E. C. Snow 
63 
is —'794 by the first method and —"64)7 by the second, while for females the 
figures are — - 798 and - - 878, these comparing with corresponding values for the 
English districts ranging from —-12 to -18. Thus the criterion which we take 
as the measure of the intensity of selection was for the Prussian cohorts of 1881 
and 1882 about five or six times as large as that for the English cohorts of 1870, 
1871 and 1872. We can assert with some confidence a considerably greater- 
selective effect of the mortality of the first two years of life on that of the next 
three in the case of Prussian rural districts than of the English rural districts in 
the epochs considered, and this fact is concomitant with a far greater stringency of 
infantile conditions in the former than in the latter. This is seen from the follow- 
ing figures. 
Mean Number of Deaths in 
First Two Years divided by 
Mean Number of Births 
Mean Number of Deaths in 
the nest Three Years divided 
by Mean Number of Births 
minus Deaths in First Two 
Years 
Male 
Female 
Male 
Female 
f 1870 
England \ 1871 
I 1872 
r, • I 1881 
Prussia -j 18g2 
•185 
•174 
•164 
•241 
■260 
■157 
•146 
•139 
■214 
•228 
■037 
•038 
■041 
•069 (T02) 
•075 (-107) 
•036 
■037 
•041 
•067 (T02) 
•074 (-107) 
The figures in brackets give the corresponding numbers for Prussian districts 
for the mortality in the eight years following the first two. 
The data for English rural districts do not allow us satisfactorily to follow the 
cohorts beyond their first five years of life, so that we cannot assert that the 
intensity of selection is generally less in England than in Prussia for the popula- 
tions considered, but we can point out definitely that the effect of selection in the 
first five years of life was much greater in the latter country than in the former. 
Whether or not the English cohorts make up the leeway at later ages can only be 
a matter of speculation. We can at present merely state that whereas a district 
in England which had an excess of 100 male survivors above the mean for all 
districts at the end of the first two years of life had, on the average, about 14 of 
these survivors killed off in the next three years, a similar district in Prussia lost 
more than 70 of the 100 in the same period. 
When, for the Prussian data, we come to deal with the results of including the 
eight years following the first two, we find that the regressions for the 1881 cohort 
have increased appreciably, but those for the 1882 cohort have not done so. There 
is nothing incongruous in this, as in one cohort selection might well be felt more 
in the 3rd, 4th and 5th years than in the other, and in this latter the effect would 
