strange position of the labellum, perched on the summit of the column, ought to have shown me that here was the place for experiment. I ought to have rejected the notion that the labellum was thus placed for no good purpose. This plain guide was overlooked, and for a long time I completely failed to understand the structure of the flower.
We have seen that when the pollinium is ejected and swings upwards, it adheres by the viscid surface of the disc to any object projecting beyond the edge of the labellum directly over the column. When thus attached, it forms an irregular hoop, with the torn-off anther-case still covering the pollen-masses which are close to the disc, but protected from adhering to it by the dependent curtain. Whilst in this position the projecting and bowed part of the pedicel would effectually prevent the pollen-masses from being placed on the stigma, even supposing the anther-case to have fallen off. Now let us suppose the pollinium to be attached to an insect's head, and observe what takes place. The pedicel, when first separated from the rostellum, is damp; as it dries, it slowly straightens itself, and when perfectly straight the anther-case readily drops off. The pollen-masses are now naked, and they are attached to the end of the pedicel by easily ruptured caudicles, at the right distance and in a proper position for their insertion into the adhesive stigma, as soon as the insect visits another flower. Thus every detail of structure is now perfectly adapted for the act of fertilisation.
When the anther-case drops off, it has performed its triple function; namely, its hinge as an organ of sense, its weak attachment to the column as a guide causing the pollinium at first to swing perpendicularly upwards, and its lower margin, together with the curtain of the
