CM

XI E> RAR.Y

OF THE

UNIVERSITY

OF ILLINOIS

590.5
FI

v. 37-38

BIOLOGi

latest Date stamped below

L161— O-1096

Early Devonian Vertebrates from the

Knoydart Formation of Nova Scotia

Robert H. Denison
Curator of Fossil Fishes

Along the shores of Northumberland Strait, near Arisaig, Nova
Scotia, is a Silurian section that has been studied and described by
many geologists because of its completeness, good exposures, and
abundant fossils. This is overlain by relatively barren, mainly red,
sandstones, siltstones, and shales that were named the Knoydart
Formation by Ami in 1901. Its Devonian age was first inferred on
geological grounds by Honeyman in 1866. In 1886, T. C. Weston
and J. A. Robert, working for the Canadian Geological Survey,
succeeded in collecting from the Knoydart Formation at McAras
Brook a few fossils, which were later submitted by Ami to the
British Museum (Natural History) for identification. A. S. Wood-
ward (Ami, 1901, pp. 309, 311-312) identified the following verte-
brates: Onchus murchisoni Agassiz, Pteraspis cf. crouchi Lankester,
Psammosteus cf. anglicus Traquair, and Cephalaspis sp., probably
new. On the basis of these fossils, Woodward correlated the Knoy-
dart Formation with the "Lower Old Red Sandstones-Cornstones"
of the west midlands of England.

In July, 1952, I visited McAras Brook, but was unable to find
the locality from which Weston and Robert obtained their fossils.
They came from just south of the shore road on the west bank of
McAras Brook, and in bed 44 of the section measured by Hugh
Fletcher in 1897 (Ami, 1901, p. 308). I was fortunate enough to
discover at a slightly higher horizon some fairly well-preserved fishes
referable to Pteraspis, Cephalaspis, and Acanthodii. The only
group in Woodward's list not discovered is that represented by the
fragment he compared to Psammosteus anglicus. The British speci-
mens ascribed to this species are now referred to Traquairaspis,
and its presence in this fauna requires confirmation, since in England
it occurs at an earlier horizon than Pteraspis. The occurrence of
the 1952 collection is as follows:

449

450 FIELDIANA: ZOOLOGY, VOLUME 37

Locality. — A small knoll on the west side of the McAras Brook
road, approximately 130 yards south of the shore road, and about
three miles southwest of Arisaig, Antigonish County, Nova Scotia.

Horizon. — Knoydart Formation, Early Devonian (Lower Ditto-
nian); about 100 feet above bed 44 of Fletcher's section, thus
presumably in his bed 53; or probably in bed 22 of the section of
Leonard (1951, p. 72). This horizon is about 560 feet above the
lowest exposure of the Knoydart Formation.

Occurrence. — Most of the fossils came from a 1-13^ inch layer
composed of subangular to subrounded, usually flattened pellets of
red siltstone in a matrix of grayish, impure, dirty, calcareous sand-
stone with angular grains. This grades upward into a reddish-gray,
cross-bedded, ferruginous sandstone similar to the matrix of the
pellet layer, and 23^-3 inches in thickness. Locally, a thin pellet
layer occurs in the sandstone. The sandstone is overlain by and the
pellet layer is underlain by thin red siltstones or mudstones, similar
in lithology to the pellets, and having an irregular or subconchoidal
fracture. Fossils are rare in the red sandstone and siltstone; in
the pellet layer they are common, mostly fragmentary, but include
a number of complete shields or plates, oriented with the convex
side up in general, but with imperfect directional orientation.

The sediments indicate a short period of quiet water deposition
during which the lower red silt accumulated, followed by deposi-
tion in currents sufficiently strong to break up and transport frag-
ments of the silt, as well as to carry and orient Pleraspis and Cepha-
laspis shields. Conditions probably remained the same during the
deposition of the cross-bedded sandstone overlying the pellet layer,
except that for the most part there were no red silt pellets available
locally for transport, and most of the exposed vertebrate remains
had already been transported and buried. A temporary return to
quiet conditions followed, as indicated by the overlying siltstone.

