( 383 )

XVIII.—On the Structure and Affinities of Tristichopterus alatus, Egerton. By RAMSAY H. TRAQUAIR, M.D., F.G.S., Keeper of the Natural History Collections in the Museum of Science and Art, Edinburgh. (Plate XXXII.) (Read April 5, 1875.) Concerning the affinities and systematic position of this very remarkable fish, there has hitherto prevailed very great uncertainty. The two original specimens, discovered by Mr C. W. PEACH, in the Old Eed Sandstone of John O'Groat's, Caithness, and described by Sir PHILIP EGERTON,* left us in complete ignorance as to the osteology of the head and the dentition, while the evidence they afforded as to the structure of the pectoral fins was by no means so clear as might have been wished for. To quote from Sir PHILIP'S descrip- tion:—"The bones of the head, with the exception of a small fragment of the operculum, are wanting, but the impressions left upon the matrix show that they were sculptured in rather a bold pattern, not unlike the ornament on some of the cranial bones of some of the Holoptychii, and consequently differing in this respect from the corresponding parts in Dipterus. The pectoral fins are very indistinctly seen. They appear to have had a short obtuse lobe forming the base, and extending therefrom a set of numerous fin-rays more elongated than those forming the pectoral fin in Dipterus." To Dipterus, however, in Sir PHILIP EGERTON'S opinion, its affinities pointed, as far as could be gathered from the structure of the body as displayed in the specimens, his description concluding as follows:—" The absence of all evidence as to the dental apparatus of Tristichopterus is much to be regretted. On other points the affinities between this genus and Dipterus are so striking that they cannot be classified in separate families.t Accordingly he assigned to Tristichopterus a place along with Dipterus in the family of " Ccelacanthi," the term being used in its former extended sense, not as now restricted to the peculiar genera Ccelacanthus, Undina, Holophagus, and Macropoma. Professor HUXLEY, at the conclusion of his Essay on the Classification of the Devonian Fishes,^ published in the same Decade of the Geological Survey, makes the following statement regarding the genus in question:—"In the absence of a full knowledge of the head, of the paired fins, and of the dentition, * Deo. Geol. Survey, x. 1861, pp. 51-55, pi. v. t Loc. cit. p. 55. + Dec. Geol. Survey, x. 1861, p. 40. VOL. XXVII. PART III. 5 H 384 DR TRAQUAIR ON THE STRUCTURE AND AFFINITIES it would be hazardous to form any decided opinion as to the affinities of Tristichopterus; I strongly suspect, however, that it will turn out to be the type of a new family allied to the Ctenododipterini and Coelacanthini." How- ever, at page 24 of the same essay, he places it along with Dipterus in PANDER'S family of Ctenododipterini, though with a mark of interrogation. But when, in 1871, Dr GUNTHER* pointed out the close relationship between Dipterus and the recent and Lepidosiren, and the consequent desirability of transferring the first named genus to the group of Dipnoi, the question of course came up as to whether Tristichopterus should also accompany it thither. But to this no satisfactory answer could be given, so long as our knowledge on so many essential points of its structure was deficient. After referring to the manner in which the innumerable fine and closely-placed rays of the fins overlapped " with their proximal ends, the extremities of the inter- spinous bones, as in the Dipnoi," and the peculiar form of the tail, which "represents a most curious intermediate condition between the diphycercal tail of the Sirenidce and the heterocercal of Dipterus," Dr GUNTHER con- cludes :—" Unfortunately, the head and the base of the paired fins are destroyed in the only two specimens known; and it is chiefly the last-named character which prevents me from associating this genus with the Dipnoi." No further description of the structure of Tristichopterus having appeared since the publication of the Tenth Decade, I felt great satisfaction when Mr PEACH communicated to me a number of additional specimens, collected by him in the years 1864^65, and which throw a very great amount of the light desired on those points of its structure previously unknown to us.t These, specimens exhibit in a clear and unmistakable manner the character of the dentition, the structure of the paired fins, and the leading features of the osteology of the head, and completely prove that Tristichopterus has no special affinity either with Dipterus or Coelacanthus. Before, however, passing on to discuss the question of its real affinities and systematic position, I shall first proceed with the description of the new facts disclosed. General Proportions.—One of the specimens, the counterpart of which is in the British Museum, is quite entire, though the axis of the body is so curved as to render the dorsal margin considerably concave, the ventral corre- spondingly convex. The entire length of this specimen, carefully ascertained with a flexible measure, is 10£ inches, of which the head occupies about £th part; the greater depth of the body, just behind the subacutely lobate pectorals, being 2 inches. The general form of the fish is thus rather slender, and the fins are crowded towards the posterior aspect of the body,—the first dorsal commencing 6 inches, and the second 1\ inches from the tip of the snout, * Description of Ceratodus, Phil. Trans. 1871. t These specimens are now in the Museum of Science and Art, Edinburgh. OF TE.ISTICHOPTERTJS ALATUS. 385 while opposite these two dorsals are respectively placed the ventrals and the anal. The lower lobe of the very peculiarly shaped caudal fin commences at 8^ inches from the front, and on the dorsal aspect the rays of the upper lobe begin to be apparent a few lines further back. Another specimen, crushed vertically, and lying on its back,—or more correctly, the counterpart of a specimen in that position, is also very nearly entire; the front and a considerable part of the right side of the head being unfortunately cut off by the edge of the slab, and the extremity of the tail being rather distorted and indistinct. If we add half an inch to complete the head in front, the length of this example would also be about 10 inches; the pectorals arising 2\ inches, the ventrals 6 inches, and the anals 7| inches from the supposed extremity of the snout. These two specimens thus closely correspond with each other, and with the more complete of the two examples figured by Sir PHILIP EGERTON, and show the accuracy with which he allowed for its missing portions. Some of the more fragmentary specimens before me show, however, that the fish must sometimes have attained a considerably larger size, one head measuring, from the tip of the snout to the hinder .margin of the gill cover, no less than 3J inches, which would give over 15 for the entire length of the fish. The Head.— The head was protected above by a cranial "buckler" (Plate XXXII. figs. 1 and 2, C.B.), which in the leading features of its configuration recalls to our minds that of the Saurodipterini, though its external sculpturing is very different. As in that family, it tends to divide across into two portions,— a posterior or parietal, and an anterior or fronto-ethmoidal; here, however, the anterior moiety is the longer, exceeding the other by nearly ^ of its length. The hinder division of this buckler is on the whole quadrate in form, but broader behind than in front, the posterior and wider margin being also somewhat con- cavely excavated. The front portion forms anteriorly a rounded depressed snout, and seems on each side to be excavated to take part in the formation of the upper boundary of the orbit, though this part of the margin is not so clearly defined as might be wished; nor are the nasal openings seen in any of the specimens, which is not strange, taking into account their position in Osteolepis and Diplopterus, so close to the margin of the upper lip. It is not possible to map out the ossifications entering into the composition of this buckler; probably their arrangement would not depart much from that which is to some extent traceable in the Saurodipterini. As far as the anterior portion is concerned, the impression of two distinct frontals, entering largely into its composition, is distinctly seen in the specimen represented in fig. 2, and the presence of a small conical tooth on the labial margin of the snout in another (fig. 1), leads us to conclude that the premaxillary [p. mx) was also here represented. The entire outer surface of the buckler, as indeed of all the external bones of the head, 386 DR TRAQUAIR ON THE STRUCTURE AND AFFINITIES is sculptured with tolerably fine, irregular, angularly contorted, and interrupted rugae, with intervening furrows and pits, the pattern assuming sometimes almost a tubercular aspect. Along the posterior margin of the cranial shield are three plates (s. t, figs. 1 and 2), one mesial, somewhat polygonal in form, and two lateral, each apparently of a triangular shape. These are obviously the representatives of the three plates, which occur in a similar position in Osteolepis, Glyptolcemus, Megalichthys, &c, and of which different interpretations have been given by different authors. In Professor HUXLEY'S description of Glyptolcemus,* the mesial one is marked " supra-occipital," the two lateral, " epiotic." Mr PARKER! has, however, pointed out that they are dermal bones, and not to be considered homologous with those other deeper ossifications of the cranial cartilage. By PANDER\ they are in Osteolepis simply designated " Hautknochen," and considered to be equivalent to the five little plates, which in the recent Polypterus occur immediately behind the transverse row of supra-temporals, and between the pair of upper supra- claviculars (supra-scapulars), being in reality the first scales of the back. On the other hand, he considered the transverse chain of small plates (supra- temporal) which lie immediately behind the parietals of Polypterus, to be represented in Osteolepis microlepidotus by the narrow portion of the cranial shield, which in that species is marked off near the hinder margin by a more or less interrupted superficial transverse groove. I am myself very much inclined to the belief that the three dermal bones in question are in reality equivalent to the transverse supra-temporal chain in Polypterus and Lepidosteus, and which have their representatives as well in the amphibian Labyrinthodonta as in most Teleostean fishes; the transverse grooving across the posterior part of the cranial shield in many Saurodipterines being probably only of the nature of superficial markings. Eegarding the condition of the side walls, or of the base of the cranium, not the smallest information is yielded by any of the specimens. The facial bones in their general form and arrangement also remind us very much of those in both the " Saurodipterini" and " Glyptodipterini." The gape extends very far back, so that the lower jaw is as long as the cranium proper, and the anterior margin of the operculum comes to be inclined obliquely down- wards and backwards. The operculum (op) is of a somewhat trapezoidal form, the anterior and inferior margins being the longest, the upper the shortest, while the posterior superior angle is obtuse and rounded. Below it is in contact with the suboperculum (s. op), which is also trapezoidal, but not quite so large; its anterior superior angle is produced a little way upwards into a narrow sharp-pointed process overlapped by the operculum. In front of the

