A simple arrangement of the ducts from the liver and gall-bladder is that found in the common Dog-Fish (Scyllium canicula). In this Elasmobranch a cystic duct leaves the gall-bladder, and, after receiving several hepatic ducts from the lobes of the liver, becomes the bile-duct and opens into the commencement of the intestine. In the Myxinoids and in the Dipnoi (e.g. Protopterus), there are but two hepatic ducts, one from each lobe of the liver; these unite and then meet the cystic duct to form the bile-duct (Fig. 155, A). The number of hepatic ducts may, however, be considerably increased, as, for example, in the Siluroid Amiurus,^[262] where 8-10 separate ducts join the cystic duct. In a few instances one of the hepatic ducts opens directly into the intestine, independently of that which unites with the cystic duct in forming the bile-duct. In the Dipnoi (e.g. Protopterus),^[263] and in some Teleostomi (e.g. Lepidosteus),^[264] the bile-duct receives the duct from the pancreas before opening into the intestine.

The Pancreas.—In the Cyclostomes (e.g. Petromyzon, Bdellostoma, Myxine) a rudimentary pancreas is apparently present, but the evidence as to its identity is not wholly conclusive. A well-developed pancreas occurs in Elasmobranchs, in at least one of the Dipnoi, and probably in most Teleostomi.^[265]

In Elasmobranchs the pancreas is a compact structure, uni- or bi-lobed, and entirely distinct from the liver. In Scyllium canicula (Fig. 153), the bilobed gland lies in the angle between the distal limb of the stomach and the adjacent portion of the intestine, and from the smaller of its two lobes the duct issues to pass to its intestinal aperture near the commencement of the spiral valve. In most of the Teleostomi in which its existence has hitherto been recorded, the pancreas is a singularly diffuse gland; and usually a considerable portion, or even the whole of it, is embedded in the substance of the liver, its lobules accompanying the ramifications of the hepatic artery and duct, and the portal vein. The pancreatic duct usually opens into the intestine near the aperture of the bile duct (e.g. Amiurus); sometimes the two ducts open on the apex of a common papilla (e.g. Acipenser and Amia), or by their union form a common duct (e.g. Lepidosteus). Among the Dipnoi a well-developed pancreas is present in Protopterus,^[266] embedded in the wall of the stomach and intestine, internal to the peritoneal investment of these organs, and extending even into the first fold of the spiral valve. The gland is traversed by fine ductules which unite together and open into the bile-duct just before the latter enters the intestine. In the remaining Dipnoi the existence of a pancreas has yet to be ascertained. Developmentally, the pancreas resembles the liver, and, histologically, is very similar to that of the higher Vertebrates, consisting of terminal glandular alveoli continuous with intermediary tubular portions, and eventually with the finer ductules, which, by their union, form the main efferent duct.

The Pyloric Caeca.—These structures are caecal outgrowths from the intestine, and are situated close to the pyloric extremity of the stomach and the intestinal apertures of the bile and pancreatic ducts. Wholly wanting in the Cyclostomata and Dipnoi, and, unless represented by a pair of caeca opening into the long, tubular, non-valvate anterior portion of the intestine in the Greenland Shark (Laemargus borealis),^[267] in the Elasmobranchs also, they are very generally present in the Teleostomi, although extremely variable both in number and arrangement in different families. In Amia there is no trace of pyloric caeca. Polypterus has a single short caecum with a thick muscular wall. In Acipenser, Polyodon, and Lepidosteus, on the contrary, pyloric caeca are unusually well developed. In Acipenser the caeca are not only numerous, but are so connected together by connective tissue and blood-vessels, and so invested externally by the peritoneum, as to form a large, compact, gland-like mass, communicating with the intestine by a single wide duct. In Polyodon the organ is essentially similar, but is lobed externally. In Lepidosteus (Fig. 155, B, py.c), the caeca are also very numerous, but relatively short, and, although united into a compact mass, open by four pit-like orifices into the intestinal cavity. In Teleosts the caeca are subject to extraordinary variations in number, size, and arrangement.^[268] In some families, and even in groups of higher taxonomic value, they are entirely absent, as is the case with the Siluridae, Esocidae, Cyprinodontidae, Labridae, Plectognathi, and Lophobranchii. The "Sand-eel" (Ammodytes) has but a single caecum; the Turbot (Rhombus maximus) two, and other Pleuronectidae three to five; and the Perch (Perca), three (Fig. 160, py.c).

