The sacrum is formed by the ninth vertebra, but there are a few interesting exceptions. Pelobates, Pipa, and Hymenochirus possess two sacral vertebrae; and, neglecting individual abnormalities, these three genera form the only exception amongst recent Amphibia. In the three genera the coccyx is fused with the second sacral vertebra, and such a fusion occurs elsewhere normally only in Bombinator with its single sacral vertebra. The morphologically oldest condition is normally represented by Pelobates, the sacral vertebrae being the tenth and ninth. One case has been recorded by Boulenger of Bombinator pachypus "with eleven segments," the last carrying the ilium. Individual lop-sided abnormalities have been described in Bombinator and Alytes, where the right ilium articulated with the tenth, the left ilium with the ninth vertebra. This shifting forwards of the ilium to the extent of one metamere has been continued further in Pipa, in which the sacrum is formed by the ninth and eighth vertebrae, their diapophyses fusing on either side into extra broad wing-like expansions. In old specimens of Palaeobatrachus fritschi the seventh vertebra is in a transitional condition, the ilium being carried by the ninth and eighth, and slightly also by the diapophyses of the seventh vertebra; and in P. diluvianus the diapophyses of all these vertebrae are united into one broad plate to which the ilia are attached. Lastly, in Hymenochirus the first sacral is the sixth vertebra, and this creature has thereby reduced the pre-sacral vertebrae to the smallest number known.

This shifting forwards of the iliac attachment implies the conversion of original trunk into sacral vertebrae, and the original sacral vertebra itself becomes ultimately added to the urostyle. The second sacral, the tenth of Pelobates, the ninth of Pipa, and the tenth on the right side of the abnormal Bombinator, are still in a transitional stage of conversion. In Discoglossidae the tenth is already a typical post-sacral vertebra, and is added to the coccyx, but it still retains distinct, though short, diapophyses. In the majority of the Anura the tenth vertebra has lost these processes, and its once separate nature is visible in young specimens only. In Bombinator even the eleventh vertebra is free during the larval stage. In fact the whole coccyx is the result of the fusion of about twelve or more vertebrae, which from behind forwards have lost their individuality. We conclude that originally, in the early Anura, there was no coccyx, and that the ilium was attached much farther back; and this condition, and the gradual shifting forwards, supply an intelligible cause of the formation of an os coccygeum. The fact that the sacral vertebrae of the Anura possess no traces of ribs as carriers of the ilia, is also very suggestive. The ilia have shifted into a region, the vertebrae of which had already lost their ribs. By reconstructing the vertebral column of the Anura, by dissolving the coccyx into about a dozen vertebrae, so that originally, say the twenty-first vertebra carried the ilia, we bridge over the enormous gap which exists between the Anura and Urodela. That whole portion of the axial continuation behind the coccyx, more or less coinciding with the position of the vent, is the transitional tail.

The disappearance of both notochord and spinal cord, and the conversion of the cartilaginous elements into a continuous rod in the case of the os coccygeum, find an analogy in the hinder portion of the tail of Dipnoi and Crossopterygii, and in the tail-end of most Urodela, portions which are not homologous with the os coccygeum. The term urostyle should be restricted to such and similar modifications of the tail-end, and this latter happens to be lost by the Anura during metamorphosis.

Strictly speaking, or rather in anatomical parlance, the Vertebrate tail begins with the first post-sacral vertebra. In the Anura that portion of the whole tail has retained most cartilage, and has become the coccygeum, which is required as a "backbone" for the often enormous belly. This requirement is an outcome of the great shortening of the trunk proper (if the trunk be defined as ending with the pelvic region), and this shortening of the trunk is again intimately connected with the jumping mechanism, enlargement of the hind-limbs, elongation of the ilia, and throwing the fulcral attachment forwards as much as possible. The pre-acetabular ilio-sacral connection is carried to the extreme in the Anura.

The shoulder-girdle and "sternum" are more complete than in the Urodela, there being also a pair of clavicles, fused with the precoracoidal bars. The whole apparatus presents two types. In the arciferous type the coracoids and precoracoids retain a great amount of cartilage in their distal portions, and these cartilages (the epicoracoids of some authors) overlap each other movably on one another, the right usually lying ventrally upon the left. The epicoracoidal cartilage of each side, by connecting the distal end of the coracoid with the precoracoid of the same side, forms an arc, hence "arciferous." In the firmisternal type the epicoracoidal cartilages are much reduced, and, instead of overlapping, meet in the middle line and often fuse with each other, forming thereby a firm median bar, which connects the ventral ends of the precoracoids with those of the coracoids. This type is morphologically the higher and more recent, and passes in the larval stage through the arciferous condition. It is restricted to the Ranidae, Engystomatinae, and Aglossa. Although these two types afford an excellent distinctive character for the main divisions of the Anura, they are to a certain extent connected by intermediate forms in such a way, that, for instance, in Bufo and among Cystignathidae in Ceratophrys, the two opposite epicoracoidal cartilages begin to unite at the anterior end.

In many Engystomatinae the precoracoids together with the clavicles are much reduced, sometimes to thin ligaments, being in this case mostly curved back and lying closely against the coracoids; or they may be lost completely. Very rarely the precoracoidal bars are actually much stronger than the coracoids, and the median symphysial bar of cartilage is lost; this is the case in Hemisus.

The scapula is always large and curved into transverse, dorsally broadening blades, the dorsal greater portion of which, the so-called supra-scapula, does not ossify but calcifies.

