Full text of "Bulletin of the British Museum (Natural History)"

See other formats

sai :SEU

IORYI

.'35

Bulletin of the

British Museum (Natural History)

Geology series Vol 37 1983-4

British Museum (Natural History)
London 1985

Contents
Geology Volume 37

Page

No 1 Taxonomy of the arthrodire Phlyctaenius from the Lower or
Middle Devonian of Campbellton, New Brunswick, Canada.
V. T. Young 1

No 2 Ailsacrinus gen. nov., an aberrant millericrinid from the Middle
Jurassic of Britain.
P. D. Taylor 37

No 3 Miscellanea 79

Glossopteris anatolica sp. nov. from uppermost Permian strata

in south-east Turkey.

S. Archangelsky & R. H. Wagner 81

The crocodilian Theriosuchus Owen, 1879 in the Wealden of

England.

E. Buffetaut 93

A new conifer species from the Wealden beds of Feron-

Glageon, France.

H. L. Fisher & J. Watson 99

Late Permian plants including Charophytes from the Khuff

Formation of Saudi Arabia.

C. R. Hill & A. A. El-Khayal 105

British Carboniferous Edrioasteroidea (Echinodermata).

A. B. Smith 113

A survey of Recent and fossil Cicadas (Insecta, Hemiptera-

Homoptera) in Britain.

P. E. S. Whalley 139

The Cephalaspids from the Dittonian section at Cwm Mill, near

Abergavenny, Gwent.

E. I. White&H. A. Toombs 149

No 4 The relationships of the palaeoniscid fishes, a review based on new
specimens of Mimia and Moythomasia from the Upper Devonian
of Western Australia.
B.G.Gardiner.. 173

Dates of publication of the parts

Nol 30 June 1983

No 2 28 July 1983

No 3 24 November 1983

No 4.., ...29 November 1984

ISSN 0007-1471

Printed in Great Britain by Adlard & Son Ltd, Bartholomew Press, Dorking, Surrey

Bulletin of the

British Museum (Natural History)

Taxonomy of the arthrodire
Phlyctaenius from the Lower or Middle
Devonian of Campbellton,
New Brunswick, Canada

V. T. Young

Geology series Vol 37 No. 1 30 June 1983

The Bulletin of the British Museum (Natural History), instituted in 1949, is issued in four
scientific series, Botany, Entomology, Geology (incorporating Mineralogy) and Zoology,
and an Historical series.

Papers in the Bulletin are primarily the results of research carried out on the unique and
ever-growing collections of the Museum, both by the scientific staff of the Museum and by
specialists from elsewhere who make use of the Museum's resources. Many of the papers are
works of reference that will remain indispensable for years to come.

Parts are published at irregular intervals as they become ready, each is complete in itself,
available separately, and individually priced. Volumes contain about 300 pages and several
volumes may appear within a calendar year. Subscriptions may be placed for one or more of
the series on either an Annual or Per Volume basis. Prices vary according to the contents of
the individual parts. Orders and enquiries should be sent to:

Publications Sales,

British Museum (Natural History),
Cromwell Road,

London SW75BD,
England.

World List abbreviation: Bull. Br. Mus. nat. Hist. (Geol.)

The Geology Series is edited in the Museum's Department of Palaeontology
Keeper of Palaeontology: Dr H. W. Ball
Editor of the Bulletin: Dr M. K. Howarth

Assistant Editor: Mr D. L. F. Sealy

ISSN 0007-1471

Geology series

British Museum (Natural History) Vol 37 No 1 pp 1-35

Cromwell Road
London SW7 5BD Issued 30 June 1983

Taxonomy of the arthrodire Phlyctaenius from the
Lower or Middle Devonian of Campbellton,
New Brunswick, Canada

V. T. Young

Department of Palaeontology, British Museum (Natural History), Cromwell Road, London
SW7 5BD

Contents

Synopsis ....

Introduction

Materials and methods

Taxonomy

Identification of the species
Diagnoses

Genus Phlyctaenius
P. acadicus (Whiteaves)
P. atholi (Pageau) .
P. stenosus sp. nov. .

Comparative description .

Discussion

Acknowledgements .

References

Index

Synopsis

The species of the Lower/Middle Devonian arthrodire Phlyctaenius are reviewed in the light of new
material. The new species P. stenosus is described and diagnoses of previous species are emended. A
multivariate analysis is used to support species recognition. It is suggested that some features hitherto used
to distinguish species are invalid, since they are shown to be highly variable. It is also suggested that
previous restorations of Phlyctaenius are composites of more than one species: a new restoration is given
here.