HETEROSTRACI

Pteraspis (Simopteraspis) whitei,1 sp. nov.

Type.— CNHM-PF 1002. This specimen (fig. 109) contains a
well-preserved dorsal shield and a ventral disc, but since the nature
of the occurrence makes it improbable that they belong to the same
individual, the dorsal shield alone is designated the type.

1 Named in honor of Dr. Errol I. White, of the British Museum (Natural
History), who has contributed greatly to our knowledge of the Pteraspidae.

Fig. 109. Pteraspis whitei, sp. nov.f PF 1002, the type; dorsal shield with
associated ventral disc; Xl^.

451

452 FIELDIANA: ZOOLOGY, VOLUME 37

Referred specimens. — Incomplete dorsal shields, PF 1191, 1193,
1194; ventral disc, PF 1192; also a number of isolated rostral, orbital,
lateral, branchial, and cornual plates, dorsal spines, and scales,
PF 1195-1211.

Horizon and locality. — As given above.

Diagnosis. — A Simopteraspis of moderately large size, with the
dorsal shield 52 mm. long in the type; the rostrum relatively long
for the subgenus, with the rostral length /total length of the dorsal
shield =0.24; dentine ridges relatively fine, numbering 86 per
centimeter in the midline of the dorsal disc in the type.

Discussion. — Woodward says of the original shields of Pteraspis
from McAras Brook that they "are so much like P. crouchi that if
these Nova Scotia fossils had been found in western England we
should have referred them to the latter species." (Ami, 1901, p.
311.) White (1935, p. 444) realized that they were quite distinct
from P. crouchi, and to avoid confusion named them P. novae-scotiae.
In this paper and in a later one (1950b, p. 86), he noted resemblances
to the English species, P. leathensis. The type of P. novae-scotiae
in the British Museum (Natural History), P 9117a, is an incomplete,
poorly preserved ventral disc. Since it does not exhibit any com-
pletely diagnostic characters, the identity of the new material cannot
be demonstrated. For this reason it is considered best to describe
the 1952 material as a new species, P. whitei, with the understanding
that identity with P. novae-scotiae may exist but may never be
demonstrable.

It is interesting to discover in the first well-preserved Pteraspis
from North America a representative of the primitive subgenus,
Simopteraspis. It partakes of all the characters of the subgenus,
as defined by White (1950b, p. 76). The size is that of the larger
species referred here by White, about that of a large individual of
P. leathensis, and perhaps a little smaller than P. vogti. The rostrum
(fig. 110, A, ro) is longer and tapers to a more narrowly rounded tip
than in the blunt-snouted species referred by White to the subgenus.
The rostral length (RL) in the type is 12.5 mm., and RL/total
length of dorsal shield is 0.24, compared to 0.20 in P. leathensis,
0.22 in P. gosseleti, and 0.23 in P. primaeva and P. vogti (latter
measured from Kiaer's restoration, 1928, fig. 1). The pineal plate
(fig. 110, A, pi) is small and triangular in shape, but somewhat
larger than in P. leathensis. It is separated on either side by slightly
more than its own width from the orbital plates, which latter (fig.
110, A, C, or) have little more prominent median processes than in

DENISON: EARLY DEVONIAN VERTEBRATES

453

Fig. 110. Restorations of Pteraspis whitei, sp. nov.; X 1^. A, dorsal view
of dorsal shield; B, ventral view of ventral disc; C, lateral view, br, branchial
opening; brp, branchial plate; co, cornual plate; dd, dorsal disc; dsp, dorsal spine;
iol, infraorbital sensory line; Ic, lateral portion of ventral transverse sensory
commissure; Idl, lateral dorsal sensory line; hi, lateral ventral sensory line; mc,
medial portion of ventral transverse sensory commissure; mdl, median dorsal
sensory line; or, orbital plate; pi, pineal plate; pil, pineal sensory line; prl, pro-
fundus sensory line; ro, rostral plate; sol, supraorbital sensory line; tc 3-5, dorsal
transverse sensory commissures; vd, ventral disc.

