76 
Inheritance of the Duration of Life 
secular acceleration. It is something we run up against at once, almost as soon as 
we examine a mortality table. But if natural selection be at work and can be 
seen at work we are still far from the end of our researches. We have got beyond 
the stage of those who still persist iti speaking of natural selection as a hypo- 
thetical source of change, but admitting once for all that it is continuously modify- 
ing races, we have still to investigate whether it can effectively differentiate races, 
i.e. be the source of the origin of species. This, whether we attack it from the 
standpoint of wild life or of laboratory experiment, is again a problem in statistics. 
Its solution depends on the relative fertility of different sections of a race with 
each other and with themselves. It cannot possibly be answered by observation 
of isolated cases, or by general reasoning. As we have said elsewhere : sine 
numero nihil demmistrandum est. 
(10) We may note lastly one or two general points bearing on the laws of 
inheritance which arise from a consideration of Table A. 
In the first place inheritance is stronger in members of the same sex, e.g. the 
correlation in the duration of life is greater between two brothers or two sisters, 
than between brother and sister ; between father and son and mother and daughter, 
than between father and daughter or mother and son. The daughter seems to be 
more closely related in duration of life to her parents than the son. This is some- 
what opposed to the result obtained for eye-colour*; but is I think quite explicable 
if we note how much more liable the male is to accidental death; thus the non- 
selective death-rate is intensified in his case. It is remarkable how much influence 
the death of a minor daughter has on the expectation of life of the mother as com- 
pared with that of the death of a minor son. One would have suspected that the 
higher mortality in childbed in the latter case would have made the reverse true. 
Finally, I may draw attention to the fact that the women are sensibly more 
variable in their age at death than the men. This confirms the general conclusion 
reached elsewhere-f- as to the greater variability of women. 
(11) Summary of Results. 
(a) Material for the inheritance of duration of life ought to be collected on 
a much larger scale, e.g. by assurance offices, than is possible for two isolated 
workers. We only publish our data as the best available at present. 
(b) Such material is of great interest, not only for actuarial but for biological 
problems. 
(c) Of actuarial interest we may especially note the results : 
(i) That the elder members of a family live sensibly longer than the 
younger. 
* Phil. Trails., Vol. 195, p. 117. 
t Pearson, Chances of Death, Vol. i. p. 373. 
