In approaching the topic of women in African traditional religions, several methodological issues immediately arise: how to define religion, how to sample religious systems, and how to cope with fragmentary information. Each of these issues demands a rather arbitrary decision which implies that any analysis based upon such decisions must be considered suggestive rather than conclusive. If one defines religion as beliefs and practices associated with spiritual beings, one is likely to neglect important ritual domains dealing with status transformations that involve great and lengthy ceremony but rarely notions of transcendental beings. Moreover, such ritual situations of status transformation frequently involve women in important ritual responsibilities. Consequently, I begin with a Tylorian definition of religion as concern with spiritual beings in discussing religious norms and ideals for women, and move to a more extensive consideration of ritual in my analysis of religious organization and practice. In considering traditional sub-Saharan African religious systems for study, one is confronted not only by the multiplicity of potential units of study but by the diversity in societal complexity ranging from the Southwestern San bands to the Western Sudanic empires. In selecting societies for study, I decided to use Robert M. Marsh's index of social differentiation. 2 Within sub-Saharan Africa, Marsh classi1 This paper was originally prepared for the Wellesley Conference on Women and Development, held at Wellesley College in June 1976. I am grateful to the Radcliffe Institute for support and to Carol Troyer-Shank for assistance while writing the paper. 2 Robert M. Marsh, Comparative sociology. New York: Harcourt, Brace and World 1967, 329-365. Marsh's index of social differentiation in preindustrial societies is based on George P. Murdock's classification of social stratification and population size of the political unit. In the "World Ethnographic Sample" (American Anthropologist 59, 1957, 664-687), "Murdock coded each society according to five categories for each variable." Marsh assigns numerical scores to each of Murdock's categories ranging from o to 4, or from least to most differentiated. Marsh's "Index of Differentiation score for any given society in Murdock's sample is the sum of its score for population size of the political unit and its score for social stratification."