Was There a “Thirteenth-and-Fourteenth-Century Turning-Point”? Population, Land, Technology, and Farm Management

Although the historiography of China’s mainland has long been relatively isolated from Western scholarship and the Chinese approach to history is quite different from that found in the West, the idea that a turning point in the Chinese economy occurred between middle and late imperial times is common to both.1 This shared conviction is no mere coincidence. Both traditions of scholarship presume that China’s economy was the foremost in the world during the midimperial period but then lost its vigor and slid into “stagnation.” Although scholars dispute exactly when and why this turning point took place, most date it to the period extending from the closing decades of the Southern Song to the f irst decades of the Ming, that is to say, from the early thirteenth to late fourteenth century. As I hope to demonstrate in this chapter, although the notion of a “fourteenth-century turning point” in Chinese economic history seems reasonable, it does not hold up to closer scrutiny.

First, historians disagree over the def inition of “turning point”. Some regard it as a great change in patterns of economic growth. They posit that the Song experienced “growth accompanied by the invention of new techniques of production,” but in the late imperial period “invention was almost entirely absent” (Elvin 1973: 203). Other scholars identify the turn in terms of a change in the speed of economic growth. That is, although there was no signif icant change in growth patterns, China’s economic growth slowed markedly over the course of the thirteenth and fourteenth centuries (Fu Yiling 1982; Qi Xia 1987: 32).

Second, the idea of a turning point is premised on hypotheses about two other stages of Chinese economic history, the “Song revolution” (or “medieval economic revolution”) and the “Ming-Qing stagnation” (or “late imperial stagnation”). Since these two stages differ qualitatively, by def inition the transition between the two during the Song-Yuan-Ming period becomes a turning point. Thus, the concept of a “turning point” inevitably hinges on our understanding of what came before and what comes after. Without either the Song revolution or Ming-Qing stagnation, a fourteenth-century turning point would be unthinkable.

Third, certain presumptions about the Yuan Dynasty are crucial to the idea of a fourteenth-century turning point. The Yuan has been seen as a dark age in Chinese history, a “conquest dynasty” that engaged in misrule and large-scale violence. Recently, however, such negative views have been challenged by a small but increasing number of Yuan historians who argue that China’s economy grew substantially during the period (Han Rulin 1986: 4, 378, 387; Shi Daogang 1984). Recent research has also shown that the disruption and disorder arising from the rise and fall of the Yuan were not as severe as once thought.

Finally, the idea of a fourteenth-century turning point is incompatible with new developments in the study of premodern Chinese agricultural history. The consensus among Chinese agricultural historians is that the Yuan, together with the Song, was a stage of “great agricultural progress in South China,” whereas the Ming and Qing were a stage of “intensive development” (Guo 1988: 179, 343). If this indeed were the case, it would be diff icult to speak of a turning point.

Although other new developments in the f ield of Chinese history have undermined the idea of a fourteenth-century turning point, it is necessary to address the issue directly. To judge whether there really was such a shift, we need not only to make a detailed analysis of the changes that happened during these two centuries but also to look at them from a wider perspective. There is no doubt that a comprehensive analysis of economic changes across a huge and diverse country like China is a daunting, if not impossible, task. Rather than attempt a comprehensive survey of the Chinese economy during the Song-Yuan-Ming transition, this chapter focuses on Jiangnan 江南 agriculture, a subject for which we have relatively good data. 2Jiangnan was not only the most economically advanced area of China from the late Tang down to modern times but also the most culturally prosperous. Hence it boasts rich historical records. My focus is trained on agriculture because despite the development of other sectors, agriculture was the basis of the Jiangnan economy during the period under study. Thus, to determine whether there really was a turning point, we must examine changes within the agricultural sector. I will concentrate on population, cultivated land, technology, and farm management.

POPULATION CHANGE

Few reliable demographic data exist for thirteenth-and fourteenthcentury Jiangnan. Many historians have argued that the region was beset by acute population pressure during these two centuries.3 Yet, demographers have typically described this period as an era of depression marked by a disastrous loss of life. These two opinions so sharply contradict each other that both would seem to be discredited. After reviewing the factors affecting demographic change, I will offer new estimates for Jiangnan’s population in the late Southern Song, Yuan, and early Ming. These f igures will enable us to chart short-and long-term trends in demographic change and provide a basis for evaluating changes in the size of Jiangnan’s agricultural labor force during the Song-Yuan-Ming transition.

Causes of Population Decline

Almost every Malthusian “positive check” occurred during the Song-Yuan-Ming transition: military conquest, civil war, famine, and epidemic. Each event caused a great loss of life. Natural disasters also decimated China during this period. According to climatic historians, these two centuries experienced colder temperatures than the preceding or succeeding eras (Zhu 1973). The unfavorable climate led to more frequent droughts and floods (Shuilibu 1984: 207, 209-15, 236-37). There were, in addition, many outbreaks of epidemic diseases during this period (Cao 1995).4 Jiangnan did not escape these misfortunes. Large-scale famines and epidemics took many lives and forced much of the population to become refugees. Although order was restored and large-scale violence ceased with the founding of the Ming Dynasty, the forced emigration of wealthy Jiangnan families imposed by the Ming founder also caused population loss. Qiu Shusheng and Wang Ting (1983) estimate that the total population of the middle and lower Yangzi Valley provinces fell by 40 percent during the Yuan-Ming transition. There is no doubt that Jiangnan suffered considerable mortality during this period, but a closer look at the demographic change in the region shows that the population loss has been greatly exaggerated.

During the past two decades, there has been a tendency to attribute massive mortality to an outbreak of the bubonic plague introduced by Mongol horsemen, who are believed to have brought the Black Death to Europe. J. D. Durand (1960: 233) argues that in the last phase of Mongol rule “the pandemic of bubonic plague raged no less f iercely in China than it did in Europe and caused a great number of deaths.” William McNeill (1976: 141-42, 259-69) suggests that the plague played at least as great a role as Mongol ferocity in reducing China’s population (see also Cao 1995). Yet the hypothesis that during this period the devastation wrought by the bubonic plague in China was comparable to the destructions of the Black Death in Europe and the Middle East remains unproven. Robert Hymes (1997) found no decline in life expectancy among his sample of elite households over the period 1040-1374, which is hardly consistent with a severe decline in population.

Moreover, Jiangnan escaped the worst of the disasters that afflicted other parts of China. By the time the Mongol troops conquered Jiangnan, they had by and large abandoned their tactic of massacre, and the, occupation of South China was relatively bloodless (Ge 1991: 213-18, 222). During the seven decades between the Yuan conquest and the outbreak of the Red Turban rebellion in the 1350s, the loss of life due to war and violence was minimal (Shi Weimin 1996: 264). Yuan’s rule brought stability to Jiangnan. Indeed, the period was regarded as an “age of peace and prosperity” by many contemporaries.5 Even during the Yuan-Ming transition, the loss of life was not as great as once thought. Generally speaking, Jiangnan enjoyed stability and peace until 1355. In that year, Yuan rule in Jiangnan collapsed and Zhang Shicheng’s 张士诚 troops marched from Gaoyou 高邮, north of the Yang-zi, to Suzhou, occupying the city without any resistance. Zhang soon eliminated his opponents in eastern Jiangnan and established the Wu 吴 kingdom, with Suzhou as the capital. That same year, Zhu Yuanzhang founded his Western Wu 西吴 kingdom at Nanjing. Since both Zhang and Zhu had to win over local people, they did not wreak havoc in areas under their rule. Most of the military action took place on the periphery of the region, and the loss of life in Zhu’s rise to power was not great (Lin 1983).

Nor did disasters play the same role in decimating Jiangnan’s population as they did in other parts of China. The frequency of droughts and floods in the plains of Jiangnan was only 21.2 per 100 years in 1224-1403, in contrast to 37.0 in 1011-1223 and 41.6 in 1404-1900 (Zheng Zhaojing 1987: 256). In addition, there is little convincing evidence that epidemics reduced the population of Jiangnan as they may have done elsewhere in China during the thirteenth and fourteenth centuries. The incidence of epidemics was not nearly as high in Jiangnan, and deaths from epidemics appeared fairly moderate compared to the situation in the Southern Song (Cao 1995). In short, Malthusian checks did not reduce Jiangnan’s population as much as is commonly believed.

Zhao Wenlin and Xie Shujun (1988: 310, 326-27) estimate that the population of Jiangsu province increased from 3.5 million in 1210 to 8 million in 1290, and that of Zhejiang from 7 to 12 million over the same period. Their f igures for 1210 are probably too low, but Zhao and Xie’s estimates imply that population loss during the Song-Yuan transition had no signif icant influence on the growth of the population in this period. Moreover, population loss in Jiangnan during the Yuan-Ming transition was far less severe than in China as a whole. Zhao and Xie (1988: 338, 374) estimate that the population of Jiangsu fell from 9 million in 1351 to 7 million in 1381, and the population of Zhejiang was reduced from 11 to about 10 million. If these estimates are valid, the population loss during the Yuan-Ming transition was less than 10 percent.

Despite some population loss during the two episodes of dynastic transition, Jiangnan enjoyed net population growth across the thirteenth and fourteenth centuries. Zhao and Xie suggest that Jiangnan saw considerable population growth in the f irst half of the thirteenth century and that population growth resumed again after 1290. The Yuan government’s administrative reclassif ications of Jiangnan counties also provide evidence of population growth. In the 1295 adjustment, Haining 海宁 and Chongde 崇德 in Jiaxing prefecture, Kunshan 昆山 and Wujiang 吴江 in Suzhou prefecture, Jiading 嘉定 in Songjiang prefecture, and Yixing 宜兴 and Wuxi 无锡 in Changzhou prefecture were promoted from “low-level counties” (xiazhou 下州) to “midlevel counties” (zhongzhou 中州). This reclassif ication implies that the number of registered households in each county rose from the 40,000-50,000 to the 50,000-100,000 range (Yang Peigui 1975: 106). On the national level, population growth halted in the early 1340s, but in Jiangnan it lasted a decade longer, since this region did not suffer large-scale disorders until the mid-1350s (Ge 1991: 218). Finally, population growth recovered in the last decades of the fourteenth century. For example, in spite of the forced emigration of wealthy households, Suzhou’s population rose from 1.95 million in 1371 to 2.16 million in 1376 and then to 2.36 million in 1393; an increase of 20 percent within two decades (Fan and Xia 1993: 62-63).

We can conclude that although Jiangnan suffered some loss of population in the Song-Yuan-Ming transition, the decline was not as large as elsewhere. The two centuries also saw long periods of population growth that offset these losses. Jiangnan’s population fluctuated over the course of the two centuries, but the fluctuations fell within a relatively narrow range.

Secular Trends in Jiangnan’s Population

Shiba Yoshinobu has projected demographic trends for Jiangnan at the pre-fectural level during the four centuries between 980 and the 1390s. Table 1 provides rough estimates of Jiangnan’s population for the late Southern Song, the Yuan, and the early Ming based on Shiba’s reconstructions (1988: 144-46), modif ied with data from Liang Fangzhong (1980: 181-82) and Cong Hanxiang (1984). 6

Data for all three periods are available only for Suzhou, Hangzhou, and Jiangning (Nanjing). The total population in these three prefectures was 4,191,000 during the late Southern Song, 5,267,000 in the early Yuan, and 4,727,000 in the early Ming. The ratio of population for the three periods is 100: 126: 113. The ratio for all seven prefectures between the Yuan and early Ming is 100: 88, which is roughly equal to the Yuan: Ming ratio of 100: 90 for Suzhou, Hangzhou, and Jiangning. This ratio of 100: 126: 113 seems closer to reality than alternative estimates. 7This ratio implies that the total population of early Ming Jiangnan was a tenth larger than that of the late Southern Song and a tenth less than that of the early Yuan. Accordingly, we can infer that the population of Jiangnan as a whole was about 8 million in the late Southern Song, 10 million in the early Yuan, and 9 million in the early Ming.

The Yuan and early Ming f igures (especially the latter) are based on more complete information and thus are more reliable. The estimate of 8 million for late Southern Song Jiangnan, while derived from only three pre-fectural f igures, is consistent with other estimates that reckon the region’s population in the range of 5.2-6.0 million in the late Northern Song (Fu Zongwen 1988: 125; Liang Fangzhong 1980: 153), rising to about 6.3-7.4 million by the 1220s. 8

The long-term demographic trends in Jiangnan are shown in Tables 2 and 3, which report off icial population f igures (in households) for the Tang, Song, Ming, and Qing. These data reveal two contrasting trends before and after the Song-Yuan-Ming transition. First, during the Tang and Northern Song, the population grew far more rapidly in eastern Jiangnan, particularly in Suzhou, Songjiang, and Jiaxing, than in the hilly and mountainous west. By the early Ming, the population of the four eastern prefectures of Suzhou, Songjiang, Jiaxing, and Changzhou accounted for 60 percent of the total population of the region, a ratio that persisted up to the mid-Qing. These prefectures accounted for only 30 percent of the region’s total population in the early Tang and 50 percent in the late Northern Song. Second, by the early Ming nearly all the region’s arable land had been brought under cultivation, and population growth gradually slowed. Worth noting is that the deceleration of population growth mainly resulted not from Malthusian positive checks but from artif icial population control, already evident in the thirteenth and fourteenth centuries (Li Bozhong 1994b).

Major Features of Demographic Change in Jiangnan

Three major features characterize the changes in Jiangnan’s population during this period. First, if the estimates given above are valid, Jiangnan’s total population fluctuated over the course of the two centuries, within a comparatively narrow range of ±10 percent. Moreover, despite these fluctuations, Jiangnan’s population showed a net increase. Although the increase does not seem large—only 10 percent over two centuries—it was still considerable compared to that found in other parts of China or Europe during the same period.9

Second, the population increase was not spread evenly throughout the region. The population of Suzhou and Jiangning rose dramatically from the late Southern Song to the early Ming, but that of Hangzhou and Zhenjiang fell sharply. In addition, although the population of Jiaxing increased slightly from the Yuan to early Ming, it decreased in Changzhou and Huzhou. These trends demonstrate that the increase in population over the course of the two centuries took place mainly in the low-lying plains of eastern Jiangnan, a continuation of a trend underway since the Tang. The decrease in population occurred mostly in prefectures located in the upland areas of western and northern Jiangnan. By 1400, the region’s population was heavily concentrated in the delta’s alluvial plain.

Third, the pattern of population growth also suggests that the major dynamic behind this growth was intrinsic to the regional economy. The presence of political and military centers was not as crucial to growth as is commonly thought. The movement of the imperial capital from Hangzhou to Dadu (Beijing) after the Mongol conquest and back to Jiangnan with the founding of the Ming had little influence on the distribution of population within the region. The underlying causes of demographic change were rooted in structural changes in the regional economy rather than in ephemeral political fortunes.

