SK \ \ \ \ ~ - _ ~ < \ Cornell Aniversity Library THE GIFT:OF .~; ‘ a os a University Library QE 851.E13 “NTA York formation New York State Education Department New York State Museum Joun M. Crarke, Director Memoir 10 DEVONIC FISHES OF THE NEW YORK FORMATIONS BY CHARLES R, EASTMAN PAGE PAGE Introduction - : : : - 7 | Summary and conclusions - 173 Conspectus of species, arranged ac- Zoological conclusions - 178 cording to their geological occur- Geological conclusions with remarks rence - z A S - - 12 on the distribution of Devonic Tabular key to systematic descriptions 23 fishes - : - 183 Systematic account of Devonic fishes, Explanation of plates - - - 195 principally from New York and Index - - - - 3 7 - 225 Pennsylvania - - - - - 24 ALBANY NEW YORK STATE EDUCATION DEPARTMENT 1907 + 4K tots Ig!7 1908 IgI4 1gI2 1918 1glO IQl5 1gIl 1909 1916 STATE OF NEW YORK EDUCATION DEPARTMENT Regents of the University With years when terms expire WuuireLtaw Reip M.A. LL.D. Chancellor 2 2 Sr Crain McKetway M.A. LL.D. V7ce Chancellor Dace Bracn Pi.D. Ld... « - - - - Pry T. Sexrow LL.B. LL.D, 5 - - - T. Guitrorp SmitrH M.A. C.E. LL.D. - : Wittiam NotrincHam M.A. Ph.D. LL.D. a Cuaritres A. Garpbiner Ph.D. L.H.D. LL.D. D.C.L. ALBERT VANDER VEER M.D. M.A. Ph.D. LL.D. = Epwarp LautersacH M.A. LL.D. - Evucene A. Puitsin LL.B. LL.D. - 2 : = Lucian L. SHEDDEN LL.B. - - - - : Commissioner of Education AnpREw S. Draper LL.B. LL.D. Assistant Commissioners Howarp J. Rocers M.A. LL.D. First Assistant Epwarp J. Goopwin Lit.D. L.H.D. Second Assistant New York Brooklyn Watkins Palmyra Buffalo Syracuse New York Albany New York New York Plattsburg Aucustus S. Downtne M.A. Pd. D. LL.D. Third Asszstant Secretary to the Commissioner Hartan H. Horner B.A. Director of State Library Epwin H. Anperson M.A. Director of Science and State Museum Joan M. Crarke Ph.D, LL.D. Chiefs of Divisions Accounts, Wi_ttiam Mason Attendance, James D. SuLLIvAN Educational Extension, Witt1am R. Eastman M.A.M.L.S. Examinations, CHarLes F. Wueetock B.S. LL.D. Inspections, Frank H. Woop M.A. Law, Tuomas E. Finecan M.A. School Libraries, Cuartes E. Frrcn L.H.D, Statistics, Hiram C. Case Visual Instruction, DeLancey M. Euuis New York State Education Department Sceence Division, May 26, 1906 Fon. Andrew S. Draper LL.D. Commesstoner of Education My pear sir: I transmit herewith, with the recommendation for publication as a memoir of the State Museum, a monograph on the Devonic fishes of New York which has been prepared at my urgent request by Dr Charles R. Eastman. I take especial satisfaction in presenting this important document as it is the result of the first attempt made to acquire a full knowledge of the fishes buried in our geological formations. In the previous study of New York paleontology this interesting field has been largely overlooked and this work is the first of its kind prepared for publication in any of our states. Very respectfully Joon M. Crarke Drrector and State Paleontologist Approved for publicatzon, June 6, 1906 Lobinrepet Commissioner of Education DEVONIC FISHES OF THE NEW YORK FORMATIONS INTRODUCTION Since the publication by Professor John Strong Newberry, in 1889, of an extensive Monograph on the Palaeozoic Fishes of North America® our knowledge of this group of fossil organisms has been vastly increased. Devonic fishes, in particular, have furnished important information concern- ing the organization and relations of primitive chordates, besides throwing much new light on the early history of fishes and fishlike vertebrates. If incertitude still exists regarding the larger problems of ancestry, we have at least progressed further than any one would have deemed possible a genera- tion ago, and have been able to determine with reasonable accuracy not only the progressive stages of evolution passed through by various groups, but also the manner in which certain fundamental structures, such as the paired limbs, dentition, dermal covering, etc. arose. With increasing wealth of material, some of it magnificently preserved, we have become almost as well acquainted with the structure of certain genera, sometimes even includ- ing the soft tissues, like the muscle fibers of Cladoselache, as with the corresponding parts of modern fishes. All this notable advance has been made within comparatively few years, and the results of modern paleontological research are scattered throughout a wide series of publications in this country and abroad, not all of them readily accessible. It is perhaps unfortunate that no single treatise has yet been prepared on the Devonic fish fauna either of this country or of Europe, since the quest for information regarding the fossil assemblages of any particular region, or concerning the structure of particular genera, involves no little expenditure of time and effort. A revised compendium, therefore, along the lines of the now somewhat antiquated monograph by Newberry would be an exceedingly welcome and useful contribution, and it is hoped that some such work may yet be undertaken. *U.S. Geol. Sur. Monogr. v. 16, 1889. 8 NEW YORK STATE MUSEUM The memoir which is now presented, and whose inception is due largely to active interest of the present State Geologist of New York, is intended partially to supply the need of a general conspectus dealing with the fauna of a definite area and within a definite formation It must be understood, however, as being of much more modest pretensions and more limited in scope, than the aforementioned monograph by Professor New- berry. As the title indicates, it is concerned chiefly with the Devonic fish fauna of the New York-Pennsylvania region, revised descriptions being offered of every genus and nearly every species entering into its composi- tion. Many of the larger groups have also been redefined in accordance with the present state of our knowledge respecting them. Obviously, how- ever, our understanding of the forms occurring within the boundaries of two political divisions would be incomplete except as they are viewed in relation to other species, genera and faunas, irrespective of their geographi- cal distribution. Moreover, a large proportion of the remains found in one state, owing to inferior preservation or other causes, would be obscure, per- haps even unintelligible, but for the occurrence of identical or kindred species in some other region. Let it be asked, for example, how much would be known of the structure of Bothriolepis if we were unable to extend our acquaintance beyond the extremely fragmentary specimens found in the New York and Pennsylvania Devonic? But as soon as com- parisons are undertaken with the exquisitely preserved Canadian examples, even the poorest fragment from the former region becomes invested with new light and interest. Accordingly, in the following pages, we have not hesitated to take considerable notice, and even to introduce descriptions of extralimital forms, wherever such procedure appeared conducive to a more perfect understanding of the species comprised by the local fauna. And finally, at the end of the systematic descriptions, some general questions are discussed which deal with geographical distribution and with the relations of the fauna as a whole. In the matter of formation names, preference has been given through- DEVONIC FISHES OF THE NEW YORK FORMATIONS 9 out to the classification of the New York series adopted by the geologists of this State,‘ and to Prosser’s classification of the Ohio Devonic.? In the latter system it will be noted that the local names of Columbus” and “ Delaware limestones” are used for the lower Mesodevonic rocks instead of Onondaga (—Corniferous) of the New York classification, for the reason that it is probable these Ohio formations should be correlated with a later horizon than the Onondaga—namely, the lower part of the Erian series of New York. We may also remark in this connection that the weight of evidence furnished by fossil vertebrates is decidedly in favor of grouping the formation ‘“Corniferous” of older geologists together with the Hamilton in a single division of the Mesodevonic. For the conven- ience of those who may not be perfectly familiar with the relations of the different systems, the following tables may be here inserted: * Science, Dec. 8, 1899; Ainerican Geologist, Feb. 1900; N Y. State Mus. Mem. 3. 1900; Univ. State N. Y. Handbook 19. 1903; N. Y. State Mus. Bul. 81, 82. 1905; 99. 1906; 107. 1907. ? Nomenclature of the Ohio Geological Formations. Jour. Geol. 1903. 11: 519-46. Io NEW YORK STATE MUSEUM Summary of classification of the New York Devonic according to John M. Clarke DESCENDING ORDER System Group Stage Carbonic ( Pennsylvanian Olean conglomerate Knapp beds Oswayo beds (Panama conglomerate) Cattaraugus beds incl, Wolf Creek con- ) Mississippian glomerate Devonic ( | Chemung beds ( Chautauquan | (Catskill sandstone, local Proruige beds | facies) | (Naples, Ithaca, Oneonta, | DeéNeCan, Sherburne beds, local facies) Genesee shale Tully limestone Neodevonic Hamil- Ludlowville shale ton Skaneateles shale Cardiff shale Marcellus shale Erian ( ( Moscow shale Mesodevonic AK Onondaga limestone Schoharie grit Ulsterian \ Esopus grit Oriskany beds Port Ewen limestone Becraft limestone New Scotland beds Coeymans limestone Helderbergian Paleodevonic Oriskanian ( | ( L c DEVONIC FISHES OF THE NEW YORK FORMATIONS II Classification of the New York Devonic according to James D. Dana DESCENDING ORDER : { Chemung Upper Devonian... Chemung........... | Portage eS . -S | Middle Devonian.. Hamilton........... Hamilton, Tully Marcellus > o , plate do not extend to the smooth articular border, but are separated from it by a band of sculpture. The pre- median plate is crossed by a groove which presents an Fig.13 Bothriolepis minor Newberry. Antero-dorsome- abrupt loop backwards at the middle.” dian plate with denuded ex- ee ee As is commonly the case in this class of remains, i sandstone; Jelaware county, N.Y, the antero-dorsomedian plate is of more frequent occur- rence than any other, though seldom found entire. Its orientation may be easily determined by means of the V-shaped sensory canals which diverge from about the center of the plate, or, in case the superficial ornament is denuded, by means of the longitudinal keel which traverses the median line of the plate for about three fourths of its length. A specimen of this character is illustrated in the accompanying text figure. Formation and locality. Chemung group of Bradford county, Pennsyl- vania, and Catskill of Delaware county, New York. According to New- berry, a species similar to this, and perhaps identical with it, occurs in the Devonic of Belgium. DEVONIC FISHES OF THE NEW YORK FORMATIONS 53 Bothriolepis coloradensis Eastman 1904 Bothriolepis coloradensis C. 2. Eastman. Am. Jour. Sci. [4] 18: 254, text fig. 2, 4 This, the largest known American species of Bothriolepis, is chiefly interesting on account of its geographical and geological occurrence, and for the evidence it furnishes regarding Asterolepid distribution. Although the length of the appendages has not been determined in this form, the general configuration of the ventral surface, and particularly the lozenge- like form of the ventromedian plate, sufficiently warrant its reference to this genus. Its relations seem to be rather with the Canadian than with either of the Chemung-Catskill species occurring in the eastern United States, Add to this the fact that no Asterolepid remains are known from the Upper Devonic horizons of the middle west, although other Chemung forms are present in abundance, it is difficult to trace the line of migration so as to connect the Rocky mountain species with those in New York and Pennsyl- vania, On the assumption that the former is a European immigrant, two other routes are available, one lying to the northward, by way of Canada, and the other to the southward, around the end of the Appalachian chain. A further discussion of this question may be reserved for a subsequent section of this memoir. In the original description of this species, two figures were introduced for the purpose of comparison with Scottish specimens which bore the designation of Bothriolepis major, and were reputed to have been derived from the well known locality southward of Elgin. Recently, a more aientive examination of these specimens‘ and of the matrix in which they are inclosed has led the writer to believe that they could not have been collected at Elgin, but most likely from the Upper Old Red sand- stone of the immediate vicinity of Nairn. Now these Nairnshire plates, which in former years were universally known under the name of Pter- * The median ventral in particular is broader and more ovately lozenge-shaped than in Bothriolepis major, and the specimen shown with all the ventral plates in apposition is one of the best ever found of this species, 54 NEW YORK STATE MUSEUM ichthys major, have been found by Dr Traquair to belong to ‘‘a very different creature from the Pterichthys (Bothriolepis) major of the adjoining district of Elgin,” and‘are determined by him as Astero- lepis maxima. The same author further observes that ‘it is remark- able that while Asterolepis maxima is unknown in those strata round Elgin which are characterized by Bothriolepis major, nota single remnant of any species of Bothriolepis has ever been found in the Asterolepis-bearing beds at Nairn. Other fish remains are very rare in these beds, but such as I have obtained leads me to suspect that we have here to deal with quite a different fauna from that of the closely adjoining district of Elgin.” Accordingly, the identification of the two figured specimens referred to should be amended so as to read ‘‘Asterolepis maxima, from Nairn.” Formation and locality. Elbert formation (Upper Devonic); Rock- wood and Devon Point, Col. Class PISCES Subclass ELASMOBRANCHII Order PLEUROPTERYGII The only known representative of this order, and at the same time, as indicated by the condition of its paired fins and other features, the most primitive type of Elasmobranch yet discovered, is the Upper Devonic and Lower Carbonic genus Cladoselache. Amongst all fossil fishes this genus is regarded as furnishing the most important testimony in favor of the lateral fin fold hypothesis. In view of its extreme importance from a mor- phological and phylogenetic standpoint, it seems desirable to offer the follow- ing account of its organization, compactly drawn up by Smith Woodward: The fish is elongated and round-bodied, with a short blunt snout and forwardly placed eye. The precise characters of the cranium are unknown: but the olfactory capsules are large, placed close together, and terminal. The mouth is also terminal, the upper and lower jaw being similar in shape and size, and supported by a slender elongated hyomandibular. The teeth are largest, longest, and most acutely pointed at the symphysis of the jaw, smallest and shortest at the angle of the mouth. The transverse series are closely apposed, and not separated as in the modern Chlamydoselachus ; DEVONIC FISHES OF THE NEW YORK FORMATIONS 55 they are indeed tightly wedged together, and the cusps are frequently much abraded by wear. Every tooth has a principal cusp with variable smaller lateral cusps, and the broad base of each is overlapped by its suc- cessor behind. The number of gill arches is uncertain, but five are known, and the presence of one or two others is suggested by some specimens. The neural and haemal arches of the axial skeleton have been observed only in the caudal region, where they are short tapering rods of cartilage, bifurcated at the base and distinctly corresponding in number with the calcified remains of the muscle plates. Intercalary cartilages are wanting. The small basal cartilages of the paired fins seem to be embedded in the body wall, and the unjointed radial cartilages extend directly outwards to the edge of the membrane. There are no claspers in the pelvic fins, and both these and the pectoral fins were probably mere balancers, directed somewhat downwards. Two low dorsal fins have been observed, both destitute of an anterior spine,’ but strengthened by simple cartilaginous rays extending to the edge of the fin membrane. No anal fin has been Fig. 14 Cladoselache fyleri Newberry, Cleveland shale (Upper Devonic); Ohio, Lateral aspect, anterior dorsal fin spine omitted. x 1-20 (From Dean) distinguished. The caudal fin [text fig. 14] is strongly heterocercal and very remarkable. The neural arches seem to persist even to the end of the upturned axis, directly supporting the thick radial cartilages of the superior lobe of the caudal fin. The dermal membrane of the inferior lobe of the caudal fin is supported by simple cartilaginous rays only in its lower portion where they extend quite to the margin. The eye is surrounded by a double series of small dermal plates; but the remainder of the fish is covered only with minute lozenge-shaped denticles, which are apparently not enameled. The latter are slightly enlarged at the angles of the mouth, where they approximate in size and shape to the smallest of the teeth. The lateral line extends along the trunk between two series of the shagreen- like granules, and was thus presumably an open canal. A short dermal expansion forms a horizontal keel on each side of the caudal lobe just in advance of its upturned end. The largest known examples measure nearly 2 meters in length. We have said that this genus is looked upon as furnishing important «This statemenc requires to be amended, Dy Dean having discovered at least one specimen in which the anterior dorsal fin is provided with a powerful Ctenacanthus like spine. 56 NEW YORK STATE MUSEUM support for the doctrine of the evolution of fins, which now ranks as a fairly well demonstrated principle. In the first place, it is important to bear in mind that the paired fins of Cladoselache are the oldest known structures of the kind which are clearly observable; and secondly, they approach more closely than any others to the hypothetical primitive type from which all paired limbs have been derived. Briefly stated, the lateral fin fold theory assumes that fishes originally possessed on each side of the body a con- tinuous fold of skin, strengthened by parallel cartilaginous rods extending outwards from the body wall, this fold becoming subdivided into the pairs of pectoral and pelvic (ventral) fins as we know them in modern forms. Now it has been shown by Dr Bashford Dean that in Cladoselache the paired fins were mere balancers with a more extended base line than is usual. The series of parallel cartilaginous rods, which in a primitive con- dition supported the lateral fin fold, exist practically unmodified in the pelvic fins, simply clustered and partly fused within the body wall of the pectoral fins. Dean, Cope, and others are of the opinion that there is a tendency in the pectoral fin for the hinder end of the row of basals to rotate outwards—a process which would reduce the point of.attachment of the fin to what was originally its front angle. The outwardly turned row of basals would in this case correspond with the median axis of the well known paddle in Ceratodus and Pleuracanthus, and one may without diff- culty conceive of a fringe of cartilaginous rays developing quite secondarily along the hinder border of this axis. Hence, as argued by Dean and Cope, the modern tribasal or dibasal shark’s fin can not have evolved from the paddlelike ‘‘archipterygium,” but these two kinds of fin must have arisen independently from the “ ptychopterygium,” as the arrangement has been appropriately termed by Cope." ‘For interesting discussions of this subject, one may consult the following: Mollier, S. Die paarigen Extremitaten der Wirbeltiere. I. Das Ichthyopterygium. Anatom. Hefte, 3: 1893; Dean, B. Contributions to the Morphology of Cladoselache. Jour. Morphol. 1894. 9: 87-114; A new Cladodont from the Ohio Waverly. New York Acad. Sci. Trans. 1894. 13: 115-119; The Fin-fold Origin of the Paired Limbs, in the Light of the Ptychopterygia of Palaeozoic Sharks. Anatom. Anzeig. 1896. 11: 673-79; Woodward, A. S. The Problem of the Primeval Sharks. Nat. Sci. 1895. 6: 38-43; /drd. 1892. 1: 28-35. DEVONIC FISHES OF THE NEW YORK FORMATIONS 57 Cladoselache sp. ind. Plate 8, figure t The few species of Cladoselache that are known have been found only in the Cleveland shale (Upper Devonic) and Waverly (Lower Car- bonic) of Ohio. This statement is made, however, without taking into account the numerous teeth known as Cladodus, some of which may possi- bly belong to the Pleuropterygii. Under these circumstances it is interest- ing to note the undoubted occurrence of the genus in the Portage beds of western New York, as indicated by a well preserved pectoral fin belonging to the Buffalo Society of Natural Sciences. This was obtained a few years ago by Mr F. K. Mixer, of Buffalo, from a concretion in the Naples shale of Eighteen Mile creek, near the shore of Lake Erie, and was by him sub- mitted to the present writer for examination. Later it was placed in the hands of Dr Dean for study, who has promised a more detailed description of it in connection with newly discovered Ohio material. We may there- fore content ourselves at present with a mere indication of the occurrence of Cladoselache in a new horizon and locality, and with remarking upon the decidedly primitive condition of the fin. This is indicated by its acutely triangular, almost spinelike form together with the lack of crowding and concentration of the numerous rays. The apex is more acutely pointed than in any form hitherto described, and the radials, upwards of 70 in num- ber, apparently extend inside the body wall. Characters of this nature are regarded by Dean" as suggestive of Acanthodian affinities, but it appears tolerably certain that the two types of fins can have but a very remote connection. The reasons for the latter view are thus stated by Dr A. S. Woodward :? In conclusion, one word of protest against the American idea that the paired fins of Cladoselache can be compared with those of an Acanthodian. We venture to maintain that these fins are fundamentally different in every respect. In Cladoselache the cartilages of the internal skeleton are well developed and support the whole fin membrane; in Acanthodians, what- *The Fin-fold Origin of the Paired Limbs, in the Light of the Ptychopterygia of Palaeozoic Sharks. Anatom. Anzeig. 1896. 11: 678. 2 The Problem of the Primeval Sharks. Nat. Sci. 1895. 6: 42. 58 NEW YORK STATE MUSEUM ever view we may adopt as the naming of the parts, these cartilages are as much reduced as in a modern herring. Dr Dean speaks of the “ radials” of Cladoselache as if, by fusion, they might readily become a fin spine like that of the Acanthodian Parexus; but the former are cartilage and endo- skeletal, the latter is merely the ordinary dentine and therefore presumably exoskeletal. The problem of the primeval sharks continues to present endless difficulties, but these are only multiplied by such comparisons. In the present writer's opinion, the pectoral of Cladoselache is more remotely connected with that of the Acanthodians than is that of a modern Siluroid with the pectoral of the Devonic Holoptychius. Everything still tends to show that the very highest Elasmobranchs lived simultaneously with almost the lowest in late Paleozoic times; while sharks and skates nowadays are a comparatively degenerate race. Formation and locality. Naples shale (Portage beds); Eighteen Mile creek, near Buffalo, N.Y. A tooth of an undetermined species of Cladodus is also recorded by J. M. Clarke,' from the Chemung beds at High Point (High Point sandstone), near Naples, N. Y. A mass of confused dermal plates, or shagreen, with prominent denticulations and possibly referable to Cladoselache has been obtained by Dr Clarke from the Marcellus shale of Leroy, N. Y., the original being preserved in the State Museum. Order ICHTHYOTOMI This order is represented in the Devonic only by detached teeth similar to those occurring in well known Pleuracanthid genera, and consist- ing of two or more sharp cusps attached to broad bases. No satisfactorily preserved remains, other than teeth, are known from an earlier period than the Lower Carbonic, and the most complete are of Permian age. It is customary to refer to this order most of the detached teeth known as Cladodus, although it is certain that teeth of the same form were common to several primitive genera— perhaps even to more than one family and order. Family PLEURACANTHIDAE Genus DIPLODUS Under this name are comprised a number of Paleozoic species known only by the evidence of detached teeth, hence the status of the genus is "U.S. Geol Sur. Bul. 16. 1885. p. 72. DEVONIC FISHES OF THE NEW YORK FORMATIONS 59 more or less provisional. This fact being understood, it is convenient to retain the generic title pending such time as our knowledge of the entire organization remains sub judice, notwithstanding the prior employment of Diplodus amongst bony fishes (porgies, bass). S. A. Miller, in his work on North American Geology and Palaeontology, has proposed its replacement amongst fossils by Dissodus; and Dr O. P. Hay prefers the earlier synonym of Dittodus. Diplodus priscus Eastman Plate 1, figures 7, 8 1899 Diplodus priscus C. R. Eastman. Jour. Geol. 7: 490, pl. 7, fig. 1, 2 1899 Diplodus priscus S. Weller, Jour. Geol. 7: 484 Teeth minute; the two principal cusps of dental crown divergent and slightly inclined backward, robust, conical, round in section, without lateral carinae; coronal surface marked with relatively few, prominent, slightly curved striae extending from the base nearly to the extremities on the anterior face, but shorter and usually fainter on the posterior face. Median denticle slender, sometimes much reduced, or in one specimen wanting altogether. Anterior border of root slightly produced downward; lower surface concave, elliptical in outline; posterior button developed. This species, together with the one immediately to be described, is known by a number of examples of striated teeth from a peculiar deposit of the Upper Devonic in the vicinity of Elmhurst, Ill. As described by Dr Stuart Weller, who first noted the unconformable relations of the beds, the Devonic fossils are deeply buried in fissures of Niagaran limestone, appearances indicating that the joints were open, and became filled with sand and fossiliferous remains during the late Devonic. The sort of unconformity presented by this singular occurrence has also been observed and described by Clarke at Buffalo, N. Y. between the uppermost Siluric and Oriskany sandstone. It is designated by Dr Weller as ‘subterranean unconformity.” Formation and locality. Upper Devonic; Elmhurst, III. 60 NEW YORK STATE MUSEUM Diplodus striatus Eastman Plate 1, figures 10, 1 1899 Diplodus striatus C. R. Eastman. Jour. Geol. 7: 490, pl. 7, fig. 3, 4 1899 Diplodus striatus S. IVeller. Jour. Geol. 7: 484 Of this species only a few fragmentary specimens were obtained by Dr Weller from the same locality as the last, the largest and most perfect being represented in plate 1, figure 10. It attains apparently about twice the size of the preceding form, and is distinguished from it by its finer striation, shallower root, and somewhat compressed section of its principal cones. The striae on the anterior face all curve uniformly in a spiral direction, but on the posterior face the tendency is to curve out- ward on either side of the median line to the lateral margin of the cones, where they terminate, exactly as in some species of Cladodus. None of the specimens show the full length of the median denticle, but it was apparently long and slender. Formation and locality. Upper Devonic; Elnihurst, Ill. Genus PHOEBODUS St John & Worthen Teeth differing from those of Diplodus in having three principal cones of about equal size, and from one to three very small intermediate cones. Phoebodus politus Newberry Plate 1, figure x2 1889 Phoebodus politus /. S. Mewberry. U.S. Geol. Sur. Monogr. 16: 173, pl. 27, fig. 27-28a 1899 Phoebodus politus C. &. Zastman. Jour. Geol. 7: 491, pl. 7, fig. 5 Newberry’s original description of this rare and interesting form is as follows: Teeth small, robust, breadth between tips of lateral cusps 6 to 12 mm, hight from 4 to 8 mm, base broadly elliptical, thick, with a broad bilobed, padlike prominence in the middle portion of the upper surface, concave below, with a narrow arch between the cusps; cusps three, of nearly equal size, with minute rudimentary ones in the angles between them, all strongly recurved, flattened in front with salient, acute angles, rounded behind; surface smooth and polished, or bearing a few short, coarse striations. DEVONIC FISHES OF THE NEW YORK FORMATIONS 61 The relations of this species have been commonly supposed to be with Cladodus, rather than with Diplodus, but recent discoveries of more perfect examples which display the characters of the root, in this and other species, leave no doubt as to the Pleuracanthid nature of these teeth. An illustra- tion is given in the accompanying plates of one of the most perfect speci- mens of P. politus that has yet been obtained. The original is preserved in the Museum of Comparative Zoology at Cambridge, Mass. Formation and locality. Cleveland shale (Upper Devonic) ; Ohio. Family pLADODONTIDAE An indefinable family, apparently closely related to the Pleuracanthidae. Genus cLapopus Agassiz This typically Carbonic genus occurs sparingly in the Neodevonic, but only two instances are known of its occurrence in the Mesodevonic, the species immediately to be described being found in the Columbus limestone of Ohio, and the next following from the Hamilton of Milwaukee, Wis. It is quite possible that some detached teeth of similar nature which range upward into the Carbonic belong in reality to Pleuracanthid sharks, their origin being traceable to forms like Protodus and Doliodus in the Lower Devonic. There are strong reasons for believing that the teeth of Lower Carbonic Cladodonts belonged to spineless sharks, and until recently it was supposed that the Devonic Cladoselache was also unarmed. Cladodus prototypus sp. nov. Plate 3, figure 15 Founded upon a single robust tooth, the crown consisting of a stout erect median cone and five lateral denticles on either side, not much com- pressed, and all delicately striated. The outer pair of lateral denticles is much the largest, and nearly circular in cross-section. The median cone is elliptical in transverse section, slightly recurved, without sharp lateral edges, and very wide across the base. Its total hight, when complete, is estimated to have been about 3 cm. The general appearance of this tooth is suggestiveof Cladodus 62 NEW YORK STATE MUSEUM striatus Agassiz, from the British Lower Carbonic limestone.