A Guide to the Study of Fishes, Volume 1 (of 2)

by David Starr Jordan

FOOTNOTES:

[152] This group was first called by Cope Ostracodermi—a name preoccupied for the group of bony trunkfishes, Ostracidæ. The still earlier name of Placodermi, chosen by McCoy (1848), was intended to include Arthrodires as well as Ostracophores. Rohon (1892) calls the group Protocephali, and to the two orders he assigns the names Aspidorhini and Aspidocephali. These groups correspond to Heterostraci and Osteostraci of Woodward. Another name of early date is that of Aspidoganoidei, given by Professor Gill in 1876, but not defined until 1896. These fishes are, however, not "Ganoids" and the name Ostracophori seems to receive general preference. The group Peltacephalata of Patten corresponds essentially to Ostracophori, as does also the order Hypostomata of Gadow.

[153] According to Professor Patten's view, the close resemblance of the shields of Pteraspis to those of contemporaneous Eurypterids indicates real affinity. But the Eurypterids are related to the spiders and to Limulus. The only reason for thinking that Pteraspis is a fish at all lies in its resemblance to Cephalaspis, which is in several ways fish-like, although its head shield is much like that of Limulus. All these resemblances in Patten's view indicate real affinity. Patten considers the Pteraspids as derived from primitive arachnid or spider-like forms having a bony carapace as Limulus has. From Pteraspis he derives the other Ostracophores, and from these the sharks and other vertebrates, all of which appear later in time than the earliest Ostracophores. This view of the origin of vertebrates is recently urged with much force by Professor Patten (Amer. Nat., 1904, 1827). Most naturalists regard such resemblances in specialized structures on the outside of an animal as parallelisms due to likeness in conditions of life. The external structure in forms of really different nature is often similarly modified. Thus certain catfishes, pipefishes, sea-moths, and agonoid fishes are all provided with bony plates not unlike those of ganoid fishes, although indicative of no real affinity with them. Commonly the ancestry of vertebrates is traced through enteropneustans to soft-bodied worms which have left no trace in the rocks.

In the same connection, Professor Patten suggests that the lateral fold from which many writers have supposed that the limbs or paired fins of vertebrates is evolved is itself a resultant of the fusion of the fringing appendages on the sides of the body. Such appendages are found in the primitive mailed arachnoids and in Limulus. They are shown very plainly in Patten's restoration of Cephalaspis. About thirty of them of a bony nature and jointed to the body occur on either side between the gill opening and the vent.

[154] Called Cœlolepidæ by Pander and Traquair, but Cœlolepis is a later synonym of Thelodus.

[155] This name, inappropriate or meaningless, is older than Tremataspis.

[156] The earlier name of Pterichthys has been already used for a genus of living fishes.

---

[CHAPTER XXXIII]
ARTHRODIRES

The Arthrodires.—Another large group of extinct fishes mailed and helmeted is included under the general name of Arthrodira[157] (ἄρθρος, joint; δεῖρα, neck), or Arthrognathi (ἄρθρος, γνάθος, jaw), the latter term recently framed by Dr. Dean with a somewhat broader application than the former.

These fishes differ from the Ostracophores, on the one hand, in the possession of jaws and in the nature of their armored covering. On the other hand, the nature of these jaws, the lack of differentiation of the skeleton, and the uncertain character of the limbs separate them still more widely from the true fishes. Their place in the system is still unknown, but their origin seems as likely to be traceable to Ostracophores as to any other group.

The head in all the species is covered with a great bony helmet. Behind this on the nape is another large shield, and between the two is usually a huge joint which Dr. Dean compares to the hinge of a spring-beetle (Elater).

As to the presence of limbs, no trace of pectoral fin or anterior limb has been found. Dean denies the existence of any structures corresponding to either limb, but Woodward figures a supposititious posterior limb in Coccosteus, finding traces of basal bones which may belong to it.

