Separation of the Artificial and Natural Methods in Ichthyology.— It has already been said that all so-called artificial methods of classification must be natural, at least as to the narrowest members of the system; thus the artificial Linnæan method is natural as to species, and even as to genera. And on the other hand, all proposed natural methods, so long as they remain unmodified, are artificial as to their characteristic marks. Thus a Natural Method is an attempt to provide positive and distinct characters for the wider as well as for the narrower natural groups. These considerations are applicable to zoology as well as to botany. But the question, how we know natural groups before we find marks for them, was, in botany, as we have seen, susceptible only of vague and obscure answers :-the mind forms them, it was said, by taking the aggregate of all the characters; or by establishing a subordination of characters. And each of these answers had its difficulty, of which the solution appeared to be, that in attempting to form natural orders we are really guided by a latent undeveloped estimate of physiological relations. Now this principle, which was so dimly seen in the study of vegetables, shines out with much greater clearness when we come to the study of animals, in which the physiological relations of the parts are so manifest that they cannot be overlooked, and have so strong an attraction for our curiosity that we cannot help having our judgments influenced by them. Hence the superiority of natural systems in zoology would probably be far more generally allowed than in botany; and no arrangement of animals which, in a large number of instances, violated strong and clear natural affinities, would be tolerated because it answered the purpose of enabling us easily to find the name and place of the animal in the artificial system. Every system of zoological arrangement may be supposed to aspire to be a natural system. But according to the various habits of the minds of systematizers, this object was pursued more or less steadily and successfully; and these differences came more and more into view with the increase of knowledge and the multiplication of attempts.

Bloch, whose ichthyological labors have been mentioned, followed in his great work the method of Linnæus. But towards the end of his life he had prepared a general system, founded upon one single numerical principle; the number of fins; just as the sexual system of Linnæus is founded upon the number of stamina; and he made his subdivisions according to the position of the ventral and pectoral fins; the same character which Linnæus had employed for his primary

division. He could not have done better, says Cuvier," if his object had been to turn into ridicule all artificial methods, and to show to what absurd combinations they may lead.

Cuvier himself, who always pursued natural systems with a singularly wise and sagacious consistency, attempted to improve the ichthyological arrangements which had been proposed before him. In his Règne Animal, published in 1817, he attempts the problem of arranging this class; and the views suggested to him, both by his successes and his failures, are so instructive and philosophical, that I cannot illustrate the subject better than by citing some of them.

"The class of fishes," he says," "is, of all, that which offers the greatest difficulties, when we wish to subdivide it into orders, according to fixed and obvious characters. After many trials, I have determined on the following distribution, which in some instances is wanting in precision, but which possesses the advantage of keeping the natural families entire.

"Fish form two distinct series;-that of chondropterygians or cartilaginous fish, and that of fish properly so called.

"The first of these series has for its character, that the palatine bones replace, in it, the bones of the upper jaw: moreover the whole of its structure has evident analogies, which we shall explain.

"It divides itself into three ORDERS:

"The CYCLOSтOMES, in which the jaws are soldered (soudées) into an immovable ring, and the bronchiæ are open in numerous holes. "The SELACIANS, which have the bronchiæ like the preceding, but not the jaws.

"The STURONIANS, in which the bronchiæ are open as usual by a slit furnished with an operculum.

"The second series, or that of ordinary fishes, offers me, in the first place, a primary division, into those of which the maxillary bone and the palatine arch are dovetailed (engrenés) to the skull. Of these I make an order of PECTOGNATHS, divided into two families; the gymnodonts and the scleroderms.

"After these I have the fishes with complete jaws, but with bronchia which, instead of having the form of combs, as in all the others, have the form of a series of little tufts (houppes). Of these I again form an order, which I call LOPHOBRANCHS, which only includes one family.

p. 108.

14

Règne Animal, vol. ii. p. 110.

“There then remains an innumerable quantity of fishes, to which we can no longer apply any characters except those of the exterior organs of motion. After long examination, I have found that the least bad of these characters is, after all, that employed by Ray and Artedi, taken from the nature of the first rays of the dorsal and of the anal fin. Thus ordinary fishes are divided into MALACOPTERYGIANS, of which all the rays are soft, except sometimes the first of the dorsal fin or the pectorals;-and ACANTHOPTERYGIANS, which have always the first portion of the dorsal, or of the first dorsal when there are two, supported by spinous rays, and in which the anal has also some such rays, and the ventrals, at least, each one.

"The former may be subdivided without inconvenience, according to their ventral fins, which are sometimes situate behind the abdomen, sometimes adherent to the apparatus of the shoulder, or, finally, are sometimes wanting altogether.

"We thus arrive at the three orders of ABDOMINAL MALACOPTERYGIANS, of SUBBRACHIANS, and of APODES; each of which includes some natural families which we shall explain: the first, especially, is very numerous.