*’ The present example differs from the latter species, however, in the lesser com- pression of the crown, fewer lateral cones, and peculiar arrangement of the coronal striae. As many as 35 fine parallel nonbifurcating striae are visible on the external face of the principal cone, those of the middle portion run- ning vertically, and those along the sides curving gradually outwards and terminating in a faint ridge along the lateral margin. The apexes of the median and nearly all of the lateral cones have unfortunately been broken away in the type specimen, and the root, also, is wanting. The width across the base line, in its present condition, is 3.6 cm. The unique specimen answering to the above description forms part of the collection purchased from the late Professor James Hall by the American Museum of Natural History in New York, and bears the museum catalogue number of 4257. It is labeled as having been derived from the ‘“Corniferous” (Columbus) limestone in the vicinity of Columbus, O., and the characteristic appearance of the matrix leaves no doubt as to the authenticity of the record. Its geological antiquity is greater than that of any other known species. Formation and locality. Columbus limestone (Ulsterian) ; Columbus, O. Cladodus monroei Eastman Plate 1, figure 5 1900 Cladodus monroei C. &. Lastman. Jour. Geol. 8: 36, text fig. 2 The type of this species is a small, imperfectly preserved tooth found by Mr Charles E. Monroe in the Hamilton limestone of Milwaukee, Wis. The median cone is robust, very thick at the base, and indistinctly striated. The external denticles are also stout in proportion to the size of the prin- cipal cone, but the three intermediate denticles of either side are excessively small. The total hight of the median cone probably amounted to less than 1.5 cm, and the width across the base 2.5 cm. Formation and locality. Hamilton limestone (Erian); Milwaukee, Wis. « The teeth recently described under this name by E. B. Branson from the Salem lime- stone of Indiana are clearly distinct, and require a new specific name. DEVONIC FISHES OF THE NEW YORK FORMATIONS 63 Cladodus coniger Hay 1889 Cladodus carinatus /. S. Newberry. U.S. Geol. Sur. Monogr. 16: 103 1899 Cladodus coniger O, P. Hay. Am. Nat. 33: 783 In the meager description given of this species by Newberry, the teeth are stated to be “less than half an inch in breadth and hight, the base narrow, and bearing one central and four lateral cones, the exterior pair larger than the intermediate ones, but all much lower than the central denticle.” The distinctive feature of the species is said to consist in four relatively strong carinetions on the flattened posterior surface of the principal cone. Teeth evidently belonging to this species are by no means uncommon in the Chemung of Warren county, Pennsylvania, but as a rule are imper- fectly preserved. The general form is similar to that of C. concinnus Newberry, from the Cleveland shale of Ohio; and, as in that species, the pair of intermediate denticles is sometimes absent. The external cones also diverge outwards at a considerable angle. The principal cone is strongly compressed, more or less recurved, with sharp lateral edges, and prominently striated on both faces. The number of carinae along the flattened, moderately convex anterior face is not limited to two pairs, how- ever, some specimens displaying twice that number. These are symmet- rically arranged on either side of the median line, the two innermost being nearly vertical and rather more prominent than the rest. Formation and locality. Chemung beds (Chautauquan); Warren, Pa. Cladodus sp, ind. A single specimen of Cladodus, associated with other undetermined fish remains, is reported by Dr John M. Clarke from the Chemung fauna (High Point sandstone) at High Point, a mountain situated about 3 miles northwest of the village of Naples, N. Y. |U.S. Geol. Sur. Bul. 16. 1885. p- 72] ; Order ACANTHODII Two families of Acanthodian fishes are known, one characterized by the presence of a single, the other by two dorsal fins, and both having 64 NEW YORK STATE MUSEUM representatives in the American Paleozoic. The Acanthodidae range from the Lower Devonic to the Lower Permian inclusive, the Diplacanthidae are confined to the Upper Siluric and Lower Devonic. The detached fin spines known as Machaeracanthus, Gyracanthus, and similar forms are regarded with much probability as belonging to fishes of this order. Acanthodes ? pristis Clarke 1885 Acanthodes ? pristis /. M. Clarke. U.S. Geol. Sur. Bul. 16, p. 42 1891 Acanthodes ? pristis A. S. Woodward. Cat. Foss. Fishes Brit. Mus. pt 2, P. 15 1902 Acanthoéssus? pristis O. P. Hay. U.S. Geol. Sur. Bul. 179, p. 273 Founded upon a considerable portion of the shagreen of an undoubted Acanthodian fish, the family position of which, owing to the nonpreserva- tion of fins, is uncertain. In the solitary specimen known, the greater number of scales have retained their natural, exceedingly compact arrange- ment, and are so minute as to indicate a species of very diminutive size. Their characters are given in the original description as follows. “ The scales are very small, measuring .5 mm on the edge, square or slightly sub- rhomboidal in outline, and one fourth as thick as wide. The adjacent edges at about two thirds the distance from the upper surface are strongly grooved by a single deep furrow. The upper surface of the scales is smooth and slightly convex.” The type specimen, which belongs to the United States National Museum, was obtained by Dr J. M. Clarke from the bituminous layers of the Portage beds (Rhinestreet shale) exposed in a Delaware and Lacka- wanna railroad cutting near the town of Sparta, N. Y. In Woodward's and Hay’s catalogues, cited in the above list of references, the species is erroneously reported as from the Genesee shales of New York in which Palaeoniscid scales of the same character are known to occur. Order SELACHII To this order, which includes the sharks and skates, numerous Paleo- zoic teeth and fin spines have been ascribed, but only on account of their close resemblance to the corresponding hard parts of recent forms, So far DEVONIC FISHES OF THE NEW YORK FORMATIONS 65 as dependence can be placed upon analogy of this nature, forms allied to the modern Port Jackson shark (Cestracion or Heterodontus, text fig. 15) were present during the Devonic, and extremely abundant during the Lower Carbonic. Tamiobatis, known by a unique skull from the Devonic of Kentucky, is interesting in that, although probably a shark, its form was decidedly raylike. No true rays, however, appear earlier than the Mesozoic. Entire skeletons of undoubted Selachii are known first from the Lower Lias. As for the detached dermal spines and tubercles occurring in various Paleozoic horizons, we are without means for determining their precise relations, hence they are conveniently grouped together under the head of ‘“Ichthyodorulites.” These are treated in the present contribution immediately after the Holocephali. Fig. 13 Port Jackson shark, Cestracion philippi (female). Recent; Australia. x 1-10 (From Dean, after Garman) Subclass HOLOCEPHALI The following general remarks on the Holocephali and Chimaeroids, taken chiefly from Smith Woodward, may profitably find a place here: Dental plates essentially similar to those of the existing Chimaeroid fishes are met with in rocks as early as the Middle Devonic; but there is still no evidence of any member of the Holocephali which can not be included in the surviving order of Chimaeroidei. Some of the early forms were certainly armed with dermal plates; but paleontology as yet lends no support tor the theory that the Chimaeridae are degenerate descendants of fishes once possessed of membrane bones. The earliest known complete skeletons are unfortunately only Liassic. 66 NEW YORK STATE MUSEUM Order CHIMAEROIDEI In all known Chimaervoids, whether recent or extinct, the notochord is persistent and at most only partially constricted, the calcifications in the sheath, when present, consisting of slender rings more numerous than the neural and haemal arches. The pectoral fins are abbreviate, without seg- mented axis; and the pelvic fins in the male are produced into a pair of claspers. In the extinct forms there is no trace of any dermal plate developed in the opercular flap. The only clear evidence of evolution hitherto observed concerns the development of the peculiar dental plates, In each of the four known families the dentition consists of a few large plates of vascular dentine of which certain areas (‘‘tritors”) are specially hardened by the deposition of salts within and around groups of medullary canals, which arise at right angles to the functional surface. In most cases there is a single pair of such plates in the lower jaw, meeting at the sym- physis, while two pairs (the so-called vomerine and palatine plates) are arranged to oppose these above. A permanent pulp remains under each plate, and growth thus takes place continually within as the oral surface is triturated by wear. In the surviving family of Chimaeridae these dental plates are much thickened, while the hinder upper pair (“ palatines””) are both closely apposed in the median line and considerably extended backwards. The dental plates named Ptyctodus, from the Devonic of Russia, the Eifel district and North America, are essentially similar to those of modern Chimaeroids, but there are no representatives of the vomerine pair. The tritors, one only in each plate, are well differentiated, consisting of hard, punctate, superimposed laminae, arranged obliquely to the functional sur- face. The contemporaneous teeth known as Rhynchodus and Palaeomylus, however, exhibit more indefinite tritoral areas, or none. The symphysial facet is always distinct. Spines which may be compared with those of Chimaeroids are also known from the Devonic and Carbonic systems, and Harpacanthus and Cyrtacanthus may perhaps be cited as examples of head spines. No DEVONIC FISHES OF THE NEW YORK FORMATIONS 67 Chimaeroid skeletons, however, have hitherto been satisfactorily deter- mined from Paleozoic rocks, save for the possible exception of Menaspis. The problematical remains known as Dictyorhabdis priscus, from the Lower Siluric of Colorado, are very doubtfully of Chimaeroid nature. The most recent discussion of the relations between fossil and living Chimaeroids is that of Dr Bashford Dean, in the publications of the Car- negie Institute of Washington, 1906. Some changes in the classification are proposed by Mr C. Tate Regan in the Proceedings of the Zoological Society of London for the same year. Family PTYCTODONTIDAE A family at present indefinable, of doubtful ordinal position, known only by remains of the dentition and possible dermal ossifications. A single pair of large, laterally compressed dental plates present in each jaw, united at the symphysis, and either with trenchant oral margin, or develop- ing one or more tritoral areas posteriorly. Rhynchodus undoubtedly rep- resents the most primitive condition of dental plates, the tritors Ptyctodus and Palaeomylus arising subsequently, and indicating a more specialized stage, Genus RuHyNcHODUs Newberr y Oral margin of dental plates simply trenchant, without tritoral areas. Upper and lower dental plates of similar form, except that in some species the symphysial margin is produced downwards into a spiniform process. Indications of cartilaginous union between each pair of dental plates are present on the inner symphysial facets. Rhynchodus secans Newberry 1873 Rhynchodus secans /. S. Mewberry. O. Geol. Sur. Rep’t. v. 1, pt 2, p. 310, pl. 28, fig. 1; pl. 29, fig. 1, 2 1889 Rhynchodus secans J. S. Mewberry. U.S. Geol. Sur. Monogr. 16: 47, pl. 28, fig. 1-3 1898 Rhynchodus secans C. &. Eastman, Am. Nat. 32: 485, 546 This species, which is the type of the genus, is not uncommon in the Columbus and Delaware limestones cof Ohio and is interesting for having 68 NEW YORK STATE MUSEUM furnished a group of four teeth preserved in natural association, There is but little difference in the form of upper and lower dental plates, and both terminate anteriorly in prominent beaks. It is probable that the latter character is a generic one, it having been observed in several other species as well, but never in Ptyctodus. An excavation is noticed just back of the beak in the upper dental plate where the terminal point of the lower came in contact with it, thus proving that the relations between the two pairs were the same as in Ptyctodus. The lower dental plate is deeper than the upper, and marks of wear are frequently observed along the inner side of the func- ticnal margin. Rhynchodus excavatus Newberry 1877 Rhynchodus excavatus /,8. Mewberry. Geol. Wis. 2: 397 1878 Rhynchodus excavatus /. 8S. Newberry. N.Y. Acad. Sci. Ann. 1: 192 1878 Rhynchodus occidentalis /. S. Mewberry. N. Y. Acad. Sci. Ann. I: 192 1889 Rhynchodus excavatus /. S. Mewberry, U.S. Geol. Sur. Monogr. 16: 50, pl. 29, fig. 1 1898 Rhynchodus excavatus C. &. Lastman. Am. Nat. 32: 486 1898 Rhynchodus occidentalis C. &. Lastman, Am. Nat. 32: 485 A comparison of the type specimen of the so called R. occi- dentalis, which remains as yet unfigured, with an extensive series of dental plates belonging to the earlier described R. excavatus, leads to the conclusion that the two forms are identical. Indeed, it is certain that Newberry would himself have recognized their identity, had he been acquainted with more perfect specimens, such as are now accessible. Any one may satisfy himself on this point who will take the trouble to verify the original description of R. occidentalis, which reads as follows: Teeth of small size, much compressed. Anterior margin slightly curved, but nearly vertical. Superior margin gently arched downward from the prominent anterior point, forming a much compressed triturating surface or edge. Posterior portion of upper margin acute edged. Exterior lateral surface striated obliquely backward. Basal margin formed by the edges of external and internal laminae, of which the edges are broken and irregular. From the Hamilton limestone, Waverly, la. DEVONIC FISHES OF THE NEW YORK FORMATIONS 69 Little need be added by way of supplementing the above description. The total length of the dental plates rarely exceeds 5 cm, and there is close similarity between those of the upper and lower jaws, excepting that the symphysial margin of the lower is produced into a long and slender descend- ing spine, and the cutting edge of the upper is somewhat less arched or “excavated” than in the lower dental plates. As shown by marks of wear, the tips of the lower dental plates closed outside and slightly behind those of the upper, in the same manner as in Ptyctodus. Dental plates differing but little from those of the present species have been described from the Eifel Devonic, the complete dentition being known in the so called “ Ram- phodus” [— Rhynchodus major] of Jaekel.t_ An extensive series of specimens, collected many years ago by Mr Orestes St John from the Cedar Valley limestone of Iowa, is now preserved in the Museum of Com- parative Zoology at Cambridge. Formation and locality. Hamilton limestone; Milwaukee, Wis., and Cedar Valley limestone of Iowa. Rhynchodus pertenuis Eastman Plate 2, figure 5 1904 Rhynchodus pertenuis C. &. £astman. Am. Nat. 38: 297, text fig. 2 Dental plate narrow and elongate, with sharp and extended functional margin and knife blade cross-section; anterior beak prominent, no sym- physial spiniform process, external surface smooth. The unique dental plate upon which this species is founded was obtained from the Chemung of Franklin, in Delaware county, New York, and is now preserved in the State Museum at Albany. The general out- line and proportions of this plate differ from those of all other species, and the absence of a spiniform symphysial process is a very unusual feature. But for the trenchant functional margin and narrow cross-section, the specimen might readily be mistaken for a lower dental plate of Ptyctodus, instead of Rhynchodus. That it is properly a mandibular element, and *Jaekel, O. Ueber Ramphodus, etc. - Sitzungsber Ges. naturf. Freunde, Berlin, 1903. p- 383-93. Cf also Am. Nat. 1904. 38: 296; Centralblatt ftir Mineral. 1900. p. 177. 70 NEW YORK STATE MUSEUM referable to the latter genus, seems to admit of no question. The hollow along the base indicates the extent to which the plate was inserted in the supporting cartilage of the jaw. The total length is 9 cm. Formation and locality. Chemung beds; Franklin, N. Y. Rhynchodus sp. ind. Dental plates of an undetermined species of Rhynchodus are reported by Dr J. M. Clarke* as occurring in the High Point (Chemung) fauna near Naples, N. Y. Genus prycropus Pander Oral surface triturating, the single tritoral area of each dental plate well differentiated, and consisting of hard, punctate, superimposed laminae arranged more or less obliquely to the functional surface. Lower dental plates with symphysial beak, which, as shown by marks of wear, closed against the outer margin of the upper dental plates. Ptyctodus punctatus sp. nov. This species is readily distinguished from all others previously described by the comparative coarseness and peculiar arrangement of the medullary canals passing through the tritoral area. The sides and functional surface of the latter are completely covered by a network of small polygonal pit- tings, formed by the termini of the medullary canals. As these are not grouped in parallel lines, there is no surface indication of laminar structure. The simpler arrangement of the medullary canals in this species, taken in connection with its greater antiquity as compared with other known species, Mig: 158 Phy “todee punctate suogests that it is a primitive character. Formation and locality. The single detached tritor upon which the above description is founded was obtained from the Onondaga limestone (Ulsterian) of Leroy, N.Y. The type specimen is preserved in the Museum of Comparative Zoology at Cambridge, Mass. and another example, from the corresponding rocks of Ohio, is in the collections of the American Museum of Natural History. U.S. Geol. Sur. Bul. 16. 1885. p. 72. DEVONIC FISHES OF THE NEW YORK FORMATIONS 71 Ptyctodus calceolus Newberry & Worthen 1866 Rinodus calceolus Mewberry & Worthen. Pal. Ill. 2: 106, pl. 10, fig. 10 1870 Ptyctodus calceolus Newberry & Worthen. Geol. Sur. Ill. Rep’t, 4: 374 1875 Ptyctodus calceolus /. S. Newberry. O. Geol. Sur. Rep’t. v. 2, pt 2, P. 59, Pl. 59, fig. 13 1898 Ptyctodus calceolus C. R. Zastman. Ta. Geol. Sur. Rep’t. 7: 115, text fig. 10a 1898 Ptyctodus calceolus C. &. Eastman. Am, Nat. 32: 476, fig. 1-17 1899 Ptyctodus calceolus S. Weller. Jour. Geol. 7: 484 Dental plates compressed into a thin cutting edge shortly behind the symphysis, but widening gradually, becoming more or less outwardly curved, and the oral surface occupied for nearly its entire width by the tritoral area, the inner margin of which is more strongly curved than the other. Laminar structure of the tritors indicated superficially by fine punctae arranged in parallel rows running obliquely across the functional surface. The compressed edge in advance of the tritor in the lower dental plate slopes rapidly upward and terminates in a strong anterior beak. Upper dental plates similar to the lower, except that the symphysial border is rounded and not produced into a beak. In spite of the extraordinary abundance of this species in various Mid- dle and Upper Devonic localities, as indicated by detached tritors, complete dental plates are very rare, their structure being on the whole very fragile. Of widespread distribution in the Hamilton, the species is most prolific in the Upper Devonic State Quarry beds near Iowa City, a single exposure only a few yards square having yielded thousands of abraded tritors. No specimens have been hitherto reported from New York State, nor has the genus itself been known to occur in rocks of earlier age than the Hamilton in this country. Recently, however, the writer has had the opportunity of examining a nearly perfect lower dental plate, evidently of this species, which was collected by Mr Edgar E. Teller from the Genesee shale at Eighteen Mile creek, near Buffalo. The precise level at which this dental plate was found is from near the base of the Conodont bed, the exposure 72 NEW YORK STATE MUSEUM being referred to as “section H” and photographed in plate 4 of Grabau’s Geology and Palacontology of Eightcen-Mile Creek.‘ The specimen in ques- tion is excellently preserved, measures 3 cm. in length, and lacks only the extreme tip of the symphysial beak. A single large dental plate, appar- ently referable to this species, has also been obtained by Dr John M. Clarke from the Goniatite or Parrish limestone (Portage) near Naples, N.Y. Formation and locality. Middle Devonic of Iowa, Illinois, Missouri, Wisconsin and Manitoba. Upper Devonic of Iowa and Elmhurst, Ill. Genesee shale of Eighteen Mile creek, and Portage of Naples, N. Y. Ptyctodus compressus Eastman 1898 Ptyctodus compressus C. &. Eastman. Am, Nat. 32: 479, fig. 18-27 The tritors in this species are relatively longer and narrower than in the preceding, and the oral margin in advance of the triturating surface is developed into a long, sharp cutting edge. In all other species this trench- ant margin is shorter than the tritoral area, but in the present form it is invariably longer, sometimes exceeding the length of the tritor by one rourth. The dental plates are as a rule less curved than those of P. calceolus, and the symphysial beak less produced. Formation and locality. State Quarry beds (Upper Devonic) ; North Liberty, Johnson county, Ia. Ptyctodus ferox Eastman 1898 Ptyctodus ferox C. R&. Eastman. Am. Nat. 32: 480, fig. 35-40 1899 Ptyctodus ferox C. R. Eastman, Jour. Geol. 7: 282 Dental plates unusually large and massive, attaining a total length of about 14 cm, and exhibiting but slight lateral curvature. Lower dental plate with a stout symphysial beak, the front margin projecting downward into a long spiniform process, evidently for strengthening the symphysial union. Anterior margin of upper dental plate uniformly rounded, not pro- duced into a beak or spiniform process, and showing on the outer face * Buffalo Soc. Nat. Sci. Bul. 1898. 6:5. DEVONIC FISHES OF THE NEW YORK FORMATIONS 73 marks of contact with the opposing dentition. Inner surface of both upper and lower dental plates with a roughened triangular symphysial facet. The fact that the smaller and younger plates belonging to this species have a decidedly Rhynchoduslike aspect indicates that the functional mar- gin in immature teeth was simply trenchant, the tritors not being developed until a comparatively late period. Rhynchodus, therefore, represents a more primitive stage than either Ptyctodus or Palaeomylus in the develop- ment of dental plates, its relations to contemporary types being closely par- alleled by Rhinochimaera amongst recent genera. There is considerable reason for believing that Ptyctodonts were provided with dermal ossifica- tions, and in particular it is difficult to avoid the suspicion that the fin spines known as Heteracanthus or “ Gamphacanthus” were borne by mem- bers of this family, possibly by the very species under discussion. The dental plates of P. ferox are accompanied not only by dermal spines, but by tuberculated dermal plates not unlike those associated with Myriacan- thus. One of these is illustrated in plate 1, figure 6, of the present memoir. As yet, however, no specimens have been found which demon- strate the supposed specific identity of these various bodies. Formation and locality. Hamilton limestone (Erian) ; Milwaukee, Wis. Cedar Valley limestone (Mesodevonic) of Iowa and Illinois. State Quarry beds (Neodevonic) ; Johnson county, Iowa. ICHTHYODORULITES Genus oncHus Agassiz Spines of small size, laterally compressed; sides of exserted portion ornamented with smooth or faintly crenulated longitudinal ridges; no posterior denticles. Represented in this country by two Upper Siluric (O. clintoni and O. pennsylvanicus) and one Devonic species. An undescribed form is also known from the Niagaran of Cumberland, Md. : Garman, S. Chimaeroids, especially Rhinochimaera and its Allies. Museum Comp. Zool. Bul. 1904, 41: 246. Dean, B. Chimaeroid Fishes and their Development. Carnegie Inst. Wash. 1906, Pub. 32, p. 126. 74 NEW YORK STATE MUSEUM Onchus rectus Eastman Text figure 16 1899 Onchus rectus C. &. Eastman, N.Y. State Geol. 17th An. Rep’t, p. 323, fig. 4 Spines attaining a total length of about 5 cm, nearly rectilinear, and gradually tapering to an acute point. Inserted portion round in cross-section, very delicately striated ; exserted portion laterally com- pressed, without trace of posterior denticles, the sides traversed by about so fine longitudinal ridges. The latter are nonbifurcating, regularly spaced, and of uniform size with the exception of the one along the anterior margin; this is twice the width of the others, and is of triangular cross-section. Formation and locality. Chemung group; Ontario and Western Railroad tunnel between Merrickville and North Walton, Delaware co., N. Y. Genus HOMACANTHUS Agassiz This genus, which is evidently closely allied to Ctenacanthus, is thus defined by Smith Woodward: ‘ Dorsal fin spines of small size, slender, more or less arched, laterally compressed, and gradu- ally tapering distally; sides of exserted portion ornamented with few, large, smooth, widely spaced longitudinal ridges; a similar ridge also forming a large anterior keel; posterior face with a double series of large, downwardly curved denticles.” American Fig. 16 Onchus species that have been referred to Homacanthus have, with but rectus Eastm. : b Daatfn tWO exceptions, since been removed to other genera, but at least Pine. , one Devonic representative of this genus seems to be indicated by Nat. size the spines described in the following paragraph. DEVONIC FISHES OF THE NEW YORK FORMATIONS 75 Homacanthus acinaciformis Eastman Plate 1, figure 16 1903 Homacanthus acinaciformis C. &. Zastman. Mus. Comp. Zool. Bul. 39: 218, pl. 5, fig. 58 Spines comparatively small, slender, gradually tapering, gently and uniformly arched; lateral surface with five or six continuous longitudinal ridges ; posterior denticles slender, rather widely spaced. The small spines known under this designation are more strongly curved than those found in the Lower Carbonic limestone, but according to J. W. Davis, there is considerable variation in this respect, even amongst spines belonging to a single species. Some resemblance is to be noticed between the present form and the Upper Devonic spines from Ohio described as Hoplonchus parvulus by Newberry. Formation and locality. Chemung group; Warren, Pa. Genus CTENACANTHUS Agassiz Dorsal fin spines robust, often attaining a large size, laterally com- pressed; sides of exserted portion ornamented with longitudinal ridges, usually crenulated or denticulated, rarely smooth; posterior face flat or concave, with a series of small denticles upon each margin. It is certain that spines of this character were possessed by sharks of more than one genus. Reference has already been made to the occurrence of these defenses in Cladoselache, and Newberry was strongly inclined to suspect, from a discovery in the Waverly of Ohio, that Ctenacanthus and Orodus may be synonymous. Of primary importance in the distinction of species is the general conformation of the spines, especially their curvature, form of cross-section, and length of inserted portion. Next in order are to be considered the number, shape and direction of the longitudinal costae, with the special ornamentation of the same; and still further distinctive characters are to be found in the nature of the posterior face and anterior margin, or “cutwater.” A class of spines agreeing in their abbreviate, :O. Geol. Sur. Rep’t 1874. v. 2, pt 2, p. 54. 76 NEW YORK STATE MUSEUM stumpy proportions, and inserted very obliquely in the integument, is prob- ably to be correlated with the posterior dorsal fin. They contrast strongly with the group of slender, elongated and tapering spines which were undoubtedly situated in advance of the first dorsal fin.’ Ctenacanthus wrighti Newberry 1884 Ctenacanthus wrighti /. S. Mewberry. N. Y. State Mus. 35th Rept, p. 206, pl. 16, fig. 12-14 1889 Ctenacanthus wrighti /. S. Newberry. U.S. Geol. Sur. Monogr. 16: 66, pl. 26, fig. 4 An examination of the peculiarly shaped and unusually large fin spine upon which this species is founded, now preserved in the American Museum of Natural History in New York, has served to confirm the correctness at all points of Newberry’s description, which is as follows : Spine of large size, long triangular in outline ; anterior margin straight, laterally compressed ; medullary cavity large, open posteriorly to the middle of the spine; posterior face traversed above by a strong rounded ridge; [posterior | denticles small; surface of exposed portion entirely covered with pectinated ridges of nearly uniform width on the front and sides, becoming narrower and less distinctly pectinated near the posterior margin. The spines of this species are very striking in their characters as regards both form and markings. The anterior margin seems to have been abso- lutely straight from base to summit. Along the line of junction between the enameled and buried portions the spine must have been 2 inches wide, but it tapered rapidly upward, terminating in a slender, acute point. The exposed surface is more completely covered with ridges similar in character, and the pectination is more crowded than in any other species known to me. In its broad base and its general and uniform ornamentation this spine has some resemblance to C. speciosus St J. & W., specimens of which have been in my hands, but the line of demarcation between the ornamented and buried portions is less oblique, showing that the spine was more erect [and hence referable to the anterior dorsal fin]; the ridges are considerably coarser and the form is straighter.. The pectination is also less oblique and closer, compared with the coarseness of the ridges. Formation and locality. Hamilton limestone (Erian), “near the middle of the Moscow shale; Cashong creek, Yates county, N. Y.” tScience. nm. s. 1901. 