This general picture of Jiangnan population growth coincides with Shiba’s study of subregional demographic change in Huzhou prefecture. Huzhou prefecture is typical of the human ecology of Jiangnan as a whole: the plains in the east and the hills in the west each comprised half of the prefecture’s total area, and Huzhou had no large cities serving as an imperial or provincial capital. The prefecture numbered about 200,000 registered households in the early Southern Song and about 220,000 in the early Ming, an increase of one-tenth. In the Southern Song the eastern plains and the western hills each accounted for half of the total population, but by the early Ming 70 percent of the total was concentrated in the plains (Shiba 1988: 382-86). Shiba’s study of this subregion corroborates our f indings for the region as a whole.

Urban Population

Although he coined the term “ (Song) revolution in urbanization,” Mark Elvin displays caution in his estimates of the degree of urbanization in the Song. Elvin (1973: 176) believes that a f igure of at least 10 percent is reasonable for the empire as a whole, and that the percentage may have been considerably higher in more advanced areas. Kang Chao (1995: 76), convinced that the Southern Song was “the summit of urbanization” in premodern Chinese history, claims that the urban population was as high as 22.5 percent of the total in the Southern Song. Chao’s estimate appears to be too high, but it is true that a handful of extant thirteenth-century f igures suggest that the urban proportion of Jiangnan’s population was very high, as much as 48 percent of the total in Hangzhou prefecture ca.1265-74, and 37 percent in Dantu 丹徒 county of Zhenjiang prefecture ca.1208-24. But Hangzhou, the imperial capital, and Dantu, a major garrison town for defense against the Jurchen, cannot be considered representative of the region as a whole. If military and logistical personnel and their families were excluded, the number of urban residents would have been far smaller. For example, in Jiankang 建康 (Nanjing) soldiers and their families comprised about half of the city’s 300,000-400,000 residents (Chen Zhen 1996: 133-34). Katō Shigeshi (1952-53 b: 336) estimated the population of Hangzhou, the Southern Song capital, at about 900,000. Of this number, members of the imperial household, off icials, military men, and wealthy families accounted for three-quarters and ordinary citizens for only one-quarter. Including the city’s suburbs, the population of Hangzhou perhaps totaled t. s million, with ordinary citizens constituting only 40 percent of the total (Shiba 1988: 247). Thus the degree of urbanization was much lower than it appears.

Except for the three great cities of Hangzhou, Jiankang, and Suzhou, Southern Song Jiangnan cities were relatively small in size. It is true that the number of market towns increased in the late Southern Song.10 The largest market towns, such as Ganpu 澉浦 in Jiaxing prefecture and Linping 临平 in Hangzhou prefecture, had a thousand or more households, but they seem to have been exceptional. The smallest ones, like Zhuxiaogang 竹筱港 and Shepogang 佘婆冈 in Jiangning prefecture, had only dozens of residents, even fewer than those of a large village (Fu Zongwen 1988: 183-84). Generally speaking, the level of urbanization of Jiangnan as a whole is unlikely to have been much more than 10 percent in the late Southern Song.

The level of urbanization in Jiangnan declined modestly during the Yuan and substantially in the Yuan-Ming transition.11 The early Ming population of Hangzhou city was much lower than during Yuan times.12 Suzhou city also witnessed a considerable decrease, although not as great as that of Hangzhou (Fan and Xia 1993: 164-65). But these decreases were offset to some extent by the increase in the population of Nanjing, the new imperial capital. Nanjing’s population doubled within the f irst three decades of the Ming, to approximately 700,000 residents (Fan 1996: 157). The total urban population of Jiangnan in the early Ming was lower than the Yuan peak, but not by much.

In sum, during the two centuries from the late Southern Song to the early Ming, both the total population of Jiangnan and the ratio of the urban population to the overall population fluctuated, but within fairly narrow margins. The rural population does not appear to have changed much at all, which suggests that the agricultural labor force remained quite stable.

CHANGES IN CULTIVATED LAND

The changes in cultivated land in Jiangnan during the thirteenth and fourteenth centuries were not only quantitative but also qualitative. Many scholars have argued that since the population increased during this period and cultivated land did not, the labor/land ratio grew more unfavorable (Qi Xia 1987: 73, 109; Fu Zongwen 1988: 315; Liang Gengyao 1984: 17; Fan and Xia 1993: 63; Cong 1984; Chao 1986). But as pointed out above, the population increase in this period was moderate at best. If the quantity of cultivated land increased or if its quality improved, the labor/land ratio would ease, not worsen.

Changes in the Amount of Cultivated Land

Among all pre-1950 f igures on cultivated land in Jiangnan, those of the 1393 survey are believed to be second only to the 1578-82 survey data in reliability.13 But the 1393 survey has numbers only for the f ive southern Jiangsu prefectures of Suzhou, Songjiang, Changzhou, Zhenjiang, and Jiangning (then known as Yingtian 应天). For the northern Zhejiang prefectures of Hangzhou, Jiaxing, and Huzhou, we have to rely on the 1578 f igures. 14A few Southern Song and Yuan prefectural f igures can be found in other sources. In most cases, the pre-1382 Ming data for Jiangnan was copied from Yuan records with only minor changes (Ho 1988: 49). Table 4 lists the prefectural f igures that I have collected.15

These data suggest that cultivated acreage in Jiangnan remained fairly stable between the Yuan and early Ming, with the most notable expansion of arable found in Suzhou and Songjiang.16 The Ming f igures in Table 5 represent the taxable land quotas established on the basis of the cadastral surveys of 1393 and 1578 (with a few changes) and thus can be seen as reflecting the actual amount of cultivated land in Ming Jiangnan.

The declines in arable shown by the f igures in Table 5 can be explained by changes in the boundaries of administrative units or in classif ication categories. For example, the Southern Song f igure for Changshu is much greater than the Ming f igure because the county was divided in two with the creation of Taicang太仓 county in 1315. The Southern Song f igures for Jiangning, Shangyuan, and Lishui (all in Jiangning prefecture) are considerably smaller than the Ming numbers because the Song data include only one category of land, tian 田(paddy), whereas the Ming encompassed four categories of land: tian, di 地 (dry f ields), shan 山 (hills and hill f ields), and dang 荡 (marshes). Zhou Shengchun (1981: 31-52) estimates that the total amount of cultivated land in the Lake Tai basin in the Southern Song was about 28 million mu, which is close to the region’s present-day total.17 Among these 28 million mu of cultivated land, about to million were diked enclosures or “polders” (weitian 围田, 圩田), most opened for cultivation during the Southern Song. From Zhou’s research, it is clear that in the main agricultural areas of Jiangnan—the Jiangnan plain and the Danyang丹阳 plain in Zhenjiang prefecture-most arable land had already been brought under cultivation by the end of the Southern Song. Contemporary observers conf irm that little uncultivated land remained. 18

Improvement of Arable

Although the amount of cultivated land in Jiangnan increased only slightly, signif icant changes took place in its quality. The improvement resulted, on the macro-level, mainly from the construction of effective flood control and irrigation systems and, on the micro-level, from increases in the productivity of the soil. Effective irrigation and drainage networks were crucial to land improvement, given the particular geography and ecology of Jiangnan.

Topographically, Jiangnan consists of three parts: the plains in the east, lying less than 6 meters above sea level; the Ningzhen 宁镇 (Jiangning-Zhenjiang) hills to the northwest, generally 20-40 meters or more above sea level; and the Zhexi 浙西 (Western Zhejiang) mountains to the southwest (Li Bozhong 1985 b). In most of western Jiangnan, in particular the Ningzhen hills, rainfall provides the main source of water. Since rainfall fluctuates considerably year by year, local farmers store rainwater to secure an adequate supply of water (Qi Yanping 1963). The number of water conservancy projects carried out in the Ningzhen hills during the Southern Song and Yuan was far greater than during any previous era. This increase was related to the reclamation of f ields in the hills and mountains (Zhang 1994). Apart from building weirs and mid- or largescale reservoirs, small projects within the means of a single family—such as ponds—also were an effective means of storing rainwater (Osawa 1993: 107). In the Yuan and early Ming, the technique of digging ponds was greatly improved and spread so widely that it was recorded in a popular encyclopedia.19 Moreover, the method of digging ponds described in this encyclopedia was more eff icient than that found in Chen Fu’s 陈旉 Treatise on Agronomy 农书 (1149), which utilized bamboo-tube pumps to obtain water from deeply dug ponds (Li Bozhong 1985 b;1998a: 64-66). The invention and widespread use of this technique surely increased the supply of water and thus land productivity.

The hydrography of most of eastern Jiangnan was quite different from that of the western part. In the prefectures of Suzhou, Huzhou, Changzhou, and Jiaxing, observed by one Yuan author, 70 percent of all cultivated land lay less than six meters above sea level.20 Because of the high water table and the numerous rivers and lakes throughout this area, these low f ields were susceptible to inundation. Heavy rainfalls or discharges of water from upstream areas often threatened the region with floods. Yang Ju 杨矩, writing in the Southern Song, stated: “The damage caused by floods is greatest in Suzhou, Xiuzhou (Jiaxing), and Huzhou, where the lands lie lowest.”21 Even when they were not covered by water, low f ields suffered from excessive dampness, which was harmful to the soil and reduced its fertility (Gao and Li 1992: 391). For these reasons, agricultural production in the Lake Tai basin was unstable. This was true even in the Northern Song, before large-scale reclamation and dike building began, when the natural hydrographic system was in relatively good condition (Han Maoli 1993: 103).

The two key methods used to avoid floods and reduce the water content of the soil were building dikes and dredging rivers.22 But their utilization varied over time. In the Southern Song, dike building was the most important method of water control. Polders were built to reclaim marshes and lake bottoms.23 But polders interfered with the natural hydrography, eliminating the flood plain and intensifying the problem of poor drainage in the Lake Tai basin. To solve this problem, the second method, dredging rivers to increase their runoff capacity, was crucial. Although some attempts to dredge rivers were made in the Southern Song, it was in the Yuan and especially the early Ming that more serious efforts were mounted (Zheng 1987: 16, 27). In the Yuan, the Wusong River 吴淞江and adjoining lakes and swamps were dredged to drain the summer floodwaters from Lake Tai. The dredging of the Liu River 浏河 and its branches leading to the Yangzi also began at this time. Thanks to these efforts, the frequency of floods and droughts was far lower in the Yuan than in the Southern Song. In the early Ming, even greater efforts were made in dredging the Wusong, Liu, and Huangpu 黄浦 rivers and Lake Dianshan 澱山湖. Canals were built to allow water from these major watercourses to drain into the Yangzi or the sea. The largest project involved draining water from the Wusong River into the Yangzi via the Liu River. Completed in 1403, it provided a new and better outlet for excess water in the Lake Tai basin and soon became the main outlet for all of the delta and reduced the incidence of floods in the region (Zheng 1987: 16, 17, 29).

The state played an active role in water control. In the Southern Song, the government organized seven large-scale projects that mobilized as many as 200,000 laborers (Liang Gengyao 1984: 165). In the f irst three decades of Yuan rule, water control was neglected by the court, and many dikes and floodgates fell into disrepair.24 The attitude of the central government changed in the middle of the Yuan, however. A special off ice (Xing dushui-jian 行督水监) was instituted to supervise water control works. Under the leadership of Ren Renfa 任仁发 and other hydrologists, several large-scale projects were carried out, continuing into the late 1330s.25 Local governments also established construction standards for dikes and supervised dike-building projects. In the early Ming, larger-scale water control projects were planned and supervised by the government under the leadership of famous statesmen such as Xia Yuanji 夏元吉 (1366-1430).

The shift in water control strategies from dike building in the Southern Song to river dredging in the Yuan and early Ming reflected a signif icant change from quantitative expansion to qualitative improvement in land utilization in Jiangnan. The enclosure of swamps and lakes greatly increased the amount of cultivated land, but many of these new f ields were prone to water logging that lowered fertility (Adachi 1985). Their damp soils were unsuitable for dry-land crops such as wheat, beans, and rape. For this reason, double-cropping of rice and winter dry-land crops was not common in the Jiangnan plain in the Southern Song. Even the productivity of rice paddies proved unstable in Southern Song Jiangnan because of frequent flooding and damp soils (Osawa 1996: 83). Hamashima Atsutoshi (1989) dates the shift from land clearance, which he calls“extensive development,” to land improvement, or “intensive development,” to the mid-f ifteenth century. Yet there is no doubt that the “drying of the f ields”started much earlier-at least by the Yuan-on a comparatively smaller scale, and that it expanded over time.

Drying the f ields involved two kinds of activity. On a macro-level, the f irst task was to dredge rivers and increase their capacity not only to discharge floodwaters in rainy years but also to drain excess water from low f ields in ordinary times. On a micro-level, however, “dividing polders” (fenwei 分围,分圩) was most crucial. The polders built in the Song were quite large and diff icult to drain using the available technology of pallet-chain pumps driven by men or water buffalo (Watanabe and Sakurai 1984: 198) Given the limits of pump eff iciency, “dividing polders” became a popular method of coping with the drainage problem. Although the term fenwei f irst appeared in Ming records, the practice itself can be traced to the Northern Song,26 In the Yuan, this method, known as “enclosure” (guitian 柜田), was already common. Wang Zhen’s 王桢 Treatise on Agronomy 农书 (1313) contrasts these small enclosures with the much larger polders (weitian). 27

The effort to dry the f ields also benef ited from advances in hydrological technology achieved in the Yuan and early Ming. Some of the tools, equipment, and machines listed in Ren Renfa’s Questions and Answers on Water Control (Shuili yida 水利议答) cannot be found in earlier literature and most likely are Yuan innovations.28 Among them, the square-pallet pump driven by a windmill is especially noteworthy. This is the earliest record in Chinese history of the use of wind-driven pumps in water control. Since wind power was much cheaper and usually stronger than human or animal power, it was far more eff icient.

Drying the f ields was a long-term process. In contrast to Hamashima’s opinion that the process began in the mid-Ming and ended at the close of the Ming, Shiba suggests that it took place over a much longer period, from the late Southern Song to the late Ming. All of Huating, for example, was swampland in the Southern Song. “What you see for over one hundred leagues is swamp, marsh, and lakes. When the busy season comes in spring, norias and other machines for removing water are placed everywhere. Only after all efforts are made to drain the f ields can rice then be planted.”29 In the Yuan, Shiba argues, drying the f ields spread to Suzhou and Jiaxing prefectures, the core area of the Yangzi Delta. This development resulted in a tripling of government-owned lands (guantian 官田) between the end of the Southern Song and the early Ming, by which time the swamps had been completely transformed into fertile arable (Shiba 1988: 43, 185). According to my own research, the process of land improvement continued well into the nineteenth century (Li Bozhong 1997;1998: 26-27). Both the early Yuan and the early Ming saw a great deal of land improvement (Kitada 1988: chap.2).