other Simopteraspis. The dorsal disc in the type, in PF 1191, and in
PF 1193 is less highly vaulted, relatively broader, and rather more
convex laterally than in P. leathensis, but this is probably due
largely or entirely to crushing. Posteriorly, under the dorsal spine,
it has a longer, sharper, posterior projection than in other species
of the subgenus. The cornual plates are smaller and less prominent

454 FIELDIANA: ZOOLOGY, VOLUME 37

than in other Simopteraspis. The dorsal spine (fig. 110, A, C, dsp)
is small and recumbent, similar to that of P. leathensis, but is more
posterior relative to the branchial openings, more as in P. primaeva.
The ventral disc (fig. 110, B) is similar to that of P. vogti as figured
by Kiaer (1928, fig. 2); it is slightly larger and relatively broader
than the type of P. novae-scotiae. In PF 1002 (fig. 109) the somewhat
crushed ventral disc has a length of 43 mm. and a maximum width
of 26 mm. The anterior margin is quite concave in PF 1192, but
none of the specimens indicate the presence of a median notch at
the posterior end for the insertion of a ventral ridge scale, such as
was described by White (1950b, p. 78) in P. leathensis.

The canals of the lateral line system (fig. 110), which have been
exposed by the removal of the basal layer of the exoskeleton in PF
1192-1194, have a similar pattern to that described by Kiaer (1928,
figs. 1, 2) in P. vogti, and by White (1950b, p. 81) in P. leathensis.
Characteristic of Simopteraspis is the fact that on the dorsal shield
the pineal (interorbital) canals (fig. 110, A, pil) form a V-shaped
loop on the dorsal disc and that the median dorsal lines do not
extend anteriorly to meet them, as in typical Pteraspis. The pattern
on the ventral disc agrees with that of P. vogti and P. leathensis;
the number of ventral transverse commissures (fig. 110, B, Ic) is
probably six, as in the latter species.

The ornamentation is similar to that of P. leathensis, with rather
sharply crested, crenulated ridges. In the median line of the dorsal
disc there are about 85 ridges per centimeter, indicating a somewhat
finer ornament than in the British species.

In conclusion, the character by which P. whitei is most sharply
distinguished from other species of Simopteraspis is its relatively
longer and more gradually tapering rostrum. In this feature it is
presumed to be slightly more advanced than P. leathensis, P.
gosseleti, P. primaeva, and P. vogti. It is, however, clearly more
primitive than typical Pteraspis such as P. crouchi and P. roslrata.

OSTEOSTRACI

Cephalaspis novaescotiae,1 sp. nov.

Type.— CNHM-PF 1212, the dorsal half of the cephalic shield,
exposed on the inner side, in counterpart (fig. 111).

Referred specimens. — Incomplete dorsal cephalic shields, PF
1213-1216; also cornua, PF 1217-1219.

1 After the province of Canada in which it was discovered.

DENISON: EARLY DEVONIAN VERTEBRATES 455

Horizon and locality. — As given above.

Diagnosis. — A small Cephalaspis with the maximum length of
the cephalic shield in the median line about 32 mm., and the maxi-
mum width, at midlength of the cornua, 39 mm. Shield rather
high, rising about 15 mm. at the posterior angle above the ventral
rim. Lateral margins of the shield slightly and evenly convex,

Fig. 111. Cephalaspis novaescotiae, sp. nov., PF 1212, the type; ventral aspect
of dorsal cephalic shield; X 1 J^.

narrowing gradually to a broadly rounded rostral margin. Cornua
rather slender, 12 mm. long, or in length 0.30 of the distance from
their tips to the rostral margin; they project almost directly back-
wards, with a slight inward curvature at the tips, and extend behind
the posterior angle of the interzonal part; they are provided with
small denticles on the inner side. The interzonal part is relatively
wide and short; its width, 20.5 mm., is 0.54 of the maximum width
of the shield; its postero-lateral angles (fig. 112, A, pla) are promi-
nent, and the posterior angle (fig. 112, A, pa) is sharp and projects
behind the postero-lateral angles. The pectoral sinuses (fig. 112,
A, ps) are rather shallow and narrow. The lateral fields (fig. 112, A,
If) are 30 mm. long and extend posteriorly onto about one-quarter
of the length of the cornua, where they terminate in a rounded
point. The dorsal field (fig. 112, A, df) is 9.5 mm. long and 3.0
mm. wide, is rounded posteriorly, and is bounded anteriorly by the

