spirally Tivistcd Races. 543 



malformations, as soon as the culture was given sufficient 

 room, nothing hut small local tvvistings occurred in this 

 case, in spite of highly favorahle conditions of growth. 

 From this I concluded that an eversporting variety anal- 

 ogous to my Dipsacus sylvcstvis torsiis cannot be raised 

 from these seeds. 



§ 20. SPIRALLY TWISTED RACES. 



The often mentioned race of Dipsacus sylirstris tar- 

 sus (iM'gs. 123-125, pp. 529-534, and Plate VI) has now 

 completed its ninth generation. It consists of about 40'/r 

 of individuals with fine torsions and about 60% of ata- 

 vists with, as a rule, decussate, bu^, occasionally ternary, 

 arrangement of the leaves. The atavists and the twisted 

 plants with incomplete or interrupted spirals have alwavs 

 been destroyed before flowering. The completely twisted 

 individuals, however, were left to be fertilized bv one 

 another through the agency of insects. 



This very stringent process of selection has had the 

 result of maintaining the race at a fairly constant level. 

 During the first two generations I did not know the proper 

 conditions and sowed the seed much too thick, and con- 

 sequently only obtained twisted individuals at the edges 

 of the beds, and even here in small numbers. Rut with 

 the improvement of these conditions the proportit^n of 

 twisted stems at once mounted to 34%, and the heredi- 

 tary value has since remained about the same, althougli 

 subject to fluctuation due to the greater or less favorable 

 climatic conditions obtaining in successive years. But 

 no advance could be discovered. 



In order to exhibit graphically the result of this cul- 



