368 Pigmentation, Selection and Anthropometric Characters 
the mortality is greater. Each class or category is, according to this argument, 
supposed to have the same number of diseases, but one class to show less resistance 
and to suffer a larger mortality ; and, therefore, when the two classes are examined 
at a later period, one is found to have more individuals who have suffered from 
disease. But this assumption may be false ; that class which has the higher 
mortality may very likely have also a larger number of cases of disease ; for a 
higher mortality might indicate a smaller power, not only to recover from disease, 
but also to resist the attacks. Thus that class which had the higher mortality 
would also have the larger number of cases. Therefore when considering the 
two classes, we might expect to find an equal number of individuals in both 
categories who had had disease. If this reasoning be correct, such results, as 
those we have reached in this memoir, may fail entirely to show whether 
selection is taking place or not. It can only be said that these results make it 
unlikely that pigmentation is a basis of selection ; but definite results can only 
be obtained from data which include not merely the surviving members of the 
population, but those which have died as well. 
In his recent memoir Macdonald* deals with the same problem. His figures 
come from Glasgow ; and in comparing his results with those reached in this paper, 
it must first be remembered that the populations in Glasgow and Birmingham are 
by no means of similar composition. The Birmingham population is probably as 
homogeneous as that of any large English city. The foreign element is extremely 
small ; children with foreign names in the schools are found very rarely. In 
Glasgow, on the contrary, there are large Irish and foreign elements ; and this fact 
may be of great importance in comparing the data derived from these two cities. 
Macdonald obtained figures from Fever Hospitals giving the pigmentation of 
children who were admitted for scarlet fever, diphtheria, measles and whooping 
cough. The pigmentation was measured by Tocher's method f . The children in 
the hospitals were then compared with Tocher's results from his "Pigmentation 
Survey of Scottish School Children." Differences in the proportions of various 
classes of pigmentation between these two populations were attributed to the 
greater liability of those classes of pigmentation, which were found to be in excess 
in the hospitals, to contract disease. Before examining Macdonald's results, it 
may be pointed out that the fact that the pigmentations of the general and fever 
populations were not measured by the same persons in the same survey makes 
it possible that the difference in the pigmentation classes in the two populations 
was due to personal equation. It is difficult to avoid such sources of error when 
the determination of such characters as hair and eye colour is not made by some 
very strict scale. 
Putting aside the question whether this fact could affect the value of 
Macdonald's results, his conclusions are at variance with those we have reached. 
He finds that the medium haired and medium eyed child is more liable to be 
* Macdonald, Biomctrika, Vol. vm. p. 13 et seq. 
t Tocher, Biomctrika, Vol. vi. p. 147. 
