482 'DIFFERENT FORMS OF FLOWERS [1862. 



[The following is an extract from the letter given in 

 part at p. 477, and refers to Dr. Gray's article on the sexual 

 differences of plants :] 



C. Darwin to Asa Gray. 



November 26 [1862]. 



.... You will think that I am in the most unpleasant, 

 contradictory, fractious humour, when I tell you that I do 

 not like your term of *' precocious fertilisation " for your 

 second class of dimorphism [/. e. for cleistogamic fertilisa- 

 tion]. If I can trust my memory, the state of the corolla, of 

 the stigma, and the pollen-grains is different from the state 

 of the parts in the bud ; that they are in a condition of spe- 

 cial modification. But upon my life I am ashamed of myself 

 to differ so much from my betters on this head. The tempo- 

 rary theory* which I have formed on this class of dimorphism, 

 just to guide experiment, is that the perfect flowers can only 

 be perfectly fertilised by insects, and are in this case abun- 

 dantly crossed ; but that the flowers are not always, especially 

 in early spring, visited enough by insects, and therefore the 

 little imperfect self-fertilising flowers are developed to ensure 

 a sufficiency of seed for present generations. Viola canina 

 is sterile, when not visited by insects, but when so visited 

 forms plenty of seed. I infer from the structure of three 

 or four forms of Balsaminece.^ that these require insects ; at 

 least there is almost as plain adaptation to insects as in the 

 Orchids. I have Oxalis acetosella ready in pots for experi- 

 ment next spring ; and I fear this will upset my little theory. 

 . . . Campanula carpathica^ as I found this summer, is abso- 

 lutely sterile if insects are excluded. Specularia speculum is 

 fairly fertile when enclosed ; and this seemed to me to be 

 partially effected by the frequent closing of the flower ; the 

 inward angular folds of the corolla corresponding with the 

 clefts of the open stigma, and in this action pushing pollen 



* 



This view is novv^ generally accepted. 



