, who uses Chayanov's arguments to elaborate a model of China's long-term development. Because Huang believes that one of the principal characteristics of China's economy in the late imperial period was diminishing returns to labor, he calls his model "economic involution," or alternatively, "involutionary growth," terms meant to convey the importance of increasing labor intensification and the propensity for per capita incomes to stagnate even as total output increased. Although I cannot fully address here all the issues raised by the concept of involution, I should note that I follow both Chayanov and Huang in my analysis of peasant decision making.^[3]

[2] Michael Lipton, "The Theory of the Optimizing Peasant," Journal of Development Studies 4, no. 3 (April 1968): 327–51.

[3] A. V. Chayanov's main works are Peasant Farm Organization and "On the Theory of Noncapitalist Economic Systems," both translated and edited by Daniel Thorner, Basile Kerblay, and R. E. F. Smith, in A. V. Chayanov on the Theory of the Peasant Economy (Homewood, Ill., 1966). For an excellent summary of Chayanov's arguments, see Mark Harrison, "Chayanov and the Economics of the Russian Peasantry," Journal of Peasant Studies 2.4 (July 1975): 389–417. Philip C. C. Huang has written two books, The Peasant Economy and Social Change in North China (Stanford, 1985) and The Peasant Family and Rural Development in the Yangzi Delta, 1350–1988 (Stanford, 1990) advancing his argument on involution. I will discuss the pros and cons of Huang's concept of involution at much greater length in a book manuscript now in preparation.

Specifically, I use Chayanov's idea that the producer-consumer dualism of peasant households affects their approach to both labor allocation and marketing activity. In my view, however, this dualism did not prevent some Chinese peasants from introducing innovative production techniques or from achieving higher levels of consumption; for this reason, the term involution used by Huang to describe the late imperial economy may be misleading, seeming to connote an inevitable process of universal, inward-looking decline. Nevertheless, I do share Huang's view that the producer-consumer dualism of peasant households precluded substantial gains in material welfare for the vast majority of Chinese peasants by the early twentieth century. Under conditions of declining land-labor ratios coupled with little technological change, most Chinese peasants approached markets as vehicles to improve their short-term consumption needs and found no leeway to pursue the expanded circulation and accumulation of capital that could have led to innovative long-term investment.

In this chapter, I will present early twentieth-century evidence from the Lower Yangzi county of Wuxi that supports this producer-consumer view of peasant rationality. I shall show that Wuxi peasants engaged in sericulture as a form of supplementary-income-earning activity under conditions of extensive commercialization, dense population, and unfavorable land-labor ratios. Cash-cropping in mulberry has often been analyzed as a way that some rural districts increased their wealth in the early twentieth century, and I do not dispute that Wuxi was one of China's most rapidly developing counties at that time. I shall also show, however, that most peasants who produced mulberries and raised silkworms in Wuxi did so out of economic necessity, that the income per labor day earned from those activities was far smaller than for rice/wheat farming, and that it was difficult to make improvements in silk production overall as long as relatively poor peasant households were responsible for innovations in sericulture technique. For these reasons, although peasants in Wuxi behaved quite rationally, they had trouble contributing to long-term plans of silk industry leaders for the improvement of silk quality and became an important obstacle to long-term silk industry growth.

Qing Trends in Commercialization and Demographic Growth

To understand the dynamics of farming and sericulture in Wuxi in the early twentieth century, it is useful to explore trends already under way in Wuxi's local economy. Situated in the heart of the Lower Yangzi regional core, Wuxi had been experiencing a long process of commercial growth and

population increase for at least two centuries before 1900.^[4] This prior experience set the stage for the rapid spread of sericulture as a supplementary income-earning activity among Wuxi peasants.

The dynamics of Qing period commercialization in Wuxi had two parts. First of all, Wuxi became an important rice-marketing center from the Qianlong period (1736–95) onward. Second, Wuxi peasants became heavily involved in cotton cloth production and marketing to supplement the income-earning capacity of their households. I shall explore these two forms of commercial activity in turn.

During the early eighteenth century, Wuxi was well on its way to becoming a major center of rice trade and transport within the Lower Yangzi region.^[5] By the early Qianlong years, areas to the west and south of Wuxi, in Anhui and Jiangxi provinces, were experiencing grain surpluses, while areas to the southeast of Wuxi, in Jiangsu and Zhejiang provinces, were evoling as a grain-deficit region because of escalating population growth, arable land becoming fully occupied, and regional urbanization.^[6] Under these conditions, Wuxi's strategic position on the Grand Canal, just south of its confluence with the Yangzi, and on the northeast bank of Lake Tai, an important transport link to northern Zhejiang, made the county a place where large-scale grain shipping, storage, and marketing began to take place.^[7]

At this point, state policy was also a dynamic factor, as Wuxi became an important concentration point in central China for the collection of tribute grain (caoliang ) before it was sent via the Grand Canal northward to Beijing. This had two consequences. First of all, as surcharges on tribute grain mounted and certain portions of the total tax were converted to cash equivalents, officials responsible for these levies increasingly collected them in cash.

[4] The term "Lower Yangzi regional core" is derived from G. William Skinner's influential work on economic "macroregions" in China as they developed in the late imperial period. The Lower Yangzi macroregion encompassed parts of Jiangsu, Zhejiang, and Anhui provinces and was one of China's richest agricultural areas from the tenth century onward. The "regional core" centered on the highly commercialized and densely populated area extending from the Shanghai-Ningbo coastal region inland to the area straddling the Jiangsu-Zhejiang-Anhui border. For a more detailed discussion of China's macroregions, see G. William Skinner, ed., The City in Late Imperial China (Stanford, 1977).

[5] I am indebted to Zhou Guangyuan of the Institute of Economics, Chinese Academy of Social Sciences, Beijing, for help in locating several sources dealing with eighteenth-century trends and in developing an overall picture of Wuxi's early patterns in commercialization and demographic growth.

[6] Shehui Jingji Yanjiusuo, Wuxi mishi diaocha (Shanghai, 1935), preface, pp. 1–2; Dwight H. Perkins, Agricultural Development in China, 1368–1968 (Chicago, 1969), pp. 140–51.

[7] Sun Jingzhi et al., Economic Geography of the East China Region (Shanghai, Kiangsu, Anhwei, Chekiang ), published in Chinese, Nov. 1959, citation from the English translation by JPRS (Washington, D.C., 1961), pp. 4, 98; Wuxi Difangzhi Bianji Weiyuanhui, ed., Wuxi gushixuan (Wuxi, 1959), pp. 36–40.