It is very doubtful if the Anura possess a true sternum, if by sternum we understand a medio-ventral apparatus which owes its origin to the ventral portions of ribs. The so-called sternal apparatus of the Anura consists of two pieces. One, anterior, variously named episternum, presternum, or omosternum, rests upon the united precoracoids and extends headwards, being either styliform or broadened out. Sometimes it is partly ossified, with a distinct suture at its base; this is the case especially in the Firmisternia; in many Arcifera the omosternum remains cartilaginous and is continuous, without a sutural break, with the cartilage of the precoracoids, indicating thereby its genetic relation to the shoulder-girdle. Hence omosternum is the preferable name. It is frequently much reduced, even absent, for instance in most Bufonidae and in the Engystomatinae. The posterior so-called sternal part may be termed metasternum. It forms the posterior counterpart of the omosternum. It is attached behind to the epicoracoidal cartilages, or fusing with them forms their posterior continuation. It appears mostly in the shape of a style, which is frequently ossified, and broadens out behind into a cartilaginous, partly calcified blade. In the Discoglossidae only it diverges backwards into two horns, assuming a striking resemblance to the typical xiphisternum of the Amniota. In young Anura the metasternal cartilage is intimately connected with the pericardium, an indication of its being derived not from ribs but from the shoulder-girdle.

The glenoid cavity is always formed by the coracoids and by the scapula, but the precoracoid often takes part in its formation, for instance in Bufonidae, Hylidae, and Discoglossidae.

In the fore-limb the humerus has a crest, stronger in the males than in the females; it assumes extraordinary strength in some Cystignathidae, notably in the male Leptodactylus. Radius and ulna are fused into one bone. The carpalia are originally nine in number: radiale, ulnare, two centralia, and five carpalia distalia, the fifth of which is reduced to a tiny nodule or to a ligamentous vestige. The primitive condition still prevails in the Discoglossidae. In most of the other Anura the fourth and third distal carpalia, in any case very small, fuse with the enlarged ulnar centrale; the radial centrale comes, in the Bufonidae and Pelobatidae, into contact with the radius, so that the forearm articulates with three elements as in the Urodela, but with this difference, that the intermedium of the Urodela has been lost by the Anura. There are five metacarpalia and five fingers, but the elements of the first or thumb are nearly vestigial, so that the pollux is reduced to one or two nodules, scarcely visible externally. The normal number of the phalanges of the second to fifth fingers is 2, 2, 3, 3. The distal phalanges are generally straight, either pointed or expanded or with Y or T-shaped ends; but in the Hylidae, in Hylambates amongst the Ranidae, and in Ceratohyla, one of the Hemiphractinae, the terminal phalanges are produced into curved claws which support the adhesive finger-discs. There are, however, many genera of different families, which possess finger-discs and have no claw-shaped phalanges. The Hylidae, and many of the climbing members of the Ranidae with adhesive discs, possess an extra skeletal piece intercalated between the last and last but one phalanges of the fingers and toes. This piece, a mere interarticular cartilage in Hyla, is in the following Raninae developed into an additional phalanx, so that their numbers are 3, 3, 4, 4 in the hand and 3, 3, 4, 5, 4 in the foot: Cassina, Hylambates, Rappia, Megalixalus, Rhacophorus, Chiromantis, Ixalus, and Nyctixalus. All the other Ranidae are without this additional phalanx, irrespective of the presence or absence or size of digital expansions.^[12]

The pelvic girdle looks like a pair of tongs (see Fig. 4, p. 22). The ilium is enormously elongated and is movably attached to the sacral diapophyses. This connection is always pre-acetabular in position. The ilium and ischium co-ossify completely, and make up nearly the whole of the pelvis; the pubis is very small, and remains cartilaginous unless it calcifies. It rarely possesses a centre of ossification, for instance in Pelobates, where the osseous nodule is excluded from the acetabulum, recalling certain Labyrinthodonta, whose ossa pubis likewise do not reach that cavity. The latter is open or perforated in young Anura and remains so in the Discoglossidae, but in the others it becomes closed up as in the Urodela. The ventral halves of the pelvis, besides forming a symphysis, closely approach each other, just leaving room for the passage of the rectum and the urino-genital ducts.

The hind-limbs are in all cases longer than the fore-limbs. The femur is slender, the tibia and fibula are fused into one bone. The tarsus is much modified by the great elongation of the two proximal tarsalia (there being no intermedium) into an astragalus and a calcaneum, both of which fuse together distally and proximally, or completely as in Pelodytes; in the latter case the limb assumes a unique appearance, since it consists of three successive and apparently single bars of nearly equal length. The other tarsal elements, especially the more lateral ones, are practically reduced to pads. The Anura have thereby acquired two well-marked joints, one cruro-tarsal, the other tarso-metatarsal; this shows a high stage of specialisation in comparison with the Urodelous and Stegocephalous type of still undefined joints.

The Anura possesses five well-developed toes with normally 2, 2, 3, 4, and 3 phalanges, and the rudiments of a sixth digit, the so-called prehallux, which consists of from two to four pieces, including the one which represents its metatarsal. This prehallux, as a vestige of a once better developed digit, is exactly like the elements on the radial side of the wrist, which, we are certain, are the remnants of a once complete finger, namely the pollex. The only weighty difficulty against its interpretation as a prehallux lies in the fact that hitherto no six-toed Stegocephali have been found; but the fact that there are no Stegocephali known with more than four fingers could be used as an argument against there being a pollex-vestige in recent Anura with just as little reason.

The skull of the Anura differs from that of the other recent Amphibia in the following features:–

The orbital region of the primitive cranium remains cartilaginous, but further forward the cranial cavity is closed by the unpaired sphenethmoid, which forms a ring round the anterior portion of the brain-cavity, hence called "os en ceinture" by some anatomists. The frontals and parietals fuse into one pair of fronto-parietal bones, and these again can fuse together in the middle line; as in Aglossa and Pelobates. The palatal portion of the palato-quadrate cartilage is complete, reaching forwards to the sides of the ethmoid region. The curved arch, formed by this cartilage, is covered by the following bones: (1) the quadrato-jugal, reduced to a thin splint which connects the quadrate and squamosal with the posterior end of the maxilla; (2) the pterygoid, always strong, extending from the distal inner corner of the quadrate to the maxilla, sometimes also to the palatine, and with a broad, median process to the parasphenoid, this process covering ventrally most of the otic region; (3) the palatines, which vary considerably in shape and size; they are placed transversely and meet in the middle line; in Bombinator and Pelodytes they are absent.

