What evidence does the author of "The Prehistory of Warfare" provide to support his view about the prevalence of warfare in preliterate societies? Why is it that the greater the proportion of a society is composed of unmarried young men, the greater the likelihood of war?