In other Teleosts, on the contrary, these structures are much more numerous. In Labrus labrax there are about 60, in the Whiting (Gadus merlangus) 120, while in the Mackerel (Scomber scombrus) there are no fewer than 191. If few in number the caeca open separately into the intestine, but when numerous, more or fewer of them may unite to form a smaller number of efferent ducts, as in the Whiting, where four such ducts are formed. In some instances, as in the Tunny (Thunnus), the union of the caeca by connective tissue leads to the formation of a compact mass. As regards their arrangement, the caeca may either be disposed in a whorl round the intestine, as in the Whiting, or in a linear series, as in the Salmon (Salmo) and in some of the Clupeidae.

The mucous membrane lining the anterior pyloric caeca is often developed into a network of ridges, limiting crypt-like or tubular depressions; and not infrequently the epithelium is ciliated.

The precise function of these organs, whether digestive or absorptive, is still uncertain.^[269] That they may be digestive is suggested by the presence of certain amylolytic and proteolytic enzymes, but this obvious conclusion is to some extent vitiated by the close proximity of these organs to the stomach, and more especially to the intestinal orifice of the pancreatic duct. It is by no means improbable, however, that the caeca are both digestive and absorptive organs. An attempt has been made to show that the pyloric caeca and the spiral valve vary inversely as regards the extent of their development in different groups of Fishes.^[270] To some extent the reciprocal variation of these structures supports this view, but it is also evident that there are obvious objections to its unqualified acceptance. Thus, in some Teleostomi (e.g. Acipenser, Polyodon), exceptionally well-developed and numerous caeca and a spiral valve are both present. Amia with an almost vestigial spiral valve has no trace of pyloric caeca, and in Teleosts the absence of a spiral valve is associated with the complete suppression of the caeca in many large and important groups.

The Rectal Gland.—The "rectal" gland, or appendix digitiformis, is a small organ of unknown function with complex glandular walls, and a central duct opening dorsally into the terminal portion of the intestine.^[271] The organ is generally present in Elasmobranchs (Fig. 153, rct.gl), in which group the intestinal orifice of its duct may either be close to the termination of the spiral valve, or, as in Chlamydoselachus,^[272] near the cloacal outlet of the gut. An apparent representative of the gland, the "caecum cloacae," is also present in the Dipnoi,^[273] but communicates directly with the cloaca (Fig. 155, A, cl.c). The "rectal" gland is perhaps homologous with the intestinal caecum which is to be found in some Teleosts (e.g. Box vulgaris), and possibly also with the "caecum" (caecum coli), and its vermiform appendix in the higher Vertebrata.^[274] The caecum cloacae, on the contrary, is morphologically a urogenital sinus, formed as a dilatation of the fused hinder portions of the mesonephric ducts, and probably comparable with the sperm sacs of male Elasmobranchs, and also with the urinary bladder of Teleostomes.^[275]

CHAPTER X

THE RESPIRATORY ORGANS

The principal respiratory organs consist of a series of pairs of branchial clefts in the form of perforations in the side walls of the throat, which place the pharynx in free communication with the exterior. The first and most anterior of these clefts, the mandibulo-hyoid cleft or "spiracle," is situated between the mandibular and hyoid arches; the second, the hyo-branchial or hyoidean cleft, between the hyoid arch and the first branchial arch; and the remaining clefts between the succeeding branchial arches. On the anterior and posterior walls of more or fewer of the clefts highly vascular plate-like, or variously shaped filamentous outgrowths of their lining membrane are developed, which subserve the purpose of exposing the blood to the influence of the oxygen-containing water, and are termed branchial lamellae or "gills." In addition to their usual respiratory organs, the gills, a few Fishes utilise the air-bladder either as a functional lung or as an oxygen reservoir, and in others accessory breathing organs of various kinds are developed.

