As to the causes which may have determined the shape and mutual relations of scales interesting suggestions have been made.^[164] Scales bear a segmental relation to the subjacent muscle-segments or myotomes, sometimes being disposed in oblique transverse rows coinciding with the latter, or the rows may be so far increased as to be multiples of the myotomes. From mechanical considerations depending on the sigmoid shape and interdigitating relations of the myotomes and their separating fibrous septa or myocommata, and the attachment of the myocommata to the dermis, the contraction of the myotomes during the lateral flexions of the trunk in swimming has a tendency to wrinkle the skin into definitely circumscribed rhombic areas, thus determining the shape, limits, and disposition of the scales which are developed in those areas. The rhombic was probably the primitive shape of scales, and is certainly characteristic of the palaeontologically older types of scaly Fishes. Generally the rhombic condition is associated with a peg-and-socket articulation between the upper and lower margins of adjacent scales. But a rhombic squamation is not without disadvantages, and would certainly impose some restriction on the lateral flexures of the body in swimming, and hence in the different groups of Fishes it may happen that, in the more specialised forms, an imbricated cycloid squamation supersedes a rhombic condition, and with the change the Fish acquires greater lateral mobility. Even in the same Fish the gradual substitution of the cycloid for the rhombic type may be observed. In the Australian Aetheolepis,^[165] a fossil genus related to the European Liassic Dapedius, there is a gradual transition along the sides of the body between the articulated rhombic scales of the relatively immobile trunk and the cycloid overlapping scales of the flexible tail; and it may be mentioned that, even where a typical rhombic squamation exists, the peg-and-socket articulation may be wanting in the caudal region, so as to ensure greater freedom of movement. Mechanical considerations may also explain the overlapping of cycloid scales. From the mode of attachment of the myocommata to the dermis, the contractions of the myotomes, through the pull which they exert on the former, tend to deflect or depress the scale-areas, particularly at their anterior margins.

In the surviving Crossopterygii, as in Polypterus, the scales are rhomboidal and thick, and they only slightly overlap. They articulate with one another by means of marginal peg-and-socket articulations (Fig. 106, B). A thick layer of hard, glistening, enamel-like substance or "ganoin" forms the outer layer of the scale; the inner layer consisting of bone in which dentinal tubules as well as bone-cells are present. In the numerous fossil members of the group the scales are either rhomboidal or cycloid.

The oldest representatives of the Chondrostei, the Palaeoniscidae (Fig. 283) possessed a complete armature of rhombic scales, but in all the surviving members of the group the scales have undergone considerable modification in some respects, and in others are degenerate. In the Sturgeon (Acipenser)^[166] the primitive rhombic squamation is retained only on the sides of the terminal part of the tail, and there they are in close apposition in oblique rows. The rest of the body is traversed by five widely-separated longitudinal rows of large bony scutes, which, like the rhombic scales, are furnished with ridges and projecting spines (Fig. 102). Between the rows of large scales there are numerous denticle-like structures arranged in oblique rows. Each of these consists of a basal plate imbedded in the dermis, and of one or more projecting spines which perforate the epidermis. All the scales have the same minute structure, consisting mainly of bone; but the surface layer and the spines seem to be composed of a hard laminated substance from which bone-cells are absent (ganoin). In Polyodon the scutes are wanting, but vestigial denticles are retained.

Among the Holostei the scales are very different in the two surviving members of the group. In Lepidosteus (Fig. 103) the scales are rhombic, and both in arrangement and structure, as well as in their method of articulating with one another, they closely resemble those of Polypterus. In Amia,^[167] on the contrary, the relatively thin scales are cycloid in shape, and in their imbricated arrangement, in their enclosure in pouches of the dermis, and in the absence of any superficial covering of ganoin, they are very similar to the scales of the more typical Teleosts (Fig. 295). The resemblance extends even to histological structure, for each scale consists of an outer layer of bone, which gradually passes into an inner fibrous stratum.

