as to form but an indifferent union, which, as already observed, Professor Martius proposes to amend by the elevation of this section to a distinct order.
The fourth and last section, Symphonieae, of Choisy and DeCandol'le's arrangement is also objectionable for the same reasons as the preceding, viz., its quinary proportion, besides which Canella, one of the genera referred to it, has alternate leaves. The stamens in this tribe are united into a tube as in Meliaceae, but differ from that order in the extrorse dehiscence of the anthers, and in so far might perhaps be advantageously separated from both to form a new order, the more so, as their properties are totally different from either ; those of Canella alba one of the tribe, being intensely aromatic. Species presenting differences so marked no Botanist would ever think of combining in the same genus, and I cannot understand on what principle genera, in which they occur, should be admitted into the same natural order, since, such combinations can only tend to prove the futility of the name by setting natural affinities at defiance.
On the principle of preserving simplicity of character, and at the same time conformity to that character among the genera referred to the order, I propose, though closely allied in habit, to separate the genus Stalagmitis or Xanthochymus from Guttiferce. A binary arrangement of the parts of the flower, (2 and its multiples) forms the essential characteristic of the order, 2-4-6 sepals and petals, 2-4-6-8 carpels or cells of the ovary,' &c. but in Xantho- chymus a quinary one prevails, 5 sepals and petals, 5 fascicules of stamens a 5, or by abortion 3, ceiled ovary, form the characteristics of that genus.
The want of uniformity between the characters of the genus and of the order is here most striking, and is such as to render it next to impossible for any one unacquainted with the genus to refer it, by its characters, to the order in which it is placed. To such anomalies much of the difficulty attending the study of the natural system of Botany is owing. Giving due weight therefore to characters derived from the number and arrangement of parts, it follows, that this genus must be removed from the order, and referred to some one in which a quinary arrange- ment prevails, such as Hypericinece, or be made the type of a separate order. To me the for- mer course seems the preferable one, since the only point of difference between the characters of the genus and the order, consists in the solitary ovules of the first, which in the last are usually, though not always, numerous, and in the structure of the embryo which partakes more of the character of Guttiferce than Hypericinece. In both a quinary order of parts exist, in both the fila- ments are united, forming androphores, and in both, we find a 5 celled ovary, with occasionally de- finite ovules. In a word the genus is much more closely allied to Vismice in its characters, than to any tribe of Guttiferae, and to that section of HypeHcineae I think it ought to be removed.
On the principle here insisted upon, that no genus be admitted into the order in which a binary order of parts is not found to exist throughout the whole floral organization, it follows, that the order as constituted by both Cambessides and Meisner, its most recent expositors, must be entirely broken down, and the portion left to bear its name, reduced within very narrow limits. As I have not Cambessides' memoir to refer to, I follow iVleisner's exposition of his (Cambessides) dis- tribution of the order. Here Choisy's tribes are retained, but some of the genera transposed — Mesua for example, one of Choisy's Calophylleae, is removed to Clusieae, and though strictly binary in its structure, is placed between two genera, in which the quinary proportion prevails. Gynotroches, a genus of Blume, allied in many respects to Garcinia, and certainly I think referable to the same section in a natural distribution of the order, is placed somewhat artificially, in Symphonieae, and in direct opposition to the original character of that tribe which is to have the stamens united into a tube. To Garcinieae the genus Stalagmitis, including Roxburgh's Xanthochymus, and the larger half of Choisy and DeCandolle's genus Garcinia, (a most incongruous assem- blage) is referred, in place of to Calophylleae, in which it and Xanthochymus had been placed by Choisy. Calophylleae thus shorn of three of the four genera originally referred to it still remains, and to supply the place of those removed, two others, which have, since the publication of Choisy's memoir been added to the order, viz., Kayea, Wall, and Apolerium, Blume, are given.
In these remarks I have confined myself to the Indian genera of the order, with which only I am well acquainted, but, judging from the characters of some of the Ame- rican ones placed here, I cannot but think that most of them ought to be excluded, and the order limited in a great measure to Asiatic species, and I feel but little doubt, when it has been subjected to a thorough revision, that such will be the result. Hitherto, it seems to
