The purpose of this essay is to present a brief description and evaluation of Mircea Eliade's interpretation of religious symbols with an eye to certain methodological difficulties. Professor Eliade characterises the procedure of the historian of religions as being uniquely preoccupied with religious symbols, that is, with those symbols which are involved in both the experience of homo religiosus and in his conception of the cosmos. As an historian of religions Eliade's own approach has consistently been to study religious symbols from a position that is not reductionist. Although admitting the influence of cultural and environmental factors upon the formulation of religious symbols, he has been primarily concerned to discern the typical structures or types of religious experience and expression by a consideration of religious phenomena. In his own words, he has attempted to study ‘the conceptions of being and reality that can be read from the behaviour of man of the premodern societies’.

Mr D. H. Mellor, in his article of this title in Religious Studies, Vol. 5 (December 1969), distinguishes three senses of words such as ‘probable’ which might be used in a religious context, especially in that of attempted theistic proofs: statistical, subjective, and inductive probability. In each case he concludes that it is misleading to use these words in such contexts at all. With his discussion of the second I do not wish to quarrel; but there seem to me to be serious defects in his discussions of the first and third.

In a recent symposium on Descartes' ontological argument, Norman Malcolm has restated a rather ingenious version of St Anse1m's ontological argument.1 The purpose of the present paper is to assess the merits of this particular version of the ontological argument.

Donald Evans, for one, thinks not. J. L. Austin, it could be argued, would have disagreed with him. But regardless of what Austin might have said on the question (someone is bound to suggest that he would have taken it back after he said it) I think that Evans is wrong.

A characteristic feature of religious discourse is its tendency to use expressions which are not only metaphorical but ‘irreducibly metaphorical’, that is to say, which have no literal equivalent. Sophisticated believers freely point out theology's tendency toward the ‘symbolical’, ‘analogical’, or ‘mysterious’; while equally sophisticated sceptics complain that faith is ‘anthropomorphic’ or ‘merely poetic’. When this discussion between believers and seeptics is confronted with what we may call the ‘reducibility thesis’, that metaphors should be reducible to literal statements, a philosophical puzzle results: religious metaphors should be reducible, but they are not.