432 Tricotylous Races. 
tetracotylous parent with 51% were planted out. Here 
again both the tricotylous and tetracotylous plantlets were 
used, but this time not in separate lots. Seeds of 37 
plants were saved and sown, and they gave values which 
did not differ essentially from those obtained in the fore- 
going year. They were distributed fairly regularly be- 
tween 16 and 52% and had a mean of 35%, i. e., the 
mean value had undergone a considerable increase. 
Antirrhinum majus. It was with this plant that I 
made one of my first experiments in the production of 
tricotylous races, and the fact that the progress in this 
case was much slower than in later cultures, may well 
be ascribed to the less extensive experience which I had 
at that time. I am now inclined to regard the practice 
of selecting from an insufficient number of seed-parents, 
and also of inadequate manuring as some of the causes. 
Nevertheless I shall describe the experiment because it 
ultimately led to a genuine tricotylous race. On the other 
hand it may not be unimportant to show 7 that, if a repe- 
tition of my tricotylous cultures is made, the attainment 
of the end must not always be expected to be reached 
in two or three generations. 
I have attempted to isolate tricotylous races from two 
varieties of the Snap-dragon. First from the striped 
variety mentioned on page 120 of this volume, and figured 
in Plate I ; but as the success attained with this after 
four generations (1892-1896), was not so great as in 
the other case, and 45% had not been exceeded, I did 
not proceed further with it. Therefore I shall confine 
my description to the latter. It was a dark red half- 
dwarf variety. 
In 1892 I had four tricotylous plants from bought 
seed, and from their seeds obtained in the following 
