mine. It should be observed, that although I looked carefully, not a grain of pollen could be found on the stigmas of any of these flowers, and their ovaria had not swollen. During a subsequent year, several plants were again covered by a net, and I found that the rostellum lost its power of explosion in about four days; the viscid matter having turned brown within the loculi of the rostellum. The weather at the time was unusually hot, and this probably hastened the process. After the four days the pollen had become very incoherent, and some had fallen on the two corners, and even over the whole surface of the stigma, which was penetrated by the pollen-tubes. But the scattering of the pollen was largely aided by, and perhaps wholly depended on, the presence of Thrips—insects so minute that they could not be excluded by any net, and which abounded on the flowers. This plant, therefore, is capable of occasional self-fertilisation, if the access of winged insects be prevented; but I have every reason to believe that this occurs very rarely in a state of nature.
That insects do their work of cross-fertilisation effectually is shown by the following cases. The seven upper flowers on a young spike with many unexpanded buds, still retained their pollinia, but these had been removed from the ten lower flowers; and there was pollen on the stigmas of six of them. In two spikes taken together, the twenty-seven lower flowers all had their pollinia removed, and had pollen on their stigmas; these were succeeded by five open flowers with the pollinia not removed and without any pollen on the stigmas; and these were succeeded by eighteen buds. Lastly, in an older spike with forty-four fully expanded flowers, the pollinia had been removed from every single one; and there was pollen,