The process of drying the f ields during the Yuan and early Ming greatly improved the physical, chemical, and structural character of the soil, enhanced the productivity of the land, and generated higher and more stable yields. “Wet paddy is the most fertile land in the world, and the best paddy is found in the Yangzi Delta,” wrote Zhou Wenying.30 Another result of land improvement was the spread of double-cropping of rice and winter crops (wheat, beans, rape). The soil also benef ited from the rotation of wet and dry-land crops (Li Bozhong 1994 a). The conclusion that the productivity of Jiangnan’s cultivated land improved most dramatically in the Yuan and early Ming is aff irmed by Kitada Hideo’s research on long-term changes in Jiangnan’s ecological environment from the Song to the Qing (Kitada 1988: chap.3).

Many scholars, including Fu Zongwen, Liang Gengyao, Fan Jinmin, Xia Weizhong, and Cong Hanxiang, have argued that Jiangnan’s farmland no longer could support a growing population after the Southern Song. But the reverse seems to be true. Since the population and the amount of cultivated land increased only moderately between 1200 and 1400, whereas the quality of cultivated land improved markedly, the labor/land ratio did not worsen but improved.

CHANGES IN AGRICULTURAL TECHNOLOGY

Scholars have often based their evaluation of the level of farm technology for a particular period on the appearance of a few important techniques.31 Rarely have they considered how widely (or well) these new techniques were employed. Nor do they pay much attention to the ancillary improvements needed to apply these techniques eff iciently. For example, double-cropping of rice and wheat could become widespread only when improvements had been made in farm tools, rice varieties, and fertilizer use, in addition to resolving the problem of soil waterlogging. Overlooking spatial differences, some scholars lump together advances that appeared in different areas of China and piece them into a general system that they present as the commonly used technology of the day.32 An understanding of technological change must incorporate an awareness of how given techniques f it into the technological regimes of particular places at particular times. Among the technological advances in Jiangnan during the thirteenth and fourteenth centuries that merit consideration in terms of the longterm development of the region’s agricultural system are (1) varieties of rice; (2) tools; (3) fertilizer; and (4) cropping patterns.

Varieties of Rice

We know little about the quantitative changes in Jiangnan crop varieties during these two centuries. What we can say, based on extant materials, is that the number of varieties of rice increased, but probably only slightly. Excluding repetitions, 51 varieties of rice are recorded in the six extant gazetteers of Southern Song Jiangnan; of these about three-quarters were new strains, appearing for the f irst time in the historical record.33 The sole Yuan gazetteer of Jiangnan mentions 22 varieties of rice.34 In the specif ic case of Suzhou, 42 varieties were recorded in three Southern Song gazetteers, and a gazetteer of 1506 enumerates 29 varieties. Of course, the information in gazetteers is incomplete.35 In general, however, a temporal comparison of these works reveals a trend toward an increase in the number of rice cultivars during the Song-Yuan-Ming transition (You 1995: 86-91; Liang Jiamian 1989: 403).

Although the diversity of varieties of rice grew only slightly, signif icant changes occurred in the relative importance of the three main subspecies, xiandao 秈稻 (indica), jingdao 粳稻, and nuodao 糯稻 (glutinous rice), and in the mixture of early, intermediate, and late rice. Jiangnan farmers planted all three main subspecies. Nuodao was mainly used in brewing wine, and xiandao and jingdao were staple foods. Because of climatic changes since the Northern Song, jingdao has become the most important subspecies in Jiangnan. It is more resistant to cold and thus better suited to the lower temperatures that prevailed in the thirteenth and fourteenth centuries (You 1995: 42-44). Both jingdao and xiandao subspecies can be further subdivided into early, intermediate, and late strains on the basis of their seasonal patterns of growth. Through a long process of natural and artif icial selection that began in the Song, most jingdao varieties cultivated in Jiangnan were intermediate-and late-ripening types, whereas most xiandao varieties were early-ripening types (see Amano 1979: 104-05; You 1986; Li Bozhong 1994a).

In addition to their adaptability to adverse climatic conditions, the intermediate- and late-ripening jingdao types have other advantages over xiandao, not the least of which is their superior quality.36 Further improvements in intermediate and late strains of jingdao would have required a sophisticated understanding of the particular characteristics of the various strains of rice. Southern Song gazetteers did not differentiate between the two subspecies, jingdao and xiandao, but in later dynasties, knowledge vastly improved. Sixteen varieties of jingdao and six varieties of xiandao were recorded in the 1332 gazetteer of Zhenjiang. Huang Shengceng’s early sixteenth-century work on rice varieties of the Lake Tai basin lists 38 kinds—21 varieties of jingdao, 4 of xiandao, and 13 of nuodao.37 The increase in jingdao varieties is clearly indicative of advances in breeding techniques.

Suzhou farmers knew the difference between early and late rice as early as the Northern Song.38 In the Southern Song, Hangzhou farmers distinguished between early, intermediate, and late jingdao, but the differences between them were not specif ied.39 In neighboring Ningbo 宁波, it was said that in normal years early rice ripened around the f irst day of the seventh lunar month, intermediate rice around the f ifteenth day of the seventh month, and late rice in the eighth month.40 In the fourteenth century, Jiangnan farmers employed a more protracted schedule of planting and harvesting. A late Yuan encyclopedia states:

Early rice is transplanted after the Festival of Vernal Equinox [the fif-teenth day of the second lunar month] and harvested after the Festival of Great Heat [the fifteenth day of the sixth month]; intermediate rice is transplanted after the Festival of Grain in Ear [the first day of the fifth month] up to the Festival of Summer Begins [the fifteenth day of the fifth lunar month]; late rice is transplanted within ten days after the Festival of Summer Begins [the fifteenth day of the fifth month] and harvested after the Festival of Cold Dew [the first day of the ninth lunar month].41

In Southern Song Jiangnan, early rice was still dominant. In the Ming, by contrast, early rice varieties were far less numerous than intermediate and late rice varieties (You 1986). The improved understanding of rice varieties in this period was based on advancements in seed selection and breeding. The increase in intermediate and late jingdao varieties came to play an important role in the agricultural development of Jiangnan, not only in increasing yields but also in changing cropping regimes (Li Bozhong 1994a).

The introduction and spread of Champa rice from the Mekong River delta had been regarded as one of the key factors in the Song agricultural revolution and the main cause behind the secular changes in land utilization and grain production in the past millennium (Ho 1959: chap.8). Champa rice is commonly believed to be far superior to native varieties because of its tolerance of drought and its short growing period, merits native rice lack. Scholars have argued that Champa rice enabled the expansion of rice cultivation to upland areas notoriously short of water and made possible the breeding of new early-ripening rice varieties. The development of early-ripening varieties in turn secured the success of the double-cropping regime (Katō 1952-53a; Ho 1959).

This conventional narrative has been challenged, however. You Xiuling (1983) points out that the historical records on Champa rice are confusing, ambiguous, and even contradictory. Also, most scholars have not systematically analyzed these accounts from a modern agro-biological perspective. In addition, the two major advantages of Champa—its tolerance of drought and poor soil—were not signif icant in Jiangnan, where most f ields were fertile and had a suff icient water supply. A short growth period, the third advantage of Champa thought to have been crucial to the introduction of double-cropping of rice and winter crops, is not true of all Champa varieties. Moreover, a short growth period in itself did not lead to the double-cropping of rice and winter dry-land crops in Jiangnan.42 Double-cropping of rice and winter dry-land crops became practical only after new strains of intermediate and late rice were developed (Li Bozhong 1994a). Thus, it is a mistake to exaggerate the role of Champa rice. Champa rice did not “revolutionize” Jiangnan agriculture. Instead, productivity increases resulted from a long-term, gradual, and probably accelerating improvement of native varieties.43

Farm Tools

Among the new farm tools that appeared in Jiangnan during the Song-Yuan-Ming transition were the tieda 铁搭 (a pronged drag hoe that looks like an iron-toothed rake), the yundang 耘荡 (for weeding the paddy), the tiandang田荡 (a smooth-edged rake for leveling the soil before transplanting), and the pingban 平板 (a plate to smooth the soil after harrowing). The use of these tools, f irst recorded in Wang Zhen’s treatise, accompanied the important innovations in agriculture: the double-cropping of rice with dry-land crops and the transplantation of rice.

The most important tool was the tieda. Though a simple device, the tieda enabled sticky and wet paddy soil to be plowed (Osawa 1993). As You Xiuling (1995: 145) explains, hoes and even ox-driven plows fail to plow rice paddies deeply, whereas the hand-held tieda is far more effective. Deep plowing is crucial to rice cultivation. It loosens and expands the cultivated layer of soil, which not only creates more room for root growth but also allows the soil to retain more fertilizer, water, and air (Chen Hengli 1983: 26). The tieda also played an important role in the cultivation of winter dry-land crops. It could be used for digging ditches to drain excess water and for making ridges between paddies on which wheat and other dry-land crops could be sown. The yundang was much more eff icient than the hoe or hands for weeding paddies and considerably lessened the amount of labor needed.44 By the mid-Ming it had become a basic tool of every Jiangnan farmer. The tiandang, used to level and smooth the soil after the paddy was plowed, and the yangtan 秧弹, a rope made of bamboo strips used to space the rows of seedlings evenly when transplanting rice, appear to have been Yuan inventions (Wang Yuhu 198: 249-50). The appearance of the yangtan is also the earliest concrete evidence of rice transplanting, which benef its productivity.45

Fertilizer

In the Song-Yuan-Ming transition period, fertilizing techniques improved in two respects. There was an increase in variety and an improvement in processing methods.

It is widely believed that great advances were made in fertilizer use in Song Jiangnan. Chen Fu, for example, mentions various kinds of fertilizers, but most were used only in rice seedling beds, vegetable gardens, or mulberry orchards, not in open paddy f ields. The only fertilizers used in paddy f ields were plant ashes and rice stubble, both of which were plowed under to serve as a base fertilizer. Grass weds in f ields were used as a topdressing.46 Lou Shu’s 楼 Illustrated Lyrics on Tilling and Weaving (Gengzhi tu shi 耕织图诗; ca.130-62) discussed the use of fertilizer in seedling beds but did not mention the application of fertilizer in open f ields at all.47 Chen Fu advocated the use of huofen 火粪 (“burned” fertilizer, mainly plant ashes) in open f ields, but made clear that huofen was mostly used as a base fertilizer for winter dry-land crops of beans and wheat. This method, however, was not in wide use, since doublecropping was not common at this time. Thus, contrary to conventional views, the application of fertilizer to open f ields was quite limited in Southern Song Jiangnan.

Methods of fertilizer processing were similarly limited. The two major processing methods were burning and composting. In burning, fallen leaves, garbage, and cooking stove ashes were collected in a special “fertilizer shed” to be burned. The remains were mixed with nightsoil to form huofen. The eff icacy of this method is debatable. Plant ashes are rich in potassium, and nightsoil is rich in nitrogen. But if both are mixed, the nitrogen is released into the air before it can be absorbed by the plants. In composting, hogwash, kitchen wastewater, rice husks, and fallen leaves were placed into a tank and left to rot. Compost would then be made from the decomposed matter (Osawa 1993: 243). From a modern agronomist perspective, this is a reasonable method, but the amount of fertilizer it yielded must have been limited. Indeed, Chen Fu states that the compost was used only in mulberry orchards.

Yuan sources attest to the growing use of fertilizer in rice paddies. Wang Zhen categorized the different fertilizers commonly used in open f ields in Jiangnan into four basic types: green manure (caofen 草粪), plant ash (huofen), human and animal excrement (fenfei 粪肥), and river mud (nifen 泥粪). He also described the processing and application of these fertilizers in some detail (Wang Yuhu 1981: 37). Despite these advances, however, the application of fertilizer in open f ields was still not common enough to merit mention in Zhao Mengfu’s 赵孟 (1254-1322) Illustrated Lyrics on Tilling and Weaving.48

Progress in developing fertilizing techniques continued in the early Ming. First, applying fertilizer to open f ields had become so commonplace by the mid-Ming that the practice was listed in a popular agricultural handbook, Kuang Fan’s 邝璠 (1458-1523) Illustrated Almanac of Aid to the People (Zhuzhici xiaoyong 竹枝词小咏), as one of the basic agricultural tasks. Second, this period saw the early use of a new and important kind of fertilizer, bean cakes. Earlier agronomic works mention only sesame cakes. Since the production of sesame was never great in Jiangnan, sesame cakes were generally used to make huofen, which was then applied to rice seedling beds. In the Ming, however, new kinds of oil cake fertilizers-bean cakes, cottonseed cakes, and rapeseed cakes-gradually became common. The Illustrated Almanac of Aid to the People shows that by the mid-Ming the three basic kinds of fertilizer applied to open f ields were bean cakes, human/animal excrement, and river mud.49 The widespread use of bean cakes marked one of the most signif icant innovations in premodern Chinese agricultural history (Li Bozhong 1998: chap.8). In this sense, the early Ming can be seen as the start of a new era in Jiangnan rice cultivation.

Cropping Patterns

The most signif icant change in cropping patterns was the increasing use of double-cropping of rice and winter dry-land crops (wheat, barley, oats, beans, and rape). From the perspective of modern agronomy, this cropping pattern has many advantages over alternative cropping systems and best f its Jiangnan’s ecological conditions.50 In other cropping regimes, the soil stays wet throughout the year, which reduces its fertility.51 A regular rotation of wet rice and dryland crops in the same plot ensures periodic drying of the soil, a practice that maintains and improves its productivity (Fang et al.1984: 40).

Double-cropping of rice and winter dry-land crops f irst appeared in Jiangnan in the Tang (Li Bozhong 1982). It is commonly believed to have been a widespread, if not the dominant, cropping regime in Song Jiangnan (Sudō 1961; Liang Gengyao 1984: 153-54). But this assumption has been increasingly challenged by recent research. Kitada (1988: chaps.1, 3), for example, distinguishes between the double-cropping patterns prevalent in Song-Yuan times and the “new double-cropping system” that appeared in the Ming-Qing period. He concludes that the most important differences were (1) geographically, the old double-cropping system was practiced mainly in upland areas of western Jiangnan, whereas the new system was found in the eastern plains; and (2) in the old system, early rice was the main crop and wheat was the secondary crop, but in the new system late rice was the main crop, and rotated with wheat, beans, and rapeseed. The f irst point has been reiterated by Adachi Keiji (1985) and Osawa Masaaki (1993: 245-48), who note that although double-cropping gradually spread from the late Tang on, the dominant system in the Song was still a single annual (or even bi-or triennial) crop of wet rice. It was not until the thirteenth and fourteenth centuries that double-cropping became widespread, expanding from upland areas to the delta plains.