14: 795. DEVONIC FISHES OF THE NEW YORK FORMATIONS V7 Ctenacanthus randalli Newberry 1889 Ctenacanthus randalli 7. S. Mewberry. U. S. Geol. Sur. Monogr. 16: 105 This species, which has never been illustrated, is founded upon the proximal portion of an extremely large spine, estimated to have been at least 30 cm in length. The original description is as follows: Dorsal fin spines 12 inches or more in length by 1% inches in width at base of ornamented portion; form slightly curved backward, sides com- pressed, basal portion conical, smooth, or finely striated longitudinally ; line of demarcation between ornamented surface and base strongly marked, inclined downward and forward at an angle of 30° with the axis of the spine; ornamented surface near base formed by about 4o fine, parallel, subequal, closely crowded ridges on each side of the median line, and these bear small, rounded, closely approximated tubercles. Formation and locality. Olean conglomerate (Chemung group); near Warren, Pa. Ctenacanthus chemungensis Claypole Piate 7, figure 3 1885 Ctenacanthus chemungensis £. IV. Claypole. Am. Ass’n Adv. Sci. 33d meeting, Phila. Proc. 1884. p. 490 (name only) This name was applied by Professor Claypole without definition or illustration to small fin spines obtained by him from the Chemung of Brad- ford county, Pennsylvania, whose length was stated to be “less than half that of Ct. vetustus.” To this species may probably be referred a number of small, gently arcuate, finely ornamented spines which were collected by the late Professor Charles E. Beecher and others from the Chemung of Warren county, Penn- sylvania, examples of which are preserved in the New Haven and Cam- bridge Museums. Spines of similar nature have also been obtained from two or three localities in New York. One in the State Museum at Albany bearing the locality label 1783 is from the Chemung between Friendship and Nile, N. Y., and displays the delicate ornamentation very clearly. Another, belonging to the Museum of Comparative Zoology at Cambridge, is from the same horizon along the bank of Fall creek, near Ithaca, N. Y., and was collected many years ago by Dr Richard Rathbun. 78 NEW YORK STATE MUSEUM None of these spines appear to have exceeded 10 cm in length, and they are frequently much shorter, their form being narrow and gradually tapering. The flattened sides are covered with numerous threadlike costae, as many as twenty being observed toward the base, and these are for the most part continuous; when new longitudinal ridges are formed, they arise by implantation. The costae are finely pectinated at intervals varying from twice to three times their own width, thus giving rise to a finely punctate appearance when seen in impression, as is usually the case. Formation and locality. Chemung group; New York and Pennsylvania. Genus ACANTHOLEPIS Newberry Spines of large size, very much laterally compressed, thin walled, internally hollow, in form gently arcuate, wide at the base, and gradually tapering to an acute point. Sides of exserted portion nearly flat, orna- mented with numerous fine stellate tubercles; these are either irregularly arranged, or in some cases exhibit a tendency toward concentric arrange- ment along lines of growth. Tubercles along either side of the posterior margin enlarged so as to forma double row of denticles directed at right angles to the margin. Appearances of at least one specimen suggest that the basal portion is segmented. The spines which are provisionally included under this title’ were first confused with Oracanthus by Newberry, but subsequently interpreted by him as dermal plates or scutes of ‘‘ Placoderms,” and supposed to be of similar nature to those which have received the name of Acanthaspis. Still later, their resemblance to the triangular ichthyodorulites of Psammosteus was pointed out by Smith Woodward. The conclusion reached by the present writer is that they are dermal defenses of Chimaeroids, probably dorsal fin spines. Their extremely thin walls, as seen under low magnify- ing power in worn specimens, exhibit the structure of vasodentine, and are traversed by coarse, branching longitudinal canals. In external characters *For this designation, preoccupied amongst insects, S. A. Miller has proposed the not altogether pleasing sobriquet of “ Eczematolepis,” an example of “index-learning that turns no student pale.” DEVONIC FISHES OF THE NEW YORK FORMATIONS 79 they bear considerable resemblance to the spines described as Phlyctaena- canthus and Apateacanthus. Considering the uncertain status of these and similar “ genera” founded upon isolated fragments, it hardly appears advisable to change their provisional designations in those cases where they happen to be preoccupied. Acantholepis fragilis Newberry Plate 3, figure s 1857 Oracanthus fragilis, O. granulatus, and O. abbreviatus /. S. Newberry. Nat. Inst. Proc. n. s. 1: 126 1875 Acantholepis pustulosus /. S. Newberry. O. Geol. Sur. Rep’t, v. 2, pt 2, p. 38, pl. 56, fig. 1-6 1889 Acantholepis pustulosus /. S. Newberry. U.S. Geol. Sur. Monogr. 16: 34, pl. 31, fig. 5 1889 Eczematolepis pustulosus S. 4. Miller. North Am. Geol. & Pal. p- 586, text fig. 1098 1891 Acantholepis pustulosus 4. S. Woodward. Cat. Foss. Fishes Brit. Mus, pt 2, p. 129 Detached spines sometimes attaining a length of 25 cm and exhibiting the characters of the genus. Inserted portion not observed, but some specimens show indications of having been expanded laterally at the base, as well as produced in an anteroposterior direction. Evidence is lacking that the spines occurred in pairs, or were serially arranged. According to Professor Newberry, the exserted portion of the spine is peculiar in being composed of two pieces which are united by an oblique suture, certainly a most unusual condition amongst ichthyodorulites. Only one of Newberry’s originals, however, is known to have shown this feature, and as none have been observed elsewhere, it is perhaps open to doubt whether the exceptional spine was entirely normal in this respect. The confirmation of other specimens would certainly be desirable before admit- ting the segmented character of the base. Fragmentary spines apparently referable to this species have been col- lected by Mr F. K. Mixer and Dr H. U. Williams from the Onondaga limestone near Buffalo, and numerous examples have been obtained from 80 NEW YORK STATE MUSEUM strata of the same horizon at Leroy, N. Y. One fairly well preserved spine belonging to the New York State Museum is labeled as having been derived from a “ravine 2% miles south of Truxton Corners, East branch,” the horizon being probably the same as the foregoing. Its characters are somewhat indistinctly shown in plate 3, figure 1. Typically the species occurs in the Columbus and Delaware limestones of Ohio, but is appar- ently also represented in the Hamilton of Milwaukee, Wis. Horizon. Ulsterian; New York and Ohio. Erian; Wisconsin. Genus PHLYCTAENACANTHUS Eastman Arcuate spines of large size, with flat lateral surfaces, very broad at the base, and gradually tapering to an acute point. Ornamentation and other characters as in Acantholepis, except that the exserted portion is cer- tainly unsegmented. Inserted portion triangularly expanded, thin walled and hollow, recalling the conditions in the corresponding portion of Stethacanthus. Phlyctaenacanthus telleri Eastman 1898 Phlyctaenacanthus telleri C. &. Zastman. Am. Nat. 32:551, text fig. 49 1899 Phlyctaenacanthus telleri C. &. Lastman. Jour. Geol. 7: 283 The spines assigned to this species are scarcely distinguishable from those of Acantholepis, their form, structure and ornamentation being prac- tically identical. The exserted portion, however, has been definitely ascer- tained to consist of but a single piece, and the inserted portion (unknown in Acantholepis) appears to have been embedded in the soft parts in similar fashion as in Stethacanthus. Admitting these remains to be fin spines, it would certainly seem that a like interpretation should be applied to Acantholepis. Genus APATEACANTHUS Woodward Fin spines elongated, slender, gradually tapering, extremely com- pressed laterally ; sides ornamented with irregular series of fine tubercula- tions; posterior border with one (or two ?) close series of acute, hook- shaped, downwardly pointed denticles. The genus is apparently closely related to the preceding. DEVONIC FISHES OF THE NEW YORK FORMATIONS 81 Apateacanthus vetustus (Clarke) Plate 3, figure 5 1885 Pristacanthus vetustus J. M, Clarke. U.S. Geol. Sur. Bul. 16, p. 42, pl. 1, fig. 7 1889 Pristacanthus vetustus J. S. Mewberry. U. S. Geol. Sur. Monogr. 16: 61 1891 Apateacanthus vetustus 4. S. Woodward. Cat. Foss. Fishes Brit. Mus. pt 2, p. 119 Known:‘only by the type specimen, which is extremely imperfect, but presents characters suggestive of Chimaeroid affinities. Spine composed of an extremely thin layer of vasodentine, probably hollow internally, lateral surface nearly flat, of uncertain width, without costae, and covered with sparsely scattered stellate tubercles of extremely minute, almost microscopic size. Posterior border set with closely spaced hook-shaped denticles of relatively large size, their apexes directed downwards. It is unfortunate that the specimen which affords the only information we possess of this genus and species should have been considerably injured in fossilization, both extremities being broken away, and also a part of the sides, including the whole of the anterior margin. The width, therefore, of the lateral surface, as well as the total length of the spine, is indeterminate. However, the preserved portion indicates a spine of large size, gradually tapering, and perhaps slightly arcuate in form. In two respects the spine may be said to be quite remarkable; first, for the large size and close approximation of the posterior denticles, 14 of which remain intact; and secondly, for the rudimentary nature of the ornamentation. The tubercles, still faintly stellate, are almost imperceptible to the unaided vision, and are without regular arrangement. The flat sides, tuberculated ornament, and absence of longitudinal costae are characters which suggest affinity with the two preceding genera, hence the form may be looked upon as a late sur- vival of Mesodevonic Chimaeroids. The type specimen is preserved in the United States National Museum. Formation and locality. Upper portion of the Cashaqua shale (Port- age); near Milo, Yates county, N. Y. 82 NEW YORK STATE MUSEUM Genus HETERACANTHUs Newberry Gamphacanthus J/7/ler Spines of moderate size, broad and laterally compressed at the base, but soon becoming subtriangular in section, nearly rectilinear, and gradually tapering toward the acute apex. Internal cavity relatively large, extending nearly to the apex, and open on the posterior margin for a considerable distance. No anterior keel; posterior denticles not observed. Exserted portion ornamented with finely crenulated longitudinal ridges which become smooth and flat with wear, and are separated by fine, denticulate, intercostal grooves, Heteracanthus politus Newberry 1889 Heteracanthus politus /. S. Mewderry. U. S. Geol. Sur. Monogr. 16: 66, pl. 21, fig. 4, 5 1892 Gamphacanthus politus S. 4. Afiler. North Am. Geol. & Pal. p. 715 1898 Heteracanthus politus C. R&R. Zastman. Am. Nat. 32:552 1899 Heteracanthus politus C. R. Zastman, Jour. Geol. 7: 282 Spines attaining a total length of abont 20 cm, very broad at the base, and with Ctenacanthuslike ornamentation. The longitudinal ridges, which are rather numerous and closely apposed, become perfectly smooth when worn, their presence being indicated only by the fine and deep intercostal grooves — the so called “sinuous or denticulate longitudinal sutures” of Newberry. More or less variation in size and number of the costae is to be observed amongst different examples. The basal portion seems to be regularly expanded, without forming an asymmetrical “shoulder” as in Physonemus, Stethacanthus etc. Bilaterally symmetrical as these spines undoubtedly are, their position must have been in the median line of the body, and not, as suggested by Newberry and others, along the anterior margin of the pectoral fins. That they belonged to Chimaeroids seems very probable, and they are an invari- able accompaniment of the large dental plates described as Ptyctodus ferox in the western Devonic. A characteristic Mesodevonic species, it is interesting to note its presence in the fauna of the Portage shale, as DEVONIC FISHES OF THE NEW YORK FORMATIONS 83 indicated by one rather large spine preserved in the State Museum at Albany, from western New York. Formation and locality. Hamilton limestone (Erian) ; Milwaukee, Wis. Cedar Valley limestone and State Quarry beds; Iowa. Portage shale (Senecan) ; New York. Heteracanthus uddeni Lindahl 1897 Heteracanthus uddeni /. S. Lindahi. Cinn. Soc. Nat. Hist. Jour. 19: 95, pl. 6 1898 Heteracanthus uddeni C. R. Zastman. Am. Nat. 32:557 1899 Heteracanthus uddeni C. &. Zastman, Jour. Geol. 7: 282 Spines of moderate size having the distal portion essentially as in the preceding species, but the basal portion curving forward so as to forma rounded anterior projection or “shoulder,” somewhat similar to that in Stethacanthus. The longitudinal costae are finer and more closely crowded than in the type species, and exhibit a sigmoidal curvature toward the base. Formation and locality. Cedar Valley limestone (Mesodevonic) ; Buf- falo, Waterloo and Waverly, Ia. Hamilton limestone (Erian); Milwaukee, Wis. Genus MACHAERACANTHUS Newberry Syn. Machaerius Roualt ; Dinacanthodes Fritsch Spines, so far as known, elongated, tapering, more or less curved, and somewhat laterally compressed, with sharp edges and a very large longi- tudinal ridge on each side; central cavity extending nearly to the apex; external surface covered with a thin layer of ganodentine, smooth, finely punctate, or longitudinally striated. The resemblance between the spines referred to this genus and those occurring in advance of the pectoral fins in Acanthodes, together with the fact that the proximal portion indicates a corresponding position, has been interpreted as pointing to Acanthodian affinities. 84 NEW YORK STATE MUSEUM Machaeracanthus peracutus Newberry 1857 Machaeracanthus peracutus /. S. Newberry. Nat. Inst. Bul. p. 125 1862 Machaeracanthus peracutus /. S. Mewberry. Am. Jour. Sci. [2] 34: 74, text fig. 2 1873 Machaeracanthus peracutus /. S. Newberry. O. Geol. Sur. Rep’t, v. I, pt 2, p. 305, pl. 29, fig. 6 1889 Machaeracanthus peracutus /.S. Mewberry. U.S. Geol. Sur. Monogr. 16: 40, pl. 29, fig. 6 and woodcut Spines of moderate or large size, slightly curved, gradually tapering, and traversed along the middle by a prominent ridge, which is distinctly triangular in cross-section on one side of the spine throughout its entire length, and more or less rounded on the other. Lateral edges sharp, apex acute, external surface smooth or with extremely delicate striae which are convergent inward at a slight angle with the external margins. Formation and locality. This species, which is the type of the genus, has been previously supposed to be confined to the “ Corniferous” (Colum- bus and Delaware) limestone of Ohio. A few examples, however, have been obtained within recent years from the Onondaga limestone of Leroy and Lime Rock, in New York State. Machaeracanthus sulcatus Newberry Plate 3, figure 6 1843 “Ichthyodorulite” /. Hall. Nat. Hist. N. Y. pt 4, p. 174, woodcut fig. 69 1857 Machaeracanthus sulcatus /. S. Mewberry. Nat. Inst. Bul. p. 6 1870 Machairacanthus sulcatus £&. R. Lankester. Geol. Mag. 7: 398, wood- cut fig. 3 1873 Machaeracanthus sulcatus /. 8S. Newberry. O. Geol. Sur. Rep’t, v. 1, pt 2, Pp. 305 1886 Machaeracanthus sulcatus 7. A. Lennox. Can. Inst. Proc. 3: 120 1889 Machaeracanthus sulcatus /. S. Mewberry. U.S. Geol. Sur. Monogr. 16: 40, pl. 29, fig. 5 Spines of about the same size as the type, and differing from it princi- pally in that the less strongly carinated face bears several prominent longi- DEVONIC FISHES OF THE NEW YORK FORMATIONS 85 tudinal sulci, separated by narrow ridges, which terminate before reaching the distal three fourths of the spine. Regarding the distribution of this species, Newberry offers the follow- ing remarks: It would seem that the sharks that carried these spines were more numerous in those portions of the Corniferous sea which covered western New York and southern Canada than in the more open waters of the area now occupied by Ohio. In the exposures of the Corniferous limestone on Kelly’s island, Lake Erie, at Sandusky, Delaware and Columbus, O., frag- ments or complete spines of Machaeracanthus major and M. pera- cutus are not at all uncommon, but though collecting extensively myself in those localities I never obtained there a specimen of M. sulcatus. Formation and locality. Onondaga limestone (Ulsterian); various places in New York State and Canada, and in the corresponding horizon at Milford, O. Machaeracanthus longaevus sp. nov. Plate 2, figure 8 Spines attaining considerably larger dimensions than the type and differing principally in their greater width and conformation of the median carina. The latter is sharply triangular on one side in the distal portion only of the spine, gradually becoming widened and flattened until in the proximal portion it is nearly rectangular in cross-section. No longitudinal sulci. This species, the only one known to persist as late as the Hamilton, is founded upon an interesting specimen belonging to the Buffalo Society of Natural Sciences, and derived from the so called ‘ Trilobite bed” on the shore of Lake Erie near Eighteen Mile creek, New York. It was kindly placed in the hands of the writer for description by its discoverer, Mr F. K. Mixer, of Buffalo. Portions of two spines are preserved in counterpart on the same slab, embedded in which are pygidia of Phacops rana, and tests of Spirifer mucronatus and Ambocoelia umbonata, characteristic fossils of the Hamilton beds. The present example is interesting in that it is one of the few in which spines of both pectoral fins are preserved in natural association. That this 86 NEW YORK STATE MUSEUM is the case, instead of there being merely a single, large broken spine, is evident from the similar proportions and general appearance of the two spines, one of which clearly represents the proximal and middle portions, and the other a section extending from about the middle for a considerable distance beyond in the direction of the epex. The form of the cross-section leaves no doubt that both spines present the same aspect, presumably the outer or external face. On the opposite, or internal face, the median carina appears to be gently rounded throughout its entire length. One of the distinguishing characteristics of this species, however, is that the axial ridge on the side which is presumed to be external remains sharply triangu- lar only in the distal half of the spine, becoming widened into a broad flat elevation, smooth or but faintly striated, and nearly rectangular in cross- section, toward the base of the spine. The general surface is smooth, save for the usual delicate striae, slightly convergent toward the apex, and pos- sibly of the same nature as growth lines. The absence of deep longitudinal grooves distinguishes this species from M. sulcatus and M. bohemi- cus, the latter having been originally referred to Ctenacanthus. A few doubtful fragments, suggestive either of this species or the preceding, have been obtained from the Hamilton of Milwaukee, Wis. Formation and locality. Hamilton limestone (Erian); Eighteen Mile creek, N. Y. (and Wisconsin ?). Genus GYRACANTHUS Agassiz The paired fin spines of this genus,* which are not known to be associ- ated with other types of dermal defenses in the rocks of this country, are thus described by Smith Woodward : Spines of two distinct types, the one evidently connected with the fins, the other free. 7 spznes elongated, robust, more or less arched, irregu- larly rounded or oval in transverse section, except towards the unworn apex, which is compressed ; base of insertion large, with the internal cavity "Woodward, A. S. On a Carboniferous Fish Fauna from the Mansfield District, Vic- toria. Nat. Mus. Melbourne. Mem 1, 1906. p. 1-32. In this paper the relations of the ‘Gyracanthidae” are definitely proved to be with Acanthodian sharks, upon the evidence of acomplete skeleton of the remarkable Australian genus Gyracanthides [text fig. 17]. DEVONIC FISHES OF THE NEW YORK FORMATIONS 87 open for a considerable extent posteriorly. The longitudinal mesial line of the anterior face, except near the unworn apex, defined only by the superficial ornament, which consists of par- allel, oblique, transverse, ridges, diverging in pairs from this line and inclined towards the inserted extremity ; posterior face with a nar- row unornamented area, sometimes bounded by a series of denticles on one side; unworn apex also destitute of ornament. Gyracanthus sherwoodi Newberry Plate 3, figure 7 1889 Gyracanthus sherwoodi J. S&. New- berry. U.S. Geol. Sur. Monogr. 16: 119, pl. 18, fig. 4 Fin spines attaining a length of about 15 cm, laterally much compressed, gently arched and gradually tapering toward the apex; internal cavity open as a narrow groove for a considerable distance posteriorly ; sides ornamented with numerous fine, _ parallel curved ridges, either irregularly pectinated or faintly nodose. The only examples hitherto known of this species are those described by Newberry from a single boulder thought by him to have been derived from the Catskill of Tioga county, Pennsylvania. A few fragmentary spines displaying the same ornamentation and agreeing in form with Newberry’s originals have been obtained from the Chemung group near Warren, in the same state, one or two being preserved in the Museum of Compara- tive Zoology at Cambridge, Mass. Several Fig.17 Gyracanthides murrayi A. S. Woodward. Restored drawing of fossil, much reduced, head and abdominal region seen from below, tail twisted to exhibit side view (After Smith Woodward) specimens from the Chemung of southern New York are also preserved in the State Museum at Albany. One particularly fine impression, shown in 88 NEW YORK STATE MUSEUM plate 3, figure 7, is labeled as having come from the top of Learn hill, in the northern part of Ischua township, Cattaraugus county, and there are others from near Portville and southeast of Olean, in the same State. Formation and locality. Chemung beds (Chautauquan); New York and Pennsylvania. Catskill sandstone; Tioga county, Pennsylvania, Subclass DIPNEUSTI Fishes with partially ossified skeleton, numerous membrane or dermal bones, and persistent notochord; skull autostylic; dentition confined to inner bones of the mouth; gill clefts feebly separated, opening into a cavity with membranous operculum ; paired fins archipterygial or reduced; median fins often subdivided ; exoskeleton consisting of true bony tissue; sensory canals well developed. The few existing Dipnoan' species, comprised by the fresh-water genera Neoceratodus, Protopterus and Lepidosiren, form a well nigh inap- preciable remnant of a once flourishing and highly diversified race of lung fishes, whose maximum development and specialization occurred during the Devonic. One remarkable order comprising huge armored fishes passed entirely out of existence during the lowermost Carbonic, without leav- ing descendants. Such, at least, appears to be the most satisfactory inter- pretation of the group now commonly known as Arthrodires. Another order, that of Ctenodipterines, was conspicuous throughout the Paleozoic, and attained a higher degree of specialization along certain lines than is evinced by later forms. The geological history of the Sirenoid order, to which Ceratodus and its modern descendants belong, is not traceable with certainty earlier than the Triassic, although it is possible that some Paleo- zoic remains, known chiefly by the dentition, properly belong here. That primitive members of this order were in existence at least as early as the t As pointed out by Haeckel, Boulenger and others, the term Dipnoi, first applied by Johannes Miiller in 1845 for the group of lung fishes, is improperly so used, having been previously chosen by Leuckart asa name for Amphibians. There is no objection, however, to retaining the name Dipnoan as a vernacular equivalent of Dipneusti, and it is here employed in that sense. DEVONIC FISHES OF THE NEW YORK FORMATIONS 89 Lower Devonic follows as a necessary inference from regarding Sirenoids as ancestral to Arthrodires and Ctenodipterines. This view of their relations, however, is novel, and the considerations which make for its acceptance, and compel us to look upon Neoceratodus as an archaic survival of the primal Dipnoan stock, will be discussed under the general treatment of Arthrodires. Order ARTHRODIRA Dipnoans having a reduced number of dermal bones forming the cranial roof, arranged essentially after the same pattern as in Ceratodonts, and their dentition also paralleling modern forms. Dermal armor of abdominal region consisting of large plates, either in simple apposition with the head shield, or articulated with its posterior border by a pair of movable ginglymoid joints placed dorsolaterally. Column notochordal, but with distinct neural and haemal arches. Tail apparently diphycercal in the best known forms; paired fins rudimentary or absent; pelvis repre- sented by a pair of sigmoidal or club-shaped plates, sometimes with an anterior ventral projection. The remarkable group of armored Coccosteuslike fishes was originally united with Asterolepids by M’Coy, in 1848, in a single “family Placo- dermi,” and for more than 4o years this arrangement was adhered to by writers generally, save for slight changes in the rank assigned to the main divisions. To Professor Cope belongs the merit of being the first naturalist to recognize the heterogeneous nature of this assemblage, and to initiate its disruption. In 1889, he proposed the removal of Asterolepids from the class of fishes altogether, and at the same time referred Coccosteans pro- visionally to the Crossopterygii.* Soon afterwards, however, following Smith Woodward's suggestion, the several families of Coccosteans were grouped, under the new name of Arthrodira, in a separate order of Dip- noans.? This arrangement obviously implied, though it had not then been demonstrated, that the Arthrodiran skull was truly autostylic, and that a * Cope, E. D. Synopsis of the Families of Vertebrata. Am. Nat. 1889. 23: 856. 2 Lbid, 1891. 25:647; also Syllabus of Lectures on Geology and Palaeontology. Phila. 1891. p. 14. go NEW YORK STATE MUSEUM maxillary arch was not developed. Another feature which influenced the novel association of Arthrodires with Dipnoans was the parallelism, pre- viously noted by Newberry,’ between the dentition of Dinichthys and that of Protopterus. The absence of any indication of a hyomandibular bone, even in the most admirably preserved specimens, and of more than a single ossification in the lower jaw, were considered sufhcient reasons for excluding Arthrodires from Teleostomes. This provisional classification of Arthrodires with Dipnoans met with an indifferent reception on the part of most paleontologists, and was afterwards rejected by some of its early supporters, notably Traquair and Bashford Dean. It was even conceded by Smith Woodward himself a few years later, that ‘the systematic position of this extinct order is indeed doubtful.”? Traquair’s defection dates from 1900, when he declared, in his Bradford address,? in favor of considering Arthrodires as ‘‘ Teleostomi belonging to the next higher order, Actinopterygii.” The following year Dean expressed the radical view that they were not fishes at all, but representatives of a distinct class, called by him Arthrognathi, and conceived to have possible kinship with Ostracophori.* It was even allowed that subsequent researches might demonstrate a union between Arthrognaths and Ostracophores, whereby the time-honored group of Placodermata would be restored. This “ce last was a complete reversal of his former view that the ‘jaws, specialized dentition, fin spines, and highly evolved pelvic fins at once separate this group [Arthrodira] from the lowly Ostracoderms.” By far the most amplitudinous extension of the term Placodermata is that proposed by Otto Jaekel, in 1902, whereby the Pteraspids, Trematas- pids, Cephalaspids, Asterolepids and Coccosteans were all embraced within "Newberry, J. S. O. Geol. Sur. Rep’t Pal. 1875. 2:15. The suggestion is here advanced that Protopterus and Lepidosiren are lineal descendants of “ Placoderms.” 2Woodward, A. S. Outlines of Vertebrate Palaeontology. 1898. p. 64. 3Traquair, R. H. Vice Presidential Address. Brit. Ass’n Adv. ’Sci. Bradford meet- ing, Rep’t. 1900. p. 779. 4Dean, B. Palaeontological Notes. New York Acad. Sci. Mem. 1901, 2: 113. sIdem, Fishes, Living and Fossil, N. Y. 1895. p. 130. DEVONIC FISHES OF THE NEW YORK FORMATIONS oI a single group.’ This assemblage was modified a twelvemonth later, how- ever, in that the two last named divisions were bracketed together under the head of * Temnauchenia,” all of the others, together with Drepanaspids, Coelolepids and Birkeniidae, being collectively designated as “ Holau- chenia.”* Placoderms in this broadened sense were all considered by Jaekel to belong to fishes proper, and it was further maintained by him that Coccosteans were ancestral to Chimaeroids, an opinion in which he clearly stands alone. ‘The only other author who has ventured to recognize any descendants from Arthrodires whatever is Newberry, who, as we have seen, imagined Protopterus to be a modern survival of Dinichthys. We may now pass rapidly in review the minor fluctuations of opinion that are apparent during the last few years. Dr O. P. Hay, in his Cata- logue of Fossil Vertebrata of North America, employs the term Placodermi for both Arthrodires and Asterolepids, placing them in the same subclass with Dipnoans. Arthrodires and Ostracophores are awarded each the rank of a separate subclass in the English edition of von Zittel’s Textbook of Falacontology, the author having disapproved of an association between Coccosteans and Dipnoans. In Mr C. T. Regan’s remarkable paper of 1904, already referred to, the Placodermi are reestablished so as to include Coccosteidae, Asterolepidae and Cephalaspidae, all being united in a single order of Teleostomes.? During the same year Professor Bridge expressed the view, in the volume on /7shes in the Cambridge Natural History, that Coccosteans are “a highly specialized race of primitive Teleostomi,” and compared their cranial roofing plates with those of typical bony fishes. The idea of a relation between them and lung fishes is dismissed in the follow- ing passage, found at page 537 of the work in question: The Arthrodira have been regarded as armored Dipneusti, a view which is mainly based on their supposed autostylism and the nature of the ‘Jaekel, O. Ueber Coccosteus und die Beurtheilung der Placodermen. Sitzungsber. Gesellsch. Naturforsch. Freunde. 