AMERICAN MUSEUM NOVITATES Published by Number 1316 Tim AMERICAN MUSEUM of NATURAL HISTORY April 3, 1946 New York City

AMERICAN MUSEUM NOVITATES Published by Number 1316 Tim AMERICAN MUSEUM OF NATURAL HISTORY April 3, 1946 New York City

THE ANTERO-SUPRAGNATHAL OF GORGONICHTHYS

BY DAVID H. DUNKLE' AND PETER A. BUNGART'

INTRODUCTION

Newberry proposed the generic name identification of the upper mouth parts of Titanichthys in 1885 for the remains of a Titanichthys was announced by Dr. Louis spectacularly large arthrodiran fish from Hussakof (1930) before the twenty-first the upper Devonian Ohio Shales formation. annual meeting of the Paleontological During the ensuing 60 years, bones refer- Society at Washington, D. C. Un- able to the genus have proved to be as fortunately neither figures nor formal commonly encountered in collecting description of the specimen have subse- throughout the northern Ohio field as are quently appeared. A meager abstract, as those of the various, more widely and more the only published account of this report, popularly recognized species of Dinichthys has remained a source of intriguing inter- which occur contemporaneously in the est to all students of the Arthrodira. same formation. Titanichthys, in conse- Recently, Mrs. John Bull, working quence, has become well established as a among the collections of fossil fishes in the distinctive member of the fish fauna of the American Museum of Natural History Ohio Shales, and although no thorough under the direction of Dr. Edwin H. Col- revisionary work has been attempted for bert, came upon two associated bones many years, six species are tentatively which bear the identifying labels, re- assigned to the genus at the present time. spectively, of a "left antero-supra-gnathal" Despite this, the details of titanichthyid and indeterminate "cranial bone" of structure are still very imperfectly under- Titanichthys. These were submitted to stood. Peculiarly modified in a bottom- and gratefully received by the present feeding adaptation (Dunkle and Bungart, writers. Careful comparisons with the 1942), many of their skeletal elements are, abstract indicate the bones to be those on in addition, so excessively thin that they which Hussakof based his original report. cannot be removed from the matrix. An They (A.M.N.H. No. 7908) were collected inability to observe either the presence or by Peter A. Bungart in 1910 from exposures absence of the usually characteristic sensory of the Cleveland Shale member of the canals and the number and nature of over- Ohio Shales formation on the shore of Lake lap areas by adjacent bones often makes Erie, 1/2 mile east of the Avon Plant of the impossible the determination, as to particu- Cleveland Electric Illuminating Company lar skeletal part, of isolated specimens. in Avon Township, Lorain County, Ohio. More important, these conditions of preser- The specimen is here described and figured vation and unfamiliar modification have led for the first time because the characters of to a complete failure to recognize for the supragnathal element preclude its Titanichthys any of the supragnathal ele- reference to the genus Titanichthys, as do ments and some of the dorsal and ventral also those of the "cranial plate" which is body armoring plates. revealed as a fragmentary anterolateral In connection with one of these latter unsolved structural issues, the first definitive plate. These specimens are here referred to Gorgonichthys. ' The Cleveland Museum of Natural History, Cleveland, Onio. These writers sincerely appreciate the 2!i AMERICAN MUMSEUM NOVITATES [No. 1316 pQurtesy of the officials of the American in permitting this study and for the privi- M{useum of Natural History, New York, lege of publication in this series.

DESCRIPTION The present specimen consists of two section, and is also convex, although to bones: one, a left antero-supragnathal; much less degree, in any parasagittal plane. and the other, as here interpreted, a frag- Its internal surface is concave in both of mentary right anterolateral plate. The these respective sections. outlines and relationships between the The lateral lamina of the antero-supra- separate parts of these two bones are gnathal comprises the main body of the illustrated in figures 1 and 2, respectively. element. With regard to general shape, it A

Fig. 1. Gorgonichthys clarki Claypole (A.M.N.H. No. 7908). Left antero-supragnathal in (A) lateral aspect, (B) mesial aspect, and (C) frontal sections. Reproduction approximately X 1/e. a-a', Plane of frontal section (C) indicated on B; al, mesially directed anterior lamina of the antero- supragnathal; avc, anteroventral cusp; b-b', plane of frontal section indicated on B; f. ig., incised area worn by the anterodorsal cusp of the inferognathal; f. sn., fossa for the articulation with the sub- nasal element; in, internal surface of the antero-supragnathal; 11, longitudinally directed lateral lamina of the antero-supragnathal; n, notch on posteroventral margin of the element worn by the oral margin of the opposing inferognathal plate; p. ev., external vertical ridge marking the thickening of the anteroventral cusp; and p. pd., backwardly directed posterodorsal extremity of the antero- supragnathal.