Kitada’s second point is even more signif icant. Beans and rape have certain advantages over wheat as secondary crops in the double-cropping system (Fang et al.1984: 306-8). As for the main crop of rice, the late-ripening jingdao rice of the new Ming-Qing system was far superior to the early-ripening xiandao rice of the Song both in quality and in yield. Early rice cannot be double-cropped with wheat because their growing seasons overlap in summer. Apparently in Song texts, “early rice” referred not only to early rice but also to subvarieties of intermediate rice that could be double-cropped with wheat.52

Other conditions in Southern Song Jiangnan were not favorable to the double-cropping of rice and winter dry-land crops, such as the lack of good irrigation and drainage systems, insuff icient supplies of fertilizer and labor, and ineffective farming techniques for tilling both wet-and dry-land crops. All of these are key requirements for the double-cropping regime.53 These unfavorable conditions make it clear why the people of the Southern Song held views on wheat cultivation different from those of their Ming-Qing descendants, for whom there was no question that wheat cultivation required less labor than rice. For example, in 1225, the magistrate of Yuhang 余杭 county in Hangzhou prefecture tried to persuade the locals to plant dry-land crops, only to discover that the peasants were reluctant to plant wheat because they thought wheat cultivation required too much labor and resulted in too little income.54 The damp f ields of the Jiangnan plain during the Southern Song indeed made it diff icult to plant wheat. Similarly, in Jiaxing prefecture, even during the early Yuan, little wheat was planted, and peasants had to obtain flour from shops in exchange for rice.55

In addition to drying the f ields, the key innovation that enabled the planting of wheat in paddy f ields was the qilun 起 (ridge-sowing) technique. We learn from Huang Zhen 黄震 (1213-80) that sowing wheat on ridges was not practiced in the late Southern Song.56 In the Yuan, however, Wang Zhen observed:

Rice ripens earlier in upland fields. Such fields can be dried out and plowed to plant wheat and barley. The planting method involves making ridges of soil with a furrow (quan 畎) between the ridges. After a section of field is plowed, a ditch called a ‘waist ditch’ (yaogou 腰沟) is dug across the plot, and water is discharged into the field. Once the wheat and barley is harvested, all the ridges and furrows are leveled. The field is flooded and then plowed to plant rice. This field is a double-cropping field. (Wang Yuhu 1981: 23)

The spread of the technique of sowing in ridges was closely tied to the use of the tieda. Only with the aid of the tieda could farmers dig deep ditches def iciently and make ridges in the sticky and wet paddies (You 1995: 148).

These developments engendered an increase in the double-cropping of rice and wheat in the Yuan and early Ming (Kitada 1988: 12-19). But the practice was still limited to low-lying areas in the plains. A typical example is eastern Huzhou, where in the late Southern Song the f ields were too low and “wheat and beans cannot be planted.”57 Such was still the case in the Yuan, as we can see in Zhao Mengfu’s Illustrated Lyrics on Tilling and Weaving, which describes rice and wheat being planted in different plots, rather than in the same plot.58 In general, as Wang Zhen notes, the practice of double-cropping was still largely conf ined to upland areas. In the early Ming, we see a moderate expansion of double-cropping of rice and winter dry-land crops into the plains, but wheat still occupied only a minor place in Jiangnan agriculture. The wheat quota in Suzhou prefecture’s grain tax, for example, was only 60,000 shi, less than 3 percent of its rice quota (2.15 million shi). Moreover, even this small amount of in-kind payment in wheat typically was commuted to cotton cloth. Kuang zhong 况钟, prefect of Suzhou in the Xuande reign (1426-35), stated that “wheat is seldom planted in the f ields of the various counties of this prefecture. Thus, when land taxes were assessed at the beginning of the dynasty, the Board of Revenue allowed the wheat tax to be paid in cotton cloth at a conversion rate of 1.2 shi wheat to 1 bolt of cloth. In this way the inhabitants can pay their wheat tax.”59

In sum, it took centuries for the late-rice/winter-wheat cropping pattern to prevail in Jiangnan. Although it appeared as early as the Tang, it did not predominate until the seventeenth century and completely replaced earlier cropping patterns only in the mid-nineteenth century (Li Bozhong 1998: 50-51). Therefore, the period down to the mid-Ming constituted the formative stage of development for this cropping pattern, which expanded steadily but moderately.

The dissemination of new techniques, or the adaptation of an existing tech-nique to a new environment, is probably more signif icant than the appearance of a new tool or technique. The distinction between invention and transmission of technology becomes blurred if we use a def inition of technology broad enough to include minor adaptations of known technologies. This approach to technological change allows us to appreciate not only the “technological breakthroughs”but also the real economic impact these breakthroughs had on the agrarian economy over the long term.

CHANGES IN FARM MANAGEMENT

Along with changes in the agricultural labor force, cultivation, and farm technology, a signif icant change in farm management took place during the two centuries from late Song to early Ming. These changes were reflected in the tenancy system, patterns of landowning, and the introduction of new industrial crops. Here I focus on two aspects—farm size and grain yield—that have often been neglected or misunderstood in previous studies. In general, farm management in Jiangnan can be characterized as “extensive” before the Tang and “intensive” from the Song on. The major indications of this intensif ication were (1) decreasing farm size, (2) increasing labor and capital inputs per mu of cultivated land, and (3) an increase in output per mu. This pattern of intensif ication continued into the Qing.

Farm Size

From at least the Tang on, the basic unit of production in Jiangnan agriculture, both for large estates and for small holdings, was the family farm, run by a nuclear family. Although there certainly were great differences among households in the amount of land owned, the variation in the amount of farmland worked by the individual cultivator—be he a tenant farmer, an independent proprietor, or even a bondservant—was not great. Thus, in the following discussion, the term “farm size” refers to the amount of farmland cultivated, not owned, by a peasant family. Changes in farm size did not necessarily result from shifts in the labor/land ratio.

Changes in ecological environment, family composition, cropping patterns, and the farmer’s labor capacity all affected farm size. As we shall see, changes in family size and the organization of production were the crucial factors in determining farm size.

Liang Gengyao has collected the most comprehensive data on the average amount of farmland per household at the county level for Southern Song Jiangnan (Table 4.6). From these f igures we can see that the largest farms were found in the low-lying plains of eastern Jiangnan (Huating and Changshu). Farm size varied widely in Jiangning prefecture in western Jiangnan, with small farms prevailing in the more hilly counties of Lishui and Liyang and much larger farms in lowland Shangyuan and Jurong. Farm sizes would be greater than farmland per household, but probably not by much, since the level of urbanization most likely did not exceed to percent in the Southern Song.

Estimates of farm size in Song Jiangnan vary widely. Fu Zongwen (1988: 306) suggests that a peasant could farm 20-30 mu of paddy in Jiangnan, and Yanagida (1957) and Kusano (1969) have argued that the minimum amount of paddy worked by a peasant family was 30 mu. At one point Qi Xia (1987: 74, 218-20, 331, 1204) argues that in most cases the acreage of cultivated land farmed by a peasant family in Song Jiangnan was in the range of 19.5-25 mu. But in some of his examples, farms were much larger, as much as 60-80 mu. Liang Gengyao writes that “in the Southern Song, a peasant could manage 30 mu of land, but in fact the acreage that many peasants farmed was much smaller.”Liang concludes that “in populous regions like Jiangsu, Zhejiang, Fujian, and Sichuan, ... the amount of cultivated land worked by a family was as small as several or ten-plus mu” (1984: 129, 155). But this estimate is contradicted by the f igures in Table 4.6, which mostly range from 28 to 48 mu, with an average of around 30 mu. Generally speaking, these estimates suggest that in late Southern Song Jiangnan the average amount of land cultivated by a peasant family ranged from 30 to 50 mu, or around a mean of 40 mu. The f igure of 40 mu seems closer to the historical reality.

Few data on farm size are available for Yuan Jiangnan. However, the relatively complete and reliable data for cultivated land and households for Zhenjiang prefecture yield an average of 34 mu per household.60 From other extant sources, it seems the amount of farmland worked by a peasant family in the Jiangnan plain was even smaller. For example, Fang Hui 方回, writing in the early Yuan, declared that a peasant could farm 30 mu in low-lying areas of Jiaxing.61 In neighboring Songjiang, according to Yuan Jie 袁介, a tenant peasant family working on government lands farmed 30 mu in the early fourteenth century.62 Compared to the average amount of 48 mu per household in the mid-Southern Song in the same county (Huating), the Yuan average is lower by more than one-third.

Based on the early Ming prefectural f igures for population and cultivated land in Tables 4.1 and 4.4 and the assumption that each household averaged f ive persons in the early Ming (see below), I have arrived at the estimates for farm size in early Ming Jiangnan shown in Table 4.7. Other data suggest that the amount of land farmed by a peasant family was even lower. According to off icial data from 1393, farmland per household was 21 mu in Suzhou. Farmland per farm family surely was greater in this highly urbanized prefecture, but on government lands, which accounted for 70 percent of total farmland in Suzhou, the average amount of land worked by tenant families was less than 20 mu. An allotment of government land to 8,986 households in Taicang in the late fourteenth century averaged 16 mu per adult male.63 When the government allocated government lands to soldiers in Changshu in 1398, the standard was 12-15 mu per person.64 The upper limit set by the state for government lands leased to a tenant family was 30 mu.65 A rare extant land registration document from the early Ming shows that the Sun family in Songjiang county farmed 19.8 mu, a f igure comparable to the 1393 Suzhou standard of government land allotments but only two-thirds of the amount worked by a family in Yuan Songjiang.66

The limited data at our disposal suggest that there was a decline in farm size between 1200 and 1400, from about 40 mu in the late Southern Song to around 30 mu in the Yuan and then to approximately 20 mu in the early Ming. This decline in farm size was most evident in the plains of eastern Jiangnan. In the Southern Song, farmland per household was much greater in the plains than in the uplands of western Jiangnan. But the situation was reversed by the early Ming. As Table 4.7 indicates, farmland per household was much smaller in the three eastern prefectures of Suzhou, Songjiang, and Jiaxing than in the four western prefectures of Zhenjiang, Jiangning, Hangzhou, and Huzhou.67 Farmland per household fell slightly from 37 to 31 mu in Jiangning prefecture from the late Southern Song to the early Ming, but it rose from 34 to 37 mu in Zhenjiang prefecture between 1332 and 1393. Farm size seems to have changed less dramatically in upland areas than it did on the plains. It shrank most dramatically in eastern Jiangnan, where population growth was concentrated.

Family Size

Since farms in Jiangnan were usually worked by a nuclear family in the middle and late imperial periods, farm size was determined to a large extent by the size of the family. A typical peasant family consisted of a married couple, their children, and in many cases the husband’s aged parents. Generally speaking, the average peasant family in late Southern Song Jiangnan had more than f ive, most likely six, members, of whom three were able-bodied workers.68

Family size does not seem to have changed much in the Yuan. In Zhenjiang, “native households” (tuzhuhu 土著户) had 6.1 persons and “civilian households”(minhu 民户) 5.7 persons.69 Households in Changzhou prefecture averaged 5.3 persons in 1350 (Liang Fangzhong 1980: 442). In general, it seems that the size of a Jiangnan peasant family in both the late Southern Song and the Yuan was more than f ive and very likely six persons.

Population data for the early Ming is much more precise. According to off icial f igures, households in Suzhou prefecture averaged 4.1 in 137,4.3 in 1367, and 4.8 persons in 1393. In Changzhou, households averaged 4.4 persons in 1377 and 5.1 persons in 1393. In 1393, the household average was 4.9 persons in Songjiang prefecture but only 3.8 in Hangzhou. In contrast, households in the west and north were much larger, averaging 7.3 persons in Jiangning and 6.0 in Zhenjiang (Liang Fangzhong 1980: 203-4, 436, 442, 446). Thus, in eastern Jiangnan a typical family had 4-5 members, and families in western Jiangnan were somewhat larger, with 6-7 persons. In general, we can say that the “f iveperson-family” predominated in Ming Jiangnan. In most cases, such a family had two able-bodied adults, the husband and wife (Li Bozhong 1996).

The decline in peasant family size had a signif icant impact on farm size. A smaller labor force reduced the amount of land a family could cultivate. In other words, the decline in farm size during the Song-Yuan-Ming transition resulted from the decline in family size as well as increasing population pressure on the land. In the three eastern prefectures of Suzhou, Songjiang, and Jiaxing, the number of households rose by 50 percent between the late Southern Song and early Ming, while the amount of cultivated land increased only slightly. Consequently, farmland per household would have shrunk by about one-third, from 40 to 27 mu. As shown above, however, farm size in fact shrank by roughly half during this period—from 40 to 27 mu. But if family size decreased by onesixth (from 6 to 5 persons per family) as suggested here, farm size per capita would have decreased by 40 percent. The amount of farmland per adult worker declined even less dramatically, from 13 to 10 mu.

Labor Capacity

In addition to changes in the labor/land ratio and family size, other factors contributed to the decline of farm size in Jiangnan, especially in the eastern part. Among the most important were changes in the working capacity of a peasant family. Since the capacity of a peasant family varied under different farming systems, farm size was likely to be affected by changes in the cropping system. Under the single-cropping regime in late Ming Jiangnan, a peasant family with a two-person labor force—husband and wife—could work about 25 mu without a water buffalo. But under the regime of double-cropping of rice and winter dryland crops, only the husband worked in the f ields, and he could work only to mu without the aid of a buffalo.70 Thus, the working capacity of a peasant family placed constraints on farm size—about 25 mu under single-cropping of rice and about 10 mu under double-cropping of rice and winter crops (Li Bozhong 1986,1996). Although single-cropping remained predominant, double-cropping was widely practiced by the early Ming. Given the changes in the working capacity of the family due to the change in cropping regimes, it is no surprise that a decline in farm size accompanied this shift.

Grain Yield

Even when farm size shrinks, a farmer can increase production by raising the yield per mu of cultivated land. This issue assumed particular importance in eastern Jiangnan during the Song-Yuan-Ming transition, when farm size was declining sharply. Scholars vary in their estimates of grain yield in Southern Song Jiangnan, but all of them tend to agree that yields were high, about 2 shi of husked rice per mu (Liang Gengyao 1984: 145; Yu 1980; Wu Hui 1985: 160; Shiba 1988: 90-91, 137-41) or even more: Min Zongdian (1984) and Fu Zongwen (1988: 310) suggest 2.5-3 shi, and Qi Xia (1987: 2, 26, 138) places Song yields as high as 6-7 shi. These estimates, unfortunately, have not been well documented, with the exception of that by Shiba. As Shiba points out, the yield of 2 shi seems to have been the upper limit of ordinary yields. His analysis of tenancy contracts for government-owned lands in Changshu county dating from 1237 shows that of 114 yield f igures, 53 percent fall below 0.62 shi per mu, 27 percent in the range of 0.62-1.20 shi, and 20 percent in the range of 1.20-2.25 shi. The average is 0.65 shi (Shiba 1988: 91, 142-43).

