The Origin of Crepis and Related Genera 47 



behaves like a distinct species, it probably could not survive 

 under natural conditions unless isolated from the parental 

 species. This was the first experimental verification of the hypo- 

 thesis proposed by Winge,^'' that interspecific hybridization fol- 

 lowed by doubling of the chromosome number might account 

 for the arithmetical series of chromosome numbers which have 

 been shown to exist in various genera of plants. Since then, 

 numerous amphidiploid hybrids between distinct species have 

 been produced experimentally and many of them behave like 

 "good" species although many others are more or less sterile and 

 irregular in reproduction. One of the most convincing examples 

 is that of Raphanobrassica, which was produced by Karpe- 

 chenko^" by crossing Rap h anus sativus with Brassica oleracea. 

 The amphidiploid race obtained is constant and fertile under 

 experimental conditions, and can be crossed with the parental 

 species only with great difficulty; hence it must be considered a 

 potential new species. Furthermore, numerous constant inter- 

 specific hybrids have been discovered in nature, and these pre- 

 sumably arose through amphidiploidy.^'' ^' *° 



In Crepis, cytological evidence has been found to indicate the 

 important role played by interspecific hybridization and amphi- 

 diploidy in the evolution of this genus (pp. 29, 38, 41). It may 

 be objected that the experimental evidence from Crepis does not 

 support this idea. It happens, however, that the three amphidi- 

 ploids thus far obtained in Crepis resulted from crosses between 

 nearly related species, so that there was sufficient homology 

 between the chromosomes of the parents to cause much irregu- 

 larity in distribution of the chromosomes to the gametes, with 

 resulting sterility."^ The fact that amphidiploids were produced 

 from hybrids between such species may be looked upon as sup- 

 porting the hypothesis here proposed. 



