VIEWS OF DE VRIES ON SELECTION 



73 



thrips, the plant being scarcely capable of self-pollination. 

 This explains why constant selection is required to maintain 

 a high standard. Hybridization constantly occurs and for 

 this reason fully stable types cannot be obtained. 



De Vries is also led to adverse conclusions concerning se- 

 lection as an agency in producing racial changes by experi- 

 ments of his own, one of the most extensive of which was an 



Fig. 14. Variation of the buttercup (Ranunculus bulbosus) in number 

 of petals preceding and following selection. H 1887, variation curve of 

 unselected race. E 1891 and 1892, curves for successive generations of the 

 selected race. A 1891, curve for parent plants of the 1892 generation. 

 (After De Vries.) 



attempt to increase by selection the number of petals in the 

 common meadow buttercup {Ranunculus bulbos^is). This 

 regularly has five-petaled flowers, but an occasional flower 

 contains one or more extra petals. See Fig. 14. When this 

 plant was cultivated in his garden, De Vries found the aver- 

 age number of petals to be 5.6. After five successive selec- 

 tions the average was raised to 8.6, the upper limit of 

 variation from eight to thirty-one, and the mode (or com- 

 monest condition) from five to nine. De Vries concludes that 

 the change thus produced could Be maintained only by con- 

 tinued selection, and that further progress could probably 

 not be made. This conclusion seems to me unwarranted, but 

 I state it as illustrative of the general view of De Vries, who 

 maintains that when a permanent racial change occurs it is 

 due to something different from fluctuating variability, viz., 

 to a discontinuous variation or sport, a process which De Vries 



