classification can be correct. The rostella are small and widely separated from each other. The viscid matter on the under side of the disc forms an oval ball which is enclosed within a small pouch. The upper membrane to which the caudicle is attached is of large size relatively to the whole disc, and is freely exposed to the air. Hence probably it is that the pollinia when removed from their cases do not become depressed until, as Mr. T. H. Farrer has observed, twenty or thirty minutes have elapsed. Owing to this long interval, I formerly thought that they did not undergo any movement of depression. Supposing a pollinium to be attached to the head of an insect, and to have become depressed, it will stand at the proper angle, vertically, for striking the stigma. But from the lateral position of the anther-cells, notwithstanding that they converge a little towards their upper ends, it is difficult at first to see how the pollinia when removed by insects are afterwards placed on the stigma; for this is of small size and is situated in the middle of the flower between the two widely separated rostella.
The explanation is, I believe, as follows. The base of the elongated labellum forms a rather deep hollow in front of the stigma, and in this hollow, but some way in advance of the stigma, a minute slit-like orifice (n) leads into a short bilobed nectary. Hence an insect, in order to suck the nectar with which the nectary is filled, would have to bend down its head in front of the stigma. The labellum has a medial ridge, which would probably induce an insect first to alight on either side; but, apparently to make sure of this, besides the true nectary, there are two spots (n′ n′) which secrete drops of nectar on each side at the base of the labellum, bordered by prominent edges, directly
