PART I

AMPHIBIA

CHAPTER I

AMPHIBIA

CHARACTERS AND DEFINITION–POSITION OF THE CLASS AMPHIBIA IN THE PHYLUM VERTEBRATA–HISTORICAL ACCOUNT OF THE CLASSIFICATION OF AMPHIBIA

A bird is known by its feathers, a Beast by its hairs, a Fish by its fins, but there is no such obvious feature which characterises the Amphibia and the Reptiles. In fact, they are neither fish, flesh, nor fowl. This ill-defined position is indicated by the want of vernacular names for these two classes, a deficiency which applies not only to the English language. All the creatures in question are backboned, creeping animals. Those which are covered with horny scales, and which from their birth breathe by lungs only, as Crocodiles, Tortoises, Lizards, and Snakes, are the Reptiles. The rest, for instance, Newts or Efts, Frogs and Toads, are the Amphibia. Their skin is mostly smooth and clammy and devoid of scales; the young are different from the adult in so far as they breathe by gills and live in the water, before they are transformed into entirely lung-breathing, terrestrial creatures. But there are many exceptions. Proteus and Siren the mud-eel, always retain their gills; while not a few frogs undergo their metamorphosis within the egg, and never breathe by gills. If we add the tropical limbless, burrowing Coecilians, and last, not least, the Labyrinthodonts and other fossil forms, the proper definition of the class Amphibia,–in other words, the reasons for grouping them together into one class, separated from the other backboned animals,–requires the examination of many other characters.

So far as numbers of living species are concerned, the Amphibia are the least numerous of the Vertebrata. There are about 40 limbless, burrowing Apoda; 100 Urodela or tailed two- or four-footed newts, and about 900 Anura, or tailless, four-footed frogs and toads; in all some 1000 different species. Few, indeed, in comparison with the 2700 Mammals, 3500 Reptiles, nearly 8000 Fishes, and almost 10,000 Birds. But we shall see that the Amphibia have not only "had their day," having flourished in bygone ages when they divided the world, so far as Vertebrata were concerned, between themselves and the Fishes, but that they never attained a dominant position. Intermediate between the aquatic Fishes and the gradually rising terrestrial Reptiles they had to fight, so to speak, with a double front during the struggle of evolution, until by now most of them have become extinct. The rest persist literally in nooks and corners of the teeming world, and only the Frogs and Toads, the more recent branch of the Amphibian tree, have spread over the whole globe, exhibiting almost endless variations of the same narrow, much specialised plan. The greatest charm of the Anura lies in their marvellous adaptation to prevailing circumstances; and the nursing habits of some kinds read almost like fairy-tales.

Characters of the Amphibia.^[1]

We can, therefore, very easily define all the Amphibia, both recent and extinct, by a combination of the characters enumerated above. For instance, by the combination of numbers 2, 3 or 4 with either 7, 8, 9, 11, 13 or 15.

Amphicondylous Anamnia would be an absolutely correct and all-sufficient diagnosis, but it would be of little use in the determination of adult specimens; and the tetrapodous character is of no avail for Apoda. Amphicondylous animals without an intracranial hypoglossal nerve is a more practical diagnosis.

In the case of living Urodela and Anura the absence of any scales in the skin affords a more popular character; it is unfortunately not applicable to the Apoda, many of which possess dermal scales, although these are hidden in the imbricating transverse rings of the epidermis; and the frequent occurrence of typical scales of both ecto- and meso-dermal composition in many of the Stegocephali forces us to discard the scales, or rather their absence, as a diagnostic character of the class Amphibia. The same applies to the mostly soft, moist, or clammy, and very glandular nature of the skin.

The position of the class Amphibia in the Phylum Vertebrata.–There is no doubt that the Amphibia have sprung from fish-like ancestors, and that they in turn have given rise to the Reptilia. The Amphibia consequently hold a very important intermediate position. It was perhaps not a fortunate innovation when Huxley brigaded them with the Fishes as Ichthyopsida, thereby separating them more from the Sauropsida (= Reptilia and Aves), than is justifiable,–perhaps more than he himself intended. The connecting-link, in any case, is formed by the Stegocephali; all the recent Orders, the Apoda, Urodela, and Anura, are far too specialised to have any claims to the direct ancestral connections. The line leading from Stegocephali to fossil Reptiles, notably to such Proreptilia as Eryops and Cricotus, and even to the Lepospondylous Prosauria, is extremely gradual, and the steps are almost imperceptible. Naturally, assuming evolution to be true, there must have lived countless creatures which were a "rudis indigestaque moles," neither Amphibia nor Reptilia, in the present intensified sense of the systematist. The same consideration applies equally to the line which leads downwards to the Fishes. But the great gulf within the Vertebrata lies between Fishes and Amphibia, between absolutely aquatic creatures with internal gills and "fins," and terrestrial, tetrapodous creatures, with lungs and fingers and toes. On the side of the fishes only the Dipnoi and the Crossopterygii come into consideration.