* Dec. Geol. Survey, x. p. 2. t Shoulder Girdle and Sternum, p. 19. J Ueber die Saurodipterinen, &c, des Devonischen Systems, St Petersburg, I860, p. 11-12. O F TRIST1CH0PTERUS ALATUS. 387 opercular bones, and covering a great portion of the cheek, is a large oblong plate [p. op), which comes into close contact below with the hinder part of the maxilla, and articulates in front with two smaller plates separating it from the orbit, while above it is in contact with the side of the posterior part of the cranial shield. Its outer aspect is granulated; the inner is smooth, and shows, along its oblique and slightly curved posterior margin, a narrow shallow groove, evi- dently for articulation with the hyomandibular element of the suspensorium. The correspondence of this plate with the similarly placed one in Osteolepis, Diplopterus, Gyroptychius, &c, and with the great cheek plate of Polypterus, is at once evident. In those fishes it is by most authors reckoned to be the prce- operculum, though by Professor HUXLEY it has been marked " supra-temporal," seeing that by its remarkable extension forwards on the cheek it differs so greatly from the true preeoperculum in the Lepidosteid Ganoids, and in ordinary Teleostei. The orbit is situated very far forwards, and has connected with it a distinct chain of osseous plates (s. o, figs. 1 and 11). Of these there is a small one placed at the posterior superior angle of the orbit, lying for a small distance along the outer margin of the anterior division of the cranial shield, immediately in front of the transverse line, which marks that off from the posterior division. This is followed by two plates of considerable size, forming the posterior bound- ary of the orbit, and interposed between it and the great cheek plate. Of these the lower one is the larger; it is in contact below with the maxilla, and sends a process forwards forming a portion of the inferior boundary of the orbit, when it apparently joins another, or prce-orbital, plate, whose conformation is not, however, very distinctly seen. The maxilla (moo) forms a long and narrow bar, slightly expanded posteriorly, placed below the great cheek plate and the sub- orbitals, and extending forwards to the small premaxilla, which is firmly united with the anterior, or ethmoidal, part of the cranial shield. The oral margin of the maxillary bone is distinct enough, though in no specimen is its upper boundary shown with the clearness that might be desired. Internally, along its anterior two-thirds, it was certainly very firmly united to the outer margin of the palato- quadrate arch, an interval being left posteriorly for the passage of the muscles of the lower jaw. The palato-suspensory apparatus {p. q, figs.2 , 3, and 10) is formed by a broad and extensive bony lamina, extending forwards from the opercular bones, and articulation of the lower jaw to the ethmoidal part of the skull. Its upper margin is connected posteriorly with the squamosal region of the cranium; in front of this its plane becomes a little twisted, so that its outer surface comes to look more upwards, and the previously upper margin is directed inwards to the base of the skull. Its posterior margin, gently curved with a posteriorly directed con- vexity, passes obliquely downwards and backwards to the very posteriorly situated articulation of the lower jaw; this margin, apparently corresponding to the hyo- VOL. XXVII. PART III. 5 I 388 DR TRAQUAIR ON THE STRUCTURE AND AFFINITIES mandibular element, articulates behind with the operculum and suboperculum, and externally with the great "prse-opercular" cheek plate along the narrow groove, already described, on the hinder edge of the inner aspect of that bone. Lastly, by the anterior two-thirds of its outer margin, it is immovably fixed to the maxilla, and then recedes a little inwards from that bone before passing back to the quadrate articulation, so that, as already mentioned, an interval is left posteriorly for the passage of the masticatory muscles. In spite of the most careful examination, I have not been able to discover the lines of demarcation between the probable constituent elements of this apparatus, which in its general relations, though considerably broader in form, corresponds closely with that in the recent Polypterus. The lower jaw (Mn) is long, and, though pretty robust, is more slender than in the Saurodipterini; near the front, its lower margin is slightly excavated. Seen from the side it is tolerably straight; from below it is, as we might expect, gently curved inwards towards the symphysis. Not much can be made out regarding its component elements, though the dentary is evidently powerfully developed, and a small narrow detached plate (fig. 5) lying on one of the slabs near a head, and having on one margin a few minute conical teeth, is in all pro- bability the splenial. The space between the two mandibular rami, on the under surface of the head, is occupied by two very narrow jugular plates (/), each of which is acutely pointed in front; there is no trace either of an azygos jugular, or of lateral ones. Dentition.—The dentition of Tristichopterus is clearly enough exhibited in several of the specimens. The lower jaw of one shows two stout sharp conical teeth, each measuring \ inch in length by a little more than ^ inch in diameter at the base; they are both so split that the external surface is seen only towards their apices. Another specimen exhibits three mandibular teeth of nearly the same dimensions, and entire except at their bases, besides one maxillary tooth with its apex broken off, and the complete impression of another. A detached tooth, quite entire (PI. XXXII. fig. 8, magnified), found in the same bed with the other remains, and from its shape and markings undoubtedly belonging to the same species, measures nearly ^ inch in length by ^ in diameter at the base. Besides these larger teeth, the presence of smaller ones, some being very minute, is obvious; some of these are seen on the premaxillary of the head represented in fig. 1. All these teeth are of an acutely conical form, slightly incurved, and very sharp; their external surface is brilliantly polished, but not smooth in every sense, being closely fretted all over with very minute and short longitudinal indentations, passing, indeed, into fine striae at the base. A distinct fluting is also observed at the bases of the larger teeth. In the detached palato-quadrate arch represented in fig. 3, a portion of the upper jaw has remained attached to its outer margin, showing the bases of some OF TKISTICHOPTERUS ALATUS. 