1902. p. 103. 2Jdem. Ueber die Organisation und systematische Stellung der Asterolepiden, Zeitschr. deutsch. geol. Gesellsch. Mai-Protokoll. 1903. 55:58. 3Regan, C. T. The Phylogeny of the Teleostomi. Mag. Nat. Hist. Ann, ser. 7, 1904. 13: 346. 92 NEW YORK STATE MUSEUM dentition. But this autostylism has yet to be verified, and, if proved, the possibility that it may be a secondary feature, associated with the evolution of a peculiar dentition, must not be forgotten. Much more may be said for their claim to be regarded as a highly specialized race of primitive Teleo- stomi. Besides a well developed lower jaw, bones comparable to the ele- ments of a secondary upper jaw are known, and in a general way the disposition of the cranial roofing bones, and the arrangement of the endo- skeletal elements of the pelvic fins, tend to conform to the normal Teleo- stome type. In fact, Dr Traquair has expressed the opinion that the Arthrodira are Teleostomi and Actinopterygii." in his description of some Dinichthyid remains from Ohio, published in 1905, Mr L. Hussakof? refers to them as ‘“ Placoderms,” apparently using the term in its familiar acceptance. Their position is also left undetermined by E. Ray Lankester, in his interesting lectures on Axtznct Anzmals, recently published in book form.* Dr Lucas’s popular treatise on Anzmals before Man tn North America places them in association with lung fishes, in accordance with Smith Woodward's idea. One other popular handbook claims attention, not only because it is an extremely useful work covering the whole subject of fishes, but because of the author’s extensive acquaint- ance with fossil as well as recent forms. We refer to President Jordan’s Guide to the Study of Fishes [|N. Y. 1905], in the first volume of which, at page 582, the relations of Arthrodires are discussed in following wise: 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 turtlelike 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. ‘In his latest reference to this subject, however, he admits that they are of uncertain subclass. Compare, for instance, his description of Coccosteus angustus in Roy. Soc. Edinburgh Trans. 1903. 40: 732. ?Hussakof, L. Notes on the Devonic “ Placoderm,” Dinichthys intermedius Newb. Am. Mus. Nat. Hist. Bull. 1905. 21: 27-36. Also, more recently, Studies on the Arthro- dira, in the memoirs of the same institution. 1906. 9: 105-54. 3Lankester, E. R. Extinct Animals. N. Y. 1905. p. 256. DEVONIC FISHES OF THE NEW YORK FORMATIONS 93 Nor is there any sure foundation to the view adopted by Woodward, that they are to be considered as armored offshoots of the Dipnoans. Again, at page 445 of the same volume, this passage occurs: These creatures have been often called ganoids, but with the true ganoids like the garpike they have seemingly nothing in common. They are also different from the Ostracophores. To regard them with Wood- ward as derived from ancestral Dipnoans is to give a possible guess as to their origin, and a very unsatisfactory guess at that. Finally, reference may be made to two papers published lately by the present writer,’ in which he endeavored to show that the dentition of Arthrodires belongs distinctly to the Dipnoan type, and that real homolo- gies exist between their dermal cranial plates and those of the living Neo- ceratodus. Indeed, the modern form was held to bear as intimate structural resemblance to Coccosteans on the one hand, as to Ctenodipterines on the other, although conforming more nearly than either in certain respects to the hypothetical ancestor from which all three types — Ceratodonts, Arthro- dires and Ctenodipterines -— have been derived. The position maintained in this last communication is adhered to in the present memoir, and it is believed that sufficient evidence has now been accumulated to sustain its correctness. Heretofore, in default of positive evidence, writers have been unable to demonstrate the truth of any one of the various conjectures put forward to explain the nature of Arthrodires. However plausible one or another of these may have appeared, however widely they have gained acceptance, it must be remembered that a sug- gestion remains a suggestion, and a hypothesis a hypothesis, until its cor- rectness is fully demonstrated. Rightly is it said in one of Plato’s Socratic dialogues :* “ Mere beliefs or opinions are, like the statues of Daedalus, runaway things; not until they have been tied down by the chain of causal sequence do they stand fast and become in the true sense knowledge.” What constitutes this ‘“‘reasoned interconnection” in the present case may be told in few words. In the first place we must anticipate a little by t Eastman, C. R. Dipnoan Affinities of Arthrodires. Am. Jour. Sci. ser. 4, 1906. 21:131-43. dem. Mus. Comp. Zool. Bul. 1906... 50: 1-30. 2 Meno, 159 D. 94 NEW YORK STATE MUSEUM saying that valuable enlightenment concerning the dentition of Arthrodires is furnished by the type of a new genus from the Portage of western New York, hereinafter described under the name of Dinomylostoma. This interesting form, transitional between Dinichthys and Mylostoma, acquaints us for the first time with the nature of the articulation between the lower jaw and head shield amongst Coccostean fishes; it displays portions of the Meckelian cartilage, still preserved in natural position; and the relations of this structure to the ossified mandibular elements— splenial and dental plate proper —are seen to be precisely as in modern Dipnoans. The lower jaw of Arthrodires is therefore proven to conform strictly to the Dipnoan type, a fact of cardinal importance for their classification. The upper den- tition, also, is comparable in a general way to young stages of Neocera- todus; hence it may be affirmed with entire propriety that the dentition of the recent genus passes through an early Mylostomid stage. Further com- parisons between Neoceratodus and typical Arthrodires, such as Coccosteus, Dinichthys and Mylostoma, establish a sound basis for the following propositions : 1 The dermal plates forming the cranial roof in Arthrodires and primi- tive Ceratodonts have undergone corresponding reduction, and become arranged after essentially the same pattern. 2 Neoceratodus recalls throughout its entire organization, save only for the absence of dermal armoring, the principal features of Arthrodires ; and no theory of parallelism is competent to explain the aggregate of these resemblances. 3 It is impossible to regard Neoceratodus as the degenerate descend- ant of do¢2 Ctenodipterines and Arthrodires, nor of either of them singly, since it partakes of the characters of both. Community of origin is there- fore necessarily presupposed for all three groups, there being no alternative explanation of their relations. 4 Arthrodires and Ctenodipterines may be regarded as specialized offshoots which diverged in different directions from the primitive Dipnoan DEVONIC FISHES OF THE NEW YORK FORMATIONS 95 stem, only the more generalized descendants of the latter having survived until the present day. 5 The primordial stock must have been autostylic, diphycercal, without a maxillary arch and dentigerous dentary elements, and with a Uronemus- like or Dipteruslike dentition ; characters which do not permit us to ascribe the ultimate origin of Dipnoans to the Crossopterygii, but suggest rather a descent from Pleuracanthid sharks. 6 The recognition of Arthrodires as an order of armored Dipneusti precludes their association with Ostracophores in any sense whatever. The recently revived group of ‘‘Placodermata” is, therefore, an unnatural assemblage, and should be abandoned. 7 In the light of present information, the progressive modifications amongst early Dipnoans may be represented after some such scheme as follows: Neoceratodus Ceratodus | sturtt C pee. | ee Titanichthys COvronemus Coccosteus, Dintchthys Paar plein Larne ‘ylostoma, Dinomylostoma Sca WUE eb Lae meds a - Macropetalichthys Primitive Ceratodonts It would be superfluous to present here a fresh demonstration of the close structural agreement between Arthrodires and Ceratodonts. One 96 NEW YORK STATE MUSEUM point only need be developed a little more fully, perhaps, and that is the fact that Dinichthys and Mylostoma are related to each other in the same way with respect to their dentition as are Protopterus and Neoceratodus. This will appear from the following considerations. Dentition of Dinichthys and Neoceratodus compared. It is recognized by all students that Neoceratodus, in comparison with other surviving lung fishes, represents a relatively early larval stage of development; nor does any one question that the trenchant dental plates of Protopterus and Lepid- osiren are other than a mere variant of the Ceratodont type. This much being granted, it is but a short step further to see that the dentition of Dinichthyids has been similarly derived. Certainly no difficulty is offered by the so called “ premaxillary” teeth of Dinichthys, which are the precise equivalent of the vomerine pair in modern Dipnoans. As for the character- istic crushing plates in the upper jaw of Ceratodonts, these occur normally in Mylostoma, but in Dinichthys become rotated so as to stand more or less vertically, their outer denticulated margins functioning against one another in opposite jaws like the blades of a pair of shears. An inkling as to how this modification was brought about is offered by the Triassic Ceratodus sturii, which may be taken to represent an incipient stage of metamor- phosis. The dental plates of this form are observed to be turned consid- erably on edge, their marginal corrugations interlocking in opposite jaws when the mouth is closed, and a rudimentary beak being developed in front which recalls the well known forward projection in Dinichthyid mandibles. As for the so called “ maxillary” or “shear tooth” of Dinichthys, this corresponds clearly to the triturating upper (palato-pterygoid) dental plate of Ceratodonts, turned rather more upright than in C. sturii, and its anterior process or “shoulder” is represented by the forwardly placed ascending process of modern forms. In Arthrodires, as in other Dipnoans and higher forms, the functional lower jaw is formed by membrane plates which have ossified around the Meckelian cartilage. Distinct angular and articular elements appear to be wanting in Dinichthys,* but the splenial is *Dean, B. Fishes, Living and Fossil. N.Y. 1895. p. 133. DEVONIC FISHES OF THE NEW YORK FORMATIONS 97 strongly developed, supporting the dental plate sewsu s¢rzcto in front, and being articulated posteriorly with the cranium by cartilage, as in Neocera- todus. But one interpretation can be given of the conspicuous groove which extends forwards along the inferior border of the splenial, passing underneath and to the inner side of the dental plate proper, and terminat- ing near the symphysis. In it were lodged remnants of the Meckelian cartilage, precisely as in the living Protopterus. Intermandibular teeth have not been definitely proved to occur, although their presence would be in strict accord with embryological evidence, and the appearance of certain specimens has led some investigators to suspect that such teeth were developed; moreover, in Coccosteus, Diplognathus etc., we are confronted with a denticulated symphysial margin, the significance of which has not been explained. Finally, it should be noted that an approach to the Dinichthyid form of mandible is made even amongst Ctenodipterines,’ and some of these also develop cutting edges, asin Sagenodus pertenuis, for example. Dentition of Mylostoma and Neoceratodus compared. Our knowledge of Mylostoma received important additions a few years ago as the result of Dr Bashford Dean’s investigation of the type species.3 Since then, how- ever, fresh material has been brought to light which shows that this author was not entirely successful in determining the relations of the palato- pterygoid dental plates in the unique specimen studied by him. Their true disposition has recently been ascertained beyond doubt to be as rep- resented in the accompanying text figure 18, a similar arrangement holding true also for Dinomylostoma, and, interestingly enough, being repeated in early stages of Neoceratodus. This we have already alluded to in a pre- ceding paragraph, by saying that Neoceratodus passes through a Mylo- stomid stage. One can hardly fail to appreciate the significance of this *Atthey, T. On Ctenodus obliquus, etc. Mag. Nat. Hist. Ann. ser. 4, 1875. 15: 300, pl. rg, fig. 2. 2Eastman, C. R. A Peculiar Modification amongst Permian Dipnoans. Am. Nat. 1904. 37: 493-95- 3Dean, B. Palaeontological Notes. N.Y. Acad. Sci. Mem. tgor. 2: 101-9, OO NEW YORK STATE MUSEUM 9 fact in considering the relations between fossil and recent forms. It has been shown by Semon? that the dental plates of the modern genus arise through concresence of conical denticles, which are at first disposed so as to form two pairs of palato-pterygoid plates, these afterwards fusing into one; and, thanks to this observation, we can readily account for what would appear at first sight to be a supernumerary pair in Mylostoma. Moreover, it will be noted that the peculiar posterior contour of the hindermost pair of plates in the fossil form is scarcely to be explained except as we regard it in association with, and conformable to, the usual pattern of palato- pterygoid cartilage found in all Dipnoans. Amongst Ctenodipterines this Fig.18 Mylostoma variabile Newberry. Cleveland shale; Sheffield, O. Palato-pterygoid dental plates, arranged in natural position, but belonging to different individuals. The vomerine pair is supposed to have been situated imme- diately in front, but no specimens undoubtedly assignable to this position are at present known. x 24 “ element is ossified, and passes under the name of “upper dentigerous bone,” yet it preserves essentially similar outlines, and bears the stamp of a persistent feature. The mandibles of Mylostoma betray unmistakable traces of a Cerato- dont origin, for the dental plate properly speaking is even more sharply distinguished from the splenial than in Dinichthys or Dinomylostoma. The presence of intermandibular teeth would be in complete harmony with embryological evidence, and as a matter of fact certain detached teeth have ‘Semon, R. Die Zahnentwickelung des Ceratodus forsteri. Zool. Forschungsreisen in Australien etc. Jena Denkschr. rgor, 4: I15~33. DEVONIC FISHES OF THE NEW YORK FORMATIONS 99 been somewhat doubtfully interpreted as such by Newberry.t The pair figured by him, however, have recently been ascertained to represent imper- fect mandibular plates of an undescribed species of Mylostoma, and are proved to have been firmly united in the median line. Under these con- ditions intermandibular structures are definitely excluded. Amongst Ctenodipterines, Synthetodus furnishes the only known example where symphysial teeth are permanently retained, thus paralleling an evanescent condition in Neoceratodus. Indeed, this very character leads one to sus- pect that the form in question belonged to primitive Ceratodonts, rather than to the Ctenodipterine order of Dipnoans. Systematic arrangement of Arthrodires. Various stages of evolution are recognizable throughout the group, the most important being typified by (1) Macropetalichthys, (2) Homosteus and (3) different genera comprised by the families Coccosteidae, Mylostomatidae and Selenosteidae in order of progressive specialization. In the system of classification proposed by Bashford Dean, these stages find expression through the erection of three new “orders,” the first and most primitive of which includes Macropetali- chthys and Asterosteus, the second Homosteus, and the third all of the more highly specialized forms. Dean’s arrangement of families, and for the most part, of genera as well, is probably as judicious as present known facts will allow; hence we have adopted it in the following section of the report, without, however, taking account of his larger divisions. His defi- nition of one of them, as pointed out by Jaekel,’ includes structural charac- ters which do not exist; and we may be reasonably confident that in none of them did the jaws operate by rotary movements, on which point Dean, and following him Hussakof, lay great stress, Family MACROPETALICHTHYIDAE Cranial shield much arched from side to side, completely inclosing the orbits, and extending over the nuchal region posteriorly. External surface covered with fine stellate tubercles which conceal the underlying sutures between dermal plates. Median series of plates but two in number, narrow tNewberry, J. 5S. U.S. Geol. Sur. Monogr. 1889. 16: 165, pl. 16, fig. 4. 2Jaekel, O. Referat in Neues Jahrb. f. Min. Jahrg. 1903. 1: 342. 100 NEW YORK STATE MUSEUM and elongate ; external occipitals large; centrals divided, the two pairs on either side not meeting in the middle line. Pineal foramen inconspicuous, situated slightly in advance of a line joining the anterior borders of the orbits. Sensory canals forming large tubular excavations in the bone, opening at the external surface by a continuous narrow slit or by a double row of pores. Parachordal cartilage and notochordal sheath ossified. Dental plates and abdominal armor unknown. The typical genus of this family is Macropetalichthys, known by a single American, and one or two European species. Some others, as yet undescribed, are reported by Jaekel from the Eifel Devonic, and at the same time the form described by Kayser as M. pruemensis is made the type of a new genus... The members of this family evidently stand in somewhat close relation with Homosteus, all of these forms having the head shield prolonged posteriorly over the nuchal region, and exhibiting a decidedly primitive arrangement of cranial plates. In this latter respect it is interesting to note that the median series of plates are reduced to the same number as in Neoceratodus. Genus MACRO_ETALICHTHYS Norwood & Owen This genus, certainly one of the most primitive of Arthrodires, is repre- sented in this country by the typical species, which was first described under the name of M. rapheidolabis. The history of our acquaintance with the genus is interesting, and may be recapitulated as follows: Specimens of the head shield seem to have been first discovered by Mr Joseph Sulli- vant, of Columbus, O., as early as 1836, though they were not brought to the attention of students until several years later. In 1846 Messrs Nor- wood and Owen published a Descrzptzon of a new Fossil Fish, from the Palacozow Rocks of Lndiana,? in which the generic and specific titles were properly proposed. The type specimen of M. rapheidolabis was exhibited before the American Association for the Advancement of Science *Jaekel, O. Ueber Coccosteus und die Beurtheilung der Placodermen. Sitzungsber. Gesellsch. Naturforsch. Freunde. rg02. p.113. did. Jahrg. 1906. 73-85. «Am. Jour. Sci. [2] 1846. 1: 367-71. DEVONIC FISHES OF THE NEW YORK FORMATIONS IOI at the Cincinnati meeting, in 1851, where it was examined by Professor Louis Agassiz.t Newberry, in 1862, mentions having seen a cast of the same, but states that the original itself had disappeared.» However, it still remained in existence, and was finally acquired by the University of Mis- souri. In 1891 it was redescribed by Cope,? most fortunately for science, since only a few months after its return to Missouri it was destroyed by fire together with the entire contents of the museum to which it belonged. The same year that Norwood and Owen described the type species, Hermann von Meyer‘ published a few notes on the fossil fishes from the Eifel Devonic, one of which received the name of Placothorax agas- sizi. This was soon afterwards figured and described by the same author,’ but it was reserved for Newberry, in 1857, to point out its identity with Macropetalichthys.6 Von Meyer’s specimen was entirely denuded of its superficial ornamentation, a circumstance which prevented him, as he tells us, from associating with it a second specimen which he described under the name of Physichthys hoeninghausii;’ nevertheless he conceded that they might possibly prove to be identical. He also pointed out that fragments of the same nature and from the same locality as these had been described by Agassiz under the title of Asterolepis hoen- inghausii. Von Meyer’s type specimens of “ Physichthys” afterwards passed into the possession of Dr L. Schultze, from whom they were pur- chased in 1871 by the Museum of Comparative Zoology at Cambridge, Mass. Here they were examined by Dr A. Smith Woodward, who was the first to demonstrate the true nature of ‘‘ Physichthys.”® A second European species was brought to light in 1880, and named M. prue- tAm., Ass’n Adv. Sci. Proc. 1851. 5: 179. 2 Am. Jour. Sci. [2] 1862. 34:76; U.S. Geol. Sur. Monogr. 1889. 16: 27. 3 U.S. Nat. Mus. Proc. 1891. 14: 449-56. + Neues Jahrb. fiir Min. 1846. p. 596-97. 5 Palaeontographica 1847. 1: 102-4, pl. 12, fig. 1. 6 Nat. Inst. Bul. 1857. p. 119. 7 Palaeontographica 1855. 4: 80-83, pl. 15, fig. 1-5. 8 Geol. Mag. [3] 1890. 7:459; also Cat. Foss. Fishes Brit. Mus. 1891. pt 2, p. 303. 102 NEW YORK STATE MUSEUM mensis by Dr E. Kayser.*. It was further described and illustrated by Professor A. von Koenen in 1895, together with an imperfectly preserved Bohemian cranial shield, designated as Holopetalichthys novaki. Quite recently, as we have already seen, Dr Otto Jaekel expressed the opinion that the type of M. pruemensis should be held to represent a distinct genus. We may now trace the progress that has been made in investigating the typical American species, which had not ceased to attract attention. Professor Newberry contributed to the first number of the Annals of Scvence, published October 15, 1852, a brief account of “ Fossil Fishes from the Cliff Limestone,” in which he described and figured an unusually per- fect specimen of Macropetalichthys. Although the position of the principal plates and their centers of ossification are clearly portrayed in this early woodcut, the significance of these features was not understood, and no account was taken of them in later restorations. Newberry’s views as to the relations of his fossil are thus stated: ‘This is evidently the cranium of a ganoid fish, allied to the Asterolepis of Europe, and is probably identical with that described by Owen and Norwood. I say prodadly, for although the general outline of the cranial plates seems to establish an identity, their descrzption would indicate entire distinctness — a discrepancy which I am inclined to attribute to the imperfect preservation of their specimen.” Other crania were exhibited by Newberry before the American Asso- ciation for the Advancement of Science at the Cleveland meeting in 1853, and in 1857 descriptions were given of two new species, the so called “Agassichthys sullivanti and A. manni,” which were at that time considered to be generically distinct from Macropetalichthys.? A few years later, however, Newberry returned to the belief that the two species of “ Agassichthys,” and also the type of von Meyer's “ Placothorax,” all belonged to Norwood and Owen’s genus. Subsequently he found it diffi- * Zeitschr, deutsch. geol. Ges. 1880. 32: 678. 2 Nat. Inst. Bul. 1857. p. 119-24. 3 Am Jour. Sci. [2] 1862. 34: 74, 75. DEVONIC FISHES OF THE NEW YORK FORMATIONS 103 cult to maintain the distinction between M. sullivanti and M. manni, and was ‘disposed to consider the differences which they exhibit as prob- ably due to age or sex.”* A final statement of Newberry’s views regarding the structure and relations of Macropetalichthys appeared in his Mono- graph on Palaeozoic Fishes of North America, published in 1889, the genus being compared with existing sturgeons, and sensory canals being still mis- taken for sutures. Cope,? in his review of his friend’s monograph, denied that this genus was in any way related to modern ganoids, and redescribed its characters the following year. Some additional details were pointed out by the present writer in 1897, and an attempt was made by Dean, in his important memoir of 1901, to homologize the cranial plates of Macro- petalichthys with those of other forms. Aside from the reviews of this memoir which appeared in this country and abroad, nothing has been contributed to the literature of the American species within the last few years, Macropetalichthys rapheidolabis Norwood & Owen Plate g, figure 5; plate 11; text figures 19-21 1846 Macropetalichthys rapheidolabis Norwood & Owen. Am. Jour. Sci. [2] 1: 371 1851 ‘Buckler of ganoid fish” Z. Agasstz. Am. Ass’n Adv. Sci. Proc. 5: 179 1852 Macropetalichthys sp. 7. S. Mewberry, Ann. Sci. 1, p. 12 1857 Agassichthys sullivanti /. S. Mewdberry. Nat. Inst. Bul. p. 124 1857 Agassichthys manni /. S$. Mewderry. Nat. Inst. Bul. p. 122, woodcut 1862 Macropetalichthys manni and M. rapheidolabis. Am. Jour. Sci. [2] 34:75, 76, woodcut 1873 Macropetalichthys sullivanti 7. §. Mewberry. O. Geol. Sur. Rep’t Palaeont. v. 1, pt 2, p. 294, pl. 24, 25, fig. 1 1889 Macropetalichthys sullivanti /. S. Newberry, U.S. Geol. Sur. Monogr. 16: 44, pl. 38, fig. 1, 2 1890 Macropetalichthys sullivanti £. D. Cope. Am. Nat. 24: 846 r89t Macropetalichthys sullivanti and M. rapheidolabis £& D. Cope. U.S. Nat. Mus. Proc. 14: 449-56, pl. 29, 30, fig. 5 t Ohio Geol. Sur. Rep’t. Palaeont. 1873. v. 1, pt 2, p. 295. * Am, Nat. 1890. 24:844-47. 1O4 NEW YORK STATE MUSEUM 1897 Macropetalichthys sullivanti and M. rapheidolabis C. &. Eastman, Am. Nat. 31: 493-99, pl. 12 tg01 Macropetalichthys sp. &. Dean. N. Y. Acad. Sci. Mem. v. 2, pt 3, p- 119g, text fig. 12 Head shield suboval, regularly arched from side to side, attaining a maximum length of about 25 cm, and width across the posterior border of about 17cm. Ornamentation consisting of fine, closely crowded stellate tubercles, sometimes displaying concentric arrangement. Of the two pairs of small centrals, which are separated from each other by the median occip- ital, the anterior takes part in the orbital border and is not traversed by sensory canals. Pineal plate pierced by an inconspicuous foramen, and apparently equivalent to the so called anterior median element or “ meseth- moid” of Neoceratodus. Parasphenoid much expanded in front, posteriorly produced, resembling in a general way that of Ctenodipterines and Sire- noids, but considerably less ossified. Preorbital sensory canals lyrate, and confluent in the middle line with the sharply angulated exoccipito-central system. The postorbital canal extends from the inferior border of the orbits to the center of the marginal plates, where it turns abruptly inwards and continues in a straight line to meet the exoccipito-central canal at the point of its angulation. The latter disappears beneath the surface of the external occipital plate close to the hinder margin of the head shield. Speaking entirely within bounds, it is not too much to say that the characters of this long misunderstood genus and species fail of comprehen- sion, or at least of satisfactory analysis, save as they are brought into rela- tion with those of modern Dipnoans, and interpreted through comparison with them. Many students have puzzled over the cranial osteology of Macropetalichthys and the allied genus Asterosteus, of which only the median series of plates are known; but accumulation of details has resulted only in greater perplexity. Were a moral to be drawn from this state of affairs, and others like it, it would be this: However diligently facts may be collected, however attentively studied, they possess of themselves no intrinsic value; their usefulness lies only in the measure that we are able to appre- DEVONIC FISHES OF THE NEW YORK FORMATIONS 105 ciate their significance, and to draw from them generalizations. Intolerably barren must be those lines of inquiry which result in no broad conclusions. Natural science, however, and especially paleontology, imposes upon us this difficulty: for an hour or two of synthesis, years of patient application in the study of facts are required. That which has hitherto been puzzling in Macropetalichthys and Aste- rosteus is the absence of a standard of comparison or other clues by means of which their characters acquire significance; they must needs remain unintelligible until brought into adjustment with other definitely known facts. Newberry, as we have seen, went widely astray in imagining these forms to be ancestral to modern sturgeons. Cope’s keen insight led him immediately to perceive the resemblances between Macropetalichthys and Dinichthys; and in suggesting a comparison between the former and Neoceratodus, he actually hit upon—though without adequately realizing it —a solution of the whole matter. Holding within his grasp the key to a correct interpretation of the Arthrodiran skull in all particulars, one regrets that he did not consistently apply it, instead of attempting vainly to estab- lish homologies with the skull of Stegocephalians. Only in one respect does he point out similarity of structure between Macropetalichthys and modern Dipnoans, and this relates to the underside of the head. The parasphenoid was correctly identified as such, and observed to have the usual Dipnoan outline; but he was less happy in determining the relations of the so called ‘cerebral chamber” of Newberry, and the structures termed by him “nuchal elements.” Cope’s “nuchal plate,” or “dorsal plate” as it was called by Dean and Eastman, was further misinterpreted by the last named authors in that it was held to represent collectively the dorsal body plates of other Arthrodires, Dean’s definition of ‘““Anarthrodira” being based upon this view. Indeed, it must be said frankly that serious misapprehension has existed in the minds of all students, including the present writer, concerning the structures seen within the interior of the head shield in Macropetalich- thys. As inthe case of the cranial buckler itself, they only become intelli- gible through comparison with surviving Dipnoans, as will presently be shown. 