Five
Infanticide and Family Planning in Late Imperial China: the Price and Population History of Rural Liaoning, 1774–1873

James Lee, Cameron Campbell, and Guofu Tan

Between 1700 and 1900 China's population more than tripled, increasing from 150 million to almost 500 million. This dramatic rise in population is probably the most frequently noted achievement of Qing society. It is also one of the most important elements in any explanation of Qing economic performance.^[1] Nevertheless, despite considerable research on the economic and demographic history of late imperial China, we have yet to devise precise demographic or economic measures for much of this period. In consequence, we have very little detailed quantitative knowledge about either the economy or the population; we also have little understanding about how population, as a variable, actually interacted with the economy during the eighteenth and nineteenth centuries.

Detailed population and price records do, however, survive in the historical archives of Taiwan and the People's Republic of China.^[2] This paper is

a preliminary attempt to use such materials to reconstruct the price and population history of Daoyi District, a rural suburb of Shenyang in Liaoning Province, for approximately 100 years, from 1774 to 1873. First, we summarize the results of an ongoing study of the population history of Daoyi and demonstrate that mortality and fertility differed sharply by sex. Second, we reconstruct the price history of five food grains (rice, millet, sorghum, wheat, and soybeans) and determine the degree of price integration in the prefectural market. Finally, we analyze the relationship between grain prices and demographic rates in order to prove that the differential rates by sex were the product of a systematic pattern of infanticide according to household situation as well as economic conditions. We conclude that in Liaoning both mortality and fertility were highly responsive to changes in economic circumstances.

The study of food prices and population has, of course, long been a central topic both in the historical demography and in the economic history of the preindustrial world. This is only natural, for population was everywhere one of the most dramatic and dynamic economic variables and fluctuations in population were at least in part a function of harvest variations. Numerous studies in European history have repeatedly discovered relatively strong positive correlations between food prices and mortality and even weak negative correlations between food prices and fertility.^[3] As we shall see, in Liao-

ning correlations between vital rates and prices were just as strong, but the patterns of population behavior were fundamentally different from European ones. Indeed, because of the widespread use of infanticide as a method of family planning, the strongest correlations link prices with fertility, not mortality. The Chinese apparently regarded infanticide as a form of postnatal abortion through which they could choose the number, spacing, and sex of their children in response to short-term economic conditions as well as their long-term family-planning goals.

Our demographic data for Liaoning come from an ongoing study of over 12,000 Chinese peasants who lived between 1774 and 1873. So far as we can tell, these peasants were direct descendants of an earlier Ming garrison.^[4] The Qing government certainly classified the population as Han Chinese and in the early seventeenth century organized them as members of the Han banner armies.^[5] Two thirds were farmers who lived in three villages (Baodao tun , Daoyi tun , and Dingjia fangshen ) in Daoyi District, a northern suburb of Shenyang. The rest were farmers originally from Daoyi, who had since moved to nearby villages. Because almost all of these villages were located near the provincial capital, the vast majority of these farmers undoubtedly produced food for the city market as well as for their own consumption. We have, however, almost no information on the specific structure of the village economy or the nature of these market relations during the eighteenth and nineteenth centuries.

What we do have are 85,000 individual records and 12,000 household records on the demography of this population throughout this period. This information is preserved in 25 triennial registers. These registers provide a nominative list of the families that received state banner land and in turn were liable for special corvée and military banner service. Specifically, they record for each person his or her name, age, occupation, family and lineage

(Table continued on next page)

(Table continued from previous page)

relationships, birth date, recent demographic events, and village of residence. The registers survive in the Liaoning Provincial Archives and were coded into machine-readable form at the Liaoning Population Research Institute and the California Institute of Technology.^[6] To the best of our knowledge, no other material records a Chinese peasant population before 1900 with such continuity and detail. We present a crude summary profile of all 25 registers in Table 5.1.

The sources are a product of the Eight Banner registration system.^[7] In Liaoning, Qing officials relied heavily on such records for civilian and military administration. They accordingly devised a remarkable system of internal cross-checks to ensure consistency and accuracy. First, they assigned every person in the banner population to a residential household (linghu ) and registered them on a household certificate (menpai ). Then they organized these households into clans (zu ) and compiled annually updated clan genealogies (zupu ). Every three years local authorities compared these genealogies with the household certificates to compile the population registers. Thanks to such efforts, the banner registers provide far more comprehensive and accurate data than the population registration system (baojia ) common elsewhere in China.