The piscine descent of the Amphibia is still proclaimed by the following features.–(1) The possession by the heart of a long conus arteriosus, provided with, in many cases, numerous valves, or at least (Anura) one series at the base, another at the beginning of the truncus where the arterial arches branch off; (2) the strictly symmetrical arrangement of these arches; (3) the trilocular heart is still like that of the Lung-fishes or Dipnoi; (4) the occurrence of as many as four or even five branchial skeletal arches in the larval stage; (5) the glottis is supported by cartilages which themselves are derivatives of posterior visceral arches; (6) the development of the vertebrae (Stegocephali and Urodela) from four pairs of arcualia, and the formation of the intervertebral joints by a split across the intervertebral ring of cartilage: this feature is unknown in Reptilia, but it occurs also in Lepidosteus, most probably also in Polypterus; (7) the hypoglossal still retains the character of a post-cranial or cervical spinal nerve; (8) the presence of lateral sense-organs; (9) the possession of external gills is of somewhat doubtful phylogenetic value, although such gills occur amongst fishes only in Dipnoi and Crossopterygii. It is not unlikely that in the Amphibia these organs owe their origin to entirely larval requirements, while the suctorial mouth of the larvae of the Anura and many fishes has certainly no ancestral meaning, but is a case of convergent development.

The usual diagnoses of the Amphibia contain the statement that they, or most of them, undergo a metamorphosis, or pass through a larval stage. The same applies to various fishes; while, on the other hand, the larval (not ancestral) stage has become permanent in the Proteidae and Sirenidae; and lastly, we cannot well speak of larvae in the viviparous Salamandra atra.

The evolution of an adequate classification of the Amphibia has been a long process. Even their recognition as a class, separate from, and of equal rank with that of, the Reptilia, was by no means generally accepted until comparatively recent times. A historical sketch of the laborious, often painful, striving for light, in France and Germany, then in England, and lastly in America, is not without interest.

CHAPTER II

SKELETON OF URODELA AND ANURA–SKIN–COLOUR-CHANGING MECHANISM–POISON-GLANDS–SPINAL NERVES–RESPIRATORY ORGANS–SUPPRESSION OF LUNGS–URINO-GENITAL ORGANS–FECUNDATION–NURSING HABITS–DEVELOPMENT AND METAMORPHOSIS

Skeleton of the Urodela

The vertebral column.–The number of vertebrae is smallest in the terrestrial, greatest in the entirely aquatic forms, and is exceptionally large in the eel-shaped Amphiuma. In the following table the sacral vertebra is included in those of the trunk.

	Trunk.	Tail.
Siren lacertina	22	35 +
Necturus maculatus	19	29
Proteus anguinus	30	28 +
Cryptobranchus alleghaniensis	20 or 21	24 +
C. scheuchzeri	21
C. japonicus	22	22 to 26
Amphiuma means	63	35 +
Amblystoma tigrinum	17 or 16	32 +
Salamandra maculosa	17	27
Triton cristatus	17	36
Triton taeniatus	14 or 15	36 +
Triton palmatus	14	23 to 25
Salamandrina perspicillata	15	32 to 42
Spelerpes fuscus	16	23