In Teleosts the usually thin and flexible scales are primarily developed from dermic papillae, but subsequently they come to lie in pockets or pouches in the dermis. As a rule no spines are developed, and so far no trace of an enamel organ has been detected during their development. The scales in their dermal pouches are disposed obliquely to the surface of the body, so that the hinder part of one scale overlaps the anterior portion of the scale next behind it (Fig. 104). Only the free hinder part of each scale has an epidermic investment (Fig. 105). In minute structure each scale consists of an outer layer of bone, which, like the bone of the endoskeleton, may either be homogeneous except for a feeble lamination, or it may contain bone-cells arranged in successive layers, parallel to the surface of the scale. In addition, there is an inner fibrous stratum in which the fibrous bundles in any one plane cross those in planes above or below them. The scales are either "cycloid," that is, they have smooth, unbroken margins (Fig. 105), or the free margin of each scale is produced into a series of tooth-like spines, and the scale is said to be "ctenoid" (Fig. 106, A).

Some Teleosts, however, have scales which are neither cycloid nor ctenoid, and in certain features seem to be intermediate between ordinary Teleostean scales and dermal denticles. Thus, on certain parts of the body of Centriscus,^[168] each scale consists of a rhombic basal plate, produced into a curved, backwardly-inclined spine, the axis of which contains a pulp-cavity opening on the inner surface of the basal plate (Fig. 107). Some Malthidae (e.g. Malthe^[169]) have similar scales, but with round basal plates and solid spines (Fig. 108, B). Similar scales (Fig. 109), sometimes rhombic in shape, with one or more spines, which may be simple or branched, are also found in the Sclerodermi (e.g. Balistes, Monacanthus, Triacanthus).^[170]

Amongst some of the usually scaleless Siluroid Fishes the scales assume a very peculiar structure. In Hypostoma^[171] (Plecostomus) the sides and back of the Fish are covered by large bony plates, but on the under surface and on the fins these are replaced by much smaller ones. Both kinds, however, carry numerous small movable spines implanted in sockets (Fig. 110), a fact which suggests comparison with a stage in the development of the scales of Lepidosteus, when the independently formed and evanescent spines have not yet fused with the basal plates.

In other Teleosts, as in the Agonidae and some Triglidae, the body is completely cuirassed with large keeled bony plates. The singular appearance of many of the Plectognathi is largely the result of the curious modifications which their scales undergo. In some of the Coffer-Fishes (Ostracion) these structures assume the form of polygonal bony plates which suturally articulate with one another and enclose the trunk in a rigid cuirass, from which the scaleless tail protrudes behind (Fig. 438); while in some Globe-Fishes and Porcupine-Fishes (e.g. Tetrodon, Diodon) the prolongation of the scales into strong erectile spines equally well serves the purpose of protection (Fig. 439).

Most Teleostomi have the scales along the "lateral line" perforated by single or multiple apertures, through which the sensory canal communicates with the exterior.

In a few Teleosts scales are entirely absent, as in most Siluridae; or they exist only as microscopic vestiges hidden in the skin, as in Eels; or, as in such naked forms as Antennarius marmoratus and Lepadogaster, and in some Siluridae, they become reduced to mere papillae of the dermis.

The concentric rings observable (Fig. 105) on the surface of many Teleostean scales are an index to the age of the Fish.^[172] The formation of these rings depends on the fact that the lines of growth on the surface of the scale are more widely separated from one another on that portion of the scale formed during summer, and relatively closer together on that part which is formed during the winter; the more rapid growth in the warmer season probably being due to favourable conditions as to food and temperature, and the retarded growth of the colder season to the reverse. Hence, by counting the alternating zones of close-set winter lines and less closely approximated summer lines of growth, a reliable clue may be gained as to the age of the Fish.

In the Dipnoi,^[173] as in Teleosts, the scales are enclosed in dermal pockets, and exhibit a regular, imbricated disposition in oblique rows (Fig. 304, A). In shape they are nearly cycloid, or slightly oval, with the long axis coinciding with that of the body. Structurally, also, they bear some resemblance to Teleostean scales, although differing in details. On the outer surface of the scales there are numerous small conical spines. No significance, other than as an example of evolutionary convergence, can be attached to the resemblance between the scales of Fishes so widely separated as the Dipnoi and the Teleosts.

All known fossil Dipnoi had scales of a similar character, although differing greatly in size in different genera. In some (e.g. Dipterus) a layer of enamel-like substance invests the exposed portions of the scales.

CHAPTER VIII

THE SKELETON

All Fishes possess an internal skeleton which, in order that it may be distinguished from the more superficial scaly exoskeleton described in Chapter VII., is termed the endoskeleton. The latter consists (i.) of an axial part, including the vertebral column and the skull; and (ii.) of an appendicular portion, consisting of the skeleton of the limbs and their supporting pectoral and pelvic girdles.

