Early Devonian fishes from Utah : Osteostraci

i

o

II B HAHY

OF THE
UNIVERSITY
Of ILLINOIS

550.5
Fl

GEOLOli*

UNIVERSITY OF

ILLINOIS LIBRARY

ATURBANA-CHAMPA1GN

GEOLOGY

Return this book on or before the
Latest Date stamped below.

GEOLOGY LIBRARY

University of Illinois Library

dec 30 \m :mi

lie st9iis »ar!5

DEC 29 *l68
FES 3 1967

JUL J j

mftt

<s

v

\

\

198^

>..

MAY

3268

J972

DeL&sW

?,

HOVj^'

982

«!

172004

L161— H41

EARLY DEVONIAN FISHES
FROM UTAH

PART I. OSTEOSTRACI

ROBERT H. DENISON

Curator of Fossil Fishes

FIELDIANA: GEOLOGY

VOLUME 11, NUMBER 6

Published by

CHICAGO NATURAL HISTORY MUSEUM

JUNE 16, 1952

THE LIBRARY OF THE

JUL 22 1952

PRINTED IN THE UNITED STATES OF AMERICA
BY CHICAGO NATURAL HISTORY MUSEUM PRESS

l\dj

INTRODUCTION

The Early Devonian fishes of northeastern Utah are the first
vertebrate fauna of this age discovered in the United States. They
were mentioned in an unpublished thesis by Cooley (1928), in brief
abstracts by Branson (1929) and Branson and Mehl (1930), and
finally described by Branson and Mehl (1931). The fishes upon
which the above reports were based came from Blacksmith Fork
canyon, southeast of Logan, in Cache County. They were collected
in loose blocks of talus derived from what these authors considered
to be the lower part of the Jefferson formation. Since the speci-
mens were obtained on a slope corresponding closely with the dip
of the formations, the determination of the horizon can be con-
sidered to be reasonably accurate.

From the lowermost part of the formation Branson and Mehl
(1931) described the following fishes: Camptaspis utahensis, con-
sidered to be an "acanthaspid" but, as will be shown below, really
Cephalaspis; " Aspidichthys" sp., an undeterminable phlyctaenaspid ;
Cardipeltis wallacii, a peculiarly specialized member of the Heter-
ostraci ; and Glossoidaspis giganteus, believed by Bryant to belong to
Pteraspis. These forms suggest an Early Devonian age. From a
higher part of the slope, and probably from a stratigraphic level
200 feet higher, they reported Dinichthys{1) jeffersonensis and
Ptyctodus cf. calceolus; these suggest a Middle or Late Devonian
age.

Williams (1948, pp. 1138-1141) divided the Devonian rocks in
this region into two formations, separated by a distinct discon-
formity. The lower unit was called the Water Canyon formation
and was assigned to the Early Devonian on the basis of its only
identifiable fossils, the fishes. For the upper and thicker Devonian
unit, the name Jefferson formation was retained; its invertebrates
suggest a Late Devonian age. It is probable that the Dinichihys{t)
and Ptyctodus described by Branson and Mehl came from the lower
part of this formation. Williams further divided the Water Canyon
formation into two members, apparently distinguished on lithological
grounds alone. As far as can be determined from his report, the
:wo members do not occur together in any one locality, so it is

265

266 FIELDIANA: GEOLOGY, VOLUME 11

probable that they are not distinct chronologically but represent
different facies deposited more or less simultaneously.

The fishes to be described in this and subsequent articles are
those of the Early Devonian, Water Canyon formation. They were
obtained for Chicago Natural History Museum by me in June,
1949, and by a party consisting of Mr. Bryan Patterson, Mr. and
Mrs. William D. Turnbull, and myself in June and July, 1950.
Collections were made at the following localities in Cache County,
Utah, all except the last one in the Bear River Range.

A. Blacksmith Fork, east side of Left Fork, NE. 34, sec. 3,
T. 10 N., R. 2 E.; also SE. %, sec. 34, and NW. M, sec. 35, T. 11
N., R. 2 E. Overlying the Silurian Laketown dolomite, at the base
of the Water Canyon formation, is a white-weathering, ledge-forming
limestone, then a poorly exposed, grayish-buff weathering limestone
with fish remains, in turn overlaid by a resistant, white-weathering
limestone. The canyon wall slopes at about 25°, corresponding
approximately with the dip of the strata. The less resistant, buff-
colored, fossiliferous limestone rarely outcrops, so that most of the
fishes were obtained from loose pieces of rock lying on the slope.
Abundant but fragmentary fish remains were obtained in place in
a small gully about 200 yards southwest of the northeast corner
of section 3, in a 4-inch shaly limestone and a 4-inch limy sandstone
lying below the upper white-weathering limestone.

B. Blacksmith Fork, immediately south of forks, south of
center, sec. 3, T. 10 N., R. 2 E. A small slope, approximately
corresponding with the dip, is formed by the resistant Laketown
dolomite, with a thin veneer of the lower, white-weathering limestone
of the Water Canyon formation, and near the top some of the
grayish-buff, fish-bearing limestone. A few specimens were obtained
from loose blocks of the latter.

