Something similar occurs in plants which exhibit alternation of generations, for instance the ferns, in which the sexual generation, the so-called prothallium or prothallus, contributes nothing to the multiplication of the plant, since only a single egg-cell is developed; and the same is true of the mosses. In both cases multiplication depends solely on the asexual generation, which, as the so-called 'moss-fruit' or 'fern-plant proper,' produces an enormous number of spores, in addition to multiplying by runners.
To sum up: we have seen that self-fertilization does occur in hermaphrodite animals, where otherwise the species would be in danger of extinction, but this is never the sole and exclusive mode of fertilization[25], for hermaphrodite species have always the possibility of securing inter-crossing of individuals, and that in various ways, whether by the intervention of 'primordial males' or by an occasional or a periodic alternation of self-fertilization and mutual fertilization. Pure parthenogenesis enduring through innumerable generations does appear to occur, but in most cases unisexual generations alternate with bi-sexual, so that a stereotyping of the germ-plasm with complete uniformity of ids is obviated.
[25] As to the cases Maupas has brought into notice, of permanent and apparently exclusive self-fertilization in Rhabditidæ (round worms), it seems fair to say that they have not been as yet sufficiently investigated to admit of a secure appreciation of their value in their theoretical bearings. Cf. Arch. Zool. Exper., 3rd ser., vol. viii, 1900.
We must now briefly consider the higher plants with reference to the maintenance of diversity in the germ-plasm through crossing.
We saw in an earlier lecture that most flowers are hermaphrodite, but that they do not fertilize themselves, and are adapted for crossing, since the pollen of one flower is carried by insects to the pistil of another, which cannot be reached by its own pollen, either because it ripens too early or too late, or because the stigma, notwithstanding its proximity, is so placed as to be out of reach of the pollen from the adjacent stamens. I showed, following the fundamental investigations of Sprengel, Charles Darwin, Hermann Müller, and other successors of Darwin, that the flowers may in a sense be regarded as the resultants of the insect-visits, since all their accessory adaptations—large coloured petals, fragrance, nectar, and even little minutiæ of colour and markings (honey-guides)—as well as their detailed shape, as seen in 'landing stages,' corolla tubes, and so on, are only intelligible when we refer their existence to natural selection. We assume that each of these adaptations secured some advantage for the species concerned, and that therefore their first beginnings as slight germinal varieties were accepted, and were brought gradually to their full expression by the united operation of germinal and personal selection. This at least is how we should express ourselves now that we have become acquainted with the factor of germinal selection. The advantage secured by every such improvement in a flower's means of attracting insects is obvious, as soon as it is established that cross-fertilization is more advantageous for the species than self-fertilization.
We have discussed this already; we saw that experiments instituted by Darwin proved that seedlings which had arisen through cross-fertilization were superior to those arising through self-fertilization, and that in many cases the mother-plant itself produced fewer seeds when self-fertilized than when cross-fertilized. This discovery afforded an explanation of the cross-fertilization of flowers by insects which Sprengel had previously observed. We understand how the flowers must have become so adapted through processes of selection that they were unable to fertilize themselves, but attracted insects, and, so to speak, compelled these to dust them with pollen from another plant of the same species. We also understand how self-fertilization remained possible for many flowers in the event of cross-fertilization through insects not being effected, since after a certain period of waiting, a curvature of the stamens or the pistil may take place and lead to the stigma being dusted with the pollen of the same flower. Obviously the development of fewer seeds is preferable to complete sterility. It is a well-known fact that peculiar inconspicuous and closed flowers, designed solely for self-fertilization, may occur along with the open flowers, as in the case of the so-called cleistogamous flowers of the violet (Viola) and the little dead-nettle (Lamium amplexicaule), and the phyletic origin of these becomes intelligible as soon as it is established that cross-fertilization is more advantageous than self-fertilization.
Now, however, it seems as if the fundamental proposition of this theory of flowers will have to be rejected. Not only do the cleistogamous flowers just mentioned exhibit a great fertility, not at all less than that of the open flowers of the same species which are adapted for cross-fertilization, but there is a small number of plants which produce seeds by self-fertilization alone. Thus in Myrmecodia cross-fertilization is absolutely prevented by the fact that the flowers never open, and according to Charles Darwin Ophrys apifera also reproduces by self-fertilization alone, and is nevertheless a thoroughly vigorous plant. There are several other cases of this sort, and particularly among the orchids, though the whole of the structure of their flowers is specially adapted for pollination by insects. Many of them are only rarely visited by insects, some not at all, we know not why, but it is readily intelligible that in such cases they should have adapted themselves to self-fertilization wherever that was possible. For this no great variation was necessary; it was enough that the pollinia, which formerly only became detached from their attachment at a touch or a push from an insect, should free themselves spontaneously. And this, according to Darwin, is what happens, for instance, in Ophrys scolopax, which at Cannes is frequently self-fertilizing. For the development of seed, however, it is not enough that the pollen should reach the stigma; the pollen-grain has to send out its tube and penetrate into the ovary, and in many orchids this does not happen; they are infertile with their own pollen. Various other plants are also non-fertile with their own pollen, for instance the common corydalis, Corydalis cava, or the meadow cuckoo-flower, Cardamine pratensis (Hildebrand).
How are we to reconcile these apparently absolutely contradictory facts? On the one hand, the innumerable devices for securing crossing lead us to conclude that it is necessary, or at least advantageous, and on the other we find a small number of plants which reproduce continually by self-fertilization and yet remain strong and vigorous. And again there are many plants which yield seed when fertilized with their own pollen, and others which remain absolutely sterile in the same circumstances, yielding no seed or very little, and there is indeed one on which its own pollen has the effect of a poison, for if it reaches the stigma the flower dies. If there is anything injurious in self-fertilization (Darwin), we can understand that it will be avoided, but how can it be continued so long in many cases, and even become in others the exclusive method of fertilization without visible evil results?
It seems to me that in these facts, established by observation, the results of two quite different processes have been confused, and that we can only gain clearness by studying them apart from one another; I mean the processes involved in the mechanism of fertilization and those involved in the mingling of the germ-plasms.
In many cases self-fertilization is said to yield less seed and weaker seedlings. Let us for the present take this statement as the basis of our consideration; it does not seem to me conceivable, though here I am not in agreement with views that have been expressed by others, that both effects should depend upon the same causes, for the smaller number of seeds cannot possibly depend upon the mingling of the two parental germ-plasms, and thus not upon the process of amphimixis itself, since the effect of the mingling does not make itself felt until the organism of the offspring is being built up. Of course the plant seed is the embryo of the young plant, but it will hardly be thought probable that its development could be absolutely prevented by the too close relationship of the two germ-cells, and thus the number of the developing seeds cannot depend on the quality of the ids co-operating in the segmentation-nucleus, but presumably on the number of ova awaiting fertilization in the ovary, which are reached by a pollen-tube and then by a paternal sex-nucleus. This again will depend upon the impelling and attracting forces of the pollen-grain on the one hand, and of the stigma and 'embryo-sac' of the flower on the other. In other words, the fertility of a flower with its own pollen will depend upon whether the two products of the flower are adapted for mutual co-operation, and in what degree they are so. We are here dealing not with the primary reactions of the germ-plasms, which are as they are and cannot be varied, but with secondary relations, which may be thus or thus—in short, with adaptations.