THE CYATHASPIDIDAE A FAMILY OF SILURIAN AND DEVONIAN JAWLESS VERTEBRATES ROBERT H. DENISON FIELDIANA: GEOLOGY VOLUME 13, NUMBER 5 Published by CHICAGO NATURAL HISTORY MUSEUM APRIL 24, 1964 THE CYATHASPIDIDAE A FAMILY OF SILURIAN AND DEVONIAN JAWLESS VERTEBRATES ROBERT H. DENISON Curator of Fossil Fishes FIELDIANA: GEOLOGY VOLUME 13, NUMBER 5 Published by CHICAGO NATURAL HISTORY MUSEUM APRIL 24, 1964 Library of Congress Catalog Card Number: 6^-1863^ PRINTED IN THE UNITED STATES OF AMERICA BY CHICAGO NATURAL HISTORY MUSEUM PRESS CONTENTS PAGE Introduction 311 Morphology 313 Dorsal shield 313 Ventral shield 318 Branchial plates 318 Suborbital plates 319 Lateral plates 321 Oral, oral-lateral, and postoral plates 321 Posterior part of body and caudal fin 323 Lateral line sensory system 325 Histology 330 Internal structure 341 Relationship of Heterostraci 347 Systematic Revision 349 Cyathaspididae 350 Tolypelepidinae 351 Tolypelepis 351 Cyathaspidinae 354 Ptomaspis 354 Cyathaspis 357 Archegonaspis 362 Seretaxpis 366 Vernonaspis 367 Pionaspis 385 Listraspis 390 Irregulareaspidinae 396 Dikenaspis 396 Dinaspidella 399 Irregulareaspis 401 Poraspidinae 402 Poraspis 403 Americaspis 411 Homalaspidella 420 Ariaspis 424 Anglaspis 428 Allocryptaspis 434 Ctenaspidinae 438 Ctenaspis 438 Cyathaspididae indet 442 309 310 CONTENTS PAGE Geological Range 449- Habitat 452 Adaptation 455 Evolution and growth 457 Phylogeny 462 Summary 466 References 468 The Cyathaspididae A Family of Silurian and Devonian Jawless Vertebrates INTRODUCTION The Cyathaspididae, a family of the Order Heterostraci, is the earliest group of vertebrates that is known from more than fragmen- tary remains. Its range extends from the Middle Silurian through the Early Devonian, and it occurs in North America, Europe, and Asia. As interpreted in this work, the family includes 63 named species grouped in 19 genera. The first cyathaspids were described in 1856 and were included with Pteraspis in the Pteraspididae. It was not until relatively recently that they were recognized as a distinct higher category. Jaekel in 1911 (p. 33) subdivided his order Hetero- straci into the suborders Palaeaspidi and Pteraspidi, the former cor- responding to the Cyathaspididae of this paper. This usage was not followed until Kiaer in 1932 presented the first broad and com- prehensive classification of this group. Kiaer's suborder Cyathaspida corresponds to the family Cyathaspididae of this paper, while his families (Poraspidae, Palaeaspidae, Dinaspidae, Anglaspidae, Ctena- spidae, Cyathaspidae, and Tolypaspidae) correspond approximately to the subfamilies used in this paper. It was in Kiaer's paper that the name Cyathaspididae (in the form Cyathaspidae) was first employed. The history of the study of the Heterostraci, and in particular of the Cyathaspididae, was reviewed in considerable detail by Kiaer and Heintz in 1935 (pp. 19-39). Since that time a number of works have dealt with this family. A general account has been presented by Heintz (1938, pp. 50-54). Classification has been discussed by Flower and Wayland-Smith (1952, pp. 356-372), Denison (1953, pp. 291-293), and Tarlo (1962a, pp. 254-258). General morpholog- ical studies have been made by Watson (1954, pp. 9-13), Stensio (1958, pp. 294-321; 336-401), and Heintz (1962, p. 24). The lateral line system has been considered by Save-Soderbergh (1941, pp. 530- 539), Holmgren (1942, pp. 8-13) and Stensio (1958, pp. 401-407), and the histology of the shield has been described by Bystrow (1955, pp. 480-491; 1959, pp. 60-62) and Gross (1961, pp. 90-96; 108-109). 311 312 FIELDIANA: GEOLOGY, VOLUME 13 The evolution of Heterostraci has been discussed by White (1935, pp. 434-436), Obruchev (1945, pp. 263-266), Stensio (1958, pp. 294- 321), and Tarlo (1962a, pp. 268-277, 1962b). The ecology has been considered by Denison (1956, pp. 416-417). New Cyathaspididae from the United States have been described by Bryant (1935, pp. 112- 119), Flower and Wayland-Smith (1952, pp. 372-385), Denison (1953, pp. 294-304; 1960, pp. 555-567), and Beerbower and Hait (1959, pp. 201-203). New cyathaspids from Canada have been described by Denison (1963, pp. 108-132), from England by Wills (1935, pp. 428-435), from Russia by Obruchev (1938, pp. 36-38; 1958, p. 43) and Bystrow (1959, pp. 59-62), and from China by P'an (1962, pp. 407-408). Much of the material upon which this work is based is in the col- lections of Chicago Natural History Museum, designated by the ini- tials "CNHM" before specimen and slide numbers. In addition I have been able to borrow for study North American material from a number of institutions. I wish to express my appreciation to the following: Dr. Glenn L. Jepsen and Dr. Donald Baird of the Depart- ment of Geology, Princeton University, for the loan of material from New Jersey, New York, Pennsylvania, and the Yukon; Dr. Alfred S. Romer and Miss Nelda Wright of the Museum of Comparative Zool- ogy at Harvard University for the loan of the type material from the Vernon Shale of New York; Mr. Clinton F. Kilfoyle of the New York State Museum for the loan of specimens from the Silurian of southeastern New York; Dr. Gordon Edmund of the Royal Ontario Museum for the loan of the type of Diplaspis acadica; the Cali- fornia Standard Company of Edmonton, Alberta, and particularly Dr. Alfred Lenz, for the gift of specimens from northern Yukon, northwestern British Columbia and Northwest Territories, Canada (the types and representative specimens have been deposited in the National Museum of Canada, the rest retained in the collection of Chicago Natural History Museum); Dr. J. R. Beerbower of Lafay- ette College for the gift of specimens from the Silurian of Pennsyl- vania and Maryland; Dr. Shelton P. Applegate of the Los Angeles County Museum for the loan of Claypole's syntypic material of Americaspis americana; and Dr. W. P. Leutze of New Orleans for the gift of specimens from the Silurian of Maryland and West Virginia. In 1953-54, thanks to a grant from the John Simon Guggen- heim Foundation, I had the opportunity to study much of the Euro- pean material at the Paleontologisk Museum in Oslo, Naturhistoriska Riksmuseet in Stockholm, and the British Museum (Natural His- DENISON: THE CYATHASPIDIDAE 313 tory) and Geological Survey and Museum in London. I wish par- ticularly to thank Dr. Anatol Heintz for permission to cast important specimens in the collections at Oslo, and for providing in an exchange , ;Vi--»V-i"nir ?~ :-M Fig. 90. Restoration of Anglaspis heintzi (from Kiaer, 1932); lateral view (X about 3/2). Spitsbergen specimens that have been extremely useful. I wish also to thank Dr. Tor 0rvig of Stockholm for the loan of the type of Seretaspis zychi, and Dr. Errol I. White of London for help in obtain- ing copies of hard-to-find literature. The following figures have been drawn, many from my original sketches, by Dr. Tibor Perenyi, staff artist at Chicago Natural His- tory Museum: figures 91, 93, 97-102, 103,A, 113, 127, 130, 132, 155 and 161. MORPHOLOGY The mineralized skeleton of the Cyathaspididae is entirely der- mal, and consists of a shield or carapace covering the anterior part of the body, and of scales covering the posterior part and tail (fig. 90). The shield (fig. 91) consists of relatively few elements, of which the most important are the dorsal shield, the ventral shield, and the paired branchial plates. Dorsal Shield The dorsal shield (rostrodorsal disc of Stensio, 1958, p. 295) is elongate and variously arched. It ranges in length in known cya- thaspids from 20 mm. in Homalaspidella nitida and Anglaspis insig- nis to 110 mm. in Allocryptaspis utahensis. Its anterior border may 314 FIELDIANA: GEOLOGY, VOLUME 13 be smoothly convex (figs. 109, 113, 146, 149, ro), or may have a rounded median rostral process (figs. 117,A, 123, C, 135, B, mrp). The rostral border is folded under to form the maxillary brim (maxillary plate of Kiaer, 1928, p. 123; maxillar brim of Kiaer and Heintz, 1935, p. 44; subrostral lamella of Stensio, 1958, p. 352), which forms the dorsal or anterior border of the mouth (figs. 91, B, 92, 105, 132, mxb). Stensio (1958, fig. 194) shows in Poraspis a postrostral ascending lamella extending backward and upward from the maxillary brim, but this is not present in sections of this genus figured by Kiaer and Heintz (1935, pi. 38). The orbits notch the sides of the anterior part of the shield, and are usually delimited anteriorly by marked preorbital processes (figs. 91, 113, 123, pop). The dorsal shield is generally curved down at the sides to meet the branchial plates, but may develop distinct, laterally directed brims (Cyathaspis, Lis- traspis (fig. 132, Ibr), Anglaspis, Ctenaspis). In Ariaspis (fig. 149, Ibr) there are small but distinct, downwardly directed lateral laminae. There is a single external branchial opening on each side, and its position is usually indicated on the dorsal shield by the downward flexure of the margin to form the postbranchial lobe (fig. 91, pbl). The postbranchial lobes are absent in Ariaspis and Ctenaspis, and are small or absent in Anglaspis. They have an abruptly truncate anterior edge in the Irregulareaspidinae and Pionaspis; in the latter (figs. 127, 130) this edge is turned inward to form the posterior boundary of the external branchial opening. The postero-lateral corners of the dorsal shield are usually rounded, but are pointed in Anglaspis, Listraspis (fig. 132) and Ariaspis (fig. 148). The poste- rior edge of the shield is usually convex, and may have a median lobe or point. In only one cyathaspid, Ariaspis (fig. 148, ms), is there a ridge scale incorporated to form a median dorsal spine. Cyathaspis, Listraspis, and Ctenaspis have a median crest on the posterior part of the dorsal shield. The dorsal shield appears on its inner surface to be a single ele- ment, but externally there may be apparent subdivision into areas distinguished by differences in the superficial ridge pattern, and sometimes by what appear to be sutures. These areas are called epitega, a name introduced by Stensio (1958, p. 297). The anterior area is the rostral epitegum (fig. 112,A, rep). Paired lateral epitega (fig. 112, A, lep) (marginal prebranchial epitega of Stensio, 1958, p. 297) extend along the sides from in front of the orbits posteriorly as far as the external branchial openings, or beyond in Listraspis. The large central area of the shield, the central epitegum (fig. 112, A, pm ro soc mxb m0 orb cd, mdl ldl cd4 ds drs mo brP lvl bro pbl Fig. 91. Shield and anterior scales of Poraspis (after Kiaer and Heintz, 1935, with modifications). A, dorsal view; B, ventral view; C, lateral view. bro, branchial opening; brp, branchial plate; cd\-i, first to fourth dorsal trans- verse sensory commissures; cso, transverse supraorbital sensory commissure; cvl, lateral ventral transverse sensory commissure; cvm, medial ventral transverse sensory commissure; dls, dorso-lateral scale; drs, median dorsal scale; ds, dorsal shield; ifc, infraorbital sensory canal; ki-*, positions of inner impressions of gill pouches, numbered according to the theory of Stensio; lap, lateral plate; ldl, lat- eral dorsal sensory canal; lk2-s, position of inner impressions of extrabranchial atria or exhalant ducts; lvl, lateral ventral sensory canal; mdl, medial dorsal sen- sory canal; mo, mouth; mxb, maxillary brim; olp, oral-lateral plate; orb, orbit; orp, oral plate; pbl, postbranchial lobe; pm, pineal macula; pol, postoral sensory canal; pop, preorbital process; ro, rostrum; sec, semicircular canal; soc, supraorbital sensory canal; sop, suborbital plate; vis, ventro-lateral scale; vrs, median ventral scale; vs, ventral shield. 315 316 FIELDIANA: GEOLOGY, VOLUME 13 cep), lies between the lateral epitega, and extends from the rostral epitegum to the posterior border. A pineal macula (figs. 91, 124, pm) is usually marked by its ridge pattern and a prominence, but there is no foramen. There is sometimes a distinct postrostral field (fig. Fig. 92. AUocryptaspis laticostata, CNHM, PF 1811; ventral side of incom- plete rostrum, showing maxillary brim, mxb (X 3). 112,A, prf) (Stensio, 1958, pp. 299-300 = pineal triangle of Denison, 1963), including the pineal macula posteriorly, and expanding for- ward to its meeting with the rostral epitegum; this may not be a distinct epitegum, but only a part of the central epitegum that is demarked by the supraorbital sensory canals. Except in Ctenaspis, the surface of the shield is covered with den- tine ridges, sometimes elongate, and at other times broken into short lengths or denticles. The ridges are arranged according to a pattern that is taxonomically important in general, though it may be variable in detail. In some genera the ridges are arranged in scale-like areas that presumably represent fused scales of an ancestral condition. The fineness of the ridges is also of taxonomic importance, and has been represented as the number of ridges per millimeter. This is usually measured in the central part of the dorsal shield; near the lateral borders the ridges are usually finer, and on the ventral shield they are usually slightly coarser. In a few cyathaspids there are coarser and higher ridges separated by a number of lower and finer ridges. Dentine ridges also cover the maxillary brim, usually ar- ranged parallel to the anterior edge, though they have an antero- posterior arrangement in AUocryptaspis laticostata (fig. 92). The crests of the ridges may be flat, or roundly or angularly convex, and tend to be higher and sharper near the lateral borders, especially DENISON: THE CYATHASPIDIDAE 317 over the orbits and branchial openings. Between the ridges are nar- row intercostal grooves that are comparable to the canals of the pore- canal system of some other fishes (Gross, 1956, p. 135). Fig. 93. Dorsal shield of a cyathaspid, showing measurements employed. A, dorsal view, and B, lateral view. ML, median length; OL, orbital length; OW, orbital width; PEL, postbranchial length; PL, pineal length; W, maximum width. The following measurements (fig. 93) and ratios have been used for the dorsal shield : Median length (or length) Maximum width (or width) Orbital width (or width at orbital notches) Orbital length (or median length from anterior border to a line between the orbital notches) Pineal length (or median length from anterior border to the center of the pineal macula) Postbranchial length (or median length from posterior border to a line between the external branchial openings) maximum width , , , , , . , . Tr. , TT . . Width ratio = -medi-an length (breadtiiJen^h index of Kiaer and Heintz, orbital width Orbital width ratio = m-edian length orbital length , . , , ,,. , TT . . ,„„ .„. Orbital ratio = m^dianiength (rostral index of Kiaer and Heintz, 193o, p. 46) Pineal ratio = Postbranchial ratio pineal length median length postbranchial length median length 318 FIELDIANA: GEOLOGY, VOLUME 13 Ventral Shield The ventral shield (fig. 91, vs) has relatively few diagnostic char- acters compared to the dorsal shield. It extends from the posterior borders of the oral or postoral plates to the posterior end of the cara- pace, and covers the area between the branchial plates. Anteriorly it is relatively flat, but posteriorly it is deeply vaulted, usually more so than the dorsal shield. Its anterior border is most commonly slightly concave, but in some genera it is transverse or slightly con- vex; in Allocryptaspis laticostata (figs. 152,C) it is notched for the re- ception of postoral plates. At its antero-lateral corners there may be shallow notches for the reception of lateral plates. The lateral edges abut against or are slightly overlapped by the ventral edges of the branchial plates (Kiaer and Heintz, 1935, figs. 53-54). In Cten- aspis the lateral margins are formed as brims. The postero-lateral corners are usually rounded, but are pointed in Ctenaspis. The pos- terior edge is usually similar to that of the dorsal shield, and may be transverse or convex, with or without a median lobe or point. In Allocryptaspis (fig. 152,B) the dorsal and ventral shields meet behind the branchial openings; in other cyathaspids (fig. 91, C) they are separated here by the branchial plates. Allocryptaspis is excep- tional in lacking free branchial plates and in having the branchial openings notch the ventral as well as the dorsal shield. This may be true in Ctenaspis also. Most of the dentine ridge pattern is either longitudinal or elliptical on the ventral shield. Anteriorly the longi- tudinal ridges commonly radiate fan-wise toward the antero-lateral corners. In some cases, notably in Cyathaspis (fig. 110,B), there are bands of ridges parallel to the anterior edge, and to the anterior part of each lateral edge. Stensio (1958, p. 308) distinguished these and other areas as "hypotega," comparable to the epitega of the dorsal shield. There is reason to doubt, however, that these "hypotega" have any great significance as structural or growth units, as is pointed out below (p. 460). The only measurements made on the ventral shield, besides dentine ridges per millimeter, are median length and maximum width. The only ratio used is the width ratio, or maxi- mum width/median length. Branchial Plates These have been recognized in less than half of the genera of Cyathaspididae. They are elongate, slender plates (fig. 91, brp) that normally articulate with the lateral edges of the ventral shield, and of the dorsal shield behind the orbits. They are curved antero- DENISON: THE CYATHASPIDIDAE 319 posteriorly to follow the curvature of the dorsal and ventral shields, and are arched dorso-ventrally, or have a sharp lateral angulation in Anglaspis. Their structure is best known in Poraspis, thanks to the complete description and excellent figures of Kiaer and Heintz (1935, pp. 46, 118-125, figs. 51-54, pi. 33, figs. 4-6). In this genus the attachment to the dorsal shield is loose and separated by a skin fold anterior to the branchial opening; the attachment behind the branchial opening and to the ventral shield is partially overlapping and presumably firmer. The dorsal border of the branchial plate of Poraspis (fig. 91, brp) has a well-marked notch that forms the lower border of the branchial opening. The notch is less distinct in Homa- laspidella (Kiaer and Heintz, 1935, pi. 30, fig. 1), Archegonaspis (Heintz, 1933, figs. 1-3), Vernonaspis (Ruedemann, 1916, pi. 32, fig. 7), and Anglaspis (Wills, 1935, pi. 1, figs. 9-11). The anterior margin in Poraspis, Anglaspis, and Listraspis is divided by an angu- lation into two parts, of which the upper is an overlapping margin for the suborbital plate, while the lower presumably articulates with a lateral plate. The dentine ridges are predominantly longitudinal, especially on the dorsal part of the plate, but may be strongly diag- onal on the ventral part. In three genera of cyathaspids the branchial plates are fused or probably fused to the dorsal shield. In Listraspis (fig. 132) each of them is attached to the medial edge of the ventral surface of the lat- eral brim, and the branchial opening is preserved as a slit between it and the brim in approximately its usual position. In Allocrypt- aspis (fig. 153, 11) the presumed equivalents of the branchial plates form ventro-mesially directed laminae on the sides of the dorsal shield, entirely anterior to the branchial openings. Similar laminae in Ctenaspis (fig. 155, vll) may represent the branchial plates. Suborbital Plates Articulated cyathaspids that preserve the smaller plates and scales in their normal position are extremely rare. Two specimens of Anglaspis heintzi (Paleontologisk Museum, Oslo, D 382, 384) fur- nished the information for Kiaer's (1932, fig. 11) well-known re- construction, but they have not been completely described. An articulated but imperfect specimen of Irregular easpis hoeli (Kiaer, 1932, pi. 5) preserves the suborbital plate, but not the plates of the mouth region. The type of Archegonaspis Integra (Heintz, 1933, fig. 2) is articulated, but the small anterior plates, if preserved, have not been figured or described. 320 FIELDIANA: GEOLOGY, VOLUME 13 A orP. __--ro Fig. 94. A, Anglaspis heintzi; ventral view of anterior part of articulated specimen, Paleontologisk Museum, Oslo, D 384 (X 3). B, Protopteraspis vogti; reconstruction of ventral side of anterior part of shield ( X 4). (Both from Heintz, 1962.) brp, branchial plate; lap, lateral plate; obp, orbital plate; olp, oral-lateral plate; orp, oral plate; ro, rostrum; rop, rostral plate; sop, suborbital plate; vs, ventral shield. Kiaer's reconstruction (fig. 90) shows the suborbital of Anglaspis to be a small, curved plate forming the whole lower border of the orbit. Anteriorly it abuts against the preorbital process, and poste- riorly it meets the upper facet of the anterior edge of the branchial plate, as well as the postorbital border of the dorsal shield. The sub- orbital plate of Irregulareaspis (Kiaer, 1932, pi. 5, fig. 2) is similar but relatively larger. In Listraspis (fig. 132, sop) the suborbital plate is fused to the dorsal shield, and forms the part of the ventral surface of the lateral brim that lies between the branchial plate and the equivalent of the preorbital process. No articulated specimens of Allocryptaspis are known, but numerous small plates of A. lati- costata have been identified with more or less certainty (Denison, DENISON: THE CYATHASPIDIDAE 321 1960, pp. 563-564), and among them is the suborbital (fig. 95,F). In this species the suborbital is a subrectangular plate forming only about one-quarter of the orbital border; as restored (fig. 152, C), the antero-ventral boundary of the orbit is formed by the oral-lateral plate which lies immediately in front of the suborbital. A plate of Homalaspidella nitida, tentatively identified by Kiaer and Heintz (1935, pp. 131-132, pi. 30, fig. 1, pi. 33, fig. 1) as a lateral plate, shows some similarities to the presumed suborbital of Allocryptaspis. Lateral Plates In pteraspids a variable number of small plates lying between the dorsal shield and the antero-lateral corners of the ventral shield have been called lateral plates (Kiaer, 1928, p. 123; White, 1935, p. 408) or orogonial plates (Stensio, 1958, p. 258). Probably a single pair of lateral plates is present in Anglaspis (fig. 94, A, lap) . A small lat- eral plate is found attached to the anterior end of the branchial plate of Listraspis (fig. 132, lap). A number of lateral plates have been provisionally identified in Allocryptaspis laticostata (fig. 95,C, D, E) ; these have been named antero-lateral, lateral postero-lateral, and medial postero-lateral (Denison, 1960, p. 563). As restored (fig. 152), the postero-laterals abut on one side against notches in the corner of the ventral shield, and on the other side adjoin the suborbital and branchial plates; the antero-lateral is placed ventro-mediad to the suborbital. The plate of Homalaspidella nitida, tentatively identi- fied as a lateral plate by Kiaer and Heintz (1935, pp. 131-132), may be a suborbital. Oral, Oral-lateral, and Postoral Plates Oral plates are preserved in the articulated specimens of Angl- aspis heintzi (fig. 94, A, orp), and have recently been described by Heintz (1962, p. 24, fig. 7). This species has relatively few, large oral plates, as compared to the 15 small plates that form the oral cover in the pteraspid, Protopteraspis vogti (fig. 94,B). Isolated small, nar- row plates of Allocryptaspis laticostata (fig. 95,H) have been doubt- fully identified as oral plates (Denison, 1960, p. 564). In pteraspids (White, 1935, pp. 408-412; Heintz, 1962, pp. 19-24) each oral plate has on the inner face of its anterior end a projecting oral tooth lamella (="oral tooth plate" of Kiaer), whose denticu- lated surface faces the adjacent oral tooth lamella. As shown by Heintz, this indicates that the oral plates did not work against the maxillary brim as believed by Kiaer. The oral plates are imbricated, 322 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 95. Allocryptaspis laticostata; small plates from anterior part of ventral side of shield (X 4). A, right postoral, PF 1768; B, right postoral, inner side, PF 1763; C, lateral postero-lateral, PF 1782; D, medial postero-lateral, PF 1781; E, anterolateral, PF 1829; F, suborbital, PF 1777; G, oral-lateral, PF 1780; H, possible oral plate, PF 1776. (From Denison, 1960; specimens in Chicago Natural History Museum.) which suggests that they were set in infoldings of the skin. This arrangement may have permitted the oral plates to have been pro- truded, forming a scoop for feeding in bottom sediments. Heintz has suggested the possibility that when the oral plates were pro- truded, water rich in organic particles may have been drawn into the mouth, and that when the mouth was closed the denticles of the oral tooth lamellae and of the maxillary brim may have served to filter out the food particles. It is also possible that the protruded oral scoop may have had limited ability to pick up selected food particles from the bottom (Denison, 1961, pp. 179-180). In pteraspids an oral-lateral plate (fig. 94,B, olp) lies on either side of the row of oral plates, and is bounded laterally by the rostrum and posteriorly by a lateral plate. No oral-lateral plates are shown in Heintz's figure (1962, fig. 7) of Anglaspis heintzi. A plate has been tentatively identified (Denison, 1960, p. 564) as oral-lateral in Alio- DENISON: THE CYATHASPIDIDAE 323 cryptaspis laticostata (fig. 95,G); as placed in the restoration (fig. 152,C), it extends posteriorly to form a corner of the orbit. In Protopteraspis vogti (fig. 94,B) and Anglaspis heintzi (fig. 94,A), there are no postoral plates, and the oral plates articulate directly with the anterior border of the ventral disc. However, postoral plates are well developed in Allocryptaspis laticostata (fig. 95,A-B), and were probably arranged much as in Pteraspis rostrata (White, 1935, figs. 41-47, 85). The pair of main postoral plates of Allo- cryptaspis attached to the ventral disc posteriorly, to lateral plates at the side, and probably met each other in the midline (fig. 152,C). Between them postero-medially lay two or more small, median pos- torals, indicated by the shape of the main postorals, and by notches in the anterior margin of the ventral shield. Posterior Part of Body and Caudal Fin Behind the carapace, the posterior body and tail are covered with scales that are best known from Kiaer's reconstruction of Anglaspis heintzi (fig. 90), and from Kiaer's and Heintz's (1935, pp. 108-118, figs. 41-50) thorough description of scales of Poraspis. In Allo- cryptaspis (Denison, 1953, pp. 302-304, figs. 61, 65,B; 1960, p. 567, fig. 123) the scales are similar to those of Poraspis. Anteriorly the scales are relatively very large (fig. 91, C), and are arranged in six longitudinal series: median dorsal, paired dorso-lateral, paired ven- trolateral, and median ventral. The median scales (fig. 96,B-D), dorsal and ventral, are symmetrical, relatively broad and flat ante- riorly, and become more and more slender posteriorly. The dorso- lateral scales (fig. 96,A) are extremely large, each usually covering more than half of its side of the ring of scales. The ventro-lateral scales (fig. 96, F) are much smaller, except in Irregulareaspis hoeli, where they are nearly equal to the dorso-lateral scales in size (Kiaer, 1932, p. 17, pi. 5). The scales are covered with predominantly longi- tudinal dentine ridges, though commonly there are one or more ante- rior ridges at right angles. Each scale, as is usual in fishes, overlaps the scale behind and also the one below. The overlapped margins are anterior brims without dentine ridges, while the posterior or pos- tero-ventral edges that overlap the scales behind have an inner mar- gin with dentine ridges continued from the external surface. In Pteraspis the row of ventral ridge scales is broken by a gap that may indicate the position of the anus (White, 1935, p. 418, figs. 65, 85). In Anglaspis the anal position may be indicated by a discon- tinuity posterior to the fourth ventral ridge scale (fig. 90). 324 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 96. Scales of Allocryptaspis laticostata. A, dorsolateral scale, PF 1817 (X3); B, anterior median scale, PF 1814; C, median scale from mid-length, PF 1816; D, posterior median scale, PF 1815; E, caudal scale, PF 1820; F, ventro- lateral scale, PF 1821. (B-F X9/2.) (From Denison, 1960; specimens in Chicago Natural History Museum.) On the tail the scales become small and lose the arrangement in longitudinal rows, except that the dorsal and ventral median scales continue as long, slender fulcra (fig. 96,D). The caudal fin itself is somewhat hypocercal in Anglaspis with a posteriorly directed lobe projecting ventrally. Scales are known also in Tolypelepis (Pander, 1856, pi. 6, fig. 32), Cyathaspis, Archegonaspis (Heintz, 1933, figs. 1-2), Vernonaspis, Americaspis, and Homalaspidella. In Tolypelepis they are similar to the scale-units that have grown together to form the posterior part of the dorsal shield. DENISON: THE CYATHASPIDIDAE 325 There are never any paired fins in Heterostraci, and their only median one is the caudal fin. There are two reasons for considering this to be a primitive vertebrate condition: (1) paired fins are absent also in primitive Osteostraci, Anaspida, and probably in Coelolepida; (2) it is unlikely from a functional point of view that the hetero- stracian ancestors would have lost such useful structures had they ever acquired them. On the other hand, Stensio (1958, pp. 409-413) sees in the cornual plates of pteraspids the vestiges of an ancestral paired fin fold, and in the dorsal and ventral ridge scales the rem- nants of median fin folds. The cyathaspids must have been relatively poor swimmers com- pared to most modern fishes. The carapace made the anterior part of the body inflexible, while the relatively large scales of the poste- rior part must have resulted in limited flexibility of the only organ of propulsion. In the absence of paired, dorsal, and anal fins, sta- bility in swimming must have been slight, and steering and fine con- trol difficult. However, the large ridge scales, and the dorsal crests and lateral brims of some cyathaspids must have acted as keels to help promote stability. Lateral Line Sensory System In the Cyathaspididae the lateral lines are canals in the cancel- lous layer of the dermal skeleton (fig. 103, A) . These are connected with the surface by small tubes leading to pores, which are usually the only part of the system that is apparent. In some cases, notably in Poraspis, the pores are large and prominent, but in a number of genera, for example, Americaspis, they are small and difficult to see. On the ventral shield of Anglaspis insignis the lateral lines consist of open grooves (fig. 150,B). The pattern of the canals in this family (fig. 97-99), usually an almost diagrammatically simple combination of longitudinal canals and transverse commissures, has been considered as a primitive ar- rangement from which could be derived that of other agnathans and gnathostomes. For this reason it has been studied by numerous paleontologists and zoologists, but there is still considerable differ- ence of opinion about the homologies of many canals. Certain iden- tification is possible only if the embryology and innervation are known. This is impossible in Heterostraci, so reliance must be placed on comparison of patterns, and on topographical relationships to the few internal morphological landmarks. Since neither of these pro- cedures is likely to be decisive, the names employed in this paper Fig 97 Dorsal lateral line patterns of Cyathaspidinae Irregulareaspidinae andFCtenaspSrnr::! A, Tolypeleprs undulata %^J%^%^ B, Ptomaspis canadensis; C, Dinaspidella robusta Wm, "JMg. °£ /nJ dJnteto (from Kiaer, 1930); E, Dikenaspis yukonensis (pattern restored), J? , , canal connecting pulp canal to intercostal groove; dr, dentine ridge; ig, intercostal groove; pc, pulp canal; pu, pulp chamber; rl, reticular layer. 335 336 FIELDIANA: GEOLOGY, VOLUME 13 only by rows of slit-like pores between the partially contiguous scal- loped edges of adjacent dentine ridges. The intercostal grooves are probably homologous to the pore- canal system of Osteostraci, Crossopterygii, and Dipnoi. In these groups, the system consists of mesh-canals (Maschenkanale) that form a network typically just below the surface of the dermal bones, and of pore-canals (Porenkanale) that connect the mesh-canals with the surface (Gross, 1956). The intercostal grooves resemble this system in their superficial position, in their cross-section, and in their vascular supply (Gross, 1961, p. 145). In Osteostraci, Crossoptery- gii, and Dipnoi, the pore-canal system is related to the lateral lines, and for this reason is believed to be a sensory system, probably sen- sitive to currents acting on the body (Denison, 1947, pp. 350-353; Bolau, 1951, pp. 38-39; Gross, 1956, p. 136; Dijkgraaf, 1963, pp. 78- 81). The lateral lines of cyathaspids lie in the cancellous layer, well below the intercostal grooves. However, White (1935, p. 421, fig. 66a) has shown in pteraspids that the lateral lines communicate with the intercostal grooves, and a thin section of Irregulareaspis sp. (fig. 103,A) shows a lateral-line canal passing into two adjacent inter- costal grooves. The intercostal canals may have been filled with mucous, but the production of mucous was not their chief function (Gross, 1956, p. 96). The recticular layer is a relatively thin part of the dermal skeleton lying below the dentine ridges and superficial to the chambers of the cancellous layer. It is composed of aspidine, but the manner of its formation is not certain. According to Gross (1961, p. 145), the vessels passing through this layer are surrounded by fine lamellae that form skeletal tubes comparable to primary osteons. I have been unable to confirm this, except in the case of the intercostal grooves, which are surrounded by aspidine lamellae continuous with the den- tine lamellae. On the other hand, an obliquely tangential section of Allocryptaspis laticostata (fig. 104,C) shows this layer to be composed of cross-fibered aspidine continuous with that of the cancellous layer; there is no clear indication of any lamellation around the canals, and the arrangement of the fibers does not appear to be governed by the canals. In Ctenaspis dentata (fig. 101, C), where the superficial layer is absent, aspidine of the reticular layer forms the surface and is clearly lamellate parallel to the surface, rising in the tubercular elevations and sinking in the depressions between. In skeletons of juvenile individuals the reticular layer may be mostly or completely absent, a condition that at times has been used mistakenly as a sys- DENISON: THE CYATHASPIDIDAE 337 tematic character. In the absence of this layer, the pulp canals open widely and directly into the chambers of the cancellous layer. cc2 dr- Fig. 103. A, oblique section through superficial, reticular and cancellous layers of dermal shield of Irregulareaspis sp., CNHM, slide 4590 (X 25); B, tan- gential section through superficial layer of Anglaspis sp. (from Gross, 1961) ( X 50). ca, chamber of cancellous layer; cch longitudinal canal connecting pulp cham- bers; cci, canal connecting pulp chamber to intercostal groove; cc:i, canal connect- ing pulp chambers of adjacent dentine ridges; dr, dentine ridge; ig, intercostal groove; pc, pulp chamber; sc, lateral line sensory canal. The outstanding feature of the reticular layer is its numerous canals which bring vascular and nervous supply to the superficial structures. These canals are: (1) Pulp canals of the dentine ridges (figs. 100-103, pc) ; when these canals are paired, as in Allocryptaspis (fig. 104,A) and Sanidaspis, there are numerous cross connections between the members of a pair; or where the pulp canals are divided into chambers, as in Tolypelepis and Anglaspis, there are longitudinal canals (fig. 103, B, cc\) connecting the chambers. (2) Canals lateral from pulp canals to the intercostal grooves; these are abundant and regular in Poraspis (fig. 102, A), Americaspis (fig. 102,B, cc2), and Sanidaspis; they emerge from each of the pulp chambers of Anglaspis (fig. 103, B, cc2); they are relatively few in number in Allocryptaspis (fig. 102,C, cc2). (3) Canals passing laterally from the pulp canals and continuing under the intercostal grooves; in Anglaspis (fig. 103, B, ccs) they may connect directly with the pulp canal of the neighboring ridge; in Allocryptaspis they meet vertical or oblique canals leading from the cancellous layer to intercostal grooves. 