A. M. Carr Saunders 
355 
great to be accounted for by the normal darkening of the hair, and the most 
probable explanation is that the black haired type has been less subject to elimi- 
nation*. But the most interesting result obtained is due to Pearson, which I am 
allowed to quote, but which has not yet been published elsewhere. He found 
a correlation of "19 between health and hair colour, and of "07 between health 
and eye colour in the case of data relating to 2317 boys. A very similar result 
was obtained from data relating to a nearly equal number of girls. These results 
of Pearson and Pfitzner tend to confirm the theory referred to above, that the 
darker types are less subject to disease. The data used by Macdonald are in some 
respects comparable with those employed in this paper, and they bear on very 
similar problems in both cases. All discussion and comparison must be reserved, 
however, until the results of the present investigation have been set forth. 
Data and Methods. The data used in this work are derived from the informa- 
tion collected as the result of the Medical Survey of School Children carried out 
to meet the requirements of the Act of 1907. The data come entirely from 
Birmingham, and the fact that they were obtained for use in the Biometric 
Laboratory, is due to the kindness of Dr Auden, medical superintendent of the 
Birmingham Education Committee. The information with regard to each child is 
contained in a separate schedule, and includes a large number of particulars. The 
facts which have been of use in the present enquiry are those collected with regard 
to the hair and eye colour of the children, and to certain selected diseases. The 
hair colour was classified under six headings : red, fair, light brown, medium 
brown, dark brown and black, and the eye colour under four headings: blue, light, 
neutral and dark. The examination of the children was actually carried out by a 
number of doctors working with the help of nurses, all of whom Dr Auden had 
trained to estimate the pigmentation with the aid of a colour-meter. The classi- 
fication of hair and eye colour always presents great difficulties ; in the case of 
hair colour the most satisfactory method is to construct a scale with specimens 
of actual hair, and to compare the hair of each individual examined with the 
specimens on the scale. The colour-meter employed by Dr Auden consisted of 
different shades of coloured glass ; and those responsible for the filling up of the 
schedules gained by its use a general idea of the categories of hair colour, but 
did not directly compare the hair of the children with their standard scale. Such 
a method of estimation is necessarily somewhat rough ; but the results obtained 
are no doubt of sufficient exactness to enable broad conclusions to be drawn from 
them. 
The six specified diseases are measles, whooping cough, scarlet fever, diphtheria, 
mumps and chicken pox. The schedules also contain information about other 
diseases, but this is not employed here. 
Table I on the following page exhibits some of the weaknesses of the scheme of 
hair-classification employed. Taking the children of 13 years of age, it is clear 
* See Biometrika, Vol. in. 1904, p. 463. 
45—2 
