192 MH. C. DABWIN ON THE SEXUAL EEIATIOITS OF 



general grounds of my belief that with all organic beings distinct in- 

 dividuals at least occasionally cross together, and reciprocal dimor- 

 phism is plainly one most efficient means for ensuring this result. 



and in the pollen-tubes which proceed from the grains within the anthers in 

 V. canina, and from within the lower anthers of Oxalis acetoseJla, having the 

 wonderful power of directing their course to the stigma. If these plants had 

 produced the minute closed flowers alone, the proof would have been perfect 

 that they could never have crossed with other iiulividuals. I am aware that 

 in some of these cases it has been stated that the perfect flowers never produce 

 any seed ; as far as jimphicarfma is concerned, I hear from Professor Asa Gray 

 that the petaliferous flowers certainly sometimes yield seed. The completely 

 enclosed flowers of that curious grass, the Leersia on/zoides, as described by M. 

 Duval-Jouve (BuU. Soc. Bot. de France, tom. x. 1863, p. 194), apparently offer 

 the best case of perpetual self-fertilization ; for when perfect flowers arc pro- 

 truded from the culms, they are, as far as is yet known, always sterile. In a 

 number of plants kept by me in pots in water, not one single perfect flower has 

 protruded, but the enclosed flowers produced plenty of seed. Without wishing 

 to throw any doubt on M. Duval- Jouve's excellent observations, I may add that 

 witli the enclosed flowers borne by my plants, the act of fertiUzation, that is, 

 the penetration of the stigma by the poUen-tubes, took place in the air and not 

 in fluid within tlie glumes. With the exception of the Leersia^ as the case now 

 stands, I cannot see how the production of the small, imperfect flowers invali- 

 dates my doctrine that no species is perpetually self-fertilized, more than the 

 multiplication of many plants by bulbs, stolons, &c. As I observe that the pro- 

 duction of seed by the perfect flowers of Viola is spoken of as something capri- 

 cious and accidental, I may state that, although it varies much in diflerent years, 

 it depends exclusively on the visits of bees ; I ascertained this by marking many 

 flowers thus visited, and finding that they produced capsules, and by covering up 

 many flowers which (excepting a few that I artificially fertiUzed) did not, when 

 thus protected, produce a single capsule. After bees have visited these flowers, the 

 pollen may be seen scattered on the papiUfe and on the stigma itself, and they can 

 liardly fail thus to cross distinct individuals. These remarks apply to V. canina, 

 Mrta, and odorata ; with V. tricolor the case is somewhat different ; but I 

 must not enlarge any more on this subject. The production by so many plant.s 

 of perfect and expanded, as well as of imperfect and closed flowers, seems to me 

 to throw much light on many points; it shows how extraordinarily little 

 pollen is necessary for fuU fertilization, for I ascertained with V. canina that the 

 perfect and imperfect flowers (tlie latter producing so few pollen-grains) yielded 

 the same average number of seeds ; it shows us that fertilization can be perfected 

 in closed flowers ; it shows us that large, liighly coloured petals, perfume, and 

 the secretion of nectar are by no means indispensable for this act, even in those 

 species which properly possess these characters. It seems to me that the neces- 

 sity of an occasional cross with a distinct individual of the same species explains 

 the universal presence of at least some expanded flowers, at the expense of injury 

 from rain and the loss of much pollen by innumerable pollen-robbing insects ; 

 it explains the enormous superfluity of pollen from its Uabdity to loss from these 

 causes and during conveyance from flower to flower ; it explains the use of a 

 gaily coloured corolla, perfume, and nectar, namely, to attract insects, except in 

 those comparatively few eases in which wmd is the agent, and in these the last- 

 named attributes are deficient. 