338 FIELDIANA: GEOLOGY, VOLUME 13 (4) Vertical canals from the cancellous layer to pulp canals. (5) Ver- tical or oblique canals from the cancellous layer to intercostal grooves; these are very numerous in Allocryptaspis laticostata (fig. 102,C). A thin section of Ctenaspis dentata (fig. 101, C) is remark- able for the scarcity of canals in the reticular layer. The cancellous layer is the weakest part of the dermal skeleton, and is often crushed. In Allocryptaspis laticostata, for example, only in pyritized specimens is it preserved uncrushed. This layer con- sists of large chambers (fig. 100-104, ca), commonly polygonal in tangential section, separated by more or less vertical septae, floored by the basal layer, and covered by the recticular layer. In those specimens that have a well-developed reticular layer, the cancellous layer usually occupies 55-72 per cent of the total thickness of the central part of a shield. Allocryptaspis laticostata (fig. 102,C) has an unusually deep cancellous layer, which may occupy 78 per cent of the thickness of the shield. Near the margins of a shield, the chambers may become very small. The width of the chambers is usually 200-400 M, but Bystrow (1959, fig. 2) shows them 800 M wide in Sanidaspis, and some are at least 670 n wide in Allocryptaspis. The chambers are surrounded by lamellar aspidine. This is clearly shown in sections of Ctenaspis dentata (fig. 101, C) and of Irregulare- aspis sp. (CNHM, slides 4589-90), also in figures of Anglaspis sp. (Gross, 1961, fig. 6, C, D). A thin section of Allocryptaspis lati- costata (CNHM, slide 4594) shows a cross-fibered structure continu- ous with that of the reticular layer, but the lamination is only faintly suggested by color differences. In younger cyathaspids there are relatively few laminae, and the septae between the chambers are thin. The septae become thicker with age by the formation of addi- tional laminae. They have been found most commonly to be 30-70 n thick, but in Allocryptaspis laticostata they have been measured as thin as 20 ^ and as thick as 130 yu. The chambers are irregular in shape, size, and arrangement in Dikenaspis (fig. 101,A) and Irregu- lareaspis (figs. 101, B, 103, A). In Poraspis they are usually regular and hexagonal in section, except near the margin of the shield. In Anglaspis (fig. 102, D) they are regularly arranged in rows under the dentine ridges, one chamber corresponding in width to one ridge, and with the septae underlying the intercostal grooves. In Allo- cryptaspis laticostata (figs. 102,C, 104,C-D) they are also arranged under the dentine ridges, but the regularity is less than in Anglaspis, and the correspondence of ridges and chambers is not perfect. In most cyathaspids more than one dentine ridge overlies a chamber, Fig. 104. Tangential or obliquely tangential sections through dermal shield of Allocryptaspis laticostata. A, Dentine ridge and two intercostal grooves, show- ing clusters of dentine tubules arising from pulp chambers; CNHM, slide 4594 (X 90). B, Oblique section through superficial layer (at top) and reticular layer, showing parts of five dentine ridges and six intercostal grooves, the latter with small slit-like openings at the surface; CNHM, slide 4597 (X 30). C, An oblique section through the reticular layer; CNHM, slide 4594 (X 30). D, A tangential section through the cancellous layer; CNHM, slide 4596 (X 30). 339 340 FIELDIANA: GEOLOGY, VOLUME 13 and a linear arrangement of chambers is lacking. There are com- monly perforations in the septae giving communication between ad- jacent chambers. There are numerous pores in the roof of the chambers, leading into the many canals of the reticular layer. Lat- eral line canals, 150-360 n in diameter, lie in the upper half of the cancellous layer; their external connections are with pores and inter- costal grooves by means of branch canals (fig. 103, A, sc). The basal layer consists of lamellae of aspidine, some of which are part of the layers concentric around overlying chambers (fig. 101, C). In well-preserved material Sharpey's fibers may be seen penetrating it (Gross, 1961, fig. 6). The basal layer is thin or absent in newly forming shields, and thickens with age by the addition of laminae. Bystrow (1955, fig. 11) shows it very thin in an incompletely devel- oped Tolypelepis, and very thick in sections of Poraspis sericea (200 ii, op. cit., fig. 17) and Sanidaspis (120 ix, Bystrow, 1959, fig. 2). Ver- tical canals pass through the basal layer to supply the cancellous and overlying layers. Comparisons. — In the microstructure of the dermal shield, the Pteraspididae are very close to the Cyathaspididae. Both families are characterized by the possession of a cancellous layer with large chambers separated by more or less vertical septae. This is in strik- ing contrast to most other Heterostraci (Psammosteidae, Astraspis, Eriptychius, Cardipeltis, Weigeltaspis, and Tesseraspis) in which the middle layer is a more typical spongiosa with smaller cavities sep- arated by irregular partitions. In Corvaspis large chambers of the cyathaspid type occur at the center of plates, and a spongy bone at the margins (Dineley, 1953, fig. 16). In Traquairaspis (Wills, 1935, pi. 6) large chambers occur under the smooth area of the central disc, while the middle layer of the rest of the shield is spongy. The initial stage in the development of the pteraspid shield is similar to that of cyathaspids. Thin sections of ventral discs of juvenile Protaspis sp. (CNHM, slides 4069-70) about 12 mm. long show only a thin cap- ping of dentine on the ridges and a thin trabecula of aspidine sur- rounding each intercostal groove. Later stages of pteraspid growth have not yet been studied in detail histologically; however, in addi- tion to growth of individual plates around their periphery, the plates apparently grow in thickness by the addition of laminae of aspidine to the inner surface of the basal layer. This leads eventually to the development of a thicker basal layer than is typically found in cya- thaspids. DENISON: THE CYATHASPIDIDAE 341 Internal Structure No member of the Heterostraci has yet been found with any mineralized internal skeleton, and it is not likely that such was ever developed in this order. For this reason there can be little direct information about the internal anatomy of the group. There are a few external indications of the position of internal structures: the mouth, orbits, pineal macula, and external branchial openings. In addition, a number of cyathaspids and a few pteraspids, especially the smaller ones, preserve impressions of internal organs on the inner side of the dermal shield (figs. 106, 132, 151,A, 156). Those struc- tures that are quite certainly interpretable are parts of the brain, pineal organ, vertical semicircular canals, and gills. Some years ago Stensio (1932, fig. 65) produced a reconstruction of the internal anat- omy of cyathaspids, and recently (1958) he has attempted a different and more elaborate one. Since there is such limited factual basis for a reconstruction, Stensio's are only understandable as attempts to adapt the structure of a myxinoid to a heterostracian shield. Some points of his later reconstruction will be considered critically in the following discussion of the cyathaspid internal anatomy. Mouth and Nostrils: The ventral border of the rostrum of cyath- aspids forms a maxillary brim (subrostral lamina of Stensio) that is variously ornamented with ridges and denticles of dentine (fig. 105, mxb). This is generally believed to be the dorsal (or anterior) bor- der of the mouth. The ventral (or posterior) border of the mouth is formed by a series of oral plates (fig. 94) which lie opposite to the maxillary brim in a few articulated specimens of pteraspids (Proto- pteraspis vogti, Kiaer, 1928, pi. 12; Heintz, 1962, figs. 1-6; Pteraspis rostrata, White, 1935, figs. 41-47) and cyathaspids (Anglaspis heintzi, Heintz, 1962, fig. 7). In his latest attempt to adapt the Heterostraci to the myxinoid plan, Stensio (1958, pp. 342-350) has found it neces- sary to insert between the oral plates and maxillary brim a palato- subnasal lamella, which separates a prenasal sinus from the buccal cavity. There is absolutely no evidence for this structure, and it is clearly impossible to fit it in such an articulated, relatively uncrushed specimen as the figured examples of Protopteraspis vogti (fig. 105). It is generally believed that the nasal sacs were situated in paired depressions on the inner side of the dorsal shield just posterior to the maxillary brim (Heintz, 1962, p. 25). The presence or absence of external nares has not been clearly demonstrated. Kiaer and Heintz (1935, p. 69, pi. 26, fig. 1) showed in the ventral border of a preorbital process of a specimen of Poraspis polaris a notch which they inter- 342 FIELDIANA: GEOLOGY, VOLUME 13 preted as a nasal opening. Watson (1954, p. 12) described but did not figure rounded notches at the lateral ends of the maxillary brim of Anglaspis macculloughi; these he interpreted as nostrils, and com- '— ds orp — -"""'^ ^ -ds - vs Fig. 105. Sagittal sections through anterior part of shield of Protopteraspis vogti (from Heintz, 1962). A, as preserved; B, as restored. ds, dorsal shield; mxb, maxillary brim; orp, oral plates; ort, oral tooth lamella; ro, rostrum; vs, ventral shield. pared their position to that of lungfishes. Stensio (1958, fig. 193) showed similarly placed notches in Poraspis pompeckji, but he inter- preted them as probably housing tentacles. Nostrils in the position of these notches would be bounded medially by oral-lateral or pos- sibly lateral plates. Openings in this position are not apparent in any articulated pteraspids or cyathaspids, and so their existence is doubtful. If external nostrils were absent, the olfactory organs pre- sumably opened into the anterior part of the buccal cavity. In Stensio's latest reconstruction (1958, figs. 190, 200), the nasal cap- sules are unpaired, lie just in front of the pineal organ, and are con- nected with the exterior by the hypothetical naso-hypophysial duct and prenasal sinus. Brain. — The pineal organ, usually indicated by the pineal macula externally (fig. 91, A, pm), and by a corresponding pit on the inner side of the dorsal shield (figs. 106, 132,A, 151,A, 155, 156, pfo), is the most important landmark in the anterior part of the brain. It is logical to assume that it lay, as in lampreys, at the anterior end of the diencephalon, and that the telencephalon lay antero-ventrally DENISON: THE CYATHASPIDIDAE 343 to it. There is no evidence that the forebrain was crowded as in embryonic myxinoids, as claimed by Stensio (1958, p. 374). The position of the myelencephalon is indicated on the inner side of the shield by a relatively broad groove that tapers posteriorly into the groove for the spinal cord (figs. 106, 132,A, 155, br). Semicircular canals. — The pair of V-shaped markings on either side of the myelencephalon indicate the position of the anterior and posterior vertical semicircular canals (figs. 106, 132,A, 151, A, 155, 156, sec). It has generally been assumed that the horizontal semicir- cular canals were absent as in cyclostomes; however, because of their deeper position they could not have left any mark on the dermal shield, so it is perfectly possible that they were present in Hetero- straci. Gills and Visceral Arches. — The series of paired impressions on either side of the dorsal and ventral shields has been recognized since Woodward's identification (1891, p. xvii) as indicating gill chambers (fig. 91, k2-ks) . Between the gills we may safely assume the presence of a visceral skeleton, possibly of cartilage. The points requiring discussion are the homology, number, type, and functioning of the gills and their arches. Save-Soderbergh (1941, p. 531) made the assumption that the first postorbital visceral arch, that is, the one lying immediately an- terior to the first gill pouch, was the segment innervated by the tri- geminal nerve. Stensio (1958, p. 378) agreed with this interpretation. Watson's (1954, p. 15) identification of gill pouches in Anglaspis is confusing because apparently he identified as the first pair some markings between the orbits labeled "de" by Wills (1935, p. 433, fig. 2B). I have been unable to find these markings on Chicago Nat- ural History Museum specimens of Anglaspis macculoughi. There is a marking here only on one side of the figured specimen of A. insignis (Kiaer, 1932, pi. 6, fig. 1). These markings are definitely absent on A. expatriata (fig. 151, A), Poraspis (fig. 106), Seretaspis, "Archegon- aspis" drummondi (fig. 156), and Pteraspis, so it is unlikely that "de" represents gill pouches. Watson's "third" pouch is lateral to the ampulla of the anterior vertical semicircular canal, the "second" pouch he identifies as "hyoidean or spiracular," and so the visceral arch anterior to it would be the mandibular arch; it is the one so identified by Save-Soderbergh and Stensio. Homology of individual gills with those of gnathostomes can best be determined by innervation, and in Heterostraci that can only be suggested by topographical relations to other cranial landmarks. Of 344 FIELDIANA: GEOLOGY, VOLUME 13 the latter, the best is the vestibular region as marked by the semi- circular canals. Typically in vertebrates, the facial nerve issues an- terior to the vestibular region, and the glossopharyngeal nerve poste- rior to it. Assuming that these nerves sent their post-trematic branches nearly directly laterad to the adjacent gill arch (which is not necessarily so), either the second or the third visceral arch of cyathaspids is hyoidean. I know no evidence which permits a more precise determination. The second arch is the one identified by Save-Soderbergh, Watson, and Stensio as hyoidean. If, on the other hand, the third is hyoidean, then both mandibular and premandibu- lar arches would be present in Heterostraci. The number of gills in Heterostraci can be determined only by the number of impressions on the dorsal and ventral shields. Since these are sometimes incompletely preserved, and commonly not indi- cated at all, it is difficult to generalize about the total number. Sten- sio (1958, pp. 384, 387-388) considered the number of gills as a character of taxonomic importance, and claimed that there are 5 to 11 innervated by the vagus, meaning 8 to 14 in all. It will be neces- sary to consider in detail the evidence for Stensio' s counts. Before doing this, it should be pointed out that Stensio labels the first pouch "k2," so that the total number is one less than the labelled number of the most posterior gill. Stensio presents the following figures as evidence: Poraspis pompeckji, fig. 179 A, a photograph. k2-k9 are labeled, but k% is the most posterior one visible, and the most posterior one shown in his drawing, fig. 193, and in another photograph, fig. 206. Poraspis cylindrica, fig. 179B, a photograph. k2-k9 are labeled, but &9 does not show on his photograph, nor on a cast of this speci- men (my fig. 106). Homalaspidella nitida, fig. 180B, a photograph. k2-k8 are labeled, but only k2-k7 appear on the photograph (see also Kiaer and Heintz, 1935, pi. 30, fig. 3). Anglaspis insignis, fig. 180A, a photograph, and fig. 205B, a draw- ing. k2-k9 are labeled, and ki0-kn are added in the drawing, but A:8 is the most posterior shown on the photograph and on a cast of this specimen. Anglaspis macculloughi, fig. 205A, a drawing. k2-ki2 are shown. Watson (1954, p. 10) says there are 10-11 gills in this species, but Wills (1935, p. 433, fig. 2B) indicates that it has only seven (k2-kg). Seretaspis zychi, fig. 204, a photograph. Ar2— A:9 are labeled, but the photograph and the specimen show only k2-ks. DENISON: THE CYATHASPIDIDAE 345 Archegonaspis Integra, fig. 202, a drawing. Shows k2-k9. Protopteraspis primaeva, fig. 197B, a drawing. Shows k2-k9f but photographs, figs. 197A and 199A, show only k2~ks. Fig. 106. Poraspis cylindrica; dorsal shield preserved largely as an internal impression and showing molds of gill pouches, pineal fossa, semicir- cular canals and cranial cavity (from Kiaer, 1932). Type, Paleontologisk Museum, Oslo, D 205 ( X about 2). It will be noted that no evidence is presented for more than seven pairs of gills (that is, k2-ks). Precisely the same seven pairs are present in "Archegonaspis" drummondi (fig. 156), Listraspis canaden- sis (fig. 132, A), Vernonaspis sekwiae, V. vaningeni, and Anglaspis expatriata (fig. 151, A). Only six can be seen in Allocryptaspis ellip- tica (Bryant, 1935, fig. 1, and pi. 5), and in Homalaspidella niiida (see p. 344). The seventh gill (k8 of Stensio) lies just in advance of of the external branchial opening, and it is probable that there were no gills behind it, and no posterior branchial ducts or other related structures inferred by Stensio (1958, pp. 393-394). The impressions of the gills on the dorsal and ventral shields in- dicate that the gill chambers were transversely arranged, very high 346 FIELDIANA: GEOLOGY, VOLUME 13 dorso-ventrally, and short antero-posteriorly (Watson, 1954, pp. 10- 11, figs. 4, B, C). A series of corresponding small impressions lateral to the main gill impressions (fig. 91,A, ZAr2_s) have been interpreted as representing extrabranchial atria by Stensio (1958, pp. 389-391), but Watson (loc. cit.) believed that they were formed by the exhalant ducts of each gill. The atria or ducts probably opened on either side into a common branchial duct, sometimes marked by a groove on the inner side of the branchial plate, and perhaps on the adjacent part of the lateral epitegum. The common branchial duct opened to the exterior at the external branchial opening, which lies close behind the most posterior gill chamber. Watson believed that the gills were pouches of adult cyclostome pattern, but there are important differ- ences. In cyclostomes the gills are relatively small, rounded pouches, surrounded by muscles and by peribranchial sinuses. In cyathaspids the gills occupied a large percentage of the volume within the shield, and because of their impressions on the dermal shield, could not have been covered by either muscles or sinuses. Stensio (1958, pp. 375, 399) believed that the gills were not pouch-like, but were more of the type inferred in Osteostraci. He interpreted a series of grooves cross- ing the gill markings of Poraspis, Seretaspis, and Protopteraspis as imprints of gill lamellae (op. cit., p. 366, figs. 199A, 203A, 204). This interpretation is questionable, since it is hard to imagine how gill lamellae could have left markings on the dermal skeleton. It is possible that these grooves are indications of a visceral endoskeleton, or of blood vessels. In conclusion, it appears that too little is known about heterostracian gills to compare them in any detail with those of other groups. Watson (1954, pp. 10-12) and Stensio (1958, pp. 395-397) have both considered the manner in which cyathaspid gills might have functioned. They pointed out that the carapace was effectively of fixed volume, and that this made impossible inspiration and expira- tion by means of movements of the whole pharyngeal region. Ciliary currents were considered to be inadequate, and muscles that could contract the gills were presumably absent. Therefore, they inferred the presence of a muscular pump to induce the respiratory current, and have considered that it may have been a velum such as pumps water to the gills of Myxine. The visceral endoskeleton may have been formed of cartilage, or perhaps merely of stiffened connective tissue. It is not necessarily the coherent structure with extensive connections to the endocranium that Stensio reconstructs (1958, fig. 198). If, as I have suggested DENISON: THE CYATHASPIDIDAE 347 above, there were no gills posterior to the branchial opening, then the visceral skeleton did not extend as far posteriorly as shown by Stensio. He shows a clearly marked posterior limit for the visceral endoskeleton in Poraspis cylindrica (1958, fig. 179B, I. vise, p); this marking is not present in the original figure of this specimen (Kiaer, 1932, pi. 3, fig. 3), or in a Chicago Natural History Museum cast of it (fig. 106). Stensio also shows the posterior limit of the visceral en- doskeleton in a drawing of Anglaspis macculloughi (1958, fig. 205A), but this does not appear in Wills' (1935) figures, or in Chicago speci- mens of this species. It is clearly absent in the type of Anglaspis expatriata (fig. 151,A) in the position shown by Stensio. RELATIONSHIP OF HETEROSTRACI Since the Heterostraci, to which the Cyathaspididae belong, are the earliest known vertebrates, their relationship to other vertebrate groups is of considerable significance. There has been, and still is, a divergence of opinion regarding their systematic position. The history of that opinion has been reviewed by Kiaer (1924, pp. 113- 119) and Kiaer and Heintz (1935, pp. 19-31), so only the most sig- nificant contributions need be mentioned here. In 1889, Cope re- ferred both the living cyclostomes ("Marsipobranchii") and fossil "Ostracodermi" (including Heterostraci) to the Class Agnatha. In 1932 Kiaer subdivided the Agnatha into two different groups: (1) the Monorhina, including the Osteostraci, Anaspida, and Cyclosto- mata, all characterized by an unpaired nasal opening, and (2) the Diplorhina, including the Heterostraci and Coelolepida, with paired nasal openings. A somewhat different form of this classification had previously been offered in 1924 by Kiaer. In 1927 and 1932 Stensio proposed another grouping: the Osteostraci, Anaspida, and Petro- myzonida were grouped in one subclass (Cephalaspidomorphi), while the Heterostraci and Myxinoidea were grouped in another subclass (Pteraspidomorphi). This classification has been defended in his recent (1958) interpretation of heterostracian structure. Today there is broad agreement, at least among paleontologists, that the Osteo- straci, Anaspida, and Petromyzonida are related, but there is lack of agreement about the relationship of the Heterostraci, and about the possibility of a biphyletic origin of Cyclostomata. Some points bearing on these questions will be discussed here. One of the most striking morphological features of the Petromy- zonida is the presence of an unpaired naso-hypophysial opening on 348 FIELDIANA: GEOLOGY, VOLUME 13 the dorsal side of the head. This unusual arrangement results from the great enlargement during development of a post-hypophysial fold or "upper lip," which comes to form the rostral part of the head. The naso-hypophysial opening is dorsally placed in Osteostraci and Anaspida, and so it is probable that these groups also had a similar development of the anterior head region. In the Myxinoidea the arrangement is somewhat different. The post-hypophysial fold en- larges, but not as much, and the single nostril finally becomes ter- minal rather than dorsal in position. The naso-hypophysial duct acquires an opening into the pharynx, and the post-hypophysial fold forms a "secondary palate" separating this duct from the buccal cavity. It is probable, but not yet demonstrated, that the myxinoid condition had a common ancestry with the forerunners of Petromy- zonida in which there was enlargement of the post-hypophysial fold. There is no evidence that Heterostraci partook of these features at all. The nostrils were probably paired and opened into the anterior buccal cavity, and the nasal sacs are believed to have been dorsal to them. This being so, there was no enlargement of the post-hypo- physial fold, and the rostral region was fundamentally different from that of Cyclostomata, Osteostraci, and Anaspida, and similar to that of gnathostomes. The Petromyzonida have only two pairs of semicircular canals, and Myxinoidea have only one, though the latter may each repre- sent the two canals of Petromyzonida. Only two pairs of vertical canals leave impressions in Heterostraci, but this can hardly be taken as evidence of the absence of the horizontal canals; as pointed out above (p. 343), the latter are too deep to leave impressions on the dermal skeleton. The gills of Heterostraci have been compared to those of cyclostomes, but they surely had little resemblance to the gill pouches of adult lampreys and hagfishes. Actually there is little evidence about the type of gills present in Heterostraci, though their size and extent are indicated. The single pair of external bran- chial openings is a point of resemblance to Myxine, but not to other cyclostomes. The long list of characters cited by Stensio (1958, pp. 414-415) in which the Heterostraci are supposed to resemble cyclostomes in general, and myxinoids in particular, are largely char- acters of his reconstruction, and not of Heterostraci at all. I would therefore agree with Kiaer (1924, 1932), White (1935), Obruchev (1945), Wangsjo (1952), Watson (1954), Balabai (1956), Tarlo (1962a), and Heintz (1962) that there is no close relationship of DENISON: THE CYATHASPIDIDAE 349 Heterostraci to Cyclostomata, Osteostraci, and Anaspida. These groups can be classified as follows: Class Agnatha Subclass Diplorhina ( = Pteraspidomorphi) Order Heterostraci ?Order Coelolepida Subclass Monorhina Superorder Hyperotreti Order Myxinoidea Superorder Hyperoartii Order Petromyzonida Order Osteostraci Order Anaspida The Monorhina are unquestionably more specialized than the Diplorhina in the structure of the anterior head region. The Diplo- rhina have their specializations also, but in a number of important features the Cyathaspididae are extremely primitive vertebrates. The gill chambers and the presumed intervening arches form a very uniform series, not crowded posteriorly by the formation of jaws as in gnathostomes, and unmodified by the development of a huge oralo-branchial chamber as in Osteostraci, or of a rasping tongue as in lampreys. It is possible that a relatively unmodified preman- dibular arch is present. The lateral line system is often quite com- pletely developed, and has a simple pattern from which that of other agnathans and gnathostomes could be derived. The presence of a dermal skeleton and the absence of a calcified endoskeleton may be primitive (Denison, 1964, p. 150). The absence of paired fins, and probably of dorsal and anal fins, is surely primitive (for a contrary opinion, see Stensio, 1958, pp. 409-413). It is regrettable that there is no preservable endoskeleton in Heterostraci from which more of the structure of these primitive vertebrates could be determined. SYSTEMATIC REVISION This work is intended primarily as a taxonomic revision of North American Cyathaspididae. Although new diagnoses are presented for all genera, I have restricted myself in most cases to no more than nomenclatorial revision of European species. The family Cyath- aspididae, as used here, is equivalent to the suborder Cyathaspida of Kiaer (1932), and the subfamilies of this paper are approximately equivalent to the families of Kiaer. The characteristics used to dis- tinguish the various categories are mostly derived from the dorsal shield. 350 FIELDIANA: GEOLOGY, VOLUME 13 Subfamilies are distinguished largely by: (1) presence or absence of apparent scale components in the shield; (2) presence or absence of distinct epitega; (3) the pattern, length and uniformity of the super- ficial dentine ridges. Certain subfamilies are distinguished in part by the proportions and shape of the shield, the fineness of the dentine ridges, the branching of the lateral lines, and the absence of the super- ficial layer. Genera are distinguished by: (1) distinctness of epitega, postros- tral field, and pineal macula; (2) pattern, length, shape, coarseness, and uniformity of dentine ridges; (3) shape of the shield, the presence or absence of a median rostral process, and the development of the preorbital processes and postbranchial lobes; (4) proportions, (par- ticularly the width), orbital, pineal, and postbranchial ratios; (5) pat- tern, subdivision, reduction, and branching of the lateral line canals. Other characters used in certain genera are: presence of lateral brims or laminae, and of median crests and spines on the dorsal shield; development and ornament of maxillary brim; presence or absence of postoral plates; ornament when superficial layer is absent; his- tology; and size, if unusually large or small. Species are distinguished primarily by: (1) size, usually measured by the median length of the dorsal shield; (2) coarseness of dentine ridges, as measured by the number of ridges per millimeter, prefer- ably on the central part of the dorsal shield; and (3) proportions, especially the width, orbital and pineal ratios. In addition, in some cases I have also used the rostral ridge pattern, the shape of the dentine ridge crests, and the shape of the rostral and posterior edges of the dorsal shield. CYATHASPIDIDAE This family includes small Heterostraci whose carapace consists of a dorsal shield, a ventral shield, paired branchial, oral, lateral, and suborbital plates. There may be postbranchial lobes but no cornual plates. A dorsal spine is typically absent. The rostrum is not extended, and the ventrally-placed mouth is subterminal. The orbits are bounded above by the dorsal shield and below by suborbital plates. The branchial openings typically lie between the dorsal shield and the branchial plates. There are probably only seven pairs of gills. Scales are of relatively large size, and include median dorsal, dorso-lateral, ventrolateral, and median ventral rows. Except in Ctenaspis, the external surface is ornamented with dentine ridges DENISON: THE CYATHASPIDIDAE 351 that have smooth or gently scalloped edges. The chambers of the cancellous layer are usually large and separated by more or less ver- tical septae. The lateral lines are canals in the cancellous layer that open to the surface by pores; they are arranged in a simple pattern consisting of two pairs of longitudinal canals dorsally and ventrally, transverse commissures, infraorbital, and supraorbital, and perhaps profundus lines. Fig. 107. dorsal shield 1932.) orb, orbit. Tolypelepis undulata, (X 2). (From Kiaer, Tolypelepidinae The dorsal shield appears to be composed in part of fused scales, and is divided into distinct epitega. Most dentine ridges are short, and the central ridges of the scale components are broader and higher than the surrounding ridges. The central ridge pattern is elliptical, the rostral pattern is transverse, and a postrostral field may be dis- tinguished. Tolypelepis. Tolypelepis Pander Type-species. — Tolypelepis undulatus Pander. Tolypelepis Pander, 1856, Mon. foss. Fische Sil. Syst. russ.-balt. Gouvern., pp. 60-61; Lindstrom, 1895, Bihang K. Svensk. Vetensk.-Akad. Handl., 21, Afd. 4, Nr. 3, p. 10; Strand, 1934, Folia Zool. Hydrobiol., 5, pp. 327- 328; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 370; Denison, 1953, Fieldiana: Geol., 11, no. 7 p. 293; Bystrow, 1955, 352 FIELDIANA: GEOLOGY, VOLUME 13 Akad. Nauk. SSSR, Mem. Vol. A. S. Berg, pp. 481-486; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 297-307; Gross, 1961, Acta Zool., 42, p. 145. Tolypaspis F. Schmidt, 1893, Neues Jahrb. Min. Geol. Pal., 1893, vol. 1, p. 100; Rohon, 1893, Mem. Acad. Imp. Sci. St. Petersbourg, (7), 41, no. 5, pp. 76-79, 119-120; Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, p. 83; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 25. Tolpyaspis (in error) Rohon, 1893, Mem. Acad. Imp. Sci. St. Petersbourg (7), 41, p. 76. Oniscolepis (in part) Pander, 1856, Mon. foss Fische Sil. Syst russ.-balt. Gouvern., pp. 56-58. Tolepelepis (in error) Tarlo, 1960, Palaeontology, 3, p. 223. Diagnosis. — The shield is broad and the preorbital length is rather short (orbital ratio=.14). On the central epitegum and postrostral field the dentine ridges are short and are grouped into scale-like areas, with narrower, lower ridges arranged around a coarser, higher central ridge. Tolypelepis undulata Pander. Figures 97,A, 107. Tolypelepis undulatus Pander, 1856, Mon. foss. Fische Sil. Syst. russ.-balt. Gouvern., p. 61, pi. 6, figs. 24a-d. Tolypelepis undulata Denison, 1953, Fieldiana: Geol., 11, no. 7, p.292; Bystrow, 1955, Akad. Nauk. SSSR, Mem. Vol. A. S. Berg, pp. 481-485, figs. 10-12; Stensio, 1958, Traite de Zool., 13, fasc. 1, figs. 169A-B, 170, 215; Gross, 1961, Acta Zool., 42, pp. 108-109, figs. 12M-N. Tolypaspis undulata F. Schmidt, 1893, Neues Jahrb. Min. Geol. Pal., 1893, vol. 1, p. 100; Rohon, 1893, Mem. Acad. Imp. Sci. St. Petersbourg, (7), 41, no. 5, pp. 79-88, fig. 17, pi. 1, fig. 45, pi. 2, figs. 54, 56; Hoppe, 1931, Palaeontogr., 76, Abt. A, p. 58; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 25, pi. 10; Gross, 1950, Abh. Deutsch. Akad. Wiss. Berlin, Math.-Nat. Kl., 1949, Nr. 1, p. 51. Tolypaspis schmidti (in error) Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, p. 83. Oniscolepis magnus Pander, 1856, Mon. foss Fische Sil. Syst. russ.-balt. Gouvern. pp. 56, 58, pi. 6, fig. 32a-c. Oniscolepis magna Rohon, 1893, Mem. Acad. Imp. Sci. St. Petersbourg (7), 41, pp. 79, 81-82, 89; Gross, 1947, Palaeontogr., 96, Abt. A, p. 96; 1961, Acta Zool., 42, p. 108. Type. — Shield fragment figured by Pander (1856, pi. 6, fig. 24). Occurrence. — Late Silurian, Upper Oesel Group (K4), Ohesaare- Pank, Oesel. Diagnosis. — The length of the dorsal shield is 33 to 36 mm. and its width ratio is .67. The dentine ridges have rounded crests and average about 4 per mm., but the largest are about 2 per mm., and the smallest on the lateral epitega are about 8 per mm. DENISON: THE CYATHASPIDIDAE 353 Discussion. — Because Pander's original description of Tolype- lepis undulata was based on a small shield fragment, Schmidt (1893) introduced the new name Tolypaspis in describing a nearly complete shield of the same species. One species of Oniscolepis, 0. magna Pander, has been referred to Tolypelepis by Rohon (1893, pp. 81-82) and by Gross (1947, p. 96; 1961, p. 108). However, Pander's other species of Oniscolepis, 0. crenulata, 0. dentata, and 0. serrata, are not cyathaspids, and may be related to Strosipherus (Gross, 1961, pp. 100-101). The reference to this species by Rohon (1893, p. 79) of specimens from the Ludlow Bone Bed of England is doubtful; his reference (loc. cit.) to "Tolypaspis" of specimens from the "cornstones" of Ledbury, England is probably incorrect. Tolypelepis cf. undulata Tolypelepis cf. undulata Gross, 1961, Acta Zool., 42, pp. 76, 108. Occurrence. — Late Silurian, glacial erratics of Beyrichienkalk, North Germany. Tolypelepis timanica Kossovoy and Obruchev Tolypelepis cf. undulata Sokolov, 1962, 2nd Internat. Arbeitstagung iiber die Silur/Devon-Grenze und die Stratigraphie von Silur und Devon, Bonn- Bruxelles, 1960, Symposiums-Band, p. 252. Tolypelepis timanica Kossovoy and Obruchev, 1962, Doklady Akad. Nauk SSSR, 147, p. 1149 (nomen nudum). Occurrence. — Early Devonian, probably Late Downtonian, Ep- tarmenskaya beds, Velikaya River, northern Timan (approximately 67° 21' N., 48° 33' E.), USSR. Discussion. — This species has only been listed. Tolypelepis sp. Tolypelepis sp. White, 1946, Quart. Jour. Geol. Soc. London, 101, p. 212. Occurrence. — Late Downtonian (zone of Traquairaspis pococki), Herefordshire, England. Tolypelepis n. sp. A Thorsteinsson, 1958, Geol. Surv. Canada, Mem. 294, p. 47. Occurrence. — Middle Silurian (Early Wenlockian, zone of Mono- graptus riccartonensis) , near top of Allen Bay formation on Shella- bear Creek, and member C of Cape Phillips formation on north coast, Cornwallis Island, Canada. 354 FIELDIANA: GEOLOGY, VOLUME 13 Cyathaspidinae Epitega are distinct, but there is little or no trace of scale com- ponents on the shield. Most of the dentine ridges are long. The ridge pattern is slightly elliptical or nearly longitudinal on the cen- tral epitegum, and transverse, in part at least, on the rostral epite- gum; the postrostral field may be distinct. Ptomaspis, Cyathaspis, Archegonaspis, Seretaspis, Vernonaspis, Pionaspis, Listraspis. Ptomaspis Denison Type-species. — Ptomaspis canadensis Denison. Ptomaspis Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 113-114, 140. Diagnosis. — The dorsal shield is of moderate proportions; its or- bital notches are anteriorly placed and shallow, and its posterior margin is gently convex. The dentine ridges include some that are slightly higher between others that are lower and sometimes slightly narrower; they are subdivided into short lengths or even into small denticles. Locally, but especially on the posterior part of the dorsal shield, the ridges are grouped into scale-like areas with parallel ridges. The rostral epitegum has short transverse ridges anteri- orly, and grades through a denticulate area into antero-posterior ridges in the postrostral field. Ptomaspis canadensis Denison. Figures 97,B, 108, 109. Ptomaspis canadensis Denison, 1963, Fieldiana: Geol., 14, no. 7, p. 116, figs. 64-65. Type. — Princeton 17090, a nearly complete dorsal shield (figs. 108, 109). Occurrence. — Probably Early Devonian (Early Downtonian) lime- stones and graptolitic shales,1 Beaver River, southeastern Yukon. Diagnosis. — The length of the dorsal shield is 61 mm. The width ratio is .58. Dentine ridges average about 5 per mm. 1 Tentatively assigned to the Middle Ludlovian in Denison (1963); see below (pp. 450-451). Fig. 108. Ptomaspis canadensis, type; dorsal shield, Princeton 17090 (X 3). (From Denison, 1963.) cd, pores of dorsal transverse sensory commissure; Idl, mdl, pores of lateral and medial dorsal sensory canals; or, orbit; pm, pineal macula; rep, rostral epitegum; soc, pores of supraorbital sensory canal. 