come into existence? And no answer is forthcoming. Now turning to Fig. 31, b, we see one use at least. The weight of the bee must be very great; and the curious shape of the lip, with its lateral ridges, is evidently not only an excellent landing-place, but is so constructed as to bear that weight. Moreover, the

two

walls slope off, and are

gripped by the legs of

a

b

the bee, so that it evidently can secure an excellent purchase, and can thus rifle the flower of its treasures at its ease.

Irregular corollas are very numerous, but certain principles, traceable to insect action, govern their forms. In the first place, the side upon which the insect rests, or at least upon which its weight is thrown, is always enlarged, and mostly forms the landing-place. It is almost always the

anterior petal; if, however, the pedicel or ovary has been too slender to support it, then it has sometimes become twisted, and the flower is said to be resupinate, so that the posterior petal becomes anterior in position, and is now the larger one, since it supplies the landing-place for insects, as in Orchis. Fumaria might be called semi-resupinate, as the corolla has only rotated through 90°. A slight modification occurs in the "Bee-orchis," Ophrys apifera, which is usually described as having a twisted ovary like a true Orchis; but in this species it has scarcely if any twist at all; the flower, however, is bent over to the opposite side of the stem, so that while the posterior petal is still the labellum, the ovary has itself remained perfectly straight.

The next point to notice is that when the anterior petal is enlarged, the posterior one or more often enlarges also, while a corresponding tendency to atrophy affects the lateral ones. This is seen in many species of Leguminosae, Scrophularineæ, and Labiata, and in zygomorphic flowers generally. It occurs thus in the wing petals of many papilionaceous flowers, as is particularly well seen in Onobrychis. The immediate causes, I repeat, I would recognize in the weight of the insect in front, the local irritations behind, due to the thrust of the insect's head and probing for nectar, coupled with the absence of all strains upon the sides. In some papilionaceous flowers the wing petals form a landing-place, as in Indigofera and Phaseolus. Whenever this is the case, they too are enlarged, as the lateral ones are in Fig. 31, and undertake the duty impressed upon them.

When, therefore, one finds as an invariable rule how the front petals enlarge when flowers are compacted and visited only from the front, and thus become irregular; and as such often occur in orders where flowers are normally regular, as Iberis, Centaurea, Heracleum, etc.; and, moreover, when the same phenomena appear in orders having no affinity between them, as in Labiata and Orchidea; and are, indeed, to be found throughout the length and breadth of the floral world, one is justified in attributing such irregularities to a common cause, that being, according to my theory, the responsive power of protoplasm to the irritations from without, set up by insect and other agencies.

Many other special cases might be described from the different orders of plants, but the above will suffice to illustrate this principle of responsive action with resulting correlations to insect agency. I would here, however, call the reader's attention to the mechanical arrangement of forces as shown in Lamium and Echium, where it will be seen that the

adhesions of the stamens to the corolla furnish the fulcra, the cohesion of the petals into a tube affording a greatly increased power of resistance; the weight of the insect on the labellum or declinate stamens is, of course, vertically downwards, and the line of the resultant, which the lip in Lamium and the stamens whenever declinate have to exert, passes through the point of meeting of the first two, and so sustains the insect while visiting the flower. Other and analogous instances will be described hereafter.

Good illustrations of the occurrence of great thickenings just where the strain will be most felt, may be seen in the slipper-shaped flowers of Calceolaria (Fig. 32), Coryanthes, and Cypripedium. Thus Calceolaria Pavonii

possesses a thick ridge along the upper edges of the curved basal part, which carries the inflated end upon which the bee stands, and which it depresses to get the honey. In this species it may be noticed the anther-cells are separated (a),

vonii (after Kerner).

so that they can oscillate as they do in Fig. 32.-Calceolaria PaSalvia. In Cypripedium the edge is folded inwards, thus strengthening the same part; while in Coryanthes the lower portion is enormously enlarged, thus acting as a powerful spring which forces the anterior end of the labellum to be in close contact with the column.

THE ORIGIN OF IRREGULARITY IN THE ANDRŒCIUM.-As it is with the perianth, so is it with the androecium: if the petals are regular the stamens are usually regular also; but when irregularity occurs in the corolla the staminal whorl follows suit, and the position and form of the stamens are equally correlated to the effectual pollination of the flower, Thus, as hypertrophy affects the anterior side of the

flowers of Labiatae, the anterior stamens are almost invariably the larger pair. On the other hand, atrophy has affected the posterior side of the staminal whorl, causing the total loss of the fifth stamen, and, to some extent, a reduction in length of the next pair of filaments.

When the weight of the insect is thrown upon the stamens, they either hang downwards, and the insect is suspended upon them, as in Epilobium angustifolium, or else they become declinate and then the anterior petal, being relieved, does not enlarge, either remaining of the same size as the rest, or else diminishes, and may even vanish altogether. Thus Vallota, with its perfectly regular perianth and spreading stamens, may be compared with Amaryllis, which has declinate stamens and a small anterior petal. The terminal flower of a "thyrse "of the Horse-chestnut, like the terminal flower of a "truss" of Pelargonium, is often regular with spreading stamens, whereas the normal flowers

Fig. 33.-Dictamnus (after Tieghem).

have declinate stamens, and

usually only four petals, the fifth or anterior one being altogether suppressed.

In some flowers the stamens are dependent at first, but their anthers rise up when dehiscing, and so the filaments become declinate in the pollinating stage. This is the case with Delphinium, Epilobium angustifolium, and Dictamnus (Fig. 33). In this flower the anterior petal is

of much the same size as the others, but is often displaced (Fig. 33), and not immediately below the stamens,-this

lateral displacement of the anterior petal being not always carried out, as it is in the next flower to be described.

In Epilobium angustifolium (Fig. 34) and Godetia, which have no anterior petals, the bees cling to the dependent stamens, while the petals have become permanently displaced, the two lower being somewhat raised, so that the angular distances are not the same. In Azalea and Rhododendron there is no anterior petal, but the posterior one is slightly enlarged, and this alone possesses extra colouring and the "path-finder." The stamens, being declinate, carry the insect without the aid of the corolla, so that the antero-lateral

Fig. 34.-Epilobium angustifolium.

Fig. 35.-Veronica Chamædrys (after Müller). pair of petals, not sharing in the support of the insect, are not enlarged at all.

In Circea and Veronica Chamædrys (Fig. 35), the insect clings to the two stamens and style; and the anterior petals are not enlarged, while in the latter flower it is, as usually the case, the smallest, the stamens of Veronica being attached to the lateral petals have to supply the fulcra for leverage, and consequently these have now become relatively hypertrophied.

In many flowers which have sub-declinate stamens, the latter lie in a more or less boat-shaped anterior petal, showing that the action of the insect has somewhat affected both the whorls together, as they have each some share in carry