Tricotyloiis Half Races. 383 
can be isolated immediately; if it is not, no amount of 
selection will bring it about. The experiments in which 
isolation succeeded will be described in 6; but the ex- 
periments which failed fall into two groups, according 
to whether it could be decided immediately, or not until 
much later, that an intermediate race could not be raised. 
In the former group fall those cultures in which there 
was no progress at all, or so small a one that I had to 
give them up after three or four generations. For if, 
after this time, a proportion of no more than from 1 to 
2 or at most 4% is reached, or if in the case of higher 
proportions the ratio fluctuates greatly but does not ex- 
hibit a regular increase, how many years of work would 
it take before we can be certain that nothing can be at- 
tained? These briefer experiments will form the sub- 
ject of this section. 
In the second group fall two cultures which I have 
continued for a considerable time, namely Amarantus 
speciosus through nine generations and Scrophularia no- 
dosa through ten. Neither now leaves any hope of ever 
becoming successful (see 5) ; but, as I have already 
said in the first part of this volume (p. 227), it is just 
the experiments in selection that fail, which give us the 
deepest insight into the nature of elementary characters. 
I shall now proceed to the description of my ex- 
periments; and I shall confine myself to those which 
were instituted with the express hope of breeding a 
tricotylous intermediate race, and were continued for 3 
or 4 generations with this sole object in view, until it 
became evident that only a half race was present. The 
extent of the cultures varied greatly, according to the 
importance which I attached to them at the beginning. 
In the first place I shall deal with Ocnothcra rubrinervis. 
