dition of the flowers after they have been open for a day or two, and have been visited by insects. The fork soon withers.
Whilst the flower is in bud, the back of the boat-formed disc is covered with a layer of large rounded cells, so that the disc does not strictly form the exterior surface of the back of the rostellum. These cells contain slightly viscid matter: they remain unaltered (as may be seen at fig. E) towards the upper end of the disc, but at the point where the pollinia are attached they disappear. Therefore I at one time concluded that the viscid matter contained in these cells, when they burst, serve to fasten the threads of the pollinia to the disc; but, as in several other genera, in which a similar attachment has to be effected, I could see no trace of such cells, this view may be erroneous.
The stigma lies beneath the rostellum, and projects with a sloping surface, as may be seen at B in the side-view: its lower margin is rounded and fringed with hairs. On each side a membrane (cl, B) extends from the edges of the stigma to the filament of the anther, thus forming a membranous cup or clinandrum, in which the lower ends of the pollen-masses lie safely protected.
Each pollinium consists of two leaves of pollen, quite disconnected at their lower and upper ends, but united for about half their length in the middle by elastic threads. A very slight modification would convert the two pollinia into four distinct masses, as occurs in the genus Malaxis and in many foreign Orchids. Each leaf consists of a double layer of pollen-grains, joined by fours together, and these united by elastic threads, which are more numerous along the edges of the leaves, and converge at the