389 of the maxillary teeth cut in transverse section. But a still clearer illustration of the structure and arrangement of the teeth is seen in another specimen (fig. 7), being a horizontal section of a portion of a jaw in a piece of grey limestone from the same locality, and which was found naturally polished by the waves. The internal structure of the teeth here exhibited corresponds so closely with that seen in the last-mentioned unquestionable specimen of Tristichopterus, that we are tolerably well justified in assuming the specimen to belong to the same species.* Here the bases (A. A.) of two of the larger teeth are seen in section, each having the empty socket of another beside it, and in addition we find no less than 26 smaller teeth cut through, the latter becoming gradually smaller towards the middle of the interval between two larger ones. In both of those specimens the transverse sections of the bases of the larger teeth measure from ^ to ^ inch in diameter; in some the section is quite round, in others slightly oval. A small central pulp cavity is shown with the dentine around it arranged in a few simple plications, most of which, though not all, reach the central cavity; in this manner the pulp cavity appears in the section to send out a number of narrow radiating prolongations towards the periphery, a few of which are seen to bifur- cate. Further up in the body of the tooth the dentinal folds become shorter, and the pulp cavity proportionally larger till towards the apex the latter has become perfectly simple. The pulp cavities of the smaller teeth appear to be perfectly simple throughout. Whether or not Tristichopterus was possessed of palatal teeth is not dis- coverable from any of the specimens under description. The Shoulder-Girdle and Paired Fins.—None of the specimens show very distinctly the upper attachment of the shoulder-girdle to the skull, or the form of the supra-claviculars. In one, evident traces are seen of a powerful second supra-clavicular (s. cl, fig. 10) extending downwards and backwards to articulate with the next or clavicular element, but too crushed and indistinct for special description. The clavicle (cl, figs. 1, 6, 9, and 10) is well marked in all; it is a stout, broad, oblong plate, expanded and produced a little forwards below; its outer surface is marked with the characteristic ridged-granular ornament, while its smooth internal aspect shows three peculiar rounded impressions, probably for the attachment of the coraco-scapular elements of the base of the pectoral fin. Articulated with the front of the lower end of the clavicle is another smaller plate (i.cl), the interclavicular of PARKER,—the " accessorisches Clavicularstuck" of GEGENBAUR. The pectoral fin itself (PI. XXXII. figs. 9 and 10) is large, obovately fan- shaped, terminally rounded, and consists of very numerous slender rays, attached * Since the above lines have been in type, the Museum has acquired from Mr Peach an addi- tional and nearly perfect specimen of Tristichopterus, in which, near the front of the head, the base of one of the large teeth is seen broken, or cut, across in transverse section. The transverse section of this tooth is £ inch in diameter, and displays a structure absolutely identical with that described above. 390 DR TEAQUAIR ON THE STRUCTURE AND AFFINITIES to each side, and round the extremity, of a central scaled lobe, which is less than half the length of the fin, the latter being thus subacutely lobate. In a specimen measuring 2§ inches from the tip of the snout to the posterior margin of the operculum, the entire pectoral fin measures If inch, and the lobe f inch in length. The rays are slender and closely set, short at their commencement on each side of the lobe, but becoming very rapidly elongated; they commence to bifurcate very soon after their origin, and the process is frequently repeated, till at their terminations the longitudinal divisions are very delicate. They are also divided all along their course by transverse articulations; the joints being, how- ever, seldom less than four or five times as long as they are broad. The pectoral fin is not absolutely complete in any one example; a comparison, however, of its appearance in the various specimens of the suite shows that it would have pretty much the form shown in the restored figure given in PI. XXXII. fig. 11. The lobe is covered externally with scales similar to those to be presently described as covering the body, though smaller, and which generally completely obscure the supporting skeleton within. Nevertheless, in the specimen represented in fig. 9, some insight is derived into the nature of this internal skeleton, though, unfortunately, not to so full an extent as might be desired. There is first, at the lower and posterior part of the clavicle (cl), some obscure bony matter (a), which may possibly represent the remains of ossification in the scapulo-coracoid cartilage. This is followed by a central basal stem for the fin, consisting of at least two flattened oblong pieces (b 1, b 2) articulated end to end. On one side (from the possible twisting round of the otherwise displaced fin, it is hard