SELECTION ON THE YAKIABIIJTY AND C'OEEELA'JTON 
OF ORGANS. 
tlieorem to special cases, but we must note that if two organs were uncorrelated with 
each other, it still might be possible by selecting a third, or a third and fourth, to 
jDi’oduce correlation between them. Further, by selection of one or more organs, two 
non-directly selected organs can have their existing correlation increased, lessened or 
even chano;ed in sinn. 
( 4 .) A primary difficulty will of course arise in the case of natural selection. Flow 
are we to determine which are the directly and which are the indirectly selected 
organs ? With artificial selection by man, we know vdrich organs have been selected 
fairly well; attention has been paid to colour, size, proportion of parts, &c. Even in 
the case of the medical examination of the recruit, it is chest, lungs, heart, stature, 
Ac., which form the basis of the acceptance or rejection. If the head or foot be not 
absolutely deformed, little if aii}^ attention is paid to them, so with hair-colour, 
probably eye-colour, and a mass of other details. No doubt the direct medical 
selection indirectly selects these, but we could roughly class the selected and non- 
selected organs or characters and investigate the changes in the correlations of the 
latter owing to the indirect selection. But how are we to form these classes in the 
case of natural selection ? 
The investigations may look difficult, and even tfom the standpoint of arithmetic 
appalling, but it seems to me that the differentiation of organs into directly and 
indirectly selected classes is the keynote to the problem of evolution by natural 
selection. 
Let us look at a simple case and see whether it will throw any light on the 
problem of distinguishing between directly and indirectly selected organs. Suppose we 
have two organs only, with means rn.^, standard deviations o-j, (Xo, correlation 
and let the first l^e selected so as to have a mean value riii fi- A],, and standard 
deviation Let No 1 )e the standard deviation of the second organ and 140 the 
correlation of the two organs after selection, and + .vb the mean of the non- 
selected organ. 
Then by (xxxii.) : 
X -2 — 740 — h 
°’i 
and it will be shown later (see p. 23 ) that 
and 
Hence Ave have : 
N,44.. 
cr., — r,o. 
~ cr. 
V 