C. Blacksmith Fork, immediately northwest of the forks, at
about the center of sec. 3, T. 10 N., R. 2 E. The canyon slope
on this side of the stream cuts across the section. A few specimens
were obtained from loose blocks of grayish-buff weathering limestone,
presumably from the same fossiliferous horizon as at localities A
and B, but none were obtained in place.

D. Blacksmith Fork, Left Fork, in the two "hollows" or side
canyons downstream from Charlie's Hollow, in SW. x/i, sec. 26,
T. 11 N., R. 2 E. As in localities A and B, fishes were obtained in
loose rocks, mostly on dip slopes. They occur in the same grayish-

DENISON: EARLY DEVONIAN FISHES 267

buff weathering limestone lying between the more resistant white-
weathering limestones at the base of the Water Canyon formation.

E. Green Canyon, northwest side, in NW. Y/i, sec. 16, T. 12 N.,
R. 2 E., about three miles northeast of Logan. A few fragments of
fishes were found in place near the base of the Water Canyon forma-
tion.

F. Water Canyon, a tributary of Green Canyon, and the type
locality of the formation. The only fossils collected were fragments
of fishes in stream boulders, presumably from this formation, in
NE. YA, sec. 9, T. 12 N., R. 2 E.

G. Cottonwood Canyon, east side, about six and one-half miles
northeast of Logan, north-central part, sec. 19; W. 3^, sec. 18; and
W. Y2, sec. 7; all in T. 13 N., R. 3 E. The lower part of the Water
Canyon formation resembles that in Blacksmith Fork, with fishes
occurring in a buff-weathering limestone, lying between harder,
white-weathering limestones. A few poorly preserved fish fragments
occur in a reddish, sandy limestone and sandstone at a slightly
higher level. Fossils were obtained in place as well as in loose
blocks on the dip slope that forms the east side of this canyon.

H. Crest of ridge, one-half to one mile south-southeast of
Naomi Peak, about seven miles northeast of Logan, and in E. y%,
sec. 1, T. 13 N., R. 2 E. Fossil fishes were obtained for the most
part in place in what is presumed to be the higher levels of the
Water Canyon formation, 100 feet or more above the base. The
strata are mostly sandstones and sandy limestones.

I. Northwest of Dry Lake, on the east edge of the Wellsville
Range, about four miles southwest of Hyrum, in SW. \i, sec. 29,
T. 10 N., R. 1 W. The stratigraphic sequence was not determined
here, but fishes were obtained in a buff-weathering, sandy limestone,
similar to that in which they are found at other localities.

The fishes of the Water Canyon formation will be described in
a series of papers of which this first one will deal with the Osteostraci.
Others on the Heterostraci, Arthrodira, Crossopterygii, and Dipnoi
are being prepared and will appear at a later date. A general con-
sideration of the fauna will be postponed until the various groups
have been described.

Comparison with other Early Devonian fishes has involved
careful study of the nearly related fauna from Beartooth Butte,
Wyoming, described by Bryant (1932, 1933). A number of additions
and corrections to Bryant's descriptions have appeared, and will
be incorporated in these publications.

268 FIELDIANA: GEOLOGY, VOLUME 11

At this time I wish to acknowledge the kindness of Dr. Glenn L.
Jepsen in lending for study the specimens of Cephalaspis wyoming-
ensis in the collection of Princeton University.

THE GROWTH OF OSTEOSTRACI

The question of whether Osteostraci could continue to grow after
their exo- and endo-skeletons had formed continues to enter into
any study of the group. Even in a purely systematic study, where
it is often overlooked, it has an important bearing on the range of
size to be expected in any species. Those who have examined this
problem (Gross, 1935, p. 9; Heintz, 1939, p. 25; Westell, 1942, p.
118, and 1945, pp. 345-346; Denison, 1947, pp. 358-365) have come
to the conclusion that the shield was formed only after the animal
was fully grown. This inference is derived mainly from the fact
that the osteostracian shield is a sutureless capsule in which indica-
tions of resorption and redeposition of bone are rare. The growth
of the Osteostraci has been reconsidered in connection with the
present work, and evidence of at least limited growth in certain
Osteostraci will be discussed.

Gross (1935, p. 9) has called attention to the limited size range
of individuals belonging to a single species of Cephalaspidae and
has considered this to be evidence against the growth of the cepha-
laspid shield. If there were growth, it should be demonstrable
statistically, but since adequate data are scarce, it has been found
necessary to use a rather crude statistical approach. A coefficient
of variation (V) has been computed (in Tremataspis mammillata
and Hemicyclaspis murchisoni) or estimated from the observed range
and size of sample, using the method proposed by Simpson (1941,
p. 800). The data have been derived from Heintz (1939), Stensio
(1927, 1932), Woodward (1891), and from original observations.
The results are presented below:

Number of Number of Coefficient of

Species specimens occurrences variation (V)