to say if it really be the medial or lateral) are two distinct radiate (r 1, r 2) set at acute angles to the axial stem, and of which the second arises at the place of junction of the first with the second axial segment. Attached to the extremity of the second axial segment are two additional pieces, one of which (b 3) seems to represent a third division of the axis, the other (r 3), slightly diverging, may be considered as a third radial. There is no clear evidence of radials on the other side of the axis, but as fin rays are also here present it is hard to suppose that they were destitute of similar supporting elements, and it seems meanwhile probable that defective preservation is the cause of their not being distinctly recognisable. Were radial elements present on both sides of the axis, the skeleton of the pectoral of Tristichopterus would present an abbreviated form of that of the Ceratodus limb, in which we have an elongated segmented central axis [Archypterygium of GEGENBAUR) set with segmented radials on both sides. Dr GUNTHER has suggested an analogy between the doubly fringed acutely lobate pectoral of Ceratodus, and the diphycercal tail with elongated axis in that genus, and many fossil ones; the pectoral of Tristichopterus, and probably of all other Crossopterygidse with subacute and obtusely lobate structure, would in like manner correspond with the diphycercal tail with shortened axis OF TRISTICHOPTERUS ALATUS. 391 seen in Polypterus* On the other hand, the pectoral limb in the Selachii, in the Sturionidee, Lepidosteidse, and Amiadse, and in modern Teleostei, offers analogies with the heterocercal tail with shortened axis, the axis being abbre- viated, and the radials developed principally, or (in most cases) entirely, on one side of it. The ventral fins, whose rays are similar in structure to those of the pec- torals, are smaller than the latter, and are very slightly lobate at the base. They are better shown in one of the specimens from which Sir PHILIP EGERTON'S figures were taken, than in any of the present suite. I have not observed the pelvic bones. The Internal Skeleton of the Trunk.—An osseous vertebral column is trace- able from the cranium to the extremity of the attenuated prolongation of the body through the upper part of the caudal fin. The bodies of the vertebrae were certainly ossified, but whether they were completely so, or remained more or less in the condition of " ring-vertebrae," is not discoverable from any speci- men I have seen, as they are in every case so compressed that no view is afforded of their anterior or posterior surfaces. A series of distinct neural arches, passing upwards into spines, the latter flattened laterally in the fore part of the body, are seen appended to the dorsal aspect of the vertebral centra. Corresponding haemal arches and spines are seen in the caudal region, though anteriorly I have not been able to detect any trace of ribs, and the extreme caudal termination of the column seems to be formed by centra alone. The interspinous bones of the azygos fins have been well described by Sir PHILIP EGERTON. Those of the first dorsal are small and obscure, but there are three very prominent ones supporting the second dorsal and anal fins respec- tively, and which are in turn supported above and below by a large flattened bony piece, considered by Sir PHILIP EGERTON to be probably a composite spinous apophysis, " formed by the union of three or more spines." Similar interspinous ossicles are seen supporting the lower lobe of the caudal, disappear- ing, however, before the extreme termination of the vertebral column is reached; their number is given by Sir PHILIP EGERTON as eight or ten. The ossicles sup- porting the anterior rays of the upper lobe of the caudal fin are certainly neu- rapophyseal in their nature. The Azygos Fins.—The first dorsal, placed opposite the ventrals, is small and narrow, being somewhat elliptic-lanceolate in shape, and frequently found more or less adpressed towards the back; the second is larger and of a more expanded triangular form, closely resembling the opposed anal in general pro- portions and form. The bases of all three are very short, and slightly lobate; their most anterior rays are short, but rapidly increase in length till the apex * The tail of Polypterus is not, however, absolutely diphycereal, though conforming more to that type than to any other. VOL. XXVII. PART III. 5 K 392 DK, TRAQUAIR ON THE STRUCTURE AND AFFINITIES of the fin is reached, from which they become gradually shorter, but increase in delicacy and in obliquity of direction—the most posterior ones being quite horizontal in position. The form of the caudal is indeed remarkable, presenting as Dr GUNTHER observes, " a most peculiar intermediate condition between the diphycercal tail of the Sirenidse and the heterocercal of Dipterus." It is large and fan-shaped, nearly truncated posteriorly, the hinder margin being only slightly excavated. A prolongation of the body axis, becoming very rapidly attenuated, and then tapering to a fine point, runs right through it, but as the lower margin of this prolongation slopes much more rapidly upwards from the commencement of the anterior rays of the lower part of the fin, its termination comes to the posterior margin of the latter much above the middle, so that, as in heterocercal forms, the larger part of the caudal fin is developed on the lower aspect of the vertebral column. The rays of the upper part of the caudal com- mence a little further back than those of the lower; in both cases they are short at first, but become rapidly longer till the upper and lower apices are reached, from which they become gradually shorter, finer, and more oblique in their origin from the body axis. Those arising from the extreme point of this axis are very delicate, and project beyond the margin of the rest of the