This paper is premised on the view that family systems determine both the residential position of the elderly relative to their offspring and also explain the extent to which aged parents are likely to be naturally dependent upon their children. Following Hajnal (1982), family systems can be distinguished according to whether and by what means they provide mechanisms for adjustment of population growth in response to varying economic conditions. In pre-industrial north-west Europe, the ‘rules’ of family formation permitted adjustment through the timing of marriage. According to these rules, marriage marked the establishment of an independent household, for which a prior accumulation of capital was necessary. Many young people acquired this capital through employment in domestic service.1 Times of economic hardship saw the period of service extended and marriage delayed. Similar (probably more stringent) controls of new household formation existed in pre-industrial Japan, associated with the high value accorded preservation of the family land-holding and descent line and perhaps also with collective village responsibility for land tax. In this way, Japan experienced near constancy of rural population over generations.2 No comparable mechanism of response and adjustment exists for the joint family formation systems once prevalent elsewhere in Europe, and still so in most of contemporary Asia and the Middle East. There, the timing of marriage is not coincident with the formation of an independent household: a newly-married couple typically is sheltered for a period as a member of an elder’s household. Nor are there economic constraints on early marriage or household establishment in traditional, lineage-dominant family systems of Africa-nor appreciable signs of such constraints appearing as these systems are increasingly stressed.