The generic characters are very long,—that of the first genus, for instance, Allamanda, occupies 54 lines, and contains details which cannot be essential to the genus, nor can have been verified in the majority of species; such, for instance, as "Rostellum . . . circa quintuplo cotyledonibus brevius," when the seeds have only been seen in three out of ten species. It is true that imperfectly known species must often be referred to a genus without verifying all its characters, subject to a subsequent removal, if found to differ in essential points; but it surely cannot be intended that any species the radicle of whose embryo should be 13rd, instead of 15th the length of the cotyledons, should on that account alone be generically separated from other Allamandas.
Practically, however, these generic characters must be taken as descriptive, rather than diagnostic; and the inconvenience of their great length for distinctive purposes is, in the present work, obviated by an excellent synoptical table of genera, remarkably clear, both as to matter and type. But no such assistance is afforded in the case of species. In the larger genera, even after their subdivision into groups, there remain series of ten, twelve, or more species, without any contrasted characters, to guide the reader, but such as he can glean from so-called diagnoses, which, far from being confined to the Linnean limits of twelve words, have an average length of twelve to fourteen lines, and are, in fact, detailed descriptions in the ablative case, of almost all characters, except colour and dimensions, which, in true orthodox style, are specially reserved for a separate paragraph in the nominative case. This is a growing evil in almost all modern systematic works, and for which we see no remedy but a rigid return to the Linnean rule, accompanied by repeated sectional subdivisions, or a total abandonment of the system of technical diagnoses, substituting synoptical tables, followed by detailed descriptions.
In the generic arrangement of the Brazilian Apocynaceæ, Dr. Müller appears to have much improved on his predecessors. The general division of the order into two main groups, founded on the structure of the anthers, is, we believe, new. It appears natural; and, if duly verefied on the Asiatic, as well as on the American genera, is in every way satisfactory. The tribes are, in other respects, nearly those of the Prodromus. We are glad to see, also, that several genera which we had already set down as not natural, are here suppressed. Collophora (Mart), and Hortsmania (Miq.), are identified respectively with Couma (Aubl.), and Condylocarpon (Desf.), Peschiera and Bonafousia (A. DC.), are reduced to Tabernæmontana; Robbia (A. DC.),to Malouetia, and Thysanthus (Benth.), to Forsteronia. On the other hand, eight new genera are proposed, besides three more South American, but not Brazilian genera, which the author has described at the same time, and published in Mohl and Schlechtendahl's Botanische Zeitung, 1860, p. 21. Of these eleven genera, Elytropus, founded on a Chilian plant, appears to be very distinct. Zschokkea, allied to Ambelania, Couma, and Hancornia, must be maintained, so long as the three latter are kept distinct, although it is not improbable that on a careful comparison