106 NEW YORK STATE MUSEUM Cranial osteology. The arrangement of dermal roofing plates in the cranial buckler of M. rapheidolabis is shown in the, restoration given herewith and also observable in the head shield represented in plate 11. Referring to text figure 24 for comparison, it will be seen that the disposition of these elements corresponds in a general way to the pattern presented by Neocer- atodus. Especially is this true of the median series of plates, the hindermost of which is elon- gated nearly to the same extent asin Homosteus. Other points of agreement between the form under discussion and Homosteus consist in the elongation of the external occipitals, and inclosure of the orbits within the head shield. In more specialized forms, the preorbital and postor- bital plates are merely notched Fig. 19 Restoration of Macropetalichthys showing arrangement of . cranial plates and course of sensory canals, x4. C', C?=divided by the orbits, but in Macropetal- centrals ; £O=external occipital ; A7—=marginal ; JfO—median occipi- tal; P=combined pineal and rostral, corresponding to anterior median ichthys, Homosteus, and presum- unpaired plate in Neoceratodus; PO=preorbital ; P¢O—postorbital ably also in Asterosteus, these two plates unite to form their external border. A conspicuous difference between Macropetalichthys and other Anthro- dires, one which has proved a stumbling-block to a correct understanding of the cranial osteology, lies in the fact that the central elements are divided so as to form ¢wo small plates on either side immediately back of the preor- bitals ; these are placed one behind the other, the two pairs being separated from contact with each other in the median line by the elongated median occipital plate, very much in the same way as in Neoceratodus. That the DEVONIC FISHES OF THE NEW YORK FORMATIONS 107 plates here called centrals are correctly identified as such is evident from the following reasons: first, the two pairs together occupy the usual posi- tion of the centrals with reference to the preorbital plate in front, and to the postorbital and marginal plates externally. Secondly, they are proved to be such by the disposition of sen- sory canals. Imagining the suture line as obliterated between the two independ- ently ossified plates on either side, we shall have a single pair of elements occupying the same rela- tive position as in Homo- Toe heen oanedy a steus. The _ posterior pile Wee eae moiety of this plate is seen Pe ere eae to be traversed by three different canal systems, identifiable with those called by Dean the preorbital, postorbital and occipital. Now, in the head shield of all Arthrodires so far as known, the central is the only plate traversed by all three of these canals. The preorbital and postorbital systems are sometimes con- fluent in other forms, but the occipital (or more prop- erly, the exoccipito-central) does not unite with these other systems save only in Macropetalichthys. A Fig. 21 Detached median occipital detached specimen of the median occipital show- plate of Macropetalichthysshowing | 2 ‘ sensory canals and partially de- ING confluent canal systems 1S represented in text nuded ornamentation, Original in Mus. Comp. Zool. Nat. size figure 21. Evidence of the primitive nature of this genus, as compared with other Arthrodires, is furnished by the following characters, which point to a sur- vival of embryonic or ancestral conditions: (1) continuity of the sensory 108 NEW YORK STATE MUSEUM canal systems; (2) discreteness of the central elements and their separation on either side of the middle line by the median occipital; (3) reduced num- ber of median series of plates; and (4) absence of any evidence of articula- tion or other connection between the head shield and dorsal body plates. We have next to speak of the configuration of the basis cranii, by which term is understood, of course, not the inner surface of the cranial buckler, but the floor and posterior wall of the chondrocranium, the passage- way provided for the initial portion of the vertebral axis, ossified para- chordal cartilages, and the greatly elongated parasphenoid which forms the roof of the mouth. In the first place it should be noted that a bony exten- sion of the external occipitals is developed along the posterior margin of the head shield in the form of a thin lamina or septum which extends nearly vertically downwards into the soft parts until reaching the level of the parasphenoid, with whose hinder extremity it joins. It is along the plane of this septum that the exoccipito-central canals penetrate downward and inward from the outer surface by means of funnel-shaped openings, the interior of which is filled with cancellated bony tissues. The function of the posterior septum seems to have been to impart rigidity to the arch of the head shield, and to serve as a partial support for the parasphenoid The last named element resembles in a general way the familiar lozenge-shaped bone in Dipterus and modern Dipnoans, but is remarkable for its great expansion in front, where it occupies nearly the entire width of the head shield. Becoming narrower in the occipital region, it is con- tinued backward as far as the posterior margin in the form of an arched laminar plate, not unlike that of Neoceratodus in form, upon which rests the parachordal cartilages and notochordal sheath. This hinder portion of the parasphenoid was interpreted by Cope as consisting of a pair of distinct elements, called by him the ‘ lateral alae of the axis,” or, in another place, ‘descending osseous laminae”, but it is clear from well preserved speci- mens that only a single ossified element is attached to the floor of the carti- laginous cranium. A right understanding of this feature shows that in the form under discussion the parasphenoid is produced posteriorly to the same DEVONIC FISHES OF THE NEW YORK FORMATIONS 109g extent as in Neoceratodus and Lepidosiren; hence Cope’s statement requires rectification when it is said that the parasphenoid in modern forms is abnormally produced behind.t. The extreme thinness of the bone in its anterior portion stands in decided contrast to the solidly ossified plate of Ctenodipterines, and it is further noteworthy that no specimen has yet enlightened us as to its relations with the palato-pterygoid cartilages. Near the point of its greatest constriction, in what corresponds to the position of the quadrate bone, is a well marked oval concavity, described by Cope as a “glenoid fossa”; and this may not improbably be looked upon as having served for attachment of the mandibular suspensorium. Nothing whatever is known of the characters of the latter, nor of the dentition; and accord- ingly these structures are inferred to have been cartilaginous. Reference has been made to the tubular sheath which inclosed the notochord, and this we have stated to be supported below by the narrow pos- terior projection of the parasphenoid, miscalled by Cope “ descending axial alae.” The region traversed by the chordal sheath within the interior of the head shield has been variously interpreted, Newberry having described it asa “cerebral,” and Cope as a “nuchal” chamber, both authors evidently regarding it as completely closed on all sides. As a matter of fact, the space designated by these writers as a “chamber” was closed at either end by thin osseous partitions, and above by the cranial roof ; but it was clearly open below on either side of the constricted portion of the parasphenoid. We can not imagine this partially inclosed space to have been the seat of any organ, although likely enough it contained fatty matter. Bearing in mind these differences of interpretation, it may be profitable at this point to examine Cope’s account of the conditions observed by him in the holo- type of the species, since unfortunately destroyed by fire. The following extract occurs at page 452 of the article already quoted: Turning now to the inferior aspect of the skull, we observed, at the « “ The parasphenoid in both Lepidosiren and Ceratodus is produced abnormally, and it is only necessary to imagine this part to be reduced to its normal length to have the conditions found in Macropetalichthys.” [Cope, E. D. On the Characters of some Palaeozoic Fishes. U.S. Nat. Mus. Proc. 1891. 14: 455] IIo NEW YORK STATE MUSEUM middle line of the inferior-posterior border, a wide, upward excavation, look- ing backwards and downwards. It rapidly contracts into a groove with an angular superior middle line. Whether this groove is part of a tube can not be ascertained, owing to the loss of the bony tissue on each side and below, but it may be only the apical angle of a roof-shaped space, whose lateral slopes are produced on each side, sloping well downwards and out- wards. ‘These sloping faces of the matrix represent a pair of osseous plates, which descended on each side from the sheath of the myelon and chorda dor- salis, for the latter occupied this position in the groove already described. Such a structure would indicate the presence of a number of fixed vertebral elements, such as exists in the chimaeras, the rays, and the sturgeons. . At the point of junction of the parasphenoid with the lateral alae of the axis is situated what I suppose to be the foramen magnum. It is the direct continuation of the groove already described, and, being floored by the parasphenoid, has a triangular section. There is no trace here of a fossa for the chorda dorsalis, nor of an occipital condyle, nor is it probable that either existed at this point. The parasphenoid is thin, and there are no indications of teeth to be observed on it. It remains for us to state that the thin, vaulted and backwardly curved partition which closed the so called “cerebral chamber” in front is in reality nothing more nor less than the posterior wall of the chondrocranium. It is proved to be such by the notochordal opening above described ; by its form and position, which are extremely suggestive of Dipterus;* and by its sus- pension from the cranial roof ina manner recalling that in Neoceratodus, It was, however, but slightly ossified, the bone substance being everywhere thin, and no distinct exoccipital plates being formed about the foramen mag- num. Nothing can be stated of the lateral walls of the chondrocranium, for the reason that they have not been observed in any specimen yet dis- covered; and it is quite likely that they were unossified. Appearances in Chelyophorus and Homosteus lead us to anticipate that further details may be forthcoming in regard to the chondrocranium and parachordal cartilages than is possible to learn from the less ossified condition of these parts in Macropetalichthys. Abdominal armoring. Neither in this nor other species of Macropet- alichthys, nor in the allied genus Asterosteus, have plates been found in *Traquair, R. H. On the genera Dipterus, Palaedaphus, etc. Ann. Mag. Nat. Hist. ser. 5. 1878. 2:5, pl. 3, fig. 1. DEVONIC FISHES OF THE NEW YORK FORMATIONS III natural association with the head shield, or even in the detached condition, which can be interpreted with any degree of assurance as belonging to the dermal covering of the trunk. Analogy with other Arthrodires is indeed suggestive of the development of body armor; but on the other hand regard must be paid to the fact that no indications of overlap nor of an articular joint are visible along any portion of the posterior margin, and that the sensory canals stop short of it, whereas in other genera they are continued across the cleft between head shield and body armor. Conclusions based upon negative evidence are liable at any moment to be overturned by new discoveries; but in the present case this is the only class of evidence avail- able, and the inference to be drawn from it is that the most primitive Arthrodires resembled Ceratodonts in being unprotected by abdominal armor. In all groups the development of an extensively ossified exoskeleton is looked upon as evidence of specialization. Its nondevelopment amongst forms that are clearly primitive in other respects would be, therefore, in nowise startling. Moreover, it should be remembered that within a single family of Ostracophores, Pterichthys is scaled, Bothriolepis naked. In view of the remarks that have just been offered it is scarcely neces- sary to say that the view formerly entertained by the present writer, and more fully developed by Dean in his definition of ‘“Anarthrodira,” accord- ing to which all that portion of the head shield lying posterior to the chon- drocranium corresponds to the dorsal body plates of typical forms, is now rejected as erroneous. The median and external occipital plates in this species are not superimposed upon a separate system of underlying plates, and whatever notion may have existed to the contrary may be set down to deceptive appearances. Thus, the plates called by Newberry “ parietal,” “squamosal,” and “epiotic,” in the region covered by the external occipital, have no real existence, the supposed sutures between them being merely the impressions of oblique ridges and other markings of the single large plate which forms the posterolateral border of the head shield. The disso- ciated elements belonging to a single individual are shown in plate 9, figure 5. 112 NEW YORK STATE MUSEUM Formation and locality. Onondaga limestone; Leroy, N. Y. Colum- bus and Delaware limestones; Ohio. ‘‘Corniferous” limestone; Indiana and (?) Canada. Macropetalichthys agassizi (von Meyer) 1845 Asterolepis hoeninghausii ZL. Agassiz (errore). Poiss. Foss. V. G. R. p- 130, 147, pl. 30a, fig. 10 1846 Placothorax agassizi HA. von Meyer. Neues Jahrb. fiir. Min. p. 596 1847 Placothorax agassizi A. von Mever. Palaeontogr. 1: 102, pl. 12, fig. 1 1855 Physichthys hoeninghausii ZA. von Meyer. Palaeontogr. 4: 80, pl. 15, fig. 1-5 (won figs. 6-11) 1857 Agassichthys agassizi /. S. Mewberry. Nat. Inst. Bul. p. 119 1873 Macropetalichthys agassizi /. S. Newberry. O. Geol. Sur. Rep’t, Pal. v. 1, pt 2, p. 291 1895 Macropetalichthys agassizii A. won Koenen. Abhandl. Ges. Wiss. Gottingen, 4o: 22, pl. 4, fig. 3 Reference has already been made to the fact that the type of Hermann von Meyers Physichthys hoeninghausii is now the property of the Museum of Comparative Zoology at Cambridge, Mass. Although incomplete, the part that is preserved shows several of the cranial elements very distinctly, and also the narrowed posterior portion of the parasphenoid which supports the notochordal sheath. This is deeply channeled in the median line for the passage of the notochordal sheath, and is concentrically striated in the same manner as the vertical lamina which descends from the posterior margin, The sheath itself exhibits no trace of segmentation, and, like that in the type species, is of remarkably small diameter. Formation and locality. Middle Devonic; Eifel district, Germany. Genus ASTEROSTEUS Newberry An imperfectly definable genus, known only by the median series of cranial plates, but apparently closely approaching Macropetalichthys in structure. Ornamentation consisting of large, rounded, stellate tubercles, very irregular in size and arrangement; sensory canals indistinct ; 2 pineal foramen, and also a pair of large oval openings of unknown significance present in the interorbital region. DEVONIC FISHES OF THE NEW YORK FORMATIONS I13 Asterosteus stenocephalus Newberry 1875 Asterosteus stenocephalus /. S. Newberry. O. Geol. Sur. Rep’t, Pal. v. 2, pt 2, p. 36, pl. 54, fig. x 1889 Asterosteus stenocephalus /. S. Mewberry. U. S. Geol. Sur. Monogr. 16: 45, pl. 30, fig. 1 1891 Asterosteus stenocephalus 4. S. Woodward. Cat. Foss. Fishes Brit. Mus. pt 2, p. 312 Only a few examples are known of this species, none of which display the entire width of the head shield; accordingly, previous descriptions which represent the latter to be narrow and elongate are to be understood as applying only to the median series of plates, between which sutures are not distinctly traceable. The large, mesially placed openings between the orbits are regarded by Woodward as nasal, an interpretation which we are inclined to consider extremely doubtful in view of the invariable position of the nasal openings in other Arthrodires, Sirenoids and Ctenodipterines. The posterior flangelike projections on either side, called by Newberry ‘‘condyloid prominences,” correspond to the semicircular ridges just back of the quadrate region on the inferior aspect of the skull in Macrope- talichthys, and may have served for the attachment of the mandibular sus- pensorium. Jaws and dental elements unknown. Formation and locality. Columbus and Delaware limestones; Ohio. Possibly also in the Onondaga limestone of New York. Family COCCOSTEIDAER Cranial shield consisting of few elements, viz, a median and two external occipital plates, in front of which is a single pair of large centrals more or less in contact along the median line; these are in turn preceded by a pair of large preorbitals, which are either wholly or partially separated from each other by the azygous pineal and rostral (“‘ethmoid”) elements; pos- terolateral border of the shield formed by the marginals and postorbitals, Orbits not completely inclosed within the shield, bounded inferiorly by a single suborbital plate, behind which occur one or two opercular elements. Upper dentition consisting of a pair each of vomerine and palato-pterygoid II 4 NEW YORK STATE MUSEUM elements (commonly interpreted, however, as “ premaxillaries” and ‘“max- illaries”), the latter with trenchant functional margin, often serrated or denticulated. Lower dental plate, when ossified, intimately fused with the forward portion of the mandible, turned in upright position with sharp cut- ting edge, often serrated or denticulated like the upper. Symphysial margin also sometimes denticulated. The structure of the typical genus, Coccosteus, is so well known from the classic researches of Pander, Traquair and later writers, that further description of it here would be superfluous. It furnishes a most valuable standard of comparison for referring the detached plates of other Arthro- dires to their proper position, and for correlating the numerous minor variations observed in different members of the group. Interest centers in it also from the fact that this is commonly looked upon as the progenitor of later, much larger, and more highly specialized genera, occurring chiefly in this country, the best known examples of which are Dinichthys and Titanichthys. Owing to their intimate relations to Coccosteus, these two genera are retained in the same family with it. We have already had occa- sion to note the close structural agreement between Coccosteids and existing Dipnoans, more particularly as regards their dentition and cranial osteology, hence we need not dwell upon these points further in this connection. No traces of axial segmentation have been observed in any member of this family, nor are there ossified ribs, Jaekel’s representation to the con- trary notwithstanding.?. There is one dorsal fin, the tail tapers gradually and to all appearances was diphycercal, but definite information concerning the caudal and anal is still lacking. Traces of pelvic fins occur, and a pair of short deep plates lying immediately in front of the ventral armor in Coccosteus has been interpreted as representing the pectoral arch. Never- theless, positive evidence of the existence of pectoral fins has not as yet been forthcoming, and the so called pectoral spine or “ Ruderorgan "in Coccosteus and Brachydirus is possibly the equivalent of the fixed spinous » Jaekel, O. Ueber Coccosteus und die Beurtheilung der Placodermen. Sitzungsber. Ges. Naturf. Freunde. 1902. p. 107, restoration. DEVONIC FISHES OF THE NEW YORK FORMATIONS 115 process in Phlyctaenaspis, Acanthaspis etc. At the same time it is worthy of remark that both in form and position it is not wholly dissimilar to the rodlike operculum inferzus, as Fiirbringer calls it, of Neoceratodus. This author’s observation that the two opercularia of the recent form are some- times fused into a single piece acquires significance on recollecting that only a single opercular element is known to occur in most Arthrodires.* Jaekel is the only author who has reported the occurrence in Coccosteus of two opercula, this being the normal number amongst Dipnoans, as already stated. Genus coccosTeus Agassiz Of the four American species that have been referred to this genus, only one, C. canadensis Woodward, is satisfactorily known, the others being represented by detached plates exclusive of the head shield. To C. occidentalis, described in the first instance by Newberry from the Corniferous limestone of Ohio, are possibly to be referred a few isolated fragments occurring in the New York Mesodevonic, and it has been further surmised by the original author that the dental plates known as Liognathus spatulatus belong to the same species. In plate 9, figure 3, the dorso- median plate of the type specimen is refigured to show the continuation of sensory canals over part of its surface, their presence having escaped New- berry’s attention. No figures have been published of the form described by Cope from the Chemung of Leroy, Pennsylvania, under the name of C. macromus, but it is said to be distinguished from C. occidentalis by its coarser tuberculation. In plate 1, figure 9, is shown a dorsomedian plate of a small Coccostean fish from the Onondaga limestone of Clifton Springs, N. Y., the original of which is preserved in the State Museum at Albany [cat. no. 181]. A small portion of the bone substance is preserved intact on the right-hand side, and shows a coarsely tuberculated ornament. The remainder of the plate is broken away, leaving an impression of the under surface. There is also visible a strong longitudinal keel which extends along the median line, but : Fiirbringer, K. Beitrage zur Morphologie des Skeletes der Dipnoer etc. Semon’s Zool. Forschungsreisen in Australien. Jena Denkschr. 1904. 4: 493. 116 NEW YORK STATE MUSEUM terminates abruptly at some distance in advance of the posterior margin. The latter is continued backward in triangular fashion, but is not produced into a long spine as in C. occidentalis. This unique specimen, which probably belongs to a distinct species, is interesting in view of the horizon from which it was obtained, and on account of the great rarity of Coccosteus remains in this country. It seems inadvisable to propose a new specific title for it, however, until other parts of the skeleton are known. Fig. 22 Coccosteus decipiens Agassiz. Lower Old Red sandstone; Scotland. Lateral aspect, restored by Dr A. S. Woodward, partly after Traquair. x 1%. The caudal fin is here conjecturally restored as heterocercal, but more probably it was diphycercal. Coccosteus macromus Cope 1892 Coccosteus macromus Z, DV. Cope. Am. Soc. Phil. Proc. 30: 225 Of this species no other examples are known except those obtained by Professor Cope, whose original description follows : Fragments of this species are abundant in the Chemung rocks at Leroy [Pa.], and I select as typical of it a pair of supraclavicular and adjacent pieces, which display its characters best. The supraclavicle has lost the condylar articulation. Both extremities display the unsculptured surface, and the usual groove extends obliquely across the sculptured portion at about two fifths the length from one of the extremities. The sculpture con- sists of obtuse tubercles with delicate radiate-grooved bases, which are usu- ally separated by spaces equal to their own diameters, sometimes by narrower spaces, but never by spaces which are wider. At some points they have a linear arrangement. This sculpture is coarser than in the C. americanus [zc C. occidentalis] Newberry [sce The Palaeozoic Fishes of North America, by this author], but resembles that of C. decip- iens Agass. of Scotland. From this species the C. macromus differs in the elongate form of the supraclavicle which is relatively short and wide inthe C. deci piens [sce Agassiz, in the Poissons du Vieux Grés Rouge, and Zittel, Handbuch der Palaontologie}, Length of supraclavicle - - - - 35mm Width just above condyle - - - . * 296°" Formation and locality. Chemung beds (Chautauquan); Leroy, Pa, DEVONIC FISHES OF THE NEW YORK FORMATIONS 117 Genus ACANTHASPIS Newberry An imperfectly definable genus of Coccostean fishes, known only by detached plates resembling the antero-ventrolaterals of Phlyctaenaspis. Acanthaspis armata Newberry Plate 2, figure 2 1875 Acanthaspis armatus /. 8. Newberry. Geol. Sur. O. Rep't. Pal. v. 2, pt 2, p. 37, pl. 53, fig. 1-6 1889 Acanthaspis armatus /. S. Newberry. U S. Geol. Sur. Monogr. 16: 36, pl. 31 1894 Acanthaspis armata R&R. H. Traquair. Ann. Mag. Nat. Hist. ser. 6, 14: 371 1896 Acanthaspis armata JZ. II. Claypole. Am. Geol. 17: 354 The spiniferous plates which are known under this provisional designa- tion have not been found in natural assemblage with other parts of the skeleton of forms accompanying them in the same formation, hence their precise relations are indeterminable. Theoretical associations are for the most part valueless, except as they rest upon strongest possible presumptive evidence; and in the present case sufficient data for comparison is lacking. The superficial ornament somewhat resembles that of Macropetalichthys, but according to all appearances that genus was destitute of dermal body armor- ing. For the present we are obliged to regard Acanthaspis as a provisional genus, whose relations may be assumed to be in the vicinity of Phlyctae- naspis, and not, as was supposed by Newberry, with any member of the Cephalaspidae. None of the plates in question that have thus far come to light seem to be preserved in their entirety, but the small specimen shown in plate 2, figure 2, is perhaps as complete as any. The lateral spine is always immovably attached to its supporting plate by an oblique suture, which is more conspicuous in larger examples. The interior of the spinous portion is hollow throughout, and along the line of its attachment with the supporting plate is sometimes to be seen a double row of perforations which communi- cate with the internal cavity of the spine. 118 NEW YORK STATE MUSEUM It is inconceivable that this fixed spinous appendage is any way homolo- gous with the jointed pectoral limbs of Asterolepids. Plates not unlike those of the present species have been described by Traquair from the Lower Devonic of the Eifel district under the name of A. pruemensis. The so called A. decipiens of Smith Woodward, from the Lower Devonic of Spitzbergen, appears to be of different nature. Formation and locality. Onondaga limestone; Leroy, N. Y., and equivalent formation of Ohio. Genus pDinicHrHys Newberry It is difficult to frame a satisfactory diagnosis of this genus which shall enumerate its principal characters and at the same time enabie one to draw a rigid distinction between its various species and those of Coccosteus. The fact is, the two genera are most intimately related, and though their terminal members are sufficiently well characterized, they are connected by insensible gradations. The typical species of Dinichthys represent unques- tionably a later and more advanced stage of specialization than that with which we are familiar in Coccosteus decipiens, for example; but between these extremes lies a host of intermediate forms. Evidence of specialization in forms like D. herzeri, D, terrelli ete. is strikingly apparent in their gigantic size, the head shield measuring nearly a meter across, and their massive and cumbersome armor being unrivaled amongst fishes. As a necessary accompaniment of increase in size, the cranial plates become more intimately fused in the adult, the articulations between head shield and abdominal armor more complicated, and various minor modifica- tions are to be seen in the dentition and arrangement of the body plates. Nevertheless, in spite of hypertrophic enlargement of all the parts, there is everywhere a surprising conformity to the basal type of Coccosteus. It is necessary to insist upon this close correspondence, which has hitherto escaped attention (owing to faulty restorations and other reasons), and possibly even now might be questioned by those who have not actually compared the different parts. DEVONIC FISHES OF THE NEW YORK FORMATIONS 11g Adopting for our thesis the obvious fact that Dinichthys and Cocco- steus are constructed upon the same general type, we may now take note of such differences as exist between them, other than mere size; and for immediate purpose of comparison it will be convenient to choose D. herzeri, D. terrelli and D. intermedius as representatives of the former, and C. decipiens as representative of the latter genus. First, as to the abdominal armor. Here we observe a perfect correspond- ence throughout, both as to number, arrangement and general form of the plates [cf text fig. 23]. The dorsomedian of Dinichthys is relatively shorter than that of Coccosteus, more emarginate in front and rounded behind, and its inferior keel is developed into a stout terminal process. There are not usually traces of sensory canals upon its external surface. In Coccosteus, the anterolateral is bounded in front and below by a distinct plate, the interlateral, which forms the connection between the dorsolateral and ventral portions of the abdominal armor. In Dinichthys, the interlateral does not occur as a distinct plate, but may be represented by the forklike prolonga- tion of the so called “clavicular,” the broad upper portion of which cor- responds, of course, to the anterolateral. The median ventrals of Cocco- steus are always separate: in Dinichthys they are occasionally fused. Secondly, as to the head shield. Again there is notable correspondence of plate for plate, and in the arrangement of sensory canals. The bones are more intimately fused in Dinichthys than in Coccosteus, and the sutures are less wavy. The median occipital is generally pointed in front [cf text fig. 23], the pineal plate is lancet-shaped and produced backward so as to come in contact with both the centrals and preorbitals, instead of being inclosed by the latter alone, and the plates forming the sides of the head shield are narrower than in Coccosteus. None of these differences, how- ever, can be considered other than of secondary importance, and their aggre- gate does not in the least obscure identity of structural plan common to both genera. The imaginary plate interpreted by Newberry* with some * Newberry, J. S.. U.S. Geol. Sur. Monogr. 1889. 16: 334, pl. 52, fig. 2. 