Therefore the little bud-like flower which produced this capsule probably was as destitute of pollen as were those which I examined.
Juncus bufonius and Hordeum.
All the species hitherto mentioned which produce cleistogamic flowers are entomophilous; but four genera, Juncus, Hordeum, Cryptostachys, and Leersia are anemophilous. Juncus bufonius is remarkable by bearing in parts of Russia only cleistogamic flowers, which contain three instead of the six anthers found in the perfect flowers. (8/22. See Dr. Ascherson's interesting paper in 'Botanische Zeitung' 1871 page 551.) In the genus Hordeum it has been shown by Delpino that the majority of the flowers are cleistogamic, some of the others expanding and apparently allowing of cross-fertilisation. (8/23. 'Bollettini del Comizio agrario Parmense.' Marzo e Aprile 1871. An abstract of this valuable paper is given in 'Botanische Zeitung' 1871 page 537. See also Hildebrand on Hordeum in 'Monatsbericht d. K. Akad Berlin' October 1872 page 760.) I hear from Fritz Muller that there is a grass in Southern Brazil, in which the sheath of the uppermost leaf, half a metre in length, envelopes the whole panicle; and this sheath never opens until the self-fertilised seeds are ripe. On the roadside some plants had been cut down, whilst the cleistogamic panicles were developing, and these plants afterwards produced free or unenclosed panicles of small size, bearing perfect flowers.
Leersia oryzoides.
It has long been known that this plant produces cleistogamic flowers, but these were first described with care by M. Duval-Jouve. (8/24. 'Bulletin Bot. Soc. de France' tome 10 1863 page 194.) I procured plants from a stream near Reigate, and cultivated them for several years in my greenhouse. The cleistogamic flowers are very small, and usually mature their seeds within the sheaths of the leaves. These flowers are said by Duval-Jouve to be filled by slightly viscid fluid; but this was not the case with several that I opened; but there was a thin film of fluid between the coats of the glumes, and when these were pressed the fluid moved about, giving a similarly deceptive appearance of the whole inside of the flower being thus filled. The stigma is very small and the filaments extremely short; the anthers are less than 1/50 of an inch in length or about one-third of the length of those in the perfect flowers. One of the three anthers dehisces before the two others. Can this have any relation with the fact that in some other species of Leersia only two stamens are fully developed? (8/25. Asa Gray 'Manual of Botany of the United States' 1856 page 540.) The anthers shed their pollen on the stigma; at least in one instance this was clearly the case, and by tearing open the anthers under water the grains were easily detached. Towards the apex of the anther the grains are arranged in a single row and lower down in two or three rows, so that they could be counted; and there were about 35 in each cell, or 70 in the whole anther; and this is an astonishingly small number for an anemophilous plant. The grains have very delicate coats, are spherical and about 5/7000 of an inch (.0181 millimetres), whilst those of the perfect flowers are about 7/7000 of an inch (.0254 millimetres) in diameter.
M. Duval-Jouve states that the panicles very rarely protrude from their sheaths, but that when this does happen the flowers expand and exhibit well-developed ovaries and stigmas, together with full-sized anthers containing apparently sound pollen; nevertheless such flowers are invariably quite sterile. Schreiber had previously observed that if a panicle is only half protruded, this half is sterile, whilst the still included half is fertile. Some plants which grew in a large tub of water in my greenhouse behaved on one occasion in a very different manner. They protruded two very large much-branched panicles; but the florets never opened, though these included fully developed stigmas, and stamens supported on long filaments with large anthers that dehisced properly.