They then purchased grain to meet the portion of the total tax still demanded in kind by the central government in Beijing. Thus, rice circulated ever more freely as a commodity for purchase and sale in various central China locales, and Wuxi emerged as a leading center for this activity.^[8] Second, the development of an extensive regional shipping network also proceeded at a faster pace as tribute grain came toward the Grand Canal via adjacent river routes from the eight central provinces responsible for providing it. Strategically placed at a regional transport hub, Wuxi took full advantage of this situation.^[9]

As commercial opportunities increased, so too did population. Estimating the precise rate of population growth in the eighteenth century is difficult because of China's methods of enumeration. In a study of Chinese population from the Ming dynasty through the mid-twentieth century, Ping-ti Ho explains that the prevailing system of population reporting in Qing times was based on the ding , a unit of taxation that originally referred to all adult males of working age. As is typical for the era, the only existing population figure for Wuxi in the early eighteenth century is a gazetteer report of 142,509 ding in 1726.^[10] Ho also points out that during the eighteenth century, the ding figures no longer bore any relation to the number of adult males in a given population, but rather had evolved over the two centuries or so of their usage into pure fiscal units corresponding only to a locale's total tax liability. On the basis of extensive examination and evaluation of such figures, Ho concludes that the ding represented "substantially less than one-half" the real population.^[11]

By the late eighteenth century, population reporting was beginning to correspond to reality more closely. Earlier in the century, the Kangxi emperor introduced a series of tax reforms referred to as tanding rudi or tanding rumu , literally, "spreading the ding into the land," with the intention of eliminating the ding system of tax accounting and instituting a tax based purely on land ownership at a permanent fixed rate.^[12] These reforms spread slowly, however, and were only brought to fruition by the Qianlong emperor in the mid 1770s, with decrees that permanently ended the ding assessment and called for an annual accounting of true population figures and real grain

[8] Harold C. Hinton, The Grain Tribute System of China (1845–1911) (Cambridge, Mass., 1956), pp. 1–15.

[9] The eight provinces responsible for tribute grain payments were Shandong, Henan, Anhui, Jiangsu, Zhejiang, Jiangxi, Hubei, and Hunan. See Hinton, Grain Tribute System , p. 7. On Wuxi's role as a transport and marketing center in this process, see Shehui Jingji Yanjiusuo, Wuxi mishi diaocha , preface, 1–2.

[10] Wuxi-Jinkui xianzhi (A gazetteer of Wuxi and Jinkui counties), 1881 edition, 8:5.

[11] Ping-ti Ho, Studies on the Population of China, 1368–1959 (Cambridge, 1959), p. 35.

[12] I am grateful to Chen Qiguang of the Institute of Economics in Beijing for explaining Kangxi's reforms that began in 1712; Ping-ti Ho also discusses the reforms in Studies , p. 25.

output.^[13] The effect of imperial reforms on population enumeration in Wuxi was that the revised ding figures reported for 1795 totaled 566,217, a closer measure of the true number of males within the county.^[14]

To derive population figures and to calculate a rate of demographic growth for Wuxi, I follow Ho's guidelines for the early eighteenth century and the suggestions of Chinese population expert Liang Fangzhong that male ding figures for Lower Yangzi region around the turn of the nineteenth century were approximately 53 percent of the total population.^[15] Therefore, I consider the 1726 ding figure to have been only around 30 percent of the real population, a rough estimate to achieve Ho's description of the ding as being "substantially less than one-half" the population, to arrive at an approximate population of 475,030; and following Liang, I take the 1795 ding figure to be 53 percent of the total, for an estimated population in that year of 1,068,334. This translates into an annual rate of population growth of 1.2 percent for the period 1726 to 1795.^[16] For China as a whole, Ho has estimated the rate of annual increase to have been 0.9 percent during this period.^[17] It is also useful to compare my calculations with Dwight H. Perkins's estimate that population doubled in Jiangsu from the mid-eighteenth century into the early nineteenth, a rate of increase that corresponds roughly to his national-level estimate that the country reached an all-time high in population, increasing from 200–250 million in 1750 to 400 million by the early years of the nineteenth century. When translated into an annual rate of population growth, Perkins's figures produce estimates very similar to Ho's, from 0.8–0.9 percent. Because of Wuxi's special conditions of rapid commercialization and the development of rural industry, it is possible that population there grew more quickly than either Ho's or Perkins's estimate suggests. But for lack of better population data for the early eighteenth century, it seems impossible for the moment to provide any more accurate calculations than these.^[18]

[13] Ping-ti Ho, Studies , pp. 47–48.

[14] Wuxi-Jinkui xianzhi , 8:5–6.

[15] Liang Fangzhong, Zhongguo lidai hukou, tiandi, tianfu tongji (China's dynastic statistics on household population, land, and land taxes) (Shanghai, 1980), pp. 440–41. Liang's figures are for Songjiang prefecture immediately to the east of Wuxi.

[16] Calculations follow George W. Barclay, Techniques of Population Analysis (New York, 1958), pp. 28–33, for continuous population growth.

[17] Ping-ti Ho, Studies , p. 64.

[18] Perkins, Agricultural Development , pp. 202–9. On trying to answer why Wuxi's annual rate of population growth may have been higher during the eighteenth century, studies of demographic growth conducted for various European locales offer intriguing hypotheses. The work of Franklin F. Mendels on Flanders in the eighteenth century, for example, argues that rapid commercialization and the development of rural industry altered nuptiality patterns—people married earlier because they could afford to—thus accelerating the regional rate of population growth (Industrialization and Population in Eighteenth-Century Flanders [New York, 1981]). Historians of Qing China have been interested in testing such theories for various regions of China, but the reconstruction of the demographic data is a formidable task due to the problems in population enumeration that I have discussed here. Ongoing work by James Lee, R. Bin Wong, and William Lavely promises to shed future light on China's unresolved demographic issues.

In any case, the rate of growth of Wuxi's population is not as important to the argument at hand as the degree of population density that Wuxi experienced by the end of the eighteenth century. In the 1790s, when figures on population and land had become far more reliable, Wuxi had only 1.3 mu of arable land per capita, just 0.1 hectare, or slightly more than one sixth of an acre.^[19] At the same time, as Perkins has demonstrated, there were no radically new developments occurring in agricultural technology during this period that would have led to substantial gains in grain output.^[20] Therefore, to comprehend how peasants in densely populated regions such as Wuxi continued to support themselves, we must turn to an exploration of trends in subsidiary activities designed to raise the potential earning power of the peasant household.

Significantly, it was also during the eighteenth century that gazetteer reports for Wuxi began to note the importance of mianbu or huabu , handicraft cotton cloth, which Wuxi peasants produced in order to secure additional grain supplies. A 1752 gazetteer explains how these exchange relationships operated among Wuxi peasant households:

Of five counties in Changzhou prefecture, Wuxi is the only one that does not cultivate cotton of its own; yet cotton cloth is even more important here than in the other counties. Wuxi peasants get only enough grain from their fields for three winter months' consumption. After they pay their rents, they hull the rice that is left, put some in bins, and take the rest to pawnshops to redeem their clothing. In the early spring, entire households spin and weave, making cloth to exchange for rice to eat because by then, not a single grain of their own is left. When spring planting is underway in the fifth month, they take their winter clothing back to the pawnshops in order to get more rice to eat. . . . In the

[19] Land per capita is calculated from a population figure estimated from the ding statistics for 1795 in Wuxi-Jinkui xianzhi , 8:5–6, and land statistics for the mid 1770s in Wuxi-Jinkui xianzhi , 9:1 and 10:1.

