396 Tricotylous Races. 
process. As a matter of fact neither one nor the other 
has occurred. If we would speak of a thoroughbred 
race I only obtained a thoroughbred half race with a 
mean value of no more than 10 to 15% which remains 
dependent on the selection of seed-parents with about 
25%, and is perhaps capable of some very small further 
improvement. The second hypothesis is so remote from 
facts that as yet it hardly admits of being tested ; my 
experiments at any rate, lend no support to it. Rather 
might they be taken to be in favor of the third view ; 
for the progress does actually seem to become gradually 
more rapid after the first few years. But then it should 
be remembered that selection is at first a very difficult 
matter, the tricotyls being still very rare, and for a large 
part delicate and unfit for further cultivation. In sub- 
sequent years there are hundreds of tricotyls from which 
the strongest may be selected ; and we can even limit 
ourselves to the best specimens produced by the best 
parents and grandparents, and thus carry out a much 
more stringent selection. In reality the acceleration of 
the progress is thus brought about by a practical im- 
provement in method and not by a biological increase in 
variability. 
At first I entertained the hope that even if these ex- 
pectations were not justified some relation between varia- 
bility and mutability might perhaps exist. 1 I imagined 
that the capacity for producing mutations might be con- 
ditioned by external influences and therefore might it- 
self be a variable character. The diversity amongst the 
mutation-coefficients of my Oenotheras seemed to sup- 
port this view. 2 Moreover the external conditions which 
1 See my Intraccllular Pangenesis. 
2 See Vol. I, p. 337. 