The antero-supragnathal possesses an is roughly triangular in either lateral or over-all length of 207 mm. The maximum lingual views with the slightly concave depth of the element, as preserved, is 194 posteroventral margin the longer of the nmn. Its ventral cusp, however, is incom- three sides, and the posterodorsal angle plete, and the total restored height should truncated. This plate-like expanse of approximate 230 mm. Typically arthro- bone is so situated that its broad internal diran in development, the plate may be and external surfaces are vertically in- said, for purposes of description, to be com- clined. The thin, although relatively posed of anterior and lateral laminae which deep, posterodorsal extremity shows little are united anterolaterally in an externally mesial curvature, but is extended virtually rounded angle of less than 90 degrees. straight backward. Anteroventrally, a The anterior lamina, quadrangular in stout cusp is developed whose thickening is outline from forward view, exhibits a depth marked by a vertical ridge, externally. of more than three-quarters the total re- Worn internally by the ventrally opposing stored height of the bone. The width of inferognathal, the cusp is triangular in this mesiad extension is less than one- frontal section. This cusp development, fourth the maximum depth. The external in so large a plate, is restricted to-a surface of the part is acutely convex in frontal prisingly low ventral area. Immediately ANTERO-SUPRAGNATHAL OF GORGONICHTHYS 3 above the external vertical ridge of the row, and deeply incised area forward from thickened cusp a depression is displayed the ventral cusp of the antero-supragnathal whose margins diverge dorsally across both resulting from wear by the anterodorsal the anterior and lateral laminae of the bone. cusp of the inferognathal. Finally, the Similar concavities occur on the antero- posteroventral margin of the plate is supragnathals, where known, of all the di- angularly notched just behind the antero- nichthyid arthrodires. They mark the areas ventral cusp in a manner indicative of an A

Fig. 2. Gorgonichthys clarki Claypole (A.M.N.H. No. 7908). Fragmentary right anterolateral plate in (A) lateral aspect, (B) mesial aspect, and (C) transverse section. Reproduction approximately X 1/6. a-a', Plane of transverse section (C) indicated on A; b, restored posterodorsal expansion of the anterolateral plate; ex, external face of the element; f. il., rugose surface on the external surface of the mesial arm for the articulation of the interolateral plate; pe, lateral anteroventrally projecting arm; pm, mesial anteroventrally projecting arm; and x, unexplained rugosities on the inner face of the mesial arm. for articulation with the paired sub-nasal externally overlapping opposition upon a elements which will be fully described and decidedly obtuse inferognathal oral margin. determined as anterior ossifications of the As regards the second of the two bones primary palatoquadrate complex in a forth- comprising the present specimen, the de- coming paper by the present writers. termination of fragmentary arthrodiran Lingually, the surface of the lateral lamina bones as to particular skeletal part is often is featureless except for a very high, nar- admittedly hazardous. In this case, how- 4 AMERICAN MUSEUM NOVITATES [No. 1316 ever, the recognition of definite structures this particular specimen, and support our is possible, and this redetermination of the convictions as to the identity of the frag- erstwhile "cranial plate" as an antero- ment. It may be noted that the lateral lateral element is considered of unquestion- arm vestige, depicted in figure 2, is very able validity. weakly developed and that the main body The preserved crescentic portion repre- of the element, paralleling the concave sents the anteroventrally projecting limbs anterodorsal margin, is only moderately of an anterolateral plate of the type known thickened. At present the rugosities dis- to be possessed by Gorgonichthys, Heintz- played on the inner face of the mesial arm ichthys, and Titanichthys (Dunkle and are not explained, although they are un- Bungart, 1940; Heintz, 1931). In these doubtedly concerned with the cartilaginous genera the external arm of the anterolateral primary shoulder girdle. Worthy of men- elements, well developed and bearing spinal tion, also, is the general delicate structure plates in Coccosteus and the members of the of this anterolateral plate compared with Dinichthys terrelli group, is reduced to a the truly massive and powerful dentition laterally upfolded flange, in varying de- which the associated antero-supragnathal grees of completeness, on the ventral *denotes. This apparently reflects a tend- margin of the mesial arm. In addition, ency toward the further reduction of the these genera invariably exhibit the rugose dorsal body armor following the phylo- overlap area on the lateral face of the mesial genetic loss, in turn, of the connection be- arm which marks the articulatory surface tween dorsal and ventral armor, the spinal for the interolateral plate. Both of these elements, and, finally, the lateral antero- structural evidences are to be observed on ventral arm of the anterolateral plates.