[20] Perkins, chaps. 3, 4. Perkins's main argument is that for China as a whole during the fourteenth through the twentieth centuries, grain output rose steadily, but mainly through the better use of "traditional" technology—the opening of new lands, the use of better and more extensive irrigation techniques, the introduction of double-cropping to new areas, and the use of a few new crops such as sorghum, sweet potatoes, and corn. Twentieth-century surveys show that no double-cropping of rice was attempted in Wuxi (although another form of double-cropping was achieved through planting a winter crop of wheat, which was treated as a cash crop) and that early-ripening varieties of rice were rarely used. As I shall argue below, the preferred method of supporting a burgeoning population in Wuxi seems to have been for peasants to develop various methods to earn cash income to purchase rice imported from surrounding areas.

autumn, with the slightest rainfall the sound of looms permeates the villages once again and the peasants take their cloth [to market] to exchange it for rice to eat. Although there are sometimes bad crop years in Wuxi, as long as other places have good cotton harvests, our peasants have no great difficulties.^[21]

From this passage, we can see that by the middle of the eighteenth century, Wuxi peasants had developed a complex pattern of cotton spinning and weaving, pawning of clothing items made of cotton, selling of cotton cloth, and rice buying. As commercialization proceeded and population grew, it was important for peasant families to develop methods that would adequately support their consumption demands. Although it is difficult to document the process of peasant decision making more precisely for the eighteenth century (I shall be more exact with twentieth-century data), it is likely that household labor was used more effectively by combining cotton spinning and weaving with the regular farming enterprise. Especially important was that the labor of women and children could be tapped and also that spinning and weaving could proceed during the slack agricultural periods of late fall, winter, and early spring. Even though peasants had to use relatively scarce rice resources immediately after the harvest to redeem their winter clothing and to acquire cotton, operating in this fashion must have meant that the income-earning potential of the family as a whole, or, stated another way, the "total product" of their labor, was raised. With an active and growing rice market within the county, the possibility that peasant demand for grain could be satisfied by rice imported from other areas helped to promote this particular pattern of agrarian development.^[22]

With this brief sketch of Qing period commercialization in Wuxi, I shall turn now to a discussion of changing conditions in the late nineteenth century. I shall argue that the large-scale undertaking of sericulture by Wuxi peasants at this time should be seen not only as an adaptation to the new market demand for Chinese raw silk but also as a continuation of seeking ways of supplementing household income through diversification.

Sericulture After the Taiping Rebellion

Sericulture became an attractive option for Wuxi peasants around 1870 because of two converging sets of circumstances: the "opening" of Shanghai by the British as a result of the Opium War and subsequent developments in the international silk market, and the chance for peasants to make innovative

[21] Xi-Jin shi xiaolu (A brief record of what is known about Wuxi and Jinkui counties) (Wuxi, 1752), 1:6–7.

[22] Further discussion of these developments in Xi-Jin shi xiaolu , 1:8, refers to the rice purchased by peasant households in Wuxi as kemi , literally "guest rice," meaning that this rice was coming into the county from other areas.

adaptations in their patterns of subsidiary activity following the devastation of central China in the Taiping period. I shall begin this discussion with the international situation.

In the 1840s, Shanghai became a port open to foreign trade, and an international community of foreign businessmen and Chinese merchants began to congregate there. Among the foreigners were men interested in promoting the development of machine-based spinning of silk yarn and its export to expanding silk markets in both Europe and the United States. The first modern steam-operated silk filatures, plants where this yarn was processed, were built in Shanghai in the early 1860s, with capital and equipment provided by European investors.^[23] A chronic problem plagued these early filatures, however. A marketing network to secure cocoons for filatures had not yet been established in rural areas adjacent to Shanghai. Moreover, silk filatures faced an uphill battle in establishing such a network, because they were thrown into direct competition for cocoons with well-entrenched production and marketing networks for handicraft silk products. For lack of adequate cocoon supplies, all of Shanghai's early filatures had great difficulties remaining open.^[24]

Meanwhile, just to the west of Shanghai, a domestic war of great intensity was building. In 1861, the Taiping rebels made their way to Nanjing, and a four-year period of intense warfare within the immediate environs ensued. The consequences of this war for many rural areas were devastating. Peasants and landlords alike fled the region, while many tens of thousands who stayed behind died as a result of the fighting. After the Taiping defeat in 1865, the area was badly in need of resettlement and restoration to its full productive capacity.^[25]

Because of these converging circumstances, the time was now ripe for peasants in Wuxi to act. As the Taiping period ended and peasants migrated to the area, they converted portions of former paddy land to mulberry fields and began to raise cocoons each spring for sale to newly developing Shanghai filatures. Although data for this period are insufficient to determine the precise start-up costs of these activities, we do know that members of elite society in Wuxi often encouraged this switch to sericulture. In some cases, gentry

[23] Yin Liangying, Zhongguo canye shi (A history of Chinese sericulture) (Nanjing, 1931), p. 12; E-tu Zen Sun, "Sericulture and Silk Textile Production in Ch'ing [Qing] China," in W. E. Willmott, ed., Economic Organization in Chinese Society (Stanford, 1972), pp. 103–4; Lillian M. Li, China's Silk Trade: Traditional Industry in the Modern World (Cambridge, 1981), p. 164.

[24] Yin Liangying, Zhongguo canye shi , p. 12; Wuxi Difangzhi Bianji Weiyuanhui, ed., Wuxi gushixuan , p. 41. For an extended discussion of the competition for cocoons between producers of handicraft silk and the new steam-powered filatures in Shanghai, see Lynda S. Bell, "Merchants, Peasants, and the State: The Organization and Politics of Chinese Silk Production, Wuxi County, 1870–1937" (Los Angeles, 1985), chap. 2.

[25] Li Wenzhi, "Lun Qingdai houqi Jiang-Zhe-Wan sansheng yuan Taiping Tianguo zhanlingqu tudi guanxi de bianhua," Lishi yanjiu 6 (Dec. 15, 1981): 82–86.

members bore the initial costs of mulberry purchase and early dissemination of sericulture technique. Undoubtedly, they did this in the hope that they could attract new tenants to their land with the promise of substantial cash earnings.^[26] From this point onward, sericulture replaced cotton cloth production as the primary subsidiary activity of peasant households, and Wuxi began to develop as the cocoon-marketing capital of the central China region.^[27]

Although ecologically it was possible to raise both mulberries and cocoons in Wuxi, it also became clear as sericulture developed that there were relatively higher risks there than in other locales. Spring weather in Wuxi was not quite as warm as in older sericulture areas closer to the coast and further south, and so Wuxi peasants often experienced seasons in which they lost their silkworm crops entirely to the vagaries of rapidly changing humidity, cool snaps, and the subsequent growth of incurable bacterial infections.^[28] Moreover, sericulture in Wuxi was tied directly and solely to the international market for machine-spun raw silk. None of its cocoons entered into the domestic handicraft market, making Wuxi peasants especially vulnerable to drops in the international price of raw silk and the filature closings that inevitably resulted.^[29] A kind of boom-bust atmosphere prevailed within the Wuxi cocoon market, and the risks associated with cocoon production in Wuxi were likewise relatively high.^[30]

[26] Yan Jinqing, ed., Yan Lianfang yigao (Wuxi, 1923), 10:9, Gao Jingyue and Yan Xuexi, "Wuxi zuizao de sangyuan,"' Wuxi xianbao (Aug. 20, 1980): 4; Zhang Kai, "Mantan lishishang Jiangsu de canye," Canye keji 3 (Oct. 1979): 53; Li, China's Silk Trade , pp. 131–38.

[27] On cocoon marketing in Wuxi, see Bell, "Merchants, Peasants, and the State," chap. 3.

