L I B RAFLY 
 
 OF THE 
 
 UNIVERSITY 
 
 OT ILLINOIS 
 
 N ATURAL IIigTCRY -STRyEr" : ' 
 
 550.5 
 FI 
 
 v.AjCop.Q. 
 
 REMOTE STORAGi 
 
LIBRARY 
 
 UNIVERSITY OF ILLINOIS 
 
 URBANA 
 
Field Museum of Natural History 
 
 Publication 232 
 
 Geological Series Vol. IV, No. 4 
 
 ON THE HEAD OF THE MACROPETALICHTHYIDS 
 
 with certain remarks on the head of 
 the other arthrodires 
 
 BY 
 
 Erik A: son Stensio 
 Royal State Museum, Stockholm 
 
 Oliver Cummings Farrington 
 Curator, Department of Geology 
 
 Chicago, U. S. A. 
 
 October, 1925 
 
PREFACE 
 
 The Museum specimen (P 1154) on which the following paper by 
 Dr. Stensio is chiefly based, is from the Onondaga (Corniferous) 
 Middle Devonian limestone of Leroy, Genesee County, New York. 
 The original collector is not known to the Museum. The specimen 
 together with another (P 11 55), from the same locality, was obtained 
 by the Museum from Ward's Natural Science Establishment, Roches- 
 ter, New York, in 1894. Its remarkably complete preservation has 
 enabled Dr. Stensio to make the profound anatomical study that is 
 presented in the following pages. 
 
 The matrix of this specimen when freshly broken is seen to be a 
 dark brown, partially crystallized limestone. This limestone is some- 
 what argillaceous and bituminous, and slightly magnesian. Here and 
 there it exhibits the inclusions of chert which are characteristic of 
 Corniferous limestones. On weathered surfaces the matrix is chalk- 
 white in color, Upon which the dull black parts of the fossil fish stand 
 out in contrast. 
 
 In specimen P 11 55, only the dermal bones of the cranial roof of a 
 Macropetalichthys are preserved. This head is larger than that of 
 P 1 1 54, being 19 centimeters in length as compared with 12 centimeters 
 in P 1 1 54. The width (12 centimeters) is about the same for both. 
 The plates preserved in the larger specimen are of a dark, calcareous 
 substance, about 1 mm. thick. Except for a black pigment, this sub- 
 stance dissolves readily in dilute hydrochloric acid. On removal of the 
 remains of these plates, imprint of a pustulose external surface can be 
 plainly seen on the overlying rock. In the limestone matrix of both 
 specimens, numerous aggregates of more or less comminuted brach- 
 iopod shells and crinoid stems are visible. Among the brachiopods so 
 preserved, individuals of the species Leptaena rhomb oidalis and A try pa 
 reticularis can be identified. 
 
 Oliver C. Farrington. 
 
 87 
 
ON THE HEAD OF THE MACROPETALICHTHYIDS 
 
 WITH CERTAIN REMARKS ON THE HEAD OF THE OTHER ARTHRODIRES 
 
 BY 
 
 Erik A: son StensiS 
 
 CONTENTS 
 
 Introduction 91 
 
 Description of the Macropetalichthyids 92 
 
 Macropetalichthys rapheidolabis 92 
 
 Primordial neurocranium : 
 
 General remarks 92 
 
 Occipital region 95 
 
 Labyrinth region 103 
 
 Labyrinth cavity 116 
 
 Orbitotemporal region 119 
 
 Ethmoidal region 129 
 
 Cavum cerebrale cranii and brain 134 
 
 Dermal bones of the primordial neurocranium 140 
 
 Dermal bones of the cheek 143 
 
 Visceral skeleton 144 
 
 Dentition 146 
 
 Sensory canals of the head 146 
 
 Macropetalichthys agassizi? 150 
 
 Macropetalichthys pelmensis 152 
 
 Epipetalichthys wildungensis, gen. nov., sp. nov. 
 Primordial neurocranium: 
 
 General remarks 152 
 
 Microscopic structure of the bone 153 
 
 Occipital region 154 
 
 Labyrinth and orbitotemporal region , 155 
 
 Ethmoidal region 156 
 
 Cavum cerebrale 157 
 
 Dermal bones of the cranial roof 157 
 
 Sensory canals of the head 158 
 
 Some general remarks on the Macropetalichthyids 160 
 
 Certain remarks concerning the other non-Macropetalichthyid 
 
 Arthrodires 164 
 
 The Phlyctenaspids 165 
 
 The Coccosteids 1 70 
 
 The Homosteids 180 
 
 The Mylostomids 180 
 
 The Ptyctodontids 181 
 
 The Jagorinids 182 
 
 Some general remarks on the non-Macropetalichthyid Arthrodires. 186 
 
 Concluding remarks on the affinities of the Arthrodires 187 
 
 Bibliography 191 
 
 Explanation of Plates 198 
 
 89 
 
INTRODUCTION 
 
 As the holder of a scholarship from the University of Upsala and 
 with support from the Palaeontological Museum of Kristiania, I under- 
 took in 1922 a journey to the United States of America for the study 
 of certain groups of fossil vertebrates, especially Ostracoderms and 
 Arthrodires. 
 
 In examining the collection of the Arthrodires in Field Museum 
 of Natural History at Chicago I found a specimen of Macropetcdichthys 
 raphcidolabis, (Field Museum Cat. No. P 1154), that if prepared in a 
 proper way was likely to show the primordial neurocranium rather com- 
 pletely preserved. Through the courteous assistance of Mr. H. W. 
 Nichols, Associate Curator of Geology of the Museum, permission was 
 secured for me to carry out the preparation of the specimen which I 
 found necessary for my studies. While carrying on this work it soon 
 became evident that the specimen was of much greater importance from 
 the anatomical point of view than I at first thought; for most of the 
 canals for nerves and vessels, the entire cavum cerebrale and a part of 
 the labyrinth cavity were in such a good state of preservation that they 
 could be examined in detail. I spent ten days working with the speci- 
 men and during this time the photographs reproduced in Pis. XIX- 
 XXVI were made by the photographer of the Museum. (Cf. Annual 
 Report of the Director, Field Museum of Nat. Hist., Publication 213. 
 Report series, Vol. VI, No. 2, p. 121.) 
 
 The present paper is chiefly based on the specimen of Macropeta- 
 UcJithys rapheidolabis, in Field Museum of Natural History at Chicago, 
 that has just been referred to. But as this specimen was imperfect 
 regarding the occipital region, this was studied on certain specimens in 
 the American Museum of Natural History in New York, a few of 
 which — those figured in this paper — were sent to me in 1923 in Sweden 
 for a re-examination in certain respects. 
 
 The specimen of the new, interesting form Epipetalichthys wildun- 
 gcnsis described in the present paper was, in the most courteous way, 
 placed at my disposal by Professor O. Jaekel of Greifswald, and for 
 the opportunity to examine the specimen described below under the 
 name of Macropetcdichthys agassisi? I am indebted to Mr. S. Junker- 
 mann of Bielefeld. 
 
 9i 
 
92 Field Museum of Natural History — Geology, Vol. IV. 
 
 It is a pleasant duty for me to express here my most sincere thanks 
 to Mr. H. W. Nichols, of Field Museum of Natural History, for all 
 the help I received from him and for the kindness he showed to me 
 during my stay at Chicago. Further I wish to express hearty thanks 
 to Dr. W. K. Gregory and Dr. W. D. Matthew for their help in 
 facilitating my studies at the American Museum of Natural History 
 in New York, and for their great kindness in sending material to me 
 in Sweden. Finally, I am also highly indebted to Professor O. Jaekel 
 of Greifswald, both for the material which I received from him for 
 this paper and for important information concerning the Wildungen 
 Arthrodires upon which he is working. 
 
 All the drawings for this paper were made by Mr. G. Liljevall, 
 of Stockholm. 
 
 DESCRIPTION OF THE MACROPETALICHTHYIDS 
 
 MACROPETALICHTHYS RAPHEIDOLABIS 
 Norwood & Owen 
 
 (Pis. XIX-XXVII, PI. XXVIII, figs. 3-5J PI. XXX, fig. 2; PI. XXXI, fig. 1) 
 PRIMORDIAL NEUROCRANIUM 
 
 General Remarks. The primordial neurocranium is strikingly 
 broad and low, its maximum width almost equalling the length and its 
 maximum height constituting only about one-seventh of the length and 
 between one-fifth and one-sixth of the maximum width (Figs. 1, 3, 
 5, 8; Pis. XIX-XXII; PI. XXIV, figs. 1, 3). The maximum height 
 is situated in the posterior and the maximum width in the anterior half 
 of the labyrinth region. 
 
 As is seen from text figs. 4, 5, 8, 9, and PI. XXIV, figs. 1, 3, the 
 primordial neurocranium is so arched both in a transversal and longi- 
 tudinal direction, that its dorsal surface is convex and its ventral sur- 
 face correspondingly concave. 
 
 Concerning the shape of the different divisions of the primordial 
 neurocranium, it may be especially pointed out in this connection that 
 the occipital and ethmoidal regions have a considerable length, while 
 the labyrinth and orbitotemporal regions are short. The latter region 
 is even strikingly short and at the same time very broad, its length 
 being only about one-half of the width, a condition which is due partly 
 to the slight development of the orbits and partly to the considerable 
 width of the skull as a whole. With regard to the orbits, it is in addi- 
 tion noteworthy that they have shifted some distance backwards and 
 medially and that they are much directed upwards. 
 
Macropetalichtiiyids and other Arthrodires — Stensio 93 
 
 The primordial neurocranium consisted of cartilage and bone, the 
 latter containing fairly numerous cell spaces. The bone occurs, how- 
 ever, only as thin layers, which covered the cartilage on the cerebral 
 surface, within the labyrinth cavity, to a large extent on the external 
 surface, and in addition also formed a lining membrane" for all the 
 canals for the vessels and nerves piercing the walls of the neurocra- 
 nium. Accordingly, there may be distinguished an internal bone-layer, 
 a labyrinth bone-layer, an external bone-layer and canal bone-layers. 
 More closely defined, these different layers of bone, which evidently 
 are perichondral, have the following extension and relations: 
 
 i. The internal layer forms a continuous covering of the cerebra/ 
 surface of the primordial neurocranium. In other words, it lines ihe 
 whole cavum cerebrale cranii. 
 
 2. The labyrinth layer forms a continuous lining membrane of the 
 whole labyrinth cavity (cavities for the semicircular canals included). 
 
 3. The external layer covers the outside of the primordial neuro- 
 cranium ventrally, laterally and posteriorly, but dorsally, on the con- 
 trary, only in the posterior narrow division of the occipital region and 
 probably in a short anterior part of the ethmoidal region. The remain- 
 ing parts of the dorsal side of the occipital and ethmoidal regions, as 
 well as the whole dorsal side of the labyrinth and orbitotemporal re- 
 gions, are destitute of a perichondral bone-layer, so that the cartilage 
 must have appeared there to the extent shown in Fig. 3. The por- 
 tion of the external layer covering the posterior side of the anterior 
 broad part of the occipital region, is connected with the inner layer by 
 a few, chiefly sagitally placed, laminae of bone, which passed through 
 the cartilage, and which naturally are enchondral. 
 
 4. The canal layers line all the canals traversing the cranial walls, 
 even canals and branches of canals with a very narrow diameter. In 
 those cases where the canals pass from the cavum cerebrale to the out- 
 side, the layers lining them are continuous with and connect the external 
 and internal bone layers with one another. In a similar way the internal 
 layer is continuous with the labyrinth layer by layers lining such canals 
 which run from the cavum cerebrale to the labyrinth cavity. 
 
 The different layers of bone just described actually form by their 
 connections with each other a single large bone extending throughout 
 the length of the primordial neurocranium. That this large bone was 
 formed by the coalescence of a few smaller ones seems not improbable, 
 since vestiges of ossification centres perhaps occur in certain places. 
 
94 Field Museum of Natural History — Geology, Vol. IV. 
 
 
 elf- 
 
 
 s.pal - - - 
 
 
 c.pal 3 
 
 
 N^S^ 
 
 
 JU '-%* 
 
 c. 
 
 car.ext-M« 
 
 chy- 
 
 -f.arthnv 
 
 ^cr.Sp 
 
 hcem. 
 
 Fig. i. Macropetalichthys rapheidolabis 
 
 Restoration of primordial neurocranium in ventral aspect. Occipital region 
 after specimens in the American Museum of Nat. Hist., New York, other regions 
 after a single specimen in Field Museum of Nat. Hist., Chicago. The dimensions 
 of the occipital region probably not fully correct in relation to those of the other 
 regions. Cartilage dotted. X^- 
 
 c.car.ext, canal for the arteria carotis externa; c.hy, canal for the vena 
 hyoidea; ch.w, ridge caused by the most anterior part of the notochord; c.pah, 
 opening for the r. palatinus facialis ; era, canal for the radix aortae ; cr.ol, crista 
 occipitalis lateralis ; cr.sp, cranio-spinal process ; dix, probable foramen for the 
 dorsal branch (lateralis branch) of the n. glossopharyngeus ; f.art.hm, probable 
 place of articulation for the dorsal end of the hyoid arch ; ham, haemal groove ; 
 ju, canal traversed in the anterior part by the vena mandibularis ; in the longer 
 posterior part by the v.jugularis (behind the confluence between this and the 
 vena mandibularis) ; jiix and juz, anterior and posterior openings of the canal ju, 
 the anterior one traversed by the vena mandibularis ; na, nasal opening ; nf, nasal 
 fontanelle; n.l, canal for the n.lineae lateralis; olf, olfactory capsule; pr, postero- 
 lateral process of the anterior broad division of the occipital region; s.pal, groove 
 for the r.palatinus facialis ; s.ra, groove for the radix aortae ; Sx, groove leading 
 some distance anteriorly from the external opening of the vagus canal; IX, canal 
 for the n. glossopharyngeus; Xxn, canal for the n.vagus and the vena cerebralis 
 posterior. 
 
Macropetalichthyids and other Arthrodires — Stensio 95 
 
 The position of these presumed centres is seen in Figs. I and 3, 
 and will be further dealt with in the subsequent description of the dif- 
 ferent regions of the primordial neurocranium. 
 
 Occipital Region 
 
 The occipital region is very long, its length probably being continued 
 between two and three times into the length of the primordial neuro- 
 cranium (Figs. 1, 3, 5; PI. XXVII; PI. XXVIII, figs. 4, 5). With 
 regard to its shape otherwise it may be considered as composed of two 
 divisions, an anterior one and a posterior one. 
 
 The posterior one of these divisions (Figs. 1, 3, 5; PI. XXVII; 
 PI. XXVIII, figs. 3-5; PI. XXX, fig. 2) is much longer than the 
 anterior one but, on the other hand, considerably narrower and lower. 
 With the longer anterior part it does not by far extend up to the 
 cranial roof. At its posterior end it has a paired laterally and dorsally 
 projecting process (cr-sp), which is coalesced with its fellow of the 
 other side medially and so connected on the anterior side with a 
 descending lamina from the posterior bones in the dermal cranial roof 
 that it forms a support for these. The process much resembles the 
 cranio-spinal process in Acipenscr, Polyodon and Saurichthys (Huxley 
 1864, Fig. 82; Parker 1882, PI. 15, fig. 13; PI. 16, figs. 1, 3; Iwanzow 
 1887, PI. 1, fig. 2; Gegenbaur 1912, pp. 319-325; Stensio 1925) and 
 may therefore be termed a cranio-spinal process although it obviously 
 has evolved independently of that in the fishes enumerated. 
 
 The most postero-ventral part of the posterior division situated 
 beneath that from which the cranio-spinal process issues is not found 
 preserved in any of the specimens investigated, a fact which seems to 
 indicate that it lacked the external bone layer and thus was entirely 
 cartilaginous when seen from the outside (Figs. 1, 3, 5; PI. XXVII; 
 PI. XXVIII, figs. 3-5; PI. XXX, fig. 2). It probably had a paired 
 condyle or a paired fossa for articulation with the vertebral column, 
 as a cranio- vertebral joint might have been present as in ordinary 
 Arthrodires. 
 
 Two transverse sections through the posterior division are shown 
 in text fig. 2. As is understood from these and Fig. 5, the pos- 
 terior division is much flattened in a dorso-ventral direction, its height 
 being only about one-half of the width. As a consequence of the flat- 
 tening, the external surface has no lateral fields, but only a dorsal and 
 a ventral field separated from each other by a paired, well pronounced 
 lateral crista, the crista occipitalis lateralis (cr.ol). The dorsal one of 
 the two fields, which, as mentioned above, is situated deep below the 
 
96 Field Museum of Natural History — Geology, Vol. IV. 
 
 dermal cranial roof except at the posterior end, is provided with a 
 longitudinal unpaired crista, the crista occipitales dorsalis (cr.od), while 
 the ventral field, on the contrary, has a wide and deep longitudinal 
 groove. This groove (ham, Figs, i, 2; PI. XXVIII, figs. 3, 4) 
 constitutes undoubtedly an anterior, cephalic portion of the haemal 
 groove of the vertebral column, and may therefore properly be termed 
 the haemal groove. Towards the transition of the anterior division of 
 the region, it becomes gradually shallower and seems to divide there 
 into two shallow branches (Fig. 1; PI. XXVII, fig. 4), a right and 
 a left one, each of which is continued forwards by the groove s.ra 
 which will be further dealt with below in another connection. 
 
 The part of the cavum cerebrale cranii situated within the pos- 
 terior division is long and low and narrow in comparison with those 
 in front of it. It is, however, somewhat narrower in its anterior, than 
 in its posterior portions, as is well shown by Fig. 10 and PI. XXVII, 
 fig. 1. A transverse section through its posterior half is seen in Fig. 2. 
 
 The notochord extended into the basal part of the neuro-cranium, 
 but was very slightly developed there (ch, Fig. 2) and reached 
 forwards only to about the transition between the anterior and posterior 
 divisions of the occipital region (PI. XXVIII, fig. 4). It was, thus, 
 in the neurocranium, limited to the posterior division of the occipital 
 region, while the parts of it that originally existed further forward 
 as far as to the dorsum sellae have become reduced. The persistent 
 part was surrounded by a thin bone layer of its own, and this bone 
 layer, which is well preserved in the fossil, forms a fine longitudinal 
 tube, a tube which is coalesced with the inner bone layer dorsally and 
 the external bone layer (for explanation of these bone layers see p. 
 93 ventrally at the bottom of the haemal groove (Fig. 2). In the 
 anterior part of the haemal groove the notochord has pressed the 
 external bone layer downwards as a ridge, the notochordal ridge (ch.w), 
 which is conspicuous when the haemal groove is seen from below or in 
 transverse section (Figs. 1, 2B ; PI. XXVIII, fig. 4). 
 
 The slight development of the cephalic portion of the notochord of 
 the fish, forms, it seems to me, a support for the view advanced above, 
 that there was a kind of articulation between the head and the body as 
 in typical Arthrodires. (Adams 1919, pp. 123-127; Dean 1901, Figs, 
 pp. 116-117; Eastman 1908a, pp. 113-149; 1908b, pp. 158-205; Jaekel 
 1902, p. 106; 1907, p. 171; 1919, pp. 96-108; etc.) Such an articula- 
 tion required of course a considerable weakening of .the notochord at 
 the transition between the vertebral column and the occipital region of 
 the neurocranium, and if such a weakening took place the cephalic por- 
 
Macropetalichthyids and other Arthrodires — Stensi6 97 
 
 tion of the notochord must evidently have lost its importance and 
 decreased in width and extension, just as it has done in recent Chi- 
 maeroids, which, as we know, have a sort of cranio-vertebral articula- 
 tion. 
 
 From its very considerable length it would be expected that the 
 posterior division of the region consisted of a rather large number of 
 coalesced segments, and, accordingly, that it was pierced by numerous 
 canals for spino-occipital nerves. No canals of this kind could, how- 
 ever, be found on the material investigated, and it is therefore impos- 
 sible at present to decide anything with certainty concerning the num- 
 ber of segments composing the division. 
 
 The anterior division of the region, if we now proceed to this 
 (Figs, i, 3, 5; Pis. XIX-XXIV; PI. XXV, figs, i, 2; PI. XXVI, figs. I, 
 2; PI. XXVII, fig. i), is, as already pointed out, much shorter, but, on 
 the other hand, considerably higher and broader than the posterior one. 
 Anteriorly, its dimensions are equal to those in the adjacent part of the 
 labyrinth region, so that it is really in no way marked off from this. 
 
 Mi i f . \_i iti iTif > rr nr 
 
 heem. 
 
 Fig. 2. Macropetalichthys rapheidolabis 
 
 A. Transverse section through the anterior part of the posterior half of the 
 posterior division of the occipital region. 
 
 B. Transverse section through the anterior half of the same division (accord- 
 ingly taken posterior of A). 
 
 After specimens in the American Museum, New York. Bone layers with 
 black, continuous lines. Cartilage dotted. Xi- 
 
 ch, cavity for the chorda dorsalis; ch.w, ridge in the haemal groove caused by 
 the chorda dorsalis; cr.od, dorsal occipital crista; cr.ol, lateral occipital crista; 
 cv, cavum cerebrale cranii; ham, haemal groove. 
 
 The postero-latero-dorsal corner of each side is produced backwards 
 and laterally to a rather large process supporting the lateral parts of 
 the dermal cranial roof. 
 
 For the detailed description we may conveniently consider that the 
 anterior division of the region is composed of five walls, which are as 
 follows : a ventral, a posterior, a right and a left lateral and a dorsal. 
 Of these the ventral, the dorsal, and the two laterals are, as is evident 
 from the figures, directly continuous with, and in no way marked off 
 from, the correspondingly situated ones in the labyrinth region. 
 
 The ventral wall, which is the smallest one of the five, is as a 
 whole narrow in comparison with the dorsal, although it rapidly in- 
 
98 Field Museum of Natural History — Geology, Vol. IV. 
 
 creases in width forwards (Fig. i; Pis. XXI-XXIII). As is well 
 shown in PI. XXIII; PI. XXV, figs, i, 2, and PI. XXVI, figs. 1, 2, 
 ic is strikingly thin, especially in the median parts, and, as already 
 pointed out, the notochord was entirely reduced in it. It is not pierced 
 by any canals and has its externa^ (ventral) surface, the shape of which 
 may be understood from text fig. 1, well bounded along each lateral 
 border by a sharp edge. ( 
 
 The posterior wall (Fig. 3; PI. XXIV, fig. 3; PI. XXV, fig. 2; 
 PI. XXVII, fig. 1) is large and mostly rather thick. Its ventro-median 
 parts are coherent with the posterior division of the region, its dorso- 
 median parts are a little inclined forwards and its lateral parts stand 
 approximately vertical or may, most laterally, be inclined a little back- 
 wards. The wall is covered by the external bone layer on the outside 
 and the internal bone layer on the inside, as are also the other walls, 
 but, as mentioned above, it is, in addition, provided with a few sagittally 
 running bone laminae (Is, PI. XIX; PI. XXV, fig. 2), which connected 
 the internal and external bone layers with one another. These bone 
 laminae in the fresh specimens traversed the cartilage. 
 
 The external surface of the posterior wall has, as is seen from 
 PI. XXIV, fig. 3, a considerable extension, which faces a little upward 
 in the dorso-medial parts and straight backward or backward and a 
 little downwards in the lateral parts. It is concave both in the trans- 
 versal (Figs. 1, 3, 5; Pis. XIX-XXIV; PI. XXVII, fig. 1) and dorso- 
 ventral directions and by an anterior extension of the crista occi- 
 pitalis dorsalis (cr.od) dorsally, and the posterior, narrow division 
 of the region, vertically, it is divided into a right and a left half. The 
 internal surface of the same wall, that is lower and narrower than 
 the external one, faces forward and downward (Fig. 11; PI. XX; 
 PI. XXIV, fig. 3; PI. XXV, figs. 1, 2; PI. XXVI, figs. 1, 2) and, like 
 the external one, is slightly concave both in transversal and dorso- 
 ventral directions. 
 
 The posterior wall is perforated by a paired, rather wide, postero- 
 laterally running canal (n.l, Figs. 1, 10, 11; PI. XXIV, fig. 3; PI. 
 XXVII, fig. 1), which has its external opening rather far laterally 
 on the external surface of the wall. This canal is, as will be evident 
 from facts given below, a branch from the wide vagus canal (X vn ) 
 and transmitted the n. lineae lateralis. 
 
 The dorsal wall (Fig. 3; Pis. XIX, XX; PI. XXIV, figs. 1, 3; 
 Pi. XXV, figs. 1, 2; PI. XXVI, figs. 1, 2) has a very considerable 
 breadth in relation to its length. In its median part it is rather thin, 
 but grows rapidly thicker towards the lateral parts. It is much arched 
 
Macropetalichthyids and other Arthrodires — StensiG 99 
 
 in transversal direction so that the convexity faces upwards. As has 
 already been pointed out, it has no external bone layer (Fig. 3) 
 and no definite external boundary. Despite this, it is, however, fully 
 evident that its upper surface had no fossae or depressions for por- 
 tions of the trunk muscles penetrating from behind, beneath the dermal 
 bones, for the external bone layer of the posterior wall has no emar- 
 ginations in its upper border and reaches in all its breadth quite up to 
 the lower surface of the dermal bones above it. (PI. XXIV, fig. 3.) 
 The sensory canals of the cranial roof being, as we shall see, situated 
 in rather strongly developed ridges on the lower side of the dermal 
 bones, the upper surface of the dorsal wall of the primordial neuro- 
 cranium must have had grooves for these ridges in the way shown 
 in text figs. 3 and 5. These grooves in the subsequent description will 
 be referred to as sensory canal grooves. (In the Field Museum speci- 
 men (Pis. XIX, XX; PI. XXIV, fig. 1) the dermal bones of the cranial 
 roof with the exception of those parts situated deepest, namely, the 
 sensory canal ridges, have weathered away, so that we can from the 
 conditions there easily see that the sensory canal ridges must have been 
 situated in grooves on the upper side of the primordial neurocranium) . 
 
 The canal for the n. lineae lateralis (n./.), during its passage through 
 the posterior wall of the division, gives off several fine branches in a # 
 more or less straight dorsal direction. These fine branches (Id^-ldg, 
 Pis. XIX, XX; PI. XXIV, fig. 3; PI. XXVII), which number at least 
 six and are surrounded in their total length by a thin bone layer, pass 
 upward through the dorsal wall and open into an overlying portion 
 of the sensory canal system, thus having a course that fully proves 
 them to have transmitted lateralis fibres and that the canal from which 
 they are given off was traversed by a thick lateralis nerve. As the 
 latter canal is the only one that issues backward from the vagus canal, 
 and, in addition, as we shall see, forms the sole possible way backward 
 to the abdominal region for all the lateralis fibres that accompanied 
 the vagus roots proper at the exit from the cavum cerebrale cranii, it 
 undoubtedly transmitted the whole n. lineae lateralis. The portion of 
 the sensory canal system of the head innervated from it is thus the 
 cephalic division of the main lateral line (Cf. the description of the. 
 sensory canal system below). In the Field Museum specimen, in which 
 the dermal bones of the cranial roof have weathered away and certain 
 parts of the underlying dorsal and posterior walls of the primordial 
 neurocranium have split off, the canal for the n. lineae lateralis and its 
 dorsal branches are, as may be seen in the figures mentioned, beauti- 
 fully displayed. 
 
ioo Field Museum of Natural History — Geology, Vol. IV. 
 
 Some distance medially of these dorsal branches, a rather fine, 
 paired canal (cl, Figs. 3, 5 ; Pis. XIX, XX) goes upward through 
 the dorsal wall and, after arriving close to the upper surface of this, 
 turns abruptly medially, continuing in this direction almost until it 
 meets its fellow of the opposite side in the median line. During this 
 superficial, medially directed, ^art of its course it sends out several 
 branches both forward and backward. It could not be traced quite to 
 its ventral end and hence a positive explanation of its function can 
 not be given. It seems most probable, however, that it transmitted 
 bundles of communis and general cutaneous fibres from the vagus to 
 the cranial roof, and thus transmitted a ramus supratemporalis vagi. 
 Besides this it may perhaps also have been traversd by vessels and, in 
 addition, perhaps by a branch of lateralis fibres, as it communicated with 
 the canal for the n. linese lateralis by a fine branch. The destination 
 and morphological importance of this possible bundle of lateralis fibres 
 will be discussed below in connection with the description of the sen- 
 sory canal system. 
 
 The lateral walls of the division (Figs. 1, 5; Pis. XXI-XXIII; 
 PI. XXIV, figs. 1, 2; PI. XXV, figs. 1, 2; PI. XXVI, figs. 1-4; PI. 
 XXVII, fig. 1) are thick throughout their extension, the thickness 
 being, however, not equal in all parts but gradually increasing upwards. 
 They are covered by the external bone layer on the outside and the 
 internal bone layer on the inside. Each has its external surface directed 
 laterally and much downwards and its internal surface almost straight 
 medially. A distinct but rather narrow groove (s.ra) runs along the 
 ventral edge of the external surface, a groove, which, as we shall see, 
 continues forward on to the external surface of the lateral wall of the 
 labyrinth region and backwards to the haemal groove, at the anterior 
 end of which it meets its fellow of the opposite side in the median 
 line. This groove undoubtedly lodged the radix aortse (lateral dorsal 
 aorta) of its side, as will be evident from the account of the labyrinth 
 region given below. 
 
 At the transition to the labyrinth region, each lateral wall is pierced 
 by the vagus canal. This canal (X vn , Figs. 5, 10, 11, 12, 13; Pis. 
 XX, XXI; PI. XXIV, fig. 2; PI. XXV, figs. 1, 2; PI. XXVI, figs. 
 1-4; PI. XXVII, fig. 1), which is wide and on account of the thickness 
 of the lateral wall at this place has a fairly considerable length, runs 
 almost straight laterally to the external surface of the lateral wall, 
 where it opens with a wide, rather forward facing foramen (X VD , 
 Figs. 1, 5; Pis. XXII, XXIII; PI. XXIV, figs. 1, 2). A proximal 
 short part of it is much higher than broad (PI. XXV, figs. 1, 2; PI. 
 
Macropetalichthyids and other Arthrodires — Stensio ioi 
 
 XXVI, figs. I, 2) and exhibits no certain evidence of subdivision by 
 longitudinal septa. 
 
 On the other hand the remaining distal, and by far longer, part, has 
 a more circular section than the former and is subdivided into two 
 incompletely separated divisions, a narrow and a wide one, by two 
 longitudinal ridges, divisions which perhaps in the living animal were 
 totally separated from each other by a membrane of connective tissue ex- 
 tending between the ridges. In the proximal part of the distal half of 
 
 -cr.sp 
 
 Fig. 3. Macropetalichthys rapheidolabis 
 
 Primordial neurocranium in dorsal view. Occipital region after specimens in 
 the American Museum, New York, and a specimen in Field Museum, Chicago; 
 other regions after the latter specimen. Perichondral bone with a dark tone. 
 Cartilage dotted. XV2. 
 
 cl, canal probably for a dorsal branch of the vagus and a bundle of lateralis 
 fibres; cr.od, dorsal occipital crista; cr.ol, lateral occipital crista; cr.sp, cranio- 
 spinal process ; d.end, canal for the ductus endolymphaticus ; f.end, fossa endo- 
 lymphatica; olf, olfactory capsule; orb, orbit; par, pineal opening; pfg, cavum 
 precerebrale ; pr, postero-lateral process of the anterior broad division of the 
 occipital region. 
 
 
io2 Field Museum of Natural History — Geology, Vol. IV. 
 
 the canal the narrow division (X v , Fig. u; PI. XXIV, fig. 2) is 
 situated dorsally of the wide one (X n , PI. XXIV, fig. 2), but during 
 the passage outwards they gradually change their mutual position so 
 that the former finally lies behind the latter and occupies the posterior 
 portion of the common external opening, which is distinctly broader 
 than high. (PI. XXIV, fig. rf{ 
 
 From conditions in other fishes we may conclude with great prob- 
 ability that in M. rapheidolabis the narrow one of the two divisions just 
 described in the vagus canal transmitted the vena cerebralis posterior 
 (v. encephalica posterior) and the wide one the vagus proper and in 
 the proximal half the n. linear lateralis too. For the vena cerebralis 
 posterior in all fishes which have been closely investigated with regard 
 to this, arises by the confluence of anterior and posterior branches from 
 the dorsal side of the medulla oblongata and enters the cranial wall 
 dorsally of the vagus roots. It then within the cranial wall usually 
 turns a little forward and crosses the vagus roots on the dorsal side 
 so that when reaching the outer surface it is situated just in front of 
 these roots and not as in M. rapheidolabis posterior to them. In cer- 
 tain forms it runs through the same canal with the vagus roots, in 
 other forms it is more or less completely separated from them by a 
 septum of bone, cartilage or connective tissue. (Cf. Stensio 1922, p. 
 172; 1925, p. 21; Grosser 1907, Fig. 4; O'Donoghue 1914, p. 442; 
 Pollard 1892a, PI. XXIX, fig. 23; Allen 1905, pp. 87-89; Rex 1891, 
 Pis. 15, 16; etc.) Similar conditions were also found in a specimen of 
 Chimcera monstrosa, which was dissected by me. 
 
 During the passage through the lateral wall, the vagus canal in 
 M. rapheidolabis gives off from its proximal half two branches, one on 
 the posterior and the other on the anterior side. The former of these 
 O.Z, Figs. 10, 11; Pis. XXI, XXIII; PI. XXIV, fig. 3; PI. XXVII, 
 fig. 1), which issues rather low, is the canal for the n. lineae lateralis 
 and has already been dealt with above in connection with the descrip- 
 tion of the posterior and dorsal walls. As pointed out there, it runs 
 postero-laterally to the outer surface of the posterior wall, sending out 
 several fine branches upwards to the cephalic division of the lateral line. 
 The other branch (d x , Figs. 10, 11; PI. XXIV, fig. 2), which is 
 much narrower and issues much higher up and more proximally, runs 
 some distance almost straight dorsally, then not so very deep below the 
 dorsal surface of the dorsal wall, bifurcating into an antero-lateral (d x& , 
 PI. XXIV, fig. 2) and an antero-medial (d xh , PI. XXIV, fig. 2) 
 ramus, each of which has a superficial course beneath a sensory canal 
 division, to which fine ramuli are sent out in a dorsal direction (PI. 
 
Macropetalichthyids and other Arthrodires — Stensio 103 
 
 XXIV, fig. 2). And, as it is fully evident that these ramuli must have 
 transmitted nerves to the sensory canal organs, the anterior branch 
 (d x ) from the vagus canal is consequently proved to have transmitted 
 a dorsal branch from the n. lineae lateralis. The relations of this 
 dorsal lateralis branch will be treated also in the description of the 
 sensory canal system below. 
 
 Into the proximal half of the vagus canal and, more exactly, into 
 the dorsal part of this just as it leaves the cavum cerebrale, there opens 
 a wide canal {v. lb, Figs. 10, 11 ; PI. XX, XXIV, fig. 2; PI. XXV, 
 figs. 1, 2; PI. XXVI, figs. 1, 2, 3), which comes from the postero- 
 dorso-medial corner of the labyrinth cavity. The function of this canal 
 is not fully clear, but it seems very probable that it transmitted a vein 
 from the labyrinth to the vena cerebralis posterior, because in the 
 Lorcati — the only teleostomous fishes investigated in detail with regard 
 to the veins of the head — there is a small vein which goes from the 
 labyrinth in a similar way back to the v. cerebralis posterior (Allen 
 1905, PI. Ill, figs. 23a, 23, 25). This vein in the Loricati receives 
 blood, as it seems, merely from the ampulla posterior, while the pre- 
 sumed vein in question in M. rapheidolabis , to judge from the large 
 width of the canal v.lb, must have received blood from the larger part 
 or the whole of the membranous labyrinth. 
 
 At the transition between the anterior and posterior divisions of 
 the region, the external bone layer on both the ventral and dorsal side 
 seems to show a certain very faint radiation from a median unpaired 
 centre. Hence, if this observation is true, there would be two unpaired, 
 vestigial centres of ossification in the region, the dorsal one of which 
 would correspond most closely to the centre of a supraoccipital bone, 
 the ventral one to the centre of a basioccipital bone. Better preserved 
 material than that investigated by the author is, however, needed for 
 a positive decision in this case. 
 
 The part of the cavum cerebrale cranii enclosed in the anterior 
 division of the region increases rapidly both in height and breadth 
 forwards. Its shape is well shown by Figs. 10, 11, 12, 13, PI. XX, 
 PI. XXIII, PI. XXIV, fig. 3; PI. XXV, figs. 1, 2 and PI. XXVI, 
 figs. 1, 2. 
 
 Labyrinth Region 
 
 The labyrinth region is very short and broad, its length being con- 
 tained about two and a half times in the maximum breadth. It is also 
 rather low, as its maximum height amounts only to between a fourth or 
 a fifth of the maximum breadth. The maximum height is situated far 
 back, almost at the transition to the occipital region, the maximum 
 
104 Field Museum of Natural History — Geology, Vol. IV. 
 
 breadth, on the other hand, in the anterior half of the region (Figs. I, 
 3, 4, 5; Pis. XIX-XXI, XXIII; PI. XXIV, figs, i, 3; PI. XXV, figs. 
 1, 2; PI. XXVI, figs. 1, 2). As shown by the figures, the region is 
 arched both transversally and longitudinally, so that the dorsal side is 
 convex and the ventral side concave, the arching being, however, 
 stronger in the former than in the latter direction. The antero-lateral 
 corner of each side projects forwards as a large postorbital process, 
 forming not only a posterior but also to a large extent a lateral boundary 
 of the orbit, a disposal which is due to a postero-medial displacement of 
 the orbit. The postorbital process is ventrally, throughout its length, 
 continuous with the extensive orbital floor (Figs. 3, 8). 
 