Changshu was one of the major rice-producing areas in Jiangnan. As early as the late Tang, the county was known for its bumper harvests, earning it the name Changshu, “Constant Bounty.” In 1249, Cheng Gongxu 程公许 wrote, “The Liangzhe 两浙 circuit [i. e, the Jiangnan region] is located in the southeast. Here water is more plentiful than land, and by irrigating the f ields the inhabitants render them fertile. Among all the prefectures of Liangzhe, Suzhou is the richest in agricultural produce, and among all the counties of Suzhou, Changshu is the richest. This is evident from the county’s name.”71 The average yield in Changshu must have ranked among the highest in Jiangnan. Moreover, the evidence suggests that the yield on school lands was at least equal to mid-quality private lands, because most of the school lands ranked in the middle grade (Qi Xia 1987: 363). If the average yield of school lands in Changshu was only 0.65 shi, it seems inconceivable that the average yield of all f ields of Jiangnan was as high as 2 shi. Two sets of Changshu land rent data for yiyitian 义役田 (public lands set aside to defray the costs of hired service labor) dating from 1210 and 1238 testify to average yields per mu of 1.36 shi in Guizheng 归政 township (438 mu) and 1.0 shi for the county as a whole (51,310 mu).72 In Huating, rents on yiyitian (140,631 mu) were much lower, averaging only 0.45 shi.73 It seems most likely that ordinary yields in eastern Jiangnan were around 1.0 shi.

An average yield of 2 shi was surely too high for f ields in upland areas of western Jiangnan, where land was reputed to be far less productive.74 In Zhenjiang and Changzhou, for example, yields of 0.8-1.0 shi were common. The f igure might have been even lower-less than 0.5 shi—in parts of Zhenjiang and Jiankang.75 Huang Zhen reported that “it is common knowledge that in Huzhou and Jiaxing no more than 50-60 percent of cultivated land yield 1 shi of rent (per mu). Farther north, in Changzhou and Zhenjiang, the land lies higher and the f ields are poorer. Rents there are 3-4, or at most 5, dadou 大斗 per mu.”76 These rent f igures imply yields in the range of 0.8-1.3 shi.77 Taken together, these data suggest that overall yields in Jiangnan may have been lower than, but were most likely close to, 1 shi per mu.

How, then, can we explain the high yields of 2 or more shi mentioned in some Southern Song sources? First, one of the main sources of information on yields has been gazetteers, which may have recorded only very good or very bad harvests. Second, another source of claims of high yields are essays written by Song off icials to promote Jiangnan farming practices in other parts of South China (quannongwen 劝农文).78 These off icials often exaggerated the benef its to be reaped from adopting these methods. Third, most citations of high yields are concentrated in Suzhou, Jiaxing, and Huzhou; other prefectures are rarely mentioned (Shiba 1988: 138, 140-141). But the Changshu school lands and yiyitian data, which indicate yields in the range of 0.65-1.3 shi per mu, seem far more reliable than such round generalizations. References to high yields of 2 or more shi per mu undoubtedly represent exceptional harvests and cannot be taken as averages.

Few data on yields for Yuan Jiangnan have survived. Yu Yefei (1980) and Wu Hui (1985: 166) estimate both Southern Song and Yuan yields at 2 shi; Li Gan (1985: 11) offers estimates from 3-4 shi to as high as 7-8 shi. But the evidence they provide is weak.79 Rent records for two sets of school lands provide more reliable data.80 According to these accounts, the rents per mu for properties owned by the West Lake Academy 西湖书院 were 0.29 and 0.78 shi in Hangzhou, 0.8 shi in Renhe 仁和 county (Hangzhou), 0.65 shi in Wucheng乌程 (Huzhou) and Kunshan (Suzhou) counties, and 0.48 shi in Yixing county (Changzhou), or an average of 0.58 shi for 2,184 mu; these f igures imply an average yield of about 1.2 shi. Since the lands of the West Lake Academy were spread throughout four prefectures, these yields should be fairly representative of Jiangnan as a whole. They suggest that Yuan yields were roughly comparable to, or slightly greater than, those of the late Southern Song.

For data on average yields for early Ming Jiangnan, we must turn to land tax records for government lands. The rent assessed on government lands, which constituted as much as half of all landholdings in Jiangnan in the early Ming,81 was usually 50-80 percent of rents for private holdings.82 I take 70 percent as an average. In Suzhou and Songjiang prefectures, where rents were most onerous, the average rents per mu of government lands were 0.44 shi and 0.31 shi, respectively (Mori 1960). Accordingly, the average yields for government and private lands would be 1.3 shi in Suzhou and 0.9 shi in Songjiang. The maximum rent for government lands was only 0.5 shi in Jiaxing prefecture and 0.7 shi in Haiyan county (Jiaxing), and lands of ordinary quality generated rents of 0.4 shi in Wujin 武进 county and 0.3 shi in Yixing county (both in Changzhou prefecture), and only 0.2 shi in Shangyuan county (Jiangning) (Wu Dan’ge 1982: 18). Accordingly, the yields in these places must have been lower than in Suzhou and Songjiang. Although there are some reports of much higher rents in early Ming Jiangnan, they seem to have been exceptional. In general, the average yield probably was about 1 shi, or more likely a little higher. For Changshu county in mid-Ming times, “the harvest from a mu of paddy generally was 2 shi for uppergrade f ields, 1.5 shi for mid-grade f ields, and 1 shi for lower-grade farms.”83 These estimates seem valid for Jiangnan generally.

Consequently, yield per mu was a little less than 1 shi in the late Southern Song and a little more than 1 shi in the Yuan and early Ming. Yields given above are expressed in nominal terms, however. The shi measure of capacity increased from 66 liters in the Song to 95 liters in the Yuan and 103 liters in the Ming, but the mu measure changed only slightly (Li Bozhong 1990: 12-13). If we take these changes into account, real yields in the Ming probably were 60 percent higher than in the Song. In Changshu, the margin of increase was even greater.84

By combining the data presented above on average farm size, the number of laborers per family, and yields, we can arrive at a rough estimate of output per family and per worker (Table 4.8). Although farm size shrank from the Song to the Ming, labor productivity rose. If winter dry-land crops were included, the margin of increase in productivity would be even greater.

Small farm size has been regarded as a cardinal feature of Jiangnan agriculture from the beginning of the Song to the mid-twentieth century. The small size of farms commonly has been deemed “backward” because it was believed to reduce labor productivity and to have caused a secular decline or stagnation in farm output. The assumption that small farms lowered labor productivity has never been adequately documented, however. The evidence adduced here contradicts this assumption: although farm size decreased from the Song to the Ming, labor productivity rose. Moreover, the decreasing size of farms was a positive response to changes in production factors and not the reverse, as is thought by those who argue that population pressure led to steadily diminishing returns to labor from the Song onward.

Spatial Features of Agrarian Change

The changes discussed above were much more pervasive in eastern Jiangnan than in the uplands to the west. From the late Southern Song to early Ming, most of Jiangnan’s population growth was concentrated in Suzhou, Songjiang, and Jiaxing prefectures, whereas the population in the western prefectures declined. As a result of this demographic shift, the eastern plains became the most populous part of the region. The improvements in land quality, too, took place mostly in the plains, which benef ited from the building of water control works and the process of “drying the soil” (Shiba 1988: 204-12). Most advances in farm technology were similarly related to the exploitation of paddy f ields in eastern Jiangnan. Finally, the reduction in farm size and the expansion of yield were most evident in eastern Jiangnan as well. Therefore, the transformation of Jiangnan agriculture during this period occurred principally in the Jiangnan plain.

This spatial feature of agrarian change was consistent with the historical process of settlement in Jiangnan. Conventional wisdom suggests that agricultural exploitation of virgin frontiers usually begins in the plains, and then expands to the hills and f inally the mountains as the landscape f ills up. Thus, it is commonly thought that the settlement of Jiangnan began in the plains, then expanded to the Ningzhen Hils, and f inally reached the Zhexi Mountains. Shiba Yoshinobu has corrected this misunderstanding and provided us with a better understanding of Jiangnan settlement patterns. He suggests that settlement initially was concentrated in the small alluvial valleys of the Ningzhen Hills and Zhexi Mountains, then proceeded to the plateau on the northern periphery of the Yangzi Delta, and f inally shifted to the plains east of Lake Tai (Shiba 1988: 167-225). Adachi (1985) and Osawa (1996: 239-40, 245-48) concur with Shiba’s view that in the Southern Song land use was less intensive in the lowland plains east of Lake Tai than in the upland areas west of the lake. Some areas east of Lake Tai (such as the outskirts of Suzhou city) already ranked among the most developed parts of Jiangnan in the Tang, if not earlier. But these cases were exceptions to the more general picture of the exploitation of Jiangnan described by Shiba, Adachi, and Osawa. The transformation of Jiangnan agriculture during the Song-Yuan-Ming transition largely resulted from a spatial shift in agricultural exploitation from the upland valley to the lowland plains.

Temporal Features of Change

Some of the agricultural changes of the thirteenth and fourteenth centuries clearly manifest cyclical patterns. In other changes, the temporal trend is less clear. Moreover, not all patterns of change were correlated with one another. Demographic change clearly moved in a cyclical pattern. The decline in population in the Song-Yuan transition was followed by a rise in the Yuan, a second decline in the Yuan-Ming transition, and a second rise in the early Ming. The development of market towns paralleled these cycles of demographic change. According to Fu Zongwen (1988: 355), the number of rural market towns increased markedly in Jiangnan between the Song and the Ming, but this trend was interrupted twice, f irst during the Song-Yuan transition and again in the Yuan-Ming transition. This cyclical pattern prevailed throughout the late imperial period. Market towns in Jiangnan stagnated in the early Ming and early Qing, but flourished remarkably in the late Ming and mid-Qing (von Glahn in this volume; Liu I978). Since the proliferation of market towns was a main feature of the Jiangnan economy in late imperial times, the cyclical tempo of market town development can be seen as reflecting changes in the agrarian economy.

This general pattern of cyclical change can also be seen in long-run economic changes at a macro-level. Shiba (1988: 75-94) divides the process of economic change in Jiangnan from the early Northern Song to the early Ming into seven periods that correspond more generally to a cyclical pattern of accelerated growth from 950 to 1200, a decline from 1200 through the fall of the Yuan, and a renewal of growth in the early Ming. Shiba’s assessment of a pronounced decline in the thirteenth and fourteenth centuries appears to be unfounded. Moreover, there is little evidence of prosperity in Jiangnan in the early Ming. Instead, it is more likely that both the Yuan and the early Ming were marked by slow growth with a sharp, short-term decline during the dynastic transition.

It is important to emphasize the cyclical character of agrarian change in Song-Yuan-Ming Jiangnan. It is true that population and agricultural production stagnated or even fell during the dynastic transitions. However, such relatively short-term intervals of stagnation or decline should not be generalized as characteristic of the whole period 1200-1400.

The Historical Signif icance of the Song-Yuan-Ming Transition

The historical signif icance of agrarian change in Jiangnan during the Song-Yuan-Ming transition must be viewed in the context of the ten centuries from the late Tang to the mid-Qing. The concept of a fourteenth-century turning point rests on two implicit assumptions: an economic revolution in the Song, and protracted stagnation in the Ming and Qing. It should come as no surprise that the thirteenth and fourteenth centuries are then regarded as a turning point in Chinese economic history, given the conventional view that in these two centuries the Chinese economy regressed from the “revolutionary” Song to the “stagnant”Ming-Qing. But recent research demonstrates that neither the Song revolution nor the Ming-Qing stagnation paradigm is consonant with the reality of agrarian changes in Jiangnan. Many scholars believe that the Song represented the apogee of economic development in premodern Jiangnan history. Much evidence for this view can be found in the rapid pace of commercialization and urbanization. But agricultural change proceeded much more gradually. Moreover, the agricultural achievements of the Song era have been exaggerated. Osawa has enumerated four fallacies in the “Song agricultural revolution” hypothesis: (1) the expansion of poldering greatly increased the acreage of irrigated f ields; (2) Champa rice was introduced and cultivated widely, leading to the development of double-cropping of rice and wheat; (3) population growth and rising grain tax quotas resulted from increased production; and (4) agricultural technology advanced rapidly, as evidenced by Chen Fu’s Treatise on Agriculture. In a thorough analysis, Osawa refutes each of these claims and concludes that the socalled Song revolution in agriculture is nothing but an “illusion” (Osawa 1996: 236-49).

If we look more closely at technological changes in Jiangnan agriculture during the millennium preceding the mid-Qing, we can see that although most of the technological breakthroughs f irst appeared as early as the Tang, Jiangnan paddy farming did not attain its maximum productivity until the mid-Qing. The Tang witnessed three great breakthroughs: an advanced kind of paddy (the Jiangdong plow), rice transplanting, and double-cropping of rice and wheat (Li Bozhong 1990: chap.3). But eff icient utilization of these techniques still depended on the invention of new tools for paddy cultivation (such as the tieda and yundang), breeding new and more productive rice varieties (especially late-ripening intermediate and early-ripening late rice), and the introduction of more effective fertilizer (in particular bean cakes). Most of these improvements were not realized until the mid-Qing. The main evidence for a Song revolution in agricultural technology is drawn from treatises like Chen Fu’s Treatise on Agriculture and Lou Shu’s Illustrated Poems on Farming and Weaving. However, the advanced techniques described in these works were used mainly in upland areas of western Jiangnan. In general, the farming techniques commonly used in the lowland plains during the Song did not surpass the level that they reached in the Tang (Osawa 1993: 40-44;1996: 239, 242, 248; Adachi 1985). The techniques described in Ming-Qing agricultural handbooks, in contrast, were derived primarily from farming practices in the eastern plains.85 These techniques were far superior not only to those used in the same region during the Song but also to those described by Chen Fu and Lou Shu. From a long-run perspective, all the incremental changes in farming did lead toward the same end—the elaboration of a skill—oriented farm technology focused on enhancing the productivity of both land and labor.