[28] "Sican zhaiyao," Jiangnan shangwu bao 9 (Apr. 21, 1900), commercial raw materials section, p. 2; and Dierlishi Dang'anguan, file no. 3504, "Liangnianlai bensheng [Zhejiang] canzhong zhizao ji qudi jingguo gaikuang, Minguo ershinian—Minguo ershiernian," p. 4; Sun Guoqiao, "Wuxi zhidao yu yangcan shiye zhi yaodian," Nongye zhoubao 1, no. 25 (Oct. 16, 1931): 985; "Wuxi jianshi jianse," Minguo ribao , May 21, 1916, sec. 3:10; "Canxun yin yushui guoduo shousun," Minguo ribao , May 11, 1921, sec. 3:11; "Wuxi chunjian shoucheng jianse," Minguo ribao , June 5, 1921, sec. 2:8.

[29] The link between the Wuxi cocoon market and the central China filature industry is treated at length in Bell, "Merchants, Peasants, and the State," chaps. 1 and 3. For the problem of periodic filature closings, see Minami Manshu Tetsudo Kabushiki Kaisha, Shanhai Jimusho Chosashitsu (hereafter SMR, Shanghai Office), Mushaku kogyo jittai chosa hokokusho (Shanghai, 1940), pp. 85–87, 94; Yinhang zhoubao 13, no. 34 (Sept. 3, 1929), weekly commerce section, pp. 2–3, and 13, no. 37 (Sept. 24, 1929), weekly commerce section, p. 4; Gongshang banyuekan 2, no. 17 (Sept. 1, 1930), legislation section, pp. 4–6; and Shenbao , Jan. 19, 1937, sec. 4:14. For filature closings in Wuxi during the depression, see Gongshang banyuekan 2, no. 3 (Feb. 1, 1930), commercial news section, p. 16, and 4, no. 19 (Oct. 1, 1932), national economy section, pp. 1–2; He Bingxian, "Minguo ershiyinian Zhongguo gongshangye de huigu," Gongshang banyuekan 5, no. 1 (Jan. 1, 1933), articles section, p. 18; and Chushi Kensetsu Shiryo Seibi Jimusho, ed., Mushaku kogyo jijo (Shanghai, 1941), p. 40.

[30] This aspect of the modern silk industry in central China is often referred to in the literature as its speculative nature. See, for example, Wuxi Difangzhi Bianji Weiyuanhui, ed. Wuxi gushixuan , pp. 45–46; Chen Tingfang, "Juyou fengjian de maiban xingzhi de Zhongguo saosiye," in Chen Zhen, et al., eds., Zhongguo jindai gongyeshi ziliao (Beijing, 1959–61), 4:113; Zhuang Yaohe, interview with author, Wuxi, May 27, 1980; Chushi Kensetsu Shiryo Seibi Jimusho, ed., pp. 39–43; and SMR, Shanghai Office, Mushaku kogyo jittai chosa hokokusho , pp. 85–87.

Despite the riskiness of the endeavor, once Wuxi peasants plunged into sericulture, they persisted with great tenacity. Throughout the decades of the late nineteenth and early twentieth centuries, the proportion of land devoted to mulberries climbed steadily in Wuxi, to reach peaks of 20–30 percent in various locales by the early 1920s.^[31] And yet in many ways, such a decision on the part of Wuxi peasants remained something of an anomaly. Why did they continue to engage in sericulture, even after it became obvious that serious uncertainties existed both in weather and marketing conditions? Were the profits they reaped in good years enough to offset the possibility that, in bad years, they would lose their cocoon crop entirely? Or were there some other, even more compelling reasons for Wuxi peasants to persist in sericulture?

In the remainder of this chapter, I shall try to answer these questions using detailed survey materials on twentieth-century Wuxi agriculture. As I introduce the world of peasant-household decision making, I shall attempt to relate this discussion to the earlier trends I have described as evolving in Wuxi since the eighteenth century—a high level of commercialization, dense population, and the consequent need for peasants to develop strategies to supplement their income-earning potential through subsidiary household activity.

Land and Labor Allocation in three Wuxi Villages

Three villages in Wuxi were surveyed in 1940 by the Shanghai Office of the Japanese South Manchurian Railway Company (SMR), a research organization active in areas of China then occupied by the Japanese government.^[32] These villages lay within a one kilometer radius of the market town of Rongxiang zhen , approximately eight kilometers from the west gate of Wuxi City.^[33] This was a highly commercialized and densely populated area, yet the large majority of households still depended primarily upon rice/wheat agriculture and household-based subsidiary activities, with sericulture the

[31] Lu Guanying, "Jiangsu Wuxixian ershinianlai zhi siye guan," Nongshang gongbao 8, no. 1 (Aug. 15, 1921), articles and translations section, p. 45; Gongshang banyuekan 2, no. 15 (Aug. 1, 1930), investigation section, p. 3; and SMR, Shanghai Office, Kososho Mushakuken noson jittai chosa hokokusho (Shanghai, 1941), p. 11.

[32] SMR, Shanghai Office, Kososho Mushakuken . For a broader discussion of the research activities of the South Manchurian Railway Company in China, see Philip C. C. Huang, The Peasant Economy and Social Change in North China (Stanford, 1985).

[33] SMR, Shanghai Office, Kososho Mushakuken , pp. 14–15.

most important among these, for the bulk of their yearly incomes. Of the 80 households in these villages, 75 engaged in farming; they constituted the sample with which the Japanese researchers worked.^[34]

Although there is some cause for concern as to whether or not the conditions in these villages still reflected economic patterns as they had evolved before the effects of the depression and the Japanese occupation, the researchers themselves were highly sensitive to this issue. They explored available secondary sources on the percentage of land devoted to mulberries before the depression and found estimates ranging from one fifth to one third for various locales.^[35] In the survey villages in 1940, 22.5 percent of arable land was devoted to mulberries, a finding within the range of predepression figures.^[36] Even though mulberry acreage declined during the years when the depression affected silk prices most severely, that is, from 1930 to 1932, by 1940 peasants were restoring mulberry fields to their former position within the agrarian cropping regime.^[37] On the issue of prices for agricultural goods, the researchers reported not only 1939 data but also 1933 data collected in Wuxi by the Nationalist government in an effort to determine effects of inflationary trends caused by the Japanese occupation.^[38]

As elsewhere in Wuxi, most families in the Japanese-surveyed villages combined rice/wheat farming with mulberry cultivation and cocoon rearing. Rice was grown once yearly during the summer months. Then, after the fall harvest, peasants pumped their rice paddies dry and planted winter wheat on all or some of the land. Mulberries were planted in "field fashion" in Wuxi. Trees were not grown on embankments between rice fields, as in the older sericulture districts to the south of Wuxi, in Jiangsu and Zhejiang. Rather, a portion of arable land that could have been used for grain cultivation was used instead for mulberry trees. While the trees could be pruned and fertilized in off-peak periods of late fall, winter, and early spring, the first monthlong busy period for mulberries, during which the trees were stripped of their leaves for silkworm feeding, came from late April to late May, coinciding precisely with rice seedling preparation. In June, cocoons were marketed, wheat harvested, and rice seedlings transplanted, giving the peasants no time to recover from their extremely busy monthlong period of silkworm feeding and cocoon raising. Peasants raised a second cocoon crop in summer or, ideally, in early fall, when the mulberry regrowth was fuller. Fall crops

[34] Ibid., pp. 25–26.

