happening in early youth, are readily repaired. The tail diminishes rapidly behind in vertical dimension, till it ends in a thin point. The entire tail is surrounded by a broad, vertical fin, which commences on the back behind the middle of the trunk, and is supported by innumerable fine cartilaginous rays. There are two fore and two hind paddles, similar to each other in shape and size, and very different from the fins of ordinary fishes. They are covered with small scales along the middle from the root to their extremity, and surrounded by a rayed fringe similar to the vertical fin. These paddles are flexible in every part and in every direction, and too feeble to assist in locomotion on land; they may be of use when the animal crawls in water over the muddy bottom of a creek; but the principal organ for locomotion is the tail, as in tailed Batrachians and the majority of fishes.

The nasal openings and the dentition can be seen only after the mouth has been slit open. The situation of the former within the cavity of the mouth, two on each side, is a very important character, which hitherto had been known in Lepidosiren. The number and form of the teeth has been noticed above, and we have only to add that, beside those molar-like teeth, there are a pair of incisor-like teeth in the fore part of the palate, obliquely inserted in the vomer, and without corresponding teeth in the lower jaw. Knowing the kind of food taken by the Barramunda, we can readily perceive that the incisors will assist in taking up or tearing off leaves, which are then partially masticated between the undulated surfaces of the molars.

With regard to the scales, we may add that a slight difference in their number has been observed in the specimens examined, a difference which, on a more extended examination, may prove to be not constant. The specimens from the Burnett River, to which first the name Ceratodus forsteri has been given, have the middle of the trunk surrounded by eighteen series of scales, whilst there are twenty-one of these series in examples from the Mary River. Consequently the scales appear conspicuously smaller and more numerous in the latter form, which I have named C. miolepis.

The skeleton is cartilaginous; where ossification appears, it is in the form of a more or less thin covering enveloping the cartilaginous substratum, but never taking its origin in the interior or by transmutation of the cartilage. Instead of a vertebral column we find a simple long, tapering chord, without any segmentation, but supporting a considerable number of apophyses. Twenty-seven of them are abdominal and bear well developed ribs. A positively defined boundary between the notochord and the skull does not exist, but in a vertical

section the tapering end of the former may be traced wedged into the basal portion of the skull. The skull is a completely closed cartilaginous capsule, nearly entirely covered with superficial bones, to which, again, some other cartilaginous elements are appended. In the former the confluence of cartilage is so complete that no distinct divisions are traceable by sutures. The tegumentary bones may be designated as, 1, ethmoid; 2, a pair of ossa frontalia; 3, a pair of ossa pterygo-palatina, bearing the upper molars; 4, a single sclero-parietal; 5, an os basale; 6, an os quadratum ; 7, operculum; 8, suboperculum; and 9, mandible. Distinct maxillary and inter-maxillary elements are not developed, but replaced by facial cartilages which pass into the suborbital ring.

On the whole the structure of the skeleton reminds us much of that of the Sturgeons, Chimara, and Lepidosiren, and of all the modifications by which it differs from these allied fishes none is of greater interest than the peculiar structure of the paddles (fig. 2). The fore-paddle is joined to the scapular arch by an oblong cartilage (forearm) and by a broad basal cartilage (carpus). The central part, which we have found externally to be covered with scales, is supported by a jointed axis of cartilage extending from the carpus to the end of the paddle; each joint bears a pair of three, two, or one-jointed branches. The skeleton of the hind-paddles is formed on exactly the same plan. This singular structure is of interest from several points of view : 1. The analogy of this framework to that of the caudal portion of the vertebral column is obvious, Ceratodus being not only truly diphycercal as far as the termination of the body is concerned, but also with regard to the extremity of its paired fins. The many-jointed pectoral axis may be compared to the series of neural and hæmal apophyses, both forming the base to a system of superadded processes (here two or three-jointed branches, there neural and interneural, hæmal and interhæmal spines), which are destined to serve as supports to the surrounding soft parts, and more especially the rays of the fin. Further, as the heterocercal tail of the sturgeon is justly considered to be a later development of the diphycercal form, so the pectoral fin of those fishes proves to be nothing but the heterocercal modification of the diphycercal Ceratodont paddle. 2. The singular filamentary limbs of Lepidosiren prove to be typically the same as the paddles of Ceratodus, but there they are reduced to the simple central cartilaginous axis, with the addition of only rudimentary rays in the African species. 3. Professor Huxley has already drawn attention to the affinity existing between the limbs of Lepidosiren and certain fossil Ganoids, of which impressions f paddles with scaly centres have been preserved (fringed or

lobate fins). Ceratodus clearly proves the correctness of this view, and we are fully justified in supposing that those extinct fishes with lobate fins had them provided with a similar internal skeleton.