The vertebrae of the Urodela and those of the Apoda differ from those of all the other Tetrapoda^[11] by possessing no special centra or bodies. That part which should correspond with the centrum is formed either by the meeting and subsequent complete co-ossification of the two chief dorsal and ventral pairs of arcualia (tail-vertebrae), or entirely by the pair of chief dorsal arcualia. There is consequently no neuro-central suture. Moreover, the central region of each vertebra is strongly pinched in laterally, widening towards the ends. Another feature of the vertebral column of the Urodela is the possession of a considerable amount of intervertebral cartilage, by which the successive vertebrae are held together. This cartilage does not ossify, and it either remains continuous, serving in its entirety and owing to its flexibility as a joint, or it becomes more or less imperfectly separated into a cup and ball portion, the cup belonging to the posterior end of the vertebra. Such joints are called opisthocoelous, and occur in the Desmognathinae and Salamandrinae. In the adult the cup and ball frequently calcify, and the chorda dorsalis or notochord is completely destroyed. Those vertebrae between which the intervertebral cartilage remains unbroken, are called amphicoelous, since in them, most obviously in macerated or dried skeletons, the vertebrae appear hollowed out at either end. In such amphicoelous vertebrae a considerable amount of the chorda always remains, running in an unbroken string through the whole length of the vertebral column. Towards adult life the chorda becomes constricted, and is ultimately squeezed out or destroyed, in the middle of the vertebra, by the invasion of cartilage from the chief arcualia. This intravertebrally situated cartilage has been described erroneously as chordal cartilage.

The development of the vertebrae proceeds as follows. First appear a pair of basidorsalia and a pair of basiventralia (Fig. 1, 1, B.D, B.V), blocks of cartilage, imbedded in and resting upon the thin sheath of the chorda dorsalis. Next appears a pair of interdorsal blocks, immediately behind the basidorsals; and somewhat later appears a pair of interventral blocks. These four pairs of cartilages or "arcualia" each meet, above or below the chorda, and form semi-rings, which again by extending upwards or downwards fuse into complete rings, in such a way that the interdorsal and interventral elements form the intervertebral mass spoken of above. The basidorsals fuse with the basiventrals, and form the body of the vertebra, the fusion being effected chiefly by the calcification and ossification of the lateral connecting portion of the skeletogenous layer. The basidorsalia form the neural arches with their unpaired short spinous or neural, and the paired anterior and posterior zygapophysial processes. Concerning the basiventralia we have to distinguish between the trunk and the tail. In the latter they produce a pair of ventral outgrowths or haemapophyses, which ultimately enclose the caudal blood-vessels. In the trunk the basiventral blocks of cartilage are suppressed; they appear in the early larvae, but disappear during or even before metamorphosis.

Towards the end of the tail the vertebrae diminish in size, and their constituent cartilages assume a more and more indifferent shape, until they become confluent into a continuous rod of cartilage, resembling in this respect the Dipnoi and Holocephali. A periodical revival of this rod, at least of its connective tissue, appears in the tail-filament of the male Triton palmatus during the breeding-season.

The first vertebra, called the atlas, because it carries the head, is remarkable for the possession of an odontoid process. The latter is formed by a pair of cartilages and represents part of a vertebra, the dorsal portion of which seems to have been added to the occipital part of the cranium.

All the trunk-vertebrae, with the exception of the atlas, carry ribs, at least vestiges thereof. Owing to the early disappearance of the basiventral cartilages the capitular portions of the ribs are much reduced, and are mostly represented by strands of connective tissue only. The ribs develop therefore occasionally at some distance from the vertebral column, and that portion of the rib which in the metamorphosed young newt looks like the capitulum is to a great extent really its tuberculum. Witness the position of the vertebral artery, which still indicates the true foramen transversarium. The homologies of these parts are still more obscured by the fact that a new process grows out from the rib, by which the latter gains a new support upon a knob of the neural arch. Thus an additional foramen is formed, sometimes confounded with the true transverse canal. The meaning which underlies all these modifications is the broadening of the body, the ribs shifting their originally more ventral support towards the dorsal side. The whole process is intensified in the Anura; it is an initial stage of the notocentrous type of vertebrae. The transverse ossified processes of the adult are often much longer than the vestiges of the ribs themselves, and are somewhat complicated structures. They are composed first of the rib-bearing cartilaginous outgrowths of the neural arches; secondly, of a broad string of connective tissue which extends from the ventro-lateral corner of the perichordal skeletogenous layer to the ribs.

The shoulder-girdle is extremely simple. It remains almost entirely cartilaginous, and the three constituent elements are not separated by sutures. Ossification is restricted to the base of the shaft of the scapula, and may extend thence over the glenoid cavity. The coracoids are broad, loosely overlap each other, and are "tenon and mortised" into the triangular or lozenge-shaped cartilaginous sternum, which latter has no connection with the ribs. The precoracoid is a large, flat process, directed forwards, not meeting its fellow; it is absent in Siren.