Introduction

In 1971 the British Museum (Natural History) received a collection of fossil fishes collected by
Dr W. Graham-Smith and Professor T. S. Westoll from the Devonian of Campbellton, New
Brunswick, Canada. Amongst the collection there were many specimens of the arthrodire
placoderm Phlyctaenius, including twelve skull roofs and many isolated but well-preserved
thoracic plates.

It became evident that three species of Campbellton Phlyctaenius could be recognized: the
type species P. acadicus (Whiteaves), P. atholi (Pageau) and a new species described below. It
also became apparent that the most comprehensive description (Heintz 1933) of P. acadicus is
based on information from more than one species. Further anatomical information is also
provided by these additional specimens. Since P. acadicus is often cited in discussions of
arthrodire phylogeny it is desirable to revise the taxonomy and to attempt a new restoration of
the Campbellton species. These are the primary objectives of this paper.

Bull. Br. Mus. not. Hist. (Geol.) 37 (1): 1-35

Issued 30 June 1983

2 V. T. YOUNG

Specimens of Phlyctaenius from Campbellton were first described by Whiteaves (1881, 1888)
as Coccosteus acadicus. Further material was described by Traquair (1890a,b,c, 1893) who
proposed the name Phlyctaenius; Woodward (1891, 1892a,b) commented on these early
descriptions. Heintz (1933) provided a detailed description, used by subsequent authors
(Denison 1958, 1978; Goujet 1975; Miles & Dennis 1979; Dennis & Miles 1979«, b, 1980; Young
1979, 1980, 1981; Young & Goiter 1981) as the basis for comparisons with presumed relatives
(Dicksonosteus, groenlandaspids). Since Heintz' description many species from the Devonian
of Europe have been assigned to Phlyctaenius, but these are now placed in other genera
(Denison 1978: 60).

Materials and methods

The material used in this study is in the British Museum (Natural History), London, the Royal
Scottish Museum, Edinburgh, and the National Museum of Canada, Ottawa. Specimens in the
two latter institutions are referred to by register number, prefixed by RSM or NMC respectively.
Specimens in the British Museum (Natural History) are referred to simply by register number.
All of the material is from Campbellton, New Brunswick, Canada. The early collections are not
precisely localized, but the material collected by Graham-Smith and Westoll comes from half a
mile west of Campbellton. The similarity of the matrix between this material and that previously
collected suggests that all of it is from the same locality.

The fish material comes from the Gaspe Sandstone Series (described by Logan 1846, 1863),
the age of which is uncertain (Alcock 1935, McGerrigle 1950, Pageau 1968), being considered by
some authors Lower Devonian and others Middle Devonian. Alcock (1935) proposed that the
base of the Gaspe Sandstone marks the beginning of the Middle Devonian; McGerrigle (1950)
divided the middle part of the Gaspe Sandstone into the older York River and younger Battery
Point Formations, which together he believed were of Middle Devonian age. According to
Pageau (1968) the fish fauna occurs in the Battery Point Formation, which he suggests belongs to
the lower part of the Middle Devonian, with the Gaspe Sandstone Series crossing the Lower/
Middle Devonian boundary. The Gaspe Sandstone Series contains abundant, well-preserved
spores (McGregor 1973, 1977) which indicate that the Lower/Middle Devonian boundary lies in
the upper part of the Battery Point Formation (McGregor 1973: fig. 4). The fish fauna is
therefore considered to be of latest Lower or earliest Middle Devonian age.

In the following study the specimens were sorted into three groups and original observations
were tested by a multivariate analysis. As with many primitive arthrodires the skull roofs tend to
remain intact and, in this instance, provide the bulk of the evidence for species recognition. The
abbreviations of the separate plates are mainly those of Miles (1971); see list below.
Explanation of abbreviations used in text and figures:

ADL anterior dorsolateral

AL anterior lateral

AMV anterior median ventral

AV anteroventral

AVL anterior ventrolateral

C central

IL interolateral

M marginal

MD median dorsal

Nu nuchal

P pineal

PDL posterior dorsolateral

PL posterior lateral

postmarginal

posterior median ventral

postnasal

paranuchal

preorbital

postorbital

posterior ventrolateral

rostral

rostralo-pineal

submarginal

suborbital

spinal

The specimens were prepared using a mechanical drilling tool. Where impressions only
remain, positive casts were made using silicone or latex rubber. Casts of ornament tubercles
were studied with the aid of a scanning electron microscope. Because of the difficulty of

TAXONOMY OF PHLYCTAENIUS 3

examining large bone fragments in theoSEM, 'Araldite' casts were made from latex rubber
moulds. The casts were coated with 360 A (36 nm) of gold palladium. Drawings were made with
the aid of a Grant Projector and with the drawing tube attached to the Wild stereo microscope.

