innate Aryan aversion to endogamous marriages, so strong even in the days of civilised Athens as to afford the Orestes of Æschylus with the defence that the mother whom he had killed was not of his blood, cannot be explained as a survival from times when kinship was counted exclusively through the female line. The savage practice must have its roots in some equally crude and savage theory. What the Aryan theory was we can hardly hope to discover, but we may conjecture that it was at least as barbarous as that which leads savages to eat their dead kinsmen, and European peasants to eat corpse-cakes, in the belief that thereby "the virtues and advantages of the departed ... and the living strength of the deceased passed over ... into the kinsman who consumed them, and so were retained within the kindred" (Mr. E. S. Hartland in _Folk Lore_, III. ii. 149). The _Leichen-nudeln_ of the Bavarian peasant, or the beans of the primitive Italian funeral feasts, would, when eaten, qualify the next of kin to wed the heiress and to raise up seed to the dead kinsman. Before leaving the subject of the levirate we may note that the joint undivided family survived in historic times at Athens and in Sparta, and that in both places brothers lived on the joint-estate as well after the death as during the life of their father. In Sparta, if one only of the brothers had a son, that son was naturally heir to the joint-estate, and was considered the son of all. Amongst the Hindoos, too, Vasishtha says (xvii. 10), "If amongst many brothers who are begotten by one father, one have a son, they all have offspring through that son" (_cf._ Vishnu, xv. 42).[130] Now, a casual observer, ignorant of the nature and constitution of the joint undivided family, might thus easily draw the mistaken inference that the wife of one brother was common to them all; and this may be the origin of Cæsar's statement with regard to the polyandry of the ancient Britons, and of Polybius' with regard to the Spartans. Or, again, it is possible that the joint undivided family may in these instances have given rise to this form of polyandry. It is thus not safe to infer that where polyandry is, the matriarchate must previously have been.