[35] Ibid., p. 11.

[36] Ibid., table 1, following the text.

[37] Ibid., pp. 9–11, 18–19. Figures reported by the SMR research group indicate that about 378,000 mu were devoted to mulberry in 1927, 30 percent of all cultivated land in Wuxi. This figure fell steadily until 1932, when only 84,000 mu were cropped to mulberries. By the late 1930s, mulberry land was up again to 240,000 mu .

[38] Ibid., pp. 9–10. I shall also say more about the reconstruction of prices in note 43 below.

first became possible in Wuxi in the late 1910s, with the introduction of refrigeration and delayed incubation of silkworm eggs. The fall round of silkworm raising usually came in late August and ended in mid-September, long before the late October rice harvest. Since late summer/early fall was a less busy time for grain cultivation, this round of cocoon raising was not nearly as taxing for peasant families as in the spring.^[39]

To relate data from the Japanese survey to the previous discussion of peasant rationality, let me pose the question I am most interested in exploring: Precisely what did peasant families gain by engaging in this particular work regime? When considering this issue, I found it useful to construct Table 7.1, comparing income, production costs, and labor usage for rice/wheat farming and mulberry culivation combined with silkworm raising.^[40] The main point I wish to make with the comparison is that mulberry cultivation with silkworm raising brought slightly higher returns per mu than rice/wheat farming (9.25 yuan for rice/wheat versus 11.96 yuan for coccons) but fewer returns per labor day (0.27 yuan for rice/wheat versus 0.19 yuan for cocoons). Moreover, earnings from wage labor in agriculture were also higher, averaging 0.25 yuan per labor day.^[41] We see a situation, therefore, much like that observed by economists elsewhere—peasants who worked at certain tasks, in this case those of sericulture, for less than optimal returns to labor.^[42] Should we conclude from this finding that peasants were "irra-

[39] Ibid., pp. 55, 61, 69–70, 72, 74.

[40] Table 7.1 originally appeared in Bell, "Merchants, Peasants, and the State," p. 122; most of the data are derived from the SMR survey report. For a full accounting of all the calculations, see "Merchants, Peasants, and the State," pp. 122–24. Philip C. C. Huang (Peasant Family and Rural Development , p. 127) has argued that my original calculation for labor days spent in sericulture in this table was too low because I failed to take into account a sufficient number of days for pruning the trees and also the days in the growing cycle of silkworms when they rested. He then adds 28 days to my original estimate of 52 days to come up with a total of 80 days for mulberry cultivation and silkworm raising. I agree with Huang that the days silkworms rested (an additional 11 days for the spring and fall cycles combined) should be added to the total and I have done this in my calculations here. However, I am less sanguine about his decision to add an additional 17 days for pruning the trees. Villagers in neither survey discussed here (for more on the second survey, see the next section) said that they spent a total of 30 days yearly (Huang's estimate) working the mulberry fields; by contrast, they regularly gave a number in the range of five to thirteen days (see Table 7.1 and Table 7.5, labor for mulberry cultivation). What the villagers did say very often when questioned about their work in sericulture is that they spent 30 days in the spring season raising silkworms, an estimate which I interpret to have included all the work involved, including the stripping (or "pruning") of the trees to get leaves to feed to their silkworms. What seems to have happened in Huang's method of calculating, therefore, is that he counts the days during the silkworm raising cycles when leaves were stripped from the trees as labor days for both sericulture and work in the mulberry fields, to get a total of 80 days rather than my estimate of 63.

[41] The figure of 0.25 yuan for day labor in agriculture is the average found for central China's rice/wheat region by John Lossing Buck's 1929–33 survey of agricultural conditions throughout China. It is cited in SMR, Shanghai Office, Kososho Mushakuken , p. 93.

[42] Chayanov, Peasant Farm Organization .

tional"? I would argue that we should not, and at this point, take into careful account the problems of population density and scarcity of other options for earning cash income to understand why peasant families in Wuxi undertook sericulture.^[43]

[43] In order to dispel doubts that price figures in Table 7.1 may have been atypical and hence relevant only for the postdepression period in Wuxi, I have also computed price figures for a sample of 146 households in Wuxi in 1928 and 1929, selected from the Guoli Zhongyang Yangiuyuan survey. This second survey is explained more fully in the next section. Prices for rice, wheat, and cocoons were all slightly higher in 1928 and 1929 than in 1939, but the crucial comparison of earnings per labor day between rice/wheat farming and cocoon rearing remains valid. In fact, it tips even more dramatically in favor of rice/wheat farming producing higher returns to labor, with the average being 0.71 yuan as opposed to 0.28 yuan for cocoons. The even larger margin arises primarily because wheat prices were higher by about 55 percent in 1928–29 than in the postdepression period, while the prices for rice and cocoons were higher by only about 20 percent. Another point worth noting concerning calculations in Table 7.1 is that I have purposely excluded land price as a production cost, because price figures for rice/wheat land and mulberry land were nearly identical. I have confirmed this fact via an analysis of variance test on land prices from the Guoli Zhongyang Yanjiuyuan survey of 1929, which showed no significant difference between prices for the two types of land. Finally, interplanting mulberry with other crops, a strategy that might have lowered land costs for mulberry culture by raising total yield, was rare in Wuxi because local farmers planted their mulberry trees close together, leaving no room for other crops. This is substantiated by the Guoli Zhongyang Yanjiuyuan survey, which demonstrates that other crops such as peas, beans, and potatoes were not grown with mulberries, but rather were raised on small supplemental plots.

First of all, let us consider population issues and average size of farm holdings. Although gazetteer figures indicate that Wuxi lost as much as one half to two thirds of its population during the Taiping period, by the 1920s former levels of population density had been completely restored.^[44] The amount of available arable land per capita in Wuxi in 1929 stood at 1.3 mu , a figure identical to that reported for the late eighteenth century. This made Wuxi the second most densely populated county within Jiangsu Province.^[45] An important by-product of dense population was the small size of peasant farming units. In the SMR-surveyed villages, average farm size was 2.5 mu , approximately 0.2 hectare, or slightly more than one-third of an acre. Moreover, no single farm was larger than 7 mu , and 72 percent of the farming households owned 3 mu or less.^[46] Even by the standards of the Lower Yangzi region, where land was fertile and well irrigated and thus capable of supporting larger populations than in many other areas, these were very small farming units.^[47]

The small size of farms in Wuxi caused the Japanese researchers to consider carefully what other methods were used to augment family incomes. What they found was a fairly consistent pattern of subsidiary activities undertaken by village households. In terms of income-generating capacity, sericulture was most important, usually accounting for 50 percent or more of a family's cash earnings.^[48] There were other important trends under way as

[44] The estimate of population loss in Wuxi during the taiping period is calculated from cadastral figures found in Wuxi-Jinkui xianzhi 8:6–7. For a more precise accounting of the figures involved, see Bell, "Merchants, Peasants, and the State," p. 84.

[45] Chen Huayin, "Jiangsusheng renkou yu yiken tianmu zhi xilian," Tongji yuebao 1, no. 3 (May 1929):44–48.

[46] SMR, Shanghai Office, Kososho Mushakuken , pp. 23, 88.