The quadrates are directed transversely and backwards, in conformity with the wide gape of the mouth. The squamosal is always well developed, covering the whole of the quadrate on its outer side; it has a forwardly directed process which ends freely in Rana, meets a corresponding process of the maxilla and forms a bony arch with it in Discoglossus, Pelobates, and others, or is scarcely developed at all, for instance in Bufo. In Pelobates cultripes the squamosal is very wide and forms a junction with the fronto-parietals, thus producing a broad bridge across the temporal fossa.

The nasal bones are large and meet in the middle line. Frequently they leave a space between them and the diverging anterior portion of the fronto-parietals, through which gap appears part of the dorsal surface of the ethmoid cartilage. A fontanelle between the frontals occurs in most Hylidae, many Cystignathidae, some few Bufonidae, in Pelodytes amongst the Pelobatidae, and in the Discoglossidae.

The tympanic cavity is bordered in front, above, and below by the squamosal and quadrate, behind by the musculus depressor mandibulae, internally by the otic capsule, and by the cartilage of the cranium between this and the lateral occipital bone. The cavity communicates, however, by the wide and short Eustachian tube with the mouth, the passage being bordered anteriorly by the pterygoid, posteriorly by soft parts. Partly imbedded in these soft tissues is the styloid process or stylohyal, which is attached to the cranium, mostly behind the otic region, and is continued downwards into the anterior horn of the hyoid. The whole partly cartilaginous, ligamentous, and osseous string is, in fact, the entire ventral half of the hyoid arch, while the dorsal half or hyomandibular portion of this, the second visceral arch, is modified into the columellar or auditory chain. The inner end of this chain, the stapes, is inserted into and around the fenestra ovalis of the otic capsule, while the outer end is somewhat T-shaped, and is loosely attached to or near the upper rim of the tympanic ring and to the middle of the tympanic disc. In many Anura this terminal bar can be seen from the outside. The middle portion of the columellar chain is ossified, the rest remains cartilaginous. But the whole chain exhibits various modifications in different genera, especially in the number and the extent of the processes sent out by the outer cartilaginous portion; these are attached in various ways to the tympanum and its rims. The tympanic disc is carried by a cartilaginous ring, which rests against a special process sent out by the quadrate, and is probably itself a differentiation of this element.

In some very aquatic genera, but also in Pelobates, the tympanic cavity is much reduced, for instance in Bombinator, Liopelma. In Batrachophrynus not only the cavity, but also the Eustachian tubes are suppressed. In the Aglossa only the two tubes are united into one short but wide median canal, opening at the level of the pterygoids on the roof of the mouth.

The lower jaw is remarkable for the possession of mento-Meckelian cartilages, absent only in the Aglossa and Discoglossidae. At first they are much longer than the rest of the jaw; during the larval life they indeed form the functional jaw, and they are now covered with horny sheaths instead of teeth. Owing to the absence of teeth on them, these mento-Meckelian cartilages are later not invested by bone, although in many Anura they ultimately ossify, either retaining their separate nature or fusing partly with the dentary bones. The bulk of the lower jaw, the Meckelian cartilage, becomes invested by the dentary, a small articulare, and an inner angulare, while a splenial element is absent. The dentary itself is mostly reduced to a small dentigerous splint, while the angulare forms by far the greater part of the bony jaw.

Teeth are more restricted in their occurrence than in the Urodela. On the jaws they always stand in one row. With the exception of the Hemiphractinae, Amphignathodontinae, Ceratobatrachinae, and Genyophryninae, no recent Anura carry teeth on the lower jaw, and even in these genera they are mostly much reduced in size and firmness, having all the appearance of vanishing structures. The premaxillae and maxillae are frequently furnished with teeth, except in the Dendrobatinae, Genyophryninae, Engystomatinae, Dendrophryniscinae, Bufonidae, Pipa, and Hymenochirus. The vomers mostly carry a series of teeth on their posterior border; when these teeth are absent, as in many species of Bufo, a kind of substitute sometimes occurs on the palatines in the shape of a row of tuberosities. The palatines carry teeth in Hemiphractinae. The parasphenoids are rarely toothed, e.g. Triprion, Diaglena, Amphodus, and occasionally in Pelobates.

A few Anura possess peculiar substitutes for teeth in the anterior portion of the lower jaw, namely, a pair of conical bony processes, sometimes rather long, but always covered by the dense gums, or investment of the jaws; e.g. Lepidobatrachus, several Rana, e.g. R. adspersa, R. khasiana, R. kuhli, and Cryptotis brevis.

Cranial dermal ossifications are developed in some species of Bufo, still more in the Hemiphractinae, and above all in Pelobates cultripes and in the Cystignathoid genus Calyptocephalus.

