The grasses constitute a very well defined natural group of plants, but the division of this family into tribes and subtribes is a difficult problem. We know that ROBERT BROWN divided the family into the Panicaceae and the Poaceae. BENTHAM already indicated that in the former the tendency to imperfection lies in the lower flowers of the spikelets, whereas in the Poaceae the tendency is in the opposite direction, but he observes at the same time that this principle is too indefinite to serve as a practical character to recognize both groups. In combination with other characters, especially those taken from the fruits (the caryopsis, enclosed by the scales), these two groups become however more stabile. KUNTH gave us no less than 13 tribes, many of them indeed very natural and accepted in recent works. The earlier agrostologists have given a considerable importance to the presence or absence of awns on the back or on the apex of the flowering glume (lemma). We know however at present that this character, although important to recognize species, is not very valuable for the different tribes and must be used with great reserve.