These monstrous creatures have been considered by Woodward and others as mailed Dipnoans, but their singular jaws are quite unlike those of the Dipneusti, and very remote from any structures seen in the ordinary fish. The turtle-like mandibles seem to be formed of dermal elements, in which there lies little homology to the jaws of a fish and not much more with the jaws of Dipnoan or shark.

The relations with the Ostracophores are certainly remote, though nothing else seems to be any nearer. They have no affinity with the true Ganoids, to which vaguely limited group many writers have attached them. Nor is there any sure foundation to the view adopted by Woodward, that they are to be considered as armored offshoots of the Dipnoans.

According to Dean we might as well refer the Arthrodires to the sharks as to the Dipnoans. Dean further observes ("Fishes Living and Fossil"):

Fig. 366.—Coccosteus cuspidatus Agassiz, restored. Lower Devonian. (After Traquair, per Woodward.)

"The puzzling characters of the Arthrodirans do not seem to be lessened by a more definite knowledge of their different forms. The tendency, as already noted, seems to be at present to regard the group provisionally as a widely modified offshoot of the primitive Dipnoans, basing this view upon their general structural characters, dermal plates, dentition, autostylism. But only in the latter regard could they have differed more widely from the primitive Elasmobranch or Teleostome, if it be admitted that in the matter of dermal structures they may be clearly separated from the Chimæroid. It certainly is difficult to believe that the articulation of the head of Arthrodirans could have been evolved after dermal bones had come to be formed, or that a Dipnoan could become so metamorphosed as to lose not only its body armoring, but its pectoral appendages as well. The size of the pectoral girdle is, of course, little proof that an anterior pair of fins must have existed, since this may well have been evolved in relation to the muscular supports of plastron, carapace, trunk, and head. The intermovement of the dental plates, seen especially in Dinichthys, is a further difficulty in accepting their direct descent from the Dipnoans."

Fig. 367.—Jaws of Dinichthys hertzeri Newberry. Upper Devonian. Ohio. (After Newberry.)

Occurrence of Arthrodires.—These fishes occur in abundance from the Silurian times to the Mesozoic. In the Devonian their gigantic size and thick armor gave them the leading position among the hosts of the sea. Among the genera there occurred "series of forms most interesting as to their evolution." "It is found more and more evident," says Dr. Dean ("Fishes, Living and Fossil," pp. 135, 136) "that the Arthrodirans may have represented the dominant group in the Devonian period, as were the sharks in the Carboniferous, or as are the Teleosts in modern times. There were forms which, like Coccosteus, had eyes at the notches of the head-buckler; others, like Macropetalichthys, in which orbits were well centralized; some, like Dinichthys and Titanichthys, with the pineal foramen present; some with pectoral spines(?); some with elaborately sculptured dermal plates. Among their forms appear to have been those whose shape was apparently subcylindrical, adapted for swift swimming; others (Mylostoma) whose trunk was depressed to almost ray-like proportions. In size they varied from that of the perch to that of a basking shark. In dentition they presented the widest range in variation, from the formidable shear-like jaws of Dinichthys to the lip-like mandibles of Titanichthys, the tearing teeth of Trachosteus, the wonderfully forked tooth-bearing jaw-tips of Diplognathus, to the Cestraciont type, Mylostoma. The latter form has hitherto been known only from its dentition, but now proves to be, as Newberry and Smith Woodward suggested, a typical Arthrodiran."

Classification of Arthrodira.—Our knowledge of the systematic relations of the Arthrodira is mostly of recent origin. Woodward refers most of the remains to the best known genus Coccosteus, and recognizes as families the Coccosteidæ, Mylostomidæ, Asterosteidæ, and Phyllolepidæ.

Fig. 368.—An Arthrodire, Dinichthys intermedius Newberry, restored. Devonian, Ohio. (After Dean.)