The Vertebral Column.^[174]—The individual segments or vertebrae which, arranged in a linear series, collectively form the vertebral column, are highly complex structures, each being formed by a number of vertebral elements, the sum total of which constitutes a vertebra. Perhaps the best conception of the nature of vertebral elements is to be gleaned from the study of such primitive Fishes as the Elasmobranchs, in which not only are all the vertebral components present, but they are less modified by suppression and fusion than in most other Fishes, and on this account they afford a convenient introduction to the study of the puzzling eccentricities of vertebral structure in other groups. Selecting any common Dog-Fish, such as Scyllium canicula, and starting with an early embryonic stage, it may be stated that the first indication of a vertebral column is the formation of the notochord, which, invested by its chordal sheath, extends from the tip of the tail to a point on the under surface of the brain just behind the hypophysis or pituitary body.

Subsequently, a number of cartilaginous pieces are developed in connexion with the dorsal and ventral surfaces of the notochord, which, as they form portions of a system of dorsal and ventral arches, are termed "arcualia" (Fig. 111). On the dorsal side there are: (i.) a series of paired basi-dorsal cartilages (neurapophyses or neural arches), the two elements of each pair contributing to form the side walls of the neural canal in which the spinal cord is lodged (Fig. 112, A); (ii.) a series of inter-dorsal cartilages (intercalary neural arches), regularly alternating with the preceding, and completing the walls of the neural canal by filling up the intervals between the basi-dorsals; and (iii.) a series of supra-dorsal elements, typically also in pairs, but in the Dog-Fish fused to form single median cartilages. Of the latter there are two sets—one the supra-basi-dorsals, or neural spines, are situated over the basi-dorsals; and the other, supra-inter-dorsals, alternating with the former, lie over the inter-dorsals, the two series forming the keystones of the dorsal arches, and thus completing the roof of the neural canal. On the ventral side of the notochord this arrangement is substantially repeated by a series of ventral arcualia, which, however, are somewhat differently arranged in the trunk and tail. Thus, in the trunk there are: (i.) a series of basi-ventral or haemal cartilages, corresponding with the basi-dorsals above, which grow out laterally into short processes, the parapophyses or transverse processes, and terminate in (ii.) short, slender cartilages—the costal elements or ribs—which may perhaps be regarded as the ventral equivalents of supra-basi-dorsals. The ribs project outwards into the dorsal wall of the coelom and end in the myocommata separating the myotomes of the body-wall. In the tail the basi-ventrals lose their ribs and, growing downwards into ventral prolongations, they unite in pairs beneath the caudal artery and vein, and so form a series of inverted arches (haemal arches) enclosing a haemal canal (Fig. 112, B). The apex of each arch is prolonged into a median process or haemal spine. Although not recognisable in the Dog-Fish, paired inter-ventral cartilages, corresponding with the inter-dorsals above, are present in some Elasmobranchs and alternate with the basi-ventrals. In the caudal region of others (e.g. Skates and Rays) ventral counterparts of the supra-interdorsals are present, and are termed infra-ventral cartilages. Much in the same way that their dorsal equivalents enclose a neural canal, so the ventral arcualia partially surround the viscera-containing coelom in the trunk; and in the tail, but more completely, the vestigial coelom of that region or the haemal canal.

The different vertebral components are by no means of equal morphological value. The basi-dorsals and basi-ventrals, and the inter-dorsals and inter-ventrals, are the primary elements and the most important. The supra-dorsals are merely cartilages segmented off from the basi-dorsals and inter-dorsals, while the ribs and the infra-ventrals are similarly derived from the basi-ventrals and inter-ventrals respectively. As to the vertebral elements which collectively form a vertebra in the Dog-Fish, it would seem from evidence afforded by the neuromeres,^[175] and more especially by the facts of development, that each complete skeletal segment or vertebra consists of a pair of basi-dorsals with the preceding pair of inter-dorsals, and of a pair of basi-ventrals with the next succeeding pair of inter-ventrals. It must be emphasised, however, that, considered as a joint or segment in a flexible back-bone, a vertebra is a physiological unit, the morphological value of which may differ widely in different Fishes. Hence, in other Fishes, the grouping of vertebral components to form individual vertebrae may be quite different to that which takes place in the Dog-Fish, and may even be accompanied by their more or less complete fusion.

