The marriage of a woman to a woman, found in many African societies, has not been given the attention it warrants, and is still imperfectly understood. Herskovits imputed to it sexual overtones that are foreign to the institution when, after stating quite definitely that such marriage did not imply a homosexual relationship, he went on to add ‘although it is not to be doubted that occasionally homosexual women who have inherited wealth…utilize this relationship to the women they marry to satisfy themselves’ (Herskovits, 1938, 1: 319–20). He made no attempt to substantiate his statement. And Lucy Mair, for all the clarity and grasp displayed in her excellent book on marriage, seems to have failed to appreciate the nature of the institution when she says, ‘According to Evans-Pritchard's account of the Nuer it is usually barren women who make such marriages, and indeed it is hard to imagine a woman who had her own children doing so’ (my italics; Mair, 1971: 60). In actual fact it is usually married women with children of their own who contract such marriages, except, it would appear, among the Nuer. By woman-marriage we mean the institution by which it is possible for a woman to give bridewealth for, and marry, a woman, over whom and whose offspring she has full control, delegating to a male genitor the duties of procreation.

The aim of this article is to interpret and understand certain Sewa Mende socio-spatial categories and at the same time to attempt to reveal an order in the otherwise confusing complex of supernatural spirits. This limited aim may be justified in so far as the Mende are among the most documented of West African societies, and yet an examination of the principal works (Little, 1951; Harris and Sawyerr, 1968) would suggest that Mende society consists of a constellation of unconnected elements. Thus their secret societies have received considerable attention, but what has a secret society (hale) to do with, for example, medicine (hale), an object for swearing by (hale), or a magical thief-catching device (hale)? Little has described them as ‘“broadly speaking” patrilineal’ (1951: 98) though his data hardly support even this vague contention. And to what are the diverse modes of supernatural spirit related? Mende culture as a whole may indeed show a remarkable lack of coherence and Harris and Sawyerr (1968: xii) have commented on the cultural variations among the different groups that constitute the Mende nation; consequently it is pertinent to add that this article is based wholly on ethnographic observations made among one of these groups.

The Shona-speaking complex of tribes covers most of present day Southern Rhodesia, stretching to the Zambian border along the Zambezi River and into Mozambique in the north and to the Indian Ocean in the east. The Korekore group is the major northern division of the Shona-speaking peoples. The original Korekore appear to have been a band of Karanga invaders from the south who entered the middle Zambezi Valley in the fifteenth century and established the Mwene Mutapa confederacy. Today the Korekore group occupies an area extending south from the Zambezi between longitude 28° and longitude 33° east, including the low-lying valley floor and the middle and high veldt or plains to the south where most of the population is concentrated (Garbett, 1966). The valley floor in this region has been considered a poor habitat by African and White settlers alike. The climate is generally poor, there are few cattle because of the tsetse fly, the area is economically undeveloped, and the population is sparse and scattered. Because of their poorer habitat the valley dwelling Korekore have come to be distinguished from their more fortunate highland neighbours to the south and are disparagingly known as Goba or poor lowlanders. In times past they have also been known as Banyai or vassals to the stronger Shona chiefs on the more populous cattle-rich highlands, and also as Chikunda or followers of the Portuguese and half-caste traders operating along the Zambezi in the nineteenth and earlier centuries. This historical background of changing political relations symbolized by changing ethnic identities will be described elsewhere (Lancaster, n.d.).

A recent paper by Igor Kopytoff (Kopytoff, 1971) argues that Africanists have been led by their Western conceptual ethnocentrism into creating a false dichotomy between ‘elders’ and ‘ancestors’ in their descriptions and analyses of African societies. He asserts that this has obscured the fact that ‘elders’ and ‘ancestors’ are conceptually as well as terminologically united in many African cultural and social traditions. There are flaws in his assertion, perhaps the most serious of which has been ably countered by James Brain in a recent article (Brain, 1973) where he demonstrates that many, if not most, African societies make explicit and clear-cut distinctions both linguistically and conceptually between ‘elders’ (both living and dead) and ‘ancestral spirits’. Nevertheless Kopytof's article effectively argues that Western scholars have had a tendency to overlook indigenous conceptual continuities between living and dead elders, and this prompted me to re-examine my data on the Tiriki of Kenya and the Irigwe of Nigeria which deal with the traditional relations of the living and the dead (including the not yet born). I found that these relationships, though strikingly different in the two societies, are nevertheless in each case patterned in ways that are congruent with relationships between both adjacent and alternate generations.

The Yalunka of north-eastern Sierra Leone are predominantly Muslim. Their religion seems to be a straightforward variant of contemporary West African Islam. They have been exposed to Islam for several centuries and although they had powerful Muslim neighbours during the eighteenth and nineteenth centuries, for most of this period they resisted conversion and remained pagan (Laing, 1825; Donald, 1968) until 1882, when they were conquered by one of the armies of Samory and forcibly converted. After the establishment of British and French control over their area in the 1890s many Yalunka reverted to paganism, but Islam remained viable and the number of adherents increased steadily. By the 1950s nearly all Yalunka in Sierra Leone were at least nominal Muslims.