Anthropological analyses of rites can be divided into three types on the basis of their theoretical and methodological approaches. One of these has paid particular attention to the psychological1 and social2 functions of rites, while a second has classified rites according to their purposes.3 These two approaches are combined when rites are considered as means of serving purposes beyond themselves; the ritual action in which a series of representations is manifested is not studied in terms of its own logic but considered as the expression of other logics (psychological, economic, etc.). A third approach, centred on certain rites or collections of rites, has subordinated their functional character to the search for a classification which would reveal their mechanisms-the arrangements and processes they develop for effecting transformation. The works of Hertz (1960) on funerary rites, Van Gennep (1960 [1909]) on rites of passage, and Hubert and Mauss (1964 [1898]) on sacrifice express the various tendencies which mark this approach. Hertz studies the treatment of different components of death, especially the task of mourning during second funerals in one particular region, South-East Asia. He introduces a comparison between these rites and representations of death in other societies. Van Gennep is more ambitious: on the one hand, he extracts from the corpus of life-cycle rituals three main stages, processes, or sequences (rites of separation, of the margin, and of reaggregation) through which the main ritual actor passes in order to change his status, rank, or social position; on the other hand, he shows how rites of passage serve as models for understanding other rites (agricultural rites, warrior rites, etc.). Some have questioned the importance and the degree of elaboration of various

‘preliminary’, ‘liminary’, and ‘postliminary’ rites in the ceremonies studied here. One might wonder, as Mauss (1968:553-5) has, whether Van Gennep was too hasty in generalizing his propositions, thus emptying his concept of ‘passage’ of any precise content. However, the problems inherent in his analysis do not detract from the fact that he has posed an important question: Do all the rites in a given society have similar mechanisms? Do they have the same weight, or are some so important that they inflect the forms assumed by the others? Hubert and Mauss’s analysis of sacrifice seems at first sight to fall half-way between Hertz’s approach and Van Gennep’s. They have exposed the mechanisms common to sacrificial rites in all their diversity: on the one hand, the distinctions between sacrificer, sacrifier, victim, and divinity and the way in which they are related through the mediation of consecration, immolation, and sharing of the victim and rites of liberation. Sacrifice thus becomes a ‘procedure [which] consists in establishing a means of communication between the sacred and the profane worlds through the mediation of a victim, that is, of a thing that in the course of the ceremony is destroyed’ (Hubert and Mauss 1964 [1898]: 97). Although Hubert and Mauss do not attempt to apply their model to other types of rites, their analysis nevertheless enabled Durkheim to clarify his distinction between the sacred and the profane and to define his general perspective for the study of Australian religions and rites.