544 Heritable Spiral Torsions. 



tnre, which extended over 18 years, I shall give here a 

 brief summary of the nine successive generations. 



The mean of the six generations was therefore about 

 35%. 



We see from this table that the continued isolation 

 and selection of the finest torsions maintained the race 

 at its level, but did not improve it to any considerable 

 extent. 



Two points about this result should be noted. In the 

 first place the percentage figures given do not denote the 

 proportion of twisted individuals, but relate solely to 

 those with torsions on the main stem. In some experi- 

 ments, however, instead of pulling up the atavists, I only 

 removed all their inflorescences before the flowers opened, 

 and by this means I found that several, and often many of 

 them, had the power of developing more or less fine local 

 torsions on their lateral branches. In 1902 I made as 

 accurate an account of these as possible, and found 71 

 of them or 14%. Of twisted i..dividuals there w^ere, 

 therefore, 41-]- 14=55% altogether, i. e., somewhat more 

 than half the individuals produced torsions. It seems 

 of some interest to note that this percentage is the same 



^ The percentage figures for the 5th, 6th and 7th generations are 

 hardly decisive on account of the small size of the cultures. 