Few present-day scholars would agree with the older view that the agrarian economy simply stagnated in the Ming-Qing period, but the early and middle Qing are still regarded as a period of minimal development. In my recent work, however, I have identif ied important advances in land improvement and utilization, agricultural technology, farm management, and re-allocation of household labor that took place in the seventeenth and eighteenth centuries. As a result, both land and labor productivity rose considerably, a rise that belies the stagnation hypothesis (Li Bozhong 1998: chap.9). As noted above, one of the most signif icant transformations in Jiangnan agricultural history during the millennium before 1850 was the shift in land exploitation from the uplands of western Jiangnan to the lowland plains. This shift began during the Tang and was largely completed by the end of the Southern Song. After the Southern Song, new advances in cultivation focused on raising the productivity of the land in the plains of eastern Jiangnan. In general, farming in Jiangnan can be characterized as extensive before the Tang, and intensive from the Tang onward. The major features of this intensif ication were (1) an increase in labor and capital inputs per unit of cultivated land; (2) increased output; and (3) a decrease in farm size (Li Bozhong 1990; chap.4). This pattern of intensif ication continued into the Ming and Qing (Li Bozhong 1984, 1985a, 1994a, 1996). Therefore, the classic model of premodern Jiangnan agriculture—a small farm with high yields—f irst appeared in the Tang and matured in the Qing. Chinese agronomists repeatedly applauded the benef its of smaller farms and more intensive farm management.86 Finally, other advances related to more intensive farm management, including the commercialization of farming, the planting of industrial crops, and the development of rural industries, also f irst appeared in the Tang and peaked in the Qing.

This shift can be seen clearly in the spatial concentration of water control projects. Before the Tang, most flood and irrigation works were built in western Jiangnan, especially in Jiangning, Zhenjiang, and Huzhou. But in the Tang and Song, more and more were built in the Jiangnan plain, particularly east of Lake Tai (Li Bozhong 1990: 76, 86; Shiba 1988: 204-8). In the Ming and Qing the vast majority were concentrated in the Jiangnan plain. In the Tang and Song, land exploitation involved large-scale reclamation, primarily through the enclosure of swamps and marshes to expand the amount of land under cultivation. After the Song, land improvement focused more on raising the productivity of the land, for example, drying the f ields in order to plant winter crops of wheat or beans. In most cases there was no hard and fast line between land reclamation and land improvement. In fact, land improvement to some degree took place as soon as land was opened. In any case, both the quantitative increase and the qualitative improvement increased farm output. The shift from quantitative increase to qualitative improvement of cultivated land, in this sense, cannot be considered a def inite turning point in the pattern of land use.

In contrast to the Song revolution and the Ming-Qing stagnation paradigms, changes in farm technology and land exploitation reveal a long process of gradual improvement and dissemination from the Song to the Qing. The development of Jiangnan agriculture throughout the Song-Yuan-Ming transition does not represent a departure from the more long-run trends during the millennium leading up to the mid-Qing. Only after we discard the paradigms of the Song revolution and the Ming-Qing stagnation can we assess the place of the Song-Yuan-Ming transition in Jiangnan’s agrarian history. The evolution of agriculture in the thirteenth and fourteenth centuries was marked not by a sudden or dramatic rupture but by cyclical fluctuations and spatial variations. In other words, the Song-Yuan-Ming transition was not a turning point but a phase in a continuous process of development.

References

Abel, Wilhelm. Agricultural Fluctuations in Europe, from the Thirteenth to the Twentieth Centuries. 1980. London: Methuen.

Adachi Keiji. Sōdai Ryōsetsu ni okeru suitōsaku no seisanryoku suijun. Kumamoto daigaku bungakubu ronsō 17. 1985. 80-100.

Amano Motonosuke. Chūgoku nōgyōshi kenkyū. 1979. Rev. Ed. Tokyo: Ochanomizu shobo.

Cao Shuji. Dili huanjing yu Song-Yuan shidai de chuanranbing. Lishi dili 1995. 1995. No.12: 183-92.

Chao, Kang. Man and Land in Chinese History: An Economic Analysis. 1986. Stanford: Stanford University Press.

——. (Zhao Gang). Zhongguo chengshi fazhan shi lunji. 1995. Taibei: Lianjing chuban.

Chen Heng. Bunongshu yanjiu. 1963. Rev. Ed. Beijing: Nongye chubanshe.

——. Bunongshu jiaoshi. Beijing: Nongye chubanshe.

Chen Zhen. Songchao shiqi de Nanjing jingji. In Nanjing jingjishi. 1996. Ed. Nanjingshi renmin zhengfu yanjiushi. 122-40. Beijing: Zhongguo nongye keji chubanshe.

CongHanxiang. Lun Mingdai Jiangnan diqu de renkou midu jiqi dui jingji fazhan de yingxiang. Zhongguoshi yanjiu 1984. 1984. No.3: 41-53.

Durand, J. D. The Population Statistics of China, AD 2-1953. Population Studies 13. 1960. No.3: 209-56.

Elvin, Mark. The Pattern of the Chinese Past. 1973. Stanford: Stanford University Press.

Fan Jinmin. Mingchao shiqide Nanjing jingji. In Nanjing jingjishi. 1996. Ed. Nanjingshi renmin zhengfu yanjiushi. pp. 162-196. Beijing: Zhongguo nongye keji chubanshe.

Fan Jinmin and Xia Weizhong. Suzhou diqu shehui jingjishi (Ming-Qing volume). 1993. Nanjing: Nanjing daxue chubanshe.

Fang Zaihui, Tang Qilin, and Chen Mingda. Zhejiangde gengzuo zhidu. 1984. Hangzhou: Zhejiang kexue jishu chubanshe.

Fu Yiling. Lun Ming Qing shehui jingji de fazhan yu chizhi. In Fu Yiling Ming Qing shehui jingji shi lunji. 1982. 103-08. Beijng: Renmin chubanshe.

Fu Zongwen. Songdai caoshizhen yanjiu. 1988. Fuzhou: Fujian renmin chubanshe.

Gao Liangzhi and Li Lin. Shuidao qixiang shengtai. 1992. Beijing: Nongye chubanshe.

Ge Jianxiong. Zhongguo renkou fazhanshi. 1991. Fuzhou: Fujian renmin chubanshe.

Guo Wentao. Zhongguo nongye keji fazhan shilüe. 1988. Beiing: Zhongguo kexue jishu chubanshe.

Hamashima Atsutoshi. Tudi kaifa yu keshang huodong: Mingdai zhongqi Jiangnan dizhu zhi touzi huodong. In Zhongyang yanjiuyuan dierjie Guoji Hanxue huiyi lunwenji 1. 1989. 101-22. Taibei: Acade mia Sinica.

Han Maoli. Songdai nongye dili. 1993. Taiyuan: Shanxi guji chubanshe.

Han Rulin. Yuanchao shi. 1986. Beijing: Renmin chubanshe.

He Zhongli. Songdai hubu renkou tongi wentide zaitantao. In Songshi lunji. 1993. 37-62. Xuchang: Zhongzhou shuhuashe.

Ho, Ping-ti (He Bingdi). Studies on the Population of China, 1368-1953. 1959.

Cambridge, Mass.: Harvard University Press.

Nan Song zhijin tudi shuzi de kaoshi he pingjia. Zhongguo shehui kexue. 1985. No.2: 133-85; No.3: 125-47.

Zhongguo gujin tudi shuzide kaoshi he pingjia. 1988. Beijing: Zhongguo shehui kexue chubanshe.

Hollingsworth, T. H. Historical Demography. 1969. Ithaca, N. Y. : Cornell University Press.

Hymes, Robert. Some Thoughts on Plague, Population, and the Sung-Yuan-Ming Transition: The McNeill Thesis after Twenty Years. Paper present at the conference The Song-Yuan-Ming Transition in Chinese History. 1997. Lake Arrowhead, Calif., June 5-11.

Katō Shigeshi. Shina ni okeru Senjō-ine saibai no hattatsu ni tsuite. In Katô, Shina keizaishi kōshō 2. 1952-53a. 659-75. Tokyo: Tōyō bunko.

——. Nan Sō no shuto Rin’anfu no toguchi ni tsuite. In Katō, Shina keizaishi kōshō 2. 1952-53b. 404-20. Tokyo: Tōyō bunko.

Kitada Hideo. Sō Gen Min Shin Chōgoku Kōnan sankakushū nōgyö no shinka to nōson shukōgyō ni kansuru hatten kenkyū. 1986-1987 nendo kagaku kenkyū hiho jokin kenkyū seika hōkokusho. (ippan kenkyū C). 1988.

Kusano Yasushi. Sōdai minden no densaku keitai. Shisō 10. 1969. 12-112.

Li Bozhong. Woguo daomai fuzhongzhi chanshengyu Tangdai changjiang liuyu kao. Nongye kaogu 1982. 1982. No.2: 65-72.

——. Ming Qing shiqi Jiangnan shuidao shengchan jiyue chengdude tigao. Zhongguo nongshi 1984. 1984. No.1: 24-37.

——. Sang zheng dao tian’yu Jiangnan nongye shengchan jiyue chengdude tigao. Zhongguo nongshi 1985. 1985. No.1: 1-12.

——. Ming Qing Jiangnan nongye ziyuande heli liyong. Nongye kaogu 1985. 1985. No.2: 150-63.

——. Ming Qing Jiangnan zhongdao nonghu shengchan nengli chutan. Zhongguo nongshi 1986. 1986. No.3: 1-12.

——. Tangdai Jiangnan nongyEde fazhan. 1990. Beijing: Nongye chubanshe.

——. Jianlun Jiangnan diqu’ de jiEding. Zhongguo shehui jingjishi yanjiu 1991. 1991. No.1: 100-7.

——. Tian, di, ren’ de bianhua yu Ming Qing Jiangnande shuidao shengchan. Zhongguo jingjishi yanjiu 1994. 1994a. No.4: 103-21.

——. Kongzhi zengzhang, yi bao fuyu: Qingdai qianzhongqi Jiangnande renkou xingwei. Xin shixue 5. 1994b. No.3: 25-72.

——. Ren geng shi mu’ yu Ming Qing Jiangnan nongminde jingying guimo. Zhonguo nongshi 1996. 1996. No.1: 1-14.

——. Funü banbiantian’ juese de xingcheng: Ming Qing Jiangnan nongjia funü laodong wenti tantao zhier. Zhongguo jingjishi yanjiu 1997. 1997. No.3: 10-22.

——. Agricultural Development in Jiangnan, 1620-1850. 1998. Houndmills, Eng.: Macmillan Press Ltd. and New York: St. Martin’s Press.

——. Forthcoming. Fazhan yu zhiyue: Ming Qing Jiangan shengchanli yanjiu. Taibei: Lianjing chuban.

Li Gan. Yuandai shehui jingji shigao. 1985. Wuhan: Hubei renmin chubanshe.

Li Longqian. Ming Qing jingji shi. 1998. Guangzhou: Guangdong gaodeng jiaoyu chubanshe.

Liang Fangzhong. Zhongguo lidai renkou, tiandi, tianfu tongji. 1980. Shanghai: Shanghai renmin chubanshe.

Liang Gengyao. NanSong de nongcun jingji. 1984. Taibei: Lianjing chuban.

Liang Jiamian. Zhongguo nongye kexue jishu shigao. Ed. 1989. Beijing: Nongye chubanshe.

Lin Jinshu. Mingchu Wuzhong diqu shehui jingji zhuangkuang chutan. In Mingshi yanjiu luncong 2. 1983. Ed. Zhongguo shehui kexueyuan lishi yanjiusuo Mingshi yanjiushi. 197-219. Nanjing: Jiangsu guji chubanshe.

Lin Shiji. Ming Qing shidai Jiangnan diqu shizhen zhi shuliang fenxi. 3 pts. Shihuo yuekan. 1978. n. s. ,8 , No.6: 27-42, No.7: 30-41, No.8: 15-30.

Maddison, Angus. Chinese Economic Performance in the Long Run. 1998. Paris: Centre of the Organisation for Economic Co-operation and Development.

McNeill, William H. Plagues and Peoples. 1976. New York: Doubleday, Anchor Books.

Min Zongdian. Song Ming Qing shiqi taihu diqu shuidao muchanliang de tantao. Zhongguo nongshi 1984. 1984. No.3: 37-52.

Miyazawa Tomoyuki. NanSō kannōbun—nōmin shihai no ideorogii. In Chūgoku shizō no

zaikōsei—kokka to nōmin. 1983. Ed. Chugokushi kenkyukai. 213-254. Tokyo: Bunrikaku.

Mori Masao. Minsho Kōnan no kanden ni tsuite—Sōshū, Sōkō nifu ni okeru sono gutaizō. Tōyōshi kenkyū 19. 1960. No.4: 1-18.

Osawa Masaaki. Tōdai Kōnan no suitōsaku to keiei. In Chūgoku shizō no zaikōsei—kokka to nōmin. 1983. Ed. Chūgokushi kenkyūkai. pp. 129-170. Tokyo: Bunrikaku.

——. Shin Fu nōsho no kenkyū. 1993. Tokyo: Nosan gyoson bunka kyokai.

——. Tō Sō henkakuki nōgyō shakaishi kenkyū. 1996. Tokyo: Kyūko shoin.

Postan, M. M. Medieval Agrarian Society in Its Prime: England. In The Cambridge Economic History of Europe, 2d Ed I. 1966. Ed. M. M. Postan. 548-632. Cambridge, Eng.: Cambridge University Press.

Qi Xia. Songdai jingishi. 1987. Shanghai: Shanghai renmin chubanshe.

Qi Yanping. Su’nan qiuling gangdi shuidao gongshui wenti. Dili 1963. 1963. No.1: 102-4.

Qiu Shusen and Wang Ting. Yuandai hukou chuyi. In Yuanshi luncong 2. 1983. Ed. Yuanshi yanjiuhui. III-24. Beijing: Zhonghua shuju.

Shi Daogang, Sun Yili, and Wang Chaozhong. Cong sanbu nongshu kan Yuanchao nongye shengchan. In Nanjing nongye daxue, Lishixi, Yuanshi yanjiushi. Ed. Yuanshi lunji. 1984. 278-98. Beijing: Renmin chubanshe.

Shi Weimin. Yuandai shehui shenghuo shi. 1996. Beijing: Zhongguo shehui kexue chubanshe.

Shiba Yoshinobu. Sōdai Kōnan keizaishi no kenkyū. 1988. Tokyo: Tokyo daigaku, Tōyō bunka kenkyūjo.

Shuilibu, Huanghe shuili weiyuanhui, Huanghe shuilishi shuyao bianxie zu. Huanghe shuili shi shuyao. 1984. Beijing: Shuili dianli chubanshe.

Sudō Yoshiyuki. Nan-Sō inasaku no chiikisei. Shigaku zasshi 70. 1961. No.6: 727-79.

Wang Chaosheng. Zhongguo gudai gengzhi tu. Ed. 1995. Beijing: Zhongguo nongye chubanshe.

Wang Yuhu. WangZhen nongshu. Ed. 1981. Beijing: Nongye chubanshe.

Watanabe Tadayo and Sakurai Yumio. Chūgoku Kōnan no inasaku bunka—sono gakus-aiteki kenkyū. Eds. 1984. Tokyo: Nihon hōsō shuppan kyōkai.

