Food

Basic caloric intake is one of the areas where we can be most confident that Chinese were no worse off than their European counterparts. Using data on the diets of landless agricultural labourers contained in seventeenth-century agricultural treatises—which refer mostly to conditions in the advanced Yangzi Delta—Ming-te Pan calculates that they represent a daily diet of 4,700 calories during the working months.’ Overall figures are of course lower, but estimates of per capita grain consumption in the eighteenth century average about 2.2 shi of rice equivalent per person (including both sexes and all ages).² If the estimate of Chinese grain output c.1753 by Guo Songyi is even close to accurate, this is a very substantial underestimate,³ and the large rations and heavy labour of Pan's farm workers were actually closer to the mark.

This converts to ’,837 calories per person from rice alone. If the age structure of the population was about the same as in the ’930s,⁴ this would work out to 2,386 calories per adult equivalent of grain alone. This would compare favourably with the various estimates for workers in late eighteenth- and nineteenth-century England (the richest part of Europe) cited by Clark, Huberman and Lindert, which range from ’,500 to 2,400 calories per person from all foods, and would be close to most of the figures they later cite for more prosperous periods (Clark, Huberman, and Lindert ’995: 223—6).

given for various parts of continental Europe: most figures I have found for eighteenth-century France, for instance, are between 1,800 and 2,500 calories per day. In the southern Netherlands (Belgium) r.1800, the average food intake is estimated at between 2,180 and 2,440 calories per capita per day; the poor fared worse (Lis and Soly 1979: 182). J. C. Toutain estimated average caloric intake in France at only 1,800 calories per capita per day in the eighteenth century.⁵ R. J. Bernard collected twelve ‘food pensions' from one region, which were supposed to provide for all the food needs of the recipient: only two of the twelve diets in question reached the Chinese average of caloric intake from rice alone, and eight out of twelve were below 2,000 calories per day (Bernard 1975: 35, 39).

Estimates of per capita grain consumption, relief rations, and so on, do not seem to vary much between richer and poorer parts of China, though these figures are admittedly sparse and not always reliable. The daily ration for convicts in prison or on their way to exile in central Asia was roughly 2,800 calories per day of grain (plus a small allowance for other food) before 1775, with the same ration being given (oddly enough) to men, women, and children. After 1775, it was reduced to roughly 2,333 calories per adult (Waley Cohen 1991: 118—19) which still compares favourably with the diet of the English poor at the end of the century. The famine relief ration distributed in kind r.1740 was much less than this—a mere 1,400 calories worth, which would barely support survival—but since ‘the usual rule' was to distribute half of the relief in kind and the other half in cash, these rations too probably averaged around 2,800 calories for adults in the eighteenth century (Will 1990: 132—3).

The stability of the basic grain supply, however, did vary enormously by region. Except for a period of flooding in the 1830s and 1840s (of which more later), the Yangzi Delta suffered relatively few natural catastrophes, and its position on the coast and at the mouth of a river system which drained close to one-third of China meant that though the area always relied on imports for some of its food, widespread famine was virtually unknown there, between the catastrophes at the end of the Ming (1644) and during the Taiping Rebellion (1851—64). Famine also appears to have been rare in the rice bowls of the middle and upper Yangzi, the southern Manchurian frontier, and Taiwan. These were areas that generally produced large surpluses, and both merchants and the state regularly moved grain out of these regions. However, there was nothing quite comparable to the absolute priority accorded to Paris, Madrid, and other early modern European capitals, which forced continued exports even during harvest shortfalls, and thus made what were normally grain-surplus regions among the places most at risk of famine.⁶ Much of north and northwest China, on the other hand, suffered from China's least reliable rainfall, its most disaster-prone rivers, and relatively poor transport facilities; in north China these conditions were complicated by dense population. Though it was in these areas that the state mounted its most impressive efforts at food supply stabilization and famine relief, they never achieved the same degree of food security as either the Yangzi Delta or some of the grain-exporting areas; and, as we shall see, the state's efforts to alleviate this problem became far less reliable in the nineteenth century.