355 356 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 109. Ptomaspis canadensis, type; dorsal shield, Princeton 17090 (X 3/2). A, ventro-lateral view; B, rostral view. bro, branchial opening; orb, orbit; pbl, postbranchial lobe; ro, rostrum. Discussion. — The outstanding characteristic of Ptomaspis is the retention of scale-like areas of ornamentation on the dorsal shield. This primitive feature relates it to Tolypelepis, but in the latter the scale-like areas are more distinct and extend over the whole central epitegum and onto the postrostral field (fig. 107), while in Ptomaspis they are clearly marked only on the posterior part of the dorsal shield (fig. 108). In the presence of higher and sometimes broader ridges surrounded by lower and narrower ridges, Ptomaspis resembles a number of other Cyathaspididae — Cyathaspis, Tolypelepis, and some Archegonaspis. This is also believed to be a primitive characteristic, but is well along on the road to reduction and loss in Ptomaspis. The pattern of the lateral line system (fig. 97,B) is indicated by dis- tinct pores and appears to be typical of the family; the lateral and medial dorsal lines are probably nearly continuous instead of being broken into short lengths as in Tolypelepis. Ptomaspis resembles Archegonaspis in lacking a median rostral process (fig. 109,B) and in having long but slightly developed postbranchial lobes (fig. 109, A, pbl). It is intermediate between the Tolypelepidinae and Cyathas- pidinae. DENISON: THE CYATHASPIDIDAE 357 Cyathaspis Lankester Type-species. — Pteraspis banksii Huxley and Salter. Pteraspis (in part) Huxley and Salter, 1856, Quart. Jour. Geol. Soc. London, 12, p. 100. Cyathaspis Lankester, 1865, Rept. Brit. Assoc. Adv. Sci., 1864, Not. and Abstr., p. 58; 1868, Fishes Old Red Sandstone, Mon. Pal. Soc, p. 26; Von Alth, 1886, Beitr. Pal. Geol. Osterr.-Ung., 5, Heft 3, pp. 63-69, 72, 73; Wood- ward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 170; Claypole, 1892, Quart. Jour. Geol. Soc. London, 48, pp. 543-546; Lindstrom, 1895, Bihang K. Svensk. Vetensk.-Akad. Handl., 21, Afd. 4, no. 3, p. 9; Leriche, 1906, Mem. Soc. Geol. Nord, 5, pp. 22-25; Zych, 1931, Fauna Ryb De- wonu i Downtonu Podola, p. 84; Kiaer, 1932, Skr. Svalbard Ishavet, 52, pp. 21-22; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, p. 32; Heintz, 1938, Naturwissensch., 28, p. 51, fig. 2c; Flower and Wayland- Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 365; Denison, 1953, Field- iana: Geol., 11, no. 7, pp. 292-293; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 307-310, 312, 314-316, 319. Scaphaspis (in part) Lankester, 1865, Rept. Brit. Assoc. Adv. Sci., 1864, Not. and Abstr., p. 58. Diplaspis Matthew, 1887, Bull. Nat. Hist. Soc. New Brunswick, no. 6, pp. 69- 73; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 25; Flower and Wayland- Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 371; Denison, 1953, Fieldi- ana: Geol., 11, no. 7, p. 293; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 312-314. Diagnosis. — The dorsal shield is broad and truncate posteriorly. The central epitegum is strongly arched, and the lateral epitega form broad, flat brims. There is a postero-median crest on both the dor- sal and ventral shields. The orbits are situated anteriorly (orbital ratio=.ll-.14), but the pineal organ is rather posterior (pineal ra- tio=.24-.28). On the central epitegum and central part of the ven- tral shield there are a few coarse dentine ridges separated by 1 to 5 finer ridges; elsewhere the ridges are of uniform size. The ridges have a rounded convex crest. The ridge pattern is elliptical on the central epitegum and central part of the ventral shield. There is usually a band of ridges parallel to the anterior and lateral edges of the ven- tral shield. Discussion. — In addition to its shape, one of the most distinctive characters of the Cyathaspis shield is the presence on the central areas of coarser ridges separated by one or a few finer ridges (fig. 100, A, dr, dri). Among the Cyathaspididae this feature is most strongly developed in Tolypelepis where each scale-like area has a coarse cen- tral ridge. It is shown to a slight extent in Ptomaspis and Arche- gonaspis integra. Stensio (1958, p. 310) states that it is present also 358 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 110. Cyathaspis banksi (from Kiaer, 1932); specimens in collection of Geological Survey of Great Britain (X about 3). A, dorsal shield; B, ventral shield. orb, orbit. in Archegonaspis ludensis but this is apparently in error; the speci- men to which he refers (Brit. Mus. P. 3241) is Cyathaspis banksi. Kiaer (1932, p. 22) stated that Cyathaspis had a "real dorsal spine" comparable to that of Pteraspis, but judging from his own figure (op. cit, pi. 8, fig. 2), and according to Lankester (1868, p. 26), this is not a distinct element, but only a crest on the central epitegum. Cyathaspis banksi (Huxley and Salter). Figures 100,A, 110. Pteraspis banksii Huxley and Salter, 1856, Quart. Jour. Geol. Soc. London, 12, p. 100, pi. 2, fig. 2a-d; Huxley, 1858, Quart. Jour. Geol. Soc. London, 14, pp. 274-277, pi. 15; von Alth, 1886, Beitr. Pal. Geol. Osterr.-Ung., 5, Heft 3, p. 64. Pteraspis truncatus Huxley and Salter, 1856, Quart. Jour. Geol. Soc. London, 12, p. 100, pi. 2, fig. la-d; Salter, 1859, Ann, Mag. Nat. Hist., (3), 4, pp. 45-47, fig. 2. Cyathaspis banksii Lankester, 1865, Rept. Brit. Assoc. Adv. Sci., 1864, Not. and Abstr., p. 58; 1868, Fishes Old Red Sandstone, Mon. Pal. Soc, pp. 26-27, pi. 2, figs. 9-11, pi. 4, fig. 6. Scaphaspis truncatus Lankester, 1865, Rept. Brit. Assoc. Adv. Sci., 1864, Not. and Abstr., p. 58; 1868, Fishes Old Red Sandstone, Mon. Pal. Soc, pp. 24- 25, pi. 2, figs. 1-3. DENISON: THE CYATHASPIDIDAE 359 Cyathaspis banksi Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, pp. 171-172, pi. 9, fig. 3; Leriche, 1906, Mem. Soc. Geol. Nord, 5, pp. 23-24, fig. 5; Jaekel, 1927, Zeits. Ges. Anat., 3 Abt., Ergebn. Anat. Entwickl., 27, p. 877, fig. 25; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 22, pi. 8; White, 1950, Bull. Brit. Mus. (Nat. Hist.), Geol., 1, p. 54; 1958, Studies on Fossil Vertebrates, ed. Westoll, p. 218; Stensio, 1958, Traite de Zool., 13, fasc. 1, fig. 171. Type. — Dorsal shield, not designated. Occurrence. — Late Silurian (Late Ludlovian, Upper Whitcliffe Beds) and Early Devonian (Early Downtonian, Ludlow Bone Bed and Downton Castle Sandstone), Shropshire and Herefordshire, England. Diagnosis. — The length of the dorsal shield ranges between 38 and 45 mm., and its width between 26 and 38 mm. There are about 5 dentine ridges per millimeter. Cyathaspis acadica (Matthew). Figure 111. Pteraspis (7) acadica Matthew, 1886, Canadian Rec. Sci., 2, pp. 251-252, 1 fig. Diplaspis acadica Matthew 1887, Bull. Nat. Hist. Soc. New Brunswick, no. 6, pp. 69-73, 1 fig.; 1888, Trans. Roy. Soc. Canada, 6, sec. 4. pp. 49-52, pi. 4, figs. 1-4; Claypole, 1892, Quart. Jour. Geol. Soc. London, 48, p. 547. Cyathaspis acadica Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 173; Hay, 1902, Bull. U. S. Geol. Surv., 179, p. 340; 1929, Carnegie Inst. Washington, Publ. 390, vol. 1, p. 635; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, p. 34. Type. — Royal Ontario Museum 1117, incomplete but associated dorsal and ventral shields (fig. Ill), originally figured by Matthew (1888, pi. 4, figs. 1-4). Occurrence. — Late Silurian, Jones Creek formation,1 Cunning- ham Brook, near Nerepis, Kings County, New Brunswick. Diagnosis. — The size is somewhat smaller than that of C. banksi. There are 7 to 8 dentine ridges per millimeter. Discussion. — Matthew (1888, pp. 52-53) recognized that this species was close to Cyathaspis, but he referred it to a new genus, Diplaspis, because he thought that its "lateral cornua" (= lateral epitega) were divided, and that its "ocular plates" (= supraorbital parts of lateral epitega) were wanting. However, the lateral epitega are formed essentially as in other Cyathaspidinae; their division in the type is the result of breakage, and each does include a supra- 1 Listed in error as Long Reach formation by Denison (1956, p. 384); according to the map published by MacKenzie (1951), this is well within the underlying Jones Creek formation. 360 FIELDIANA: GEOLOGY, VOLUME 13 orbital part (fig. 111). The latter is to be identified as the anterior end with curved ridges of the plates labeled b and b' in Matthew's - :'■ rCP v % — ■ -- . lepi ,lep, Fig. 111. Cyathaspis acadica, type, parts of dorsal and ventral shields; Royal Ontario Museum 1117 (X 5/2). cep, central epitegum; lep, lateral epitegum; rep, rostral epitegum; vs, ventral shield. figure (1888, pi. 4, fig. 4). In its known characters there is nothing to distinguish this species from Cyathaspis and very little to distin- guish it from C. banksi. The type of ornamentation is precisely the same, with round-topped dentine ridges of two sizes on the central parts of the dorsal and ventral shields (fig. 159,A) and of uniform size elsewhere. The pattern of the ridges is similar also. The pat- tern of the postrostral field has not been described in Cyathaspis banksi; in C. acadica the pineal macula is marked by circular ridges around a tubercle (Matthew, 1888, pi. 4, fig. 2), and in front of it ridges radiate toward the rostral epitegum. The ventral shield is surrounded anteriorly and laterally by a band of uniform-sized ridges; Matthew (1888, p. 51) described an anterior triangular area of irreg- ular ridges on it, comparable to the postrostral field of the dorsal shield ; this area is not now preserved on the type nor is it recogniz- able on Matthew's figures. It is possible that the fragment e of Matthew's figure (1888, pi. 4, fig. 1) represents part of a branchial plate; this part of the type is no longer preserved. The shape, pro- portions, and size of C. acadica cannot be accurately determined from material available at present. DENISON: THE CYATHASPIDIDAE 361 During the process of preservation, the type specimen has not only been flattened, but has also been broken into separate parts. The disarticulation of the ventral shield and branchial plates is not remarkable, since this has occurred in nearly all known specimens of cyathaspids. However, the elements of the dorsal shield — the ros- tral, lateral, and central epitega — are also separated in the type. This is the result of crushing, but the breaks have taken place be- tween the epitega, at what are presumably lines of weakness. Kiaer and Heintz (1935, p. 34) concluded that the shield consisted of "sev- eral wholly isolated portions," that is, of separate plates, and this specimen is support for this view. In Chicago Natural History Museum there are two specimens from the type locality that probably belong to this species. They differ from the type in having generally coarser dentine ridges (5-6 per mm.). PF 1799 is a fragment of the anterior part of a ventral shield showing coarser and finer ridges over much of its surface. Its ridge pattern shows the convergence of the anterior part of the ellip- tical central pattern, and most anteriorly a divergence or radiation, unsymmetrically developed ; there is no indication of an anterior band of transverse ridges. PF 1800 is probably a complete Cyathaspis cf . acadica contained in a coprolite. This is spread over an oval area about 40 by 50 mm. parallel to the bedding, and contains a crowded mass of scales and plate fragments. A shield fragment about 5 mm. wide covered with uniform (5 per mm.) straight ridges probably belongs to a lateral epitegum. Fragments, presumably from the central epitegum or ventral shield, display the typical coarser and finer ridges. Other shield fragments, possibly from the postrostral field, the lateral part of the rostral epitegum, or the anterior part of the ventral shield show a local subdivision of the ridges into short lengths or into denticles. A long narrow ridge scale is covered with finer (6-7 per mm.), high-crested ridges. Other smaller scales have broader ridges with lower crests. A thin section (fig. 159,A) shows this to be a juvenile specimen, with the reticular, cancellous, and basal layers not yet formed. Cyathaspis(?) miroshnikovi Obruchev Cyathaspis miroshnikovi Obruchev, 1958, Sovietskaya Geologiya, 1958, no. 11, p. 43 (nomen nudum). Occurrence. — Early Devonian (lower Zubova horizon), near No- rilsk, Nats. Okrug, Krasnoyarskiy Territory, Siberia. 362 FIELDIANA: GEOLOGY, VOLUME 13 Cyathaspis sp. Cyathaspis sp. ind. Obruchev, 1938, Central Geol. & Prosp. Inst., Mater., Gen. Ser., 2, pp. 36-38, 42, pi. 1, fig. 1; 1958, Sovietskaya Geologiya, 1958, no. 11, p. 40. Occurrence. — Silurian (?Wenlockian), River Siren-gupan, Basin of River Belaya, southern Urals, USSR. Discussion. — This is based on a shield fragment showing dentine ridges of two sizes, coarser ones separated by three finer ones. An- teriorly there is an area ornamented with small denticles. The refer- ence to Cyathaspis is not certain. Archegonaspis Jaekel Type-species. — Cyathaspis (Pteraspis) integer Kunth. Pteraspis (in part) Salter, 1859, Ann. Mag. Nat. Hist., (3), 4, pp. 44-55. Scaphaspis (in part) Lankester, 1868, Fishes Old Red Sandstone, Mon. Pal. Soc, p. 25. Cyathaspis (in part) Kunth, 1872, Zeits. Deutsch. Geol. Ges., 24, pp. 1-8; Geinitz, 1884, Zeits. Deutsch. Geol. Ges., 36, pp. 855-857; von Alth, 1886, Beitr. Pal. Geol. Osterr.-Ung., 5, Heft 3, pp. 63-69, 72-73; Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 173; Lindstrom, 1895, Bihang K. Svensk. Vetensk.-Akad. Handl., 21, Afd. 4, Nr. 3, pp. 1-10; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 36-37. Palaeaspis (in part) Jaekel 1911, Die Wirbeltiere, pp. 32-33. Archegonaspis Jaekel, 1927, Zool. Anz., 70, p. 282; 1927, Zeits. gesamte Anat., Abt. 3, Ergeb. Anat. Entwickl., 27, pp. 854, 877, 927; 1928, Pal. Zeits., 9, p. 260; Kiaer, 1932, Skr. Svalbard Ishavet, 52, pp. 22-23; Heintz, 1933, Zeits. Geschiebeforsch., 9, pp. 125-130; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 365-366; Denison, 1953, Fieldiana: Geol., 11, no. 7, pp. 292-293; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 307-314, 384, 386, 389, 393; Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 108, 113, 140. Aequiarchegonaspis Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 307, 308, 310-315. Lauaspis Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 307-316, 319. Diagnosis. — The shields are of moderately broad proportions (width ratio= .61-.62). The orbits and pineal organ are placed more posteriorly than in most Vernonaspis (orbital ratio=.13-.16; pineal ratio=.23-.26). There is no median rostral process (fig. 113), but the preorbital processes are well developed. The postbranchial lobes are small. The posterior edge of the shield is nearly transverse. There are 4-5 dentine ridges per millimeter, most of them round- topped. The ridge pattern is transverse on the rostrum. The post- DENISON: THE CYATHASPIDIDAE 363 rostral field has a fanned or irregular pattern with the ridges occur- ring as short lengths or denticles. On the ventral shield the pattern is longitudinal, with an anterior area of fanned ridges. Archegonaspis integra (Kunth) Cyathaspis (Pteraspis) integer Kunth, 1872, Zeits. Deutsch. Geol. Ges., 24, pp. 3-7, pi. 1, figs. 1-6. "Pteraspis" integra Lindstrom, 1895, Bihang K. Svensk. Vetensk.-Akad. Handl., 21, Afd. 4, no. 3, p. 9. Cyathaspis integer Lankester, 1873, Geol. Mag., 10, p. 243; Jentzsch, 1879, Zeits. Deutsch. Geol. Ges., 31, p. 793; Roemer, 1885, Palaeont. Abh., 2, pp. 131-132, pi. 10, fig. 1; von Alth, 1886, Beitr. Pal. Geol. Osterr.-Ung., 5, Heft 3, pp. 65, 73; Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 173; Leriche, 1906, Mem. Soc. Geol. Nord, 5, p. 24; Jaekel, 1927, Pal. Zeits., 8, pp. 177-180, fig. 5; Stensio, 1927, Skr. Svalbard Nordishavet, 12, p. 317, fig. 87. Palaespis integer Jaekel, 1911, Die Wirbeltiere, figs. 15, 17. Archegonaspis integer Jaekel, 1927, Zool. Anz., 70, p. 282, fig. 4; 1927, Zeits. gesamte Anat., Abt. 3, Ergeb. Anat. Entwickl., 27, p. 854, 927, figs. 14, 54; 1928, Pal. Zeits., 9, pp. 177-179, fig. 5; Kiaer, 1932, Skr. Svalbard Ishavet, 52, pp. 23-24, pi. 9, fig. 2; Heintz, 1933, Zeits. Geschiebeforsch., 9, pp. 127- 130, fig. 2; Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 310, fig. 173 A, B, 202. Type. — Partially articulated dorsal and ventral shields, branchial plates and scales, figured by Kunth. Occurrence. — Late Silurian (Early Ludlovian), glacial erratics of Graptolithengestein, Schoneberg in Berlin, Germany; also Bydgoszcz (Bromberg), Poland. Diagnosis. — The length of the dorsal shield is about 41 mm. On the central epitegum a few of the ridges stand slightly higher than the others. Archegonaspis lindstromi Kiaer. Figures 100, B, 112. Cyathaspis cf. schmidti Lindstrom, 1895, Bihang K. Svensk. Vetensk.-Akad. Handl., 21, Afd. 4, no. 3, pp. 1-10, fig. 1, pis. 1-2; Leriche, 1906, Mem. Soc. Geol. Nord, 5, p. 23. Archegonaspis lindstromi Kiaer, 1932, Skr. Svalbard Ishavet, 52, pp. 23-24, pi. 9, fig.l; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 366. Archegonaspis lindstromi Heintz, 1933, Zeits. Geschiebeforsch., 9, pp. 130-131, fig. 3; Spjeldnaes, 1950, Ark. Min. Geol., 1, Nr. 8, p. 213; 0rvig, 1951, Ark. Zool., (2), 2, No. 2, fig. 5B. Cyathaspis lindstromi Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 36-37. 364 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 112. Archegonaspis lindstromi, type (from Lindstrom, 1895); Stockholm, Naturhistoriska Riksmuseet, C 1564 ( X 3/2). A, dorsal shield; B, ventral shield. cep, central epitegum; lep, lateral epitegum; orb, orbit; prf, postrostral field; rep, rostral epitegum. Lauaspis lindstromi Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 307, figs. 172A,B. Type. — Associated dorsal and ventral shields, Naturhistoriska Riksmuseet, Stockholm, C 1564. Occurrence. — Late Silurian (Early Ludlovian), Marl Shales of Hemse Group, Lau Canal, Gotland, Sweden.1 Diagnosis. — The length of the dorsal shield is 47 mm., and that of the ventral shield is 44 mm. The dentine ridges are strongly convex and of uniform size. The shields of the type appear some- what broader than those of other species as a result of crushing. Archegonaspis cf. lindstromi Onchus windti Weigelt, 1930, Zeits. Geschiebeforsch., 6, pp. 6-10. Archegonaspis cf. lindstromi Heintz, 1933, Zeits. Geschiebeforsch. 9, pp. 130- 131, fig. 1. Type. — Associated branchial plates and scales; Heimatmuseum at Kothen, Germany. 1 Listed in error from Skane, south Sweden by Kiaer (1932, p. 23). DENISON: THE CYATHASPIDIDAE 365 Occurrence. — Late Silurian (Early Ludlovian), glacial erratic of Graptolithengestein, Muntzeschen quarry, Kothen, Saxony, Germany. cep — ^*- Fig. 113. Archegonaspis schmidti, rostral view, drawn from cast of type speci- men (X 3). cep, central epitegum; lep, lateral epitegum; orb, orbit; pop, preorbital process; rep, rostral epitegum; ro, rostrum. Archegonaspis ludensis (Salter) Pteraspis ludensis Salter, 1859, Ann. Mag. Nat. Hist., (3), 4, pp. 45-47, fig. 1. Scaphaspis ludensis Lankester, 1868, Fishes Old Red Sandstone, Mon. Pal. Soc, p. 25, pi. 2, figs. 4, 4a. Cyathaspis ludensis Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 173. Archegonaspis ludensis Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 24; White, 1958, Studies on Fossil Vertebrates, Westoll ed., p. 218. Type. — Ventral shield, Geological Survey and Museum, London, 49106. Occurrence. — Late Silurian (Middle Ludlovian), highest Lower Leintwardine beds, (according to Dr. J. D. Lawson, in litt.) Church Hill, Leintwardine, Herefordshire, England. Diagnosis. — The length of the ventral shield is about 38 mm. The dentine ridges are of uniform size. The reference of this species to Archegonaspis is not firmly established. Archegonaspis schmidti (Geinitz). Figure 113. Cyathaspis schmidti Geinitz, 1884, Zeits. Deutsch. Geol. Ges., 36, pp. 854-857, pi. 30; Roemer, 1885, Palaeont. Abh., 2, p. 132. (?)Cyathaspis schmidti Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 173. Archegonaspis integer var. schmidti Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 24. Aequiarchegonaspis schmidti Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 307, fig. 174. 366 FIELDIANA: GEOLOGY, VOLUME 13 Type. — Dorsal shield, Rostock University Museum (fide Wood- ward, 1891). Occurrence. — Late Silurian (Early Ludlovian), glacial erratic of Graptolithengestein, Rostock, northern Germany. Diagnosis. — The length of the dorsal shield is 38 mm. The den- tine ridges are of uniform size, and perhaps less strongly convex than in A. Integra and A. lindstromi. Archegonaspis sp. Archegonaspis sp. White, 1958, Studies on Fossil Vertebrates, Westoll ed., p. 218. Occurrence. — Late Silurian (Middle Ludlovian) and (?) Early De- vonian (Bone Bed), Radnorshire, Wales. Seretaspis Stensio Type-species. — Seretaspis zychi Stensio. Seretaspis Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 314, 315, 316, 318, 366, 371, 384, 386, 387, 391, 393. Diagnosis. — Seretaspis resembles Archegonaspis in most of its known characteristics. Its epitega, including the postrostral field, are clearly distinct. Its dentine ridges are long and of quite uniform size. There is a slight median convexity, but no lobe, on the rostral border. The dentine ridges are very fine. Seretaspis zychi Stensio Seretaspis zychi Stensio, 1958, Traite de Zool., 13, fasc. 1, fig. 204. Type. — Naturhistoriska Riksmuseet, Stockholm, C 1608a, b, in- complete dorsal shield in counterpart. Occurrence. — Early Devonian (Gedinnian), presumably Czortkow stage (labeled "Downtonian"), Podolia. Diagnosis. — The length of the dorsal shield is not known, but its maximum width is estimated to have been 18.5 mm., and its pineal length is 8.8 mm. There are 8-9 dentine ridges per millimeter, and their crowns are gently convex, or sharp-crested near the lateral border. Discussion. — Stensio (1958, p. 366, fig. 204) considered Seretaspis to be related to Anglaspis. Its most striking characteristics, as interpreted by him, and its presumed similarities to Anglaspis, are related to the shortness of the dorsal shield posteriorly. The DENISON: THE CYATHASPIDIDAE 367 latter results in a relatively small number of gill pouches and gill arches, and in a posterior position for the branchial openings. How- ever, the type and only specimen of Seretaspis zychi is clearly incom- plete posteriorly; this is suggested by the asymmetry of the posterior edge shown in Stensio's figure, and is clearly demonstrated by the counterpart of this specimen. Thus there is no way to determine the length or proportions of the dorsal shield, or to count the number of gill pouches and gill arches. In addition, it is probable that the ex- ternal branchial openings lay just posterior to the preserved part of the shield, that there were the usual postbranchial lobes behind them, and that Seretaspis was not strikingly different from typical cyath- aspids in the position of these openings. As far as it is known, the pattern of the dentine ridges is typical of the Cyathaspidinae. On the rostral epitegum the anterior ridges are parallel to the anterior edge and thus more or less transverse. More posterior rostral ridges radiate from a median point on the epitegum's posterior border, and are thus antero-posteriorly directed centrally, and nearly transverse laterally near the posterior edge. The ridges of the postrostral field radiate from the pineal macula, except for one or two ridges that run parallel to the anterior edge of the field. Little is shown of the pattern of the central epitegum, and minor irregularities make it difficult to determine whether the overall pattern is longitudinal or elliptical. On the lateral epitega, the ridges curve over the orbits, are irregular or diagonal behind the orbits, and are longitudinal posteriorly. Little of the lateral lines can be determined. The pores are large and are commonly encircled by a ridge. There is no suggestion of the branchings and irregularities that typify the Irregulareaspidinae. The known characteristics of Seretaspis show little to distinguish it from Archegonaspis. It is considerably younger than known spe- cies of Archegonaspis, and has much finer ridges. This is hardly enough for generic distinction, but Seretaspis is provisionally retained as distinct, pending the discovery of more complete material. Vernonaspis Flower and Wayland-Smith Type-species. — Vernonaspis allenae Flower and Wayland-Smith. Cyathaspis (in part) Bryant, 1926, Proc. Amer. Phil. Soc, 65, pp. 266-270. Eoarchegonaspis (nomen nudum) Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 24. Vernonaspis Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zoo!., 107, pp. 373-374; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 307, 312-315, 319; Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 108-109, 113, 140. 368 FIELDIANA: GEOLOGY, VOLUME 13 Diagnosis. — The shields are moderately broad to narrow (width ratio of dorsal shield= .54-64. The orbits and pineal organ are typ- ically forward in position (orbital ratio= .10-. 14; pineal ratio= .20-22 or more). The postbranchial lobes are long (postbranchial ratio= .34-.45), but their development is weak to moderate. The preorbital processes and the median lobe on the rostral edge of the dorsal shield are strongly developed. The dentine ridges are fine to moderately coarse (5-9 per mm.); they usually have a gently convex crown, though they may be flat-topped anteriorly and sharply crested at the lateral margins of the dorsal shield behind the orbits. The ros- tral epitegum is clearly marked, usually with ridges arranged in part transversely. The postrostral field is not distinct from the cen- tral epitegum. Discussion. — Vernonaspis appears to be widespread and common in the Late Silurian rocks of North America. It is most closely sim- ilar to the contemporary European genus, Archegonaspis, from which it differs in the following characters: its shield is usually narrower; its orbits and pineal organ are usually more anterior; its dentine ridges are usually finer; it has a well-developed medial rostral proc- ess (figs. 117,A, 123,C) and its postrostral field is not distinct from the central epitegum. Differentiation of species in Vernonaspis is difficult. The ridge pattern, which is one of the most obvious features, shows great varia- tion in detail and for this reason cannot be relied on entirely. Shape and proportions are rarely accurately determinable because of in- completeness and crushing of specimens. The species have been distinguished rather arbitrarily, mostly on the basis of size and on the coarseness of ridges. It has been found that two species com- monly occur together, one of them larger and with coarser ridges, and the other smaller and with finer ridges; in each of two localities three species have been determined. Vernonaspis allenae Flower and Wayland-Smith. Figure 114. Vernonaspis allenae Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 375-376, pi. 1, pi. 2, figs. 3, 8. Vernonaspis leonardi (in part) Flower and Wayland-Smith, Bull. Mus. Comp. Zool., 107, pi. 2, figs. 6-7, pi. 5, fig. 2. cf. Vernonaspis leonardi Flower and Wayland-Smith, Bull. Mus. Comp. Zool., 107, pp. 378-380, pi. 3, figs. 2-4, pi. 4, pi. 5, fig. 1. Type.— Museum of Comparative Zoology 8871 (fig. 114), a dorsal shield figured by Flower and Wayland-Smith (1952, pi. 1., pi. 2., figs, 3, 8). DENISON: THE CYATHASPIDIDAE 369 Fig. 114. Vernonaspis al- lenae, photograph of rubber impression of type dorsal shield; Museum of Compara- tive Zoology 8871 (X 3). Referred specimens. — Mus. Comp. Zool. 8873, a ventral shield, and 8874, a dorsal shield. Occurrence. — Late Silurian, Vernon shale (Salina group), 2 miles southeast of Kenwood, Oneida County, New York. Doubtfully from the Wills Creek formation, cut on Western Maryland Railroad at Round Top, near Hancock, Maryland. Diagnosis. — The length of the dorsal shield is 32-33 mm. There are 6-7 dentine ridges per millimeter. The dorsal shield usually has a rounded median lobe on its posterior edge. It is rather narrow, with a width ratio of .54-. 57. Discussion. — The ridge pattern of the type (fig. 114) is as follows: On the rostral epitegum 3 ridges are parallel to the anterior edge; the posterior ridges are transverse but curve forward toward the sides to follow the posterior edge of the epitegum, and the central 370 FIELDIANA: GEOLOGY, VOLUME 13 ridges are fanned or radiate toward the front. The ridges of the postrostral field radiate fan-wise from the pineal organ, then curve laterally and backward into the longitudinal ridges of the central epitegum. The latter are arranged somewhat sinuously and slightly elliptically. The ventral shield referred by Flower and Wayland-Smith to V. leonardi (Mus. Comp. Zool. 8873) has an estimated length of 26 mm., which is about what it should be in V. allenae, and there are 6 ridges per millimeter, as in the latter. The posterior edge has an obtuse median angle. The ridge pattern is essentially longitudi- nal, but the anterior ridges radiate toward the antero-lateral corners, and there appear to be about 4 ridges that may extend transversely across the anterior end. The fine cracks and ridges on the inner sur- face have nothing to do with the lateral line system as believed by Flower and Wayland-Smith (1952, fig. 1). The dorsal shield in counterpart referred by Flower and Wayland- Smith to cf. Vernonaspis leonardi (Mus. Comp. Zool. 8874) is also referred to V. allenae. Its length is 33.5 mm., about as in the type of V. allenae, and there are about 6 dentine ridges per millimeter. It is considerably broader than the type of V. allenae, but this is attributable to flattening. Less certainly referable to Vernonaspis allenae are specimens from the Wills Creek formation at Round Top, near Hancock, Maryland. Four poorly preserved dorsal shields collected by J. R. Beerbower (CNHM, PF 3650-3653) agree in size, proportions, and in coarse- ness of ridges. One of these (PF 3650), but apparently not all, differs from the usual Vernonaspis situation in having all of the rostral ridges transverse. A ventral shield (CNHM, PF 2156) collected by Willard P. Leutze at the same locality agrees in size and ridge coarse- ness, but differs from known Vernonaspis in the somewhat uneven ridge width, the prominence of the lateral line pores, and in having an unornamented brim along the sides and around the antero-lateral corners. Vernonaspis leonardi Flower and Wayland-Smith. Figure 115. Vernonaspis leonardi (in part) Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 376-378, pi. 3, fig. 5. Type. — This was not specifically designated by Flower and Way- land-Smith, but their description was based primarily on a dorsal shield, now Museum of Comparative Zoology 8872 (fig. 115), which is here named lectotype. DENISON: THE CYATHASPIDIDAE 371 Occurrence.— Late Silurian, Vernon shale (Salina group), 2 miles southeast of Kenwood, Oneida County, New York (the lectotype). Late Silurian, High Falls formation, east side of Route 23, 0.9 miles (1.4 by road) south-southeast of Duttonville, Montague town- ship, Sussex County, New Jersey (CNHM, PF 840, 844, 846). Fig. 115. Vernonaspis leonardi, photograph of rubber impression of anterior part of type dorsal shield (from Flower and Wayland-Smith, 1952); Museum of Comparative Zoology 8872 (X 4). Also from High Falls formation, cut on Delaware Water Gap- Millbrook road, opposite Shawnee Island, Warren County, New Jersey (CNHM PF861). Diagnosis. — The length of the dorsal shield is 27-29 mm. The dentine ridges are very fine, 7}^-9 per mm. The shield is moderately narrow, with a width ratio of .54. Discussion. — This species is smaller than V. allenae and about the size of V. vaningeni, but has much finer dentine ridges than either. The gently convex crowns of the dentine ridges of the referred speci- mens are typical of Vernonaspis, but in the type they are flat, with a faint, narrow ridge centered on the crest; this peculiarity is believed to be the result of crushing. Flower and Wayland-Smith distin- guished V. leonardi from V. allenae by shape (but this is incomplete 372 FIELDIANA: GEOLOGY, VOLUME 13 and altered by crushing), by smaller pustules on the surface (a pecu- liarity of preservation), by the thicker cancellous layer (a difference I cannot observe), and by the arrangement of the ridges in whorls extending from the pineal organ (but the pattern is much as de- scribed above in V. allenae) . Three specimens from Montague township, New Jersey (CNHM, PF 840, 844, and 846) show close agreement with the typical V. leo- nardi. One specimen from near Delaware Water Gap, New Jersey (CNHM, PF 861) shows parts of three dorsal shields of about the size of V. leonardi and with very fine ridges. Epitega are distinct except for the postrostral field, and in one dorsal shield the central ridges continue longitudinally through the postrostral field. Vernonaspis vaningeni (Bryant). Figures 116-118. Phyllocarida (in part) Clarke, 1907, Bull. N. Y. State Mus., 107, p. 310, pi. 8, figs. 14-17. Anatifopsis wardelli (in part) Ruedemann, 1916, Bull. N. Y. State Mus., 189, pp. 102-105, pi. 32, figs. 1, 9, 11. Cyathaspis van ingeni Bryant, 1926, Proc. Amer. Phil. Soc, 65, pp. 269-270, pi. 2, fig. 1, pi. 4, fig. 2; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 366-370. Cyathaspis wardelli (in part) Bryant, 1926, Proc. Amer. Phil. Soc, 65, pp. 266-269, pi. 2, figs. 5-6. Eoarchegonaspis (nomen nudum) wardelli Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 24, fig. 12. Archegonaspis van ingeni Beerbower and Hait, 1959, Proc. Pennsylvania Acad. Sci., 33, p. 203, fig. 2. Lectotype. — Bryant did not designate a type in his original de- scription, but Princeton 12916 (fig. 116) is marked as type and is hereby so named. It is an incomplete dorsal shield, figured by Bry- ant (1926, pi. 2, fig. 1). Occurrence. — The material studied by Clarke, Ruedemann, and Bryant all came from near Otisville, Orange County, New York, but the precise localities and horizons are difficult to determine. Cyath- aspididae have been found in three localities along the Erie Railroad tracks in this vicinity. One of these, at Shin Hollow, is clearly in the Longwood shale and contains Americaspis, Vernonaspis sp., and un- determined cyathaspids. The material described by Clarke (1907) probably came from the same locality as that described by Ruede- mann (1916). Both lots were collected by H. C. Wardell and are lithologically similar — greenish-gray siltstones with brown iron stains. Clarke gives the horizon as "gray shales above the grit" in the upper DENISON: THE CYATHASPIDIDAE 373 part of the Shawangunk formation, while Ruedemann says his speci- mens came from "a bed of greenish shaly sandstone above the eu- rypterid-bearing strata." The horizon is the upper member of the Shawangunk formation, named the Otisville Shale member by Swartz and Swartz (1931, p. 651). It was listed as Guymard Quartzite by Kilfoyle (1954, pp. 499-500), but this name has not been widely used. The precise locality is probably not now determinable. Ruedemann (1916, p. 105) gives it as. ". . . railroad cut north of Otisville, on the eastern slope of Shawangunk Mountain ..." This is evidently in error as there is no railroad cut north of Otisville and no Shawangunk formation on the eastern slope of the mountain (Kilfoyle, in litt., Dec. 7, 1961). The material probably came from the western slope of the mountain, possibly near the stone crusher which was formerly located about 134 miles southwest of Otisville. The syntypic material (Princeton 12916, 12919-12921) described by Bryant (1926) was collected by Prof. Gilbert Van Ingen of Prince- ton University. Bryant gives the locality as Guymard, and the speci- men labels say either "near Guymard" or "just N. of Guymard." This is in the vicinity of Graham Station on the Erie Railroad as it descends the western slope of Shawangunk Mountain, and is pos- sibly about two miles south of the locality where Clarke's and Ruede- mann's material was obtained. The horizon is given as Guymard Quartzite, which is probably equivalent to the Otisville Shale mem- ber of the Shawangunk formation. Bryant (1926, pp. 259-260), writing from Van Ingen's notes, says that near its top the Guymard formation contains thin red shales prophetic of the overlying Long- wood shale; it was from the lowermost of these red shales (or more properly siltstones) that the material described by Bryant was ob- tained. The horizon is thus about the same as or perhaps slightly higher than that of Clarke's and Ruedemann's specimens. Additional localities: High Falls formation, east side of route 23, 0.9 miles (1.4 by road) south-southeast of Duttonville, Montague township, Sussex County, New Jersey (CNHM, PF 847-854, 3340). Upper Bloomsburg formation, cuts on U. S. highway 611, about 200 yards north of toll station, Delaware River bridge, Monroe County, Pennsylvania (CNHM, PF 3647-3649, collected by J. R. Beerbower) . Diagnosis. — The length of the dorsal shield is 26.5-28.5 mm. There are 5lA~§lA dentine ridges per millimeter. A median lobe may be absent on the posterior edge of the dorsal shield. The width ratio is about .56. 