456

FIELDIANA: ZOOLOGY, VOLUME 37

pineal plate, which was a distinct element. The pineal fissure is
about 12 mm. from the rostral margin. The orbital openings are
small and oval, 3.6 mm. long and 3.0 mm. wide; they are rather
close, 3.7 mm. apart, and are placed considerably nearer the rostrum

Fig. 112. Restorations of Cephalaspis novaescotiae, sp. nov.; X 2. A, ventral
aspect of dorsal cephalic shield, with part of ventral rim on right; B, lateral view
of impression of inner side of dorsal cephalic shield, df, dorsal field; If, lateral
field; m, margin of oralo-branchial fenestra; nh, naso-hypophyseal foramen; or,
orbit; pa, posterior angle; pla, postero-lateral angle; ps, pectoral sinus; pi, fissure
occupied in life by pineal plate.

than the posterior margin. The nasal and hypophyseal openings
are united (fig. 112, A, nh). The ornament of the surface is finely
tubercular dorsally and finely reticular on the ventral rim of the
cephalic shield. The superficial layer of the exoskeleton was certainly
not continuous.

DENISON: EARLY DEVONIAN VERTEBRATES 457

Discussion. — The above diagnosis is based almost entirely on
the type. None of the available specimens exhibit the outer dorsal
surface of the cephalic shield, and the nature of the ornament was
determined only from a small preparation from the inner side of
PF 1216.

Cephalaspis novaescotiae is not very sharply characterized, yet it
is clearly distinct from any previously described species. It is
comparable to some of the British Dittonian Cephalaspis (none has
yet been described from the British Downtonian), as well as to a
few of the species from the Red Bay Series of Spitsbergen, particu-
larly those of the lower or Fraenkelryggen Division of Downtonian
and Early Dittonian age. Judged largely by Wangsjo's (1952, pp.
572-573) comparison of the Red Bay and succeeding Wood Bay
Series Cephalaspis, the following characters of C. novaescotiae are
primitive for the known members of the genus: (1) small size; (2)
cephalic shield relatively narrow, as compared to the broad, flattened
shields of many Wood Bay Series Cephalaspis; (3) narrow pectoral
sinus and wide interzonal region; (4) relatively slender cornua.

The Nova Scotia Cephalaspis shows many resemblances to a
group of five species thought by Wangs jo (1952, p. 287) to be related,
namely, C. oreas, C. heintzi, and C. pygmaea of the Fraenkelryggen
Division of the Red Bay Series of Spitsbergen, and C. langi and C.
heightingtonensis of the Lower and Middle Dittonian of England.
Of these, it is perhaps closest to C. langi (Stensio, 1932, p. 134),
which species differs in its smaller size, relatively broader cephalic
shield, more laterally projecting cornua, less prominent posterior
angle, and relatively larger orbits. C. novaescotiae may also be
compared to another related assemblage of Red Bay Series species,
C. dissimulata from the Fraenkelryggen Division, and C. hoeli, C.
exilis, and C. retusa from the overlying Ben Nevis Division. C. dis-
simulata has a relatively narrower shield, an angulate rostral border,
a more projecting posterior angle, and a lower shield. The Ben
Nevis species lack an independent pineal plate, have a shorter inter-
zonal part, and differ in other features. Other comparable species
from the Fraenkelryggen Division are: C. hyperboreus, which differs
in its larger size, absence of denticles on the cornua, rounded pos-
terior angle, and relatively smaller dorsal field; C. verruculosa, which
differs in its larger size, deeper pectoral sinus, and longer interzonal
part; C. excellens, which may be distinguished by its short, broad
dorsal field and separate nasal and hypophyseal openings; C. vogli,
which has larger cornua; and C. sinuata, which differs in its larger
size, its low cephalic shield, and broader interzonal part. Two more

