446 Tricotylous Races. 
the earlier seeds give different values from those given 
by the later ones? Has the character of the year in 
which the harvest is made any such influence amongst 
perennial plants ? Obviously these questions must be 
answered if the individual harvests are to be limited in 
the interest of the experiment. 
In the last two sections of the first part of this vol- 
ume we have seen that semi-latent characters manifest 
a certain periodicity in their distribution over the plant, 
and also that the choice of seeds on the plant plays some 
part in the process of selection. The question is, how 
do the tricotylous races behave in this respect? 
The general rule seems to be that a bud, whether of 
a branch or of a flower, is more likely to reproduce an 
anomaly, the more vigorous it is (p. 324). Therefore 
with an increase in the degree of branching the expecta- 
tion of the occurrence of anomalies decreases (p. 329). 
The first or lower fruits of an inflorescence will be 
stronger than the higher ones, and the fruits on the 
weak lateral sprigs of the primary and secondary branches 
may as well be thrown away. 
There is no ground for supposing that the flower buds 
behave differently from the vegetative parts. The best 
instance of the phenomena in question is afforded by 
profusely branched specimens of the twisted race. Dip- 
saciis sylvestris forsus. The torsion affects the middle 
portion of the main stem, but neither its upper nor lower 
extremity. Tt is repeated on the strongest branches situ- 
ated on the middle of the stem, and on these only in 
their middle parts, and excluding the weaker ones. The 
stronger a stem or a branch is, the greater is the extent 
of the twisted part. Tn the branches it is always con- 
fined to sinHe internodes, whereas the stem may fre- 