These registers do not, however, record the entire population. They have two related defects. On the one hand, nine of the registers from the century under observation are damaged or lost. We therefore have incomplete information on deaths and to a lesser extent births for 27 years.^[8] On the other hand, even when registers do exist, registration is still incomplete in the very early age groups. Almost no one below 2 sui is registered. Indeed, the mean age at first appearance for both sexes is 6 sui , which is slightly less than five Western years of age.^[9] Many children who died before 6 sui simply do not

appear in our records. Girls are more likely than boys not to be registered, especially after 1840. Our estimations of vital rates accordingly fall considerably short of the actual levels of fertility and mortality.

Fortunately, with the exception of female underregistration, these omissions, at least before 1840, do not appear to follow any selective bias. Such bias as there is appears to be uniform over time and household position. Thus, mean age at first appearance is quite consistent by sex, time, and household relationship, so that the data, although incomplete, are nevertheless sufficient to document a variety of distinct patterns of mortality and fertility behavior in Daoyi.^[10] These patterns by sex, age, household type, and family relationship, as well as economic condition, are far too consistent to be the product of underregistration. We summarize three particularly pertinent examples.

First, although the level of mortality during this century was moderate, deaths were distributed highly unevenly by sex. We contrast the overall mortality experiences of males and females in Table 5.2. We then present the experiences of the two sexes in periods of high and low overall mortality in Tables 5.3 and 5.4. The contrasts reveal significantly higher levels of mortality among females than among males but also far greater fluctuations in mortality among males than among females. The pattern of mortality, in other words, suggests a system of resource allocation wherein the female share, although relatively constant, was smaller than the male share. Males consumed more resources; but because they relied on the harvest surplus for their larger share, they were more vulnerable to harvest fluctuations than females. In Daoyi, in other words, the price of privilege was economic insecurity.

Second, the patterns of birth spacing and birth stopping strongly suggest that most married couples controlled their fertility to a considerable extent. Two very useful indications of such conscious limitation are the length of time between births and the age of women at the birth of their last child. Studies of many historical European populations show that in the absence of fertility control, birth intervals beginning with the second birth are rarely

much more than two years long.^[11] Moreover, the mean age at last birth is almost always within one year of age 40. In our population, by contrast, the mean age at last birth was only 34 (35.5 sui ). Furthermore, the mean birth interval beginning with the second birth was almost five years. As we can see in Figure 5.1, in striking contrast with European populations, where birth intervals increase by order of child, birth intervals in Daoyi actually decrease. According to these data, most couples decided to have fewer children than their natural limit and tended to space these children far apart.

Third, people in Daoyi appear to have used some form of sex-selective fertility control.^[12] Table 5.5 analyzes the birth histories of almost 1,000 com-

pleted marriages and computes the sex ratios by birth order and completed family size, that is, the total number of births registered to the parents by the time the mother reached age 45. The numbers are certainly exaggerated because of the underregistration of females, but it is the unusual pattern that is important. In single-child families there were 576 boys for every 100 girls. For families with two children ever born there were 211 boys per 100 girls at

[Full Size]

Fig. 5.1.
Preceding Birth Interval by Order of Child in Daoyi and France, 1774–1873

Note: Based on 1774–1873 data for Daoyi and 1500–1820 data for France.
Data for France are from Michael W. Flinn, The European Demographic System (Baltimore, 1981), p. 330.

the first birth and 450 boys to 100 girls at the second birth. For families with three children ever born the ratio was 156 boys to 100 girls at the first birth, 294 boys to 100 girls at the second birth, and 324 boys for every 100 girls at the last birth. This highly unnatural pattern continues through all other completed family sizes. The closer a girl's birth order was to the completed family size, the less likely she was to survive to registration. The pattern is too systematic to be explained by random underregistation.

