Tricotylous Races Do Not Arise by Selection. 409 
in the best cases 2.4 to 4.1 and 5.4%. The counts were 
made on 700, 800 and 2000 seedlings. These three plants 
had, however, decussate leaves on their main stems, and 
this shows that the disposition of leaves on the stem is 
of subordinate value in selection. In the later genera- 
tions I have always selected the ternary individuals where 
possible, but have not found that they are any better than 
the others. 
In the summer of 1894 tricotylous seedlings of the 
seed-parent with 5.4% were planted out. I saved the 
seeds of 25 plants separately, and in the following spring 
examined from 300 to 500 seedlings in each crop. The 
values were essentially the same as in the previous year, 
and varied between 0.5% and 5.5% with a mean of 2% ; 
and the five best gave 3.6 3.8 4.0 4.2 and 5.5%; 
the three last values are based on counts of 1000, 1500 and 
2000 seedlings. It was only the offspring of these parents 
that were planted out in 1895, and only those with three 
cotyledons and with a first whorl of three leaves (Fig. 
73, p. 366). 
By planting them out in three groups in the following 
spring, a selection was made possible which related to the 
grandparents. Those with 4.0 to 4.2% gave offspring 
with 1.3 to 5% with a mean of 3%; (from 12 parents 
and from counts of groups of 300 seedlings). The 
grandparent with 5.5% proved to be considerably better. 
Its twelve offspring had 2 to 8%, with a mean of 4.5%. 
Seedlings of this group only were planted out in 1896 
and only tricotylous specimens with ternary lower whorls, 
from parents with 6 6 and 8%. They were in all 72 
strong plants each of which gave a sufficient harvest for 
the calculation of their hereditary values in groups of 
300 seedlings. The parent with 8% gave 2 to 15%, with 