 For the detailed description we may consider the region to be 
 composed of four walls: a ventral, a dorsal and a paired lateral. 
 
 The ventral wall, if we first turn to it (Figs. 1, 4, 5; Pis. XXI, 
 XXIII; PL XXV, figs. 1, 2; PI. XXVI, figs. 1, 2, 3), is comparatively 
 thin throughout almost its entire extension, but is thinnest in the 
 median parts. At the transition to the occipital region it is fairly 
 narrow, but grows rapidly broader forwards, finally attaining a very 
 considerable width. It is covered both by the internal and external 
 bone layers and perforated by certain canals in the lateral parts at 
 the transition to each lateral wall. These canals will, however, not be 
 described here, but in connection with the lateral walls, which are also 
 perforated by them. 
 
 Fig. 4. Macropetalichthys rapheidolabis 
 
 Transverse section through the posterior part of the labyrinth region. Dermal 
 bones marked with vertical lines. Layers of substitution bone with continuous black 
 lines. Cartilage dotted. XM- 
 
 Mi, L 3 , P 2 , S, bones of the dermal cranial roof (approximate extensions). 
 For their position cf. text fig. 15. c.pp, sensory canal commissure, probably cor- 
 responding to the posterior head line of pit organs in fishes in general; cv, 
 cavum cerebrale cranii; d.end, canal for the ductus endolymphaticus (ventral 
 part) ; d™, the antero-lateral branch of the canal d\ which transmitted a lateralis 
 branch to the above-lying sensory canal ; lab.cav, labyrinth cavity ; Ic, cephalic 
 division of the lateral line; s.ra, groove for the radix aortae (lateral dorsal 
 aortse) ; St, groove leading some distance forward from the external opening of 
 the vagus canal. 
 
Macropetalichthyids and other Arthrodires — Stensio 105 
 
 The dorsal wall of the region (Figs. 3, 4, 5; Pis. XIX, XX, 
 XXIV; PI. XXV, figs. 1, 2; PI. XXVI, figs. 1-3) is broad throughout 
 its extension, its breadth being, however, larger in the anterior than in 
 the posterior half. With the exception of its lateral parts, situated 
 above the labyrinth cavities, the wall must have been rather thick, and, 
 as pointed out already in the general description of the neurocranium, 
 it was covered only by the internal bone layer, so that the dermal bones 
 of the cranial roof rested on its cartilage. The external (dorsal) sur- 
 face must have had rather pronounced grooves for the sensory canal 
 ridges on the lower sides of the dermal bones of the cranial roof, as 
 indicated in text figs. 3, 4 and 5. Far back on the dorsal surface there 
 may, in addition, have been a paired pit (f.end) at the bottom of which 
 the canal for the ductus endolymphaticus (d.end, Fig. 3; PI. XX; 
 PI. XXVI, fig. 3) opened, for this canal does not seem to extend fully 
 up to the ventral side of the overlying dermal bones (Pis. XIX, XX). 
 In other words, there seems to have been a fossa endolymphatica. 
 
 Besides by the canal for the ductus endolymphaticus mentioned, 
 which will be described more in detail below in the account of the 
 labyrinth cavity, the dorsal wall must have been pierced by several 
 fine canals for vessels and nerve branches. Two of these, both for 
 lateralis branches to a portion of a sensory canal, are well seen in the 
 Field Museum specimen (in, Pis. XX, XXII). 
 
 The lateral wall of each side is very thick, especially in its dorsal 
 and anterior parts, and is covered by both the internal and external 
 bone layers (Figs. 1, 3, 4, 5; Pis. XX, XXI, XXIII, XXIV). 
 In its interior it contains the large labyrinth cavity (lab.cav, Fig. 4; 
 PI. XX; PI. XXVI, figs. 3, 4; PI. XXVII), which is completely sep- 
 arated from the cavum cerebrale cranii by a rather thick septum of 
 cartilage, covered by the internal bone layer on the cerebral surface 
 and the labyrinth bone layer on the labyrinth surface. The internal 
 surface of the wall is directed almost straight medially. The external 
 one consists of two fields, an anterior and a posterior one, the latter 
 of which faces ventrally to a marked degree, while the former, which 
 forms the posterior surface of the orbital cavity, faces mainly for- 
 wards. The two fields are separated from each other by a rounded 
 edge running from the antero-dorsal corner of the postorbital process 
 a short distance in a ventral direction. By another, although indistinct 
 dorso-ventral edge the posterior field is in its turn subdivided into an 
 anterior and a posterior part, the former of which is convex both in 
 dorso-ventral and antero-caudal directions and faces almost straight 
 downwards, while the latter, on the contrary, is concave in the same 
 
106 Field Museum of Natural History — Geology, Vol. IV. 
 
 directions and faces much less downwards, but in addition a little 
 backwards (Figs, i, 4, 5; Pis. XXI, XXIII, XXIV). On the 
 latter part we find along the ventral border the anterior continuation 
 of the groove s.ra (Figs. 1, 4, 5; Pis. XXI-XXIV) already re- 
 ferred to above in the description of the lateral wall of the anterior 
 division of the occipital region, and at the transition to the anterior 
 part of the same field this groove deepens and is suddenly transformed 
 into a closed canal (era), which goes forward in the wall. We shall 
 below return to this canal (era) and the groove (s.ra) leading to it 
 and try to make out what their functions were. Before doing this we 
 must, however, examine the canals for cranial nerves that perforate 
 the lateral wall, beginning with the glossopharyngeus canal and pro- 
 ceeding forwards. 
 
 The glossopharyngeus canal (IX, Figs. 1, 5, 7, 10, 11; Pis. XXI- 
 XXIII; PI. XXIV, figs. 1, 2; PI. XXV, figs. 1, 2; PI. XXVI, figs. 
 1, 2, 4; PI. XXVII), which is fairly narrow, has its internal opening 
 situated rather far anterior of the vagus canal and much nearer to 
 the floor than to the roof of the cavum cerebrale cranii. It seems to 
 go almost straight laterally to the posterior lower part of the labyrinth 
 cavity, continuing from this cavity with about the same direction as 
 before to the external surface of the lateral wall, where it opens in the 
 anterior part of the groove s.ra, and accordingly close to the ventral 
 edge of the surface. The external opening is incompletely subdivided 
 into two divisions (Figs. 1, 5; PI. XXIV, fig. 1), an anterior and 
 a posterior one, the latter of which is slightly smaller than the former. 
 
 Somewhat dorsally of the external opening of the glossopharyngeus 
 canal there are traces of another, although finer, canal (d ix , Fig. 1 ; 
 Pis. XXI, XXIII; PI. XXIV, figs. 1, 2; PI. XXVII), which seems to 
 run upwards to a portion of a sensory canal, piercing during its passage 
 the dorsal part of the lateral wall and the overlying lateral part of 
 the dorsal wall. Accordingly, this canal has a course which seems to 
 indicate that it may have contained a lateralis nerve. And as it is situ- 
 ated almost straight above the external opening of the glossopharyngeus 
 canal, we may conclude with a rather high degree of probability that 
 this lateralis nerve arose from fibres which had their passage through 
 the lateral wall, together with the n. glossopharyngeus, as is the case in 
 several other recent and fossil fishes, e.g. Selachii. (Ewart and Cole 
 1895, pp. 475-476; Norris and Hughe 1920, p. 358, Figs. 51, 52, 53), 
 Polypterus (Pollard 1892a, p. 397, PI. 28, fig. 13; Lehn 1918, pp. 
 395-396; Allis 1922a, pp. 283-284), Lepidosteus (Veit 1907, p. 187; 
 191 1, PI. D, figs. 2, 6, 10), Amia (Allis 1897, p. 684), Acipenser, 
 
Macropetalichthyids and other Arthrodires — StensiG 107 
 
 Polyodon (Allis 1920, pp. 138, 142; observations made by the author: 
 cf. also Stensio 1924), Saurichthys (Stensio 1925) and certain 
 Teleostei (Herrick 1901, pp. 207-208, PI. XIV). In other fishes in 
 which the conditions are known, the corresponding lateralis fibres pass 
 through the vagus canal associated with the n. lineae lateralis. 
 
 The possible lateralis branch which has just been dealt with may 
 perhaps be thought to have caused the bipartition of the external open- 
 ing of the glossopharyngeus canal. And, if so, it seems probable that the 
 n. glossopharyngeus proper passed through the larger anterior and the 
 lateralis nerve through the smaller posterior division of the opening. 
 
 The canals for the n. acusticus (VIII) and the n. facialis (VII) 
 begin proximally in a common, distinct recess (afr, Figs. 10, IX J 
 PI. XXIII; PI. XXV, figs. 1, 2; PI. XXVI, fig. 2) on the ventral 
 part of the lateral side of the cavum cerebrale. This recess, which may 
 be properly termed the acustico-facialis recess, communicates in the 
 fossil freely with the cavUm cerebrale, but was perhaps in the fresh 
 specimens separated from this by a membrane belonging to the dura 
 mater. For comparison, it may here be mentioned that a recess of a 
 similar kind occurs in several recent fishes, but that in these it often 
 has a larger extension forward so that not only the acusticus and 
 facialis canals but also the trigeminus canal take their origins from it. 
 On account of this fact it is called in such forms the acustico-trigemino- 
 facialis recess, and it may further be noted that it sometimes is sep- 
 arated there from the cavum cerebrale by a thin lamina of connective 
 tissue or bone, a lamina formed by the dura mater. (Cf. Allis 1909a, 
 pp. 44, 4<>47> i 2 4; i9!4a, pp. 232-236, 239, 240, 243, 246-248; 1914b; 
 1919a; 1922a, p. 228.) 
 
 The acusticus canal (VIII) of M. rapheidolabis, which is rather 
 wide, goes from the posterior part of the acustico-facialis recess almost 
 straight laterally to the labyrinth cavity, piercing the ventral part of 
 the septum that separates the labyrinth cavity and the cavum cerebrale 
 from one another. Its position in relation to the labyrinth cavity is 
 shown by PI. XXVI, fig. 3, in which the septum separating that cavity 
 from the cavum cerebrale, however, has been almost entirely removed. 
 
 The facialis canal (VII), which is of about the same width as the 
 acusticus canal, issues from the anterior part of the acustico facialis 
 recess (Figs. 10, 11 ; PI. XXV, figs. 1, 2; PI. XXVI, figs. 1, 2). On 
 account of the thickness of the lateral wall at this place it gets a con- 
 siderable length. It runs first for some distance forward and slightly 
 laterally close to the antero-ventral part of the labyrinth cavity 
 (Figs. 10, 11 ; PI. XXV, fig. 1 ; PI. XXVI, fig. 3), then at the antero- 
 
108 Field Museum of Natural History — Geology, Vol. IV. 
 
 ventro-medial corner of this cavity it suddenly takes a more lateral and 
 ventral direction (Figs. 10, n; PI. XXV, figs, i, 3; PI. XXVI, 
 figs. 3, 4) and continues with this course to the external opening (text 
 fig. 1; Pis. XXI-XXII; PI. XXIV, fig. 1), close posterior to the 
 antero-ventral corner of the posterior field of the external (lateral) 
 surface of the lateral wall. The^more laterally and ventrally running 
 part is situated immediately anterior of the labyrinth cavity between 
 this and the orbit. As may be understood from PI. XXVI, fig. 3, 
 the part of the lateral wall separating the orbit and labyrinth cavity 
 from one another is rather thin. 
 
 Approximately at the place where it turns more laterally and down- 
 ward, the facialis canal gives, off from its antero-dorsal parts a rather 
 wide branch in an antero-dorsal direction. This branch (c.ophth.lat, 
 Figs. 8, 10, 11; PI. XXIII; PI. XXV, figs. 1-3) soon enters the 
 orbitotemporal region, ascending there after a short distance to the 
 dorsal wall and continuing forward into the ethmoidal region rather 
 close below the upper surface of this. In almost the whole orbito- 
 temporal region and throughout the ethmoidal region it goes rather 
 straight beneath the supra-orbital sensory canal to which fine rami issue 
 from it {rm.ophth.lat, Pis, XXI, XXIII), a fact which fully shows 
 that it lodged the n. ophthalmicus lateralis. It may, therefore, properly 
 be called the canalis n. ophthalmici lateralis. In the subsequent descrip- 
 tion we shall return to it and its importance for the understanding of 
 the trigemino- facialis ganglionic complex. 
 
 Not far distally of the canalis n. ophthalmici lateralis, a second 
 branch issues from the facialis canal, but on the antero-ventral side. 
 This branch {c.bucc.lat, Figs. 10, 11), which has almost the same 
 calibre as the canalis n. ophthalmici lateralis, goes forward and 
 opens into the trigeminus canal, and, as the trigeminus and facialis 
 canals lie close to each other at this place, it is rather short. As will 
 be evident from my subsequent account of the trigemino-facialis gan- 
 glionic complex, the canal transmitted certain lateralis fibres, viz., those 
 that formed the n. buccalis lateralis. 
 
 A third, wide, but short branch (c.pal 1 , Figs. 10, 11) is given 
 off from the facialis canal close to the external opening. This branch 
 runs forward to the postero-ventro-lateral corner of the orbit, and, 
 as will be shown below, the conditions are such that it must have been 
 pierced by the r. palatinus facialis. 
 
 It has already been pointed out above, that the fine but pronounced 
 
 1 This branch was discovered by preparation after the photographs had been 
 taken and is therefore not shown in the plates. 
 
Macropetalichthyids and other Arthrodires — Stensio 109 
 
 groove s.ra, along the ventral border of the external (lateral) surface 
 of each lateral wall deepens and is transformed into a canal a short 
 distance anterior to the external opening of the glossopharyngeus canal, 
 the canal arising in this way {era, Figs. 1, 5, 7; Pis. XXI-XXIII) 
 being fairly narrow. Not far dorsally of this canal is found the 
 posterior opening (ju 2 ) of another much wider canal (ju, Figs. 1, 
 5. 7; Pis. XX-XXIII). These two canals, which we, after their index 
 letters in the previous figures, may refer to as era, and ju respectively, 
 go forward within the lateral part of the lateral wall, where they soon 
 meet and join to a wide sinus (si, text fig. 7; Pis. XXI-XXIII). In 
 this sinus there may be distinguished a wide, dorsal division correspond- 
 ing to the canal ju and a narrow, ventral division corresponding to the 
 canal era. Although very imperfectly separated from one another, 
 these divisions indicate that the structures traversing the two canals ju 
 and era, retained their independence and original positions in the 
 sinus, si. 
 
 The ventral division of the sinus si, is continued forward by a 
 canal which is lettered ecom in the figures (Fig. 7; Pis. XXI, 
 XXIII) and which runs in an anterior and slightly medial direction 
 at the transition between the lateral and ventral walls of the region. 
 It is also noteworthy that this canal (ecom) has a very superficial 
 position, as it is situated wholly within the external bone layer. After 
 a short course forward it divides into two branches of about equal 
 calibre, a lateral one (ecar.ext) and a medial one (ecar.int). The 
 former of these runs latero-dorsally through the lateral wall and opens 
 close to and postero-ventrally of the facialis canal (Figs. 1, 7; 
 Pis. XXI, XXIII), while the latter takes a course forward and slightly 
 medially to the orbitotemporal region. The former of these branches 
 is throughout its extension situated within the external bone layer 
 (Fig. 8), as is also the latter till it reaches the anterior end of 
 the orbitotemporal region (Fig. 8), where it suddenly leaves the 
 external bone layer and ascends into the cartilaginous interior of the 
 ventral wall in the manner that will be described below in the account 
 of the orbitotemporal region. As we shall find from its course, the 
 latter branch (ecar.int) must have lodged the arteria carotis interna 
 and is therefore termed the internal carotid canal. This being the 
 case, it is easy to understand that the other branch (ecar.ext) must 
 have been traversed by the arteria carotis externa. (Cf. Allis 1897, 
 PP- 497-50O; 1908a; 1908b; 1909a, pp. 51-55, 185-187; 1909b; 1911a; 
 1911b; 1912a, b, c, d; 1914a; 1914b; 1919a, 1922a, pp. 266-268; 1922b; 
 Allen 1905, pp. 51-62; Danforth 1912, pp. 435-445; Greil, 1913, 
 
1 10 Field Museum of Natural History — Geology, Vol. IV. 
 
 PI. LV, fig. 2.) Hence this may properly be termed the external 
 carotid canal. 
 
 The branches of the canal c.com, having had these functions, it is 
 evident that the canal c.com itself must have lodged the arteria carotis 
 communis. From the canal c.com, the arteria carotis communis may 
 have continued a short distance backwards into the ventral division of 
 the sinus si, there probably, as we shall see, having received the arteria 
 efferens hyoidea and having become the radix aortae (lateral dorsal 
 aorta). The radix aortae must then have traversed the remaining pos- 
 terior part of the ventral division of the sinus si behind this, continuing 
 through the canal era, and the groove s.ra, to the anterior end of the 
 haemal groove (hczm). At the anterior end of this it probably met its 
 fellow of the opposite side and joined with it to the unpaired aorta 
 dorsalis, which in its backward course occupied the haemal groove. 
 The efferent arteries from the two or three anterior branchial arches 
 may probably have emptied into the radix aortae of their side, while 
 those from the remaining posterior branchial arches probably opened 
 into the anterior portion of the aorta dorsalis. My view concerning 
 the course and mutual relations of the arterial trunks mentioned is 
 further elucidated by the diagrammatic sketch reproduced in Fig. 6, 
 and as is evident from this the arterial system of the head seems to 
 
 fend 
 
 c.ophth.sup.V? 
 
 cr.'ol 
 
 V 
 
 ju-i c.hy IX S.ra, 
 
 Fig. 5. Macropetalichthys rapheidolabis 
 
 Primordial neurocranium in lateral view. Occipital region after specimens 
 in the American Museum, New York, other regions after a specimen in Field 
 Museum, Chicago. Bone dark, cartilage dotted. XH- 
 
 a.opt, division of the distal part of the opticus canal, probably for the arteria 
 optica; c.hy, canal for the v. hyoidea; c.ophth.sup.V ?, canal probably for the 
 r. ophthalmicus superficialis trigemini ; cr.od, crista occipitalis dorsalis ; cr.ol, 
 crista occipitalis lateralis ; cr.sp, cranio-spinal process ; /wi, anterior opening of 
 the canal ja. The opening was traversed by the vena mandiburalis ; jui, posterior 
 opening of the canal ju; na, nasal aperture; orb, orbit; s.ra, groove for the radix 
 aortae ; vy, canal of doubtful importance ; either for vessel or nerves or for 
 both ; II, opticus canal ; IX, glossopharyngeus canal ; Xvn, canal for the n. vagus 
 and the vena cerebralis posterior. 
 
 
Macropetalichthyids and other Arthrodires — Stensio III 
 
 resemble mostly that in Chimcera (Allis 1912a). The arteria carotis 
 interna will be further treated below in the description of the orbito- 
 temporal region. 
 
 Turning again to the canal ju, we find that anterior to the sinus si 
 (Fig. 7; Pis. XXI, XXIII) it takes a more lateral direction than 
 in its posterior part. It opens with its anterior opening (juy) on the 
 
 L-— <z.pal 
 
 cuc.int 
 
 
 a.coru F^5f \ 
 
 
 
 
 ..-a.effhy 
 
 
 " a " rt " ]^jfeS?N - ' V J U 
 
 —v.hy 
 
 rSk\^z^ 
 
 \ J$^<w\ ueffI 
 
 
 i& ^p 1 \ 
 
 1 1 
 
 v.oerebr.jjost ~ 
 
 
 • 
 
 u-dtors 
 Fig. 6. Macropetalichthys rapheidolabis 
 
 Diagrammatic sketch of the main, arterial and venous trunks of the primordial 
 neurocranium. Outlines of neurocranium in ventral view. Arteries red, veins 
 blue. The parts of the different vessels situated in canals or cavities indicated 
 by transversal bands of color. X/^. 
 
 a.com, arteria carotis communis; a.c.ext, arteria carotis externa; a.c.int, arteria 
 carotis interna; a.eff.hy, arteria efferens hyoidea; a.effJ-a.eff.V, efferent arteries 
 of the branchial arches I-V ; a.dors, aorta dorsalis ; a.pal, possible palatine branch 
 of the a. carotis interna ; r.aort, radix aortae ; v.cerebr.post, vena cerebralis pos- 
 terior; v.hy, vena hyoidea; v.ju, vena jugularis; v.m, vena mandibularis (from 
 the dorsal parts of the mandibular arch to the jugular vein). 
 
ii2 Field Museum of Natural History — Geology, Vol. IV. 
 
 external (lateral) surface of the lateral wall rather close laterally to 
 the facialis canal. With the part situated anterior of the sinus si, it 
 joins on the medial side a fairly wide canal jg, which comes from the 
 postero-latero-ventral corner of the orbit, perforating in its course back- 
 wards the ventral portion of the postorbital process. This canal, jg, 
 does not, however, run straight^backwards but postero latero-ventrally 
 and crosses in its course the distal part of the facialis canal on the 
 dorsal side. Its position may be understood from PI. XXI, PI. XXII, 
 PI. XXV, figs, i, 3, and Figs. 6, 7. (In PI. XXIV is seen a section of it 
 just after it has crossed the facialis canal; in PI. XXV, fig. 1, it is 
 seen crossing the facialis canal on the dorsal side, and it is also obvious 
 that the bone layer lining it is continuous proximally with the part of 
 the external bone layer that covers the posterior surface of the orbit; 
 PI. XXV, fig. 3, shows the same part of it as the preceding figure but 
 from behind.) 
 
 From the conditions in recent fishes it is not difficult to conclude 
 what the functions of the canals ju and jg were. Both must have been 
 traversed by veins. The vein that traversed the canal jg evidently col- 
 lected the blood from the eye bulb, the eye muscles and probably also 
 from the anterior parts of the cavum cerebrale cranii. Hence it must 
 have corresponded to the vena jugularis interna according to the 
 terminology mostly employed hitherto. Allis has, however, recently 
 proposed to call this vein simply the jugular vein, a term that I shall 
 adopt here as I have done in my other works. (Cf. Allen 1905, pp. 
 78-90, PL I, figs. 1, 5; PL II, figs. 13, 15; PL III, figs. 23-25; Allis 
 1897, pp. 500-506; 1903, p. 93; 1908b, pp. 219-222; 1909a, pp. 40, 
 45, 50, 187, etc.; 1914a; 1914b; 1919a, p. 222, and in several other 
 places throughout the paper; 1922a, pp. 260-264; 1922b, p. 153; Greil 
 1913, PI. 54, fig. 1 ; PI. 55, figs. 2, 8 (vena capitis lateralis) ; Grosser 
 1907; Lehn 1918, pp. 365-372; O'Donoghue 1914, pp. 438-443, an- 
 terior cardinal sinus -f- postorbital sinus; Parker, T. J., 1887, pp. 711- 
 713; Stensio 1921, p. 178; 1923, pp. 1244, 1260-1268; 1925.) 
 
 A vein of a considerable size must have entered the canal ju through 
 the anterior opening ju ± . As far as can be judged, this vein collected 
 the blood from the palatoquadrate and perhaps also from certain parts 
 of the ethmoidal region; in other words, it seems to have drained a 
 large dorsal part of the mandibular arch and nearest surroundings. 
 Hence it may be properly termed the vena mandibularis. 
 
 The vena mandibularis of the fish so defined, after a short passage 
 backwards within the canal ju, met and joined the jugular vein to a 
 common large stem which then passed backwards through the dorsal 
 
 
Macropetalichthyids and other Arthrodires — Stensio 113 
 
 division of the sinus si and the posterior part of the canal ju behind 
 this sinus, then continuing along the external surface of the lateral 
 wall of the neurocranium received during this latter part of its course 
 the vena cerebralis posterior as it passed the external opening of the 
 vagus canal. Despite the fact that the vena mandibularis was much 
 larger than the vena jugularis, the common stem described, that arose 
 by its confluence with the latter, must for morphological reasons be 
 considered as the backward continuation of this, and hence throughout 
 its extension it must retain the name of jugular vein. The course and 
 relations of the juglar vein, the mandibular vein and the vena cerebralis 
 posterior seem to have been as indicated in the diagrammatic sketch 
 in Fig. 6. 
 
 In the dorsal division of the sinus si — thus, after what we have 
 just found, the division for the jugular vein — there opens on the lateral 
 side a short but rather wide canal c.hy (Figs. 1, 5, 7; Pis. XX- 
 XXIII; Pi. XXIV, figs. 1, 3), which leads from the external surface 
 of the lateral wall of the region, or, to define it more closely, from the 
 postero-dorsal corner of the anterior part of the posterior field of the 
 lateral surface of the region. From its course and position we may 
 conclude with a rather large degree of probability that it was traversed 
 by a vein. This supposed vein must have come from the hyoid arch 
 and emptied into the jugular vein, thus corresponding partly to the 
 sinus hyoideus of Selachians. I call it here the vena hyoidea. 
 
 In Selachians the sinus hyoideus is a large vein which runs upward 
 on the posterior side of the hyoid arch, receiving blood not only from 
 this arch, but also from the dorsal and middle parts of the mandibular 
 arch (Parker, T. J., 1887, PI. 35, fig. 10; O'Donoghue 1914, pp. 441- 
 442), a condition which explains the fact that a mandibular vein is 
 lacking. In Polypterus a hyoid vein seems also to occur (Allis 1922a, 
 p. 264), but in Ceratodus (Greil 1913, Pis. 54, 55) and the Loricati 
 (Allen 1905, pp. 79-80), the dorsal parts of this vein seem to be 
 lacking. At least in the Loricati the venous blood of the hyoid arch 
 goes chiefly forward to the mandibular vein, arriving through this into 
 the vena jugularis. In Ceratodus the conditions in this respect are not 
 known with certainty. The venous system of other fishes besides those 
 mentioned, is not known in detail and accordingly nothing can be said 
 of the veins in the mandibular and hyoid arches there. 
 
 The occurrence of both the vena mandibularis and the vena hyoidea 
 in Macropetalichthys rapheidolabis is, it seems to me, to be considered 
 a primitive character, since we may presume that the venous system 
 of fishes originally had a segmental arrangement. 
 
ii4 Field Museum of Natural History — Geology, Vol. IV. 
 
 If the interpretation given here of the canal c.hy of Macropetalich- 
 thys rapheidolabis is correct, and if the vena hyoidea during the passage 
 to this canal went approximately as it does in Selachians, it is evident 
 that the dorsal end of the hyoid arch must have articulated against the 
 lateral wall of the region anterior to the external opening of the canal 
 c.hy. What this means for views concerning the homologies of 
 the hyomandibular in Selachians and Teleostomes will be dealt with 
 below in the account of the visceral skeleton. It has been mentioned 
 above that the arteria efferens hyoidea ought to have persisted in 
 M. rapheidolabis, and if so, it probably, as in recent fishes in which 
 it is found, went upwards along the posterior side of the hyoid 
 arch and emptied into the anterior part of the radix aortae, which, as 
 we have seen, was situated here within the sinus si. After leaving the 
 hyoid arch, the a. efferens in order to reach this sinus must have 
 entered the external opening of the canal c.hy, as no other foramen 
 can be found for it. Perhaps it did not, however, traverse the entire 
 canal c.hy, but took near the proximal end of this a more anterior 
 course, entering the rather fine canal c.ef (Fig. 7; Pis. XX, XXII), 
 which, at least on the left side of the Field Museum specimen, runs 
 from the proximal part of the canal c.hy forward, medially and down- 
 ward to the anterior end of the sinus si. If well arrived into the sinus 
 si in this way, the artery must have crossed the jugular vein and passed 
 downwards to the ventral division of the sinus, joining there the radix 
 aortae at the transition to the arteria carotis communis. It may, how- 
 ever, be possible that the canal c.ef, instead of being traversed by the 
 arteria efferens hyoidea, as here supposed, transmitted the truncus 
 hyoideomandibularis facialis, and if this was the case, the a. efferens 
 hyoidea must of course have traversed the entire canal c.hy, and have 
 entered the sinus si much farther back. (Allen 1905, pp. 44-62 ; Allis 
 1908a, b; 1909a, pp. 51-52, 183-184; 1909b; 1911a, b; 1912a, b, c, d; 
 1922a, pp. 264-268; Danforth 1912, pp. 435-445.) 
 
 In most fishes in which there is a canal for the jugular vein in 
 the lateral wall of the neurocranium, the jugular vein is generally 
 during its passage through this canal accompanied for some distance 
 by the truncus hyoideomandibularis, which enters the canal either from 
 in front through the anterior opening or from the medial side. In 
 the former case the truncus hyoidemandibularis has its exit through 
 the cranial wall anterior to, in the latter case into the canal for the 
 jugular vein. (Cf. Allis 1897, pp. 492-497; 1903, pp. 87-95, 2 ^6; 
 1908b, pp. 219-222; 1909a, pp. 44-45; 1911a, p. 291; 1914a; 1914b; 
 
Macropetalichthyids and other Arthrodires — StensiS 115 
 
 19x9a; 1922a, pp. 263-264; 1922b; Lehn 1918, pp. 363-372; Pollard 
 1892a, PI. 30, figs. 27-30; Stensio 1921, p. 178; 1923, pp. 1244, 1261, 
 1262-1268; 1925; etc.) 
 
 Concerning M. rapheidolabis , we cannot say at present whether the 
 truncus hyoidemandibularis after its exit from the facialis canal defin- 
 itely left the cranial wall or whether it again entered the cranial wall 
 accompanying the vena jugularis for some distance. If the former of 
 these alternatives were true, the truncus hyoidemandibularis would have 
 left the cranial wall rather far anterior of the hyoid arch and would 
 thus have reached this from in front in the same way as in Selachians. 
 (Cf. Ruge 1897, Figs. 3, 12, 13, 14; Goodrich 1909, Fig. 44; Allis 
 1915; 1918b; cf. also van Withe 1882, PI. 15, figs. 1, 3, 4, 6; etc.) 
 If, on the other hand, the latter were true, it is at once evident that the 
 truncus hyoideomandibularis cannot have entered the anterior opening 
 of the canal jg for the jugular vein, (as we have found, this opening 
 is situated in the orbit, while the external opening of the facialis canal 
 lies on the lateral surface of the lateral wall behind the orbit), but 
 would, together with the vena mandibularis, have passed through the 
 anterior opening (/*j) of the canal ju. In other words, the truncus 
 hyoideomandibularis would, with this alternative, have reached the 
 vena jugularis in the anterior part of the canal ju, after traversing the 
 part of this canal situated anterior to the canal jg. 
 
 Assuming that it did so, it may during its course further back- 
 ward either have traversed the entire dorsal division of the sinus si 
 and behind this the posterior part of the canal ju, or may soon after its 
 entrance in the sinus si have turned latero-dorsally, entering and 
 traversing the canal c.ef and the distal portion of the canal c.hy. If 
 it took the former of these courses it would have left the cranial wall 
 considerably behind its arch, the hyoid arch, which, as we have seen, 
 articulated against the lateral wall of the region anterior to the external 
 opening of the canal c.hy. It would thus have gone a considerably 
 roundabout way to its destination, which seems rather improbable. If, 
 on the contrary, it passed from the sinus si through the canal c.ef and 
 the canal c.hy, it would, of course, have had its exit from the cranial 
 wall in closer relation to its arch, although still behind this. 
 
 The conditions on the external surface of the lateral wall of the 
 region between and around the external opening of the facialis canal 
 and the anterior opening of the canal ju, would, perhaps, if they were 
 known in detail, enable us to decide with considerable degree of 
 probability what course the truncus hyoideomandibularis took after 
 
n6 Field Museum of Natural History — Geology, Vol. IV. 
 
 leaving the facialis canal 1 . But, unfortunately, the external surface of 
 the wall is not sufficiently well preserved to enable me to determine 
 this in any of the specimens so far investigated by me. 
 
 In dealing with the trigeminus and facialis canals below we shall 
 see that the r. oticus lateralis must have originated in the proxifnal part 
 of the facialis canal, which it ""left either through the external opening, 
 or through a canal of its own. In the latter case it would have issued 
 from the dorsal side of the distal part of the facialis canal. A few 
 fine canals, which were traversed by rami from the r. oticus lateralis to a 
 sensory canal portion in the dermal cranial roof are seen in the Field 
 Museum specimen (in, Pis. XX, XXII). These fine canals could, 
 however, not be traced downwards to their origin, and, therefore, we 
 cannot say whether they come from a larger canal within the cranial 
 wall or whether each of them went independently down and opened on 
 the lateral surface of the lateral wall. 
 
 Labyrinth Cavity 
 
 The labyrinth cavity is completely preserved only in the Field 
 Museum specimen. Traces of it are, however, seen also in one of the 
 New York specimens (PL XXVII). It (lab.cav, Fig. 4; PL XX; PL 
 XXVI, figs. 3, 4; PI. XXVII) is large and, as pointed out, lined 
 throughout by a thin layer of perichondral bone, the labyrinth bone 
 layer. As has also been mentioned, it is completely separated from the 
 cavum cerebrale cranii by a thick septum (PL XX) which consisted 
 of cartilage lined by the labyrinth bone layer on the labyrinth side and 
 by the internal bone layer on the cerebral side. 
 
 The division for the sacculus (sac, PL XXVI, fig. 3) is not well 
 exposed, but, as far as can be judged, it cannot have been very large. 
 It is not distinctly separated from the remaining divisions of the 
 cavity. From its dorsal part a canal (d.end, Fig. 4; PL XX; 
 PL XXVI, fig. 3) issues in a dorso-postero-lateral direction to the 
 dorsal surface of the primordial neurocranium. This canal seems, 
 however, to end some distance below the dermal bones of the cranial 
 roof, a fact which probably indicates that the cartilage at this place 
 did not reach quite up to the dermal bones, i.e., that there probably 
 was a shallow fossa into which the canal had its dorsal opening 
 
 1 The presence of a groove between the two openings would of course have 
 indicated that the r. hyoideomandibularis after its exit from the facialis canal 
 went backward along the lateral surface and entered the canal ju through the 
 anterior opening j'ih. The absence of such a groove would of course point in 
 the opposite direction. If the two openings should be found situated in a com- 
 mon pit this must probably also indicate that the tr. hyoideomandibularis entered 
 the canal ju through the anterior opening jui. 
 
 
Macropetalichthyids and other Arthrodires — Stensio 117 
 
 (Fig. 3; Pis. XIX, XX). From its course and relations it is evi- 
 dent that the canal in question must have lodged the ductus endo- 
 lymphaticus, which, accordingly, opened on the dorsal surface of the 
 primordial neurocranium in a fossa endolymphatica as in Selachians 
 (Retzius 1881). 
 
 Slightly postero-ventrally of the canal for the ductus endolym- 
 phaticus, another, but much wider canal, issued from the division for 
 the sacciilus. This canal (v.lb, Figs. 10, 11 ; PI. XX; PI. XXIV, 
 fig. 2; PI. XXV, figs. 1, 2; PI. XXVI, figs. 1, 2, 3), which runs back- 
 ward and somewhat dorsally to the dorsal division of the vagus canal, 
 has already been described in another connection, and, as pointed out 
 there, it probably transmitted a large vein, that drained the whole, or 
 at least the main part, of the labyrinth. 
 
 The division for the utriculus (utr, PI. XX; PI. XXVI, figs. 3, 4) 
 appears to be relatively very large, but as it is not distinctly marked 
 off from the other divisions, its size may, perhaps, be 'over-estimated. 
 As is seen from the plates, its largest extension is from above down- 
 
 VII c.car.ext 
 
 19 \ ! 
 
 Fig. 7. Macropetalichthys rapheidolabis 
 
 Diagrammatic sketch showing the position of certain canals in the lateral 
 wall of the labyrinth region. Certain ventral parts of the lateral wall removed. 
 Bone layers indicated with shading and when in cross section with black lines. 
 Cartilage dotted. XH- 
 
 c.com, canal for the arteria carotis communis ; c.ef, canal perhaps for the 
 arteria efferens hyoidea ; c.hy, canal for the vena hyoidea ; era, canal for the 
 radix aortse behind the sinus si: jg, canal for the vena jugularis through the 
 postorbital process; ju, canal behind jg, for the vena jugularis, anterior of jg, 
 for the vena mandibularis jui, pit, anterior and posterior openings of the canal ju ; 
 si, sinus arisen by confluence of the canals ju and era; s.ra, groove for the 
 radix aorta (lateral dorsal aorta); VII, canal for the n. facialis; IX, canal for 
 the n. glossopharyngeus. 
 
n8 Field Museum of Natural History — Geology, Vol. IV. 
 
 wards and forwards, which must mean that the utriculus did not have 
 a horizontal position, but was inclined in the same way as in Selachians 
 (cf. Retzius 1881). Just at the antero-ventral end, there is, immedi- 
 ately anterior of the place of entrance of the acusticus canal (VIII) 
 a rather pronounced bulge (ru, PI. XXVI, fig. 3), which, to judge 
 from its position, probably contained the recessus utriculi. Immedi- 
 ately in front of this bulge lies the proximal portion of the facialis 
 canal (VII). 
 
 The divisions for the semicircular canals are very imperfectly 
 known, which is due partly to the fact that they could not be laid 
 bare without danger to other important structures, and partly to the 
 circumstance that they probably are not well differentiated from the 
 other parts of the labyrinth cavity. It seems, however, not unlikely 
 that the part designated with the letters c.sem.ant, in PI. XX and PI. 
 XXVI, figs. 3, 4, was occupied by the canalis semicircularis anterior, 
 and if this is true the canalis semicircularis anterior would have had 
 a position almost transversal to the longitudinal axis of the head. This 
 position would, as is easily understood, have been caused by the back- 
 ward shifting of the orbit. What parts of the divisions for the other 
 two semicircular canals are represented by the fragments denoted by 
 c.sem in PI. XXI and PI. XXVI, fig. 4, is difficult to say with cer- 
 tainty, but at least what is seen in PI. XXI seems most probably to 
 be a part of the division for the canalis semicircularis externus, which 
 thus would have been situated as in Elasmobranchs. 
 
