Page:On the Various Contrivances by Which British and Foreign Orchids are Fertilised by Insects, and on the Good Effects of Intercrossing.djvu/33

 failed to visit the flowers. The penetration of the pollen-tubes, however, is apparently more important in retaining the pillars of pollen in their proper places, so that the insects, in crawling into the flowers, might get dusted with pollen. This imperfect self-fertilisation is habitually and largely aided by insects, which may carry a flower's own pollen on to its stigma; but an insect thus smeared with pollen could not fail likewise to cross distinct individuals. From the relative position of the parts, it seems indeed probable (but I omitted to prove this by the early removal of the anthers, so as to observe whether pollen was brought to the stigma from other flowers) that an insect would more frequently get dusted by crawling out of a flower than by crawling into one; and this would of course facilitate a union between two distinct individuals. Hence Cephalanthera offers only a partial exception to the almost universal rule that the flowers of Orchids are fertilised by the pollen of other flowers.

Goodyera repens.

This genus, in most of its characters with which we are concerned, is rather closely related to Epipactis. The shield-like rostellum is almost square, and projects beyond the stigma; it is supported on each side by sloping sides rising from the upper edge of the stigma, in nearly the same manner as we shall presently see in Spiranthes. The surface of the protuberant part of the rostellum is rough, and when dry can be seen to be formed of cells; it is delicate, and, when slightly pricked, exudes a little milky viscid fluid; it is lined by a layer of very adhesive matter, which quickly sets hard when exposed to the air. The protuberant surface of the rostellum, when gently rubbed upwards, is easily removed, and carries with it a strip of membrane, to the hind part of which the pollinia are attached. The sloping sides which support the rostellum are not removed at the same time, but remain projecting up like a fork and soon wither. The anther is borne on an elongated broad filament; a membrane on both sides unites this filament to the edges of the stigma, forming an imperfect cup or clinandrum. The anther-cells open in the bud, and the pollinia become attached by their anterior faces, just beneath their summits, to the back of the rostellum; ultimately the anther opens widely, leaving the pollinia almost naked, but partially protected within the membranous cup or clinandrum. Each pollinium is partially divided length-ways; the pollen-grains cohere in subtriangular packets, including a multitude of grains, each grain consisting of four granules; and these packets are tied together by strong elastic threads, which at their upper ends run together and form a single flattened brown elastic ribbon, of which the truncated extremity adheres to the back of the rostellum.

The surface of the orbicular stigma is remarkably viscid, which is necessary in order that the unusually strong threads connecting the packets of pollen should be ruptured. The labellum is partially divided into two portions; the terminal portion is reflexed, and the basal portion is cup-formed and filled with nectar. The passage into the flower between the rostellum and labellum is contracted. Since my examination of Spiranthes, immediately to be described, I have suspected that the labellum moves further from the column in mature flowers, in order to allow insects, with the pollinia adhering to their heads or probosces, to enter the flowers more freely. In many of the specimens received, the pollinia had been removed by insects, and the fork-shaped supporting sides of the rostellum were partially withered.

Goodyera is an interesting connecting link between several very distinct forms. In no other member of the Neotteæ have I seen so near an approach to the formation of a true caudicle, like that in the Ophreæ; and it is curious that in this genus alone (as far as I have seen) the pollen-grains cohere in large packets, as in the Ophreæ. If the nascent caudicles had been attached to the lower ends of the pollinia, and they are attached a little beneath the summit, the pollinia would have been almost identical with those of a true Orchis.

In the rostellum being supported by sloping sides, which wither when the viscid disc is removed,—in the membranous cup or clinandrum between the stigma and anther,—and in some other respects, we have a clear affinity to Spiranthes. In the anther having a broad filament we see a relation to Cephalanthera. In the structure of the rostellum, with the exception of the sloping sides, and in the shape of the labellum, we see