Witches, mysteries, rumours, dreams and bones: Tensions in the subjective reality of witchcraft in the Mpumalanga lowveld, South Africa

Abstract

Evans-Pritchard's classical text Witchcraft, Oracles and Magic Among the Azande (1937) lay the foundations for contemporary scholarly understandings of witchcraft. Yet the author's central contention that witchcraft presents a logical explanation for misfortune has been less inspirational than his suggestion than that witchcraft accusations express regularly recurring socio-structural conflicts [2]. This idea was developed most fully by Marwick (1970) who argued that witchcraft accusations present a social "strain-gauge". This formulation is based on two closely related assumptions. First, that at a general level, the distribution of witchcraft accusations, between persons standing in various relationships, reveals tension points in the social structure. Anthropologists and historians have contended that witchcraft accusations indicate different sorts of tensions in different social contexts. Witchcraft accusations have been shown to cluster between different matrilineal segments among the Chewa of Northern Rhodesia (Marwick 1965), agnates and affines among the Zulu of South Africa (Gluckman I960), youths and elders among the Gisu of Uganda (Heald 1986), competing work parties among the Hewa of New Guinea (Steadman 1985), commoners and new state elites in Cameroon (Geschiere 1988), and between men and women in colonial Peru (Silverbladt 1987). Second, the social strain hypothesis assumes that tense relations are the prime determinants of whom the accused shall be. For example, Macfarlane argues that in sixteenth century Essex witchcraft accusations arose from quarrels over gifts and loans, rather than strange events.

This article critically reexamines the relationship between social tensions and witchcraft. It draws on fieldwork conducted between 1990 and 1995 in Green Valley, a village situated in the lowveld of Mpumalanga, South Africa. In 1991 Green Valley had a population of approximately 20 000 Northern Sotho and Tsonga-speakers [4]. In the article I aim to focus on how individuals subjectively inferred the existence of witchcraft and the identity of alleged witches, rather than to explore the quantitative distribution of witchcraft accusations. From this perspective, I suggest that social tensions by themselves are less accurate predictors of witchcraft attributions and accusations than the literature may lead us to believe. The article is divided into two parts. The first considers the ontological status of witchcraft in local knowledge. I argue that the perception of witchcraft as a transcendent reality immunizes the belief against disproof. Yet in specific situations the occurrence of mysterious events, circumstantial evidence, revelations through divination and dreams, and confessions attested to the reality of witchcraft. Part two provides a detailed analysis of five case studies, and critically scrutinizes the role of social tensions relative to other types of evidence. I argue that social tensions were neither a sufficient, nor even a necessary, condition for witchcraft accusations. Villagers did perceive a conflictual relationship between the victim and the accused, prior to the advent of misfortune, as a motive for witchcraft. Tensions were therefore part of the wider framework of evidence they used to justify particular accusations. But villagers believed that witches often struck without motive.