The Isolation of Tricotylous Intermediate Knees. 423 
SELECTION OF THOROUGHBRED TRICOT YLol'S RACKS. 
HEREDITARY YAI.L'ES OF THE BEST TKICOTYI.oUS SEED-PARENTS OF Till: 
FOURTH GENERATION. 
THIRD 
GENER- 25 30 35 40 45 50 55 60 65 70 75 80 85 
ATION 
Merc -uriahs an nua 55% 122121231 
Silene inflata 55% 10234444711 
It will be seen that regression did not accompany this 
..election, because the parent plants deviated too little as 
yet from the mean of the new race. On the other hand 
the high values of 75-85% were immediately reached, 
and this obviously indicates a considerable advance. These 
instances, taken together with the table which will be 
given on page 439, are sufficient to indicate the method 
by which tricotylous intermediate races, if they are at 
all present, can be isolated and improved. They will 
facilitate the understanding of the detailed description 
of my experiments. These obviously present, according 
to the species investigated, greater or less deviations from 
the instances given, but as a rule they are not essential 
ones. Therefore I shall give these descriptions as briefly 
as possible. 
But before doing so I wish once more to lay emphasis 
on the fact that these cases are thoroughly distinct from 
those in which only half races are present, and from 
which, after several generations, often no more than 2 
to 4% and only very rarely as much as 20% was reached, 
as maximum figures (see table on page 392). 
Oenothera kirtclla. Under this name I cultivate a 
tricotylous race which I obtained by chance from bought 
seed. In 1895 I was growing some samples of seed of 
Oenothera Lciinarckiana and allied species which had 
