In the Bampanasi Stool House of the Oyoko royals of Ashanti, now at Breman on the outskirts of Kumasi, ten black stools are preserved, arranged as shown in Fig. 1. There is also a bell representing the famous Golden Stool of Ashanti, which is regarded as the black stool of its first king, Nana Osei Tutu. I am grateful to the Asantehene, Nana Sir Osei Agyeman Prempeh II for permitting photographs of these stools to be taken, and for the invaluable information he has provided about them. The black stools are of historical importance, in that they commemorate the reigns of several of the Ashanti kings. They are also of great social significance, enshrining the patriotic sentiments of the people and serving as a bond between the living, the dead, and those yet to be born. This bond is the very essence of Akan culture.

It is not uncommon in West Africa for lineage heads and chiefs in patrilineally organized societies to retain important bridewealth rights in their daughters or marriage wards and their descendants. My own field-work, on which this paper is based, was done among the Bangwa, a Bamileke group of the Cameroons. The Bamileke have been described as practising a special form of marriage, known as ngkap marriage. The principle is simple: a girl is given in marriage without transfer of bridewealth from her husband; as a result, the marriages of the daughters of this union are arranged by their maternal grandfather, who becomes known as their tangkap (ta ngkap = father [of bride] wealth). In ‘ngkap marriage’, unlike marriage with normal bridewealth transfers between a husband and his wife's kin, husbands only obtain rights in uxorem (in the woman as a wife) and not rights in genetricem (in the woman as a mother). If the tangkap continues to give his wards in marriage without receiving bridewealth from their suitors, he continually increases his capital stock of marriage wards. Chiefs among the Bamileke, heads of large polygynous compounds, give daughters and wards to their impecunious subjects, sons, and servants who cannot afford the marriage payments. Hurault estimated that in one chief-dom with a population of approximately 20,000, the chief had 1,500 female wards and was linked, through ngkap marriage, to the majority of his subjects. On a lower level lineage heads are able to increase the size of patrilineages through ngkap marriage, since wards and their children are incorporated into the group of their tangkap, not that of their father.

The Sandawe of central Tanzania speak a click language which shows no relationship with the languages of their Bantu-speaking neighbours, nor with any of the other non-Bantu languages in the neighbourhood; rather, it may be remotely related to the Khoisan languages of South Africa, in particular to Nama Hottentot. Physically the Sandawe differ to a degree from their neighbours, and their closest affinities may again be with Hottentot peoples. Sandawe material culture also differs to a degree from the cultures of their neighbours; like them, the Sandawe have an economy which largely depends on cattle-keeping and horticulture, but it is less sophisticated and their reliance on food-gathering and hunting is still considerably greater. Considering this difference in background it would be not at all surprising if their system of beliefs also showed differences. Comparisons cannot yet be profitably made, however, because little has so far been published about Sandawe religion, except a paper by van de Kimmenade and some details which can be found in the writings of Dempwolff and Bagshawe. In his ethnographic survey Huntingford draws our attention to the lack of knowledge of Sandawe religious beliefs, pointing out that these have been imperfectly recorded; yet he recognizes that the moon (láb′so or !áoso) and the sun (//′akásu) occupy a central position in Sandawe religion, which he summarizes as follows:

It appears that the sun and the moon are regarded as supreme beings, and that propitiatory sacrifices are made to the ancestral spirits who can do both good and evil to mankind.

Given that one is attempting to collect information as an academic observer among a people other than one's own, the data on many aspects of that society presented in standard ethnographic analyses must inevitably be thin and inadequate by the current standards of the social sciences. This situation tends to make for a one-sided approach. With a paucity of material at one's disposal, there is an overwhelming tendency to stress the analysis of ‘norms’ (and the more abstract ‘structures’ based upon them) since verbal statements of ‘expected behaviour’ are often more easily gathered than observations of ‘actual’ transactions. On this level, too, the limited number of one's respondents can be bolstered up by resort to linguistic parallels (‘from one informant you can analyse a language’) or by the comforting assumption, nourished by much sociological theory, that simpler societies are more ‘homogeneous’.

Can an anthropologist work with evidence supplied by historical sources? Or rather, can he deal with sources written by foreigners to a culture, data so obviously different from his usual reliance on his own participant observation? The question becomes more relevant each day as more historical data are becoming available about African and Asian societies. In addition trends in theoretical anthropology require more and more diachronic evidence to test or refine existing structural or functional theories.