The hyoid apparatus of the Anura is complicated. It is originally composed of the hyoidean and four branchial arches, with one median, copular piece. The branchial arches form in the early life of the tadpole the elaborate framework of the filtering apparatus mentioned on p. 44. During metamorphosis the whole filter disappears, owing to resorption of the greater part of the branchial arches; only their median portions remain, and fuse with the enlarged copular piece and the hyoidean arches into a broad shield-shaped cartilage (corpus linguae), whence several lateral processes sprout out, the posterior pair of which are generally called thyrohyals or thyroid horns. The true hyoid horns give up their larval lean-to articulation with the quadrate, become greatly elongated, and gain a new attachment on the otic region of the cranium. The transformation of the whole apparatus has been studied minutely by Ridewood, in Pelodytes punctatus.^[13]

Skin

The epidermis of the young larvae of Amphibia is furnished with cilia, which later on are suppressed by the development of a thin hyaline layer or cuticula, but clusters of such cilia remain, at least during the larval life and during the periodical aquatic life of the adult, in the epidermal sense-organs. In the frog, currents are set up by the ciliary action at an earlier stage, and are maintained to a later stage than in the newt. In the latter the tail loses its ciliation, whereas in the frog it remains active almost up to the time of the metamorphosis. In tadpoles of 3-10 mm. nearly the whole surface is ciliated (Assheton).^[14] The cilia work from head to tail, causing the little animal, when perfectly quiet, to move forwards slowly in the water. Beneath the cuticula, in the Perennibranchiata and the larvae of the other Urodela, lies a somewhat thicker layer of vertically striated cells, the so-called pseudo-cuticula, which disappears with the transformation of the upper layers of the Malpighian cells into the stratum corneum. The latter is very thin, consists of one or two layers of flattened cells, and is shed periodically by all Amphibia in one piece. In the Urodela it generally breaks loose around the mouth, and the animal slips out of the delicate, transparent, colourless "shirt," which during this process of ecdysis or moulting becomes inverted. In the Anura it mostly breaks along the middle line of the back, the creature struggles out of it, pokes it into its mouth, and swallows it. Urodela also eat this skin. As a rule the first ecdysis takes place towards the end of the metamorphosis, preparatory to terrestrial life. So long as the animal grows rapidly, the skin has to be shed frequently, since this corneous layer is practically dead and unyielding. Adult terrestrial Urodela do not seem to moult often, mostly only when they take to the water in the breeding season. Anura, on the other hand, moult often on land, at least every few months. The surface of the new skin is then quite moist and slimy, but it soon dries and hardens.

The Malpighian stratum consists of several layers, thickest in the Perennibranchiata; in them it contains mucous cells throughout life, in others such slime-cells are restricted to larval life. Later, regular slime-glands are developed, which open on the surface. They are very numerous, and more evenly distributed, over most parts of the body, than the specific or poison-glands, which are restricted to certain parts, often forming large clusters, especially on the sides of the body. They reach their greatest development in the "parotoid glands" of the Anura. Both kinds of glands are furnished with smooth muscle-fibres, which are said to arise from the basal membrane underlying and forming part of the Malpighian layer; these muscle-cells extend later downwards into the corium. For the action of the poison, see p. 37.

The stratum corneum is mostly thin, but on many parts of the body, especially in Anura, the epidermal cells proliferate and form hard spikes or other rugosities, generally stained dark brown. With these may be grouped the nuptial excrescences so frequent in the Anura, especially on the rudiment of the thumb, and on the under surface of the joints of the fingers and toes. In many Anura, less frequently in the Urodela, the tips of the fingers and toes are encased in thicker horny sheaths, producing claws or nails. They are best developed among newts in Onychodactylus, among the Anura in Xenopus and Hymenochirus. The horny covering of the metatarsal tubercles reaches its greatest size in the digging spur or spade of Pelobates. In most of these cases the cutis is elevated into more or less wart-like papillae, covered, of course, by the proliferated and cornified epidermis. In the female of Rana temporaria nearly the whole surface of the body becomes covered with rosy papillae during the breeding season. Similar nuptial excrescences are common, and are most noteworthy in the male of the Indian Rana liebigi.

The epidermis also contains sense-organs. They attain their highest development in the larvae; later on they undergo a retrogressive change. Each of these sense-organs is a little cup-shaped papilla, visible to the naked eye. It is composed of elongated cells which form a mantle around some central cells, each of which ends in a stiff cilium perforating a thin, hyaline membrane which lines the bottom of the cup, and is perhaps the representation of the cuticula. These ciliated cells are connected with sensory fibres, the nerve entering at the bottom of the whole organ. The cilia are in direct contact with the water, but the outer rim of the whole apparatus is protected by a short tube of hyaline cuticula-like secretion. These sense-organs are, in the larvae, scattered over the head, especially near the mouth and around the eyes, whence they extend backwards on to the tail, mostly in three pairs of longitudinal rows, one near the vertebral column, the others lateral. They are supplied by the lateral branch of the vagus nerve. They disappear during the metamorphosis, at least in the Anura, with the exception of Xenopus, in which they form conspicuous white objects. The white colour is caused by the tubes becoming choked with the débris of cells or coagulating mucous matter, so that it is doubtful if these organs, which moreover have sunk deeper into the skin, are still functional. In the terrestrial Urodela these organs undergo a periodical process of retrogression and rejuvenescence. During the life on land they shrink and withdraw from the surface, and their nerves likewise diminish, but in the breeding season, when the newts take again to aquatic life, they revive, are rebuilt and become prominent on the surface. They are an inheritance from the fishes, in which such lateral line organs are universally present.

The cutis of most Amphibia is very rich in lymph-spaces, which, especially in the Anura, assume enormous proportions, since the so-called subcutaneous connective tissue forms comparatively few vertical septa by which the upper and denser layers, the corium proper, are connected with the underlying muscles. The spaces are filled with lymph, and into some of them the abnormally expanded vocal sacs extend, notably in Paludicola, Leptodactylus, and other Cystignathidae, and in Rhinoderma.

