I W. R. Macdonell 205 
And if we compare our coefficients of variation as a whole with those obtained 
for other parts of the body by various investigators, we shall find a general 
agreement in the results. 
For instance, Miss Whiteley has shown that the coefficients of variation for the 
first joints of the fingers range from 4'695 to 5"361 {Proceedings of the Royal 
Society, Yo\. 65, p. 129); and Professor Pearson has found that these coefficients 
for the long bones (Aino and French) range from 4'655 to 5'425. 
From these figures it appears that 3 to o'o are representative values for 
variability in man, while in plants it may run to 40 ! 
I may direct attention to some of the general results of the table of correlation 
(Table 19); e.g., the close relationship of finger and foot, of cubit and foot, which I 
think is novel and interesting, of height and finger, and cubit and finger. It will 
be noticed that the high correlations between finger, cubit, foot and height agree 
generally with the correlations between height and long bones as calculated by 
Professor Pearson (Phil. Traiis. Vol. 192, A., p. 181, Table III.). For instance, he 
gives the coefficient of correlation between height and radius as ■6956, as compared 
with "7999 the coefficient of correlation between height and cubit in criminals. 
Turning now to the head measurements. Head Length, Head Breadth and Face 
Breadth, we observe that our results confirm the view that parts of the head are 
less highly correlated together than other parts of the body. It is also to be 
noticed that comparatively little correlation exists between the head and the other 
four characters, and that Head Length is more closely coi'related with Cubit than 
is Head Breadth or Face Breadth, wliile Face Breadth is more nearly correlated 
with Foot than is Head Length or Head Breadth. Long head associated with long 
arm (i.e. cubit) rather suggests the gorilla type, but I can think of no type 
suggested by the association of long foot and broad face. 
PART III. 
Application of these results to Criminal Identification. 
(16) The conditions to be satisfied in selecting a series of organs for criminal 
identification are : 
(i) Comparative ease and accuracy of measurement. 
(ii) Small correlation between them, so that fairly few organs will provide a 
reliable index to the criminal population. 
Head Length, Head Breadth and Face Breadth can be fairly accurately deter- 
mined ; Cubit and Finger with less accuracy ; Height changes about ^" per 
10 years from about 27 years of age*, and is not veiy reliable ; Foot is a doubtful 
measurement. 
* Biometrika, Vol. i. p. 49. 