As in the other Ganoid and Plagiostomous fishes, the heart of Ceratodus (fig. 4) is provided, in addition to the ordinary two divisions of the fish-heart, with a third contractile chamber, the conus arteriosus. The internal structure of the ventricle and atrium is extremely similar to that of Lepidosiren, but the valvular arrangement in the interior of the conus arteriosus differs considerably, inasmuch as the valves are disposed in two or three transverse rows, of which, however, one only is fully developed.

We have mentioned above that the Barramunda can breathe free air as well as air dissolved in water, and we may infer this from the perfect development of the gills and of a lung. There are four gills on each side; they are broad, lamellated membranes, free from each other, but attached to the outer walls of the gill cavity, which peculiarity is clearly an approach to the fixed gills of the sharks and rays. The pneumatic apparatus may be described either as a single lung, with symmetrical arrangement of its interior, or as two lungs confluent into a single sac without a dividing longitudinal septum. Its interior is divided into about thirty compartments on each side, formed by strong transverse septa and cellular at the bottom; it is evident that by this arrangement the respiratory surface is much increased in extent. This pulmonary sac extends from one end of the abdominal cavity to the other, and terminates anteriorly by a short duct in a glottis, viz. a slit with a valve in the ventral side of the oesophagus, somewhat to the right of the median line. In this respect the Barramunda shows itself to be a true fish, in spite of the presence of a lung, inasmuch as it is obliged to receive the air through the mouth, whilst nearly all Batrachians have the palate perforated by the nostrils, which form a distinct passage for the air used in breathing. When the fish sojourns in pure water, and breathes by the gills, the lung does not differ from the air bladder of other fishes; it then receives arterial blood, returning venous blood, like all the other organs of the body; but when the respiratory function rests with the lung, the pulmonary vein carries purely arterial blood to the heart, where it is mixed with venous blood and distributed to the various organs.

Externally the intestinal tract appears as a wide straight sac without divisions or circumvolutions. Internally it is traversed throughout by a spiral valve performing nine gyrations; the cavity before the commencement of the valve must be regarded as the stomach.

The liver does not show any peculiarity.

The kidneys are paired, the ureters entering a very small urine bladder at the back of, and partly confluent with, the rectum.

The generative organs are paired. Their products pass outwards by a paired oviduct or vas deferens. These ducts are entirely separate from the ovaries or testicles, each having a distinct abdominal orifice immediately below the diaphragma. They accompany the ureters in their posterior course, but are nowhere confluent with them, and terminate in a common opening into the cloacal dilatation, immediately in front of the uretral orifice. The ovaries are elongate bands, their outer surface being crossed by a great number of lamellæ, the bearers of the stroma in which the ova are developed. The ova are in very great number, and when mature drop into the cavity of the peritoneum, as in Salmonoids; but instead of being expelled at once by the peritoneal openings, they travel along a shallow gutter forwards, enter the much convoluted oviducts, where they receive a gelatinous covering, like the eggs of Batrachians, and are finally expelled through the cloaca. The testicles and vasa deferentia are analogous to the female organs with regard to position, form, and orifices.

Such are the principal points in the organisation of the Barramunda; and it remains now to add some remarks on its affinities and on the bearings which the acquaintance of this singular type has upon the advancement of science.

1. When we direct our enquiry at first to recent fishes, there cannot be any doubt as regards the close relationship between Ceratodus and Lepidosiren. The latter had been regarded by Joh. Müller (and by most subsequent ichthyologists) as the type of a separate sub-class-Dipnoi-which he distinguished from the Ganoids by the presence of a pair of longitudinal valves in the conus arteriosus of the heart, the valves being arranged in transverse series in Ganoid and Plagiostomous fishes. We see now that the valvular arrangement in Lepidosiren is merely a modification of the Ganoid heart, and that the characteristic feature of the latter consists in the presence of a pulsating third division-the conus arteriosus. Therefore we are compelled to abandon the sub-class, Dipnoi, and to refer it as a sub-order to the Ganoids, with its definition altered thus: The Dipnoi are Ganoid fishes with the nostrils within the mouth, with paddles supported by an axial skeleton, with lungs and gills, with a notochordal skeleton, and without branchiostegals.

2. But it appears to me that Ganoids and Chondropterygians (sharks and rays) are much more closely allied to each other than either of them to the sub-class of Teleosteans, which comprises the majority of the fishes of the present epoch, and the members of which have, instead of a contractile conus arteriosus, simply a non-contractile swelling of the aorta, separated