REMARKS As may be gathered from the above de- deep notch, and an obtuse cutting edge. scription, the characters of the present two These structures suit the conditions known bones are identical with those of the cor- for Gorgonichthys. In contrast, we have responding elements of the genus Gorgon- failed ever to observe conclusive evidences ichthys Claypole (1892), as recently rede- of functionally worn inferognathals among fined (Dunkle and Bungart, 1940). Briefly the titanichthyids. Further, a belief discussed below, the most pertinent of these (Dunkle and Bungart, 1942) has been ex- characters would be clearly distinguishable pressed that the inferognathal plates of from Titanichthys even without prior Titanichthys have always been incorrectly knowledge of Gorgonichthys, and show that interpreted-upside down and inter- these bones cannot possibly be referred to changed from side to side. By the pro- the former genus. posed re-orientation, the anterior extremity The antero-supragnathal plate exhibits of the titanichthyid inferognathal is di- three areas of extensive wear by the oppos- rected ventromesially and is, in conse- ing, inferognathal element. These are: quence, non-cuspidate orally. the high, short, and deeply incised area on As explained above, the structure of the the internal f4ce of the bone, anteriorly; fragmentary anterolateral plate is similar the worn inner surface of the anteroventral in basic pattern to that of Gorgonichthys, cusp; and the deep, square notch behind Heintzichthys, and Titanichthys. The over- the cusp on the posteroventral margin of all dimensions of the bone, vestigial condi- the plate. Assuming that functionally tion of the lateral anteroventrally project- worn supragnathal bones must necessarily ing arm, and the moderate development reflect functionally abraded inferognathals, along the longitudinal axis of the mesial the above evidences of wear would seem to anteroventral arm are details at variance indicate the oral margin of an opposing in- with the anterolateral elements possessed ferognathal which bore below this antero- by the last two genera, and definitely de- supragnathal an acute anterior "tooth," a note an identity with Gorgonichthys. We ANTERO-SUPRAGNATHAL OF GORGONICHTHYS 5 are convinced of the correctness of the pres- edge, and no posterior "denticles" were ent determination for this specimen. considered aberrant Cleveland Shale deriva- While regrettable, it seems necessary to tives of the Huron Shale herzeri type. conclude that Hussakof's original assign- Hussakof (1906) suggested the possibility ment of this material was erroneous and to that the blunt and acute cutting-edged state that the supragnathal bones of Titan- inferognathal types represented two dis- ichthys remain unknown. tinct developmental lines throughout the The initial proposal (Dunkle and Bun- Ohio Shales formation, but discarded the gart, 1940) for the reestablishment of the theme in favor of the above pattern. An Elaypole term Gorgonichthys was based on arbitrary practice thus obtained, by which composite but fairly complete materials in the D. terrelli-like forms were restricted to the Cleveland Museum of Natural His- the Cleveland Shale and any dinichthyid tory, but pertained as well to all the speci- specimen found in the Huron Shale was mens correctly referred to the clarki species automatically referred to D. herzeri. which had been assigned to the genus This usage is no longer tenable. A gradual Dinichthys by Eastman (1900) and Hussa- accumulation of more complete specimens kof (1905). As stated without elabora- has resulted through years of collecting. tion in that preliminary study, a number of These, when studied in the light of the ex- important questions concerned with the cellent advances made by our European interpretation of structure and with the colleagues in the fundamental interpreta- phyletic relationships of the form were tion of arthrodiran structure, have shown left unanswered. Continued study on the that the inferognathal elements merely arthrodiran fauna of the Ohio Shales has express conditions of basic modification not appreciably forwarded the solution of whose explanations must be found in any of these questions. Some clarification other, associated skeletal parts. of the problems has resulted, however, and In the above regard and based solely on this moment seems opportune for the state- the characteristics of the inferognathal ment of certain of these unsolved points plates, the more common dinichthyid regarding the status of Gorgonichthys. arthrodires from the Ohio Shales appear In brief survey, the few attempts at re- separable into two groups. The first, construction of the phylogeny of the Ohio which may be called the Dinichthys herzeri Shales Arthrodira have been based pri- group, exhibits obtuse cutting-edged marily on the characteristics of the gnathal mandibular elements and includes, in addi- elements, and of these the inferognathal tion to D. herzeri, Gorgonichthys clarki, plates have been used most widely. From Heintzichthys gouldii, and Holdeniys,holdeni. the tentative groupings of forms that have The second, here referred to as the,`Jinich- been discussed in the literature, one gains thys terrelli group, possesses acuite cutting- the impression that the principal develop- edged lower mouth parts, and includes, ment among the typical dinichthyid besides D. terrelli, the speciesiintermedius arthrodires proceeded from older forms in and curtus. the lower Huron Shale member of the With the exception of TH enius in which formation with obtuse cutting-edged, the upper mouth parts remain unknown, posteriorly denticulate inferognathals of th-e herzeri group all possess antero-supra- the Dinichthys herzeri type to culminating gnathal elements whose posterodorsal ex- forms in the upper Cleveland Shale member tremities are thin, high, and directed with acute cutting-edged, posteriorly non- straight backward. In contradistinction, denticulate inferognathals of the D. ter- the posterodorsal extremities of the antero- relli type. In this arrangement, the supragnathals of the Dinichthys terrelli inferognathals of the clarki species with group are mesially recurved and greatly excessively high anterior cusp, short, blunt expanded. The significance of this feature cutting edge, and massive, posterior has previously been touched on (Dunkle "denticles" and those of Heintzichthys and Bungart, 1940) and is here illustrated with low anterior cusps, long, blunt cutting in figure 3. Reminiscent of the relations 6, AMERICAN MUSEUM NOVITAPES [No. 1316 in acanthodians between the five small margin excavated by a pair of deep con- anterior ossifications with the anterior basal cavities. From a correct alinement of all plate (Watson, 1937), the five supragnathal associated elements, the expanded postero- elements of certain dinichthyid arthrodires dorsal extremities of the terrelli type of (single mediognathal, and paired antero- antero-supragnathal are indicated to have supragnathals and postero-supragnathals) had a direct abutting articulation into are intimately associated with a para- these pits. In contrast, as known for at sphenoid-like plate in the ventral cranial least one of the herzeri group, Heintzichthys basis. This latter bone, previously re- gouldii, the anterior portion of the anterior C