The arrangement of the branchial clefts and the gills may be conveniently studied first in the Elasmobranchs. Excluding the spiracles, there are usually in this group (Fig. 161, A), five pairs of branchial clefts, but in certain primitive members of the group the number may be larger. Thus, in Notidanus griseus (Hexanchus) and in Chlamydoselachus there are six, and in Notidanus cinereus (Heptanchus), seven clefts. The pharyngeal apertures of the clefts are relatively wide, but their external openings, which are freely exposed on the lateral surface of the head between the eye and the pectoral fin, are usually narrow and slit-like.

The successive clefts are separated from one another by a series of inter-branchial septa, each of which consists of the lining membrane of two contiguous clefts and a median fibrous sheet; it is further strengthened on its pharyngeal margin by a branchial arch, and more externally by the fringe of cartilaginous rods (branchial rays) with which the outer convex edge of each arch is provided. The anterior and posterior walls of each septum are produced into a number of outwardly-radiating vascular plates or folds (branchial lamellae or "gills"), which by their free edges project into the cavity of the cleft (Fig. 161, A). Although slightly free at their outer extremities, the lamellae do not extend so far as the external margin of the septum to which they are attached (Fig. 164, B). Each series of lamellae is termed a "hemibranch," and, from what has been said, it is obvious that each inter-branchial septum and its supporting branchial arch carry two hemibranchs, an anterior and a posterior, the two forming a complete biserial gill or "holobranch." The hyoid arch, however, has only a single hemibranch, viz. that pertaining to the anterior wall of the hyo-branchial cleft, and as the fifth or last cleft has a hemibranch only on its anterior wall, the fifth arch is gill-less.^[276] The spiracle is a vestigial cleft. At an early stage of embryonic growth it differs but little from its fellows, but subsequently degenerating it is represented in the adult by a tubular passage between the oral cavity and the exterior, which, however, is often complicated by the development of caecal outgrowths.^[277] The anterior wall of the spiracle often retains a rudiment of a hemibranch in the shape of more or fewer vascular lamellae, which, as they are supplied with arterial blood, and not with venous blood like the ordinary gills, are said to form a mandibular or spiracular "pseudobranch." The spiracle varies greatly in size in different families, being largest in the Trygons and Torpedos, and very small, or even absent in the Lamnidae. Its pseudobranch is best developed in the Notidanidae, where it has the essential structure of a true hemibranch, and, as in other Elasmobranchs, but to a greater extent, probably aids in the additional aeration of the blood which is distributed to the eye and brain. The characteristic opercular covering of the external apertures of the gill-clefts in the Teleostomi and Dipnoi is wanting in Elasmobranchs. It is interesting to note, however, that in Chlamydoselachus^[278] curious frilled cutaneous folds are developed as extensions of the outer edges of the inter-branchial septa, as well as of the hyoid region, and, like a series of incipient opercula, project backwards over the successive branchial clefts (Fig. 252).

While in many respects more primitive than in Elasmobranchs the branchial system of the Cyclostomata presents certain special and peculiar features. The branchial clefts assume the form of oval, antero-posteriorly flattened pouches or sacs, varying, however, in number, and in their mode of communicating with the exterior, in different genera. In the Lamprey (Petromyzon) there are seven pairs of obliquely-disposed gill-sacs opening externally by small rounded orifices, and by similar apertures, not directly into the pharynx, but into a branchial canal (Fig. 162, r.t), which underlies the oesophagus, and, while ending blindly behind the last pair of sacs, communicating in front with the oral cavity.^[279] The first of the series of gill-sacs corresponds to the hyo-branchial or hyoidean cleft of Elasmobranchs and other Fishes. Spiracles are absent in the adult, but in the embryo are represented by pouch-like outgrowths of the hypoblast of the oral cavity, which subsequently undergo singular changes.^[280] Thus, the outgrowths become converted into the lateral halves of a complete ciliated circum-oral groove, which is retained even in the Ammocoetes stage, and recalls the ciliated peripharyngeal ring of Ascidians. Another archaic feature is also to be noted in the continuity of the groove with a ciliated mid-dorsal pharyngeal ridge, which has been compared to the "dorsal lamina" of Ascidians, and to the equally characteristic hyperbranchial groove of Amphioxus.^[281] Ventrally also, the lateral halves of the groove unite to form a single groove, which, after receiving the median aperture of the thyroid rudiment,^[282] is continued backwards in the mid-ventral line of the pharyngeal wall as far as the last branchial arch. No trace of these ciliated structures is, however, to be met with in the adult.

