as of generic importance, many of the most natural groups in those orders have been broken up, and split down almost to single species, classed into purely artificial tribes and sub-tribes. So, also, a character generally important may, in some instances, separate a single species from a large order with which it may agree in every other respect. The dismemberment of such exceptional species from that order,—as, for instance, that of Phryma from Verbenaceæ,—becomes then purely artificial, and contrary to all principles laid down for natural classification.
This introduction of artificial arrangements, under the disguise of a strict adherence to the rules of the natural system, is much promoted by a tendency to which we systematists are all very liable. It has happened to most close observers to have on some occasion brought forward some character till then comparatively neglected, but which has proved to be eminently useful for establishing natural groups in particular genera, orders, or classes. Such a character is then apt to assume an undue importance in the observer's mind, and to be applied by him indiscriminately throughout the vegetable kingdom. The arrangement of the parts of the floral whorls with relation to the main axis of inflorescence, the aestivation of the floral envelopes, the relative attachment of the floral whorls, and consequent modifications in form of the torus, disk, or floral receptacle; the numbers absolute and relative of the parts in the several floral whorls, the position of the ovary with relation to the rest of the flower, that of the ovules with relation to the ovary, the structure of the fruit, and even the most important of all, the relation of the embryo to the seed, and, the seat of deposit of starch for supplying the first nutriment to the growing embryo—whether as albumen around it, or in its cotyledons, or in the intermediate point (the collet) between the radicle and cotyledons—all characters which more or less generally mark out large and highly natural orders, have nevertheless, each in their turn, on some occasion or other, been applied too strictly, so as to dissever groups otherwise most natural.
On the other hand, however closely we follow natural indications, our system must be to a certain degree artificial. A purely natural method of arranging species and genera is impossible; at least, none has ever been brought forward. The affinities and cross-affinities of plants are so complicated and intertwined, that we have no method of representing them either by a linear series, or by mapping them out on a plane surface. Many of the most natural groups have no definite limits; and yet, to form any clear idea of them for the purpose of study, we must assign limits. The truly German idea of taking one species or genus as a normal type of a genus or order, and grouping others around it as more perfect, or reduced, or collaterally aberrant forms, leads to no practical results. However well it may read in chamber speculations, it produces nothing but confusion when applied to the actual grouping of species. There is no plant which arguments like those usually brought forward may not show equally well to be an aberrant form of almost any number of different types. The absurdity of such a system appears to me never to have been so fully exemplified as in an elaborate