Dr. Bashford Dean in different papers has treated these fishes in great detail. In a recent paper on the "Relationships of the Arthrognathi"[158] he recognizes the group as a class coordinate with Cyclostomi and Elasmobranchii. This class, which he calls Arthrognathi, is first divided into two suborders, Anarthrodira, without joint at the neck, and Arthrodira, with such a joint. The former comprises one order, Stegothalami, and the latter two orders, Temnothoraci and Arthrothoraci. The following is Dr. Dean's definition of these orders and their component families:

Arthrognathi.—"Chordates whose anterior body region is encased in dermal elements, and divisible by a more or less definite partition into head and trunk. Dermal plates which surround the mouth function as jaws. No evidence of branchial arches. Column notochordal, showing no traces of centra; well-marked neural and hæmal elements. Paired limbs [absent or uncertain]. Dermal plates consisting typically of two layers, the superficial tuberculate, the inner bony with radiating lamellæ. Orbits situated near or at the margin of the head-shield and separated from one another by fixed integumental plates. A pineal funnel present situated in a fixed plate. A mucous system whose canals radiate from the preoccipital region."

Anarthrodira.—"Arthrognaths in which the cranial and dorsal regions are separated by a fixed partition whose dorsal rim is overlapped and concealed by superficial plates. Of these a large median dorsal element is present which extends backward superficially from the region near the pineal funnel. Also a pair of elements which overlie the position of the external occipital joint. Suborbital plates apparently absent. Jaw elements undescribed."

Stegothalami (στέγος, roof; θάλαμος, chamber).—"Anarthrodires in which the cranio-dorsal septum is vertical and deep, its height equal apparently to that of the arch of the head-shield. By this deep partition the latter appears to inclose two chambers (whence the ordinal name). Orbits inclosed by pre- and postorbital plates. Mucous system lacks a postorbital canal."

One family, the Macropetalichthyidæ, thus defined:

"Stegothalami with large orbits and well-arched cranio-dorsal shield. Dorso-central shield long, wide, gomphoidal, extending backward to the hinder margin of the shield and bordered by all plates save the postorbitals and marginals. Pineal funnel small and obscure." Macropetalichthys sullivanti from Ohio Devonian rocks, and Macropetalichthys agassizi from the Devonian of Germany, are important species of this group.

The Asterosteidæ perhaps constitute a second family in this order. The single species Asterosteus stenocephalus is from the Devonian of Ohio.

Arthrodira.—"Arthrognaths in which the dorsal armoring is separated into dorsal and cranial elements, the latter attached to the former movably by means of a pair of peg-and-socket joints. The interval lying between cranial and dorsal armoring does not appear to have been protected by plates, and in the median line, instead of the cranio-central of the Anarthrodires, there are separate elements, median occipital, median dorsal, and perhaps others. Suborbital plates present. Jaws of three pairs of elements. Ventral armoring of two pairs of lateral and two median elements."

Temnothoraci (τέμνω, to cut; θώραξ, thorax).—"Arthrodires whose cranial and dorsal shields are closely apposed, separated only by a transverse fissure-like interval (whence the ordinal name); interarticulation of cranial and dorsal shields little developed. Head-shield elliptical in outline as far as the line of the transverse division. The anterior rim of the shoulder-shield flattened at its sides, suggesting a rudiment of the vertical partition of the Anarthrodira. Suborbital plate is present, but takes no part, apparently, in the ventral boundary of the orbit, this being formed, as in the Anarthrodira, by the pre- and postorbital elements. Jaws, ventral armoring, and endoskeleton not definitely known."

One family, Chelonichthyidæ, thus defined:

"Temnothoraci with orbits relatively small in size and situated well forward in the head-shield. Occipital elements produced antero-posteriorly, the external occipital forming the posterior lateral angle of the head, no projection of the head occurring in the region of the marginal plate. Median occipital trapezoidal. Centrals take part in the median boundary of the orbits, and embrace the pineal plate. Median dorsal with poorly developed keel and terminal process."

