375 
manifested in the sober realm of fact. Moreover, among 
methods possible in nature, those which are the easiest 
and the cheapest will probably prove the most efficient 
in the long run. But however this may be, it can be 
shown that, taking the processes of Natural Selection for 
granted, there are at the least three methods of isolating 
similar variants, the emergence and successful operation 
of which in nature cannot be prevented by Natural 
Selection. 
But all these methods of isolating for breeding purposes, 
similar birth variations, have one insuperable difficulty 
attaching to them. They all go on the assumption that 
the variations necessarily associated with reproduction can 
be produced pure and simple and altogether unaffected by 
the external conditions. In the next chapter we shall 
show that this is impossible. If that were possible, I con 
tend that it would be just as easy to isolate favourable 
birth variations by physical isolation, by social segregation 
or by physiological differentiation, as by life and death. 
As it is not possible, such contention is only an argumen 
tum ad hominem which it is useless to pursue, because it is 
based upon assumptions which do not correspond with the 
facts of nature. It may not be unprofitable, however, if we 
endeavour to show how difficult it is to suppose that 
physical isolation, social segregation and physiological 
differentiation can act as the agents of selection with 
respect to purely sexual variations; and, on the other 
hand, how powerful their action may be, as transforming 
influences, if we allow for the direct action of outward 
conditions. 
In seeking for possible methods of “selection ” in nature, 
the first thing which strikes us is to ask whether there is 
in nature anything analogous to that physical separation 
by which artificial selection is, for the most part, effected.