Transcriber's Note:

The cover image was created by the transcriber and is placed in the public domain.

GUIDE TO THE STUDY OF FISHES

VARIATIONS IN THE COLOR OF FISHES
The Oniokose or Demon Stinger, Inimicus japonicus (Cuv. and Val.), from Wakanoura, Japan. From nature by Kako Morita.
Surface coloration about lava rocks.
Coloration of specimens living among red algæ.
Coloration in deep water; Inimicus aurantiacus (Schlegel).

A GUIDE
TO
THE STUDY OF FISHES

BY

DAVID STARR JORDAN

President of Leland Stanford Junior University

With Colored Frontispieces and 507 Illustrations

IN TWO VOLUMES

Vol II.

"I am the wiser in respect to all knowledge

and the better qualified for all fortunes

for knowing that there is a minnow in that

brook."—Thoreau

NEW YORK

HENRY HOLT AND COMPANY

1905

Copyright, 1905

BY

HENRY HOLT AND COMPANY

Published March, 1905

CONTENTS
VOL. II.

LIST OF ILLUSTRATIONS
VOL. II.

ERRATA[^[1]]
Vol. II

The adoption of the Code of the International Congress of Zoology necessitates a few changes in generic names used in this book.

[1]. For most of this list of errata I am indebted to the kindly interest of Dr. B. W. Evermann.

CHAPTER I
THE GANOIDS

Subclass Actinopteri.—In our glance over the taxonomy of the earlier Chordates, or fish-like vertebrates, we have detached from the main stem one after another a long series of archaic or primitive types. We have first set off those with rudimentary notochord, then those with retrogressive development who lose the notochord, then those without skull or brain, then those without limbs or lower jaw. The residue assume the fish-like form of body, but still show great differences among themselves. We have then detached those without membrane-bones, or trace of lung or air-bladder. We next part company with those having the air-bladder a veritable lung, and those with an ancient type of paired fins, a jointed axis fringed with rays, and those having the palate still forming the upper jaw. We have finally left only those having fish-jaws, fish-fins, and in general the structure of the modern fish. For all these in all their variety, as a class or subclass, the name Actinopteri, or Actinopterygii, suggested by Professor Cope, is now generally adopted. The shorter form, Actinopteri, being equally correct is certainly preferable. This term (ακτίς, ray; πτερόν or πτερύξ, fin) refers to the structure of the paired fins. In all these fishes the bones supporting the fin-rays are highly specialized and at the same time concealed by the general integument of the body. In general two bones connect the pectoral fin with the shoulder-girdle. The hypercoracoid is a flat square bone, usually perforated by a foramen. Lying below it and parallel with it is the irregularly formed hypocoracoid. Attached to them is a row of bones, the actinosts, or pterygials, short, often hour-glass-shaped, which actually support the fin-rays. In the more specialized forms, or Teleosts, the actinosts are few (four to six) in number, but in the more primitive types, or Ganoids, they may remain numerous, a reminiscence of the condition seen in the Crossopterygians, and especially in Polypterus. Other variations may occur; the two coracoids sometimes are imperfect or specially modified, the upper sometimes without a foramen, and the actinosts may be distorted in form or position.

Fig. 1.—Shoulder-girdle of a Flounder, Paralichthys californicus (Ayres).

The Series Ganoidei.—Among the lower Actinopteri many archaic traits still persist, and in its earlier representatives the group approaches closely to the Crossopterygii, although no forms actually intermediate are known either living or fossil. The great group of Actinopteri may be divided into two series or subclasses, the Ganoidei, or Chrondrostei, containing those forms, mostly extinct, which retain archaic traits of one sort or another, and the Teleostei, or bony fishes, in which most of the primitive characters have disappeared. Doubtless all of the Teleostei are descended from a ganoid ancestry.

Even among the Ganoidei, as the term is here restricted, there remains a very great variety of form and structure. The fossil and existing forms do not form continuous series, but represent the tips and remains of many diverging branches perhaps from some Crossopterygian central stock. The group constitutes at least three distinct orders and, as a whole, does not admit of perfect definition. In most but not all of the species the tail is distinctly and obviously heterocercal, the lack of symmetry of the tail in some Teleosts being confined to the bones and not evident without dissection. Most of the Ganoids have the skeleton still cartilaginous, and in some it remains in a very primitive condition. Usually the Ganoids have an armature of bony plates, diamond-shaped, with an enamel like that developed on the teeth. In all of them the pectoral fin has numerous basal bones or actinosts. All of them have the air-bladder highly developed, usually cellular and functional as a lung, but connecting with the dorsal side of the gullet, not with the ventral side as in the Dipnoans. In all living forms there is a more or less perfect optic chiasma. These ancient forms retain also the many valves of the arterial bulb and the spiral valve of the intestines found in the more archaic types of fishes. But traces of some or all of these structures are found in some bony fishes, and their presence in the Ganoids by no means justifies the union of the Ganoids with the sharks, Dipnoans, and Crossopterygians to form a great primary class, Palæichthyes, as proposed by Dr. Günther. Almost every form of body may be found among the Ganoids. In the Mesozoic seas these fishes were scarcely less varied and perhaps scarcely less abundant than the Teleosts in the seas of to-day. They far exceed the Crossopterygians in number and variety of forms. Transitional forms connecting the two groups are thus far not recognized. So far as fossils show, the characteristic actinopterous fin with its reduced and altered basal bones appeared at once without intervening gradations.

The name Ganoidei (γάνος, brightness; εἶδος, resemblance), alluding to the enameled plates, was first given by Agassiz to those forms, mostly extinct, which were covered with bony scales or hard plates of one sort or another. As the term was originally defined, mailed catfishes, sea-horses, Agonidæ, Arthrodires, Ostracophores, and other wholly unrelated types were included with the garpikes and sturgeons as Ganoids. Most of these intruding forms among living fishes were eliminated by Johannes Müller, who recognized the various archaic characters common to the existing forms after the removal of the mailed Teleosts. Still later Huxley separated the Crossopterygians as a distinct group, while others have shown that the Ostracophori and Arthrodira should be placed far from the garpike in systematic classification. Cope, Woodward, Hay, and others have dropped the name Ganoid altogether as productive of confusion through the many meanings attached to it. Others have kept it as a convenient group name for the orders of archaic Actinopteri. For these varied and more or less divergent forms it seems convenient to retain it. As an adjective "ganoid" is sometimes used as descriptive of bony plates or enameled scales, some-in the sense of archaic, as applied to fishes.

Are the Ganoids a Natural Group?—Several writers have urged that the Ganoidei, even as thus restricted, should not be considered as a natural group, whether subclass, order, or group of orders. The reasons for this view in brief are the following:

1. The group is heterogeneous. The Amiidæ differ more from the other Ganoids than they do from the herring-like Teleosts. The garpikes, sturgeons, paddle-fishes likewise diverge widely from each other and from the Palæoniscidæ and the Platysomidæ. Each of the living families represents the residue or culmination of a long series, in some cases advancing, as in the case of the bowfin, sometimes perhaps degenerating, as in the case of the sturgeons.

2. Of the traits possessed in common by these forms, several (the cellular air-bladder, the many valves in the heart, the spiral valve in the intestine, the heterocercal tail) are all possessed in greater or less degree by certain Isospondyli or allies of the herring. All these characters are still better developed in Crossoptergyii and Dipneusti, and each one disappears by degrees. Of the characters drawn from the soft parts we can know nothing so far as the extinct Ganoids are concerned.

3. The optic chiasma, thus far characteristic of Ganoids as distinct from Teleosts, may have no great value. It is urged that in closely related species of lizards some have the optic chiasma and others do not. This, however, proves nothing as to the value of the same character among fishes.

4. The transition from Ganoids to Teleosts is of much the same character as the transition from spiny-rayed to soft-rayed fishes, or that from fishes with a duct to the air-bladder to those without such duct.

Admitting all this, it is nevertheless natural and convenient to retain the Ganoidei (or Chrondrostei if the older name be discarded on account of the many meanings attached to it) as a group equivalent to that of Teleostei within the class or subclass of Actinopteri. It comprises the transitional forms between the Crossopterygii and the bony fishes, and its members are especially characteristic of the Mesozoic age, ranging from the Devonian to the present era.

Of the extensive discussion relating to this important question we may quote two arguments for the retention of the subclass of Ganoids, the first by Francis M. Balfour and William Kitchen Parker, the second from the pen of Theodore Gill.

Balfour and Parker ("Structure and Development of Lepidosteus," pp. 430-433) thus discuss the

Systematic Position of Lepidosteus.—"Alexander Agassiz concludes his memoir on the development of Lepidosteus by pointing out that in spite of certain affinities in other directions this form is 'not so far removed from the bony fishes as has been supposed.' Our own observations go far to confirm Agassiz's opinion.

"Apart from the complete segmentation, the general development of Lepidosteus is strikingly Teleostean. In addition to the general Teleostean features of the embryo and larva, which can only be appreciated by those who have had an opportunity of practically working at the subject, we may point to the following developmental features[^[2]] as indicative of Teleostean affinities:

"(1) The formation of the nervous system as a solid keel of the epiblast.

"(2) The division of the epiblast into a nervous and epidermic stratum.

"(3) The mode of development of the gut.

"(4) The mode of development of the pronephros; though the pronephros of Lepidosteus has primitive characters not retained by Teleostei.

"(5) The early stages in the development of the vertebral column.

[2]. The features enumerated above are not in all cases confined to Lepidosteus and Teleostei, but are always eminently characteristic of the latter.

"In addition to these, so to speak, purely embryonic characters there are not a few important adult characters:

"(1) The continuity of the oviducts with the genital glands.

"(2) The small size of the pancreas, and the presence of numerous so-called pancreatic cæca.

"(3) The somewhat coiled small intestine.

"(4) Certain characters of the brain, e.g., the large size of the cerebellum; the presence of the so-called lobi inferiores on the infundibulum, and of tori semi-circulares in the mid-brain.

"In spite of the undoubtedly important list of features to which we have just called attention, a list containing not less important characters, both embryological and adult, separating Lepidosteus from the Teleostei, can be drawn up:

"(1) The character of the truncus arteriosus.

"(2) The fact of the genital ducts joining the ureters.

