Chap. VII.] THEORY OF NATURAL SELECTION. 269 



some cases, and in other cases the lower ones alone are 

 fertilised; and he suggests that this probably depends 

 on the direction in which the pollen-tubes enter the 

 ovarium. If so, the position of the ovules, even when 

 one is erect and the other suspended within the same 

 ovarium, would follow from the selection of any slight 

 deviations in position which favoured their fertilisation, 

 and the production of seed. 



Several plants belonging to distinct orders habitually 

 produce flowers of two kinds, — the one open of the 

 ordinary structure, the other closed and imperfect. 

 These two kinds of flowers sometimes differ wonder- 

 fully in structure, yet may be seen to graduate into 

 each other on the same plant. The ordinary and open 

 flowers can be intercrossed; and the benefits which 

 certainly are derived from this process are thus secured. 

 The closed and imperfect flowers are, however, mani- 

 festly of high importance, as they yield with the utmost 

 safety a large stock of seed, with the expenditure of 

 wonderfully little pollen. The two kinds of flowers 

 often differ much, as just stated, in structure. The 

 petals in the imperfect flowers almost always consist of 

 mere rudiments, and the pollen-grains are reduced in 

 diameter. In Ononis columnas five of the alternate 

 stamens are rudimentary; and in some species of Viola 

 three stamens are in this state, two retaining their 

 proper function, but being of very small size. In six 

 out of thirty of the closed flowers in an Indian violet 

 (name unknown, for the plants have never produced 

 with me perfect flowers), the sepals are reduced from 

 the normal number of five to three. .In one section of 

 the Malpighiaceae the closed flowers, according to A. de 

 Jussieu, are still further modified, for the five stamens 



