spikelet are really modified bracts and some differentiate the flowering glumes from the empty ones, by giving them different names. The first two empty glumes are called glumes by all agrostologists. Some in Europe call the flowering glume lower palea to distinguish it from the real palea which they call the upper palea. Some American Authors have recently adopted for the flowering glume the term lemma introduced by Piper.

Considerable variation is met with in the case of the empty glumes. Generally they are unequal, the first being smaller. Very often the first glume becomes very small and it may be altogether absent. In some species of Panicum the first glume is very small, in Digitaria it is very minute and in Paspalum and Eriochloa it is entirely suppressed. The flowering glumes are generally uniform when there are many. In the spikelet having only four glumes the fourth glume differs from the others mainly in texture. Instead of being thin and herbaceous it becomes rigid and hard, smooth or rugose externally as in Panicum. Flowering glumes instead of being like empty glumes, become very thin, shorter and hyaline in Andropogon. Sometimes the flowering glumes are awned. All of them may be awned as in Chloris or only the fourth glume as in Andropogon.

The palea is fairly uniform in its structure in many grasses, but it is also subject to variation. It becomes shorter in some and is absent in others. Instead of having two nerves, it may have one and rarely more than two. The palea can easily be distinguished from the glume, because its insertion in the spikelet is different from that of the glume.

Fig. 18.—Flower of Chloris.
1. lodicules; 2. stamens; 3. ovary.

The lodicules are small organs and they are the vestiges of the perianth. In most grasses there are only two, but in Ochlandra and other bamboos we meet with three lodicules. There are also some species with many lodicules. In shape they are mostly of some form referable to the cuneate form. They are of somewhat elongated form in Aristida and Chloris. The function of the lodicules seems to be to separate the glume and its palea so as to enable the stamens to come out and hang freely at the time of anthesis. So it is only at the time of the opening of the flowers that the lodicules are at their best. Then they are fairly large, fleshy and thick and conspicuous. In the bud stage they are usually small and after the opening of the flower they shrivel up and are inconspicuous. There are also species of grasses in which the lodicules are not found.

The stamens are three in number in the majority of grasses and six are met with in Leersia, Hygrorhiza and Bamboos. Each stamen consists of a very delicate long filament and an anther basifixed to the filament. But as the anthers are long and the connective is reduced to its minimum, they appear as if versatile when the stamens are out. When there are three stamens one stands in front of the flowering glumes and the other two in front of the palea, one opposite each edge of the palea. The relative positions of the parts of the floret are shown in the floral diagrams. (See figs. 18 and 19.)

Fig. 19.—Floral diagrams.
The first is that of Chloris, second of Panicum and the third of Oryza.

The pistil consists of an ovary and two styles ending in plumose stigmas. The ovary is 1-celled and 1-ovuled. It is one carpelled according to the views of Hackel and his followers and there are also some who consider it as 3-carpelled because of the occurrence of three styles in the pistil of some bamboos.

The rachilla is usually well developed and elongated in many-flowered spikelets, while in 1-flowered spikelets it remains very small so that the flower appears to be terminal. It often extends beyond the insertion of the terminal flower and its glume, and then lies hidden appressed to the palea. This may be seen in the spikelets of the species of Cynodon. This prolonged rachilla sometimes bears a minute glume, which is of course rudimentary. Usually the glumes are rather close together on the rachilla so that the internodes are very short; but in some grasses, as in Dinebra arabica, the glumes are rather distant and so the internodes are somewhat longer and conspicuous. In some species of Panicum the rachilla is jointed to the pedicel below the empty glumes, whereas it is articulated just above these glumes in Chloris barbata. Sometimes the rachilla is articulated between the flowers. This is the case in the spikelet of Dinebra arabica.

Pollination in most grasses is brought about by wind, though in a few cases self-pollination occurs. The terminal position of the inflorescence, its protrusion far above the level of the foliage leaves, the swinging and dangling anthers, the abundance of non-sticking pollen and the plumose stigmas are all intended to facilitate pollination by wind. Furthermore the stamens and the stigmas do not mature at the same time. In some grasses the stamens mature earlier, (protandry) while in others the stigmas protrude long before the stamens (protogyny). As the result of the pollination the ovary developes into a dry 1-seeded indehiscent fruit. The seed fills the cavity fully and the pericarp fuses with the seed-coat and so they are inseparable. Such a fruit is termed a caryopsis or grain. Though in the vast majority of grasses the pericarp is inseparable, in a few cases it is free from the seed-coat as in Sporobolus indicus and Eleusine indica.