"(3) The presence of vasa efferentia in the male carrying the semen from the testes to the kidney, and through the tubules of the latter into the kidney-duct.

"(4) The presence of a well-developed opercular gill.

"(5) The presence of a spiral valve; though this character may possibly break down with the extension of our knowledge.

"(6) The typical Ganoid characters of the thalamencephalon and the cerebral hemispheres.

"(7) The chiasma of the optic nerves.

"(8) The absence of a pecten, and presence of a vascular membrane between the vitreous humor and the retina.

"(9) The opisthocœlous form of the vertebræ.

"(10) The articulation of the ventral parts of the hæmal arches of the tail with the processes of the vertebral column.

"(11) The absence of a division of the muscles into dorso-lateral and ventro-lateral divisions.

"(12) The complete segmentation of the ovum.

"The list just given appears to us sufficient to demonstrate that Lepidosteus cannot be classed with the Teleostei; and we hold that Müller's view is correct, according to which Lepidosteus is a true Ganoid.

"The existence of the Ganoids as a distinct group has, however, recently been challenged by so distinguished an ichthyologist as Günther, and it may therefore be well to consider how far the group as defined by Müller is a natural one for living forms, and how far recent researches enable us to improve upon Müller's definitions. In his classical memoir the characters of the Ganoids are thus shortly stated:

"'These fishes are either provided with plate-like angular or rounded cement-covered scales, or they bear osseous plates, or are quite naked. The fins are often, but not always, beset with a double or single row of spinous plates or splints. The caudal fin embraces occasionally in its upper lobe the end of the vertebral column, which may be prolonged to the end of the upper lobe. Their double nasal openings resemble those of Teleostei. The gills are free, and lie in a branchial cavity under an operculum, like those of Teleostei. Many of them have an accessory organ of respiration, in the form of an opercular gill, which is distinct from the pseudobranch, and can be present together with the latter; many also have spiracles like Elasmobranchii. They have many valves in the stem of the aorta like the latter, also a muscular coat in the stem of the aorta. Their ova are transported from the abdominal cavity by oviducts. Their optic nerves do not cross each other. The intestine is often provided with a spiral valve, like Elasmobranchii. They have a swimming-bladder with a duct, like many Teleostei. Their pelvic fins are abdominal.

"'If we include in a definition only those characters which are invariable, the Ganoids may be shortly defined as being those fish with numerous valves to the stem of the aorta, which is also provided with a muscular coat, with free gills, and an operculum, and with abdominal pelvic fins.'

"To these distinctive characters he adds, in an appendix to his paper, the presence of the spiral valve, and the absence of a processus falciformis and a choroid gland.

"To the distinctive set of characters given by Müller we may probably add the following:

"(1) Oviducts and urinary ducts always unite, and open by a common urogenital aperture behind the anus.

"(2) Skull hyostylic.

"(3) Segmentation complete in the types so far investigated, though perhaps Amia may be found to resemble the Teleostei in this particular.

"(4) A pronephros of the Teleostean type present in the larva.

"(5) Thalamencephalon very large and well developed.

"(6) The ventricle in the posterior part of the cerebrum is not divided behind into lateral halves, the roof of the undivided part being extremely thin.

"(7) Abdominal pores always present.

"The great number of characters just given are amply sufficient to differentiate the Ganoids as a group; but, curiously enough, the only characters, amongst the whole series which have been given, which can be regarded as peculiar to the Ganoids are (1) the characters of the brain, and (2) the fact of the oviducts and kidney-ducts uniting together and opening by a common pore to the exterior.

"This absence of characters peculiar to the Ganoids is an indication of how widely separated in organization are the different members of this great group.

"At the same time, the only group with which existing Ganoids have close affinities is the Teleostei. The points they have in common with the Elasmobranchii are merely such as are due to the fact that both retain numerous primitive vertebrate characters,[^[3]] and the gulf which really separates them is very wide.

[3]. As instances of this we may cite (1) the spiral valve; (2) the frequent presence of a spiracle; (3) the frequent presence of a communication between the pericardium and the body-cavity; (4) the heterocercal tail.

"There is again no indication of any close affinity between the Dipnoans and, at any rate, existing Ganoids.

"Like the Ganoids, the Dipnoans are no doubt remnants of a very primitive stock; but in the conversion of the air-bladder into a true lung, the highly specialized character of their limbs,[^[4]] their peculiar autostylic skulls, the fact of their ventral nasal openings leading directly into the mouth, their multi-segmented bars (interspinous bars) directly prolonged from the neural and hæmal and supporting the fin-rays of the unpaired dorsal and ventral fins, and their well-developed cerebral hemispheres, very unlike those of Ganoids and approaching the Amphibian type, they form a very well-defined group and one very distinctly separated from the Ganoids.

[4]. Vide F. M. Balfour, "On the Development of the Skeleton of the Paired Fins of Elasmobranchs," Proc. Zool. Soc., 1881.

"No doubt the Chondrostean Ganoids are nearly as far removed from the Teleostei as from the Dipnoans, but the links uniting these Ganoids with the Teleostei have been so fully preserved in the existing fauna of the globe that the two groups almost run into each other. If, in fact, we were anxious to make any radical change in the ordinary classification of fishes, it would be by uniting the Teleostei and Ganoids, or rather constituting the Teleostei into one of the subgroups of the Ganoids, equivalent to the Chondrostei. We do not recommend such an arrangement, which in view of the great preponderance of the Teleostei amongst living fishes would be highly inconvenient, but the step from Amia to the Teleostei is certainly not so great as that from the Chondrostei to Amia, and is undoubtedly less than that from the Selachii to the Holocephali."

Gill on the Ganoids as a Natural Group.—Dr. Gill observes ("Families of Fishes," 1872): "The name Ganoides (or Ganiolepedoti) was originally framed by Prof. Agassiz as an ordinal term for fishes having the scales (when present) angular and covered with enamel; and in the group so characterized were combined the Ganoids of subsequent authors as well as the Teleostean orders Plectognathi, Lophobranchii, and Nematognathi, and (subsequently) the genus Sudis (Arapaima), the last being regarded as a Cœlacanth. The group has not been accepted with these limits or characters.

"But the researches of Prof. Johannes Müller on the anatomy and classification of the fishes culminated at length in his celebrated memoirs on those fishes for which he retained the ordinal name Ganoidei; those memoirs have left an impression on ichthyology perhaps more decided than made by any other contributions to science, and that published in extenso will ever be classical; numerous as have been the modifications since introduced into the system, no forms except those recognized by Müller (unless it be Dipnoi) have been interjected since among the Ganoids.

"It has been objected that the Ganoids do not constitute a natural group, and that the characters (i.e., chiasma of optic nerves and multivalvular bulbus arteriosus) alleged by Müller to be peculiar to the teleostomous forms combined therein are problematical, and only inferentially supposed to be common to the extinct Ganoids so called, and, finally, such objections couched in too strong language have culminated in the assertion that the characters in question are actually shared by other physostome fishes.

"No demonstration, however, has been presented as yet that any physostome fishes do really have the optic chiasma and multivalvular bulbus arteriosus, and the statement to the contrary seems to have been the result of a venial misapprehension of Prof. Kner's statements, or the offspring of impressions left on the memory by his assertions, in forgetfulness of his exact words.

"But Prof. Kner, in respect to the anatomical characters referred to, merely objects: (1) that they are problematical, are not confirmable for the extinct types, and were probably not existent in certain forms that have been referred to the Ganoids; (2) the difference in number of the valves of the bulbus arteriosus among recent Ganoids is so great as to show the unreliability of the character; (3) a spiral valve is developed in the intestine of several osseous fishes ('genera of the so-called intermediate clupeoid groups'), as well as in Ganoids; and (4) the chiasma of the optic nerves in no wise furnishes a positive character for the Ganoids.

"It will be noticed that all these objections (save in the case of the intestinal spiral valve) are hypothetical and vague. The failure of the intestinal spiral valve, as a diagnostic character, has long been conceded, and in this case only have the forms that prove the failure been referred to; in the other cases, where it would be especially desirable to have indicated the actual types falsifying the universality or exclusiveness of the characters, they have not been referred to, and the objections must be met as if they were not known to exist.

"(1) The characters in question are, in the sense used, problematical, inasmuch as no examination can be made of the soft parts of extinct forms, but with equal force may it be urged that any characters that have not been or cannot be directly confirmed are problematical in the case of all other groups (e.g., mammals), and it can only be replied that the coordination of parts has been so invariably verified that all probabilities are in favor of similar coordination in any given case.

"(2) There is doubtless considerable difference in the number of valves of the bulbus arteriosus among the various Ganoids, and even among the species of a single family (e.g., Lepidosteidæ), but the character of Ganoids lies not in the number, more or less, but in the greater number and relations (in contradistinction to the opposite pair of the Teleosts) in conjunction with the development of a bulbus arteriosus. In no other forms of Teleostomes have similar relations and structures been yet demonstrated.

"(3) The failure of the spiral intestinal valve has already been conceded, and no great stress has ever been laid on the character.

"(4) The chiasma of the optic nerves is so common to all the known Ganoids, and has not been found in those forms (e.g., Arapaima, Osteoglossum, and Clupeiform types) agreeing with typical physostome Teleosts in the skeleton, heart, etc., but which at the same time simulate most certain Ganoids (e.g., Amia) in form.

"Therefore, in view of the evidence hitherto obtained, the arguments against the validity of title, to natural association, of the Ganoids, have to meet the positive evidence of the coordinations noted; the value of such characteristics and coordinations can only be affected or destroyed by the demonstration that in all other respects there is (1) very close agreement of certain of the constituents of the subclass with other forms, and (2) inversely proportionate dissimilarity of those forms from any (not all) other of the Ganoids, and consequently evidence ubi plurima nitent against the taxonomic value of the characters employed for distinction.

