Page:Primitive Culture Vol 1.djvu/29

Rh once settled, new cases recorded by new observers simply fall into their places to prove the soundness of the classification. There is found to be such regularity in the composition of societies of men, that we can drop individual differences out of sight, and thus can generalize on the arts and opinions of whole nations, just as, when looking down upon an army from a hill, we forget the individual soldier, whom, in fact, we can scarce distinguish in the mass, while we see each regiment as an organized body, spreading or concentrating, moving in advance or in retreat. In some branches of the study of social laws it is now possible to call in the aid of statistics, and to set apart special actions of large mixed communities of men by means of taxgatherers' schedules, or the tables of the insurance office. Among modern arguments on the laws of human action, none have had a deeper effect than generalizations such as those of M. Quetelet, on the regularity, not only of such matters as average stature and the annual rates of birth and death, but of the recurrence, year after year, of such obscure and seemingly incalculable products of national life as the numbers of murders and suicides, and the proportion of the very weapons of crime. Other striking cases are the annual regularity of persons killed accidentally in the London streets, and of undirected letters dropped into post-office letter-boxes. But in examining the culture of the lower races, far from having at command the measured arithmetical facts of modern statistics, we may have to judge of the condition of tribes from the imperfect accounts supplied by travellers or missionaries, or even to reason upon relics of prehistoric races of whose very names and languages we are hopelessly ignorant. Now these may seem at the first glance sadly indefinite and unpromising materials for scientific enquiry. But in fact they are neither indefinite nor unpromising, but give evidence that is good and definite so far as it goes. They are data which, for the distinct way in which they severally denote the condition of the tribe they belong to, will actually bear comparison with the