This book is bound to arouse considerable interest among students of Aboriginal kinship in general and of Yolngu1 culture and society in particular. Miwuyt Marriage is based substantially on Shapiro's Ph.D. thesis (1969a), revised in the light of further fieldwork, as well as a decade of reflection no doubt. The thesis was 'the basis also of numerous articles on Yolngu kinship and marriage, so that the book can be taken to represent the author's considered views on the subject, and to present some kind of synthesis. Shapiro's main quarrel is with descent and alliance approaches to Yolngu affinity. According to his earlier book (1979:89-91), an alliance view holds that (1) sibs2 are united in themselves and distinguished by patrilocal residence; (2) sibs are allied through marriage; (3) sibs act corporately in arranging individual marriages; (4) in these inter-sib relations females are the objects, males the givers and recipients of these objects; (5) these relationships involve exchange as well as alliance; and (6) the size of the alliance network depends on the number of patri-sequences in the relationship terminology. Shapiro argues in the present book that the data are better ordered by appealing to a notion of *endogamous kindred' rather than one of exogamous descent groups and their interconnections (p. 1). Among various alliance approaches to Yolngu affinity, Shapiro's particular targets are Leach (1961) and Maddock (1969). Northeast Arnhem Land society, he argues, has nothing much in common with asymmetric alliance systems of southeast Asia, such as the Kachin, but is a member of a class of Aboriginal social orders which includes the neighbouring Gidjingali. Levi-Strauss (1969) placed the 'Murgnin' system between Australian restricted exchange systems and generalized exchange systems such as Kachin; Leach (1961) transformed northeast Arnhem Land 'patrisibs' into local descent groups', linking (alliance to the descent theory of Fortes and others, a feat duplicated by Berndt (1955). The analyses of both Leach and Berndt stressed local corporations allied asymmetrically (pp. 1-2). Furthermore, contrary to Maddock's assertion that each patrilineal clan (sab) stands at the centre of an exchange network, the other elements of which are also patrilinea! clans (1974:71; 1969:24), 'the transaction is indeed part of an exchange network, but its elements are individuals and aggregates of matrikin' (p. 102). Maddock takes individual sib affiliations, which Aboriginal people sometimes use in discussing such networks, •to indicate initersib political relationships; supposing, in the face of the evidence, that the sibs are corporate groups naturally differentiated from each other (and hence likely to enter into such relationships); ignoring the evidence of specifically uterine kinship and representing it instead as agnatically inherited affinity (p. 102). In demolishing what he takes to be an alliance model, what kind of structure does Shapiro erect to take its place? It seems to have five main components. First the clan or sib is devalued as a corporate descent group, and relegated to ritual and expressive functions. Thus, kiin-ties are taken as providing 'the most salient grounds for social action', although partly 'expressed' in terms of sibs (p. 152). Second, the tendency for affinal relations between sibs to be asymmetric