u 
Ifiheritance of the Duration of Life 
the reduction in the correlation of heredity as given in this case will enable us to 
find approximately the proportions of the selective and non-selective death-rates. 
As we have already pointed out, the case of brethren is better suited than the 
case of parent and offspring to test the proportions of the selective and non- 
selective death-rates. 
Now if we take parents and adult children, the average correlation from 
Table A for all combinations of sexes is "ISGS. For adult brethren, including 
brothers and sisters, it is •2831. If ^iV^ be the number of deaths in N cases 
which are non-selective, i.e. do not depend in any way on the constitution of 
the individual, then (1—^) will represent the chance of any individual dying 
under the selective death-rate, and (1 — jj)- the chance that both the members 
of a pair so die. Hence the ratio of the selective death-rate cases to the 
whole number of such cases is (1 — -p^N to N, or if r be the observed 
correlation, and the correlation to be expected from the laws of inheritance, 
we should expect rjrf, = {1 — pf*. We have therefore: 
•1365 ^(1 -pfn, 
•2831 =(1 -pYr,, 
where i\, and are the parental and fraternal correlations. The following table 
is based upon these equations ; we have given a range of values to i\ and 1\ 
covering the actually observed numbers lor other characters. 
TABLE C. 
Selective Death-Rate in Man deduced from Ages at Death of Relatives. 
Parental Correlation 
Fraternal Correlation 
Value 
Selective 
Death -Rate 
Noil- Selective 
Death-Rate 
Value 
Selective 
Death-Rate 
Non-Selective 
Death-Rate 
•3 
•4 
■45 
67'5 p.c. 
58^4 p.c. 
55-1 p.c. 
32-5 p.c. 
41-6 p.c. 
44-9 p.c. 
•4 
•45 
■5 
84^1 p.c. 
79-3 p.c. 
75-2 p.c. 
15-9 p.c. 
20-7 p.c. 
24-8 p.c. 
This table is in good agreement with the general results reached in the earlier 
paper, namely, that judging from parental inheritance the selective death-rate is 
63 to 70 per cent., and from fraternal inheritance about 80 per cent, of the total 
death-rate The drop between the results derived from fraternal and parental 
correlation is due, as we have before remarked, to the environment changing 
much more from parent to offspring than it does from brother to brother. Even 
the environment of two sisters is usually much moi'e alike than that of two 
* Phil. Trans, a, Vol. 192, p. 277. 
+ R. S. Proc, Vol. 65, p. 293. 