The branchial lamellae are represented by a series of vascular horizontal and parallel ridges radiating outwards along the roof, floor, and lateral walls of each gill-sac, and invested by an epithelium which is partially ciliated. The inter-branchial septa are much thicker than in Elasmobranchs, and include not only the walls of adjacent sacs and the branchial muscles, but also contain cavernous peribranchial lymph-sinuses. The cartilaginous branchial skeleton is situated wholly external to the gill-sacs, the so-called branchial arches lying between the external apertures of the sacs, and directly beneath the superficial skin, or, in other words, on the outer margins of the inter-branchial septa, and not on the inner, as is invariably the case with the branchial arches of Fishes.

In the Hag-Fish (Myxine) (Fig. 163), there are usually six, very rarely seven, pairs of gill-sacs, all of which open directly into the pharynx, and not into a branchial canal as in the Lampreys. On the other hand, Myxine is unique in having the outer extremities of its gill-sacs produced into a corresponding number of tubular canals which, after a longer or shorter course obliquely backwards and outwards, unite to form on each side a ventrally-situated external aperture (Fig. 163). In the same genus a short canal, or oesophageo-cutaneous duct, passes from the pharynx behind the last gill-sac of the left side, and opens externally with the common external branchial aperture of that side.

In Bdellostoma there are usually six or seven pairs of gill-sacs, but some species have ten or even fourteen pairs.^[283] They agree with those of the Lamprey in having independent external apertures, but resemble the corresponding organs in Myxine in opening directly into the pharynx. An oesophageo-cutaneous duct is also present.^[284]

In the Holocephali there are but four branchial clefts, the fifth cleft being closed. Spiracles are absent in the adult, although present in the young of Chimaera. The branchial lamellae resemble those of Elasmobranchs, but the inter-branchial septa are somewhat shorter, so that the lamellae project slightly beyond their outer margins (Fig. 164, B). A hyoidean hemibranch is present. A noteworthy feature is the development of a cutaneous fold from the outer surface of the hyoid arch, which grows backwards over the gill-clefts, and, uniting above and below with the body-wall, terminates in a free posterior margin, just behind the last gill-cleft. By the growth of this opercular fold the gills become enclosed in a spacious branchial cavity, and the clefts communicate with the exterior through a slit-like opening between the free margin of the fold and the body-wall.

The reduction in the extent of the inter-branchial septa which is initiated in the Holocephali is carried to a still further extent in the Teleostomi. Commencing with the Chondrostei, and passing thence to the more specialised Teleostei, the septa become gradually reduced in length, and the branchial lamellae project freely beyond their outer margins to an increasing extent.

This modification, least marked in Acipenser (Fig. 164, C) and Polyodon, attains its maximum in the Teleosts (Fig. 164, D and E), where the branchial lamellae take the form of a double series of free filaments disposed along the convex outer margin of each branchial arch, and attached by their bases only to the reduced and inconspicuous septa. As a general rule each of the first four arches supports two hemibranchs,^[285] forming a biserial gill or holobranch. In shape the branchial filaments are usually somewhat triangular, and consist of an axial supporting cartilage or bone, invested superficially by a highly vascular mucous membrane. As in most of the preceding groups the fifth branchial arch is gill-less. All Teleostomi possess a well-developed movable operculum, supported by a more or less complete series of opercular bones, with or without the addition of branchiostegal rays (Fig. 161, B). The size of the external branchial aperture varies considerably. Usually the hinder and lower margins of the operculum are free, and then the aperture is spacious. Not infrequently, however, the more or less extensive fusion of the ventral and hinder edges of the operculum with the body-wall reduces the aperture to a narrow slit, as in the Eels and some Siluridae, or to a small upwardly directed pore, as in the "Sea-Horse" (Hippocampus). In the Symbranchidae the branchial apertures close dorsally, but fuse ventrally, leaving a single median orifice on the under side of the throat.

