In the previous chapter we examined models of the influence of fertility and mortality on the size of households. Two major (and quite deliberate) simplifying assumptions underlying all these models are (i) that fertility and mortality are exogenous to the household and (ii) that all households in a community are exposed to exactly the same demographic conditions. This assumption is patently false, even if it is convenient for some analytical purposes. Starting in the present chapter, we reverse the causality, asking whether mortality, fertility and migration differ among households owing to material conditions (especially those influencing food availability) that are peculiar to each individual household. Do all households in a community have the same level of, for example, early childhood mortality – and, if not, why not? I will argue that an important part of the answer is that feedbacks operate within the household involving food production and the household’s demographic life-cycle, and these feedbacks are powerful enough on their own to differentiate households even in the absence of clear social and political differences dividing the community as a whole. The critical linkage in this feedback, I believe, involves the relationship between dietary adequacy and various aspects of human physiology. If I am right about the importance of this relationship, it would point to a basic (and perhaps ever-present) form of population regulation operating at the level of the subsistence farming household. But before we can examine this claim, we need to review what is currently known about the relationship between dietary adequacy and the basic forces of demography – fertility, mortality and migration. And before we do that, we need to think more carefully about what we mean by “dietary adequacy” and how to measure it. This chapter does both. Because little of the evidence I marshal in this chapter is explicitly organized at the household level, households will play a somewhat covert role in this treatment. But, while contemplating the relationship between under-nutrition and, say, fertility or mortality, the reader should really be thinking about the role that under-nutrition plays as a potential impediment to the household demographic enterprise. Subsequent chapters return the household to center stage.

In discussing Richard Longhurst’s model of the energy trap in the previous chapter, we speculated that differences among households in nutritional status could be amplified, via their effects on work capacity, into semipermanent, inter-generational differences in a family’s material well-being. But the energy trap, if it is to operate at all, must act on preexisting differences in household well-being. Thus far, we have said nothing about where those initial differences come from. A large part of the answer, of course, is that traditional farmers, like all people, vary among themselves in such things as industriousness, cleverness, prudence, fecklessness, even sheer dumb luck – not to mention the political nous to turn a temporary advantage into something more enduring. In addition, environmental unpredictability can be a potent source of differential household success, even in communities that appear to us to be egalitarian. In this chapter, however, we focus on one possible source of economic differentiation that is inherent in the household demographic enterprise – the household’s built-in demographic life cycle, the ebb and flow of household size and age–sex composition resulting from births, deaths, marriages and other forms of inter-household migration (see Chapter 7). More specifically, inter-household differentiation arises from two facts: (i) that the life cycles of all the households in a community are never in perfect synchrony, and (ii) that an element of randomness always plays a part in every household’s life cycle. No two life cycles are ever quite the same in their intensity and timing – and these dissimilarities can have important economic consequences.

Traditional farming – farming in the absence of fossil fuels, electricity, commercial seed, tractors and combines and other industrial inputs such as inorganic fertilizers, pesticides and herbicides – has sustained much of the human species for ten millennia. And not just sustained: the global emergence of farming led to a thousand-fold increase in the size of the human population by the beginning of the Industrial Revolution in the late eighteenth century (Cohen, 1995: 96). Traditional farming provided the foundation for early civilizations, cities and states, all of which evolved along with it. By some estimates, traditional farming or something very like it was still feeding a third of the world’s people in the second half of the twentieth century (Haswell, 1973).

In every preindustrial farming operation, some of the most important and insurmountable limits to a household’s food supply are imposed by the size and demographic makeup of the work force it can muster to produce it. Apart from the energy for photosynthesis provided by the Sun, the energy for food production in subsistence farming is almost entirely biological in nature – mostly human and secondarily animal (in some places), with wind or water power very occasionally playing a minor role. Almost universally, the labor needed for household food production is organized by the household itself and is indeed made up overwhelmingly of its own members. Granted, as will be discussed in more detail in Chapter 12, voluntary work groups that draw in members of several neighboring households may be mobilized at certain times of year for particularly heavy tasks that need to be done quickly, often in return for some kind of payment in kind (perhaps a part of the harvest, a feast, or a beer-drinking party) or a promise of reciprocation; in addition, individual labor contributions by close relatives living in nearby households may be requested on a day-to-day basis. In this chapter and the next, however, I shall make the model assumption – one of those tactical “lies” discussed in Chapter 1 – that labor on the household farm is provided exclusively by the household itself. I do this to show some of the built-in limitations to household-based food production, limitations that require special social arrangements – the expenditure of “social capital” – if they are to be overcome. The idea that households are wholly on their own when it comes to farm labor may be a tactical lie, but the truth is that most labor in traditional farming is provided by the household that runs the farm (Figure 9.1). The subsistence-oriented household provides by far the largest contribution of nonsolar energy for its own food production.

In 1975). Keyfitz suggested that few of the most firmly established facts of demography are based on observations alone – reality is far too messy for that. Instead, it is the interplay of theoretical models and observations that underpins what we know about population – and sometimes it is theory alone that provides the most reliable pointers to the nature of reality. As Keyfitz famously put it, “no model, no understanding” – perhaps the most fundamental of all the facts of demography.

