It is certainly through an error of observation that some writers maintain that in the age-class of children who have survived their first year we find no weaklings. Even the strongest infant will not survive to enter this age-class if its environing conditions are too unfavourable. It often happens that an infant survives an illness, and yet survives in a damaged condition. A high death-rate is a consequence of a high disease-rate; but of the infants affected with disease, only a certain proportion succumbs, whilst the others survive with damaged constitutions, and constitute a favourable soil for fresh inroads of disease. In the twentieth century, in the civilised countries of Europe, of one hundred children dying during the first year of life, barely twenty die in consequence of inborn defects or congenital diseases (such as congenital debility, atrophy, congenital scrofula, tuberculosis, &c.). If it were true that infant mortality exercised a selective function, we should find a high infantile death-rate associated with a low death-rate in the case of children past infancy, and conversely. But everywhere we find that the infantile death-rate and the death-rate amongst children at ages one to five vary directly, and not inversely. Moreover, the variations in the death-rates in children during different years of life are determined by the fact that the younger the child, the less are its powers of resistance; thus the danger to health resulting from unfavourable conditions of life varies inversely as the child’s age; and even within the limits of the first year of life, the infantile death-rate is higher in proportion as the time which has elapsed since birth is less. Among the lower classes of the population, the infantile death-rate is higher than it is among the upper classes. If a high infantile death-rate exercised a selective influence, we should find that among the lower classes the death-rate among children more than one year old would be less than the death-rate of upper-class children of corresponding ages. But this is nowhere the case.