[47] The SMR researchers had already surveyed villages in three other Jiangsu counties, where they found the average size of landholdings to be 6.4 mu (Jiading), 5.3 mu (Changshu), and 9.6 mu (Songjiang). They were quite surprised by the small size of farming units they observed in Wuxi. See SMR, Shanghai Office, Kososho Mushakuken , p. 23.

[48] SMR, Shanghai Office, Kososho Mushakuken , pp. 103–4.

well, with a high proportion of village men working away from home on a permanent basis, often as clerks or factory workers in Shanghai. But their yearly cash incomes were still lower on average than those generated through cocoon sales.^[49] Men could also remain in the villages to hire themselves out as agricultural wage labor, but because land was scarce, this was done only at peak seasons among village households as a means of satisfying temporary labor needs. The number of days men could work at such tasks was extremely limited and, in comparison to sericulture, provided only small amounts of supplementary cash income.^[50]

Given the constraints of peasant life in Wuxi, it thus appears that sericulture enabled peasants to better meet their subsistence requirements by increasing the total product of their family labor. First of all, sericulture employed women in an occupation that produced relatively higher cash earnings than cotton cloth production. When we compare the daily income from sericulture with that from cotton weaving, we see, on average, that women were able to earn only 0.02 yuan per day making cotton cloth versus 0.19 yuan from sericulture.^[51] In turn, peasants used most of their cash income from sericulture to purchase rice and other items necessary to satisfy immediate household consumption demands. Sixty percent of all cash income was spent on food, and most of the rest went to purchase clothing and other household goods. There is no evidence to suggest that these households had accumulated any substantial savings, and only 0.8 percent of cash income went to productive investments in agriculture.^[52] Rent obligations also figured into this picture for many families, putting a further strain on the capacity for self-sufficiency in grain production. Overall, about 25 percent of the land in these villages was rented, and 10 percent of the rice produced yearly went to rent payments.^[53]

The issue of what constitutes "survival capacity" for peasant households, or, put another way, their relative level of subsistence, is a highly volatile topic in the literature on peasant decision making. I cannot attempt to give the definitive answer to the question here. But to get at least some sense of the importance of sericulture in Wuxi in assuring what I would call "basic subsistence," I shall consider evidence concerning grain consumption both in China at large and in Wuxi in particular.

In the early 1930s, John Lossing Buck documented yearly adult grain consumption for China as a whole: it ranged from 390 to 952 jin . His figures

[49] Ibid., pp. 99–104.

[50] Ibid., pp. 88–95; tables 2 and 3, following the text.

[51] The figure for cotton weaving comes from Guoli Zhongyang Yanjiuyuan, selected questionnaires, table 11. The sericulture figure is found in Table 7.1 above.

[52] SMR, Shanghai Office, Kososho Mushakuken , pp. 122–24; table 15, part 2, following the text.

[53] Ibid., pp. 25, 144–47.

include not only the main cereal grains of rice, wheat, and millet but also potatoes, beans, and peas. In many areas of China, these latter three items often made up a large proportion of daily consumption for many peasant households. This could be because of the relative infertility of the land or because rent payments were usually demanded in the highest quality grain a given locale produced, leaving the poorest of local peasant families to consume coarser foodstuffs. Commenting on these figures and comparing them to provincial-level data for the 1950s, Dwight H. Perkins has concluded that about 400 jin of grain per adult per year is a fairly accurate estimate of China's "minimum subsistence" requirements.^[54]

In the SMR-surveyed villages in Wuxi, the preferred pattern of grain management and consumption for most households was to preserve as much of the rice they produced themselves as they could for their own use. As we have seen, rent obligations took about 10 percent of the rice produced, so that overall the village households had about 29,550 jin of their total production of 32,833 jin left for consumption purposes. They also purchased rice with cash income from sericulture, wheat sales, and other subsidiary employments, amounting to an additional 32,250 jin . When converted to the average amount of rice consumed yearly, this works out to approximately 824 jin of rice per household. Since the average household size was 4.1 members, this meant an average per capita rice consumption figure of only 201 jin . If we convert this to an adult equivalent for purposes of comparison with Buck's data (using Buck's estimate of 77 adult equivalents for a total population of 100), we arrive at a figure of 261 jin for the yearly rice consumption per adult villager. To come up to the minimum subsistence levels suggested by Perkins, Wuxi peasants supplemented their diets with broad beans and peas, which were both locally grown.^[55]

When viewed from this comparative perspective, we find that in terms of quantity of grain consumed, Wuxi villagers were quite near the lower end of the consumption range observed by Buck for China as a whole in the 1930s. Of course, their diet, which had a large proportion of white rice, was of quite high quality by Chinese standards. Relatively speaking, then, just how well off were these peasant households?

It seems fairly clear that as long as the market for Chinese raw silk was doing well, Wuxi peasants would rarely have faced hunger. Moreover, the

[54] This discussion of "minimum subsistence" is drawn from Perkins, Agricultural Development , pp. 14–15, 300–301. In the latter pages, Perkins cites Buck, Land Utilization in China: Statistical Volume (Nanking, 1937), at length.

[55] SMR, Shanghai Office, Kososho Mushakuken , pp. 144–49. The average size of a farming household is found in this same work, table 1 following the text. For Buck's "adult equivalent," see Perkins, Agricultural Development , p. 301. According to Guoli Zhongyang Yanjiuyuan, selected questionnaires, table 1, broad beans (candou ) and peas (wandou ) were both grown in Wuxi in 1929, broad beans being far more common.

grain they had available to them was of high quality, and they themselves must have felt quite satisfied that these sorts of options were available to them. However, when the market for raw silk was doing poorly, this rather fragile, near-subsistence-level system would have collapsed. To demonstrate this, we can consider what would have happened if all mulberry land within the SMR-surveyed villages had been converted back to rice/wheat farming, and sericulture abandoned. Many families in Wuxi took this option during the worst years of the depression. The additional rice that could have been produced on former mulberry land amounted to only 12,300 jin , a figure that falls far short of the 32,250 jin that villagers normally purchased.^[56] This means that if sericulture had been abandoned, peasant families would have had to find other ways to earn supplementary cash income to continue purchasing grain at their accustomed levels. Options that provided returns comparable to sericulture were not readily available within the village itself and, as we have seen, often meant out-migration for male members of individual farming families.^[57]

What I would argue on the basis of these considerations is that in Wuxi by the early twentieth century, sericulture had become a crucial link in a complex system that, in good years, provided moderate levels of subsistence for most farming families. Even though individual workers' earnings were below those to be had through rice/wheat farming, the total product of peasant family labor was raised via sericulture. Peasants tolerated this situation because under conditions of dense population and scarce land, finding ways to better employ peasant women had become essential. Since financial reserves were small or nonexistent and most cash income was spent on food and other basic living expenses, a sudden drop in income caused by falling silk prices would depress living standards for many peasant families in Wuxi and would push some households below basic subsistence requirements.^[58]

A Further Look at Wuxi Villages

To substantiate this picture of peasant family farming, some additional evidence can be garnered from a second rural survey in Wuxi. This second survey was carried out in the summer of 1929 by Chinese researchers of the Social Science Research Institute of the newly formed Academia Sinica. I

[56] I have made this calculation from figures on average rice output per mu and the number of mu currently cropped to mulberry. See "Merchants, Peasants, and the State," pp. 148–49.

