388 MEANS OF CROSS-FERTILISATION. Chap. X. 



fertilisation for a long series of generations. I think 

 so, not from the evil which manifestly follows from 

 self-fertilisation, in many cases even in the first gene- 

 ration, as with Viola tricolor, Sarothanmus, Nemo- 

 phila, Cyclamen, &c. ; nor from the probability of the 

 evil increasing after several generations, for on this 

 latter head I have not sufficient evidence, owing to the 

 manner in which my experiments were conducted. But 

 if plants bearing small and inconspicuous flowers were 

 not occasionally intercrossed, and did not profit by the 

 process, all their flowers would probably have been 

 rendered cleistogamic, as they would thus have largely 

 benefited by having to produce only a small quantity 

 of safely-protected pollen. In coming to this con- 

 clusion, I have been guided by the frequency with 

 which plants belonging to distinct orders have been 

 rendered cleistogamic. But I can hear of no instance 

 of a species with all its flowers rendered permanently 

 cleistogamic. Leersia makes the nearest approach to 

 this state ; but as already stated, it has been known 

 to produce perfect flowers in one part of Germany. 

 Some other plants of the cleistogamic class, for instance 

 Aspicarpa, have failed to produce perfect flowers during 

 several years in a hothouse ; but it does not follow that 

 they would fail to do so in their native country, any 

 more than with a Vandellia and Viola, which with me pro- 

 duced only cleistogamic flowers during certain years.* 

 Plants belonging to this class commonly bear both 

 kinds of flowers every season, and the perfect flowers 

 of Viola canina vield fine capsules, but only when 

 visited by bees. We have also seen that the seedlings 

 of Ononis minutissima, raised from the perfect flowers 

 fertilised with pollen from another plant, were finer 



* These cases are given in ch. viii. of iny ' Different Forms of Flowers,' 