fin, so as to produce the appearance described by Sir PHILIP EGERTON as " forming a kind of supplemental fin, projecting beyond the terminal margin of the true caudal fin." It must, however, be observed that the rays in question, though projecting in that remarkable manner, form a perfectly continuous series with those of the rest of the caudal (see restored figure, PI. XXXII. fig. 11). All these fins are composed of slender, closely-set rays, repeatedly dichotomising, and divided by frequent transverse articulations, the joints being, however, as in the case of the paired fins, always considerably longer than broad. Those of the anterior part of the lower lobe of the caudal are, as Sir PHILIP EGERTON has pointed out, stouter than the others; they are, in fact, the stoutest fin rays in the entire structure of the fish. There are no traces of fulcral scales on the anterior margins of the fins, and, in this respect, Tristichopterus differs from Gyroptychius and from the Saurodipterini, where such scales are present, though differing rather in form from the pointed imbricating fulcra of the Pcdceoniscidce and Lepidosteidce. The statement of Sir PHILIP EGERTON, borne out by one of his figures, that the anterior rays of the upper lobe of the caudal are "short and fulcral, the anterior ones being short, and forming a marginal fringe along the upper edge of the fin," I do not find corroborated by a very beautifully preserved tail in the series of specimens from which the present description is taken. The fin-rays overlap the extremities of the supporting ossicles, and are more numerous than the latter elements—characters occurring also in many other Crossopterygidae, and in the Palseoniscidse. Scales of the Body.—The scales are of moderate size, rounded, thin, and OF TRISTICHOPTERTJS ALATUS. 393 deeply imbricating—the exposed area being smaller than that which is over- lapped. The exposed surface is ornamented by very fine closely-set raised strise, often interrupted, and branching and anastomosing; their general direction is from before backwards, though they are usually seen to converge slightly towards the median line of the scale. On very close examination these striae may frequently be observed to be decussated obliquely by still more delicate lines, seen in the intervals between them, and which seem to radiate from the central point of the scale. The covered area is very minutely granular, a concentric linear arrangement of the granules being also usually to be observed. The impression of the under surface of a scale, in one specimen, shows evidence of a small elevated central boss or elevation. The scales of the lateral line seem to be perforated by a slime-canal, and to be slightly notched posteriorly. Conclusion.—The structural characters of Trislichopterus, as far as they have as yet been ascertained, may be summed us as follows:— Body slender, elongated; scales cycloidal, thin, imbricating, delicately striated. External bones of the head sculptured; cranial roof bones united into a buckler of two principal parts, anterior and posterior; snout depressed, rounded; orbit placed far forward; a large prseopercular plate covering the cheek in front of the well-developed operculum and suboperculum, as in the Saurodipterini. Gape extending far back; maxilla narrow, closely united internally to the bony palate; mandible stout; jugular plates two, narrow; no median or lateral jugulars. Teeth in both jaws conical, sharp, slightly incurved, of different sizes; the larger teeth having the dentine thrown internally into a few simple folds at the base, which is also fluted externally. Vertebral column with ossified centra, attenuated posteriorly. Shoulder-girdle provided with interclaviculars; pectoral fins subacutely lobate; ventrals very slightly lobate. Dorsal fins two, placed far back and opposite the ventrals and anal respectively, the posterior dorsal and anal being each supported by three prominent inter- spinous bones. Caudal intermediate in general form between the heterocercal and diphycercal type; large, pointed above and below; the rays affixed to extreme termination of body-axis projecting beyond the line of the nearly vertical posterior margin of the rest of the fin. Rays of all the fins slender, articu- lated, closely set, overlapping their supporting ossicles; fulcral scales absent. This assemblage of characters renders it evident that Tristichopterus has, as stated in the introduction, no affinity with Dipterus, nor any special relationship with Ccelacanthus, but that, on the other hand, its place in Professor HUXLEY'S classification of the Crossopterygian Ganoids is in the cycliferous division of his family of Glyptodipterini. This group may, however, be very advan- tageously subdivided, as there are structural differences between some of its members, which, as it seems to me, are of importance enough to rank as family 394 DR TRAQUAIR ON THE STRUCTURE AND AFFINITIES distinctions. Dr LUTKEN* has already proposed to unite those genera having rhombic scales with the Saurodipterini, under the name of Rhombodipterini, an arrangement in which I am much inclined to concur—the remainder with rounded imbricating scales being designated Cydodipterini. But among these Cyclodipterini there occur fishes with two somewhat different types of pectoral fin—those in which that member is very acutely lobate, like that of Dipterus, or of the recent Ceratodus {Holoptycliius, ), and those in which it is only subacutely so, the lobe terminating considerably before the end of the fin, which consists of more elongated rays, especially in the central part (Tristi- chopterus, Rhizodus, Rhizodopsis). Following Dr GUNTHER in associating the Ctenododipterini {Dipterus, ) with the Dipnoi, I would therefore, pro- visionally at least, propose the following modification in the arrangement of the remaining Crossopterygidae:—