"And it is true that there is a greater superficial resemblance between the Hyoganoids (Lepisosteus, Amia, etc.) and ordinary physostome Teleosts than between the former and the other orders of Ganoids, but it is equally true that they agree in other respects than in the brain and heart with the more generalized Ganoids. They all have, for example, (1) the paraglenal elements undivided (not disintegrated into hypercoracoid, hypocoracoid, and mesocoracoid); (2) a humerus (simple or divided, that is, differentiated into metapterygium and mesopterygium); and (3) those with ossified skeletons agree in the greater number of elements in the lower jaw. Therefore, until these coordinates fail, it seems advisable to recognize the Ganoids as constituents of a natural series; and especially on account of the superior taxonomic value of modifications of the brain and heart in other classes of vertebrates, for the same reason, and to keep prominently before the mind the characters in question, it appears also advisable to designate the series, until further discovery, as a subclass.

"But it is quite possible that among some of the generalized Teleosts at least traces of some of the characters now considered to be peculiar to the Ganoids may be discovered. In anticipation of such a possibility, the author had at first discarded the subclass, recognizing the group only as one of the 'superorders' of the Teleostomes, but reconsideration convinces him of the propriety of classification representing known facts and legitimate inferences rather than too much anticipation.

"It is remembered that all characters are liable to fail with increasing knowledge, and the distinctness of groups are but little more than the expressions of our want of knowledge of the intermediate forms; it may in truth be said that ability to segregate a class into well-defined groups is in ratio to our ignorance of all the terms."

CHAPTER II
THE GANOIDS—Continued

Classification of Ganoids.—The subdivision of the series of Ganoidei into orders offers great difficulty from the fact of the varying relationships of the members of the group and the fact that the great majority of the species are known only from broken skeletons preserved in the rocks. It is apparently easy to separate those with cartilaginous skeletons from those with these bones more or less ossified. It is also easy to separate those with bony scales or plates from those having the scales cycloid. But the one type of skeleton grades into the other, and there is a bony basis even to the thinnest of scales found in this group. Among the multitude of names and divisions proposed we may recognize six orders, for which the names Lysopteri, Chondrostei, Selachostomi, Pycnodonti, Lepidostei, and Halecomorphi are not inappropriate. Each of these seems to represent a distinct offshoot from the first primitive group.

Order Lysopteri.—In the most primitive order, called Lysopteri (λυσός, loose; πτερόν, fin) by Cope, Heterocerci by Zittel and Eastman, and the "ascending series of Chondrostei" by Woodward, we find the nearest approach to the Chondropterygians. In this order the arches of the vertebræ are more or less ossified, the body is more or less short and deep, covered with bony dermal plates. The opercular apparatus is well developed, with numerous branchiostegals. Infraclavicles are present, and the fins provided with fulcra. Dorsal and anal fins are present, with rays more numerous than their supports; ventral fin with basal supports which are imperfectly ossified; caudal fin mostly heterocercal, the scales mostly rhombic in form. All the members of this group are now extinct.

The Palæoniscidæ.—The numerous genera of this order are referred to three families, the Palæoniscidæ, Platysomidæ, and Dictyopygidæ; a fourth family, Dorypteridæ, of uncertain relations, being also tentatively recognized. The family of Palæoniscidæ is the most primitive, ranging from the Devonian to the Lias, and some of them seem to have entered fresh waters in the time of the coal-measures. These fishes have the body elongate and provided with one short dorsal fin. The tail is heterocercal and the body covered with rhombic plates. Fulcra or rudimentary spine-like scales are developed on the upper edge of the caudal fin in most recent Ganoids, and often the back has a median row of undeveloped scales. A multitude of species and genera are recorded. A typical form is the genus Palæoniscum,[^[5]] with many species represented in the rocks of various parts of the world. The longest known species is Palæoniscum frieslebenense from the Permian of Germany and England. Palæoniscum magnum, sixteen inches long, occurs in the Permian of Germany. From Canobius, the most primitive genus, to Coccolepis, the most modern, is a continuous series, the suspensorium of the lower jaw becoming more oblique, the basal bones of the dorsal fewer, the dorsal extending farther forward, and the scales more completely imbricate. Other prominent genera are Amblypterus, Eurylepis, Cheirolepis, Rhadinichthys, Pygopterus, Elonichthys, Ærolepis, Gyrolepis, Myriolepis, Oxygnathus, Centrolepis, and Holurus.

[5]. This word is usually written Palæoniscus, but Blainville, its author (1818), chose the neuter form.

Fig. 2.—Palæoniscum frieslebenense Blainville. Family Palæoniscidæ. (After Zittel.)

The Platysomidæ.—The Platysomidæ are different in form, the body being deep and compressed, often diamond-shaped, with very long dorsal and anal fins. In other respects they are very similar to the Palæoniscidæ, the osteology being the same. The Palæoniscidæ were rapacious fishes with sharp teeth, the Platysomidæ less active, and, from the blunter teeth, probably feeding on small animals, as crabs and snails.

The rhombic enameled scales are highly specialized and held together as a coat of mail by peg-and-socket joints. The most extreme form is Platysomus, with the body very deep. Platysomus gibbosus and other species occur in the Permian rocks of Germany. Cheirodus is similar to Platysomus, but without ventral fins. Eurynotus, the most primitive genus, is remarkable for its large pectoral fins. Eurynotus crenatus occurs in the Subcarboniferous of Scotland. Other genera are Mesolepis, Globulodus, Wardichthys, and Cheirodopsis.

Fig. 3.—Eurynotus crenatus Agassiz, restored. Carboniferous. Family Platysomidæ. (After Traquair.)

Some of the Platysomidæ have the interneural spines projecting through the skin before the dorsal fin. This condition is found also in certain bony fishes allied to the Carangidæ.

The Dorypteridæ.—Dorypterus hoffmani, the type of the singular Palæozoic family of Dorypteridæ, with thoracic or sub-jugular many-rayed ventrals, is Stromateus-like to all appearance, with distinct resemblances to certain Scombroid forms, but with a heterocercal tail like a ganoid, imperfectly ossified back-bone, and other very archaic characters. The body is apparently scaleless, unlike the true Platysomidæ, in which the scales are highly developed. A second species, Dorypterus althausi, also from the German copper shales, has been described. This species has lower fins than Dorypterus hoffmani, but may be the adult of the same type. Dorypterus is regarded by Woodward as a specialized offshoot from the Platysomidæ. The many-rayed ventrals and the general form of the body and fins suggest affinity with the Lampridæ.

Fig. 4.—Dorypterus hoffmani Germar, restored. (After Hancock and Howse.)

Dictyopygidæ.—In the Dictyopygidæ (Catopteridæ), the body is gracefully elongate, less compressed, the heterocercal tail is short and abruptly turned upwards, the teeth are sharp and usually hooked, and the bony plates well developed. Of this group two genera are recognized, each containing numerous species. In Redfieldius (= Catopterus Redfield, not of Agassiz) the dorsal is inserted behind the anal, while in Dictyopyge this is not the case. Redfieldius gracilis and other species are found in the Triassic of the Connecticut River. Dictyopyge macrura is found in the same region, and Dictyopyge catoptera and other species in Europe.

Order Chondrostei.—The order Chondrostei (χόνδρος, cartilage; ὀστέον, bone), as accepted by Woodward, is characterized by the persistence of the notochord in greater or less degree, the endoskeleton remaining cartilaginous. In all, the axonosts and baseosts of the median fins are arranged in simple regular series and the rays are more numerous than the supporting elements. The shoulder-girdle has a pair of infraclavicular plates. The pelvic fins have well-developed baseosts. The branchiostegals are few or wanting. In the living forms, and probably in all others, a matter which can never be ascertained, the optic nerves are not decussating, but form an optic chiasma, and the intestine is provided with a spiral valve. In all the species there is one dorsal and one anal fin, separate from the caudal. The teeth are small or wanting, the body naked or covered with bony plates; the caudal fin is usually heterocercal, and on the tail are rhombic plates. To this order, as thus defined, about half of the extinct Ganoids belong, as well as the modern degenerate forms known as sturgeons and perhaps the paddle-fishes, which are apparently derived from fishes with rhombic enameled scales. The species extend from the Upper Carboniferous to the present time, being most numerous in the Triassic.

At this point in Woodward's system diverges a descending series, characterized as a whole by imperfect squamation and elongate form, this leading through the synthetic type of Chondrosteidæ to the modern sturgeon and paddle-fish, which are regarded as degenerate types.

The family of Saurorhynchidæ contains pike-like forms, with long jaws, and long conical teeth set wide apart. The tail is not heterocercal, but short-diphycercal; the bones of the head are covered with enamel, and those of the roof of the skull form a continuous shield. The opercular apparatus is much reduced, and there are no branchiostegals. The fins are all small, without fulcra, and the skin has isolated longitudinal series of bony scutes, but is not covered with continuous scales. The principal genus is Saurorhynchus (= Belonorhynchus; the former being the earlier name) from the Triassic. Saurorhynchus acutus from the English Triassic is the best known species.

The family of Chondrosteidæ includes the Triassic precursors of the sturgeons. The general form is that of the sturgeon, but the body is scaleless except on the upper caudal lobe, and there are no plates on the median line of the skull. The opercle and subopercle are present, the jaws are toothless, and there are a few well-developed caudal rays. The caudal has large fulcra. The single well-known species of this group, Chondrosteus acipenseroides, is found in the Triassic rocks of England and reaches a length of about three feet. It much resembles a modern sturgeon, though differing in several technical respects. Chondrosteus pachyurus is based on the tail of a species of much larger size and Gyrosteus mirabilis, also of the English Triassic, is known from fragments of fishes which must have been 18 to 20 feet in length.

Fig. 5.—Chondrosteus acipenseroides Egerton. Family Chondrosteidæ. (After Woodward.)

The sturgeons constitute the recent family of Acipenseridæ, characterized by the prolonged snout and toothless jaws and the presence of four barbels below the snout. In the Acipenseridæ there are no branchiostegals and a median series of plates is present on the head. The body is armed with five rows of large bony bucklers,—each often with a hooked spine, sharpest in the young. Besides these, rhombic plates are developed on the tail, besides large fulcra. The sturgeons are the youngest of the Ganoids, not occurring before the Lower Eocene, one species, Acipenser toliapicus occurring in the London clay. About thirty living species of sturgeon are known, referred to three genera: Acipenser, found throughout the Northern Hemisphere, Scaphirhynchus, in the Mississippi Valley, and Kessleria (later called Pseudoscaphirhynchus), in Central Asia alone. Most of the species belong to the genus Acipenser, which abounds in all the rivers and seas in which salmon are found. Some of the smaller species spend their lives in the rivers, ascending smaller streams to spawn. Other sturgeons are marine, ascending fresh waters only for a moderate distance in the spawning season. They range in length from 2½ to 30 feet.