 The n. acusticus, as is seen from the position of its canal (VIII), 
 enters the labyrinth cavity at the very bottom, as in certain Elasmo- 
 branchs (Scyllicum, Raja, cf. Retzius 1881), and is rather short. 
 The glossopharyngeus, as has already been mentioned, traversed the 
 postero-ventral parts of the labyrinth cavity. The facialis went first 
 for some distance along the antero-ventro-medial part of the labyrinth 
 cavity, then turned laterally and passed along the antero-ventral part 
 of the same cavity between it and the orbit. (PI. XXV, figs. 1, 3; 
 PI. XXVI, figs. 3, 4.) 
 
 From the description now given of the labyrinth cavity it seems 
 rather probable that this was larger than the membranous labyrinth en- 
 closed in it, a condition which makes it difficult to conclude anything 
 with certainty about the finer details of the latter. But what is, how- 
 ever, known so far of the larger divisions of the labyrinth cavity, 
 indicates beyond doubt, as we have seen, that the membranous labyrinth, 
 at least in certain of its general characters, was of the Elasmobranchian 
 type. 
 
Macropetalichthyids and other Arthrodires — Stensio 119 
 
 A more detailed knowledge of the labyrinth cavity would seem evi- 
 dently to be of great importance for the understanding of the affinites 
 of Macropetalichthys to other fishes, and it is therefore to be hoped 
 that new material, fit for a detailed investigation of the labyrinth 
 cavity, will soon be found. 
 
 Orbitotemporal Region 
 
 The orbitotemporal region (Figs. 1, 3, 5, 8; Pis. XIX-XXIII; PI. 
 XXIV, fig. 1 ; PI. XXV, figs. 1, 2; PI. XXVI, figs. 1, 2) is strikingly 
 short and broad, its breadth being almost three times as great as its 
 length. The breadth is, however, much greater at the bottom of the re- 
 gion than higher up, a condition which is caused by the presence of the 
 very extensive floors for the orbits (Figs. 3, 8; Pis. XIX, XXI, XXIII). 
 The height is greatest at the posterior end of the region, where it is 
 about equal to the length and a third of the breadth, decreasing slowly 
 and gradually forwards throughout the region. On account of the orbital 
 floors, the region is, at the bottom, not marked off from the adjacent 
 regions, otherwise being, however, considerably narrower than these. 
 
 The orbitae are rather small and lie far apart. While in Arthrodires 
 in general they are situated fairly far forward, and have the entrances 
 facing almost straight laterally, they have here shifted a certain dis- 
 tance in a postero-medial direction and have their entrances directed 
 much upwards (Figs. 3, 5, 8; Pis. XIX, XX; PI. XXIV, fig. 1). 
 On account of the shifting, they are much more completely bounded 
 by the primordial neurocranium than is generally the case in fishes, 
 for, in fact, they have not only a medial, a posterior and an anterior, 
 but, in addition, a ventral and to a large extent a lateral wall formed 
 by this. The ventral wall, which has been already referred to as the 
 orbital floor, is a rather thin plate which is continuous laterally and 
 posteriorly with the postorbital process, medially with the ventral part 
 of the interorbital wall and anteriorly with the ethmoidal region. It 
 consists of cartilage covered both on the upper (orbital) and lower 
 surfaces by the external bone layer. The lateral wall, as far as it is 
 present, is formed by the large postorbital process, which reaches so 
 far forward that it leaves only a rather limited opening between its 
 anterior end and the ethmoidal region (Figs. 3, 5; Pis. XIX, XX; 
 PI. XXIV, fig. 1). A transverse section through the posterior half of 
 the region that shows the relations between the different walls of the 
 orbit at this place as well as the shape of the orbit itself is seen in 
 Figure 8. The orbital entrance is of an elliptical shape, with the 
 
120 Field Museum of Natural History — Geology, Vol. IV. 
 
 largest axis directed so as to converge a little towards that of the other 
 side in a postero-dorso-medial direction (Fig. 3; Pis. XIX, XX; 
 PI. XXIV, fig. 1). 
 
 For detailed description we may consider the region to be com- 
 posed of four walls — a ventral, a, dorsal, and a right and a left lateral 
 — all of which, of course, are continuous with the correspondingly 
 situated walls in the adjacent regions. 
 
 The ventral wall (Figs. 1, 5, 8; Pis. XXI, XXII, XXIII; PI. 
 XXV, figs. 1, 2; PI. XXVI, figs. 1, 2) is characterized in the first 
 place by its very considerable breadth, which remains almost unchanged 
 throughout the region. As a whole it is thin, and especially is this 
 the case in the lateral (orbital floors) and median parts. Like the 
 corresponding walls in the adjacent regions, it is covered by the inter- 
 nal bone layer dorsally and the external bone layer ventrally and is 
 curved so that the ventral surface is concave in transversal direction. 
 It has no fenestra hypophyseus or fenestration whatever and the fossa 
 hypophyseus is only very slightly developed on its dorsal (cerebral) 
 side. 
 
 The canal for the internal carotid artery of each side (c.car.int, 
 Fig. 8; Pis. XXI-XXIII; PI. XXV, fig. 2) enters the ventral wall 
 of the region from behind and perforates the lateral part of this that 
 forms the orbital floor. Its direction is forward and a little medially 
 and, in the posterior larger part of its course, it is here, as in the 
 labyrinth region, situated entirely within the external bone layer, thus 
 very superficially. In the anterior part of the region it penetrates 
 deeper into the wall, curving postero-dorso-medially. With this latter 
 course it arrives up and into the lateral wall of the region, where it 
 finally opens into the opticus canal (PI. XXIII; PI. XXV, fig. 2), its 
 opening into this being situated on the anterior side close to the orbit. 
 Just at the place where it curves from the anterior to the postero- 
 dorso-medial direction, it lies somewhat anterior of the orbit, thus 
 really in the basal part of the antorbital process, a fact which is cer- 
 tainly explained by the backward shifting of the orbits. We shall have 
 the opportunity of returning to it again in connection with the de- 
 scription of the opticus canal. 
 
 The orbital floor of each side is perforated by a second canal (c.pal 2 , 
 Pis. XXI, XXII) which begins in the orbit and runs antero-ventrally 
 through the floor, opening into the anterior portion of the suborbital 
 part of the internal carotid canal. At the place where the latter canal 
 curves dorso-postero-medially it has an opening (c.pal 3 , Fig. 1), 
 leading to the ventral surface of the neurocranium, and from this 
 
Macropetalichthyids and other Arthrodires — Stensio 121 
 
 opening a shallow groove (s. pal, Fig. i) goes forward. As we 
 shall see from the description of the trigemino-facialis ganglionic 
 complex below, it is not difficult to conclude that the canal c.pal 2 was 
 traversed by the r. palatinus facialis, which passed down to the anterior 
 portion of the suborbital part of the internal carotid canal, then left 
 this and arrived at the ventral side of the neurocranium by the open- 
 ing c.pd 3 . 
 
 The dorsal wall (Figs. 3, 8; Pis. XIX, XX; PI. XXIV, fig. 1; 
 PI. XXV, figs. 1, 2; PI. XXVI, figs. 1, 2) is broad and short like the 
 ventral one, but on the other hand considerably thicker. As far as one 
 can judge from its appearance in the fossil it had no fontanelle, and, 
 as already ^pointed out, it is devoid of the external bone layer. Its 
 upper (external) surface is slightly convex in transversal direction 
 and provided with grooves for the ridges on the lower side of the 
 dermal bones in which the sensory canals are enclosed. In its anterior 
 part it is, as already mentioned above, traversed by the canal for the 
 n. ophthalmicus lateralis, which arrives there from below from the 
 lateral wall. Otherwise there are in it only several very fine canals, 
 the most important of which issue from the canal for the n.ophthalmi- 
 cus lateralis and pass upwards to the supraorbital sensory canal. 
 
 Finally, the lateral walls (Figs. 1, 5, 8; Pis. XXII-XXIV; PI. 
 XXV, figs. 1, 2; PI. XXVI, figs. 1, 2), although thick, are by no means 
 as thick as those in the labyrinth region. Their thinnest place is situated 
 around the opticus canal (II). They are covered both by the internal 
 and external bone layers. Their external surface is concave both in 
 dorso-ventral and antero-caudal direction. Their internal surface has 
 a rather complicated appearance, as may be understood, from the figures 
 quoted and as will be further dealt with below. 
 
 Each lateral wall is perforated by several canals. We begin here 
 with the most posterior one, the trigeminus canal, proceeding forward 
 from this. 
 
 The trigeminus canal (V, V 2 , 3 , Figs. 10, 11; PI. XXIII; PI. 
 XXV, figs. 1, 2; PI. XXVI, figs. 1, 2) leaves the cavum cerebrale 
 cranii a short distance anterior to the facialis canal and, as pointed out 
 above, its internal opening has nothing to do with the acustico- facialis 
 recess. It runs antero-laterally and somewhat ventrally to the postero- 
 ventro-medial corner of the orbit, and, on account of the fact that the 
 cranial wall is much thinner at this place than immediately behind the 
 orbit, it naturally becomes considerably shorter than the facialis canal. 
 The external opening, as is easily understood, is situated rather far 
 medially of that of the facialis canal. 
 
122 Field Museum of Natural History — Geology, Vol. IV. 
 
 In the trigeminus canal there may properly be distinguished two 
 parts, a proximal short and wide one (V) and a distal rather long 
 and narrower one (V 2 , 3 ), the former of which actually forms a recess 
 on the lateral side of the cranial wall, a recess which may be called 
 the trigeminus recess. 
 
 From the antero-dorsal part of the trigeminus recess, as thus de- 
 fined, there issues a rather wide canal (c.ophth.sup.V) which runs for- 
 ward and a little upward, crossing the canal for the n. ophthalmicus 
 lateralis (c.ophth.lat) on the ventro-medial side and opening into the 
 postero-ventro-medial part of the orbit (PI. XXIII; PI. XXV, fig. 2; 
 PI. XXVI, figs. 1, 2). Slightly ventrally of this canal a second, but 
 considerably narrower one, (c.ophth.prof), leaves the trigeminus re- 
 cess, and taking its course ventro-medially of the canal for the n. oph- 
 thalmicus lateralis, goes almost straight forward to the orbit, where it 
 opens somewhat postero-ventrally of the former. 
 
 With the knowledge we now possess of the conditions in recent 
 fishes, it is easy to conclude what the functions of the canals just de- 
 scribed were, and, in addition, to make out to a certain degree how 
 the trigemino-facialis ganglionic complex and its roots were developed. 
 For comparison the reader may be referred to the works of Herrick 
 (1899; 1900; 1901), Johnston (1898; 1901), Kingsbury (1897), 
 Allis (1897; 1901 ; 1903; 1909a; 1918c; 1922), Landacre (1916), 
 Norris and Hughe (1920) and Strong (1895), etc. 
 
 As mentioned in another connection, it is fully evident that the 
 canal designated by the letters c.opht.lat transmitted the n. ophthalmicus 
 lateralis (cf. above) and as this canal does not communicate with the 
 trigeminus canal or the fine canals issuing from the latter, we may also 
 conclude with full certainty that it did not give passage to any general 
 cutaneous fibres. These fibres to the orbit or dorsal side of the neu- 
 rocranium must therefore undoubtedly have emerged into the orbit 
 through the canals c.ophth.prof and c.ophth.sup.V, and it is easy to 
 understand that those of them which passed through the canal c.ophth. 
 prof formed the r. ophthalmicus profundus trigemini, or a somewhat 
 equivalent nerve, while the others which passed through the canal 
 c.ophth.sup.V formed the r. ophthalmicus superficialis trigemini. In 
 other words, the general cutaneous fibres which formed the r. ophthal- 
 micus profundus were transmitted to the orbit through the canal c.ophth.- 
 prof, and those which formed the r. ophthalmicus superficialis trigemini 
 through the canal c.ophth.sup.V , while the lateralis fibres which formed 
 the n. ophthalmicus lateralis never entered the orbit but passed in the 
 interorbital wall through a canal of their own which is given off from 
 
Macropetalichthyids and other Arthrodires — Stensio 123 
 
 the facialis canal and has no connection with the trigeminus canal- (in- 
 cluding the trigeminus recess). A separate course of the r. ophthalmi- 
 cus lateralis and r. ophthalmicus profundus trigemini is, besides in the 
 form now under consideration, found in Acipenscr, Polyodon (Stensio 
 1925), Saurichtliys (Stensio 1925) and Siluroids (Workman 1900, 
 pp. 403-407; Herrick 1901, pp. 201-203). 
 
 The r. ophthalmicus superficialis trigemini and the r. ophthalmicus 
 profundus trigemini having issued from the trigeminus recess — the 
 proximal part of the trigeminus canal — it is fully evident that this 
 recess must have contained at least the main parts of the ganglion 
 gasseri and the profundus ganglion, i.e., the main parts of the ganglia 
 of the general cutaneous fibres. Only small parts of these ganglia 
 may perhaps have reached into the cavum cerebrale cranii. From the 
 origin of the r. ophthalmicus superficialis and the r. ophthalmicus pro- 
 fundus trigemini, we can further conclude that the distal narrower 
 part of the trigeminus canal (V 2 , 3 ) transmitted the r. maxillaris and 
 the r. mandibularis trigemini after their origin from the ganglion 
 gasseri. That these two latter rami were accompanied by certain 
 lateralis fibres and perhaps also by certain communis fibres, we shall see 
 from the subsequent account. 
 
 During their passage through the cranial wall, the trigeminus roots 
 proper (i.e., the general cutaneous roots and the motor root) of recent 
 fishes are, as a rule, accompanied by a lateralis root, which leaves the 
 medulla oblongata postero-dorsally of the trigeminus root proper run- 
 ning outward postero-dorsally or dorsally of these, and having its 
 ganglion dorsally or postero-dorsally of the ganglion gasseri. From 
 the ganglion of this lateralis root the n. ophthalmicus lateralis and 
 n. buccalis lateralis take their origin. 
 
 In Macropetalichthys rapheidolabis, in which the canal for the 
 n. ophthalmicus lateralis issues from the facialis canal and in which, 
 as we have seen, there is no communication between the proximal 
 parts of the latter canal and the trigeminus canal, the lateralis root 
 which gave rise to the n. ophthalmicus lateralis and the n. buccalis 
 lateralis must evidently have left the cranial cavity through the facialis 
 canal, traversing the proximal part of this canal and having its ganglion 
 there. The n. buccalis lateralis thus also arose within the facialis canal. 
 
 In order to reach the trigeminus canal, the n. buccalis lateralis in 
 M. rapheidolabis must, soon after its origin, have turned forward, 
 traversing the short canal c.buc.lat, described above (Figs. 10, 11), 
 which leaves the facialis canal slightly distally and somewhat ven- 
 trally of the canal for the n. ophthalmicus lateralis and goes to the 
 
124 Field Museum of Natural History — Geology, Vol. IV. 
 
 distal narrow part (V 2 , 3 ) of the trigeminus canal. The n. buccalis 
 lateralis after arriving into this part of the trigeminus canal, probably 
 joined more or less intimately the r. maxillaris trigemini and passed 
 distally with this in the normal way. 
 
 From the considerations given the following is evident in M. 
 rapheidolabis concerning the trigemino-facialis ganglionic complex and 
 its roots : i . Into the trigeminus recess there entered only the general 
 cutaneous roots and the viscero-motor root of the mandibular arch. 
 2. The ganglion gasseri and the profundus ganglion must have been 
 situated in a proximal wide part of the trigeminus canal, called 
 here the trigeminus recess. 3. Into the facialis canal there entered all 
 the lateralis fibres and all the communis fibres of the ganglionic com- 
 plex and, further, the viscero-motor root of the hyoid arch. 4. The 
 ganglia of the fibres entering the facialis canal, i.e., in this case the 
 ganglion geniculi, the ganglion of the dorsal lateralis root and the 
 ganglion of the ventral lateralis root, must have been situated in the 
 proximal part of this canal and partly, perhaps, in the acustico-facialis 
 recess. 
 
 The lateralis ganglion of the dorsal lateralis root (from which the 
 n. ophthalmicus lateralis and n. buccalis lateralis arose) must in the 
 facialis canal have had its position antero-dorsally of the two other 
 ganglia, the ganglion geniculi and the ganglion of the ventral lateralis 
 root. Although separated from the ganglion gasseri by a rather thick 
 wall, it thus had, as far as can be judged, about its normal position in 
 relation to this, and the fact that it was situated in the facialis canal 
 is therefore of hardly any importance from a morphological point of 
 view. Similar arrangements would certainly arise in many recent forms 
 if the septa of connective tissue that there more or less completely 
 separate the ganglia in question and their roots from each other 
 developed more strongly and chondrified or ossified. Moreover, we 
 find in certain recent forms that ail the roots of the trigemino-facialis 
 ganglionis complex may run close together through a single canal in 
 the cranial wall, and that their ganglionic formations may be very 
 intimately connected with each other. Thus the wall between the 
 trigeminus and facialis canals in certain cases is reduced. That varia- 
 tions occur within certain bounds in the trigemino-facialis ganglionic 
 complex of fishes is accordingly not so remarkable. 
 
 Whether in M. rapheidolabis any general cutaneous fibres went 
 backward to the facialis branches and any communis fibres forward 
 to the trigeminus branches it is not possible to say with certainty from 
 the conditions in this fossil, but it is not impossible that this was the 
 
Macropetalichthyids and other Arthrodires — Stensio 125 
 
 case, for these fibres can very well be thought to have traversed the 
 canal c.buc.Iat. Possible communis fibres to the n. ophthalmicus lateralis 
 must have entered the canal for this nerve directly from the facialis 
 canal. 
 
 The r. oticus lateralis, which, as is known, is a branch from the 
 n. buccalis lateralis, certainly arose within the facialis. canal, probably 
 just as the n. buccalis lateralis on its way forward entered the canal 
 c.buc.Iat. Whether the r. oticus lateralis then traversed the entire re- 
 maining part of the facialis canal or sooner or later perforated its 
 dorsal wall, entering a canal of its own, it is impossible to decide at 
 present. 
 
 Apart from the r. oticus facialis, the distal part of the facialis 
 canal (distally of the point of origin of the ophthalmicus lateralis 
 canal) transmitted communis fibres — the viscero-motor fibres — for the 
 hyoid arch, lateralis fibres for the n. mandibularis externus and pos- 
 sibly also a few general cutaneous fibres. All these fibres formed to- 
 gether a large facialis stem which, soon after its origin from the 
 ganglia within the canal, must have branched into the r. palatinus and 
 the truncus hyoidemandibularis. 
 
 The r. palatinus, after its origin in the facialis canal, must first 
 have turned forward and traversed the canal c.pal x which leads from 
 the facialis canal to the orbit (cf. above). Well within the orbit the 
 nerve continued forward for some distance in a groove on the orbital 
 floor, then penetrated downwards in the floor through the canal 
 c.pal 2 to the suborbital part of the internal carotid canal. In the last 
 mentioned canal it went only a short distance forward, leaving it 
 through the opening c.pal 3 and arriving at the lower surface of the 
 primordial neurocranium. When emerging from the opening c.pal 3 
 it was perhaps accompanied by a fine branch from the arteria carotis 
 interna (Cf. text fig. 6). 
 
 The correctness of the course of the r. palatinus facialis of the 
 fish as sketched here is proved by the conditions in recent fishes, in 
 which the nerve in question also runs in a similar way and also often 
 more or less closely accompanies the a. carotis interna for a correspond- 
 ing distance. (Allis 1897, pp. 498, 685, PI. XXXVI, fig. 61 ; PI. 
 XXXVII, figs. 62, 63; 1901, p. 182; 1909a, pp. 43-55; 88, 185-186; 
 1911a, p. 290; 1919a; 1922a, pp. 280-281; 1923, p. 213; Allen 1905, 
 pp. 51-56; Greil 1913, PI. LV, fig. 2; etc.; cf. also Stensio 1923, 
 pp. 1241-1248; 1925.) 
 
 If in M. raphcidolabis there was an anastomosis between the r. 
 palatinus facialis and the r. palatinus glossopharyngei, this anasto- 
 
126 Field Museum of Natural History — Geology, Vol. IV. 
 
 mosis must have been situated in the anterior suborbital part of the 
 internal carotid canal and the r. palatinus glossopharyngei must have 
 arrived there through the canal era, the sinus si, and the canal c.cotn 
 (Cf. Allis 1897, Pis. XXXVI, fig. 61; PI. XXXVII, figs. 62, 63; 
 1909a, p. 186; 1911a, p. 289-290). The canal for the n. ophthalmicus 
 lateralis has already been dealt/ with, and I call attention here only 
 to the fact that from its origin in the lateral wall of the labyrinth 
 region it goes forward and upward to the lateral wall of the orbito- 
 temporal region, . from which it soon arrives up and into the dorsal 
 wall, continuing into this through the entire ethmoidal region. It gives 
 off several rami in dorsal direction to the sensory canal situated dor- 
 sally of it, i.e., the supraorbital sensory canal. 
 
 After having dealt with the trigeminus and facialis nerves and 
 their roots and ganglionic complex, we can now understand better than 
 before the relation between these structures and the jugular vein, and 
 that my view of the course of the latter ought to be correct. As I 
 suppose this vein to have run, it must, after its origin in the orbit, 
 have passed backward below the r. maxillaris and the r. mandibularis 
 trigemini. During its passage through the basal part of the postorbital 
 process it then crossed the facialis canal on the dorsal side, thus pass- 
 ing ventrally of the trigeminus and dorsally of the facialis. Although 
 it ran rather far laterally of the trigemino-facialis ganglionic complex 
 and did not, as in most recent forms, traverse this, it had, however, 
 obviously retained its normal relations to the trigeminus and facialis 
 nerves. (Cf. Allis 1922a, pp. 260-264.) 
 
 Neither for the n. trochlearis nor for the n. abducens have canals 
 so far been found. The trochlearis canal was probably very narrow, 
 and may, therefore, have been overlooked, especially as the medial 
 wall of the orbit is very imperfectly known in detail at the place where 
 this canal ought to have been situated. That the abducens canal is 
 not found may perhaps be explained by concluding that there was in 
 reality no independent canal for the n. abducens, for this nerve may 
 perhaps have emerged into the orbit more or less closely associated 
 with the r. maxillaris and r. mandibularis trigemini, as it does in 
 certain recent forms, for instance, certain Selachians. (Allis 1901, 
 p. 132; etc.) 
 
 A rather wide canal for the n. oculomotorius opens into the orbit 
 somewhat in front of the canals for the n. ophthalmicus profundus and 
 the r. ophthalmicus superficialis (V, Figs. 8, 10, 11; PI. XXIII; PI. 
 XXV, fig. 2; PI. XXVI, figs. 1, 2). A short distance in front of that 
 canal, the wide opticus canal (II, Figs. 5, 10, 11 ; PI. XXIII ; PI. XXV, 
 
Macropetalichthyids and other Arthrodires — StensiO 127 
 
 fig. 2) pierces the lateral wall in an anterolateral direction, opening 
 into the anterior half of the orbit not far above the floor. In its most 
 distal (orbital) part it receives on the anterior side the canal for the 
 arteria carotis interna (c.car.int), which, during its way upwards from 
 the ventral wall, makes a long curve forward, so that for a short 
 distance it even traverses a posterior part of the ethmoidal region. 
 
 We thus find that the arteria carotis externa entered the distal 
 part of the opticus canal. From there it evidently must have traversed 
 the canal in medial direction to reach the cavum cerebrale. But just 
 as it turned medially to do this it probably gave off the branch known 
 as the arteria optica, for the external opening of the opticus canal 
 has on its anterior side distally of the confluence with the internal 
 carotid canal a distinct emargination (a.opt), which cannot have had 
 any other function than to transmit a vessel. (Cf. Allen 1905, pp. 56- 
 57; Allis 1897, p. 498; 1908a, p. 258; 1908b, p. 223; 1909a, p. 186; 
 1911a, p. 290; 1911b, p. 518; 1912a; 1912b, pp. 489-490; 1912c, p. 
 586; 1914a; 1922a, p. 168; 1923, PI. XIX; Danforth 1912, pp. 442- 
 445 etc.) 1 
 
 The way in which the arteria carotis interna of the fish entered the 
 cavum cerebrale may perhaps at first glance appear rather strange, as 
 this artery in fishes generally pierces the ventral wall of the primordial 
 neurocranium so that it arrives immediately into the cavum cerebrale 
 and not first into the opticus canal. But a closer study makes it clear 
 that the difference is really not so important, for in certain forms, 
 e.g., Saurichthys (Stensio 1925), the canal for the internal carotid 
 opens either in the proximal part of the opticus canal or at the tran- 
 sition between this and the cavum cerebrale cranii. We have there a 
 stage intermediate between the one in Macropetalichthys raphcidolabis 
 just described and the one generally occurring in fishes. In this con- 
 nection it is also noteworthy that the arteria carotis interna in Polyptcrus 
 (Allis 1908b, loc.cit; 1922a, loc.cit; Lehn 1918, p. 380) runs exter- 
 nally of the interorbital wall so far forward that it enters the cavum 
 cerebrale either together with or close behind the n. opticus, and that 
 in Elasmobranchs the same artery within the lateral wall of the neuro- 
 cranium goes a long distance forward and upward in direction towards 
 the opticus canal before entering the cavum cerebrale. (Allis 1912b, 
 loc.cit; 1914a, pp. 229-230, 236, 238, etc.) 
 
 'Another interpretation is possible, viz. : — that the arteria pseudobranchialis 
 efferens reached the a. carotis interna just in the opticus canal, and that the 
 a. ophthalmica magna was given off to the orbit through the emargination on 
 the anterior side of the opticus canal. This emargination would consequently in 
 this case have been traversed by both the a. ophthalmica magna and the a. pseudo- 
 branchialis efferens. 
 
128 Field Museum of Natural History — Geology, Vol. IV. 
 
 Whether a canal for the arteria pseudobranchialis efferens was 
 present or not in the lateral wall could not be decided from the material 
 available, and it is therefore not possible to conclude anything with 
 certainty concerning the relations between the arteria pseudobranchialis 
 efferens and the arteria carotis interna. (Cf. the footnote on the pre- 
 ceding page.) 
 
 As is seen from the description given of the trigeminus and facialis 
 canals, there is no real trigemino-facialis chamber in Macropetalichthys 
 raphcidolabis. 
 
 A posterior myodome also does not occur in M. rapheidolabis, and 
 it does not seem probable that there was an anterior one either. A 
 satisfactory decision in the latter respect is, however, not as yet pos- 
 sible, since the conditions of the anterior wall of the orbit are very 
 imperfectly known. 
 
 Since the postero-ventro-medial corner of the orbit as well as the 
 postero- ventral part of the interorbital wall were partly covered by 
 matrix, which could not be removed without hurting the fossil as a 
 specimen, the pituitary canal (canal for the pituitary vein) could not 
 be observed, but I do not doubt that it actually exists, for it is, as 
 far as I know, without exception found in all fishes, although in those 
 forms which have a posterior myodome it is a part of this and has 
 been much widened and transformed. (Cf. Allis 1909a, pp. 183-208; 
 
 cophih.lat 
 
 C-ophth.Su.pV *« 
 
 cophthprqf 
 
 o.ca-r.ext cca.T-.ini 
 
 Fig. 8. Macropetalichthys rapheidolabis 
 Transverse section through the posterior part of the orbitotemporal region 
 and the antero-lateral parts (postorbital processes) of the labyrinth region. Layers 
 of substitution bone drawn with black lines, dermal bones with vertical striation. 
 Cartilage dotted. X-K- 
 
 Li, L 2 , Mi, Pi, dermal bones of the cranial roof (boundaries approximately 
 drawn); c.car.cxt, canal for the arteria carotis externa; c. car. hit, canal for the 
 arteria carotis interna; c.ophth.lat, canal for the n. ophthalmicus lateralis; c. 
 ophth.prof, canal for the r. ophthalmicus profundus trigemini; c.ophth.sup.V , 
 canal for the r. ophthalmicus superficialis trigemini ; cv, cavum cerebrale cranii ; 
 ifc, infraorbital sensory canal ; jg, canal for the vena jugularis through the 
 postorbital process (somewhat diagrammatically drawn as it really would not 
 be visible to this extent on a single transverse section) ; ju, canal in the part 
 seen in the section for the vena mandibulars, behind this place for the vena 
 jugularis; orb, orbit; soc, supraorbital sensory canal; III, canal for the n. 
 oculomotorius ; VII, facialis canal. 
 
Macropetalichthyids and other Arthrodires — Stensio 129 
 
 1911a, p. 291; 1914a; 1918a; 1919a; 1922a, p. 228; 1922b; Gegenbaur 
 1872, pp. 75-79, canalis transversus Allen 1905, p. 81 ; O'Donoghue 
 1914, pp. 440-441; Stensio 1921, pp. 62, 177-180; 1922, p. 179; 
 1923, pp. 1244, 1262-1268; 1925, Lehn 1918, p. 380.) 
 
 Owing to reasons similar to those which prevented study of the 
 pituitary canal, it was not possible to see any canal for the vena cere- 
 bralis anterior, which in fishes usually perforates the upper part of 
 the interorbital wall and goes to the orbit. (Cf. Allis 1897, PI. 21, 
 figs. 9-1 1, acvfr; 1903, PI. 4, fig. 9, acvfr; 1909a, p. 39; 1922a, pp. 
 261-262 Lehn 1918, p. 380, figs. 3, 4; Stensio 1921, pp. 168-169; 
 1923, pp. 1246-1247; 1925.) 
 
 Ethmoidal Region 
 
 As compared with the labyrinth and orbito-temporal regions the 
 ethmoidal region is fairly long, its breadth being, however, also con- 
 siderable, and even so considerable that it exceeds the length. On 
 account of the presence of the extensive orbital bottom, the region is 
 not marked off from the orbitotemporal region ventrally. Higher up 
 above the bottom of the orbit it is, however, much broader than the 
 latter region, forming the anterior wall of the orbits and projecting on 
 each side as a preorbital process in the normal way (Figs. 1, 3, 5; Pis. 
 XIX-XXI; PI. XXIV, fig. 1; PI. XXV, figs. 1, 2). 
 
 Both the height and breadth of the region decrease anteriorly, the 
 latter, however, proportionally somewhat more than the former. The 
 anterior end is imperfectly preserved, but it is probable that it was 
 truncated, and, to judge from the conditions in Epipetalichthys de- 
 scribed below, it is likely that it was, in addition, a little concave. 
 
 Close to the anterior end of the region there projects from the 
 basal parts of each side a strong, broad process (olf, Figs. 1, 3, 5; 
 Pis. XXI-XXIII; PI. XXIV, fig. 1) in a lateral and somewhat ven- 
 tral direction. This process has its distal end truncated and provided 
 with a large oval opening (na, nf), which, as is evident from the de- 
 scription of Epipetalichthys below, must at least in part be the nasal 
 opening. Accordingly, the process contains the nasal cavity and rep- 
 resents the olfactory capsule. We thus have here an olfactory capsule 
 that, as in Elasmobranchs, is fairly independent and projects laterally 
 in relation to the other internasal parts of the ethmoidal region. (Cf. 
 Allis 1923, pp. 126-147.) 
 
 A closer examination of the nasal opening shows that this, by a 
 process of bone that projected forwards from the posterior wall, is 
 incompletely subdivided into a dorso-lateral (na) and a ventro-medial 
 
130 Field Museum of Natural History — Geology, Vol. IV. 
 
 (»/) portion. Both of these portions are directed ventro-laterally, but 
 the latter considerably more so than the former, so that it in fact is 
 situated on the ventral side of the nasal capsule (Figs, i, 5; 
 PI. XXI ; PI. XXIII ; PI. XXIV, fig. 1). I was at first inclined to con- 
 sider them as corresponding to the two nasal apertures generally found 
 in fishes, and to think that the process that separates them would be 
 homologous with a part of the ala nasalis of Elasmobranchs (Allis 
 1923, pp. 126-147), and that the ala nasalis would be continuous with, 
 and form an integral part of, the nasal capsule. But after having read 
 the fine memoir on Chlamydoselachus recently published by Allis 
 (1923) I arrived at the conclusion that what has been called above the 
 ventromedial portion («/) of the nasal opening, really represents a lat- 
 eral part of the nasal fontanelle of Chlamydoselachus and certain other 
 Selachians and that, accordingly, the other portion (na) alone is the 
 nasal opening. The ala nasalis or its possible homologue, if there were 
 one, was probably entirely without ossification. If this opinion is true, 
 the vein commissure that in Selachians (Allis 1923, pp. 126-147), 
 Saurichthys (Stensio 1925), and probably in most fishes (Allen 1905, 
 PI. 1, fig. 1; PI. Ill, figs. 17, 18), connects the orbito-nasal vein and 
 the anterior facial vein with each other must, obviously, have traversed 
 the posterior part of the nasal fontanelle. Whether this fontanelle 
 was entirely occupied by a membrane of connective tissue or whether 
 it was more or less occupied by cartilage it is not possible to decide. 
 
 From what has been set forth regarding the nasal capsule, it is 
 clear that M. rapheidolahis in this regard much resembles the Elas- 
 mobranchs. 
 
 The cavum cerebrale cranii extends only a short distance into the 
 posterior part of the region, which, therefore, apart from certain 
 canals, in the whole of its interior consisted of cartilage. On account 
 of this fact we cannot conveniently consider it to be composed of a 
 number of walls, as in the case of the other regions, but will have to 
 deal with it as a solid body with five external surfaces which are well 
 bounded from each other. These surfaces are the following : a ventral, 
 a posterior, a dorsal, and a right and a left lateral. 
 
 The ventral surface is broad and large and is posteriorly continuous 
 with the ventral surface of the orbitotemporal region (Fig. 1 ; 
 Pis. XXI-XXIII). Unfortunately it is not fully preserved in its 
 anterior parts. As is seen from the figures (the transversal section 
 reproduced in Fig. 9 and PI. XXIV, fig. 1) it is slightly concave 
 both in transversal and rostro-caudal direction. And, as has been 
 pointed out above, it is covered by the external bone layer 
 
Macropetalichthyids and other Arthrodires — Stensio 131 
 
 throughout its extension. Not far from each lateral border it has a 
 distinct sulcus {s.pal, Figs. 1, 9; PI. XXII), which leads forward 
 for some distance from the opening c.pal z (Fig. 1; PI. XXII). 
 This sulcus was evidently developed for the r. palatinus facialis, and 
 may, perhaps, in addition, also have lodged a small arterial branch 
 given off from the arteria carotis interna through the foramen c.pal 3 . 
 So far as can be judged there is no indication of an articulation facet 
 for the palatoquadrate on the ventral surface of the region. 
 
 Close, medially, to each groove for the r. palatinus just described, 
 and about midway between the anterior end of the region and the 
 foramen c.pal z , the external bone layer seems to show faint indications 
 of an ossification centre, a centre which, if it really exists, is so situ- 
 ated that the bone which it represents corresponds at least to the 
 exethmoid (lateral ethmoid) of fishes in general, and perhaps, in addi- 
 tion, to the preethmoid of certain fishes. (Cf. Allis 1897, PI. 21, figs. 
 8-10, SMX; 1898, pp. 446-450; 1909a, pp. 17-22; Stensio 1921, pp. 
 61, 93-94; I9 2 2, PP- 184-186; 1923, pp. 1247-1248; 1925, etc.) A 
 positive decision in this respect is, however, not possible until more 
 and better preserved material is available for investigation. 
 
 The posterior surface of the region, if we next turn to this, is, 
 by the anterior part of the orbitotemporal region above the orbital 
 floor, subdivided into a right and a left half, each forming the anterior 
 surface of the orbit of its side. As it could not be cleaned from 
 matrix, I can state little concerning it. It may have had a foramen 
 through which the r. ophthalmicus superficialis after traversing the 
 orbit entered the region, but nothing can be said about the posi- 
 tion of this foramen. As has already been pointed out above, there 
 probably was no anterior myodome developed, but a positive de- 
 cision in regard to this point is evidently not possible until material is 
 found which- will enable us to undertake a detailed examination of the 
 posterior parts of the region. The olfactory nerve did not traverse 
 the orbit and accordingly it had no foramen on the posterior surface 
 of the ethmoidal region. 
 
 The dorsal surface, like the ventral one, is large and broad and 
 posteriorly continuous with the dorsal surface of the orbitotemporal 
 region (Figs. 3, 5, 9; Pis. XIX-XX; PI. XXIV, fig. 1; PI. XXV, 
 figs. 1, 2; PI. XXVI, figs. 1, 2). It is slightly convex, both in trans- 
 versal and rostro-caudal direction (Fig. 9), and slopes slightly for- 
 ward. In its most anterior parts it is not preserved in the specimen 
 examined, being either crushed together with the dermal cranial roof 
 or destroyed in some other way. From the closely allied Epipetalichthys 
 
132 Field Museum of Natural History — Geology, Vol. IV. 
 
 to be described below, we are, however, justified in concluding that the 
 conditions were about as shown in Fig. 3, i.e., that there probably 
 was a deep, broad, antero-dorsally directed depression (pfg) at the 
 anterior end of the dorsal surface, and that this depression was lined 
 throughout by a perichondral bone layer. A similar depression seems 
 among adult fishes to occur only in Elasmobranchs (Cf. Gegenbaur 
 1872, Pis. IV-VIII, etc.; Allis 1913; 1923, pp. 127-130; Fritsch 
 1895, figs. 228, 230; Woodward 1916-1919, PI. II, fig. 1, PI. XXVI, 
 fig. 3; Daniel 191 5, PI. I, fig. 1, etc.), and we would accordingly have 
 here a very remarkable agreement between Epipetalichthys and Macro- 
 petalichthys on the one hand, and the Elasmobranchs on the other. As 
 we shall see from the account given below, a depression of the same 
 sort probably occurred in all Arthrodires. In Elasmobranchs the de- 
 pression, which is described by Allis under the name of cavum pre- 
 cerebrale, is filled by a fatty tissue and this was probably the case 
 also in the Arthrodires, since the dermal bones in these do not bend 
 down so that they could have been directly superimposed on the bottom 
 of the depression, but are situated on a level with the upper borders 
 of this. This relatively loose position of the dermal skeleton in rela- 
 tion to the primordial neurocranium at the anterior end of the ethmoidal 
 region explains why the dermal skeleton is generally in a very im- 
 perfect state of preservation there. While in Elasmobranchs the cavum 
 precerebrale is separated from the cavum cerebrale in most cases merely 
 by membranous tissue, there was here in Macropetalichthys rapheido- 
 labis and in Epipetalichthys certainly a thick wall of cartilage between 
 them. In Lcemargus borealis I have also found that the wall between 
 the cavum precerebrale and the cavum cerebrale is to a large extent 
 formed by cartilage. This is also partly shown in White (1892, PI. I, 
 
 %. 5)- 
 
 From the dorsal side of the anterior part of the cavum cerebrale 
 cranii a rather wide canal goes dorsally and forward to the upper sur- 
 face of the region, where it opens with a large oval foramen (par, 
 Fig. 3; Pis. XIX-XX; PI. XXVI, fig. 1). This canal, as we shall 
 see, lodged the pineal or parietal organ or both these structures. 
 