Tremataspis mammillata 23 1 6.2

Aceraspis robustus 40 1 7

Micraspis gracilis 30 1 9

Thyestes egertoni 40 1 8

Didymaspis grinrodi 16 3 9

Hemicyclaspis murchisoni 8 1 or more 16

Cephalaspis salweyi 14 4 17

Cephalaspis whitei 21 4 or more 11

Cephalaspis pagei 32 5 24

C. powriei v. brevicornis 8 3 16

Cephalaspis hoeli 19 9 18

Cephalaspis heintzii 3 3 36

LIBRARY

UNIVERSITY OF ILLINOIS

DENISON: EARLY DEVONIAN FISHES 269

In Tremataspis mammillata, Aceraspis robustus, Micraspis
gracilis, Thyestes egertoni, and Didymaspis grinrodi the coefficient of
variation is relatively low, and well within the usual range for pure
samples of mammals of uniform age (Simpson and Roe, 1939,
p. 123). Clearly there is no indication of exoskeletal growth here.
In the first four species, however, the individuals may have been
obtained from a single stratigraphic horizon at one locality, so
there is a possibility that in each case they represent a school of
fishes belonging to a single age group. The latter is highly improbable
in the case of Didymaspis grinrodi, since the sixteen specimens were
obtained from three localities and probably from different strati-
graphic levels. The relatively high value of the coefficient of varia-
tion in Hemicyclaspis murchisoni suggests that the shield may have
grown, but the data are inadequate to demonstrate this clearly. In
the few species of Cephalaspis where there are sufficient data to
estimate the coefficient of variation, the value is moderately to very
high, but where it has been possible to obtain measurements of a
number of specimens from a single occurrence, the estimated value
of the coefficient of variation is low (between 5 and 8). The purity
of the samples is questionable, therefore, especially in C. hoeli and
C. heintzii, where the few specimens were distributed through as
much as 2,300 feet of sediments. In conclusion it may be said that
the few available measurements suggest that Tremataspis, Aceraspis,
Micraspis, Thyestes, and Didymaspis did not continue to grow after
the shield had formed; they are inadequate or non-committal in the
cases of Hemicyclaspis and Cephalaspis.

The problem of the growth of the osteostracian shield may be
approached from a different angle. If there were growth of the
exoskeleton, it would have to be accomplished by extensive resorption
and redeposition of the hard tissues, and evidences of these processes
should appear in thin sections. Indications of resorption have been
noted as rare occurrences by Gross (1935, p. 9), Heintz (1939, p. 25),
and Westoll (1945, p. 345), but it is probable that this process was
more common than has been suspected. In an earlier paper (Denison,
1947, pp. 358-361), I described four sections that were believed to
represent successive stages of growth of the Tremataspis exoskeleton.
A re-examination of these sections with Dr. Rainer Zangerl has
convinced me that a somewhat different interpretation is more
probably correct. The first stage (op. cit., fig. 11, A), with its
thin superficial layer and with bone only around the sensory canals,
s surely an early stage, as it was considered to be before. But the
ast two stages (fig. 11, C-D), and possibly the last three, are now

270 FIELDIANA: GEOLOGY, VOLUME 11

considered to show varying degrees of resorption rather than of
deposition. The bony part of the exoskeleton was formed by
deposition around the canals of the middle layer, and then by apposi-
tion to the inner surface of successive laminae forming the basal layer.
At full development (fig. 50, A), the laminae were penetrated by
relatively narrow canals for nerves and blood vessels. Subsequently
resorption took place along these canals, resulting in the formation
of large vacuities in the basal layer (fig. 50, B), and perhaps in its
nearly complete removal (fig. 50, C). The resorptive origin of these
large vacuities is indicated by their roughened edges.

Some of the bone of the middle layer was also eroded, but in
numerous sections of Tremataspis the superficial layer is little
touched by this process. This is an essential point, because the super-
ficial layer is continuous in this genus, and growth could not take
place until this layer had been penetrated to allow redeposition in
the zone of penetration, or to allow overgrowth and new deposition
external to it. Whatever the physiological basis of the resorption
may be, its net effect is a lightening of the armor, which, of course,
would allow the individual to swim more actively.

Since the superficial layer of Tremataspis was not resorbed to
any great extent, it may be concluded that the exoskeleton did not
grow in this genus. But the same difficulty does not exist in those
Osteostraci whose superficial layer is discontinuous or absent. In
these cases resorption and redeposition could be limited to the bony
part of the exoskeleton and could be accompanied by an outward
growth of the dermis over the original external surface of the skeleton.
This process would result in an increase in size, and could properly
be referred to as growth. That it does occur in Osteostraci is clearly
demonstrated by a few examples, of which the most striking is a
section of Cephalaspis sp. (fig. 51) recently figured by Orvig (1951).
This is a tuberculate form with dentine restricted to the tubercles.
Clearly shown is the partial resorption of dentine tubercles of an
earlier generation (fig. 51, t{) and the new formation of dentine
external to them (fig. 51, k). Similar examples have been figured
by Stensio (1932, pi. 63, fig. 3) in Hemicyclaspis{t) lighibodii, and
(op. cit., pi. 66, fig. 1) in Cephalaspis salweyi. Wills (1935, pi. 5,
fig. 2) figures an old tubercle overgrown by a new one in Tesseraspis
tessellata, a form that he considered to be a drepanaspid, although
it is more probably a cephalaspid.

