108 GRAMINACEiE. [Endogens. 



be regarded as consisting merely of bractese, the real perianthium of the order must be 

 looked for in those mmute scales, wliich, in the greater part of its genera, are found 

 immediately surrounding the sexual organs. The scales are, in most cases, only two in 

 number, and placed collaterally \vithia the inferior valve of the proper envelope. In 

 their real insertion, however, they alternate with the valves of this envelope, as is obvi- 

 ously the case in Ehrharta and certain other genera ; and their collateral approximation 

 may be considered as a tendency to that confluence which miifonuly exists in the parts 

 composing the upper valve of the proper envelope, and which takes place also between 

 these two squamae themselves, in some genera, as Glyceria and Melica. In certam other 

 genera, as Bambusa and Stipa, a thu-d squamula exists, which is placed opposite to the 

 axis of the upper valve of the proper envelope, or, to speak in conformity with the -view 

 already taken of the structm-e of this valve, opposite to the conjunction of its two com- 

 ponent parts. With these squamse the stamina m triandrous Grasses alternate, and 

 they are consequently opposite to the parts of the proper envelope ; that is, one stamen 

 is opposed to the axis of its lower or outer valve, and the two others are placed opposite 

 to the two nerves of the upper valve. Hence, if the mner envelope be considered as 

 consistmg of bractese, and the hypogjmous squamee as forming the perianthium, it seems 

 to follow, from the relation these parts have to the axis of inflorescence, that the outer 

 series of this perianthium is wanting, Avhile its corresponduig stamina exist, and that 

 the whole or part of the inner series is produced while its corresponding stamina are 

 generally wanting. This may, no doul>t, actually be the case ; but as it would be, at least, 

 contrary to every analogy in Monocotyledonous plants, it becomes in a certain degree 

 probable that the inner or proper envelope of Grasses, the calyx of Jussieu, notwithstand- 

 ing the obliquity in the msertion of its valves, forms in reality the outer series of the 

 true perianthium, whose mner series consists of the minute scales, never more than 

 three in number, and in which an u^regularity in some degree analogous to that of the 

 outer series generally exists. It is necessary to be aware of the tendency to suppression 

 existing, as it were, in opposite du-ections in the two floral envelopes of Grasses, to com- 

 prehend the real structure of many u'regular genera of the order, and also to imderstand 

 the limits of the two great tribes into wliich I have proposed to subdivide it. One of 

 these tribes, which may be called Panicete, comprehends Ischsemum, Holcus, Andro- 

 pogon, Anthistiria, Saccharum, Cenchrus, Isachne, Panicum, Paspalum, Reimaria, 

 Anthenantia, Monachne, Lappago, and several other nearly related genera ; and its 

 essential character consists in ha^-ing always a locusta of two flowers, of which the lower 

 or outer is uniformly imperfect, being either male or neuter, and then not unfrequently 

 reduced to a single valve. Ischeemum and Isachne are examples of this tribe in its 

 most perfect form, from which Anthenantia, Paspalum, and Reimaria, most remarkably 

 deviate, in consequence of the suppression of certain parts : thus Anthenantia (which is 

 not correctly described by Palisot de Beauvois) differs from those species of Panicum 

 that have the lower flower neuter and bivah-ular, in being deprived of the outer valve 

 of its gluma ; Paspalum differs from iVnthenantia in the want of the imier valve of its 

 neuter flower, antl from those species of Panicum whose outer flower is univah^lar, in 

 the want of the outer valve of its gluma ; and Reimaria differs from Paspalum in being 

 entirely deprived of its gluma. That this is the real structure of these genera may be 

 proved by a series of species connecting them with each other, and Panicum with Paspa- 

 lum. The second tribe, which may be called Poacese, is more numerous than Panicese, 

 and comprehends the greater part of the European genera, as well as certain less exten- 

 sive genera pecuhar to the equinoctial countries ; it extends also to the highest latitudes 

 in which Phaenogamous plants have been fomid ; but its maximum appears to be in the 

 temperate chmates, considerably beyond the tropics. The locusta in this tribe may 

 consist of 1, 2, or of many flowers ; and the 2-flowered genera are distinguished from 

 Panicese by the outer or lower flower being always perfect, the tendency to imperfection 

 in the locusta existing in opposite directions in the two tribes. In confoiTuity with this 

 tendency in Poacese, the outer valve of the perianthium in the smgle-flowered genera 

 is placed within that of the gluma, and in the many-flowered locusta the upper flowers 

 are frequently imperfect. There are, however, some exceptions to this order of suppres- 

 sion, especially in Arundo Phragmites, Campulosus, and some other genera, in which 

 the outer flower is also imperfect : but as all of these have more than two flowers in 

 then* locusta, they are still readily distuiguished from Panicese." Broion in Flinders^ 580, 

 According to this view, in a locusta of several florets, the scales at its base, or glumes, 

 are bracts, and each floret consists of a calyx formed of one sepal remote from the 

 racliis, and two cohering by their marguis and next the rachis ; the little liypog\Tious 

 scales are the nidiments of two petals, and the stamens alternate with these in the 

 normal manner. This may be rendered more clear by the following diagram, in 

 which the triangle ABB represent the outer series, or palese, or calyx, A being 

 the inferior valve, and B B the superior, fomied of two sepals united by their con- 



