Kari. Pearson 
369 
general population of any species. Those who desire to test the efficiency of 
Darwinian theory must examine nature in the first place from this standpoint. 
(2) Again, recent work on evolution seems to liave very largely disregarded 
the direct possibility of testing in wild life, or under artificial conditions, the 
destruction of the less fit. The correlations between death-rate and environment, 
and between somatic characters and death-rate, are essentially problems which can 
be treated by biometric methods, and which are vital to further progress in our 
study of evolution. It is true that human vital statistics present an immense 
material for the relationship of death-rate to environment ; but they have not 
hitherto provided the much needed measures of the somatic characters of the 
selected. What is much needed is a widening of the breeder's outlook, — that he 
should not only find what is inherited, but that the age at death and the fertility of 
his stud should be recorded with special relation to the somatic characters of its 
members, and to marked differences in environment. Light, warmth, food are all 
in a high degree within his power of differentiation, and his greatest difficulty, 
environmental growth changes, ought to be surmountable by the use of standard 
populations. For the purpose of evolutionary study, once the intensity of heredity 
has been determined, the physiological process of heredity, important as it is 
for other branches of science, is secondar}' ; the next important step is the 
correlation of death-rate with somatic characters. I venture to think that more 
experimental work than has been recently attempted might well be under- 
taken in this direction. 
(3) Lastly I would refer to the third important factor of evolution — the 
absence of a differential fertility. This point, it seems to me, we again and again 
overlook. It is extremely hard to believe that fertility could by any mechanism 
come to be highly correlated with the peculiar type of character fittest to survive 
in a given environment. The only easy way to suppose it lies in the case of those 
species in which length of life is at once a measure of fitness and a measure of 
fertility. We cannot apply this consideration, however, to the case of species 
which breed only once. If on the otlier hand fertility were inherited, and were 
at the same time correlated with other characters, it is difficult to believe that 
a selective death-rate could perform its functions, except in the case where the 
maximum of fertility coincided with the optimum of the selected characters. 
Personally, the truth that natural selection presupposes the practical absence of 
genetic selection, has only gradually been forced upon me by the discovery of case 
after case in which there was little or no demonstrable inheritance of fertility. 
I began with the conception that fertility would be found to be a markedly 
hereditary character ; I expected to find genetic selection masking or even rever sing 
natural selection. Then, as very small or even insensible values came out for man, 
horse, swine and mice, I have been forced to the conclusion that the smallness of 
the hereditary factor in fertility is an essential feature of Darwinian evolutiun. 
I should be surprised to find a large inheritance of fertility now, just as I should 
be surprised not to find a large inheritance of any somatic character. And on the 
