States of North America, and may well be borne in mind in connection with geological speculations, as may also the extraordinary proportion of Ferns. Of these there are in New Zealand 31 genera, including no fewer than 121 species, of which nearly a dozen may be styled Tree-ferns.

This latter ferny peculiarity is one of the most striking in the Flora, and we shall be glad to see it referred to by Dr. Hooker in his promised essay.

LXVI.-AGARDH'S CLASSIFICATION OF PLANTS.

THEORIA SYSTEMATIS PLANTARUM, &c. Auctore J. G. Agardh, Botanices in Academia Lundensi Professore. Lundae, 1858.

THE work of the younger Agardh, to which we now propose to direct the attention of our readers, though published in 1858, has not been noticed in our pages. It seems to us, indeed, that it has by no means attracted that amount of attention, in this country at least, which it deserves. Its subject is the classification of plants, and though its principles will certainly not be accepted as a final solution of the question, yet as the views which it contains are not only in a great measure new, and on that account deserving of careful study, but are based upon a very complete knowledge of the structure of plants, and are everywhere illustrated by references to facts, the work is valuable as a contribution to our knowledge, and should certainly be carefully studied by all systematists.

J. G. Agardh, who is one of the most learned Algologists of the day, had found in the course of his researches on the classification of that extensive family of Cryptogams, in which there is so great a diversity of both vegetative and reproductive organs, that the greatest possible differences of opinion existed among systematists as to the value of different characters as indicative of affinity. The uncertainty, which thus met him as it were on the threshold of his attempts to classify Algae, led him to institute an inquiry into the principles of biological classification in general, which, in its results, led him to the conclusion, that in botany, at least, our systematists are pursuing a wrong course, which tends more and more in the direction of that artificial system, from which we had all hoped that we had emancipated

ourselves, instead of leading to the discovery of a truly natural classification of which we have been in search since the days of Ray and Linnaeus.

The work before us, which is written in Latin, and extends to about 400 octavo pages, not being intended as an elementary work, but addressed to advanced students of the classification of plants, does not give detailed characters of the natural families. These are enumerated in succession, some of their more essential characters are briefly indicated, their various relationships as they appear to the author pointed out, and where they differ from the ordinary view, fully explained, illustrated by examples, and supported by arguments. The number of distinct families enumerated is greater than usual, being about 480, of which 400 are Dicotyledons, and the remainder Monocotyledons. The essential distinction between Professor Agardh's system and those in ordinary use is, as we shall have occasion to explain in detail, that he considers the different lesser groups which are commonly included in many of our larger (generally thought) natural families, not to be closely related to each other, but only to present analogical resemblances. These groups, which we call tribes or sub-families, assume with him the most prominent position.

There is no attempt at collecting the families into large groups, unless, indeed, certain lines of separation drawn between the groups in a tabular arrangement of affinity at the end of the book be supposed to indicate the first sketches of a higher classification. This grouping of the families into classes is, to say the least, very vague, as if the author thought that the time had not come for carrying it out successfully. This vagueness is probably the cause of the unmerited neglect with which the work has been received. Had the families been grouped together into classes, according to a clear and well-defined system, however artificial, it would, we think, have attracted much more attention. This, however, would have been, as we shall see, too repugnant to Prof. Agardh's principles. Though ready to admit the value, and even to advocate the employment of an artificial system for the determination of the names of plants, he strongly condemns its use in any part of a truly natural system.

It is somewhat startling at the present day to be told that all the larger groups above the natural families, in the sense in which Professor Agardh understands that term, are quite artificial. That, for instance, there is no near affinity between Cruciferae and Papaveraceae, between Rhamnaceae and Celastraceae, between Umbelliferae

and Corneae, between Campanulaceae and Compositae, between Coniferae, Gnetaceae and Cycadeae, nay even, though this is stated with some hesitation, between the different minor groups which are commonly classed together in Rosaceae and Leguminosae. There is no doubt, however, that these conclusions, however paradoxical, are no crude hypotheses, but are based upon an extremely careful study of structure. In every part of the work we have ample proof of our author's complete familiarity with his subject. Even, therefore, though his readers should, as is most likely, concur with us in declining to accept many or even any of his conclusions, there is much to be learned from the work. There are occasional happy approximations of Orders hitherto considered to occupy distant places in the system, and in almost every page useful hints from which the systematist may derive much profit.

To some of these we may, should space permit, return below, were it but to give a few examples to induce our readers to study the work. Our first object must be to give a general idea of the principles of Professor Agardh's system, as explained by himself in a series of somewhat unconnected essays prefixed to the work, and extending to 96 pages.

