spirally Tzvistcd Races. 547 



Twisted stems 37 =34% 



Stems with ternary whorls J2 =11% 



Atavists with decussate arrangement of weaves 58 =55% 



Total 107 



It is very important to notice that the number of 

 twisted stems increased as the result of tlie greater (hs- 

 tances between the plants, not only relatively (i e., per 

 100 individuals), but absolutely (i. e., per square meter). 

 In the third generation there were 50 plants per square 

 meter, of which 4% (1-7%, from 1-4 specimens) were 

 spirally twisted. In the fourth generation, however, 

 there were 37 twisted individuals on 4 square meters or 

 9 per square meter. 



I selected the seven plants with the finest torsions 

 on this bed as seed-parents. They all, moreover, pre- 

 sented local torsions on several branches. I isolated 

 them before they flowered. 



The fifth generation (1893-1894) was not so success- 

 ful. Only 25% with twisted stems were produced. The 

 seed to produce this generation had not as hitherto been 

 sown on the beds but in seed-pans in tlie greenhouse 

 attached to my laboratory. This method has since proved 

 more convenient and safer, and has been employed in the 

 two generations which follow. 



The seed saved in September 1892 was sown in the 

 middle of March 1893. In the middle of ]\Iay the best 

 seedlings were planted separately in 10 centimeter pots 

 with good, richly manured garden soil, and planted in 

 tlie bed in the middle of Alay, at the same distance a]")art 

 as were the ]')lants in the preceding generation {22 i)]ants 

 ])er square meter). In tlie following year (18^4). all 

 the stems develoi)ed and were recorded. The result was 

 as follows : 



