Karl Pearson and David Heron 
307 
any reasonable assumption with regard to x, the tetrachoric process suggests not 
perfect but zero association ; the perfect association is only reached as a limiting 
case, although Mr Yule's coefficient gives an unhesitating unity*. 
* In the Table selected by Mr Yule to exhibit the variations of his coefficient of association {Phil. 
Trans. Vol. 194 A, p. 277) 51 out of the 164 values shown are ± 1 owing to the occurrence of a zero 
quadrant. With the introduction of a -5 into this quadrant — quite reasonable on the basis of a con- 
tinuous variate — his values maybe reduced from +1 to +"72 or swing over from - 1 to +-98, etc., 
etc.! What is the value in such cases of the statement that there is "perfect association"? 
39—2 
