THE INHERITANCE OF DURATION 181 



Snow's results for English and Prussian rural dis- 

 tricts are set forth in Table 22. From this table it is 

 seen that in every case the correlations are negative, and 

 therefore indicate that the mortality of early life is selec- 

 tive. Furthermore, the demonstration of this fact is 

 completed by showing that the observed coefficients are 

 from 3 to 10 times as great as they would be if there were 

 no selective character to the death rate. The coefficients 

 for the Prussian population, it will be noted, are of a 

 distinctly higher order of magnitude than those for the 

 English population. This divergence is probably due 

 chiefly to differences in the quality of the fundamental 

 statistical material in the two cases. The Prussian ma- 

 terial is free from certain defects inherent in the English 

 data, which cannot be entirely got rid of. The difference 

 in the coefficients for the two successive Prussian cohorts 

 represents, in Snow's opinion, probably a real fluctua- 

 tion in the intensity of natural selection in the one group 

 as compared with the other. How significant Snow's 

 results are is shown graphically in Figure 45. 



Snow's own comments on his results are significant. 

 He says: 



* 



The investigations of this memoir have been long and laborious, and 

 the difficulties presented by the data have been great. Still, the general 

 result cannot be questioned. Natural selection, in the form of a selective 

 death-rate, is strongly operative in man in the early years of life. Those 

 data which we believe to be the best among those we have used the Prus- 

 sian figures show very high negative correlation between the deaths in 

 the first two years of life and those .in the next eight, when allowance is 

 made for difference in environment. We assert with great confidence that 

 a high mortality in infancy (the first two years of life) is followed by a 

 corresponding low mortality in childhood, and conversely. The English 

 figures do not allow such a comprehensive survey to be undertaken, but, 

 so far as they go, they point in the same direction as the Prussian ones. 

 The migratory tendencies in urban districts militate against the detection 

 of selective influences there, but we express the belief that those influences 



