had it not been for the membranous expansions on each side of the column forming the clinandrum, within which they lie safely.
When an insect inserts its proboscis or head into the narrow space between the upright labellum and the rostellum, it will infallibly touch the little projecting viscid mass, and as soon as it flies away it will withdraw the pollinia. I easily imitated this action by inserting any small object into the tubular flower between the labellum and rostellum. When the insect visits another flower, the very thin pollen-leaves attached parallel to the proboscis, or head, will be forced into the pocket-like stigma with their broad ends foremost. I found pollinia in this position glued to the upper membranous expansion of the rostellum, and with a large number of pollen-tubes penetrating the stigmatic tissue. The use of the thin layer of viscid matter, which coats the surface of the rostellum in this genus and in Microstylis, and which is of no use for the transportal of the pollen from flower to flower, seems to be to keep the leaves of pollen fixed in the narrow stigmatic cavity when their lower ends have been inserted by insects. This fact is rather interesting under a homological point of view, for, as we shall hereafter see, the primordial nature of the viscid matter of the rostellum is that which is common to the stigmatic secretion of most flowers, namely, the retention of the pollen, when placed by any means on its stigma.
The flowers of the Malaxis, though so small and inconspicuous, are highly attractive to insects. This was shown by the pollinia having been removed from all the flowers on the spikes which I examined, excepting from one or two close under the buds. In some old flower-spikes every single pollinium had
