E. Y. Thomson, J. Bell and K. Ppjakson 
63 
concern the total length of the wings and the lengths of cells h,f, and d (see Fig. I, 
p. 49). If we confine our attention to these homologous lengths of the wasp's wings 
we see that the range '911 to "964 is quite comparable with the values obtained 
by Whiteley and Pearson* for the first joint of the human fingers, ■904 to '934; 
with those obtained by Whiteley and Lewenz-f* for homologous bones of the two 
hands, '793 to '990; and by Dr Warrenj for homologous long bones in man, "86 to 
•97. The right and left wings of the wasp appear as closely related as the right 
and left members in man. 
(8) Conclusions. 
The results reached in this paper really centre on the determination of what 
meaning we are to give to the high variability ratio between the population of 
a single nest and the general population. The relatively close agreement of the 
means of this individual nest with those of the general population for the indices, 
does not seem to indicate that we are to look for an explanation of the difficulty 
in dififei-entiated environmental effect on the different nests. A priori reasonable 
values of the heredity coefficients do not appear to allow of a reduction of variability 
to 50% of tlie general population value. Once the source of tliis reduction in 
variability is accounted for, the correlation changes between nest and general 
population queens are seen to be of the order which should theoretically be 
associated with it. The only comparable material is that of Dr Warren for Natal 
termites, which gives a variability reduction of approximately the same value ; 
and this reduction is considered by him to be due in that case to differential 
environmental effect. It will be of interest to obtain, if possible, a series of 
autumn queens from the general population, and compare the variability of these 
with that of the spring queens dealt with in this paper. 
Note. It occurred to me after completing the above memoir tliat selection of 
midparents might give a greater variability than the selection of actual parents in 
each generation. Testing, however, for the case of immediate parents, this does 
not appear to be the case. The variability of the offspring : 
/(l-r^'-ri-e'' + 2r,r.,€) H 2r,Ml-e) 
' = "V ^ TTT^ ^ = -V^ — r^^" 
if the parental correlations be equal. Thus : 
by p. 54, if poi be the correlation of the offspring with the first midparent. Thus 
the midparent simplification of the analysis does not appear to account for the 
theoretically large variability in the single nest as compared with observation. 
K. P. 
* -R. S. Proc. Vol. 65, pp. 349—350. 
t Biometrika, Vol. i. p. 358. 
t Phil. Trans. Vol. 189, B, p. 178. 