All are used as food, although the flesh is rather coarse and beefy. From their large size and abundance they possess great economic value. The eggs of some species are prepared as caviar.

Fig. 6.—Common Sturgeon, Acipenser sturio Mitchill. Potomac River.

The sturgeons are sluggish, clumsy, bottom-feeding fish. The mouth, underneath the long snout, is very protractile, sucker-like, and without teeth. Before it on the under side of the snout are four long feelers. Ordinarily the sturgeon feeds on mud and snails with other small creatures, but I have seen large numbers of Eulachon (Thaleichthys) in the stomach of the Columbia River sturgeon (Acipenser transmontanus). This fish and the Eulachon run in the Columbia at the same time, and the sucker-mouth of a large sturgeon will draw into it numbers of small fishes who may be unsuspiciously engaged in depositing their spawn. In the spawning season in June these clumsy fishes will often leap wholly out of the water in their play. The sturgeons have a rough skin besides five series of bony plates which change much with age and which in very old examples are sometimes lost or absorbed in the skin. The common sturgeon of the Atlantic on both shores is Acipenser sturio. Acipenser huso and numerous other species are found in Russia and Siberia. The great sturgeon of the Columbia is Acipenser transmontanus, and the great sturgeon of Japan Acipenser kikuchii. Smaller species are found farther south, as in the Mediterranean and along the Carolina coast. Other small species abound in rivers and lakes. Acipenser rubicundus is found throughout the Great Lake region and the Mississippi Valley, never entering the sea. It is four to six feet long, and at Sandusky, Ohio, in one season 14,000 sturgeons were taken in the pound nets. A similar species, Acipenser mikadoi, is abundant and valuable in the streams of northern Japan.

Fig. 7.—Lake Sturgeon, Acipenser rubicundus Le Sueur. Ecorse, Mich.

In the genus Acipenser the snout is sharp and conical, and the shark-like spiracle is still retained.

Fig. 8.—Shovel-nosed Sturgeon. Scaphirhynchus platyrhynchus (Rafinesque). Ohio River.

The shovel-nosed sturgeon (Scaphirhynchus platyrhynchus) has lost the spiracles, the tail is more slender, its surface wholly bony, and the snout is broad and shaped like a shovel. The single species of Scaphirhynchus abounds in the Mississippi Valley, a fish more interesting to the naturalist than to the fisherman. It is the smallest of our sturgeons, often taken in the nets in large numbers.

In Scaphirhynchus the tail is covered by a continuous coat of mail. In Kessleria[^[6]] fedtschenkoi, rossikowi, and other Asiatic species the tail is not mailed.

[6]. These species have also been named Pseudoscaphirhynchus. Kessleria is the earlier name, left undefined by its describer, although the type was indicated.

Order Selachostomi: the Paddle-fishes.—Another type of Ganoids, allied to the sturgeons, perhaps still further degenerate, is that of the paddle-fishes, called by Cope Selachostomi (σέλαχος, shark; στόμα, mouth). This group consists of a single family, Polyodontidæ, having apparently little in common with the other Ganoids, and in appearance still more suggestive of the sharks. The common name of paddle-fishes is derived from the long flat blade in which the snout terminates. This extends far beyond the mouth, is more or less sensitive, and is used to stir up the mud in which are found the minute organisms on which the fish feeds. Under the paddle are four very minute barbels corresponding to those of the sturgeons. The vernacular names of spoonbill, duckbill cat, and shovel-fish are also derived from the form of the snout. The skin is nearly smooth, the tail is heterocercal, the teeth are very small, and a long fleshy flap covers the gill-opening. The very long and slender gill-rakers serve to strain the food (worms, leeches, water-beetles, crustaceans, and algæ) from the muddy waters from which they are taken. The most important part of this diet consists of Entomostracans. The single American species, Polyodon spathula, abounds through the Mississippi Valley in all the larger streams. It reaches a length of three or four feet. It is often taken in the nets, but the coarse tough flesh, like that of our inferior catfish, is not much esteemed. In the great rivers of China, the Yangtse and the Hoang Ho, is a second species, Psephurus gladius, with narrower snout, fewer gill-rakers, and much coarser fulcra on the tail. The habits, so far as known, are much the same.

Fig. 9.—Paddle-fish, Polyodon spathula (Walbaum). Ohio River.

Fig. 10.—Paddle-fish. Polyodon Spathula (Walbaum). Ohio River.

Fig. 11.—Psephurus gladius Günther. Yangtse River. (After Günther.)

Crossopholis magnicaudatus of the Green River Eocene shales is a primitive member of the Polyodontidæ. Its rostral blade is shorter than that of Polyodon, and the body is covered with small thin scales, each in the form of a small grooved disk with several posterior denticulations, arranged in oblique series but not in contact. The scales are quadrate in form, and more widely separated anteriorly than posteriorly. As in Polyodon, the teeth are minute and there are no branchiostegals. The squamation of this fish shows that Polyodon as well as Acipenser may have sprung from a type having rhombic scales. The tail of a Cretaceous fish, Pholidurus disjectus from the Cretaceous of Europe, has been referred with doubt to this family of Polyodontidæ.

Order Pycnodonti.—In the extinct order Pycnodonti, as recognized by Dr. O. P. Hay, the notochord is persistent and without ossification, the body is very deep, the teeth are always blunt, the opercular apparatus is reduced, the dorsal fin many-rayed, and the fins without fulcra. The scales are rhombic, but are sometimes wanting, at least on the tail. Many genera and species of Pycnodontidæ are described, mostly from Triassic and Jurassic rocks of Europe. Leading European genera are Pycnodus, Typodus (Mesodon), Gyrodus, and Palæobalistum. The numerous American species belong to Typodus, Cœlodus, Pycnodus, Hadrodus, and Uranoplosus. These forms have no affinity with Balistes, although there is some resemblance in appearance, which has suggested the name of Palæobalistum.

Fig. 12.—Gyrodus hexagonus Agassiz. Family Pycnodontidæ. Lithographic Shales.

Woodward places these fishes with the Semionotidæ and Halecomorphi in his suborder of Protospondyli. It seems preferable, however, to consider them as forming a distinct order.

Fig. 13.—Mesturus verrucosus Wagner. Family Pycnodontidæ. (After Woodward.)

Order Lepidostei.—We may place, following Eastman's edition of Zittel, the allies and predecessors of the garpike in a single order, for which Huxley's name Lepidostei may well be used. In this group the notochord is persistent, and the vertebræ are in various degrees of ossification and of different forms. The opercles are usually complete, the branchiostegals present, and there is often a gular plate. There is no infraclavicle and the jaws have sharp teeth. The fins have fulcra, and the supports of the fins agree in number with the rays. The tail is more or less heterocercal. The scales are rhombic, arranged in oblique series, which are often united above and below with peg-and-socket articulations. This group contains among recent fishes only the garpikes (Lepisosteus). They are closely allied to the Palæoniscidæ, but the skeleton is more highly ossified. On the other hand they approach very closely to the ancestors of the bowfin, Amia. One genus, Acentrophorus, appears in the Permian; the others are scattered through Mesozoic and Tertiary rocks, the isolated group of gars still persisting. In the gars the vertebræ are concavo-convex, with ball-and-socket joints. In the others the vertebræ are incomplete or else double-concave, as in fishes generally.

For the group here called Lepidostei numerous other names have been used corresponding wholly or in part. Rhomboganoidea of Gill covers nearly the same groups; Holostei of Müller and Hyoganoidea of Gill include the Halecomorphi also; Ginglymodi of Cope includes the garpikes only, while Ætheospondyli of Woodward includes the Aspidorhynchidæ and the garpikes.

Fig. 14.—Semionotus kapffi Fraas, restored. Family Semionotidæ. (After Fraas, per Nicholson.)

The Semionotidæ (Stylodontidæ) are robust-bodied Ganoids, having the vertebræ developed as rings, the jaws with several rows of teeth, those of the outer row styliform.

Semionotus bergeri is a well-known species, with the body moderately elongate. Semionotus agassizi and many other species occur in the Triassic of the Connecticut valley and in New Jersey. The body is very deep in the related genus Dapedium, and the head is covered with strong bony plates. Dapedium politum is a well-known species of the English Triassic. Tetragonolepis (Pleurolepis) is a similar form, very deep and compressed, with strong, firm scales.

In the extinct family of Lepidotidæ the teeth are conical or chisel-shaped, while blunt or molar teeth are on the inside of the mouth, which is small, and the suspensorium of the mandible is vertical or inclined forward. The body is robust-fusiform, covered with rhomboid scales; the vertebræ form rings about the notochord; the teeth are either sharp or blunt. The dorsal fin is short, with large fulcra.

The best known of the numerous genera are Lepidotes, rather elongate in body, with large, blunt teeth. Of the many species of Lepidotes, Lepidotes elvensis abounds in the English and German Triassic, and Lepidotes minor in the English Triassic. Another well-known European species is Lepidotes mantelli.

Fig. 15.—Dapedium politum Leach, restored. Family Semionotidæ. (After Woodward.)

The Isopholidæ (Eugnathidæ) differ from the families last named in the large pike-like mouth with strong teeth. The mandibular suspensorium is inclined backwards. The body is elongate, the vertebræ forming incomplete rings; the dorsal fin is short with large fulcra.

Isopholis dentosus is found with numerous other species in the British Triassic. Caturus furcatus is especially characteristic of Triassic rocks in Germany. Ptycholepis marshi occurs in the Connecticut valley.

Fig. 16.—Tetragonolepis semicinctus Brown. Lias. Family Semionotidæ. (After Woodward.)

The Macrosemiidæ are elongate fishes with long dorsal fin, the numerous species being found in the Triassic, Jurassic, and Cretaceous of Europe. Macrosemius rostratus has a very high, continuous dorsal. Macropistius arenatus is found in the Cretaceous of Texas, the only American species known. Prominent European genera are Notagogus, Ophiopsis, and Petalopteryx.