These few examples are a clear demonstration of at least limited
growth of the exoskeleton in certain Osteostraci. They indicate

Fig. 50. Transverse sections of exoskeleton of dorsal shield of Tremataspis
mammillata (X 100). A, Fully developed exoskeleton. B, Resorption cavities
formed in basal layer. C, Basal layer, mostly eroded. AVC, ascending vascular
?anal; BL, basal layer; DC, tubules in dentine-like layer; ML, middle layer;
R, resorption cavity; RC, canal of lower vascular plexus; SC, sensory canal; SL,
;uperficial layer; VCS, canal of subepidermal vascular plexus; X, septum of
ensory canal.

271

272

FIELDIANA: GEOLOGY, VOLUME 11

os can.asc

Fig. 51. Transverse section of rim of cephalic shield of Cephalaspis sp. (X 60)
from Orvig, 1951, fig. 11A. U, dentine tubercle of first generation; h, tubercle
of second generation.

that one must expect considerable intraspecific variation in size,
at least in species where the superficial layer is not continuous.

Cephalaspis utahensis (Branson and Mehl)

This species was originally described as Camptaspis utahensis
(Branson and Mehl, 1931, p. 510) and considered to be an "acan-
thaspid" in which the small plates (polygonal areas) had not yet
fused to form the usual arthrodire plates. New material shows
beyond any question that Camptaspis is not an arthrodire, but a
cephalaspid indistinguishable from Cephalaspis.

Occurrence and material. — The original description was based
upon two specimens from Blacksmith Fork, Utah, in the University
of Missouri collection, the type, no. 592, and a doubtfully referred
specimen, no. 598; these specimens have not been available to me
in this study. In the Chicago Natural History Museum collection
there are approximately forty specimens, obtained from the lower
part of the Water Canyon formation at Blacksmith Fork (Localities
A, B, D), Water Canyon (Locality F), and Cottonwood Canyon
(Locality G), all in Cache County, Utah. Of these, the following
have been particularly useful: PF 472, the right half of the dorsal
shield, preserving the median posterior spine but lacking the medial
part anteriorly (fig. 52, B) ; PF 474, a much-weathered central part
of a cephalic shield, lacking the rim but preserving parts of both
dorsal and ventral shields (fig. 52, A); PF 323, 473, shield rims
exposed on the ventral side (fig. 52, E). In addition there are

Fig. 52. Cephalaspis utahensis. A, Weathered and incomplete dorsal shield,
PF 474 (X M). B, Right half of dorsal shield, PF 472 (X M). C, Ornamentation
of dorsal shield, PF 478 (X 5). D, Base of left cornu, ventral view, PF 468 (X H).
E, Rim of shield, ventral view, PF 473 (X H).

273

274 FIELDIANA: GEOLOGY, VOLUME 11

a number of cornua, parts of shield rims, and fragments of dorsal
and ventral shields.

Description. — Since no specimen is complete enough to furnish
the desired measurements, the following dimensions are estimated
from the reconstruction, figure 53, based mainly on PF 472, 473,
474, and 481:

mm.

Maximum length of dorsal shield 110

Maximum breadth (near middle of cornua) 122

Length of cornu (tip to pectoral sinus) 62

A few specimens differ noticeably in size. Thus, from locality A,
PF 323 is about 7 per cent smaller, and PF 457 is probably larger;
from locality G, PF 634 and 463 are about 20 per cent smaller.
These differences are not sufficient to warrant specific differentiation,
especially since it is probable that the cephalaspid shield can grow.

The shield (fig. 53) is of moderate width. The anterior end is
sharply rounded, with a fairly distinct rostral angle. The lateral
margin is gently rounded, but with less curvature opposite the
orbits. The cornua are long, moderately robust, blunt-tipped, and
with a strong inward curvature. The interzonal region of the dorsal
shield is rather long and terminates in a rounded point, surmounted
by a low median crest (fig. 52, B). The pectoral sinuses are rather
deep, well marked medially, but with no trace of a pectoral angle
on the cornua.

The position of the orbits, as shown in the restoration (fig. 53)
is indicated by PF 472 and 474 (fig. 52, A, B). Their shape is
shown by a small fragment of a cephalic shield, PF 461, in which
the orbit is 8.9 mm. long and 7.4 mm. broad. The orbit is surrounded
by a pronounced, elevated ridge that is extended anteriorly around
the circumnasal fossa, and posteriorly along the dorsal field. Neither
the pineal nor the hypophysial foramina are known, although the
pineal canal is indicated on PF 474. The latter specimen, though
much weathered, shows the position of the anterior end of the dorsal
field (fig. 52, A). The lateral fields are of moderate length for
Cephalaspis; although they do not extend as far anteriorly as in
some species, posteriorly they extend onto the cornua a considerable
distance, terminating in a rounded point.

The ornamentation of the dorsal shield (fig. 52, C) is character-
istic of the species. The surface is divided into irregular, curved-
sided, polygonal areas by circumareal canals. Each area is occupied

DENISON: EARLY DEVONIAN FISHES

275

Fig. 53. Restoration of dorsal shield of Cephalaspis utahensis (X %).

by about four to seven projections, generally irregular in shape,
often deeply notched, but sometimes partly or completely isolated
as distinct, rounded tubercles. In thin sections the minute structure
is poorly preserved, but apparently the projections and tubercles
are capped with dentine. It is possible that intra-areal canals lay
in the depressions between the projections and tubercles. In certain
parts of the dorsal shield the ornamentation is modified. Thus, in
the raised rim around the orbits the projections, though still irregular
in shape, have a radial arrangement. The posterior dorsal crest is

ined with three rows of rather regularly elongated tubercles. On

he dorsal side of the shield rim and cornua the projections and

ubercles are arranged in a distinct pattern.

