Karl Pearson and David Heron 
171 
elude that the Boas-Yuleau method indicated no practically important relationship 
between mental derangement and blindness and that there was no trustworthy sign 
of any modification of this relationship with age. The point is not whether such 
a relationship really exists or not, but that one method advocated by Mr Yule 
shows definite results (high association), although a second, equally strongly 
advocated, fails to give any association of practical value and is thus in direct 
opposition to the first*. Mr Yule has omitted to indicate which method is the 
proper one to use in such cases. Are both right or both wrong ? One method 
Mr Yule states may be applied to all 2 x 2-fold tables, the other should be used 
for " discrete " quantities. When is a quantity " discrete " ? Mr Yule confuses 
" discreteness " in the class-index— a mere verbalism — with discreteness in the 
attribute classified under it, and this reduces his investigations from the plane 
of practical statistics to the field where we originally placed them, that of 
theoretical logic. 
Diagram I. Diagram showing absence of any relationship of practical value between 
Blindness and Mental Defect. 
Blindness and Mental Defect. 
Average. 
(4) On Association and the Boas-Yulean Coefficient. 
Again we reach an interesting point which Mr Yule has failed to elucidate. 
It is best illustrated by an example. Take the following table for lengths of ivy 
leaves on the same spray — one of the examples selected by Mr Yule : 
* See also p. 204 below. 
22—2 