B ( n s D | ~~~~~~~~~AbAi

Fig. 3. Diagrams showing the arrangement of the supragnathal elements and anterior basal plate in dorsal aspect and in transverse section of (A and B, respectively) Dinichthy8 terrelli and (C and D, respectively) Heintzichthys gouldii. Reproduction of A and B approximately X 1/e; and of C and D approximately X 2/3. a-a', Planes of the transverse sections (B and D) indicated on A and C; Ab, anterior basal plate; Asg, antero-supragnathal plate; fp, concavities on the forward margin of the anterior basal plate; Mg, mediognathal plate superimposed in dashed lines in A but omitted on C; p. pd., posterodorsal extremity of the antero-supragnathal plate; and Psg, postero-supragnathal element. marked by Hills (1936) for Coccosteus, by basal plate is a proportionately much Stensi6 (1934; 1945) for various other narrower, thin lamina of bone, strongly arthrodires, and completely described for arched dorsally. From observations on Dinichthys terrelli in a forthcoming paper reconstructed specimens, the flange-like by us, is apparently to be homologized with posterodorsal extremities of the Heintzich- the anterior basal element of the acantho- thys antero-supragnathal could not have dians. This element in D. terrelli is had any direct articulation with that thickened anteromesially and its forward member, but apparently possessed a type 19461 ANTERO-SUPRAGUNAPHAL OF 6/ORGONICHTHYS 7

