Some exceptions to the rule that Totem-clans are exogamous may, perhaps, be explained by supposing that certain bands had not adopted Exogamy when they became totemists—may not this have happened amongst the Arunta?—and that they have since learnt Exogamy under other conditions from neighbouring people, or by conquest. Amongst the other possible conditions there are the remarkable Marriage-classes, or phratries that prevail in Australia, North America, and less regularly in other parts of the world. These Marriage-classes are considered by some ethnologists to be of deliberate institution—a reform for the regulation of Exogamy; and Spencer and Gillen, who knew the Australians so intimately, thought such a law—though its intricacy astonishes most Europeans and stumbles many—was not beyond their power of excogitation. Upon that point it would be absurd of me to have any opinion. But to suppose deliberate intention is contrary to the usual method of interpreting savage institutions: a savage language is a system of customs much more intricate and refined than the Marriage-classes, and not generally believed to have originated with the deliberate enactment of rules of grammar. I conjecture that those classes resulted at first from a grouping of exogamous Totem-clans which grew up by custom, and that the only deliberate work consisted in some minor adjustments of clan-relations. Marriage of cousins is prevented in Central and North Australia by the recognition of eight sub-classes; but the same result is obtained in other tribes by a custom which merely forbids it. Exogamous classes having been established in some tribes, may have been imitated in others. One result of the classificatory marriage system, however it grew up, is the extended use of names of kinship to denote all individuals of the same or corresponding generation—“father” for all men who might by custom have married your mother; “brother” and “sister” for all men and women descended from those whom your father or mother might have married; “husband” or “wife” for any man or woman whom, according to custom, you might have married. When further progress has been made in culture, distinctive names are used to express blood-relationship and class-relationship. It may be worth considering whether, in Australia, some of the customs which have been supposed to bear witness to an original state of promiscuity are not really results of the classificatory system of naming relationship: marital rights having been claimed on the ground of the names “husband” and “wife” as used in that system. § 6. THE CLANSMAN AND HIS TOTEM