The humerus articulates with both radius and ulna, and these two bones of the forearm remain separate. The elements which compose the wrist and hand exhibit an almost ideally simple arrangement, slightly varied by the frequent fusion of two or more neighbouring carpalia into one, and by the reduction of the number of fingers. Most frequently the intermedium and the ulnar carpal element fuse together, and there is more often one centrale instead of two. The wrist and hand of the Urodela represent, however, no longer the entirely primitive pentadactyle type, owing to the loss of one finger together with its metacarpal and carpal element. Comparison with the Anura makes it probable that the Urodela have lost the pollex, their four fingers being consequently the 2nd, 3rd, 4th, and 5th. Siren has four or three fingers; Proteus has only three fingers and three large compound carpal cartilages. In Amphiuma, with either three or two fingers, the ulnare, intermedium, and carpale are fused together, the radiale with the neighbouring carpale. The number of phalanges in the four-fingered species is generally 2, 3, 3, 2 respectively.

The pelvic girdle.–The ilium stands vertically to the vertebral axis, slanting slightly forwards and downwards. It is attached by means of a rib to only one vertebra, and this ilio-sacral connection is acetabular in its position, i.e. it lies in the same transverse plane with the acetabulum, in other words vertically above it. The ventral portion of the pelvis is formed by one large continuous mass, the united pubo-ischia, the anterior or pubic portion of which extends forwards in the shape of a broad triangle (Necturus) or as a slender, stalked, Y-shaped cartilage, the epipubis, which is often movably jointed at its base. The lateral portion of the pubic cartilage is always perforated by the nervus obturatorius. Ossification is restricted to the ischium and to the middle of the shaft of the ilium. The acetabular fossa for the femur is closed. The tibia and fibula remain separate. The foot is still more primitive than the anterior extremity, as the majority of Urodela possess the full complement of five toes, with 2, 2, 3, 3, 2 phalanges respectively. Concrescence of the tarsalia applies most frequently to the fourth and fifth distal and to the two centralia; exceptional, for instance, in Cryptobranchus japonicus, are as many as three centralia, but this is an individual, even a one-sided variation, as shown for instance by a specimen in the Cambridge Museum. Loss of the fifth toe occurs sporadically in genera of different groups, namely, in Salamandrella, Batrachyperus, Salamandrina, Necturus, Manculus, Batrachoseps. In Amphiuma the number is reduced to three or two; in Proteus to two; and in Siren the hind limbs, with their girdle, are altogether absent. Lastly, in some species of Spelerpes and Batrachoseps both fore and hind limbs have become so small as to be practically without function, parallel cases being found among various Scincidae and other Lizards.

The hyoid apparatus is still very primitive in many, especially in larval, Urodela. Besides the hyoid there are as many as four pairs of branchial arches, which, however, decrease in size and completeness, so that the last two have lost their connection with the median copular piece, and become attached in various ways to the second branchial arch. This is the arrangement apparently in all larvae, but four pairs of branchials persist in the adult Siren, Amphiuma, and Cryptobranchus alleghaniensis. The whole branchial apparatus is reduced to three pairs of arches in Necturus and Proteus, to two in the adult Cryptobranchus japonicus and in the Salamandridae. Of considerable interest is the vestige of a fifth pair of arches in the larvae of Triton and Salamandra, in the shape of a pair of tiny cartilages, which lie in front and on each side of the opening of the trachea, and give rise to the formation of the laryngeal cartilages, better developed in the higher Vertebrata.

The following are noteworthy characters of the skull of Urodela. The articulation of the skull with the vertebral column is not always effected entirely by the two condyles of the lateral occipital bones, but the median basal cartilage often possesses a pair of facets for the odontoid-like process of the first vertebra; such additional facets are perhaps best developed in Cryptobranchus and in the Salamandrinae.

The middle portion of the primitive cranium, from the exit of the optic nerve to the ethmoid cartilage, is formed by a pair of separate bones, the orbito-sphenoids. The parietal and frontal bones remain separate. One or more periotic bones exist, besides the prootic, in the aquatic families.

A pair of prefrontal bones is present in most Salamandridae, e.g. Salamandra, Triton, Amblystoma, especially in the larva, and in Cryptobranchus; these bones are absent in Amphiuma, Necturus, Proteus, and Siren.

The lacrymalia are still separate in some Amblystomatinae, e.g. Ranidens and Hynobius. A pair of nasalia are generally present, but are absent in Necturus, Proteus, and Siren. The parasphenoid is furnished with teeth in the Plethodontinae and Desmognathinae.

