Work-parties, i.e. occasional labour groupings organized for specific economic tasks, are not as yet a topic of marked concern, either for social anthropology in general or for economic anthropology in particular. This is borne out by the fact that there is thus far no study which discusses them in any depth, or crossculturally. Yet work-parties are a fairly widespead phenomenon in Africa.

Since 1960 several million Africans have been relocated in South Africa, mainly from the common (or ‘white’) area to the so-called ‘homelands’. This process of relocation is still under way, and precise enumeration of the numbers of people who have been and will be affected is impossible. Estimates vary according to the definition of ‘relocation’: it is apparent, for instance, that official sources define the process as narrowly as possible in order to minimize the numbers involved. There is abundant evidence, however, that relocation has taken place on a vast scale, and that the process has generally resulted in increased poverty and misery for its victims (Surplus People Project, 1983). Kane-Berman has recently argued that the South African Bantustans are, at present, not so much reservoirs of a reserve army of labour as dumping grounds ‘for people who have little chance of obtaining employment anywhere’ (1981: 29). From having been sub-subsistence dormitory areas for labour migrants, they are rapidly becoming places to which the structurally unemployed are being permanently consigned.

This story begins with Robin Horton's thoughtful and stimulating article ‘African conversion’, published in Africa in 1971. My own ‘Conversion reconsidered’ followed in 1973. This was in part a response to Horton, though concentrating upon Islamic conversion, which had been only a supporting feature in Horton's chiefly Christianity-based argument. I also explored a little the possibilities for Muslim/Christian comparison in black Africa, a fruitful and promising field which has not figured prominently in the subsequent discussion, but to which I return briefly towards the end of the present article. Horton's own rely (‘On the rationality of conversion’, in 1975) being essentially rather acid in tone, it seemed best at the time to let the matter rest, which I did.

This article examines the growth of the Anomabo settlement pattern. Two main perspectives are used: one geographical, in which the number and configuration of villages are investigated; the other historical, in which village origins are examined, largely through the use of oral evidence collected from local informants. A discussion of Anomabo's geography begins the study.

The idea that the knowledge of an individual is an articulation of a culturally shared paradigmatic framework of concepts is now familiar in anthropology. It is one derived from the work of T. S. Kuhn (1962), who argued that, for natural scientists, the definitions and solutions of problems normally take place within an existing framework provided by the paradigmatic work of a great scientist. Since then several attempts have been made to indicate that, in fields other than the strictly scientific, the knowledge which people have of their physical and social environments is similarly an articulation of paradigmatic ideas (Ardener, 1971, 1978; Barnes, 1969; Buckley, 1976, 1982, 1983; Willis, 1967, 1972, 1978). This article will attempt to develop this methodology in the field of Yoruba religion. Its focus will be Sònpònnó, the Yoruba god of smallpox, a ferocious and vindictive god who not only causes smallpox, but who is also responsible for a form of madness.