8 Result of Crossing Japanese Waltzing with Albino Mice 
not involve a complexity in the constitution of the waltzing parent, leading to 
the assumption that its coat-colour depends not on one allelomorph, but on a 
number. This might be conceivable, but it would involve a reduction in the 
proportion of albino young among children of hybrids incompatible with that 
observed in the next generation. 
We see therefore that characters of hybrids are not compatible with, still less 
explained by, Mendelian pi-inciples ; it still remains to account for them. 
Parental correlation. 
If we classify the hybrids into those in which the albino parent was pure- 
bred, and those in which it was cross-bred, i.e. in which it had some pigmented 
ancestors, and determine the correlation between the colour of the young and the 
purity of the albino parent, we find that it is about -0-15, and the correlation 
between the amount of whiteness in the hybrids and purity of albino parent about 
— 0"2. These are quite sensible correlations ; they show that there is a distinct 
relation between the pigmentation of the hybrids and of the ancestors of their 
albino parent : they show therefore that the albino cannot be said to be gametically 
pure in respect of its whiteness, but that in order to predict the character of its 
young a knowledge of its ancestry is necessary. The character of the hybrids is 
therefore influenced by the ancestry of the albino parent : but it is also affected by 
the character of the waltzing parent. As has already been stated, pictures of all 
the waltzing parents of hybrids have not been made, but the waltzing parents of 
210 out of the 340 hybrids obtained are represented by accurate diagrams, and 
these have been divided into (a) those with less extent of coloured fur than the 
mouse figured in Fig. 1, and (/3) those with more ; from waltzers of category (a) 113 
young have been produced ; and from {fi) 97. The coat-colour and degree of 
whiteness in these young is shown in Tables IV. and V. (page 38). It will be 
seen that there is considerable correlation between the amount of white in the 
waltzing parent and that in the resultant hybrid. Table V. shows the correlation 
between coat-colour of hybrids and whiteness in their waltzing parent. In both 
these characters the coefficient of parental correlation is very nearly \. 
The amount and nature of the colour of a hybrid are therefore not constant, but 
exhibit great variability, (a) The hybrid does not show dominance of the character 
of either parent, and (6) its variability is sensibly correlated, not only with the actual 
character of the " dominant " parent, but also with the character of the ancestry 
of the " recessive " albino parent. Whatever theory these facts are to be reconciled 
with, I maintain that it is not with the theory of gametic purity. How far they 
are reconcilable with other theories of heredity is a more difficult question. 
In order to find out whether the results obtained are in accordance with 
the law of ancestral inheritance, as stated by Pearson (cf Biometrika, Vol. ll. 
pp. 211-229), it would be necessary to examine more closely than is here possible 
the effect produced upon all the coefficients of parental, grandparental, and other 
ancestral correlation, and upon the paternal correlation by the rigid selection of 