In the more primitive types of vertebral column, such as are characteristic of many fossil and not a few existing Fishes, arcualia alone are present, and remain associated with a persistent notochord which has grown with the growth of the animal. In the more specialised Fishes, on the contrary, the need of an axial support for the body, which, while retaining the necessary flexibility, must possess greater strength, has resulted in the development of a series of solid cartilaginous, calcified or bony, discoidal joints or segments, the centra, which surround and more or less completely replace the notochord, and, while supporting, form also a bond of connexion between the dorsal and ventral arches. Notwithstanding their superficial resemblance, an important developmental distinction is to be noted in the mode of formation of centra in different Fishes, which enables one kind to be distinguished as "chorda-centra," and another as "arch-centra."^[176] Chorda-centra are centra formed by the conversion of the chordal sheath into a series of ring-like cartilaginous segments, which subsequently, by a process of inward thickening, become biconcave, disc-like structures, and more or less completely replace the notochord, except in the spaces between them. Arch-centra, on the other hand, owe their formation to the growth of the bases of the primary arcualia round the notochord, external to the chordal sheath, and their subsequent fusion to form annular segments, which, later, become biconcave centra. Of Fishes which possess vertebral centra the Elasmobranchs alone have chorda-centra; the Holostei and the Teleostei, and very probably the Crossopterygii also, having arch-centra. The Dipnoi and the Holocephali, and the Chondrostean Teleostomi are acentrous—that is, they are devoid of vertebral centra and possess a persistent notochord. Neither in their embryonic development nor in their evolution in time are the different vertebral components synchronous in their appearance. Developmentally, the arcualia are the first to be formed, and of these those on the dorsal aspect of the notochord appear earlier than their representatives on the ventral side, while the centra are the last of all; and in a general way the palaeontological sequence agrees with the embryological.

The independent evolution of a more specialised vertebral column from a more primitive one may often be traced within the limits of the same group of Fishes when the more ancient genera are compared with the more recent. In the Elasmobranchs and the Crossopterygii, for example, the oldest known types were acentrous, while the more recent have acquired calcified or bony centra, and altogether they have reached a more advanced stage of vertebral evolution. Some Fishes, like the Chondrostei and the Dipnoi, seem, however, to exhibit comparatively little advance in vertebral structure, since both the Palaeozoic and the living representatives of these groups agree in being acentrous.

Some of the more notable features in the structure of the vertebral column in the Cyclostomata and Fishes will now be briefly considered.

In the Cyclostomata the acentrous vertebral column is more primitive than in any other Craniates, and in the Lamprey it consists of a persistent notochord, supporting a series of isolated cartilages on each side of the spinal cord.^[177] As two pairs of these cartilages are included in each neuromere it is possible that they represent alternating basi-dorsals and inter-dorsals. There are no ventral arcualia in the trunk and no ribs. In the Hag-Fish (Myxine) the dorsal cartilages are restricted to the tail.

The description of the vertebral column of the Dog-Fish may be taken as fairly applicable to Elasmobranchs in general, and hence only certain notable features in some other members of the group need be referred to here. The most primitive Elasmobranchs, the Palaeozoic genera Cladoselache and Pleuracanthus were acentrous, although calcified rings have been observed in a Permian species of the latter genus and scattered calcifications in others. Some of the earlier Mesozoic genera (e.g. Hybodus) were also devoid of centra, at least in the trunk-region. The first indication of complete centra occurs in the Lower Lias Cestraciont, Palaeospinax.^[178] All the later extinct, as well as all existing forms, have more or less well-developed centra, hardened by the deposit of lime salts in their primitively cartilaginous substance, but never in the form of true bone.

The mode in which the lime is deposited is marked by certain peculiarities which are characteristic of particular families^[179] (Fig. 113). In some genera, as in the extinct Palaeospinax and the living Acanthias and Scymnus, the calcified portion of each centrum takes the form of a cylinder constricted across the middle, like two cones joined apex to apex (cyclospondylic). This condition is probably the most primitive, but it may be modified in other genera by the further addition of calcic salts in two different ways. Thus, the deposit may take place by the simple addition of concentric layers to the original constricted cylinder (tectospondylic), as in the Skates and Rays; or it may take the form of a series of longitudinal plates radiating outwards from the cylinder, and giving rise to a star-like pattern in cross-section (asterospondylic), as in Scyllium and Lamna. In most living Elasmobranchs (e.g. Scyllium), but not in such genera as Notidanus, Heterodontus, and Squatina, the bases of the dorsal and ventral arches grow round the centra and meet, or even fuse, so that the latter become surrounded by rings of cartilage which, after a fashion, suggest incipient arch-centra (Fig. 112, A). The caudal portion of the vertebral column is often described as "diplospondylic," that is, there are two centra, two pairs of basi-dorsals, two pairs of inter-dorsals, and two pairs of basi-ventrals, or in other words, two vertebrae to each neuromere^[180] (Fig. 111, B).

