392 
Tjricotylous Races. 
tion of tricotyls was 1 to 2% and reached 3 and 4% 
so rarely that these numbers must perhaps be regarded as 
the extreme results of the errors of observation which 
are bound to occur in such countings. In two species 
the ratio was about 10 to 15%, but these were both 
perennial forms which, in my annual cultures set but little 
seed. They were Lychnis fulgens, from 1892 to 1895, 
with 5- -13- -11 and 8 19% tricotyls and Penstemon 
gentianoides in the same years with 3 12- -11 and 15%. 
They should perhaps be excluded from further considera- 
tion. Summarizing my data therefore (with the omis- 
sion of these cases) we obtain a very uniform picture 
of the inheritance of tricotyly in half races under con- 
tinued selection. 
INHERITANCE OF TRICOTYLY IN HALF RACES UNDER 
CONTINUAL SELECTION. 
L 
MAXIMUM VALUES IN THE 
v W O 
>^ ^^ 
SPRINGS OF 
fe W ^ 
' 
s 
o 
1890 
1891 
1892 
1893 
1894 
1895 
1896 
Oenothera rubrinervis . 
1892 



2.8 
1.1 
3.7 
2 
Chenopodium album . 
1889 
1 
1 





Dracocephalum moldavicum 
1892 



0.2 
0.4 
0.4 
0.7 
Polygomnn Convolvulus 
1888 
1 
2.4 
2 
2 
2.8 
2 

Silene conic a 
1892 
__ 


3 
1 
2 
4 
" conoidea .... 
1892 



3 
3 
4 
Spinacia oleracea .... 
1892 



0.6 
2 
2 
4 
Thus we see that tricotylous half races exist which 
even under the most stringent selection can produce only 
small percentages of this anomaly. So far as we can 
conclude from indirect data, half races of this kind ap- 
pear to be widely distributed in the vegetable kingdom. 
Samples of seed, whether they be bought or collected in 