Wu Dan’ge. Mingdai tudi zhidu he fuyi zhidude fazhan. 1982. Fuzhou: Fujian renmin chubanshe.

Wu Hui. Zhongguo lidai liangshi muchanliang yanjiu. 1985. Beijing: Nongye chubanshe.

Yanagida Setsuko. Sōdai gyōson no kadōko ni tsuite. Tōyō gakuhō 40. 1957. No.21: 113-50.

Yang Guozhen and Chen Zhiping. Mingshi xinbian. 1993. Beijing: Renmin chubanshe.

Yang Peigui. Yuandai difang zhengfu. 1975. Taibei: Haohan chubanshe.

Yao Wenhao. Zhexi shuili shu jiaozhu. Ed. Wang Jialun. 1984. Beijing: Nongye chubanshe.

You Xiuling. Zhanchengdao zhiyi. Nongye kaogu 1983. 1983. No.1: 25-32.

——. Songdaide shuidao shengchan. Zhongguo shuidao kexue 1. 1. 1986.

——. Zhongguo daozuo shi. 1995. Beijing: Zhongguo nongye chubanshe.

Yu Yefei. Zhongguo lidai liangshi pingjun muchanliang kaolüe. Chongqing shifan xueyuan xuebao 1980. 1980. No.3: 8-20.

Zhang Fang. Ning-Zhen-Yang diqu lishishangde tangba shuili. Zhongguo nongshi 1994. 1994. No.2: 32-42.

Zhao Gang. Zhongguo chengshi fazhan shi lunji. 1995. Taibei: Lianjing chuban.

Zhao Wenlin and Xie Shujun. Zhongguo renkou shi. 1988. Beijing: Renmin chubanshe.

Zheng Zhaojing. Taihu shuili jishu shi. 1987. Beijing: Nongye chubanshe.

Zhou Shengchun. Lun Songdai weitiande lishi diwei yu yingxiang. 1981. M. A. thesis, Hangzhou University, Department of History.

——. Shilun Songdai Jiangnan shuilitiande kaifa he dizhu suoyouzhide tedian. Zhongguo nongshi 1995. 1995. No.1: 1-11.

Zhu Kezhen. Zhongguo jin wuqiannian lai qihou bianqiande chubu yanjiu. In Zhu Kezhen kepu chuangzuoxuanji. 1973. 95-102. Beijing: Kexue puji chubanshe.

---

1　Among Chinese scholars, Fu Yiling (1982) believes that China’s economy was ahead of the world economy before the fourteenth century but fell behind in the f ifteenth and sixteenth centuries. Qi Xia (1987: 2-3, 30-31) likewise claims that a “reversal” occurred in the late Song and Yuan. Before then China had been the most economically advanced country in the world. But, with the invasions by the Jurchens and Mongols, Chinese economic growth slowed and stagnated. A recent example in Western scholarship is the work of Angus Maddison (1998), who argues that China’s per capita GDP rose substantially during Song times but thereafter stagnated until 1820.

2　Geographically, “Jiangnan” is here def ined as the southern half of the Yangzi Delta, a region comprising the prefectures of Jiangning (Nanjing), Zhenjiang, Changzhou, Suzhou, Songjiang, Huzhou, Jiaxing, and Hangzhou in the southern part of present-day Jiangsu province and the northern part of Zhejiang province. For the logic of this def inition, see Li Bozhong 1991.

3　For example, Fu Zongwen (1988: 315) thinks that there was a large “surplus population” in this region as early as the early Southern Song. Liang Gengyao (1984: 17) also believes that population pressure was a serious problem in Jiangnan during the Southern Song. Cong Hanxiang (1983) argues that the three provinces of Jiangsu, Zhejiang, and Jiangxi suffered from severe population pressure throughout the Ming. Other scholars concur, asserting that “the contradiction between too much population and too little land” became acute in the Song (Qi Xia 1987: 73, 178) or the early Ming (Fan and Xia 1993: 63).

4　According to Ge Jianxiong (1991: 192, 210, 222, 236-37), the total population of China might have been near 70 million (including the Mongols and their allies) in the early thirteenth century. The Yuan population reached a new high of about 85 million in the early 1340s and then plunged to about 54 million in 1370. It took until the early 1390s for the population to rise to about 70 million. In contrast, Zhao Wenlin and Xie Shujin (1988: 279, 310, 357) estimate that China’s population was 108 million in 1210, 55 million in 1279, about 87 million in 1350, and 60 million in 1381. In other words, China’s population stagnated according to Ge but dropped by more than one-third according to Zhao and Xie. In my opinion, Ge’s estimate is more persuasive.

5　A typical example is Ye Ziqi (d.1385?), a Jiangnan native who lived through the Yuan-Ming transition. Ye wrote: “The whole country enjoyed peace for six or seven decades after Khubilai Khan had united China. During this time, penal laws were not severe, taxation not heavy, and weapons seldom used; the living could be supported, the dead buried, and travelers could go for a long distance and f ind sanctuary on the road no less than at home. It was really a prosperous age” (see Ye Ziqi, Caomuzi [Beijing: Zhonghua shuju, 1959], 3A: 47).

6　Shiba’s f igures (1988: 124, table 4) are rendered as persons per kilometer. Here I have converted them into population totals by multiplying the number of persons by the area of each prefecture. Because there are no f igures for Yuan Changzhou and Ming Jiaxing-Songjiang in Shiba’s table, I have taken the off icial census f igures for these two prefectures for the two periods from Liang Fangzhong 1980: 181-82 and Cong 1984.

7　For example, according to Zhao and Xie’s (1988: 594-95) estimates cited above, the ratio of population in Zhejiang for 1210, 1290, and 1381 is 100: 172: 157. But an increase of 72 percent over the period 1210 to 1290 is implausible.

8　According to Shiba’s (1988: 146) calculation, between 1080 and 1279 population density grew more rapidly in Jiangnan than in adjacent areas such as eastern Zhejiang. For example, population density increased from 135 to 261 persons per square kilometer in Hangzhou, from 39 to 137 in Suzhou, and from 81 to 159 in Zhenjiang. In contrast, population density increased only from 81 to 96 persons per square kilometer (in 1199) in Mingzhou (Ningbo) and from 77 to 137 in Shaoxing.

9　According to Ge Jianxiong’s (1991: 236) estimate, China’s population fell by as much as 30 percent in the Yuan-Ming transition. The fourteenth century saw a decrease in population in most of Europe. The population of England, for example, decreased by 30-50 percent in the second half of the fourteenth century (Abel 1980: 94; Postan 1966: 562). In France, the Black Death alone killed one-third of the population and left the total population in the 1470s one-third less than that of a century before (Abel 1980: 93).

10　See Richard von Glahn’s chapter in this volume (“Towns and Temples: Urban Growth and Decline in the Yangzi Delta, 1100-1400”, in Paul Jakov Smith, Richard von Glahn eds: The Song-Yuan-Ming Transition in Chinese History, the Harvard University Asia Center, 2003): 176-211.

11　Because the imperial capital was no longer in Jiangnan and the military confrontation between the north and south of China had ended, the number of off icials and military men was much smaller in Yuan Jiangnan than in the Southern Song. Thus, the urban population decreased, but the number of “real” urban residents seems not to have changed much. A typical example is Hangzhou. In the Southern Song, civil servants and military men and their families accounted for the majority of the urban residents of Hangzhou. Despite the loss of population due to the removal of the imperial capital to the north, Hangzhou’s urban population apparently did not decline drastically in the Yuan, if Marco Polo’s description is reliable (cf. Hollingsworth 1969: 247). The off icial population f igures shown in Table 4.1 also conf irm this point: the population of Hangzhou prefecture declined by only 8 percent between the Southern Song and the Yuan.

12　In contrast to the relatively smooth Song-Yuan transition, in the f inal years of the Yuan 60-70 percent of the urban residents of Hangzhou city starved to death, according to Tao Zongyi’s contemporary observation (Tao, Nancun chuogeng lu) [SKQS ed.], 1: 19b-20a. Table 4.1 shows that the population of Hangzhou prefecture declined by 67 percent between the Yuan and the early Ming. It seems likely that this decrease resulted mainly from the loss of urban residents.

13　The 1578 f igures are more reliable. The 1393 numbers are a bit exaggerated, because they include a considerable amount of uncultivated land that would soon (or so Zhu Yuanzhang assumed) be brought under cultivation (Li Bozhong 1996).

14　Well before the famous 1580-82 national cadastral survey was carried out under Zhang Juzheng’s administration, efforts had been made in some Jiangnan localities, especially Suzhou and Songjiang, to reconcile the off icial f igures with real cultivated acreage (see Li Longqian 1988: 178-82; Yang Guozhen and Chen Zhiping 1993: 232-35). In the case of Jiangnan, the 1580-82 survey f igures, which are not extant, should be close to the 1578 f igures. For the northern Zhejiang prefectures of Hangzhou, Jiaxing, and Huzhou, the 1578 prefectural f igures are not available today, but the “Wanli”f igures in Zhejiang tongzhi (1683), juan 17, should be from the 1578 or 1580-82 set. The Wanli survey is likely to have been carried out in 1580-82 or earlier. Here I treat them as the 1578 f igures.

15　As Ping-ti Ho (1988: chap.3) points out, these off icial f igures are not necessarily reliable. The Yuan f igures are the most suspect; no cadastral surveys were conducted during the Yuan, and most Yuan f igures were simply copied from Southern Song records. Early Ming f igures typically are larger than Yuan ones, but the increase reflected the registration of previously unrecorded lands as well as the opening of new lands to cultivation (Fan and Xia 1993: 62). Moreover, the early Ming surveys found that the f igures in earlier records were inflated; these were adjusted downward (Ho 1988: 49). Despite these defects in off icial statistics, it seems clear from the evidence in Tables 4.4 and 4.5 and Ho’s research that the amount of cultivated land increased moderately.

16　The very high 1346 f igure for Huzhou, which is greater than the 1578 f igure, is problematic. But the general picture of the economic and demographic development of Huzhou given in Shiba 1988 suggests neither expansion nor dramatic decline of arable during this period.

17　Some of the f igures that Zhou used to support his estimate date not from the Southern Song but from the Yuan or even early Ming. Zhou needs to account for changes in the amount of cultivated land in these places between the Southern Song and early Ming.

18　For example: “There is no fallow land anywhere [in Suzhou]. All land, high and low, has been opened and made into f ields” (Fan Chengda, Wujun zhi [1192]: 2.7b). “In the Zhejiang region, no land is fallow” (Huang Zhen [1213-80], Huangshi richao [SKQS ed.], 78: 50b). “In Fujian and Zhejiang. land is limited, and the population dense. No land is fallow, and the people need not be admonished to be diligent in their husbandry” (Xu Yinglong [1168-1248], Dongjian ji [SKQS ed.], 13: 5b). “In the Liangzhe circuit, no land is fallow” (Wang Xunyou [ca.1142], cited in Ma Duanlin, Wenxian tongkao [Shanghai: Shangwu yinshuguan, 1936], 5: 63a).

19　Jujia Biyong Shilei, Reprinted, Kyoto: Chūbun shuppansha, 1984, 9: 7b-8a, “chushui fa” 畜水法.

20　Zhou Wenying, “Sanwu shuili”, cited in Yao 1984: 87-89.

21　Yang Ju, “Chongkai Guhuipu ji”, cited in Yao 1984: 32-34.

22　See Fan Chengda, “Shuili tuxu”, cited in Yao 1984: 39-41; Wu Zhizhong, “Shundao shuishi”, citedin Yao 1984: 74-76; and He Yi, “Shuili celüe”, cited in Yao 1984: 98-100. Ren Renfa, a famous hydrologist of Yuan Jiangnan, cites Fan’s opinion and adds: “Building dikes, dredging rivers, and creating floodgates are equally important. None of them should be neglected” (Ren Renfa, Shuili yida, cited in Yao 1984: 61-64). But another Yuan hydrologist argued that “building floodgates across rivers to store water is not a good or a permanent solution after all” (Zhou Wenying, “Sanwu shuili”). Generally speaking. the major water storage methods used in Southern Song and Yuan Jiangnan were building dikes and dredging ponds.

23　In the Southern Song. floodgates were built to drain large areas for poldering (Shiba 1988: 379-80). In the Yuan, by contrast, river dredging eclipsed floodgates as the primary method of water control.

24　Ren Renfa, Shuili yida.

25　See the documents collected in Yao 1984: 59-89.

26　A leading Northern Song hydrologist, Jia Dan, reported that the state asked local people in Jiangnan to build small polders that enclosed the f ields of only ten to twenty households. At the time, existing enclosures encompassed 13,000-26,000 mu of f ields on average (Zheng 1987: 115-16). Throughout the Northern and Southern Song. even though large polders predominated, small enclosures also were being built.

27　“Guitian are earthen dikes for protection against floods. They look like weitian but are smaller in size ... If floods come, the small size and high, solid dikes prevent inundation, and water within the f ields can be easily pumped out ... [the size of the area within the guitian] is one hundred mu at most” (Wang Yuhu 1981: 188).

28　The earliest extant published version of Ren’s text dates from 1640. See Xu Guangqi, Nongzheng quanshu jiaozhu, ed. shi Shenghan, Shanghai: Shanghai guji chubanshe, 1979: 321.

29　Yang Ju, “Chongkai Guhuipu ji.”

30　Zhou Wenying, “Sanwu shuili.”

31　For example, Qi Xia (1987: 178) argued that “intensive farming reached a new high in the area centering on the Lake Tai basin” during the Song because of (1) the invention of the so-called Jiangdong plow and (2) intense population pressure on the land. The Jiangdong plow, which actually was invented in the Tang, in theory, enabled farmers to plow more deeply. However, this plow did not work very well in Jiangnan paddy-f ields, and the tieda, a Yuan innovation, proved far superior for this task (Chen Hengli 1963: 129-30; Osawa 1993: 64).

32　An example is the synthesis of Southern Song paddy technology constructed by Liang Gengyao (1984: 155-56) based on studies by Amano Motonosuke, Sudō Yoshiyuki, Zhao Yashu, Chen Liangzuo, and others.

33　The six gazetteers are Qinchuan zhi (1254); Yufeng zhi (1251); Wujun zhi (1192); Wuxing zhi (1201); Ganshui zhi (1203); and (Xianchun) Lin’an zhi (1268). The Qinchuan zhi and Yufeng zhi record 35 and 34 varieties, respectively; the other gazetteers only record several varieties each. Excluding repetitions, the total is 51 (You 1995: 87-89).

