A. D, Darbishire 
7 
children of h3'brids ; the offspring of the hybrids crossed with albinos, children of 
hybrids and albino. These terms are perfectly simple, but it is impossible to be 
too explicit about terms such as these. 
The hybrids are invariably dark-eyed ; their coat is never completely white, 
and they never waltz ; the two facts about them which strike one immediately are 
(i) that whereas the parents are all like the other members of their race, the hybrids 
are by no means all alike ; they exhibit much variability in coat-colour and in the 
distribution of their colour ; (ii) that not only do they differ among themselves, 
but they are absolutely different from either parent ; the character of neither 
parent can be said to be dominant over the other in the sense that it exists in the 
hybrid to the exclusion of the other, which is what Mendel meant by dominance. 
To make up, however, for this deficiency in dominance, the characters of both 
parents are recessive, in so far that neither of them appear in the hybrid. This is 
plainly not a case of simple Mendelian dominance, though I am perfectly ready to 
admit that it may be an instance of the mysterious properties of heterozygotes, 
confessing at the same time that I do not see how the admission of this fact helps. 
It seems to me that we have not got any further in this direction than Darwin 
had when he called phenomena of this kind reversions to ancestral condition. 
The eye-colour and the amount and nature of coat-colour are not the only new 
characters which appear in the hybrids : there is a correlation between the size of 
the patches of colour and the intensity of the colour in sucli a way that the lighter 
the colour the more space it occupies. The measure of this correlation is between 
— 0 2 and — 0'3. This character of the hybrids is clearly absent in the parents ; 
for, while the waltzers vary slightly in the amount of colour which they possess, 
the colour itself does not vary : the albinos have no colour at all*. 
We have up to now merely stated the fact that the hybrids are variable ; it 
remains to try to account for this variability. How Mendelians will account for it 
I do not know, but I have facts to show that it is due to ancestry of the parent 
forms. 
Fraternal correlation. 
From the Mendelian point of view the families should be all similar ; and this 
would lead to fraternal correlation =0. Fraternal correlations are, however, 
remarkably like those observed in fraternities which are not cross-bred ; taking 
either the colour of the coat or the extent of the white patches as the character to 
be examined, we find a fraternal correlation veiy nearly equal to 0'5. This shows 
that there is (1) a sensible range of variation within each family, and (2) a con- 
siderable range in the mean characters of different families. The variation within 
the family seems to me incompatible with any Mendelian hypothesis which does 
* The obvious want of "normality" in the distribution of coat-colour and of whiteness in all 
the generations observed renders the determination of the correlation coefficients ver}' diflicult. The 
process adopted, which is due to Professor Pearson, will be explained and justified in a future paper 
by him. 