 Apart from this canal, several fine ones for vessels and nerve 
 branches naturally opened on the dorsal surface of the region. In 
 Pis. XXI and XXIII we can see one of this sort (rm.ophth.lat), that 
 goes from the canal for the r. ophthalmicus lateralis (c.ophth.lat) a 
 short distance upwards and forwards to the anterior part of the supra- 
 orbital sensory canal (soc). This sensory canal being, like the others, 
 situated in a ridge on the lower side of the dermal bones, there must 
 
Macropetalichthyids and other Arthrodires — Stensio 133 
 
 evidently have been a paired groove corresponding to the ridge, a 
 groove running as is shown by text fig. 3. 
 
 Finally, the lateral surface of each side (Figs. 5, 9; Pis. XXI- 
 XXIII; PI. XXIV, fig. 1; PI. XXV, figs. 1, 2) is strikingly long, 
 but not particularly high. It faces laterally and slightly downwards 
 and is somewhat concave in dorso-ventral direction. From its antero- 
 ventral part projects the large nasal capsule (olf) already described, 
 but, as is seen from text fig. 5 and PI. XXIV, fig. 1, this does not 
 occupy more than the ventral half of the height of the surface at this 
 place. As far as can be seen, there is no articulation facet for the 
 palatoquadrate, either on the surface itself or on the posterior or ventral 
 parts of the nasal capsule. This, together with the fact pointed out 
 above that there was no articulation facet on the ventral surface either, 
 seems to indicate that the palatoquadrate actually did not articulate 
 against the region, but was suspended beneath this with ligaments in 
 about the same manner as in the Selachians. 
 
 Ll =»„■-#* 
 
 g±?^. 
 
 a.pal 
 
 Fig. 9. Macropetalichthys rapheidolabis 
 
 Transverse section through the posterior half of the ethmoidal region, slightly 
 anterior to the anterior end of the cavum cerebrale cranii. Layers of substitution 
 bone denoted by strong black lines ; cartilage dotted ; dermal bones with vertical 
 striation. XK- 
 
 Li, Mi, dermal bones of the cranial roof (their boundaries not fully correct) ; 
 soc, supraorbital sensory canal ; s.pal, groove for the r. palatinus facialis and 
 perhaps for an arterial branch too; vx, proximal part of a canal running up- 
 wards in the cartilage, probably for some venous vessel; I, canalis tractus ol- 
 factorii. 
 
 In this connection it may also be pointed out that the postnasal 
 part of each lateral surface is very large, a condition probably caused 
 by the insertion of a powerfully developed part of the musculus con- 
 strictor superficialis (I) homologous to the muscle of the Selachians and 
 Teleostomes generally, known as the m. levator labii superioris (Vetter 
 1874, etc.), levator maxilla; superioris (Allis 1897, pp. 552-556), ab- 
 ductor B (Allis 1901, PI. II, fig. 4; PI. 12, fig. 5) or preorbitalis 
 (Luther 1909, pp. 36-39). If this were not the case, it would be 
 very difficult to find a satisfactory explanation for the considerable 
 post-nasal extension of the lateral surfaces. 
 
134 Field Museum of Natural History — Geology, Vol. IV. 
 
 On the posterior part of each lateral surface there are found the 
 anterior openings of two fairly wide canals (c.ophth.sup.V ? and vy, 
 Fig. 5 ; PI. XXII ; PI. XXIII ; PI. XXIV, fig. i ) which evidently issue 
 from the orbit. But as the position of the posterior openings of these 
 canals and their course within the preorbital process are so far un- 
 known, it cannot be positively decided what their functions were. The 
 dorsal one {c.ophth.sup.V?) possibly transmitted the r. ophthalmicus 
 superficialis trigemini, the ventral one either vessels or the r. maxillaris 
 trigemini and the n. buccalis lateralis or both these nerves and vessels. 
 
 From the anterior end of the cavum cerebrale cranii, a paired, 
 rather low but wide canal (I, Figs. 9, 10, 11, 12, 13; Pis. XXIII, 
 XXVII) issues in a forward and somewhat lateral direction to the nasal 
 fossa of its side, being during the entire length situated very low in 
 the region. This canal, which narrows anteriorly, lodged, as far as I 
 can judge, the tractus olfactorius and, at the most anterior, probably 
 somewhat widened end, the lobus olfactorius. It may, therefore, be 
 called the canalis tractus olfactorii. That it would lodge a long olfactory 
 nerve and that the lobus olfactorius would have been situated at the 
 front end of the telencephalon is not, as will be set forth below, in 
 accordance with what we know concerning the features of the brain 
 otherwise. From its dorso-medial side there are given oft" in the 
 Field Museum specimen two canals of finer calibre which run antero- 
 dorso-laterally within the interior of the region {yx, vx\, PI. XXVI, 
 fig. 1), and which probably were traversed by veins draining the car- 
 tilage and dorsal surface. 
 
 Close below the groove for each supraorbital sensory canal, we 
 find the canal for the r. ophthalmicus lateralis (c.ophth.lat, Fig. 9; 
 Pis. XXI, XXIII), from which, as already pointed out, branches 
 (rm.ophth.lat) run upward to the supraorbital sensory canal. 
 
 Cavum Cerebrale Cranii and Brain 
 
 Although being partly much narrower than the primordial neuro- 
 cranium, the cavum cerebrale cranii (cv, Figs. 2, 4, 8, 10, 11 ; Pis. XX, 
 XXI, XXIII; PI. XXIV, fig. 3; PI. XXV figs. 1, 2; PI. XXVI, figs. 
 1, 2; PI. XXVII) has nevertheless a rather considerable size. Ob- 
 serving its shape, it may be considered to consist of three divisions, 
 which in the subsequent description are called the posterior, middle and 
 anterior divisions respectively. 
 
 The posterior division is limited entirely to the posterior narrow 
 division of the occipital region, and, in accordance with this, it is 
 much narrower and lower than the other two {pd, Figs. 10, 11; PI. 
 
 
Macropetalichthyids and other Arthrodires — Stensio 135 
 
 XXVII). Its narrowest part is situated anteriorly, just at the tran- 
 sition to the middle division. Behind this place it grows slowly and 
 gradually. broader and a little lower backwards, but whether it did so 
 quite to the posterior end or not is unknown. Two transverse sections 
 through its posterior half somewhat anterior to the cranio-spinal 
 processes are seen in text fig. 2. 
 
 The middle division (md, Figs. 10, 11) extends through the 
 anterior division of the occipital region, the whole labyrinth region and 
 a short posterior part of the orbitotemporal region (Cf. also Figs. 
 2, 4, 8; Pis. XX, XXIII; PL XXIV, fig. 3; PL XXV, figs. 1, 2; PL 
 XXVI, figs. 1, 2), and is the largest one of the three with regard to 
 height and breadth, but its length is about equal to that of the pos- 
 terior division. Its maximum height is situated approximately at the 
 transition between the occipital and labyrinth regions, decreasing from 
 there both forward and backward, although only very slightly for- 
 ward. The bulge caused by the vagus and the other nerves, not taking 
 into consideration the width, attains its maximum in the front end of 
 the division, in the most anterior part of the orbitotemporal region, 
 from which it narrows slightly both backward and forward. 
 
 The anterior division (ad, Figs. 10, 11) extends through the an- 
 terior larger part of the orbitotemporal region and a very short 
 posterior part of the ethmoidal region, where, at the very anterior end, 
 it divides into a right and a left half, each of which anteriorly is con- 
 tinued by the canalis tractus olfactorii (Figs. 10, 11; PL XXIII; 
 PL XXVI, figs. 1, 2). It is well bounded from the middle division, 
 being both lower and narrower than the anterior part of this. The 
 height gradually diminishes forward throughout the length of the divi- 
 sion, finally becoming equal to that of the posterior end of the canalis 
 tractus olfactorii, while the breadth, on the contrary, increases slightly 
 in the same direction. The length is only about half as great as in 
 the other two divisions. 
 
 The floor of the middle division is slightly convex, that of the 
 posterior division slightly concave, and so is also that of the anterior 
 division, although the concavity is somewhat larger there (Figs. 
 10, 13; PL XXVI, figs. 1, 2) and forms a shallow, indistinct fossa. 
 In the median parts, this fossa has, in its turn, a small, very slightly 
 indicated depression (fhy, Fig. 10), which is so situated that it 
 must have lodged at least the most ventral part of the hypophysis. 
 The lobi inferiores probably had their position laterally of this de- 
 pression, which may represent the last remnant of an originally more 
 pronounced and large fossa hypophyseos. 
 
136 Field Museum of Natural History — Geology, Vol. IV. 
 
 ,.JII ..c.opkth.SupV 
 
 VII 
 
 Fig. 10. Macropetalichthys rapheidolabis 
 
 Shape of the cavum cerebrale cranii in ventral view. With the exception 
 of the posterior division, which is restored from specimens in the American 
 Museum, entirely after the Field Museum specimen. X24. 
 
 ad, anterior division of the cavum cerebrale cranii; afr, acustico-facialis 
 recess ; a.opt, division of the external opening of the opticus canal, probably for 
 the arteria optica; c.bucc.lat, canal for the lateralis fibres, representing the n. 
 buccalis lateralis and possibly also for certain general cutaneous fibres to the 
 n. facialis ; c.car.int, canal for the arteria carotis interna ; c.ophth.lat, canal for 
 the n. ophthalmicus lateralis ; c.ophth.prof, canal for the r. ophthalmicus pro- 
 fundus; c.ophth.sup.V , canal for the r. ophthalmicus superficialis trigemini; c.pah, 
 canal for the r. palatinus facialis from the facialis canal to the orbit; dx, canal 
 for a dorsal branch from the n. linese lateralis ; fhy, depression probably for the 
 
Macropetalichthyids and other Arthrodires — Stensio 137 
 
 hypophysis; md, middle division of the cavum cerebrale; n.l, canal for the 
 n. lineae lateralis ; pd, posterior division of the cavum cerebrale cranii ; v.lb, canal 
 leading from the labyrinth cavity to the vagus canal, probably for a large vein; 
 I, canalis tractus olfactorii ; II, division of the opticus canal for the n. opticus ; 
 III, canal for the n. trochlearis; V, proximal wide part of the trigeminus canal, 
 forming a trigeminus recess ; V»,», distal narrow part of the trigeminus canal 
 traversed by the r. maxillaris, r. mandibularis trigemini and the n. buccalis 
 lateralis ; VII, facialis canal ; VIII, acusticus canal ; IX, glossopharyngeus canal ; 
 Xm, vagus canal transmitting the n. vagus and the v. cerebralis posterior. 
 
 In the posterior part of the roof of the anterior division, there is, 
 in the Field Museum specimen, a median, unpaired diverticle (ep) 
 situated as shown by Fig. 11, i.e., it extends a short distance 
 upward into the dorsal wall of the orbitotemporal region. Somewhat 
 in front of this diverticle, the canal par to which reference has already 
 been made issues from the same division in an antero-dorsal 
 direction to the dorsal surface of the posterior part of the ethmoidal 
 region, opening there with a rather large, oval foramen (par, Figs. 
 3, 11-13; Pis. XIX, XX; PI. XXVI, fig. 1). 
 
 According to the observations recently made by Woodward (1922, 
 p. 31, fig. 3) the pit that regularly occurs on the lower side of the 
 pineal plate in non-Macropetalichthyid Arthrodires is paired in Tita- 
 nichthys, a fact which causes Woodward to consider that the epiphysis 
 there was a paired, symmetrically situated organ. We are thus in 
 Titanichthys concerned with a form in which the parietal and pineal 
 organs probably were situated beside one another (Cf. Goodrich 1909, 
 p. 25 ; Plate 1923, pp. 743-753) and it seems to be a matter of course 
 that in all typical non-Macropetalichthyid Arthrodires, as, for instance, 
 Dinichthys and Coccosteus, they must have been situated exactly as 
 in Titanichthys. 
 
 In Macropetalichthys rapheidolabis the bone-plate M x (text fig. 15) 
 of the dermal cranial roof with its posterior part evidently corresponds 
 to the pineal plate (Pi, Fig. 14) of Titanichthys and other typical 
 non-Macropetalichthyid Arthrodires. The canal by which it is 
 said by Eastman (1908a, p. 104; 1908b, p. 169) to be perforated 
 seems to have such a position that it must form the distal continua- 
 tion of the canal par, and at least ventrally correspond to the pit on 
 the lower side of the pineal plate in the non-Macropetalichthyid Arthro- 
 dires. These facts, together with the circumstance that Macropetalich- 
 thys rapheidolabis in most other respects is fairly closely allied to the 
 other Arthrodires, make it highly probable that the parietal and pineal 
 organs in it were situated as in these. If, therefore, these organs in 
 the other Arthrodires occupied a position beside one another, they 
 would obviously in M. rapheidolabis both have been situated in the 
 canal par, and the diverticle ep would thus have lodged no part of the 
 
138 Field Museum of Natural History — Geology, Vol. IV. 
 
 brain and might in such a case very well be thought to be occasionally 
 developed in the Field Museum specimen. That the diverticle ep can- 
 not probably have lodged the pineal organ is also, as kindly pointed 
 out to me by Professor N. Holmgren of Stockholm, indicated by its 
 position very far back. 
 
 Leaving this question, we turn to the canals that issue from the 
 three divisions of the cavum cerebrale cranii, beginning posteriorly and 
 proceeding forward. 
 
 It would be expected that several canals for spino-occipital nerves 
 had left the posterior division of the cavum cerebrale cranii, but, 
 as pointed out above, no such canals could be observed in the 
 material so far investigated, a fact which of course does not neces- 
 sarily imply that they are lacking, for the possibility is to be taken 
 into consideration that they are very minute and on that account can 
 only exceptionally be distinctly seen, and then only in especially well- 
 preserved specimens. 
 
 From the middle division of the cavum cerebrale issues most pos- 
 teriorly the wide vagus canal (X vn Figs. 10, 11, PI. XX; PI. XXIII; 
 PI. XXV, figs. 1, 2; PI. XXVI, figs. 1, 2). Some distance anterior of 
 this appears the glossopharyngeus canal (IX) and most anteriorly, 
 rather close to each other, the acusticus canal (VIII), the facialis canal 
 (VII) and the trigeminus canal- (V). The acusticus and facialis canals 
 take their origin from a common recess on the side of the division, a 
 recess which I have termed the acustico-facialis recess (afr). All the 
 canals enumerated leave the basal parts of the division. Besides these 
 there are, however, three other canals situated higher up, all of which 
 must have been traversed by vessels. The most posterior one of these, 
 (v.lb, Fig. 11; PI. XX; PI. XXV, figs. 1, 2; PI. XXVI, figs. 1, 2), 
 which is very wide and goes from the postero-dorso-medial parts of 
 the labyrinth cavity to the dorsal part of the vagus canal at the tran- 
 sition between this canal and the division in question of the cavum 
 cerebrale, probably transmitted a vein that drained the labyrinth. The 
 two remaining canals (v lt v 2 , Fig. 11; PI. XX; PI. XXV, figs. 
 1, 2; PI. XXVI, figs. 1, 2, 3) are both narrow, connect the dorsal parts 
 of the division above the acustico-facialis recess with the labyrinth 
 cavity and most probably may have been traversed by arteries. 
 
 From the anterior division of the cavum cerebrale cranii, the oculo- 
 motorius canal (III, Fig. 11; PI. XXIII; PI. XXV, figs. 1, 2; PI. 
 XXVI, figs. 1, 2), the opticus canal (II) and the canal for the tractus 
 olfactorius (I) take their origin, the one last mentioned from the an- 
 terior end and the one first mentioned far posteriorly, not so very far 
 
Macropetalichthyids and other Arthrodires — Stensio 139 
 
 from the middle division. Besides these canals there issue from the 
 anterior division, as we have already seen, the canal par and the diver- 
 ticle ep. 
 
 Guided by the detailed knowledge we possess of the canals for the 
 cranial nerves, we are to a certain extent able to draw conclusions con- 
 cerning the brain and its position in the cavum cerebrale. From the 
 place of exit and direction of the acusticus facialis and trigeminus 
 canals, it is obvious that the roots of the n. acusticus, n. facialis and 
 n. trigeminus must have left the medulla oblongata in the anterior half 
 of the middle division of the cavum cerebrale and, accordingly, that 
 the medulla oblongata reached far forward in the middle division (myl. 
 Figs. 12, 13). The most anterior part of this division was, how- 
 ever, certainly occupied by the whole, or, at least, the larger posterior 
 portion of the mesencephalon (tnes, Figs. 12, 13). Dorsally of the 
 mesencephalon and the anterior portion of the medulla, there probably 
 
 c.ophth.Jat par 
 
 c.ophih.sup.V 
 op \ 
 
 Fig. 11. Macropetalichthys rapheidolabis 
 
 Shape of the cavuum cerebrale cranii in lateral view. After the Field Museum 
 specimen, with the exception of the posterior division, which is restored after a 
 specimen in the American Museum. X^- 
 
 ep, canal of doubtful importance, probably an occasional structure ; par, pineal 
 canal, probably for both the parietal and pineal organs (at least in the basal half) ; 
 '1. v», canals probably for arteries from the cavum cerebrale cranii to the 
 labyrinth cavity ; vx, vx,, canals probably for veins. Other letters of reference 
 as in text fig. 10. 
 
 existed a rather well-developed cerebellum (met, Figs. 12, 13) and 
 the anterior half of the middle division may accordingly have been 
 rather well filled, while the posterior half of the same division, which 
 lodged only the posterior longer portion of the medulla oblongata, 
 cannot have been filled by this. The conditions there must have been 
 about as seen in Figs. 12, 13. The medulla was undoubtedly long, 
 but it is impossible to say with certainty how far back it extended. 
 
 The anterior division of the cavum cerebrale must, as is easily 
 understood, have lodged the telencephalon and diencephalon (Figs. 
 12, 13), and in addition probably a short anterior portion of the mesen- 
 
140 Field Museum of Natural History — Geology, Vol. IV. 
 
 cephalon. The diencephalon (die) seems to have been long, probably 
 reaching forward to about the middle of the division, but, may as far as 
 can be judged, not have been so high as this. The hypophysis, and, if 
 they were developed, the saccus vasculosus and the lobi inferiores as 
 well, may have been situated in the shallow concavity in the floor of 
 the anterior division approximately in the way shown in text fig. 13. 
 The hypophysis occupied, however, a small depression of its own in the 
 middle part of the concavity (fhy, Fig. 10). The canal par, prob- 
 ably lodged both the parietal and pineal organs, while the importance 
 of the diverticle ep is still doubtful. 
 
 The telencephalon (tel, Figs. 12, 13) probably occupied about 
 the anterior half of the anterior division of the cavum cerebrale. It 
 must have been comparatively broad and low and anteriorly it was 
 probably bilobated, a portion of it on each side extending into the 
 proximal part of the canalis tractus olfactorii of its side. From each 
 of the two anterior lobes the tractus olfactorius (I) went through its 
 long canal anteriorly and laterally to the lobus olfactorius (l.olf), 
 which, as far as can be judged, was situated close to the nasal pit. 
 
 From what has now been said it will be evident that the cavum 
 cerebrale cranii only to a certain extent reflects the shape of the brain. 
 But despite this it seems to be rather clear that the brain cannot have 
 resembled that in Dipnoi, Crossopterygii or Actinopterygii, but that it 
 may have been Elasmobranchian-like. At least with regard to certain 
 anterior parts it probably was much as in Hcptanchus (Cf. Gegenbaur 
 1898, Fig. 455) and Chalmydoselachus (Allis 1923, PI. 22, fig. 59). 
 
 The restoration of the brain given in text figs. 12 and 13 is based 
 chiefly on the conditions in Heptanchus and is of course to a large extent 
 rather arbitrary. 
 
 DERMAL BONES OF THE PRIMORDIAL NEUROCRANIUM 
 
 The ventral side of the primordial neurocranium was entirely with- 
 out dermal bones, the parasphenoid of Cope's (1891, p. 453, PI. 29, 
 fig. 2; PI. 30, fig. 1) and Eastman's (1908a, pp. 108-109; 1908b, pp. 
 174-175) descriptions being simply the external bone layer on the 
 ventral side of the primordial neurocranium. 
 
 The dorsal side of the primordial neurocranium, on "the contrary, 
 is, as known, completely covered by a cuirass of dermal bones, a cuirass, 
 which, in this paper, is generally referred to as the dermal cranial roof. 
 On account of the postero-medial displacement of the eyes, the orbital 
 openings have become entirely encircled by the dermal cranial roof. 
 
Macropetalichthyids and other Arthrodires — Stensio 141 
 
 Fig 12. Macropetalichthys rapheidolabis 
 
 Hypothetical shape of brain from above in relation to the cavum cerebrale 
 cranii. Brain with a dark tone, outlines of the cavum cerebrale indicated by a 
 continuous line. X?4- 
 
 die, diencephalon ; l.olf, lobus olfactorius; mes, mesencephalon; met, cerebel- 
 lum ; myl, medulla oblongata ; nac, nasal pit ; par, parietal or pineal organs or 
 both; tel, telencephalon; I, tractus olfactorius; II, n. opticus; III, n. oculo- 
 motorius; V, roots of the trigeminus proper; VII, facialis roots + lateralis 
 roots of the entire trigemino-facialis complex; VIII, root of the n. acusticus ; 
 IX, root of the n. glossopharyngeus ; X, roots of the n. vagus and the n. lineae 
 lateralis. 
 
142 Field Museum of Natural History — Geology, Vol. IV. 
 
 The general shape of the dermal cranial roof may be seen in 
 Fig. 15. The restoration given of it by Dean (1901, Fig. 12, p. 116) 
 is almost correct, while the one published six years later by Eastman 
 is wrong with regard to its general proportions (Eastman 1908a, 
 Fig. 19), as becomes evident by comparing Eastman's restoration with 
 the specimen he figured in PI. rrof his description. 
 
 Concerning the number of bones composing the dermal cranial roof, 
 1 have nothing new to add here to what is known from earlier descrip- 
 tions. The boundaries between the various bones, on the other hand, 
 I have in certain places found to be somewhat different from those 
 shown in Dean's (1901) and Eastman's (1908a) restorations. But 
 my own restoration (Fig. 15) does not claim to be fully accurate 
 in this respect, as the material investigated by me was both very 
 limited and imperfectly preserved with regard to the dermal cranial 
 roof. 
 
 Fig. 13. Macropetalichthys rapheidolabis 
 
 Hypothetical shape of brain from the left side in relation to the shape of the 
 cavum cerebrale cranii. Brain with a dark tone, outlines of the cavum cerebrale 
 indicated by a continuous line. X24- 
 
 hyp) hypophysis and other ventral parts of the diencephalon. Other letters 
 of reference as in text fig. 12. 
 
 As in a paper which I expect to publish later, I think I can show 
 that the dermal bones of the cranial roof in Arthrodires have evolved 
 independently of those in the Telcostomi, or, to put it differently, that 
 they are not directly comparable to those in the Teleostomi, I have let- 
 tered them here in text fig. 15 with indifferent letters. 
 
 Each plate 5" has, posteriorly, close to the posterior border, a ver- 
 tically descending lamina which is rigidly fixed to the cranio-spinal 
 process of its side. In the specimens investigated by me, this lamina 
 is unfortunately very imperfectly preserved, so that its extension and 
 relations could not be established with certainty. In M. pelmensis it 
 is, however, well-preserved, and it seems there to cover the posterior 
 surface of the cranio-spinal process and to be very large and meet its 
 fellow of the opposite side in the median line. 
 
Macropetalichthyids and other Arthrodires — Stensio 143 
 
 Hennig in his description of M. pehncnsis, did not perceive the 
 true nature of this lamella, but considered it to be a paired lateral 
 occipital (Hennig 1907, pp. 585-586, Fig. 3, OL) 1 . This ventral lamella 
 is necessary for the support of the much lengthened, posterior parts of 
 the dermal cranial roof. A parallel to this is found in sturgeons 
 (both Polyodontids and Acipenserids) in which a lamella issues from 
 each extrascapular bone downwards in such a way that it covers and 
 is fixed to the anterior surface of the cranio-spinal process of its side. 
 Accordingly, it is not homologous with that in Macro petalichthys. 
 In Saurichthys, in which also a strong cranio-spinal process, very sim- 
 ilar to that in sturgeons, occurs, a descending lamina from the dermal 
 cranial roof is, on the contrary, entirely lacking (Stensio 1925). 
 
 The plate M is said by Eastman (1908a, p. 104; 1908b, p. 169) 
 to be pierced by a fine canal just above the canal par in the roof of the 
 primordial neurocranium. The material investigated by me was not 
 satisfactorily preserved at this place, and I have therefore not been 
 able to verify the correctness of Eastman's statement as to this point. 
 
 Fig. 14. TlTANICHTHYS AGASSIZI, NEWB. 
 
 Dermal cranial roof and infraorbital plate. After Eastman (1008a. fig. 29). 
 Sensory canal grooves omitted. Much diminished. Pi, pineal plate; R, rostral 
 plate; orb, orbital opening. 
 
 DERMAL BONES OF THE CHEEK 
 No remains of dermal bones were found which, with certainty, be- 
 longed to the cheek, and it is rather probable that these bones were 
 either much reduced or entirely lacking in the fish. That the plate P lf 
 situated latero-ventrally of the orbital opening, is no infraorbital bone, 
 is evident both from the fact that the eye has shifted a rather consider- 
 able distance postero-medially and from the course of the sensory canals, 
 as we shall understand from the account of these given below. 
 
 'The description which Hennig (1007) gives of M. pelmensis is also in most 
 other respects very incorrect, the anatomical features of the fish having been 
 for the most part entirely misunderstood. 
 
144 Field Museum of Natural History — Geology, Vol. IV. 
 
 VISCERAL SKELETON 
 
 Remains of the visceral skeleton have hitherto not been found. 
 Despite this we are, however, able to draw certain conclusions about it. 
 
 The palatoquadrate was not coalesced with the primordial neuro- 
 cranium, but was an independent element. As far as we can judge, 
 it did not articulate with the primordial neurocranium, either anteriorly 
 or posteriorly, and, if this be true, it must have been suspended merely 
 by ligaments below the ethmoidal region and by means of the hyoid arch 
 below the labyrinth region. Accordingly, its relations to the primordial 
 neurocranium seem to have been about as in sturgeons and recent sharks 
 (except the Notidanide, in which, as known, there exists a postorbital 
 articulation between the palatoquadrate and primordial neurocranium). 
 
 cpp 
 
 Fig. 15. Macropetalichthys rapheidolabis 
 
 Restoration of the dermal cranial roof. Course of the sutures probably in 
 places not fully correct. Sensory canals marked with double dotted lines and 
 colors. Ornament of the bones omitted. XM- 
 
 Li, L 2 , Ls, Mi, M 2 , Pi, P2, S, dermal bones of the cranial roof ; c.pp, sensory 
 canal commissure, probably corresponding to the posterior head lines of pit origin 
 in recent fishes (red) ; ifc, infraorbital sensory canal (blue) ; Ic, cephalic divi- 
 sion of the lateral line( red) ; olf, nasal capsule; orb, orbital opening; poc, pre- 
 opercular sensory canal (yellow) ; soc, supraorbital sensory canal (green). 
 
Macropetalichthyids and other Arthrodires — Stensio 145 
 
 Probably there was in relation to the palatoquadrate, a large musculus 
 preorbitalis (Cf. the description of the ethmoidal region above). 
 
 If my interpretation of the canal c.hy as the canal for the vena 
 hyoidea is correct, it is obvious that the hyoid arch articulated against 
 the primordial neurocranium closely anterior to the external opening 
 of this canal, or, at least that the hyoid arch, if it did not articulate 
 against the primordial neurocranium, had its dorsal end situated closely 
 anterior to this place, for it must of course have been situated an- 
 terior to the hyoid vein. Or to express it in another way, the hyoid 
 vein must have passed upward to the jugular vein along the pos- 
 terior side of the hyoid arch. Whether the dorsal end of the hyoid 
 arch was situated ventrally of the vena jugularis, as in Selachians, or 
 dorsally of the same vein as in the Teleostomi, it is impossible to decide 
 from the specimens investigated (Cf. Allis 1915, 1918b). 
 
 Of much interest in this connection is the fact that the jugular 
 vein as well at the upper end of, as anterior and posterior to the hyoid 
 arch, went within the lateral wall of the primordial neurocranium and 
 that, accordingly, a shifting of the place of articulation of the dorsal 
 end of the hyoid arch against the primordial neurocranium could take 
 place in a dorsal or ventral direction without affecting the jugular vein. 
 As the jugular vein in Eusthenopteron 1 also passed within the lateral 
 wall of the primordial neurocranium to about the same extent as in 
 Macropetalichthys and probably did also in other Crossopterygians 
 (Cf. Stensio 1922, p. 201, the long intramural course of the truncus 
 hyoidemandibularis facialis) one suspects that this was originally the 
 case in all fishes. 
 
 If this suspicion be true, the mainly extramural course of the jugu- 
 lar vein in most fishes would be due to reduction of certain lateral 
 parts of the lateral walls of the primordial neurocranium. And, more- 
 over, it would be very easy to understand how the place of articulation 
 of the hyoid arch could have shifted upwards to the dorsal side of 
 the jugular vein in the Teleostomi. Finally, the hyomandibular of the 
 Teleostomi, would not, as Allis (1915, 1918b) supposes, have incorpo- 
 rated at its dorsal end certain parts lacking as such in the hyomandib- 
 ular of the Elasmobranchs. The conditions found in Macropetalich- 
 thys and Eusthenopteron with regard to the vena jugularis and the vena 
 hyoidea thus seem to make certain of the views recently advanced by 
 Allis (1915, 1918b) concerning the homologies of the hyomandibular 
 of fishes very doubtful. 
 
 'Observation made on a well preserved specimen in the collection of the 
 Palseontological Institution of Upsala. 
 
146 Field Museum of Natural History — Geology, Vol. IV. 
 
 The first branchial arch followed not very far behind the hyoid 
 arch. On account of the cranio-vertebral articulation found in most 
 or all Arthrodires, there is no reason to believe that the branchial basket 
 extended backwards into the abdominal region of the trunk, but was 
 limited to the suboccipital part q£ the head. 
 
 DENTITION 
 Unfortunately the dentition is so far entirely unknown. 
 
 SENSORY CANALS OF THE HEAD 
 The sensory canals are found preserved on the head only in the 
 dermal cranial roof. They consist there of tubes which are situated 
 within the bones, or, more exactly, in ridge-like thickenings on the 
 lower side of the bones (Figs. 4, 8, 9). Accordingly they have a 
 very deep position (Cf. Kaer 1915, p. 12), a fact which explains that 
 there are frequently traces of them preserved even in specimens in 
 which the dermal bones have almost entirely weathered away, as, for 
 instance, in the specimen reproduced in Pis. XIX, XX. At least in 
 the larger parts of their extension they open outwards with a double 
 series of fine tubuli. These tubuli occur in great number and are 
 situated fairly close to each other, conditions which seem to indicate 
 that they were much more numerous than the sensory canal organs. 
 In any case they are much more numerous than the nerve-canals lead- 
 ing to the sensory canals. , 
 
 On account of the fortunate circumstance that the fine canals for 
 the nerve branches to the sensory canal organs are lined by a thin 
 layer of bone, they could be preserved, and thus enable us to get a 
 rather complete knowledge of the manner in which the sensory canals 
 were innervated. With the help of this knowledge it is not difficult 
 to homologize the sensory canals in Macropetalichthys rapheidolabis 
 with those in recent fishes, as will be evident from the subsequent ac- 
 count. With regard to the terminology, I shall follow Allis (1889; 
 1900; 1901 ; 1903; 1905; 1923) with the exception only that in accord- 
 ance with Ewart (1892), Ewart and Cole (1895), Ewart and 
 Mitchell (1892), and Cole (1896a, b; 1898) I limit the infraorbital 
 sensory canal to those parts innervated by the n. buccalis lateralis and 
 n. oticus lateralis. 
 
 The supraorbital sensory canal (soc. Figs. 8, 9, 15; Pis. XIX, 
 XX, XXI, XXIII) has in its present state of preservation its an- 
 terior end at the anterior border of the bone-plate L lt but in the living 
 fish it continued forward into the soft tissue anteriorly and laterally 
 to this plate. How it may have passed in this soft tissue it is of course 
 
Macropetalichthyids and other Arthrodires — Stensio 147 
 
 not possible to say positively, but it seems to me most probable that 
 it was there as in Coccostens, and forms allied to this (Cf. Jaekel 1902 
 fig. to page 107; 1906a, text figs. 2, 3), i.e., that it very soon turned 
 backwards and went dorsally and behind the nasal opening down to 
 the infraorbital canal. It would thus, if I am right, have had the 
 same course in its anterior parts as in Elasmobranchs. (Cf. Garman 
 1888; Allis 1901, pp. 104-107, 1 18-128; 1905, p. 419, pp. 470-478; 
 1923; Ewart 1892, pp. 66-74; Hawkes 1906, pp. 960-961, PI. 68; cf. 
 also Cole 1896 a, b.) Within the bone-plate L x it runs first backward 
 and slightly laterally to about the center of the ossification of the plate, 
 then fairly suddenly curves and takes a backward and slightly medial 
 direction. With this direction it continues backward to the plate M 2 , 
 which it traverses to about the center of ossification, and here joins its 
 fellow of the opposite side and a cross comissural sensory canal between 
 the cephalic division of the lateral lines in the way shown in text fig. 
 15. No anastomosis is found between it and the posterior part of the 
 infraorbital canal of its side. 
 
 To the supraorbital sensory canal go several fine branches from 
 the canal for the n. ophthalmicus lateralis. The number of these 
 branches could not, however, be established. The posterior ones issue 
 dorso-postero-medially, the anterior ones antero-dorso-laterally. In 
 Pis. XXI and XXIII one anterior branch (rm.ophth.lat) is clearly 
 seen issuing from the canal for the n. ophthalmicus lateralis (c.ophth. 
 lat) to the supraorbital sensory canal (soc). 
 
 Its considerable extension backwards and its direction in the pos- 
 terior parts, indicate with certainty that the supraorbital sensory canal 
 comprises not only the supraorbital sensory canal as found, for in- 
 stance, in Amia and many other fishes, but also the homologue of the 
 anterior head line of pit organs of these fishes. (Cf. Allis 1889, pp. 
 505-506; 1900, pp. 445; 1903, p. 187; 1905; Herrick 1901, pp. 222- 
 223; Stensio 1921, pp. 218, 263; 1922, p. 192; 1923, p. 1258; 1925.) 
 
 The infraorbital sensory canal (ifc, Figs. 8, 15; Pis. XIX, XX, 
 XXII; PI. XXIV, fig. 1) has its posterior end situated at about 
 the ossification center of the plate P 2 ; being there continuous with the 
 cephalic division of the lateral line of its side. From this place it runs 
 forward and slightly medially into the plate P lt retaining the same di- 
 rection as behind this and being during this part of its course situated 
 latero-ventrally of the orbital opening. It does not, however, traverse 
 the bone-plate P x , in the entire length, but goes to the lateral border 
 of this some distance behind the anterior end as is shown in text fig. 
 15. After leaving the cranial roof at this point, it must have passed 
 
148 Field Museum of Natural History — Geology, Vol. IV. 
 
 forward and downward on the cheek, probably through soft tissue, and 
 it is likely that it extended as far forward as to the rostrum, and, that 
 it, as pointed out, anastomized with the supraorbital canal behind the 
 nasal opening. 
 
 To the part of the infraorbital canal situated within the bone- 
 plates P 1 and P 2 lead a few fine canals which pierce the lateral parts 
 of the postorbital process and which very probably transmitted lateralis 
 branches from a nerve corresponding to the r. oticus lateralis of recent 
 fishes. Two of these branches (in) are partly seen in Pis. XX, XXII, 
 and it is easy to understand from their direction that they in any case 
 must have been traversed by lateralis branches that were given off from 
 a prootic nerve, i.e., a nerve that had its exit through the cranial wall 
 anterior to the labyrinth. 
 
 Before leaving the infraorbital canal, it is finally necessary to point 
 out that its course in the dermal cranial roof is incorrect in the restora- 
 tion published by Dean (1901, text fig. 12). 
 
 The cephalic division of the lateral line (Ic, text figs. 4, 15; Pis. 
 XIX, XX, XXIV) enters the dermal cranial roof at the posterior 
 border of the bone-plate S, traverses this bone-plate in anterior and 
 slightly medial direction, and continues with the same direction to the 
 ossification center of the bone-plate L 3 . There it suddenly takes an 
 antero-lateral course and goes with this to the ossification center of the 
 bone-plate P 2 , joining there, as already mentioned, the infraorbital canal. 
 
