MURDOCK HAS REFERRED To a large number of societies characterized by double unilineal descent of various kinds (Murdock I940); but as this article has arisen out of attempts to analyse the social structure of the Mbembe peoples of the Cross River, south-eastern Nigeria,2 I shall be concerned only with those societies which, like the Mbembe, have systems of double descent in which the patrilineal groups form the residential units, and the matrilineages although necessarily dispersed yet form corporate groups in the sense used by Radcliffe-Brown (I950, p. 41). Such combinations of descent groups have not frequently been recorded, but in Africa such systems have been described in some detail from the LoDagaba (Goody I956; I957), the Yako (Forde I950) who are the neighbours of the Mbembe, and from certain of the tribes of the Nuba Hills (Nadel I 947; I 950) . In addition, such a system has been more briefly described from the Ovimbundu (McCulloch I952), and the Chamba Daka at Gandole; and the (Chamba) Lamja of the Cameroons-Northern Nigeria border (Meek I93I, pp. 395-6) may also have such a double unilineal organization.3 At first sight these peoples appear to have little in common; they are for the most part sedentary cultivators and, except for the LoDagaba (and the Chamba for whom there is no information on this point) they have, or had before Europeans imposed peaceful conditions, large, nucleated settlements. But they live in very different natural environments, their economies differ considerably, and they have, apparently, different political systems. One thing however seems characteristic of these societies, or at least for those of them for which we have the necessary information, and that is the shortness of their remembered genealogies, for it seems generally true that the typical individual can quote his genealogy in either line for only two to four generations. It is not suggested that this is a feature restricted to systems of double descent, but the fact that it is characteristic of them has been pointed out by Goody (I957, p. 90), who notes that this phenomenon is found not only among the LoDagaba, but has also been observed by Forde and Nadel. Goody considers that this 'limited depth' appears to be associated with double descent itself, and suggests, as a reason, that 'an extensive calculation of specific genealogical ties in both lines would be highly complicated' (Goody I957, p. 90). However, although the Mbembe also have very short genealogical memories, a fact which lends support to the contention that there is a relation between double descent and genealogies of limited depth, it seems more probable that such genealogies may be the result of 'structural amnesia' which is usually found where larger lineage segments are not significant. If so, then the question that must be asked is not, 'Why are the genealogies in these societies so short ?', but rather, 'Is there anything incompatible between the recognition of two lines of descent and the recognition of many levels of segmentation in one or more of the lines ?'. 86