The ornamentation of the ventral shield is usually quite different

figs. 52, D; 54). Around the edge of the rim are low, elongate
lidges composed of the superficial layer. The ridges are deeply
(rimped, and in certain regions the projections on adjacent ridges
j )in, so that only pores are left in the superficial layer. In the

276

FIELDIANA: GEOLOGY, VOLUME 11

Fig. 54. Restoration of ventral shield of Cephalaspis utahensis (X %).

rostral region the ridges parallel the rim of the shield; lateral to the
rostral area are paired V-shaped areas in which the ridges are ar-
ranged in a V-shaped pattern; posterior to the V-shaped areas, as
far back as the base of the cornua, the ridges are diagonally placed
and are directed postero-laterally (fig. 52, D) . In general, the medial
edges of the ventral side of the rim and cornua are set with fine
denticles capped by the superficial layer. At the postero-lateral
corner of the oralo-branchial chamber the surface is broken into
polygonal areas, larger in size posteriorly and laterally (fig. 54).
On the ventral shield of a few individuals (PF 323, 324, 468, and 469)
the superficial layer and probably part of the middle layer are clearly
absent (because of weathering, non-deposition, or resorption); in
these specimens there are irregular tubercles or projections such as
occur on the dorsal side of the shield.

The canals of the lateral line system have not been recognized.

Comparisons. — Cephalaspis utahensis is distinguished from other
species of the genus by its ornament, its long, moderately robust,

DENISON: EARLY DEVONIAN FISHES 277

incurved cornua, and its rather long, broad interzonal region,
crowned posteriorly by a low median crest. None of the other
species described from North America resemble it at all closely.
The British species of Cephalaspis, with one exception, have relatively
shorter cornua (Stensio, 1932) . The exception, C. jacki (White, 1935,
fig. 5), is similar in general shape and proportions, in its long cornua,
its low postero-dorsal spine, and possibly in its ornament (the latter
is compared to that of C. salweyi, which shows some resemblance
to that of C. utahensis). But C. jacki differs in its smaller size,
more slender cornua, shorter interzonal region, and shallower pectoral
sinuses. Among the species of Cephalaspis described from Spits-
bergen (Stensio, 1927), those that have long cornua are smaller in
size and are distinguished by other features. Thus the cornua of
C. isachseni and C. spitsbergensis are relatively more slender, while
the cornua of C. acuticornis, C. eukeraspidoides, C. vogti, and C.
gracilis are straighter.

Cephalaspis brevirostris, sp. nov.

Type.— C.N.H.M.-PF 328, a dorsal shield lacking the orbital
and postero-medial region (figs. 55, 56).

Horizon. — Early Devonian, near base of Water Canyon formation.

Locality. — Blacksmith Fork (Locality A), NE. %, sec. 3, T. 10 N.,
R. 2 E., Cache County, Utah.

Referred specimen. — In addition to the type, there is only a single
specimen, PF 480, from Cottonwood Canyon (Locality G) that may
belong to this species. It is a weathered fragment of an orbit,
resembling the type in ornament.

Description. — This species is of moderate size for the genus. The
type has the following dimensions:

mm.

Length from rostrum to tip of cornu 105

Length of cornu (tip to pectoral angle) 44

Length of lateral field 54

Maximum width (at posterior part of cornua) 88

Estimated length in midline 70-75

C. brevirostris has a moderately slender shield, characterized
particularly by the incipient rostrum that protrudes slightly but
loticeably at the anterior end. Rostral processes occur also in
7. campbelltonensis and C. acutirostris, as well as in Hoelaspis and
Soreaspis, but in all of these they are pronouncedly elongate. In

278

FIELDIANA: GEOLOGY, VOLUME 11

Fig. 55. Cephalaspis brevirostris, sp. nov., type, PF 328, restored (X 1).

its outline the shield rim has slight concavities lateral to the rostrum,
then is gently convex to the tip of the cornua. The cornua are
very long and slender, tapering gradually to a moderately sharp
point; they are gently incurved and directed posteriorly. In their
shape and proportions they are most nearly comparable to the
cornua of C. jacki. The interzonal region, not preserved in the
type, was apparently slightly narrower than in C. utahensis. The
pectoral sinuses are of moderate breadth and show a barely per-
ceptible pectoral angle on the cornua. The lateral fields are similar
to those of C. utahensis, rounded anteriorly, and extending a short
distance onto the cornua, where they terminate in a rounded point.
The ornamentation of the shield consists of polygons from 1.5
to 2.4 mm. in diameter, set with minute tubercles. The tubercles
appear to lack dentine and enamel caps, and so apparently the
superficial layer is entirely absent. The polygons are, of course,

DENISON: EARLY DEVONIAN FISHES

279

Fig. 56. Type of Cephalaspis brevirostris, sp. nov., PF 328 (X 1).

demarked by circumareal sensory canals, but the presence of intra-
areal canals cannot be determined. In the rostral area the division
of the shield into polygonal areas is absent, and this is also true on
the shield rim laterad to the lateral fields as well as on the cornua
(fig. 55). The rim is delimited by a clearly marked raised edge in
the rostral area. The cornua have thick lateral and mesial rims
separated by a depressed area that is ornamented with V-shaped
"idges and rows of tubercles. No lateral line canals have been
•ecognized.