Fig. 17.—Isopholis orthostomus (Agassiz). Lias. (After Woodward.)

Fig. 18.—The Long-nosed Garpike, Lepisosteus osseus (Linnæus). Fox River, Wisconsin. (From nature; D. S. Jordan and M. L. McDonald, 1874.)

Intermediate between the allies of the gars and the modern herrings is the large extinct family of Pholidophoridæ, referred by Woodward to the Isospondyli, and by Eastman to the Lepidostei. These are small fishes, fusiform in shape, chiefly of the Triassic and Jurassic. The fins are fringed with fulcra, the scales are ganoid and rhombic, and the vertebræ reduced to rings. The mouth is large, with small teeth, and formed as in the Isospondyli. The caudal is scarcely heterocercal.

Fig. 19.—Caturus elongatus Agassiz. Jurassic. Family Isopholidæ. (After Zittel.)

Fig. 20.—Notagogus pentlandi Agassiz. Jurassic. Family Macrosemiidæ. (After Woodward.)

Fig. 21.—Ptycholepis curtus Egerton. Lias. Family Isopholidæ. (After Woodward.)

Of Pholidophorus, with scales joined by peg-and-socket joints and uniform in size, there are many species. Pholidophorus latiusculus and many others are found in the Triassic of England and the Continent. Pholidophorus americanus occurs in the Jurassic of South Dakota. Pleuropholis, with the scales on the lateral line, which runs very low, excessively deepened, is also widely distributed. I have before me a new species from the Cretaceous rocks near Los Angeles. The Archæomænidæ differ from Pholidophoridæ in having cycloid scales. In both families the vertebræ are reduced to rings about the notochord. From fishes allied to the Pholidophoridæ the earliest Isospondyli are probably descended.

Fig. 22.—Pholidophorus crenulatus Egerton. Lias. (After Woodward.)

In the Aspidorhynchidæ the snout is more or less produced, the mandible has a distinct presymphysial bone, the vertebræ are double-concave or ring-like, and the fins are without fulcra. This family constitutes the suborder Ætheospondyli. In form these fishes resemble Albula and other modern types, but have mailed heads and an ancient type of scales. Two genera are well known, Aspidorhynchus and Belonostomus. Aspidorhynchus acutirostris reaches a length of three feet, and is found in the Triassic lithographic stone of Bavaria. Other species occur in rocks of Germany and England.

Belonostomus has the snout scarcely produced. Belonostomus sphyrænoides is the best known of the numerous species, all of the Triassic, Jurassic, and Cretaceous.

Family Lepisosteidæ.—The family of Lepisosteidæ, constituting the suborder Ginglymodi (γιγγλυμός, hinge), is characterized especially by the form of the vertebræ.

These are opisthocœlian, convex in front and concave behind, as in reptiles, being connected by ball-and-socket joints. The tail is moderately heterocercal, less so than in the Halecomorphi, and the body is covered with very hard, diamond-shaped, enameled scales in structure similar to that of the teeth. A number of peculiar characters are shown by these fishes, some of them having often been regarded as reptilian traits. Notable features are the elongate, crocodile-like jaws, the upper the longer, and both armed with strong teeth. The mandible is without presymphysial bone. The fins are small with large fulcra, and the scales are nearly uniform in size.

All the species belong to a single family, Lepisosteidæ, which includes the modern garpikes and their immediate relatives, some of which occur in the early Tertiary. These voracious fishes are characterized by long and slender cylindrical bodies, with enameled scales and mailed heads and heterocercal tail. The teeth are sharp and unequal. The skeleton is well ossified, and the animal itself is extremely voracious. The vertebræ, reptile-like, are opisthocœlian, that is, convex in front, concave behind, forming ball-and-socket joints. In almost all other fishes they are amphicœlian or double-concave, the interspace filled with gelatinous substance. The recent species, and perhaps all the extinct species also, belong to the single genus Lepisosteus (more correctly, but also more recently, spelled Lepidosteus). Of existing forms there are not many species, three to five at the most, and they swarm in the lakes, bayous, and sluggish streams from Lake Champlain to Cuba and along the coast to Central America. The best known of the species is the long-nosed garpike, Lepisosteus osseus, which is found throughout most of the Great Lake region and the Mississippi Valley, and in which the long and slender jaws are much longer than the rest of the head. The garpike frequents quiet waters and is apparently of sleepy habit. It often lies quiet for a long time, carried around and around by the eddies. It does not readily take the hook and seldom feeds in the aquarium. It feeds on crayfishes and small fishes, to which it is exceedingly destructive, as its bad reputation indicates. Fishermen everywhere destroy it without mercy. Its flesh is rank and tough and unfit even for dogs.

In the young garpike the caudal fin appears as a second dorsal and anal, the filamentous tip of the tail passing through and beyond it.

The short-nosed garpike, Lepisosteus platystomus, is generally common throughout the Mississippi Valley. It has a short broad snout like the alligator-gar, but seldom exceeds three feet in length. In size, color, and habits it agrees closely with the common gar, differing only in the form of the snout. The form is subject to much variation, and it is possible that two or more species have been confounded.

Fig. 23.—Alligator-gar, Lepisosteus tristœchus (Bloch). Cuba.

The great alligator-gar, Lepisosteus tristœchus, reaches a length of twenty feet or more, and is a notable inhabitant of the streams about the Gulf of Mexico. Its snout is broad and relatively wide, and its teeth are very strong. It is very destructive to all sorts of food-fishes. Its flesh is worthless, and its enameled scales resist a spear or sometimes even shot. It breathes air to a certain extent by its lungs, but soon dies in foul water, not having the tenacity of life seen in Amia.

Embryology of the Garpike.—Mr. Alexander Agassiz has given an account of the embryology of the garpike, of which the following is an abstract:

"The garpike comes up the St. Lawrence in May, lays its eggs about the 20th, and then disappears. The eggs are large, viscous, stick fast in an isolated way to whatever they fall upon, and look much like those of toads, having a large outer membrane and a small yolk. Artificial fecundation failed, but about 500 naturally-laid eggs were secured, of which all but 30 perished through mold. The young began to hatch in six days. Out of 30 young hatched, 27 lived until the 15th of July. Connection with the sharks appears in the similarity of the branchial arches and by the presence of the lateral fold in which the pectoral fins are formed; the way the tail is developed is very like that of the bony fishes. Among the Ganoids it appears, as well as in ordinary fishes, the dorsal cord is straight at first, then assumes a slightly upward curve at the extremity, when finally there appears the beginning of a lobe underneath, pointing to a complete heterocercal tail. All this is as in the bony fishes, but this is the permanent condition of the garpike, while in the bony fishes the extremity of the dorsal cord becomes extinct. The mode of development of the pectoral lobe (very large in this species) furnishes another resemblance. In the brain, and in the mode of formation of the gills, a likeness to the sharks is noticeable. The young garpikes move very slowly, and seem to float quietly, save an exceedingly rapid vibration of the pectorals and the tip of the tail. They do not swim about much, but attach themselves to fixed objects by an extraordinary horseshoe-shaped ring of sucker-appendages about the mouth. These appendages remain even after the snout has become so extended that the ultimate shape is hinted at; and furthermore, it is a remnant of this feature that forms the fleshy bulb at the end of the snout in the adult. The investigations thus far show that the young garpike has many characteristics in common with the sharks and skates, but it is not so different from the bony fishes as has been supposed."

Fossil Garpikes.—A number of fossil garpikes, referred by Cope to the genus Clastes and by Eastman and Woodward to Lepidosteus, are found in the Eocene of Europe and America. The most perfect of these remains is called Lepisosteus atrox, upward of four feet long, as large as an alligator-gar, which the species much resembles. Although found in the Eocene, Dr. C. R. Eastman declares that "it has no positively archaic features. If we inquire into the more remote or pre-Eocene history of Lepidosteids, palæontology gives no answer. They blossom forth suddenly and fully differentiated at the dawn of the Tertiary, without the least clue to their ancestry, unheralded and unaccompanied by any intermediate forms, and they have remained essentially unchanged ever since."

Another fossil species is Lepisosteus fimbriatus, from the Upper Eocene of England. Scales and other fragments of garpikes are found in Germany, Belgium, and France, in Eocene and Miocene rocks. On some of these the nominal genera Naisia, Trichiurides, and Pneumatosteus are founded. Clastes, regarded by Eastman as fully identical with Lepisosteus, is said to have the "mandibular ramus without or with a reduced fissure of the dental foramen, and without the groove continuous with it in Lepisosteus. One series of large teeth, with small ones external to them on the dentary bone." Most of the fossil forms belong to Clastes, but the genus shows no difference of importance which will distinguish it from the ordinary garpike.

Fig. 24.—Lower jaw of Amia calva Linnæus, showing the gular plate.

Order Halecomorphi.—To this order belong the allies, living or extinct, of the bowfin (Amia), having for the most part cycloid scales and vertebræ approaching those of ordinary fishes. The resemblance to the Isospondyli, or herring group, is indicated in the name (Halec, a herring; μορφή, form). The notochord is persistent, the vertebræ variously ossified. The opercles are always complete. The branchiostegals are broad and there is always a gular plate. The teeth are pointed, usually strong. There is no infraclavicle. Fulcra are present or absent. The supports of the dorsal and anal are equal in number to the rays. Tail heterocercal. Scales thin, mostly cycloid, but bony at base, not jointed with each other. Mandible complex, with well-developed splenial rising into a coronoid process, which is completed by a distinct coronoid bone. Pectoral fin with more than five actinosts; scales ganoid or cycloid. In the living forms the air-bladder is connected with the œsophagus through life; optic chiasma present; intestine with a spiral valve. This group corresponds to the Amioidei of Lütken and essentially to the Cycloganoidei of Gill. The Protospondyli (προτός, before; σπόνδυλος, vertebra) of Woodward contains essentially the same elements.

