Raymond Pearl 
73 
The brain- weiglit correlations with two exceptions are smaller than the 
corresponding skull capacity correlations. A portion at least of this difference is 
probably to be accounted for on the ground that brain- weight fluctuates as a result 
of causes which do not at all atfect skull capacity. Possibly the whole of the 
difference is to be accounted for in this way, but until there is available a more 
reliable collection of material on the brain-weight side the question cannot be 
definitely settled. Until such material is available we shall be justified, I think, 
in assumino' that the values of the coefficients for the correlation between brain- 
weight and skull characters given by the male series probably represent more 
nearly the normal relation than do those of the female series. 
Making all due allowance for the shortcomings of the material the following 
points are clearly brought out : (a) Brain-weight is positively correlated with 
both skull length and skull breadth, (b) The correlations here are higher than 
any of the other correlations of brain-weight with bodily characters, (c) The 
regression of brain-weight on these skull characters is approximately linear. 
(d) Indirect selection of skull characters is capable of modifying brain-weight to 
a considerable extent. 
The use of equations (31) to (36) in selection problems has been illustrated 
above (p. 49). One other example may be given because it bears very directly on 
a question of perennial interest in certain quarters. The problem to which I have 
reference may be stated for our present purpose in this form : Does dolichocephaly 
imply a greater brain-weight than brachycephaly or vice versa* ? Mere inspection 
of equations (35) and (36) shows that neither dolichocephaly nor brachycephaly will 
have a great advantage over the other in respect to brain- weight, but the matter 
can be made somewhat more anschaulich by taking a particular example. Con- 
sequently I have chosen the following figures. In Macdonell's Tables I and II 
{loc. cit.) are given the values for cephalic index, skull-length and skull-breadth 
for a dolichocephalic race (the Wliitechapel English series) on the one hand and 
a brachycephalic race (the Altbayerisch) on the other. Those values are as 
follows : 
? 
English 
Altbayerisch . . . 
English 
>i ... 
Altbayerisch . . . 
Cephalic index = 74-34 
„ = 83-20 
Length =189-06 
Breadth =140-67 
Length =180-58 
Breadth =150-47 
English 
Altbayerisch . . . 
English 
Altbayerisch ... 
Cephalic index = 74-73 
„ = 83-10 
Length =180-36 
Breadth =134-68 
Length =173-45 
Breadth = 143-98 
* Evidently the real essence of the question, so far as it has what might be called "humanistic" 
interest, rests on the assumption of a reasonably high positive correlation between brain-weight and 
intellectual ability. "Brains" not brain-weights are the things! Up to this point I have carefully 
avoided any discussion of the validity of such an assumption as that meutioued, deeming it eutirely 
irrelevant to the main purpose of the paper. The point is discussed in the concluding section. 
Biometrika iv 10 