 At the place in the bone-plate L,, where the cephalic division of 
 the lateral line turns laterally, it gives off an antero-medial branch 
 which goes to the ossification center of the plate M 2 , joining there with 
 its fellow of the opposite side and forming in this way an angulated 
 transverse commissure (c.pp, text figs. 4, 15; Pis. XIX, XX; PI. 
 XXIV, fig. 3). At the angle in the median line this commissure 
 anastomozes with the posterior end of the supraorbital sensory canals 
 just after these have joined one another. 
 
 A short anterior portion of the cephalic division of the lateral line 
 just behind the infraorbital sensory canal was probably innervated by 
 the presumed lateralis branch that traversed the cranial wall, together 
 with the n. glossopharyngeus. The remaining part of the division 
 anterior of the transverse commissure, as well as the adjacent half 
 of this commissure, was supplied by a nerve that issued from 
 the n. linese lateralis already within the vagus canal and from there 
 went upward through the canal d x , (PL XXIV, fig. 2), and its 
 branches d xa and d xb (PL XXIV, fig. 2), thus by a ramus supra- 
 temporalis from the n. lineae lateralis. Finally, the part of the division 
 
Macropetalichthyids and other Arthrodires — Stensio 149 
 
 behind the transverse commissure was innervated by a number of fine 
 branches which independently of each other issued from the n. lineae 
 lateralis, each of them probably supplying only a single sense organ. 
 In the anterior division of the occipital region where these nerve 
 branches pierced the primordial neurocranium, six of these canals are 
 seen in the Field Museum specimen (Id^ld* Pis. XIX, XX ; PI. XXIV, 
 fig- 3). 
 
 In the description of the anterior division of the occipital region it 
 was pointed out that a fine branch from the n. lineae lateralis prob- 
 ably went up to the dorsal surface of the primordial neuro- 
 cranium through the canal cl (text fig. 3; Pis. XIX, XX). This fine 
 lateralis branch thus had, if it really existed, such a course that it 
 must have innervated one or a few sensory canal organs situated 
 medially of the cephalic division of the lateral line posterior to the 
 cross commissural canal c.pp, but, as it is not a closed sensory canal 
 there, the presumed sensory canal organs supplied by it would have 
 formed a short, transverse pit-line. Unfortunately the material inves- 
 tigated did not enable me to decide whether there was a groove for 
 such a pit-line on the dorsal side of the dermal bones. The possible 
 non-existence of the groove need not, however, indicate absolutely 
 the absence of the pit-line, for this may in such a case very well be 
 thought to have been situated entirely in the skin and thus to have 
 been wholly unrelated to the underlying dermal bones. 
 
 If the pit-line in question really could be shown to exist, it would, 
 according to its position and innervation, evidently be homologous with 
 the supratemporal commissure in recent fishes, while the cross com- 
 missural canal c.pp would represent the posterior head line of pit organs 
 in these, a view which is also supported by the position and direction 
 of each half of the latter canal. (Cf. Allis 1889, PI. 42 ; 1901, pp. 101- 
 103, 116-117, PI. 10; 1903, pp. 186-187, PI. Ill; 1905, pp. 471, 474; 
 Ewart 1892, PI. II, fig. 2; Stensio 1923, p. 1258, fig. 4.) Below, 
 in the description of Epipetalichthys, we shall find additional facts 
 which point in the same direction. 
 
 According to the restoration published by Dean (1901, text fig. 
 12) a canal, which, in its position, corresponds very well to the 
 preopercular canal, issues in latero-ventral and slightly posterior di- 
 rection from the point where the infraorbital canal joins the cephalic 
 division of the lateral line. This canal is not found in Eastman's 
 restorations in 1908 (Eastman 1908a, text fig. 19; 1908b, text fig. 24), 
 and the material investigated by me was not preserved sufficiently well 
 to show with certainty whether Dean or Eastman is right in this 
 
150 Field Museum of Natural History — Geology, Vol. IV. , 
 
 respect. As the canal in question is found in a specimen described as 
 Macro pctalichthys agassizif below, and also occurs in most other 
 Arthrodires, it seems very likely that it is not missing in Macropeta- 
 lichthys rapheidolabis. On account of this it was drawn in text fig. 15, 
 in which it is designated by the letters foe. If. it really was present, it 
 must have continued down to -the cheek, where it chiefly, or, perhaps, 
 even entirely, had its passage through soft tissue. 
 
 REMARKS ' 
 
 On account of the flat, broad shape of the primordial neurocranium 
 as a whole and the position and direction of the orbital openings, there 
 can be no doubt that M. rapheidolabis, like other species of Macropeta- 
 lichthys must have been a decided benthonic form. 
 
 The description given here makes it fully evident that many of the 
 strange anatomical features which Eastman thought he had observed 
 in M. rapheidolabis are mistakes. 
 
 MACROPETALICHTHYS AGASSIZI? (H. v. Meyer) 
 (PI. XXIX, figs, i, 2) 
 
 Through the courtesy of Mr. S. Junkermann of Bielefeld, Ger- 
 many, I had the opportunity of examining an imperfectly preserved 
 neurocranium of a Macropetalicthyid, which, as far as it is preserved, 
 agrees so well with M. rapheidolabis, that it at least provisionally seems 
 correct to refer it to the genus Macro pctalichthys. Among the species 
 described so far of this genus it seems to resemble most closely M. 
 agassizi, and on this account I have, with some hesitation given it this 
 specific name. 
 
 The primordial neurocranium is imperfectly preserved, being rep- 
 resented only by certain parts of the occipital region (PI. XXIX, fig. 
 1), which are much fractured and crushed. As far as can be seen, 
 it seems to have been less ossified than that of M. rapheidolabis, but 
 at least in the posterior, narrow division of the occipital region, I found 
 both the internal and external bone layers. Further forward the in- 
 ternal bone layer appears to have been entirely lacking. With regard 
 to its general shape, the primordial neurocranium seems to have con- 
 siderably resembled that in M. rapheidolabis. 
 
 In text fig. 16 we see two transversal sections through the most 
 anterior part of the posterior narrow division of the occipital region. 
 As is evident from these sections, the haemal groove did not extend so 
 far forward as in M. rapheidolabis, and it is further worthy of notice, 
 that, contrary to what is the case in that species, there was no bone 
 
Macropetalichthyids and other Arthrodires — Stensio 151 
 
 tube around the notochord. In the anterior one of the two sections 
 figured, the postero-lateral-dorsal corners of the anterior broad division 
 of the region are seen in section, as they are extended considerably in 
 postero-lateral direction (pr, text fig. 16a). Of the dermal cranial 
 roof we find the plates M 2 , L 3 , P % and 5" (text fig. 16a; PI. XXIX, 
 fig. 2) more or less well-preserved. Their ornaments consist of small, 
 numerous, rather densely placed tubercles, which are stellate at their 
 bases. 
 
 The sensory canals are closed tubes situated deeply in the dermal 
 bones, as in M. rapheidolabis. They open outward with numerous, 
 straight, unbranched tubuli, but while these tubuli in M. rapheidolabis 
 are arranged in a double series, they form here only a single one from 
 each canal (PI. XXIX, fig. 2). It is quite probable that the canals for 
 
 ,cr.od 
 
 «— er.ol 
 
 Fig. 16. Macropetalichthys agassizi? 
 
 Transverse section through the most anterior part of the posterior narrow 
 division of the occipital region. The section A taken only a short distance an- 
 terior to the section B. Cartilage dotted, layers of substitution bone with strong 
 black lines, dermal bones of the cranial roof with vertical striation. X'/s 
 
 cr.od, crista occipitalis dorsalis ; cr.ol, crista occipitalis lateralis ; ch, probable 
 position of the space for the notochord ; cv, cavum cerebrale cranii ; Ic, cephalic 
 division of the lateral line; pr, postero-laterally projecting corner of the anterior 
 broad division of the occipital region. 
 
 nerves leading from below to the sensory canals were much fewer in 
 number than the tubuli, a fact which enables us to conclude with a 
 rather great degree of probability, that the sensory canal organs did 
 not correspond in number to the tubuli but were much less numerous 
 than these. 
 
152 Field Museum of Natural History — Geology, Vol. IV. 
 
 MACROPETALICHTHYS PELMENSIS, Hennig 
 
 The description given by Hennig (1907) of this species is, as I 
 found by examination of his type specimen, in several respects very 
 incorrect. 
 
 What Hennig (1907, pp. 585-586) takes to be a paired lateral 
 occipital seems in fact to be a ventral, paired process from the bone- 
 plate S, of the dermal cranial roof, according to the terminology em- 
 ployed by me in the present paper. This process, which meets its 
 fellow of the opposite side in a median suture and extends very low 
 downwards, covers the posterior surface of the probably very large 
 cranio-spinal process of its side. 
 
 The structure of the dermal cranial roof was also in other respects 
 very much misinterpreted by Hennig, a circumstance which is, how- 
 ever, so obvious that it will not be necessary to deal with it here. 
 
 Eastman's description of the species in 1908 (1908b, pp. 176-177) 
 is, with regard to the posterior parts of the neurocranium, also very 
 erroneous, as Eastman on that occasion did not even correctly recog- 
 nize the main divisions of the neurocranium, a fact which lead him to 
 certain very strange conclusions about the position of the ductus endo- 
 lymphaticus and the relations between this and the sensory canal sys- 
 tem (Cf. Kaer 1915, pp. 11-12). 
 
 It may finally be emphasized in this connection that the primordial 
 neurocranium probably is preserved in the specimen, and that a further 
 preparation of it would therefore be very desirable. 
 
 EPIPETALICHTHYS WILDUNGENSIS, Jaekel, gen. nov., sp. nov. 
 
 PI. XXVIII, figs. 1, 2; PI. XXIX, figs. 3, 4; PI. XXX, fig. 1; 
 PI. XXXI, fig. 2. 
 
 This species, which is a new one and also represents a new genus, 
 is from the upper Devonian of Wildungen, Germany. It was placed 
 at my disposal by Professor O. Jaekel of Greifswald. On the labels 
 accompanying it Professor Jaekel had named it Epipetalichthys wild- 
 ungensis. So far, only a single specimen, represented by the imper- 
 fectly preserved neurocranium, has been found of it. To Judge from 
 this specimen, the species must have attained about the same size as 
 Macro pet alichthys rapheidolabis. 
 
 PRIMORDIAL NEUROCRANIUM 
 General remarks. The primordial neurocranium, if we first turn 
 to this, is partly rather badly crushed and fractured, and, in addition, 
 also, weathered in places, so that it is not by far as well adapted for a 
 
Macropetalichthyids and other Arthrodires — Stensio 153 
 
 detailed examination as that in the Field Museum specimen of Macro- 
 petalichthys rapheidolabis. In its general shape it must obviously have 
 resembled that of the last-mentioned species, but, as we shall see, it 
 differs distinctly from this in certain details. 
 
 Its degree of ossification was considerably less than that in M. 
 rapheidolabis and less also than in M. agassisif, a fact which deserves 
 a special attention, as we are here concerned with an upper Devonian 
 form, while the other two just mentioned specimens are both from 
 earlier divisions of the Devonian. At least this is beyond question the 
 case with M. rapheidolabis, which occurs only in middle Devonian strata. 
 
 On account of the imperfect state of preservation in which the 
 primordial neurocranium is found, it is not possible to make out with 
 full exactness the extension of its different layers of substitution bone, 
 but the following may give an idea of the most important differences 
 from M. rapheidolabis in this respect. 
 
 1. There is no evidence of the labyrinth bone layer. 2. The inner 
 bone layer seems to be entirely lacking, except in the posterior, narrow 
 division of the occipital region (text fig. 17). 3. The external bone 
 layer surrounds the posterior, narrow division of the occipital region, 
 and extends from this forward also on to the lateral and ventral sides of 
 the anterior division of the same region and the ventral side of the 
 labyrinth region. Whether, however, it continued from the latter region 
 to the ventral side of the orbitotemporal region, cannot be decided, as 
 the ventral parts of this region have been destroyed. On the ethmoidal 
 region it is, however, found again on the ventral side and, in addition, 
 at least partly, on the lateral sides and probably also on the anterior 
 end. Accordingly, it is lacking on the larger anterior part of the dorsal 
 side, quite as in M. rapheidolabis, but, contrary to what is the case in 
 that form, also probably on the lateral sides of the entire labyrinth and 
 orbitotemporal regions and the posterior side of the ethmoidal region. 
 If it was lacking on the ventral side of the orbitotemporal region, this 
 region would obviously have consisted entirely of cartilage. 4. The 
 canal layers seem to have become entirely reduced. 
 
 As far as I can see, there are no traces of ossification centers either 
 in the occipital region or in the ethmoidal region. 
 
 Microscopic Structure of the Bone 
 The substitution bone of M. rapheidolabis has generally been much 
 infiltrated by a dark bitumen that is difficult to remove, and, accord- 
 ingly, no good microscopic sections can be made from it. I could, 
 however, see rather distinctly that it has numerous cell spaces (PI. 
 XXXI, fig. 1). In the species under consideration the bone tissue is 
 
i54 Field Museum of Natural History — Geology, Vol. IV. 
 
 also not favorably preserved for microscopic investigation, but it is 
 fully obvious that it there too contained cell-spaces in fairly large 
 number. In the posterior division of the occipital region we can 
 further see that from the external and internal bone layers, numer- 
 ous, irregular trabecular (PI. XXXI, fig. 2) extend a short distance 
 into the parts of the cranial walls that were occupied by cartilage, so 
 that there really is certain evidence of cancellous bone, although this 
 bone was very slightly developed. By means of short bone trabecular 
 of about the same kind as those just mentioned, the bone layer sur- 
 rounding the notochordal space (ch.l, PI. XXXI, fig. 2) is connected 
 with the internal and external bone layers. Otherwise, no details can 
 be observed. 
 
 Occipital Region 
 
 As in M. rapheidolabis and M. agassizi? the occipital region is 
 composed of two divisions, a posterior, long and narrow one, and an 
 anterior, short and broad one. 
 
 ft 
 
 A cent 
 
 Fig. 17. Epipetalichthys wildungensis Jaekel 
 
 Transverse section through the anterior half of the posterior division of the 
 occipital region. Layers of substitution bone with strong black lines, cartilage 
 dotted. The trabecular of substitution bone penetrating inwards into the cartilage 
 from the bone layers not drawn. XVs 
 
 ham, haemal groove; ch, space for the notochord; cv, cavum cerebrale cranii. 
 
 The posterior one (text fig. 18; PI. XXVIII, fig. 2) is represented 
 merely by an anterior part, and no certain statement can therefore be 
 given of its length. It is narrow and, contrary to that in M. rapheido- 
 labis and M. agassizi f, it equals the other divisions of the primordial 
 neurocranium in height. Further, it differs from that in the two species 
 mentioned in the fact that it does not have the slightest traces of the 
 crista occipitalis lateralis, which in them is such an important feature. 
 The haemal groove (hcem, text figs. 17, 18; PI. XXVIII, fig. 2) is 
 rather distinct and it is noteworthy that we find in the part of it pre- 
 
Macropetalichthyids and other Arthrodires — Stensio 155 
 
 served two rather shallow depressions (seen in text fig. 18), which, 
 perhaps, may be segmentally arranged. Otherwise I only wish to 
 emphasize that the division was narrowest anteriorly at the transition 
 to the anterior division, gradually increasing somewhat in breadth back- 
 ward from this place. 
 
 A transverse section through the part of the division present (text 
 fig. 17) shows that the cavum cerebrale cranii (cv) within this part 
 was very high and narrow and that there was below the cavum cere- 
 brale a space for the notochord (ch). This space is, like that in Macro- 
 pctalichthys raphcidolabis, bounded by a bone layer of its own, a bone 
 layer that is continuous dorsally with the inner bone layer (bottom of 
 the cranial cavity), and ventrally with the external bone layer (bottom 
 of the haemal groove), as already pointed out above (PI. XXXI, fig. 2). 
 
 No traces of canals for spino-occipital nerves could be seen in the 
 part of the division present. 
 
 --i--c.rv. 1 
 
 Fig. 18. Epipetalichthys wildungensis 
 
 A posterior part of the primordial neurocranium from the central side. X$4- 
 c.rai, c.rat, anterior and posterior openings respectively of the canal for the radix 
 
 aorta: (lateral dorsal aorta); ham, haemal groove; s.ra, groove for the radix 
 
 aortae lateral dorsal aorta, anterior of the canal era. 
 
 Of the anterior broad division of the region only the ventral parts 
 are accessible for investigation (text fig. 18; PI. XXVIII, fig. 2). As 
 far as can be seen from these its shape must have been very much the 
 same as in Macropetalichthys rapheidolabis. 
 
 With regard to the details, we find that in this division the pos- 
 terior part of the groove for the radix aortae forms a closed canal in 
 the primordial neurocranium. This canal has its anterior and posterior 
 openings (c.ra lt c.ra 2 ) situated as shown by text fig. 18 and PI. XXVIII, 
 fig. 2. 
 
 Nothing is so far known concerning the vagus canal and its branches. 
 
 Labyrinth and Orbitotemporal Region 
 The labyrinth region (PI. XXVIII, fig. 2) is somewhat longer than 
 in M. rapheidolabis. It is very fragmentarily preserved, only parts of 
 the external bone layer of its ventral side being left. 
 
156 Field Museum of Natural History — Geology, Vol. IV. 
 
 The orbitotemporal region is still more imperfectly preserved, and 
 the only statement that can be made about it is that it certainly was 
 rather short and that the orbits are situated far apart and in such a 
 way that the eyes must have been directed much upwards, quite as in 
 M. rapheidolabis. 
 
 Ethmoidal Region 
 
 On account of the fact that, as pointed out, it has the external bone 
 layer persisting to a considerable extension, the ethmoidal region has 
 been fairly well preserved. A large part of its ventral side, an anterior 
 part of its lateral side, an anterior part of its dorsal side and the 
 anterior end are available for investigation (text fig. 19; PI. XXVIII, 
 fig. 1; PI. XXIX, fig. 3; PI. XXX, fig. 1). 
 
 From its broad, posterior end, it becomes rapidly narrow forward, 
 but does not taper to a point, its anterior end being truncated and 
 rather concave. The breadth at this end is about one-third of that at 
 the posterior end. 
 
 From the antero-ventral part of each lateral side the olfactory cap- 
 sule (olf, text fig. 19; PI. XXIX, fig. 3; PI. XXX, fig. 1) projects 
 laterally exactly as in M. rapheidolabis. Unfortunately it is not so well 
 preserved as in that species, its anterior and external parts being prac- 
 tically destroyed, and on account of this nothing can be said with cer- 
 tainty about its external opening and whether there was any nasal fon- 
 tanels on the ventral side of its lateral part. 
 
 The ventral surface of the region (PI. XXIX, fig. 3) is rather 
 concave in transversal direction and has in the postero-lateral part 
 somewhat medially of the lateral margin a longitudinal, shallow groove 
 (s.pal) for the r. palatinus facialis, quite as in M. rapheidolabis. Each 
 one of the lateral surfaces (PI. XXVIII, fig. 1), is concave in dorso- 
 ventral direction, and has a long post-nasal portion as in M. rapheidola- 
 bis. The dorsal surface (text fig. 19; PI. XXX, fig. 1) is exposed 
 only most anteriorly, the most anterior of the bones in the dermal 
 cranial roof being weathered away. We find in this part of it a very 
 deep, anteriorly and dorsally open depression (pfg), which, as far as 
 can be judged from the conditions in the fossil, seems to have been 
 lined by the external bone layer. 
 
 This depression, which, as already pointed out in the description 
 of M. rapheidolabis, is obviously the homologue of the similarly 
 situated depression in the skulls of Elasmobranchs, and which therefore 
 may be properly called the cavum precerebrale, was, contrary to that 
 in these fishes, separated from the cavum cerebrale by a thick septum 
 of cartilage, presumably lined with the external bone-layer on the an- 
 
Macropetalichthyids and other Arthrodires — Stensio 157 
 
 terior side. There is thus no fenestra precerebralis. Laterally the 
 cavum cerebrale is also well separated from the nasal capsules. Its 
 floor is not perforated by any foramina. 
 
 Cavum Cerebrale 
 
 Since the internal bone layer, as pointed out, is lacking-, except in 
 the posterior division of the occipital region, the shape of the cavum 
 celebrale is known merely in this, where it, as seen from text fig. 17, 
 is very high and narrow. 
 
 The canal for the pineal and parietal organs seems not to have per- 
 forated the dermal bones of the cranial roof. At least no traces of 
 such a perforation could be seen in the parts of the dermal bones 
 preserved. 
 
 DERMAL BONES OF THE CRANIAL ROOF 
 
 Parasphenoid or other dermal bones were not present on the ven- 
 tral side of the primordial neurocranium. 
 
 The dermal bones of the dorsal side .of the primordial neurocranium 
 (text fig. 19; PI. XXX, fig. 1) are the same as in M. rapheidolabis, 
 and their extensions and relations, too, are about the same as in this, 
 with the exception, however, that the plates M t and M 2 do not meet, 
 but are separated by the plates L lf which extend to and suture with 
 one another in the median line. 
 
 The plate L s is, in the posterior part, perforated by a rather large, 
 oblong foramen (d.end), which certainly must be the external opening 
 of the canal for the ductus endolymphaticus. This foramen lies defin- 
 itely behind the place where in M. rapheidolabis the fossa endolym- 
 phatica is situated, and it seems therefore, rather probable that the duc- 
 tus endolymphaticus from this fossa extended somewhat backward be- 
 neath the dermal cranial roof, thus making a bend and continuing back- 
 ward quite as it does in Chlamydoselachus (Goodey 1910, PI. 43, figs. 
 7, 8) and several other Elasmobranchs (Retzius 1881). In this con- 
 nection it deserves also to be emphasized that the position of the ex- 
 ternal opening of the ductus endolymphaticus in the form under con- 
 sideration has, in relation to the sensory canal system, exactly the same 
 position as in Chlamydoselachus (Cf. Allis 1923, PI. II). 1 
 
 The ornament on the dermal bones of the cranial roof consists of 
 short, rather low, but nevertheless distinct, ridges, which, on the plates 
 5" and M 2 , seem to lie with their longitudinal axis about parallel with 
 
 1 In Chlamydoselachus the ductus endolymphaticus has its external opening 
 posterior of the supratemporal commissure of the main lateral lines; in Acanthias, 
 Mustelus and others, on the contrary, it has this opening anterior of the same 
 commissure (Cf. Allis 1901, PI. 10). 
 
158 Field Museum of Natural History — Geology, Vol. IV. 
 
 the longitudinal axis of the head. On the other bones, excepting the 
 plate M lt on which their arrangement is unknown, the ridges are, on 
 the contrary, arranged concentrically with the edges of the bones (PL 
 XXIX, fig. 4). In the Macropetalichthys species known so far, the 
 ornament consists of fairly sparse tubercles, which are, in the main, 
 irregularly scattered. 
 
 THE SENSORY CANALS OF THE HEAD 
 The sensory canals of the head are known merely in the bones of 
 the dermal cranial roof, in which they form closed canals. They are 
 finer than in M. rapheidolabis ; and the ridges for them on the lower 
 
 Pff 
 
 Fig. 19. Epipetalichthys wildungensis 
 
 Head in dorsal view with the anterior parts of the bone plate Mi removed to 
 show the cavum precerebrale. The sensory canals marked with dotted outlines 
 and colours. XJ4- 
 
 Li, L 2 , L 3 , Mi, Mi, Pi, P2, S, dermal bones of the cranial roof (the anterior 
 larger part of the plate Mi, removed) ; d.cnd, external opening of the canal for 
 the ductus endolymphatic^ ; ifc, infraorbital sensory canal (blue); Ic, cephalic 
 division of the main lateral line (red); na, nasal aperture; olf, nasal capsule; 
 orb, orbital entrance ; pfg, cavum precerebrale ; st.com, supratemporal commissure 
 between the cephalic division of the main lateral lines (red) ; soc, supraorbital 
 sensory canal (green). 
 
Macropetalichthyids and other Arthrodires — Stensio 159 
 
 sides of the bones are also less strong than in that species. They 
 open outwards with numerous short and generally unbranched tubuli 
 (PI. XXX, fig. 1), which, however, contrary to the case in M. rapheido- 
 labis are arranged like those in M. agassisi? and M. pelmensis in a 
 single series. 
 
 The supraorbital sensory canal (soc, Fig. 19; PI. XXX, fig. 1) 
 does not extend as far backwards as in M. rap held olabis, and, on this 
 account, does not meet its fellow of the opposite side in the median 
 line nor does it anastomose with the commissure (s.com) between the 
 cephalic divisions of the main lateral lines, but ends some distance 
 anterior of this. Behind the part where it ends as a closed canal it 
 may, however, probably have continued a short distance further back- 
 wards as a pit line. Its anterior parts are not known, but it seems 
 very probable that anterior of the most anterior part preserved in the 
 fossil it curved strongly backwards and went first dorsally and then 
 posteriorly of the nasal opening down to the infraorbital sensory canal, 
 as it does in Coccosteus and Elasmobranchs. All the tubuli of the 
 preserved portion of the canal issue in lateral direction, as is clearly 
 shown by PI. XXX, fig. 1. 
 
 The infraorbital sensory canal (ifc, Fig. 19; PI. XXX, fig. 1) 
 is, in the main, as in M. rapheidolabis. 
 
 The cephalic division of the lateral line (Ic, Fig. 19) ; PI. XXX, 
 fig. 1) has also the same course as in M. rapheidolabis, and is, as in 
 this, in communication with its fellow of the opposite side by a trans- 
 versal commissure (st.com, Fig. 19; PI. XXX, fig. 1), which 
 traverses the plate L 3 and the plate M 2 to the plate L 3 of the opposite 
 side. This commissure is, however, contrary to that in M. rapheidolabis, 
 almost straight, and, as already mentioned, it does not communicate with 
 the posterior ends of the supraorbital canals. It seems even to be 
 situated behind the point in which the prolongations of these canals 
 appear to meet each other. Consequently it has such a position and 
 shape that in M. rapheidolabis it would seem to correspond to the 
 presumed line of pit organs behind the cross commissural canal and 
 not to this canal, and it thus appears to really represent the true supra- 
 temporal commissure of fishes. If this opinion is true, the homologue 
 of the commissural canal of M. rapheidolabis would either be entirely 
 lacking or be represented by a paired line of pit organs corresponding 
 to the posterior head line of pit organs in recent fishes. (Cf. Allis 
 1889, pp. 505-506; 1903, p. 187; 1905; 1923, pp. 195-201; Herrick 
 1901, pp. 222-223; Stensio 1921, pp. 218, 263; 1922, p. 192; 1923, 
 p. 1258; 1925; etc.) 
 
160 Field Museum of Natural History — Geology, Vol. IV. 
 
 The most anterior portion of the cephalic division of the lateral line 
 situated between the supratemporal commissure and the infraorbital 
 canal, does not open outward and has no tubuli at all. The remaining 
 portion, on the contrary, is provided with numerous short tubuli, which 
 issue on the lateral side and in ^lateral direction (PI. XXX, fig. i). 
 The supratemporal commissure' also has numerous tubuli, but these 
 seem, as a rule, to perforate the bones in a straight upward direction 
 (PI. XXX, fig. i). 
 
 Whether there was a preopercular canal or not, it is at present 
 impossible to decide. 
 
 Remarks 
 
 As is obvious from this description, the form described here as 
 Epipetalichthys wildungensis is well separated from other Macropeta- 
 lichthyids known hitherto, and the erection for it of a new genus and 
 species is therefore fully justified. 
 
 SOME GENERAL REMARKS ON THE 
 MACROPETALICHTHYIDS 
 
 With the facts previously known, it has not been possible to arrive 
 at any certain conclusion concerning the relationships of the Macropet- 
 alichthyids, but from what has now been determined some deductions 
 may be made. Thus, there cannot be the slightest doubt that the 
 Macropetalichthyids are true fishes, and, moreover, it is also clear that 
 among recent forms they most resemble the Elasmobranchs. The 
 most important characters which they have in common with these 
 are the following: (i) The general shape of the primordial neuro- 
 cranium, especially the tendency to broadening of the ventral surface, 
 partly at the expense of the lateral surfaces. (This is especially the 
 case in the labyrinth region in which the sacculus and perhaps the 
 canalis semicircularis externus had about the same relation to the 
 ventral surface as in Chlamydoselachus and most other Selachians). 
 (2) The position and relations of the olfactory capsule. (3) The 
 presence of a nasal fontanelle on the lower side of the nasal capsule 
 as in Chlamydoselachus and certain other Selachians. (4) The pres- 
 ence of the cavum precerebrale. (5) The general shape of the laby- 
 rinth, especially with regard to the position of certain of its main parts, 
 as, for instance, the utriculus. (6) The presence of the ductus endo- 
 lymphaticus and the fact that there probably was a distinct fossa 
 endolymphatica on the dorsal surface of the primordial neurocranium 
 beneath the dermal bones. (7) The fact that the ductus endolympha- 
 
Macropetalichthyids and other Arthrodires — Stensio 161 
 
 ticus perforated the dermal cranial roof and had an external opening 
 situated as in certain primitive Selachians (Chlamydoselachus). (8) 
 The general shape of the brain as far as this can be restored from the 
 exit of the nerve canals and the shape of the cavum cerebrale. (9) To 
 a certain extent the course and arrangement of the blood-vessels and the 
 presence of certain important trunks, as, for instance, the vena hyoidea. 
 (10) The fact that the palatoquadrates as far as can be judged did not 
 articulate with the ethmoidal region but must have been suspended be- 
 low this by ligaments. (11) The probable course of the anterior part 
 of the supraorbital sensory canal. (It should also be mentioned that 
 the sensory canals must have had rather numerous sense organs and 
 that they opened outward with very many tubuli.) 
 
 There are, of course, many differences between the Macropetalich- 
 thyids and Elasmobranchs, but these differences seem in general from 
 the morphological point of view to be of much less importance than 
 the agreements. The fact that bone occurs to a considerable extent in 
 the Macropetalichthyids, while it is, as we know, totally lacking in 
 recent Elasmobranchs, ought at first perhaps to be considered as an 
 important difference, but from what is now known of the occurrence 
 of bone in the vertebrate series, even this must be considered to be 
 a character of minor importance. In order to illustrate this the fol- 
 lowing facts are given : 
 
 Typical bone tissue is found in the Cephalaspids, which, according 
 to my recent investigations not yet published, are lowly organized 
 vertebrates, without jaws, with three distinct prootic branchiomeres 
 (hyoid segment, mandibular segment, and premandibular segment) and 
 indications of still more. Further, there are in them also certain evi- 
 dences that there were two segmentally arranged nerves anterior to the 
 facialis, and we are able to conclude that they were in these respects 
 much more primitive even than the recent Cyclostomes. As, in addition, 
 more or less osteoid tissues occurs in a number of other very lowly 
 organized Ostracoderms, we conclude that bone must be a tissue 
 that appeared very early and very low down among the vertebrates, 
 in fact already in the common stem of fishes and Cyclostomes. 
 It is also well worthy of notice in this connection that in the Cephala- 
 spids both the brain and the ears were situated in the bone of the 
 cephalic shield, which means that there is in them no difference be- 
 tween dermal bone and substitution bone. 
 
 In the earliest appearing Dipnoans, Crossopterygians and Actin- 
 opterygians the primordial neurocranium and primordial skeleton in 
 general are regularly well ossified, but the degree of ossification grad- 
 
1 62 Field Museum of Natural History — Geology, Vol. IV. 
 
 ually decreases during the course of the geological periods, the sub- 
 stitution bone in many forms being finally entirely lost. We thus 
 find that the primordial neurocranium in the Devonian Dipnoi 
 Scaumenacia (my own observations) and Dipterus (Traquair 1878, 
 p. 5; Watson and Day 1916, p. 33; observations made by the author 
 in the Royal Scottish Museum, 'Edinburgh) is very well ossified, while 
 it is not found to be ossified in any of the Carboniferous or post- 
 Carboniferous forms, if we except the small ossification in the 
 most posterior part of the occipital region of the recent Ceratodus (Cf. 
 Watson and Gill 1923 ; K. Furbringer 1904, PI. 38, fig. 7 ; Teller 
 1891 ; etc.). Among the Crossopterygians the process of reduction 
 of the bone tissue is especially well displayed in the Coelacanthida, in 
 which the Devonian form Diplocercidcs has the primordial neurocran- 
 ium comparatively well ossified (Stensio 1922; 1923, pp. 1259-1269), 
 while the post-Devonian forms all have it to a very considerable extent 
 consisting of cartilage, the bone being limited only to certain spots 
 (Stensio 1921, pp. 53-61, 91-94, 120-122; Watson 1921 ; cf. also 
 Stensio 1923, pp. 1241-1259). According to information which I 
 have received from Professor D. M. S. Watson of London, the pri- 
 mordial neurocranium is, further, much more ossified in the Carbonifer- 
 ous Palseoniscids than in the Triassic ones Birgeria and Acrorhabdus 
 (Stensio 1921, pp. 152-175, 182-186, 187-198, 211, 224, etc.; pp. 238- 
 250). Similar conditions are also met with in the Saurichthyids, in 
 which those from the lower Triassic have their primordial skeleton 
 very completely ossified, while primordial bone seems to be almost en- 
 tirely lacking in those from the lower Jurassic (Stensio 1925). In 
 the Chondrosteids, Acipenserids and Polyodontids, which are very 
 closely related to the Palaeoniscids and Saurichthyids, the primordial 
 neurocranium, as we know, is, as a rule, without bone. (We find 
 bones only in very old specimens which indicates that the bone has 
 lost its importance and that the parts still remaining therefore develop 
 onto-genetically very late.) Finally, in a newly discovered ganoid 
 from China, which with regard to its cranial anatomy much resembles 
 Amia, the primordial skeleton is much more ossified than in Anna. 
 
 A similar reduction affects in several forms also the dermal bones. 
 In the Dipnoi from the Devonian the dermal bones all lie superficially, 
 are ganoine-covered and well developed (Cf. Watson and Gill 1923). 
 In many post-Devonian Palaeozoic Dipnoans on the contrary, we see 
 that, for instance, the most anterior and posterior bones of the dermal 
 cranial roof sink down beneath the surface of the dermis, lose their 
 ganoine covering and the grooves for head lines of pit organs, and, in 
 
Macropetalichthyids and other Arthrodires — StensiS 163 
 
 the recent Ceratodus, this procedure has gone so far that the anterior 
 and posterior parts of the dorsal side of the primordial neurocranium 
 are uncovered by dermal bones (Cf. Teller 1891, Figs. 2-4; cf. also 
 K. Furbringer 1904, pp. 498-500, who strangely enough arrives at a 
 quite opposite conclusion). The same procedure as the one described 
 in the Dipnoi has taken place in several other Teleostomes, as, for in- 
 stance, in the Chondrosteids, the Acipenserids and Polyodontids, which 
 all are certainly descendants from forms with a very completely de- 
 veloped dermal skeleton. 
 
 From the facts now given we thus find that most groups of Tele- 
 ostomous fishes in which the skeleton has been investigated in detail, 
 undoubtedly appear to be degenerating with regard to the degree of 
 ossification. And in this connection it is worthy of notice that the 
 Macropetalichthyids themselves too seem to represent a degenerative 
 series in this respect, as the primordial neurocranium of the middle 
 Devonian Macropetalichthys rapheidolabis is much more completely 
 ossified than that of the upper Devonian Epipetalichthys wildungensis. 
 
 If we turn to the Elasmobranchs, it has generally been considered 
 that these do not have real bone tissue. This may be true for the 
 more typical later forms, but if we go back to the Acanthodians the 
 conditions seem to be different. Reis (1896, pp. 179-184, PI. VII, 
 fig. 1) and with him most other authors certainly say that in these too 
 there is no bone, but that the primordial skeleton is calcified. But, as 
 is evident from Reis' description and figures, the manner of "calcifica- 
 tion" is there of quite another kind from that in the typical Elasmo- 
 branchs and in reality everything indicates that in the so-called calcified 
 cartilage of Acanthodians we are concerned with true bone. That 
 Reis 1 did not recognise it as bone is probably due simply to the fact 
 that he expected to find the bones of fishes with the structure occurring 
 in higher vertebrates, i.e., throughout, or almost throughout, with a 
 lamellary structure and the lamellae in the interior of the skeletal ele- 
 ments surrounding Haversian canals. This, however, is often not 
 the case, the bone of lower vertebrates generally being entirely without 
 Haversian canals and lamellae and even rather often without cells. 
 Unfortunately the knowledge of the microscopic structure of bone in 
 
 'According to Reis the primordial skeleton of the Acanthodes consists peripher- 
 ally of a rather lamellary layer containing numerous cell spaces. Internally to 
 this follows a cancellous layer and most centrally a vacuity occupied in the fossil 
 by stone but which in the living animal certainly consisted of cartilage. In the 
 cancellous layer there are, if Reis is correct, no cell spaces. A microscopic inves- 
 tigation of the primordial skeleton of the Acanthodians, from the new points of 
 view advanced here would be very desirable, but unfortunately I had no mate- 
 rial with which to undertake one myself. 
 
1 64 Field Museum of Natural History — Geology, Vol. IV. 
 
 lower vertebrates is still very imperfect. The conditions in several 
 other Elasmobranchs from the Palaeozoic also seem to indicate that 
 the Elasmobranchs in general originally had an osseous skeleton (Cf. 
 Woodward 1924). Finally, it ought, in connection with this also, to 
 be pointed out that, according to the recent interesting investigations by 
 Weiden reich (1923, pp. 4 1 5-4I 9, 461-462), there is really no funda- 
 mental difference between calcification and ossification. 
 