Open spiracles are wanting in most adult Teleostomi, but are, nevertheless, retained in the Crossopterygii (Polypterus), and in the Chondrostei (Acipenser and Polyodon). They have been observed, however, in the embryos of some Teleosts, as in the Salmon (Salmo),^[286] and even in the adults of Amia,^[287] Lepidosteus, and a few Teleosts^[288] are represented by pouch-like recesses of the oral cavity. A few vestigial branchial lamellae may be developed on the anterior wall of each spiracle in Acipenser and Polyodon, but are wanting in Polypterus, and, as in Elasmobranchs, represent a mandibular or spiracular pseudobranch.

The structure usually regarded as a hyoidean hemibranch in the Teleostomi differs greatly in its development in different members of the group. In Acipenser it is undoubtedly the hemibranch of the hyoid arch and is a true gill, receiving venous blood from the ventral aorta and returning arterial blood to the dorsal aorta, as in Elasmobranchs. In Polyodon and in Polypterus the hemibranch is suppressed. Lepidosteus,^[289] on the other hand, has two series of lamellae on the inner surface of the operculum, a dorsal and a ventral series meeting at an angle (Fig. 197). The ventral lamellae are supplied with venous blood, the dorsal with arterial,^[290] so that while the former retain their primitive character as a functional hyoidean hemibranch, the latter is a pseudobranch. It is interesting to note, however, that the development of this pseudobranch and its blood-vessels proves that it does not represent any portion of a true hyoidean hemibranch, but is really a spiracular pseudobranch.^[291] In most other Teleostomi a degenerate hemibranch occupies a similar position. In Amia^[292] it is very feebly developed, and is lodged in a canal communicating with the branchial cavity by a small aperture, and situated directly anterior to the dorsal end of the first branchial arch. Its blood supply is arterial, and the organ is therefore a pseudobranch. In Teleosts the hemibranch is invariably a pseudobranch; nevertheless, its primitive condition as a gill is indicated either by its structure or by its embryonic history. In some genera the pseudobranch consists of short free lamellae, as in some Pleuronectidae; or it is partly free and partly concealed, as in some of the Horse Mackerels (Caranx) and in Salmo; or it may be completely hidden beneath the oral epithelium, as in the Cod (Gadus), where the organ is very degenerate, and is little more than a "rete mirabile" of blood-vessels. The nature of the Teleostean pseudobranch is not in all cases quite clear. In Salmo it is said that there is no hyoidean hemibranch, and that the pseudobranch is really a persistent spiracular pseudobranch;^[293] hence it is probable that a like significance must be attached to this singular structure in other Teleosts. The evidence of the cranial nerves on this point is conflicting. If the pseudobranch pertains to the spiracular cleft its nerve supply should be derived from the nerve of that cleft—viz. the seventh or facial nerve; but if it represents a hyoidean hemibranch, then one would expect it to be innervated by the ninth or glossopharyngeal nerve. As a matter of fact, however, the organ is said to be supplied by the seventh in some Teleosts, and in others by the ninth nerve.

In the Dipnoi the branchial system is best developed in Neoceratodus, the increasing importance of the lungs as respiratory organs in Protopterus and Lepidosiren being associated with a corresponding reduction in the structural and functional development of the gills. There is no trace of spiracles in the adult.