Taxonomy

Identification of the species

The specimens of skull roofs studied here belong to more than one morphological group. They
were divided visually into three groups, based on proportions and characters, corresponding to
Phlyctaenius acadicus, P. atholi and the new species P. stenosus.

A multivariate analysis was carried out, using qualitative characters and measurements
(numbered 1-10, Fig. 1) of the skull roofs. The characters used in the analysis were those visible
in the majority of specimens. They were coded for their alternative states and are discussed in
the text below. These characters include the nature of the sutures; the nature of the occipital
cross commissural groove and position of the growth centre and the shape of the posterolateral
margin of the PNu; the skull roof length relative to the width; the presence or absence of the
median postpineal and the nature and arrangement of the ornament tubercles. Ten measure-
ments (Fig. 1) were used; the raw data are deposited on file in the Palaeontology Library, British
Museum (Natural History).

The percentage similarity between each pair of specimens was calculated using Gower's
(1971) coefficient of similarity. The results were used in a principal co-ordinates analysis to
produce a two-dimensional plot in which the distances between the points (representing
specimens) most closely correspond to the calculated similarities between the specimens.

The first plot (Fig. 2) used qualitative characters and measurements for 41 specimens out of a
total of 43 (two very fragmentary specimens were excluded). From the plot it seems that the
specimens may be divided into three groups. One incomplete specimen (P6573d, Fig. 2), at first
assigned to P. acadicus, appears to have a higher similarity to P. atholi: it is uncertain to which
group it belongs. A second specimen (P6573e, Fig. 2), probably belonging to P. stenosus sp.
nov. , appears to have a higher similarity to P. acadicus or P. atholi, but it is incomplete and the
lack of data may account for its unexpected position on the plot.

Some of the specimens included on this first plot have many missing values and it was felt that
this may have introduced distortion. So a second plot was prepared for the 34 most complete
specimens; that is, excluding those specimens with more than 10 missing values. The resulting
plot is almost identical to the first.

A third plot (Fig. 3) was constructed from measurements only, to test whether the grouping
was entirely due to the qualitative characters and whether or not it was biased. Specimens with
more than four missing values were excluded. Although this left only 22 specimens, the result is
a grouping pattern similar to that of the first two plots, and corresponding to the groupings
originally recognized.

Thus from the plot it seems that P. acadicus, P. atholi and P. stenosus are distinct groups. The
specimens grouped as P. atholi include two described by Pageau (1969), both with a median
postpineal, and eight others showing the anterior area all without this plate. This suggests that
the presence of a median postpineal is a variation and not a specific character (see pp. 12-13).

For each dimension measured (Fig. 1), a separate sheet of tracing paper was laid over Fig. 3
(the plot incorporating measurements only), and using this as a basis the measured value for
each specimen, represented by a point on the figure, was plotted on the sheet at the relevant
point.

By doing this, certain trends in the value of each dimension were indicated. The first axis of
Fig. 3 shows, from left to right, increase in the value of each dimension corresponding with the
size of the specimen: the second axis indicates differences in shape between the specimens and
separates the groups. Relatively, specimens of P. stenosus tend to have medium to high values
for measurements 1-4, and low to medium values for measurements 5-10. Specimens of
P. acadicus tend to have medium to high values for measurements 1-3 and high values for
measurements 4-10. Specimens of P. atholi tend to have low to medium values for measure-

V. T. YOUNG

TAXONOMY OF PHLYCTAENWS 5

ments 1-4 and medium to high values for measurements 5-10. These character tendencies,
interpreted as a result of overlaying measurements on the plot, as described above, can be
summarized as follows:

1 . Skull roofs of P. acadicus tend to be larger overall than those of the other two species;

2. Skull roofs of P. acadicus and P. stenosus tend to be relatively longer than those of P.
atholi;

3. Skull roofs of P. atholi tend to be relatively narrower anteriorly compared with those of the
other two species;

4. Skull roofs of P. atholi and P. acadicus tend to be relatively wider posteriorly than those of
P. stenosus; and

5. The growth centre of the PNu occurs in the same transverse plane as that of the Nu in
P. stenosus but further posteriorly in the other two species.