120 NEW YORK STATE MUSEUM hesitation as a parietal” in Dinichthys does not occur, an examination of many well preserved cranial shields by the present writer having determined Fig. 23 Dinichthys intermedius Newberry. Cleveland shale; Ohio Restoration of head shield and dorsal armor- ing of trunk, x 1-5. 4DZ£=antero-dorsolateral; C=central; DA7=dorsomedian; LO=external occipital; AZ7= marginal; MO-= median occipital; P=pineal; PDZ=postero-dorsolateral; PO=preorbital; P¢O=postorbital; R—=rostral, sometimes identified as ‘‘nasal’’ or ‘*mesethmoid.’’ Sensory canals shown by double, overlap by single dotted lines. The markings on the outer (lower) border of the lateral plates indicate the extent of the overlap by the clavicular, which is omitted from this diagram. (From Hussakof, 1905) this point conclusively. In both genera a sclerotic ring is present, the sub- orbital plate is distinct, and followed behind by one, or possibly two, DEVONIC FISHES OF THE NEW YORK FORMATIONS 121 opercular elements, which have been variously interpreted by different writers. Finally, and this is the most important test of all, we have to consider the dentition. A serious difficulty in the way of correlating the several parts, and of understanding accounts of earlier writers, is removed the moment we perceive that the suborbital plate has nothing whatever to do with the dentition. Its identification in Coccosteus on the part of some writers as a “superior maxillary” is not only erroneous, but misleading, since it implies the development of a secondary upper jaw amongst Arthro- dires, and a secondary upper jaw does not occur. The suborbital is simply a cheek plate, and corresponds to the three nonfused ossicles forming the lower border of the orbit in Neoceratodus, The dental elements of both jaws are precisely alike in Coccosteus and Dinichthys, as regards number, form and arrangement. The whole matter of the dentition can be summed up by saying that Coccosteus and Dinichthys exhibit the same close paral- lelism with Protopterus and Lepidosiren as is observed between Mylostoma and Neoceratodus. Coccosteus and Dinichthys represent a modification of Dipnoan dentition adapted for cutting, and closely corresponding modifica- tions exist in the modern fauna. Mylostoma and Dinomylostoma adhere to the ancestral type of crushing or triturating dentition, which has been preserved essentially unchanged until the present day amongst the most generalized descendants of the primordial stock. It will be seen that the foregoing interpretation is at variance with the one recently proposed by Jaekel in his paper of 1902, to which frequent reference has been made. The dental plates of Coccosteus are identified in that contribution as maxillae and premaxillae, and their arrangement is expressed graphically by means of a restoration, concerning which the author speaks as follows: Ich bemerke dazu, dass ich dieses Bild durch sorgfaltige Praparation einer ganzen Anzahl von Exemplaren erzielt habe, und dass sich dasselbe von dem Traquairschen Entwurf namentlich unterscheidet durch den Nachweis der Maxillen und Praemaxillen, durch die abweichende Lage der Nasenéffnungen und der unteren Rumpfpanzerelemente, den Nachweis des Beckens und den Hinweis auf paarige Extremitaten. haz NEW YORK STATE MUSEUM When it comes to a consideration of the dental elements in Coccosteus, one is somewhat surprised to find that they are compared directly with Lac- ertilians, much in the same way as Cope compared Macropetalichthys with Stegocephalians. The continuation of the above passage reads as follows :* Ueber die Bezahnung der Kiefertheile habe ich noch keine volle Klarheit, indess scheint mir Folgendes zu rechtfertigen. Die Bezahnungs- form scheint durchaus tibereinstimmend mit der von Sphenodon. Von den Zahnspitzen des Unterkiefers diirften mindestens die vordersten dem Den- tale, die hinteren, mehr einwarts gelegenen vielleicht dem Spleniale zuzu- rechnen sein. Die Praemaxillen scheinen entsprechend dem Vorderrand des Unterkiefers mit je einer Spitze versehen gewesen zu sein. . . Die Maxillen sind kleine schmale mehrzackige Knochenstiicke die dem Innenrand des suborbitalen Astes des beilformigen Oberkieferstiickes eingeftigt sind. Palatina und Transversa habe ich noch nicht finden konnen vermuthe aber ihre Existenz hinter den genannten Elementen. We may now give special consideration to the arrangement of dental elements in Dinichthys. In the upper jaw there are always two pairs of dental plates, which are clearly shown to be of dermal origin. It is denied “by Dean that these plates, which are commonly known as * premaxillaries ” and ‘shear teeth,” can be homologized with any structures within the mouth of other fishes, hence he proposes to call them by the noncommittal names of ‘rostro-gnathals” and “ orbito-gnathals” respectively, the term “ gnathal” being made synonymous with mandible. According to the view adopted in this report, the anterior pair of upper dental plates in Dinichthys is to be interpreted as vomerine, and the posterior as palato-pterygoid, thus recog- nizing definite homologies between them and the like named structures — which are also of dermal origin —in modern Dipnoans. The vomerine pair is situated close to the median line, one on either side of the plate answering in part to the dermal mesethmoid in Neoceratodus, and the tumid basal expansion is received into a slight concavity on the visceral surface of the preorbital. The exposed, or functional portion of the vomerine teeth is cleft so as to form two vertical prongs or “beaks” of unequal length, the "Ueber Coccosteus und die Beurtheilung der Placodermen. Sitzungsber. Ges. Naturforsch. Freunde. 1902. p. 106. DEVONIC FISHES OF THE NEW YORK FORMATIONS 123 shorter of which is inwardly placed and is scarcely separated from its fellow of the opposite side. At least one Dinichthyid individual is known in which the vomerine pair is solidly fused with the supporting bones of the head shield The specimen referred to forms part of the collection of Ohio fishes brought together by Dr William Clark and eventually acquired by the British Museum of Natural History. It is catalogued under the number P. 9490, and in the opinion of Dr A. S. Woodward, keeper of the Geological Depart- ment, properly belongs to the genus we are now considering. Another interesting specimen having the upper and lower dental elements preserved in situ, is preserved in the same collection under the catalogue number P. 9340. By means of this and similarly preserved material, and also by examining the marks of wear in upper and lower dental plates, one may readily determine the manner in which the dental parts came together when the mouth was closed. The vomerine teeth are seen to have protruded just a trifle in advance of the mandibular beaks, which closed within the angle formed by the prongs of the opposing elements. A few plates of Dinich- thyid armor are also known, which have been bitten by creatures of their own kind in such wise as to leave the impress of teeth; and these markings confirm in an unexpected way the observations we have just made concern- ing the arrangement of jaw parts, besides affording indications of the dis- tance apart of the symphysial beaks. The best illustration that has been offered of the manner in which the upper and lower dentition came together in front will be found in the frontispiece of Dr Bashford Dean’s work on Fishes, Living and Fossil. Some novel suggestions concerning the inter- play of the jaw parts, which are assumed to have been freely mobile, are contained in Dr L. Hussakof’s Studzes on the Arthrodtra, published in volume IX of the Memoirs of the American Museum of Natural History, 1906. The vomerine teeth are succeeded almost immediately behind by the cleaverlike palato-pterygoid plates, called by Dean ‘‘orbitc-gnathals,” but commonly known as “shear teeth,” in allusion to their mode of working 124 NEW YORK STATE MUSEUM against the trenchant margin of the lower dental plates like the blade of a pair of shears. The manner of their operation remains the same even in those species where the opposing margins are denticulated, a condition which is regarded as more primitive than that with simply sharpened, or beveled edges. Traces of an original denticulation, which once extended along the entire functional margin, are often observed in the form of tubercles, or denticles, whose position is confined in specialized species to the extreme posterior margin of the tooth. It is to be noted that this posterior margin is usually narrower and more rounded than the anterior. Slightly in advance of the middle portion of the tooth, along its superior margin, there is given off from this upper margin a well marked, inwardly curved ascending process or ‘‘shoulder,” which corresponds without question to the similarly placed process of Ceratodont dental plates. Notwithstand- ing the large size and evidently great efficiency of the shear teeth, they do not appear to have been rigidly attached to the head shield, but rather to have been held in place by cartilage against the prominent inferior ridge which extends forwards as far as the orbital region from the posterolateral angles, in a direction parallel with the sides of the head shield. Precisely similar conditions are observed in Neoceratodus, where the ridges referred to serve as a support for the upper dental plates, and relieve the strain imposed by the action of the jaws in mastication. It is interesting to note that these ridges along the underside of the head shield in Dinichthys acquire greater solidity in proportion as the dental plates become more massive and powerful. Finally, attention may be directed to the lower dental plates strictly speaking, the term of ‘dental plate” being properly limited to the func- tional, anterior portion of the mandibular arch, with which element it is fused. The relations between the lower dental plate and the supporting ossification, or splenial, and the manner in which it is supposed to have become rotated so as to stand upright in the jaws, are matters that have been sufficiently explained in the general account of Arthrodires [cf supra, p- 96]. In almost all species of Dinichthys, the anterior extremity of DEVONIC FISHES OF THE NEW YORK FORMATIONS 125 the mandible is developed into a powerful cutting beak, usually much facetted by wear. The marks of contact with the opposing pair of vomerine elements were attentively examined by Newberry, who concluded that they closed together in the following manner: “The incisorlike teeth of the pre- maxillaries interlock with and shut over the projecting poe of the turned up mandibles, which are received into their concavities.” Again, in describ- ing these elements in D. terrelli, he observes: ‘The inner [—poste- rior] side is concave and frequently much worn and excavated by the promi- nent extremity of the mandible, over which it shuts.” * Some distance behind the symphysial beak, the functional margin of the Jower dental plate rises again into a prominent projection, shorter, how- ever, and less massive than the first, and appearing on the inner aspect as a distinct riblike swelling, nearly vertical, and evidently in the nature of a rudimentary tooth. From this point backward along the functional margin, the lower dental plate is compressed into a thin edge, beveled somewhat on the outer face by contact against the opposing element of the upper jaw. In the majority of species, the margins of both upper and lower dental plates are smooth and bladelike; but a few, including the type, have them denticulated as in Coccosteus. Vestigial remnants of a primitive Cerato- dontlike denticulation occur in the lower dental plates along the abrupt downward slope of their posterior margin; that is to say, in a position cor- responding to that in which they are seen in the upper. The line of fusion between the lower dental plate, in the strict sense of the term, and the supporting ossification, which we take to be the splenial, is marked externally by a prominent constriction, starting from a point slightly behind the functional margin, and extending in a curved line down- ward and forward uatil reaching the bottom point of the anterior or sym- physial margin. Seen from the external aspect, the lower dental plate thus appears to be set off from the slender, elongate shaft of bone which we iden- tify as splenial, by means of a conspicuous shoulder, underneath which there runs a groove for the lodgment of portions of the Meckelian cartilage. * Newberry, J.S. Ohio Geol. Sur. Rep’t. Pal. 1875. v. 2, pt 2, p. 5, 29. 126 NEW YORK STATE MUSEUM The correctness of our interpretation of the groove in question is clearly established through comparison with Protopterus and Lepidosiren.: Only in one respect is there a marked distinction between the mandibles of Dinichthys and those of Coccosteus; no species of the former genus are known to have denticulations along the symphysial margin. On the other hand, a series of tubercles is sometimes present along the outer face of the vomerine teeth in Dinichthys, and this fact suggests the probability of tubercles or denticles having been present in the lower dental plates as well, at an earlier period in the history of these forms. The differences between Dinichthys and Titanichthys will be considered in the discussion of the latter genus. Dinichthys halmodeus (Clarke) Plate 2, figure 7; plate to, figure 1; text figure 24 1894 Coccosteus (?) halmodeus /. M. Clarke. N.Y. State Geol. Rep’t 1893, 1: 162, pl. 1 1900 Dinichthys halmodeus C. R. Eastman. Jour. Geol. 8: 34 1906 Dinichthys halmodeus ZL. Hussakof. Am. Mus. Nat. Hist. Mem. 9, p. 140, text fig. 22C, 244. A primitive species of small size, the head shield having a total length of about 11 cm, and very similar to Coccosteus in the configuration of plates, arrangement of sensory canals, and character of the superficial ornamentation. The suture lines, however, are less undulating than in Coccosteus, the articulation with the abdominal armor is much stronger, the pineal is partly in contact with the centrals, and the dentition is char- acteristically Dinichthyid, with strongly developed vomerine teeth. The anterior margin of the lower dental plates is developed into a prominent beak, and the superior or functional margin is strongly denticulated; the posterior extremity of the splenial is broad and spatulate. The suborbitals are unusually wide and massive, and the rostral seems to have been laterally expanded in front. The dorsomedian bears the usual inferior keel, its *Wiedersheim, R. Morphologische Studien. Heft 1, p. 55, pl. 2, fig. 3, 8 Jena. 1880; Fiirbringer, K. of. cit. p. 481, pl. 39, fig. 28; Bridge, T. W. Morphology of the Skull in Lepidosiren. Zool. Soc. Trans. 1898. 14: 342, pl. 28, fig. 7, 8. DEVONIC FISHES OF THE NEW YORK FORMATIONS 127 terminal process being given off at a slight distance in advance of the posterior margin. In the minute and in most respects very accurate description given of Fig. 24 Dinichthys halmodeus {¢Clarke). Erian; Livonia Salt Shaft, N.Y. Restoration of head shield and dorsal armoring of trunk. x 344. 4ADL=antero-dorsolateral ; C=central ; D/¢/=dorso- median; £O=external occipital; 4/=marginal; 47O=median occipital; P=pineal: PDL=postero- dorsolateral ; R=rostral ; SO=suborbital the type specimen by the original author, it is clearly stated that the denti- tion presents ‘‘an aspect highly similar” to that of Dinichthys, comparisons being made between the present species and D. herzeri, which also has 128 NEW YORK STATE MUSEUM the functional margin denticulated. The observation is of sufficient weight, in our opinion, to justify the transfer of this species from Coccosteus, where it was doubtfully placed by Dr Clarke, to Dinichthys. Of special signifi- cance from a systematic standpoint are the vomerine teeth, which are described as follows by the same author: ‘Each is concave on the inner surface, convex externally, and bore a somewhat extended apophysis, which in each case has been broken off. There is no evidence of denticles or a tuberculated surface; the lower edge is, however, rather sharp and would have served a cutting purpose.” Amongst other minor peculiarities of the head shield may be mentioned the subquadrate outline of the median occipital, and the strong outward flexure of the preorbital sensory canals -— characters which one may regard as of specific importance. Associated with the head shield, and possibly belonging to the same individual, certainly to the same species, is a specimen interpreted by Dr Clarke with some reservation as a dorsomedian plate, and considered by him to be divided by sutures into three parts. In the explanations of plates, however, it is conceded that ‘the surface shows no satisfactory evidence of sutures,” and after a careful inspection of the original, the present writer is convinced that the plate is normal in all respects, except that the external surface has been somewhat injured. That others may judge of the char- acters of this plate, an illustration is given of it in plate 2, figure 7. It has a total length of 9 cm, and apparently did not exceed 2 mm in thick- ness; the carinal process has been broken away. The locality from which both of these interesting specimens were obtained was the Livonia salt shaft, in Livingston county, New York, at a depth of 828 feet below the surface. They are preserved in the New York State Museum at Albany. Another interesting cranial fragment belonging to the same museum, and from the same horizon as the preceding, is shown in plate ro, figure 1. In size, it indicates an individual about one third larger than the type of D. haimodeus, and the tuberculation is relatively coarser, but the form of the median occipital element agrees so closely with that of the latter species, it is difficult to regard this as other than an adult example; DEVONIC FISHES OF THE NEW YORK FORMATIONS 129 besides, it is a well ascertained fact that the tubercles increase in size with age. The median occipital has an extreme length of 5.5 cm, and thickness of about 3mm. The specimen is labeled as having been found “ Sept. 9, 1851, in the Goniatite limestone‘ at Hendrick’s ledge, west of Manlius, N.Y.” There is no catalogue number. Formation and locality. Marcellus division (Erian) ; New York. Dinichthys lincolni Claypole Plate 7, figures 4-6 1893 Dinichthys lincolni & IV. Claypole. Am. Geol. 12: 277, text fig. p. 276 1906 Dinichthys lincolni Z. Hussakof. Am. Mus. Nat. Hist. Mem. g, p. 117, 142 Known only by a single right vomerine tooth of about the size of the corresponding element in D. intermedius Newberry, and very similar to it in general form. The external surface, however, not only along the symphysial margin, but also over nearly all of the exposed portion, is cov- ered with enlarged conical tubercles, or even denticles, which are arranged . in more or less regular vertical series. The externolateral process which serves for the attachment of the tooth to the preorbital, is well developed, but distally somewhat compressed. The unique tooth upon which this species is founded possesses a num- ber of interesting features. In the first place, as already noted by Claypole, it is singular in having the entire outer face strongly tuberculated ; and it is noteworthy that the tubercles are most conspicuously developed, so that they become in fact denticles, along the inner or symphysial margin, thus placing the species in close relation with D. herzeri. The dermal origin of the different dental plates in Arthrodires could not be more distinctly indicated than by these vestigial remnants of Uronemuslike tuberculation. Another point worthy of attention relates to the marks of contact with the lower dental plates, such as are plainly visible on the inner or poste- rior face of the tooth. It is evident that the larger and outer (ectad) prong has been considerably worn down by use, and its lower extremity blunted; *The Agoniatite limestone, which in the section referred to lies between the Marcellus and Cardiff shales, 130 NEW YORK STATE MUSEUM but on examining its inner aspect, and also that of the secondary (entad) prong, one may determine the exact position occupied by the anterior beak of the mandible when the jaws were closed. Furthermore, the beveling of the inner edge is so regular, and there is such close conformity of all the parts, that it appears practically certain that the teeth were held rigidly in place against the head shield, and incapable of motion either forwardly or laterally. Bashford Dean, however, has expressed the opinion that not only each of the mandibular rami, or ‘“‘gnathals” as they are termed by him, but the various parts of the upper dentition as well, were capable of a consider- able amount of independent motion. At the same time he acknowledges that these conditions are absolutely unparalleled amongst chordates. In cases like this, the improbable is always to be distrusted. Finally it is to be noted that a portion of the tooth is somewhat deformed — whether as the result of accident during life of the creature, or through pressure during the process of fossilization, it is difficult to say — but whatever the cause, it did not operate so as to crush the more slender prong, or rend it asunder. Formation and locality. ‘The specimen was found in the Marcellus shale at the foot of Slate Rock fall, near Geneva, Ontario co., N. Y., 25 feet below the basal limestone of the Hamilton group, in October, 1890.” The original is now preserved in the New York State Museum at Albany, 6 and bears the catalogue number ~2**. Dinichthys pustulosus Eastman Plate 2, figure 6; plate 5, figure 2, 3: plate 12, text figure 25 1897 Dinichthys pustulosus C. &. Lastman. Mus. Comp. Zool. Bul. 31: 38, pl. 3, fig. 4 1898 Dinichthys pustulosus C. &, ELastman. Am. Nat. 32: 748, text fig. 1, 2 t900 Dinichthys pustulosus C. & Lastman, Jour. Geol. 8: 32, text fig. 1 1gor Dinichthys pustulosus B&B. Dean. N.Y. Acad. Sci. Mem. 2: 122 1902 Dinichthys pustulosus O. A. St John, Am. Nat. 36: 657, text fig. 1, 2 1906 Dinichthys pustulosus LZ. Hussakof. Am. Mus. Nat. Hist. Mem. 9; p. 142, text fig. 22D A primitive species somewhat smaller than D. terrelli, generally of about the size of D. intermedius, and distinguished from both by its DEVONIC FISHES OF THE NEW YORK FORMATIONS 131 fine tuberculation, wavy suture lines, and more Coccosteuslike aspect. Lower dental plates with a simple trenchant margin, behind which there is an abrupt downward slope beset with rudimentary denticles. Shear teeth with convex functional margin, simply trenchant, and without posterior denticles so far as known. Vomerine teeth resembling those of D. inter- medius. Visceral surface of occipital region without prominent ridges, the posterior pit of the median occipital scarcely divided. Pineal plate apparently in contact with centrals, and with inconspicuous foramen. Of this species, which appears to have been rather abundant and widely distributed in the Mesodevonic of this country, nearly the entire dermal armor is known, and the whole of the dentition. Amongst the primitive characteristics of the form in question may be reckoned its fine, Coccosteus- like tuberculation, sinuous suture lines, remnants of an original denticulation along the sloping posterior margin of the lower dental plates, and the com- paratively slight development in the latter of a toothlike projection at no great distance behind the symphysial beaks. At the same time, however, it must be acknowledged that this species marks a considerable advance over typical Coccosteuslike conditions, inasmuch as the functional margin of the dental plates is no longer serrated, the dorsomedian plate has a strongly developed inferior carina and posterior process, being also emarginate in front, and a clavicular occurs of the usual Dinichthyid type. The occurrence of D. pustulosus in the Black slate of Kentucky, a horizon corresponding approximately to the Genesee of New York State, favers the supposition that it was the immediate progenitor of forms like D. intermedius and D. terrelli of the Cleveland shale, which have retained a similar form of dentition. According to this view, D. herzeri and other species in which the functional margin of the dental plates is denticulated, form a separate series, descended along collateral lines from Coccosteus, and characterized by the persistence of this distinctly Cocco- steuslike feature. We should therefore be inclined to look upon D. hal- modeus as standing in the same ancestral relations to the type species of this genus as does D. pustulosus to D. terrelli. It is noteworthy 132 NEW YORK STATE MUSEUM that D. pustulosus does not occur in the rocks of New York State, nor, so far as known, in any district eastward of Kentucky, until very late in the Devonic (Oneonta beds); whereas in the Mississippi valley region it is tolerably abundant throughout the Mesodevonic. Its advent, then, in the Hamilton limestone of the central western ‘states, is probably to be explained through immigration by way of Manitoba and Canada from Eurcpe. The arrangement of dermal plates in the head shield of this species is shown in the accompanying text figure 25, and adjoining it is placed, for 26 Fig. 25 Dinichthys pustulosus Eastm. Middle Devonic; Iowa. Restoration of the head shield, dorsal aspect, x¥%. C=central; EO = external occipital; A/ = marginal ; 470 = median occipital; P = pineal; PO = preorbital; PfO= postor bital; R = rostral, sometimes identified as nasal or mesethmoid. Sensory canals represented by double dotted lines. Fig. 26 Neoceratodus forsteri (Krefft). Dorsal aspect of cranial roof, drawn as if flattened out to same extent as in Dinichthys. Cartilaginous portions dotted, and dermal plates lettered to correspond with those of Arthrodires. The undivided, anterior median plate is commonly termed mesethmoid, x4 sake of comparison, one showing the cranial roof of Neoceratodus. Making proper allowance for the fact that the preorbital plates remain cartilaginous in the recent form, and that the anterior median element (‘‘ dermal meseth- moid”) is undivided, as it is also in Macropetalichthys, the correspondence in pattern will be sufficiently obvious. The significance of these resem- blances can hardly fail to be appreciated, after attention has once been fixed upon them, and they are considered’ in connection with other points of agreement throughout the entire skeleton. We need not dwell upon these matters further here than to say that all available evidence goes to show DEVONIC FISHES OF THE NEW YORK FORMATIONS 133 that the Arthrodiran skull was constructed upon essentially the same model as in Neoceratodus; and the latter, accordingly, furnishes a most valuable criterion for interpreting structural details of the former. Formation and locality. The remains of this species occur abundantly in the Hydraulic limestone of Milwaukee, Wis., and are not uncommon in the Mesodevonic limestone of Iowa and Illinois. Characteristic fragments have been found in the Black slate (—New Albany, equivalent to the Genesee) near Lexington, Ky.; and portions of the dentition, including more than one well preserved mandible, have been obtained from the Oneonta beds (Senecan) near Delphi and Oxford, N. Y. Dinichthys newberryi Clarke Plate 6, figure 2 1885 Dinichthys newberryi /. M. Clarke. U.S. Geol. Sur. Bul. 16, p. 17, pl. 1, fig. 1 1889 Dinichthys newberryi /. 8. Newberry. U.S. Geol. Sur. Monogr. 16: 153 1897 Dinichthys newberryi C. R&. Eastman. Mus. Comp. Zool. Bul. 31: 30, pl. 1, fig 2 1906 Dinichthys newberryi ZL. Hussakof. Am. Mus. Nat. Hist. Mem. g, p. 145 Mandibles attaining a total length of 28% cm in the type specimen, with very prominent anterior beak, simple trenchant margin, and closely resembling that of D. pustulosus in general outline. There are, how- ever, no denticulations or tubercles along the downward slope immediately behind the cutting margin, and the other plates associated with the type specimen have a smooth external surface. A single dorsomedian plate from the same horizon as the type, and considered by Dr Clarke to be specifically identical with it, is thus described by him: In the same Styliola layer as it outcrops on the east side of Canan- daigua lake, near Genundewah, 6 miles from the Bristol locality, I had earlier discovered a dorsomedian plate belonging presumably to the same species. Its dimensions are as follows: length, 12% cm (broken) ; width anteriorly, 1334 cm; hight of carinal process, 5 cm. . . . The posterior edge of this plate in D. newberryi is broken and has apparently lost 3 or 4 cm from its length. The smallness of the bones of D. newberryi does 134 NEW YORK STATE MUSEUM not indicate immature growth of an individual of either of the other species (ie. D. terrelli or D. herzeri). The discovery in outcrops of the same horizon, in localities separated by a distance of several miles, of bones of different individuals, all of which seem to agree with one another in their relative proportions, is at least presumptive evidence that these individuals had attained maturity and that the size of the bones given above is that of normal full growth. The entire ventral armor of a small Dinichthyid, doubtfully referred to this species, and also a pair of mandibles preserved on a single slab, were obtained some years ago by Mr F. K. Mixer from the Black Portage (Rhine- street) shale at Sturgeon Point, on the shore of Lake Erie near Buffalo. The originals, which have been already described by the present writer, are now preserved in the collection of the Buffalo Society of Natural Sciences. Formation and locality. Typically from the Styliola layer (Genundewa limestone) of the Genesee shale (Senecan) ; vicinity of Bristol Center and Canandaigua lake, New York. Possibly also from the Portage beds near Buffalo, New York. Dinichthys ringuebergi Newberry 1884 Dinichthys minor & WM. S. Ringueberg (errore). Am. Jour. Sci. [3], 27: 476, text fig. 1, 2 1889 Dinichthys ringuebergi /. S. Mewberry. U.S. Geol. Sur. Monogr. 