W. R. Macdonell 
187 
TABLE 6. 
1306 and 3000 Criminals. 
Coefficient of Correlation 
D 
S 
Head Length and Head Breadth 
Head Breadth and Height 
Finger and Height 
•00586 
■01420 
•00735 
•01884 
•02154 
•01245 
From these figures it is apparent that there is no sensible difference between 
the 1306 and the 3000 ; D is not in any case, even in the means, twice to thrice *S'. 
We may therefore treat our material for practical purposes as normal and homo- 
geneous. 
Before leaving this group of 1306, I may add that having formed correlation 
tables for Head Length and Left Middle Fin^^er, Head Length and Face Breadth, 
I find the coefficients of correlation to be as follows : 
TABLE 7. 
1306 Criminals. 
Coefficient of Correlation 
Head Length and Finger 
Head Length and Face Breadth 
•2861f0171 
•4074 ±-0156 
(7) Comparison of classes from luhich criminals are drawn with the educated 
classes of the community. It will now be interesting to compare our 3000 criminals 
with an entirely different social class of the population, and for this purpose I have 
prepared frequency correlation Tables VII., VIII. and IX. showing the correlation 
between Head Length and Head Breadth, Height and Head Breadth, Height and 
Head Length, in the case of 1000 male students at Cambridge. The figures are 
taken from cards in the possession of the Cambridge Anthropometrical Committee*, 
and are given in inches, but for Head Length and Head Breadth I have reduced 
the results to centimetres in order to facilitate comparison. 
* Thanks are due to the Committee for their courtesy in allowing the use of their cards. 
16—2 
