K. Pearson 
339 
(ii) Differentiation exists between pairs of brothers, and therefore I ought to 
have included the correlation of differentiated like organs in forming my average. 
My reply is that I was dealing with types of life in which differentiation 
between pairs of brothers is not sensible, and therefore I was perfectly justified in 
seeking the correlation of undifferentiated like organs, my homotypes. 
Mr Bateson wants to know what I should do if I had to deal with fraternal 
correlation in a community of ants. I reply that nobody at present knows 
anything whatever about heredity in a confraternity of ants, and that until 
some attempt has been made to apply the exact methods of biometry to such 
communities it is impossible for him to assert either that differentiation is so 
imperfect that it cannot be determined, or still more that if it exists without 
being sensible it would have any sensible effect on fraternal correlation*. Mr 
Bateson will I hope pardon me if I say that a quantitative study of variation 
and heredity in ants, starting with those genera where differentiation can be 
detected, would be a far more valuable test of my views, than any amount of 
appeal to what may or may not be the case under circumstances which nobody 
has tested. 
(iii) Mr Bateson charges me with being compelled " to pick and choose " 
my cases, and he puts this charge in a manner which anyone reading his 
paper without studying mine — and there will be many such among biologists — 
would undoubtedly interpret to signify doctoring of returns. 
" In plain language, we shall have to pick and choose our cases, and the 
value of our coefficient of homotyposis will depend entirely on how we do it. 
Has not Professor Pearson himself been so compelled in more than one of his 
examples, notably in that of Nigella ? " (p. 202). 
Any reader of this passage would think that Nigella and other things had 
been discarded by me after I had found their coefficient of homotyposis to be 
out of keeping with my theory. But what are the facts clearly stated in my 
paper ? Why, that after the first few Nigella were examined they luere recognised 
to he differentiated, long before their coefficient was found. That again I was 
distinctly warned not to include Asperula odorata and Scolopendriimi vulgare by 
well-known botanists before I had even begun to collect them ; that Malva rotun- 
difolia was collected with a full knowledge that it had spread by stolons, and that 
I did collect and measure the homotyposis in all these things, because I wanted to 
appreciate the effect of differentiation and connnon origin on the coefficient of 
homotyposis. And having done this, did I put them on one side as I really ought 
* It is possible that tlie correlation between pairs of brothers both in the army and pairs one in and 
one out might differ slightly for men. Mr Bateson might be puzzled to know how to rate a pair one of 
whom was a volunteer. I happen to know the effect of volunteering on fraternal correlation. It is 
sensible but of no practical importance. That one quantitative fact is of more value I take it than 
Mr Bateson's sweeping statement about ants whose heredity nobody has yet studied by exact methods, 
i.e. "average fraternal correlation, I think, has no meaning, still less an ascertainable value in these 
cases" (p. 201). 
35—2 