374 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 116. Vernonaspis raningeni, type; incomplete dorsal shield, Princeton 12916 (X 3). A, dorso-lateral view; B, lateral view of right side. orb, orbit; pbl, postbranchial lobe; pm, pineal macula; pop, preorbital process. Discussion. — The considerable taxonomic confusion that has de- veloped in the history of this species and of "Cyathaspis" wardelli has been reviewed but not resolved by Flower and Wayland-Smith. The syntypic material of Ruedemann's species, Anatifopsis wardelli, comprises the twelve specimens that he figured (1916, pi. 32, figs. 1- 12), but he did not designate a holotype. A lectotype was named by Flower and Wayland-Smith, and they intentionally selected an un- recognizable fragment (New York State Mus. 9613). This made future use of the specific name, wardelli, impracticable, and also effectively prohibited use of the generic name, Eoarchegonaspis, which was based on this species. Since the lectotype of Anatifopsis wardelli exhibits no diagnostic characters, the other eleven specimens cannot be identified with it and are available for systematic restudy. In my opinion this suite of specimens includes two species of Ver- nonaspis, as well as some undeterminable fragments. A Vernonaspis DENISON: THE CYATHASPIDIDAE 375 of smaller size and with finer ridges is described below as a new spe- cies, V. bryanti. A larger form with broader ridges (New York State Mus. 9612, 9620, 9622, 9624-5, 11470-1) agrees with the species described by Bryant as Cyathaspis van ingeni. Bryant also described pop '"""^J! mrp Fig. 117. Vernonaspis vaningeni, anterior part of dorsal shield; New York State Museum 9625 (X 4). A, rostral view; B, dorsal view. mrp, median rostral process; pop, preorbital process. and figured specimens which he referred to Cyathaspis wardelli. The characters by which Bryant distinguished these two species are at- tributable either to distortion in preservation or to misinterpretation of their structure. V. vaningeni was considered to have a narrower shield, but the type (fig. 116) is incomplete and is missing nearly all of the left side posteriorly; if restored it would have a moderately broad shield comparable to that of specimens referred to C. wardelli. Besides, the type of vaningeni is compressed laterally while most of the specimens that Bryant referred to wardelli are compressed dorso- ventrally. The differences he mentions in proportion, arching, and depression of the snout may all be attributed to differences in pres- ervation. Therefore, two of the specimens referred by Bryant to Cyathaspis wardelli (Princeton 12920-1) are here referred to Ver- nonaspis vaningeni, while another (Princeton 12922) belongs to V. bryanti. Kiaer (1932) and Beerbower and Hait (1959) have pre- viously expressed the belief that Bryant's two species were really identical. This species has the characteristics of Vernonaspis as defined above, and is clearly different from Cyathaspis and Archegonaspis. It is about the size of V. leonardi but has coarser ridges. It is smaller than V. allenae, with which it agrees in coarseness of ridges. The ridge pattern is similar in general to that described above in V. allenae but is variable in detail. The central ridges of the rostral epitegum may be longitudinal, fanned, denticulate (in the type), or irregular (New York State Mus. 11470, 9625, fig. 117). The rostral margin is denticulate in Princeton 12919. A pineal macula, rarely seen in 376 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 118. Vernonaspis vaningeni, ventral shield, CNHM, PF 853 (X 3). A, ventral view; B, lateral view of left side. Vernonaspis, consists of an elongate oval of three ridges in the type (fig. 116, pm). The ridge pattern of the lateral epitegum is typical of the genus: the ridges curve over the orbit and converge with trans- verse rostral ridges anteriorly on the preorbital process; behind the orbit a few sharp-crested ridges run parallel to the edge, while medial to them the ridges are arranged diagonally anteriorly and become longitudinal more posteriorly. The ventral shield (fig. 118) is deeply arched posteriorly, has a small postero-median lobe (CNHM, PF854), and a median notch on the anterior border (CNHM, PF853). The dentine ridges are sinuously longitudinal except that they radiate fan-wise at the an- terior end, and a few ridges run parallel to the lateral borders. Vernonaspis bryanti, new species. Figures 119-120. Phyllocarida (in part) Clarke, 1907, Bull. N. Y. State Mus., 107, pi. 8, figs. 20-21. Anatifop&is wardelli (in part) Ruedemann, 1916, Bull. N. Y. State Mus., 189, pp. 102-105, pi. 32, figs. 3, 6-8. DENISON: THE CYATHASPIDIDAE 377 Fig. 119. Vernonaspis bryanti, type; dorsal shield, CNHM, PF 838 ( X 4). lep, lateral epitegum; orb, orbit; pm, pineal macula; rep, rostral epitegum. Cyathaspis ivardelli (in part) Bryant, 1926, Proc. Amer. Phil. Soc, 65, pp. 266- 269, pi. 4, fig. 1. Type.— CNHM, PF 838, a nearly complete dorsal shield (fig. 119). Occurrence. — Late Silurian, High Falls formation, east side of New Jersey Highway 23, 0.9 miles (1.4 by road) south-southeast of Dutton- ville, Montague township, Sussex County, New Jersey (CNHM, PF 838, 839, 841, 842, 845, 855). Otisville Shale member of Shawangunk formation, near Otisville and Guymard,1 Orange County, New York (New York State Mus. 9614, 9617-9619, 9626; Princeton 12922, 18004). Upper Bloomsburg formation, cuts on U. S. Highway 611, about 200 yards north of toll station, Delaware River bridge, Monroe 1 See note on this occurrence under V. vaningeni (p. 373). 378 FIELDIANA: GEOLOGY, VOLUME 13 VPrf Fig. 120. Vernonaspis bryanti, anterior part of type; dorsal shield, CNHM, PF838 (X 6). cep, central epitegum; lep, lateral epitegum; orb, orbit; pro, position of pineal macula, removed on this specimen; prf, postrostral field; rep, rostral epitegum. County, Pennsylvania (CNHM, PF 3646, collected by J. R. Beer- bower) . Diagnosis. — The length of the dorsal shield is 23-24 mm. Its dentine ridges are very fine, 8-9 per mm. The median lobe on the posterior edge of the dorsal shield is slight or absent. The width ratio is .60-. 61. Discussion. — Vernonaspis bryanti is the smallest species of the genus, and is distinguished from all except V. leonardi by its very fine dentine ridges. It may be identical to the species described by Flower and Way land-Smith as Archegonaspis drummondi, which agrees in its small size and fine ridges. However, as is indicated below (p. 444), the latter does not display characters permitting either generic or specific identification, so identity cannot be proved. The type dorsal shield is relatively broader than most other Ver- nonaspis, but the breadth may have been increased by crushing. The postbranchial lobe, well shown in the type and in PF 839, is relatively long (postbranchial ratio=.40) but is weakly developed. The ridge pattern is much as in V. vaningeni and V. allenae, that of the rostral epitegum and postrostral field being quite variable. In the rostral epitegum the central ridge pattern is elliptical in PF 838 (fig. 120), a pair of whorls in PF 839, fanned in Princeton 12922, and irregular and denticulate in New York State Mus. 9617; there DENISON: THE CYATHASPIDIDAE 379 Fig. 121. Vernonaspis bamberi, anterior part of dorsal shield of type (from Denison, 1963); Princeton 17081 (X 4). lep, lateral epitegum; or, orbit; pm, pineal macula; rep, rostral epitegum. are posterior transverse ridges, commonly broken into short lengths, and anterior ridges parallel to the anterior edge. In the postrostral field transverse ridges in the type (fig. 120, prf) curve back into longi- tudinal central ridges; in Princeton 12922 the ridges of the postrostral field are antero-posteriorly and diagonally arranged, and in Prince- ton 18004 there are irregular convolutions. In the ventral shields referred here, the dentine ridges are slightly coarser (7-8 per mm.). The anterior edge has a median concavity, and the posterior edge has a slight median lobe in PF 855, though this is absent in PF 3646. The ridges are mainly longitudinal, but radiate anteriorly toward the antero-lateral corners. One branchial plate (New York State Mus. 9618) is referred here. It is 11.5 mm. long, has a maximum depth of 2.8 mm., and has nearly 10 dentine ridges per millimeter. The ridges are predominantly longitudinal and mostly flat-topped, but are sharp-crested near one edge. This plate is bluntly terminated at the anterior end and tapers toward the posterior end. Vernonaspis bamberi Denison. Figure 121. Vernonaspis bamberi Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 110-113, figs. 60-61. 380 FIELDIANA: GEOLOGY, VOLUME 13 Type. — Princeton 17081, a nearly complete though crushed and distorted dorsal shield (fig. 121). Referred specimens. — Princeton 17082-87; 17091. Occurrence. — Probably Early Devonian (Early Downtonian) lime- stones and graptolitic shales/ Beaver River, southeastern Yukon. Diagnosis. — The length of the dorsal shield is 43-46 mm. There are 5-6 dentine ridges per millimeter. The posterior edges of the dorsal and ventral shields are evenly convex, without median lobes. The width ratio is .56. Discussion. — See under Vernonaspis major. Vernonaspis major Denison. Figure 122. Vernonaspis major Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 110-113, figs. 62-63. Type. — Princeton 17104, a complete dorsal shield (fig. 122). Referred specimens. — Princeton 17105-07, 17376. Occurrence. — Same as V. bamberi. Diagnosis. — The length of the dorsal shield is 57 mm. There are 5-6 dentine ridges per millimeter. The dorsal shield has a rounded median lobe on its posterior edge. Discussion. — The two species from the Yukon are larger than the species from the northeastern United States but are in most respects typical members of the genus. Both commonly have circular ridges around the pores of the supraorbital canal (fig. 122, soc). V. major, especially, shows a tendency to fragmentation of the rostral ridges, and in the type (fig. 122, rep) the anterior part is denticulate, even the ridges that usually run parallel to the anterior edge are present only as denticles. As in other Vernonaspis the pineal organ is usually not marked by any ridge pattern, though there is commonly a small elevation at this point. The ridge pattern of the postrostral field is variable: in the type of V. major the ridges extend a short distance anteriorly from the pineal elevation, curve laterally, then curve back- ward into the longitudinal ridges of the central epitegum; in another specimen (17106) the ridges of the central epitegum continue longi- tudinally through the postrostral field to meet the transverse ridges of the rostral epitegum. In V. major the postbranchial lobes appear to be more strongly developed than in other Vernonaspis, and are 1 Tentatively assigned to the Middle Ludlovian in Denison (1963); see below (pp. 450-451). X 381 Fig. 123. Vernonaspis sekwiae, type; incomplete dorsal shield, National Mu- seum of Canada 10036 ( X 3). A, dorsal view; B, lateral view of left side; C, rostral view. bro, branchial opening; mrp, median rostral process; orb, orbit; pop, preorbital process; rep, rostral epitegum. 382 DENISON: THE CYATHASPIDIDAE 383 terminated quite abruptly anteriorly, that is, at the posterior borders of the branchial openings. Fig. 124. Vernonaspis sekwiae, type; anterior part of dorsal shield, National Museum of Canada 10036 (X 5). cep, central epitegum; lep, lateral epitegum; orb, orbit; pm, pineal macula; rep, rostral epitegum; soc, pores of supraorbital sensory canal. Vernonaspis sekwiae new species. Figures 100, C, 123-124. Type. — National Museum of Canada 10036, dorsal shield, com- plete anteriorly but preserved only as an internal impression in the posterior two-thirds (figs. 123-124). Referred specimens. — CNHM, PF 3673, crushed dorsal shield; PF3674, incomplete dorsal shield; PF 3675-3676, incomplete ven- tral shields; PF 3676-3678 and National Museum of Canada 10037, shield fragments. Occurrence. — Late Silurian, probably Late Ludlovian, top of Mt. Sekwi, beside Keele River, Northwest Territories, Canada; 63° 28' N; 128° 40' W. (California Standard Company locality Z-29-61; hori- zon 4994 '-5080'). Diagnosis. — The dorsal shield has a length of about 33 mm., and ^Yr-^Yi dentine ridges per millimeter. Its posterior edge is nearly transverse, probably with little or no median lobe. Its width ratio is about .64. The orbital and pineal ratios are greater than in other species. Discussion. — This species is of about the same length as Vernon- aspis allenae but is relatively broader. Proportionate width is usually difficult to determine reliably, but two dorsal shields (National Mu- seum of Canada 10036 and PF 3674) are little flattened or distorted, 384 FIELDIANA: GEOLOGY, VOLUME 13 while known specimens of V. allenae are narrower even though flat- tened. The dentine ridges are somewhat coarser in V. sekwiae, especially on the ventral shield, which has only 4-5 per mm. V. sekwiae differs from other Vernonaspis in the more posterior posi- tion of the orbits and pineal organ; the orbital ratio is .14 and the pineal ratio is .24, both within the range of Archegonaspis. How- ever, these ratios may have been increased slightly by the fact that the median rostral process of the type has been bent up so that it projects somewhat anteriorly. On the rostral epitegum the ridges are all transverse (fig. 124, rep) ; there is no suggestion of any antero- posterior ridges and little tendency toward denticulation here, such as occurs in other Vernonaspis. A pineal macula (fig. 124, pm) is clearly indicated in the type by an oval ridge pattern, elongate laterally, but is not present in other specimens. The postrostral field is completely indistinguishable from the central epitegum, as is typical of Vernonaspis. In front of the pineal macula the ridges are transverse, and laterally these ridges curve backward into the some- what elliptical central pattern. There are many minor irregularities in the central pattern of PF 3673. The ventral shield is incompletely known. PF 3675 suggests that it was probably relatively flat anteriorly and deeply vaulted poste- riorly. The ridges are somewhat coarser than on the dorsal shield and are arranged mostly longitudinally. Anteriorly they radiate toward the antero-lateral corners, and there may be an anterior band of transverse ridges. This species resembles Archegonaspis in its relative broadness, in the posterior position of its orbits and pineal macula, and in its com- pletely transverse rostral pattern. However, the presence of a strong median rostral process (fig. 123, C, mrp) and the fact that the post- rostral field cannot be distinguished from the central epitegum, show its relationship to Vernonaspis. Vernonaspis sp. CNHM, PF 863 (fig. 125) is a crushed and incomplete dorsal shield from the Late Silurian, Longwood shale, along the Erie Rail- road tracks, Shin Hollow, about 1% miles south-southeast of Graham Station, Orange County, New York. The estimated total length is 45-50 mm. The median rostral and preorbital processes are moder- ately well developed, and the postbranchial lobes are weak. The dentine ridges are 5-6 per mm., flat-topped on the rostrum and over the orbits, round-crested centrally, and sharply crested along the DENISON: THE CYATHASPIDIDAE 385 lateral edges where they are somewhat finer. The rostral epitegum is covered with denticles and short ridges; antero-medially there are irregularly arranged denticles and short ridges; behind this is a band Fig. 125. Vernonaspis sp., anterior part of dorsal shield, CNHM, PF 863 (X4). of short transverse ridges, then denticulations posteriorly. Some pores of the supraorbital canals are surrounded by circular ridges. This specimen shows many similarities to Vernonaspis major but is smaller and has weaker postbranchial lobes. Princeton 17043 is an incomplete dorsal shield in counterpart from the Late Silurian, High Falls formation, road cut on Dela- ware Water Gap-Millbrook road, opposite Shawnee Island, Warren County, New Jersey. The length of the dorsal shield is about 36 mm. and there are about 6 dentine ridges per millimeter. The epitega and ridge pattern are similar to other species of Vernonaspis. How- ever, this specimen is somewhat larger than other species from this region, and the median rostral process is small. Pionaspis1 new genus Type-species. — Pionaspis planicosta, new species Diagnosis. — This genus includes large-sized Cyathaspidinae with rather broad shields. The epitega are distinctly separated by ridge pattern and apparent sutures, but the postrostral field is not distin- guished from the central epitegum. The dorsal shield lacks a median rostral process, has deep orbital notches, and has short, deep post- branchial lobes with their anterior edges abruptly truncate. The 1 From iriov, plump; and aa-ins, shield. 386 FIELDIANA: GEOLOGY, VOLUME 13 ro - ro — • popi ' H ?'(r'' orb ra Is *& if B Fig. 126. Pionaspis planicosta, type; incomplete dorsal shield, National Mu- seum of Canada 10035 (X 1). A, dorsal view; B, lateral view of left side. bro, branchial opening; orb, orbit; pbl, postbranchial lobe; pop, preorbital process; ro, rostrum. orbital ratio is about .14. The ridge pattern of the central epitegum, including the postrostral field, and of the ventral shield, is semi- elliptical, with the ridges converging only slightly posteriorly, but converging strongly anteriorly to form a continuous curve around the anterior end. Pionaspis planicosta, new species. Figures 126-128. Type. — National Museum of Canada, 10035, incomplete dorsal shield (figs. 126-127). Referred specimen.— CNHM, PF 3681, ventral shield (fig. 128). Occurrence. — Early Devonian, about 30 miles northwest of Mun- cho Lake, British Columbia, 59° 07' 30" N.; 126° 22' W. (locality R5-61, 2100' of California Standard Company). Diagnosis. — The length of the dorsal shield is 77 mm., and of the ventral shield is about 65 mm. The width ratio of the dorsal shield is .62. On the dorsal shield there are 53^-6 dentine ridges per milli- meter, and they are flat-topped, except that laterally and postero- Fig. 127. Pionaspis planicosta, restoration of dorsal shield based on type (X 1). A, dorsal view; B, lateral view of left side. bro, branchial opening; orb, orbit; pbl, postbranchial lobe; ro, rostrum. Fig. 128. Pionaspis planicosta, ventral shield, CNHM, PF 3681 (X 1). A, ventral view; B, lateral view of right side. 387 388 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 129. Pionaspis acuticosta, type; incomplete dorsal shield, National Mu- seum of Canada 10034 (XI); the pores of the lateral line canals have been marked with white ink. A, dorsal view; B, lateral view of right side. bro, branchial opening; lep, lateral epitegum; mdl, medial dorsal sensory canal; pbl, postbranchial lobe; ro, rostrum. laterally the inner edges are elevated above adjacent ridges. On the central part of the ventral shield there are 4-5 flat-topped den- tine ridges per millimeter. Discussion. — See Pionaspis acuticosta. Pionaspis acuticosta new species. Figures 129-130. Type. — National Museum of Canada, 10034, an incomplete dor- sal shield (figs. 129-130). Occurrence. — The same as Pionaspis planicosta. Diagnosis. — The length of the dorsal shield is estimated to be about 80 mm., and the width ratio is estimated to be .75. On the dorsal shield the dentine ridges are 63^-7 per mm. centrally, and 73^-8 per mm. laterally. The ridges are sharp-crested. Discussion. — With the exception of Allocryptaspis the two spe- cies of Pionaspis are the largest known cyathaspids. Their post- branchial lobes (figs. 127,B, 130,B, pbl) resemble those of the Irregu- DENISON: THE CYATHASPIDIDAE 389 -/—pbl Fig. 130. Pionaspis acuticosta, restoration of dorsal shield based on type (X 1). A, dorsal view; B, lateral view of right side. bro, branchial opening; cdi-z, dorsal transverse sensory commissures; cso, trans- verse supraorbital sensory commissure; Idl, mdl, lateral and medial dorsal sensory canals; orb, orbit; pbl, postbranchial lobe; soc, supraorbital sensory canal. lareaspidinae in being relatively short and deep, with the anterior edges abruptly truncate. However, Pionaspis apparently lacks the other characteristic features of that subfamily — the very fine den- tine ridges, the much branched lateral line canals, and the posteri- orly placed orbits. As far as is known, there is nothing to debar Pionaspis from the subfamily Cyathaspidinae. As is commonly the case in Vernonaspis, two species of Pionaspis are found together. The length of the dorsal shields is about the same in the two species, but the breadth of the dorsal shield of P. acuticosta is considerably greater. The type of the latter is crushed and somewhat flattened, but this cannot account for the difference in the width ratio. A transverse measurement following the surface of the shield across the postbranchial lobes is about 75 mm. in P. acu- ticosta, but only about 60 mm. in P. planicosta. Moreover, the dif- ference in shape and coarseness of the dentine ridges indicates that these are two distinct species. The dorsal shield is only slightly arched. The median rostral process is absent, but there is a slight median convexity on the ros- 390 FIELDIANA: GEOLOGY, VOLUME 13 tral margin of P. planicosta. The anterior border of the postbran- chial lobe is turned inward to form a small process that bounds the branchial opening posteriorly (figs. 127,B, 130,B, bro); this process is ornamented with fine tubercles on its external part. The posterior margin of the dorsal shield of P. planicosta is slightly convex with a small median lobe. Little is known of the rostral epitegum. The preorbital process (fig. 126,B, pop), which is formed in part by the lateral part of this epitegum, is ornamented with large, flat denticles. Near the anterior margin, the rostral epitegum of P. planicosta is covered with transversely arranged short ridges and denticles. The lateral and central epitega are more distinct in P. acuticosta than in most cyathaspids; they are separated by an apparent suture, and their dentine ridges meet at an acute angle. The lateral epitegum tapers to a point and ends over the branchial notch. There are prominent irregularities in the ridge pattern of P. acuticosta, but these are perhaps restricted to the region along the lateral lines. The latter are indicated in P. acuticosta by prominent pores, each surrounded by a circular ridge. The pattern of canals (fig. 130), as far as it is determinable, is quite complete and regular. The only ventral shield (fig. 128) is the one that has been referred to P. planicosta because it agrees with the latter in its type of dentine ridges. This shield was probably highly vaulted originally, but it has been somewhat crushed in preservation. Its anterior edge is approximately transverse. A small notch in one antero-lateral cor- ner is probably for a lateral plate. The lateral edge is gently con- cave back to the post-branchial region, where it becomes gently convex; it has a distinct inwardly turned margin, ornamented with fine, short ridges laterally, but unornamented medially where it was presumably overlapped by the branchial plate. The posterior edge is transverse and nearly straight except for a small median lobe that has been broken off in this specimen. Listraspis1 new genus Type-species. — Listraspis canadensis, new species. Diagnosis. — On the dorsal shield the lateral epitega form brims, which are ornamented on both dorsal and ventral surfaces and which extend posteriorly past the branchial openings to form projecting points at the postero-lateral corners of the shield. The branchial plates are attached to the ventro-mesial edges of the lateral brims. 1 From Xiarpov, a spade; and aa-n-is, shield. DENISON: THE CYATHASPIDIDAE 391 The orbits are completely surrounded by bone as a result of the fusion of the suborbital plates to the dorsal shield. The dentine ridges are very fine, with narrow crests and with lateral projections at their bases. Fig. 131. Listraspis canadensis, type; incomplete dorsal shield exposed on inner side, National Museum of Canada 10030 (X 2). h, fossa for sixth gill pouch; Ibr, lateral brim; orb, channel leading to orbit, exposed by removal of ventral rostral lamina; pfo, fossa for pineal organ; ro, ros- trum; sec, depressions for semicircular canals. Listraspis canadensis, new species. Figures 131-133. Type. — National Museum of Canada, 10030, incomplete dorsal shield exhibiting the inner side (fig. 131). Referred specimens. — Incomplete dorsal shields, National Museum of Canada, 10031-10032 and CNHM, PF 3682-3685; ventral shields, National Museum of Canada, 10033, and CNHM, PF 3687-3693. Occurrence. — Early Devonian, about 30 miles northwest of Mun- cho Lake, British Columbia, 59° 07' 30" N., 126° 22' W. (locality R5-61, 2100' of California Standard Company). Diagnosis. — The length of the dorsal shield is 26-32 mm., its width ratio is usually .82-. 89, its pineal ratio is .21-24, and its or- bital ratio is .13-. 15. The length of the ventral shield is 24-28 mm. On the dorsal shield there are 13-14, and on the ventral shield 8-10 dentine ridges per millimeter. 392 FIELDIANA: GEOLOGY, VOLUME 13 Discussion. — This strikingly distinct cyathaspid shows some su- perficial resemblances to pteraspids, and at first glance appears to represent at least a distinct subfamily. The orbits (figs. 131, 133, A, orb) are completely encircled by the dorsal shield, but this is only the result of fusion of the usually separate suborbital plates to the shield. The postero-lateral points on the dorsal shield (fig. 133, A, pip) resemble the cornual plates of pteraspids, but are only extensions of the lateral epitega, not separate elements. A postero-median proc- ess on the dorsal shield (figs. 132,A, 133, A) is only a lobe of the cen- tral epitegum, not a distinct element, as is the dorsal spine of ptera- spids and Ariaspis to which it is analogous. The fine dentine ridges are narrow-crested, and have at the base numerous projections at right angles, alternating with those of adjacent ridges; they are un- like other cyathaspid dentine ridges, and show some resemblance to the crenate ridges of pteraspids. On the dorsal shield there are well- developed lateral brims (fig. 132, Ibr) which are ornamented with den- tine ridges on both dorsal and ventral surfaces; they are roughly comparable to sharply folded branchial plates of pteraspids, though in Listraspis they are, of course, formed by the lateral epitega and have the branchial plates attached to their ventro-mesial edges. These resemblances of Listraspis to pteraspids are clearly superficial. This genus is definitely a cyathaspid, and is probably best considered as a specialized member of the subfamily Cyathaspidinae. It re- sembles the latter in having distinct epitega, long dentine ridges, elliptical ridge pattern on the central epitegum, and transverse pat- tern on the rostral epitegum. Listraspis canadensis is a small cyathaspid with a broad, gently arched dorsal shield, rivaled in its width ratio only by Ctenaspis. One specimen (PF 3683a, fig. 133, A) is relatively narrower; its width ratio is .78, but this is a small, probably juvenile, and perhaps dis- torted specimen. There is also inconsistency in the width ratio as used here, for it, or more exactly the total length measurement, is in- fluenced by the postero-median process, which is not always present. Fig. 132. Listraspis canadensis, dorsal shield, drawn largely from CNHM, PF 3682 (X 3). A, ventral view, with inner face restored in large part from other specimens; B, ventro-lateral view. br, impression of cranial cavity; bro, branchial opening; brp, branchial plate; ifc, infraorbital sensory canal; &>_s, impressions of gill pouches; lap, lateral plate; Ibr, lateral brim; mxb, maxillary brim; pfo, fossa for pineal organ; pip, pos- tero-lateral point; ro, rostrum; sec, impressions of semicircular canals; sip, sutural area for lateral plate; sop, suborbital plate; vlr, ventral lamina of rostrum. axb pfo 393 394 FIELDIANA: GEOLOGY, VOLUME 13 The dorsal shield is clearly divisible into rostral, lateral, and cen- tral epitega. The rostral epitegum is short and has transverse den- tine ridges. The anterior and lateral edges of the rostrum are turned under to form a ventral lamina (fig. 132, A, vlr). The postero-lateral parts of this lamina have lateral line pores (fig. 132,A, ifc) and are equivalent to the preorbital processes of other cyathaspids. The rest of the underturned rostral lamina is ornamented with fine ridges in short lengths, arranged parallel to the margin. The rostral mar- gin has a slight median concavity, reflected by a corresponding con- vexity on the inner margin of the lamina. The maxillary brim proper (fig. 132, mxb) is a nearly vertical, transverse narrow face at right angles to this antero-ventral lamina, and is ornamented with fine, crenate ridges. The central epitegum has a median, longitudinal crest in its cen- tral part. Its dentine ridges are arranged more or less elliptically (fig. 133, A), but posterior to the midpoint, the ridges meet at an angle at the midline, and anteriorly they form continuous curves through the postrostral field, which is not distinct in this genus. The pineal prominence (fig. 131, 132,A, pfo) is large and is only thinly covered by the shield. PF 3683a (fig. 133, A) and National Museum of Canada 10031 have a distinct postero-median process; the ridges of the central epitegum curve into it, showing that it is a part of this epitegum. Each lateral epitegum has its ridges arranged slightly obliquely so that they meet the ridges of the central epitegum at a low angle. Anteriorly, the ridges curve up over the orbits, as is usual in cyath- aspids. At the lateral margin of the shield, each lateral epitegum is folded under to form its ventral ornamented surface. The lateral margin narrows slightly but abruptly at the level of the branchial opening, and then continues posteriorly into the postero-lateral points (fig. 132, pip). The fact that the dentine ridges of the lateral epitegum are usually or mostly continuous into the postero-lateral point indicates that the latter is probably only a posterior extension of the lateral epitegum. Each branchial plate attaches to the medial edge of the ventral surface of the lateral brim, and is directed ventrally or slightly medio- ventrally (fig. 132, brp). The anterior end of the branchial plate is bluntly pointed, and the posterior end, which terminates at the base of the postero-lateral point of the lateral epitegum, is obliquely trun- cate. The branchial opening (fig. 132, bro) is preserved as a slit be- tween the branchial plate and the ventral surface of the brim formed DENISON: THE CYATHASPIDIDAE 395 fo --~-^= Fig. 133. Listraspis canadensis (X 2). A, incomplete dorsal shield, CNHM, PF 3683a; B, ventral shield, largely preserved as an impression, CNHM, PF 3687. orb, orbit; pip, postero-lateral point; ro, rostrum. by the lateral epitegum, and is situated about as is usual in cya- thaspids, that is, well in advance of the posterior end of the shield. The ventral surface of the brim of the dorsal shield continues as a distinct ventral lamina anterior to the branchial plate, under the orbits, and into the much narrower ventral lamina of the rostrum. Discontinuities in the ridge pattern of PF 3682 indicate that the suborbital part of this lamina, which shows pores of the infraorbital canal, is probably a distinct plate (fig. 132, sop) homologous to the suborbital plate of Anglaspis heintzi (fig. 94,A, sop). The part of the ventral lamina anterior to the suborbital plate is equivalent to the preorbital process of other cyathaspids. A small plate with a single lateral line pore (fig. 132, A, lap), lying just in front of the anterior end of the branchial plate, is comparable to a lateral plate of pteraspids. The internal impressions of pineal organ, brain, semicircular ca- nals, and gill pouches are much as in other cyathaspids (fig. 132, A). There are seven pairs of gill pouches (k2-k8 by Stensio's terminol- ogy), with k8 lying only slightly anterior to the branchial opening. The ventral shield (fig. 133, B) has a concave anterior border, and a strongly convex posterior border, the latter sometimes with a slight median lobe. In known specimens, this shield is only gently arched. The dentine ridge pattern is elliptical, except that the ridges flare anteriorly into a band of ridges parallel to the anterior edge. 396 FIELDIANA: GEOLOGY, VOLUME 13 Very little of the dorsal lateral line pattern is discernible. That of the ventral shield agrees with the typical cyathaspid condition as far as it can be made out, and has the medial transverse commissures arranged in Vs. Thin sections of both dorsal and ventral shields have been made, but because of crushing the structure of the basal and middle layers cannot be determined. The superficial layer differs from that of other cyathaspids in its narrow, widely spaced dentine ridges. Irregulareaspidinae Epitega may or may not be indicated, and there is no trace of scale components on the shield. The shield is relatively broad, with the orbits and usually the pineal organ posteriorly placed. The post- branchial lobes are short, deep, and truncate anteriorly. The dentine ridges are very fine and commonly irregular. The lateral line canals are sometimes branched in a complex manner. The dermal skeleton is relatively thin. Dikenaspis, Dinaspidella, Irregular easpis. Dikenaspis Denison Type-species. — Dikenaspis yukonensis Denison. Dikenaspis Denison, 1963, Fieldiana: Geol. 14, no. 7, pp. 116-120, 140. Diagnosis. — The dorsal shield narrows anteriorly toward the or- bits. The pineal organ is intermediate in position (pineal ratio= .24) . The rostrum has a strong median lobe on its anterior margin (fig. 135,B, mrp). The dorsal shield is divided by the ridge pattern into rostral, central, and paired lateral epitega. The ridges are mostly transverse on the rostral epitegum; they converge at the midline an- teriorly on the central epitegum and in the postrostral field. The lateral line canals are branched in a complex fashion; there are two pairs of lateral lines on the rostrum. Dikenaspis yukonensis Denison. Figures 97, E, 101,A, 134-135. Dikenaspis yukonensis Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 117-120, figs. 66-67. Type. — Princeton 17088, a nearly complete dorsal shield and the posterior part of a ventral shield (figs. 134-135). Occurrence. — Probably Early Devonian (Early Downtonian) lime- stones and graptolitic shales,1 Beaver River, southeastern Yukon. 