This dissertation clarifies the anthropological concept of religion by crafting the most useful scientific definition. The first chapter illustrates the confusion surrounding this term, and argues that the comparative method of anthropology requires that this imprecision be removed. Reviewing the four types of definitions of religion extant in the literature -- Content, Behavioral/Performative, Mental, and Functional -- Chapter 1 concludes that the best definition will be a generative functional definition. The second chapter supports this conclusion by finding that an independent discipline, American constitutional law, has arrived at the same understanding.

Chapter 3 details the nature of definitions and the method of their proper construction, and uses these insights to define of religion based upon its function to alleviate death anxiety. Chapter 4 examines whether an inverse relationship between religiousness and death anxiety actually pertains, while Chapter 5 argues that the discerned relationship is not merely correlational, but in fact technically functional. The sixth chapter demonstrates the superior utility of the proposed definition as compared to content definitions by considering which generates better hypotheses about religious conversion.

The remaining chapters anticipate possible criticisms of the proposed definition. Perhaps because the highlighted inverse relationship is not supported unanimously by the empirical evidence, it is unsuitable for a definitional criterion. But Chapter 7 counters that imperfect results arise from a faulty methodological equation between personal religiousness and professions of institutional affiliation. To illustrate the independence of theisms and religiousness, instances of theistic nonreligions (superstitions, folktales) and nontheistic religions (political and scientific religions) are discussed.

The universality of the proposed definition could be more forcefully argued if the relationship between religiousness and death anxiety were tied to other human universals such as the psychobiological substratum of possession mediumship. Chapter 8 first shows that the most popular sociological interpretation of mediumship, that by I.M. Lewis, inadequately fits the data for Brazil. Chapter 9 supplements Lewis' model by linking mediumship to genetic talents for dissociation. Such links, if they exist, would serve as support for the underlying premises which predicted the results, and thereby the probable universality of the proposed definition of religion. [Ph.D. dissertation]