The cutis frequently forms papillae and prominent folds, sometimes regular longitudinal keels on the sides of the back; but dermal, more or less calcified or ossified scales are restricted to the Stegocephali and to the Apoda, q.v., pp. 79, 87. We conclude that the Urodela and Anura have entirely lost these organs. Dermal ossifications, besides those which now form an integral part of the skeleton, like many of the cranial membrane-bones, are rare, and are restricted to the Anura. They are least infrequent on the head, where the skin is more or less involved in the ossification of the underlying membrane-bones, for instance in Triprion, Calyptocephalus, Hemiphractus and Pelobates. The thick ossifications in the skin of the back of several species of Ceratophrys are very exceptional. In Brachycephalus ephippium these dermal bones enter into connection with the vertebrae; small plates fuse with the dorsal processes of the first to third vertebrae, while one large and thick plate fuses with the rest of the dorsal vertebrae. Simple calcareous deposits in the cutis are less uncommon, for instance, in old specimens of Bufo vulgaris. We are scarcely justified in looking upon these various calcifications and even ossifications as reminiscences of Stegocephalous conditions.

The skin contains pigment. This is either diffuse or granular. Diffuse pigment, mostly dark brown or yellow, occurs frequently in the epidermis, even in the stratum corneum. The granular pigment is stored up in cells, the chromatophores, which send out amœboid processes, and are restricted to the cutis, mostly to its upper stratum, where they make their first appearance. Contraction of the chromatophores withdraws the pigment from the surface, expansion distributes it more or less equally. The usual colours of the pigment are black, brown, yellow, and red. Green and blue are merely subjective colours, due to interference. A peculiar kind of colouring matter is the white pigment, which probably consists of guanine, and is likewise deposited within cells; cf. the description of the white spots in the skin of Hyla coerulea.

Most Amphibia are capable of changing colour, the Urodela, however, far less than the Anura, some of which exhibit an extraordinary range and adaptability in their changes.

The mechanism by which the change of colour is produced in frogs has been recently studied by Biedermann.^[15] If we examine the green skin of the common Tree-frog, Hyla arborea, under a low power and direct light, we see a mosaic of green, polygonal areas, separated by dark lines and interrupted by the openings of the skin-glands. Seen from below the skin appears black. Under a stronger power the black layer is seen to be composed of anastomosing and ramified black pigment-cells. Where the light shines through, the skin appears yellow. The epidermis itself is quite colourless. The mosaic layer is composed of polygonal interference-cells, each of which consists of a basal half which is granular and colourless, while the upper half is made up of yellow drops. Sometimes the tree-frog appears blackish, and if then the black pigment-cells are induced to contract, for instance, by warming the frog, it appears silver-grey; in this case the pigment in the yellow drops is no longer diffuse, but is concentrated into a round lump lodged between the interstices of the granular portions; the black pigment-cells are likewise balled together. These black chromatophores send out numerous fine branches, which occasionally stretch between and round the polygonal cells. When each of these is quite surrounded and covered by the black processes, the frog appears black. On the other hand, when the black pigment-cells withdraw their processes, shrink up, and, so to speak, retire, then the light which passes through the yellow drops is, by interference, broken into green.

Stoppage of the circulation of the blood in the skin causes the black chromatophores to contract. Carbon dioxide paralyses them and causes them to dilate. This is direct influence without the action of nerves. But stimulation of the central nerve-centres makes the skin turn pale. Low temperature causes expansion, high temperature contraction, of the chromatophores. Hence hibernating frogs are much darker than they are in the summer. Frogs kept in dry moss, or such as have escaped into the room and dry up, turn pale, regardless of light or darkness, probably owing to a central, reflex, nerve-stimulus.

Tree-frogs turn green as the result of the contact with leaves. Dark frogs will turn green when put into an absolutely dark vessel in which there are leaves. This is reflex action, and blinded specimens do the same. The principal centres of the nerves which control the chromatophores, lie in the corpora bigemina and in the optic thalami of the brain. When these centres are destroyed, the frog no longer changes colour when put upon leaves, but if a nerve, for instance the sciatic, be stimulated, the corresponding portion of the body, in this case the leg, turns green. Rough surfaces cause a sensation which makes the frog turn dark. Rana seems to depend chiefly upon temperature and the amount of moisture in the air, so far as its changes of colour are concerned. Biedermann concludes that the "chromatic function of frogs in general depends chiefly upon the sensory impressions received by the skin, while that of fishes depends upon the eye."

All this sounds very well, but the observations and experiments are such as are usual in physiological laboratories, and the frogs, when observed in their native haunts, or even when kept under proper conditions, do not always behave as the physiologist thinks they should. There is no doubt that in many cases the changes of colour are not voluntary, but reflex actions. It is quite conceivable that the sensation of sitting on a rough surface starts a whole train of processes: roughness means bark, bark is brown, change into brown; but one and the same tree-frog does not always assume the colour of the bark when it rests, or even sleeps upon, such a piece. He will, if it suits him, remain grass-green upon a yellow stone, or on a white window-frame. I purposely describe such conditions, changes, coincidences, and discrepancies in various species, notably in Hyla arborea, H. coerulea, Rana temporaria, Bufo viridis, to show that in many cases the creature knows what it is about, and that the eye plays a very important part in the decision of what colour is to be produced. The sensory impression received through the skin of the belly is the same, no matter if the board be painted white, black, or green, and how does it then come to pass that the frog adjusts its colour to a nicety to the general hue or tone of its surroundings?

Boulenger^[16] has given us a summary of the action of the poison of Amphibia:

It is well known to all who have handled freshly-caught newts, and certain toads, especially Bombinator, that their secretion acts as a sternutatory, and causes irritation of the nose and eyes, the effects produced on us by Bombinator being comparable to the early stages of a cold in the head. Many collectors of Batrachians have learned, to their discomfiture, how the introduction of examples of certain species into the bag containing the sport of their excursion may cause the death of the other prisoners; for although the poison has no effect on the skin of individuals of the same species, different species, however closely allied, may poison each other by mere contact. But when inoculated the poison acts even on the same individual.