1 Tentatively assigned to the Middle Ludlovian in Denison (1963); see (p. 450). DENISON: THE CYATHASPIDIDAE SOC. 397 Fig. 134. Dikenaspis yukonensis, type (from Denison, 1963); incomplete dorsal shield, Princeton 17088 (X 4). cd, branched dorsal transverse sensory commissure; Idl, mdl, lateral and me- dial dorsal sensory canals; or, orbit; pc, pores of ?profundus sensory canal; pm, position of pineal organ; rep, rostral epitegum; soc, ?supraorbital sensory canal. Diagnosis. — The length of the dorsal shield is about 30 mm.; its proportions are moderately broad, with a width ratio of .63. There are 9-10 dentine ridges per millimeter. Discussion. — The relationship of Dikenaspis to the Irregulare- aspidinae is indicated by a number of characters, especially the pos- terior position of the orbits (orbital ratio=.17), the short, deep postbranchial lobes with abrupt anterior truncations (fig. 135, pbl), the complexly branched lateral lines (figs. 97,E, 134), the fine den- tine ridges, and the thinness of the shield (fig. 101, A). But Diken- :*!is FIELDIANA: GEOLOGY, VOLUME 13 Fig. 135. Dikenaspis yukonensis, type; incomplete dorsal shield, Princeton 17088 (X 3). A, lateral view of left side; B, rostral view. bro, branchial opening; mrp, median rostral process; pbl, postbranchial lobe; pop, preorbital process. aspis differs from other Irregulareaspidinae in the presence of distinct epitega, indicated especially by the more or less transverse ridges of the rostral epitegum (fig. 134, rep), and the semi-elliptical arrange- ment of the anterior ridges of the combined central epitegum and postrostral field (fig. 134). An unusual feature of the lateral line system, unknown in other Cyathaspididae, is the presence of two pairs of longitudinal sensory canals on the rostrum. It is conceiv- able that this is a result of the excessive branching of the sensory canals. However, it may be of more fundamental significance. Some pteraspids have, in addition to the supraorbital canals, a pair of small branches directly over the orbits (e.g., Protopteraspis pri- maera, Kiaer, 1928, fig. 1). Traquair aspis pococki (White, 1946, fig. 40) also has two pairs of anterior sensory canals. Holmgren (1942, p. 12) has suggested that the canals immediately over the orbits may have been innervated by the profundus nerves, and if DENISON: THE CYATHASPIDIDAE 399 this is so, the more medial pairs (figs. 97,E, 134, soc) would be homol- ogous to the supraorbital canals of other fishes. A ventral shield (Princeton 17377) referred here has a length of 26 mm., a width ratio of .67, and fine dentine ridges (8 or more per mm.). The anterior border is concave, the posterior border is convex with a rounded median point, and the lateral borders lack the deep incisions for branchial plates that occur in Dinaspidella and Irregulareaspis. The shield is deeply arched posteriorly but nearly flat anteriorly. The dentine ridges are longitudinally ar- ranged posteriorly, but converge toward the midline anteriorly. The most anterior and lateral ridges are, except for irregularities, continuous around the anterior and lateral parts of the shield. Dinaspidella Strand Type-species. — Dinaspis robusta Kiaer. Dinaspis Kiaer (not of Leonardi 1911), 1932, Skr. Svalbard Ishavet, 52, p. 18; White, 1935, Phil. Trans. Roy. Soc. London, (B), 225, p. 437; Save-Soder- bergh, 1941, Zool. Bidr. Uppsala, 20, p. 530; F0yn and Heintz, 1943, Skr. Norges Svalbard Ishavs-Unders0k., 85, p. 43. Dinaspidella Strand, 1934, Folia Zool. Hydrobiol., 5, p. 327 (new name for Dinaspis Kiaer); White and Moy-Thomas, 1940, Ann. Mag. Nat. Hist., (11), 6, p. 99; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 363, 368; Denison, 1953, Fieldiana: Geol., 11, no. 7, pp. 292, 296; 1963, Fieldiana: Geol., 14, no. 7, pp. 117, 120; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 314-317, 404. Diagnosis. — The dorsal shield is of rather uniform breadth. The pineal macula is situated posteriorly (pineal ratio=.30). The dorsal shield lacks a median rostral lobe and has a very small median lobe on its posterior edge. No epitega are apparent. The rostral region has transverse ridges anteriorly and irregular whorls posteriorly; the rest of the dorsal ridge pattern is mainly longitudinal, but with irreg- ularities. The sides of the ventral shield are deeply incised for the branchial plates. The lateral line pattern (fig. 97,C) is much as in Poraspis except for a transverse commissure in front of the pineal macula, a tendency toward side branches, and minor irregularities. Discussion. — The name Dinaspis is preoccupied, and Dinaspi- della is not available nomenclatorially because both specific names are nomina nuda. Dinaspidella robusta (figs. 97,C, 136) and D. parvula from the Early Devonian (Gedinnian), Fraenkelryggen division, Red Bay ser- ies, Spitsbergen, were merely listed by Kiaer (1932, p. 18) as new 400 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 136. Dinaspidella robusta, type (from Kiaer, 1932); incomplete dorsal shield, Paleontologisk Museum, Oslo, D 454 (X about 2). A, lateral view of left side; B, dorsal view. orb, orbit; pbl, postbranchial lobe; pm, pineal macula. species of Dinaspis. A dorsal shield (fig. 136) in the Paleontologisk Museum, Oslo, D 454 (op. cit., pi. 4, figs. 2-3) was designated as type of D. robusta. cf. Dinaspidella sp. Referred specimen. — CNHM, PF 3871, broken and incomplete dorsal and ventral shields, associated. Occurrence. — Probably Early Devonian, approximately 70 feet above top of "Ronning" dolomite, immediately west of Snake River, northern Yukon; 65° 22' N., 133° 30' W.; (locality Z Fx 36A of Cali- fornia Standard Company). Description and discussion. — In the absence of sufficient diag- nostic characters, the reference of this specimen to Dinaspidella is uncertain. The dorsal shield does not have distinct epitega. Its length and proportions are not measurable, but the orbits and in- DENISON: THE CYATHASPIDIDAE 401 distinct pineal macula probably have a relatively posterior position. The postbranchial lobes are not preserved. The canals of the lateral line system cannot be seen, but its pores are relatively large and distinct on the dorsal shield. The orbits are well marked. The den- tine ridges are very fine, 11 to 12 per mm. dorsally, and as many as 13 per mm. ventrally. The ridge pattern is largely longitudinal, but has whorls and irregularities in the anterior part of the dorsal shield, and shows a flaring towards the antero-lateral corners of the ventral shield. The ventral shield has concave lateral margins similar to those described in Dinaspidella and Irregular -easpis by Kiaer (1932). On a broken edge of the dorsal shield the chambers of the cancellous layer appear irregular, but it cannot be determined whether they have been crushed. The specimen is associated with Corvaspis, and with undeter- mined cyathaspids. These Heterostraci occur about 40 to 50 feet above a limestone containing brachiopods and corals, which are prob- ably of Late Silurian age according to Alfred Lenz (in litt., Dec. 20, 1963). Irregulareaspis Zych Type-species. — Irregular easpis stensioi Zych. Irregulareaspis Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, pp. 83-84; White and Moy-Thomas, 1940, Ann. Mag. Nat. Hist., (11), 6, p. 99; Denison, 1953, Fieldiana: Geol., 11, no. 7, p. 292; 1963, Fieldiana: Geol., 14, no. 7, pp. 117, 120. Irregularaspis (in error) Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, p. 82; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 363. Irregulariaspis (in error) Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 296, 312, 317-318, 404. Dictyaspis Kiaer (not of Haeckel, 1887), 1932, Skr. Svalbard Ishavet, 52, p. 18; White, 1935, Phil. Trans. Roy. Soc. London, (B), 225, p. 437. Dictyaspis (Irregularaspis) F0yn and Heintz, 1943, Skr. Norges Svalbard Ishavs-Unders0k., 85, p. 44; Friend, 1961, Proc. Yorkshire Geol. Soc, 33, pt. 1, no. 5, p. 112. Dictyaspidella Strand, 1934, Folia Zool. Hydrobiol., 5, p. 327 (new name for Dictyaspis Kiaer); Fowler, 1947, Not. Nat. no. 187, p. 4. Dictyonaspis (in error) Berg, 1940, Trav. Inst. Zool., Acad. Sci. USSR, 5, livr. 2, p. 110. Diagnosis. — The dorsal shield is of rather uniform breadth. The indistinct pineal macula is situated posteriorly (pineal ratio=.27- .30). The posterior edges of both dorsal and ventral shields are 402 FIELDIANA: GEOLOGY, VOLUME 13 sharply pointed medially. Epitega are probably not indicated. The dentine ridge pattern is very irregular, composed largely of whorls, and with transverse ridges only along the rostral margin. The ven- tral shield has deep lateral incisions for the branchial plates. The lateral line canals (figs. 97,F, 99,E) are much branched, irregular, and form a network over the shield; their pores are numerous. The dorso-lateral and ventro-lateral scales are subequal. Irregulareaspis stensioi Zych Irregalareaspis stensib'i Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, pp. 83-84, drawings 46-47, photograph 5. Irregulaspis stensioi Brotzen, 1936, Ark. Zool., 28A, no. 22, p. 6. Type. — Not designated. Occurrence. — Early Devonian (Gedinnian), Passage Beds at top of Czortkow stage, Podolia. Diagnosis (based on Zych's figures). — The width ratio=.65. The pineal ratio=. 27. The orbital ratio=. 21. There are about 10 den- tine ridges per millimeter. Irregulareaspis complicata (fig. 97,F), I. hoeli (fig. 99, E), and I. prisca, from the Early Devonian (Gedinnian), Ben Nevis division, Red Bay series, Spitsbergen, were listed by Kiaer (1932, p. 19, figs. 9-10, pi. 5, pi. 6, figs. 2-3) as new species of Dictyaspis. They are nomina nuda. /. hoeli is known from a nearly complete, articulated specimen (Paleontologisk Museum, Oslo, D 474; op. cit., pi. 5). Irregulareaspis sp. Irregulariaspis sp. Stensio, 1958, Traite de Zool., 13, fasc. 1, fig. 175. Occurrence. — Early Devonian (listed as "Downtonien," but pre- sumably from the Czortkow stage, Dittonian), Podolia. Poraspidinae Epitega are indicated faintly or not at all and no scale components are evident. The dentine ridges are long and the ridge pattern is mostly longitudinal, although it is commonly radiating on the an- terior parts of the dorsal and ventral shields, and diagonal on the lateral parts of the dorsal shield. Poraspis, Americaspis, Homalaspi- della, Ariaspis, Anglaspis, Allocryptaspis. DENISON: THE CYATHASPIDIDAE 403 Poraspis Kiaer Type-species. — Holaspis sericeus Lankester. Holaspis Lankester (not of Gray, 1863) 1873, Geol. Mag., 10, pp. 242-243; Claypole, 1892, Quart. Jour. Geol. Soc, London, 48, pp. 546, 551. Cyathaspis (in part) von Alth, 1874, Abh. Geol. Reichanst., Wien, 7, p. 46; Leriche, 1906, Mem. Soc. Geol. Nord, 5, pp. 22-25. Palaeaspis (in part) Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 169; Stensio, 1926, Ark. Zool., 18A, no. 19, pp. 2-3, 12; Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, p. 84; Brotzen, 1933, Palaeobiol., 5, pp. 431-432; White, 1935, Phil. Trans. Roy. Soc. London, (B), 225, p. 437. Poraspis Kiaer, 1930, Skr. Svalbard Ishavet, 33, p. 4 (new name for Holaspis Lankester); 1932, Skr. Svalbard Ishavet, 52, pp. 13-14; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 51-59; Heintz, 1938, Naturwiss., 28, pp. 50-51; Save-Soderbergh, 1941, Zool. Bidr. Uppsala, 20, p. 534; Holm- gren, 1942, K. Svensk. Vetensk.-Akad. Handl., (3), 20, no. 1, p. 12; Wangsjo, 1952, Skr. Norsk Polarinst., 97, p. 562; Flower and Wayland- Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 360-361; Denison, 1953, Fieldiana: Geol., 11, no. 7, pp. 293, 296, 303-304; 1960, Fieldiana: Geol., 11, no. 10, pp. 556, 558, 559, 567; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 296, 314-318, 352-353, 366, 384, 386, 391, 393, 404. Diagnosis. — The rostral region narrows somewhat in front of the orbits, and its anterior border has a slight median convexity. The maxillary brim is broad and is covered with ridges parallel to the anterior edge. The postbranchial lobes of the dorsal shield are long and strongly developed. The posterior edges of the dorsal and ven- tral shields have pronounced, rounded, median lobes. The dentine ridges are fine, 7-8 per mm. Epitega are faintly or not at all indi- cated. The dentine ridge pattern is mostly longitudinal, except that it is sometimes fanned or irregular on the rostrum and on the ante- rior triangle of the ventral shield, and there are generally one or more ridges parallel to the anterior edges of the shields. The pineal macula is distinct. The lateral line pores and canals are large. Discussion. — Poraspis is most closely related to Homalaspidella and Americaspis. It is the only genus of cyathaspids that is well enough known so that the course of its evolutionary changes may be determined. In the Spitsbergen species, which range throughout the Red Bay series, the following evolutionary trends are indicated (Kiaer and Heintz, 1935, pp. 125-126): (1) the size increases; (2) the dentine ridge pattern becomes more longitudinal; (3) the lateral line canals tend to form a more completely united network, and most particularly, the posterior ends of the supraorbital canals unite with the medial dorsal canals. Among the Spitsbergen and western Euro- 404 FIELDIANA: GEOLOGY, VOLUME 13 pean species two groups may be distinguished; the Podolian species are grouped separately because they are inadequately known. Group A: Small, primitive species with the dorsal shield less than 40 mm. long, with the orbits anterior (orbital ratio=.14-.16), and with the shields rather broad (width ratio usually more than .50). Poraspis subtilis Kiaer and Heintz Poraspis subtilis Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14 (nomen nudum) ; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 81-82, fig. 23, pi. 21, fig. 1, pi. 22. Type. — Paleontologisk Museum, Oslo, D 1904, dorsal shield. Occurrence. — Early Devonian (Gedinnian), Plant horizon, Fraen- kelryggen division, Red Bay series, Spitsbergen. Diagnosis. — A small species (length dorsal shield =26 mm.) with a rather narrow shield (width ratio=.52) and with the orbits rather posterior (orbital ratio= .16) . The dorsal shield narrows rather evenly in front of mid-length, and the dentine ridges on the rostrum are fanned. Discussion. — This is the earliest known species of Poraspis. The narrow proportions and the rather posterior orbits suggest that it may be ancestral to the species of Group B. Poraspis brevis Kiaer and Heintz Poraspis brevis Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14, pi. 2 (nomen nudum); Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 78-79, figs. 20-22, pis. 14-15. Type. — Paleontologisk Museum, Oslo, D 304, dorsal shield. Occurrence. — Early Devonian (Gedinnian), Primaeva and Polaris horizons, Fraenkelryggen division Red Bay series, Spitsbergen. Diagnosis. — A small species (length dorsal shield= 25-28 mm.) with a broad shield (width ratio= .56-. 66) and anterior orbits (orbital ratio=.14). Discussion. — P. brevis differs from P. polaris only in its smaller size. Poraspis polaris Kiaer. Figures 98,A, 99,A, 102,A, 137. Poraspis polaris Kiaer, 1930, Skr. Svalbard Ishavet, 33, fig. Sa-b; 1932, Skr. Svalbard Ishavet, 52, p. 14, figs. 1-2, pi. 1; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 59-78, figs, 2, 3, 5, 7-14, pis. 1-8, pi. 9, fig. 2, pi. 10, fig. 1, pis. 11-13, pi. 26, fig. 1, pi. 31, fig. 1, pi. 32, fig. 2, pi. 34, figs. 1-5, pis. 35-37, pi. 38, fig. 1; Save-Soderbergh, 1941, Zool. Bidr. Up- DENISON: THE CYATHASPIDIDAE 405 Fig. 137. Poraspis polaris, type (from Kiaer and Heintz, 1935); dorsal shield, Paleontologisk Museum, Oslo, D 665 (X 3). A, dorsal view; B, lateral view of right side. orb, orbit; pbl, postbranchial lobe; pm, pineal macula. psala, 20, fig. 5; Holmgren, 1942, K. Svensk. Vetensk.-Akad. Handl., (3), 20, no. 1, p. 10, fig. 7. Type. — Paleontologisk Museum, Oslo, D 665, dorsal shield (fig. 137). Occurrence. — Early Devonian (Gedinnian), Primaeva, Polaris, and Anglaspis horizons, Fraenkelryggen division, Red Bay series, Spitsbergen. 406 FIELDIANA: GEOLOGY, VOLUME 13 Diagnosis. — The length of the dorsal shield= 30-40 mm., width ratio=.45-.65. The orbital ratio=.15-.16. The rostrum is faintly constricted anterior to the orbits. The rostral ridge pattern is usually longitudinal but is often fanned in earlier forms. The postbranchial part is long. Discussion. — Kiaer and Heintz (loc. cit.) distinguished a forma angusta and a forma lata of P. polaris, and considered that the great difference in the width ratio might be a secondary sexual character. It is suggested below (p. 415) that many of these proportional dif- ferences may be the result of distortion. Poraspis intermedia Kiaer and Heintz Poraspis intermedia Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14 (nomen nudum); Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 79-81 pis. 16-17. Type. — Paleontologisk Museum, Oslo, D 1308, dorsal shield. Occurrence. — Early Devonian (Gedinnian), Anglaspis horizon, Fraenkelryggen division, Red Bay series, Spitsbergen. Diagnosis. — The length of the dorsal shield= 26-28 mm., and its width ratio= .53-.65. The orbits are anterior (orbital ratio= .14) and the rostrum is not constricted anterior to the orbits. The postbran- chial region is short. Group B: Larger species (dorsal shields longer than 36 mm.) with more slender shields (width ratio=.50 or less), and with the orbits farther back (orbital ratio=.16-.20). The supraorbital canals tend to unite with the medial dorsal canals. Poraspis elongata Kiaer and Heintz Poraspis elongata Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14 (nomen nu- dum); Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 82-84, figs. 24-26, pi. 9, fig. 3, pi. 10, fig. 2, pis. 18-20, pi. 21, fig. 2, pi. 27, fig. 1; Denison, 1953, Fieldiana: Geol., 11, no. 7, p. 300. Type. — Paleontologisk Museum, Oslo, D 141a, dorsal shield. Occurrence. — Early Devonian (Gedinnian), Primaeva and Polaris horizons, Fraenkelryggen division, Red Bay series, Spitsbergen. Diagnosis. — The length of the dorsal shield is 36-40 mm., and its proportions are relatively slender (width ratio= .44-.46). The orbits are posterior (orbital ratio=.18), and the postbranchial part is long. The dentine ridges are mostly longitudinal, but curve slightly antero- laterally on the rostrum. DENISON: THE CYATHASPIDIDAE 407 Discussion. — This species approaches the larger and more elon- gated species of the overlying Ben Nevis division, such as P. rostrata. Poraspis cf. elongata has been reported by White (in Clarke, B. B., 1951, Trans. Woolhope Nat. Field Club, 33, pp. 236-237, fig. 9) from the Early Dittonian (zone of Protopteraspis leathensis), Dinmore Hill, Herefordshire, England. Poraspis rostrata Kiaer and Heintz Poraspis rostrata Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14, pi. 3, figs. 1-2 (nomen nudum); Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 85-88, figs. 27-28, pis. 23-25, 39; Holmgren, 1942, K. Svensk. Vetensk.- Akad. Handl., (3), 20, no. 1, pp. 9, 10. Type. — Paleontologisk Museum, Oslo, D 124, dorsal shield. Occurrence. — Early Devonian (Gedinnian), lower part of Ben Nevis division, Red Bay series, Spitsbergen. Diagnosis. — A relatively large species (length of dorsal shield = 45-51 mm.) with a narrow shield (width ratio=.44-.45), posterior orbits (orbital ratio=.20), and a long postbranchial part. The den- tine ridges are fanned on the rostrum. The medial dorsal canals coalesce with the supraorbital canals, and the latter nearly meet in the midline. Poraspis barroisi (Leriche) Cyathaspis barroisi Leriche, 1906, Ann. Soc. Geol. Nord, 35, pp. 339-340, figs. 1-2; 1906, Mem. Soc. Geol. Nord, 5, pp. 18, 19, 21, 25-26, figs. 6-7, pi. 1, figs. 1-5. Pteraspis barroisi Zych, 1927, Serv. Geol. Pologne, Trav., 2, livr. 1, p. 52. Poraspis barroisi Kiaer, 1930, Skr. Svalbard Ishavet, 33, p. 4; 1932, Skr. Sval- bard Ishavet, 52, p. 14; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 101-104, figs. 36-37. Lectotype. — Selected by Kiaer and Heintz (1935) ; internal mold of dorsal shield in collections of University de Lille, figured by Leriche (1906, pi. 1, fig. 2). Occurrence. — Early Devonian (upper part of Early Gedinnian), Psammites de Lie>in, Pas de Calais, France. Diagnosis. — The dorsal shield has a length of 42-44 mm., and a width ratio of about .50. The orbital ratio is .17-. 18. The rostral ridges are mostly longitudinal, but irregular anteriorly. Poraspis cylindrica Kiaer and Heintz. Figure 106. Poraspis cylindrica Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14, pi. 3, fig. 3 (nomen nudum); Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 408 FIELDIANA: GEOLOGY, VOLUME 13 88-92, figs. 29-30, pi. 26, fig. 2, pi. 27, fig. 2, pi. 29, fig. 1; Stensio, 1958, Traite de Zool., 13, fasc. 1, fig. 179B. Type. — Paleontologisk Museum, Oslo, D 205, dorsal shield, mostly an internal impression (fig. 106). Occurrence. — Early Devonian (Gedinnian), upper part of Ben Nevis division, Red Bay series, Spitsbergen. Diagnosis. — The dorsal shield has a length of 46-53 mm. and a width ratio of .43-. 50. It is slightly constricted at the orbits and the orbital ratio is .17. The rostral ridges are fanned. The medial dor- sal canals meet the supraorbital canals, but the transverse commis- sures do not meet the lateral dorsal canals. Discussion. — This species is similar to P. rostrata and P. barroisi. Poraspis magna Kiaer and Heintz Poraspis magna Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14. (nomen nudum); Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 92-97, figs. 31-32, pis. 28, 29, figs. 2, 4, pi. 38, fig. 2; Save-Soderbergh, 1941, Zool. Bidr. Uppsala, 20, fig. 4. Type. — Paleontologisk Museum, Oslo, D 203, dorsal shield. Occurrence. — Early Devonian (Gedinnian), upper part of Ben Nevis division, Red Bay series, Spitsbergen. Diagnosis. — The length of the dorsal shield is 52-62 mm. and the width ratio is .44-. 50. The orbital ratio is .18, and there is no abrupt constriction at the orbits. The dentine ridge pattern is longitudinal except for a slight fanning on the rostrum. The lateral line pattern is similar to that of P. cylindrica. Discussion. — This species may have been derived from P. cylin- drica, from which it differs in its larger size. Poraspis sericea (Lankester) Holaspis sericeus Lankester, 1873, Geol. Mag., 10, pp. 241, 331-332, pi. 10. Palaeaspis sericea Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 169; Stensio, 1926, Ark. Zool., 18A, no. 19, pp. 1, 7, fig. 6. Palaeaspis (Poraspis) sericea Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, fig. 18. Poraspis sericea Kiaer, 1930, Skr. Svalbard Ishavet, 33, p. 4; 1932, Skr. Sval- bard Ishavet, 52, pp. 13, 14; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 98-101, figs. 6, 33-35; Holmgren, 1942, K. Svensk.Vetensk.- Akad. Handl., (3), 20, no. 1, pp. 9, 10; White, 1950, Bull. Brit. Mus. (Nat. Hist.), Geol., 1, p. 56; Bystrow, 1955, Akad. Nauk SSSR., Mem. Vol. A. S. Berg, pp. 488-491, figs. 16-18; Ball and Dineley, 1961, Bull. Brit. Mus. (Nat. Hist.), Geol., 5, p. 202. DENISON: THE CYATHASPIDIDAE 409 Poraspis (Palaeaspis) sericea White, 1935, Trans. Woolhope Nat. Field Club, 1930-1932, pt. 3, pp. 179-180, figs. 3-4. Type. — British Museum (Natural History), P 4117, dorsal shield. Occurrence. — Early Devonian (Middle Dittonian), Monmouth- shire and Herefordshire, Great Britain. Diagnosis. — The length of the dorsal shield is 70-72 mm., and its width ratio is .44. The orbital ratio is .19. The lateral line sys- tem is very regular and complete, with all the branches united. Discussion. — This is the largest species of Poraspis and may be related to P. magna. Group C: The four species described from Podolia resemble Group B in being rather large and slender, but they have not been adequately characterized. Poraspis also occurs south of the Dniester in equivalent deposits in Bucovina and Bessarabia. Poraspis simplex (Brotzen) Palaeaspis simplex Brotzen, 1933, Palaeobiol., 5, p. 432, pi. 24, fig. 3; 1936, Ark. Zool., 28A, no. 22, p. 6. Poraspis simplex Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 52, 107. Type. — A dorsal shield, not designated, in Geologisch-paleontolo- gisch Institut, Berlin. Occurrence. — Early Devonian (Gedinnian), Czortkow stage, Po- dolia. Diagnosis. — The length of the dorsal shield is 50 mm., and the width ratio is .46. The dentine ridges are mostly longitudinal, irreg- ular on the rostrum. Poraspis sturi (von Alth) Cyathaspis sturi von Alth, 1874, Abh. Geol. Reichanst., Wien, 7, p. 46, pi. 5, figs. 1-2; 1886, Beitr. Pal. Geol. Osterr.-Ung., 5, Heft 3, pp. 64, 72; Siemiradzki, 1906, Beitr. Pal. Geol. Osterr.-Ung., 19, Heft 4, p. 214. "Pteraspis" (Cyathaspis) sturi Zych, 1927, Serv. Geol. Pologne, Trav., 2, livr. 1, p. 52, pi. 2, figs. 1-2. Poraspis sturi Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 52, 104-105, fig. 38; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 386, 401. Palaeaspis sturi Brotzen, 1936, Ark. Zool., 28A, no. 22, p. 6. Type. — Not designated. 410 FIELDIANA: GEOLOGY, VOLUME 13 Occurrence. — Early Devonian (Gedinnian), Czortkow stage and ?01d Red, Stage I, Podolia. Diagnosis. — The dorsal shield is long (length about 60 mm.) and narrow (width ratio=.39). Discussion. — The specimen referred here by Zych (1927, pi. 2, fig. 1) does not belong, according to Kiaer and Heintz. Poraspis siemiradzkii (Zych) Palaeaspis (Poraspis) siemiradzkii Zych, 1931, Fauna Ryb Dewonu i Down- tonu Podola, drawings 37-39, photographs 3-4, 6-7; Brotzen, 1936, Ark. Zool., 28A, no. 22, p. 6. Poraspis siemiradzkii Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, p. 106, fig. 39; Holmgren, 1942, K. Svensk. Vetensk.-Akad. Handl., (3), 20, no. 1, p. 10. Type. — Not designated. Occurrence. — Early Devonian (Gedinnian), Passage beds of Czort- kow stage, Podolia. Diagnosis. — The length of the dorsal shield is 56 mm., and its width ratio is .44. The orbital ratio is about .15. The rostral ridge pattern is irregular. The lateral line pattern is very completely de- veloped. Poraspis pompeckji (Brotzen) Palaeaspis pompeckji Brotzen, 1933, Palaeobiol., 5, pp. 433-435, fig 2, pi. 24, figs. 1, 2, 4; 1936, Ark. Zool., 28A, no. 22, p. 6. Poraspis pompeckji Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 52, 107, fig. 40; Holmgren, 1942, K. Svensk. Vetensk.-Akad. Handl., (3), 20, no. 1, p. 9; Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 386, figs. 179A, 193, 194A-C, 203A, 206, 209. Type. — Not designated, in Geologisch-paleontologisch Institut, Berlin. Occurrence. — Early Devonian (Gedinnian), Old Red, Stage I, Podolia. Diagnosis. — The dorsal shield has a length of 54-58 mm., a width ratio of .45-.48, and is narrowed at the orbits. The orbital index is .17. The dentine ridges are longitudinal, except for fanning on the rostrum. Poraspis sp. Poraspis sp. Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, photographs 8, 9, 12, 13. DENISON: THE CYATHASPIDIDAE 411 Occurrence. — Early Devonian (Gedinnian), Czortkow stage and Old Red I, Podolia. Poraspis sp. White, 1950, Bull. Brit. Mus. (Nat. Hist.), Geol., 1, pp. 56, 74-75; 1961, Bull. Brit. Mus. (Nat. Hist.), Geol., 5, p. 246; Ball and Dineley, 1961, Bull. Brit, Mus. (Nat. Hist.), Geol., 5, pp. 201, 202, 219, 221, 226, 229, 231, table 1. Occurrence. — Early Devonian (Late Downtonian to Middle Dit- tonian), Shropshire, England. Americaspis White and Moy-Thomas Type-species. — Palaeaspis americana Claypole. Glyplaspis Claypole, Nov. 1884, Geol. Mag., (3), 1, p. 520 (nomen nudum). Palaeaspis Claypole (not of Gray, 1870), Dec. 1884, Amer. Nat., 18, p. 1224 1885, Quart. Jour. Geol. Soc. London, 41, p. 62; 1892, Quart. Jour. Geol Soc. London, 48, pp. 549-561; Woodward, 1891, Cat. Fossil Fishes Brit Mus. (Nat. Hist.), 2, p. 169; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 15 Americaspis White and Moy-Thomas, 1941, Ann. Mag. Nat. Hist., (11), 7 p. 397 (new name for Palaeaspis Claypole); Fowler, 1947, Not. Nat no. 187, p. 3; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool. 107, pp. 361-362; Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 312. Diagnosis. — The margin of the rostrum has a small median lobe separated by concavities from the preorbital processes. The orbits are placed anteriorly (orbital ratio=.12-.14). There is no pineal macula. The postbranchial lobes of the dorsal shield are long but weakly developed. The posterior edges of the dorsal and ventral shields have small, rounded median lobes. The general pattern of the dentine ridges (fig. 140) is much as in Poraspis, but it is usually modified by sinuosities or irregularities. The canals and pores of the lateral line system are small; the pattern of the lateral lines (fig. 98, C) resembles that of Poraspis, but both the longitudinal and transverse canals are interrupted, and the commissures between the infraorbital and supraorbital canals are absent. Discussion. — Americaspis is very similar to Poraspis, from which it is distinguished as follows: the median lobe on the anterior margin of the rostrum is somewhat stronger; the orbits are usually more anterior; the dorsal shield does not narrow abruptly in front of the orbits; there is no pineal macula; the posterior edges of the dorsal and ventral shields have only small median lobes; the dentine ridge pat- tern (fig. 140) is characteristically sinuous or irregular, and shows on the rostrum a greater tendency toward radiation or transverseness than in most species of Poraspis; the ridge pattern on the central epitegum is slightly elliptical; the lateral epitega, with their diag- 412 FIELDIANA: GEOLOGY, VOLUME 13 onally arranged ridges, are more distinct; the lateral line system differs as stated in the diagnosis. Americaspis americana (Claypole). Figures 98, C, 138-140, 160. Glyplaspis elliptica Claypole, Nov. 1884, Geol. Mag., (3) 1, p. 520 (nomen nudum). Glyptaspis bitruncata Claypole, Nov. 1884, Geol. Mag., (3) 1, p. 520 (nomen nudum). Palaeaspis americana Claypole, Dec. 1884, Amer. Nat., 18, p. 1224 (nomen nudum); 1885, Quart. Jour. Geol. Soc. London, 41, p. 62, fig. 7; 1892, Quart. Jour. Geol. Soc. London, 48, pp. 549-561, figs. 1-8; Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 170; Eastman, 1907, New York St. Mus., Mem., 10, pp. 29-30; 1917, Proc. U. S. Nat. Mus., 52, pp. 239-240; Bryant, 1926, Proc. Amer. Phil. Soc, 65, pp. 261-264, pi. 1, fig. 1, pi. 3, fig. 1; Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, drawing 13; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 15. Palaeaspis bitruncata Claypole, Dec. 1884, Amer. Nat., 18, p. 1224 (nomen nudum); 1885, Quart. Jour. Geol. Soc. London, 41, pp. 62-63, fig. 8; Bryant, 1926, Proc. Amer. Phil. Soc, 65, pp. 264-265, fig. 1, pi. 2, fig. 3, pi. 3, figs. 2-5; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 15. Americaspis americana Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 361-362; Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 312. 1 879 Type. — Los Angeles County Museum 6393> a dorsal shield lack- ing the superficial layer. Flower and Wayland-Smith (1952, p. 362) concluded that Claypole's type specimens were probably destroyed in 1899 in a fire at Buchtel College, where he taught for a number of years. However, when Claypole moved from Buchtel to Throop Polytechnic Institute in Pasadena, California in 1898, apparently he took some of his collections with him. Throop Polytechnic later became the California Institute of Technology, and in 1957 its ver- tebrate paleontologic collections were transferred to the Los Angeles County Museum. A drawer full of "Palaeaspis" with a few labels, probably in Claypole's handwriting, has come to light recently in the latter institution. One of the original labels reads "counterpart of 'type,' " and specimen 6393 is quite certainly to be identified as the counterpart of the dorsal shield figured by Claypole in 1885 (fig. 7). Both part and counterpart of this specimen were figured in lateral view by Claypole in 1892 (fig. 3), when they were identi- fied as the type. Only the counterpart is known to be preserved, and this shows none of the dentine ridges. Referred specimens. — Los Angeles County Museum 6385, an in- complete dorsal shield impression, probably the one figured by Claypole in 1892 (fig. 4); 6388 and 6405, ventral shield impres- sions. CNHM, dorsal shields UF 938, PF 832, 837, 3269-3270, DENISON: THE CYATHASPIDIDAE 413 3272, 3274, 3276-3277, 3280, 3284-3286, 3305-3307, 3309-3311; ven- tral shields, PF 3273, 3275, 3278-3279, 3281, 3308; dorsal and ventral shields, PF 3268, 3321-3322. Occurrence. — Late Silurian, Landisburg Sandstone member of the Wills Creek formation, Landisburg, Alinda, Andersonburg, and New Bloomfield in Perry County, Pennsylvania. The type material came from the same sandstone (then called the Bloomfield Sandstone) from unknown localities in Perry County. Diagnosis. — The length of the dorsal shield ranges from 50 to 68 mm. There are 5>2 to 8 dentine ridges per millimeter. Discussion. — When Claypole first described his two species of Palaeaspis, the structure of the Heterostraci was poorly known. After a controversy between Lankester, Kunth, Schmidt, von Alth, and others had concluded with the demonstration that pteraspids had ventral as well as dorsal shields, Woodward (1891, p. 170) and Claypole (1892, p. 550) decided that Palaeaspis bitruncata was the ventral shield of P. americana. This conclusion is surely correct, although Bryant (1926, p. 264) has disputed it, claiming that Clay- pole figured P. bitruncata upside down, and that it really was a dorsal shield and a valid species; neither Claypole's nor Bryant's own fig- ures support this contention. I have assembled a considerable number of cyathaspids from the Landisburg Sandstone at nine localities in Perry County, and with the exception of a few small shields,1 I believe that all may be re- ferred to Americaspis americana. This is in spite of the fact that the type and other remnants of Claypole's collection show very few diag- nostic characters, and the precise locality from which they came is not known. The length of the type is 67 mm., while the referred dorsal shields range from 50 to 68 mm. The specimens referred to A. americana show great variation in their proportions; the width ratio of the type is .53, and in the referred dorsal shields it ranges from .36 to .65, so that the extremes appear to be quite distinct spe- cies. Patten (1912, fig. 244B, C) figured a broad form and a slender form from Perry County, and suggested that the difference might be sexual. Bryant (1926, p. 264) referred more slender forms to a separate species, Palaeaspis bitruncata, and broader forms to P. amer- icana. A comparable situation occurs in the Spitsbergen Devonian from which Kiaer and Heintz (1935, pp. 53-55) described angusta and lata forms of species of Poraspis, and concluded that the pro- 1 Claypole (1892, p. 560) mentioned several small shields not over one inch long; these are also probably a distinct species. Fig. 138. Americaspis americana, slab of Landisburg sandstone from near Andersonburg, Pennsylvania, with impressions of numerous shields; CNHM, PF3321 (X 1/4). Fig. 139. Americaspis americana ( X 3/2). A, dorsal shield, CNHM, PF 3286; B, ventral shield, CNHM, PF 3279. Both photographs are of rubber impressions of original natural molds. 414 DENISON: THE CYATHASPIDIDAE 415 portional differences were secondary sexual characters. It is possible, however, that in some of the Spitsbergen examples the proportions have been altered by distortion resulting from tectonic deformation. This is suggested by the slab figured by Kiaer and Heintz (1935, pi. 1), on which the more slender specimens are oriented in one di- rection, and the broader ones more or less at right angles. The same appears to be true on some slabs of Americaspis americana (PF 3321, fig. 138) from a locality west of Andersonburg, Pennsylvania. Meas- urements of Perry County specimens support the theory that the great variation in proportions is the result of deformation; the longer specimens are narrower and have finer dentine ridges, the shorter ones are broader and have coarser ridges, and all intermediates are present. In the referred specimens, the shape (excepting proportions) and dentine ridge pattern agree with Claypole's descriptions and figures of this species. It should be noted that the "lateral plates" described by Claypole (1892, pp. 551-553, fig. 3) are the postbranchial lobes of the dorsal shield, and that the "fins" he illustrates (op. cit., pp. 553- 557, figs. 6-7) are probably scales or branchial plates, as recognized by Jaekel (1894). Branchial plates were not recognized by Claypole, but are possibly represented by three imperfect specimens in his col- lection (Los Angeles County Museum 6386, 6402, 6409). The pattern of the dentine ridges of Americaspis americana (fig. 140) is quite variable, particularly anteriorly. The postrostral field is not distinct, and the combined rostral-postrostral area shows a number of different patterns: the ridges may be fanned, curving an- tero-laterally and laterally toward the orbits; the pattern may be in part transverse, or it may be irregular or partly denticulate with in- distinct pattern. Commonly there are two or three ridges parallel to the rostral border. On the central epitegum, the ridges, though pre- dominantly longitudinal, usually show a slightly elliptical pattern, indicated by the angle at which the ridges of the two sides meet at the midline. There are usually prominent irregularities in the an- terolateral parts of the central epitegum. On the lateral epitega there are about four marginal ridges parallel to the edge; medial to these the ridges behind the orbits are commonly diagonal, becoming longitudinal more posteriorly. On the ventral shield (fig. 139, B) most ridges have a longitudinal or slightly elliptical pattern, com- monly with a fanning toward the antero-lateral corners in the ante- rior triangle; the pattern in this anterior triangle is quite variable, however, and may consist of denticles or of ridges disposed in whorls. 416 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 140. Americaspis americana, restoration of dorsal shield in dorsal view (about X 2). Ridge pattern shown in a generalized way, with characteristic irreg- ularities omitted; ridges actually finer than drawn. orb, orbit. The crests of the ridges are generally somewhat convex, possibly slightly flatter on the ventral shield, and higher and sharper laterally on the dorsal shield. Americaspis claypolei,1 new species. Figure 141. Type. — CNHM, PF 866, an incomplete and crushed dorsal shield (fig. 141). 1 In honor of Edward W. Claypole (1835-1901) who described the first Palae- aspis. Fig. 141. Americaspis claypolei; A, type, incomplete and crushed dorsal shield, CNHM, PF 866, (X 4/3); B, anterior part of type, (X 4). 