IS RrsifU of Cro.<^inff Japaiiesr Waltzing iritli Albino Miee 
and a zygote of elements G-^ only, or only can never be produced without 
assuming that remarkable elimination of some elements from every gamete 
suggested by Mr Bateson (Proc. Gamh. Phil. Soc. xil.), and so treating the 
phenomena of colour inheritance in a manner not only foreign to all Mendel's 
conceptions, but leading to obvious absurdity if applied to any set of characters 
other than colours. 
A further difficulty involved in any of these hypotheses is the correlation 
between coat-colour and eye-colour. Pink eyes are confined, as has been said, to 
mice with yellow, fown-yellow, or lilac coats, and all lilac mice whose eye-colour 
is certainly known have pink eyes ; but dark-eyed mice with yellow or fawn- 
coloured coats occur in this and in the preceding generation, and their presence 
is difficult to explain on either of the hypotheses examined. 
The difficulties we find when we try to describe the various classes of indi- 
viduals found among the offspring of hybrids are increased when we consider the 
great variability both in the colour of the coloiu'ed patches and in their extent 
among the individuals of any class, the amount of which may be gathered from 
Tables I.— VII. 
The correlations between the various characters exhibited by the offspring of 
hybrids and the characters of their ancestors are quite inexplicable on any 
hj-pothesis involving the purity of the albinos. The correlation between indi- 
viduals of this generation and the purity or impurity of their albino grandparents — 
that is the measure of the effect of ancestral pigmentation transmitted through 
the albino parent — is evident from Tables XX. — XXIII. These correlations are 
all negative, — the sign, taking the scales in the directions given in the tables, 
showing that the amount of whiteness is less among the individuals descended 
from pure-bred albino grandjaarents, the coat-colour itself being darker ; so that 
the relation between the albinos used in the original cross and the offspring of 
hybrids has the same sign as that between the albinos and the hybrids them- 
selves ; the correlation, however, is apparently larger, instead of being smaller as 
might have been expected, the value of r for both colour and whiteness being 
about — O'S (Tables XX., XXI.). A correlation so large as this is far beyond the 
probable error of the determination, and it is of course absolutely fatal to any 
theory which involves the " gametic purity " of the albinos. 
The only character of the pure-bred waltzing grandparents which varies 
enough to permit of its correlation with the offspring of hybrids being deter- 
mined, is the extent of the coloured patches. Dividing the waltzing grandparents 
into a and /3 as already explained, the correlations are fairl}' uniform, the coefficient 
for colour and for amount of whiteness being about 0"3 (see Tables XXIV. and 
XXV.), as in the previous generation ; the sign of the correlation between colour 
in the hybrids and amount of whiteness in their waltzing parent (which of course 
depends upon the order in which the colours are arranged upon the arbitrary 
scale) is vegative, showing that the waltzing mice with a greater amount of white 