Miss Ormerod, to personally test the effect, pressed part of the back and tail of a live Crested Newt between the teeth. "The first effect was a bitter astringent feeling in the mouth, with irritation of the upper part of the throat, numbing of the teeth more immediately holding the animal, and in about a minute from the first touch of the newt a strong flow of saliva. This was accompanied by much foam and violent spasmodic action, approaching convulsions, but entirely confined to the mouth itself. The experiment was immediately followed by headache lasting for some hours, general discomfort of the system, and half an hour after by slight shivering fits."

Numerous experiments have shown that the poison of toads, salamanders, and newts is capable, when injected, of killing mammals, birds, reptiles, and even fishes, provided, of course, that the dose be proportionate to the size of the animal. Small birds and lizards succumb as a rule in a few minutes; guinea-pigs, rabbits, and dogs in less than an hour.

This poison of Amphibia is not septic, but acts upon the heart and the central nervous system. That of the common toad has been compared, in its effects, to that of Digitalis and Erythrophlaeum. Some authorities hold that the poison is an acid, others regard it as an alkaloid.

Phisalix^[17] has come to the conclusion that toads and salamanders are possessed of two kinds of glands, different both anatomically and physiologically. These are, first the mucous glands, spread over the greater part of the body, with an alkaloid secretion, which acts as a narcotic; secondly, specific glands, as the parotoids and larger dorsal glands, the secretion of which is acid, and acts as a convulsive.

The Indians of Colombia are said to employ the secretion of Dendrobates tinctorius for poisoning their arrows. The poison is obtained by exposing the frog to a fire, and after being scraped off the back is sufficient for poisoning fifty arrows. It acts on the central nervous system, and is used especially for shooting monkeys. Concerning the use of this poison for "dyeing" parrots, see p. 272.

The milky secretion of toads protects them against many enemies, although not always against the grass-snake. A dog which has once been induced to bite a toad, suffers so severely that it will not easily repeat the experiment. The handling of tree-frogs also irritates both nose and eyes. The hind limbs of the Water-frog, Rana esculenta, have a very bitter, acrid taste. In short, most, if not all, Amphibia are more or less poisonous, and it is significant that many of the most poisonous, e.g. Salamandra maculosa, Bombinator, Dendrobates, exhibit that very conspicuous combination of yellow or orange upon a dark ground, which is so widespread a sign of poison. Other instances of such warning colours, protective in a defensive sense, are the Wasps and Heloderma, the only poisonous lizard.

Nerves

Spinal nerves.–Each spinal nerve issues originally immediately behind the neural arch of the vertebral segment to which it belongs. This intra-vertebral position is ultimately modified into a more inter-vertebral one, owing to the predominant share of the neural arches, basidorsalia, in the composition of the whole vertebra. Consequently the nerves issue behind their corresponding vertebra.

The first spinal nerve, or N. suboccipitalis, is exceptional in several respects. It develops a dorsal and a ventral root like a typical spinal nerve, but the dorsal root soon degenerates in all Amphibia, while in the Phaneroglossal Anura the whole nerve disappears. The first spinal nerve reduced to its ventral half persists therefore only in the Apoda, Urodela, and the Aglossal Anura. It issues originally between the occiput and the atlas, but in the adult it is partly imbedded in the anterior portion of the atlas. Its own vertebra is lost, having probably been added to the cranium.

In the Urodela the first spinal nerve either remains separate, or it joins the second spinal, forming with it and with a branch from the third nerve the cervical plexus, which supplies the muscles of the cervical region. The third, fourth, and fifth nerves, and sometimes also the sixth, form the brachial plexus.

In the Aglossal Anura N. spinalis I. mostly sends a fine thread to the second spinal nerve, the rest supplies chiefly the M. levator scapulae, in Pipa the abdominal muscles also. In all the other Anura this N. spinalis I. is lost; occasional vestiges have been reported in Bufo vulgaris and Rana catesbiana, and remnants of it may possibly be found in Pelobatidae and Discoglossidae. The first actually persisting nerve of the Phaneroglossa is consequently N. spinalis II.

The brachial plexus is composed as follows:–Pipa, N. spinalis II. and III.; Xenopus and Phaneroglossa, N. spinalis III. and IV., with a small branch from the second; the next following three nerves, numbers V., VI., and VII., behave like ordinary trunk nerves.

The pelvic plexus of the Phaneroglossa is formed in Rana by the VIII. + IX. + X. + XIth nerves, the tenth issuing between the sacral vertebra and the coccyx. In Bufo and Hyla the plexus is composed of five nerves, the seventh spinal sending a branch to it. Occasionally the twelfth nerve contributes a small branch to the posterior portion of the plexus. This and the eleventh nerve leave the coccyx by separate holes, thereby indicating its composition. The rest of the spinal cord gives off no more recognisable nerves, owing to its reduction during the later stages of metamorphosis; its terminal filament passes out of the posterior end of the coccygeal canal.

Concerning the cranial nerves it is necessary to draw attention to one point only. The last nerve which leaves the cranium of the Amphibia is the vagus or tenth cranial nerve. There is consequently no eleventh, and no twelfth or hypoglossal, pair of cranial nerves. Their homologues would be the first and second spinal nerves, but the whole tongue of the Amphibia, with its muscles, is supplied by the glossopharyngeal, or ninth cranial pair, and is morphologically not homologous with the tongue of the Amniota.