Comparisons. — Cephalaspis brevirostris is distinguished from all
other species of the genus by its small rostral process. Other espe-
< ially characteristic features are the long, slender cornua, and the
ornamentation of the dorsal shield.

280

FIELDIANA: GEOLOGY, VOLUME 11

_-NA

Fig. 57. Type of Cephalaspis wyomingensis, Princeton 13479 (X 1), rein-
terpreted from Bryant's published photograph. DF, dorsal field; HY, hypophysial
foramen; NA, nasal opening; OR, orbit; PI, pineal foramen.

Cephalaspis wyomingensis Bryant

Occurrence. — Early Devonian. Beartooth Butte formation, Bear-
tooth Butte, Park County, Wyoming; ? Water Canyon formation,
Blacksmith Fork (Locality A) and Cottonwood Canyon (Locality
G), Cache County, Utah.

Material. — The original description of this species (Bryant, 1933,
p. 312) was based upon two specimens collected by the 1932 Princeton
expedition. A fragment of the dorsal cephalic shield (Princeton
13473) was designated as type, and considering its incompleteness
it is not surprising that Bryant misinterpreted it. A specimen in
Chicago Natural History Museum (PF 148) shows beyond any doubt
that what Bryant considered to represent the rostral region is
actually the posterior part of the cephalic shield, and that he had
reversed the orientation. A reinterpretation of the type, based on
the published photograph (Bryant, 1933, pi. 20, fig. 1) is given in
figure 57. The second specimen studied by Bryant (1933, pi. 20,

DENISON: EARLY DEVONIAN FISHES 281

fig. 2; Princeton 13576), and incorrectly designated as a cotype,
consists of the left cornu and the greater part of the shield rim.
It is probable that this belongs to the same species. A third specimen
mentioned by Bryant (1933, p. 314; Princeton 13653) as showing
part of a lateral field, appears to be a fragment of the eurypterid,
Strobilopterus.

Three additional specimens have been discovered since Bryant's
description was published. Two of these are in the Princeton
Museum: 16132, showing details of the orbital region, and 13748,
showing the central part of the dorsal cephalic shield. The third,
C.N.H.M.-PF 148, is comparable in extent to the type but is in
a better state of preservation (fig. 58).

Description. — The orbits of C. wyomingensis are small relative
to the size of the cephalic shield, round-oval in shape, and widely
spaced. In PF 148 they are 12.3 mm. in antero-posterior diameter,
11.7 mm. in transverse diameter, and spaced 19 mm. apart (fig. 58).
Within the orbits is a wide sclerotic ring, but the preservation is
not sufficiently good to show whether the ring was a single ossifica-
tion, as is often the case in Osteostraci. The strongly protruding
corneal openings, 6.3 by 5.4 mm. in size, face upward and very slightly
antero-laterally.

Connecting the orbits in most Osteostraci is a pineal groove,
occupied, where preservation has been good, by a pineal plate. In
C. wyomingensis this region is continuously roofed over by the
exoskeletal shield, and as a consequence there is no pineal plate.
The pineal foramen lies in the midline somewhat posterior to a line
connecting the middle of the orbits (fig. 57, PI). On the shield,
extending from the orbits toward the pineal foramen are slight
depressions (shown on PF 148 and Princeton 16132) that may
represent the position of the groove for the pineal plate of other
Osteostraci. Lying in the midline anterior to the pineal foramen are
two openings in the cranial roof. The more posterior of these, an
elongate-oval opening lying between the anterior part of the orbits,
is the unpaired nasal opening (fig. 57, NA). In front of it is the
keyhole-shaped hypophysial foramen (fig. 57, HY), identified as the
"pineal prominence" by Bryant. In all other Osteostraci where this
region is known, these two openings are united to form a single naso-
hypophysial foramen, so it is probable that they have been secondarily
subdivided in C. wyomingensis. In PF 148 (fig. 58) both the nasal
and hypophysial foramina are on slight prominences, probably
endoskeletal, that rise in the exoskeletal openings somewhat above

282

FIELDIANA: GEOLOGY, VOLUME 11

Fig. 58. Cephalaspis wyomingensis, incomplete dorsal shield, PF 148 (X 1).

the surface. In the same specimen, the nasal and hypophysial
openings lie within a slightly concave area of the shield, corresponding
to the fossa circumnasalis of other Osteostraci.

The dorsal field is well shown in PF 148, where it appears to be
relatively short (26 mm.) and broad (18 mm.), although its relative
length may correspond to that of other species of Cephalaspis. The
anterior border of the dorsal field is also recognizable in the type
(fig. 57, DF). The extent of the lateral fields is not known, although
a part of the lateral border of one of them is preserved in Princeton
13576. This specimen shows the left cornu and lateral margin in
ventral view but has the bone removed in its anterior part, where
it exhibits an impression of the dorsal side of the rim laterad to the
lateral field.