417 418 FIELDIANA: GEOLOGY, VOLUME 13 Referred specimens. — Incomplete dorsal shields, CNHM, PF 862, 3641, 3643; fragments of dorsal shields, PF 865, New York State Mus. 11469. Fig. 142. Americaspis cf. claypolei, anterior half of dorsal shield, New York State Museum 10678 (X 3). Occurrence. — Late Silurian, Longwood shale, Shin Hollow, along Erie Railroad tracks about 1% miles south-southeast of Graham Station, Orange County, New York. Diagnosis. — The length of the dorsal shield is about 40 mm., and it is of rather narrow proportions. The dentine ridges are fine, about 63^-8 per mm., with the coarsest ones on the rostrum and over the orbits, and the narrowest ones near the lateral margins. The ridges are slightly convex centrally, sharp-crested laterally, and broad and flat-topped anteriorly on the rostrum. Discussion. — The incompleteness and poor preservation of most specimens makes it difficult to characterize this species and to deter- mine its relationships. In its known characters it agrees well with Americaspis americana, except in its smaller size and in its broad, flat ridges on the anterior part of the rostrum. It has a small median rostral lobe and strong preorbital processes. The postbranchial lobes are long and distinctly marked anteriorly, but are rather weakly de- DENISON: THE CYATHASPIDIDAE 419 veloped. The orbits are placed anteriorly, and a pineal macula ap- pears to be absent. The dentine ridges of the central epitegum are longitudinal or somewhat sinuous. On the lateral epitegum a few Fig. 143. Americaspis sp., dorsal shield and anterior half of ventral shield, CNHM, PF834 (X 1). (4-7) ridges are parallel to the edge, while behind the orbits the others are diagonal anteriorly and longitudinal posteriorly. The post- rostral area is probably not distinct; the ridges of the postrostral area are continuous with those of the central epitegum, and perhaps in some cases with those of the rostral epitegum. The ridge pattern of the rostral epitegum is extremely variable; usually there are 3-5 broad, flat ridges parallel to the anterior edge; behind these the ridges may be longitudinal (PF 862, 866), or denticulate anteriorly and longitudinal posteriorly (PF865a), or all denticulate (PF3641), or in one case mostly transverse (PF 865b). New York State Mus. 10678 (fig. 142), the anterior part of a dor- sal shield from the same locality, is doubtfully referred to this species. On the rostral epitegum there are almost no ridges parallel to the anterior edge, and most of the rostral ridges radiate sinuously. This specimen has been identified as Cyathaspis "wardelli" by Flower and Wayland-Smith (1952, pi. 2, fig. 4) and by Kilfoyle (1959, p. 108). Americaspis sp. CNHM, PF833 and 834, from the Late Silurian, Landisburg sandstone, % mile north-northwest of Landisburg, Perry County, Pennsylvania, were originally one block of sandstone containing six 420 FIELDIANA: GEOLOGY, VOLUME 13 shields and partial shields, all similarly oriented. PF 834 (fig. 143) contains a dorsal shield and the anterior part of a ventral shield pre- pared to show their outer surfaces. The length of the dorsal shield is 42 mm., its width ratio is .78, and there are 5-6 dentine ridges per millimeter. It is thus smaller, broader, and coarser-ridged than A. americana, with which it is associated. The proportions may not be significant, but the dentine ridges differ from those of A. americana in being divided into short lengths, and in possessing flat crests, ex- cept for those along the lateral margins. PF 3283 from the Landis- burg sandstone at Landisburg village may belong to the same species. These specimens agree with A. claypolei in size, but differ from it in their broader proportions and in their short, flat-crested ridges. Three other specimens (CNHM, PF 831, 3271, and UF 937) from the Landisburg sandstone of Perry County, Pennsylvania, resemble A. americana except in their smaller size; the length of their dorsal shields is 42-44 mm. Their width ratio is .62-.65, and they have 6-7 dentine ridges per millimeter. They lack the broad, flat-topped anterior rostral ridges that characterize A. claypolei. CNHM, PF 3282, a ventral shield from the Landisburg sandstone of Landisburg village, agrees with A. americana in all respects except that its dentine ridges are coarse (43^-5 per mm.) and flat-topped. Americaspis appears to be present also in the Late Silurian, High Falls formation, exposed in outcrops along the Delaware Water Gap- Millbrook road, opposite Shawnee Island, Warren County, New Jer- sey. Three poorly preserved, obviously distorted dorsal shields (CNHM, PF 858, 860, and Princeton 18005) are referred to this genus, but are not determined specifically. Homalaspidella Strand Type-species. — Homalaspis nitida Kiaer. Homalaspis Kiaer (not of Reinhard 1860, nor of Milne Edwards 1863), 1932, Skr. Svalbard Ishavet, 52, p. 14. Homalaspidella Strand, 1934, Folia Zool. Hydrobiol., 5, p. 327 (new name for Homalaspis Kiaer); White and Moy-Thomas, 1940, Ann. Mag. Nat. Hist., (11), 6, p. 101; Strand, 1942, Folia Zool. Hydrobiol., 11, p. 384; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 361; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 314-318, 353, 369, 386, 404; Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 123, 140. Homaspis Kiaer and Heintz (not of Foerste 1868, nor of Skuse 1888), 1935, Skr. Svalbard Ishavet, 40, pp. 127-128 (new name for Homalaspis Kiaer); Save-Soderbergh, 1941, Zool. Bidr. Uppsala, 20, pp. 533-534. DENISON: THE CYATHASPIDIDAE 421 Homolaspis (error) White, 1935, Phil. Trans. Roy. Soc. London, (B), 225, p. 437; Holmgren, 1942, K. Svensk. Vetensk.-Akad. Handl., (3), 20, no. 1, p. 10. Diagnosis. — Horn alas pidella includes small Poraspidinae with shields moderately narrow (width ratio= .43-54), rather uniform in breadth, and little vaulted. The rostrum is broadly rounded and lacks any median lobe (fig. 146,B). The preorbital processes and orbital notches are slightly developed (fig. 146,A), and the preorbital length is short (orbital ratio=.10-14). The postbranchial lobes of the dorsal shield are short. Dentine ridges are of moderate width (5-73^ per mm.) and are mostly flat-topped. Epitega are little or not at all indicated, and the ridge pattern (figs. 145, 147) is essen- tially longitudinal, except for a belt of ridges that runs parallel to the anterior and lateral edges, and for fanned or irregular ridges an- teriorly on the ventral shield. The dorsal transverse commissures of the lateral line system are poorly developed. Discussion. — Homalaspidella is quite similar to Poraspis. The latter may be distinguished in its dorsal shield by the longer pre- orbital region (orbital ratio=.14-.20), stronger development of the preorbital processes and orbital notches, narrowing of the dorsal shield in front of the orbits, and longer postbranchial lobes. The simple, essentially longitudinal ridge pattern of Homalaspidella, and the absence of indications of epitega is not typical |for Poraspis, though it is approached by some specimens. Homalaspidella nitida (Kiaer). Figures 98,D, 99,C, 144. Homalaspis nitida Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 14, text fig. 6, pl. 4, fig. 1. Homaspis nitidus Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 128- 132, text-figs. 55-57, pis. 30, 31, fig. 2, pl. 32, fig. 1, pl. 33, fig. 1, pl. 34, fig. 6, pl. 40; Holmgren, 1942, K. Svensk. Vetensk.-Akad. Handl., (3), 20, no. 1, p. 10. Homaspis nitida Save-Soderbergh, 1941, Zool. Bidr. Uppsala, 20, fig. 6 (p. 533). Homalaspidella nitida Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 361; Stensio, 1958, Traite de Zool., 13, fasc. 1, figs. 180B, 203B; Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 123-126. Type. — Paleontologisk Museum, Oslo, D 156, dorsal shield (fig. 144). Occurrence. — Early Devonian (Gedinnian), Ben Nevis division, Red Bay series, Spitsbergen. Diagnosis. — The length of the dorsal shield is 20-26 mm., and the orbital ratio is .14. The rostral margin is broadly rounded, and 422 FIELDIANA: GEOLOGY, VOLUME 13 the posterior edge of the dorsal shield is produced into a rounded median lobe. There are about 7 dentine ridges per millimeter, and a broad, flat pineal macula is present. Fig. 144. Homalaspidella nitida, type (from Kiaer, 1932); dorsal shield, Paleon- tologisk Museum, Oslo, D 156 (X 3). orb, orbit. Discussion. — The diagnosis includes var. robusta (Kiaer and Heintz, 1935, p. 132), which differs from typical H. nitida only in its slightly larger size (length of dorsal shield= 23-26 mm.) and broader proportions (width ratio= .54 as compared to .42-.52 in typ- ical forms). Kiaer and Heintz (op. cit., p. 129) indicated that there are only 5 dentine ridges per millimeter, but measurements of their figures and a cast of the type show about 7. Their reconstruction of the dorsal lateral line pattern (fig. 98,D) shows that the dorsal trans- verse commissures were reduced, but that the longitudinal canals were continuous. In the ventral shield (fig. 99,C), the postoral canals continue into medial ventral canals, the latter appearing, anteriorly at least, as longitudinal canals rather than as transverse commissures. This species is not far removed from Poraspis, with which it is asso- ciated in the Spitsbergen deposits. Homalaspidella borealis Denison. Figures 145-147. Homalaspidella borealis Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 123-127, figs. 70-73. DENISON: THE CYATHASPIDIDAE 423 Fig. 145. Homalaspidella borealis, type (from Denison, 1963); dorsal shield incomplete posteriorly, Princeton 17101 (X 4). orb, orbit. Type. — Princeton 17101, a dorsal shield, incomplete posteriorly (figs. 145, 146,B). Referred specimens: Princeton 17092, 17102, 17378-17383. Occurrence. — Probably Early Devonian (Early Downtonian) lime- stones and graptolitic shales,1 Beaver River, southeastern Yukon. Diagnosis. — The length of the dorsal shield is 28-30 mm., and the orbital ratio is .10-. 12. The rostral margin is nearly transverse, and the posterior edge of the shield is gently convex with no median lobe. There are 5.0-6.2 dentine ridges per millimeter. A pineal macula is lacking. 1 Tentatively assigned to the Middle Ludlovian in Denison (1963); see below pp. 450-451 424 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 146. Homalaspidella borealis, dorsal shield (X 3). A, lateral view of right side, Princeton 17379; B, rostral view of type, Princeton 17101. bro, branchial opening; orb, orbit; ro, rostrum. Discussion. — Though older than H. nitida, this species appears to be more advanced in the simplification of the dorsal shield out- lines and of the dentine ridge pattern. The orbital notches and preorbital processes (fig. 146,A) are so slightly developed that it is difficult at first to recognize a dorsal shield by its shape. In addi- tion, the absence of epitega and of a pineal macula, and the presence of longitudinal ridges covering most of the shield (fig. 145), make recognition even more difficult. The transverse commissures and medial longitudinal canals of the dorsal sensory line system are di- vided into short lengths. The ventral shield (fig. 147) is 25-29 mm. long, and has a width ratio of .58. Its ridge pattern resembles that of the dorsal shield: there are a few anterior transverse ridges that con- tinue around to the lateral margins, and the central ridges are longi- tudinal, becoming irregular behind the anterior transverse ridges. Ariaspis Denison Type-species. — Ariaspis ornata Denison. Ariaspis Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 120-123, 140. DENISON: THE CYATHASPIDIDAE 425 Diagnosis. — The dorsal shield is relatively broad and weakly vaulted. There is no median process on the rostral margin, but the Fig. 147. Homalaspidella borealis (from Denison, 1963) ; ventral shield, Prince- ton 17092 (X 4). preorbital processes are strongly developed. Narrow, downwardly directed lateral laminae extend from the orbits to the branchial openings and are separated from the rest of the shield by sharp an- gulations. There are no postbranchial lobes, and the postero-lateral corners of the shield are small points. The posterior edge of the dor- sal shield is nearly transverse except for a small, median, projecting scale that is incorporated. No epitega are distinguishable. Ariaspis ornata Denison. Figures 148-149. Ariaspis ornata Denison, 1963, Fieldiana: Geol., 14, no. 7, pp. 120-123, figs. 68-69. 426 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 148. Ariaspis ornata, type (from Denison, 1963); dorsal shield, Prince- ton 17103 (X 6). ms, median scale; or, orbit; pm, pineal macula. Type. — Princeton 17103, a complete dorsal shield (figs. 148-149) . Occurrence. — Probably Early Devonian (Early Downtonian) lime- stones and graptolitic shales,1 Beaver River, southeastern Yukon. Diagnosis. — The length of the dorsal shield is 21.5 mm. The width ratio is .72. The dentine ridges are fine, 7.5-8 per mm., and have sharply or roundly convex crests. 1 Tentatively assigned to the Middle Ludlovian in Denison (1963); see (p. 450). DENISON: THE CYATHASPIDIDAE 427 m«h|^^^^ pm ,*ii'Vs^ 1 / ^fy^^ Fig. 149. Ariaspis ornata, type; dorsal shield, Princeton 17103 (x 4). A, lateral view of right side; B, rostral view. Ibr, lateral lamina; orb, orbit; pip, postero-lateral point; pm, pineal macula; pop, preorbital process; ro, rostrum. Discussion. — The loss of distinct epitega and the simplification of the ridge pattern are considered to be advanced characters, com- parable to the situation in some other Poraspidinae, such as Homala- spidella and Allocryptaspis. The overall ridge pattern (fig. 148) is elliptical, but is varied by an anterior band of transverse ridges, antero-lateral patches of short, irregular ridges, and a prominent, oval, pineal macula. In addition, there is a break in the continuity of ridges, making it appear as if a pair of large dorso-lateral scales had been fused to the posterior edge. Postero-medially a dorsal ridge scale (fig. 148, ms) has been incorporated in and attached to the shield, forming a feature otherwise unknown in Cyathaspididae, though characteristic of Pteraspididae where it is known as the dorsal spine. The lateral laminae of Ariaspis (fig. 149, Ibr) are narrow, ven- tro-laterally directed bands, probably in no way comparable to the lateral laminae of Allocryptaspis, in which they appear to be formed by the fusion of the branchial plates to the dorsal shield. The re- duction of the postbranchial lobes and the formation of postero- lateral pointed projections on the dorsal shield are comparable to the situation in Anglaspis. 428 FIELDIANA: GEOLOGY, VOLUME 13 Anglaspis Jaekel Type-species. — Cyathaspis macculloughi Woodward. Cyaihaspis (in part) Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, p. 172. Anglaspis Jaekel, 1927, Zeits. Ges. Anat., Abt. 3, Ergebn. Anat. Entwickl., 27, p. 877; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 20; Holmgren, 1942, K. Svensk. Vetensk.-Akad. Handl., (3), 20, no. 1, p. 10; Flower and Way- land-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 363-364; Wangsjo, 1952, Skr. Norsk Polarinst., 97, p. 562; Watson, 1954, Phil. Trans. Roy. Soc. London, (B), 238, pp. 9-13; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 314-316, 384-385, 389-391. Fraenkelaspis Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 314-316, 384, 391. Diagnosis. — Anglaspis includes small Poraspidinae in which the dentine ridges are coarse (23/2-5 per mm. in the central parts of the shield), and the epitega are quite distinct. The dorsal shield is little arched, but is relatively broad (width ratio= .59-67) ; it is constricted at the branchial openings, in front of which the lateral epitega form brims; its postbranchial lobes are very small or absent. The pineal macula is prominent, and the maxillary brim has fine ridges parallel to the margin. The ventral shield is strongly arched posteriorly. The posterior edge of both shields is bluntly pointed. There are no separate postoral plates. The branchial plates have a sharp lateral angulation. The recticular layer is poorly developed, and very large chambers occur in rows in the cancellous layer, one row under each dentine ridge. Discussion. — Certain species of Anglaspis have recently been re- moved by Stensio (1958) to a new genus, Fraenkelaspis, which he distinguished by two features. Firstly, the ventral shield was sup- posed to have distinct anterior and posterior central hypotega. The ventral shield that Stensio figures (fig. 150,B in this paper) is the type specimen of Anglaspis insignis, and could be interpreted as having the two central hypotega. However, two other ventral shields on the same specimen (Paleontologisk Museum, Oslo, D 186) do not show this division and have a ridge pattern similar to that of typical Anglaspis. This case of apparent subdivision of the ventral shield is really nothing but an example of the great individual variability of dentine ridge patterns (cf. Wills, 1935, p. 431, pi. 2). Secondly, Stensio (1958, pp. 385-386) claims that Fraenkelaspis has three vagal visceral arches in contact with the lateral wall of the occipital region, while Anglaspis has only two. This interpretation is based, of course, only on the often vague impressions of internal structures on the exoskeleton. Stensio's reconstruction of Fraenkelaspis (op. DENISON: THE CYATHASPIDIDAE 429 cit., fig. 205, B), based on a figure of specimen D 193 in the Paleon- tologisk Museum in Oslo (Kiaer, 1932, pi. 6, fig. 1), shows important differences from a cast of this specimen in Chicago Natural History Museum (PF 1133). The eighth gill pouch ("kg") is not appar- ent on the cast, nor is the posterior limit of the visceral endoskeleton ("1. vise. p."). The seventh gill pouch ("k8") is shorter than the fifth and sixth so that the arrangement of the gill pouches is closer to that shown by Stensio in typical Anglaspis (op. cit., fig. 205, A). Stensio states that his reconstruction of the latter was based on fig- ures of Wills' (1935). However, in his figures and text Wills shows the presence of only seven gills (k2-k8) ; from his specimens it would be impossible to show definitely that any of them were in contact with the occipital region of the endocranium. There appears to be no justification for the subdivision of Anglaspis, and so Fraenkelaspis should be discarded. Anglaspis is one of the most distinctive of the Poraspidinae largely because the coarseness of its dentine ridges emphasizes their pattern, and causes the epitega to appear more distinct than in other members of the subfamily. The rostral epitegum and postrostral field are not differentiated. The ridges radiate on the rostro-postrostral area, are nearly longitudinal on the central epitegum, and diagonal on the lateral epitega. On the ventral shield, the ridges typically radiate in the anterior region and are nearly longitudinal in the central area; in the lateral areas they are parallel to the margin. Anglaspis macculloughi (Woodward). Figures 98,B, 102,D. Cyathaspis macculloughi Woodward, 1891, Cat. Fossil Fishes Brit. Mus. (Nat. Hist.), 2, pp. 172-173, pi. 9, fig. 4. Anglaspis macculloghi Jaekel, 1927, Zeits. Ges. Anat., Abt. 3, Ergebn. Anat. Entwickl., 27, p. 877, fig. 26. Anglaspis macculloughi Wills, 1935, Trans. Roy. Soc. Edinburgh, 58, pp. 429- 435, figs. 1-3, pi. 1, figs. 2-12, pi. 2, pi. 3, figs. 3-7, pi. 5, fig. 8, pi. 7, figs. 1-4; Westoll, 1945, Trans. Roy. Soc. Edinburgh, 61, p. 347, fig. 4A; White, 1946, Quart. Jour. Geol. Soc. London, 101, pp. 210, 211, 213; Wat- son, 1954, Phil. Trans. Roy. Soc. London, (B), 238, fig. 4; Bystrow, 1955, Akad. Nauk. SSSR., Mem. Vol. A. S. Berg, pp. 485-487, figs. 13-15; Stensio, 1958, Traite de Zool., 13, fasc. 1, figs. 168A-B, 205; White, 1961, Bull. Brit. Mus. (Nat. Hist.), Geol., 5, p. 246; Ball and Dineley, 1961, Bull. Brit. Mus. (Nat. Hist.), Geol., 5, p. 223; Heintz, 1962, Coll. Internat., Centre Nat. Rech. Sci., no. 104, fig. 8D. Type. — British Museum (Natural History), P 4797, imperfect dorsal shield. 430 FIELDIANA: GEOLOGY, VOLUME 13 w--1 Fig. 150. Anglaspis insignis (from Kiaer, 1932); Paleontologisk Museum, Oslo, D 186 (X 3). A, dorsal shield; B, ventral shield, designated as type. orb, orbit. Occurrence. — Early Devonian (Late Downtonian), zone of Tra- quairaspis symondsi, Shropshire, Herefordshire, Brecknockshire, and Pembrokeshire, Great Britain. This species is doubtfully reported by White (1950, p. 56) from the Early Dittonian (zone of Protopter- aspis leathensis) of Worcestershire, England. Diagnosis. — The length of the dorsal shield is 24-30 mm. Cen- trally there are 23^-33^ dentine ridges per millimeter, and their crests are sharp and high. The orbital ratio is .16-18, and the pineal ratio is .23-.26. Anglaspis insignis Wills. Figures 99,B, 150. Anglaspis insignis Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 20, pi. 7 (nomen nudum); Wills, 1935, Trans. Roy. Soc. Edinburgh, 58, pp. 429, 432-433; Denison, 1953, Fieldiana: Geol. 11, no. 7, p. 292. Anglaspis insignis var. brevis Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 20, pi. 6, fig. 1 (nomen nudum). Fraenkelaspis insignis Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 315, figs. 168C-D, 180A. Fraenkelaspis brevis Stensio, 1958, Traite de Zool., 13, fasc. 1, fig. 205B. Type. — Paleontologisk Museum, Oslo, D 186, ventral shield (fig. 150,B). DENISON: THE CYATHASPIDIDAE 431 Occurrence. — Early Devonian (Gedinnian), Primaeva, Polaris, Anglaspis, and Red Horizons, Fraenkelryggen division, Red Bay series, Spitsbergen. Diagnosis. — The length of the dorsal shield is 20-23 mm. Cen- trally there are 33^-5 ridges per millimeter; they are usually flat or slightly convex, but are sharp-crested anteriorly on the rostrum and laterally on the dorsal shield. The orbital ratio is .17 and the pineal ratio is .24. Discussion. — Unfortunately, according to the International Code of Zoological Nomenclature (article 13), the name Anglaspis insignis as proposed by Kiaer was a nomen nudum. Wills (1935, p. 429) vali- dated the name when he gave the first "statement that purports to give characters differentiating the taxon." Anglaspis heintzi Heintz. Figures 90, 94,A. Anglaspis heintzi Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 20, fig. 11 (nomen nudum); Heintz, 1933, Zeits. Geschiebeforsch., 9, p. 130, fig. 4; 1962, Coll. Internal:., Centre Nat. Rech. Sci., no. 104, p. 24, fig. 7; Wills, 1935, Trans. Roy. Soc. Edinburgh, 58, pp. 429, 430, 435; Tarlo, 1962, Acta Paleont. Polonica, 7, fig. 3. Poraspis heintzi Moy-Thomas, 1939, Palaeozoic Fishes, fig. 2A. Fraenkelaspis heintzi Stensiti, 1958, Traite de Zool., 13, fasc. 1, fig. 166A. Type. — Paleontologisk Museum, Oslo, specimen not designated. Occurrence. — Early Devonian (Gedinnian), upper part of Fraen- kelryggen division, Red Bay series, Spitsbergen. Diagnosis. — The known characters are not sufficient to differen- tiate this species. Discussion. — Anglaspis heintzi, as originally listed by Kiaer, was a nomen nudum. The recent description and figure of the mouth parts by Heintz (1962, p. 24, fig. 7) may be considered as giving "characters differentiating the taxon"; therefore Heintz unwittingly validated the name of a species proposed in his honor. Anglaspis elongata and A. platostriata, from the Early Devonian, Red Bay series, Spitsbergen, are nomina nuda, merely listed by Kiaer (1932, p. 20). Anglaspis expatriata, new species. Figure 151. Type. — National Museum of Canada 10038, a nearly complete dorsal shield exposed on the inner side (fig. 151, A). 432 FIELDIANA: GEOLOGY, VOLUME 13 Fig. 151. Anglaspis expatriata (X 2). A, rubber impression of type, inner side of dorsal shield, National Museum of Canada 10038; B, ventral shield and part of dorsal shield, inner sides, CNHM, PF 3666. ds, dorsal shield; k-.-^, impressions of gill pouches; pfo, impression of fossa for pineal organ; sec, impressions of semicircular canals; vs, ventral shield. Referred specimens. — CNHM, PF 3665, a flattened dorsal shield and scales, in counterpart; PF 3666 (fig. 151, B), incomplete dorsal and ventral shields, inner sides; National Museum of Canada, 10039-40, CNHM, PF 3667, incomplete ventral shields; CNHM, PF 3668-69, 3671, shield fragments; PF 3670, scales. Occurrence. — Early Devonian, top of Mt. Sekwi, beside the Keele River, Northwest Territories, Canada (68° 28' N.; 128° 40' W.) (California Standard Company locality Z 29-61, 4474'). Diagnosis. — The length of the dorsal shield is about 24 mm., and that of the ventral shield is 20-23 mm. On both shields there are centrally 3-33^ dentine ridges per millimeter, and their crests are sharp and moderately high. The orbits and pineal organ are rather anterior in position; the orbital ratio is .14-. 15, and the pineal ratio is .20-.21. Discussion. — Anglaspis expatriata is very similar to the British species, A. macculloughi. From the latter it differs in its more an- terior orbits and pineal organ, and usually in its smaller size. A. ex- patriata may have had rounded postero-lateral corners on the dorsal shield; in A. macculloughi these corners form pointed spurs. The DENISON: THE CYATHASPIDIDAE 433 maxillary brim of A. expatriata is incompletely preserved in PF 3666; it is about 0.7 mm. wide and has approximately 8 fine, even ridges parallel to the anterior margin; in A. macculloughi the corresponding ridges appear to be irregular (Wills, 1935, pi. 1, fig. 6, mb). In A. ex- patriata (Nat. Mus. Canada 10038) on the inner or ventral surface of the dorsal shield, at the margins where it overlies the branchial openings, there are similar bands, 0.6-0.7 mm. wide, with 6-7 nar- row, even dentine ridges; a similar band in A. macculloughi (CNHM PF 1390) is shorter, broader (1.2 mm.), and has about 12 ridges of irregular width. A. insignis differs from A. expatriata in its somewhat smaller size, in its more posterior orbits and pineal organ, and in its finer dentine ridges, most of which have flat or only slightly convex crests. The dentine ridge pattern is not well displayed in any of the specimens of A. expatriata. On the dorsal shield the ridges are longi- tudinal posteriorly, but otherwise unknown. On the ventral shield they are longitudinal on the posterior half, and fanned in the ante- rior third, with two ridges parallel to the antero-lateral and probably anterior edges. PF 3667 shows on the ventral shield a pair of lateral bands of fine ridges (about 6 per mm.) similar to those of A. mac- culloughi. Anglaspis sp. Anglaspis n. sp. Thorsteinsson, 1958, Geol. Surv. Canada, Mem. 294, p. 77. Occurrence. — Early Devonian, Snowblind Bay formation, Read Bay, Cornwallis Island, Canada. Anglaspis sp. indet. Gross, 1961, Acta Zool., 42, pp. 76, 90-96, 143-144, figs. 5,G-K, 6-7. Occurrence. — Downtonian, glacial erratics of red, quartzose Bey- richienkalk, northern Germany. Anglaspis sp. Ball and Dineley, 1961, Bull. Brit. Mus. (Nat. Hist.), Geol., 5, pp. 202-203, 221, 222, 225, 228, 230, 235. Occurrence. — Early Devonian (Late Downtonian to Middle Dit- tonian), England. Anglaspis sp. Karatajute-Talimaa, 1962, Lietuvos TSR ma Geologijos ir Geografijos Institutas Moksliniai Pranesimai, Geol.-Geogr., 14, I sas., pp. 46, 48, 53, 58. 434 FIELDIANA: GEOLOGY, VOLUME 13 Occurrence. — Late Downtonian, boreholes at Krekenava, Staci- unai and Papilvis in Lithuania. cf. Anglaspis Thorsteinsson and Tozer, 1963, Geol. Surv. Canada, Mem. 320, p. 122. Occurrence. — Silurian or Devonian, lower part of Peel Sound for- mation, 5 miles west of Cunningham Inlet, Somerset Island, Canada. Allocryptaspis Whitley Type-species. — Cryptaspis ellipticus Bryant. Cryptaspis Bryant (not of Pascoe 1872), 1934, Proc. Amer. Phil. Soc, 73, p. 154; 1935, Proc. Amer. Phil. Soc, 75, pp. 112-114; Kiaer and Heintz, 1935, Skr. Svalbard Ishavet, 40, pp. 132-133. Allocryptaspis Whitley, May 1940, Australian Nat., 10, p. 243 (new name for Cryptaspis Bryant); Denison, 1953, Fieldiana: Geol., 11, no. 7, p. 294; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 314-316. Bryantaspis White and Moy-Thomas, June 1940, Ann. Mag. Nat. Hist., (11), 5, p. 507 (new name for Cryptaspis Bryant); Romer, 1945, Vertebrate Paleontology, p. 574. Cryptaspidisca Strand, 1942, Folia Zool. Hydrobiol., 11, p. 384 (new name for Cryptaspis Bryant). Diagnosis. — Allocryptaspis includes large Poraspidinae in which the branchial plates are fused to the dorsal shield forming ventro- lateral laminae. The branchial openings lie between the dorsal and ventral shields; both shields are notched for these openings (fig. 152,B). The small orbits notch the lateral laminae and are rela- tively far forward (orbital ratio=. 11-14). The maxillary brim (fig. 92, mxb) is small and is ornamented with ridges at right angles to the rostral edge. The postbranchial lobes are short and weakly developed. Both dorsal and ventral shields have a median lobe on the posterior edge. Postoral plates (fig. 95,A-B) are present. The lateral line pattern (figs. 98,E, 99,D) is similar to that of Poraspis except that the commissures between the supraorbital and infra- orbital lines are absent. Dentine ridges are coarse, 2.3-5 per mm.; their pattern is essentially longitudinal except for fanning in the rostral region and in the anterior part of the ventral shield. No epitega are apparent. Fig. 152. Restoration of shield of Allocryptaspis laticostata (from Denison, 1960); approximately natural size. A, dorsal view; B, lateral view of left side; C, ventral view. 435 436 FIELDIANA: GEOLOGY, VOLUME 13 Discussion. — In an early paper (Denison, 1953, p. 296) I consid- ered the lateral laminae of Allocryptaspis to be differentiated parts of the dorsal shield. In a later paper (Denison, 1960, p. 558) I con- cluded that these laminae represented the branchial plates that had fused to the dorsal shield. The arguments for the latter hypothesis are as follows: (1) No branchial plates have been found. Those so identified by me (1953, p. 299) are now interpreted as fragments of lateral laminae. The failure to discover branchial plates would ordi- narily not be significant, but is in the collection of Allocryptaspis laticostata from the Holland Quarry shale, for the abundant material includes specimens of even the smallest oral and lateral plates. (2) The branchial openings notch both the dorsal and ventral shields; therefore, the openings were not bounded by separate branchial plates. (3) Except that they are entirely anterior to the bran- chial openings, the lateral laminae have the same morphological re- lations as the branchial plates of typical cyathaspids. Allocryptaspis is the latest, and with the exception of Ctenaspis and Listraspis, the most specialized member of the Cyathaspididae. Its outstanding specializations are large size, and the fusion of the branchial plates to the dorsal shield with the related modification of the boundaries of the branchial openings. Other advanced char- acters include: the absence of distinct epitega; the essentially longi- tudinal pattern of the dentine ridges; the rather complete lateral line pattern; the presence of distinct postoral plates and numerous lat- eral plates; the thick shields with wide dentine ridges and large chambers arranged in rows in the cancellous layer. Allocryptaspis may have been derived from Americaspis or Poraspis. Allocryptaspis elliptica (Bryant) Cryptaspis ellipticus Bryant 1934, Proc. Amer. Phil. Soc, 73, pp. 154-157, fig. 7, pi. 26, fig. 1 (not fig. 8 and pi. 8, fig. 4); 1935, Proc. Amer. Phil. Soc, 75, pp. 114-118, fig.l, pis. 1-5, pi. 6, fig. 1. Allocryptaspis ellipticus Denison, 1953, Fieldiana: Geol., 11, no. 7, pp. 295- 304, fig. 65A; 1960, Fieldiana: Geol. 11, no. 10, pp. 556-559. Type. — Princeton 13752, a dorsal shield. Occurrence. — Early Devonian, Beartooth Butte formation, Bear- tooth Butte, Park County, Wyoming. Diagnosis. — The length of the dorsal shield is 78-91 mm., and that of the ventral shield is 79-83 mm. The shields are narrower than in A. laticostata and A. flabelliformis (orbital width ratio=.35). There are 4-5 dentine ridges per millimeter, probably smooth-crested. DENISON: THE CYATHASPIDIDAE 437 Fig. 153. Allocryptaspis utahensis, ventral view of anterior part of type (from Denison, 1953); dorsal shield, CNHM, PF 737 (X 1). brn, position of notch bounding branchial opening; 11, lateral lamina; or, orbit. Allocryptaspis flabelliformis (Bryant) Cryptaspis flabelliformis Bryant 1935, Proc. Amer. Phil. Soc, 75, pp. 118-119, 'pi. 7. Allocryptaspis flabelliformis Denison 1953, Fieldiana: Geol., 11, no. 7, pp. 296- 299, 304; 1960, Fieldiana: Geol. 11, no. 10, p. 556. Type. — Princeton 13888, a ventral shield. Occurrence. — Early Devonian, Beartooth Butte formation, Bear- tooth Butte, Park County, Wyoming. Diagnosis. — The length of the ventral shield is about 100 mm., and it is relatively broad. There are about 4 dentine ridges per mm. Discussion. — This is a poorly characterized species, distinguished by its large size. Allocryptaspis laticostata Denison. Figures 92, 95, 96, 98,E, 99,D, 102,C, 104, 152, 159,B. Allocryptaspis laticostatus Denison, 1960, Fieldiana: Geol., 11, no. 10, pp. 555- 567, figs. 117-123. Type.— CNHM, PF 1701, a dorsal shield (Denison, 1960, fig. 117,A). 438 FIELDIANA: GEOLOGY, VOLUME 13 Occurrence. — Early Devonian, Holland Quarry shale, Holland Quarry, Monclova Township, Lucas County, Ohio. Diagnosis. — The length of the dorsal shield is 81-92 mm., and that of the ventral shield is 81-87 mm. The shields are relatively broader than in other species (orbital width ratio=.41). The den- tine ridges are very coarse, 2.3-2.8 per mm., generally flat-topped and finely tuberculate. Allocryptaspis utahensis Denison. Figure 153. Allocryptaspis utahensis Denison 1953, Fieldiana: Geol., 11, no. 7, pp. 294-304, figs. 61-64, 65,B; 1960, Fieldiana: Geol., 11, no. 10, pp. 556-559, 564. Type. — CNHM, PF 737, associated dorsal and ventral shields and scales (fig. 153). Occurrence. — Early Devonian, Water Canyon formation, Cotton- wood, Green, and Blacksmith Fork Canyons in the Bear River Range, and Dry Lake on the eastern edge of the Wellsville Range, all in Cache County, Utah. Diagnosis. — The length of the dorsal shield is about 110 mm., and that of the ventral shield is about 91 mm. The shields are more slender than in the other species (orbital width ratio=.32). There are 3.2-3.6 dentine ridges per millimeter, with their crests smooth and slightly convex. Ctenaspidinae There is no subdivision of the shield into epitega. The super- ficial layer is typically absent, and the ornamentation consists of tubercles or an irregular network of ridges. The branchial plates are believed to be fused to the dorsal shield. Postbranchial lobes are absent, and the branchial openings lie between the postero-lateral corners of the dorsal and ventral shields. Ctenaspis. Ctenaspis Kiaer Type-species. — Ctenaspis dentata Kiaer. Ctenaspis Kiaer, 1930, Skr. Svalbard Ishavet, 33, pp. 3-5; 1932, Skr. Svalbard Ishavet, 52, p. 20; Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, pp. 19, 30, 58, 74, 87; Gross, 1935, Palaeontogr., 83A, p. 11, F0yn and Heintz, 1943, Skr. Norges Svalbard Ishavs-Unders0k., 85, p. 42; Flower and Way- land-Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 364; Denison, 1953, Fieldiana: Geol., 11 , no. 7, p. 293; Stensio, 1958, Traite de Zool., 13, fasc. 1, pp. 295, 318-319, 369, 384, 386, 387, 394. DENISON: THE CYATHASPIDIDAE 439 Fig. 154. Ctenaspis dentata (X3). Museum, Oslo, D 582; B, ventral shield. orb, orbit. A, type, dorsal shield, Paleontologisk (From Kiaer, 1930.) Diagnosis. — Ctenaspis includes Ctenaspidinae in which the shield is short and relatively very broad (width ratio= .70-.90). The dorsal shield is rather flat, has the orbits far forward (orbital ratio=.08- .12), and has well-developed lateral brims. The ventral shield is more strongly arched, except for flatter marginal areas. Discussion. — The presence or absence of distinct branchial plates has not been clearly demonstrated in Ctenaspis. Kiaer (1930, p. 4) merely stated that "free branchial plates have not yet been found ..." Stensio does not discuss the matter in his text, but in a reconstruc- tion of Ctenaspis kiaer i (1958, fig. 176) each side of the dorsal shield is shown to have a ventro-lateral lamella which "repr^sente peut- etre la plaque branchiale." The branchial opening is shown behind the posterior end of the lamella at the postero-lateral corner of the shield, and is bounded above by the dorsal shield and below by the ventral shield. The situation is essentially the same as that of Allo- cryptaspis, except that Ctenaspis lacks postbranchial lobes and has a marked, toothed brim separating the dorsal face of the shield from the ventro-lateral lamella. I have been able to confirm the presence of the ventro-lateral lamellae and the position of the branchial open- 440 FIELDIANA: GEOLOGY, VOLUME 13 ings on a specimen of Ctenaspis dentata (CNHM, PF 1088, fig. 155). It shows on the ventral surface of the dorsal shield behind the lamella an ornamentation of a few fine ridges, presumably of dentine, over- lying the branchial opening. Though the orbits are far forward, the pineal organ is rather far back (pineal ratio= .23-27) . Preorbital processes are well developed, and between them is a flat maxillary brim (fig. 155, mxb), ornamented with denticles and ridges (presumably of dentine) parallel to the rostral edge. Some species have dorsally a postero-median crest which is part of the dorsal shield and not a separate element as is the dorsal spine of pteraspids. The lateral line pattern (figs. 97, D, 99, F) shows some differences from the typical cyathaspid arrange- ment which Kiaer (1930, p. 3) believed approached the pteraspid condition. The posterior ends of the supraorbital canals (= pineal canals) meet behind the pineal macula. Only two or three dorsal transverse commissures are present. Paired medial ventral canals are continuous with the postoral canals and extend about half way back on the ventral shield. A thin section of Ctenaspis dentata (PF 1088, fig. 101, C) shows that the superficial layer is absent. The surface ornament consists of tubercles of aspidine, and this is probably true also of other described species of Ctenaspis. Ctenaspis dentata Kiaer. Figures 97,D, 99,F, 101, C, 154, 155. Ctenaspis dentata Kiaer, 1930, Skr. Svalbard Ishavet, 33, p. 7, figs. 1, 2, 4a; 1932, Skr. Svalbard Ishavet, 52, p. 20; Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, drawings 29-30; Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 295, fig. 177; Friend, 1961, Proc. Yorkshire Geol. Soc. 33, pt. 1, no. 5, p. 112. Type.