The concept of the ‘total social phenomenon’, in Mauss's odd but compelling phrase, has served anthropology well, not only as a frame for thinking about the complexity of cultural forms but as a narrative device, a way of handling the transition from the exuberant and bewildering world encountered in the field to the orderly microcosm of the ethnography itself. Malinowski, it could be said, found his ethnography in the kula, as Bateson found his in the naven ceremony. If we are now less confident about the notion of totality, this is as much a matter of practical contingency as of theoretical objection: our field locations tend, increasingly, to be characterized by ideological diversity and plurality. Yet, in ritual, we still encounter powerful evocations of the whole, a semblance of totality. What are we to make of this discrepancy? Consider the Javanese case. The slametan, a ceremonial meal consisting of offerings, symbolic foods, a formal speech, and a prayer, is a very modest event by the standards of a potlatch or kula; but it has a comparable primacy within its setting and a corresponding symbolic density. Participants see it as integral to their lives as social beings and to their sense of themselves as Javanese; they regard it as the epitome of local tradition. But its ‘totality’ is deceptive. The slametan is a communal affair, but it defines no distinct community; it proceeds via a lengthy verbal exegesis to which all express their assent, but participants privately disagree about its meaning; and, while purporting to embody a shared perspective on mankind, God, and the world, it represents nobody's views in particular.