of overlapping arrangement, only, how- terrelli and herzeri groups, and the demon- ever, through the intermediary of a wide stration that the herzeri group constitutes a expanse of the cartilaginous braincase floor. natural unit. The common possession of In addition to the above differences of basically similar gnathal elements need not gnathal and endocranial structures be- mean that D. herzeri and Gorgonichthys tween the terrelli and herzeri groups, two of clarki will prove to have anterior basal the designated members of the latter, plates identical with those of Heintzichthys Gorgonichthys clarki and Heintzichthys gouldii, and the cranial shield and body gouldii, possess in common a number of armor of D. herzeri need not exhibit the fundamental characteristics which are at comparable modifications of those common wide variance with the corresponding to G. clarki and H. gouldii. However, the structures of the terrelli group. These are: structure of the antero-supragnathal of D. cranial shields lacking the massive trans- herzeri, which militates against the con- verse and longitudinal ventral thickenings, ception of any arrangement of parts similar possessing proportionately larger orbital to that of the terrelli group, and the regular emarginations, indistinct postorbital proc- possession in all members of the herzeri esses, and relatively much shorter postero- group of certain comparable structural de- lateral marginal extent; reduced ventral velopments quite prohibit the dismissal of keels on the mediodorsal plates; only the present phylogenetic suggestions with- vestigial lateral anteroventral arms on out serious consideration.' the anterolateral bones; apparent com- 'The above considerations reflect the ad- plete loss of spinal elements; and general mitted uncertainties with which Gorgon- reduction of the spinal process'es of the ichthys clarki must still be regarded. The antero-ventrolateral plates as well as'of the clarki species is without doubt generically entire ventral armor. distinct from the members of the so-called It may be remarked that comment on terrelli group. However, the as yet in- the modifications of the cranial shield and completely revealed herzeri species, from body armor of Dinichthys herzeri has been which G. clarki is presumably derived, is the omitted in the above discussion. This has genotype of Dinichthys. Until D. herzeri been a deliberate omission. Acute cutting- is more adequately understood, the re- edged inferognathal plates which can per- moval of the clarki species from the genus tain only to some member of the so-called Dinichthys and the proposal for the re- terrelli group are frequently encountered in establishment of the term Gorgonichthys the Huron Shale. In fact, all the skeletal for its reception are based on inconclusive parts of the terrelli group exhibit so uni- evidence. Despite this, the provisional form and distinctive a combination of recognition of the clarki species as a dis- characteristics that in the opinion of these tinct morphological type is advocated by writers the vast majority of all known us because of the peculiar modifications' of Huron Shale dinichthyid specimens must its skeletal complex and the measurable dif- belong to the terrelli group. This fact ferences between its gnathal elements and would extend the geologic range of the those of D. herzeri. These latter structural terrelli group throughout the Ohio Shales adaptations are, in part, illustrated by the formation, and necessitates the implication following attempted analysis of the jaw that D. herzeri is to be considered a rather mechanism. rarely collected Huron Shale form whose The gnathostomous character of placo- dermal armor, except for the mouth parts, derm jaws has been established through has not been recognized. Thus, in the many recent studies by Gross, Heintz, absence of entirely complete specimens for Stensi6, Watson, and others. Based on the several designated members of the these fundamental interpretations the herzeri group, inference has been freely but present writers have determined that the necessarily employed in establishing the first visceral arch among the dinichthyid criteria for both the separation of the arthrodires ossified in four centers. The common Ohio Shales dinichthyids into the antero-inferognathal and postero-infero- AMERICAN MUSEUM NOVITAVES [No. 1316 gnathal represent, respectively, anterior the muscles originate, in reconstructed and posterior ossifications of the Meckelian specimens where all elements are oriented segment of the arch. The presumed in correct relationship to one another. dermal inferognathal plate is borne dorsally The distinctive modifications of Gor- astride with its anterior end lateral and its gonichthys clarki may be demonstrated by posterior extremity mesial to these primary an application of the above considerations. mandibular elements. The sub-nasal bone, For purposes of comparison the recon- referred to above, and the post-suborbital structed heads of Dinichthys terrelli, Gor- element represent, respectively, anterior gonichthys clarki, and Heintzichthys gouldii, and posterior ossifications of the dorsal all reduced to a unit size, are shown in palatoquadrate portion of the arch. Simi- figure 4. larly, thickened protuberances on the In Dinichthys terrelli it may be noted lateral face of the postero-inferognathal that the orbit is small. Its length is con- and on the inner face of the post-suborbital tained five and one-half times in the great- seem to indicate a movable articulation be- est lateral length of the cranial shield from tween these two elements, and liken them to the plane of the rostral tip to the hinder- articular and quadrate, respectively. Some most limit of the postmarginal plate. The evidences exist (Dunkle and Bungart, 1945) marginal length of the shield from post- that the sub-nasal and post-suborbital orbital process to postmarginal extremity were connected in life by a lamina of is thus the longest such dimension attained cartilage. In Dinichthys terrelli this carti- by any of the Ohio Shale arthrodires. This lage must have extended, vertically in- great length is reflected in the lengths of clined, mesial to the ventrolateral borders the posterior blade portions of both its of the marginal plate, posteriorly, and to suborbital and inferognathal elements. have occupied, horizontally inclined, the Similarly the ventrolateral expansion of the prominent groove on the mesial surface of marginal plate is proportionately tre- the "handle" portion of the suborbital mendous. The restored cross-sectional element, anteriorly. In this forward part area of the adductor muscles was in conse- it would then have served as a strengthen- quence large and indicative of massive ing support dorsolaterally for the supra- muscles. The fiber length was also rela- gnathal elements which dorsomesially were tively longer. This fact is shown by the embedded in the cartilaginous braincase more feeble extension of the posterolateral floor. From these homologies it may be wings of the cranial shield, the depth of the postulated that the adductor muscles, vertically expanded blade part of the sub- originating on the ventrolaterally ex- orbital, and the degree of upward arching panded marginal plate, passed downward of the posterior blade portion of the infero- toward their insertion on the inferognathals, gnathal plate. The jaw mechanism must mesial to the suborbital check plate and therefore be considered as both powerful lateral or structurally dorsal to the re- and fast, and the terrelli group, with their stored palatoquadrate cartilage. The acute cutting-edged mouth parts, conse- lengths of the posterior blade portions of quently appear to have been the most for- both the suborbital and inferognathal midable of the upper Devonian predators. plates as well as the length of the lateral In contrast, the orbits of both Gor- margin of the cranial shield from post- gonichthys clarki and Heintzichthys gouldii orbital process to postmarginal tip may are much enlarged. Their lengths are thus be considered indices of the length of contained two and one-half times and twice, the space occupied by the adductor muscles, respectively, in the greatest lateral length as well as of the relative mechanical of the cranial shield. The marginal length, efficiency of the lever-like jaws. Some therefore, from which the adductor muscles approximation to the cross-sectional area originate is much reduced. The ventro- of these muscles may be approached lateral expansion of the marginal is cor- through the projection of the ventrolateral respondingly small. The restored cross- expansion of the marginal bone, from which sectional area indicates much smaller 19461 ANTERO-SUPRAGNATHAL OF GORGONICHTHYS 9

