Another explanation, given by Livi, seems to account adequately for the difference between city and country population, without necessitating the assumption of any considerable effect of natural selection, which presupposes an improbable correlation between mortality and fertility on the one hand, and traits like head-form and pigmentation on the other. The change of type in cities, so far as it has been observed, is of such character, that the city always shows greater resemblance to the average type of the whole large district in which it is located. If the local rural population is markedly short-headed, the general type over a larger area from which the city population is drawn more long-headed, then the city population will be more long-headed, and _vice versâ_. Unless selection can be demonstrated to occur in a sufficient number of definite families, this explanation seems simpler and adequate. Up to quite recent times no evidence of actual changes of type was available, except the observations by Ammon and those by Livi on the physical characteristics of rural and urban populations, to which I have just referred, and some others on the influence of altitude upon physical form. In the discussions of the distribution of different types of man in Europe, peculiarities of body-form in certain areas—as in the mountains of central France, in parts of Tuscany, in the province of Zealand in Holland, in southwestern Norway—have been explained as due to the survival of old racial types, to the influence of natural selection, or to the direct influence of environment, according as the necessities of the case prompted the investigator to adduce the one cause or the other, or a combination of any two or of all, as a convenient explanation of the difficult phenomenon (Ripley). It goes without saying that haphazard application of unproved though possible theories cannot serve as proof of the effectiveness of selection or of environment in modifying types. The effectiveness of selection can be proved only by an investigation of the surviving members of a type as compared to those eliminated by death, or of a shifting of population connected with the selection of a certain type. The influence of environment requires the direct comparison of parents living under one environment with children living under another environment. I cannot give any example in which the influence of selection has been proved beyond cavil. It seems plausible that in the criminal colonies of earlier periods, and in the settling of the West by the most vigorous members of our Eastern population, and in the complementary weeding-out of strong elements in some parts of New England, this principle may have been active; but we have no actual data which would connect with physical types the selection that has undoubtedly taken place.