The surface of the shield is divided by open circumareal canals
into polygonal areas, usually 1.4 to 2.2 mm. in diameter (fig. 59).

DENISON: EARLY DEVONIAN FISHES

283

Fig. 59. Restoration of Cephalaspis wyomingensis based upon PF 148, 482,
and Princeton 13479, 13576 (X %).

Around the orbits, the nasal and the hypophysial openings, they are
arranged in a concentric fashion. The preservation of the Bear tooth
Butte material is such that details of the surface ornament are not
distinct. They are shown best in Princeton 13748, where there is
no superficial layer and the surface of the exoskeleton within the
polygons is composed of small bony projections of the middle layer.
None of the canals of the lateral line system have been recognized.

Bryant's restoration of C. wyomingensis (1933, fig. 8) is far from
giving a correct picture of the shape and proportions of this species,
largely because of his misinterpretation of the posterior margin
is the rostrum. Another restoration of the cephalic shield has been
ittempted in figure 59, but it is admittedly provisional and subject
;o future correction. The short, broad cornua, as well as the lateral
nargin, are based on Princeton 13576; the central and posterior parts
<>f the shield are based on the type and PF 148. Cephalaspis wyo-
iiingensis certainly had a relatively broad shield, although the pro-
portions cannot be determined exactly from the available material.
' ^he interzonal region was relatively short and broad. A low median
(rest posterior to the dorsal field is indicated on PF 148, but is

284 FIELDIANA: GEOLOGY, VOLUME 11

completely absent on Princeton 13748. The shield was probably
quite flat, but it is not certain how much of the flatness of the speci-
mens is due to crushing.

The scale of figure 59 is based on the type and PF 148 for the
central area, and on Princeton 13576 for the cornua and lateral
margins. Princeton 13748 is approximately 18 per cent larger and
16132 is 27 per cent larger. Since it has been shown above that
the Cephalaspis shield could grow, there should be no hesitation
about referring these specimens to C. wyomingensis.

Comparisons. — Cephalaspis wyomingensis is clearly distinguished
from other well-known species of the genus by the separate nasal
and hypophysial openings, and by the absence of a pineal groove
and a separate pineal plate. It is possible that these characters are
of sufficient importance to warrant generic separation, although in
the present systematic state of Cephalaspis it does not seem advisable
to make this distinction. C. wyomingensis shows some resemblances
to C. magnifica of the Middle Old Red Sandstone of Scotland in
its broad cephalic shield, small, widely spaced orbits, and short
stout cornua, but it is considerably smaller and probably lacks the
peculiarities of shield outline of C. magnifica. Among the North
American species it may be compared with the broad-shielded C.
laticeps and C. rosamundae (which may be identical with C. laticeps)
from the Late Devonian of Escuminac Bay, Quebec. These species,
however, are very much smaller, and they have relatively larger
orbits and shorter, stouter cornua. C. jexi of the Early Devonian
(or possibly base of the Middle Devonian) of Campbellton, New
Brunswick, resembles C. wyomingensis in its short, stout cornua,
but it has a relatively longer shield and smaller, more closely spaced
orbits.

Doubtfully referred specimens. — Three fragmentary specimens
from the Water Canyon formation of Utah resemble C. wyomingensis
and may be referred to it provisionally. PF 493 from Locality A
consists of a fragment of a cephalic shield lying between the dorsal
field and the left pectoral sinus; it agrees in size and in its polygonal
areas with C. wyomingensis. PF 482 (fig. 60, A, B) from Locality A
and PF 483 from Locality G preserve portions of the lateral rim, the
lateral field, and the base of the cornu. On the dorsal side they
have a narrow undivided lateral rim separated from the lateral field
by an area of varying width showing from one to four rows of poly-
gonal areas (fig. 60, A). The provisional reference of these speci-
mens to C. wyomingensis is based mainly on the resemblance in size

DENISON: EARLY DEVONIAN FISHES

285

Fig. 60. A, B, Ceph-
alaspis cf. wyomingensis, PF
482, fragment of right shield
margin (X 1); A, dorsal view
showing posterior part of lat-
eral field; B, ventral view
showing margin of oralo-
branchial fenestra. C, Ceph-
alaspis sp., PF 487, fragment
of shield rim of right side,
dorsal view (XI). D, Ceph-
alaspis sp., PF 638, incom-
plete shield rim, ventral view
(XI).

ind appearance of their polygons. The ventral surface of the rim
is marked with a pattern of ridges separated by rows of pores,
;irranged obliquely to the edge (fig. 60, B). Near the postero-lateral
< orner of the oralo-branchial fenestra, the surface is divided into
polygonal areas.

UNDETERMINED CEPHALASPIS

A few specimens of Cephalaspis from the Water Canyon forma-
t on do not belong to the species described above. They appear
t<> represent two distinct species but are insufficiently preserved to
furnish a satisfactory specific diagnosis.

