Tricotylous Races Do Not Arise by Selection. 395 
(4) these two species were omitted from consideration 
in order to describe the experiments here in their en- 
tirety. 
The hope which I cherished at the beginning of these 
experiments has not been fulfilled, it is true; but I think 
that a brief notice of it will serve a useful purpose. The 
present form of the theory of selection would justify 
the expectation that a continued selection of the tricotyl- 
ous individuals would result in a race which should, 
year after year, produce tricotyls in continually increas- 
ing quantities, until ultimately a new variety or sub- 
species would arise, composed solely of such individuals. 
This form of the theory is very accommodating. If we 
have regard to the law of regression (Vol. I, p. 83), the 
mean of the race always lags further behind the indi- 
viduals which have been and are to be selected ; so that, 
as a matter of fact we never attain to the type of a new 
and constant race. But if we neglect this law, as is now 
frequently done, we might expect that continual and 
uniform progress, which alone could account on the 
ground of the theory of selection for the origin of spe- 
cies in the vegetable and animal kingdom. And lastly 
we might assume an increase of variability in the chosen 
direction by means of selection, an hypothesis which, as 
I have shown in the first part (p. 9), is entirely unsup- 
ported by evidence. 
The first of these hypotheses would lead us to ex- 
pect a variable tricotylous race, never becoming perfectly 
constant, a thoroughbred race in the agricultural sense 
of the word. The second would lead us to expect a con- 
tinuous and uninterrupted increase in tricotylous indi- 
viduals resulting in a constant tricotylous subspecies. 
The third would point to a gradual acceleration of the 