Respiratory Organs

A very important and characteristic feature of the Amphibia is the development of two sets of respiratory organs: Gills and Lungs. It is as well to give definitions of these organs. Lungs are hollow evaginations from the ventral wall of the pharynx, and their thin, vascularised walls enable the blood to exchange, by osmosis, carbon dioxide for oxygen from the air which enters the lungs by the mouth or the nostrils, and the windpipe. The latter is unpaired, the lungs themselves are paired. Gills are highly vascularised, more or less ramified excrescences, covered by a thin epithelium of ecto- or endo-dermal origin, which permits of the exchange of carbon dioxide for oxygen from the air which is suspended in the surrounding water. It is obvious that this definition applies to all sorts of well-vascularised organs whose thin surface comes into contact with the water. Various recesses of the pharyngeal cavity, the dorsal and ventral folds of the tail-fin, nay, even any part of the skin of the body can, and does occasionally, assume additional respiratory functions. The proper definition of gills, in Vertebrates, requires, therefore, the restriction that they must be developed upon and carried by visceral arches.

The general statement that the Amphibia breathe by lungs, and, at least during some stage of their life, also by gills, requires various restrictions. As a rule the majority of Amphibia first develop gills, later on also lungs, whereupon, during the metamorphosis, the gills are gradually suppressed, so that the perfect animal breathes by lungs only (see p. 61). But a number of Urodela retain their gills throughout life, although the lungs are also functional. These are the Perennibranchiata, not a natural group, but a heterogenous assembly, Proteidae and Sirenidae. Some species of Amblystoma remain individually Perennibranchiate (cf. Axolotl, p. 112). On the other hand, in some Anura the gills are almost or entirely suppressed, or restricted to the embryonic period only. Lastly, a considerable number of Salamandridae have lost their lungs; they breathe by gills until their metamorphosis, but have in the adult state to resort to respiration by the skin (cf. p. 46).

The general plan of the development of the branchial respiratory apparatus is as follows:–The six visceral arches, namely, the mandibular, the hyoidean, and the four branchial arches, correspond, long before they are cartilaginous, with four main arterial arches of the truncus arteriosus. The first, the arteria hyo-mandibularis, belongs to the hyoidean and mandibular segments, the second to the first branchial, the third to the second branchial, while the fourth soon splits in two for the third and fourth or last branchial arch. On the dorsal side these branchial arterial arches combine to form the radix of the dorsal aorta. These arches, especially the three branchials, appear in newts, less clearly in frogs, as transverse ridges on the sides of the future neck. Between the arches the pharynx gradually bulges out in the shape of five lateral gill-pouches; the first between the mandibular and the hyoidean arch, the second between the hyoidean and the first branchial arch, etc. These pouches soon break through to the outside and become gill-clefts, except the first pouch in Urodela. Before the breaking through of the clefts there appears upon the outside of the middle of the rim of each arch a little knob, which soon ramifies and forms an external gill. The knob owes its origin to the development of a blood-vessel which buds from the arterial arch, ramifies and breaks up into capillaries, and returns a little further dorsalwards into the arch. A secondary loop to the outside of the primary arterial arch is thus formed; and whilst this outer loop sprouts out further, driving before it the likewise proliferating skin, and thus producing the gill, the middle portion of the primary arch remains in the Urodela as a short cut, but in the Anura it partly obliterates, and henceforth acts as the internal efferent vessel of the gill. When, during metamorphosis, the gills disappear, their intrinsic afferent and efferent vessels vanish likewise, and the short cut completes the circuit. In order to do this they have, in the Anura, to form new connections with the trunks of the afferent vessels.

The arterial arches themselves are modified as follows:–The first pair become the carotids, the second form the right and left aortic arches, while the third and fourth unite and are transformed into the pulmonary arteries and "ductus Botalli," the last arterial arch having previously sent a branch into the developing lungs. In the Anura the third arch obliterates.

The gills and clefts present various modifications. The Urodela possess three pairs of gills, one each upon the dorsal half of the three branchial arches, just near the upper corners of the clefts; and the skin of the body is continued upon the stem of each gill, pigmented like the rest of the surface of the body. Such a gill is more or less like a blade, standing vertically, and is composed of a stem of connective tissue, thick at the base, and, as a rule, carrying two series of fine lamellae, which, however, do not form two opposite series, but hang downwards, being, so to speak, folded down, so that the upper surface of the stem is bare, and carries the lamellae on its under side. In the Axolotl some of these lamellae are further subdivided. In Necturus they are enormously increased in numbers, but are rather short, and they stand no longer in two rows, but are crowded into one. Those of Proteus form two rows of dendritic filaments; those of Siren are likewise much ramified.

The larvae of the Urodela have four clefts. In the adult Siren these are reduced to three, the first, namely, that between the hyoid and the first branchial arch, being closed up. In Necturus, Proteus, and Typhlomolge the clefts are further reduced to two, owing to the closing up of the first and last, only those between the first, second, and third arches remaining. Amphiuma, and usually Cryptobranchus alleghaniensis, possess only one pair of clefts, while in C. japonicus and in the Salamandridae all the clefts are abolished.

The gills of the Urodela are always uncovered, although a short operculum is formed from the posterior margin of the hyoidean arch; the halves of this fold meet below the throat, and persist in various terrestrial and aquatic species as the "gular fold." It reaches its greatest size just before metamorphosis, but scarcely ever produces a proper outer gill-chamber, and it does not cover the gills owing to their rather pronounced dorsal position. It is perhaps best developed in Typhlomolge, and even there its dorsal portion is continued upon the first of the three broad vertical and short-fringed blades which form the gills.

A description of the gills of the Apoda will be found in the systematic part.

In the Anura the gills are complicated, owing to the development of the so-called internal gills. First appear, exactly in the same way as in the Urodela, the external gills, one upon each of the first three branchial arches. In the larva of Rana esculenta, 5 mm. in length, a little protuberance appears upon the first, and then upon the second arch. In the 6 mm. larva the first gill shows four knobs, the second two, the third one knob. They are always delicate and thin, although sometimes pigmented, long, and much-ramified structures. The first pair is always the largest; well developed and persisting a long time in Rana temporaria; smaller in R. esculenta and Bufo vulgaris; very short, scarcely forked, in B. viridis and Hyla arborea. They are relatively largest in Alytes, while still in the egg. Numerous descriptions of these gills will be found in the systematic part.

