Tricotylous Half Races. 385 
12 were hemi-tricotyls and one was a tetracotylous speci- 
men. There were also some few syncotylous ones. The 
ratio from the best seed-parents was no more than from 
0.8% to 1.1%. The tricotylous offspring of these and 
of some with from 0.5 to 0.6% were planted out in the 
same way as in the previous year, provided they were 
strong little plants. This time the inflorescence of each 
plant was guarded against the visits of insects by means 
of a parchment bag and artificially fertilized in the hope 
of thus isolating individuals producing higher propor- 
tions. 
On page 354 I have plotted, in Fig. 66, a curve of the 
harvest of 1894, based on the records made in the spring 
of 1895. There were two seed-parents which manifested 
an advance on the previous year, inasmuch as their ratio 
was 3.5% and 3.7%, but the difference was only a 
very inconsiderable one. In all, there were 87 seed- 
parents. As I had planted the offspring of the individual 
grandparents of 1893 together in groups in 1894, I 
could now make a selection not only between the parents 
but also between the grandparents. The grandchildren 
of those grandparents only, whose offspring had pro- 
duced the highest mean proportion, were planted out. 
By means of such a selection of grandparents the pedi- 
gree becomes an individual one in spite of the size of the 
-ultures, embracing, in each generation the offspring of 
one selected individual only. The method, therefore, 
unites this principle with the greater certainty that can 
be obtained by extensive cultures and a double selection. 
I have also applied it to a number of other cases. 
From one grandparent with 1%, therefore, and from 
six of its offspring with from 0.9% to 2.1% 54 tricotyl- 
ous specimens in all were planted out (1895). From the 
