It has also been suggested that selection produces the sex which is deficient in numbers; but here again proofs are wanting. It has been maintained that crossing tends to breed females, while consanguineous marriages produce males; in other words, that mongrel races show an excess of female births, while races in which marriages are very consanguineous, and polyandrous tribes show an excess of males. It is much better to leave this question alone till science has furnished us with conclusive proofs. Certain results obtained with the lower animals give hope that the future may shed some light on this point. Again, marriage customs are not always in relation to the excess of one of the sexes. Races in which men predominate are not always polyandrous, and those in which women are in excess are not always polygamous; sometimes even the contrary exists. Polygamy is thus not always due to a surplus of female births, or to the death of many men, but often to religious prescripts, as among the Islamites and Mormons. In polyandry, poverty often plays a greater part than consanguineous marriages or surplus of male births. Religious prescription of the husband's continence during his wife's menstrual periods, pregnancy, and even the period of nursing, a period which often lasts from two to four years in savages, is an important cause of polyandry. At Sierra Leone, coitus of the husband with his wife before the last-born child can walk is regarded as a crime. Although very advantageous to the wife's health this custom is entirely based on religious ideas and superstitions. Many savages consider that every woman is impure and bewitched during her monthly periods, during pregnancy and suckling. If we add to this the fact that, being usually treated as beasts, the women soon grow old, we can easily understand that the men are inclined to polygamy. It is remarkable with what rapidity the savage woman grows old. She is only fresh from thirteen to twenty years; after twenty-five she is old and sterile, and a little later she has the aspect of an old sorceress. This premature senility is not so much due to early sexual intercourse as to the terribly hard work they undergo, and also to the prolonged period of suckling. Another cause of polygamy is man's natural desire for change. The negroes of Angola exchange wives. The instinct of procreation, love of glory and riches coöperate with the sterility of many women in propagating polygamy. Certain races only tolerate it when the woman is sterile, or has only daughters, which clearly proves that it is based on the fear of remaining without male descendants.