Heterosteus asmussi (perhaps to be called Ichthyosauroides spinosus) is a gigantic species from the Lower Devonian of Livonia.

Allied to this species is Homostius milleri from Scotland, celebrated as the "Asterolepis of Stromness" in Hugh Miller's "Footsteps of the Creator." Another notable species is Homostius formosissimus from the Lower Devonian of Russia.

Arthrothoraci.—"Arthrodires whose dorsal shield articulates with the head-roof by a conspicuous and movable peg-and-socket joint, and leaves a definite interval (unprotected?) between the two armorings. Orbits marginal, bounded inferiorly not by the suborbital element. In the head-shield the postero-lateral angles formed by the marginal plate (Phlyctænaspis?), the occipital border concave. A dorsal fin is present, supported by endoskeletal elements." Five families, the most important being the Coccosteidæ, thus defined:

"Arthrothoraci with head-shield hexagonal in outline. Median occipital trapezoidal, margins underlapped conspicuously by the external occipitals. Prefrontals meet below pineal plates, thus occluding this element from contact with centrals. The median dorsal plate elongated, terminating in an acute heavy point; no definite ventral keel; its anterior border approaches the head-shield more closely than in related families. Cranio-dorsal joint relatively small. Postero-dorso-lateral large." (?A pair of spines occurs in the pectoral region.) The best-known species is Coccosteus cuspidatus (decipiens) of the Lower Red Sandstone or Devonian of Scotland.

The family of Dinichthyidæ consists of "Arthrothoraci with stout trenchant jaws, whose cutting surfaces have worn away marginal teeth. Plates heavy. Head-shield with conspicuous lateral indentation to form dorsal border of orbit. Preorbitals separated by rostral and pineal elements, the latter passing backward between the anterior ends of the centrals. Cranio-dorsal joint conspicuous. Median dorsal shovel-shaped, nearing a stout keel with a large neck and with heavy gouge-shaped terminal. Postero-dorso-lateral relatively small in size." Dinichthys hertzeri and numerous other species are described from the Devonian and Carboniferous rocks of Ohio.

The Titanichthyidæ are "Arthrothoraci with slender edentulous jaws bearing a longitudinal sulcus. Plates squamous. Head-shield wide, with indentations to form dorsal border of orbit. Cranio-dorsal joint complete, but of relatively small size. Median dorsal with lateral border indented with rudimentary keel and with flat and rounded terminal. Antero-dorso-lateral with an area of overlap on median border." Titanichthys agassizi is a gigantic mailed fish from the Lower Carboniferous of Cleveland, Ohio.

The Mylostomidæ are "Arthrothoraci with dental elements in the character of crushing plates. Cranial shield wide, rounded anteriorly, deeply indented in nuchal margin; orbital rim not apparent in dorsal aspect. Central separated from marginal." Mylostoma terrelli is based on jaws from Cleveland, Ohio.

The Selenosteidæ are "Arthrothoraci with jaws studded with cuspidate teeth; the mandibular rami rounding out anteriorly or presenting diverging tips, bearing teeth in the symphysis. Cranial shield deeply concave on lateral margins, no orbital rim here apparent. Nuchal border deeply indented. (Centrals separate from marginals.) Cranio-dorsal hinges large in size. Dorsal armoring reduced antero-posteriorly, giving an almost zone-like appearance. Dorso-median crescent-shaped, with feeble keel and knob." Selenosteus glaber is described by Dean from the Cleveland shales.

Relations of Arthrodires.—To complete our account of the Arthrodira we may here summarize Dr. Dean's reasons for separating its members from true fishes on the one hand and from the Ostracophores on the other.