Europe, of course, also had a wide range of famine vulnerability by region. England (though not all of Britain) was essentially free from widespread famine from mid-seventeenth century onwards, much like the Yangzi Delta, and some other parts of north and northwest Europe were not far behind.

2.1 Beyond basic calories

For non-grain foods, comparisons are harder to make, but when we can do so, eighteenth-century China fares reasonably well. Chinese meat and (especially) dairy consumption levels were surely lower overall than European ones, given far lower ratios of livestock to population. But protein intake was nonetheless probably adequate for most people, and in some cases probably exceeded what was available to ordinary Europeans: any overall European edge is probably accounted for by very high protein consumption among a relatively small group of well-to-do people. It also seems likely that Chinese protein consumption was higher in the mid-eighteenth century than it was in the early twentieth century according to the data compiled by J. L. Buck. Unfortunately, Chinese sources consistently refer to ‘a piece' of meat, fish, or beancurd without specifying its size. However, Pan Ming-te has made plausible (and probably conservative) estimates of the size of portions referred to in the discussion of workers' diets in the Shenshi nongshu, a widely used and often reprinted seventeenth-century agricultural manual. Using Pan's estimates and the data in the manual itself (read in the most cautious way possible) I arrive at an estimate of roughly 22 g of animal protein per day (over the course of the year) and 7 g of protein from beancurd/day, to give a total of 29 g per day of actual protein from these relatively protein-rich foods.

comparability with the other studies. I will refer to it as the ‘protein portion', rather than ‘actual protein intake' to keep the two distinct. Using the data and calculations described above, the ‘protein portion' would be 124 g per day for seventeenth-century Yangzi Delta farm workers, even before we add any protein content in the rice they ate.

These figures are higher than one might expect and, as we shall see, they compare rather well with both eighteenth­century European and twentieth-century Chinese data. It is also worth noting that while rice provides very little protein per pound or calorie of consumption, what protein there is, is of high quality. Thus in a very high-calorie diet, such as that of the farm labourers described above, rice protein alone might meet the person's needs, at least according to some authorities; it would have added about 44 g of protein per day to the diet of somebody consuming the conservatively estimated ‘average adult rice ration' for China as a whole, and an impressive 85 g per day for labourers eating the diet in Shenshi nongshui Thus actual protein intake would range between 73-114 g per day, and the less accurate ‘protein portion' might be estimated as 168-209 g per day, depending on which estimate of rice consumption one uses.

Consequently, the situation would seem to compare quite favourably with John L. Buck's estimates of anywhere from 57 to 148 g of ‘protein portions' from all sources per adult male equivalent in six mostly poor counties in the 1920s (Buck 1930/1971: 374) with averages ranging from 82 to 117 g per adult male equivalent (again including all sources) in each of the eight regions into which he divided his large 1937 study of China (Buck 1937/1964: 419). The actual protein intake would of course be smaller than the ‘protein portion', but once the protein content of grain consumed was added in (giving us the 73-114 g band), the result would meet the US government's recommended daily allowance of 66 g per day of actual protein for an adult male, or the international minimum standard of an equivalent of 37 g per day of actual egg or milk protein for a 143 pound adult male.⁹ As a percentage of total calories, protein intake would be somewhere below the recommended 10-15% if the person in question was consuming the enormous quantities of grain cited in Shenshi nongshu for agricultural labourers in summer and fall, and within or slightly above that band if they were eating something closer to 2,500 or even 3,000 calories of grain.

Moreover, this was not the extent of protein consumption, even for these very poor workers. This is made clear by the absence of poultry and eggs from the diets in the agricultural manuals. Even agricultural labourers typically had access to a small plot for raising vegetables and chickens; in fact, in many parts of the country it was still customary for farm labourers to make ceremonial presents of eggs to their employers on certain holidays. (The agricultural manuals probably omit them because they are concerned with telling landowners what they need to provide for their labourers.) Interestingly, Buck found in the twentieth century that though large numbers of eggs were produced on Chinese farms, only a very small percentage of them were consumed on the farm. His 1930 survey suggested that the average farm family had about 4 chickens, but his 1937 study suggested that they consumed

only 40 eggs (per 5-member person family) per year, out of perhaps 200 produced (Buck 1930/1971: 218, 1937/1964: 258 n. 12, 411). The vast majority of eggs, Buck argued, were too valuable as a source of cash to be eaten by the producers; more and more were being exported to raise cash (Buck 1937/1964: 430). In the seventeenth and eighteenth centuries, exports of eggs were clearly not a factor, and it seems likely that most farm families had less of a need to raise cash than their warlord-era descendants: taxes were lower, fuel that one could gather was more plentiful, and probably more household goods were self-produced. Under the circumstances, it seems likely that a much larger share of eggs produced were actually consumed on the farm in earlier days, though for now this must remain speculative.