A B

C D

F ft'SAl~ ~ ~ ~ ~ ~~~A/b,tO

Fig. 4. Attempted restorations of the heads and of transverse sections through the cheek regions of (A and B, respectively) Dinichthys terrelli, (C and D, respectively) Gorgonichthys clarki, and (E and F, respectively) Heintzichthys gouldii to show the degree of variation in the proportions of homologous parts. In B, D, and F presumed dermal bone is indicated by solid black, cartilage and primary ossifications by stipples, and muscle sections by hachuring. Figures are all reduced to a unit size: that of Dinichthys being approximately X 1/14; Gorgonichthys, X 1/l2; and Heintzichthys, X '/o, of average-sized specimens. Aig, antero-inferognathal plate; Asg, antero-supragnathal plate; ax. me., anterior limit of the ventrolateral expansion of the marginal plate; C, central; Eth, restored extremity of the ethmoidal cartilage; Ig, inferognathal; M, marginal; m, expanded ventrolateral edge of the marginal plate; Mg, mediognathal; ml. ad., frontal section through the restored adductor muscles; mt. ad., transverse section through the restored adductor muscles; N, nuchal plate; n, external nare; or, orbit; P, pineal plate; Pig, postero-inferognathal; Pm, postmarginal; Pn, paranuchal; Po, postnasal; p. pto., postorbital process; pq, restored palatoquadiate cartilage; PrO, pre-orbital; Psg, postero-supragnathal; Pso, post-suborbital; PtO, postorbital; px. me., posterior limit of the ventrolateral expansion of the marginal plate; R, rostral; Sn, sub-nasal plate; So, suborbital plate; and X, sub-marginal plate. 1.0 AAIERICAN MUSEUM NOVITA TES [No. 1316

