442 Tricotylons Races, 
which is also present, for hardly ever is the intermediate 
race found pure by itself. As far as I know, at least, 
there is no species of plants which has so much as half 
of its seedlings showing three cotyledons, without being 
selected. If we study the process of separation by the 
statistical method, w T e find that two curves, a half curve 
with an apex at 0, and a bilateral curve with an apex at 
50 55%, can be distinguished from one another. Some- 
times in one or other of the transitional generations 
both curves can be more or less clearly seen side by side, 
constituting a so-called dimorphic curve. I have fre- 
quently observed this in these experiments, and have, in 
some cases referred to it. 
Dimorphic curves of this kind are best obtained in the 
transitional generations by planting out dicotylous as well 
as tricotylous seedlings ; for, as was mentioned before, 
it is very likely that many dicotyls will belong to the 
half race and most of the tricotyls to the intermediate 
race. But if the latter is once isolated, all the individuals 
belong to it, independently of the question whether they 
have two or three or cleft cotyledons. In this stage a 
dimorphism of this kind is no longer to be expected, un- 
less selection is continued in two different directions. 1 
As an instance of this let me cite the case of my cul- 
ture of Mercurialis annua in 1895 (see the table on page 
439), inasmuch as this species, being dioecious, would 
be expected to exhibit a levelling of the differential char- 
acters. The plant with a value of 55% in 1894 could 
obviously have been partly cross-fertilized. Its offspring 
had in their seeds the following hereditary values : 
1 With regard to this question compare the analogous experi- 
ments with H "licnitliits annuns syncotylcus (II. n). 
