This attempt was made four years ago by Professor Kroeber,[5] of the University of California. The line he takes is absolutely to reject the view common to both Morgan and McLennan that the nature of the classificatory system has been determined by social conditions. He explicitly rejects the view that the mode of using terms of relationship depends on social causes, and puts forward as the alternative that they are conditioned by causes purely linguistic and psychological. [5] _Journ. Roy. Anth. Inst._, 1909, xxxix, 77. It is not quite easy to understand what is meant by the linguistic causation of terms of relationship. In the summary at the end of his paper Kroeber concludes that “they (terms of relationship) are determined primarily by language.” Terms of relationship, however, are elements of language, so that Kroeber’s proposition is that elements of language are determined primarily by language. In so far as this proposition has any meaning, it must be that, in the process of seeking the origin of linguistic phenomena, it is our business to ignore any but linguistic facts. It would follow that the student of the subject should seek the antecedents of linguistic phenomena in other linguistic phenomena, and put on one side as not germane to his task all reference to the objects and relations which the words denote and connote. Professor Kroeber’s alternative proposition is that terms of relationship reflect psychology, not sociology, or, in other words, that the way in which terms of relationship are used depends on a chain of causation in which psychological processes are the direct antecedents of this use. I will try to make his meaning clear by means of an instance which he himself gives. He says that at the present time there is a tendency among ourselves to speak of the brother-in-law as a brother; in other words, we tend to class the brother-in-law and the brother together in the nomenclature of our own system of relationship. He supposes that we do this because there is a psychological similarity between the two relationships which leads us to class them together in our customary nomenclature. I shall return both to this and other of his examples later. We have now seen that the opponents of Morgan have taken up two main positions which it is possible to attack: one, that the classificatory system is nothing more than a body of terms of address; the other, that it and other modes of denoting relationship are determined by psychological and not by sociological causes. I propose to consider these two positions in turn.