34 SOCIAL SCIENTIST Princeton University Press, 1972, and Nathan Keyfitz, Applied Mathematical Demography, John Wiley, 1977. The data on ag^ distribution are given in NSS report 144 cited in footnote 12. 19 For a study on the factors affecting lactational amenorrhea and interpregnancy interval see, K Prema and others "Lactation and Fertility'*, The American Journal of Clinical Nutrition, 32, June 1979, pp 1298-1303, The authors say, "Our data indicated that, fortunately, prolonged lactation is the rule among urban low income groups. However, the educationally and economically better off segments of the population tend to wean their babies earlier'"Data presented here show that duration of lactational amenorrhea and inter - pregnancy interval are closely related variables. It is thus possible that decrease in duration of lactation might result in shorter inter-pregnancy interval. . . " Although those findings refer to urban areas they could apply equally well to rural areas. I am grateful to F G Patra for this reference, A large number of studies relating to rural Bangladesh are referred to in the literature on the subject but I have not yet seen them. 20 Tables with Notes on Couple Fertility, Seventeenth round (1961-62), NSS report No 154, p 78. 21 When households are classified according to per capita expenditure the average age at effective marriage(years) does not exhibit much variation (in respect of both husband and wife) over all but the last expenditure group (above Rs 51) in rural India It was roughly 22 years for men and 16 years for women. This observation is based on the seventeenth round data (1961-62); see the NSS report referred to in the last footnote (p 33J. The nuptiality rates for females given by the ratio of the general fertility r3,te to th-; na,ntal fertility rate in Table II do not also exhibit much variation (at 85 to 88 percent except in the lowest expenditure group where it is 82 percent). These data refer to 1964-65. However, an analysis of some small sample data relating to Kerala indicated that when households were classified according to the size of landholdings the proportion of unmarried adults tends to increase as landholding increases. We may interpret this as a phenomenon associated with the prominence of the joint family in the big holdings. 22 T N Krishnan, "Demographic Transition in Kerala, Facts and Factors" : Economic and Political Weekly, Vol XI, 1976, pp 1203-1224. Krishnan says, 'Tn the last three or four decades Kerala has brought about a number of social and economic reforms which have altered the traditional relations in family and in land ownership. These changes have given rise to the emergence of unitary families in Kerala in place of the traditional joint families". While further investigation is needed to settle this we may point out what could be wrong with this analysis. Nuclearization is associated with early rather than late marriage, as we have argued. Land redistribution could have raised the age at marriage through the reverse of the nuclearization process connected with the abolition of tenancy which has created family property in land on a large scale. One more significant fact to be noted in this context is the presence of joint as also larger-sized families among agricultural labourers in Kerala as shovvn by recent survey data (referred to in the last footnote). This could be attributed to the conferment of ownership rights on hutment land occupied by agricultural labourers. Since newly set up labour households cannot get such hutment land this reform could have promoted jointness of family as well as lengthening the age at marriage among agricultural labourers. We do not however disagree with Krishnan^s general finding that social icform has contributed to the fertility decline in Kerala. 23 The survey data for Kerala, to which we have referred in the last two footno tes, lead us to the same observation. We found that all agricultural labour households classified among the upper per capita expenditure groups consist of couples with no children below the age of 15 years. While some of them could be newly set up, ihe inverse relationship between infant mortality and per capita expenditure strongly suggests the possibility of reduction in family size through infant and child deaths.