"But this basis of division is absolutely impracticable with the Acanthopterygians; and the problem of establishing among these any other subdivision than that of the natural families has hitherto remained for me insoluble. Fortunately several of these families offer characters almost as precise as those which we could give to true orders.

"In truth, we cannot assign to the families of fishes, ranks as marked, as for example, to those of mammifers. Thus the Chondropterygians on the one hand hold to reptiles by the organs of the senses, and by those of generation in some; and they are related to mollusks and worms by the imperfection of the skeleton in others.

"As to Ordinary Fishes, if any part of the organization is found more developed in some than in others, there does not result from this any pre-eminence sufficiently marked, or of sufficient influence upon their whole system, to oblige us to consult it in the methodical arrangement. "We shall place them, therefore, nearly in the order in which we have just explained their characters."

I have extracted the whole of this passage, because, though it is too technical to be understood in detail by the general reader, those who have followed with any interest the history of the attempts at a natural classification in any department in nature, will see here a fine example of the problems which such attempts propose, of the difficul

ties which it may present, and of the reasonings, labors, cautions, and varied resources, by means of which its solution is sought, when a great philosophical naturalist girds himself to the task. We see here, most instructively, how different the endeavor to frame such a natural system, is from the procedure of an artificial system, which carries imperatively through the whole of a class of organized beings, a system of marks either arbitrary, or conformable to natural affinities in a partial degree. And we have not often the advantage of having the reasons for a systematic arrangement so clearly and fully indicated, as is done here, and in the descriptions of the separate orders.

This arrangement Cuvier adhered to in all its main points, both in the second edition of the Règne Animal, published in 1821, and in his Histoire Naturelle des Poissons, of which the first volume was published in 1828, but which unfortunately was not completed at the time of his death. It may be supposed, therefore, to be in accordance with those views of zoological philosophy, which it was the business of his life to form and to apply; and in a work like the present, where, upon so large a question of natural history, we must be directed in a great measure by the analogy of the history of science, and by the judgments which seem most to have the character of wisdom, we appear to be justified in taking Cuvier's ichthyological system as the nearest approach which has yet been made to a natural method in that depart

ment.

The true natural method is only one: artificial methods, and even good ones, there may be many, as we have seen in botany; and each of these may have its advantages for some particular use. On some methods of this kind, on which naturalists themselves have hardly yet had time to form a stable and distinct opinion, it is not our office to decide. But judging, as I have already said, from the general analogy of the natural sciences, I find it difficult to conceive that the ichthyological method of M. Agassiz, recently propounded with an especial reference to fossil fishes, can be otherwise than an artificial method. It is founded entirely on one part of the animal, its scaly covering, and even on a single scale. It does not conform to that which almost all systematic ichthyologists hitherto have considered as a permanent natural distinction of a high order; the distinction of bony and cartilaginous fishes; for it is stated that each order contains examples of both.15 I do not know what general anatomical or physiological

15 Dr. Buckland's Bridgewater Treatise, p. 270.

truths it brings into view; but they ought to be very important and striking ones, to entitle them to supersede those which led Cuvier to his system. To this I may add, that the new ichthyological classification does not seem to form, as we should expect that any great advance towards a natural system would form, a connected sequel to the past history of ichthyology;-a step to which anterior discoveries and improvements have led, and in which they are retained.

But notwithstanding these considerations, the method of M. Agassiz has probably very great advantages for his purpose; for in the case of fossil fish, the parts which are the basis of his system often remain, when even the skeleton is gone. And we may here again refer to a principle of the classificatory sciences which we cannot make too prominent;-all arrangements and nomenclatures are good, which enable us to assert general propositions. Tried by this test, we cannot fail to set a high value on the arrangement of M. Agassiz; for propositions of the most striking generality respecting fossil remains of fish, of which geologists before had never dreamt, are enunciated by means of his groups and names. Thus only the two first orders, the Placoïdians and Ganoïdians, existed before the commencement of the cretaceous formation: the third and fourth orders, the Ctenoïdians and Cyclordians, which contain three-fourths of the eight thousand known species of living Fishes, appear for the first time in the cretaceous formation and other geological relations of these orders, no less remarkable, have been ascertained by M. Agassiz.

But we have now, I trust, pursued these sciences of classification sufficiently far; and it is time for us to enter upon that higher domain of Physiology to which, as we have said, Zoology so irresistibly directs

us.

[2nd Ed.] [I have retained the remarks which I ventured at first to make on the System of M. Agassiz; but I believe the opinion of the most philosophical ichthyologists to be that Cuvier's System was too exclusively based on the internal skeleton, as Agassiz's was on the external skeleton. In some degree both systems have been superseded, while all that was true in each has been retained. Mr. Owen, in his Lectures on Vertebrata (1846), takes Cuvierian "characters from the endo-skeleton, Agassizian ones from the exo-skeleton, Linnæan ones from the ventral fins, Müllerian ones from the air-bladder, and combines them by the light of his own researches, with the view of forming a system more truly natural than any preceding one.

As I have said above, naturalists, in their

progress towards a Natural