Tricotyls as Half Races and Intermediate Races. 355 
as sharply and unalterably separated as, in the first part 
of this volume, we saw was the case in numerous in- 
stances, and especially in the five-leaved race of the red 
clover. Nothing less than a mutation can effect the 
transition between the two, but I have not yet had the 
good fortune to observe such an occurrence. 
In these experiments therefore the differences be- 
tween individuals to which attention has to be paid are 
their hereditary values ; and whether they themselves 
have two, or three or cleft cotyledons is a matter of sec- 
ondary importance. In my cultures the selection of tri- 
cotyls as seed-parents has been the general rule, since 
this practice on the one hand increases the probability 
of excluding specimens with a low hereditary capacity, 
and on the other, of including those with the high ; but 
the increase of this chance is only a small one, as the 
frequent cultures I have made from atavists clearly show 
(see below, 6). 
For a half race the curve describing these values is 
a half curve. The vast majority of individuals have 
either nothing, or little else, but dicotylous offspring; 
and the numbers of individuals with the larger numbers 
of tricotylous offspring decrease rapidly (Fig. 66). These 
curves may be improved in the same way as those of 
other half races, viz., by the selection of individuals with 
the highest value, as we have seen in Ranunculus bul- 
bosiis semiplenus (see 23, of the first part of this 
volume, p. 249). 
Curves describing these values in intermediate races 
usually have their maximum ordinates at 50% ; they are, 
however, liable to be much altered by selection and ex- 
ternal conditions. Fig. 67 is a curve of this kind for 
Oenothera hirtella, whose apex is at about 65%. If 
