of the sections Cymbopogon and Gymnandropogon, in which the awn is much reduced or obsolete. Agenium, Nees, from his character, would also refer to one of these species without prominent

awns.

23. CHRYSOPOGON, Trin. (Rhaphis, Lour., Centrophorum, Trin.), and 24. SORGHUM, Pers. (Blumenbachia, Koel.), are two genera very nearly allied to each other and differing from Andropogon, as Spodiopogon does from Pollinia, chiefly in their inflorescence; the branches of the panicle bear three spikelets at the end, a sessile one between two pedicellate ones, and occasionally only one or two pairs below on the same branch. They were both included by Linnæus, and afterwards by Brown, in Holcus, a name since restricted to that portion of the old genus which belongs to Avenaceæ. Chrysopogon, as now constituted, has nearly twenty species, chiefly tropical or subtropical, but including also the European C. Gryllus and some other temperate species. The genus may be divided into two natural sections: in the typical form the pedicellate spikelets usually contain a male flower; in the section Stipoides, exclusively American, it is reduced to a long hairy stipes rarely bearing a minute rudimentary glume. This section includes C. nutans, C. avenaceus, C. stipoides, C. Minarum, and a few others. Sorghum differs from Chrysopogon in habit, in the scarcely articulate branches of the panicle, and in the glumes of the fertile spikelets more hardened after flowering. The number of species is very uncertain, for, of the two principal ones, S. halepense is so widely spread as a tropical or subtropical weed, and S. vulgare so long and so generally cultivated in warm regions for a variety of purposes, as to have produced a great variety of forms, raised by many to the rank of species.

25. ANTHISTIRIA, Linn. fil. (Themeda, Forsk.), if taken as a whole, is a very natural genus, of about a dozen species from the warmer regions of the Old World, easily recognized by its inflorescence. The spikelets are in short dense spikes or clusters, usually seven together, of which the four lower ones (two pairs) are either empty or with a male flower in each, and are placed apparently in a whorl, forming a kind of involucre round the three inner ones, which, as in Chrysopogon, are one sessile between two pedicellate ones. In a few species the number of spikelets is raised to nine, or even to eleven, by the intervention of one or even two pairs of spikelets between the involucral and the terminal ones. These slight differences in the number or in the

pedicellation of the spikelets have induced the proposal of distinct genera for most of the species, and several of them have been adopted by Andersson in a monograph most carefully worked up in as far as the materials at his command admitted, but in which, for want of access to a sufficiently rich library, he is much mistaken as to several of the synonyms quoted. These proposed genera are:-1. Aristaria, Jungh., for A. frondosa, Br. (A. Junghuhniana, Nees), which forms the section Heterelytron of Andersson, but not Junghuhn's genus of that name. 2. Perobachne, Presl, is A. arundinacea, Roxb., forming Andersson's subsection Chrysanthistiria. 3. Andersson's subsection Euanthistiria for the common A. ciliata, Linn., and its immediate allies, to which some botanists would restrict the genus. Andersson distinguishes twelve species, adding at the same time that they might well all be reduced to varieties of a single widely-spread species. 4. Androscepia, Brongn. (Heterelytron, Jungh.), was founded originally on the A. gigantea, Cav., but became a very unnatural group when made to include A. (Androscepia) anathera, Anders., which very closely resembles A. (Euanthistiria) minuta, Anders., and a variety armata, Anders., of A. gigantea, which is much nearer to the A. (Perobachne) arundinacea. 5. Iseilema, Anders., containing two East-Indian and one Australian species, and 6. Exotheca, Anders., comprising A. abyssinica, Hochst., from tropical Africa, and A. fasciculata, Thw., from Ceylon, have each a peculiar habit and characters, sufficient to maintain them as sections. 7. Germainia, Balansa, has, perhaps, two closely allied species—A. caudata, Nees, from Khasiya and China, and the typical A. capitata fron Saigou; the latter, however, which I only know from Balansa's figure and description, is exactly like the Chinese plant, except that there appear to be rather more spikelets in the cluster.

26. APLUDA, Linn., is now universally recognized as a distinct and natural genus, limited to the two tropical-Asiatic species originally assigned to it by Linnæus, though his character was even then very imperfect, and rendered still more so by the subsequent addition of the very different American Zeugites, which Schreber afterwards restored as an independent genus. Beauvois, however, threw every thing into confusion; for it is evident from his figures that his Diectomis is A. aristata, Linn., and his Calamina is A. mutica, Linn., though in drawing up his character for the latter he combined it with some species of Anthistiria. Beauvois's Apluda is certainly different, probably a Chrysopogon.

1

Series B. POACEE.

Having already explained the difference between the two primary divisions of Gramineæ, I need only repeat here that the main characters of Poaceae consist, firstly, in the want of any articulation of the pedicel below the lower empty glumes, which remain persistent after the fruiting one has fallen away, or fall away separately, and, secondly, in the male or imperfect or rudimentary flowers, when present, being above, not below, the fertile one. The former character is all but universal; but from the latter one exceptions are not very rare, besides that, where there is only one flower without any continuation of the rhachilla beyond it, the character entirely fails. I should add that in some tribes of Poaceae there are two or more perfect flowers in the spikelet, which is not the case in Panicaceæ; and may now proceed to examine in detail the eight tribes into which this second series may be divided.

Tribe VII. PHALARIDEE.

