2G2 
On Inheritance of Coat- Colour in the Greijhound 
It is possible that the difficulty here suggested, a temiDorary influence of 
embryonic co-environment, ought to be more fully regarded when records are 
taken of breeding experiments, especially in all cases of hair-colour where the 
' adult ' colour may not be sufficiently closely approximated to until after the 
loss of the first coat*. 
Still we are not prepared to attribute all high fi-aternal correlations to such a 
source as too early record. We consider that there is really greater variation 
in fraternal than in parental resemblances. Some of it may be due to a 
permanent effect of co-environment, embryonic or not. The problem is one of 
considerable interest and yet of great difficulty, and we only hope that more light 
may yet be thrown on it from further reduction of records or direct breeding 
experiments. 
(8) General Conclusions. Our reductions of the greyhound data indicate : 
(i) That whether there be one or more pigments, no class of greyhound can 
be looked upon as purely dominant or purely recessive. The colours of the earlier 
ancestry crop up in the offspring of parents of definite colours. We do not see how 
Mendelian principles can be in any way applied to the greyhound. We publish, 
however, the whole of our tabulated data, and are quite ready to receive and 
work out reasonable suggestions for its statistical reduction on Mendelian lines. 
(ii) That as far as ancestry is concerned the biometrical statistical method 
leads to parental and grandparental correlations sensibly identical with those 
which have already been found for pigmentation inheritance in man and horse. 
There can be no question therefore whether the knowledge of the ancestry is or is 
not important. It is equallv important in all three cases, and no good prediction 
can neglect the high correlations of the ancestry beyond the parents. 
(iii) That if we deal only with the stud-book records there is no difference 
of a marked character between siblings from the same and different litters, 
and the values of the fraternal coi-relation reached are in excellent agreement 
with those found for Basset Hounds and for pigmentation and physical measure- 
ments in the case of man. But if we deal with a colour-record made with the 
litters in a very early stage, we find the highest fraternal correlations yet reached. 
The nearest approach to them are the coat-colour correlations in the case of the 
horse, followed at some distance by Dr Warren's values for Aphis and Daphnia. 
It is probable that these high fraternal values with fairly uniform parental 
correlations are due to different sources, possibly to too early recording in the horse 
and dog, almost certainly to differentiated environment in the case of the insects. 
We submit, however, that while fraternal correlation ' clusters ' between "5 and "6, 
" The percentage returns given by Virchow, Pfitzner and others for darkening of the hair with age 
are fallacious because they are not dealing with the same material at different ages. They have neglected 
(i) the selection of the fitter— health we find to be correlated with hair colour — and (ii) the children in the 
primary schools (Virchow) are compared with a totally different group and class in the high schools. We 
are returning to this matter in another paper. 