"First. The Arthrodira cannot be strictly included among the Pisces. According to the definition of the latter class its members are Craniotes possessing the following characters: a, dermal defenses which in their simplest terms can be reduced to the shagreen denticles of the Elasmobranch; b, a series of definite gill-arches whose foremost elements are metamorphosed into hyoid and mandibular apparatus; c, paired fins, or their equivalents. In the first of these regards I think it can be shown that the remarkable character of the dermal plates in the Arthrognaths approaches rather that of the Ostracophores than that of the Pisces. In certain of these forms, Trachosteus, for example, the tuberculated plates are made up of inner and outer elements, each with tubercles, which denote a distinctly different mode of origin from that of any known type of fish. The absence of remains of gill-arches in the Arthrognaths would be not a serious objection to including these forms among Pisces, especially in view of the fact that cartilaginous gill-arches are rarely preserved even in favorable fossils. But that their presence is more than doubtful is indicated by the peculiar character of the 'jaws' in these forms. For the character of these structures is such as to suggest that they are not homologous with the branchial arch jaws of the true fishes, but are rather parallel structures which owe their origin to distinctly exoskeletal elements, i.e., that they were derived from dermal plates surrounding the mouth, which became mobile, and whose edges became apposed as sectorial structures. I would in this connection call attention to the fact that the 'mandibles,' 'premaxillary,' and 'maxillary' dental plates[159] were not fixed in the sense in which these elements are in the true Pisces. On the evidence of several types, Dinichthys, Titanichthys, Mylostoma, Trachosteus, Diplognathus, and other of the American forms, Macropetalichthys[160] excepted, there is the clearest proof that each element of the jaws had a considerable amount of independent movement. On account of the mobility of these elements the name Arthrognathi is suggested. Thus the mandibular rami could change the angle of inclination towards each other, as well as their plane with reference to the vertical axis. So, too, could the 'premaxillæ' be protracted like a pair of bent fingers, and it is more than probable that the 'maxillæ' had a considerable amount of independent movement. In connection with these characters it is also important to note that the blades of the 'mandible' show nowhere the faintest trace of an articular facet for attachment to the cranium. In short, the entire plan of the mandibular apparatus in these forms is strikingly unfish-like, although one will frankly confess that it is remarkable that these forms should have paralleled so strikingly the piscine conditions, to the extent of producing mandibular rami margined with teeth, and an arrangement of toothed elements on the 'upper jaw' which resembles superficially the premaxillary and maxillary structures of teleostomes, or the vomero-palatine structures of lung-fishes and chimæras.

"In the matter of paired fins there seems little evidence to conclude that either pectoral or pelvic fins were present. In spite of the researches upon these forms during the past half-century, no definite remains of pectoral fins have been described. The so-called pectoral spines described for Dinichthys by Newberry, whatever they may be, certainly are not, as far as the present evidence goes, pterygial, nor are the similar structures in Brachydirus.[161] The sigmoid element, described as a 'pelvic girdle' by Smith Woodward, in Coccosteus, a structure which appears to occur in a small species of Dinichthys(?), may as reasonably be interpreted as a displaced element of the armor-plates of the trunk. In Coccosteus, as far as I am aware, it occurs in well-preserved condition in but a single specimen.

"In referring to the singular joint between the shoulder-plates and the hinder margin of the cranium Smith Woodward has called attention to one of the striking features of the group. It is one, however, which, as a functional structure, i.e., a joint, characterizes only a portion of its members; and in these the region in which vestiges of the joint are sought is overlaid and concealed by dermal plates. Such are the conditions in Macropetalichthys (with transitional characters in Trachosteus and in Mylostoma). For this form a special subclass (or order) may be created which we may term Anarthrodira.

