POVERTY AND FAM'LY SIZE 23 can contribute to the family income either directly by doing paid work or indirectly by attending to unpaid duties and thereby releasing adults for wage employment.3 This is a short-term calculation; in the long term, a large supply of labour power within (he family makes it possible for the family to generate savings, acquire land and move up the social scale.3 But the factual basis for the belief that the poor have, in general, high fertility levels and large families seems to have never been systematically investigated. One reason why the belief is deeply ingrained in academic thought is that poor countries have, in recent decades, experienced high rates of population growth resulting from high, and sometimes increasing, fertility and declining mortality, whereas low levels of fertility prevailing ,in the rich countries are the outcome of a historical process in which the decline in fertility is generally associated with rising income. Theories of "demographic transition" which seek to summarize and explain this and other generalizations based on the experience of Western Europe do not concern us here.4 What we need to note, however, is that they are not quite relevant to the understanding of possible demographic differentials between the "poor" and the "rich" within the poor countries. Is it true that, in general, the levels of fertility and family size arc higher among the poor than among the rich? This is the question with which we arc concerned here. We note that the question cannot be answered without an understanding of what constitutes "poverty"; for this reason we shall be concerned also with alternative ways of identifying the poor and the rich. We shall restrict ourselves to rural areas of India. The paper is based on research in progress and hence its conclusions arc tentative in nature. HOUSEHOLD SI^E AMD PER CAPITA EXPENDITURE The only data which appear to support an inverse relationship between household size and levels of living are those based on the National Sample Survey (NSS) which show that as the monthly per capita expenditure increases the average household size declines.6 This inverse relationship holds for all rounds of the NSS and for all regions of the country for which such data are available. Table I which gives the nineteenth round (1964-65) data in respect of rural India shows, for example, that the average household size declines from 5.86 in the lowest per capita expenditure group to 3.48 in the highest.