 From what we now know of the occurrence of bone in the vertebrate 
 series, we must evidently presume that bone must have occurred also 
 in the ancestors of and in the most primitive Elasmobranchs and the 
 fact that it is to be found in the Acanthodes is therefore only a matter 
 of course. 
 
 That the Macropetalichthyids cannot be closely related to any 
 Teleostomous fishes so far known is thus, despite the presence of bone 
 in their skeleton, beyond question. The development of the dermal 
 bones in their cranial roof and the lacking of dermal bones on the 
 ventral side of the primordial neurocranium, together with the Elasmo- 
 branchian characters, show definitely that they are not specialized 
 Teleostomes at all but forms that among fishes known hitherto are in 
 some way most closely related to the Elasmobranchs. In fact the indi- 
 cations as far as we can judge at present seem to be that the Macro- 
 petalichthyids evolved from some early, very primitive Elasmobranchian 
 form soon after the Elasmobranchian stem had separated from the 
 stem of Teleostomous fishes. 
 
 CERTAIN REMARKS CONCERNING THE OTHER NON- 
 MACROPETALICHTHYID ARTHRODIRES 
 
 The increased knowledge that we now possess of the anatomy of 
 the Macropetalichthyids throws of course new light on the other 
 Arthrodires too, and it is therefore of importance to give here some 
 remarks on these. I shall begin with the Phlyctenaspids, which, as 
 Jaekel has pointed out (1911, p. 46) seem to be most nearly related 
 to the Macropetalichthyids. I shall then turn to the Coccosteids, 
 Homosteids, Mylostomids, Ptyctodontids and, finally, to a form which 
 has not hitherto been recognized as an Arthrodire. 1 
 
 1 A satisfactory classification of the Arthrodires into families is difficult to 
 give at present on account of the imperfect knowledge we still possess of the 
 various forms, and because the opinions of different authors are widely at vari- 
 ance in this respect. The subdivisions given by me here are merely intended to 
 simplify the account of certain anatomical features and not to give positive 
 conclusions as to the mutual relations of all the forms. 
 
Macropetalichthyids and other Arthrodires — Stensio 165 
 
 THE PHLYCTENASPIDS 
 
 To the Phlyctenaspids are here referred among others the genera 
 Phlyctenaspis and Acanthaspis (Arctolepis, Eastman 1908 b; Kaer 
 1916). 
 
 Primordial Neurocranium 
 
 Of the primordial neurocranium no remains have been found so 
 far, a fact which indicates that it was either entirely cartilaginous 
 or at least very slightly ossified. Despite this, it is, however, possible 
 with the help of the configuration of the dermal cranial roof and the 
 conditions in the Macropetalichthyids, to obtain an idea of its general 
 shape, at least in the anterior parts. (See Fig. 20.) It is thus obvious 
 that the ethmoidal and labyrinth regions must both have been very 
 short and broad and that the labyrinth region was rather long, in any 
 case several times as long as the two first-mentioned regions. It 
 seems probable also that the occipital region was rather long, too. The 
 differences with regard to general extent of the various regions from 
 those in Macropetalichthys rapheidolabis can be clearly seen by com- 
 paring Fig. 20 with Figs. 1, 3, 5, etc. 
 
 Fig. 20. Phlyctenaspis acadica 
 
 Restoration of the outlines of the primordial neurocranium with the guidance 
 of the shape of the dermal cranial roof (Cf. Fig. 21) and the conditions in 
 Macropetalichthys rapheidolabis. Shape of the cavum cerebrale indicated by 
 broken lines. X^- 
 
 Olf, olfactory capsule; orb, orbit; par, pineal opening. 
 
1 66 Field Museum of Natural History — Geology, Vol. IV. 
 
 By the use of broken lines I have tried in Fig. 20 to indicate 
 the possible shape and extension of the cavum cerebrale cranii, which 
 if my view is correct ought to have been fairly like that in Macropet- 
 alichthys except that the canals for the tractus olfactorii must have 
 diverged very strongly and rapidly from each other. As the anterior 
 median dermal bone plate M (Fig. 21) has on the lower side close 
 to its posterior border a pronounced pit corresponding to the pineal 
 pit in Titanichthys, we are quite sure of the position of the dorsal 
 part of the canal par, and from this we are of course able to conclude 
 approximately where the anterior end of the cavum cerebrale cranii 
 was situated. Whether there was any distinct cavum precerebrale or 
 not it is not possible to decide, and, finally, it may be added that no 
 external opening of the canal for the ductus endolymphaticus has hith- 
 erto been observed. 
 
 Fig. 21. Phlyctenaspis acadica 
 
 Dermal cranial roof. Sensory canal grooves with dotted outlines. Ornament 
 not drawn. Sketch made after specimen P. 6555 in the British Museum, Lon- 
 don. XVa- 
 
 A, B, I, L, M, Pi, Pi, S, dermal bones of the cranial roof ifc, infraor- 
 bital sensory groove (blue) ; Ic, cephalic division of the lateral line (red) n, 
 nasal aperture; orb, orbital entrance; poc, preopercular sensory groove (yellow) ; 
 soc, supraorbital sensory groove (green), t, line on the plate M, perhaps a vestigial 
 suture. 
 
Macropetalichthyids and other Arthrodires — Stensio 167 
 
 Dermal Cranial Roof 
 
 The plates P lt P 2 , and S (Fig. 21) correspond, at least in the 
 main, to the plates in Macropetalichthyids lettered P lt P 2 , and 5" 
 (Figs. 15, 19), while the plate / certainly represents a posterior part 
 of the plate M 2 of the latter. The plate B (Fig. 21) seems to 
 include the homologues of the plates L 2 and L 3 , in the Macropetalich- 
 thyids, together with a part of the plate M 2 in these. The plate A 
 (Fig. 21) is, as far as can be seen at present, represented in the 
 Macropetalichthyids by the plate L x , and an anterior part of the plate 
 M 2 . The plate L and the most anterior part of the plate M, seem to 
 be without homologues in Macropetalichthyids and that in this case 
 we are concerned with reductions in the last-mentioned forms is highly 
 probable from the conditions in the other Arthrodires. The plate M 
 has probably arisen by the coalescence of two plates, an anterior and a 
 posterior one, homologous to the rostral and pineal plates respectively 
 of the Coccosteids (Fig. 24) as it is crossed by a fine transverse line 
 which seems to be the vestigial suture between the presumed two com- 
 ponents. 
 
 As far as can be judged at present from the different conditions of 
 their dermal cranial roofs, the Macropetalichthyids and Phlyctenaspids 
 must both have descended from primitive Arthrodires in which the 
 medial parts of the cranial roof were occupied by a larger number of 
 bone plates than in any Arthrodire known hitherto. 
 
 Between the plates L and P x there is in the fish a rounded notch 
 (orb, Fig. 21) indicating the position of the certainly very small 
 orbital entrance, which is thus in relation to the dermal cranial roof 
 situated in another manner than in the Macropetalichthyids and which 
 in addition appears to be directed much more laterally than in these. 
 Concerning the ventral boundaries of the orbital entrance nothing 
 certain is known at present. Between the antero-medial border of 
 the plate L and the postero-lateral border of the plate M, there is 
 found another notch in the outline of the dermal cranial roof, a notch, 
 which as far as I can see, must be the external nasal aperture (n, Fig. 
 30. 
 
 Most of the dermal bones are ornamented with tubercles and must 
 have had a superficial position. Only the most anterior bones L and 
 M seem to have been situated somewhat deeper in the skin, as at 
 least their anterior parts usually seem to be without ornament and the 
 supraorbital sensory canal anterior of the plate A probably passed 
 entirely externally of the plate L. 
 
1 68 Field Museum of Natural History — Geology, Vol. IV. 
 
 Sensory Canal System of the Head 
 
 The sensory canals are known only on the dermal cranial roof, 
 on which they form open grooves. These grooves are, however, at 
 least in several cases, deep and narrower at their external opening 
 than further inwards (Fig. 22V so that they in fact open outwards 
 only with a narrow slit. Concerning the number of sense organs 
 in the grooves nothing can be said at present. 
 
 The supraorbital groove (soc, Fig. 21), which is very short, 
 has its posterior end at about the ossification centre of the plate A. 
 From this point it passes forward and laterally to the antero-lateral 
 corner of the plate. On the plate L, there are no traces of it, but 
 it certainly continued still further forwards, although situated there 
 entirely externally of the plate L in the most external layers of the 
 skin. From the direction of the part known of the groove it seems, 
 as is obvious from text fig. 21, certain that the other hitherto unknown 
 part of the same groove passed, as has already been emphasized above, 
 antero-latero-ventrally posterior of the nasal aperture as it does in the 
 Coccosteids and Elasmobranchs. 
 
 Fig. 22. Sketch of a sensory canal groove of a Phlyctenaspid in transversal 
 section. Magnified. 
 
 The infraorbital groove (ifc, Fig. 21) is, with its posterior 
 end, which is situated approximately at the ossification centre of the 
 plate P 2 , continuous with the cephalic portion of the lateral line groove. 
 It goes forward to about the ossification center of the plate P lt where 
 it suddenly bends latero-ventrally and leaves the cranial roof posterior 
 to the orbital entrance. 
 
 Just at the bend, it is continuous with a groove (ifbr, Fig. 21) 
 which goes in an arch postero-medially to near the centre of the plate 
 B. This groove must, as is easily understood, have been innervated 
 either from the n. ophthalmicus lateralis and represent a posterior 
 part of the supraorbital sensory canal, or from the r.oticus lateralis, 
 in which case it must be considered as a branch from the infraorbital 
 groove. Which one of these two alternatives is the true one it is 
 difficult to decide from the conditions in the Phlyctenaspids, but when 
 we come to the Coccosteids we find there, as we shall see, certain facts 
 which seem to be much in favour of the latter. I have therefore 
 called the groove in question the postero-medial branch of the in- 
 
Macropetalichthyids and other Arthrodires — Stensio 169 
 
 fraorbital groove. A further support for this view is given perhaps 
 by the conditions in Lcemargus. In this form the infraorbital canal 
 with its posterior part on the dorsal side of the head, turns abruptly 
 medially and continues in this direction till it almost meets its fellow 
 of the opposite side in the median line (Ewart 1892, PI. I, fig. 1). 
 
 A groove issuing in postero-medial direction from the in- 
 fraorbital groove quite in the same manner as the postero-medial 
 branch of the infraorbital groove in the Phlyctenaspids occurs in many 
 Stegocephalians (Wiman 1914, PI. I, figs. 1, 2, 3; PI. Ill, fig. 2; PI. 
 IV, figs. 2, 6, 7; PL. V, fig. I; PI. VII, in the form figured on this 
 plate the branch in question is double; 1916, PI. XV, cf. Moodie 1908; 
 1915). A homologue to it is also found in larval forms of Triton, 
 in which, however, like the whole sensory canal system, it is repre- 
 sented merely by isolated pits arranged in a line. 
 
 The cephalic division of the lateral line (Ic, Fig. 21) is, as 
 already pointed out, like the other sensory canals, an open groove. 
 Its anterior end is situated as the ossification centre of the plate P 2 , 
 in which it is continuous with the infraorbital groove. It goes back- 
 ward to the postero-lateral edge of the plate S, its farther course back- 
 ward being unknown. There is no cross commissural groove, which 
 connects it with its fellow of the opposite side and, as far as known, 
 there is not even any pit line representing a commissure of this kind. 
 
 The preopercular groove (poc, Fig. 21) is represented by a 
 dorsal part which issues in a postero-latero-ventral direction from the 
 point at which the infraorbital groove and the lateral line groove (Ic) 
 meet one another. 
 
 Nothing is known of pit lines, but this may perhaps be due to the 
 unfavorable state of preservation in which remains of Phlyctenaspids 
 are generally found. 
 
 Dermal Covering of the Trunk 
 
 The anterior parts of the trunk of the Phlyctenaspids are, as we 
 know, enclosed in a strong armour, which, as it follows immediately 
 behind the head, obviously, at least with its anterior parts, occupies 
 the position of the dermal shoulder girdle of the TeleostomoUs fishes. 
 A further support for this view is the fact that, as set forth in the 
 description of M. raphcidolabis, the branchial basket cannot have ex- 
 tended backward into the abdominal armour, but must have been 
 limited entirely to the head. The powerful Elasmobranchian-like spine 
 that in all Phlyctenaspids is rigidly attached to the antero-lateral edge 
 of the antero-ventro-lateral and the lateral edge of the inter-lateral 
 
170 Field Museum of Natural History — Geology, Vol. IV. 
 
 of each side, thus has in fact the approximate position of the pectoral 
 fin of ordinary fishes and seems to me to represent either the entire 
 dermal skeleton of this fin or to be a spine that was attached to its 
 anterior border while the other parts of the fin have become reduced. 
 
 THE^TOCCOSTEIDS 
 
 To the Coccosteids in the broad sense in which they are taken 
 here are referred not only Coccostens, Pachyosteus, Rhinosteus, Pholi- 
 dosteus, Chelyophorus, and others, but also such genera as Dinichthys, 
 Titanichthys, and Selenosteus, the chief reason for this being the 
 characters of their dermal cranial roof. 
 
 Primordial Neurocranium 
 
 In Chelonichthys primigenius (Eichwald i860, pp. 1526-1527 PI. 
 57, figs. 1, 2) the primordial neurocranium was partly well ossified, 
 but unfortunately no details are known of it. In other forms it seems 
 as a rule to have been entirely cartilaginous or at least almost entirely 
 so, for no certain remains of it have been found so far. 
 
 Turning first to Coccosteus decipiens and the forms closely allied 
 to it, we can at once understand from the proportions of the dermal 
 cranial roof that the ethmoidal and occipital regions of the primordial 
 neurocranium in them must have been short, while the remaining two 
 regions, the orbitotemporal and labyrinth regions, had a rather con- 
 siderable length. It deserves also to be mentioned in this connection 
 that the primordial neurocranium must have been fairly high, in any 
 case higher than in the dorso-ventrally much flattened Macropetalich- 
 thys rapheidolabis. That this was the case is easily understood if we 
 look at the head, for instance, of Pholidosteus friedelii (Jaekel 1919, 
 Fig. 2) from the lateral side. In Fig. 23 A, the general shape of 
 the primordial neurocranium has been drawn as I suppose it to have 
 appeared. In this figure also the outlines of the cavum cerebrale 
 cranii and the canals for the olfactory tracts have been put in mainly 
 with the guidance of the conditions in M. rapheidolabis and the posi- 
 tions of the nostrils and the pineal pit on the lower side of the pineal 
 plate {Pi, Fig. 24). We see from this that the canals for the ol- 
 factory tracts cannot have diverged quite so strongly forward as in the 
 Phlyctenaspids. 
 
 While in Coccosteus and its nearest allies the lower side of the 
 dermal cranial roof is rather smooth, or in any case without prominent 
 ridges or formations of this kind, it has in Dinichthys a paired lamella, 
 which extends some distance downwards. The position of this lamella, 
 
Macropetalichthyids and other Arthrodires — Stensio 171 
 
 which in places is thickened or is provided with downward di- 
 rected processes or is irregular in other ways, is shown by Fig. 2, 
 PI. IV, and Fig. 1, PI. LII of Newberry's description in 1889, and by 
 Fig. 4 D of Woodward's description in 1922, and, as we understand 
 from these figures, it must have covered a dorsal portion of the outer 
 surface of each lateral wall of the primordial neurocranium. Accord- 
 ingly it shows the outlines of the dorsal side of the primordial neuro- 
 cranium, a fact that was recently pointed out by Woodward (1922, 
 
 P- 33). 
 
 With the guidance of this lamella and the conditions in the 
 Macropetalichthyids, the restoration shown in Fig. 23 B was made. 
 As we see from it, the primordial neurocranium of Dinichthys agrees 
 in several points with that in the Phlyctenaspids and Coccosteus but 
 
 Fig. 23. Restoration of the outlines of the primordial neurocranium. A, of 
 Coccosteus; B, of Dinichthys. The possible extension of the cavum cerebrale 
 cranii and the olfactory tracts is indicated by broken lines, co, occipital condyles ; 
 olf, olfactory capsule; orb, orbit. 
 
 differs, however, distinctly at least from that in the former by the 
 greater length and comparatively less considerable breadth of its orbi- 
 totemporal region. Further, it is also worthy of notice that the orbito- 
 temporal region probably was not definitely bounded from the labyrinth 
 region, but that the anterior parts of this were rather narrow too, while 
 its posterior parts were very broad. Whether the short occipital region 
 had an anterior broad division as in the Macropetalichthyids it is riot 
 possible to say, as we do not know how the vagus canal had its course. 
 According to the opinion advanced by Woodward (1922, p. 33) 
 the primordial neurocranium in Dinichthys would not have extended 
 farther back beneath the dermal cranial roof than approximately to 
 
172 Field Museum of Natural History — Geology, Vol. IV. 
 
 the transversal plane through the posterior ends of the supraorbital 
 sensory canals (cf. Fig. 24 B). In the Macropetalichthyids the 
 posterior ends of the supraorbital canals are, as is seen from PI. XX in 
 the present paper situated at a place about midway between the laby- 
 rinth cavities. In other fishes, too, ' they do not reach much farther 
 back, even in those cases whenuiey with certainty include the anterior 
 head lines of pit organs. This fact in connection with the circum- 
 stance that the paired lamella on the lower side of the dermal cranial 
 roof distinctly continues postero-laterally of this place to the very 
 postero-lateral corner of the dermal cranial roof, seems to me to indi- 
 cate with almost absolute certainty that the primordial neurocranium 
 of Dinichthys extended backward as far as the dermal cranial roof 
 and that the posterior parts of the labyrinth region were as broad as 
 shown by my restoration in Fig 23 B. 
 
 The paired bony lamella on the lower side of the dermal cranial 
 roof in Dinichthys is held by Woodward to be an ossified dorso- 
 lateral part of the lateral wall of the primordial neurocranium and we 
 would thus have here a primordial component coalesced with the der- 
 mal cranial roof. That this perhaps is the case seems probable to me 
 especially from the conditions in the Coelacanthids, in which a part of 
 the primordial skeleton in certain specialized Triassic forms has 
 coalesced with an overlying dermal bone, while in the other forms it 
 has retained its original primitive relations (Stensio 1921, pp. 53, 93, 
 97, 124; 1923, p. 1269; cf. Allis 1899; 1909a). In the Coelacanthids 
 this coalescence between dermal bone and primordial bone has, how- 
 ever, obviously taken place in connection with a process of reduction 
 of the bone of the primordial neurocranium, and, from what has been 
 emphasized above there is every reason to believe that this is also the 
 case in Dinichthys. What the paired bone is that Newberry (1889, 
 pp. 147-148) describes from the interior of the head in Dinichthys 
 tcrelly, it is not possible to decide. Perhaps it may be an ossified part 
 of the primordial neurocranium. 
 
 A common character for all the Coccosteids is the fact that their 
 orbit usually was large and that the eyes must have been directed 
 straight or almost straight laterally. 
 
 Dermal Cranial Roof 
 
 The dermal cranial roof of Coccosteus decipiens (Fig. 24 A; 25) 
 is rather similar to that in Phylyctenaspis (Fig. 21), but differs from 
 this by the somewhat different shape of the various bones and the fact 
 that the rostral plate (R) and the pineal plate (Pi) are always inde- 
 
Macropetalichthyids and other Arthrodires — Stensio 173 
 
 pendent. Further, it is also to be noticed that the orbital entrance is 
 bounded dorsally by the plate A, while in Phlyctenaspis it is bounded 
 there by the plates L and P lf the plate A there lying medially of the 
 two just-mentioned plates. The external nasal aperture is paired, sit- 
 uated rather close to its fellow of the opposite side and directed almost 
 straight forward. It is bounded medially by the rostral plate (R), 
 laterally by the plate L. 
 
 The upper Devonian Coccosteids are generally more specialized 
 than Coccostcus decipiens. (Cf. Jaekel 1906a; 1907; Dean 1901, 
 figs. 14-18; Eastman 1908a, pp. 89-156; 1908b, pp. 152-205; 
 Bryant 1918, pp. 26-105; Woodward 1922, fig. 4, pp. 30-36.) 
 The ornament of the bones in them is generally fine or is even in certain 
 cases lacking; the bones have become firmly united to one another and 
 in the American forms, as, for instance, Dinichthys , Titanichthys and 
 Stcnostens, the plate L has hitherto not been found. This indicates 
 either that it had sunk deep down in the skin and lost its connections 
 with the neighbouring plates or that it had become entirely reduced, as 
 in the Macropetalichthyids (Cf. Woodward 1922, p. 35). 
 
 In the restorations of the Dinichthy s-species exhibited in various 
 American museums, no consideration seems to have been given either 
 to the absence of the paired plate L or to the circumstance that the 
 primordial neurocranium with the olfactory capsules must have ex- 
 tended somewhat forward beyond the anterior margin of the plates A 
 (Fig. 24 B) of the dermal cranial roof. (Cf. the restoration of Dean 
 1909, Pis. 38-39; and by Bryant 1918, PI. 1, both of which are 
 incorrect in this respect.) 
 
 Dermal Bones of the Cheek 
 
 In Coccosteus decipiens and its nearest allies, the cheek is regularly 
 occupied by two bones, a large, anterior infraorbital (So, Fig. 24), 
 which forms the ventral boundary of the orbital entrance and a small 
 plate (Ps, Figs. 24, 25) \ which is situated behind the infraorbital and 
 occupies the most posterior part of the cheek. (Cf. Traquair 1890, 
 PI. 10, figs. 1, 2; Jaekel 1902, Fig. on p. 107; 1919, Fig. 2, Qj.) In 
 the American forms from the upper Devonian the last mentioned plate 
 is, as far as I know, always lacking, a fact which probably indicates 
 that it has become reduced. (Cf. Eastman 1908a, Figs. 23, 24, 29; 
 1908b, Figs. 27, 31; Hussakof 1906, Figs. iB, 12; Dean 1909, Fig. 
 56, Pis. 38-40, etc.) 
 
 x In certain cases this plate may probably be subdivided into two (Jaekel 1902, 
 Fig. on p. 107). 
 
174 Field Museum of Natural History — Geology, Vol. IV. 
 
 Visceral Skeleton 
 
 The primordial visceral skeleton seems generally to have consisted 
 chiefly of cartilage, but in Pholidostcus (Jaekel 1907, pp. 176-186; 
 1919, pp. 77-87, Figs. 4, 8) and Erromenosteus (Jaekel 1919, pp. 84- 
 86, Fig. 9) it is certainly at least partly ossified. 
 
 In Dinichthys intermedins Woodward has found a bone (Wood- 
 ward 1922, p. 35) which seems to belong to the palatoquadrate, but 
 whether this bone was an ossification in the palotoquadrate cartilage 
 itself or whether it is simply a dermal bone developed in relation to this, 
 it is at present not possible to decide. The occurrence is, however, of 
 much interest, because it shows that there was in the upper jaw of the 
 Arthrodires a palatoquadrate medially of the two external elements, 
 
 xp 
 
 Fig. 24a. Dermal cranial roof 
 
 A. Coccostcus decipiens (sketch drawn by the author from specimens in the 
 Royal Scottish Museum, Edinburgh). For explanation of lettering see next page. 
 
 and that the latter two elements on account of this must really be most 
 nearly comparable with the premaxillary and maxillary respectively. 
 An additional support for this interpretation of the two external ele- 
 ments in the upper jaw is' the fact that the anterior one of these ele- 
 ments in certain species is tuberculated on the external surface (Wood- 
 ward 1922, p. 33; Eastman 1908a, PI. 7, figs. 4-6) and accordingly 
 appears to be a superficial dermal bone. . 
 
 The Meckelian cartilage has in Pholidosteus and Erromenosteus 
 (Jaekel 1907, pp. 176-184; 1919, Figs. 4, 8, 9) its pars articularis 
 ossified, but seems, as far as we know, generally to have been entirely 
 
Macropetalichthyids and other Arthrodires — Stensio 175 
 
 cartilaginous. 1 On the medial side it was almost entirely covered by 
 the large, well-known bone (Mix, Fig. 25) which has been termed by 
 various authors mandibular, gnathal, infero-gnathal, dentary, and 
 splenial. This bone, on account of its extent and relations, obviously 
 corresponds to all the dermal bones on the medial side of the man- 
 dible in the Teleostomes and would therefore if named from this 
 point of view get a very complex title. For the sake of brevity we 
 may call it simply the mixicoronoid (Cf. Stensio 1921, p. 244). On 
 the outside of the postero-ventral part of the Meckelian cartilage, there 
 is in Pholidostcus (Jaekel 1907, Fig. 5; 1919, Fig. 4) a small dermal 
 bone which is ornamented with tubercles and which, as pointed out by 
 
 J u * 
 
 Fig. 24b. Dermal cranial roof 
 
 B. Dinichthys halmodeus (Bones after Eastman 1908a, Fig. 21, and 1908b, 
 Fig. 24). Sensory grooves according to observations made by the author on a 
 specimen in Albany, N. Y. Much diminished. 
 
 Sensory canal grooves in both A and B, indicated by a double dotted line 
 and colours. 
 
 A, B, J, L, Pi, Pi, Pi, R, S, bones in the dermal cranial roof; Ps, So, 
 bones of the cheek; ifc, infraorbital sensory groove (blue); ifbr, postero- 
 medial branch of the infraorbital sensory groove (blue) ; juc, jugal groove 
 (yellow) ; k, lateral line groove (red) ; tnp, groove, probably for the middle 
 head line of pit organs (red); na, nasal aperture; orb, orbital entrance; poc, 
 preopercular groove (yellow) ; pp, groove probably for the posterior head line 
 of pit organs (red) : soc, supraorbital groove (green) ; xp, branch from the 
 cephalic division of the lateral line (groove), probably corresponding to a part 
 of the superatemporal cross commissure (red). 
 
 *I have had the opportunity of investigating Pholidosteus friedelii in the col- 
 lection of Professor O. Jaekel at Greifswald and I fully agree with him in his 
 interpretations of the elements in the lower jaws of Arthrodires. 
 
176 Field Museum of Natural History — Geology, Vol. IV. 
 
 Jaekel, apparently is comparable to the angular of the Teleostomi. 
 The term angular may therefore properly be employed for it. 
 
 As is fully evident from the facts now known, the jaws in the 
 Arthrodires must have been homologous with those in other fishes, and 
 their movements, as emphasized by Woodward (1922, p. 35), have 
 been as normally up and downr (For the previous views on this sub- 
 ject see: Dean 1901, pp. 105-107; Hussakof 1906; Jaekel 1919, 
 pp. 96-108; Adams 1919, pp. 123-127.) 
 
 In Erromenosteus (Jaekel 1919, Fig. 9) the ventral parts of the 
 hyoid arch and perhaps also a basibranchial element appear to have 
 been ossified. 
 
 Dentition 
 
 Teeth are found in all young specimens of Coccosteids, but while 
 in certain forms, as for instance, Coccosteus and Diplognatus, they 
 are retained throughout life, they are in adult specimens of several 
 other forms such as Dinichthys, Titanichthys etc., entirely abraded by 
 use, the labial edges of the jaws becoming in these forms transformed 
 to cutting edges or adapted for crushing. 
 
 It is generally maintained (Cf. Woodward 1922, p. 35) that the 
 teeth occur in a single series on the labial margins of the jaws, but 
 the investigations undertaken by the author 1 show that on the mixi- 
 coronoid of certain forms there is at least a second and perhaps also a 
 third or more series of lower teeth medially of the labial ones. All the 
 teeth are ankylosed to the jaw bones with their bases and generally have 
 the central parts occupied by a rather large pulp-cavity. In the dorsal 
 parts this pulp-cavity is generally surrounded by ortho-dentine, in the 
 basal parts by trabecular-dentine, but in certain teeth the trabecular- 
 dentine may extend almost to the apex. This seems especially to be 
 the case in one category of large teeth which belong to the labial 
 series and are characterized by the fact that they have on their medial 
 side one, two or three, small, bluntly conical cusps. Basally the tra- 
 becular-dentine gradually turns into the bone of the mixicoronoid. 
 
 Externally of the dentine there is in the distal parts of the teeth a 
 layer of bright, shining enamel, which is perforated by the distal ex- 
 tensions of the dentine tubules, these tubules leading to and opening 
 on the outside of the enamel. Also, the cusps on the medial side of 
 the large external teeth are provided with an enamel layer of the same 
 sort. Besides the cusps on the medial side, several of the larger teeth 
 of the external series have also a small cusp on the anterior and pos- 
 
 'The investigations have been made on the mixicoronoid of a Coccosteid from 
 the Devonian of Spitzbergen. 
 
Macropetalichthyids and other Arthrodires — Stensio 177 
 
 terior side, so that, when seen from the lateral or medial side, they 
 have the appearance of shark teeth (also teeth of Acanthodians). 
 
 The occurrence of at least two or three longitudinal series of teeth 
 in their jaws, together with the circumstance that certain of the teeth 
 in the external one of these series are shark-like in shape, indicates that 
 the dentition in the Coccosteids probably has arisen from an Elasmo- 
 branchian-like one. 1 An additional support for this view is the fact 
 that the medial side of the mixicoronoid ventrally of the longitudinal 
 series of teeth appears to consist mainly of a dentine-like tissue. 
 
 Sensory Canals of the Head 
 
 The sensory canal system of the Coccosteids is, as in that of the 
 Phlyctenaspids, represented merely by grooves on the external sur- 
 faces of certain of the external dermal bones. The course of these 
 grooves is best known in Coccosteus dccipiens (Figs. 24 A, 25) and 
 in the Dinichthys species (Fig. 24 B). The subsequent account is based 
 chiefly on C. decipietis. 
 
 The supraorbital groove (soc, Figs. 24 A, B; 25) goes from the 
 plate B forward and somewhat laterally to the plate A, which it crosses 
 to the antero-lateral corner. From this point it continues downward 
 anterior of the eye and posterior to the nasal aperture on the plate L, 
 on which it probably anastomized with the infraorbital groove. It is 
 thus fully evident that with its anterior part it had the same course 
 as in the recent Elasmobranchs (Cf. Allis 1901, pp. 104-107; 1905). 
 
 The infraorbital groove (ifc, Figs. 24 A, B, 25) has on the dermal 
 cranial roof exactly the same course as in the Phlyctenaspids and has 
 also a postero-medial branch (ifbr), as in these. We can here see 
 quite clearly from its transverse course that this branch cannot 
 represent a. part of the supraorbital groove as one might perhaps be 
 inclined to think merely from its position in the Phlyctenaspids (Cf . 
 Pollard 1892b, pp. 546-548). From the dermal cranial roof the in- 
 fraorbital groove turns latero- ventrally to the cheek, on which it is 
 first situated on the infraorbital bone. On this it passes first down- 
 wards and then anteriorly in the normal way, posterior and ventrally of 
 the orbital entrance. After leaving the infraorbital it continues for- 
 ward on the plate L, probably anastomizing there with the supraorbital 
 groove and probably passing anterior of this commissure ventrally of 
 the nasal aperture. What its relations were at the very anterior end 
 is not known. 
 
 J The only description that I have found of the microscopic structure of the 
 teeth in Coccosteids was published by Agassiz 1844 (p. 26). What Agassiz 
 describes there is probably merely a basal part of a tooth. 
 
178 Field Museum of Natural History — Geology, Vol. IV. 
 
 The cephalic division of the lateral line groove (Ic, Figs. 24 A, B, 
 25) has the same course as in the Phlyctenaspids. Close to its pos- 
 terior end there issues from its medial side a branch (xp) in anterior 
 and somewhat medial direction. This branch, which goes close to 
 the medial edge of the bone-plate 6" and which, contrary to what is 
 , shown in most restorations, is Tather short and does not reach the plate 
 B, may perhaps correspond to a lateral part of the supratemporal cross 
 commissure. 
 
 The preopercidar groove (poc, Figs. 24 A, B, 25) issues in the 
 plate Pi from the point where the infraorbital groove and the lateral 
 line groove meet each other, quite as it does in the Macropetalichthyids 
 and Phlyctenaspids. It goes to or close to the postero-lateral corner 
 of the cranial roof, its passage on the cheek being unknown, as it 
 probably passed there entirely in soft tissue. 
 
 Fig. 25. COCCOSTEUS DECIPIENS 
 
 Head and abdominal armour, chiefly from Jaekel (1902). Sensory canal 
 grooves indicated by double dotted lines and colors. 
 
 A, B, J, L, Pi, P 2 , R, S, bones of the dermal cranial roof ; Ps, So, bones 
 of the cheek (So, infraorbital bone); Mx, maxillary; Pmx, premaxillary ; Mix, 
 mixicoronoid ; Spi, spinal; ifc, infraorbital groove (blue); ifbr, postero-medial 
 branch of the infraorbital (blue) ; juc, /ugal groove (yellow) ; Ic, cephalic 
 division of the lateral line (red) ; mp, groove probably corresponding to the 
 middle head-line of pit organs (red) poc, preopercular groove (yellow); pp, 
 groove probably corresponding to the posterior head-line of pit organs xp, groove 
 probably corresponding to a lateral part of the supratemporal cross commissure. 
 
 About at the place in the infraorbital plate (So) at which the in- 
 fraorbital groove turns forward from its downward direction, there 
 issues from it a posteriorly and somewhat ventrally running branch 
 (juc, Figs. 24 A, B; 25) which obviously is the homologue of the 
 
Macropetalichthyids and other Arthrodires — Stensio 179 
 
 horizontal cheek canal of the Elasmobranchs (Allis 1923, pp. 199- 
 201), the horizontal part of the jugal canal of Crossopterygians 
 (Stensio 1921, pp. 76-7711923, p. 1258), the horizontal cheek line of 
 several Teleostomous fishes (Allis 1889, p. 506, PI. 42; 1900, p. 445; 
 1905, pp. 406-407; 1923 p. 201, Pollard 1892b, p. 548) and the jugal 
 groove of the Stegocephalians (Moodie 1908, p. 515; 1915, p. 320). 
 
 Whether there was any mandibular canal is not known. 
 
 Behind the postero-medial branch of the infraorbital groove, too, 
 paired, short grooves (Figs. 24, 25) are regularly found on the bone 
 plate B in most of the Coccosteids examined by the author. The 
 anterior one (mp) of these grooves is situated transversally to the 
 longitudinal axis of the head and seems to correspond to the middle 
 head-line of pit organs in the Elasmobranchii and Teleostomi, while 
 the posterior one (pp) which is situated obliquely so that its anterior 
 end is nearer to the median line than the posterior one, probably is 
 homologous with the posterior head-line of pit organs in the same fishes. 
 (Cf. Allis 1889, pp. 502-509, PI. 42; Herrick 1901, pp. 222-223, PI. 
 14; Stensio 1921, p. 218; 1923, p. 1258; etc.) 
 
 Some Remarks on the Trunk and the Fins 1 
 
 A pectoral spine (Spi, Fig. 25) although much smaller than in the 
 Phlyctenaspids, occurs at least in most of the Coccosteids, and it is 
 beyond question that the paired plate found in Coccosteus decipiens and 
 Dinichthys gouldi somewhat behind the abdominal armour, really rep- 
 resents the pelvic girdle (Cf. Woodward 1891, pp. 282, 289; 1922, p. 
 35; Dean 1896, p. 162; 1909, pp. 282-287). There are also in C. decip- 
 iens a few endoskeletal radials in connection with the pelvic girdle. The 
 lateral line was, on the trunk at least, partly surrounded by lime-bear- 
 ing tissue, but whether this tissue was bone or not it is not possible 
 to say. There are also in certain well-preserved specimens reliable evi- 
 dences of very delicate dermal tubercles behind the abdominal armour. 
 
 Contrary to what is generally maintained, I feel inclined to think 
 that the pelvic girdle, as well as all other endoskeletal elements that 
 are found preserved in Coccosteids, consists of true bone. A more 
 detailed microscopic examination of these elements has, however, so 
 far not been made, a fact which is much to be regretted. 
 
 'The remarks given here on the skeleton of the pelvic girdle and fins of 
 Coccosteus are based on observations made by the author in the Royal Scottish 
 Museum, Edinburgh and the British Museum, London. I fully agree with Wood- 
 ward (1891) in his interpretation of the specimens in the latter museum 
 (Dean 1909, p. 285). 
 
180 Field Museum of Natural History — Geology, Vol. IV. 
 
 Some General Remarks on the Coccosteids 
 From what we know at present of the Coccosteids, it seems as if 
 they degenerated during the Devonian with regard to the degree of 
 ossification both in the exo- and endo- skeleton, a fact which to a cer- 
 tain degree was emphasized by Jaekel in 1906 (1906a p. 82). In 
 connection with this also a reduction of the pectoral spines took place. 
 
 THE HOMOSTEIDS 
 
 The Homosteids, represented by the genus Homosteus, are certainly 
 closely related to the Coccosteids. 
 
 The bones of the dermal cranial roof are the same as in the Coccos- 
 teids, but the dermal cranial roof as a whole differs from that in these 
 by the considerable lengthening of the occipital region and by the fact 
 that the orbital entrance is directed upwards and entirely situated in 
 it. It thus approaches in these respects that of the Phlyctenaspids on 
 the one hand and that of the Macropetalichthyids on the other. 
 
 The plate L is not found in Traquair's restoration (Traquair 
 1889) of the fish, a circumstance which probably indicates that it had 
 become entirely reduced. The anterior parts of the plates P x , A and R 
 lack both ornament and sensory canal grooves and must obviously have 
 been situated deep in the skin and are probably in an early stage of re- 
 duction. The other parts of the dermal cranial roof are provided with 
 a faint ornament, and this, in connection with the fact that certain 
 of the sensory canal grooves are lacking, while the remaining ones 
 are rather shallow and indistinct, seems to indicate that the dermal cra- 
 nial roof as a whole had commenced to sink down into the deeper layers 
 of the skin. 
 
 The skeleton of the cheek and the visceral skeleton are very im- 
 perfectly known and nothing can therefore be said of them here. 
 