— Paleontologisk Museum, Oslo, D 582, dorsal shield (fig. 154, A). Occurrence. — Early Devonian (Gedinnian), Ben Nevis (and ? Fraenkelryggen) division, Red Bay series, Spitsbergen. Diagnosis. — The length of the dorsal shield is 25-27 mm., its width ratio is about .80, and its lateral brims are finely serrate. The surface ornament consists of large, flat tubercles, pointed toward the rear. Ctenaspis cancellata Kiaer Ctenaspis cancellata Kiaer, 1930, Skr. Svalbard Ishavet, 33, p. 7, fig. 46; Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 295; Friend, 1961, Proc. York- shire Geol. Soc, 33, pt. 1, no. 5, p. 112. Ctenaspis conselatus (in error) Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 20. DENISON: THE CYATHASPIDIDAE mxb \ \ pfo ^ 441 Fig. 155. Ctenaspis dentata, ventral view of dorsal shield based largely on CNHM, PF 1088 (X 3); impressions on inner side of central part largely from Kiaer, 1930. br, impression of cranial cavity; bro, branchial opening; k2.s, impressions of gill pouches; mxb, maxillary brim; obr, ornamented border of branchial opening; orb, orbit; pfo, fossa for pineal organ; pop, preorbital process; sec, impression of semicircular canal; vll, ventro-lateral lamina. Type. — Paleontologisk Museum, Oslo, D 543a. Occurrence. — Early Devonian (Gedinnian), Ben Nevis division, Red Bay series, Spitsbergen. Diagnosis. — The length of the dorsal shield is about 30 mm., its width ratio is about .90, and the lateral brims are coarsely serrate. The surface ornament consists of rounded tubercles, more widely spaced than in C. dentata. Ctenaspis kiaeri Zych Ctenaspis kiaeri Zych, 1931, Fauna Ryb Dewonu i Downtonu Podola, photo 11, drawings 45a-e; Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 20; Brotzen, 1936, Ark. Zool., 28A, no. 22, p. 6; Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 295, figs. 176, 178. Lectotype. — Naturhistoriska Riksmuseet, Stockholm, C1616, dor- sal shield (according to Stensio, 1958, fig. 176; not designated by Zych). 442 FIELDIANA: GEOLOGY, VOLUME 13 Occurrence.— Eav\y Devonian (Gedinnian), passage beds of Czort- kow stage, Podolia. Diagnosis. — The length of the dorsal shield is about 23 mm., its width ratio is about .70, and its lateral brims are only faintly serrate. The surface ornament consists of a network of delicate ridges. Ctenaspis zychi Stensio Ctenaspis zychi Stensio, 1958, Traite de Zool., 13, fasc. 1, p. 295 (nomen nudum, listed as a new species from Podolia). Ctenaspis sp. Ctenaspis n. sp. aff. C. dentatus Kiaer, Thorsteinsson, 1958, Geol. Surv. Can- ada, Mem. 294, pp. 76-77. Occurrence. — Early Devonian, Snowblind Bay formation, Read Bay, Cornwallis Island, Canada. cf. Ctenaspis sp. Wills, 1935, Trans. Roy. Soc. Edinburgh, 58, pp. 428-429, pi. 1, fig. 1; White, 1946, Quart. Jour. Geol. Soc, London, 101, p. 210. Occurrence. — Early Devonian (Late Downtonian), zone of Tra- quairaspis symondsi, Shropshire, England. Ctenaspis F0yn and Heintz, 1943, Skr. Norges Svalbard Ishavs-Unders0k., 85, pp. 43, 44; Friend, 1961, Proc. Yorkshire Geol. Soc, 33, pt. 1, no. 5, p. 111. Occurrence. — Early Devonian (Gedinnian), Fraenkelryggen divi- sion, Red Bay series, Spitsbergen. Cyathaspididae indet. Indeterminable Cyathaspididae from Eastern United States Eoarchegonaspis wardelli Ruedemann Anatifopsis icardelli (in part) Ruedemann, 1916, Bull. N. Y. State Mus., 189, pp. 102-105, pi. 32, fig. 2. Eoarchegonaspis wardelli (in part) Kiaer, 1932, Skr. Svalbard Ishavet, 52, p. 24; Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, p. 370, pi. 2, fig. 5. Lectotype.—New York State Mus. 9613. Occurrence. — Late Silurian, Otisville Shale member of Shawan- gunk formation, near Otisville, Orange County, New York. Discussion. — The taxonomic and nomenclatorial problems re- lated to this species have been considered above (p. 374) in the dis- DENISON: THE CYATHASPIDIDAE 443 Fig. 156. "Archegonaspis'' drummondi, type; internal impression of dorsal shield, Museum of Comparative Zoology 8875 (X 3). k-,-s, impressions of gill pouches; pfo, fossa for pineal organ; sec, impression of semicircular canal. cussion of Vernonaspis vaningeni. Since Flower and Wayland-Smith have designated an unrecognizable fragment as lectotype, this name becomes a nomen dubium and further use of it becomes impracticable. "Archegonaspis" drummondi Flower and Wayland-Smith. Fig- ure 156. Archegonaspis drummondi Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 380-381, pi. 2. figs. 1-2. Archegonaspis cf. drummondi Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 381-382, pi. 6. Archegonaspis ? sp. Flower and Wayland-Smith, 1952, Bull. Mus. Comp. Zool., 107, pp. 382-384, fig. 2, pi. 3, fig. 6, pi. 7, pi. 8. fig. 3. Type. — Museum of Comparative Zoology 8875, impression of the inner side of a dorsal shield (fig. 156). Occurrence. — Late Silurian, Vernon shale (Salina group) ; 2 miles southeast of Kenwood, Oneida County, New York. Discussion. — In the type specimen, the length of 22.1 mm. indi- cates a considerably smaller species than the others from the Vernon 444 FIELDIANA: GEOLOGY, VOLUME 13 shale, Vernonaspis allenae and V. Leonard i. It resembles Arche- gonaspis and differs from Vernonaspis in the following characters: (1) it is relatively broad (width ratio=.64), but its breadth has prob- ably been increased by flattening; (2) the pineal organ and orbits are placed rather posteriorly, but the ratios have probably been altered slightly by measuring on an internal impression; the pineal ratio=.24 and the orbital ratio=.14. It resembles Vernonaspis and differs from Archegonaspis in having median lobes on the rostral and posterior margins. The dentine ridges, indicated by a small patch on one postbranchial lobe and by impressions along the pos- terior margin, are very fine, about 8 per mm. The postbranchial lobes are weak and the postbranchial region is shorter than in Arche- gonaspis and Vernonaspis. Seven pairs of gill pouches are indicated, corresponding to those numbered k2-k8 by Stensio (1958). The ventral shield impression (Mus. Comp. Zool. 8876) identi- fied as Archegonaspis cf. drummondi by Flower and Wayland-Smith, agrees in size (length given as 21 mm.) and in its fine dentine ridges (about 8 per mm.). The dentine ridge pattern is essentially longi- tudinal, except that in front the ridges curve toward the antero- lateral corners. Its outline is not completely preserved so its origi- nal shape is unknown. The ventral shield in counterpart (Mus. Comp. Zool. 8877) named Archegonaspis"! sp. by Flower and Wayland-Smith, can also be referred to "Archegonaspis" drummondi. It is small; its total length is estimated to be 19.5 mm. (not 24 mm. as given in the origi- nal description). The shield is badly crushed and its proportions are not determinable. The "eccentric emargination of the anterior end" is the result of a break and is not a natural character. The posterior edge has a very slight rounded median lobe. The ridges are fine, about 8 per mm. The ridge pattern is similar to that of MCZ 8876. This specimen has been etched by the original describers and displays much of the pattern of the lateral line canals. It differs from the typical cyathaspid pattern, as exemplified by Poraspis, in that the lateral ventral canals are broken into short lengths and the medial ventral lines appear as transverse commissures showing a gradual transition anteriorly into the postoral canals. Although this is surely a valid species, none of the three speci- mens is adequate to determine its generic affinities. Possibly it is identical with Vernonaspis bryanti. Heterostraci plates. — Leutze (1960, p. 215). This record was based on scales, now PF 2131-2 in Chicago Natural History Museum, DENISON: THE CYATHASPIDIDAE 445 from the Late Silurian, Wills Creek formation, on the east side of Elkhorn Mountain, near Bass, Hardy County, West Virginia. Fish scales. — Hoskins (1961, p. 98, pi. 7, fig. 2). These fragments have been found in the Late Silurian, Bloomsburg formation, at Bea- vertown in Snyder County, Mifflintown in Juniata County, Strodes Mills, Pine Glen and Mt. Union in Mifflin County, and Neffs Mills in Huntingdon County, Pennsylvania. They were compared by Hoskins to cyathaspids from eastern Pennsylvania identified by Beerbower and Hait (1959) as Archegonaspis van ingeni (here referred to Vernonaspis vaningeni). Indeterminable Cyathaspididae from the Yukon, Canada Undetermined Cyathaspididae. — Denison (1963, pp. 127-132, figs. 74- 77). The Early Devonian1 limestones and graptolitic shales on the Beaver River in the southeastern Yukon have yielded, in addition to the species described above, a number of cyathaspids that are too incomplete or poorly preserved to warrant designation by name. A dorsal shield, Princeton 17384 (op. cit., fig. 74), is of interest be- cause of its superficial subdivision into a central disc, a lateral epite- gum, a postero-lateral area, and a scale-like band along the posterior border. Another dorsal shield, Princeton 17385, is notable for its great relative breadth, its width ratio being 0.77. A fragment of a ventral shield, Princeton 17094, is characterized by distinct anterior and lateral marginal areas, and by fine ridges. Another ventral shield (fig. 157), Princeton 17387 (op. cit., figs. 75-76), has a dentine ridge pattern resembling that of pteraspids, though its lateral line sensory canals are arranged as in cyathaspids. Two other ventral shields, Princeton 17093 and 17386, are indeterminable, though the latter may belong to Poraspis. Finally a branchial plate, Princeton 17099 (op. cit., fig. 77), could belong to Vernonaspis bamberi. Poraspidinae indet. — A new locality discovered in 1963 by California Standard Company geologists near the Snake River in northern Yukon, has yielded, in addition to the IDinaspidella sp. described above (p. 400), two other cyathaspids. One (CNHM, PF 3872) is a ventral shield of a small member of the Poraspidinae. Its age is probably Early Devonian. 1 Tentatively assigned to the Middle Ludlovian in Denison (1963); see below pp. 450-451 cvm Fig. 157. Cyathaspid indet.; ventral shield, Princeton 17387 (X 2). cvl, cvm, pores of lateral and medial ventral transverse sensory commissures; hi, pores of lateral ventral sensory canal; pol, pores of postoral sensory canal. 446 DENISON: THE CYATHASPIDIDAE 447 Cyathaspidinae indet. — From the same locality in the northern Yu- kon, a shield fragment (CNHM, PF 3873) shows larger and higher dentine ridges interspersed between smaller and lower ridges. This character occurs in a number of Cyathaspidinae. UNDESCRIBED CYATHASPIDIDAE FROM ARCTIC CANADA Cyathaspididae n. gen. A and sp. B. — Thorsteinsson (1958, pp. 54, 72, 100) . This is listed from the Middle Silurian (Late Wenlockian, zone of Monograptus testis and Cyrtograptus trilleri) in the Read Bay formation (member A) on Goodsir Creek, and in the Cape Phillips formation (member C) on Snowblind Creek, Cornwallis Is- land, Canada. New genus and species of Cyathaspididae. — Thorsteinsson (1958, p. 55; 1963, pp. 227-228). This is reported from the Late Silu- rian (Early Ludlovian), Read Bay formation (member A), on Shella- bear Creek, Cornwallis Island, from the Douro formation, near Ptarmigan Lake, Douro Range, Devon Island, Canada. Cyathaspididae gen. and sp. indet. — Thorsteinsson (1958, p. 68). This is listed from the Late Silurian (? Middle Ludlovian), Read Bay formation (member C), at the head of Snowblind Bay, Cornwallis Island, Canada. Cyathaspididae gen. and sp. indet. — Thorsteinsson (1958, p. 77). This occurs in the Snowblind Bay formation on Read Bay, Cornwallis Island, Canada. Its age is probably Early Devonian. Porapsinae [sic]. — Thorsteinsson and Tozer (1963, p. 122). This is listed from the Late Silurian or Devonian, Peel Sound formation, 6 miles southeast of Cape Anne, Somerset Island, Canada. Indeterminable Fragments from Northern Europe Cyathaspides.— Lehman (1937, pp. 62-64, pi. 8, figs. 85-88). This record is based on a shield fragment resembling Cyathaspis in hav- ing higher dentine ridges separated by two or three lower and nar- rower ridges. It came from the latest Silurian or earliest Devonian Oved-Ramsasa beds in southern Sweden. It has also been figured by Stensio (1958, fig. 173,C). Poraspididae gen. indet. — Karatajute-Talimaa (1962, pp. 46, 48, 53, 55, 58, text-fig. 1:6, pi. 1, fig. 23). An incomplete scale from a deep 448 FIELDIANA: GEOLOGY, VOLUME 13 borehole in Lithuania is considered to be Early Devonian (Late Downtonian) on the basis of associated fish fragments. Fragmentary or questionable Cyathaspididae from Asia Sanidaspis siberica Bystrow Sanidaspis siberica Bystrow, 1959, Acta Zool., 40, pp. 59-62, figs. 1-3. Occurrence. — Early Devonian, Krasny Creek, a left tributary of the River Nizhni Viluikan, Siberia (60° 20' 27" N.; 107° 01' 15" E.). Description. — The shield has coarse, shiny, dentine ridges, each underlain by two pulp canals. The cancellous layer has very large chambers. Discussion. — This genus and species was based on fragments only and its affinities are uncertain. It resembles Allocryptaspis in the coarseness of its dentine ridges (.30-.35 mm.), in the thickness of its shield (1.3 mm.), and in having a pair of pulp canals in each ridge. However, Allocryptaspis generally has a linear arrangement of cham- bers in the cancellous layer under each dentine ridge, while there is no correspondence of ridges and chambers in Sanidaspis. Allo- cryptaspis has few lateral canals connecting pulp canals and inter- costal grooves, while these are numerous in Sanidaspis. Kiangsuaspis nankingensis P'an New genus of Cyathaspida: P'an, 1961, Vertebrata Palasiatica, 1961, no. 4, p. 348. Kiangsuaspis nankingensis P'an 1962, Acta Palaeontologica Sinica, 10, no. 3, pp. 407-408, text-fig. 1A, pi. 1. Type. — Incomplete ventral shield, no. V994, Museum of Geology, Ministry of Geology, Peking. Occurrence. — Possibly Late Silurian, upper part of Fentou series, Fentou, Nanking, Kiangsu Province, China. Discussion. — The ornamentation in the central part of the shield consists of widely spaced coarser and finer ridges, separated by a network of very fine, low ridges, the latter possibly formed by the reticular layer. There is nothing like it in any other cyathaspid. It is possible that Kiangsuaspis is related to the Traquairaspididae, though the ridges are not broken into short lengths as is usual in Traquairaspis. Fig. 158. Correlation chart showing occurrence of Cyathaspididae. GREAT BRITAIN SPITSBERGEN PODOLIA BALTIC CANADA WYOMING UTAH OHIO § z i o | - 2 1 i 1 < in I -z is z 1 1 1 I I- o Allc^iyptaipu tuipticj AlUiiptiipii fUbtlLformii p u < o: S So *">*-*-» * > — ^ ... C'rj Z & __ Inrgulirtupit hotb Homikipidrllj [rindi p^p^, i 3 , u in PofJlpu ^p,, i I ^ a - Angljjpu tp. b t/1 ^ = Z Ctnwpw c-mcelUu kn,W «™» NORTH GERMANY CORNVVALLIS ISLAND NORTHWEST TERRITORIES BRITISH COLUMBIA z i i s I^-Wp- o^r. u*ri (ERRATICS) □ z 3 z 3 5 Porup,. ,,. * iS z 3 Angina hnnu, PocKpu ^^nji, Angbipu im.gtus Poraspu polirrt Cutuspu .p. pMltpil e|onga[i 0 N U Po.up.. „mpk. z | A^p. .p - > fc^pb * 7 •o o Cp.thj.jnd ind«. YUKON ;- Pompi .p. Ulrupil t.n.do..u Vcmonuri b.mb*r> .. 1 p 8 sag - . 0 Angl-pa >p. OESEL < I y □ z 1 o ; ' Cv.rf.«pid md«. Z < = f hZLpT;: k,„,„ PENNSYLVANIA MARYLAND :■'. 8 Cpak-p. h^. ? Toh.p.kp,. d. «*!«. NEW ► B ■ o BRUNSWICK p1 T.lml^« »»J»Lu «- D * "' gz - : gP Cp^p. ™^ 'P' 0 - f « ¥■ '"•<«■ A»„™p» ct^fa Q CJ a 2 ' ^5 ^""""■P" -P- < 1 s Kl 3. Q «w g„. W ,p. Z 0 p q Ssi z£ z r o X 3 5 V.^,^. i ?; Zip o™ II 'cEip5£?— '■ 5 ES A*^*^. ; -: z 1 = 7 0° su A < — Cp.ihi.p.d > c £ z ~ WENLOCK- IAN - 3 I ! 2 v | <2 Toipptkpu "- •?■ A u ' LLANDO- B VER1 \N DENISON: THE CYATHASPIDIDAE 449 GEOLOGICAL RANGE The Cyathaspididae have a rather short geological range. The earliest known member of the family occurs near the base of the Middle Silurian, while the latest occurs below the top of the Early Devonian. The relative ages of most of the species are shown on the accompanying correlation chart (fig. 158). For purposes of correla- tion the British section is used as a standard, and for the Early De- vonian the stage names of the predominantly marine section of the Ardennes and Rheinischen Schiefergebirge are also used. A few of the correlations require some comment. In the British section the boundary between the Downtonian and Dittonian is here retained between the zone of Traquairaspis sy- mondsi and the zone of Protopteraspis leathensis, as is common usage (White, 1950, p. 56; Ball and Dineley, 1961, pp. 201-202). Allen and Tarlo (1963, pp. 146-148) have recently argued that the bound- ary should be placed lower, at the base of the zone of Traquairaspis pococki. The Czortkow stage of Podolia has commonly been referred to the Downtonian, and recently certain Russian stratigraphers have placed it in the Tiver stage, which they insert between the typical Ludlovian and Downtonian (Sokolov, I960; Nikiforowa and Obut, 1962). The pteraspids not only indicate that the age is Dittonian, but that it is Middle Dittonian (Denison, 1956, pp. 403-404; 0rvig, 1961, p. 523). The pteraspids also indicate that the overlying "Old Red" (Babin sandstone) includes Middle Dittonian to Breconian equivalents. A recent review of the invertebrate faunas, particu- larly the brachiopods (Boucot and Pankiwskyj, 1962), dates the Czortkow stage as Early Gedinnian. The Upper Oesel group of the Island of Oesel in the Baltic has generally been correlated with the Ludlovian, but some workers have argued for an earlier dating (Lamont, 1952, p. 30). Some Russian stratigraphers (Sokolov, 1960, p. 101; Nikiforowa and Obut, 1960, pp. 281-282) place the Upper Oesel group in the Early Ludlovian, but it should be noted that in their usage this corresponds to the entire Ludlovian of the typical section. Glacial boulders of Beyrichienkalk from northern Germany have been separated into two zones by Gross (1961, pp. 74, 76). The earlier zone is characterized by Thelodus parvidens and may be equiv- alent to the upper or K4 horizon of the Upper Oesel group. In this zone occurs Tolypelepis cf. undulata, (in litt., March 18, 1963). The 450 FIELDIANA: GEOLOGY, VOLUME 13 later zone is characterized by Thelodus cf. scoticus. This includes the so-called red Beyrichienkalk in which is found Anglaspis, Corv- aspis and Traquairaspis, an assemblage suggestive of Late Down- tonian age. On Cornwallis Island the geology has been described and the faunas listed, but not yet described, by Thorsteinsson (1958). The two cyathaspid-bearing horizons in the upper part of the Cape Phil- lips formation are dated by graptolites: Tolypelepis n. sp. A, found in the zone of Monograptus riccartonensis (Early Wenlockian), is the oldest known member of the family; Cyathaspid n. gen. A and sp. B occurs in the zone of Monograptus testis and Cyrtograptus trilleri (Late Wenlockian). Tolypelepis n. sp. A also occurs in the upper part of the Allen Bay formation, which for this reason in part is cor- related with the upper part of the Cape Phillips formation. Cyath- aspid n. gen. A and sp. B also occurs in the lower part of the Read Bay formation, which is correlated for this and other reasons with the upper part of the Cape Phillips formation. The Read Bay for- mation also contains a new genus and species of cyathaspid which is dated by brachiopods and graptolites as Early Ludlovian, and an undetermined cyathaspid which is only tentatively dated as Middle Ludlovian. Other cyathaspids have been found in the Snowblind Bay formation on Cornwallis Island. Thorsteinsson correlated their horizon with late Silurian or earliest Devonian for structural reasons. However, the presence of Pteraspis as well as Anglaspis and Ctenaspis indicates a Dittonian age. The occurrence in limestones and graptolitic shales in southeast- ern Yukon has been described by Hovdebo, Lenz, and Bamber (in Denison, 1963). The cyathaspid-bearing horizon is underlain by beds containing Early Wenlockian graptolites, and is overlain by beds with Monograptus dubius; the latter was believed to indicate that the age could not be younger than Middle Ludlovian. It has been shown recently (Jaeger, 1962, p. 112) that M. dubius ranges from the Late Llandoverian to the top of the Ludlovian. In view of this very long range, and of the fact that some species of Monograptus range well up into the Early Devonian (op. cit., table 3), it is possible that the Yukon fauna is considerably younger than Middle Ludlovian. It may well be Devonian, as is suggested by the presence of Traquair- aspis, a genus that occurs typically in the Late Downtonian, and is found as high as the base of the Middle Dittonian (Ball and Dine- ley, 1961, table 1). T. campbelli is probably the earliest known member of the genus and may be of Early Downtonian age. The DENISON: THE CYATHASPIDIDAE 451 Yukon fauna is now considered to be probably of Downtonian age, and is placed provisionally in the Early Downtonian. A new locality, discovered in 1963 by California Standard Com- pany geologists on the Snake River in northern Yukon, contains cf. Dinaspidella sp., Poraspidinae indet,. and Cyathaspidinae indet. These and the associated Corvaspis suggest an Early Devonian age. The vertebrates occur 40 to 50 feet above a limestone with brachio- pods and corals of probable Late Silurian age. Cyathaspididae have been found by California Standard Com- pany geologists at two horizons on Mt. Sekwi, Northwest Territories, Canada. The lower horizon, designated 4994 '-5080', contains, be- sides Vernonaspis sekwiae, a coral and brachiopod fauna dated as Late Ludlovian by Alfred Lenz. The higher horizon, designated 4474', contains only Anglaspis expatriata. The resemblance of this species to the British A. macculloughi is the basis for the correlation with Late Downtonian. The new cyathaspids, Listraspis and Pionaspis, from northwest of Muncho Lake in British Columbia, were also collected by Cali- fornia Standard Company geologists, at a horizon designated 2100'. This horizon lies several hundred feet stratigraphically above Middle Silurian corals. The cyathaspids themselves are specialized and so suggest Devonian rather than Silurian age. With them is found a Traquairaspis of about the size of T. symondsi, and suggestive of a Late Downtonian age. The fish-fauna of the Jones Creek formation of New Brunswick is difficult to date precisely. Its only cyathaspid is here referred to Cyathaspis, and appears to be close to C. banksi, which occurs in the latest Ludlovian and earliest Downtonian of Great Britain. The other vertebrates are coelolepids (Thelodus), a single presumed ana- spid (Ctenopleuron) , and acanthodian spines. Neither these nor the associated invertebrates, Ceratiocaris, Bunodella, and conodonts, are of much assistance in precise correlation of this formation. Accord- ing to Arthur J. Boucot (in litt., Oct. 1, 1962) it is "most likely of Ludlow age," which conforms with the indication of the Cyathaspis. The correlation of the Beartooth Butte and Water Canyon for- mations of Wyoming and Utah with the European section is based largely on the presence of Protaspis but is supported by the cyatha- spids and arthrodires. These indicate a Late Dittonian or Early Siegenian age. The Holland Quarry Shale of Ohio is believed to be slightly older but still of Early Siegenian age (Denison, 1960, p. 609). 452 FIELDIANA: GEOLOGY, VOLUME 13 These formations contain the latest and largest known members of the Cyathaspididae. In the northeastern United States a number of formations, mostly red beds, contain Vernonaspis and Americaspis. In only one case, the Vernon Shale of central New York, is there an associated inver- tebrate fauna, and in no case is there at present any very firm basis for correlation with the European section. All of the occurrences probably fall within the Cayugan Series, and it is probable that the boundary of this series with the underlying Niagaran is somewhere near the Ludlow- Wenlock boundary of England (Arthur J. Boucot, in litt, May 25, 1962). So the best that can be done at present is to place Vernonaspis and Americaspis somewhere within the Ludlovian. HABITAT In a previous paper (Denison, 1956, pp. 371, 376-387, 416-417) I reviewed the habitat of early vertebrates and concluded that Silu- rian Heterostraci, all of which were Cyathaspididae, lived mostly in the seas. It was pointed out that while an isolated occurrence of a single cyathaspid might not be significant, a large number of such occurrences conforming to a pattern did have ecological significance. The absence of Heterostraci from presumed fresh-water or brackish Silurian vertebrate occurrences (Oesel Ki; Ringerike, Norway; and Lesmahagow inlier, Scotland) furnished additional, if negative, evi- dence in support of this hypothesis. Since the publication of the 1956 paper there have been new discoveries which strengthen this view. The following Silurian and Early Downtonian cyathaspid occur- rences are unquestionably marine: Late Ludlovian, England Cyaihaspis Middle Ludlovian, England and Wales Archegonaspis Graptolithengestein, Germany and Poland Archegonaspis Hemse group, Gotland Archegonaspis Upper Oesel group (K4), Oesel Tolypelepis Allen Bay formation, Cornwallis Island Tolypelepis Read Bay formation, Cornwallis Island New cyathaspids Cape Phillips formation, Cornwallis Island Tolypelepis and new cyathaspid Limestones and graptolitic shales southeast Yukon Vernonaspis, Ptomaspis, Dikenaspis, Ariaspis, Homalaspidella Horizon 4994'— 5080', Mt. Sekwi Northwest Territories Vernonaspis DENISON: THE CYATHASPIDIDAE 453 The occurrence in southeastern Yukon is of special importance because the cyathaspids are both common and varied. They are found in dolomites in direct association with articulate brachiopods, fistuliporid bryozoans, and crinoids, and are in the midst of a series of shales that contain graptolites. The Vernonaspis from Northwest Territories is associated with a coral and brachiopod fauna. The Jones Creek formation of New Brunswick is difficult to in- terpret ecologically. This occurrence, misidentified as the Long Reach formation by me (1956, p. 384), contains Cyathaspis associ- ated with the coelolepid Thelodus and a few ceratiocarids. Other horizons contain acanthodian spines, conodonts, and abundant cera- tiocarids. Unspecified horizons have yielded a single presumed ana- spid (Ctenopleuron) , and a xiphosuran (Bunodella) , both suggestive of brackish or at least not typically marine conditions. The rock containing the Cyathaspis is very finely laminated, consisting of al- ternating thin dark carbonaceous shaly laminae, thin silts, and some- times thicker clayey layers. All of these indicate quiet, perhaps sometimes stagnant waters. The mixed fauna is suggestive of mar- ginal conditions, and lagoons come to mind as a possible site of deposition. The other Silurian cyathaspid occurrences, with Vernonaspis and Americaspis, are in sequences of red beds in the northeastern United States. In many of these, the cyathaspids are the only known fos- sils. However, in the Vernon shale of central New York, Vernon- aspis occurs in gray or buff shales intercalated in the red beds, and is associated with a varied invertebrate fauna much of which is un- doubtedly marine. Fisher (1957, p. 21) concluded that the Vernon shale as a whole was deposited in "the littoral area of a restricted semi-marine environment," and that the faunal zone that contains the cyathaspids indicates a temporary marine invasion. The Ver- nonaspis of the Wills Creek formation of Maryland occurs in associ- ation with Lingula and ostracods (Leutze, 1960, p. 215), and is thus presumably brackish or marine. Fragments of Heterostraci, pre- sumably cyathaspids, occur in the Bloomsburg formation of central Pennsylvania. The lithology and associated invertebrate fauna here indicate, according to F. M. Swartz (1946, p. 27) and Hoskins (1961, pp. 103-105), brackish waters in broad lagoons fed by a river system from the east. More eastern occurrences in the Bloomsburg forma- tion of Pennsylvania, and the approximately equivalent High Falls formation of New Jersey, and Otisville shale member and Longwood shale of New York are considered to be probably fresh-water, alluvial 454 FIELDIANA: GEOLOGY, VOLUME 13 deposits (Swartz, loc. cit.; Hoskins, loc. cit.; and Beerbower and Hait, 1959, p. 200). It should be noted, however, that the commonest fos- sil in these formations is Vernonaspis, which elsewhere is presumably an inhabitant of salt or brackish waters. In Perry County of central Pennsylvania the Wills Creek formation is topped by the Landisburg sandstone, in which Americaspis is common. The shales and limy beds in this formation contain abundant Leperditia and locally small brachiopods and gastropods, and so they are considered to be ma- rine or brackish-water deposits. The Landisburg sandstone contains only Americaspis and is assumed by some to be a fresh-water de- posit. Dr. James L. Dyson (in litt., June 29, 1963) notes characters of floodplain deposition, including obvious channels, associated with lagoonal features. He suggests that the environment was one mainly of small deltas in a lagoon. The position between marine beds of the underlying Wills Creek formation and the overlying Tonoloway for- mation indicate that this was deposited near the shoreline. In summary, the Silurian habitat of most cyathaspids, even of some of those found in or associated with red beds, was clearly in the sea. Toward the end of the period, Americaspis and some Vernon- aspis inhabited marginal habitats, perhaps both brackish and fresh- water. Most Silurian cyathaspids belong to the Tolypelepidinae and Cyathaspidinae, few of which survive into the Devonian. The Ver- nonaspis from southeastern Yukon is now believed to be Early Down- tonian and is marine. Cyathaspis occurs at the base of the Downtonian in the Downton Castle sandstone, which has been interpreted by me (1956, p. 391) and by Allen and Tarlo (1963, pp. 135-136) as a ma- rine, probably brackish deposit, but was considered to be a fresh water or brackish deltaic deposit by Ball and Dineley (1961, p. 215). The Podolian genus, Seretaspis, probably comes from the Czortkow stage, a shallow-water, marine deposit (Denison, 1956, p. 405). Pi- onaspis and Listraspis occur in deposits in which no associated in- vertebrates have been found, but the accompanying heterostracan, Traquairaspis, suggests a fresh-water habitat (op. cit., p. 417). Most Early Devonian cyathaspids belong to the Poraspidinae, and with them are sometimes associated Irregulareaspidinae and Ctenaspidinae with probably similar habitats. With the exception of Ariaspis, Homalaspidella, and Dikenaspis from the Yukon, they are not found in typical marine deposits. They do occur often in marginal marine sediments, notably the Czortkow stage of Podolia, many horizons of the Red Bay series of Spitsbergen, and the Water DENISON: THE CYATHASPIDIDAE 455 Canyon formation of Utah (Denison, 1956, pp. 400-402, 405, 414). The Beartooth Butte formation at the type locality (op. cit., p. 414) and the Holland Quarry shale (Denison, 1960, pp. 610-611) were probably deposited in arms or estuaries of the sea where the waters may have been brackish or fresh. Other Early Devonian cyathaspid occurrences are in fresh-water deposits. Surely identifiable as such are the British Dittonian (Ball and Dineley, 1961, p. 217; Allen and Tarlo, 1963, pp. 143-146), and the Old Red of Podolia, both of which are probably deltaic. Probably fresh-water, but with intercalated marine beds, are much of the Late Downtonian of Britain, the beds with Poraspis in the Psammites de LieVin of France, and some hori- zons of the Red Bay series of Spitsbergen. In summary, the early habitat of cyathaspids was in the seas. Their absence in most marine Silurian formations has been used by Romer (1946, p. 45) as an argument for a fresh-water habitat, but more probably indicates that they were largely restricted to some special marine ecological niches. That these were near-shore is sug- gested by the interpretations of many of the Silurian formations in which they are found. In the Late Silurian some cyathaspids show a trend into marginal habitats that may well have been brackish, and a few probably lived in fresh-water streams. This trend is con- tinued in the Early Devonian, at which time almost all cyathaspids lived in marginal marine or fresh-water habitats. ADAPTATION Cyathaspids have no close counterparts among living vertebrates, so it is difficult to determine their adaptation except in a general and sometimes speculative way. Most conclusions must be drawn from Kiaer's restoration based on the few articulated specimens of Ang- aspis heintzi (fig. 90). The anterior part of the body was encased in a shield that was essentially immovable. The posterior part of the body, including the tail, was covered with overlapping scales, and so was flexible; but because the body scales were so large, the range and variety of move- ments must have been distinctly limited. There were no paired fins, and ridges and keels that would increase stability in swimming were often absent and rarely strongly developed. The body and tail prob- ably formed a moderately powerful organ of propulsion. The tail itself, as restored by Kiaer, was somewhat asymmetrical and had an extended ventral lobe, stiffened by relatively large fulcral scales all 456 FIELDIANA: GEOLOGY, VOLUME 13 around the margin. It is unlikely that strictly lateral movements in ordinary swimming would have resulted in elevating or depressing this tail to any great extent. With forward movement, lift would have been given to the anterior part of the body by the upwards slope toward the rostrum of the ventral surface. But progress in different vertical or lateral directions must have resulted largely from controlled movements of the body and tail. The swimming of cyath- aspids may perhaps best be compared to that of tadpoles. They probably swam in any direction, with adequate control, quite rapidly for short spurts. The following characters indicate that cyathaspids were not highly specialized for life on the bottom: their well-rounded con- tour, especially that of the strongly arched ventral shield ; the lateral position of the eyes; the position of the mouth, which, though on the lower surface, must ordinarily have been some distance above the lowermost part of the body. It is probable, however, that cyatha- spids spent much time on the bottom, at least when resting; this would have resulted from their presumed relatively high specific gravity, due to a heavy shield and probable lack of an air bladder. The sensory equipment of cyathaspids included a pair of very small eyes that were directed laterally. The nostrils are believed to have opened into the anterior part of the roof of the mouth. It is possible that there were sensory barbels around the mouth, as sug- gested by Stensio (1958, p. 355). Very probably the usual three pairs of semicircular canals were present. The lateral line system was strongly developed, and perhaps also the pore-canal system; this suggests that cyathaspids were well equipped for getting around in turbid or poorly lighted waters. The manner of respiration of cyathaspids has been discussed above (p. 346). The main problem is the manner of production of a respiratory current, and it was concluded that there was probably a muscular pumping apparatus, perhaps derived from a velum. For feeding, it is clear that there was no mechanism suited for biting or chewing. The oral plates, set in a skin membrane and worked by small muscles, served mainly to open and close the mouth. It is not unlikely that the mouth was protrusible, perhaps forming a scoop. With it cyathaspids could have scooped, sucked or perhaps even picked up detritus from the bottom, ingulfed small inverte- brates, and swallowed small animals and plants that floated on or in the water. Much of their feeding may have been done on the DENISON: THE CYATHASPIDIDAE 457 bottom, but to accomplish this they would have had to elevate the tail and depress the snout. EVOLUTION AND GROWTH The classical theory of the origin of dermal bones is credited largely to Hertwig (1876). He considered isolated placoid scales, such as occur in elasmobranchs, as primitive, and believed that der- mal bones were formed by the enlargement and coalescence of the basal plates of such scales. This theory was applied to Heterostraci by Traquair (1899), who placed the Coelolepida, which have small, isolated scales, at the base of the phylogeny, and regarded the Psam- mosteidae and Pteraspididae (including the Cyathaspididae of the current classification) as successive stages in the fusion of the shield. There was wide acceptance of this concrescence theory. In 1903 Jaekel pointed out that the paleontological evidence did not support the progressive fusion of plates, but suggested rather a phylogenetic subdivision of an originally continuous covering. Jaekel's brief statement was largely overlooked, and it was not until his theory was revived and developed by Stensio (1927) that it be- came quite widely accepted by paleontologists. According to this theory, the Cyathaspididae, especially the Poraspidinae, were prim- itive, while the Pteraspididae, Psammosteidae, and perhaps Coelole- pida, exemplified in that order the tendency toward subdivision of the shield. In more recent years Obruchev (1945) has proposed another theory of the history of the heterostracan shield. He suggested that the primitive condition