16: 60 1897 Dinichthys ringuebergi C. R. Zastman. Mus. Comp. Zool. Bul. 31: 40 An imperfectly definable species, founded upon a detached dorsome- dian plate having a total length of 16 cm, exclusive of the well developed carinal process. The external surface is described as having a “ fine grained rugose appearance ” [— finely tuberculate ?] and the anterior border is less emarginate than in most species. The type specimen is preserved in the private collection of its first describer, Mr E. N. S. Ringueberg, of Lock- port, N. Y. A second specimen, presumably of identical nature with the type, and from the same horizon and locality, was obtained by Mr F. K. Mixer a few years ago, and is now preserved in the collection of the Buffalo Society of Natural Sciences, Formation and locality. Black Portage (Rhinestreet) shale; Sturgeon Point, near Buffalo, N. Y, DEVONIC FISHES OF THE NEW YORK FORMATIONS 135 Dinichthys dolichocephalus sp. nov. Plate 5, figure 1 The remains which we propose to describe under this title leave much to be desired in the way of preservation, hence it is impossible to gain more than a superficial idea of the peculiarities presented by the new form. Inasmuch, however, as a number of plates, including the head shield and mandibles, are here found in natural association, and as we have evidently to deal with a form partaking of the characters of both Dinichthys and Coccosteus, some notice of the material seems desirable. Taken in connec- tion with D. halmodeus, from the Marcellus stage, the present species will at least serve to emphasize two facts with which paleichthyologists seem to be unfamiliar. The first of these is that to which we have already called attention, namely, as to the close relations between Dinichthys and Coccosteus ; and the second is that, little by little, we are becoming cogni- zant of a surprising variety of Coccosteuslike forms in the Middle and Upper Devonic, by means of which it may in time be possible to trace the successive stages of evolution leading up from primitive Arthrodires to the gigantic creatures which constitute one of the most remarkable features of late Devonic fish life. Naturally one turns first of all to the dentition to determine whether the characters approach those of Dinichthys or Coccosteus, and after that one considers the cranial pattern, sensory canals, and arrangement of body plates. On the evidence furnished by the mandibles, which are the only portions of the dentition preserved, one is led to decide unreservedly in favor of an association with Dinichthys; on the evidence of the head shield, and all of the body plates excepting the clavicular, the decision would be in favor of Coccosteus. We are obliged, therefore, to recognize the transi- tional nature of the species here represented, and as insufficient characters are offered for the erection of a new genus, it may be provisionally assigned to Dinichthys, The mandibular rami are unusually long and slender, their total length 136 NEW YORK STATE MUSEUM being 4.6 cm, and the anterior beak is separated by a considerable interval from the serrated functional margin. The latter terminates posteriorly in a sharp declivity, and this portion of the dental plate is differentiated after the usual manner in Dinichthys. At least eight fine serrations are to be counted along the oral border, quite uniformly spaced. The head shield is chiefly remarkable for its relative length as com- pared with the abdominal armor, and for its extreme narrowness anteriorly. The possibility is not excluded, however, that the sinus back of the orbits, although symmetrically developed on both sides in the present condition of the specimen, may be due in part to accidental preservation ; but if normal, a resemblance is to be noted to the configuration of the head shield in Titanichthys brevis, as represented in Claypole’s restoration of the latter form. The suture lines are less undulating than those of Coccosteus, the agreement being rather with Dinichthys in this respect. The rostral plate has become dislocated from attachment with the head shield, but is preserved intact on another portion of the slab (between the right antero- dorsolateral and mandible). Its outline is represented by dotted lines in the figure, as if it were replaced in natural position, but it is impossible to assign definite limits to the pineal and contiguous plates. The median occipital develops a small process in the median line posteriorly, as in some species of Dinichthys. The dorsomedian plate is decidedly Coccosteuslike, and but little emarginate in front, a character which it shares in common with D. halmodeus. The antero-dorsolaterals have strong articular condyles, and much extended anterior margins. The postero-dorsolaterals are slender, triangular plates, not unlike those of Dinichthys. A pair of imperfectly preserved bones, which may possibly be the claviculars, occur to one side of the last mentioned plates ; whatever their true nature, they have at least a form not unlike the clavicular of Titanichthys. The ventrome- dian is narrow and elongate, the postero-ventrolaterals very similar to those of Coccosteus. In order to provide data which shall serve for a comparison of parts *Am, Geol. 1896. v. 17, pl. ro. DEVONIC FISHES OF THE NEW YORK FORMATIONS 137 between this and other species, the following table of measurements is given: Length of head shield along median line - - - - 8.3.cm Maximum width of head shield 8.3 Length of dorsomedian - - - - - > 5-2 Maximum width of dorsomedian - - - 4-4 Maximum width of antero-dorsolateral - 4.0 Length of postero-dorsolateral | - 3-8 Total length of mandible - - - 4.6 Formation and locality. The unique specimen answering to the above description was obtained by Mr F. K. Mixer from the black Rhinestreet shale (Portage) at Sturgeon Point, on the south shore of Lake Erie, near Buffalo, N. Y. It is now preserved in the collection of the Buffalo Society of Natural Sciences, and the writer is indebted to Mr Mixer for the privilege of its description. Dinichthys tuberculatus Newberry 1888 Dinichthystuberculatus J. S. Mewberry, N. Y. Acad. Sci. Trans, 72179 1889 Dinichthystuberculatus /. S. Mewberry. U.S. Geol. Sur. Monogr, 16: 98, pl. 32, fig. 3 1893 Dinichthystuberculatus &. W. Claypole. Am. Geol. 12: 277 1897 Dinichthys tuberculatus C. &. Lastman. Mus. Comp. Zool. Bul. 1: 38 oe ee tuberculatus C. &. Hastman, N.Y. State Geol. 17th An. Rep’t, p. 318 An imperfectly definable species, known only by detached plates which are remarkable for their relatively gfeat thickness, and coarsely tuberculate style of ornamentation. The known portions of the abdominal armor indi- cate a species rather less than one half the size of D. intermedius. In the present state of our knowledge, there are no reasons other than differ- ence in geological horizon to prevent assigning to this species certain heavy and coarsely tuberculated Dinichthyid plates found in the Middle and Upper Devonic of Wisconsin and Iowa; neither is it possible, except for difference in geological age, to recognize a distinction between the plates 138 NEW YORK STATE MUSEUM known under this name and the so called D. precursor Newberry, from the Corniferous limestone of Ohio. The typical locality for D. tubercu- latus is in the Chemung conglomerate of Warren, Pa., but according to Newberry, the same form occurs also in the Upper Devonic of Belgium. Formation and locality. Chemung beds (Chautauquan); Warren, Pa. Also, according to Newberry, in the Psammites de Condroz, near Liége, Belgium. Either this or a very similar species is also represented in the Middle and Upper Devonic of Ohio, Wisconsin and Towa. Dinichthys curtus Newberry 1888 Dinichthys curtus /. S. Newberry. N.Y. Acad. Sci. Trans. 7:179 1889 Dinichthys curtus /. S. Mewberry, U.S. Geol. Sur. Monogr. 16: 156, pl. 48, fig. 3; pl. 53, fig. 1-4 1893 Dinichthys curtus &. II. Claypole. O. Geol. Sur. Rep’t, 7: 606 1893 Dinichthys curtus 4. 4. I right. O. Geol. Sur. Rep't, 7: 623 tg00 Dinichthys curtus C. R&R. Zastman. Jour. Geol. 8: 33 1905 Dinichthys curtus LZ. Hussakof. Am. Mus. Nat. Hist. Bul. 21: 409, pl. 15, fig. 1; pl. 16 1906 Dinichthys curtus LZ. Hussakof. Am. Mus. Nat. Hist. Mem. g; p. 112, text fig. 5, pl. 12 Newberry’s description of this species is as follows: Fishes of moderate or small size; head a nearly equilateral triangle, measuring about a foot on a side: cranium, maxillary and mandible similar in character to those of Dinichthys intermedius, but only half to two thirds as large, and the mandible bears two subordinate prominences back of the turned up toothlike extremity ; also the posterior end of the cutting edge is set with two or three unequal denticles in place of the series of even, lancetlike points in the same position on the mandible of D. intermedius. The anterior ventrolateral plate is scimiter-shaped, 8 inches long by 2% inches wide, being relatively narrower than the corre- sponding bone in any other species known. Occurring typically in the Cleveland shale of Ohio, this species is also reported by Newberry from the Chemung of Pennsylvania, although no precise indications as to locality are given. Investigation shows that detached plates of a species fully as large as D. curtus, possibly even larger, occur in the Chemung of Warren county, Pennsylvania, but the pres- DEVONIC FISHES OF THE NEW YORK FORMATIONS 139 ent writer is unacquainted with any evidence which will enable one to state positively that the species is identical with any of the Ohio forms. It is to be hoped that more characteristic remains from the eastern region may yet be brought to light. Quite recently the Ohio material has been subjected to a renewed examination by Mr L. Hussakof, of the American Museum at New York, who gives a restoration of the ventral armoring [1905, p. 412], and discusses the arrangement of jaw parts (1906, p. 112). Formation and locality. Cleveland shale (Upper Devonic); Ohio. Presumably also in the Chemung of Pennsylvania. Dinichthys sp. ind. Amongst the series of fish remains collected by Professor Charles S. Prosser from the Chemung of Delaware county, New York, and from the Catskill beds near Palenville, occur fragments of small Dinichthyids agreeing in size with the plates of D. tuberculatus, but too poorly preserved for satisfactory determination. Figures are given of two of these plates in an appendix to Professor Prosser’s paper on the “ Hamilton and Chemung Series of Central and Eastern New York,” published in the r7¢ Annual Report of the New York State Geologist for the year 1897. In addition, there is preserved in the State Museum at Albany one very perfect example of the postero-dorsolateral plate belonging to an unknown species of Dinichthys, from the Chemung of Franklin, in Delaware county. It has a total length of about 30 cm, is finely tuberculated, and shows impressions of sensory canals on the external surface [sce pl 6, fig. 1]. The same museum also possesses an imperfect head shield, preserved so as to show the visceral aspect, of a small Dinichthyid from the Oneonta sandstone of Oxford, N. Y., its general appearance being suggestive of D. pustu- losus, but having a width of only 9 cm across the posterior border. Other obscure fragments are known from the lower Genesee of Eighteen Mile creek, near Buffalo, and from Hamilton rocks at Haight’s Quarry, Cazenovia, N. Y. The Black slate of Kentucky, a horizon just above the Hamilton, also yields a considerable number of fragments, 140 NEW YORK STATE MUSEUM Genus TITANICHTHYS Newberry Plates of head and trunk resembling those of Dinichthys, but relatively thinner, and more laterally expanded. Pineal plate elliptical, broader than long, in contact with the centrals, and pierced by one or two foramina, the latter sometimes capped by a small bony operculum. Lower dental plates long and slender, without denticulations, grooved in the anterior portion of the oral margin as if for a horny sheath, and somewhat turned upwards at the symphysis. Outer (free) portion of clavicular developed as a stout arm, rounded or semicylindrical in cross-section. The remarkable fishes comprised by this genus represent the ultimate stage of specialization attained by Dinichthyids. Unable to maintain an existence except under peculiarly favorable conditions — their gigantic size, unwieldy organization and weak dentition presupposing an estuarine habitat and abundant food supply —they survived for a relatively short period, and within a limited area. Their remains are confined, so far as known, to the Upper Devonic of Ohio. Forerunners of the genus, however, make their appearance as early as the Ulsterian, and fragmentary plates very suggestive of Titanichthys occur in the Hamilton limestone of Milwaukee, Wis. The arrangement of cranial plates conforms closely to the pattern of typical Coccosteans. The bone substance, however, is much thinner, the sutures more intimately fused, and all of the plates are relatively broader. The long axis of the pineal, too, is transverse instead of longitudinal, as in Dinichthys and Coccosteus. The enormous width and flatness of the cranial and abdominal armor plainly indicate a depressed, raylike form of body, probably correlated with bottom-feeding and generally sluggish habits. Certainly the degenerate character of the dentition does not permit us to look upon these creatures as very formidable competitors of Dinichthys and the much more agile Cladoselache. Interesting features are displayed by the abdominal armor. The mode of articulation between the head shield and antero-dorsolateral plates is less complicated than in Dinichthys, the condyle and socket of the latter form being replaced by an elongated flange and groove, which probably admitted DEVONIC FISHES OF THE NEW YORK FORMATIONS I4!I of but slight movement between the parts. In the type species at least, the antero- and postero-dorsolateral plates are completely fused, and are over- lapped to a relatively greater extent by the dorsomedian than in Dinichthys. Particular attention is claimed by the last named plate, for the reason that its relations have not been clearly understood.. It would appear that our only knowledge of the dorsomedian in Titanichthys is confined to a single specimen, identified by Newberry as belonging to T. clarki, but wrongly referred by him to the “underside of the body or head.”* A figure of it is Fig. 27 Dorsomedian plate of Titanichthys clarki Newb., fromthe Cleveland shale of Ohio. Original in American Museum, x 1-11 Fig. 28 Clavicular of Titanichthys clarki Newb., from the Cleveland shale of Ohio. Original in American Museum. X I-10 given, with the posterior margin uppermost, in plate 3, figure 1 of New- berry’s monograph, and its outlines are reproduced in the accompany- ing text figure 27. The original is now preserved in the Americar Museum of Natural History in New York, after having remained for some time inaccessible in the collections of Columbia University. Newberry’s failure to recognize this plate as a dorsomedian is probably to be explained by the fact that its visceral aspect is embedded in the matrix, thus conceal- ing the characteristic longitudinal keel. That the latter was present, how- U.S. Geol. Sur. Monogr. 1889. ‘16: 135. I42 NEW YORK STATE MUSEUM ever, is shown by a convenient fracture, although there is no evidence that it terminated in a distinct process, as in Dinichthys, Mylostoma, and other genera. Measurements show that the extreme length and width of this plate are very nearly equal, the distance between the anterolateral extremi- ties amounting to 70 cm. In an earlier restoration of T. agassizi by Fig. 29 Titanichthys agassizi Newb. Restoration of head shield and dorsal armoring of trunk, x 1-13. 4DZ=antero- dorsolateral, fused with the postero-dorsolateral ; C=central ; DAZ—dorsomedian restored after the outline of 1. clarki; £O= external occipital ; 17=marginal; 47O—median occipital ; P=pineal; R=position of rostral; SO=suborbital the present writer,' the dorsomedian is shown as being partly overlapped by the adjacent plates on either side. This is now believed to be erroneous, the relations of overlap are as in Dinichthys, and the idea that a carinal process was present was suggested by a photograph of a plate purporting to belong to T. clarki, but which, upon examination of the original, tAm. Nat. 1898. 32: 763, fig. 4. DEVONIC FISHES OF THE NEW YORK FORMATIONS 143 turns out to have been a large dorsomedian of Dinichthys. The error to which we have called attention is corrected in the new restoration given herewith, We come now to speak of the clavicular, a plate which has been curiously misinterpreted both in this genus and in Dinichthys. The frag- mentary example figured by Newberry in plate 3, figure 3 of his Monograph on Palaeozor Fishes under the name of “coracoid (?),” gives no adequate conception of what the plate was like. We accordingly present an outline drawing and section [text fig. 28] of a nearly perfect clavicular of T. clarki, the original of which belongs to the American Museum of Natural History ; and this element is also shown in its natural position, overlapping the antero-dorsolaterals, in the restoration of T. agassizi given intext figure 29. It will be seen from these figures that the external or distal portion of the clavicular is very different from the corresponding portion in Dinichthys, being developed as a stout, slightly curved arm, semicylindrical in cross- section, and not dividing into two branches, nor provided with the peculiar “knuckle joint” described by Claypole in a large specimen of Dinichthys.' The total length of the figured example, measured in a straight line, is 65 cm, and it is evident that the rodlike projection must have extended almost, if not entirely free from the body armor on either side. What function it subserved is difficult to imagine, but as likely a conjecture as any is that it furnished a support for the branchial apparatus and attach- ment for the tissues connected with the lower jaw. It is impossible to suppose that plates of such thickness as the clavicular—in the distal process it amounts to upwards of 5 cm—could have been flattened out horizontally by mechanical agencies during fossilization ;' hence, as this bone continues the plane of the head shield and body armor, we are compelled to assign to Titanichthys a depressed form of body, having in the type species the truly enormous expanse from side to side of over 1.5 meters. Nothing is known of the ventral armor of Titanichthys, unless we are permitted to assign to this genus certain large, thin plates of lanceolate : Ohio Geol, Sur, Rep’t. 1893. 7: 610, pl. 38, fig. 1; pl. 39, fig. 1. ella, 144 NEW YORK STATE MUSEUM outline, which have been found as yet only in the detached condition, but evidently represent the median ventrals of some large member of the family. One such, having a length of 29 cm, and maximum width of 12.5 cm, has been figured by the present writer in the bulletin of Museum of Comparative Zoology [31 : 26, pl. 5, fig. 1]. It is abruptly truncated in front, and bears traces of overlap by contiguous plates; accordingly if this plate is rightly interpreted as belonging to Titanichthys, there is excellent reason to suppose that the ventral armor consisted of the same number of plates as in Dinichthys. Genus PROTITANICHTHYS 7007s Primitive Coccosteans of small size, displaying composite characters of later forms. Arrangement of cranial plates in general resembling that of Coccosteus, the centrals meeting in a sinuous longitudinal suture and not in contact with the pineal; the latter is subellipitical in outline, its long axis directed transversely, and pierced by a relatively large pineal foramen; rostral plate also very broad. External surface finely tuberculated ; lateral margin of head shield apparently not much widened posteriorly ; sensory canals distinct [dentition unknown]. Protitanichthys fossatus sp. nov. Plate 10, figure 2; text figure 30 The unique specimen which is here made the type of a new genus and species is of interest in two respects; first, on account of its geological antiquity, and in the second place, because it displays synthetic characters. In the majority of its features, a close approximation is to be observed to Coccosteus, and to the more primitive species of Dinichthys, as for example, D. halmodeus. First of all, there is to be noted its small size, corre- lated with finely tuberculate ornament of the external surface; secondly, the sinuous suture line between the pair of central plates; and thirdly, the exclusion of the pineal from contact with the latter pair. The form of the head shield, too, although this can only be determined approximately, appears to have been long and narrow, without much lateral expansion at the posterior border. One striking peculiarity, however, distinguishes it DEVONIC FISHES OF THE NEW YORK FORMATIONS 145 from all species of Coccosteus and Dinichthys, and points to ancestral rela- tionships with Titanichthys, in recognition of which we have given it the generic name. Indeed, no other course is open to us than to attach signifi- cance to a character which is common to the two genera under comparison (Titanichthys and Protitanichthys), but occurs not elsewhere amongst members of the same family. We refer to the large size and transverse elongation of the pineal plate, which, according to our view, indicate an initial modification along the line of descent culminating in Titanichthys. As yet our acquaintance with intermediate forms connecting the archetype and extreme limits of specialization is very inadequate. But when we take into consideration the long geological interval separating the two genera in point of time, it is scarcely risking too much to claim this small composite form as a possible forerunner of the Upper Devonic Titanichthys. The head shield, itself imperfectly preserved, is unaccompanied by any other bones of the skeleton. Although, as a rule, little dependence is to be placed upon theoretical association of parts, the question may properly be raised whether there are any other Coccostean remains occurring in the same horizon which correspond in general proportions and superficial orna- ment to the head shield under discussion. Attention rests immediately upon two forms: the lower dental plate described as Liognathus spatulatus, and the dorsomedian plate known as Coccosteus occi- dentalis [ante, p. 115]. The possibility is not remote that all three of these detached parts which have received separate names may actually belong to a single species ; in default of proof, however, we have no other procedure than to maintain each provisionally as an independent species, or even genus. At the same time it may not be unworthy of comment that the peculiar spiniform prolongation observed in Coccosteus occiden- talis reappears in an abbreviated form in the triangular termination of the corresponding plate in Titanichthys. Such a coincidence certainly sug- gests the idea of an inherited characteristic, and confirms us in the belief that prototypes of the most highly specialized Coccosteans are to be sought as early as the Ulsterian stage in this country. 146 NEW YORK STATE MUSEUM The actual condition of the type specimen may be seen from the photo- graph reproduced in plate ro, figure 2; in addition, an outline restoration is here offered, which it is hoped will be of service in comparing the arrange- ment of cranial plates and sensory canals, as far as these are discernible, with other Coccosteans. The specific title has reference to the conspicuous pineal foramen, placed slightly posterior to the middle of the plate. Two Fig. 30 Protitanichthys fossatus sp. nov. Ulsterian; Delaware, Ohio. Restoration of head shield, x 44. C=central; EO=external occipital ; A7=marginal ; MO=median occipital; P=pineal; ?O=preorbital; /*O=postorbital ; A=rostral. Sensory canals shown by double dotted lines. Original in Mus Comp. Zool. other prominent fossae, corresponding to those in Dinichthys, occur in line with the pineal on the underside of the preorbitals, and these appear to have been bounded by a low transverse ridge in front. Whether or not vomerine teeth were attached in this vicinity can not at present be deter- mined. Analogy, our only guide, would lead us to expect that they had been. Formation and locality. Delaware limestone (Ulsterian); Delaware, O. Holotype preserved in the Museum of Comparative Zoology at Cambridge, Mass. DEVONIC FISHES OF THE NEW YORK FORMATIONS 147 Genus eryprasris Newberry A provisional genus including Arthrodires of large or even gigantic size, known only by plates of the abdominal armor. Ornamentation con- sisting of fine stellate tubercles, closely arranged and often becoming con- fluent in irregular or vermiculating ridges. Overlapping margins of plates broad, beveled to a thin edge, and sometimes striated. Glyptaspis verrucosa Newberry 1889 Glyptaspis verrucosus /. S. Newberry. U.S. Geol. Sur. Monogr. 16: 158, pl. 13, fig. 1, 2 Under this name several large plates with strongly beveled edges, and with central portion displaying peculiar ornamentation, have been described by Newberry from the Cleveland shale of Ohio. One of the two plates figured by this author is identifiable as the anterior, the other as the pos- terior ventromedian, both represented in inverted position. The larger of these, as compared with the corresponding element amongst other forms, indicates a creature fully equaling Titanichthys in size. Of the head shield, and other portions of the armor aside from the ventromedian plates, nothing is at present known, nor has the species been recognized with certainty beyond the confines of the Cleveland shale of Ohio. Glyptaspis abbreviata sp. nov. Plate 13 A few fragmentary plates are preserved in different collections from the Black slate of Kentucky (Genesee) and Portage shale of western New York, which display the vermiculating type of ornamentation peculiar to Glyptaspis, but until recently there has been insufficient material for fram- ing a satisfactory diagnosis of the species represented. The relatively smaller size of the plates, their fineness of ornamentation, and difference in geological horizon lead one to suspect that another than the type species was present during the earlier part of the Upper Devonic, and this inference would seem to be borne out by a recent fortunate discovery, which we owe to Dr J. M. Clarke. 148 NEW YORK STATE MUSEUM In plate 13 is shown an exceptionally perfect abdominal plate identi- fiable as the right antero-ventrolateral of a form closely resembling the type specics of Glyptaspis, but distinguished by its more delicate ornamentation, and by a very marked difference in the proportions of length and breadth. As indicated by the trivial title, the plate here named G. abbreviata is very much foreshortened in an anteroposterior direction, more so than in any other genus of Arthrodires with the possible exception of Trachosteus. The general outline is subtriangular, the total length being but 15 cm, and width 16cm. Although actual specimens are wanting of the corresponding plate in G. verrucosa, we are able to form a tolerably accurate concept of its appearance and general proportions, by means of its impress upon the antero-ventromedian, taken together with the configuration of the postero-ventromedian, and imprints upon it of the postero-ventrolaterals. For all practical purposes therefore, our knowledge of the ventral plates in G., verrucosa is sufficient to permit comparisons to be made with the armoring of other forms; and in the case of the extremely perfect outlines of G. abbreviata, one can see at a glance that a very different set of proportions obtained in the two species. As regards superficial ornament, the forms are related to each other much in the same way as are Bothrio- lepis nitida and B. minor amongst Ostracophores. The bone substance of the plate discovered by Dr Clarke is almost entirely denuded, leaving, however, a very clear impression of the external surface. Only the central portion of the plate is tuberculated, the surface sloping away on all sides of this area so as to form a broad, smooth peripheral margin, on which no traces of overlap are to be observed. The details of the ornamentation are shown to rather better advantage in the smaller frag- ment from the Black slate of Lexington, Ky. The specimen here regarded as the type of a new species is from near the base of the Portage beds at Valois, N. Y., and is preserved in the New York State Museum. A some- what Jarger plate, possibly identical with this, is to be seen in the collection of the Buffalo Society of Natural Sciences, and was obtained by Mr. Mixer from the Portage of Sturgeon Point, on the shore of Lake Erie. DEVONIC FISHES OF THE NEW YORK FORMATIONS 149 Formation and locality. Genesee of Eighteén Mile creek, N. Y., and Lexington, Ky.; Portage of Seneca co. N. Y. Family MYLOSTOMATIDAE Head shield and abdominal armor essentially as in the preceding family, but with dentition adapted for crushing instead of cutting. Upper dentition consisting of two pairs of Ceratodontlike palato-pterygoid dental plates, with nondenticulate margins. Vomerine teeth present in at least one genus, Genus MYyLostoma Newberry Distinguishable from Dinichthys only by characters of the dentition. Oral surface of lower dental plates broad, more or less flattened, and bearing either a rounded boss or V-shaped eminence close to the inner margin, which plays into a corresponding depression of the upper pair. Vomerine teeth as yet unrecognized. Our knowledge of Mylostoma is confined at present to three species, all from the Cleveland shale of Ohio. Of these M. variabile New- berry, which is typical of the genus, has become comparatively well known within the last few years, thanks especially to the elaborate researches of Dr Bashford Dean. On the other hand, no additional material illustrating the characters of M. terrelli has come to light since the discovery of the unique type described by Newberry, now the property of the Museum of Comparative Zoology. One of the interesting points established by Dean’s study of the type species is the close agreement between it and Dinichthys in all essential respects save for the dentition; and as regards this latter feature, the same difference is to be noted as exists between Rhynchodus and Palaeomylus amongst Chimaeroids, or between Protopterus and Neoceratodus amongst modern Dipnoans. Parallel modifications of this nature, occurring as they do in diverse groups, are doubtless to be cor- related with similar food habits. Amongst Chimaeroids, for instance, certain genera are shown by their development of tritoral dental plates to have subsisted on hard-shelled prey, such as mollusks, echinoderms and the like; whereas others, as indicated by their sharp cutting blades, were 150 NEW YORK STATE MUSEUM adapted for subsistence on soft tissues, and were probably predaceous creatures. Mylostoma and Dinichthys furnish examples of corresponding adaptations amongst Arthrodires, and we have already seen that analogous changes are sustained amongst Dipnoans of the present day. The mandibles of Mylostoma would seem to have retained with great