The first species (fig. 60, D) is represented by PF 492 and PF
6. 18 from Locality A, and PF 491 from Locality G, all fragments
o: the shield rim. The species is only slightly larger than C. brevi-

286 FIELDIANA: GEOLOGY, VOLUME 11

rostris and lacks the rostrum of the latter species, but it has a
fairly distinct rostral angle as in C. utahensis. The dorsal rim has
a narrow, smooth, solid edge, as in the specimens from the Water
Canyon formation provisionally referred to C. wyomingensis (PF
482, 483); the part of the rim medial to it is not preserved. The
ventral rim is marked with ridges separated by rows of pores, as
in the Utah specimens referred to C. wyomingensis, but the pattern
of ridges is finer than in the latter.

The second species (fig. 60, C) is represented by PF 487, 488,
and 489, fragments of rims and cornua all from Locality G. It
agrees with the above species in size, but has a broadly rounded
rostral region and a coarser ridge pattern on the ventral rim. Medial
to the narrow, undivided edge of the dorsal rim, the shield is super-
ficially divided into smooth-surfaced, but very irregularly shaped
areas. The cornua are probably short and stout.

UNDETERMINED OSTEOSTRACI

Fragmentary rims and cornua (PF 639-645), from Localities A,
B, D, and G, presumably belong to Osteostraci. Their ventral
ornamentation consists of pores, sometimes united to form grooves
and arranged in a definite pattern that in places gives way to a
fine polygonal network of canals. The dorsal side of the rims and
cornua is thickly covered with fine pores, which, on the cornua,
occur on wide ridges arranged in a V-shaped pattern, and on the
inner edge of the rim on ridges that follow its margin. The cornua
are moderately stout, long, and gradually tapering. Thin sections,
though very poorly preserved because of extensive replacement,
demonstrate the presence of a well-developed basal laminar layer
and a superficial layer composed of very small, flat-topped denticles.

REFERENCES

Branson, E. B.

1929. New localities for Devonian fishes. Bull. Geol. Soc. Amer., 40, p. 245.

Branson, E. B., and Mehl, M. G.

1930. Primitive fishes from the Devonian of Utah and Wyoming. Bull. Geol.
Soc. Amer., 41, p. 180.

1931. Fishes of the Jefferson formation of Utah. Jour. Geol., 39, pp. 509-531,
figs. 1-2, pis. 1-3.

Bryant, W. L.

1932. Lower Devonian fishes of Bear Tooth Butte, Wyoming. Proc. Amer
Phil. Soc, 71, pp. 225-254, figs. 1-6, pis. 1-10.

DENISON: EARLY DEVONIAN FISHES 287

1933. The fish fauna of Beartooth Butte, Wyoming. Part I: The Heterostraci
and Osteostraci. Proc. Amer. Phil. Soc, 72, pp. 285-314, figs. 1-8, pis. 1-21.

Cooley, I. L.

1928. The Devonian of the Bear River Range, Utah. Utah State Agric.
College, unpublished thesis offered for the degree of Master of Arts, pp. 1-23,
pis. 1-14.

Denison, R. H.

1947. The exoskeleton of Tremataspis. Amer. Jour. Sci., 245, pp. 337-365,
figs. 1-13, pis. 1-3.

Gross, W.

1935. Histologische Studien am Aussenskelett fossiler Agnathen und Fische.
Palaeontogr., 83, Abt. A, 60 pp., 30 figs., 7 pis.

Heintz, A.

1939. Cephalaspida from Downtonian of Norway. Skr. Norske Videnskaps-
Akad., I, Math. Nat. Kl., 1939, no. 5, 119 pp., 35 figs., 30 pis.

Orvig, T.

1951. Histologic studies of placoderms and fossil elasmobranchs. I: The
endoskeleton, with remarks on the hard tissues of lower vertebrates in general.
Ark. Zool., (2), 2, pp. 321-454, figs. 1-22, pis. 1-8.

Simpson, G. G.

1941. Range as a zoological character. Amer. Jour. Sci., 239, pp. 785-804.

Simpson, G. G., and Roe, A.

1939. Quantitative zoology. McGraw Hill Book Co., New York, xvii+414 pp.

Stensio, E. A.

1927. The Downtonian and Devonian vertebrates of Spitsbergen. Part I.

Family Cephalaspidae. Skr. Svalbard Nordishavet, 12, xii+391 pp., 103

figs., 112 pis.
1932. The cephalaspids of Great Britain. British Museum (Natural History).

xiv+220 pp., 70 figs., 66 pis.

Westoll, T. S.

1942. The earliest Panzer gruppen. Aberdeen Univ. Rev., 1942, pp. 114-122,
1 fig.

1945. A new cephalaspid fish from the Downtonian of Scotland with notes
on the structure and classification of ostracoderms. Trans. Roy. Soc. Edin-
burgh, 61, pp. 341-357, figs. 1-7, 1 pi.

White, E. I.

1935. Note on new discoveries of Lower Old Red Sandstone vertebrates in
Herefordshire. Trans. Woolhope Nat. Field Club, 1930-32, pt. 3, pp. 179-
181, figs. 1-5.

Williams, J. S.

1948. Geology of the Paleozoic rocks, Logan quadrangle, Utah. Bull. Geol.
Soc. Amer., 59, pp. 1121-1164, figs. 1-2, pis. 1-6.

Wills, L. 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.

'Voodward, A. S.

1891. Catalogue of the fossil fishes in the British Museum (Natural History).
Part II. xliv+567 pp., 57 figs., 16 pis.