362 Pigmentation, Selection and Anthropometric Characters 
differences in the death rate are not such as to seriously counteract the former 
factor. It would seem that the death rate can have very little influence on the 
figures in question, since the death rates for four out of the six specified diseases 
show no similarity. In whooping cough, for instance, the death rate is always 
higher for girls; while in scarlet fever it is almost always higher for boys. The 
only explanation, therefore, is that girls at the older ages suffer more from these 
specified diseases than boys; this is no real explanation, but a physiological fact. 
It does, however, rule out of account any explanation through selection by 
means of the six specified diseases which seemed at first the most probable 
It is well known that the correlation between hair colour and eye colour is 
fairly high ; it seems, in fact, to be something under "4*. Now Table VI is con- 
structed on exactly similar lines to Table IV, but with eye colour in place of hair 
TABLE VI. (See Tables XV and XVI, pp. 378—379.) 
Incidence of Disease in Eye Colour Groups. 
Eye Colour 
Boys. Age 13 
Girls. Age 13 
Cases of Disease 
Average 
Cases of Disease 
Average 
Blue 
Light 
Neutral 
Dark 
1012) 1635 
997 I 9005 
1008 | ~ uu 
1-75 1 1>77 
1-78 j 1 " 
1-831 llg7 
1-90 | 1 8/ 
804 I 2027 
1223 j ' 
1226 ' 2401 
1175 | ^ UL 
2-08| 
2-16| 1 11 
2 ' 17 ' 2-21 
2-19'f 
Totals 
3640 
4428 
Eye Colour 
Boys. Age 3 — 7 
Girls. Age 3 — 7 
Cases of Disease 
Average 
Cases of Disease 
Average 
Blue 
Light 
Neutral 
Dark 
325 i 425 
100 | 4 
144 I 302 
158 i 3Uw 
}:?( >•«» 
\Z\ 1-48 
256| 
85 f 541 
ioi i 205 
104 | " ° 
1-5 ! 1-54 
1-8 f 1 04 
1-6 * T48 
1 -4 j 1 48 
Totals 
727 
546 
colour. There is no reason to suppose a priori that, owing to the correlation between 
hair and eye colour, this table must show a similar absence of connection between 
eye colour and disease. If there be three characters A, B and C of which 
* Biometrika, Vol. in. p. 459. 
