68 VIOLA. [CHAP. 



from the anthers ; and it is therefore important that 

 the pollen should not easily be detached and carried 

 away by the wind. In the present case, on the con- 

 trary, it is desirable that it should be looser and drier, 

 so that it may easily fall into the space between the 

 stamens and the pistil. If it remained attached to 

 the anther, it would not be touched by the bee, and 

 the flower would remain unfertilised. 



3. Why is the base of the style so thin ? In order 

 that the bee may be more easily able to bend the style. 



4. Why is the base of the style bent ? For the 

 same reason. The result of the curvature is that 

 the pistil is much more easily bent than would be 

 the case if the style were straight. 



5. Finally, why does the membranous termina- 

 tion of the upper filament overlap the correspond- 

 ing portions of the two middle stamens ? Because 

 this enables the bee to move the pistil, and thereby 

 to set free the pollen more easily than would be the 

 case under the reverse arrangement. 



In Viola tricolor, the form of the stigma is very 

 different from that of V. canina^ but the reason of 

 the difference has not been satisfactorily explained. 

 Mr. Bennett considers that this species is fertilised by 

 Thrips. Mr. Darwin, however, has satisfied himself 

 that when bees are excluded, it is comparatively 

 infertile, and he has favoured me with the following 

 memorandum on the subject. 



"When," he says, " I formerly covered up a fine, 

 large, cultivated variety, it set only 18 capsules, and 

 most of them contained very few good seeds, several 

 from only I to 3 ; whereas an equally fine uncovered 

 plant, growing close by, produced 105 fine capsules. 



