416 Tricotylous Races. 
each plant are sufficient, and this is a great advantage 
as compared with the laborious process of harvesting in 
Scrophularia. 
My culture began in the summer of 1896. At that 
time the prospect of obtaining an intermediate race in 
the other two races had already become very remote. 
From some plants in our botanical garden I saved seeds 
which produced 13 tricotyls, 4 hemi-tricotyls and one 
tetracotyl amongst 300 seedlings, i. e., 6%. This figure 
was considerably higher than most of the values of my 
half races mentioned on page 392, and about as high 
as the value attained by Amarantus and Scrophularia at 
that time. At the same time six other species of Oeno- 
thcra were tested with reference to their production of 
tricotyls. I found from to 1 and 2% and therefore 
selected the 0. Berteriana. In the spring of 1897, tri- 
cotyls only were potted singly, and planted out in July 
in the beds. Their hereditary values were determined in 
the following year in the case of 1 5 plants. They fluctu- 
ated between 1 and 12% and exhibited a mean of 4%. 
In the following generation, 1898, I was able to plant out 
about 60 strong tricotyls, and since that time I have con- 
ducted the cultures on this scale with only slight modi- 
fications. 15 tricotyls from the plant with 12%, and 45 
from those with 6 to 7% were planted out. The former 
group, however, proved to be no better than the latter. 
The values of these plants varied from 2 to 16%, with 
a mean of 4%. The best of the five other groups, how- 
ever, had values varying from 6 to 16%, with a mean 
of 8% (from 10 separate crops). The remainder varied 
between 4 and 11, with a mean of 6%. I selected the 
three best parents whose values were 15, 16 and 16%, 
and I planted out the strongest tricotylous offspring of 