Diagnoses

Order ARTHRODIRA Woodward, 1891
Suborder PHLYCTAENIOIDEI Miles, 1973

Infraorder PHLYCTAENII Miles, 1973

Family PHLYCTAENIIDAE Fowler, 1947

Genus PHLYCTAENIUS Traquair, 18900

[= Phlyctaenaspis Traquair, 1890c]

DIAGNOSIS. See Denison (1978: 59).

TYPE SPECIES. Phlyctaenius acadicus (Whiteaves 1881).

REMARKS. The name Phlyctaenius was originated by Traquair (18900: 20), but he later (1890c:
144) changed it to Phlyctaenaspis since he believed the name Phlyctaenius to be preoccupied by
Phlyctaenium Zittel, a fossil sponge. However, Phlyctaenius is the valid name for this genus
(White 1969: 302 footnote; Fowler 1947: 6).

Two points in Denison's 1978 diagnosis could not be confirmed in material examined here:
the PMs were stated to be large, and the endocranium perichondrally ossified (see
p. 17). The source of the reference to the size of the PMs seems to be Denison (1958: 511, fig.
105, specimen number P5972), where he records an impression of a plate suspected to be large,
although he does not say that the PM is from an impression.

Phlyctaenius acadicus (Whiteaves 1881)
Figs 4; 5A, D; 9A; 10; 11A; 12A; 16D; 18A, D, F

1881 Coccosteus acadicus Whiteaves: 94; text-fig. 1.

1888 Coccosteus acadicus Whiteaves; Whiteaves: 93; pi. 9, figs 1, 3 (not figs 2,4); text-fig. 2.

1890a Phlyctaenius acadicus (Whiteaves) Traquair: 20.

18906 Phlyctaenius acadicus (Whiteaves) ; Traquair: 60.

1890c Phlyctaenaspis acadica (Whiteaves) Traquair: 144.

1891 Phlyctaenaspis acadica (Whiteaves) ; Woodward: 295 (in part).

1892a Phlyctaenaspis acadica (Whiteaves) ; Woodward: 5 ; pi. 1 , fig. 8 (not fig. 7).

18926 Phlyctaenaspis acadica (Whiteaves); Woodward: 481, text-fig. 1.

Fig. 1 Outline drawing of skull roof to show measurement parameters used in multivariate analysis:
1- PrO-PNu length of skull roof; 2. Length of mutual C suture; 3. Distance between growth centres
of PtOs; 4. Width of skull roof at level of PtOs; 5. Distance between growth centres of Ms; 6. Width
of skull roof at level of Ms; 7. Width of C (where possible measured on left C); 8. Distance between
growth centres of PNus; 9. Width of skull roof at level of PNus; 10. Longitudinal distance between
the growth centres of Nu and PNu. The breadth/length index (100 B/L) used in the species
diagnoses was obtained by expressing measurement 6 as a percentage of measurement 1. The
abbreviations for plate names follow those used by Miles (1971); see p. 2.

V. T. YOUNG

W D

R w

y n o

Q O 2

< I UJ

< ^ CO
CO CO CO

Z 2 Z

LLJ <
CO

TAXONOMY OF PHLYCTAENIUS

— II

9 o

< i

<-> H

OJj

Z CO
O x

LLJ <
CO

V. T. YOUNG

10mm
Fig. 4 Phlyctaenius acadicus (Whiteaves). Restoration of skull roof based on RSM GY 1897.51 . 129.

Fig. 5 A. Phlyctaenius acadicus (Whiteaves), skull roof, RSM GY 1897.51. 129. B. Phlyctaenius
atholi (Pageau), skull roof, lectotype RSM GY 1897. 51. 123. C. Phlyctaenius stenosus sp. nov.,
skull roof, hoiotype P6555. D. Phlvctaenius acadicus (Whiteaves), skull roof, visceral surface,
P6554.

