64 Inheritance of the Duration of Life 
landed gentry and of the peerage. We have the following comparative re- 
sults : 
Constant 
(1) Peerage 
(2) Landed Gentry 
Mean of (1) and (2) 
Friends 
Mr 
65-835 
65-963 
65-899 
68^370 
58-775 
60-915 
59-845 
53^490 
60-971 
60-971 
56^568 
(Tf 
14-6382 
14-4308 
14-5345 
14^6974 
17-0872 
17-0986 
17-0927 
19-5706 
16-8354 
16-8354 
19-4302 
-1149 
-1418 
•1283 
-1353 
•2602 
-2602 
-2853 
Here Mp, il/^, are the mean ages at death of fathers, sons, and brothers; 
cTp, CT.s. and o"^ are the standard deviations of the same classes respectively, and 
'>\sF> '>'bb ^ii'G the correlations of ages at death of father and son and of pairs of 
brothers. 
Now these classes are really widely differentiated. Many of the Friends belong 
to the farming and small trading, as well as to the greater commercial classes. 
We see at once that the fathers live longer and the sons live shorter lives than is 
the case with the peerage or landed gentry. The younger male Friends have also 
a greater variety in their ages at death. Yet notwithstanding these class differ- 
ences the values of r^F and ?"^^ are in substantially good agreement for the two 
investigations. The value of r,^^ for the Friends lies between those for the 
peerage and the landed gentry ; the value of Vj^p^ differs from that for the peerage 
by less than the probable error of the difference. 
If we compare the six diagrams of plotted regression polygons and their best 
fitting regression lines which accompany this paper with the three diagrams of the 
former paper we shall see that the two sets are in general agreement. We note 
that the straight line serves as well as any curve could do to give the general 
drift of the phenomena ; the chief deviations from it are as a rule at early or late 
ages, where the data themselves are very sparse. We find also the same sensible 
divergence from the theoretical values for parental and fraternal inheritance 
marking the existence of a considerable non-selective death-rate. Fig. 3 for the 
regression of fetliers on sons even suggests the same points as Fig. (i) of the earlier 
memoir*. In early manhood the inheritance of duration of life approaches much 
more nearly the slope of the theoretical line ; in middle life the correlation is 
small, the non-selective death-rate having greater influence ; while in old age the 
inheritance of longevity is again very manifest f. 
With standai'd deviations for father and son of 14'6974 years and 19^5706 
years, and a theoretical correlation of S according to the Galtonian law, the slope 
* Loc. cit. p. 29G. 
t Loc. cit. p. 299. 
