There is nothing new about the belief that the spread of education with its influence on knowledge and outlook is a central force behind the demographic transition. In 1934 Penrose wrote: ‘when a community has gained the knowledge and acquired the habits necessary to reduce the death rate it will sooner or later gain the knowledge and acquire the habits necessary to reduce the birth rate. There may be time lag between the two processes, but both of them in a large share are the outcome of education’ (Penrose, 1934). It is thus surprising that investigation of the specific influence of parental education on the mortality of children in developing countries was neglected until Caldwell's (1979) analysis of survey data from Ibadan, Nigeria, which demonstrated that mother's education was a more decisive determinant of child survival than other family characteristics such as husband's occupation and education. Other studies followed this influential investigation. The greater contribution to mortality decline of educational advance compared to health care provision, it has been argued, is a contribution to the development versus medical technology debate (Mosley, 1985). Inevitably the beginnings of a counter-reaction may be discerned. Caldwell (1986) stressed that the influence of education should not be considered in isolation from the wider context, while Cooksey et al. (1986, unpublished) consider that the effect of maternal education may have been exaggerated.