Tricotyls as Half Races and Intermediate Races. 351 
we rarely get more than low percentages ; but it is obvious 
that we should not expect all the seed-parents in the field 
to furnish the same number of tricotylous seedlings. Com- 
mercial seed is almost certain to be a mixture and to 
yield a mean proportion, which may perhaps be much 
lower than the seed of those specimens would yield, 
which were the richest in tricotylous offspring. It fol- 
lows from this that we can only form a proper estimate 
if we select a group of plants among our first sowing 
and harvest their seeds separately, if possible after tak- 
ing the precaution of preventing too much cross-fertili- 
zation by insects. It is obvious that in such a culture a 
choice will have to be made, and as a rule the tricotyls, 
using this term in its widest sense, will be selected. From 
the separate harvest of these selected specimens, individ- 
ual hereditary coefficients may then be obtained. 
In spite of every precaution, the choice of the plants 
will depend mainly on chance, for, as we shall see later, 
tricotyls are by no means more likely to reproduce the 
anomaly than some of the dicotyls. On the contrary, 
plants much better in this respect sometimes occur among 
these ; but we have at present no other means of dis- 
covering them except by growing their seed. Experience 
however has shown that if we allow a fairly large group 
to ripen their seed we may become independent of chance 
in so far that we can be fairly sure that at least some 
seedparents will give a relatively high proportion, pro- 
vided of course that the original mixture contains such 
amongst its components. It will soon be seen that in 
spite of all apparent intermediate degrees, two main 
groups can be distinguished. In one of them we find 
onlv low individual values, from to 3%, or very little 
/ / 
over ; in the second, on the other hand, besides these 