There seems little doubt, then, that, on average, actual protein intake was quantitatively sufficient—as was even the generally lower intake noted by Buck in most regions of China in the 1930s. It should be remembered, though, that such averages do not reflect various scenarios that would have had important health implications: for instance, the substantial possibility that children or pregnant women may have been deprived of adequate protein while all the meat in the diet went to men working in the fields. Moreover, it is harder to be sure of the quality than the quantity. In Buck's twentieth-century survey, most of the protein came from cereal and vegetable sources, as all animal products (including eggs) provided an average of only 77 calories per adult male equivalent per day—much less than the estimate for seventeenth-century farm labourers derived here (Buck 1937/1964: 407, 411). And while, as noted above, rice protein is of fairly high quality (but limited quantity, unless one eats a lot), this is not true for wheat or for some of the other grains consumed by poorer people, especially in the north. Under these circumstances—and using ideas about the low quality of vegetable protein that are now questioned, especially of rice—Buck and other observers had doubts about whether the quality of the protein in the average Chinese diet was sufficient for certain groups, such as growing children (Buck 1937/1964: 418). If the figures adapted from Shenshi nongshu are even close to being representative, and if recent science that has upgraded our impression of the adequacy of vegetable proteins is accurate, this had probably not been a widespread problem in the Yangzi Delta over the last 200 years. Unfortunately, we lack the data to make even these rough estimates for other parts of seventeenth- or eighteenth-century China.

Is it plausible that protein consumption was a good deal higher in the seventeenth and eighteenth centuries than in the early twentieth century? In his famous work on Chinese agriculture 1368—1968, Dwight Perkins argues that the ratio of hogs (by far the most important meat source) to human population had been roughly constant from the 1360s to 1957—though he admitted that he had very little evidence for this crucial assumption (Perkins 1969: 71—3). However, Perkins' own supply and demand schedule for hogs suggests that (1) once hogs become sufficiently numerous that they cannot be fed entirely on chaff and garbage, the cost of raising them rises rapidly, so that more hogs can be raised only if demand is strong enough to sustain

Figure 1.1 Supply and demand for hogs (reproduced from Perkins 1969: 72)

Number of hogs raised

S: Supply schedule for hogs—the cost of raising hogs rises steeply when farmers switch from feeding hogs chaff to feeding them grain. The cost of grain rises more and more rapidly as farmers cut increasingly into their own (or the nation's) grain consumption (solid line). D': Demand schedule for hogs where hogs are only used as the source of pork (dashed line). D: Demand schedule for hogs where hogs supply both pork and fertilizer (solid line).

sharply rising prices, (2) the demand schedule for hogs shifts sharply to the right insofar as hogs are an important source of fertilizer as well as meat. Thus, the number of hogs in an area such as the lower Yangzi (in which cultivated acreage in the late Ming was already close to its twentieth-century maximum) should be very sensitive to the demand for pig manure (Figure 1.1).

