06 
The Intensity of Nat tired Selection in Man 
addition to x 3 or x i} as before. This would fix, not the number of births, but the 
number of survivor's at the end of the first two years. Then the population liable 
to mortality in the second period would be the same for all districts, but the 
districts with the larger number of births would be exposed to the possibility 
of greater mortality in the first period, and therefore to the possibility of greater 
mortality of the kind which is taken to measure environment (the mortality in 
the first two years being much greater than in the next three or next eight). 
Thus an 'expected' (negative) correlation if selection were absent would not arise 
in the same manner as before, but would probably be entailed in the partial 
correlation. A priori, however, it does not appear that this method would pro- 
duce a correlation, if selection were inoperative, of the same magnitude as that 
indicated in § xxiv of the memoir, since the population rendered constant stands 
in an intermediate position to those at the beginning and end of the periods 
considered. The correlations under these new conditions have been worked out 
only for the case of the 1884 Prussian cohort, both male and female. The results 
differ but very little from those reached before, and must, I think, be taken 
as supporting the substantial accuracy of the interpretation put upon the earlier 
ones. If x s denote x a — w lt the following are the additional correlations: 
Males 
Females 
0-5 
3117-6 
3175-1 
r 6 i 
•924739 
•928746 
'•52 
•847138 
•851981 
? - 53 
•924711 
•942791 
>54 
•927020 
•934079 
5:i r i2 
■918 
-•747 
54*12 
- -780 
-•806 
The values previously found for 03 r 12 and 04 r 12 were —'917 and —'786 for males, 
and — - 768 and —"797 for females. It appears therefore to be of little account 
whether we make x c or x 0 — x 1 measure the size of the populations; we should 
probably, too, get similar results if we put x 0 — x-± — x 2 , i.e. the size of the cohorts at 
the end of our survey, constant. 
On the whole, the work of which this paper gives a short account has justified 
itself by the confirmation and emphasis it gives to the results previously obtained. 
The general impression received by a study of the results reached by the employ- 
ment of the new method of measuring environment alone is much the same as 
that derived from a survey of those by the earlier one, though individual differences 
of appreciable magnitude, occur. Apart from the emphasis it gives to the results 
of the memoir, the present work has discovered, I think, a significant difference in 
the operation of selection on the mortality of the first five years of life in Prussian 
and in English rural districts, and suggests (but, at present, no more than suggests) 
that there is some differentiation in its effect upon the two sexes. But the 
existence of a selective death-rate in the general populations dealt with admits of 
no doubt. 