[57] For a discussion of relative returns to various kinds of subsidiary occupations, see the next section of this paper.

[58] I have argued elsewhere that accounts of rural destitution during the 1930s in Wuxi reflect the impact of a collapsing world silk market on a regional economy that had become increasingly dependent on revenues from the production of cocoons and raw silk. See Bell, "Explaining China's Rural Crisis."

will present a more extensive analysis of these materials in a future book manuscript. Here I give only preliminary findings from a small sample of the total number of households surveyed. What is quite useful about these data is that they were collected before the depression, yet they reveal a nearly identical pattern of land and labor usage to that seen in the SMR survey.^[59]

When working with this second data set, my first concern has been to compare land and labor productivity in rice/wheat farming with mulberry cultivation and silkworm raising. If my original hypothesis is correct, then we would expect the results of Table 7.1 to be duplicated—namely, that overall cash earnings of the family might be raised via mulberry cultivation and silkworm raising but that labor productivity would be lower. We would also expect to see a preponderance of small peasant family farms operating on small plots, using primarily their own family labor for a combined effort in rice/wheat farming, silkworm raising, and other subsidiary occupations. In order to specify more precisely the significance of the size of farming units, I have also tried to introduce another comparative dimension into the data analysis, that is, between larger and smaller farming units, to see if there were any economies of scale at work and, if so, what size of farm produced the most efficient use of land and labor resources.

I have constructed a set of six tables from a 31-household sample selected from the three villages of Suxiang, Maocun, and Baishuidang.^[60] Table 7.2 presents the size distribution of farming units. Although these farms were slightly larger on average than those in the SMR survey, even the very largest farms worked only between 20 and 38.5 mu . In fact, in the entire sample of 800 households, only 25 households worked 20 mu or more. Where size did seem to make some difference was in the decision of what proportion of farmland would be planted in mulberry. Very small farms sometimes planted all of their land in mulberry, an indication that they may have hoped that a more intensive effort in sericulture would increase their overall earning

[59] Guoli Zhongyang Yanjiuyuan, selected questionnaires from Suxiang (Beiyanxia Township), Maocun (Kaiyuan Township), and Baishuidang (Yangming Township). In the 1950s and 1960s, the Chinese scholar Zhang Zhiyi was working through these materials; however, during the Cultural Revolution his manuscript was lost. Since that time, the materials have been housed at the Institute of Economics in Beijing. I am grateful to Professor Chen Hansheng, who led the original survey, for telling me that the materials were still at the Institute, and to Professors Dong Fureng and Peng Zeyi for arranging access to them. Shang Lie of the Modern History Section of the Institute also has helped to sort the materials, and Professor Liu Kexiang has helped familiarize me with much of the special terminology used, especially in matters of landholding.

[60] I have chosen the 31–household sample using three criteria. First, I have aimed for roughly the same proportions of large, medium, and small farms found in the entire group of 800 survey questionnaires. Secondly, these 31 households came from three ecologically different villages. And finally, these households had no missing data in the categories relevant to the argument I present below.

(Table continued on next page)

(Table continued from previous page)

(Table continued on next page)

(Table continued from previous page)

potential despite the riskiness of such a venture. This never happened among medium-size or larger farms.

Table 7.3 looks at yields per mu of different types of farmland—a rough measure of land productivity. The results among different-size farms are ambiguous, indicating that economies of scale probably did not apply in this system. Larger farms did not get consistently higher yields per mu than small farms in any crop. But what does seem of importance in this table is a higher degree of variance in both physical yields and monetary returns for mulberry cultivation as opposed to rice/wheat farming. Physical returns for Rice 1 for Maocun, for example, varied by roughly a 2 to 1 ratio, whereas physical returns for mulberry varied by 5 to 1. Likewise, financial returns varied by less than 3 to 1 for Rice 1 but varied from -4.00 yuan to 22.00 yuan for mulberry, a range that included the possibility of negative returns for some households. The reason for the greater variance in financial returns to mulberry can be attributed to at least two factors: the difficult climatic conditions for silkworm raising that prevailed in Wuxi and the vagaries of daily market prices for both cocoons and mulberries. We have seen that Wuxi's springs were cold and damp, often leading to bacterial infection among silkworms and the total destruction of some households' yearly crops. In addition, prices for cocoons and mulberry leaves varied widely from one day to another and also among marketing locations. Both these factors meant that some families spent more fertilizing their mulberry fields and buying silkworm eggs than they were able to earn by selling their cocoon crop.^[61] The result was that mulberry cultivation, with its widely varying financial returns, was much riskier than devoting land to a combination of rice and wheat farming.

Tables 7.4, 7.5, 7.6, and 7.7 explore questions of labor usage and labor productivity. Tables 7.4 and 7.5 summarize labor usage. In Table 7.5 we see that the majority of labor days in silkworm raising were performed by household labor, especially that of women.^[62] Comparisons among villages are also important. We see in Baishuidang, for example, that all households put more total labor days into mulberry cultivation and cocoon rearing than into rice/

[61] On daily and hourly fluctuations in cocoon prices see "Jianshi zaxun" [Miscellaneous reports on cocoon markets], Minguo ribao , June 4, 1917, sec. 2:7; "Wuxi—shiju bujing zhi shangshi" [Wuxi—markets in chaos], Minguo ribao , June 4, 1917, sec. 2:7; and "Zhijie shouhai zhi shangkuang" [Trade experiences immediate adverse effects], Minguo ribao , June 5, 1917, sec. 3: 10; on problems in the erratic pricing of mulberry leaves according to widely fluctuating demand, see Wang Xuexiang, "Gaishan sangye maimai jigou zhi guanjian" [My opinions on reforming marketing mechanisms for mulberry leaves], Zhongguo canci 2, no. 7 (Jan. 1937): 92–93.

[62] In the Suxiang households, there are slightly more labor days in silkworm raising performed by men. This is an unusual aspect of the labor distribution among the ten households sampled here. Calculations for a larger sample of 141 households across 13 villages show that 69 percent of silkworm raising days were performed by household women. See Lynda S. Bell, "Chinese Women and the Value of 'Surplus Labor,'" (Los Angeles, 1990) pp. 12–13.

wheat farming. This suggests that in future work with these data, it will be worth testing intensity of labor effort in sericulture against size of landholdings as well as the availability of other kinds of subsidiary employment in various villages. Each of these factors probably influenced a peasant family's decision to work more intensively at sericulture. If there was to too little land or too few opportunities for other outside employment, a family would have no choice but to work more intensively at their sericulture effort.

Table 7.6 examines the composition of family labor units. Three things are striking about this table. First is the number of adult males by size of farm. Small farms almost always had only one adult male, suggesting that division of household property in each generation (fenjia ) was an important downward leveling force. Second, only the largest farms hired laborers by the year. Since large farms were very rare in Wuxi, this is an indication that the demand for full-time agricultural laborers was quite small. Finally, most farms, regardless of size, hired some day labor. However, as in the villages of the SMR survey, this method was used only to fill temporary labor needs, usually at peak periods of planting and harvesting. The most intensive period came in May and early June, when peasants were tending to delicate rice seedlings, raising silkworms, marketing cocoons, and harvesting winter wheat. Overall, this table shows that with the conjugal family as the principal unit of production, most farms remained quite small and hired additional labor only as short-term labor needs dictated.