The close affinity of this tribe and the Oryzeæ has been generally admitted, and the two are usually placed in juxtaposition; I had even proposed their consolidation into a single one in the 'Flora Australiensis.' They have in common the important character of the scale immediately under the single perfect terminal flower being keeled or one-nerved, so as to make it a matter of discussion whether it be a glume terminal on the main axis or rhachilla of the spikelet, or a palea at the base of a secondary floral axis. The deciduous part of the spikelet of Phalarides with its four glumes (or three glumes and a palea) is precisely as in Oryzeæ; but there are in addition, below the articulation, the two persistent empty glumes characteristic of Poacea. The spikelet, therefore, in this tribe consists of six glumes (or five and a palea), the lowest pair empty below the articulation; the second pair, above the articulation, corresponding to the lowest two glumes of Oryzeæ, are usually empty and small, sometimes reduced to a small bristle, rarely enclosing each a small palea or a male flower; the upper pair (or glume and similar palea) enclosing the terminal fertile flower and fruit, without any continuation of the rhachilla above it. A slight apparent exception will be mentioned under Phalaris itself; and in the genus Cinna of Agrostideæ and a very few Bambuses the palea of the fertile flower is, at least apparently,

one-nerved, but otherwise the character of Phalarideæ is constant. They comprise the following six genera :

1. EHRHARTA, Thunb. (Trochera, L. C. Rich.), has twenty-four species, of which two are from New Zealand, two from the Mascarene Islands, and all the rest from South Africa. In them the glumes of the second pair are the largest, empty and usually awned, and the fertile flower has six stamens. 2. MICROLENA, Br., including Diplax, Hook. f., has five Australian or New-Zealand species, differing from Ehrharta only in the number of stamens reduced to four or two. 3. TETRARRIENA, Br., four Australian species, with four stamens to the flower as in Microlana, but the glumes are in less regular pairs, all unawned, and the fourth (one of the second pair) alone the largest. The panicle is also almost always contracted into a spike, not, however, so dense and cylindrical as in the following two genera.

4. PHALARIS, Linn., has nine or ten extratropical species, chiefly from the Mediterranean region, but also extending to North and South America. In this genus it is the lowest two persistent empty glumes that are the largest, usually very flat, and often winged on the keel, the second pair (like the lowest in Oryza) very narrow, sometimes reduced to small bristles, those of the upper pair thin and hyaline; and sometimes in both of them, but almost always in the uppermost one, the central nerve is very faint or quite obsolete, a character adduced as an argument that this upper one is a two-nerved palea on the floral axis, and not a glume on the main rhachilla. The two nerves are, however, very faint, and the central keel is usually marked by a line of hairs on the outside, and the question remains a moot one. In the majority of species the panicle is contracted into a dense globular or cylindrical head; but in P. arundinacea, Linn., a stout tall species, forming the genus Digraphis, Trin. (Baldingera, Gærtn., Meg., and Schrad., Typhoides, Moench), the inflorescence, though still very dense, is more or less branched or interrupted. This genus has also been supposed to be distinguished by the want of the broad wings of the outer glumes, so conspicuous in the common P. canariensis; but these wings are very narrow in P. paradoxa, Linn., and entirely disappear in P. intermedia, Bosc (P. americana, Ell.), leaving no available character to separate Digraphis generically.

5. ANTHOXANTHUM, Linn., has four or five European species, of which one is now widely spread over various regions of the

globe, but often only as an introduced weed. One at least of the glumes of the lowest pair is the largest of the spikelet, as in Phalaris; those of the second pair, though small and without flowers, have a dorsal awn. The panicle is usually cylindrical and spikelike.

6. HIEROCHLOA, Gmel. (Savastana, Schrank, Disarrenum, Labill., Torresia, Ruiz and Pav.), about eight species from the colder or mountain regions both of the northern and the southern hemispheres, is usually referred to Avenacea next to Holcus; but it appears to me to be much nearer to Anthoxanthum, from which it differs in its looser paniculate inflorescence, and in the glumes of the second pair being but little smaller than the lower ones, and frequently, but not always, enclosing each a male flower. Ataxia, Br., one or two Asiatic and two South-African species, forms a section of Hierochloa, differing slightly from the typical form in the glumes of each pair being more unequal, the lower one only of the second pair (rarely both) having a male flower. A. mexicana, Rupr., seems to connect the two sections.

Tribe VIII. AGROSTEÆ.

The large tribe Agrosteæ is one of the most difficult to circumscribe satisfactorily, or to divide into definite genera. We have taken it nearly in the sense given to it by Trinius, so as to include the Stipeæ, of which other botanists make a distinct tribe; and we have adopted thirty-seven genera, a number which some would extend to above eighty, whilst others might reduce it to about thirty. Their general character is to have a single flower in each spikelet, either apparently terminal as in Panicaceæ, or with a slight bristle-like continuation of the rhachilla beyond it; and from these Panicaces they are constantly distinguished by the pair of empty glumes persistent below the articulation of the rhachilla, without any empty glume or male flower intervening between the articulation and the flowering glume. The single flower in the spikelet, which separates the tribe from the following ones, is not so positive a character, as it occurs also in one genus of Aveneæ, in a few genera of Chlorides, and occasionally in a few exceptional species of some genera of Festuceæ, which cannot well, from inflorescence or other accessory characters, be included in Agrosteæ. There are also two species of Sporobolus which approach the Isantheæ in having frequently two flowers; and in Coleanthus the lower empty glumes are entirely deficient. LINN. JOURN.-BOTANY, VOL. XIX.

H