TAXONOMY OF PHLYCTAENIUS

: ,

" \ / ,<-*„ Y<

10 V. T. YOUNG

1893 Phlyctaenaspis acadica (Whiteaves); Traquair: 147, text-fig. 1.

1894 Phlyctaenaspis acadica (Whiteaves); Traquair: 370.

1907 Phlyctaenaspis acadica (Whiteaves); Whiteaves: 265; pi. 4, figs 3, 4 (not figs 1, 2).

1916 Phlyctaenaspis acadica (Whiteaves); Chapman: 212.

1925 Phlyctaenaspis acadica (Whiteaves); Stensio: 165; text-fig. 20 (not fig. 21).

1932 Phlyctaenaspis acadica (Whiteaves); Woodward in Zittel: 42.

1933 Phlyctaenaspis acadica (Whiteaves); Heintz: 127; (not pi. 1), pi. 2, figs 1-8; pi. 3, figs 4-6 (not
figs 1-3); text-figs 3 (in part), 5 (in part), 6 (in part) (not text-figs 1,2, 4).

1938 Phlyctaenaspis acadica (Whiteaves) ; Hussakof : 280.

1951 Phlyctaenaspis acadica (Whiteaves); 0rvig: 408; pi. 7, fig. 1.

1957 Phlyctaenaspis acadica (Whiteaves); 0rvig: text-fig. 9A.

1957 Phlyctaenaspis acadica (Whiteaves) ; Gross: 20; pi. 6, figs 5-7, 9.

1958 Phlyctaenaspis acadica (Whiteaves) ; Denison : text-figs 107B , 1 10B , 1 12K, 1 13B .

1959 Phlyctaenaspis acadica (Whiteaves); Stensio: 13.

1962 Phlyctaenaspis acadica (Whiteaves) ; Miles: 65 .

1963 Phlyctaenaspis acadica (Whiteaves); Westoll & Miles: 146; text-fig. 6a.

1964 Phlyctaenaspis acadica (Whiteaves); Lehman: 194; pi. 1, fig. E.
1966 Phlyctaenaspis acadica (Whiteaves); Gardiner: 32.

1968 Phlyctaenaspis acadica (Whiteaves) ; Miles & Westoll: 399.

1969 Phlyctaenaspis acadica (Whiteaves) ; Miles: 132 ; text-figs 9f-g (not 9h) .

1969 Phlyctaenaspis acadica (Whiteaves); Pageau: 810; pi. 25, fig. 1; pi. 29, fig. 5; pi. 30, figs 1, 2, 4-7,

10; text-figs 19.5, 20.7, 21M.

1975 Phlyctaenius acadicus (Whiteaves) ; Goujet: text-fig. IB .
1978 Phlyctaenius acadicus (Whiteaves); Denison: 59; text-fig. 42.
19790 Phlyctaenius acadicus (Whiteaves); Dennis & Miles: text-fig. 1 .

DIAGNOSIS (emended). A species of Phlyctaenius in which the skull roof breadth at the level of
the M is greater than the PrO-PNu length; 100 B/L index =110 (excluding RP); posterolateral
margin of the PNu straight or gently curved; anterior and lateral margins of the R convex; sutures
not evident; ornament tubercles uniformly arranged; tubercles small with pointed, angular
peaks; growth centre of PNu at posterolateral margin of the plate, and close to the posterior
margin; occipital cross-commissural groove on PNu clearly indicated; postoccipital para-
articular process on PNu small; infraorbital canals converge anteriorly; AL more than twice as
tall as wide.

HOLOTYPE. NMC 2774, a cranial shield. Restigouche River, Campbellton, New Brunswick,
Canada.

HORIZON AND LOCALITY. Latest Lower or earliest Middle Devonian, Campbellton, New Bruns-
wick, Canada.

MATERIAL. Specimens showing skull roofs: a cast of the holotype; RSM GY 1897. 51. 129, an
almost complete specimen; P6554, the visceral surface of the skull roof; and four other incom-
plete specimens, P5474, P6572, P56113a,b and P56115a,b. P6577a is probably P. acadicus,
and possibly also P6573d and P6577d.

Specimens showing only thoracic plates: P6576, P7083, P56131, P56137 and RSM
GY 1897.51. 123, 124, 128.

REMARKS. Batteraspis fulgens Pageau 1969, known only by an incomplete AL, may well be a
separate species of Phlyctaenius (Denison 1978: 60). I have not included it in the formal
synonymy since I have not had the opportunity to examine the specimen at first hand.

Phlyctaenius atholi (Pageau 1969)
Figs 5B; 6; 7C, D; 9B; 11B; 12D; 16B, E; 17A; 18B

1888 Coccosteus acadicus Whiteaves; Whiteaves: 94; pi. 9, fig. 2.
18906 Phlyctaenius acadicus (Whiteaves) Traquair: pi. 3, fig. 2.
1892/7 Phlyctaenaspis acadica (Whiteaves) Woodward: 481 .