muscles than those possessed by the terrelli lengths of its postero-supragnathal and group. Shorter fiber length is also indi- opposing oral margin of the inferognathal cated by the greater extension of the were increased. Gorgonichthys, on the posterolateral wings of the cranial shield other hand, with proportionately the and low suborbital blades in both genera. smallest muscles of the three genera dis- Interestingly, the decrease in mechanical cussed, reduced the length of the infero- efficiency of the jaw mechanism and these gnathal part functionally opposed to the apparent losses in both power and speed of postero-supragnathal and increased both the muscles were, however, compensated the size of the antero-supragnathal and the for in two distinct adaptations. Heintzich- height of the inferognathal anterior cusp. thys, on the one hand, with relatively This modification in attempting to concen- larger muscles than Gorgonichthys, reduced trate all of the available force of weak the anterior cusp of the inferognathal and muscles into one magnified point made the proportionate size of the antero- Gorgonichthys a veritable "saber-toothed" supragnathal. In return, the functional arthrodire.

REFERENCES CLAYPOLE, E. W. known dinichthyids. Bull. Amer. 1892. A new gigantic placoderm from Ohio. Mus. Nat. Hist., vol. 21, pp. 409-414, Amer. Geol., vol. 10, pp. 1-4, 1 text fig. 2 text figs., 2 pls. DUNKLE, D. H., AND P. A. BUNGART 1906. Studies on the Arthrodira. Mem. 1940. On one of the least known of the Cleve- Amer. Mus. Nat. Hist., vol. 9, pt. 3, land Shale Arthrodira. Sci. Publ. pp. 103-154, 25 figs., 2 pls. Cleveland Mus. Nat. Hist., vol. 8, no. 1930. (Abstract.) Dental elements in the 3, pp. 29-48, 7 text figs., 2 pls. arthrodire Titanichthys. Bull. Geol. 1942. The inferognathal plates of Titanich- Soc. Amer., vol. 41, p. 196. thys. Ibid., vol. 8, no. 4, pp. 49-60, NEWBERRY, J. S. 4 text figs. 1885. Description of some gigantic placoderm 1945. A new arthrodiran fish from the Upper fishes recently discovered in the Devonian Ohio Shales. Ibid., vol. 8, Devonian of Ohio. Trans. New York no. 8, pp. 85-95, figs. 1-3. Acad. Sci., vol. 5, pp. 25-28. STENSIO, E. A:SON EASTMAN, C. R. 1934. On the heads of certain arthrodires. I. 1900. Dentition of some Devonian fishes. Pholidosteus, Leiosteus, and acantha- Jour. Geol., vol. 8, no. 1, pp. 32-41, 7 spids. K. Svenska Vetensk. Handl., text figs. vol. 13, no. 5, pp. 1-79, 30 figs., 14 pls. HEINTz, A. 1945. On the heads of certain arthrodires. 1931. The anterior-lateral plate in Titanich- II. On the cranium and cervical thys. Ann. Mag. Nat. Hist., ser. 10, joint of the Dolichothoraci (Acantha- vol. 8, pp. 208-212, 3 figs. spida). Ibid., ser. 3, vol. 22, no. 1, HILLS, E. S. pp. 1-70, 14 figs. 1936. On certain endocranial structures in WATSON, D. M. S. Coccosteus. Geol. Mag., vol. 73, pp. 1937. The acanthodian fishes. Philos. 213-226, 6 figs., 1 pl. Trans., Roy. Soc. London, ser. B, vol. HUSsAKOF, L. 228, no. 549, pp. 49-146, 25 figs., 10 1905. On the structure of two imperfectly pls.