140 
Variation and Inheintance in Aphis 
tion. But a wide survey of the correlation of very various characters between 
parent and offspring of sexual forms lends no support to such a view. It 
should be noticed that the parental correlation for the ratio approximates more 
closely to the value usually found among sexual forms ("45 — "50)* than do the 
coefficients of either of the absolute dimensions. It is very probable that these 
discrepancies must be referred to the effect of environment. This view is favoured 
by the following consideration. I selected the frontal breadth as a criterion of 
the general size of the body and it is the size of the animal that is most affected 
by the environment. Now, the coefficients for the frontal breadth are more 
aberrant than those for the length of the antenna, and so it is reasonable to 
assume that these deviations are to be ascribed to the varying conditions of life. 
Figs. 3, 4 and 5 represent graphically the correspondence between the line of 
means of the arrays and the correlation line. Theoretically the mean of each 
array near the region of the general mean of the parents should be on the dotted 
line. Further away from the parental mean the means of the arrays are much 
less likely to approximate to the dotted line for the following reasons: (1) the 
unavoidable fewness of the more diverging individuals and the consequent impossi- 
bility of obtaining a satisfactory mean to these arrays, and (2) in the case of .skew 
curves the correlation line should really be sigmoid and not straight. 
In Fig. 3 the divergence of the uppermost point from the dotted line must be 
ascribed to the small number of individuals, there being only 7 members of this 
arraj' (Table X.). In Fig. 4 the widely deviating point between divisions + 2 and 
+ 3 (array 42'0 — 4o 9, Table VIII.) is probably due to the circumstance that 
these families may h^ve lived on particularly innutritions leaves, and this view 
is favoured by the fact that the adult individuals of these families tended to 
be small. 
(9) Cross-Inheritance. 
Although two organs A and B of an animal may be correlated yet it does not 
necessarily follow that there is a correlation between the organ A of the parent 
and B of the offspring. If, however, A and B are closely correlated in the indi- 
vidual it is probable that there will generally be found some cross-inheritance 
between A or B of the parent and B or A of the offspring. 
If A (frontal breadth) of the offspring be directly acted upon by the environ- 
ment and A and B (length of antenna) are dimensions closely correlated in the 
individual, then the irregular alteration of A in the offspring by different environ- 
ments for different broods would probably affect the correlation between organ B 
of the parent and the same organ of the offspring. If the alteration be uniform, 
that is if all the offspring are acted upon by one and the same environment which 
* Pearson, R. S. Proc. 1900, p. 157. Pearson and Lee, Phil. Trans, of the Roy. Soc, Vol. 195 (a), 
1900, p. 119. 
