394 Tricotylous Races. 
several generations. But, at the beginning of our cen- 
tury, we stand only on the threshold on which systematic 
botany must be raised from a comparative to an experi- 
mental science. 
The experiments described in the foregoing section 
(4) cover four generations as a rule, i. e., a selection 
of tricotylous plants four times repeated, and thrice re- 
peated for those with apparently the highest values. It 
may, however, be allowable to suppose that stray un- 
favorable individuals occurred amongst the selected ones 
and that a selection of longer duration might possibly be 
crowned with success. 
In order to determine this point I have, as I have 
stated above on page 383, continued the experiment with 
two species, to which I have lately added a third, through 
about ten generations. I devoted every possible care to 
the selection and carried it out on as large a scale as 
could be desired. The result was a genuine progress 
which amounted in both cases from one or two per cent to 
a value which reached in the seeds of some rare seed- 
parents, even as much as 25%. But from the character 
of an intermediate race with a mean value of 50% the 
races are still far removed, and every circumstance points 
to the conclusion that it is simply impossible to reach 
this by the method (as yet the only available one) which 
was employed. 
The two plants used in these experiments were Aina- 
rantns speciosus and Scrophularia nodosa. The former 
is an annual garden plant, much in favor on account of 
its height and its red foliage (Fig. 83) ; the second is a 
wild perennial species which is very common in this 
neighborhood. It flowers in its first year and can be 
easily cultivated as an annual. In the previous section 