 THE MYLOSTOMIDS 
 
 The Mylostomids, which probably are to be considered as an highly 
 specialized offshoot of the Coccosteids, differ from them mainly in the 
 characters of their dentition, the teeth, as we know, being in them 
 transformed to strong tritoral plates. 
 
 The dermal cranial roof is, at least in Mylostoma, said to be thin 
 and devoid of ornament and lacks most of the sensory canal grooves, 
 all of which conditions indicate that it was situated deep in the skin 
 and probably was in an initial stage of reduction. It may further be 
 mentioned that the plates homologous with L and Ps of Coccostens 
 are lacking and that the infraorbital plate (So) is narrow. The orbital 
 opening is not directly limited by any part of the dermal cranial roof. 
 
Macropetalichthyids and other Arthrodires — Stensi5 181 
 
 In the lower jaw of Dinomylostotna, the Meckelian cartilage is ossi- 
 fied in the pars articularis (Woodward 1922, p. 34; Eastman 1906, 
 pp. 25-26). Eastman's view that the tritoral dentition of the Mylos- 
 tomids would be more primitive than that in the Coccosteids (East- 
 man 1906, p. 6; 1907; 1908a, p. 95; 1908b, p. 163) must, as is easily 
 understood, be incorrect (Cf. Woodward 1922, p. 35), for the Mylos- 
 tomids not only do not pertain to the latest Arthrodires (they are 
 from the upper Devonian), but they appear in most respects to be 
 more specialized than the Coccosteids from the lower Devonian. More- 
 over it would be very difficult to explain from Eastman's point of 
 view how the tritoral plates of the Mylostomids could be transformed 
 into teeth of the type found in the more primitive Coccosteids, as, for 
 instance, in Coccostcus. 
 
 THE PTYCTODONTIDS 
 
 The primordial neurocranium of Rhamphodus is said by Jaekel 
 (1906b, p. 183) to consist of calcified cartilage, and if I understand 
 Jaekel's account correctly this would also have been the case with 
 the visceral skeleton. I am not inclined to accept this view, however, 
 but rather think that we are in this case as in the Macropetalichthyids 
 concerned with true bone-tissue. Probably this bone-tissue, as in 
 Macropetalichthys, appeared chiefly or exclusively as thin perichondral 
 layers without evidence of sutures and distinct centres, conditions which 
 of course without microscopic investigation make the detection of its 
 true nature difficult. 
 
 I have more reason to suspect that the primordial skeleton of 
 Rhamphodus was, at least in part, actually ossified, since the primor- 
 dial neurocranium in a form pertaining to the family dealt with below 
 after the Ptyctodontids, as far as I could find by a macroscopic exami- 
 nation of it, was to a considerable extent ossified, although it is said 
 by Jaekel in a recent paper (1921, p. 217) to consist of calcified 
 cartilage (cf. the Jagorinids below) 1 . 
 
 Jaekel further mentions a "Schadeldach" in Rltamphodus, which, 
 if I understand him correctly, must mean that there is in this fish a 
 dermal cranial roof (Jaekel 1906b, p. 183). 
 
 Concerning the dentition, I have nothing new to add here. It must, 
 however, be emphasized that it is highly specialized. On account of 
 its agreements with that of the Chimaeroids, the Ptyctodontids have, 
 as is well known, been considered by several authors to be closely allied 
 to the Chimaeroids (Cf. Eastman 1908b, pp. 120-123). 
 
 'In another place in the same paper Jaekel is, however, obviously aware 
 that it was ossified (Jaekel 1921, p. 218). 
 
1 82 Field Museum of Natural History — Geology, Vol. IV. 
 
 Rhamphodus has behind the head a dermal armour on the anterior 
 part of the trunk, an armour that, as is clearly shown by Jaekel, is of 
 the Arthrodiran type (Jaekel 1906b, Figs. 3, 5) although it has become 
 more reduced than is generally the case in the Arthrodires. In fact 
 this armour is represented merely by certain of the anterior plates 
 and the pectoral spine, which latter occupies its ordinary position. Its 
 agreements with a dermal shoulder girdle are obvious from Jaekel's 
 account, and from this it seems to be beyond question that the armour 
 of the trunk in the Arthrodires, at least in its anterior parts, really is 
 homologous with the dermal shoulder girdle of the Telostomi and 
 Tetrapods. 
 
 The trunk armour of the Ptyctodontids, as is seen from Jaekel's 
 figure of it, obviously formed part of the posterior boundary of the 
 branchial chamber and we thus have here an additional support for 
 my view that this chamber in Arthrodires did not extend backward 
 into the trunk. (Cf. Jaekel 1906b, Figs. 3-6; 1907, p. 184; the homo- 
 logue of the spinale also clearly understood from Jaekel's figures 3-6 
 in 1906a.) 
 
 THE JAGORINIDS 
 
 In a paper published in 192 1, Jaekel described under the name of 
 Jagorina pandora, certain remains of an upper Devonian fish, which 
 in this connection has a special interest. Through the courtesy of 
 Prof. Jaekel, I had, during a visit at Greifswald in 1922, the oppor- 
 tunity to examine the material so far available of this fish and the 
 account given below is therefore in part based on my personal ob- 
 servations. 
 
 Primordial Neurocranium 
 
 The primordial neurocranium is said by Jaekel in one part of his 
 paper, as already pointed out above, to consist of calcified cartilage, 
 while in another part it is said to have its outer and inner walls 
 feebly ossified. From the detailed account given by Jaekel and from 
 my own observations it is, however, fully evident to me that we are 
 concerned here with a primordial neurocranium that consisted chiefly 
 of cartilage, but which had the outside and inside (cerebral side) of 
 this cartilage provided with a thin, perichondral layer of bone exactly 
 as in Macropetalichthys rapheidolabis. There were, also, as in this 
 species, a labyrinth layer of perichondral bone and, at least, to a cer- 
 tain extent, canal layers too. 
 
 Unfortunately the single specimen that exhibits the primordial 
 neurocranium is preserved so that the primordial neurocranium is 
 
Macropetalichthyids and other Arthrodires — Stensio 183 
 
 available for examination chiefly from the ventral side. The shape 
 from this side is shown by Fig. 26, which has been copied after Jaekel. 
 As is seen from this figure the ethmoidal region is very short and 
 broad, with a truncated, concave anterior margin and a large antero- 
 lateral^ projecting process on each side, a process {olf, Fig. 26), 
 which obviously is so situated and has such a shape, that it agrees 
 completely with the olfactory capsule of the Macropetalichthyids, and 
 is therefore undoubtedly to be interpreted as the olfactory capsule. 
 The orbits are very large and imperfectly bounded and the long orbito- 
 temporal region is consequently not distinctly marked off from the 
 labyrinth region, which probably also is rather long. The labyrinth 
 region is further characterized by the circumstance that it is broadest 
 
 Fig. 26. Jagorina pandora Jaekel 
 
 Primordial neurocranium in ventral view. After Jaekel (1921). XH- c°, 
 occipital condyle (paired); olf, nasal capsule; orb, orbit. 
 
 in its posterior parts, quite as the one in Dinichthys (Cf. Fig. 23 B). 
 The occipital region is short and its posterior end is provided with a 
 paired condyle for articulation with the vertebral column. 
 
 As a bottom for the orbit does not occur in the fish and the interor- 
 bital wall is fairly narrow in comparison with that in the Macropetali- 
 thyids, we must expect that the carotis interna passed forward along 
 the lateral edge of the ventral side of the primordial neurocranium and 
 not in the orbital bottom as in Macropetalichthys rapheidolabis. Con- 
 sequently it is probable that this artery when it turned medially to 
 ascend through the ventral wall of the primordial neurocranium to the 
 cavum cerebrale, entered a canal which opened in one of the pits 
 
1 84 Field Museum of Natural History — Geology, Vol. IV. 
 
 situated on the ventral side of the anterior part of the primordial neuro- 
 cranium at about the transition between the orbitotemporal and ethmoid- 
 al regions. (As these pits had not been cleaned from matrix in their 
 deeper parts at the time I examined the specimen, I cannot, however, 
 say with certainty whether there were among them any that really 
 formed ventral openings for canals of this sort.) 
 
 The labyrinth cavity is well preserved and is evidently entirely 
 separated from the cavum cerebrale, quite as in M. rapheidolabis and 
 Selachians. The separating wall, which consisted of cartilage, was 
 lined by the internal bone layer on the cerebral surface and the 
 labyrinth bone layer on the labyrinth surface. A division for the 
 sacculus, divisions for semicircular canals and a dorsally ascending 
 canal for the ductus endolymphaticus are clearly distinguishable. The 
 canal for the ductus endolymphaticus opened on the dorsal side of the 
 primordial neurocranium and it seems probable also that in the living 
 fish it continued out to and opened on the external surface of the skin, 
 as it does in Epipetalichthys wildungensis, Selachians and Chimaeroids. 
 The canal for the n. acusticus issues from the cavum cerebrale very low 
 down, almost at the bottom and goes outward to the ventral parts of 
 the labyrinth cavity, probably almost as in M . rapheidolabis. 
 
 The shape of the caVum cerebrale is imperfectly known, but, as 
 far as can be judged at present, it appears to have resembled that in 
 M. rapheidolabis. 
 
 What function is to be ascribed to the wide canal that opens on, the 
 posterior side of the labyrinth region (Fig. 26) it is not possible to 
 decide at present, as its course inside the lateral wall is very imper- 
 fectly known. It appears most probable that it may have transmitted 
 the jugular vein, and perhaps, also, in its posterior parts, the vagus 
 nerve as well. 
 
 It is fully evident that the palatoquadrate cannot have been fused 
 with the primordial neurocranium, but must have been an independent 
 element. 
 
 Despite the fact that the primordial neurocranium of Jagorina 
 pandora is still in several points imperfectly known, we see, however, 
 that it is clearly of the Arthrodiran type (Cf. Fig. 23). It is therefore 
 beyond question that Jagorina pandora cannot be a sturgeon, as main- 
 tained by Jaekel, but that it must be referred to the Arthrodires. 
 
 Dermal Cranial Roof 
 The dermal cranial roof is, as far as is known, represented merely 
 by feebly developed isolated tubercules (called "Dornen" in the de- 
 scription given by Jaekel 1921, p. 219). The microscopic structure 
 
Macropetalichthyids and other Arthrodires — Stensio 185 
 
 of these tubercles is not known and it is therefore impossible to say 
 whether they consist of bone or whether they are most comparable 
 to the dermal denticles of the sharks. 
 
 This fish thus has no real dermal cranial roof of bones united with 
 each other by sutures. The lack of a cranial roof of this kind must, 
 however, as far as we can judge, be due to reduction, since a real dermal 
 cranial roof occurs in all the other Arthrodires and is generally even 
 most complete in the forms from the lower Devonian. 
 
 Dermal Bones of the Cheek 
 
 No dermal structures belonging to the cheek have been observed 
 hitherto. „ 
 
 Visceral Skeleton 
 
 Associated with the primordial neurocranium are found several 
 remains of the visceral skeleton, which was partly ossified. Among 
 these remains are surely parts of the palatoquad rates and the mandibles. 
 It is worthy of notice that the primordial neurocranium, as far as can 
 be seen, has no articulating surfaces for the anterior ends of the 
 palatoquadrates, and it therefore seems as if these had been suspended 
 by ligaments beneath the ethmoidal region, as in the Elasmobranchs. 
 
 Dentition 
 
 The dentition is, as Jaekel pointed out to me during my visit at 
 Greifswald, very interesting, as it consists of numerous, small, inde- 
 pendent 1 Selachian-like teeth, thus teeth with several cusps. The shape 
 of these teeth is in fact of such a nature that they undoubtedly would 
 have been taken for shark teeth if they had been found detached from 
 the specimen, a fact which in this connection is of much importance, 
 as it forms a strong additional support for the view advanced by me 
 in this paper that the Arthrodires are really closely related to the Elas- 
 mobranchs. The microscopic structure of the teeth is not known. 
 
 It is worthy of notice in this connection that Woodward already in 
 192 1 shortly after Jaekel had published his description of the pri- 
 mordial neurocranium of Jagorina pandora, without any knowledge of 
 the character of the dentition, arrived at the conclusion that /. pandora 
 was a primitive shark (Woodward 1921, p. 33). 
 
 Sensory Canals 
 Nothing is known of the sensory canals of the head. 
 
 Abdominal Armour 
 Close behind the head there is a girdle of dermal plates that evi- 
 dently represents the shoulder girdle. It is not known in detail, but 
 
 *Not ankylosed with the jaws 
 
1 86 Field Museum of Natural History — Geology, Vol. IV. 
 
 Jaekel says of it (1921, p. 218) that it is sturgeon-like, which may 
 mean very little, since he says the same of that in Rhamphodus (1906b, 
 pp. 183-185). 
 
 SOME GENERAL REMARKS ON THE NON-MACRO- 
 PETALICHTHYID ARTHRODIRES 
 
 From what has been set forth in the account given above the fol- 
 lowing ought to be obvious: 
 
 1. There is strong reason to believe that the primordial skeleton 
 of the non-Macropetalichthyid Arthrodires originally was rather well 
 ossified and that accordingly, a reduction of the degree of ossification 
 took place during the course of the Devonian period. 
 
 2. It is fully clear that the dermal skeleton of the non-Macropeta- 
 lichthyid Arthrodires was also during the Devonian period in process 
 of reduction. In certain, especially upper Devonian, forms the reduc- 
 tion has gone rather far and in Jagorina even so far that practically the 
 entire dermal skeleton of the head has been lost. 
 
 3. The reduction of the dermal skeleton in the non-Macropetalich- 
 thyid Arthrodires takes place in this manner : The bones sink deeper 
 into the skin, become thinner and lose their ornament and the external 
 sensory canal grooves. The sensory canals obviously must retain their 
 superficial relations to the skin and they become on this account sit- 
 uated totally externally of the bones. 
 
 4. From the facts so far known it is beyond doubt that the non- 
 Macropetalichthyid Arthrodires with regard to the degree of ossifica- 
 tion represent a degenerating series. 
 
 5. There are so many common points of agreement between the 
 non-Macropetalichthyid Arthrodires and the Macropetalichthyids that 
 we cannot doubt that they are closely related to each other and that 
 they must, as has generally been done hitherto, be referred to the 
 same group of fishes. • 
 
 6. From the structure of the primordial neurocranium, from the 
 development of the labyrinth cavity, from the character of the den- 
 tition in certain forms, and from the course and development of the 
 sensory canals, it is clear that the non-Macropetalichthyid Arthrodires, 
 like the Macropetalichthyids, must be closely related to the Elasmo- 
 branchs. In the case of Jagorina pandora this is even so clear, that 
 we should almost feel inclined to take it for a primitive Elasmobranch 
 if we did not know that it had a shoulder girdle of dermal bones. 
 
Macropetalichthyids and other Arthrodires — Stensio 187 
 
 CONCLUDING REMARKS ON THE AFFINITIES OF THE 
 
 ARTHRODIRES 
 
 After having dealt in detail with the head of the Macropetalich- 
 thyids and also given a short account of the head in the other Arthro- 
 dires, I shall here briefly summarize the results obtained and add cer- 
 tain remarks of interest otherwise. 
 
 With the comparatively good knowledge we now possess of their 
 anatomy it is quite clear that the Arthrodires are true fishes and that 
 as recently pointed out by Woodward (1922, p. 35) they have noth- 
 ing to do with Ostracoderms, which as I have been able to make out, 
 are much more lowly organized agnathous vertebrates. 
 
 We have seen that the Arthrodires from the oldest divisions of the 
 Devonian generally have the dermal skeleton more complete and 
 stronger than those from the youngest division of the same formation, 
 and we have also found that the primordial skeleton is more or less 
 ossified in several of their representatives. Concerning the primordial 
 skeleton there have been advanced several facts which indicate that it 
 must have been rather completely ossified in all primitive Arthodires. 
 Accordingly we find that the Arthrodires like the Dipnoi, the Crossop- 
 terygii, the Palaoniscida, the Saurichthyida, the sturgeon fishes, the 
 Amiada and several other fishes, form a degeneration series with re- 
 gard to their degree of ossification. 
 
 In their general organization the Arthrodires have, as recently em- 
 phasized by Woodward (1922, p. 35), little in common with the Dipnoi 
 and Crossopterygii, and it is equally evident that they are not Actino- 
 ptcrycjii. From- the investigation made here they were, however, found 
 to be at the stage of the Elasmobranchs and in fact everything seems 
 to indicate that they are really closely related to these. Their Elasmo- 
 branchian characters are as follows : 
 
 I. The general characters of the primordial neurocranium, especially 
 the tendency to broadening of the ventral surface, partly at the ex- 
 pense of the lateral surfaces. (This is above all the case in the labyrinth 
 region, the sacculus having occupied about the same position in rela- 
 tion to the ventral surface as in Chlamydoselachus and most other 
 Selachians.) 2. The position and relations of the olfactory capsule. 
 3. The presence of a nasal fontanelle on the lower side of the olfactory 
 capsule (Macropctalichtliys) as in Chlamydoselachus and certain other 
 Selachians. 4. The presence of the cavum precerebrale (EpipetalicJi- 
 thys). 5. The general shape of the labyrinth, especially with regard 
 to the position of certain of its main parts, for instance, the utriculus. 
 6. The persistence of the ductus endolymphaticus (Macropetalichthys, 
 
1 88 Field Museum of Natural History — Geology, Vol. IV. 
 
 Epipetalichthys, Jagorina) and the fact that there was probably, at 
 least in certain forms (Macropetalichthys) , a fossa endolymphatica on 
 the dorsal side of the primordial neurocranium beneath the dermal 
 bones. 7. The fact that the ductus endolymphaticus, at least in cer- 
 tain forms (Epipetalichthys), perforated the dermal cranial roof and 
 had an external opening situated about as in Chlamydoselachus. 8. The 
 general shape of the brain as far as this can be restored from the posi- 
 tion and course of the nerve canals through the cranial walls and from 
 the shape of the cavum cerebrale. 9. To a certain extent the develop- 
 ment of the blood vessels. 10. The fact that the palatoquadrate did 
 not articulate with the ethmoidal region but must have been suspended 
 beneath this merely by ligaments. II. The dentition, which in certain 
 forms, as Jagorina, consists of multicuspidated shark-like teeth. 12. The 
 course and development of the sensory canal system. 
 
 The opinion now advanced of the close relationship between the 
 Arthrodires and Elasmobranchs of course implies either that bone has 
 arisen independently in different groups of vertebrates or that it is a 
 very old sort of tissue that was present already among the most primi- 
 tive vertebrates and from these was bequeathed to the common ancestors 
 of the Arthrodires and Elasmobranchs. As has been pointed out, the 
 conditions in the Cephalaspids and Ostracoderms in general as well as 
 in the Acanthods and oldest Teleostomi indicate that the latter alter- 
 native ought to be the true one, and accordingly that the Elasmobranchs 
 as far as can be understood at present must have lost the bone-tissue. 
 In other words they too would represent a degenerating series with 
 regard to the degree of ossification. In full accordance with this view 
 is obviously the occurrence of large bone-like plates (or perhaps true 
 bone-plates) in the earliest known true forms of the Holocephali 
 (Myriacanthus and Chimeeropsis) and the fact that among the Cochlio- 
 dontids there is found a form (Menaspis) with evidence of armour on 
 the head and anterior part of the trunk (Cf. Woodward 1924) .* 
 
 Except by the presence of bone-tissue in their skeleton, a character 
 which, as we have found, is of very slight importance from a morpho- 
 logical point of view, the Arthodires seem, as far as we know, to 
 differ in no essential points from the Elasmobranchs. The different 
 shape of the vertebral column in them as compared with the recent 
 Elasmobranchs might perhaps be considered very insignificant, but if 
 we consider the fossil Elasmobranchs this difference becomes less ap- 
 
 *It is also of interest to note that Menaspis has paired lateral spines, which 
 perhaps are identical with the pectoral spines of Arthrodires. In fact it seems 
 not impossible that the Cochliodontids will reveal themselves as highly special- 
 ized Arthrodires (for Menaspis see Zittel's Grundzitge der Palaeontologie, 1923; 
 Dean 1904, and Jaekel 1891). 
 
Macropetalichthyids and other Arthrodires — Stensio 189 
 
 parent, for in the Palaeozoic forms and certain earlier Mesozoic forms 
 of these the neural arches are almost like those in the Arthrodires (Cf. 
 Dean 1909, Clamydosclache; Fritsch 1895, Xenacanthus and Pleura- 
 canthus; Brown 1900 Hybodus; Koken 1907, Hybodus). 
 
 In addition to their general Elasmobranchian-like characters above 
 enumerated the Arthrodires have also certain special Holocephalian char- 
 acters the following of which deserve to be specially mentioned here. 
 1. The joint between the head and the vertebral column. 2. The 
 structure of the branchial apparatus, inasmuch as this is covered ex- 
 ternally by a sort of gill-cover. 3. The character of the dentition in 
 the more specialized forms. 4. The dorsal extension of the pelvic 
 girdle. 5. The general shape of the body. 6. Certain characters of the 
 arterial system. 
 
 Of these characters, No. 3 is quite certainly simply a parallelism and 
 it seems highly probable that this is the case with No. 5 too. Concern- 
 ing the remaining four it is difficult to say at present how much stress 
 ought to be laid on them, and it is consequently difficult to conclude 
 whether the Arthrodires among the Elasmobranchs were most closely 
 allied to the Holocephali or to the Selachii or whether they represent an 
 independent branch from the Elasmobranchian stem. Since, as we have 
 seen, they seem to have the pectoral fins much transformed or perhaps 
 represented merely by an anterior spine, it is not probable that they 
 could have given rise to the Holocephali. If it should appear that they 
 are most closely allied to these the Arthrodires would have descended 
 from the same primitive ancestors or, if we want to express it more 
 exactly, from the primitive forms of the Holocephalian branch of the 
 Elasmobranchian stem. Under the present conditions we can, however, 
 with assurance only say that they are to be considered as an offshoot of 
 the Elasmobranchian stem of fishes and that besides their Elasmo- 
 branchian characters they have retained a few primitive ones at the 
 same time that they by specialization have acquired certain new ones. 
 
 Their main primitive characters must, as far as we understand, be 
 the presence of bone-tissue in their skeleton and the occurrence of a 
 postero-medial branch for the infraorbital sensory groove on the der- 
 mal cranial roof. 
 
 As in another paper I shall give a detailed account of the relation- 
 ship between the Arthrodira and the Antiarcha, I shall here only men- 
 tion that, as far as I can see from the investigations of material in the 
 chief American and European Museums, as well as from material re- 
 cently obtained from Scaumenac Bay, the Antiarcha seem to be a highly 
 specialized group of fishes closely related to the Arthrodira. 
 
BIBLIOGRAPHY 
 
 Adams, L. A. 
 
 1919. A memoir on the phylogeny of the jaw muscles in recent and fossil 
 vertebrates. New York. Acad. Sci., Ann., Vol. 28. 
 
 Acassiz, J. L. R. 
 
 1844. Monographic des Poissons Fossiles Du Vieux Gres Rouge ou Systeme 
 Devonien (Old Red Sandstone) Des lies Brittaniques Et De Russie. 
 
 Allen, W. F. 
 
 1905. The blood-vascular system of the Loricati, the mail-cheeked fishes. 
 Washington Acad. Sci., Proc., Vol. 7. 
 
 Allis, E. Ph. 
 
 1889. The anatomy and development of the lateral line system in Amia calva. 
 Journ. Morph., Vol. 2. 
 
 1897. The cranial muscles and cranial and first spinal nerves in Amia calva. 
 Journ. Morph., Vol. 12. 
 
 1898. On the morphology of certain of the bones of the cheek and snout of 
 Amia calva. Journ. Morph., Vol. 14. 
 
 1899. On certain homologies of the squamosal intercalar, exoccipital and 
 extrascapular bones of Amia calva. Anat. Anz., Bd. 16. 
 
 1900. The lateral sensory canals of Polypterus bichir. Anat. Anz., Bd. 17. 
 
 1001. The lateral sensory canals, the eye-muscles, and the peripheral distribu- 
 tion of certain of the cranial nerves of Mustelus Icevis. Quart. Journ. Micr. 
 Sci., Vol. 45. 
 
 1003. The skull, and the cranial and first spinal muscles and nerves in Scomber 
 
 scomber. Journ. Morph., Vol. 18. 
 1005. The latero-sensory canals and related bones in fishes. Internat. Monat- 
 
 schrift f. Anat. u. Physiol., Bd. 21. 
 1908a. The pseudobranchial and carotid arteries in Amcirus. Anat. Anz., Bd. 33. 
 1908b. The pseudobranchial and carotid arteries in Polypterus. Anat. Anz., 
 
 Bd. 33. 
 1909a. The cranial anatomy of the mail-cheeked fishes. Zoologica, Vol. 22. 
 1909b. The pseudobranchial and carotid arteries in the gnathostome fishes. 
 
 Zool. Jahrb. Abt. Anat., Bd. 27. 
 1911a. The pseudobranchial and carotid arteries in Polyodon spathula. Anat. 
 
 Anz., Bd. 39. 
 1911b. The pseudobranchial and carotid arteries in Chlamydosclachus anguineiu. 
 
 Anat. Anz., Bd. 39. 
 1912a. The branchial, pseudobranchial and carotid arteries in Chimera (Hydro- 
 
 lagus) colliei. Anat. Anz., Bd. 42. 
 1912b. The branchial, pseudobranchial and carotid arteries in Hcptanchus 
 
 (Notidanus) cinerens. Anat. Anz., Bd. 41. 
 1912c. The branchial, pseudobranchial and carotid arteries in Raja radiata. 
 
 Anat. Anz., Bd. 44. 
 I9i2d. The pseudobranchial and carotid arteries in Esox, Salmo and Gadus, 
 
 together with a description of the arteries in the adult Amia. Anat. Anz., 
 
 Bd. 41. 
 191 3. The homologies of the ethmoidal region of the Selachian skull. Anat. 
 
 Anz., Bd. 44. 
 1914a. The pituitary fossa and trigemino-facialis chamber in Selachians. Anat. 
 
 Anz., Bd. 46. 
 1914b. The pituitary fossa and trigemino-facialis chamber in Ceratodus for- 
 
 steri. Anat. Anz., Bd. 46. 
 
 191 
 
192 Field Museum of Natural History — Geology, Vol. IV. 
 
 1914c. Certain homologies of the palatoquadrate of Selachians. Anat. Anz., 
 Bd. 45- 
 
 1915. The homologies of the hyomandibula of the gnathostome fishes. Journ. 
 Morph., Vol. 26. 
 
 1918a. The myodome and trigemino-facialis chamber of fishes, etc. Washing- 
 ton. Nat. Acad. Sci., Proc, Vol. 4. 
 
 1918b. On the origin of the hyomandibula of the Teleostomi. Anat. Record, 
 Vol. 15. 
 
 1918c. The ophthalmic nerves of gnathostome fishes. Journ. of Comp. 
 Neurology, Bd. 30. 
 
 1919a. The myodome and trigemino-facialis chamber of fishes and the cor- 
 responding cavities in higher vertebrates. Journ. Morph., Vol. 32. 
 
 1919b. The homologies of the squamosal bone of fishes. Anat. Record, Vol. 17. 
 
 1920. The branches of the branchial nerves of fishes, with special reference 
 to Polyodon spat hula. Journ. of Comp. Neurology, Vol. 32. 
 
 1922a. The cranial anatomy of Polypterus, with special reference to Polyp- 
 terus bichir. Journ. of Anat., Vol. 56. 
 
 1922b. The myodome and the trigemino-facialis chamber in the Coelacan- 
 thidae, Rhizodontidae and Palaeoniscidae. Journ. of Anat., Vol. 56. 
 
 1923. The cranial anatomy of Chlamydoselachus anguineus. Acta Zoologica, 
 Bd. 4. 
 
 Brown, C. 
 
 1900. Ueber das Genus Hybodus und seine systematische Stellung. Palaeonto- 
 graphica, Bd. 44. 
 
 Bryant, H. C. 
 
 1918. Catalog of the fossil fishes in the Museum of the Buffalo Soc. of Nat. 
 Sci. Buffalo Soc. Nat. Sci., Bull., Vol. 12. 
 
 Cole, F. J. 
 
 1896a. On the cranial nerves of Chimcera monstrosa, with a discussion of the 
 lateral line system and of the morphology of the chorda tympani. Edin- 
 burgh. Roy. Soc. Trans., Vol. 38. 
 
 1896b. On the sensory and ampullary canals of Chimcera. Anat. Anz., Bd. 12. 
 
 1898. Observations on the structure and morphology of the cranial nerves and 
 lateral sense organs of fishes ; with special reference to the genus Gadus. 
 London, Linn. Soc, Trans. (Zool.), Vol. 7. 
 
 Cope, E. D. 
 
 1891. On the characters of some Palaeozoic fishes. U. S. Nat. Mus., Proc, 
 Vol. 14. 
 
 Danforth, C. H. 
 
 1912. The heart and arteries of Polyodon. Journ. Morph., Vol. 23. 
 
 Daniel, J. F. 
 
 1915. The anatomy of Heterodontus francisci. II. Journ. Morph., Vol. 26. 
 
 Dean, B. 
 
 1896. On the vertebral column, fins and ventral armoring of Dinichthys. New 
 York Acad. Sci., Trans., Vol. 15. 
 
 1901. Palseontological Notes. New York Acad. Sci., Mem., Vol. 2. 
 
 1904. In the matter of the Permian fish Menaspis. Amer. Geologist, Vol. 34. 
 1909. Studies on fossil fishes (Sharks, Chimseroids, and Arthrodires). Ameri- 
 can Mus. Nat. Hist., Mem., Vol. 9. 
 
 Eastman, Ch. R. 
 
 1906. Structure and relations of Mylostoma. Cambridge. Mus. of Comp. 
 Zool. Harvard., Bull., Vol. 50. 
 
Macropetalichthyids and other Arthrodires — Stensio 193 
 
 1907. Mylostomid dentition. Cambridge. Mus. of Comp. Zool. Harvard., 
 
 Bull., Vol. 50. 
 1908a. Devonic fishes of the New York formations. Albany. N. Y. State 
 
 Museum, 60th Annual Report, 106, Vol. 5. Appendix 7 (continued). 
 1908b. Devonian fishes of Iowa. Iowa Geol. Surv., Vol. 18. Ann. Rep. 1907. 
 
 Eichwald, C. Ed. 
 
 i860. Lethaea Rossica., 1 :2. 
 
 Ewart, J. C. 
 
 1892. The lateral sense organs of Elasmobranchs. I. The sensory canals of 
 Lcemargus. Edinburgh. Roy. Soc, Trans., Vol. 37, Pt. I. 
 
 Ewart, J. C. and Cole, F. J. 
 
 1895. On the dorsal branches of the cranial and spinal nerves of Elasmo- 
 branchs. Edinburgh. Roy. Soc., Proc., Vol. 20. 
 
 Ewart, J. C. and Mitchel, J. C. 
 
 1892. On the lateral sense organs of Elasmobranchs. II. The sensory canals 
 of the common skate (Rata batis). Edinburgh. Roy. Soc., Trans., Vol. 37. 
 
 Fritsch, A. J. 
 1895. Fauna der Gashohle u. der Kalksteine der Permformation Bohmens, 
 Bd. 3. Prague. 
 
 FURBRINGER, K. 
 
 1904. Beitrage zur Morphologie des Skelettes der Dipnoer, nebst Bemerkun- 
 gen uber Pleuracanthiden, Holocephalen and Squaliden (In Scmon, R. Zoo- 
 logische Forschungsreisen in Australien und dem Malayischen Archipel, 
 Bd. 1). Jena. Med. Nat. Gesellsch., Denkschriften 4. 
 
 Garman, S. 
 
 1888. On the lateral canal system of the Selachia and Holocephala. Cam- 
 bridge. Mus. Comp. Zool. Harvard, Bull., Vol. 17. 
 
 Gegenbaur, C. 
 
 1872. Untersuchungen zur Vergl. Anat. der Wirbelthiere. III. Das Kopfskelet 
 
 der Selachier, etc. 
 1898. Vergleichende Anatomie der Wirbelthiere mit Berucksichtigung der 
 Wirbellosen. Bd. 1. Leipzig. 
 
 1912. Ueber die occipital region und die ihr benachbarten Wirbel der Fische. 
 Gesammelte Abhandlungen von C. Gegenbaur. Bd. 3. Leipzig. 
 
 Goodey, A. T. 
 
 1910. A contribution to the skeletal anatomy of the filled shark, Clamydose- 
 lachus anguineus. London. Zool. Soc, Proc. 
 
 Goodrich, E. S. 
 
 1009. Vertebrata Craniata, Fasc. I : Cyclostomes and Fishes in Lankester's 
 Treatise on Zoology, Part IX. 
 
 Greil, A. 
 
 1913. Entwicklungsgeschicht des Kopfes und des Blutgefassystems von Cera- 
 todus fosteri. Part 2: Die epigenetischen Erwerbungen wahrend der 
 Stadien 39-48. Jena. Med. Nat. Gesellsch., Denkschr., Bd. 4. 
 
 Grosser, O. 
 
 1907. Die Elemente des Kopfvenensystems der Wirbelthiere. Verhandl. Anat. 
 Ges. Supplement. Anat. Anz., Bd. 30. 
 
 Hawkes, O. A. 
 
 1906. The cranial and spinal nerves of Chlatnydoselachus anguineus. London. 
 Zool. Soc, Proc 
 
194 Field Museum of Natural History — Geology, Vol. IV. 
 
 Hennig, E. 
 
 1907. Macropetalichthys pelmensis, n.sp., Stuttgart, Centralbl. f. Min. Geol. 
 Pal. 
 
 Herrick, C. J. 
 
 1899. The cranial and first spinal nerves of Menidia; a contribution upon the 
 nerve components of the bony fishes. Journ. of Comp. Neurology, Vol. 9. 
 
 1900. A contribution upon the cranial nerves of the codfish. Journ. of Comp. 
 Neurology, Vol. 10. 
 
 1901. The cranial nerves and cutaneous sense organs of the North American 
 siluroid fishes. Journ. of Comp. Neurology, Vol. 11. 
 
 Hussakof, L. 
 
 1906. Studies on the Arthrodira. Amer. Mus. Nat. Hist., Mem., Vol. 9. 
 
 Huxley, Th. H. 
 
 1864. Lectures on the elements of comparative anatomy. London. 
 
 Iwanzow, N. 
 
 1887. Der Scaphirhynchus. Vergleichendo-anatomische Beschreibung. Moscou. 
 Soc. Imper. des Nat., Bull. 
 
 Jaekel, O. 
 
 1891. Ueber Mcnaspis armata Ewald. Berlin. Gesellsch. Naturforsch. Freunde. 
 Sitzber. 
 
 1902. Ueber Coccosteus und die Beurtheilung der Placodermen. Berlin. 
 Gesellsch. Naturforsch. Freunde, Sitzber. 
 
 1906a. Neue Wirbelthierfunde aus dem Devon von Wildungen. Berlin. 
 
 Gesellsch. Naturforsch. Freunde, Sitzber. 
 1906b. Einige Beitrage zur Morphologie der altesten Wirbelthiere. Berlin. 
 
 Gesellsch. Naturforsch. Freunde, Sitzber. 
 
 1907. Uber Pholidosteus n.g. Die Mundbildung n. die Korperform der Pla- 
 codermen. Berlin. Gesellsch. Naturforsch. Freunde. Sitzber. 
 
 191 1. Die Wirbelthiere. Berlin. 
 
 1919. Die Mundbildung der Placodermen. Berlin. Gesellsch. Naturforsch. 
 
 Freunde, Sitzber. 
 1921. Palaeontologische Berichte. II. Schadelprobleme. Palaeontologische Zeit- 
 
 schrift, Vol. 3. 
 
 Johnston, J. B. 
 
 1898. Hind brain and cranial nerves of Acipenser. Anat. Anz., Bd. 28. 
 
 1 901. The brain of Acipenser, etc. Zool. Jahrb., Abt. f. Anat., Bd. 15. 1902. 
 
 Kaer, J. 
 
 1915. Upper Devonian fish remains from Ellesmere Land. Rep. of the Second 
 Norwegian Arctic Exped. in the "Fram", 1898- 1902. No. 33. Publ. by 
 Videnskabs-Selskabet i Kristiania. 
 
 191 6. Spitzbergens Devoniske Faunaer. i6de Skandinaviska Naturforskarmotet. 
 Kristiania. Forhandl. 
 
 Kingsbury, B. F. 
 
 1897. The structure and morphology of the oblongata in fishes. Journ. of 
 Comp. Neurology, Vol. 7. 
 
 Koken, E. 
 1907. Ueber Hybodus, Geol. Palaeontol. Abh., Vol. 5. 
 
 Landacre, F. L. 
 
 1916. The cerebral ganglia and early nerves of Squalus acanthias. Journ. 
 Comp. Neurology, Vol. 27. 
 
Macropetalichthyids and other Arthrodires — Stensio 195 
 
 Lehn, Ch. 
 
 1918. Beitrage zur Kenntnis des Primordialschadels von Polypterus. Zeit- 
 schr. f. angew. Anat. u. Konstitutionlehre, Vol. 2. 
 
 Luther, A. F. 
 
 1909. Untersuchungen uber die vom N. trigeminus innervierte muskulatur der 
 Selachier (Haie and Rochn) unter Beriicksichtigung ihrer Beziehungen zu 
 benachbarten Organen. Soc. Sci. Fennicae, Acta., Tome 36. 
 
 Moodie, R. L. 
 
 1908. The lateral line system in extinct Amphibia. Journ. Morph., Vol. 14. 
 1915 A further contribution to a knowledge of the lateral line system in 
 extinct Amphibia. Jour, of Comp. Neurology, Vol. 25. 
 
 Newberry, J. S. 
 