is shown by Tolypelepis, in which the shield appears to have been formed of a number of small plates or tesserae. Cyathaspis represents a later stage in which the tesserae have co- alesced to form a shield of large plates. The pteraspids and psam- mosteids were derived from cyathaspids by subdivision of the shield into plates which could grow and so protect the animal not only as an adult, but also in its early stages. The Poraspidinae, on the other hand, were considered to be a side branch of cyathaspids in which the armor could not grow and appeared only after full size was at- tained. As far as this theory applies to the Cyathaspididae (though not to other families), it has been largely accepted and further de- veloped by Stensio (1958), Tarlo (1960, 1962a,b), and 0rvig (1961). The paleontological evidence supports the view that a shield com- posed of many tesserae or scales was primitive in Heterostraci. The 458 FIELDIANA: GEOLOGY, VOLUME 13 scale-units, of course, are relatively large, complicated structures, not the small, simple placoid scales considered primitive in Hert- wig's theory. This was the type of shield possessed by the Ordo- vician genera, Astraspis and Pycnaspis. The earliest known cyath- aspid, Tolypelepis (fig. 107), had a dorsal shield apparently composed in large part of scale-like elements. An early Downtonian cyathas- pid, Ptomaspis (fig. 108), retained evidence of scale components in its dorsal shield. It has been assumed by Obruchev (1945, p. 264) and Stensio (1958, p. 305) that the scale-like elements of Tolypelepis grew, and that they fused when adult size was attained. It can be shown that the individual tesserae of Pycnaspis did grow; thin sec- tions demonstrate that small superficial tubercles were overgrown by larger tubercles of several generations, and concurrently new layers of aspidine were applied around the sides and base of the tes- serae. Growth of the scale-units of Tolypelepis has not yet been demonstrated, and the descriptions of their growth by Stensio (1958, pp. 297-306) are merely interpretations based on the dentine ridge pattern. Stensio assumes that this genus had both "cyclomorial" scales that grew, and "synchronomorial" plates that formed all at once after the animal had attained full size. This seems most un- likely, especially as it is possible to interpret the "synchronomorial" areas (e.g., lateral and rostral epitega) as having grown also. There are three other possible interpretations of the Tolypelepis shield: (1) the apparent tessera, scale, and plate units of the dorsal shield all formed early in life and grew, fusing at maturity; (2) the sub- division of the dorsal shield was restricted to the superficial layer, and these superficial units may have grown, but the deeper layers did not develop until maturity; (3) there was no growth of the dorsal shield, all of which formed at maturity; in this case the scale-like units would be merely relics of the earlier evolutionary stage. A more advanced evolutionary stage is exemplified by the Cyath- aspidinae, in which the shield is formed of larger units, termed by Stensio (1958, pp. 297, 308) "epitega" in the dorsal shield and "hy- potega" in the ventral shield. Except in the posterior part of the dorsal shield of Ptomaspis and perhaps Ariaspis, there is no indica- tion of scale-components that could have formed as separate units and fused at maturity. It is necessary next to examine the nature of the units, epitega and hypotega. Kiaer (1932, p. 8) and White (1935, p. 437) consid- ered the division of the shield to be merely superficial. But Kiaer and Heintz (1935, p. 40), Moy-Thomas (1939, p. 9), Obruchev (1945, DENISON: THE CYATHASPIDIDAE 459 p. 265), and Stensio (1958, p. 308) believed that in some genera at least, the epitega were distinct plates. The evidence now available does not lead to a definite answer. Epitega are already clearly marked in Tolypelepis (fig. 107). Precisely the same units are rec- ognizable in Cyathaspidinae, where they are distinguished most clearly by the pattern of the dentine ridges. Thus in Archego?iaspis (fig. 112,A), for example, the ridges of the rostral epitegum are largely transverse, those of the postrostral field radiate anteriorly, those of the central epitegum are approximately longitudinal, and those of the lateral epitega nearly parallel the lateral margins. These dif- ferences in pattern suggest that the superficial layer, at least, was formed in separate parts. Between the epitega there may be dis- tinct external grooves that suggest sutures. However, sutures have not been demonstrated on the internal surface, nor are they apparent in thin sections (fig. 100,B). If epitega are formed as separate ele- ments, one would expect to find them preserved separately, at least in an occasional juvenile individual. I know of no examples of this sort; in fact, the only specimen supporting the view that epitega are separate plates is the type of Cyathaspis acadica (fig. Ill), where flattening during preservation has caused the dorsal shield to break into its rostral, lateral, and central epitega. The manner of formation of the epitega has received relatively little attention. Could they have formed as scale-units that then fused together? This appears to be unlikely in Cyathaspidinae be- cause there is usually no indication of such scale-units. In Ptom- aspis, however, it is possible that the posterior part of the shield was formed in this fashion, though the anterior part probably was not. In those species of Archegonaspis and Cyathaspis that have dentine ridges of two sizes, it has been suggested by Obruchev (1945, p. 265) and Tarlo (1962a, p. 270) that the larger ridges belong to an earlier generation, and that the smaller ridges were formed later ontogeneti- cally. This is almost certainly not the case. A thin section of one specimen of Cyathaspis cf. acadica (CNHM, PF 1800; fig. 159, A) re- veals that the shield was in an early stage of formation and consisted only of a very thin capping of dentine on the ridges, continuing into a thin layer of aspidine around the intercostal grooves. The dentine is uniformly developed on all the ridges, high and low, and evidently had just been formed over much or all of the shield at the same time. This suggests that the shields of Cyathaspidinae formed only at ma- ture size. The superficial layer developed first, and then presumably the deeper layers. The superficial layer may have been formed in 460 FIELDIANA: GEOLOGY, VOLUME 13 units within the areas that are designated epitega, but this may not have been true for deeper layers. This manner of growth is very different from that of Pycnaspis and from that inferred by Stensio for Tolypelepis. Fig. 159. Transverse sections of dermal shields of juvenile cyathaspids ( X 75). A, Cyathaspis cf. acadica, CNHM, slide 4028; B, Allocryptaspis laticostata, CNHM, slide 4043. as, aspidine; dn, dentine; dru large elevated dentine ridge; ig, intercostal groove; pu, pulp cavity; tb, tubercle. There is some question whether the ventral shield was subdivided into hypotega equivalent to the epitega of the dorsal shield. Cyath- aspis banksi (fig. 110,B) has a band of fine ridges parallel to the an- terior edge and to the anterior parts of the lateral edges; these were designated by Stensio (1958, p. 308) "anterior postoral" and "mar- ginal" hypotega. Similar bands are formed in Cyathaspis acadica (fig. Ill) and in an undetermined cyathaspid from southeastern Yukon (Princeton 17094). But there are comparable ridges parallel to the anterior and lateral edges of both the dorsal and ventral shields of Homalaspiclella (figs. 145, 147), a genus that has gone farther than any other cyathaspid in the reduction of epitega. This and the ab- sence of any apparent subdivision of the ventral shield in most cyath- aspids makes the presence of distinct hypotega seem improbable. The anterior and posterior "hypotega" that Stensio (1958, p. 316) described in Anglaspis insignis have been shown above (p. 428) to be merely the result of an individual variation in ridge pattern. Within the Cyathaspididae, the final evolutionary stage appears to be the loss of most or all traces of epitega. It is this feature espe- cially that characterizes the Poraspidinae, and it is found also in the later genera of the Irregulareaspidinae. The significance of this change is clear: the dorsal shield, paired branchial plates, and ven- DENISON: THE CYATHASPIDIDAE 461 Fig. 160. Americaspis americana; ventral shield, CNHM, PF 3308 ( X 3/2), showing anomalous growth line (gl); photograph is of rubber impression of nat- ural mold. tral shield had become the main structural and growth units of the armor, and could have been formed only after growth had been com- pleted. This is confirmed by a thin section of Allocryptaspis lati- costata (fig. 159,B) ; though this specimen (CNHM, PF 3737) is of adult dimensions, its dorsal shield consists only of a very thin layer of dentine capping the ridges and continuing into a thin layer of aspidine surrounding the intercostal grooves. These thin layers are the earliest stage in the formation of the exoskeleton, and appear to have formed over the whole dorsal shield at once. The formation of the shield differs from that of the Cyathaspidinae only in that the superficial parts of the dorsal shield probably form as a unit rather than as separate epitega. 462 FIELDIANA: GEOLOGY, VOLUME 13 A few anomalies have some bearing on the formation of the cy- athaspid shield. PF 3308, a ventral shield of Americaspis americana, has an irregular and somewhat asymmetrical, oval area in its center, marked by its ornamentation (fig. 160, gl). Within the area the ridges are mainly longitudinal, but curve near the anterior and posterior ends to run parallel to the periphery of the area. Outside the oval area, the ridges follow the usual irregularly longitudinal pattern, but near the oval area may curve to run parallel to its margin. The boundary of the oval area resembles a growth line, but in view of the conclusion that the shield of Poraspidinae did not grow, must have some other explanation. It is comparable to the "growth lines" of Tremataspis (Denison, 1947, pp. 362-365), and may be explained in a similar fashion, that is, as reflecting an anomaly in the develop- ment of superficial soft tissues previous to calcification of the shield. Roughly comparable dentine ridge pattern anomalies subdivide the central epitegum of an undetermined cyathaspid from the Yukon (Denison, 1963, fig. 74) into: (a) a central area; (b) a postero-lateral area lying between (a) and the lateral epitegum, and forming the postbranchial lobe; and (c) a scale-like band along the posterior bor- der. The subdivision of (a) and (b) is not shown in other cyathaspids and is of doubtful significance. The scale-like posterior band is known also in the dorsal shield of the type of Ariaspis ornata (fig. 148), where a distinct median scale is also included. It is entirely possible that the scales of these two specimens formed separately and then fused to the dorsal shield. PHYLOGENY An attempt has been made in figure 161 to show the phylogeny of the Cyathaspididae, although it is necessarily generalized and pro- visional. The major phyletic subdivisions are here considered to be subfamilies, but within each subfamily details of phylogenetic his- tory are rarely determinable. The Tolypelepidinae are believed to include the ancestral stock of the family. From them was derived the Cyathaspidinae by loss of scale-components in the shield. In this subfamily Ptomaspis is the most primitive genus; from some such form could well have been derived the closely related and rather un- specialized Archegonaspis, Vernonaspis, and probably Seretaspis. Cyathaspis occupies a more isolated position within the subfamily, and this is also true of the late and specialized Pionaspis and Lis- traspis. The Irregulareaspidinae are a distinctive branch, probably derived from early Cyathaspidinae. The Early Downtonian Diken- Fig. 161. Phylogeny of the Cyathaspididae. 463 464 FIELDIANA: GEOLOGY, VOLUME 13 aspis retains epitega, but these are reduced or lost in the later genera. The Poraspidinae may represent an artificial assemblage of genera that have independently lost distinct epitega. The Silurian Americ- aspis, though possibly not directly ancestral, could be close to the stock from which Poraspis was derived. Allocryptaspis could easily have been derived from this same stock. The European Homal- aspidella is probably derived from the Poraspis stock, but the Cana- dian species could represent a distinct but parallel line. Ariaspis and Anglaspis are quite isolated within the subfamily. The most distinctive cyathaspids, the Ctenaspidinae, are of obscure ancestry. The absence of any apparent division of their dorsal shield into epi- tega has suggested relationship to the Poraspidinae (Kiaer, 1930, p. 5), but more probably results simply from the loss of the super- ficial layer. As far as present information shows, Ctenaspis could have been derived from either the Poraspidinae or Cyathaspidinae. It is unlikely that it evolved entirely independently from other cy- athaspids, as Zych (1931, p. 87) and Stensio (1958, p. 319) concluded. None of the subfamilies of Cyathaspididae are known to have sur- vived the Early Devonian, and the question arises as to whether these phyletic lines all became extinct without leaving any descend- ants. Those who have supported the theory of progressive phylo- genetic subdivision of the heterostracan shield (p. 457) have believed in general that the Poraspidinae, Cyathaspidinae, Pteraspididae, and Psammosteidae, in that order, represent morphological if not evolu- tionary stages. Obruchev (1945, p. 265) presented an improvement on this theory according to which such a form as Cyathaspis gave rise on the one hand to the Poraspidinae, a side branch that died out early, and on the other hand to the Pteraspididae and through them to the Psammosteidae. Stensio (1958, p. 252), following his lepido- morial theory, has explained the differences between the families as the result of fusions of smaller or larger elements. The cyathaspids were considered by him to be more advanced than the pteraspids because of greater fusion, and the ancestors of the pteraspids were believed to be forms like the psammosteids with belts of smaller scales separating the larger plates. Tarlo (1962a, p. 272) has followed this view. This is not the place to enter upon detailed criticism of the lepidomorial theory, but one point needs to be emphasized. It is probable that many evolutionary changes, including some within the Heterostraci, have resulted from fusions of smaller skeletal ele- ments to form larger elements. It is also probable in other cases that evolution has resulted in the subdivision of larger elements to form DENISON: THE CYATHASPIDIDAE 465 a few or many smaller elements. One cannot, therefore, automati- cally gauge the evolutionary advancement within the Heterostraci by the degree to which fusion of the shield has progressed. Specifi- cally, one cannot exclude the possibility of derivation of pteraspids from cyathaspids because subdivision of the shield is involved. There are a number of arguments in favor of a derivation of pteraspids from the Cyathaspidinae; they are: 1. The known geological ranges of the Cyathaspidinae and Pter- aspididae favor it. The former range from Early Ludlovian into Early Dittonian. The latter first appear in Late Downtonian, near the end of the range of the Cyathaspidinae. 2. The epitega, branchial plates, and ventral shield of Cyathas- pidinae are comparable to the plates of the pteraspid shield. The pineal plate of the latter may have arisen from a new center of ossi- fication in the postrostral field of cyathaspids. Derivation of the pteraspid orbital plate, which surrounds the eyes, from the cyathas- pid lateral epitegum, which is only notched for the eye, presents no real problem. It could have been accomplished either by a growth of the lateral epitegum down and around the orbit, or by fusion of the suborbital plate to the lateral epitegum. The dorsal spine of pteraspids was derived from a scale that attached to the dorsal shield, as has clearly happened in Ariaspis ornata (fig. 148) and Pter- aspis carmani (Denison, 1960, p. 578). The cornual plates of pteras- pids could have evolved from a pair of scales that attached to the postero-lateral corners of the dorsal shield. 3. The pteraspid lateral line pattern is easily derivable by minor modifications of the simple pattern of cyathaspids. 4. A thin section of a juvenile Protaspis (CNHM, slide 4070) shows that a very early stage of formation of a dorsal disc, approxi- mately 12.5 mm. long, is similar to an early stage of the shield of Cyathaspis (fig. 159,A) or Allocryptaspis (fig. 159,B). A thin layer of dentine caps the ridges and is joined by a thin layer of aspidine around the intercostal grooves, the whole developing at one time over the juvenile disc. Later, presumably, the reticular, cancellous, and basal layers are formed. The essential difference in the pteraspids is that following this early stage, the dorsal disc and other plates do not fuse as they do in cyathaspids, but are added to by successive peripheral growth increments until adult size is attained. 5. Certain cyathaspids approach pteraspids in one character or another. Some cyathaspids have a dentine ridge pattern close to that of pteraspids, transverse on the rostrum and broadly elliptical on the 466 FIELDIANA: GEOLOGY, VOLUME 13 central epitegum and ventral shield. A ridge scale is attached to the central epitegum of Ariaspis, homologous to the pteraspid dorsal spine. The suborbital plate of Listraspis is fused to the lateral epite- gum, completing the encirclement of the orbit. The narrow-crested, crenulated ridges of Listraspis are similar to those of pteraspids. These points favor the belief that the pteraspids were derived from cyathaspids. The most important advance in pteraspids was the mechanism permitting growth of the shield. This made it pos- sible for the shield to form early, thus protecting the larval pteraspid, yet permitting growth sometimes to a considerable size (White, 1958, pp. 229-232). The acquisition of dorsal spine and cornual plates by pteraspids must have given them somewhat more stability in swim- ming, but these adaptations sometimes appear in cyathaspids, though usually in a different way. The scales of the posterior part of the body are relatively smaller in pteraspids than in cyathaspids, per- mitting greater flexibility. The development of an enlarged rostrum on many pteraspids would give added lift in swimming to the ante- rior part of the body. These points all indicate that some pteraspids were better swimmers than cyathaspids, though, since no paired fins were evolved, their ability was limited. There is no strong evidence for the derivation of other families of Heterostraci from cyathaspids. Obruchev (1945, p. 265) believed that the Psammosteidae were derived from Pteraspididae, and thus more remotely from Cyathaspididae. Stensio (1958, p. 334) thought that Cardipeltis was derived from primitive cyathaspids, but its rela- tionship to the latter is unlikely. There is, at present, no very close approach between Cyathaspididae and Traquairaspididae. SUMMARY The Cyathaspididae, a family of Agnatha from the Silurian and Early Devonian, are considered from many points of view. The structure of their dermal skeleton is described, including a compre- hensive account of its histology. The evidences for internal struc- ture are discussed, and the reconstructions by Stensio of the internal anatomy are criticized. It is concluded that the order Heterostraci, to which Cyathaspididae belong, is a very primitive group of verte- brates, not closely related to modern cyclostomes, or to fossil Osteo- straci and Anaspida. The systematic revision is concerned primarily with North Amer- ican forms, but all genera and species are included. The family is divided into the following subfamilies and genera: DENISON: THE CYATHASPIDIDAE 467 Tolypelepidinae: Tolypelepis. Cyathaspidinae: Ptomaspis, Cyathaspis, Archegonaspis, Seretaspis, Vernon- aspis, Pionaspis (gen. nov.), and Listraspis (gen. nov.). Irregulareaspidinae: Dikenaspis, Dinaspidella, and 1 r regular easpis. Poraspidinae: Poraspis, Americaspis, Homalaspidella, Ariaspis, Anglaspis, and Alloc ryptaspis. Ctenaspidinae: Ctenaspis. The following generic revisions are made: Diplaspis acadica (Matthew) is referred to Cyathaspis; Cyathaspis vaningeni Bryant is referred to Vernonaspis; Aequiarchegonaspis Stensio and Lauaspis Stensio are referred to Archegonaspis; Fraenkelaspis Stensio is re- ferred to Anglaspis. The following new species are described : Vernonaspis bryanti from the Late Silurian of New Jersey, New York and Pennsylvania; V. sek- wiae from the Late Silurian, and Anglaspis expatriata from the Early Devonian of Northwest Territories, Canada; Pionaspis planicosta, P. acuticosta, and Listraspis canadensis from the Early Devonian of British Columbia; and Americaspis claypolei from the Late Silurian of New York. The geologic age of the various cyathaspid occurrences is consid- ered and displayed in a correlation chart. Their habitat is shown to have been first in the seas, though many later species lived in mar- ginal, often brackish habitats, and in fresh-water streams. The loco- motor, sensory, respiratory, and feeding adaptations are discussed. In the evolution of the family, the primitive condition is believed to be a shield formed by the fusion of numerous small tesserae. In later forms the shield is composed of large plates, formed only when the animal had reached full size. The derivation of Pteraspididae from Cyathaspididae is defended, and involves the subdivision of the shield into plates that can form early and grow with the animal. The subfamilies of Cyathaspididae are believed for the most part to be distinct phyletic lines, with the Tolypelepidinae including the ancestral stock, the Cyathaspidinae representing a central line lead- ing to pteraspids, and the other subfamilies diversely specialized side lines. REFERENCES Allen, J. R. L., and Tarlo, L. B. 1963. The Downtonian and Dittonian facies of the Welsh borderland. Geol. Mag., 100, pp. 129-155, figs. 1-4. Balabai, P. P. 1956. [On the phylogeny of the Agnatha.] Zool. Zhur., Moscow, 35, pp. 874- 890, figs. 1-12. (Russian, with English abstract in supplement, p. 9.) Ball, H. W., and Dinley, D. L. 1961. The Old Red Sandstone of Brown Clee Hill and the adjacent area. I. Stratigraphy. Bull. Brit. Mus. (Nat, Hist.), Geol., 5, pp. 177-242, figs. 1-2, 1 table, 1 map. Beerbower, James R., and Hait, Mortimer H., Jr. 1959. Silurian fish in northeastern Pennsylvania and northern New Jersey. Proc. Penna. Acad. Sci., 33, pp. 198-203, figs. 1-2. Bolau, Edmund 1951. Das Sinnesliniensystem der Tremataspiden und dessen Beziehungen zu anderen Gefasssystemen des Exoskeletts. Acta Zool., 32, pp. 31-40, figs. 1-9. Boucot, Arthur J., and Pankiwskyj, Kost 1962. Llandoverian to Gedinnian stratigraphy of Podolia and adjacent Mol- davia. 2. Internat. Arbeitstagung fiber die Silur/Devon-Grenze und die Stratigraphie von Silur und Devon. Bonn-Bruxelles (1960), Symposiums- Band, pp. 1-11, figs. 1-2. Bryant, William L. 1926. On the structure of Palaeaspis and on the occurrence in the United States of fossil fishes belonging to the family Pteraspidae. Proc. Amer. Phil. Soc, 65, pp. 256-271, figs. 1-3, pis. 1-4. 1935. Cryplaspis and other Lower Devonian fossil fishes from Beartooth Butte, Wyoming. Proc. Amer. Phil. Soc, 75, pp. 111-128, figs. 1-2, pis. 1-18. Bystrow, A. P. 1955. [The microscopic structure of the shields of jawless vertebrates of the Silurian and Devonian.] Mem. vol. A. S. Berg, Akad. Nauk, SSSR., pp. 472- 523, figs. 1-46. (In Russian.) 1959. The microstructure of skeleton elements in some vertebrates from Lower Devonian deposits of the USSR. Acta Zool., 40, pp. 59-83, figs. 1-15. Clarke, John M. 1907. The Eurypterus shales of the Shawangunk Mountains in eastern New York. Bull. New York State Mus., 107, pp. 295-310, pis. A-B, 1-8. Claypole, E. W. 1885. On the recent discovery of pteraspidian fish in the Upper Silurian rocks of North America. Quart. Jour. Geol. Soc. London, 41, pp. 48-64, figs. 1-8. 468 DENISON: THE CYATHASPIDIDAE 469 1892. On the structure of the American pteraspidian, Palaeaspis (Claypole); with remarks on the family. Quart. Jour. Geol. Soc. London, 48, pp. 542-561, figs. 1-8. Cope, E. D. 1889. Synopsis of the families of Vertebrata. Amer. Nat., 23, pp. 849-877. Denison, Robert H. 1947. The exoskeleton of Tremataspis. Amer. Jour. Sci., 245, pp. 337-365, figs. 1-13, pis. 1-3. 1953. Early Devonian fishes from Utah. Part II. Heterostraci. Fieldiana: Geol., 11, no. 7, pp. 291-355, figs. 61-85. 1956. A review of the habitat of the earliest vertebrates. Fieldiana: Geol., 11, no. 8, pp. 359-457. 1960. Fishes of the Devonian Holland Quarry shale of Ohio. Fieldiana: Geol., 11, no. 10, pp. 555-613, figs. 117-149. 1961. Feeding mechanisms of Agnatha and early Gnathostomes. Amer. Zool., 1, pp. 177-181, figs. 1-4. 1963. New Silurian Heterostraci from southeastern Yukon (with Introduction by H. R. Hovdebo, A. C. Lenz, and E. W. Bamber). Fieldiana: Geol., 14, no. 7, pp. 105-141, figs. 58-82. 1964. The early history of the vertebrate calcified skeleton. Clinical Ortho- paedics and Related Research, no. 31, pp. 141-152, figs. 1-9. Dijkgraaf, S. 1963. The functioning and significance of the lateral-line organs. Biol. Rev., 38, pp. 51-105, figs. 1-16, pis. 1-2. Dineley, D. L. 1953. Notes on the genus Corvaspis. Proc. Roy. Soc. Edinburgh, (B), 65, pp. 166-181, figs. 1-16, pis. 1-2. Fisher, Donald W. 1957. Lithology, paleoecology and paleontology of the Vernon Shale (Late Silu- rian) in the type area. Bull. New York State Mus., 364, pp. 3-31, figs. 1-2, pis. 1-3. Flower, R. H., and Wayland-Smith, R. 1952. Cyathaspid fishes from the Vernon Shale of New York. Bull. Mus. Comp. Zool., 107, pp. 355-387, figs. 1-2, pis. 1-8. Gross, Walter 1947. Die Agnathen und Acanthodier des obersilurischen Bevrichienkalks. Palaeontographica, 96, Abt. A, pp. 91-161, figs. 1-32, pis. 20-28. 1956. Uber Crossopterygier und Dipnoer aus dem baltischen Oberdevon im Zusammenhang einer vergleichenden Untersuchung des Porenkanalsystems palaozoischer Agnathen und Fische. Handl. K. Svensk. Vetenskapsakad., (4), 5, nr. 6, pp. 1-140, figs. 1-124, pis. 1-16. 1961. Aufbau des Panzers obersilurischer Heterostraci und Osteostraci Nord- deutschlands (Geschiebe) und Oesels. Acta Zool., 42, pp. 73-150, figs. 1-26. Heintz, Anatol 1933. Neuer Fund von Archegonaspis in einem obersilurischen Geschiebe. Zeits. Geschiebeforsch., 9, pp. 123-131, figs. 1-5. 1938. tlber die altesten bekannten Wirbeltiere. Naturwiss., 26, pp. 49-58, figs. 1-4. 1962. Les organes olfactifs des Heterostraci. In Problemes actuels de Paleon- tologie (evolution des vertebres). Coll. Internat., Centre Nat. Rech. Sci., 104, pp. 13-29, figs. 1-8. 470 FIELDIANA: GEOLOGY, VOLUME 13 Hertwig, Oscar 1876. Ueber das Hautskelet der Fische. Morph. Jahrb., 2, pp. 329-391, pis. 23-28. Holmgren, Nils 1942. General morphology of the lateral sensory line system of the head in fish. Handl. K. Svensk. Vetenskapsakad., (3), 20, nr. 1, pp. 1-46, figs. 1-27. Hoskins, Donald M. 1961. Stratigraphy and paleontology of the Bloomsburg formation of Pennsyl- vania and adjacent states. Bull. Penna. Topog. Geol. Surv., G 36, pp. 1-124, figs. 1-6, pis. 1-7. Huxley, Thomas H. 1858. On Cephalaspis and Pteraspis. Quart. Jour. Geol. Soc. London, 14, pp. 267-280, pis. 14-15. Jaeger, Hermann 1962. Das Silur (Gotlandium) in Thuringen und am Ostrand des rheinischen Schiefergebirges (Kellerwald, Marburg, Giessen). 2. Internat. Arbeitstagung iiber die Silur /Devon-Grenze und die Stratigraphie von Silur und Devon. Bonn-Bruxelles (1960), Symposiums-Band, pp. 108-135, tables 1-3. Jaekel, Otto 1894. [Review of Claypole, E. W., 1892.] Neues Jahrb. Min. Geol. Pal., 1894, Bd. 2, Ref., pp. 466-467. 1903. Die Organisation und systematische Stellung der Asterolepiden. Zeits. Deutsch. Geol. Ges., 55, pp. 41-60, figs. 1-8. 1911. Die Wirbeltiere. pp. i-viii+1-252, 281 figs. Berlin. Karatajute-Talimaa, V. 1962. Description of remains of Downtonian Agnatha from Lithuania. Lietu- vos TSR ma Geologijos ir Geografijos Institutas Moksliniai Pranesimai, Geol.-Geogr., 14, I sas., pp. 45-59, figs. 1-2, pi. 1 (Russian with English and Latvian summaries). KlAER, JOHAN 1924. The Downtonian fauna of Norway. I. Anaspida. With a geological introduction. Skr. Videnskapsselsk., I, Mat.-Nat. Kl., 1924, no. 6, pp. 1-139, figs. 1-50, pis. 1-14. 1928. The structure of the mouth of the oldest known vertebrates, pteraspids and cephalaspids. Palaeobiologica, 1, pp. 117-134, figs. 1-7, pis. 12-13. 1930. Ctenaspis, a new genus of cyathaspidian fishes. Skr. Svalbard Ishavet, 33, pp. 1-7, figs. 1-4. 1932. (A. Heintz, ed.) The Downtonian and Devonian vertebrates of Spits- bergen. IV. Suborder Cyathaspida. Skr. Svalbard Ishavet, 52, pp. 7-26, figs. 1-12, pis. 1-11. Kiaer, Johan, and Heintz, Anatol 1935. The Downtonian and Devonian vertebrates of Spitsbergen. V. Sub- order Cyathaspida. Part I. Tribe Poraspidei, Family Poraspidae Kiaer. Skr. Svalbard Ishavet, 40, pp. 1-138, figs. 1-57, pis. 1-40. Kilfoyle, Clinton F. 1954. Catalogue of type specimens of fossils in the New York State Museum. Supplement 4. Bull. New York State Mus., 348, pp. 1-719. 1959. Catalogue of type specimens of fossils in the New York State Museum. Supplement 5. Bull. New York State Mus., 376, pp. 1-134. DENISON: THE CYATHASPIDIDAE 471 Lamont, A. 1952. Ecology and correlation of the Pentlandian — a new subdivision of the Silurian system in Scotland. Rept., Internat. Geol. Congr., 18 (Great Brit- ain), pt. 10, pp. 27-32. Lankester, E. R. 1868-1870. A monograph of the fishes of the Old Red Sandstone of Britain. Part I. The Cephalaspidae. Mon. Palaeontogr. Soc, pp. 1-62, figs. 1-33, pis. 1-14. Lehman, Jean-Pierre 1937. Les poissons du Downtonien de la Scanie (Suede). Mem. Fac. Sci., Univ. Paris, Dipl. d'Etudes Sup., 664, pp. 1-98, pis. 1-8. Leriche, Maurice 1906. Contribution a l'etude des poissons fossiles du nord de la France et des regiones voisines. I. Les poissons siluriens et devoniens du Nord de la France. Mem. Soc. Geol. Nord, 5, pp. 13-39, figs. 1-12, pis. 1-4. Leutze, W. P. 1960. Silurian fish fossils in the Salina basin. Bull. Geol. Soc. Amer., 71, pp. 215-218, fig. 1. LlNDSTROM, GUSTAV 1895. On remains of a Cyathaspis from the Silurian strata of Gotland. Bihang K. Svensk. Vetenskapsakad. Handl., 21, Nr. 3, pp. 1-15, figs. 1-2, pis. 1-2. Mackenzie, G. S. 1951. Preliminary map: Westfield, Kings, Queens, Saint John, and Charlotte Counties, New Brunswick. Geol. Surv. Canada, Paper 51-15. Matthew, G. F. 1888. On some remarkable organisms of the Silurian and Devonian rocks in southern New Brunswick. Trans. Roy. Soc. Canada, (1), 6, sec. 4, pp. 49-62, pi. 4. Moy-Thomas, J. A. 1939. Palaeozoic fishes, pp. i-ix+ 1-149, figs. 1-32. London. Nikiforowa, Olga I., and Obut, Alexadr M. 1960. Parallelisierung des Silurs im europaischen Teil der UdSSR und in Mit- teleuropa. „ In Prager Arbeitstagung fiber die Stratigraphie des Silurs und des Devons. Ceskoslovenske Akad. Vfd, pp. 277-283, 1 chart. 1962. Zur Frage des Silur/Devon-Grenze in der UDSSR. 2. Internat. Arbeits- tagung fiber die Silur/Devon-Grenze und die Stratigraphie von Silur und Devon. Bonn-Bruxelles (1960), Symposiums-Band, pp. 175-179, 1 table. Obruchev, D. V. 1938. [Vertebrata aus dem Obersilur und Devon des Urals.] Russia, Central Geol. Prosp. Inst., Materials (gen. ser.), 2, pp. 36-43, pis. 1-2. (In Russian, German summary.) 1945. [The evolution of the Agnatha.] Zool. Zhur., Moscow, 24, pp. 257-272, figs. 1-10. (In Russian.) 1958. [On the biostratigraphy of the lower and middle Paleozoic ichthyofaunas of the USSR.] Sovietskaya Geologiya, 1958, no. 11, pp. 40-53. (In Russian, English summary.) 0rvig, Tor 1961. Notes on some early representatives of the Drepanaspida (Pteraspido- morphi, Heterostraci). Ark. Zool., (2), 12, nr. 33, pp. 515-535, figs. 1-10. 472 FIELDIANA: GEOLOGY, VOLUME 13 P'an, Kiang 1962. A new Silurian fish from Nanking, China. (Chinese with English sum- mary). Acta Palaeont. Sinica, 10, no. 3, pp. 402-409, fig. 1, pi. 1. Pander, C. H. 1856. Monographie der fossilen Fische des silurischen Systems der russisch- baltischen Gouvernements. pp. i-x+1-91, pis. 1-9. St. Petersburg. Patten, William 1912. The evolution of the vertebrates and their kin. pp. i-xxi+ 1-486, figs. 1- 309. Philadelphia. Rohon, J. V. 1893. Die obersilurischen Fisehe von Oesel. II. Selachii, Dipnoi, Ganoidei. Pteraspidae und Cephalaspidae. Mem. Acad. Imp. Sci., Saint-Petersbourg, (7), 41, no. 5, pp. 1-124, figs. 1-19, pis. 1-3. Romer, Alfred S. 1946. The early evolution of fishes. Quart. Rev. Biol., 21, pp. 33-69, figs. 1-31. RUEDEMANN, RUDOLF 1916. Paleontological contributions from the New York State Museum. Spathi- ocaris and the Discinocarina. Bull. New York State Mus., 189, pp. 98-112, figs. 34-37, pis. 32-33. Save-Soderbergh, Gunnar 1941. Notes on the dermal bones of the head in Osteolepis macrolepidoius Ag. and the interpretation of the lateral line system in certain primitive verte- brates. Zool. Bidrag Uppsala, 20, pp. 523-541, figs. 1-9, pis. 1-3. Sokolov, Boris S. 1960. Stratigraphisehe Grenzen der Systeme des alteren Palaozoikums auf dem Gebiet der russischen Tafel. In Prager Arbeitstagung iiber die Stratigraphie des Silurs und des Devons. Ceskoslovenske Akad. Ved, pp. 97-103, 1 chart. Stensio, E. A. 1926. On the sensory canals of Pteraspis and Palaeaspis. Ark. Zool., 18A, no. 19, pp. 1-14, figs. 1-7. 1927. The Downtonian and Devonian vertebrates of Spitsbergen. Part I. Family Cephalaspidae. Skr. Svalbard Nordishavet, 12, pp. i-xii+ 1-391, figs. 1-103, pis. 1-112. 1932. The cephalaspids of Great Britain. Brit. Mus. (Nat. Hist.), pp. i-xiv-(- 1-220, figs. 1-70, pis. 1-66. 1958. Les cyclostomes fossiles ou ostracodermes. In Grasse, P. -P., Traite Zool., 13, fasc. 1, pp. 173-425, figs. 107-219. Swartz, C. K., and Swartz, F. M. 1931. Early Silurian formations of southeastern Pennsylvania. Bull. Geol. Soc. Amer., 42, pp. 621-662, figs. 1-2. Swartz, F. M. 1946. Faunal development, conditions of deposition, and paleogeography of some Appalachian mid-Paleozoic sediments. An. Segundo Congr. Panamer. Engenharia, Minas, Geol., 3, pt. 2a, pp. 9-33. Tarlo, L. B. 1960. The Downtonian ostracoderm Corvaspis kingi Woodward, with notes on the development of dermal plates in the Heterostraci. Palaeontology, 3, pp. 217-226, figs. 1-7, pis. 37-38. 1962a. The classification and evolution of the Heterostraci. Acta Palaeont. Polonica, 7, pp. 249-281, figs. 1-15. DENISON: THE CYATHASPIDIDAE 473 1962b. Lignees evolutive? chez les ostracodermes heterostraces. In Problemes actuels de Paleontologie (evolution des vertebres). Coll. Internal., Centre Nat. Rech. Sci., 104, pp. 31-36, fig. 1. Thorsteinsson, R. 1958. Cornwallis and Little Cornwallis Islands, District of Franklin, North- west Territories. Mem. Geol. Surv. Canada, 294, pp. 1-134, figs. 1-5, pis. 1-8. 1963. Prince Alfred Bay. In Fortier, Y. O. et al., Geology of the north-central part of the arctic archipelago, Northwest Territories (Operation Franklin). Mem. Geol. Surv. Canada, 320, pp. 221-232, fig. 9. Thorsteinsson, R. and Tozer, E. T. 1963. Geology of northern Prince of Wales Island and northwestern Somerset Island. In Fortier, Y. O. et al., Geology of the north-central part of the arctic archipelago, Northwest Territories (Operation Franklin). Mem. Geol. Surv. Canada, 320, pp. 117-129. Traquair, R. H. 1899. Report on fossil fishes collected by the Geological Survey of Scotland in the Silurian rocks of the south of Scotland. Trans. Roy. Soc. Edinburgh, 39, pp. 827-864, figs. 1-6, pis. 1-5. Wangsjo, Gustav 1952. The Downtonian and Devonian vertebrates of Spitsbergen. IX. Mor- phologic and systematic studies of the Spitsbergen cephalaspids. Skr. Norsk Polarinst., 97, pp. 1-616, figs. 1-108, pis. 1-118. Watson, D. M. S. 1954. A consideration of ostracoderms. Phil. Trans. Roy. Soc. London, (B), 238, pp. 1-25, figs. 1-6. White, E. I. 1935. The ostracoderm Pteraspis Kner and the relationships of the agnathous vertebrates. Phil. Trans. Roy. Soc. London, (B), 225, pp. 381-457, figs. 1-97, pis. 25-27. 1946. The genus Phialaspis and the "Psammosteus Limestones." Quart. Jour. Geol. Soc. London, 101, pp. 207-242, figs. 1-60, pis. 12-13. 1950. The vertebrate faunas of the Lower Old Red Sandstone of the Welsh borders. Bull. Brit. Mus. (Nat. Hist.), Geol., 1, pp. 51-67, figs. 1-2. 1958. Original environment of the craniates. In Westoll, T. S. (ed.), Studies on fossil vertebrates, pp. 212-234, figs. 1-8. London. Wills, Leonard J. 1935. Rare and new ostracoderm fishes from the Downtonian of Shropshire. Trans. Roy. Soc. Edinburgh, 58, pp. 427-447, figs. 1-4, pis. 1-7. Woodward, A. S. 1891. Catalogue of the fossil fishes in the British Museum (Natural History). Part II. Brit. Mus. (Nat. Hist.), pp. i-xliv+1-567, figs. 1-57, pis. 1-16. London. Zych, W. 1931. Fauna Ryb Dewonu i Downtonu Podola. Pteraspidomorphi: Hetero- straci. Czesc 1A. pp. 1-91, drawings 1-52, photographs 1-31. Lwow. Publication 978