Daoyi peasants, in other words, used sex-selective methods of fertility control to determine the number and sex of their children. Generally there was a strong preference for boys, since sons had a higher utility, especially during their parents' old age. Couples targeted the number of boys they wanted and stopped having children after the desired number had been reached. This behavior produced both the low mean age at last birth and the unusual pattern in Table 5.5, where sex ratios increase steadily with birth order. Whether or not a girl born before the cutoff would be allowed to live depended on many factors, one of the most important of which was wealth. Poorer couples who only planned on one or two boys would be less willing to be burdened with girls. Wealthier couples, however, would want more children and would be more likely to allow early girls to live. As a result, the sex ratios in Table 5.5 for couples who had few children are comparatively high, while the ratios for couples who had more children are comparatively moderate.^[13]

A study of age-specific fertility by household position underlines the strong relationship between wealth and fertility. Most peasants in Daoyi lived in highly hierarchical, complex households, where resources depended to a large extent on their position within the household.^[14] We would therefore expect fertility patterns to reflect two well-known Confucian principles of household organization, whereby household heads took precedence over other family members and senior relatives took precedence over junior relatives. Indeed, the calculation of fertility by household position, presented in Figure 5.2, confirms that, at least in Daoyi, most Chinese families obeyed such principles. The hierarchy of fertility begins with the head at the top, followed by the head's brothers and sons, then his uncles, brothers' sons, cousins, and cousins' sons. The marital fertility rate of household heads is

[Full Size]

Fig. 5.2.
Male Fertility in Daoyi by Family Relationship, 1792–1873 (births per 1,000 person-years)

Note: General male marital fertility equals male births over married men 15–35. General male fertility equals male births over
all men aged 15–35.

twice that of cousins' sons. The marital fertility of sons is not only higher than that of brothers' sons, it is almost as high as that of brothers and far higher than that of cousins. Removing the control for marriage accentuates the differences between the privileged and the dispossessed. Because chances of marrying depended on household position as well, not only does the gap between the head and cousins' sons widen considerably: so do the gaps between the head and his sons and brothers. In both cases as distance from the household head increases, fertility decreases.

Given the technology of birth control available in late imperial China, such distinctive sex-selective patterns of family planning could only arise if parents consciously chose the number and sex of their children at least in part through infanticide. The patterns according to birth order, completed family size, and household position are too rational and systematic to be explained by underregistration. Parents' decisions to use infanticide were, of course, not based solely on their long-term goals and household position. Couples must have been influenced by short-term conditions. Indeed, as we shall see, couples were influenced in the short term by the price of food.

The price data come from an empirewide system to monitor food conditions that began elsewhere in China as early as the late seventeenth century but did not extend to Liaoning until the late eighteenth century. Thereafter until well into the twentieth century, magistrates in each of Liaoning's two prefectures, Fengtian and Jinzhou, reported every ten days to the provincial government on food supply conditions, including the price of all major food grains, the state of the weather, and harvest yields when appropriate. The provincial governor in turn prepared for the emperor a brief summary each lunar month of the lowest and highest prices reported in each prefecture for five food grains (rice, husked and unhusked millet, sorghum, wheat, and soybeans). It is these monthly summaries that provide our price data for Fengtian prefecture.

The principal virtue of these price summaries is the systematic spatial and temporal coverage they provide across the entire empire for over 200 years. To date, we have collected over 1,500 of these monthly price reports for the prefecture, almost two thirds of which are from the century under consideration in this paper.^[15] Even after converting these lunar data to solar-month equivalents, we have complete or almost complete information (nine months' worth or more) for 65 years and no information for only five years (1791,

1815, 1822, 1823, and 1825).^[16] Figures 5.3 and 5.4 illustrate the price curves of the monthly lowest reported price and the monthly highest reported price for all five grains from 1774 to 1873 in taels of silver per shi of grain. These price data provide a systematic measure of food availability in Daoyi and, by extension, of economic conditions, which we can correlate with vital rates. They also enable us to identify which grains were most closely tied to specific changes in population and which grains may therefore have loomed largest in individual decision making on fertility and mortality.

It is important to remember, however, that these price materials also have a number of deficiencies, especially for microanalysis at the subprefectural level.^[17] They are denominated in silver and accordingly do not accurately reflect the retail market, which commonly used copper cash. They only provide us with the highest and lowest prices in each month and tell us nothing about the overall distribution of prices within each prefecture. Finally, they do not tell us the location of the reported prices. We consequently cannot calculate an average prefectural price. We cannot even assume that the monthly low and high prices were necessarily related. We can only assume that the data represent general price trends over time. For the purposes of this analysis we shall therefore separately analyze all ten types of price at our disposal.

