the phenomena of the visible universe. Here, as in the other cases, the Australians, who possess—or till recently possessed—in a fairly complete form, customs and institutions which, among the Bantu, have fallen into decay, can elucidate many obscure points in our study of the latter. We are apt to forget that a classification of genera and species, and a sequence of cause and effect, which seem to us perfectly simple and obvious, are by no means so to children or savages. Even a person with no special scientific training would never suppose that a man could turn into a hyena, or that a leopard could become the parent of a crocodile, nor would he find it easy to realise the mental attitude of people to whom these things would seem quite natural. The leopard, says Mr. Dennett, "is looked upon as the mother of all animals [and] we cannot be wrong in saying that she is descended from Zambi through Xikamaci [Chikamasi] and her offspring Xikanga and Nxiluka, who are said to be the parents of an animal and a wooden figure" (p. 145).
Later on (p. 152), we read: "Zambici, in some of the stories, is spoken of as the mother of all animals, as if she were the immediate mother rather than the creator. This confusion is natural to degenerate people, who are apt to mistake the intermediate causes for the first cause." This is surely reversing the order of things. Such a conception belongs rather to a primitive than a degenerate stage of thought.
Mr. Dennett has not republished the etymological speculations of "The Bavili Alphabet Restored" (African Society's Journal, October, 1905), and I cannot help thinking, with M. Van Gennep, that he has been well advised. At the same time, it seems most probable that, as the last-named writer points out (Revue des Idées, Jan. 15, 1907), the "Categories" may give us the clue to the real meaning of the Bantu noun-classes, which some students were inclined to rank with things impossible to be known, while others seemed to lose themselves in fantastic conjectures. The statement that the first class contains nouns signifying persons, and the second names of trees, and so on, did not exhaust the facts, even as far as those classes were concerned, and the exceptions were too numerous to be treated as accidental. It is much to be desired that this investigation
