drum (see fig. B) which lies behind the rostellum. When the flower is expanded the two pollinia, united into a single mass, lie quite loose on the clinandrum and under the anther-case. The labellum embraces the column, leaving a passage in front. The middle portion of the labellum (as may be seen in fig. A) is thickened, and extends up as far as the top of the stigma. The lowest part of the column is developed into a saucer-like nectary, which secretes honey.
As an insect forces its way into one of these flowers, the labellum, which is elastic, will yield, and the projecting lip of the anther will protect the rostellum from being disturbed; but as soon as the insect retreats, the lip of the anther will be lifted up, and the viscid matter from the rostellum forced into the anther, gluing the pollen-mass to the insect, which will thus be transported to another flower. I easily imitated this action; but as the pollen-masses have no caudicle and lie rather far back within the clinandrum beneath the anther, and as the matter from the rostellum is not highly viscid, they were sometimes left behind.
Owing to the inclination of the base of the clinandrum, and owing to the length and elasticity of the filament, as soon as the anther is lifted up it always springs forward, over the rostellum, and remains hanging there with its lower empty surface (fig. C) suspended over the summit of the stigma. The filament now stretches across the space (see fig. B) which was originally covered by the anther. Several times, having cut off all the petals and labellum, and laid the flower under the microscope, I raised the lip of the anther with a needle, without disturbing the rostellum, and saw the anther assume, with a spring, the position