Further, beginning in the late seventeenth century, more and more Yangzi Delta farmers began to apply large amounts of imported soybean cake to their fields—a fertilizer that to a great extent replaced, rather than supplemented night soil.¹⁰ It thus seems likely that the amount of hog-derived fertilizer being used by each farmer would have declined over time as soybean cake fertilizer spread, and with it the number of (and demand for) pigs relative to farmers: a reasonable proxy, in this economy, for the availability of pork per capita. By the 1930s, the middle and lower Yangzi (grouped together in Buck's Yangzi Rice/Wheat' area) used less than two-thirds the quantity of animal manure and night soil per acre as did China as a whole (even though they had far more humans per acre), and more than double the amount of other fertilizers per acre.¹¹ Perkins himself also notes that a late Ming source gives an estimate of manure use per acre in Jiaxing prefecture (in the Yangzi Delta) that is roughly equal to Buck's estimate for south China as a whole in the 1930s (Perkins 1969: 73). This figure is, in turn, 87% higher than the 1930s figures for Buck's ‘Yangzi Rice/Wheat' region—the region which includes Jiaxing, though it is far larger, including most of the lower and middle Yangzi (Buck 1937/1964: 259). This would seem to confirm the guess that this area had had far more pigs per crop acre in the late Ming than it had by the 1930s: a period when it also had

considerably fewer humans than it would have had by the mid-eighteenth century, or the early twentieth century.’² It thus seems more likely that our relatively high estimates for seventeenth-century protein intake estimates are roughly accurate and that Perkins was right to guess that meat consumption was unchanged over centuries—though we certainly need more evidence to prove either case.’³ It would then seem likely that protein availability had already declined significantly by the late eighteenth century—as population grew and easier access to soybean cake cut into demand for hog wastes—but was probably not yet down to the levels observed in early twentieth-century Jiangnan. Li Bozhong has also provided evidence that cows and oxen, rare in twentieth-century Jiangnan agriculture, were somewhat more plentiful in the Ming and early Qing (Li 2000: 275—6, 278—80). These were work animals and would have made only a very small additional contribution to protein intake, but the dynamics causing their numbers to decline may have been similar.

If per capita meat consumption in particular was indeed declining in late imperial China, this would fit the long-term trend in early modern Europe as well: though in relatively land-rich, livestock-using Europe this decline began from a much higher base. Braudel estimates that meat consumption in Germany fell by 80% between the late Middle Ages and ’800 (Braudel ’98’: ’96). In Toulouse, the average consumption of beef fell from ’5 k per capita in ’655—9, to 7 kg in ’750 (though there was also some consumption of other meats).’⁴ The latter figure would work out to 20 g per day of meat, or barely 4 g per day of actual protein from this source—considerably less than our figures, though missing data make any systematic comparison impossible.

Indeed, the general picture of protein supply in late eighteenth-century western Europe seems no better, and often worse, than these Chinese figures—though the way that some historians calculate it complicates comparisons. Lis and Soly's estimate of total protein intake per capita in the ‘average’ southern Netherlands diet r.’800—73.2 g per person per day, of which only 22.5 g was a ‘protein portion’ from animal sources (Lis and Soly ’979: ’82)—is near the bottom of Buck's above-cited range of figures for twentieth-century rural China, and much lower than the estimates we have made for the seventeenth and eighteenth centuries. Actual protein intake in such a diet would be under 60 g per day. Of course, ’800—in the midst of the Napoleonic Wars—was a particularly bad year, but the comparison is nonetheless suggestive. Toutain's figures for Napoleonic France, which suggest about 200 kcal/day of protein—only about 50 kcal of it came from animal protein’⁵—would convert to different amounts of protein in grams depending on precisely what foods were involved. The weight of the protein-rich food however, that is, the ‘protein portion’, is unlikely to have been above 50 g per day, and the quantity of protein actually obtained at the most was ’0 g/day. Even if we add protein from the rest of the diet, we would have a total protein intake of perhaps 70 g—matching the very bottom of our range of Chinese estimates.’⁶ Bernard's figures from Gevaudan range, with one exception, from 32 to 75 g of actual protein per day, including protein obtained from cereals; the one exception is 89 g