Correlation coefficients between annual averages of different grain prices, summarized in Table 5.6, reveal a high degree of substitutability among most grains. This was especially true for the monthly low and high prices of husked and unhusked millet, as well as the monthly low and high prices of millet, rice, and sorghum. Monthly low and high prices for the same grain, however, are less strongly correlated than low-priced and high-priced grains by themselves. We suspect that this pattern occurred because these two sets of prices may represent two different regions within the prefecture.^[18] Our analysis, in other words, confirms that monthly low-priced and high-priced grains were not always substitutable.

[Full Size]

Fig. 5.3.
Low Grain Prices in Fengtian Prefecture, 1774–1873 (annual averages in taels per shi )

[Full Size]

Fig. 5.4.
High Grain Prices in Fengtian Prefecture, 1774–1873 (annual averages in tales per shi )

Nevertheless, virtually all prices rose and fell usually in tandem and occasionally by as much as a factor of two or three within the space of just one or two years. A recent study by Wang Shaowu of the climate and harvest history of Manchuria indicates that fluctuations in summer temperature were responsible for most of these sharp price variations through their impact on crop yields.^[19] Even as recently as 1954, 1957, 1969, 1972, and 1976, low summer temperatures reduced harvest yields by as much as one third. According to Wang, there were at least 26 similarly cold summers between 1774 and 1873, concentrated in the 1780s, 1810s, and 1830s.^[20] Given the low level of agricultural technology in the late eighteenth and early nineteenth centuries, the impact of these low temperatures on harvests may have been even more severe than in the twentieth century.^[21] Indeed, an examination of Figures 5.3 and 5.4 reveals that these three decades were characterized by rapid price increases. Prices, in other words, appear to accurately reflect the availability of food within Daoyi and should have had great impact on population behavior.

To what extent, then, did grain prices affect vital rates? Figures 5.5 and 5.6 summarize what information we have on annual crude birth and death rates for the century under observation. Again these data suffer from two limitations. On the one hand, we cannot calculate reliable birth or death rates for several years due to missing registers. On the other hand, because of the limitations of the registration system, we can only compute average death rates over three-year periods.^[22] We cannot compute actual annual death rates for either sex. Our mortality statistics therefore appear deceptively stable. Given the incompleteness of the vital data and the volatility of the price data, we should anticipate that the correlations of prices to population will be weaker on paper than they were in reality.

[Full Size]

Fig. 5.5.
Crude Birth Rates in Daoyi, 1774–1864 (per 1,000 married women aged 15–45)

[Full Size]

Fig. 5.6.
Crude Death Rates in Daoyi, 1771–1873 (per 1,000 population, by intercensal period)

Nevertheless, in spite of these limitations, the comparison of annual birth and death rates with annual average grain prices yields a number of truly significant results. In keeping with our previous analysis of mortality, the correlations between food prices and death rates are far stronger for men than for women. Indeed, as we can see from Table 5.7, there are no significant correlations between food prices and female death rates. The only meaningful relationships we can find are between male death rates and the annual average of monthly low prices. There are no correlations between the annual averages of monthly high prices with mortality. From these findings we can infer that the monthly low prices for the prefecture somehow reflect the availability of food in Daoyi during subsistence crises. We can also identify which of the five reported grains were the most important subsistence

crops for our population. Rice prices, for example, were uncorrelated with mortality and can be considered relatively unimportant. All other grains had significant if weak correlations with mortality. Wheat was the most important of these food crops.

However, in keeping with our analysis of fertility and family planning, correlations between grain prices and birth rates are not only far stronger than correlations between grain prices and death rates, they are also stronger for females than for males, especially in complex households. Strong negative correlations exist for virtually all grain prices regardless of the type of price (monthly low or monthly high) or the variety of grain. Figure 5.7, which plots birth rates by sex against low millet prices—millet being a common food staple—graphically illustrates the relationship between prices and fertility. When food prices were high, people had fewer children, especially fewer girls.

Fetal wastage and standard methods of family planning would have produced dramatically different results. Spontaneous abortions would have affected males as well as females.^[23] Contraception would have produced strong correlations with lagged prices as well as current prices. In fact, though, correlations between prices and fertility are not only stronger for females than for males, they are stronger for current prices than for lagged prices.^[24] Parents, in other words, made their fertility decisions in response to conditions at time of birth rather than conditions at time of conception. The unnatural response of birth rates by sex to immediate economic conditions, therefore, strongly suggests that in Daoyi many peasants limited their fertility through sex-selective neglect or infanticide.