In Neoceratodus^[294] there are five branchial clefts, including the hyobranchial. Each of the first four branchial arches carries a pair of hemibranchs, and, as in the Holocephali, the gill-lamellae are attached along nearly their whole length to a well-developed interbranchial septum (Fig. 165). A peculiarity of Neoceratodus, which has no counterpart in any other Fishes, is the extension of the branchial lamellae on to the dorsal and ventral walls of the branchial clefts, so that the hemibranchs on opposite sides of each cleft are continuous both dorsally and ventrally (Fig. 166). The fifth arch is gill-less. In addition to the normal gills there is also a hyoidean pseudobranch. As in other Dipnoi, an operculum forms the outer wall of the branchial cavity, and leaves but a narrow, slit-like external branchial aperture.

In Protopterus^[295] the number of branchial arches is increased to six, but, in consequence of the closure of the hyobranchial cleft, there are but five open clefts. The first, second, and third arches are wholly devoid of branchial filaments: the fourth and fifth support each a biserial gill, while the sixth arch retains only an anterior hemibranch, which, however, as the source of its blood supply seems to indicate, may consist of "emigrant" gill-filaments from the posterior hemibranch of the fifth arch.^[296] Interbranchial septa are practically non-existent, the flattened, leaf-like gill-lamellae being free except at their attached bases, and thus repeating a characteristic Teleostean feature. A "hyoidean" hemibranch or pseudobranch, supplied from the ventral aorta, is present, but as the hyobranchial cleft is closed it projects into the branchial cavity immediately in front of the cleft between the first and second branchial arches. In Lepidosiren^[297] the branchial arches are reduced to five and the clefts to four, the hyobranchial and fifth clefts being closed. There is a "hyoidean" hemibranch resembling that of Protopterus.

The facts furnished by the study of the numerical and structural variations in the gill-clefts, gills, and gill-arches of different groups of Fishes prove that atrophy of these structures takes place at opposite ends of the series. We have examples of this anteriorly in the suppression of the hyo-mandibular cleft and its hemibranch, and of the hyoidean hemibranch, as the result of the conversion of the mandibular and hyoid arches into jaws, or into skeletal supports for the jaws; and posteriorly, in the reduction which is evident when the generality of Fishes are compared with such primitive Elasmobranchs as Chlamydoselachus and Notidanus.

In most Fishes the concave pharyngeal margins of the branchial arches are fringed with a double series of either cartilaginous or bony tubercles or filaments, the "gill-rakers" (Figs. 161 and 164). The anterior row of gill-rakers on each arch usually interdigitate with those of the posterior row on the preceding arch, and in this way the two rows form a sieve-like mechanism to prevent any solid particles, which may enter the pharynx with the respiratory current of water, from passing into the gill clefts and clogging or otherwise injuring the branchial filaments.

In a few Fishes the gill-rakers are enormously developed, and subserve a function similar to that of the baleen plates of the Whalebone Whales in acting as a filter for straining from the water the small pelagic organisms on which the Fish feeds. This is notably the case in the great Basking Shark (Selache maxima)^[298] in which the closely-set, flattened, tapering gill-rakers may be so long as four or five inches, and, while somewhat resembling "whalebone" in appearance, have the histological structure of vascular dentine. The nature of the food, which in the stomach of one specimen examined consisted solely of an immense quantity of plankton, including Copepods and the larvae of other Crustaceans,^[299] affords clear evidence of the great value of such a filtering mechanism to this Shark, and, at the same time offers an explanation of the striking and significant reduction in the size of the teeth, which, relatively to the dimensions of the Fish, are so small as to be almost vestigial. A similar filter has been observed in an extinct Selache (S. aurata)^[300] from the Antwerp Crag, and also in an existing South African Shark (Rhinodon typicus);^[301] and in the latter, as in the Basking Shark, is associated with a marked reduction in the importance of the dentition. The long slender gill-rakers of the Chondrostean Polyodon also constitute an efficient filter, and the same may be said of several plankton-eating Teleosts.