Separate palatine bones exist in Necturus and Proteus, and in the larva of Amblystoma, but in the adult form they fuse with the vomers, producing the vomero-palatines characteristic of the majority of Urodela.

The pterygoid bones are most fully developed, so as to reach the vomero-palatines, in the Amblystomatinae, in Necturus, and in Proteus; they are reduced, so as to leave a gap, in Cryptobranchus, and still more in the Salamandrinae; they are absent in Amphiuma and in Siren.

The quadrates are directed forwards in Necturus, Proteus, and Siren, while in the other Urodela they extend transversely and almost horizontally. The hyomandibular remnant, the so-called operculum, is small, and forms a plate which fits into the fenestra ovalis, extending as a ligamentous process upon the quadrate.

The quadrato-jugal elements are reduced to ligaments. In many Salamandrinae the large orbito-temporal space is divided into an orbital and a temporal fossa by an arch which is formed by the meeting of two corresponding processes from the squamosal and frontal bones respectively. This bridge is rarely bony (Salamandrina, Triton), mostly ligamentous;–apparently a reminiscence of the Stegocephalous condition. The two premaxillary bones are liable to fuse into one, for instance in Cryptobranchus, generally in adult Tritons. They are most reduced, and are toothless, in Siren.

The two maxillary bones are absent only in Necturus, Proteus, Typhlomolge, and Siren. Their posterior end is frequently free, loosely connected by ligaments with the pterygoid in Cryptobranchus; or with the distal portion of the quadrate, and in this case either just touching it (Tylototriton), or forming a broad junction (Pachytriton).

Each half of the lower jaw consists of a dentary, articular and angulo-splenial. The splenial remains as a separate element in Siren; in others only during the larval period. There are no mento-Meckelian elements.

Skeleton of the Anura

The vertebral column.–The distinctive peculiarities of the vertebrae of the Anura are that they are notocentrous, and that about a dozen of them are modified and fused into an os coccygeum. The whole column is the most specialised found in the Vertebrata; and various stages are rapidly hurried through and obscured caenogenetically during the embryonic development. Paired cartilages appear on the dorsal side of the thin chordal sheath, and whilst tending to enclose the spinal cord in a canal, their bases grow head- and tail-wards into what will ultimately become the intervertebral region. This extension of cartilage leads to a fusion with that of the next following pair of arches, so that the axial column at this early stage consists of a right and left longitudinal ridge of cartilage which sends off dorsal processes, neural arches, in metameric succession. Next, the intervertebral cartilage increases in such a way as to constrict the chorda either laterally (Rana) or obliquely from above downwards and inwards (Bufo, Hyla). We recognise in this cartilage the interdorsalia. Ventral arcualia are late and much obscured. There is scarcely any cartilage which could represent the interventralia, the intervertebral cartilage being almost entirely made up of the interdorsalia. These fuse together and form a disc or nodule, which later fuses either with the vertebra in front, and in this case fits into a cup carried by the vertebra next behind (procoelous vertebrae), or the knob is added to the front end of the vertebra, fitting into a cup formed by the tail end of the vertebra next in front (opisthocoelous vertebrae). Much later than the two longitudinal dorsal bands there appears on the ventral side an unpaired band in which appear metamerically repeated swellings of cartilage, likewise unpaired. These swellings become confluent, in a way similar to that which produced the dorsal bands, and form the unpaired ventral band of cartilage, the hypochordal cartilage of some authors. The swellings in this band, equivalent to the basiventralia, become semilunar in a transverse view, their horns tending upwards towards the basidorsal cartilages, but there is no actual meeting. Both dorsal and ventral elements are, however, joined together and form the chief portion of the vertebrae, owing to the rapidly proceeding calcification and later ossification of the all-surrounding "membrana reuniens" or skeletogenous layer so far as that is not cartilaginous.