Finally, Table 7.7 looks at the important issue of labor productivity. The most conclusive data of the analysis thus far are found in the final column of this table: returns per unit of labor were, as anticipated, almost always lower for mulberry cultivation and silkworm raising than for rice/wheat farming. This duplicates in more precise terms my former analysis of the SMR data, confirming that participation in sericulture raised family income overall even as it lowered labor productivity in terms of net income per labor day for the household as a whole. Thus, as before, I would argue that sericulture was a form of labor intensification in which peasant families engaged to allow their continued survival under conditions of high population density and a relatively low land-labor ratio.

What this analysis has not yet provided is an answer to why sericulture was chosen over other forms of subsidiary work, including factory employment and other handicraft options. It is true, for example, that factory work brought relatively high cash returns. Data from the 1929 survey on factory wages indicate that child female factory workers could earn about 0.10 yuan per day while women made 0.40 to 0.50 yuan per day. Comparing these sums to those observed for returns to sericulture in the final column of Table 7.7 shows that earnings per day for work in sericulture were often roughly comparable to factory wages but not as reliable, with a great deal of variation possible. Despite the better wage-earning conditions, the survey also indicates

(Table continued on next page)

(Table continued from previous page)

(Table continued on next page)

(Table continued from previous page)

(Table continued on next page)

(Table continued from previous page)

(Table continued on next page)

(Table continued from previous page)

that relatively few females were employed in factories in Wuxi in 1929, a situation that may be explained by factors other than cash-earning considerations. When peasant girls and women were sent to factories, their prolonged absence meant that families were deprived of their work in domestic services such as cooking, child-care, and animal tending. For many peasant households, therefore, the loss in female services offset the benefits of a reliable cash wage from factory work. As for the possibility of substituting other cash-earning options within the household setting, compared to sericulture they all brought far fewer returns to labor: only 0.03 yuan per day for making hemp rope, 0.02 yuan per day for weaving cotton, and 0.03 yuan per day for making lace. These meager amounts made such options far less attractive than sericulture as cash-earning alternatives within the household setting.^[63]

Peasant Farming and Problems of Silk Industry Growth

Given the constraints of environment, demography, and available options for earning cash income, the Wuxi peasant decision for sericulture was entirely rational. At the same time, however, Wuxi peasants did not earn exceptionally high incomes through this route. As they intensified their labor effort in sericulture, they remained fairly close to subsistence-level living conditions, spending most of their cash income to satisfy yearly consumption demands. These facts allow me to comment briefly on the pitfalls of assuming any close relationship between peasant rationality and the potential for new forms of economic growth.

Because sericulture was primarily a method used to augment family income to satisfy immediate consumption demands, little surplus income was available for technical improvements. Although the most modern methods of silkworm breeding were known in Wuxi, the majority of small peasant family farms were not able to introduce them. These techniques included crossbreeding of silkworm egg varieties to strengthen disease resistance; refrigeration of silkworm eggs to allow for more efficient timing of summer and fall cocoon raising; close regulation of temperature during the silkworm-raising process; and use of bacterial disinfectants to protect delicate silkworms in their early stages of growth.^[64] Some households had access to improved egg cards via state-supported and privately run egg breederies, but these improved cards had to be purchased by the peasant households. They were rarely given free of charge even on an experimental basis and thus required

[63] Guoli Zhongyang Yanjiuyuan, selected questionnaires, table 11.

[64] For further discussion of problems of silkworm raising and methods of sericulture improvement, see Bell, "Merchants, Peasants, and the State," pp. 49–50, and chap. 4; and Dierlishi Dang'anguan, file no. 3504, "Jiang-Zhe-Wan-Hu qudi canzhong qingxing."

yearly cash outlays that peasant households often did not or could not make.^[65] Methods of temperature control and disinfectant use were even more difficult to implement because silkworms were reared within peasant family living quarters. The most efficient way to implement these improved methods would have been to build separate facilities designed especially for silkworm rearing.^[66] This option was simply not within the means of most Wuxi peasant households. The results were clear enough: the 1929 survey data from Wuxi show a high rate of yearly cocoon crop failure, with many households routinely losing an entire season's efforts. Moreover, the quality of cocoons remained low compared to rapidly improving Japanese varieties, and the quantity of cocoons needed to make equivalent amounts of raw silk remained high.^[67]

Poor raw material supply does not tell the whole story of the problems encountered by China's silk industry. I raise this issue here only to suggest that the purist approach to peasant rationality does not do much to explain the larger developmental issues at stake. Ultimately, a fuller analysis of silk industry growth should also include discussion of capital markets, the behavior of new bourgeois investors, political institutions, and the interrelationships of all such factors to the rural base.^[68] To begin such an analysis, a modified understanding of peasant rationality is useful as a first step. We should realize that small peasant family farming in Wuxi, a perfectly rational system of peasant behavior, may have been more of a hindrance than a help to silk industry growth.

[65] Guoli Zhongyang Yanjiuyuan, selected questionnaires, table 8, shows only a moderate rate of improved silkworm egg usage in Wuxi in 1929, far less than I believed previously on the basis of information available for the mid-1930s. See Bell, "Merchants, Peasants, and the State," p. 234.

[66] When I visited sites of silkworm raising in Wuxi in 1980, I observed that temporary thatched, windowless structures put up within mulberry fields and collectively managed by groups of peasant women were quite efficient for maintaining temperature and draft control. Coal stoves were kept burning inside constantly, and the temperature was carefully monitored. See also Dierlishi Dang'anguan, file no. 3504, "Liangnianlai," p. 4, for a discussion of the construction of silkworm rearing facilities within sericulture experimental stations designed to control drafts and temperature change. A report on a trip by a U.N.-sponsored group to sericulture areas in central China in 1979 also discussed the relative success of collective silkworm-rearing enterprises in maintaining conditions of constant temperature and humidity. See Food and Agriculture Organization of the United Nations, China: Sericulture. Report on a FAO/UNDP Study Tour to the People's Republic of China, 6 May to 4 June 1979 (Rome, 1980), pp. 38–42.

[67] According to personal accounts of former sericulture improvement specialists, it took 6 dan of dried cocoons to make 1 dan of Chinese raw silk. In Japan, because improved silkworms produced cocoons with fibers of increased length and strength, only 3.5–4 dan of dried cocoons were needed during the 1920s to produce 1 dan of raw silk. See Bell, "Merchants, Peasants, and the State," p. 217.

[68] This fuller analysis will appear in a book now in preparation.

In this chapter, I have provided a long-term view of peasant decision making in Wuxi. By looking at trends in rural development over three centuries, the switch to sericulture in Wuxi can be seen as a product not only of new opportunities brought about through world market integration but also of long-term demographic trends and the development of subsidiary activities to increase the income-earning potential of individual peasant families. Another focus here has been on the internal workings of farm households to understand more precisely what motivates their actions. I have shown that under conditions of a low land-labor ratio and few options for nonfarm employment, peasants were willing to work at sericulture despite the low and unreliable income that it provided. What they gained was a chance to better employ peasant women and increase the income-earning potential of the family as a whole. Although these actions were entirely rational, peasant families in Wuxi worked close to subsistence levels so that surplus earnings for investment were slim. These findings suggest that future studies of Chinese development should take into careful account the evolution of peasant family farming systems in various regions and the nature of their impact on long-term patterns of economic growth.