persistency typical Dipnoan conditions. Not only do the dental plates portray what may fairly be called a Ceratodontlike configuration, but they ten, Fig. 31 Upper dentition of Mylostoma variabile Newberry, based upon a single, nearly complete example from the Cleveland shale near Cleveland, O. Nat. size are more sharply demarcated from the supporting splenial than in other Arthrodires. And although marginal serrations have disappeared, the divided ridge which occurs close to the inner margin is perhaps to be regarded as a relic of one of the most persistent features of Dipnoan denti- tion. As for the upper dental plates, had they always been found in the detached condition, and were we ignorant of their association with typical Arthrodiran mandibles, they would be unhesitatingly identified with the Ctenodipterine order of Dipnoans. That these plates were supported in cartilage forming the roof of the month is perfectly evident from their rugose, slightly hollowed upper surface, and outwardly beveled edges; and the pos- terior contour of the hinder pair renders it extremely probable, at least, DEVONIC FISHES OF THE NEW YORK FORMATIONS I51 that the supporting palato-pterygoid cartilage was of the usual pattern found in all Dipnoans. This cartilage, when ossified, is commonly known as “‘upper dentigerous bone”; the fact that it is unossified in Arthrodires agrees with other evidence pointing to their lesser specialization as com- pared with Ctenodipterines. The restoration of the upper dentition of M. variabile, shown in the above figure, requires no additional explanation beyond that given in the introductory account of Arthrodires. It may profitably be compared with text figure 18, which is reproduced from a photograph of the actual dental plates. The retention throughout life of two pairs of palato-ptery- goid dental plates in this family, corresponding to an evanescent stage of Neoceratodus, is regarded as a primitive characteristic. Hence, in so far as the dentition is concerned, members of this family recall ancestral conditions more distinctly than either Dinichthys or Coccosteus, Genus DINOMYLosTomMa Eastman A genus transitional between Mylostoma and Dinichthys, as its name implies, and partaking of the characters of both. Mandibles with slightly prehensile symphysial beak, and broad, flattened, regularly concave func- tional margin, showing marks of contact with dental plates of the opposite jaw, the latter essentially like those of Mylostoma. Vomerine teeth subtrihedral, slightly prehensile. Dinomylostoma beecheri Eastman Plate r4, figures 5, 6; plate 15; text figure 32, 33 1906 Dinomylostoma beecheri C. R&. Zastman. Am. Jour. Sci. v. 21, ser. 4, p. 83, text fig. 2 (No description) 19066 Dinomylostoma beecheri C. R. Eastman. Mus. Comp. Zool. Bul. 50, p. 23, pl. x, fig. 4, 5; pl. 2, fig. 13, 14, 16, 17; pl. 4, 5 1906 Dinomylostoma beecheri JZ. Hussakof. Am. Mus. Nat. Hist. Mem. 9, P. 119, 123 The specific characters of this form are included in the foregoing generic diagnosis. It may be remarked, however, that its particularly dis- tinctive feature consists in the acute termination of both mandibles and 152 NEW YORK STATE MUSEUM vomerine teeth in front, together with the deeply concave oral margin of the lower dental plates. The functional surface of all the dental elements is narrower than in Mylostoma, thus making some approach to Dinichthys- like conditions, and a still further resemblance is to be observed in the form of the vomerine teeth. The latter, if found in the detached state, might readily be mistaken for the commonly so called “ premaxillary teeth” of the preceding family. The dorsomedian and other plates of the abdominal armor are indistinguishable from those of Dinichthys. The unique specimen upon which this species and genus are founded is extremely important for the light thrown upon the structure and relations of the group to which it belongs, besides acquainting us with an interesting connecting link between two well known genera of Arthrodires. Nearly the complete dentition is presented for study, besides portions of the head shield, suborbital, sclerotic ring, and several plates of the abdominal armor, including a nearly perfect dorsomedian. All of these parts were inclosed originally in a single block of shale; this has since been considerably broken, allowing a number of plates to be disengaged from the matrix. The history of the specimen is as follows: collected in the year 1868 from an outcrop of Portage shale upon the farm of John Pierce, near Mount Morris, Livingston co. N. Y., it was acquired for the Peabody Museum at Yale by the late Professor O. C. Marsh. Here it remained for many years stored away and apparently forgotten. At all events there is no record of its consultation by any one prior to its being called to the present writer’s attention by Professor Charles E. Beecher, shortly before the unfortunate loss to science of the latter. An opportunity for its description arising in connection with the present report, it was generously placed at the writer’s disposal by his friend Professor Schuchert, of Yale, at whose suggestion the specific title was inscribed to the memory of the lamented Beecher. Although this specimen has already been noticed in the papers above referred to, its importance warrants further detailed description. The dentition of this species presents features of absorbing interest. The mandibles bear a superficial resemblance to the lower dental plates of DEVONIC FISHES OF THE NEW YORK FORMATIONS 153 Palaeomylus, especially P. greenei, and they are constructed more nearly after the pattern of Dinichthys than of Mylostoma. This by no means allows us to suppose, however, that the Mylostomid type is a deriva- tive of Dinichthys, or of its immediate predecessor, Coccosteus, inasmuch as the upper dentition in the family we are considering is decidedly more primitive than that of Coccosteans. The present species illus- trates the manner in which the lower dental plates of Dinichthys and its allies may have become .modified from the usual form of these organs amongst Dipnoan fishes, by becoming turned upright in the jaws and Fis. 3 left vomerine tooth of Dinomylostoma acquiring a trenchant functional margin through gradual aioe ease compression. One can readily conceive the possibility Nat. size of the palatopterygoid plates undergoing corresponding modifications, which resulted in the development of shear teeth. In the mandibles of the species under discussion, the symphysial beak is rather obtuse, and but slightly prehensile, being scarcely elevated above the broad, flat, deeply excavated functional surface. The latter displays a single inconspicuous eminence or tubercle close to the external margin, situated about midway the length of the oral margin; and at a distance of about 8 cm behind the. anterior beak is a larger tubercle, rather elongate, and placed externally like the first. This posterior prominence fits accu- rately against the single large rounded boss of the opposing upper dental plate, thus determining the orientation of the latter with absolute certainty, and affording a trustworthy clue to the position of the corresponding element in M. variabile. The splenial is developed as a long slender shaft of bone, resembling that of Dinichthys, but relatively deeper. Both of these elements, right and left, are preserved in natural association with the articular cartilage. This has become more or less compressed through fossilization, but remains attached to the outer face of the splenial near its posterior extremity. Owing to the favorable preservation of the specimen, we are now enabled 154 NEW YORK STATE MUSEUM for the first time to identify positively the elements of the lower jaw in Arthrodires, and to ascertain the nature of its articulation with the head shield. That the shaft of bone which we have called the splenial is prop- erly determined as such follows necessarily from the rule that in all higher types of fishes and other vertebrates, the functional lower jaw is formed by membrane bones which become ossified around Meckel’s cartilage. In their greatest development there may be several of these bones on either side, amongst which the dentary is defined as that bone which is developed in front — usually it is dentigerous—and the splenial that which is developed further back oz the ‘nner side. The angulare, when present, extends for- ward from the angle of the lower jaw to meet the other two, and in addition a supra-angulare sometimes occurs behind the articulation of the lower jaw with the quadrate. Meckel’s cartilage itself never ossifies, but remnants of it may persist in grooves or otherwise, as has already been observed. The man- dibles of Dinomylostoma are of peculiar importance in that their constituent parts are found to correspond precisely with those of Dipnoan fishes. The vomerine teeth of the present species are readily identifiable as such, owing to their general similarity to the corresponding structures in Dinichthys. They are prehensile to about the same degree as the sym- physial beaks of the lower jaw, against which they closed. Their posterior face is smooth and slightly concave, as if they had been in direct apposition with the anterior pair of palato-pterygoid dental plates, but these latter are missing in the only known example. The hinder pair, however, is admir- ably preserved, and, its position being definitely fixed in the manner already indicated, it is not difficult to restore the general outline, at least, of the missing elements. Their external margin must have been parallel to that of the functional surface opposed to them in the lower jaw; and as they were probably in contact along the median line, the inner ma‘yin was rectilinear, asin Mylostoma variabile. An inspection of text igure 32 will facilitate an understanding of the manner in which upper and lower dental elements closed together, it being borne in mind that the tips of the vomerine and mandibular pair of plates were directly opposed. DEVONIC FISHES OF THE NEW YORK FORMATIONS 155 Measurements of the more characteristic elements may be given as follows: the vomerine teeth have a length along the inner longitudinal margin of 3 cm; their width across the posterior face is 3.5 cm, and across the external face itis 4cm. The posterior palato-pterygoid dental plates are of irregularly cruciform outline, with major and minor axes of 6.8 and 5 cm respectively. They are about 2 cm in thickness, except in the region Fig. 33 Diagram showing contour of oral margin of mandibular and posterior palato-pterygoid dental plates of Dinomy - lostoma beecheri, in front of which are shown the upturned vomerine pair, all drawn to the same scale, and the mem- bers of each pair placed in natural position with respect to each other. x¥% of the centrally placed tubercle, where the total thickness is 2.5 cm. The total length of the mandibles falls a trifle short of 20 cm, that of the func- tional margin about 8 cm, and the maximum width of the latter is 2.3 cm. The dorsomedian plate, closely similar in form to that of Dinichthys intermedius, and with well developed posterior process, is estimated to have had a total width of at least 18 cm; it is unfortunately not preserved for its entire length, The antero-ventrolaterals have approximately the 156 NEW YORK STATE MUSEUM same form, but are in the neighborhood of one fifth smaller than the corre- sponding plates in the unique example of Mylostoma variabile figured by Dean. It is necessary to bear in mind, however, in connection with this author’s restoration of the ventral armor, that the anterior and posterior pairs of lateral plates have been confused with one another. Such, at least, is the opinion of the present writer after comparison of the plates marked by him “4VZ” and “PVZL”* with the ventrolaterals of Dinichthys and noting specially their centers of ossification, growth-lines, and nutritive canals, Formation and locality. Portage (Cashaqua) shale; Mount Morris, N. Y. There is also indistinct evidence either of this or of some species of Mylostoma in the black Naples shale (Portage) at Sturgeon Point, on the south shore of Lake Erie. GENERA OF DOUBTFUL FAMILY POSITION Fragmentary remains of Arthrodires, differing from other described species as regards superficial ornament, and some of them indicating fishes of considerable size, have been described from various Devonic localities of the New York-Pennsylvania region. Most of the species are known only by detached plates, none in association with the dentition, hence their systematic position is doubtful. Coarsely tuberculated plates, suggestive of Newberry’s type of Aspidichthys clavatus, are reported from the New York Portage.* As indicated by the generic name, the ornament of Holonema consists of threadlike, radiating ridges. Two species are known, the type (H. rugosum Claypole) occurring in the Chemung of New York and Pennsylvania, and the other (H. horridum Cope) thus far known only from rocks of the same age in Bradford county, Pennsyl- vania. One specimen of H. rugosum is interesting in that it displays. the entire ventral armor. Phyllolepis is distinguished from Holonema by having the superficial rugae arranged in concentric, instead of radiating Dean, B. Palaeontological Notes: On the Characters of Mylostoma Newberry. N. Y. Acad. Sci. Mem. trgo1. 2: 108, pl. 7. 27 Williams, H.S. U.S. Geol. Sur. Bul. 41. 1887. p. 43. DEVONIC FISHES OF THE NEW YORK FORMATIONS 157 lines. The only American species known is P. delicatula Newberry, from the Chemung of Bradford county, Pennsylvania. Accompanying the latter, but of still more problematical nature, are plates with arrowheadlike ornamentation, described by Newberry under the name of Spheno- phorus lilleyi.: Certain elliptical plates having a closely similar style of ornamentation are also known from the Hamilton of Milwaukee. From this locality also are obtained a number of peculiar plates, possibly repre- senting the dorsomedian of unknown forms, two such being shown in plate 2, figures 3 and 4. Aspidichthys, Phyllolepis, and possibly also Holonema, are represented in the European Devonic, and one species from the Devonic of Manitoba is doubtfully referred to Aspidichthys (A. ? notabilis) by Whiteaves. Order CTENODIPTERINI Body fusiform, without dermal armor. Skeleton and chondrocranium partially ossified, skull autostylic, premaxillae and maxillae absent. Cranial roof bones small, and, like the squamation, with or without ganoin invest- ment. Nostrils inferiorly situated; jugular plates present or not. Tail heterocercal or apparently diphycercal (gephyrocercal). Anal fin always distinct, the remaining median fins either discontinuous or becoming coalesced. Paired fins acutely lobate. Secondary pectoral arch consisting of ossified supraclavicular and clavicle; pelvic arch present. Dentition consisting of large tritoral dental plates supported by the palato-pterygoid and splenial bones; a marginal series of teeth above and below also some- times present, but never any vomerine teeth. The structure of this singularly interesting order of Paleozoic fishes has been investigated in minute detail by Pander, Traquair and others, and particularly within the last dozen years or so its.relations to modern lung fishes have engaged profound attention amongst students of ichthy- ology and paleontology. Recent discussion has centered around two rival theories, as we have already seen. According to the first, which has steadily gained in ascendancy, the typical Dipnoans of the Devonic are t The generic name being preoccupied, Oestophorus has been proposed as a substitute by S. A. Miller. Plates of this description occur also in the Hamilton. 158 NEW YORK STATE MUSEUM supposed to have attained greater specialization than any of their known descendants in later times, those existing at the present day showing decidedly more primitive characters as compared with Ctenodipterines. Modern forms would therefore be looked upon as survivals of the more generalized primeval Dipnoan stock, rather than as the direct descendants of Dipterus and its allies. The other and newer interpretation is that of Dollo, and directly the reverse of the first. Evidence drawn from other than cranial characters is held to confirm the belief that Dipterus is the most archaic of all Dipnoans, and that modern lung fishes have been derived through successive stages of specialization. As starting point of this theory, Dollo accepts the con- clusion previously reached by Balfour and Parker that the apparently diphycercal tail of recent forms is secondary, due to abortion of the termination of the vertebral axis, and coalescence of the median fins. The less ossified condition of the skull in modern forms is likewise explained as the result of secondary reversion, through degeneration, to an apparently primitive condition. The chief objection to this view is that we are unac- quainted with any parallel example amongst vertebrates which justifies belief in the possibility of such degeneration as is here assumed. Yet another view of Dipnoan relationships, or perhaps rather to be considered as a modification of the first, is that already outlined in the present memoir. The Ceratodont type is regarded as decidedly more primitive in structure than that of Dipterus and its allies; and either it or its direct prototype is assumed to have been in existence at least as early as the lowermost Devonic, to have given rise to the highly specialized orders of Ctenodipterines and Arthrodires, and to have persisted practically unchanged ever since. Neoceratodus therefore falls within the same cate- gory as scorpions, king-crabs, Lingula, Cestracion and other archaic sur- vivals which have manifested extraordinary persistence and conservatism throughout nearly the whole sequence of geological formations. This view may be said to rest almost entirely upon the evidence of comparative anatomy, and has received as yet no confirmation through the discovery of DEVONIC FISHES OF THE NEW YORK FORMATIONS 159 actual remains which fulfil the requirements of a common ancestor to the three recognized orders of lung fishes, — Arthrodires, Ctenodipterines and Sirenoids. It may be that confirmation of this nature will never be forth- coming, owing to the imperfection of the paleontological record. Never- theless, guided by the principles of structural resemblances, and not by any exercise of the art of divination, one may project the divergent lines of descent backward until they meet in a common point; and at this point is to be sought the ancestry of the three orders with which we are now acquainted, Family DIPTHRIDAE Cranial roof bones numerous; no secondary upper jaw, and no mar- ginal series of teeth above or below; jugular plates present. Upper and lower dental plates triangular, with outwardiy radiating ridges, usually tuberculated or strongly crenulated, rarely smooth or nearly so; no vom- erine teeth. Caudal fin heterocercal; two remote dersal fins opposed to the pelvic and anai fins, separated from the caudal. Genus piprerus Sedgwick & Murchison Our knowledge of the complete form of this genus, shown in text figure 34, is dependent entirely upon the well preserved skeletons occurring in the Scottish Lower Old Red sandstone. In this country Dipterine remains are RS ZB nee LD ee eee ie MY) ee £2 ES Ts Fig. 34 Dipterus valenciennesi Sedgw. & Murchison. Lower Old Red sandstone; Scotland. Lateral aspect, restored. x34 (From Dean) confined exclusively to detached hard parts, such as dental plates, scales, and calcified labial cartilage. Under these circumstances it is sometimes difficult to determine whether teeth of a certain form should be retained in 160 NEW YORK STATE MUSEUM this genus, or referred to Sagenodus, Ctenodus, and others having a similar form of dentition. Recent discoveries having brought to light large num- bers of Dipterine teeth from the western Devonic, the leading species which are here figured and described offer valuable points of comparison with those from the eastern States. Dipterus uddeni Eastman Plate 4, figures 3, 4 1899 Dipterus uddeni /. A. Udden. Jour. Geol. 7: 494 (name only) 1900 Dipterus uddeni C. &. Eastman. Jour. Geol. 8: 37, text fig. 5 This species, the earliest to make its appearance in the Devonic rocks of this country, is founded upon a unique lower dental plate from the base of the Cedar Valley limestone (Erian) near New Buffalo, Ia. The specific title is bestowed in honor of its discoverer, Professor J. A. Udden, formerly of the Iowa State Geological Survey. It has a total length of 36 mm, is moderately convex, and remarkable for the paucity of its denticulated ridges. These are but four in number, and radiate outwards from the pos- terior angle, which is worn smooth by use. The anterior row of denticles, and inner moiety of the remaining rows are also considerably worn; but in the outer moiety of these rows the denticles are acutely conical, of large size, and well separated. There is a gradual increase in size of all the denticles from within the inner margin outwards. The coronal surface is finely punctate. Formation and locality. Base of the Cedar Valley limestone; New Buffalo, Ia. Dipterus calvini Eastman Plate 4, figure 1900 Dipterus calvini C. &. Eastman. Jour. Geol. 8: 38, text fig. 7 Lower dental plate elliptical in outline, and moderately convex in an anteroposterior direction. Eight tuberculated ridges extend from the outer margin to about the middle of the plate, the two anterior ones larger than the rest and elevated into a slight fold. Coronal surface considerably worn in the type specimen, and external margin partially broken. Tubercles DEVONIC FISHES OF THE NEW YORK FORMATIONS 161 conical and well separated, except those of the two anterior ridges, which are coalesced and worn on their summits. Total length of plate 3 cm. This species, like the preceding, is founded upon a unique dental plate from the Cedar Valley limestone of Iowa. It comes, however, from a higher level, locally known as the “ Euomphalus bed,” which lies about eight feet below the summit of the Middle Devonic in Muscatine county. Formation and locality. Cedar Valley limestone; Fairport, la. Dipterus costatus Eastman Plate 4, figure 9 1900 Dipterus costatus C. &. Hastman. Jour. Geol. 8: 39, text fig. 4° Dental plates agreeing in size and general outline with D. calvini, but with fewer and more widely separated coronal ridges, which disappear before reaching the middle of the plate. The distinguishing feature of this species consists in the elevated acute ridge extending along the entire length of the inner border, and separated from the remaining tuberculated ridges by a broad longitudinal furrow. This ridge appears to be made up of three coalesced costae, of which the third counting from the inner margin is the largest. The two innermost costae are so faint as to be almost imperceptible against the steep face of the main ridge. The latter shows no evidence of having been tuberculated, although faint and partially coalesced tubercles occur on the five remaining costae. Several examples of this species, including the type, are preserved in the Museum of Com- parative Zoology at Cambridge, Mass. Formation and locality. State Quarry beds (Upper Devonic); near North Liberty, Johnson co., Ia, Dipterus mordax Eastman Plate 4, figures 5, 6 1900 Dipterus mordax C. R. Eastman. Jour. Geol. 8: 39, text fig. 6, 8 Lower dental plates attaining a length of over 3 cm, coronal surface gently convex, with six rows of very large, discrete, conical or rounded tubercles which extend from the outer margin for a variable distance toward the posterior angle; the two posterior rows often rudimentary. Some of 162 NEW YORK STATE MUSEUM the tubercles, when worn by use, become elongated in the direction of the rows to which they belong, others in an oblique direction. The coarseness of tuberculation, in proportion to the size of the plates, is greater than in any other species yet described. Formation and locality. State Quarry beds (Upper Devonic); near North Liberty, Johnson co., Ia. Dipterus fleischeri (Newberry) Plate 7, figure 2 1897 Ctenodus fleischeri /. S. Mewberry. N.Y. Acad. Sci. Trans. 16: 302, pl. 24; figs 25 Upper dental plates thin, slightly concave, triangular in outline ; coronal surface transversed by six rows of rounded obtuse tubercles increasing in size from the posterior angle outwards. At least the anterior or innermost rows of tubercles are continuous almost to the posterointernal angle. Besides the type, only one other example of this species has yet been brought to light. This is a large upper dental plate, preserved in counter- part, now the property of the New York State Museum. It is from the Catskill of Delaware county, N. Y., and being more complete than New- berry’s original, we present a figure of it herewith. Its relations are evidently with D. uddeni and D. mordax, as shown by the small number of coarsely tuberculated costae; but in size it is much larger, the length of the inner margin being fully 5 cm. Formation and locality. Catskill sandstone (Chautauquan); Tioga county, Pennsylvania, and Franklin, Delaware co. N. Y. Dipterus sherwoodi Newberry 1875 Dipterus sherwoodi /. 8. Newberry. O. Geol. Sur. Rep’t. Palaeont. v.2, pt 2, p. 61, pl. 58, fig. 17 1889 Dipterus (Ctenodus) sherwoodi /. S. Mewberry. U.S. Geol. Sur. Monogr, 16: 118, pl. 27, fig. 32 Of this species no other examples are known but the type specimen, which is imperfect. In the text of Newberry’s monograph, this is deter- mined as “apparently one of the upper palate teeth of a species of Dip- DEVONIC FISHES OF THE NEW YORK FORMATIONS 163 terus,” but in the explanation of figures it is identified as a mandibular dental plate. It agrees with the preceding species in having a coarse tuber- culation and very few ridges, some of them indistinct. The tubercles them- selves are stated to be ‘somewhat compressed laterally, rounded, smooth, and blunt at the summit.” Formation and locality. Catskill sandstone (Chautauquan); Tioga county, Pennsylvania. CHEMUNG SPECIES OF DIPTERUS Precise determination of Dipterus teeth from the Chemung proper of New York and Pennsylvania is a matter of some difficulty, owing to imper- fection of type material upon which the various “species” are founded, and an insufficient series of specimens for illustrating the range of variation common to both sets of dental plates, upper and lower. In the case of some species, the original descriptions undoubtedly require emendation, but it would be hazardous to attempt this without having recourse to a larger and better suite of material than is at present available amongst different museums. This much, however, one feels warranted in concluding: that very probably not more than two well marked types of Dipterine teeth occur in the Chemung proper of the eastern region. These are D. fla- belliformis and D. nelsoni Newberry, the latter including New- berry’s so called D. levis (founded on worn specimens), and possibly also D. quadratus and minutus. Of Dipterus nelsoni both pala- tine and mandibular dental plates are known, a good example of the latter being represented in plate 4, figures 13, 14. For the original description of these forms, which we are willing to leave as they are for the present, one may consult Newberry’s monograph, pages 87-91. Whether the so called Dipterus ithacensis, considered by Williams‘ as ‘showing some relationship perhaps to Pterichthys,” properly belongs in this category can not be determined from the meager account given of it, unaccompanied by illustrations. ? Williams, H. S. Notes on some Fish-remains from the Upper Devonic of New York State (Abstract). Am. Ass’n Ady. Sci. joth meeting. Proc. 1882. p. 192. 164 NEW YORK STATE MUSEUM On the other hand, Dipterine remains are present in astonishing abun- dance and considerable variety in the western Neodevonic, specially the State Quarry beds of Johnson county, Iowa. A number of representative forms, some of them belonging to still undescribed species, are shown in plate 2, figure 1, and plate 4, figures 2, 7, 8,12, 15 and 16. It is to be regretted that no portions of the skeleton, other than the dentition, have yet been brought to light from this region. Subclass TELEOSTOMI The great group of fishes commonly known under the designations of Ganoids and Teleosts and first recognized by Owen as a single subclass, Teleostomi, makes its appearance in the Lower Devonic, but does not really become significant until the Carbonic. The Crossopterygian order, which predominates during the Devonic, is somewhat abundantly repre- sented toward the close of that system in this country; but except in Canada, all the remains are extremely fragmentary, consisting of detached scales, teeth, plates, and in two or three instances of imperfectly preserved skeletons. The most important American genera are Holoptychius, Saurip- terus, Onychodus and Eusthenopteron, but of these the last named alone has been found in a state bordering upon completeness. Nevertheless, it is possible to frame a tolerably accurate conception of the remaining genera through comparison of their characteristic parts with the admirably pre- served skeletons of their foreign representatives, especially those from the Scottish Old Red sandstone. In this way the detached head plates and bones of the shoulder girdle belonging to Onychodus, for instance, acquire much greater significance than would otherwise be possible. The osteology of various typical members of the Crossopterygian order is now quite satisfac- torily known, as the result especially of the researches of Pander,* Huxley” "Pander, C. H. Ueber die Saurodipterinen, Dendrodonten, Glyptolepiden, und Cheirolepiden des devonischen Systems. St Petersburg 1860. 2 Huxley, ‘T. H. Illustrations of the Structure of the Crossopterygian Ganoids. Geol. Sur. Mem. United Kingdom 1866. Reprinted also in the Scient. Mem. of T. H, Huxley, Suppl. volume, 1903. DEVONIC FISHES OF THE NEW YORK FORMATIONS 165 and Traquair. To the last named author,t also, we owe our principal knowledge of the structure of Paleoniscid fishes, these being the only Devonic representatives of the next higher order of Teleostomes, or Actinopterygii. Order 1 CROSSOPTERYGII Pectoral fins lobate, with a large basal portion covered with scales, more or less fringed with dermal rays; the ventrals usually much like the pectorals, always abdominal in position; caudal fin diphycercal or hetero- cercal. Skeleton more or less ossified, body covered with rhombic or circular ganoid scales.