458 FIELDIANA: ZOOLOGY, VOLUME 37

English Lower Dittonian species are closely comparable: C. fletti is
somewhat larger, and may be distinguished mainly by the more
rapidly tapering lateral margins of the dorsal shield, the more sharply
pointed rostrum, and the pointed posterior termination of the dorsal
field (Stensio, 1932, p. 127); C. whitbachensis is larger, has cornua
without denticles and of somewhat shorter relative length, and very
small orbits (Stensio, 1932, p. 131).

Cephalaspis novaescotiae is not closely comparable to any of the
species of this genus previously described from the Early Devonian
of North America. A small species is reported from Campbellton,
New Brunswick, but has not been figured or adequately described
(Woodward, 1892, p. 5). The two named species from Campbellton,
C. campbelltonensis and C. jexi, are much larger and distinctly charac-
terized. C. dawsoni from the Early Devonian of Gaspe" Bay, Quebec,
shows more similarities, but is larger, relatively broader (perhaps
due in part to crushing), and possesses relatively longer cornua and
wider pectoral sinuses. The species from Utah and Wyoming, C.
utahensis, C. brevirostris, and C. wyomingensis, show little resem-
blance.

ACANTHODII

The pellet layer has yielded a few fragmentary remains of acan-
thodians, including scales and two distinct types of spines. The
first type of spine is characterized by broad, closely spaced, smooth,
longitudinal ridges, and has a well-developed unornamented base
for insertion into muscles. PF 1221 is a fragment of a small spine
of this type with at least six ridges on the exposed face, while PF
1220 (fig. 113, A) is a still smaller spine with only three ridges on
one side. These must belong to Onchus or a closely related genus.
The other type of spine has relatively narrower ridges that are
separated by furrows wider than the ridges; the ridges are orna-
mented, at least near the anterior edge, with small, node-like tu-
bercles; there is no unornamented base for insertion. One nearly
complete though poorly preserved spine of this type, PF 1222, is
about 27 mm. long, is gently curved, and has seven or eight longi-
tudinal ridges on one side. A fragment of a larger spine, PF 1223
(fig. 113, B), shows this type of ornamented ridge well. PF 1225 (fig.
113, C) is a straight, fragmentary spine or jaw with this type of
nodose tubercles, but exhibiting large, widely spaced, tooth-like
projections on one border. Spines of this type occur in Climatius,
Parexus, and Nodocosta, and the Nova Scotia specimens, though
not definitely determinable, appear to be closest to Climatius.

DENISON: EARLY DEVONIAN VERTEBRATES

459

Only two acanthodian scales have been found, but their rarity
is probably due to the difficulty of seeing them in the pellet layer.
Both are of the Gomphodus type, which, according to Gross (1947,
p. 114), probably belongs with the Onchus-type spine.

Fig. 113. Fragmentary acanthodian spines. A, Onchus sp., PF 1220;
X 10. B, Climatius sp., PF 1223; X 3^. C, spine or jaw of Climatius sp., PF
1225; X 8H-

CORRELATION AND MANNER OF DEPOSITION OF THE
KNOYDART FORMATION

The vertebrates described above give for the first time a basis
for a reasonably precise correlation of the Knoydart Formation.
Pteraspis whitei belongs to the subgenus Simopteraspis that charac-
terizes the lower part of the English Dittonian. The Nova Scotia
form is slightly more advanced and may be slightly younger than
the English P. leathensis but presumably is older than P. crouchi,
the index fossil of the Middle Dittonian. The Pteraspidae are the
most useful vertebrates for Early Devonian correlations. Various
species have been used by White (1950a, fig. 1) and Croft (1953,
p. 429) to designate the stages of the Dittonian and succeeding
Breconian in England and Wales. The pteraspids of Spitsbergen
will, when completely described, show an evolutionary progression
from the small primitive Simopteraspis of the lower Red Bay Series
to the large "Gigantaspis" and the highly specialized Doryaspis of
the Wood Bay Series. The recorded history of the family is shorter

460 FIELDIANA: ZOOLOGY, VOLUME 37

in Podolia, but shows an evolution paralleling that in Britain. Thus
the evolutionary stage of Pteraspis whilei may be given considerable
weight in correlating the Knoydart Formation.

