OF ATTRIBUTES IN STATISTICS. 



311 



individual will decrease association is obvious in the case of such a defect as blindness, 

 which may be brought on by accident, cataract, or other causes having no bearing on 

 mental derangement or dumbness. It is not obvious, however, why the association 

 between blindness and mental derangement should be much smaller than the 

 association between blindness and dumbness in the later age groups. 



65. The question of what would be the effect of selection alone or whether it 

 would have any regular effect in one direction is, however, a most interesting one. 



TABLE VII. Showing the Number of Cases in which the Association Coefficient 

 decreased or increased on passing from one group of Standards to the next. 

 There is a great majority of decreases. 



(Note. Two coefficients that remained constant from one Standard group to the next have been 

 entered as decreasing.) 



The fact that partial associations in undefective universes are higher than total associa- 

 tions, combined with my first impression (unjustified I think) that associations were, on 

 the whole, higher in the healthier groups, led me at first to believe that the effect of 

 selection would be to increase associations. I still cannot help thinking that this is 

 practically, as it is formally, possible, it being remembered that association will 

 decrease or increase simply according as 



3 i s s 



** *s s t being the survival rates in the four classes AB, A/?, Sec. (vide 54). 