Parents, of course, were most likely to make such drastic decisions in response to extreme economic conditions, only adjusting their fertility when prices were exceedingly high or exceedingly low. As a result, the correlations between birth rates and food prices in Table 5.7 may not accurately reflect the full responsiveness of fertility to economic conditions. We therefore calculate the percent changes in birth rates between years of medium prices (within one standard deviation of the mean) to years of unusually high or low prices (above or below one standard deviation) in order to measure the sensitivity of birth rates to high and low price extremes. According to these calculations, summarized in Figures 5.8 through 5.11, virtually all households

[Full Size]

Fig. 5.7.
Birth Rates by Sex in Daoyi as a Function of Low Millet Prices, 1775–1840 (per 1,000 population)
Note: Male correlation coefficient = -0.59. Female correlation coefficient = -0.69

responded to high prices by reducing fertility and to low prices by increasing fertility.

Household wealth, and, by extension, household type, played a key role in the decision to let a daughter survive. Wealthy households, which were generally complex, were less affected by economic conditions than were poorer households, which were usually simple. Indeed, as we can see in Figure 5.8, when prices were high, parents in complex households reduced their female birth rate by one quarter, while parents in simple households cut theirs in half. In contrast, when prices were low, parents in complex households increased their female birth rate by one third, while parents in simple households did not change theirs at all. Simple households, therefore, were so impoverished relative to complex households that they not only kept far fewer girls than normal when times were bad but they allowed no extra ones to live when times were good.

The decision to keep a daughter depended, of course, on whether or not the couple already had a child. Indeed as we saw in Table 5.5, a girl was most likely to be kept if she was the first of several children. According to Figure 5.8, however, this was especially true in complex households. When times were bad, the birth rate for girls without older siblings dropped by less than one tenth while the rates for girls with older siblings dropped by four-tenths. In contrast, when times were good, the rate for girls without older siblings almost doubled while the rates for girls with older siblings went up only by one tenth. That complex-household parents were willing in bad times as well as good times to support a daughter so long as she was their firstborn reflects a desire to keep first children regardless of their sex. In spite of the widespread practice of infanticide, in other words, primordial affection could triumph over material concerns.

A girl's chances of surviving in a complex household, however, depended on household position as well as birth order. Figure 5.9 reveals that the less incentive parents had to produce a male heir, the more girls they allowed to live when prices were low. In good years, therefore, brothers and nephews, who were the furthest from the line of inheritance, had 50 percent more girls than usual. Heads, however, had only as many girls as they did in normal times. In contrast, all parents, whatever their relationship to the head, had at least one-third fewer girls in bad times than normal. Even in complex households, while some parents were willing to support more female children than normal in times of plenty, girls were still enough of a luxury that few parents were willing to save them when times were hard.

But baby girls were by no means the only victims of infanticide. Some parents also neglected or perhaps even killed their sons as well as their daughters when times were bad. According to the correlations for male births in Table 5.7, this was especially true in the simple, that is, poorer, households. Male birth rates had uniformly strong negative correlations with all

[Full Size]

Fig. 5.8.
Female Births by Household Type in Daoyi in Periods of High and Low Grain Prices, 1792–1840 (% change from normal rate)

Notes: Marital fertility only. For boys, periods of high and low grain prices are years where the price is more than one standard deviation
above or below the mean for the period 1792–1873. For girls, periods of high and low grain prices are years where the price is more than one
standard deviation above or below the mean for the period 1792–1840. Different years were selected for boys and girls because the mean
and standard deviation of grain prices for the period 1792–1873 were different from the mean and standard deviation for the period
1792–1840. The years of high prices were therefore 1812–1816 for girls and 1807–1817, 1823–1827, 1829, 1833, and 1836–1838
for boys. The years of low prices were 1795–1802 and 1805 for girls and 1797–1801 and 1854–1862 for boys.

[Full Size]

Fig. 5.9.
Female Births by Family Relationship in Daoyi in Periods of High and Low Grain Prices, 1792–1840 (% change from normal rate)

Note: See notes in Fig. 5.8.

food prices. Indeed, the correlations with male fertility were almost as strong as the correlations with female fertility. When prices were high, in other words, parents in these households neglected their sons as well as their daughters. In Fengtian, at least, the common assumption that Chinese parents neglected only their daughters may be untrue.^[25]

Even in simple households, of course, parents only made the decision not to keep a son under extreme economic pressure. As we can see in Figure 5.10, when prices were high, the male birth rate in simple households declined by one third, but when prices were low, it increased by only one sixteenth. Parents, in other words, did not regard their boys as luxury goods. They thought of them as necessities. As a result, male birth rates remained relatively constant so long as there was no crisis. Parents in simple households only gave up their sons when they did not have the resources to support them.

