342 Correlation of Intellect with Size and Shape of Head. 

 (c.) Breadth of Head of Honours Men. 



Honours. 



] st class. . 

 2nd class 

 3rd class 



Tot als 



Grouping first and second class together Miss M. A. Lewenz found 

 r = 0-0056 0'0475 for the correlation between great ability and 

 breadth of head ; while when the second and third classes were 

 grouped together r rose to O1689 O0478, the probable error of the 

 difference being thus sensibly less than the difference. Now if such 

 a small number as 524 can be trusted there would thus seem to be an 

 increasing correlation between ability and size of head, when we 

 isolate the exceptional men who take first-class honours.* But when 

 we note the size of the probable error of these results, and see how 

 comparatively small they are, it seems possible to attribute their 

 divergence from the previous values to other sources than the relation 

 of genius to size of head. Notably we must bear in mind that head 

 length and breadth increase throughout life ;f and that the honours 

 men as a whole are older than the poll men and older in the first than 

 in the second class a considerable number of resident dons being 

 included in this series of measurements men of high honours and 

 considerable age. Hence I think we can lay only slight stress on the 

 apparent increase of correlation as we isolate the more brilliant men. 

 We must finally conclude that : Very brilliant men may possibly have 

 a very slightly larger head than their fellows, but taking the general 

 population there is really a very insignificant association between size 

 of head and ability. For practical purposes it seems impossible, either 

 in the case of exceptionally able men or in the bulk of the population, 

 to pass any judgment from size of head to ability or vice versa. 



* The alteration of correlation with the alteration of the class divisions shows 

 that the frequency surface is not accurately normal. In the general Cambridge 

 undergraduate population, as well as in the criminal classes, the head measure- 

 ments give normal curves within the limits of random sampling. See Macdouell, 

 loo. tit., p. 181, et seq. 



f W. Pfitzner, ' Zeitsclirift fur Morphologic u. Anturopologie,' vol. 1, p. 365. 