The Mechanism of Respiration.—The aeration of the blood is effected by the rhythmical suction of water into the oral cavity, and its subsequent expulsion through the gill-clefts, bathing the highly vascular gill-lamellae in its course. In any single act of inspiration the mouth is opened, and the oral cavity enlarged by the lateral expansion of its walls. When the oral cavity is filled with water, the mouth is closed and the expiratory process begins. By the lateral contraction of the oral walls the water is driven outwards through the gill-clefts, and over the gill-lamellae. During this process the branchial arches become widely separated by the contraction of their muscles, the operculum is elevated, and the oesophagus is closed by the contraction of its muscular wall. In many Fishes the course of the expiratory water-current is controlled by special valve-like folds of the oral mucous membrane, the maxillary and mandibular "breathing-valves."^[302]

The rate of "breathing" varies considerably in different Fishes, even in allied species.^[303] In the Blue Wrasse (Labrus), and the Rockling (Motella), the number of respirations per minute is 15, in the Minnow (Leuciscus), and Stickleback (Gastrosteus), as many as 150. A deficiency of oxygen in the water accelerates the respiratory movements, and the Fish appears to "pant" or breathe hurriedly. In the Lampreys, both inspiration and expiration may take place through the external gill-apertures by the alternate expansion and contraction of the gill-sacs, more especially when the suctorial buccal funnel is used for the attachment of the animal. On the other hand, the singular habits of the Myxinoids involve a further modification of the respiratory process. In these Cyclostomata the inspiratory current enters the external naso-pituitary aperture and reaches the pharynx through the naso-pituitary canal, and thence, as an expiratory stream, traverses the gill-sacs on its way outwards. The pharynx is closed behind the last pair of gill-sacs by a constrictor muscle, which prevents the entrance of the water into the oesophagus, and converts the pharynx into a respiratory tube for the time being; but, when food is being swallowed, the pharyngeal constrictor is relaxed and the internal apertures of the gill-sacs are closed by the contraction of their own sphincter muscles.

In addition to the usual respiratory organs it is probable that in not a few Fishes the superficial skin may share with the gills the function of breathing. In this connexion may be mentioned the fact that in Periophthalmus the tail is used for respiration. Hickson^[304] observed that a species of this genus, frequenting the extensive sandy shores of the Island of Celebes, often rests with its tail in the water, the head and trunk being exposed. Under such circumstances the gills are probably of little use, and the tail is utilised as a breathing organ, principally, as Haddon^[305] subsequently pointed out, through the agency of its extremely vascular caudal fin.

Some Fishes possess larval breathing organs; others, even when provided with gills, either utilise the air-bladder, or develop special accessory organs, for aquatic or, more usually, for aerial respiration.

Larval Gills.—In early life many Fishes acquire larval gills, either as the result of the precocious growth of the normal gills, or by reason of the development of evanescent structures. In the embryos of Elasmobranchs "external gills," in the form of long filiform processes invested by hypoblast, are developed from the walls of all the branchial clefts, including the spiracles, and protrude outwards for some distance through the external apertures of the clefts (Fig. 167, B). They perhaps facilitate respiration within the egg, as they completely disappear after hatching; but there is also reason for believing that they aid in the absorption of nutriment. Similar gills are present in young Holocephali. In some larval Teleosts, as in certain genera of the Osteoglossidae and Mormyridae (e.g. Heterotis and Gymnarchus)^[306] these structures are remarkably developed (Fig. 239). The young of the Loach (Misgurnus) and of the Salmon (Salmo) also have the ordinary gill-filaments prolonged externally as filiform structures, which subsequently become reduced to their normal size.^[307] In its larval state Polypterus^[308] has a pair of pinnately-fringed ectodermal or cutaneous gills projecting from the lateral surfaces of the head behind and above the external branchial apertures (Figs. 168 and 281). Apparently as an individual peculiarity the right gill has been retained in a specimen of P. congicus so large as 22 cm. in length, although the left one had entirely disappeared.^[309] Each gill is supplied with blood from the ventral aorta by a vessel which ascends the hyoid arch, and is apparently the representative of the artery supplying the hyoidean hemibranch in Elasmobranchs. The efferent vessel of each gill joins the common trunk formed by the union of the efferent vessels of the normal gills of the same side.