I. Caudal diphycercal, but with shortened body-axis. Dorsal fin multifid, pectorals obtusely lobate, scales rhomboidal. FAM. 1.—Polypteridae (Polypterus, Calamoichthys). II. Caudal with elongated attenuated body-axis, heterocercal or diphycercal. A. Pectorals obtusely lobate, tail diphycercal, dorsal fins two, scales cycloidal, air bladder ossified. FAM. 2.—Ccelacanthidse (Caelacanthus, Macropoma, Holophagus). B. Pectorals subacutely lobate, dorsal fins two, tail heterocercal or diphycercal. a Scales rhomboidal. FAM. 3.—Rhonibodipteridse. * Scales sculptured, Sub-fam.—Glyptolaemini (Glyptoltemus, Glyptopomus). ** Scales smooth. Sub-fam.—Saurodipterini (Osteolepis, Diplopterus, Megalichthys). /S Scales cycloidal, sculptured. FAM. 4.—Cyclodipteridse (Tristichopterus, Gyroptychius (?), Ehizodus, Rhizo- dopsis, Strepsodus, Archiehthys (?) ). C. Pectorals acutely lobate, scales cycloidal. a Dorsal fins two, ventrals subacutely lobate, scales thick, sculptured. FAM. 5.—Holoptychiidae (Holoptycliius, Glyptolepis, Dendrodus(V), Gricodus(f)). /S Dorsal fin elongated, continuous with the upper part of the caudal; ventral fins acutely lobate; scales thin. FAM. 6.—Phaneropleuridse (Phaneropleuron, Uronemus).