A major goal of this book has been to reframe the scientific debate over the relationship between population and preindustrial farming. One way to see how successful I have been in meeting that goal is to turn back to the section on “Rethinking the relationship between human population and traditional agriculture” in Chapter 1, which posed a series of questions that needed to be answered before further progress could be made. It is now time to revisit those questions in order to judge how far, if at all, we have advanced toward answering them.

In this chapter I hope to put some real-world flesh on the bare theoretical bones of Malthus and Boserup, especially the latter. Although Malthus and Boserup both drew upon empirical evidence in their writings, neither did the sort of long-term fieldwork on traditional farming that would satisfy a modern-day anthropologist. I have no wish or warrant to disparage their efforts at theoretical modeling – but the comparison of model to reality is also important, not only to test the model for possible rejection but to suggest ways in which it might be improved or extended. It is worth emphasizing, however, that there is nothing to be gained by attempting to make any model perfectly realistic, even if that were possible, for to do so would be to make it too complicated to understand and would destroy its generalizability. Even as we complicate our model, we should still seek simplicity and generality: we still want the model to be a model. If complications are to be added, they should be important complications – things that significantly increase our understanding, not just things that improve the fit of the model ex post facto to a particular set of field observations or that merely satisfy an esthetic preference for holism or complexity.

We turn at last to the “classic” debate (as I’ve called it) on population and agriculture in the preindustrial world. In this and the following chapter I summarize the debate as it had evolved through the 1990s. To do justice to the authors whose views I summarize, I need to lay out the logic of the debate as they have understood it. But that poses something of a problem for me, since I think that logic has been to some degree misdirected and confused – on both sides. In particular, there has been little explicit attention paid to what scale of analysis is likely to be most productive in moving the debate forward (see Chapters 1 and 2). Choice of scale is one of the most important decisions involved in designing any empirical research, but the population/agriculture debate has bounced back and forth between scales rather heedlessly. As a crude generality, it might be said that empirical studies have been conducted mostly at the microdemographic scale (that of individual farmers and their farms), whereas theoretical models have been formulated primarily at the macrodemographic scale (the whole population or farming system). Little thought has been given to how this disjunction might confound the comparison of empirical findings and theoretical expectations. I would guess that many empirical researchers have chosen the micro-level of study not for principled reasons but because a single fieldworker or a small team of fieldworkers cannot survey a wide area or a large number of people. Intellectually, the debate has been framed mostly at the macrodemographic level – again not deliberately but merely because that was the level at which the original framers conceived it.

Two themes ran through the previous chapters. One was that subsistence farmers are intelligent, well-trained, inventive and practical-minded, and have numerous tricks for increasing net crop production; the other that their farms provide very low yields and are subject to severe physical constraints, reflecting environments that are finite, risky and indifferent to farmers’ needs. Some readers may find these two themes contradictory: if subsistence farmers are so skillful, why can’t they overcome environmental limits as well as modern farmers do? The answer, explored in detail in this chapter, is that the limits are built into the basic biology of food production. In fact, they can only be overcome, even partially and temporarily, with the massive, costly and energetically inefficient inputs provided to modern farmers by the industrial sector.

Before we can investigate the role of demographic processes in subsistence farming, we need to explore some of the basic features of such farming. As argued in the previous chapter, we can understand the impact of demography only if we can link it to specific mechanisms operating in such farming regimes – which means we need to identify what the relevant mechanisms are from real-world field observations. Most of the empirical evidence concerning subsistence farmers and their farms comes from in-depth studies of living communities in the farther reaches of the rural developing world, studies conducted by anthropologists, geographers, economists, ethnobotanists, ecologists, agricultural scientists and others. Most of this field research has post-dated 1950, which means that many of these purportedly traditional farming operations had already been “contaminated” to some degree by the modern world – by the penetration of markets, the commercialization of food, the rise of wage labor, the exhortations of agricultural extension officers. In this book, I have drawn as far as possible on studies in which the contamination is limited and to some extent “correctable.” Often this has required examining older literature – scarcely a hardship since the older material is often of very high quality. The studies I draw upon are extremely diverse, but the majority are either cross-sectional or of short duration (a few years at most) – a fact that often limits our ability to observe and understand the inherently dynamic nature of farming. Sometimes archaeological and historical reconstructions can provide greater time depth, but they are unable to recover many of the fine details of farming behavior. Even restricting attention to studies of living farmers (living, that is, at the time of study), we find little standardization of research methods, not surprising given the long time span over which these studies were done and the diverse professional backgrounds of the authors involved. The literature is huge, and I do not attempt to review it all; in this chapter I emphasize topics that are important for later chapters. In the appendix to this book, however, there is a bibliographical essay that provides pointers to the larger literature.

In preceding chapters, I have urged repeatedly, first, that the population/agriculture debate needs to be disaggregated to some spatial scale lower than the total population and, second, that the household, the most conspicuous functional group in the everyday working life of all preindustrial farming communities, is the best place to start the disaggregation. Note that I say “start”: other scales may turn out to be useful – perhaps even more useful for some purposes – but the household level would seem, as many researchers have suggested, to be the smallest scale that as a general rule captures the essential processes linking traditional farming and demography (Laslett, 1983; Fricke, 1984; Netting, 1993).