Great changes take place about the time when the fourth or last branchial arch and the pulmonary arteries are developed. This occurs in R. esculenta when the larva is about 9 mm. long. The sprouting of the gills extends gradually downwards along the arches upon their ventral halves, and these new gill-filaments or loops transform themselves into numerous dendritic bundles, resting in several thickset rows upon the hinder margin of the first to the third arch, one row only on the fourth arch, which carries no external gill. These "internal gills" look like red bolsters or thick and short-tasselled bunches. Whilst they are developing the dorsal, older gills become arrested in their growth and disappear, and at the same time a right and left opercular fold grows out from the head and covers these new gills, shutting them up in an outer branchial chamber, just like that of Teleostei and other Tectobranch fishes. This is the reason why these new gills have been called internal, and the mistaken notion has sprung up that they are comparable with the true internal gills of fishes. In reality Amphibia have only external gills. They are always covered by ectoderm, are restricted to the outside of the branchial arches, and are developed before the formation of the clefts. These gills are in many cases directly continuous with the more dorsally and more superficially placed earlier external gills; but although nearly every one who has studied their development has observed this agreement, the old error still prevails. They are morphologically as little internal as the true internal gills of Elasmobranch embryos are external gills, because these have become so elongated that they protrude out of the gill-clefts.

The fact that the Amphibia possess only external gills throws important light upon their phylogeny. Not only do the Apoda, Urodela, and Anura agree much more with each other than would be the case if the Anura possessed both internal and external gills, but the Amphibia reveal themselves also in this point as connected with the Crossopterygii and the Dipnoi, some of which fishes also possess external gills. It is of course quite possible that the Amphibia have developed these organs independently, but we understand now that the latter are accessory, and not the primitive respiratory organs; they are developed in adaptation to embryonic conditions and to prolonged larval, occasionally perennibranchiate, aquatic life (cf. the chapter on Neoteny, p. 63).

There is no valid reason for supposing that the Stegocephali had true internal gills. We know their branchial skeleton, and we can discern even gill-rakers on the arches. Such gill-rakers occur also, although but feebly developed, in Urodela. The whole branchial framework of the Urodela and Apoda undergoes simple reductions during metamorphosis (see p. 86), but in the Anura these arches are in early tadpole life transformed into a most complicated basket-work which acts as a straining apparatus or filter, to prevent any particle of food or other foreign matter from finding its way into the delicate gills, the current of water passing from the mouth through the filter, past the gills and out of the clefts. During metamorphosis this whole elaborate apparatus is again transformed, almost beyond recognition, into the hyoidean apparatus for the support of the generally very movable and much-specialised tongue. The fact that the hyoid apparatus of the Aglossa, especially that of Xenopus, is constructed upon the same lines, is a strong indication that these creatures have arrived at their tongueless condition through the loss of this organ, and this is intelligible in correlation with their absolutely aquatic life.

The opercular folds assume great dimensions in all tadpoles. They cover the whole gill-region, thereby producing on either side an outer gill-chamber. The posterior margins of the folds gradually become continuous with the rest of the surface of the body. Each gill-chamber opens at first by one lateral canal, usually called the spiracle. This condition prevails in the tadpoles of the Aglossa. In the Discoglossidae the two canals gradually converge and combine into one median opening on the middle of the belly. In all the other Anura the right opening becomes closed, or rather its canal passes over to and joins that of the left side, both opening by one short tube laterally on the left side, at a variable distance between the eye and the vent. Hence the elegant terms of Amphi-, Medio-, and Laevo-gyrinidae (γυρῖνος being the Greek for tadpole).

The external gills lead to a further consideration. Protopterus possesses a vestigial external gill on the shoulder-girdle. Lepidosiren has them on the gill-arches, resembling piscine internal gills, and Polypterus has a large biserially fringed external gill (in some cases not disappearing until the fish is adult), which starts from the mandibular arch, at the level of the spiracle or first visceral cleft, and overlaps the operculum externally. The axis of this peculiar organ is possibly based upon the homologues of the spiracular cartilages, which themselves are the branchiostegal rays of the dorsal half of the quadrato-mandibular arch. The branchiostegal rays of the hyoidean arch, at least their material, have given rise to the elaborate opercular apparatus; and, in conformity herewith, the hyomandibular itself is not known to carry a gill. Quite possibly the large external gill of Polypterus is not serially homologous with other external gills–it may not be a true gill at all, it has perhaps quite a different function–but it seems to throw light upon a mysterious pair of organs which are common in larval and young Urodela, in the larval Aglossa and in the Apoda. These are the "balancers."

In Triton taeniatus, before hatching, there appears a little protuberance behind and below the eye; it rests upon the angle of the mandibular arch, and is separated from the first transverse, externally visible ridge of the first branchial arch by the beginnings of the hyoidean arch. A few days later the arteria hyomandibularis sends a vessel into this knob, forms a vascular coil, and leaves it as a vein which, instead of returning into the arterial arch, passes into the veins of the body. Its epithelium is not covered with flat, but with cubical cells; and sensory cells have not been found in it. These organs attain some size, and are shaped like rods, with thickened ends; they are movable, and are used by the larvae as "balancers," keeping the head from sinking into the slime at the bottom. But they may have other functions besides, and it is not unlikely that they develop into sensory organs like feelers. They occur in many Salamandridae, and are not reduced until, or even after, the metamorphosis, and during this time they shift their place with relation to the eye and the mouth.