Cephalaspis novaescotiae does not furnish as precise information
about the age, since the evolution of the genus is poorly understood,
and the species may show great variability and be poorly defined.
As pointed out above, it is comparable to certain English Dittonian
species. No species have yet been described from the British
Downtonian, but at least one of these is of a similar type. The
Nova Scotia form is also close to some of the species of the Red
Bay Series of Spitsbergen, particularly of its lower or Fraenkelryggen
Division; the latter is equivalent in age to the upper part of the
Downtonian and the Lower Dittonian, while the upper or Ben Nevis
Division corresponds to much of the rest of the Dittonian. C. novae-
scotiae thus is consistent with the Lower Dittonian age of the
Knoydart Formation.

The acanthodians are of little use for correlation. About all
that can be said is that they belong to genera common in the Late
Silurian and Early Devonian of Europe.

The Knoydart Formation is underlain by the marine Arisaig
Series, whose deposition probably commenced in the Upper Llando-
very or Clinton and continued to latest Silurian times (Jones, 1926;
McLearn, 1924). Williams (1914, pp. 73, 91-92) believed there was
a marked unconformity at the base of the Knoydart Formation.
McLearn (1924, p. 8) showed that the evidence is inconclusive and
that it may be interpreted as either angular unconformity or con-
formity. Near the mouth of McAras Brook, where the outcrops of
the Knoydart Formation and underlying Stonehouse Formation are
close, there is no apparent discordance of dips. If these formations
are conformable, there may be no marked break between them,
and then the upper part of the Stonehouse Formation or the lower
part of the Knoydart should include Downtonian equivalents.

The Arisaig Series, as interpreted by McLearn (1918, pp. 130-
133), was deposited in a shallow marginal sea, largely on muddy
bottoms, and in turbid waters receiving silt from rivers. In the
upper part, beginning at the top of the Moydart Formation and
continuing at intervals in the Stonehouse Formation, red shales
and sandstones appear, but the fauna is still marine invertebrates.
In the Knoydart Formation the sediments are prevailingly red and
typical marine invertebrates are absent, or at least they have not
been detected. Only three invertebrates have been reported from

DENISON: EARLY DEVONIAN VERTEBRATES 461

this formation. In the original Weston and Robert collection,
Henry Woodward identified a fragment of the eurypterid, Pterygotus
sp. (Ami, 1901, p. 309). Leonard (1951, p. 76) reports from an
unspecified horizon "a small (1.7 mm.) biconvex brachiopod shell";
this was not identified and there is a possibility that it is an ostracod.
My own collection contains, in the pellet layer associated with the
vertebrates, a single ostracod belonging to the Leperditiidae.1 None
of these records may be used as a certain indication of either marine
or fresh-water deposition, but the absence of typical marine forms
has generally been interpreted to mean that the sediments were
deposited in a fresh-water environment. The evidence for this is,
however, entirely negative and unconclusive. It may mean that
the Knoydart Formation was deposited near the margin of the sea,
where conditions were not suitable for the life or preservation of a
marine fauna. It may mean that deposition took place off the mouth
of a large river, where the waters were brackish or fresh. It is very
likely that these are deltaic deposits where it is difficult to draw
the line between the marine and the continental realm. There is
little to support Ami's view (1901, p. 312) that the Knoydart
Formation was a lake deposit; the variable nature of the sediments,
the poor sorting and angular grains in the coarser rocks, and the
cross-bedding are all indicative of current deposition. But as far
as the present evidence indicates, the currents may have been either
in a stream or near the shore of the sea.