The Holocephali have a vertebral column essentially similar to that of other Elasmobranchs, but of a more primitive type (Fig. 114). The notochord is persistent and there are no centra; but ring-like calcifications, four or five to each neuromere, occur in the chordal sheath in Chimaera, although not in Callorhynchus. Ribs are absent. In the whip-like terminal portion of the tail the arcualia and the notochord become replaced by a slender continuous filament of cartilage.

In the more obvious features of vertebral structure the Dipnoi^[181] have much in common with the Elasmobranchs, especially with certain of the acentrous Palaeozoic representatives of that group. The notochord is persistent, centra are wanting, and the different vertebral components continue to retain their primitive distinctness. On the other hand, the basi-dorsals are much better developed than the inter-dorsals, which are either vestigial or absent. The basi-dorsals unite in pairs over the spinal cord to form complete neural arches, and each arch supports dorsally the legs of a Λ-shaped, gable-like element or neural spine, which probably represents a pair of fused supra-basidorsals. Ventrally, there are basi-ventral cartilages, fused in pairs beneath the notochord, and supporting well-developed, bone-ensheathed ribs. Inter-ventrals appear to be absent. Each neuromere corresponds with a pair of basi-ventrals, of basi-dorsals and of inter-dorsals. The haemal arches and spines are formed partly by the basi-ventrals, but mainly by the ventral union of the successive pairs of ribs. As in the Holocephali, the terminal arcualia of the tail become fused into a straight axial cartilaginous filament, transversely divided into segments, which replaces the notochord. Each segment supports a variable number of dorsal and ventral gable-pieces, or neural and haemal spines. Certain of the vertebral components, such as the ribs, and the neural and haemal spines, are ensheathed by membrane bone.

With certain modifications in details the preceding description will also apply to the vertebral column of the Chondrostei (Fig. 115). It will be noted, however, that the inter-dorsals are much better developed than in the Dipnoi, although when compared with the basi-dorsals they take but a small share in forming the walls of the neural canal. Well-developed but somewhat fragmentary inter-ventrals are present. The haemal arches and spines are formed by the downgrowth and ventral union of the basi-ventrals as in the Dog-Fish, and apparently without the aid of costal elements. In Polyodon the ribs are vestigial,^[182] but in Acipenser they are well developed. The neural arches and spines, and their haemal representatives in the tail, and also the ribs, are partially ossified, or ensheathed by bone.

In the existing Crossopterygii, Holostei, and Teleostei, popularly known as the "bony Fishes," the vertebral column assumes a more familiar character, and at the same time we meet with interesting illustrations of the different methods by which the separate component vertebral elements of the more primitive types of "backbone" are concentrated together in groups, and fused to form that complex physiological product, the complete bony vertebra.^[183] In most of these Fishes each vertebra is formed by the aggregation and fusion of a pair of basi-dorsals and a pair of basi-ventrals, and includes, in addition, a pair of inter-dorsals, which may either be the pair in front of the basi-dorsals or the pair behind, and also a pair of inter-ventrals, which, similarly, may be the pair in front or behind the basi-ventrals (Fig. 116). The product of this fusion is a series of bony vertebrae, each consisting of a biconcave arch-centrum, which includes the fused basal portions of a pair of basi-dorsals and a pair of basi-ventrals. The distal portions of the basi-dorsals form the neural arch, while the rib-bearing parapophyses are lateral outgrowths from the basi-ventrals which otherwise have become merged in the centrum. Finally, the centrum is completed by its fusion with a pair of inter-dorsals and a pair of inter-ventrals. Supra-dorsal elements may also be included as minor contributory factors. The supra-basi-dorsals co-ossify with their basi-dorsals and then unite to form the ordinary unpaired neural spine of most bony Fishes, or, as in Amia, they remain distinct from each other, and are obvious as a double spine. In Lepidosteus these elements co-ossify with the neural arches and form the post-zygapophyses. Supra-inter-dorsals have been identified in the embryo as distinct elements, but their eventual fate is not always known. In Lepidosteus they persist as distinct cartilages in the adult (Fig. 118, A). Well-developed bony ribs are usually present. The haemal arches of the tail are formed by the downgrowth of the parapophyses and their ribs, or by the latter alone, and by their ventral union to form haemal spines; consequently, each arch always includes a pair of costal elements. With such general features in common there are certain notable variations in some of these Fishes, to which brief reference may be made.