I have placed the genus Gyroptychius, M'COY, in the family of Cyclo- dipteridae, next to Tristiehopterus, with a mark of interrogation, because, according to PANDER'S elaborate description,! it seems to approach the subject of the present memoir nearer than does any other genus hitherto described, in the sculpturing of the head plates, and of the scales, which, according to that author, "sind elliptisch und bedecken sich dachziegelartig." The fulcration of the fins, the rhomboidal diphycercal form of the caudal, and the greater relative

* Ueber die Begrenzung und Eintheilung der Ganoiden; Palseontographica, Bd. xxii., Erste Lieferung, 1873, p. 47. t Op. cit. pp. 55-61. OF TRISTICHOPTERITS ALATXJS. 395 size of the exposed area of the scales, which are also mentioned as "an manchen Stellen des Korpers ins Ehomboidale iibergehend," are, however, very obvious marks of distinction. Yet, as to the true nature of M'COY'S Gyroptychius, there seem to be very considerable doubts. Classified by PANDER as one of the " Dendrodontes," a family afterwards merged by Professor HUXLEY in the cycliferous Glyptodipterini,* and originally described by M'COY as a " Coelacanth "t (i.e., cycliferous Crossopterygian), Gyroptychius was, however, placed by Sir PHILIP EGERTON among the Saurodipterini,J and by HUXLEY in the rhombiferous section of the Glyptodipterini. § And it must be owned that, although the elliptical scale, figured by Professor M'COY from the back of Gyroplychius angustus, " very much resembles," as Sir P. EGERTON says, " a scale of Tristichopterus,"|| yet, in the definition of the genus, its founder states that the scales of the flanks are "sub-rhomboidal;" and those figured by him from the side of G. diplopteroides are most decidedly Saurodipterine in form and arrangement, the surface sculpturing also very closely resembling the markings seen on many saurodipterine scales when divested of their external ganoine layer. Not having, however, yet seen the original specimens on which the genus was founded, I must be content, in the meanwhile, to leave this question as it is.

