400 Tricotylous Races. 
in the fourth generation (1892) and this has been main- 
tained since that time in spite of considerable fluctuation. 
Up till that time my selection had been limited by the 
fact that only hemi-tricotyls had survived to become seed- 
parents. From this point onwards I had both hemi- 
tricotyls and tricotyls in abundance. For the next four 
years I selected in these two directions, and maintained 
a tricotylous and hemi-tricotylous race simultaneously 
(1892-1896), but as no essential difference was mani- 
fested between the two I abandoned the hemi-tricotylous 
*> 
race, as stated in 2, and only went on with the tricotyl- 
ous one. The account which follows relates solely to this. 
But before I proceed with it I wish to call attention 
to some facts relating to the method of culture. The seed 
was sown in sterilized soil in pans ; the soil was not 
manured. As soon as the cotyledons had fully unfolded 
and before, the first leaf was visible, the seedlings were 
recorded. All, or nearly all, of the dicotylous ones were 
destroyed and only the aberrant forms saved. Of the 
latter I chose what I considered to be a sufficient number 
of the strongest specimens, and planted them singly in 
pots with heavily manured soil. The best manures for 
this purpose are nitrogenous ones. If the number per- 
mits it, two tricotyls are put in each pot, of which the 
weaker is subsequently killed out. This transplantation 
takes place in April or May ; the pots are kept under glass 
in the garden until June, during the nights at any rate. 
Then the plants are planted out into the bed at distances 
of from 20 to 30 centimeters, and the larger lateral 
branches are cut off in order that the plants may not 
interfere with one another. The plants are monoecious, 
the male and female flowers standing close together. 
Self-fertilization seems to be the rule; for isolated plants 
