K. Pearson 
337 
different relations to the same environment. I should myself be quite content to 
reject the Nigella for homotypic purposes on the basis of the continuous line, but 
further experiments were made with the offspring of the first year's Nigella to fix 
absolutely the nature of the differentiation*. 
It may be of interest to note that a purely algebraical attempt to resolve the 
1899 Nigella into its components led to the result 
r^vn - -dp, 
where r is the homotypic correlation and p the organic correlation between 
high and low capsules on the same plant. At present I have not data to 
determine p, but if its value be as I suspect from like cases in plants at least 
as high as r, then the homotyposis of Nigella will lie between "5 and '6, i.e. 
about three times its apparent value as affected by differentiation. The 
labour of such an investigation is only justified in the present case, because 
Mr Bateson appears to think that the biometrician has no power of detecting 
differentiation or having detected it of analysing his material. I think if 
Mr Bateson were better acquainted with the really large amount of work which 
has been done in detecting class and race differentiation by our modern methods 
he would speak less confidently of the difficulties which he, without having 
applied these methods, feels certain must arise in the practice of them. 
Turning now more directly to the problem of homotyposis, I believe that 
Mr Bateson would have understood my paper better had he not read his own 
in February 1901 — a date some weeks before my own paper was notified as 
accepted and ten months before it was printed and available for study. Thus he 
entirely misunderstands the relationship between fraternal cori'elation and homo- 
typic correlation. He appears to think that I am working with an analogy of 
some sort, and writes: "it would then be expected that the correlation between 
those repeated parts of the same individual would be similar to that between 
the germ cells of its parents " (p. 195). Of course the argument has nothing to do 
with " the repeated parts of the same individual." It is briefly the following : 
(i) The organ or character G in an individual A, putting aside the influence 
of environment, is determined in some way unknown to us by the characters or 
organs of the two gametes E and F from which was formed the zygote out of which 
A sprung. 
(ii) The organ or character C in another individual B will also be determined 
by the characters or organs of the two gametes E' and F' from which its zygote 
was formed. 
(iii) Hence if A has hereditary relationship to B the correlation between their 
like characters C and C must ultimately be deducible from the relationship 
between the gametes E, F, E', F'. In particular if A and B are brothers their 
correlation for any character depends upon the fact that E and E' are products of 
one gonad and F and F' products of a second gonad. 
* I have very heartily to thank Mr A. G. Tansley for taking charge of this crop. 
Biometrika i 35 