"Second. The Arthrognathi cannot well be included in any other class. It would certainly be more convenient to retain the Arthrognaths among the Ostracophores, regarding them as a fourth subclass, were it not that they differ from them in so marked a way in the presence of well-marked vertebral arches, of supports for the unpaired fin, and in the possession of 'jaws.' In these regards—add to them the (probable if not certain) absence of the paired paddle-like 'spines'—they stand certainly further from the Antiarcha than these from the Osteostraci, or than the latter from the Heterostraci. It appears to me desirable, therefore, that the Arthrodira and the Anarthrodira be brought together as a separate class. Should subsequent researches demonstrate a closer affinity with the Ostracophores, the Arthrognathi can be regarded as of rank as a subclass, with the orders Anarthrodira and Arthrodira."[162]

In a recent paper Dr. Otto Jaekel unites Arthrodires and Ostracophores under the name Placodermi. He regards Pteraspis as a larval type, Asterolepis as one more specialized. In Coccosteus he claims to find a pelvic girdle as well as a more segmented skeleton. He regards all of these as true fishes, the Coccosteidæ as ancestral, related on the one hand to the Crossopterygians, and on the other to the Stegocephali and other ancestral Amphibians.

Suborder Cycliæ.—We may append to the Arthrodira as a possible suborder the group called Cycliæ by Dr. Gill, based on a single imperfectly known species. Few organisms discovered in recent times have excited as much interest as this minute fish-like creature, called Palæospondylus gunni, discovered in 1890 by Dr. R. H. Traquair in the flagstones of Caithness in Scotland. Many specimens have been obtained, none more than an inch and a half long. Its structure and systematic position have been discussed by Dr. R. H. Traquair, by Woodward, Gill, Gegenbaur, and recently by Dean, from whose valuable memoir on "The Devonian Lamprey" we make several quotations.

Fig. 369.—Palæospondylus gunni Traquair. Devonian. (After Traquair and Dean.)

Palæospondylus.—According to Dr. Traquair: "The Palæospondylus gunni is a very small organism, usually under one inch in length, though exceptionally large specimens occasionally measure one inch and a half.... It has a head and vertebral column, but no trace of jaws or limbs; and, strange to say, all the specimens are seen only from the ventral aspect, as is shown by the relation of the neural arches to the vertebral centra.

"The head is in most cases much eroded.... It is divided by a notch ... into two parts.... The anterior part shows a groove the edges of which are elevated, while the surface on each side shows two depressions, like fenestræ, though perhaps they are not completely perforated, and also a groove partially divided off, posteriorly and externally, a small lobe. In front there is a ring-like opening ... surrounded by small pointed cirri, four ventrally, at least five dorsally, and two long lateral ones which seem to arise inside the margin of the ring instead of from its rim like the others. The posterior part of the cranium is flattened, but the median groove is still observable. Connected with the posterior or occipital aspect of the skull are two small narrow plates which lie closely alongside the first half-dozen vertebræ.

"The bodies of the vertebræ are hollow or ring-like, and those immediately in front are separated from each other by perceptible intervals; their surfaces are marked with a few little longitudinal grooves, of which one is median. They are provided with neural arches, which are at first short and quadrate, but towards the caudal extremity lengthen out into slender neural spines, which form the dorsal expansion of a caudal fin, while shorter hæmal ones are also developed on the ventral aspect."

Dr. Traquair concludes that "there seems to be no escape from the conclusion that the little creature must be classed as a Marsipobranch." "If Palæospondylus is not a Marsipobranch, it is quite impossible to refer it to any other existing group of vertebrates."

Gill on Palæospondylus.—In 1896 Dr. Gill proposed to regard Palæospondylus provisionally as the type of a distinct order of Cyclostomes to be called Cycliæ (κύκλος, circle), from the median ring on the head, whether nostril or mouth. Dr. Gill observes:

"Assuming the correctness of Dr. Traquair's description and figures, we certainly have a remarkable combination of characters. On the one hand, if the 'median opening or rim' is indeed nasal, the animal certainly cannot be referred to the class of Selachians or of Teleostomes. On the other hand, the cranium and the segmental vertebral column indicate a more advanced stage of development of the vertebrate line than that from the living Marsipobranchs must have originated. We may, therefore, with propriety isolate it as the representative not only of a peculiar family (Palæospondylidæ), but of an order or even subclass (Cycliæ) of vertebrates which may provisionally (and only provisionally) be retained in the class of Marsipobranchs.