34　Zhenjiang zhi (1332), 4: 1a-b.

35　Several gazetteer editors made clear that they had omitted certain varieties of rice. For example, the editors of the 1201 Wuxing zhi (20.3a) wrote, “If you ask the farmers, the varieties of glutinous rice far exceed this number, but the names farmers give them are crude and vulgar and do not deserve to be recorded.” The editors of the 1332 Zhenjiang zhi (4.1a) also mentioned that “varieties of rice of Jiangnan are too numerous to name in full.” You Xiuling (1995: 87) therefore concludes that “the varieties not recorded in the literature must be far more numerous than those that were.”

36　Because of the longer growth period, intermediate and late-ripening jingdao varieties are higher in nutrition than early-ripening xiandao, a quality that was well known already in the early Southern Song. Shu Dan wrote, “Jingdao rice is low in yield but high in price. After submitting [this high-quality rice] in tax payment to the government, only upper-class households can afford to eat it... The so-called xiaogu (“small rice,” a colloquial term for xiandao) produces high yields and is inexpensive; it is consumed by middle- and lower-class families” (Shu Lin, “Yu Chen Cang lun changpingcang”, Shu Wenjing ji [SKQS ed.], xia: 9b-11a).

37　Huang Shengceng, Lisheng yujing daopin (CSJC ed.): 1-7.

38　See Wujun tujing xuji (1084), shang: 7a.

39　See Lin’an zhi (1268), 8: 6b.

40　Siming zhi (1227), 4: 16b-17a.

41　Wumen shilei, cited in Suzhou fuzhi (1379), 42: 1b.

42　Because the period for transplanting early rice is earlier than the harvest of winter wheat, it is impossible to plant early rice and winter wheat in the same plot (Li Bozhong 1994a). Osawa (1993: 75-76) also points out that with the exception of Huanglü (yellow-green) rice, which was not among the major strains planted in Jiangnan, the double-cropping was diff icult given the varieties of rice and wheat grown in Southern Song Jiangnan. For this reason, double-cropping of rice and wheat was not promoted in the most important Southern Song agricultural handbook of Jiangnan, Chen Fu’s Treatise on Agriculture.

43　Five of the eleven varieties of rice identif ied in Tang sources are also mentioned in Ming and Qing agricultural handbooks. Moreover, 27 out of the 35 varieties recorded in the mid-Ming Lisheng yujing daopin and 26 of the 38 in Suzhou fuzhi (1824) were already known in the Southern Song (You 1995: 83-86, 92-96).

44　Kuang Fan, Bianmin tuzuan, ed. shi Shenghan and Kang Chengyi, Beijing: Nongye chubanshe, 1959, 8: 34; Wang Yuhu 1981: 233.

45　Other new tools served not to increase the productivity of the land but to ease the burdens of workers. Among them were the yunpa, a weeding hoe that eliminated weeding by hand; the haoma, a wooden seat used when pulling up seedlings for transplanting; the fuke, a raincoat made of coconut f ibers; and the bigou, bamboo sleeves that protected the arms when weeding or harvesting.

46　See Osawa 1993: 63-64. In a strict sense, turning the ashes and stubble into the soil cannot be regarded as using fertilizer, since they were not added to the f ields from outside.

47　Lou Shu, Gengzhi tu shi, in Wang Chaosheng 1995: 187-92.

48　Zhao Mengfu, Gengzhi tu shi cited in Wang Chaosheng 1995: 193-97.

49　Kuang Fan, Zhuzhici xiaoyong, in Bianmin tuzuan: 6.

50　Fang Zaihui (Fang et al.1984: 161-67) points out that in present-day Zhejiang double-cropping of rice and dry-land crops (especially broad beans) is superior to any of the alternatives-double-cropping of rice, double-cropping of rice and green manure crops, and single-cropping of rice. Under this regime, not only can nutrient elements in the soil be utilized more effectively, but the rotation of wet- and dryland crops improves soil structure, decreases toxicity, and increases the humus and air content in the soil. reduces the incidence of insect pests and plant diseases, and kills most weeds. All these benef its enhance soil fertility.

51　Under double-cropping of rice, the paddy is full of water almost all year long. Under single-cropping of rice or double-cropping of rice and green manure crops, the paddy is not plowed and therefore is never dried by the sun. The soil remains waterlogged, even in the winter when rice is not growing.

52　Osawa (1993: 47, 74-76) suggests the possibility that early rice with a short growing season (around ninety days) was double-cropped with wheat. But such subvarieties of early rice were poor in quality and yield and seem to have been planted only as famine relief crops.

53　Double-cropping of rice and winter dry-land crops (particularly wheat) is very labor intensive. In the case of Jiangnan, this cropping regime could be carried out only on small farms (Li Bozhong 1996). The labor/land ratio of Southern Song Jiangnan was not able to support this cropping system.

54　Song shi (Shanghai: Shanghai guji chubanshe, 1986), 173: 4178.

55　See Fang Hui, Xu gujin kao (SKQS ed.), 19: 14a-b.

56　Huang Zhen, Huangshi richao, 78: 50b-51a.

57　Yuxi zhi (ca.1241-52), cited in Shimen xianzhi (1872), 3: 105b-6a.

58　Zhao Mengfu, Gengzhi tu shi. Zhao declared that “in the eastern part of Huzhou, the paddy is fertile and productive; winter wheat is harvested, and the land is suitable for rice cultivation.” On the basis of this passage, Kitada (1988: 13) concluded that double-cropping was practiced in this area. But in his “Twelve Poems on Plowing.” Zhao wrote that Huzhou farmers plowed f ields in the second lunar month, transplanted rice in the fourth month, harvested wheat in the f ifth month, weeded f ields in the sixth month, and harvested rice in the eighth month. Clearly, given this schedule, rice and wheat could not have been planted in the same plots. Zhao makes no mention of when the wheat was planted, but he does say that in the tenth month “all f ields are empty.” Since winter wheat had to be planted by the tenth month, Zhao’s statement at the very least implies that wheat was not planted in paddy f ields. Even in the seventeenth century, no winter dry-land crops were planted in paddy-f ields of eastern Huzhou; see Chen Hengli 1983: 106.

59　Kuang Zhong. Kuang taishou ji (reprinted-Nanjing: Jiangsu renmin chubanshe, 1983): 83. Fan Jinmin and Xia Weizhong (1993: 72-73) argue that Kuang Zhong understated the amount of wheat planted in Suzhou at that time. But although double-cropping of rice and winter crops increased in the early Ming. It remained a minor practice until later times (Li Bozhong 1998: 29-33).

60　Zhenjiang zhi (1332), 3: 16b-19b, 5: 1b-2b.

61　Fang Hui, Xu gujin kao, 18: 14a-b.

62　Tao Zongyi, “Jiantian li”, Nancun chuogeng lu, 23: 8a.

63　Zhou Chen, “Yu xingzai hubu zhugong shu”, in Chen Zilong et al., eds, Huang Ming jingshi wenbian (1637; Reprinted—Taibei: Guofeng chubanshe, 1964), 21: 13a-b.

64　Kuang Zhong. “Qing juntian reng zhaoli mindian zou”, Kuang taishou ji: 8.89.

65　Ming Shilu (reprinted—Taibei: “Zhongyang” yanjiuyuan, 1962), Shizong 27: 5b.

66　“Sun Shenyi hu tie”, in Pinghu xianzhi (1627), 10: 22b-25b.

67　According to Table 4.7, farmland per household was exceptionally large in early Ming Changzhou, which probably is attributable to the heavy loss of population that this area suffered in the Yuan-Ming transition. Most battles between Zhang Shicheng and Zhu Yuanzhang occurred in the Changzhou area. When the population recovered, farmland per household in Changzhou approached the level of the other prefectures in eastern Jiangnan. The 1578 survey f igures show, for example, that farmland per household in Changzhou fell sharply from the 1393 level, from 52 to 25 mu, which was close to the 1578 Songjiang f igure of 19 mu but much smaller than the 1578 Zhenjiang f igure of 49 mu. In the 1393 survey, the amount of farmland per household (44 mu) in Zhenjiang had been signif icantly lower than in Changzhou (52 mu). See Liang Fangzhong 1980: 435.

68　Only able-bodied adult males and females (ding or chengding) were registered in the Southern Song. and off icial f igures did not count the whole population. Scholars have calculated family size from ding/chengding f igures in various ways. He Zhongli (1993) suggests that a family averaged two ding in the Song. If buchengding (minor children) were included, the total would be f ive-six persons (three males and two or three females). But Jiangnan family size was much greater than the national average, according to He. In Jinling county of Changzhou prefecture, for example, a family had f ive ding on average in the Northern Song; if buchengding were included, the family would have had more than ten persons. Fu Zongwen (1988: 305) believes that a typical peasant family had ten persons in the Song; most of the examples he gives are from Southern Song Jiangnan. Liang Gengyao (1984: 80-82) argues that family size ranged between four and ten persons, but the most reliable f igures he cites represent those registered by the government as eligible for relief during famine years. These f igures indicate a family size of six persons in Lin’an (Hangzhou) prefecture. In addition, families in Cixi county in neighboring Ningbo prefecture had 7.5 persons on average in the late Song, according to the population f igures in Hu Ju’s memorial quoted in Siming zhi (1227), 16: 12a-b.

69　Zhenjiang zhi (1332), 3: 16b-18a.

70　Under the double-cropping regime, the time gap between the winter harvest and transplanting rice seedlings was very short. The most important job during this period was land preparation, which was men’s work. After that, women joined their husbands in transplanting, weeding, and harvesting. From the late Ming on, developments in the sexual division of labor and rural industrialization drew women’s labor out of farming. Since a man could work only 10 mu, if he owned more than 10 mu, he had to rent out the surplus (Li Bozhong 1996).

71　Cheng Gongxu, “Changshu chongkai zhichuan ji”, Qinchuan zhi (1254), 12: 21a-23a; see also ibid., 13: 6a-8a.

72　Zhang Pan, “Guizheng xiang yiyi ji”, in Qinchuan zhi (1254), 12: 8b-10a; Liu Zai, “Yiyi ji”, in ibid., 12: 10a-13b.

73　Yu Zhuo, “Songjiangfu zhuyi tianliangji” (1324), Songjiang fuzhi (1521), 6: 9b.

74　Huang Zhen, Huangshi richao, 73: 5a-b, 84: 20b.

75　In 1136, the Song emperor issued an edict to recruit people to cultivate fallow lands in Zhenjiang prefecture and Jiangxi. “The rent will follow the custom for private lands: 2 dou [i.e., 0.2 shi] for upper-grade f ields, 1.8 dou for mid-grade f ields, and 1.5 dou for lower-grade f ields” (Song huiyao jigao, shihuo 2: 23b-24a.) These rent f igures suggest yields of 0.3-0.4 shi per mu, although these rents probably were set artif icially low to attract tenants.

76　Huang Zhen, Huangshi richao, 84: 20b.

77　In the Southern Song, the dadou (big peck) used in rent payments equaled 1.3 standard dou, or 0.13 shi; see Qi Xia 1987: 375.

78　On quannongwen in the Southern Song, see Miyazawa 1983.

79　Yu Yefei and Wu Hui base their estimates on general accounts of land taxes; their only direct evidence is three yield f igures found in Huang Jin, “Huzhoulu Bao’en guangxiao chansi zhi tianshan bei”, Huang Wenxian ji (SKQS ed.), 10. Li Gan did not cite any evidence to support his estimates, which are unimaginably high.

80　“Xihu shuyuan zengzhi tian bei”, in Ruan Yuan, ed., Liangzhe jinshi zhi (Shike shiliao congshu ed.), 15.37b-39a; Huang jin, “Xihu shuyuan yitian ji”, Huang Wenxian ji, 7: 56a-57a.

81　The share of government lands relative to total landholdings was 85 percent in Songjiang; 63 percent in Suzhou, 32 percent in Zhenjiang, 29 percent in Jiangning, 27 percent in Jiaxing, 24 percent in Huzhou, and 15 percent in Changzhou (Mori 1960; Wu Dan’ge 1982: 26-27). No data are available for Hangzhou prefecture.

82　In 1425, Zhou Gan reported that “in Wujiang and Kunshan counties of Suzhou prefecture, formerly [i.e., before the Ming] a mu of private land was taxed at 0.05 shi per mu; peasants who rented f ields from wealthy families paid 1 shi in rent. [In the early Ming, ] the lands of wealthy households were conf iscated by the government. The rent was reduced to 0.8 shi” (Ming shilu, Xuanzong 6: l0a). These “conf iscated f ields” (chaomotian), together with the existing government-owned lands (guantian), formed the bulk of early Ming government lands (Mori 1960). Gu Yanwu (1613-82) observed that “in the case of plain f ields, tenants who rented private f ields usually paid 1 shi of rent, whereas farmers who worked in off icial f ields paid 7 dou (0.7 shi) at most. Farmers who worked government lands paid 4 dou. The levy (on government lands) was light compared to (private) rents” (Gu Yanwu, Tianxia junguo libing shu, Vol.7, cited in Mori 1960, 446: 22).

83　Changshu xianzhi (1539), 4.

84　In Changshu, the average yield was 0.65-1.36 shi in the Southern Song; I take 1.0 shi as the mean. If converted to the Ming standard of measurement, the yield in the Song would have been 0.6 shi per mu. In the early Ming, the average yield was more than 1 shi for mid-grade farms and thus probably double that of the late Southern Song.

85　See the following: Kuang Fan, Bianmin tuzuan; Huang Shenzeng, Lisheng yujing daopin; Shenshi nongshu; Pan Zengyi, Quzhong fa, in Wang Yuhu ed., Quzhong shizhong, Beijing: Caijing chubanshe, 1955: 114-34; Jiang Gao, Pumao nongzi, reprinted—Shanghai: Shanghai tushuguan, 1964; Zhang Lüxiang, Bu nongshu, in Chen Hengli 1983; Xi Cheng, Gengxin nonghua; and Xu Danfu, Nongshi youwen, in Nanxun zhenzhi, 30: 5a-16a.

86　Chen Fu remarked that “a proverb says that ‘planting more but harvesting less is not as good as planting less but harvesting more.’ It is really true” (Osawa 1993: 231). Wang Zhen, who knew Jiangnan agriculture well, said that “when a family manages its farm, it should f irst determine its labor capacity. It is better to have a small farm run well, rather than the reverse” (Wang Yuhu 1981: 23). A late Ming author declared that “the most important thing for managing your family farm is to work hard in plowing the f ields and apply more fertilizer. Plant less land but harvest more.” He further exhorted, “If farming tasks are performed well and supervised according to the rules, it is better to plant fewer f ields but farm more carefully. You cannot plant more f ields and not farm carefully.” He also cited a local proverb to validate his point: “Planting less [land] is better than planting more, because planting less will save not only your energy, but also your land” (Chen Hengli 1983: 69, 75). “Save your land”here means that because of higher productivity, a farmer could get the same output from only a part of his holding, and the remainder could be deployed for other purposes, including leasing it to others, to bring in added income.