Pachycormidæ.—In the family of Pachycormidæ the notochord is persistent, the ethmoids and vomer fused and projecting between the maxillaries to form the prominent snout, the teeth large, the body fusiform, the dorsal short, with slender rays and few fulcra or none, and the scales are thin and rhombic. The numerous species are characteristic of the Triassic, Jurassic, and Cretaceous. In Sauropsis (longimana) the body is elongate, and the pectoral fins are large and sickle-shaped. Euthynotus has small fulcra. In Pachycormus (macropterus, esocinus, etc.) the form is robust and the ventral fins are wanting. In Hypsycormus ventrals are present, and the caudal deeply forked.

In the American family of Protosphyrænidæ the jaws are armed with very strong teeth, as in the Barracuda, which, however, the species do not resemble in other respects. Protosphyræna nitida, perniciosa, and numerous other extinct forms, some of them of large size, were voracious inhabitants of the Cretaceous seas, and are found fossil, especially in North Carolina and Kansas. Numerous species called Erisichthe and Pelecopterus are all referred by Hay to Protosphyræna. In this family the scapula and coracoids are ossified, and perhaps the vertebræ also, and, as Dr. Hay has recently suggested, the Protosphyrænidæ may really belong to the Isospondyli. In any event, they stand on the border-line between the most fish-like of the Ganoids and the most archaic of the bony fishes.

The Liodesmidæ (genus Liodesmus) are much like Amia, but the notochord is persistent, its sheath without ossification. Liodesmus gracilis and L. sprattiformis occur in the lithographic stones of Bavaria. Woodward places Liodesmus with Megalurus among the Amiidæ.

The Bowfins: Amiidæ.—The Amiidæ have the vertebræ more complete. The dorsal fin is many-rayed and is without distinct fulcra. The diamond-shaped enameled scales disappear, giving place to cycloid scales, which gradually become thin and membranous in structure. A median gular plate is developed between the branchiostegals. The tail is moderately heterocercal, and the head covered with a bony coat of mail.

The family of Amiidæ contains a single recent species, Amia calva, the only living member of the order Halecomorphi. The bowfin, or grindle, is a remarkable fish abounding in the lakes and swamps of the Mississippi Valley, the Great Lake region, and southward to Virginia, where it is known by the imposing but unexplained title of John A. Grindle. In the Great Lakes it is usually called "dogfish," because even the dogs will not eat it, and "lawyer," because, according to Dr. Kirtland, "it will bite at anything and is good for nothing when caught."

The bowfin reaches a length of two and one half feet, the male being smaller than the female and marked by an ocellated black spot on the tail. Both sexes are dark mottled green in color. The flesh of the species is very watery, pasty, much of the substance evaporating when exposed to the air. It is ill-flavored, and is not often used as food. The species is very voracious and extremely tenacious of life. Its well-developed lung enables it to breathe even when out of the water, and it will live in the air longer than any other fish of American waters, longer even than the horned pout (Ameiurus) or the mud-minnow (Umbra). As a game fish the grindle is one of the very best, if the angler does not care for the flesh of what he catches, it being one of the hardest fighters that ever took the hook.

Fig. 25.—Bowfin (female), Amia calva Linnæus. Lake Michigan.

The Amiidæ retain many of the Ganoid characters, though approaching more nearly than any other of the Ganoids to the modern herring tribe. For this reason the name Halecomorphi (shad-formed) was given to this order by Professor Cope. The gular plate found in Amia and other Ganoids reappears in the herring-like family of Elopidæ, which includes the tarpon and the ten-pounder.

Woodward unites the extinct genera called Cyclurus, Notæus, Amiopsis, Protamia, Hypamia, and Pappichthys with Amia. Pappichthys (corsoni, etc.), from the Wyoming Eocene, is doubtless a valid genus, having but one row of teeth in each jaw, and Amiopsis is also recognized by Hay. Woodward refers to Amia the following extinct species: Amia valenciennesi, from the Miocene of France; Amia macrocephala, from the Miocene of Bohemia; and Amia ignota, from the Eocene of Paris. Other species of Amia are known from fragments. Several of these are from the Eocene of Wyoming and Colorado. Some of them have a much shorter dorsal fin than that of Amia calva and may be generically different.

Fig. 26.—Megalurus elegantissimus Wagner. Family Amiidæ. (After Zittel.)

The genus Megalurus differs from Amia in the still shorter dorsal fin, less than one-third the length of the back. The body is elongate and much depressed. Megalurus lepidotus and several other species are found in the lithographic stones of Bavaria and elsewhere.

The Oligopleuridæ.—In the extinct family Oligopleuridæ the scales are cycloid, the bones of the head scarcely enameled, and the vertebræ well ossified. Fulcra are present, and the mouth is large, with small teeth. The genera are Oligopleurus, Ionoscopus, and Spathiurus, the species not very numerous and chiefly of the Cretaceous. Ionoscopus cyprinoides of the lithographic shales of Bavaria is a characteristic species.

From the three families last named, with the Pholidophoridæ, there is an almost perfect transition from the Ganoid fishes to teleosteans of the order of Isospondyli, the primitive order from which all other bony fishes are perhaps descended. The family of Leptolepidæ, differing from Oligopleuridæ in the absence of fulcra, is here placed with the Isospondyli, but it might about as well be regarded as Ganoid.

CHAPTER III
ISOSPONDYLI

The Subclass Teleostei, or Bony Fishes.—The fishes which still remain for discussion constitute the great subclass or series of Teleostei (τελεός, true; οστέον, bone), or bony fishes. They lack wholly or partly the Ganoid traits, or show them only in the embryo. The tail is slightly, if at all, heterocercal; the actinosts of the pectoral fins are few and large, rarely over five in number, except among the eels; the fulcra disappear; the air-bladder is no longer cellular, except in very rare cases, nor does it assist in respiration. The optic nerves are separate, one running to each eye without crossing; the skeleton is almost entirely bony, the notochord usually disappearing entirely with age; the valves in the arterial bulb are reduced in number, and the spiral valve of the intestines disappears. Traces of each of the Ganoid traits may persist somewhere in some group, but as a whole we see a distinct specialization and a distinct movement toward the fish type, with the loss of characters distinctive of sharks, Dipnoans, and Ganoids. In a general way the skeleton of all Teleosts corresponds with that of the striped bass (see Figs. 22, 23, Vol. I), and the visceral anatomy is in all cases sufficiently like that of the sunfish (Fig. 16, Vol. I).

The mesocoracoid or precoracoid arch, found in all Ganoids, persists in the less specialized types of bony fishes, although no trace of it is found in the perch-like forms. With all this, there is developed among the bony fishes an infinite variety in details of structure. For this reason the Teleostei must be broken into many orders, and these orders are very different in value and in degrees of distinctness, the various groups being joined by numerous and puzzling intergradations.

Order Isospondyli.—Of the various subordinate groups of bony fishes, there can be no question as to which is most primitive in structure, or as to which stands nearest the orders of Ganoids. Earliest of the bony fishes in geological time is the order of Isospondyli (ἴσος, equal; σπόνδυλος, vertebra), containing the allies, recent or fossil, of the herring and the trout. This order contains those soft-rayed fishes in which the ventral fins are abdominal, a mesocoracoid or precoracoid arch is developed, and the anterior vertebræ are unmodified and essentially similar to the others. The orbitosphenoid is present in all typical forms. In certain forms of doubtful affinity (Iniomi) the mesocoracoid is wanting or lost in degeneration. Through the Isospondyli all the families of fishes yet to be considered are apparently descended, their ancestors being Ganoid fishes and, still farther back, the Crossopterygians.

Woodward gives this definition of the Isospondyli: "Notochord varying in persistence, the vertebral centra usually complete, but none coalesced; tail homocercal, but hæmal supports not much expanded or fused. Symplectic bone present, mandible simple, each dentary consisting only of two elements (dentary and articulo-angular), with rare rudiments of a splenoid on the inner side. Pectoral arch suspended from the cranium; precoracoid (mesocoracoid) arch present; infraclavicular plates wanting. Pelvic (ventral) fins abdominal. Scales ganoid only in the less specialized families. In the living forms air-bladder connected with the œsophagus in the adult; optic nerves decussating (without chiasma), and intestine either wanting spiral valve or with an incomplete representative of it."

The Classification of the Bony Fishes.—The classification of fishes has been greatly complicated by the variety of names applied to groups which are substantially but not quite identical one with another. The difference in these schemes of classification lies in the point of view. In all cases a single character must be brought to the front; such characters never stand quite alone, and to lay emphasis on another character is to make an alteration large or small in the name or in the boundaries of a class or order. Thus the Ostariophysi with the Isospondyli, Haplomi, and a few minor groups make up the great division of the Abdominales. These are fishes in which the ventral fins are abdominal, that is, inserted backward, so that the pelvis is free from the clavicle, the two sets of limbs being attached to different parts of the skeleton. Most of the abdominal fishes are also soft-rayed fishes, that is, without consecutive spines in the dorsal and anal fins, and they show a number of other archaic peculiarities. The Malacopterygians (μαλακός, soft; πτερύξ, fin) of Cuvier therefore correspond very nearly to the Abdominales. But they are not quite the same, as the spiny-rayed barracudas and mullets have abdominal ventrals, and many unquestioned thoracic or jugular fishes, as the sea-snails and brotulids, have lost, through degeneration, all of their fin-spines.

In nearly but not quite all of the Abdominal fishes the slender tube connecting the air-bladder with the œsophagus persists through life. This character defines Müller's order of Physostomi (φυσός, bladder; στόμα, mouth), as opposed to his Physoclysti (φυσός, bladder; κλεῖστός, closed), in which this tube is present in the embryo or larva only. Thus the Thoracices and Jugulares, or fishes having the ventrals thoracic or jugular, together correspond almost exactly to the Acanthopterygians, (ακανθα, spine; πτερύξ, fin), or spiny-rayed fishes of Cuvier, or to the Physoclysti of Müller. The Malacopterygians, the Abdominales, and the Physostomi are in the same way practically identical groups. As the spiny-rayed fishes have mostly ctenoid scales, and the soft-rayed fishes cycloid scales, the Physostomi correspond roughly to Agassiz's Cycloidei, and the Physoclysti to his Ctenoidei.