"The group may be defined as Monorrhines with a continuous (?) cranium, a median nasal (?) ring, and a segmented vertebral column.

"The differences between the Hyperoartia and Hyperotreta are very great, and Prof. Lankester did not go much too far when he elevated those groups to class rank. Among the numerous distinctive characters are the great differences in the auditory organs. Perhaps the organs of Palæospondylus might be worked out in some specimen and throw light on the subject of affinities. At present even the region of the auditory organs is not exactly known and we are now at a loss to orient the several parts of the cranium. In fact, the question of the relations of Palæospondylus is a very open one."

Views as to the Relationships of Palæospondylus.—Dr. Dean thus summarizes in a convenient and interesting fashion the views of different students of fossil fishes in regard to Palæospondylus:

Huxley.—A "baby Coccosteus."

Traquair, 1890.—"Certainly not a Placoderm, its resemblance to a supposed 'baby Coccosteus' being entirely deceptive. The appearance of the head does remind us in a strange way of the primitive skull of Myxine, a resemblance which is rendered still more suggestive by the apparent complete absence of the lower jaw, or of limbs or limb-girdles."

Traquair, 1893.—"It seems, indeed, impossible to refer the organism to any existing vertebrate class, unless it be the Marsipobranchs or Cyclostomata." Does not believe it a larval form, because the possible adult is unknown, and because of the highly differentiated vertebræ. Granting his interpretation of the parts of the fossil, "there seems no escape from the conclusion that the little creature must be classed as a Marsipobranch."

Traquair, 1897.—"The question of the affinities of Palæospondylus is left precisely where it was after I had written my last paper on the subject."

Smith Woodward, 1892.—"It seems to possess an unpaired nose, lip cartilages in place of functional jaws, and no paired limbs; thus agreeing precisely with the lampreys and hagfishes, of which the fossil representatives have long been sought. It is extremely probable, therefore, that Palæospondylus belongs to this interesting category."

Dawson, 1893.—Palæospondylus suggests "the smaller snake-like Batrachians of the Carboniferous and Permian; and I should not be surprised if it should come to be regarded as either a forerunner of the Batrachians or as a primitive tadpole."

Gill, 1896.—"The group to which Palæospondylus belongs may be defined as Monorrhines with a continuous (?) cranium, a median nasal (?) ring, and a segmented vertebral column." "The cranium and segmented vertebral column indicate a more advanced stage of development of the vertebrate line than that from which the living Marsipobranchs must have originated. We may, therefore, with propriety isolate it as the representative not only of a peculiar family (Palæospondylidæ), but of an order or even subclass (Cycliæ) of vertebrates which may provisionally (and only provisionally) be retained in the class of Marsipobranchs."

Dean, 1896.—"Place it with the Ostracoderms among the curiously specialized offshoots of the early Chordates, but this position would be at the best unsatisfactory."

Dean, 1898.—"Palæospondylus should not be given a place—even a provisional one—among the Marsipobranchs." To be accepted "as the representative of the new subclass (or class) Cycliæ constituted for it by Professor Gill."

Parker & Haswell, 1897.—"There is some reason to regard that Palæospondylus is referable to the Cyclostomes." "A distinctly higher type than recent forms."

Gegenbaur, 1898.—"Discovery of Palæospondylus one of the highest importance. If this organism stands in no way near the Cyclostomes, the tentacles lose their higher importance, since they also occur in other groups." "Through Palæospondylus came also the attempt (Pollard) to deduce the presence of the tentacular condition in the higher forms." (Mem.—In this Gegenbaur has not consulted the literature accurately. At the time of founding his "Cirrhostomal Theory" Pollard was unaware of the discovery of Palæospondylus). "Ich muss sagen, das die positive Behauptung der einen wie der anderen Deutung mir sehr unsicher scheint, da auch an den übrigen Resten des Kopfskelets keine bestimmten Uebereinstimmungen mit anderen Organismen erweisbar sind. Es ist daher auch nicht zu vermuthen, dass sogar an Beziehung zu Froschlarven gedacht ward. Unter diesen Umständen möchte ich jene im Verhältniss zum Kopfe wie zum gesammten Körper bedeutende, von Cirren umstellte Eingangsöffnung als nicht einer Nase, sondern einem Munde oder beiden zugleich angehörig betrachten. Zu einem dem Cyclostomenriechorgan vergleichbaren Verhalten fehlen alle Bedingungen."

