RELATIVES ON THE SUPPOSITION OF MENDELIAN INHERITANCE. 
433 
ratio and so distinguish dominance from all non-genetic causes, such as environment, 
which might tend to lower the correlations : this is due to the similarity in siblings 
of the effects of dominance which causes the fraternal correlation to exceed the 
parental. The fact that this excess of the fraternal correlation is very generally 
observed is itself evidence in favour of the hypothesis of cumulative factors. On 
this hypothesis it is possible to calculate the numerical influence not only of 
dominance, but of the total genetic and non-genetic causes of variability. An 
examination of the best available figures for human measurements shows that there 
is little or no indication of non-genetic causes. The closest scrutiny is invited on 
this point, not only on account of the practical importance of the predominant 
influence of natural inheritance, but because the significance of the fraternal correla- 
tion in this connection has not previously been realised. 
Some ambiguity still remains as to the causes of marital correlations : our 
numerical conclusions are considerably affected according as this is assumed to be of 
purely somatic or purely genetic origin. It is striking that the indications of the 
present analysis are in close agreement with the conclusions of Pearson and Lee as 
to the genetic origin of a part of the marital correlation, drawn from the effect of the 
correlation of one organ with another in causing the selection of one organ to involve 
the selection of another. This difficulty will, it is hoped, be resolved when accurate 
determinations are available of the ratio of the grandparental to the parental correla- 
tion. From this ratio the degree of genetic association may be immediately obtained, 
which will make our analysis of the Variance as precise as' the probable errors will 
allow. 
In general, the hypothesis of cumulative Mendelian factors seems to fit the facts 
very accurately. The only marked discrepancy from existing published work lies in 
the correlation for first cousins. Snow, owing apparently to an error, would make 
this as high as the avuncular correlation ; in our opinion it should differ by little 
from that of the great-grandparent. The values found by Miss Elderton are certainly 
extremely high, but until we have a record of complete cousinships measured 
accurately and without selection, it will not be possible to obtain satisfactory 
numerical evidence on this question. As with cousins, so we may hope that more 
extensive measurements will gradually lead to values for the other relationship 
correlations with smaller standard errors. Especially would more accurate deter- 
minations of the fraternal correlation make our conclusions more exact. 
Finally, it is a pleasure to acknowledge my indebtedness to Major Leonard 
Darwin, at whose suggestion this inquiry was first undertaken, and to whose kindness 
and advice it owes its completion. 
