D. Heron 
369 
It is possible to illustrate the various fallacies which vitiate Dr Mott's conclu- 
sions regarding anticipation by considering the age at death of parent and child. 
I do not know whether it is generally recognised that it is exceedingly difficult 
to get any considerable body of data in which the ages at death of a parent and all 
his children are given, for of course the record is incomplete and biassed until the 
death of the last surviving member, and in some cases to get a complete recoi'd we 
must trace the history of a family for over 150 years. George the Ilird, for instance, 
was born in 1738 and all but one of his 15 children were still alive in 1810, 72 years 
afterwards, and the last surviving son, Duke of Cumberland and King (if Hanover, 
did not die till 1851, 113 years after his father's birth — and this is by no means 
an extreme case. In the material I am about to describe I found one case where 
the interval was 160 years. 
Another difficulty which arises is the tendency in practically all family 
histories to omit infant deaths, so that we do not get a complete record. It 
seems probable that the deaths of minors are not represented in such records 
in anything like their true proportion and that the differences are greater than 
might be expected to arise from differences of physique and nurture due to 
class. Thus records of the Landed Gentry give 31 deaths per 1000 males under 
20 years* while actual experience shows 103 to 197 per lOOOf. But in the 
records of the reigning families of Europe we get a practically complete record of 
all members and therefore from von Behr's Genealogie der in Europa regierevden 
Fiirstenhduserl , I have extracted particulars of the age at death of over 2000 
individuals — all belonging to the 18th century. There was here no selection — 
every child was entered and every family had been traced from the birth of the 
parents till the death of the last survivor. 
Now in Dr Mott's data we have already seen that cases in which the age at 
onset of insanity in parent and child is contemporaneous are most likely to be 
recorded. We can test the effect of a selection of this kind by investigating the 
effect of selecting, from our data regarding the age at death among those royal 
families, onl}' those individuals who died within a certain number of years of their 
father's death, and the results are given below in Table VI, p. 370. 
When we deal with the whole of the data, absolutely unselected, every family 
being complete and traced to the death of the last surviving member, we find that 
680 out of 1829 or 37 2°/^ died under 20 years of age. Let us now apply a very 
slight selection to the data and reject the 92 cases in which the interval between 
the deaths of father and child was at least 60 years. We find now that 680 out of 
the remaining 1737 died under 20 years of age — or 39'1 °/^. Thus the effect 
of a selection of this kind is to cause a slight increase in the proportion of deaths 
at the early ages. If we make the selection slightly more stringent, by taking only 
those who died within 40 years of their father's death, the percentage of individuals 
dying under 20 years of age rises to 46"7 and if we go still further and consider 
* See Pearson: Proc. R. S. Vol. 65, p. 291. t Statistics of Families, p. 73. 
% Tauchnitz, Leipzig, 1870. 