He begins by referring briefly to the views of authors on natural classification, and without attempting a rigid definition of a species, tells us, that we must for the present assume its existence as an axiom, without which it is impossible to reason at all on the subject of classification. In thus assuming the existence of species, he is careful to avoid expressing a belief in their permanence, and, indeed, indicates it as his opinion that species in the course of ages are capable of variation, and that those existing at the present day are the direct descendants of those found fossil throughout the whole series of geological ages.

Starting then from the fact that species, as they now exist, are definite entities, he proceeds to show that genera, families, and even higher groups are also natural, but that the higher the group the more difficult it is of definition. Such groups, however, as Salix, Eryngium, Labiatae, Umbelliferae, he says, are as natural as species. To illustrate natural groups of higher value he takes his examples from the animal kingdom only, those he gives being birds and fishes. The two largest natural associations are plants and animals. In the vegetable kingdom he thinks that the higher groups are of little or no value, that the large divisions of Jussieu and De Candolle are quite artificial, and the attempts made by Brown, Agardh, Lindley, A. de

Jussieu, and Brongniart, to bring the families together in classes are little if at all better. The distinction between the artificial system of Linnæus and the so-called natural system he states to consists only in the grouping of plants into families. In both the genera are natural.

The existence of natural groups being thus established, and admitting the facility with which natural genera and, to a great, but less, extent natural families may be recognised by the practised naturalist, he goes on to inquire according to what principles these families are to be grouped together. Rejecting altogether the hypothesis of a single continuous series, which he seems to think has found so much favour with many of those who, in our day, have tried to establish a natural system of plants, he proceeds to consider that of a net-like or reticulated series, in which a single family of plants may have numerous and very various affinities passing in different directions into many different families. To this, likewise, he is not prepared to assent, on grounds which he more fully explains elsewhere in giving his views of analogy and affinity. "The mistake," he says, "arises from the "erroneous belief, that it is possible to judge of the affinity of a large group by means of the smaller, that for the determination of the relationships of an order the characters of one genus are to be "looked to, and in the case of a genus those of a species. Though single genera of a family may appear to have intimate relations "with single genera of another, or of many others, we are not "justified in thence drawing a conclusion as to the relationships of "the families to which these genera belong. It is not Amphioxus or "Salmo, but the class Pisces which is near Amphibia, Mollusca are "contiguous to Vertebrata, the Vegetable kingdom to the Animal "kingdom."

"To me," he says, (in the next paragraph, p. xxxviii) "nature appears neither a simple nor a reticulated series, but an infinite "and innumerable multitude of series advancing from a lower to a "higher grade, from any part of which lesser series may project like rays, some diverging more, some less, much as the trunk of a tree is "divided into larger branches, these again into lesser and lesser ones, "and at last into an almost infinite number of leaves. One branch "of the tree becomes thick and strong, ramifies much, and reaches "the top of the tree, while another remains weak, and a third may "be scarcely developed at all." And again: "In determining natural "affinities we must take nature as our guide in everything. It is "not enough to take into consideration all the characters of the

"forms which we are investigating, and to develop the essence (vim "et naturam) of the family from a knowledge of all the forms belonging to it. We must further seek to discover the direction of "the evolution of each series, and inquire whether its forms are "advancing this way or that. It is only after this has been done "that we can decide to which series a plant should be referred, and "whether apparent resemblances are to be considered affinities or "analogies."

These extracts give in a condensed form the general nature of the principles on which Professor Agardh bases his system. Forms which are truly allied or present the same type of organisation, he considers to form a number, indeed a very great number, of linear series, in each of which the type appears under a variety of forms, ascending from a low degree of imperfection to a highly perfect condition. A great many resemblances may exist between organisms of different types, but these are analogical and do not indicate affinity. Though he no where makes use of the image, and, indeed, seems as we have seen above to prefer that of the branches of a tree, we should conceive that the radii of a circle, or still more those of a sphere, would best express the general nature of his views of the relationships of plants. On each of these rays the lowest plants are supposed to be nearest the centre, and the higher modifications of the same type to occupy more and more distant positions. Some of the rays, as we gather from our text, may fork after the start from the centre, but once separated no reunion like that indicated by a reticulated arrangement is possible. The lower forms of different series are closer to each other than those of higher type, and those of different type are therefore more likely to be mixed up together among the lower organisms than among the higher. In the different rays, which thus diverge more and more from each other as they recede from the centre, resemblances of analogy, not of affinity, exist between the species which have attained the same degree of progress; so that a series of concentric circles, which intersect all the rays, will, where they cut each, connect together species with certain superficial resemblances though belonging to distinct types. All the species in each of these rays belong to different degrees of development of the same type. Those on different rays, however much they may appear alike in one or more points, are not nearly related. Groups of greater or less value may of course be substituted for species, as each group for purposes of classification must be considered as a single entity, an abstraction, formed by a condensation as it were of the characters of all the subordinate parts.