But in none of these cases is the correspondence perfectly exact, and in any system of classification we must choose characters for primary divisions so ancient and therefore so permanent as to leave no room for exceptions. The extraordinary difficulty of doing this, with the presence of most puzzling intergradations, has led Dr. Gill to suggest that the great body of bony fishes, soft-rayed and spiny-rayed, abdominal, thoracic, and jugular alike, be placed in a single great order which he calls Teleocephali (τελεός, perfect; κεφαλή, head). The aberrant forms with defective skull and membrane-bones he would separate as minor offshoots from this great mass with the name of separate orders. But while the divisions of Teleocephali are not strongly differentiated, their distinctive characters are real, ancient, and important, while those of the aberrant groups, called orders by Gill (as Plectognathi, Pediculati, Hemibranchii), are relatively modern and superficial, which is one reason why they are more easily defined. There seems to us no special advantage in the retention of a central order Teleocephali, from which the divergent branches are separated as distinct orders.

While our knowledge of the osteology and embryology of most of the families of fishes is very incomplete, it is evident that the relationships of the groups cannot be shown in any linear series or by any conceivable arrangement of orders and suborders. The living teleost fishes have sprung from many lines of descent, their relationships are extremely diverse, and their differences are of every possible degree of value. The ordinary schemes have magnified the value of a few common characters, at the same time neglecting other differences of equal value. No system of arrangement which throws these fishes into large groups can ever be definite or permanent.

Relationships of Isospondyli.—For our purposes we may divide the physostomous fishes as understood by Müller into several orders, the most primitive, the most generalized, and economically the most important being the order of Isospondyli. This order contains those bony fishes which have the anterior vertebræ unaltered (as distinguished from the Ostariophysi), the skull relatively complete, or at least not eel-like, the mesocoracoid typically developed, but atrophied in deep-sea forms and finally lost, the orbitosphenoid present. In all the species the ventral fins are abdominal and normally composed of more than six rays; the air-duct is developed. The scales are chiefly cycloid and the fins are without true spines. In many ways the order is more primitive than Nematognathi, Plectospondyli, or Apodes. It is certain that it began earlier in geological time than any of these. On the other hand, the Isospondyli are closely connected through the Berycoidei with the highly specialized fishes. The continuity of the natural series is therefore interrupted by the interposition of the side branches of Ostariophysans and eels before considering the Haplomi and the other transitional forms. The forms called Iniomi, which lack the mesocoracoid and the orbitosphenoid, have been lately transferred to the Haplomi by Boulenger. This arrangement is probably a step in advance.

Ganoid traits are present in certain families of Isospondyli. Among these are the gular plate (found in Amia and the Elopidæ), doubtless derived from the similar structure in earlier Ganoids; additional valves in the arterial bulb in the cellular air-bladder of Notopterus and Osteoglossum, the spiral intestinal valve in Chirocentridæ, and the ganoid scales of the extinct Leptolepidæ.

The Clupeoidea.—The Isospondyli are divisible into numerous families, which may be grouped roughly under three subdivisions, Clupeoidea, the herring-like forms; the Salmonoidea, the trout-like forms; and the Iniomi, or lantern-fishes, and their allies. The last-named group should probably be removed from the order of Isospondyli. In the Clupeoidea, the allies of the great family of the herring, the shoulder-girdle is normally developed, retaining the mesocoracoid arch on its inner edge, and through the post-temporal is articulated above with the cranium. The fishes in this group lack the adipose fin which is characteristic of most of the higher or salmon-like families.

Fig. 27.—Leptolepis dubius Blainville, Lithographic Stone. (After Woodward.)

The Leptolepidæ.—Most primitive of the Isospondyli is the extinct family of Leptolepidæ, closely allied to the Ganoid families of Pholidophoridæ and Oligopleuridæ. It is composed of graceful, herring-like fishes, with the bones of the head thin but covered with enamel, and the scales thin but firm and enameled on their free portion. There are no fulcra and there is no lateral line. The vertebræ are well developed, but always pierced by the notochord. The genera are Lycoptera, Leptolepis, Æthalion, and Thrissops. In Lycoptera of the Jurassic of China the vertebral centra are feebly developed, and the dorsal fin short and posterior. In Leptolepis the anal is short and placed behind the dorsal. There are many species, mostly from the Triassic and lithographic shales of Europe, one being found in the Cretaceous. Leptolepis coryphænoides and Leptolepis dubius are among the more common species. Æthalion (knorri) differs in the form of the jaws. In Thrissops the anal fin is long and opposite the dorsal. Thrissops salmonea is found in the lithographic stone; Thrissops exigua in the Cretaceous. In all these early forms there is a hard casque over the brain-cavity, as in the living types, Amia and Osteoglossum.

Fig. 28.—Ten-pounder, Elops saurus L. An ally of the earliest bony fishes. Virginia.

Fig. 29.—A primitive Herring-like fish, Holcolepis lewesiensis, Mantell, restored. Family Elopidæ. English Chalk. (After Woodward.)

The Elopidæ.—The family of Elopidæ contains large fishes herring-like in form and structure, but having a flat membrane-bone or gular plate between the branches of the lower jaw, as in the Ganoid genus Amia. The living species are few, abounding in the tropical seas, important for their size and numbers, though not valued as food-fishes save to those who, like the Hawaiians and Japanese, eat fishes raw. These people prefer for that purpose the white-meated or soft-fleshed forms like Elops or Scarus to those which yield a better flavor when cooked.

The ten-pounder (Elops saurus), pike-like in form but with very weak teeth, is found in tropical America. Elops machnata, the jack mariddle, the awaawa of the Hawaiians, abounding in the Pacific, is scarcely if at all different.

Fig. 30.—Tarpon or Grande Écaille, Tarpon atlanticus Cuv. & Val. Florida.

The tarpon, called also grande écaille, silver-king, and sable (Tarpon atlanticus), is a favorite game-fish along the coasts of Florida and Carolina. It takes the hook with great spirit, and as it reaches a length of six feet or more it affords much excitement to the successful angler. The very large scales are much used in ornamental work.

A similar species of smaller size, also with the last ray of the dorsal very much produced, is Megalops cyprinoides of the East Indies. Other species occur in the South Seas.

Numerous fossil genera related to Elops are found in the Cretaceous and Tertiary rocks. Holcolepis lewesiensis (wrongly called Osmeroides) is the best-known European species. Numerous species are referred to Elopopsis. Megalops prisca and species of Elops also occur in the London Eocene.

In all these the large parietals meet along the median line of the skull. In the closely related family of Spaniodontidæ the parietals are small and do not meet. All the species of this group, united by Woodward with the Elopidæ, are extinct. These fishes preceded the Elopidæ in the Cretaceous period. Leading genera are Thrissopater and Spaniodon, the latter armed with large teeth. Spaniodon blondeli is from the Cretaceous of Mount Lebanon. Many other species are found in the European and American Cretaceous rocks, but are known from imperfect specimens only.

Sardinius, an American Cretaceous fossil herring, may stand near Spaniodon. Rhacolepis buccalis and Notelops brama are found in Brazil, beautifully preserved in concretions of calcareous mud supposed to be of Cretaceous age.

The extinct family of Pachyrhizodontidæ is perhaps allied to the Elopidæ. Numerous species of Pachyrhizodus are found in the Cretaceous of southern England and of Kansas.

The Albulidæ.—The Albulidæ, or lady-fishes, characterized by the blunt and rounded teeth, are found in most warm seas. Albula vulpes is a brilliantly silvery fish, little valued as food. The metamorphosis (see Fig. 112, Vol. I) which the larva undergoes is very remarkable. It is probably, however, more or less typical of the changes which take place with soft-rayed fishes generally, though more strongly marked in Albula and in certain eels than in most related forms. Fossils allied to Albula, Albula oweni, Chanoides macropomus, are found in the Eocene of Europe; Syntegmodus altus in the Cretaceous of Kansas. In Chanoides, the most primitive genus, the teeth are much fewer than in Albula. Plethodus and Thryptodus, with peculiar dental plates on the roof and floor of the mouth, probably constitute a distinct family, Thryptodontidæ. The species are found in European and American rocks, but are known from imperfect specimens only.

Fig. 31.—The Lady-fish, Albula vulpes (Linnæus). Florida.

The Chanidæ.—The Chanidæ, or milkfishes, constitute another small archaic type, found in the tropical Pacific. They are large, brilliantly silvery, toothless fishes, looking like enormous dace, swift in the water, and very abundant in the Gulf of California, Polynesia, and India. The single living species is the Awa, or milkfish, Chanos chanos, largely used as food in Hawaii. Species of Prochanos and Chanos occur in the Cretaceous, Eocene, and Miocene. Allied to Chanos is the Cretaceous genus Ancylostylos (gibbus), probably the type of a distinct family, toothless and with many-rayed dorsal.

Fig. 32.—Milkfish, Chanos chanos (L.). Mazatlan.

Fig. 33.—Mooneye, Hiodon tergisus Le Sueur. Ecorse, Mich.

The Hiodontidæ.—The Hiodontidæ, or mooneyes, inhabit the rivers of the central portion of the United States and Canada. They are shad-like fishes with brilliantly silvery scales and very strong sharp teeth, those on the tongue especially long. They are very handsome fishes and take the hook with spirit, but the flesh is rather tasteless and full of small bones, much like that of the milkfish. The commonest species is Hiodon tergisus. No fossil Hiodontidæ are known.

Fig. 34.—Istieus grandis Agassiz. Family Pterothrissidæ. (After Zittel.)

Fig. 35.—Chirothrix libanicus Pictet & Humbert. Cretaceous of Mt. Lebanon. (After Woodward.)

The Pterothrissidæ.—The Pterothrissidæ are sea-fishes like Albula, but more slender and with a long dorsal fin. They live in deep or cold waters along the coasts of Japan, where they are known as gisu. The single species is Pterothrissus gissu. The fossil genus Istieus, from the Upper Cretaceous, probably belongs near the Pterothrissidæ. Istieus grandis is the best-known species. Another ancient family, now represented by a single species, is that of the Chirocentridæ, of which the living type is Chirocentrus dorab, a long, slender, much compressed herring-like fish, with a saw-edge on the belly, found in the East Indies, in which region Chirocentrus polyodon occurs as a fossil. Numerous fossil genera related to Chirocentrus are enumerated by Woodward, most of them to be referred to the related family of Ichthyodectidæ (Saurodontidæ). Of these, Portheus, Ichthyodectes, Saurocephalus (Saurodon), and Gillicus are represented by numerous species, some of them fishes of immense size and great voracity. Portheus molossus, found in the Cretaceous of Nebraska, is remarkable for its very strong teeth. Species of other genera are represented by numerous species in the Cretaceous of both the Rocky Mountain region and of Europe.

