For other illustrations see multiplication of whorls, petalody; see also Moquin, loc. cit., p. 350. Engelmann, loc. cit., p. 20, § 18. Cramer, loc. cit., p. 25.

Polyphylly of the andrœcium.—An increased number of stamens frequently accompanies the corresponding alterations in other whorls, and seems, if anything, to be more frequent among monocotyledonous plants than among dicotyledonous ones; thus, we occasionally find tetramerous flowers in Crocus, Hyacinthus, Tulipa, Iris, Tigridia, &c., and more rarely in Yucca (Y. flexilis[401]).

The increased number of stamens in a single whorl may result from a development of organs usually suppressed, and constitute a form of regular peloria as in Linaria, wherein a fifth stamen is occasionally met with. Among normally didynamous plants such numerical restitution, so to speak, is not unusual; thus, in Veronica four and five stamens occur. Fresenius has seen five stamens in Lamium, Mentha, Chelone;[402] Bentham in Melittis, and other instances are cited under the head of peloria. Chorisis may also serve to account for some of these cases; thus, Eichler[403] figures a flower of Matthiola annua with five long stamens instead of four; one of the long pairs of stamens has here undergone a greater degree of repetition than usual. De Candolle[404] cites and figures a curious form of Capsella Bursa-pastoris sent him by Jacquin, and which was to some extent reproduced by seed. In the flowers of this variety there were no petals, but ten stamens; hence De Candolle inferred that the petals were here replaced by stamens, but Moquin[405] objects, and with justice, to this view, as the ten stamens are all on the same line; he considers the additional stamens to be the result of chorisis. Buchenau[406] mentions the presence of seven stamens in another Crucifer, Ionopsidium acaule. Here the supernumerary organ was placed between two of the long stamens. The effect of chorisis in producing an augmentation of parts is well seen in some plants that have some of their flowers provided with staminodes or abortive stamens, and others with clusters or phalanges of perfect stamens. Thus, in the female flowers of Liquidambar there are five small staminodes without anthers, whereas in the male flower the stamens are numerous and grouped together in phalanges, so that the relation of simple to compound stamens is in this case readily seen, as also in many Malvaceæ, Sterculiaceæ, Byttneriaceæ, Tiliaceæ, and Myrtaceæ. It is probably the idea of splitting or dilamination involved in the word chorisis that has led many English botanists to hesitate about accepting the notion. Had they looked upon the process as identical with that by which a branched inflorescence replaces an unbranched one, or a compound leaf takes the place of a simple one, the objections would not have been raised with such force. The process consists, in most cases, not so much in actual cleavage of a pre-existing organ as in the development of new-growing points from the old ones.

An illustration given by Moquin from Dunal[407] goes far to support the notion here adopted. The majority of the stamens of laurels (Laurus) have, says M. Dunal, on each side of the base of their filaments a small glandular bifid appendage; these excrescences are liable to be changed into small stamens. The male flowers have a four-leaved calyx, and sometimes eight stamens, each with two glands, four in one row, opposite to the sepals, four in a second series alternating with the first. More generally two of the stamens are destitute of glands, but have in their place a perfectly developed stamen, so that in these latter flowers there are twelve stamens.

M. Clos[408] mentions a flower of rue (Ruta) wherein there were two stamens joined together below and placed in front of a petal, as in Peganum.

Buchenau[409] mentions a flower of Lotus uliginosus in which there were eleven stamens, namely, two free and nine monadelphous; and Hildebrand describes an analogous increase in a flower of Sarothamnus scoparius in which, in conjunction with a seven-toothed calyx, there were two carinas and fourteen stamens. It would seem probable in this case that there was a coalescence of two flowers at an early date and consequent suppression of some of the parts of the flower. Whether this was the case or not in this particular illustration, it is nevertheless certain that many of the recorded instances of increased number in the organs of a flower are really the results of a fusion of two or more flowers, though frequently in the adult state but few traces of the coalescence are to be seen.

Polyphylly of the gynœcium.—Moquin[410] remarks that, as the pistils are, generally speaking, more or less subject to pressure, owing to their central position, and it may be added owing to their later development, than the other parts of the flower, they are more subject to suppression than to multiplication; nevertheless, augmentation in the number of carpels does occasionally take place, especially when the other parts of the flower are also augmented in number. Sometimes this increase in the number of carpels is due to pure multiplication, without any other change. At other times the increase is due to a substitution of stamens or other organs for carpels (see Substitutions). In other cases the augmentation seems to be due to the development of parts usually suppressed; for instance, in Antirrhinum, where there are usually only two carpels present, but where, under peculiar circumstances, five may be found—thus rendering the symmetry complete.[411] In Papilionaceæ, wherein usually only one carpel is developed, we occasionally find two, or even more, as in Wistaria, Gleditschia, Trifolium, &c. In Prunus and Amygdalus from two to five carpels are occasionally to be found,[412] in Mimosa five, in Umbelliferæ three to five; in some composites, e.g. Spilanthes, five carpels have also been noticed; in Cruciferæ three and four, in grasses three.[413] The double cocoa-nut affords an illustration of the development of two carpels out of three, one only generally arriving at perfection. Triple nuts (Corylus) also owe their peculiarity to the equal development of all three carpels which exist in the original flower, but of which, under ordinary circumstances, two become abortive. It is necessary, however, to distinguish these cases from those in which two embryos are developed in one seed.