(Bernard 1975: 37 n. 1, 38). For Britain, D. J. Oddy estimates 49 g of protein per capita per day for 1787—93, using family budgets for rural labourers—a figure which appears to be for actual intake, not ‘protein portions'. (Working backwards from his data to get an estimated ‘protein portion' for purposes of comparability, I got about 75 g per person per day for those years.) This rises to 62 g in a 1796 sample, but does not exceed 66 g until a 1937 sample.¹⁷ For vegetables (and thus many vitamins), it seems very likely that comparisons would favour China, since early modern Europeans tended to eat very few vegetables. Chinese cooking methods also preserved the vitamins in vegetables better than European ones. Again working from Shenshi nongshu, Pan estimates that the average daily ration of vegetables (again, excepting anything grown in one's own garden) would be about 200 g/day (Pan 1998: 17). By contrast, Buck's figures for the 1920s (in this case from one county in Anhui) comes to 190.3 k/year of various vegetables for families with an average of 4.88 adult male equivalents.¹⁸ This works out to only 107 g/day per person—barely half of the estimated seventeenth-century level. But consumption estimates for fruits and vegetables are likely to be particularly unreliable, both in China and in Europe, because much of this consumption did not pass through the market: even today, such estimates are almost always lower than reality (Aymard 1979: 2).

Overall, then, the food component of the standard of living seems generally comparable in eighteenth-century China and Europe, and in the most advanced regions of each. Admittedly, for the poorer and more poorly documented parts of China we must rely in part on the assumption that non-grain food consumption was no lower in the late eighteenth century than in the early twentieth century. Nevertheless, as I will try to show near the end of this chapter, it is precisely those regions for which such a claim is most compelling. Of course, a great deal of information is still missing but from what we do know, there are as many reasons to give China an edge over Europe as vice versa.

At least comparable nutritional levels are also suggested by the fact that eighteenth-century rural Chinese life expectancies—between thirty and forty in most available studies—are quite comparable to those for eighteenth­century England, and higher than those found in most studies of continental European populations.¹⁹ Moreover, since recent studies suggest that Chinese birth rates were below European ones throughout the 1550—1850 period,²⁰ while the overall rate of population growth was first faster (1550—1750) and then similar (1750—1850),²¹ we have a further indication that Chinese death rates could not have been significantly higher than those for even the wealthier countries of Europe in the late eighteenth century.

Moreover, Chinese appear to have reached these nutritional standards without spending any more of their incomes on basic foodstuffs than did their European counterparts. Fang Xing's study of Yangzi Delta farm labourers suggests that they spent 55% of their cash and kind earnings on basic grain supplies in the seventeenth century, and almost the same amount (54%) in the early nineteenth century (Fang 1996: 93, 95). This is almost exactly the same as the figure for rural poor

in the 1790s cited by Phelps Brown and Hopkins, and since agricultural labourers were the poorest non-beggars in China, the comparison seems apt (Phelps Brown and Hopkins 1981: 14). Moreover, Fang's method of calculation seems more likely to understate than overstate the consumption of these labourers (Pomeranz 2000a: 137, n. 110). Data compiled by Carole Shammas suggests that in comparison with at least southern England the results may have remained quite close until much later. She estimates that agricultural labourers spent 72% of their incomes on food in 1837—8, and finds figures fluctuating between 50% and 70% for various samples (not all of them from such poor people) until 1900 (Shammas 1990: 124—6). She also estimates that 66% of food expenditures went for cereals in south England.²² Assuming that farm labourers were able to spend as much of their food budget on non-cereals as other labourers in south England (which seems optimistic, since they generally earned much less), they would have been spending 48% of their income on grain: just very slightly below Fang's lower Yangzi estimate.

Moreover, Fang does not calculate peasant incomes directly. Instead he relies on accounts of the minimum expenditures needed for clothing, firewood, and so on (again, mostly in agricultural manuals). Thus he excludes various other kinds of expenditures: for example, occasional very large expenditures for life-cycle rituals; jewellery, which even poor women seem to have had some of (Pan 1994: 85); clothes for special occasions;²³ entertainment, and so on. Overall, then, it is much more likely that Fang has underestimated what even farm labourers could spend on things beyond basic calories than that he has overestimated it: and yet his estimates seem comparable to those for the English poor.

For the early twentieth century, we again have data from J. L. Buck, which suggests that peasants in six counties of ‘East Central China' spent 53.8% of their income on food (Buck 1937/1964: 386)—and therefore presumably less than that on basic cereals. But these were farm families of all sorts, not just landless labourers, and Buck's sample has been criticized for being skewed towards more prosperous families. It thus seems likely that by this indicator, too, the Chinese poor of the eighteenth century were somewhat better off than their early twentieth-century counterparts.

3.