Fig. 36.—Gigantic skeleton of Portheus molossus Cope. (Photograph by Charles H. Sternberg.)

The Ctenothrissidæ.—A related family, Ctenothrissidæ, is represented solely by extinct Cretaceous species. In this group the body is robust with large scales, ctenoid in Ctenothrissa, cycloid in Aulolepis. The fins are large, the belly not serrated, and the teeth feeble. Ctenothrissa vexillifera is from Mount Lebanon. Other species occur in the European chalk. In the small family of Phractolæmidæ the interopercle, according to Boulenger, is enormously developed.

Fig. 37.—Ctenothrissa vexillifera Pictet, restored. Mt. Lebanon Cretaceous. (After Woodward.)

The Notopteridæ.—The Notopteridæ is another small family in the rivers of Africa and the East Indies. The body ends in a long and tapering fin, and, as usual in fishes which swim by body undulations, the ventral fins are lost. The belly is doubly serrate. The air-bladder is highly complex in structure, being divided into several compartments and terminating in two horns anteriorly and posteriorly, the anterior horns being in direct communication with the auditory organ. A fossil Notopterus, N. primævus, is found in the same region.

The Clupeidæ.—The great herring family, or Clupeidæ, comprises fishes with oblong or herring-shaped body, cycloid scales, and feeble dentition. From related families it is separated by the absence of lateral line and the division of the maxillary into three pieces. In most of the genera the belly ends in a serrated edge, though in the true herring this is not very evident, and in some the belly has a blunt edge. Some of the species live in rivers, some ascend from the sea for the purpose of spawning. The majority are confined to the ocean. Among all the genera, the one most abundant in individuals is that of Clupea, the herring. Throughout the North Atlantic are immense schools of Clupea harengus. In the North Pacific on both shores another herring, Clupea pallasi, is equally abundant, and with the same market it would be equally valuable. As salted, dried, or smoked fish the herring is found throughout the civilized world, and its spawning and feeding-grounds have determined the location of cities.

Fig. 38.—Herring, Clupea harengus L. New York.

The genus Clupea, of northern distribution, has the vertebræ in increased number (56), and there are weak teeth on the vomer. Several other genera are very closely related, but ranging farther south they have, with other characters, fewer (46 to 50) vertebræ. The alewife, or branch-herring (Pomolobus pseudoharengus), ascends the rivers to spawn and has become landlocked in the lakes of New York. The skipjack of the Gulf of Mexico, Pomolobus chrysochloris, becomes very fat in the sea. The species becomes landlocked in the Ohio River, where it thrives as to numbers, but remains lean and almost useless as food. The glut-herring, Pomolobus æstivalis, and the sprat, Pomolobus sprattus, of Europe are related forms.

Fig. 39.—Alewife, Pomolobus pseudoharengus (Wilson). Potomac River.

Very near also to the herring is the shad (Alosa sapidissima) of the eastern coasts of America, and its inferior relatives, the shad of the Gulf of Mexico (Alosa alabamæ), the Ohio River shad (Alosa ohiensis), very lately discovered, the Allice shad (Alosa alosa) of Europe, and the Thwaite shad (Alosa finta). In the genus Alosa the cheek region is very deep, giving the head a form different from that seen in the herring.

The American shad is the best food-fish in the family, peculiarly delicate in flavor when broiled, but, to a greater degree than occurs in any other good food-fish, its flesh is crowded with small bones. The shad has been successfully introduced into the waters of California, where it abounds from Puget Sound to Point Concepcion, ascending the rivers to spawn in May as in its native region, the Atlantic coast.

The genus Sardinella includes species of rich flesh and feeble skeleton, excellent when broiled, when they may be eaten bones and all. This condition favors their preservation in oil as "sardines." All the species are alike excellent for this purpose. The sardine of Europe is the Sardinella pilchardus, known in England as the pilchard. The "Sardina de España" of Cuba is Sardinella pseudohispanica, the sardine of California, Sardinella cærulea. Sardinella sagax abounds in Chile, and Sardinella melanosticta is the valued sardine of Japan.

In the tropical Pacific occur other valued species, largely belonging to the genus Kowala. The genus Harengula contains small species with very large, firm scales which do not fall when touched, as is generally the case with the sardines. Most common of these is Harengula sardina of the West Indies. Similar species occur in southern Europe and in Japan.

Fig. 40.—Menhaden, Brevoortia tyrannus (Latrobe). Wood's Hole, Mass.

In Opisthonema, the thread-herring, the last dorsal ray is much produced, as in the gizzard-shad and the tarpon. The two species known are abundant, but of little commercial importance. Of greater value are the menhaden, or the moss-bunker, Brevoortia tyrannus, inhabiting the sandy coasts from New England southward. It is a coarse and bony fish, rarely eaten when adult, although the young in oil makes acceptable sardines. It is used chiefly for oil, the annual yield exceeding in value that of whale-oil. The refuse is used as manure, a purpose for which the fishes are often taken without preparation, being carried directly to the cornfields. From its abundance this species of inferior flesh exceeds in commercial value almost all other American fishes excepting the cod, the herring, and the quinnat salmon.

One of the most complete of fish biographies is that of Dr. G. Brown Goode on the "Natural and Economic History of Menhaden."

Numerous other herring-like forms, usually with compressed bodies, dry and bony flesh, and serrated bellies, abound in the tropics and are largely salted and dried by the Chinese. Among these are Ilisha elongata of the Chinese coast. Related forms occur in Mexico and Brazil.

The round herrings, small herrings which have no serrations on the belly, are referred by Dr. Gill to the family of Dussumieriidæ. These are mostly small tropical fishes used as food or bait. One of these, the Kobini-Iwashi of Japan (Stolephorus japonicus), with a very bright silver band on the side, has considerable commercial importance. Very small herrings of this type in the West Indies constitute the genus Jenkinsia, named for Dr. Oliver P. Jenkins, the first to study seriously the fishes of Hawaii. Other species constitute the widely distributed genera Etrumeus and Dussumieria. Etrumeus sardina is the round herring of the Virginia coast. Etrumeus micropus is the Etrumei-Iwashi of Japan and Hawaii.

Fig. 41.—A fossil Herring, Diplomystus humilis Leidy. (From a specimen obtained at Green River, Wyo.) The scutes along the back lost in the specimen. Family Clupeidæ.

Fossil herring are plentiful and exist in considerable variety, even among the Clupeidæ as at present restricted. Histiothrissa of the Cretaceous seems to be allied to Dussumieria and Stolephorus. Another genus, from the Cretaceous of Palestine, Pseudoberyx (syriacus, etc.), having pectinated scales, should perhaps constitute a distinct subfamily, but the general structure is like that of the herring. More evidently herring-like is Scombroclupea (macrophthalma). The genus Diplomystus, with enlarged scales along the back, is abundantly represented in the Eocene shales of Green River, Wyoming. Species of similar appearance, usually but wrongly referred to the same genus, occur on the coasts of Peru, Chile, and New South Wales. A specimen of Diplomystus humilis from Green River is here figured. Numerous herring, referred to Clupea, but belonging rather to Pomolobus, or other non-Arctic genera, have been described from the Eocene and later rocks.

Several American fossil herring-like fishes, of the genus Leptosomus, as Leptosomus percrassus, are found in the Cretaceous of South Dakota.

Fossil species doubtfully referred to Dorosoma, but perhaps allied rather to the thread-herring (Opisthonema), being herrings with a prolonged dorsal ray, are recorded from the early Tertiary of Europe. Among these is Opisthonema doljeanum from Austria.

Fig. 42.—Hickory-shad, Dorosoma cepedianum (Le Sueur). Potomac River.

The Dorosomatidæ.—The gizzard-shad, Dorosomatidæ, are closely related to the Clupeidæ, differing in the small contracted toothless mouth and reduced maxillary. The species are deep-bodied, shad-like fishes of the rivers and estuaries of eastern America and eastern Asia. They feed on mud, and the stomach is thickened and muscular like that of a fowl. As the stomach has the size and form of a hickory-nut, the common American species is often called hickory-shad. The gizzard-shad are all very poor food-fish, bony and little valued, the flesh full of small bones. The belly is always serrated. In three of the four genera of Dorosomatidæ the last dorsal ray is much produced and whip-like. The long and slender gill-rakers serve as strainers for the mud in which these fishes find their vegetable and animal food. Dorosoma cepedianum, the common hickory-shad or gizzard-shad, is found in brackish river-mouths and ponds from Long Island to Texas, and throughout the Mississippi Valley in all the large rivers. Through the canals it has entered Lake Michigan. The Konoshiro, Clupanodon thrissa, is equally common in China and Japan.

The Engraulididæ.—The anchovies (Engraulididæ) are dwarf herrings with the snout projecting beyond the very wide mouth. They are small in size and weak in muscle, found in all warm seas, and making a large part of the food of the larger fish. The genus Engraulis includes the anchovy of Europe, Engraulis encrasicholus, with similar species in California, Chile, Japan, and Australia. In this genus the vertebræ are numerous, the bones feeble, and the flesh tender and oily. The species of Engraulis are preserved in oil, often with spices, or are made into fish-paste, which is valued as a relish. The genus Anchovia replaces Engraulis in the tropics. The vertebræ are fewer, the bones firm and stiff, and the flesh generally dry. Except as food for larger fish, these have little value, although existing in immense schools. Most of the species have a bright silvery band along the side. The most familiar of the very numerous species is the silver anchovy, Anchovia browni, which abounds in sandy bays from Cape Cod to Brazil. Several other genera occur farther southward, as well as in Asia, but Engraulis only is found in Europe. Fossil anchovies called Engraulis are recorded from the Tertiary of Europe.

Fig. 43.—A Silver Anchovy, Anchovia perthecata (Goode & Bean). Tampa.

Fig. 44.—Notogoneus osculus Cope. Green River Eocene. Family Gonorhynchidæ.