With respect to the plants in the open ground, each long row was divided into half, so as to diminish the chance of any accidental advantage in one part of either row; and the four tallest plants in the two halves of the two rows were carefully selected and measured. The eight tallest crossed plants averaged 30.92, and the eight tallest self-fertilised 30.7 inches in height, or as 100 to 99; so that they were practically equal. But we should bear in mind that the trial was not quite fair, as the self-fertilised plants had a great advantage over the crossed in being much less crowded in their own row, owing to the large number of seeds which had perished under ground after sprouting. Nor were the lots in the two rows subjected to any mutual competition. 29. CANNACEAE.--Canna warscewiczi. In most or all the species belonging to this genus, the pollen is shed before the flower expands, and adheres in a mass to the foliaceous pistil close beneath the stigmatic surface. As the edge of this mass generally touches the edge of the stigma, and as it was ascertained by trials purposely made that a very few pollen-grains suffice for fertilisation, the present species and probably all the others of the genus are highly self-fertile. Exceptions occasionally occur in which, from the stamen being slightly shorter than usual, the pollen is deposited a little beneath the stigmatic surface, and such flowers drop off unimpregnated unless they are artificially fertilised. Sometimes, though rarely, the stamen is a little longer than usual, and then the whole stigmatic surface gets thickly covered with pollen. As some pollen is generally deposited in contact with the edge of the stigma, certain authors have concluded that the flowers are invariably self-fertilised. This is an extraordinary conclusion, for it implies that a great amount of pollen is produced for no purpose. On this view, also, the large size of the stigmatic surface is an unintelligible feature in the structure of the flower, as well as the relative position of all the parts, which is such that when insects visit the flowers to suck the copious nectar, they cannot fail to carry pollen from one flower to another. (6/7. Delpino has described ‘Bot. Zeitung’ 1867 page 277 and ‘Scientific Opinion’ 1870 page 135, the structure of the flowers in this genus, but he was mistaken in thinking that self-fertilisation is impossible, at least in the case of the present species. Dr. Dickie and Professor Faivre state that the flowers are fertilised in the bud, and that self-fertilisation is inevitable. I presume that they were misled by the pollen being deposited at a very early period on the pistil: see ‘Journal of Linnean Society Botany’ volume 10 page 55 and ‘Variabilité des Espèces’ 1868 page 158.)