Relationships of Palæospondylus.—The arguments for and against the supposition that Palæospondylus is a Cyclostome may be here summed up after Professor Dean.

The vertebral column agrees with that of the lamprey in having the notochord in part persistent. On the other hand, the vertebræ have continuous centra, showing definite processes. Those of the different regions are differentiated. These conditions are quite unlike those seen in the lamprey.

The cranium is massive, over twice as large proportionally as that of the lamprey. In the latter type the cranium forms but a small portion of the bulk of the head; in Palæospondylus, on the other hand, the cranium bears every sign of having filled the contour of the head. Moreover, if the region adjacent to the structure is admitted to be that of the eye, and few, I believe, will doubt it, then the brain-cavity must, by many analogies, have been much larger than that of a Marsipobranch. Also the auditory capsules must have been of extraordinary size. In short, there is very little about the cranium to suggest the structures of Cyclostomes.

The "oral cirri" suggest somewhat the barbels of the nose and mouth of a hagfish. They, however, resemble even as much in arrangement and greater number the buccal cirri of Amphioxus. On the other hand, similar mouth-surrounding tentacles are evolved independently in many groups of fishes, siluroids, sharks, forms like Pogonias, Hemitripterus. A possibility further exists that the "cirri" may turn out to be remnants of cranial or facial structures of an entirely different nature. In fact the very uncertain preservation of these parts renders their evidence of little definite value. In but one specimen, as far as I am aware, is there any evidence of the presence of ventral cirri.

The jaw parts in Palæospondylus are unknown. It is possible that the ventral rim of the "nasal ring" may prove to be the remains of the Meckelian cartilage (the cartilaginous core of the lower jaw).

It is possible that certain very faint ray-like markings noted by Professor Dean may be the basalia of paired fins. In such case Palæospondylus can have no affinity with the lampreys. Dr. Dean asserts that the presence of these, in view of the wide dissimilarity in other and important structures, is sufficient to remove Palæospondylus from its provisional position among the Cyclostomes. The postoccipital plates may represent a pectoral arch. It is, however, much more likely, as Dr. Traquair has insisted, that the supposed rays are due to the reflection of light from striations on the stone, and that the creature had no pectoral limbs.

The caudal fin, with its dichotomous rays, is essentially like the tail of a lamprey. This condition is, however, found in other groups of fishes, as among sharks and lung-fishes. It is, moreover, doubtful whether the rays are really dichotomous.

It is possible that Palæospondylus may be, as Huxley suggests, a larval Arthrodire. It is not probable that this is the case, but, on the other hand, Palæospondylus seems to be an immature form. According to Dr. Dean, it is more likely to prove a larval Coccosteus, or the young of some other Arthrodire, than a lamprey. Against this view must be urged the fact that the tail of Palæospondylus is not heterocercal, a fact verified by Dr. Traquair on all of his many specimens. It is more like the tail of a lamprey than that of Coccosteus. It is, however, certain that it cannot be placed in the same class with the living Cyclostomes, and that it is far more highly specialized than any of them. In a still later paper (1904) Dr. Dean shows that the fossil might as easily be considered a Chimæra as a lamprey, and repeats his conviction that it is a larval form of which the adult is still unrecognized.

We cannot go much farther than Dr. Dean's statement in 1896, that it belongs "among the curiously specialized offshoots of the early Chordates."