At first the membrane forming the exterior surface of the rostellum is continuous; but as soon as the flower opens, the slightest touch causes it to rupture in such a manner as to set free from it, and from one another, the little discs, while at the same time the anther cells themselves split in a longitudinal direction from top to bottom. In this state none of the organs are changed; they are merely set free, and ready for change, yet maintaining their normal positions. The labellum, which is the largest petal of the flower, lies on one side of the nectary tube, and the rostellum projects into it on the other; hence, if an insect alights on the labellum, and pushes its head into the tube, in order to reach the honey in the nectary with its proboscis, it cannot fail to touch and depress the rostellum (see fig. 76), so that one or both of the viscid discs carrying the pollinia will stick to it, and as the anther cells are open in front, when the insect withdraws its head, one or both of the pollinia are drawn out of their cells, and stick upright on it like horns. In that position they never could fertilize any blossom which the insect might afterwards visit; but Mr. Darwin has shown that their position is changed by a contrivance that is not surpassed in beauty in the whole vegetable world. While a pollinium is upright on the head of the insect, the little viscid disc which supports it contracts on being exposed to the air, so as to cause the pollinium to sweep through ninety degrees towards the apex of the insect’s proboscis, and this is accomplished in about thirty seconds, the time an insect would take to fly to another flower. The pollinium in this position would exactly strike the surface of the stigma when the insect inserts its proboscis into the nectary of a flower. It was long ago noticed by Robert Brown, that the stigma is very viscid, but not so viscid as when touched to pull the whole pollinium off the insect’s head, yet sufficiently viscid to break the elastic threads by which the packets of pollen grains are tied together, and leave some of them on the stigma. Hence a pollinium attached to an insect can be applied to many stigmas, and fertilize them all. Mr. Darwin mentions having seen the pollinium of Orchis pyramidalis adhering to the proboscis of a moth with the stalks alone left, all the packets of pollen having been left glued to the stigma of the flowers successively visited. It appears that insects, for the most part, only remove one pollinium at a time, and that the rostellum returns to its normal position to prevent the viscid matter of the discs of the remaining pollinia from being exposed to the air. [Illustration: Fig. 77. Orchis pyramidalis:—Front view of flower with upper sepal and petals removed; _a_, anther; _s_, stigma; _r_, rostellum; _l_, labellum.]