* Dec. Geol. Survey, x. p. 23. t British Palaeozoic Fossils, pp. 596-597, Plate 2, C, figs. 2, 3. X Qu. Journ. Geol. Soc. xvi. (1860) p. 126. § Dec. Geol. Survey, x. p. 23. || Ibid. p. 54, note.

EXPLANATION OF PLATE. In all the figures the matrix has been omitted to save space. The same letters refer to the same bones throughout. C.B, cranial buckler; /, frontal; p.mx,pre-maxillary; Mn, mandible; sp, splenial; op, operculum; s.op, sub-operculum; p.op, prseopercular cheek-plate; s.t, supra-temporals; s.o, sub-orbitals; j, jugular plate; p.q, palatoquadrate arch; cl, clavicle; s.cl, supra-clavicular; i.d, inter-clavicular; b, basal segments of archypterygium; r, radials. Figure 1. Head of Tristicliopterus alatus, the facial bones of the left side exhibited from the internal aspect. The granulated impression of the outer upper aspect of the cranial buckler is shown, a good deal of the bone of its anterior division still adhering to the matrix; and in the upper part of the figure the prseopercular plate of the right side is shown, as also the operculum, the latter partly in impression. Figure 2. Another head, seen from the right side. The prseopercular plate and circumorbital bones are gone, the palatoquadrate apparatus being thus exposed, with a portion of the maxilla attached to the anterior part of its outer margin; the supra-temporals, operculum, part of the sub-operculum, and the cranial buckler, are seen in impression of their internal surfaces, part of the externally granulated bone of the latter still remaining in its posterior division. The mandible is rather injured along its inferior margin. Figure 3. Palato-quadrate apparatus detached. A portion of the edge of the maxilla, showing bases o numerous teeth cut transversely, is seen anteriorly and externally. Figure 4. Detached prseopercular plate, seen from the inner aspect. Figure 5. Detached splenial of mandible (?). VOL. XXVII. PART III. 5 L 396 DR TRAQUAIR ON THE STRUCTURE OF TRISTICHOPTERUS ALATUS.

Figure 6. A group of facial and shoulder-bones lying detached on a slab of stone. Figure 7. Horizontal section of a detached dentigerous bone (mandible ?), showing the structure and arrangement of the teeth. Magnified a little more than 3 diameters. A. A. Transverse sections of the bases of two of the larger teeth. Figure 8. One of the large teeth of Tristichopterus, magnified 7 diameters. Figure 9. Clavicle and part of the pectoral fin, the latter displaced downwards, but showing part of the internal skeleton of the lobe. Figure 10. Anterior part of a specimen of Tristichopterus, principally to show the form of the pectoral fin and of its basal lobe. The rays of the inner side of that fin are injured and broken up, except at their origins, s.cl, Left supraclavicular; cl', portion of left clavicle; el, right clavicle, seen from the inner side ; i.cl, right interclavicular. Figure 11. Restored figure of Tristichopterus alatus. Trans. Roy Soc.Edmr Vol. XXVII, Plate XXXII.

Fy. I.

IB. f

f R

s op Mrv

Fig. 5.

/[ $$ op

n$-7-

R.H.Tra^uair, Jtinat ad* M'Tarlane k Eislcine, Lift™