The cutaneous gills of the Dipnoid Protopterus may also be included in the category of larval breathing organs. They consist of three simple unbranched filaments on each side of the head, and, as in Polypterus, are situated at the dorsal extremity of the external gill aperture (Fig. 309). Although usually represented in the relatively young or half grown specimens which, so far, have reached Europe, it is extremely probable that these organs atrophy in older individuals. Similar gills are present in the larval Lepidosiren (Fig. 311), but disappear at a much earlier stage. At no period of its development are larval gills present in Neoceratodus.^[310]

The Air-Bladder as a respiratory Organ.—In certain Fishes the air-bladder may become subservient to the function of respiration. In Amia and Lepidosteus the internally sacculated and vascular air-bladder is obviously adapted for air-breathing, and there are not wanting observations^[311] which suggest that the organ is actually used for this purpose after the fashion of a lung. According to Jobert,^[312] this is also the case with the sacculated air-bladder of certain Brazilian Teleosts, viz. Sudis gigas, Erythrinus taeniatus and E. braziliensis, since these Fishes die of asphyxia when the organ is cut off from communication with the exterior by the ligature of its ductus pneumaticus. It is in the Dipnoi, however, that the air-bladder becomes most completely a true lung. In Neoceratodus^[313] the lung is probably of the greatest use to the Fish when the rivers are low during the hot season and the water is charged with foul gases from decomposing vegetable matter, and possibly also when the water is filled with sediment in the rainy season. In Protopterus, and more especially in Lepidosiren, the partial atrophy of the gills renders it highly probable that the lungs are the principal breathing organs at all times. Nevertheless, it must be emphasised that in all these Fishes respiration by means of the air-bladder necessarily involves a transit of air to and from that organ through the ductus pneumaticus, and at present nothing is known as to the method by which such inspiratory and expiratory currents can be produced.^[314]

There is also some experimental evidence for the belief that the air-bladder of some Teleosts may be subsidiary to respiration by acting as a reservoir for the superabundance of oxygen which is taken into the blood through the gills, and subsequently reabsorbed into the blood when the Fish is in water containing relatively little oxygen.^[315] It is clear, however, that the conditions under which the air-bladder can be used in this way are by no means fully understood, for, under experiment, such Fishes died of asphyxia even though after death the air-bladder still contained upwards of fifty per cent of oxygen.

Accessory Organs of Respiration.—In certain Fishes of peculiar habits, or living under special external conditions, accessory respiratory organs are developed.

Although in this particular instance no special organs are formed, mention may first be made of the singular method of intestinal respiration in vogue in some Teleosts. In one of the Loaches (Misgurnus fossilis),^[316] air is swallowed and passed along the alimentary canal until it is finally voided at the anus. The mucous membrane of the intestine is extremely vascular, and hence the blood comes into sufficiently intimate relations with the swallowed air to admit of it exchanging carbon dioxide for oxygen. Intestinal respiration also occurs in species of the South American freshwater genera of Siluridae and Loricariidae, Callichthys, Doras, Loricaria, and Plecostomus;^[317] and in some cases the area of respiratory surface is considerably increased by the development of folds and processes of the intestinal mucous membrane.

In a few tropical Teleosts curious labyrinthiform organs are developed in connexion with certain of the branchial arches, and serve as accessory breathing organs. In the Indian "Climbing Perch" (Anabas scandens),^[318] of the family Anabantidae, the organ (Fig. 169) consists of three or more concentrically-arranged bony laminae, with wavy, crenulated margins, attached by a common bony base to the upper extremity of the fourth branchial arch, and enclosed in a special dorsal enlargement of the branchial cavity. The vascular membrane which invests the laminae is abundantly supplied with venous blood by a branch of the fourth afferent branchial artery, the equivalent efferent vessel joining the dorsal aorta. Essentially similar organs are found in several genera of Osphromenidae (e.g. Polyacanthus, Osphromenus, and Trichogaster). A simpler form of respiratory organ of somewhat the same type occurs in the Indian family Ophiocephalidae.^[319] In these Fishes there is, on each side, an accessory branchial cavity, situated above that which contains the gills, but freely communicating with it (Fig. 170). The cavity is lined by a thickened and puckered vascular membrane, but otherwise contains no special respiratory structures.