Again, in the island of Guadalcanar in the Solomons the system of relationship is just such as would result from the cross-cousin marriage. One term, _nia_, is used for the mother’s brother and the wife’s father, and probably also for the father’s sister’s husband and the husband’s father, though my stay in the island was not long enough to enable me to collect sufficient genealogical material to demonstrate these points completely. Similarly, _tarunga_ includes in its connotation the father’s sister, the mother’s brother’s wife and the wife’s mother, and probably also the husband’s mother, while the word _iva_ is used for both cross-cousins and brothers- and sisters-in-law. Corresponding to this terminology there seemed to be no doubt that it was the custom for a man to marry the daughter of his mother’s brother or his father’s sister, though I was not able to demonstrate this form of marriage genealogically. These three regions, Fiji, the southern New Hebrides and Guadalcanar, are the only parts of Melanesia included in my survey where I found the practice of the cross-cousin marriage, and in all three regions the systems of relationship are just such as would follow from this form of marriage. Let us now turn to inquire how far it is possible to explain these features of Melanesian systems of relationship by psychological similarity. If it were not for the cross-cousin marriage, what can there be to give the mother’s brother a greater psychological similarity to the father-in-law than the father’s brother, or the father’s sister a greater similarity to the mother-in-law than the mother’s sister? Why should it be two special kinds of cousin who are classed with two special kinds of brother- and sister-in-law or with the husband or wife? Once granted the presence of the cross-cousin marriage, and there are psychological similarities certainly, though even here the matter is not quite straightforward from the point of view of the believer in their importance, for we have to do not merely with the similarity of two relatives, but with their identity, with the combination of two or more relationships in one and the same person. Even if we put this on one side, however, it remains to ask how it is possible to say that terms of relationship do not reflect sociology, if such psychological similarities are themselves the result of the cross-cousin marriage? What point is there in bringing in hypothetical psychological similarities which are only at the best intermediate links in the chain of causation connecting the terminology of relationship with antecedent social conditions?