Little is at present known of the development of the vertebral column in either of the only two existing genera of Crossopterygii, Polypterus^[184] and Calamichthys, and hence the precise mode of grouping of their vertebral components to form vertebrae is unknown. The condition of the vertebral column in the fossil forms varies greatly in different families, but in none is it so specialised as in the surviving members of the group. In the Devonian Holoptychidae, and even in genera so comparatively recent as the Upper Cretaceous Coelacanth Macropoma, the persistence of the notochord and the absence of centra indicate a very primitive grade of vertebral evolution. The Devonian and Carboniferous Rhizodontidae (e.g. Eusthenopteron and Rhizodus), on the contrary, seem to have had well-ossified ring-like vertebrae.

In the caudal region of Amia the basi-dorsals and basi-ventrals, and the inter-dorsals and inter-ventrals, form separate arch-centra which remain distinct; hence each vertebra is double, and there is a regular alternation of arch-bearing "pre-centra" and arch-less "post-centra" (Fig. 117, D). In the trunk-region the pre- and post-centra have fused, and in this region the vertebrae are single.

A very primitive type of vertebral column occurs in some of the Jurassic allies of Amia, in which certain of the vertebral components, confluent in the adult Amia, retain some measure of their primitive distinctness.^[185] Thus, in the precaudal region of Eurycormus (Fig. 117, B) there is a series of alternating dorsal and ventral half-rings of bone, which between them completely invest the persistent notochord. Each ventral half-ring or "hypocentrum" represents a pair of fused and ossified basi-ventrals, and possibly also a pair of included inter-ventrals, and supports dorsally a pair of basi-dorsals, forming a neural arch, and laterally a pair of ribs. The dorsal semi-rings, or "pleuro-centra," similarly represent fused and ossified pairs of inter-dorsals. In the tail, modifications approximating to what is seen in the caudal region of Amia are to be noticed (C). By the upgrowth of the ventral arch-bearing semi-rings, and their conversion into complete rings encircling the notochord, incipient pre-centra are formed, and by a similar modification of the down-growing, archless, dorsal half-rings, structures comparable to post-centra are produced. In brief, Eurycormus, as well as such other extinct Amioid genera as Caturus (Fig. 117, A), Callopterus, and Euthynotus, retain in the adult a stage of vertebral evolution which is closely paralleled by transitory stages in the embryonic and young forms of Amia.

Lepidosteus^[186] is unique amongst existing Fishes in having opisthocoelous vertebrae; that is, the centra are convex in front and concave behind, and therefore articulate with one another by ball-and-socket joints (Fig. 118). This condition is due to the development of a series of intervertebral rings of cartilage round the notochord. The subsequent inward growth of each of these rings leads to the constriction, and ultimately to the complete obliteration, of the notochord, much in the same way as by the growth of ordinary centra. Later, this solid mass of cartilage becomes transversely divided by a cleft which is convex anteriorly and concave behind (Fig. 116, C), and of the two portions one fuses and co-ossifies with the centrum of the vertebra in front, and the other with the one pertaining to the vertebra behind. Reference to Fig. 116 will show that the grouping of the vertebral elements to form the individual vertebrae is not the same as in Amia.

In the dominant group of existing Fishes, the Teleostei, the centra are almost invariably biconcave, although in the Eels they may be flat or even slightly convex in front. Ribs are absent in the Syngnathidae and in the Plectognathi. In addition to the usual articulation between the centra, the vertebrae often articulate with one another by means of paired processes arising from the anterior margin of each neural arch, or from the centrum at the base of the arch (pre-zygapophyses), and meeting similar processes which project either from the hinder margin of the arch of the vertebra in front, or from the adjacent portion of its centrum (post-zygapophyses). The haemal arches may have similar processes (Fig. 119). One, two, or in some Teleosts, three pairs of slender intermuscular bones radiate outwards from the centra into the myocommata (epicentrals), or from the neural arch (epineurals), or from the ribs (epipleurals).