 1889. The Palaeozoic fishes of North America. U. S. Geol. Surv., Monogr., 
 No. 16. 
 
 Norris, H. W., and Hughe, Sally P. 
 
 1920. The cranial, occipital, and anterior spinal nerves of the Dogfish, Squalus 
 acanthias. Journ. of Comp. Neurology, Vol. 31. 
 
 O'Donoghue, Ch. H. 
 
 1914. Notes on the circulatory system of Elasmobranchs. I. The venous sys- 
 tem of the dogfish (Scyllium canicula). London. Zool. Soc, Proc. 
 
 Parker, W. K. 
 
 1882. On the structure and development of the skull in Sturgeons (Acipenscr 
 ruthcnus and A. sturio). London. Roy. Soc, Phil. Trans., Vol. 173, Part I. 
 
 Parker, T. J. 
 
 1887. On the blood-vessels of Mustclus antarcticus, etc. London. Roy. Soc. 
 Phil., Trans., Vol. 177. Part II (1886). 
 
 Plate, L. 
 
 1924. Allgemeine Zoologie u. Abstammungslehre. Part II. Jena. 
 
 Pollard, H. B. 
 
 1892a. On the anatomy and phylogenetic position of Polypterus. Zool. Jahrb., 
 
 Abt. f., Anat., Bd. 5. 
 1892b. The lateral line system in Siluroids. Zool. Jahrb., Abt. f. Anat., Bd. 5. 
 
 Reis, O. M. 
 
 1896. Ueber Acanthodes bronni Agassiz. Morphol. Arbeiten herausg. von 
 Schwalbe. Bd. 6., Jena. 
 
 Retzius, M. G. 
 
 1881. Das Gehororgan der Wirbelthiere, Vol. I. Stockholm. 
 
 Rex, H. 
 
 1891. Beitrage zur Morphologie der Hirnvenen der Elasmobranchier. Morph. 
 Jahrbuch, Bd. 17. 
 
 Ruge, G. H. 
 
 1897. Ueber das peripherische Gebiet des Nervus Facialis bei Wirbelthieren. 
 In Festschrift zum siebenzigsten Geburtstage von C. Gegenbaur, Bd. 3. 
 
 Stensio, E. 
 
 1921. Triassic fishes from Spitzbergen. Part I. Vienna. 
 
 1922. Uber zwei Coelacanthiden aus den Oberdevon von Wildungen. Palaeon- 
 tologische Zeitschrift, Bd. 4. 
 
 1923. Notes on certain Crossopterygians. London. Zool. Soc, Proc. 1922. 
 
196 Field Museum of Natural History — Geology, Vol. IV. 
 
 1925. Triassic fishes from Spifzbergen. Part II. Stockholm. K. Vet. Akad; 
 Handl., Ser. Ill, Vol. 2. 
 
 Strong, O. S. 
 
 1895. The cranial nerves of Amphibia. Journ. Morph., Vol. 10. 
 
 Teller, Fr. J. 
 
 1891. Ueber den Schadel eines fossilen Dipnoers, Ceratodus sturii, n.sp. aus 
 den Schichten der obern Trias der Nordalpen. Wien. Geol. Reichsanstal, 
 Abhandlungen, Bd. 15. 
 
 Traquair, R. H. 
 
 1878. On the genera Dipterus, Palcedaphus, Holodus and Cheirodus. Ann. 
 and Mag. Nat. Hist., Ser. 5, Vol. 2. 
 
 1889. On Homosteus compared with Coccosteus, Ag. Geol. Mag., n.s., del. Ill, 
 Vol. 6. 
 
 1890. On the structure of Coccosteus decipiens. Ann. and Mag. Nat. Hist., 
 Ser. 6, Vol. 5. 
 
 Veit, O. 
 
 1907. Ueber einige Besonderheiten am Primordial Cranium von Lepidosteus 
 
 osscus. Anat. Hefte, Abt. I., Bd. 33. 
 191 1. Beitrage zur Kenntnis des Kopfes der Wirbelthiere. I. Die Entwick- 
 
 lung des Primordial Cranium von Lepidosteus osseus. Anat. Hefte, Abt. I., 
 
 Bd. 44- 
 
 Vetter, B. 
 
 1874. Untersuchungen ur vergleichenden Anatomie der Kiemen- und Kierer- 
 musculatur der Fische. Teil I. Jen. Zeitschr., Bd. 8. 
 
 Watson, D. M. S. 
 
 1 92 1. On the Coelacanth Fish. Ann. and Mag. Nat. Hist., Ser. 9, Vol. 8. 
 
 Watson, D. M. S. and Day, H. 
 
 1 916. Notes on some Palaeozoic fishes. Manchester. Lit. & Phil. Soc, Mem. 
 & Proa, Vol. 60, Part I. 
 
 Watson, D. M. S. and Gill, E. L. 
 
 1923. The structure of certain Palaeozoic Dipnoi. London. Linn. Soc, Journ. 
 (Zool.), Vol. 35- 
 
 Weidenreich, F. 
 
 1923. Knochenstudien I. t)ber Aufbau u. Entwicklung des Knochens u. den 
 Charakter des Knochengewebes. Zeitschrift fur Anatomie und Entwick- 
 lungsgeschichte, Bd. 69. 
 
 WlMAN, C. 
 
 1914. Uber die Stegocephalen aus der Trias Spitzbergens. Upsala. Geol. Inst., 
 
 Bull., Bd. 13:1. 
 1916. Neue Stegocephalen funde aus dem Posidonomyaschiefer Spitzbergens. 
 
 Upsala. Geol. Inst., Bull., Bd. 13:2. 
 
 White, Ph. J. 
 
 1899. The skull and visceral skeleton of the Greenland shark, Lcemargus mi- 
 crocephalus. Edinburgh. Roy. Soc, Trans., Vol. 37. 
 
 Withe, J. W. van. 
 
 1882. Ueber das Visceralskelett und die Nerven des Kopfes der Ganoiden und 
 von Ceratodus. Niederl. Archiv. f. Zool., Bd. 5. 
 
Macropetalichthyids and other Arthrodires — Stensio 197 
 
 Woodward, A. S. 
 
 1891. Catalogue of the fossil fishes in the British Museum (Natural His- 
 tory). Part II. London. 
 1916-1919. The fossil fishes of the English Wealden and Purbeck formations. 
 London. Palaeont. Soc. 
 
 1921. Observations on some extinct Elasmobranch fishes. London. Linnean 
 Soc. Proc. Session 133. Presidential address 1921 
 
 1922. Observations on Crossopterygian and Arthrodiran fishes. London. Lin- 
 nean Soc. Proc. Session 132 Presidential address. 1922. 
 
 1924. Une nouvel Elasmobranche (Bratoselache jesuvosti gen. et spec, nov.) 
 du calcaire carbonifere inferiem de Denee. Livre Jubilaire publ. a l'oc- 
 casion du cinquantenaire de la fondat de la Soc. Geol. Belg. Liege. 
 
 Workman, I. S. 
 
 1900. The ophthalmic and eye muscle nerves of the cat-fish. Journ. of Comp. 
 Neurology, Vol. 10. 
 
EXPLANATION OF PLATES 
 
 All the photographs here shown of the specimen of Macropeta- 
 lichthys rapheidolabis in the possession of Field Museum of Natural 
 History at Chicago were made by the photographic department of that 
 Museum. The other specimens figured here were photographed by 
 myself in the photographic laboratory of the Palaeozoological Depart- 
 ment of the Royal State Museum at Stockholm. 
 
 In those cases in which no special statements as to the scale of the 
 figures are given the figures have been reproduced in natural size or 
 have been slightly diminished. 
 
 In Fig. 4, PI. XXIX and Fig. 2, PI. XXXI, the bone is brighter than 
 the stone ; in all the other figures the bone is black or, in any case, 
 darker than the stone. Sections through cavities and canals or exter- 
 nal openings of canals are generally shown in a very bright, sometimes 
 even an almost white tone. Fig. 1, PI. XXX is an exception to this 
 however, the sensory canals and their tubuh and the external opening 
 of the canal for the ductus endolymphaticus being black. 
 
EXPLANATION OF PLATE XIX 
 
 Macropctalichthys rapheidolabis. The specimen in Field Museum of Nat- 
 ural History at Chicago (Museum No. P 1154) in, dorsal view. The dermal 
 bones of the cranial roof being weathered away with the exception of the sensory 
 canal ridges, the outlines of the primordial neurocranium are rather clearly ex- 
 hibited. The specimen lacks the posterior narrow division of the occipital region, 
 and the most anterior part of the ethmoidal region is imperfectly preserved. The 
 ridge for the hindmost part of the cephalic division of the lateral line has become 
 so abraded that the canals for nerve branches to it from the canal for the n. 
 linese lateralis are clearly seen. 
 
 cl, fine canal coming from below; the canal probably transmitted a vagus 
 branch, a fine branch from the n. linear lateralis and perhaps in addition some 
 vessel ; c.pp, sensory canal commissure, probably corresponding to the posterior 
 head line of pit organs in fishes in general; ifc, infraorbital sensory canal; Ic, 
 cephalic division of the lateral line; Idi-ldz, dorsally ascending canals for nerve 
 branches to the lateral line; the canals issue from the canal (n./) for the n. 
 lineae lateralis ; orb, orbital entrance ; par, pineal foramen ; soc, supraorbital sen- 
 sory canal. 
 
FIELD MUSEUM OF NATURAL HISTORY. 
 
 GEOLOGY, VOL. IV, PL. XIX. 
 
 "I** 
 
 c.jop 
 
 ld\ 
 
LI6KAKY 
 
 UNIVERSIIY Of ILLINOIS 
 
 URBANA 
 
EXPLANATION OF PLATE XX 
 
 Macropetalichthys rapheidolabis. Same specimen as in the preceding plate 
 but with a postero-lateral part of the cranial roof removed to show in dorsal 
 view the labyrinth cavity, a part of the cranial cavity and certain canals. As 
 all these structures are lined by a thin, perichondral bone layer, which is black 
 from infiltration by bitumen, they consequently appear black in the figures ex- 
 cept when there are sections through them. 
 
 c.ef, canal perhaps traversed by the arteria efferens hyoidea on its way to 
 the radix aorta? (lateral dorsal aorta) ; c.hy, canal for the vena hyoidea (prob- 
 ably, in the external part, for the arteria efferens hyoidea too) ; c.pp, sensory 
 canal commissure, probably corresponding to the posterior head line of pit-organs 
 in fishes in general ; c.sem.ant, division of the labyrinth cavity for the canalis 
 semicircularis anterior; cv, cavum cerebrale cranii (with its lining bone mem- 
 brane) ; only a posterior part of it is seen; d.cnd, canal for the ductus endolym- 
 phaticus, ifc, infraorbital sensory canal ; in, two canals for nerve branches to 
 the infraorbital sensory canal. The direction of the two canals makes it fully 
 obvious that the nerve-branches transmitted by them must have come from a 
 prootic nerve, probably the r. oticus lateralis or its equivalent; /«, the canal 
 for the jugular vein; only the posterior part behind the sinus si (see PI. XXII), 
 shown; lab.cav, labyrinth cavity (partly exposed. Like the cavum cerebrale it 
 has its lining bone membrane preserved and is therefore black in the figure) : 
 Ic, cephalic division of the lateral line; Idi-lds, canals ascending from the canal 
 n.l. for the n. linese lateralis to the cephalic division of the lateral line; Is, lamella 
 of bone connecting canals transmitting lateralis branches for the innervation of the 
 posterior part of the internal and external bone layers with each other ; orb, orbital 
 entrance ; par, pineal foramen ; soc, supraorbital sensory canal ; v.lb, canal from 
 the labyrinth ; cavity to the dorsal part of the proximal portion of the vagus canal. 
 The canal probably transmitted a vein. On the right side the canal is partly seen 
 in section close to its posterior end; X™, the canal for the vagus and the vena 
 cerebralis posterior (proximal part). 
 
FIELD MUSEUM OF NATURAL HISTORY. 
 
 GEOLOGY, VOL. IV, PL. XX. 
 
 par 
 
 C.se/n.aat 
 
 ■ 
 
 -SOC 
 
 ,lfc 
 
 m 
 
 ** 
 
 -c.pp 
 
 Id 
 
 r\^ \ ! Xv/l 
 
 \idj i is 
 
 i** 
 
LIBKAKY 
 
 UNIVERSUY Of ILLINOIS 
 
 URBANA 
 
EXPLANATION OF PLATE XXI 
 
 Macropctalichthys raphcidolabis. Same specimen as in Pis. XIX and XX. 
 Primordial neurocranium in ventral view. Certain parts of the left side (right 
 in the figure) and of the anterior end removed. 
 
 c.car.ext, canal for the arteria carotis external (represented merely by a 
 groove, as its filling of stone and its external wall have been destroyed by 
 weathering) ; c.car.int, canal for the arteria carotis interna (represented to a 
 large extent by its filling of stone, as its external wall has been abraded). The 
 most anterior white spot marks the place at which the internal carotid curved 
 upward to the cavum cerebrale; c.com, canal for the arteria carotis communis 
 (its external wall and in part also its stone filling destroyed. In the anterior 
 part, in which the stone filling is lost, it appears as a groove) ; c.hy, canal for 
 the vena hyoidea (merely the proximal part completely preserved in the figure, 
 the distal part being, as is well seen, represented there by a groove) ; c.ophth.lat, 
 canal for the n. ophthalmicus lateralis (anterior part) ; c.paU the canal for the 
 r. palatinus facialis through the orbital floor (dorsal opening) ; era, canal for 
 the radix aortae (lateral dorsal aorta). The canal is imperfectly preserved, 
 being represented merely by a groove in the figure ; csem, division of the labyrinth 
 cavity for a semicircular canal, probably the c. semicircularis externas ; cv, cavum 
 cerebrale cranii (merely a small posterior part is shown) ; dtx, indication of a 
 canal, perhaps for the lateralis branch that accompanied the n. glossopharyngeus 
 from the cavum cerebrale; jgx, canal for the jugular vein from the orbit to the 
 mandibular vein (section) ; ju, canal, which through its anterior opening trans- 
 mitted the vena mandibularis (the anterior part not fully preserved. The 
 anterior opening was situated more laterally than in the figure) ; jih and jih, an- 
 terior and posterior opening of the canal ju; na, nasal aperture; nf, nasal fon- 
 tanelle ; n.L, canal for the n. linae lateralis ; olf, olfactory capsule ; rm, ophth. lat, 
 branch for a lateralis nerve from the canal for the n. ophthalmicus lateralis to the 
 supraorbital sensory canal ; si, sinus formed by the confluence of the canal ju and 
 the canal era (its ventral wall has been removed, so that its stone filling covered 
 by the lining membrane of bone is visible) : soc, supraorbital sensory canal (a 
 small anterior part) : s.ra, groove for the radix aortae (lateral dorsal aortae) ; Sx, 
 groove anterior of the external opening of the vagus canal ; VII, canal for the 
 n. facialis on the left side — right in the figure — the external part in section; 
 IX, canal for the n. glossopharyngeus (external opening) ; X™, external opening 
 of the canal for the vagus and the vena cerebralis posterior. 
 
FIELD MUSEUM OF NATURAL HISTORY 
 
 olf 
 
 GEOLOGY, VOL. IV, PL. XXI. 
 
 c.ophth.lat 
 
 rm.ophth.la.t 
 
 -c.car.int 
 
LIBRARY 
 cRSiTY Of ILLINOIS 
 URBANA 
 
 
EXPLANATION OF PLATE XXII 
 
 Macropetalichthys rapheidolabis. Same specimen as in the three preceding 
 plates. Imperfect impression of the ventral surface of the primordial neuro- 
 cranium with certain lateral parts of this remaining in their original position on 
 the left side. 
 
 c.car.int, canal for the arteria carotis interna (the white opening denoted 
 is the place at which the artery turned upwards and ascended to the cavum 
 cerebrale) ; c.hy, canal for the venal hyoidea (on the right side is the external 
 opening shown as impression) ; c.cf, canal perhaps for the arteria efferens hyoidea 
 on its way to the radix aortae (lateral dorsal aorta) ; c.ophthJat, canal for the 
 n. ophthalmicus lateralis; c.ophth.sup.Vf, canal perhaps for the r. ophthalmicus 
 superficialis V through the antorbital process (merely the anterior opening seen 
 as impression in the figure) ; c.pah, canal for the r. palatinus facialis through 
 the orbital floor (merely the ventral opening into the internal carotid canal is 
 seen here) ; era, canal for the radix aortae (lateral dorsal aorta) ; ifc, infraor- 
 bital sensory canal ; in, two canals for nerve branches to the infraorbital sen- 
 sory canal. The direction of the canals makes it fully evident that the nerve 
 branches transmitted by them must have come from a prootic nerve, probably 
 the r. oticus lateralis or its equivalent; jg, canal for the vena jugular is from the 
 orbit to the mandibular vein (section) : ju, canal for the jugular vein pos- 
 terior of the sinus si; ju u anterior opening of the canal ju. The opening was 
 traversed by the mandibular vein (impression) ; olf, olfactory capsule (impres- 
 sion of ventral side) ; si, sinus arisen by the confluence of the canals ju and 
 era; s.pal, groove for the r. palatinus facialis on the lower side of the primor- 
 dial' neurocranium (seen as impression here and therefore appearing as a ridge) ; 
 s.ra, groove for the radix aortae (lateral dorsal aorta). The groove is preserved 
 as an impression and has therefore the shape of a ridge; VII, canal for the 
 n. facialis (section) ; IX, canal for the n. glossopharyngeus (section close at the 
 external opening into the groove s.ra; Xvn, canal for the vagus and the vena 
 cerebralis posterior (abraded impression of the left external opening). 
 
FIELD MUSEUM OF NATURAL HISTORY. 
 
 GEOLOGY. VOL. IV, PL. XXII. 
 
 ophth.lat 
 
 jyr^j***sp 
 
 ophth 
 
 s r<x 
 
LIBRARY 
 
 UNIVERSITY OF ILLINOIS 
 
 URBANA 
 
191 
 
EXPLANATION OF PLATE XXIII 
 
 Macropetalichthys rapheidolabis. Same specimen as in the preceding plates. 
 The primordial neurocranium in ventral view with certain parts removed on 
 the left side (right in the figure) to show the cavum cerebrale and the exits 
 of the canals for the cranial nerves from this. 
 
 afr, acustico-facialis recess; a.opt, division of the external opening of the 
 opticus canal traversed by the arteria optica; c.car.ext, canal for the arteria 
 carotis externa (as its filling of stone and ventral wall have been destroyed, it 
 appears as a groove in the figure) ; c.car.int, canal for the arteria carotis in- 
 terna. On the right side (left in the figure) the suborbital part is in almost the 
 entire extension represented by the filling of stone, as the ventral wall has been 
 destroyed. The foramen somewhat in front of the stone filling is the place 
 at which the artery curved upward to ascend to the cavum cerebrale. On the 
 left side (the right in the figure) is seen the ascending part of the same canal; 
 c.com, canal for the arteria carotis communis (in its anterior part, in which the 
 ventral wall and stone filling have been destroyed, it appears as a groove, in its 
 posterior part, in which merely the external wall is lacking, the filling of stone 
 is clearly seen) ; c.hy, canal for the vena hyoidea (the distal part destroyed and 
 represented in the figure merely as a groove) ; c.ophth.lat, canal for the n. oph- 
 thalmicus lateralis (merely an anterior part and a posterior part discernible) ; 
 c.ophth.prof, canal for the r. ophthalmicus profundus (proximal part) ; c.ophth. 
 sup.V, canal for the r. ophthalmicus superficialis trigemini to the orbit (entire 
 canal with the external opening) ; c.ophth. sup. Vf, canal probably for the 
 r. ophthalmicus superficialis trigemini through the preorbital process ; era, canal 
 for the radix aortae (lateral dorsal aorta). Its external wall and filling being 
 destroyed it appears merely as a groove; cv, cavum cerebrale cranii (with its 
 lining membrane of bone — the inner bone layer) ; dn, indication of a canal to 
 the dorsal side of the cranial roof, probably for a lateralis branch that accom- 
 panied the n. glossopharyngeus from the cavum cerebrale; jih, anterior opening 
 of the canal ju (as the lateral parts are broken off the actual anterior opening 
 of the canal ju was situated more laterally than in the figure. We are there 
 in fact concerned with a section through the part of the canal ju situated an- 
 terior of the sinus si) ; jih, position of the posterior opening of the canal ju 
 (the external wall and the filling of stone being destroyed, only a groove is 
 seen at this place in the figure) ; na, nasal aperture; nf, nasal fontanelle; n.l, 
 canal for the n. lineae lateralis ; olf, olfactory capsule ; rm.ophth.lat, fine canal for 
 a branch from the n. ophthalmicus lateralis to the supraorbital sensory canal; 
 si, sinus in the lateral wall arisen by confluence of the canals ju and era; soc, 
 supraorbital sensory canal (anterior part) ; s.ra, groove for the radix aortae 
 (lateral dorsal aorta) ; S*, groove leading from the vagus canal ; I, canal for 
 the tractus olfactorius; II, canal for the n. opticus (entire canal with external 
 opening) ; III, canal for the n. oculimotorius ; V, trigeminus recess ; Vi, i canal for 
 the r. maxillaris and r. mandibularis trigemini (section through the proximal 
 part) ; VII, canal for the n. facialis (proximal part in section) ; VIII, canal for 
 the n. acusticus ; IX, canal for the glossopharyngeus (proximal part) ; Xtd, canal 
 for the n. vagus and the vena cerebralis posterior. 
 
FIELD MUSEUM OF NATURAL HISTORY. 
 
 GEOLOGY, VOL. IV, PL. XXIII. 
 cophth.'ar 
 
 ; rmophT'i I at 
 ,-Soc 
 
 7,1 
 
LIBRARY 
 UNIVERSITY OF ILLINOIS 
 
 URBANA 
 

 
 
 
 
 
EXPLANATION OF PLATE XXIV 
 Macropetalichthys rapheidolabis. Same specimen as in the preceding plates. 
 Fig. I. Primordial neurocranium seen from the right side. 
 
 Fig. 2. A transversal section through the lateral part of the primordial neu- 
 rocranium, along the canal for the n. vagus and the vena cerebralis posterior. The 
 impression of the anterior side of this canal is shown in black. In one place a 
 part of its filling of stone remains (with white external end). 
 
 Fig. 3. Primordial neurocranium in posterior aspect. The place from which 
 the posterior narrow division of the occipital region issues is well shown. Further, 
 we can see at this place the posterior end of the middle division of the cavum 
 cerebrale and a section through the cavum cerebrale at the transition to the 
 posterior narrow division (white). The canal for the n. lineae lateralis is ex- 
 posed during a rather large part of its course, as the cranial wall has been 
 destroyed at this place. 
 
 c.hy, canal for the vena hyoidea (the septum between it and the anterior 
 opening of the canal ju, partly destroyed) ; c. phth.su p. V, canal through the 
 preorbital process, probably for the r. ophthalmicus superficialis trigemini; c.pp, 
 sensory canal commissure, probably corresponding to the posterior head line of 
 pit organs in fishes in general; cv, cavum cerebrale cranii (a section of it at 
 the transition to the posterior narrow division is shown in white) ; dix. indication 
 of a canal to the dorsal side of the primordial neurocranium. The canal pro- 
 bably transmitted a lateralis branch that might have accompanied the n. glosso- 
 pharyngeus from the cavum cerebrale; dx, dorsal branch from the vagus canal 
 for a lateralis nerve; dxn, and dn>, branches of the canal ds, the former of 
 which lodged the nerve to the antero-laterally running part of the cephalic di- 
 vision of the lateral line, while the latter lodged the nerve to the sensory canal 
 commissure c.pp; ifc, infraorbital sensory canal ; jui, anterior opening of the 
 canal ju (the septum between it and the external part of the canal c.hy, partly 
 destroyed); Ic, cephalic division of the lateral line; ldi-ld e , canals for branches 
 from the n. lineas lateralis to the most posterior part of the cephalic division of 
 the lateral line ; na, nasal aperture ; nf, nasal f ontanelle ; n.l, canal for the lineae 
 lateralis (exposed in the distal part owing to the incomplete state of preservation of 
 the posterior cranial wall ; orb, orbit ; v. lb, canal from the postero-dorso-medial 
 part of the labyrinth cavity to the proximal dorsal part of the vagus canal. The 
 canal probably transmitted a vein (section) ; vy, opening of a canal of doubtful 
 importance (perhaps for some vessel) ; VII, canal for the n. facialis; XV, canal 
 for the n. glossopharyngeus ; Xvn, canal for the n. vagus and the vena cerebralis 
 posterior (the two divisions are clearly shown). 
 
LIBRARY 
 
 UNIVERSITY OF ILLINOIS 
 
 URBANA 
 

EXPLANATION OF PLATE XXV 
 Macropetalichthys rapheidolabis. Same specimen as in the preceding plates. 
 
 Fig. i. Primordial neurocranium from the left side with certain lateral parts 
 removed to show parts of the cavum cerebrale. 
 
 Fig. 2. Primordial neurocranium in the same aspect as in the preceding figure 
 but with still another lateral part removed to show the interorbital wall. 
 
 Fig. 3. The surface orp in fig 1, seen from behind. This surface is the 
 impression of the posterior surface of the orbit. 
 
 afr, acustico-facialis recess ; a.opt, division of the external opening of the 
 opticus canal for the arteria optica ; c.car.int, canal for the arteria carotis in- 
 terna (the ascending portion) ; c.ophth.lat, canal for the n. ophthalmicus lateral- 
 is (proximal part) ; c.ophth.prof, canal for the r. ophthalmicus profundus to the 
 orbit (proximal part) ; c.ophth.sup.V, canal for the r. ophthalmicus superficialis 
 trigemini to the orbit; cv, cavum cerebrale (middle division, which is covered 
 by its lining layer of bone — the inner bone layer) ; jg, canal for the jugular vein 
 from the orbit to the confluence with the mandibular vein ; na, nasal aperture ; 
 orb, orbit. In fig. 1 it is filled with stone, in fig. 2 this filling has been re- 
 moved so that its medial wall is seen partly with its external surface preserved 
 (black, as it is covered by the external bone layer) and partly in section (bright, 
 with nerve canals in it) ; orp, posterior surface of the orbit preserved as an im- 
 pression (to a large extent with the external bone layer adherent to it and there- 
 fore black in the figure) ; v.lb, canal from the postero-dorso-medial part of the 
 labyrinth cavity to the proximal dorsal part of the vagus canal. The canal 
 probably transmitted a vein (here is merely seen a section through the posterior 
 part) ; Vt, Vi, two canals leading from the cavum cerebrale to the labyrinth cavity, 
 probably for vessels ; II, canal for the n. opticus (external opening) ; III, canal 
 for the n. oculomotorius ; V, trigeminus recess ; Vs, $, canal for the r. maxillaris 
 trigemini and the r. mandibularis trigemini (section through the proximal part) ; 
 VII, facialis canal (several parts of it are seen). Figs. I and 2 show that it is 
 crossed on the dorsal side by the canal jg; VIII, canal for the n. acusticus; IX, 
 canal for the n. glossopharyngeus (proximal part) ; Xvn, canal for the n. vagus 
 and the v. cerebralis posterior. 
 

 LIBRARY 
 
 UNIVERSITY OF ILLINOIS 
 
 URBANA 
 

 
EXPLANATION OF PLATE XXVI 
 Macropetalichthys rapheidolabis. Same specimen as in the preceding plates. 
 Figs. 1, 2. The cavum cerebrale seen from the left side (partly in section). 
 
 Fig. 3. The labyrinth cavity seen from the medial side with the wall separat- 
 ing it from the orbit (the posterior surface of the orbit denoted by orp) and 
 the vagus canal (Xvn,) shown in their positions and with their real thickness. 
 
 Fig. 4. The labyrinth cavity shown from the lateral side (imperfectly ex- 
 posed). 
 
 afr, acustico-f acialis recess ; c.ophth.prof, canal for the r. ophthalmicus pro- 
 fundus to the orbit (proximal part) ; c.ophth.sup.V, canal for the r. ophthalmi- 
 cus superficialis trigemini to the orbit proximal part) ; csem, divisions of the 
 labyrinth cavity for a semicircular canal, perhaps the c. semicircularis externas ; 
 csem. ant, division of the labyrinth cavity for the canalis semicircularis anterior ; 
 cv, cavum cerebrale (with its lining bone layer — the inner bone layer — where it is 
 shown in black ; sections through it are shown in white) ; d.end, canal for the 
 ductus endolymphaticus ; ep, diverticle from the posterior part of the roof of 
 the anterior division of the cavum cerebrale (not seen in its entire length in the 
 figures) ; orp, posterior surface of the orbit; par, pineal canal; ru, division of 
 the labyrinth cavity occupied by the recessus utriculi; utr, division of the laby- 
 rinth cavity occupied by the utriculus ; sac, division of the labyrinth recess 
 occupied by the sacculus (partly seen) ; v. lb, canal from the labyrinth cavity to 
 the dorsal proximal part of the vagus canal (probably for a vein) ; v*, Vx x , canals 
 probably for vessels ; Vi, Vz, canals probably for vessels from the cavum cerebrale 
 to the labyrinth cavity ; I, canal for the tractus olfactorius ; III, canal for the 
 n. oculomotorius (proximal part) ; V, trigeminus recess ; V», », canal for the r. 
 maxillaris trigemini and the r. mandibulars trigemini (proximal part in section) ; 
 VII, canal for the n. facialis (in fig. 3 a considerable part of it is seen in the wall 
 between the labyrinth cavity and the orbit, as the ventral part of this wall has 
 been removed) ; VIII, canal for the n. acusticus ; IX, canal for the n. glosso- 
 pharyngeus; X™, canal for the n. vagus and the vena cerebralis posterior (in 
 figs. 1, 2, section through its proximal part, in figs. 3, 4, impression of its anterior 
 side. In fig. 4 also, a part of its stone filling is seen, the anterior end of which 
 is white). 
 
LIBRARY 
 
 UNIVERSITY OF ILLINOIS 
 
 URBANA 
 

 
EXPLANATION OF PLATE XXVII 
 
 Macropetalichthys rapheidolabis. Specimen 710 F in the American Museum 
 of Nat. Hist., New York. About 34 of the nat. size. 
 
 The specimen exhibits certain basal parts of the primordial neurocranium. 
 
 cv, cavum cerebrale cranii, the posterior narrow division (in section at the 
 place denoted. Behind this place covered by the internal bone layer il) ; el, the 
 external bone layer (the part of it on the ventral surface of the occipital 
 labyrinth and orbitotemporal regions and a part of it on the dorsal side of the 
 occipital region preserved. The latter part has, however, been partly removed 
 by preparation) ; il, internal bone layer (surrounding the posterior narrow divi- 
 sion of the cavum cerebrale) ; lab.cav, labyrinth cavity (in horizontal section) ; 
 Idi, canal for a nerve branch from the n. linese lateralis to the posterior part of 
 the cephalic division of the lateral line; n.l, canal for the n. linese lateralis (its 
 origin from the vagus canal clearly seen here) ; orb, orbit (its floor partly pre- 
 served on the left side) ; I, canal for the tractus olf actorius ; IX ?, perhaps the 
 canal for the n. glossopharyngeus ; X-m, canal for the n. vagus and the vena cere- 
 bralis posterior. 
 
FIELD MUSEUM OF NATURAL HISTORY. 
 
 GEOLOGY, VOL. IV, PL. XXVII. 
 
LIBRARY 
 
 UNIVERSITY OF ILLINOIS 
 
 URBANA 
 

EXPLANATION OF PLATE XXVIII 
 
 Figs. I, 2. Epipetalichthys ivildungensis. Specimen belonging to the Geologi- 
 cal Institution of the University of Greifswald, Germany. Fig. i shows the 
 parts preserved of the ethmoidal region of the primordial neurocranium from the 
 right side; fig. 2 shows the parts preserved of the occipital and labyrinth regions 
 in ventral aspect. 
 
 Figs. 3, 4. Macropetalichthys rapheidolabis. Specimen 4445 G of the Ameri- 
 can Museum of Nat. Hist., New York. Fig. 3 is in posterior, fig. 4 in ventral 
 view. The specimen displays the posterior narrow division of the occipital region 
 of a rather large animal. 
 
 Fig. 5. Macropetalichthys rapheidolabis. Specimen 280 G of the American 
 Museum of Nat. Hist. New York. Portion of the posterior narrow division of 
 the occipital region in ventral view. The ventral parts of the division and the 
 cavum cerebrale have been removed so that merely certain dorsal parts remain. 
 What is lettered el is here the anterior part of the external bone layer of the 
 dorsal side of the division. The bone piece lettered it is a posterior part of the 
 inner bone layer on the dorsal side of the cavum cerebrale. 
 
 ch, space for the notochord (just at the anterior end) ; ch.w, ridge in the 
 anterior part of the haemal groove caused by the notochord; c.rai, c.rai, anterior 
 and posterior opening of a canal for the radix aortae (lateral dorsal aorta). The 
 canal corresponds to a posterior part of the groove s.ra in M. rapheidolabis ; cr.ol, 
 crista occipitalis lateralis ; cr.sp, craniospinal process (impression of its anterior 
 surface; fragments in fig. 3); el, external bone layer; hcem, haemal groove; il, 
 internal bone layer; na, nasal aperture; orb, orbital entrance; pr, postero-laterally 
 projecting process of the anterior broad division of the occipital region. 
 
LIBRARY 
 
 UNIVERSITY OF ILLINOIS 
 
 UR8ANA 
 

 . 
 
EXPLANATION OF PLATE XXIX 
 
 Figs, i, 2. Macropetalichthys agassisif Specimen in the possession of Mr. 
 S. Junkermann of Bielefeld, Germany. Fig. i shows much crushed posterior 
 parts of the primordial neurocranium in ventral view ; fig. 2 a posterior part of 
 the dermal cranial roof with a number of sensory canal pores. 
 
 Figs. 3, 4. Epipetalichthys wildungensis. Specimen belonging to the Geo- 
 logical Institution of the University of Greifswald, Germany. Same specimen 
 as in figs. I, 2, PI. XXVIII. Fig. 3 shows the part preserved of the ethmoidal 
 region in ventral view, fig. 4 the ornament on a part of the dermal cranial roof 
 (impression with the bone tissue of the tubercles adhering to the stone and 
 appearing brighter than this). 
 
 c.pp, sensory canal commissure probably corresponding to the posterior head 
 line of pit organs in fishes in general; cr.od, crista occipitalis dorsalis (impres- 
 sion) ; d.end, dorsal opening of the canal for the ductus endolymphaticus ; Ic, 
 cephalic division of the lateral line; na, nasal aperture; olf, olfactory capsule; 
 pr, postero-laterally projecting process of the broad anterior division of the 
 occipital region; s.pal, groove for the r. palatinus facialis. 
 
LIBRARY 
 
 UNIVERSITY Of ILLINOIS 
 
 URBANA 
 
EXPLANATION OF PLATE XXX 
 
 Fig. i. Epipetalichthys wildungensis. Specimen belonging to the Geological 
 Institution of the University of Greifswald, Germany. Same specimen as in 
 figs, i, 2, PI. XXVIII and figs. 3, 4 in PI. XXIX. Head in dorsal view. The 
 head lacks a large posterior part. Anteriorly it shows the cavum precerebrale 
 and the olfactory capsule. The approximate position of the sutures between the 
 bones of the dermal cranial roof is indicated with black lines. 
 
 Fig. 2. Macropetalichthys rapheidolabis. Specimen 280 G of the American 
 Museum of Nat. Hist., New York. The posterior end of the occipital region in 
 posterior views. The cranio-spinal process {cr.sp) is well shown, but is repre- 
 sented only by the impression of its anterior side. 
 
 Mi, M t , Li, L 2 , L», Pi, P2, S, dermal bones of the cranial roof ; Mi imper- 
 iect anteriorly; cr.od, crista occipitalis dorsalis (its posterior continuation on 
 the anterior surface of the cranio-spinal process) ; cr.sp, cranio-spinal process 
 (represented merely by the impression of its anterior surface) ; d.end, the canal 
 for the ductus endolymphaticus (dorsal opening) ; ifc, infraorbital sensory canal 
 (its position indicated by its pores) ; il, internal bone layer; Ic, cephalic division 
 of the lateral line (its posterior part indicated by pores; its anterior part does 
 not open outwards); olf, olfactory capsule; orb, orbital entrance; pfg, cavum 
 precerebrale; soc, supraorbital sensory canal (exposed to a large extent by 
 weathering) st.com, cross-commissural sensory canal, probably representing the 
 true supratemporal commissure of fishes in general. 
 
FIELD MUSEUM OF NATURAL HISTORY. 
 
 GEOLOGY, VOL. IV, PL. XXX. 
 
 pSs 
 
 d.en <£■ Zc 
 
 crod 
 
 en 
 
 T-sjo 
 
 ifc 
 
LIBRARY 
 
 UNIVERSITY Of ILLINOIS 
 
 URBANA 
 
9dl 
 
EXPLANATION OF PLATE XXXI 
 
 Fig. I. Macropetalichthys rapheidolabis. Microscopic section through the 
 external bone layer of the primordial neurocranium. Magnification 700/1. 
 
 Fig. 2. Epipetalichthys wildungensis. Specimen belonging to the Geological 
 Institution of the University of Greifswald, Germany. Part of a section through 
 the anterior part of the posterior narrow division of the occipital region. Note 
 the trabecles on the inside of the external bone layer {el) and on the outside of 
 the notochordal layer (ch.l). Magnification about 30/1. 
 
 ch, space for the notochord (partly seen); ch.l, bone layer surrounding the 
 notochordal space; cs, cell space; el, external bone layer; ham, haemal groove. 
 
FIELD MUSEUM OF NATURAL HISTORY. 
 
 GEOLOGY, VOL. IV, PL. XXXI. 
 
 -Vi 
 
 CS- 
 
 -*a