In a recent paper Gluckman puts forward a general hypothesis to account for the wide variation in the frequency of divorce among some African societies. Contrasting the bilateral kinship system of the Lozi of Northern Rhodesia with the strongly agnatic one of the Zulu of Natal, he notes that divorce is common among the former, rare among the latter, and suggests the general hypothesis that stable marriage is associated with patriliny. Gluckman tests this hypothesis against a substantial body of comparative data, drawn mainly from trans-Saharan Africa, and concludes that the available evidence seems generally to support it. He notes, however, that data for many areas are inadequate and he is properly cautious in the face of the methodological difficulties involved in putting such hypotheses to decisive test:
I am aware of the difficulties of establishing the validity of the hypothesis, but even if it is wrong it may be useful. Some of the difficulties are inherent in sociological analysis, since in this there are always complicating variables. Others arise from the vague and embracing use of categories and concepts (of which I too am guilty) such as patrilineal, lineage, marriage, divorce, &c….
I am myself uncertain whether it is the stability of people's attachment to specific areas, or patriliny or father-right itself, or the agnatic lineage, or all of these together, which, whatever the other variables are, tend to be associated with a strong marriage tie.
He concludes by inviting further testing and reformulation of the hypothesis.