Nevertheless the pollen-masses were not removed nearly so cleanly as those which had been naturally removed by insects. I tried dozens of flowers, always with the same imperfect result. It then occurred to me, that an insect in backing out of the flower would naturally push with some part of its body against the blunt and projecting upper end of the anther, which overhangs the stigmatic surface. Accordingly I so held a brush that, whilst brushing upwards against the rostellum, I pushed against the blunt solid end of the anther (see fig. C); this at once eased the pollinia, and they were withdrawn in an entire state. At last I understood the mechanism of the flower.
The large anther stands above and behind the stigma, forming an angle with it (fig. C), so that the pollinia when withdrawn by an insect would adhere to its head or body in a position fitted to strike the sloping stigmatic surface as soon as another flower was visited. Hence we have not here, or in any of the Neotteæ, that movement of depression so common with the pollinia of the Ophreæ. When an insect with the pollinia attached to its back or head enters another flower, the easy depression of the distal portion of the labellum probably plays an important part; for the pollen-masses are extremely friable, and if they were struck against the tips of the petals much of the pollen would be lost; but as it is, an open gangway is offered, and the viscid stigma, with its lower protuberant part lying in front, is the first object against which the pollen-masses projecting forwards from the insect's head or back would naturally strike. I may add that in one large lot of flower-spikes, a great majority of the pollinia had been naturally and cleanly removed.
In order to ascertain whether I was right in believing