Procoelous vertebrae exist in the overwhelming majority of Anura; opisthocoelous are those of the Aglossa, the Discoglossidae, and of some Pelobatidae. The systematic value of this pro- or opistho-coelous character has been much exaggerated. We have seen that the centra of the vertebrae of the Anura are formed entirely by the interdorsal elements, hence the term "notocentrous," and these centra sometimes remain in adult specimens of Pelobates as separately ossified and calcified pieces, not fused with the rest of the vertebrae. This important discovery has been made by Boulenger, but Stannius had previously mentioned a specimen of Pelobates in which the second and fourth vertebrae are biconvex, the third, sixth, and eighth biconcave. Moreover, since the sacral vertebra, generally the ninth, in all the Anura is invariably biconvex, the eighth being biconcave in the procoelous families, opisthocoelous like the remaining seven vertebrae in the other families, it is not difficult to imagine that in the Anura the production of pro- or opistho-coelous vertebrae depends simply upon the centra or articulating knobs happening to fuse either with the hind or the front end of the vertebrae. This must of course ultimately be determined by a mechanical problem of motion.

A second type of the vertebrae amongst the Anura is the epichordal type, an exaggeration in degree of the notocentrous tendencies of the more usual perichordal arrangement. It shows, namely, the almost complete suppression of all the ventral cartilaginous elements, so that the chorda remains for a long time on the ventral surface of the axial column in the shape of a flattened longitudinal band. These two types are not unconnected. The suppression of the ventral elements applies most typically to the trunk region, while hypochordal cartilage exists in the anterior cervical vertebrae, and above all in the coccyx. Typically epichordal are the vertebrae of Pipa, Xenopus, Bombinator, Pelobates, Discoglossus and Alytes. It is significant that the epichordal often coincide with opisthocoelous vertebrae, and still more suggestive is the fact that Bombinator is eminently aquatic, Pipa and Xenopus entirely so, having lost the tympanum, at least externally. The epichordal feature is not necessarily indicative of relationship. It has probably been developed independently in various groups, in correlation with a resumption of aquatic life. Various genera of Pelobatidae and most likely some Cystignathidae, e.g. Pseudis, will not improbably connect the two types and their several correlated features, for instance, the frequent reduction of the tympanic cavity.

The os coccygeum has retained rather primitive features in so far as much dorsal and ventral cartilage is developed; but this has almost entirely lost its metameric arrangement, and the posterior half of the coccyx is formed chiefly by the ventral mass of cartilage, while the dorsal elements are more or less reduced. Only two vertebrae, generally the tenth and eleventh of the whole column, are clearly visible, each being composed of a pair of dorsal and a pair of ventral cartilaginous blocks. The sacral vertebra articulates with the coccyx by one or two convexities, but in the Aglossa, in some Pelobatidae, and a few others, the coccyx is fused with the sacral vertebra. Beyond the first and second component vertebrae of the embryonic coccyx, the cartilage is continued in the shape of two dorsal, and one ventral, bands, which soon fuse with each other. Dorsally this cartilage surrounds the spinal cord; the latter degenerates towards the end of the tadpole-stage, leaving, however, the empty spinal canal. The chorda, completely surrounded by cartilage, persists into the post-larval stage, but is destroyed long before the creature attains maturity. Ultimately the whole coccyx ossifies.

The tail proper, namely that portion which is absorbed during the metamorphosis, remains throughout its existence in an apparently primitive condition. The chorda dorsalis and the spinal cord extend through its whole length, surrounded by continuous connective tissue without any cartilage; in fact it represents a piece of typical vertebral column before the appearance of cartilage. The reduction of this swimming organ begins at the hind end.

The vertebral column of the adult.–The first vertebra (we will call it the atlas since it carries the skull) is not, as in the Urodela, provided with an odontoid process. It articulates by two cups with the condyles of the occiput. In some Anura it co-ossifies, rather incompletely, with the second vertebra, regularly in the fossil Palaeobatrachus, often in Ceratophrys, Breviceps, and occasionally in Pelobates, Bufo, Rana, and Xenopus. This is, however, no justification for looking upon the first vertebra as a complex of two vertebrae, although the atlas is frequently very thick and broad, and even carries, in the Aglossa, considerable lateral wings or diapophyses. Those of the trunk-vertebrae are often very long, acting thereby as substitutes for ribs which are absent, except on the second, third, and fourth vertebrae of the Discoglossidae, and on the second and third of the Aglossa. In the adult Aglossa these ribs fuse with the processes which carry them.

The diapophyses of the sacral vertebra carry no ribs, the ilia being attached to them directly. They are either cylindrical as in the Ranidae and Cystignathidae, or they are more or less dilated as in all the other families, most strongly in the Pelobatidae and the Aglossa. In some members of the large sub-family of the Cystignathidae the otherwise cylindrical diapophyses are slightly dilated.