sion caused, the pollen-masses are not so neatly caught by their tips as those of Listera. Thus a good deal of the friable pollen is often left behind in the anther-cells and is apparently wasted. Several plants were protected from the access of winged insects by a net, and after four days the rostella had almost lost their sensitiveness and power to explode. The pollen had become extremely incoherent, and in all the flowers much had fallen on the stigmas which were penetrated by the pollen-tubes. The spreading of the pollen seems to be in part caused by the presence of Thrips, many of which minute insects were crawling about the flowers, dusted all over with pollen. The covered-up plants produced plenty of capsules, but many of these were much smaller and contained fewer seeds than those produced by the adjoining uncovered plants.
If insects had been forced by the labellum being more upturned to brush against the anther and stigma, they would always have been smeared with the pollen as soon as it became friable; and they would thus have fertilised the flowers effectually without the aid of the explosive rostellum. This conclusion interested me, because, when previously examining Cephalanthera, with its aborted rostellum, its upturned labellum and friable pollen, I had speculated how a transition, with each gradation useful to the plant, could have been effected from the state of the pollen in the similarly constructed flowers of Epipactis, with their pollinia attached to a well-developed rostellum, to the present condition of Cephalanthera. Neottia nidus-avis shows us how such a transition might have been effected. This Orchid is at present mainly fertilised by means of the explosive rostellum, which acts effectually only as long as the pollen remains in mass; but we have seen that as the flower grows old the pollen swells and becomes friable, and is then apt to
