Much stress has been laid by anthropologists on the few instances of peoples who habitually or occasionally contract unions which we should consider criminal. They have been taken for surviving types of the primitive condition of man, proving that “sentiments such as those which among ourselves restrain the sexual instincts are not innate.”[1976] But it is obvious that they prove nothing of the kind. Students of early history have often paid too much regard to exceptions, and too little to rules, overlooking the fact that there is no rule which has no exceptions. It may be objected that no feeling of incest exists among the lower animals.[1977] According to Mr. Huth, incest “is constantly practised by animals, and habitually by those which are polygamous.”[1978] But, as we have previously seen, among species that live in families, the young, without exception, leave the family as soon as they are able to shift for themselves; and Mr. Huth has adduced not the slightest evidence for his statement that “polygamy among animals means the closest incest.”[1979] The hypothesis here advocated can, I think, account for all the facts given in the last chapter. It explains how the horror of incest may be independent of experience as well as of education; why the horror of incest refers not only to relations by blood, but very frequently to persons not at all so related; why the prohibitions of consanguineous marriages vary so considerably with regard to the prohibited degrees, applying, however, almost universally to persons who live in the closest contact with each other; and why these prohibitions are so commonly extended much farther on the one side, the paternal or the maternal, than on the other. The question now arises:—How has this instinctive aversion to marriage between persons living closely together originated? * * * * * We have seen that a certain degree of similarity as regards the reproductive system of two individuals is required to make their union fertile and the progeny resulting from this union fully capable of propagation. It might, then, be supposed that the highest degree of similarity must be the most beneficial; but in all probability this is not the case. It seems to be necessary not only that the sexual elements which unite shall be somewhat like, but that they shall be in some way different. The similarity must not be _too_ great.