Partial Variability of Tricotyly. 445 
self has two or three or merely cleft seed-leaves is a 
question of minor importance, and does no more than 
furnish a suggestion in starting the experiment. On 
the average, atavists and hemi-tricotyls are, without 
doubt, of less value in this respect than tricotyls and 
tetracotyls; but the differences are too small to warrant 
a choice of seed-parents on this ground. 
The determination of the hereditary value depends, 
therefore, largely on the extent of the crops. The larger 
this is for any one individual, the more closely will the 
value found correspond to that which would be derived 
from an imaginary harvest, consisting of an infinite num- 
ber of seeds. Therefore, the best plan would be to make 
the plants grow as vigorously as possible, and to save 
the seeds of all fruits from as many branches and sprigs 
as possible. In practice, however, this cannot be done, 
since it is far more important to grow as many indi- 
viduals as possible in a given space. The more numerous 
the individuals are, the more stringent the selection be- 
comes ; and, what is far more interesting, at least in 
working with mixtures, the greater is the prospect of 
finding any desired particular sort. Unbranched, or al- 
most unbranched, plants with terminal inflorescences only, 
obviously need much less room than much branched in- 
dividuals ; and larger harvests require relatively too much 
time to gather. The best plan, therefore, would be to 
collect only just so much seed from each individual as is 
necessary for sowing in the following year. 
These considerations suggest the question whether 
different groups of fruits are alike in regard to hereditary 
capacity or whether they exhibit, perhaps, considerable 
differences. Shall we find this capacity to be different on 
the branches from what it is on the main stem? Will 
