118 BIOMETRY 
and show no regression towards the type of the general 
population to which the group belongs. 
If we were to carry on this conception to the case of 
bisexual inheritance, we should find that the different 
pure lines would become crossed and confused together 
in a way which would be very difficult to disentangle. 
There is no reason to doubt that statistical treatment 
of such a population would yield similar results to 
those actually obtained by biometricians from the 
data at their disposal; and we may notice that a for- 
tuitous mixture of a considerable number of pure 
lines, having slightly different types, would admirably 
fulfil the conditions we have seen to be necessary in 
the case of material, to which methods based upon 
the theory of chance are to be applied. The phe- 
nomena which follow upon the crossing together of 
two or more pure lines have been found, in the majority 
of cases so far studied, to conform to those laws of 
heredity associated with the name of Mendel which 
are explained in Chapter VII. This being the case, 
there appears to be every probability that the theory 
of pure lines, in combination with the method of in- 
heritance referred to, may adequately serve to describe 
those phenomena to account for which the law of 
ancestral inheritance was called into existence. 
The conclusions to which Professor Johannsen’s 
experiments lead him may be summed up as follows : 
Individuals which differ (in size, for example) from 
the mean of a population give rise to offspring which 
differ from that mean value in the same direction 
but to a smaller extent. Selection, therefore, will 