The Ribs.—It is doubtful if the structures termed "ribs" are homologous in the different groups of Fishes. There appear to be two kinds, distinguishable as dorsal and ventral ribs (Fig. 156). Dorsal ribs are situated in the fibrous tissue separating the epiaxial from the hypaxial muscles of the body wall, and they take no part in forming the haemal arches of the caudal region. Ventral ribs, on the other hand, always lie internal to the hypaxial muscles, and directly external to the peritoneal lining of the coelom, and they usually contribute to the formation of the haemal arches. To the former belong the ribs of the Elasmobranchs, and to the latter the ribs of the Teleostomi and Dipnoi. Polypterus alone has both kinds of ribs.

The Skull.

The skull is a highly complex structure, the various components of which are as different physiologically as they are morphologically. It consists (i.) of the cranium, for the enclosure and protection of the brain; (ii.) of sense capsules, which fulfil a like function for the auditory, visual, and olfactory organs; (iii.) of certain vertebrae or vertebral elements fused with the hinder part of the cranium; (iv.) of a series of visceral arches; and (v.) of a series of paired or median cartilages developed in relation with the mouth and nostrils, which may be collectively spoken of as "labial" cartilages.

The cranium is formed in the embryo from two pairs of cartilaginous rods or plates, developed in the mesoblast of the head. Of these the posterior pair, or parachordals, underlie the hinder part of the brain, and are situated one on each side of the cranial portion of the notochord. The anterior pair or trabeculae are pre-notochordal, and lie beneath the anterior portion of the brain.^[187] Between their hinder extremities, and in front of the anterior termination of the notochord, is the pituitary body. As development proceeds the parachordals blend with each other and with the trabeculae, while the latter fuse in front to form a median plate—the mesethmoid cartilage. The hinder portions of the two trabeculae remain distinct for some time, and enclose between them the pituitary fontanelle, but later they fuse beneath the pituitary body, leaving, however, a pit for its reception—the pituitary fossa. Cartilaginous capsules are formed round the cranial sense organs. The auditory or periotic capsules fuse on each side with the parachordals. The optic capsules, either fibrous or cartilaginous, remain free, and do not fuse with the adjacent trabecular region. The olfactory capsules alone are not developed independently, but are formed as lateral outgrowths from the mesethmoid plate. Later, the parachordals and trabeculae grow upwards on each side round the brain, and to a greater or less extent they meet and fuse on its dorsal surface, thus enclosing the latter organ in a cranial cavity, leaving, nevertheless, a large foramen behind (foramen magnum) through which the brain is continuous with the spinal cord. In this condition the primitive cartilaginous cranium, with its included sense-capsules, has reached a stage which is permanently retained in such Fishes as the Elasmobranchs.

The visceral arches consist of a number of pairs of curved rods of cartilage, at first simple, but subsequently segmented, and developed in the splanchnic mesoblastic walls of the oral cavity and pharynx. Each rod is connected with its fellow by a median cartilage in the floor of the pharynx, so that the whole form a series of dorsally incomplete hoops encircling the anterior portion of the alimentary canal. No doubt all the visceral arches were originally branchial arches, and were so disposed between the successive gill-clefts as to support their walls and the vascular folds or gill-lamellae to which they gave rise. In Fishes most of the arches still retain their primitive gill-supporting function, but the first or mandibular arch has become modified to form upper and lower jaws, although in the Sharks and Dog-Fishes it may lie in front of a gill-cleft and still be associated with vestigial gills. The second or hyoid arch is less removed from the condition of a branchial arch, and generally supports either a functional or a vestigial gill, but in most Fishes it has acquired the secondary function of forming a suspensorium for the attachment of the jaws to the cranium.

The skull of the common Dog-Fish, Scyllium canicula (Fig. 120),^[188] may be studied as a type which in the adult remains cartilaginous, and has no secondary addition of cartilage- or membrane-bones. In this Fish the chondrocranium, or primary cartilaginous cranium, presents the appearance of a somewhat depressed oblong box, which has a complete roof, side-walls, and floor, but is open in front (anterior cranial fontanelle) and also behind (foramen magnum). The hinder, or parachordal portion of the cranium surrounds the foramen magnum, and there forms the occipital region. At the ventral margin of the foramen there are two prominences, or occipital condyles, for articulation with the first vertebra, and between them the remains of the notochord are traceable into the cranial floor.