It has already been shown that in about 32 out of the 55 genera in the list just referred to, the perfect flowers are irregular; and this implies that they have been specially adapted for fertilisation by insects. Moreover three of the genera with regular flowers are adapted by other means for the same end. Flowers thus constructed are liable during certain seasons to be imperfectly fertilised, namely, when the proper insects are scarce; and it is difficult to avoid the belief that the production of cleistogamic flowers, which ensures under all circumstances a full supply of seed, has been in part determined by the perfect flowers being liable to fail in their fertilisation. But if this determining cause be a real one, it must be of subordinate importance, as four of the genera in the list are fertilised by the wind; and there seems no reason why their perfect flowers should fail to be fertilised more frequently than those in any other anemophilous genus. In contrast with what we here see with respect to the large proportion of the perfect flowers being irregular, one genus alone out of the 38 heterostyled genera described in the previous chapters bears such flowers; yet all these genera are absolutely dependent on insects for their legitimate fertilisation. I know not how to account for this difference in the proportion of the plants bearing regular and irregular flowers in the two classes, unless it be that the heterostyled flowers are already so well adapted for cross-fertilisation, through the position of their stamens and pistils and the difference in power of their two or three kinds of pollen, that any additional adaptation, namely, through the flowers being made irregular, has been rendered superfluous.
Although cleistogamic flowers never fail to yield a large number of seeds, yet the plants bearing them usually produce perfect flowers, either simultaneously or more commonly at a different period; and these are adapted for or admit of cross-fertilisation. From the cases given of the two Indian species of Viola, which produced in this country during several years only cleistogamic flowers, and of the numerous plants of Vandellia and of some plants of Ononis which behaved during one whole season in the same manner, it appears rash to infer from such cases as that of Salvia cleistogama not having produced perfect flowers during five years in Germany (8/31. Dr. Ascherson 'Botanische Zeitung' 1871 page 555.), and of an Aspicarpa not having done so during several years in Paris, that these plants would not bear perfect flowers in their native homes. Von Mohl and several other botanists have repeatedly insisted that as a general rule the perfect flowers produced by cleistogamic plants are sterile; but it has been shown under the head of the several species that this is not the case. The perfect flowers Viola are indeed sterile unless they are visited by bees; but when thus visited they yield the full number of seeds. As far as I have been able to discover there is only one absolute exception to the rule that the perfect flowers are fertile, namely, that of Voandzeia; and in this case we should remember that cultivation often affects injuriously the reproductive organs. Although the perfect flowers of Leersia sometimes yield seeds, yet this occurs so rarely, as far as hitherto observed, that it practically forms a second exception to the rule.
As cleistogamic flowers are invariably fertilised, and as they are produced in large numbers, they yield altogether a much larger supply of seeds than do the perfect flowers on the same plant. But the latter flowers will occasionally be cross-fertilised, and their offspring will thus be invigorated, as we may infer from a wide-spread analogy. But of such invigoration I have only a small amount of direct evidence: two crossed seedlings of Ononis minutissima were put into competition with two seedlings raised from cleistogamic flowers; they were at first all of equal height; the crossed were then slightly beaten; but on the following year they showed the usual superiority of their class, and were to the self-fertilised plants of cleistogamic origin as 100 to 88 in mean height.