552 Heritable Spiral Torsions. 
20 over smaller portions. This means a proportion of 
about \% of the total number of the stems. The 33 
best torsions occurred on ten plants of which five had 12, 
6, 5, 3 and 2 twisted shoots, and the remaining five, one 
such anomaly each. The 20 smaller abnormalities were 
distributed at random over the plants in such a way that 
about 10% of the individuals had developed the anomaly. 
It is worth remarking that the most abnormal individuals 
occurred chiefly on the most sunny edge of the bed. 
The selected individuals mentioned were not allowed 
to flower before all the remaining plants had been re- 
moved. In the spring of 1899 I only sowed the seeds of 
the plant with 12 torsions. I planted out 180 individ- 
uals, at the same distances apart and also in other respects 
under exactly similar conditions as those which obtained 
f 
in the preceding generation. They flowered in the sum- 
mer of 1900. At the beginning of the flowering period 
I had 2246 stems, in 1246 of which the arrangement of 
the leaves was decussate, whilst in 414 it was ternary. 
Further, there were 227 stems which were twisted over 
more than half their length and 359 which were twisted 
in the upper part only, i. e.,26% twisted shoots as against 
1 % in the preceding generation, on the average 3 torsions 
per plant. According to a rough estimate well over half 
the individuals had produced one or more twisted stems. 
Two generations, therefore, had sufficed for the iso- 
lation of a genuine intermediate race from an original 
insect-fertilized sample of seed. 
The other race referred to above occurred in Vise aria 
oculata (Lychnis Coeli-rosa), a favorite garden plant. 
The plant is annual and its culture easy. In twisting, 
its stems become much shortened, and the plants become 
low, but they flowered freely (Fig. 130). In 1897 in a 