In spite of the fact that the association of Pteraspis, Cephalaspis,
and acanthodians in the pellet band of the Knoydart Formation
was due to transportation and deposition by currents, there is
evidence from other sources that these fishes occupied a similar
broad habitat during life. They are commonly, if not usually,
associated in similar deposits in Podolia, Spitsbergen, and Great
Britain. Their manner of preservation, in general, is similar to
that in the Knoydart Formation, and it is probable that in such
cases they have been transported some distance from their usual
place of life. In one case, however, at the Wayne Herbert quarry
in southwestern Herefordshire, England, numerous individuals of
these three groups are preserved entire in a small siltstone lenticle.
The lenticle probably represents the deposits in a single small pool
that dried up, and in this case no transportation was involved; the
fishes were actually associated in life. The pool may have been in
a flood plain of a river, or at the margin of the sea.

1 Probably Herrmannina Kegel 1933, according to Frank M. Swartz (in litt.).

462 FIELDIANA: ZOOLOGY, VOLUME 37

The drawings in this paper have been made from my sketches by
Miss Maidi Wiebe, Staff Artist, Chicago Natural History Museum.

REFERENCES

Ami, Henry M.

1901. Knoydart Formation of Nova Scotia. Bull. Geol. Soe. Amer., 12:
301-312, fig. 1, pi. 26.

Croft, W. N.

1953. Breconian: a stage name of the Old Red Sandstone. Geol. Mag., 90:
429-432.

Gross, Walter

1947. Die Agnathen und Acanthodier des Obersilurischen Beyrichienkalks.
Palaeontographica, 96, Abt. A, pp. 91-161, figs. 1-32, pis. 20-28.

HONEYMAN, D.

1866. Geology of Antigonish County, Nova Scotia. Nova Scotia Inst. Nat.
Sci., 1, pt. 4, pp. 106-120.

Jones, O. T.

1926. Notes on the Silurian rocks of Arisaig, Nova Scotia. Amer. Jour.
Sci., (5), 11: 119-125.

Kiaer, J.

1928. The structure of the mouth of the oldest known vertebrates, pteraspids
and cephalaspids. Palaeobiologica, 1: 117-134, figs. 1-7, pis. 12-13.

Leonard, Robert

1951. The geology of the McAras Brook area. Nova Scotia Dept. Mines,
Ann. Rept. on Mines, 1951, pt. 2, pp. 61-91, 4 figs., 3 pis.

McLearn, F. H.

1918. The Silurian Arisaig Series of Arisaig, Nova Scotia. Amer. Jour. Sci.,

(4), 45: 126-140.
1924. Paleontology of the Silurian rocks of Arisaig, Nova Scotia. Mem.

Geol. Surv. Canada, 137: 1-179, fig. 1, pis. 1-30.

Stensio, E. A.

1932. The cephalaspids of Great Britain. British Museum (Natural History).
xiv+220 pp., 70 figs., 66 pis.

Wangsjo, G.

1952. The Downtonian and Devonian vertebrates of Spitsbergen. IX. Mor-
phologic and systematic studies of the Spitsbergen cephalaspids. Skr. Norsk
Polarinst., 97: 1-611, figs. 1-108, pis. 1-118.

White, E. I.

1935. The ostracoderm Pteraspis Kner and the relationships of the agnathous

vertebrates. Phil. Trans. Roy. Soc. London, (B), 225: 381-457, figs. 1-97,

pis. 25-27.
1950a. The vertebrate faunas of the Lower Old Red Sandstone of the Welsh

borders. Bull. Brit. Mus. (Nat. Hist.), Geol., 1: 51-67, figs. 1-2.
1950b. Pteraspis leathensis White, a Dittonian zone-fossil. Bull. Brit. Mus.

(Nat. Hist.), Geol., 1: 69-89, figs. 1-25, pi. 5.

463

464 FIELDIANA: ZOOLOGY, VOLUME 37

Williams, H. Y.

1914. Arisaig-Antigonish district, Nova Scotia. Mem. Geol. Surv. Canada,
60: 1-173.

Woodward, A. S.

1892. On the Lower Devonian fish-fauna of Campbelltown, New Brunswick.
Geol. Mag., (3), 9: 1-6, pi. 1.