58 
A Second Cooperative Study of " Vespa Vidgaris " 
Case (iii). Let the correlations be equal to the somatic correlations on the 
Mendelian hypothesis. Thena = |, /3=^, 7 = 2 25 and 
1-8125 ^+-0156=0, 
giving = 1'8041, and leading to 
o-„/5„ = l-08. 
Case (iv). We assumed the fraternal correlation was as high as "7, following 
Warren's results for Aphis and Daphnia. The ratio for a = ^, yS = ^ was 
o-„/5„=l-19, 
and no sensible change resulted. 
The values of the biometric and Mendelian gametic correlations fit the facts 
better than those of the Mendelian somatic correlations. But neither give a 
sufficiently reduced variability when we pass from general population to a special 
nest. We have at most 1"24 as against 1'8 to 21. Two only, and these the 
indices t7 and K for the left wing, out of the 26 characters, give values 1'20 and 
1"22 comparable with the above theoretical values. We are therefore forced to 
conclude that (a) there is some defect in the above algebraic analysis ; the 
problem is not an easy one, and there may be a slip, or {h) the wasps, as 
Dr Warren supposes in the case of the termites, are markedly influenced by the 
local environment of the nest, and thus a considerable part of the observed large 
variability ratio has an environmental and not a hereditary source, or (c) the result 
may be due to winter selection, if we suppose that selection of a twofold character, 
i.e. that from some unexplained cause the rather larger and rather smaller queens 
have a somewhat better chance of surviving than the mediocre queens. 
Undoubtedly the more prudent course at present is not, however, to manu- 
facture hypotheses, but to wait till further material has been measured and till 
more light can be thrown on this most interesting point. Unless we were most 
unfortunate in our selection of the original nest*, Table I shows that remarkably 
little effect has been produced by either selection or growth on the mean values 
of the indices during the winter months. Yet the variability of these indices 
is much larger than we can possibly attribute in the present state of our knowledge 
to the hereditary factor, when compared with their values for an individual nest. 
We should, a priori, conceive that local environment would give the general 
population a wide range of variability by making a large range of local means. 
The crux is that our one set of local index means is extraordinarily close to the 
general population index means. Personally, we admit that we are inclined to 
attribute the result to the continual inbreeding of the wasp, but we also confess 
that without assuming values for the hereditary coefficients incompatible with our 
present knoAvledge, we cannot reach such a variability ratio as we find indicated 
for both wasps and termites. 
* I.e. just chancing to hit a nest with all its indices closely in agreement with the general population 
means. 