Male infanticide, however, was not restricted to simple households. Figure 5.11 indicates that in complex households parents at the very top and very bottom of the household hierarchy also reduced their male birth rate in response to economic pressure. Indeed, when prices were high, cousins, who were at the bottom of the hierarchy, only allowed half as many boys to live as when prices were normal. Similarly, heads who were at the top of the household hierarchy had one quarter fewer boys. Evidently, just as men paid for their privileged position relative to women with greater increases in mortality in times of crisis, so household heads paid for their privileged position by lowering their fertility to the levels of other relations. Here too the price of privilege was greater vulnerability to economic fluctuation.

Just as with girls, birth order was an important consideration for boys in complex households, but the response to economic conditions differed considerably. The birth rates for firstborn sons, unlike those for their female counterparts, were more responsive to high prices than the birth rates for later sons. Thus when prices were high, the rate for firstborn sons declined by one fifth while the rate for later-born boys changed hardly at all (see Figure 5.10). This was especially true for the children of co-resident sons, brothers, and nephews, who were neither at the top nor at the bottom of the household hierarchy. Mid-ranking couples apparently had extra boys only if they were wealthy enough to be relatively immune to economic pressure. Heads, on the other hand, were under constant pressure to produce more sons regardless of wealth. It was poorer heads who responded to economic crisis by delaying or canceling plans for extra sons.

[Full Size]

Fig. 5.10.
Male Births by Household Type in Daoyi in Periods of High and Low Grain Prices, 1792–1873 (% change from normal rate)

Note: See notes in Fig. 5.8.

[Full Size]

Figure. 5.11.
Male Births by Family Relationship in Daoyi in Periods of High and Low Grain Prices, 1792–1873 (% change from normal rate)

Note: See notes in Fig. 5.8.

But while male birth rates were not as responsive as female birth rates to changes in food prices, we should remember that male death rates were more strongly correlated with prices than female death rates. Girls, in other words, were apparently considered such luxuries that, like wives, they were permitted only when parents were confident that they could support them, albeit at a very low level. Boys, on the other hand, were valued so highly that in even the most marginal of household circumstances they were often allowed to live. Thus while females who survived infancy were unaffected by price fluctuations, males were vulnerable even as adults precisely because they were allowed to survive in poor as well as wealthy households.

Until recently virtually the only attempt at precise demographic measures of a "traditional" rural Chinese population has been the reexamination by G. W. Barclay, A. J. Coale, M. A. Stoto, and T. J. Trussell of the data collected by John Lossing Buck in his 1929–31 survey of Chinese agriculture.^[26] Relying very heavily on a body of indirect techniques of demographic estimation, these distinguished demographers concluded that in China, although marriage was both early and universal, fertility was nevertheless extremely low. They furthermore suggested that such low fertility may well have been characteristic of the late imperial period. The problem that then confronted historians of late imperial China was how to explain the sustained rapid rise of Qing population in the face of such low "natural" fertility. The present study, in relying on direct analysis of eighteenth and nineteenth century materials, suggests that such low fertility may well have been the result of fertility control and that the level of "natural" fertility in eighteenth- and nineteenth-century China may in fact have been higher.

This study further suggests that if all Chinese peasants controlled their fertility in response to economic conditions, then the rise in population during the eighteenth and nineteenth centuries may well have been a direct response to significant advances in economic growth. Our findings, in other words, seem to corroborate recent claims by Western as well as Chinese historians of unprecedented agrarian expansion beginning in the eighteenth century and increasingly rapid commercialization beginning in the nineteenth century.^[27]

Finally, this study provides some insights into the complexity of household decision making in eighteenth- and nineteenth-century Daoyi and perhaps by extension (a very wide one to be sure) in China at large. Peasants based their decisions about infanticide on a complex combination of interrelated factors, including economic conditions, household type, position within the household, number and sex of previous children, sex of child, and long-term goals for family size and composition. Food prices were an important factor in this decision-making process, but the equation differed considerably for each individual depending on the other variables. The relative importance of these factors has yet to be determined. We can already appreciate that the study of prices and population ultimately leads to a better understanding of Qing society as well as the Qing economy.