TAXONOMY OF PHLYCTAENIUS

10mm

Fig. 6 Phlyctaenius atholi (Pageau). Restoration of skull roof based on the lectotype, RSM

GY 1897.51. 123.

1916 Phlyctaenaspis acadica (Whiteaves); Chapman: pi. 21, fig. 6.

1933 Phlyctaenaspis acadica (Whiteaves); Heintz: pi. 3, fig. 1.

1969 Phlyctaenaspis atholi Pageau: 819; pi. 25, fig. 2; pi. 28, fig. 5; text-fig. 19.1, 4.

DIAGNOSIS (emended). Species of Phlyctaenius in which the skull roof breadth at the level of the
M is greater than the PrO-PNu length; 100 B/L index = 110-142 (excluding RP); posterolateral
margin of PNu divided by an angle into two sections; ornament tubercles uniformly arranged;
tubercles large with rounded peaks; growth centre of PNu inside posterolateral margin of plate
and fairly close to posterior margin; occipital cross-commissural groove on PNu clearly
indicated; postoccipital para-articular process on PNu well developed; infraorbital sensory
canals converging strongly anteriorly; anterior point of Nu approximately in line with growth
centres of Cs; Nu length greater than half PrO-PNu length of skull roof; AL equidimensional;
ADL with prominent posterior process above lateral line groove.

LECTOTYPE. Pageau based P. atholi upon two skull roofs (Pageau 1969, specimens RSM
GY 1897.51.113, 123) but he did not specify which was the holotype. I therefore here select as
lectotype the better-preserved specimen, RSM GY 1897.51.123, a skull roof from the Lower or
Middle Devonian, Campbellton, New Brunswick, Canada. Fig. 5B.

HORIZON AND LOCALITY. Latest Lower or earliest Middle Devonian, Campbellton, New
Brunswick, Canada.

MATERIAL. Specimens showing skull roofs: nine fairly complete specimens; P6556,
P6573a, P6573g, P6574, a (part and counterpart), P56120, RSM GY1887.20.44a, RSM
GY 1897.51.113, 123, RSM GY 1978.30.3. Also five others which are poorly preserved.

V. T. YOUNG

pfc P5475

10mm

Fig. 7 Outline of skull roofs showing variations of sensory lines and suture lines C/PrO in P. stenosus
and P. atholi. A. P. stenosus sp. nov. with sensory line variations: pp - posterior pit line grooves
after RSM GY 1897.51.118; pfc - profundus pit line grooves after P56125, P5475; cc- central canal
duplication after P5475. B. P. stenosus sp. nov. (RSM GY 1887.20.44). C. P. atholi with sensory
line variation: pfc - profundus pit line grooves after RSM GY 1978.30.3. D. P. atholi: C/Nu suture
tends to curve towards the C (RSM GY 1887.20.44a) in contrast to conditions in other species.

Specimens showing only thoracic plates: P6559, P6575e, P6577e,h, P56126b, P56127, RSM
GY 1887.20.45, RSM GY 1897.51. 126, 130, 131, 132, 142, 143, RSM GY 1978.30.8, 12, 13.

REMARKS. Pageau (1969: 820) specified three characters in erecting this species:

1. Ornament of large, uniformly arranged tubercles with rounded peaks;

2. RP not attached to skull roof; and

3. Presence of a median postpineal.

The first of these characters is valid but the remaining two are omitted from the emended
diagnosis presented here, for the following reasons.

The detachment of the RP is not a distinguishing character since skull roofs without attached
RP elements are also known for both P. acadicus and P. stenosus. In RSM GY 1897.51. 113, 123

TAXONOMY OF PHLYCTAENWS 13

and P6556 a median postpineal is present. This plate is hexagonal (Figs 5, 6) with the smallest
tubercles, indicating the growth centre, at the centre of the plate. However, a number of
specimens which do not possess this plate are similar to P. atholi in all other characters by which
they were compared, and group with that species in a multivariate analysis (see p. 3). On
these specimens, where the postpineal is absent, the suture line pattern between the PrOs and
Cs is very variable in P. atholi as it is in P. stenosus (Fig. 7), and the right or left PrO may extend
between the two Cs (RSM GY 1887.20.44a, RSM GY 1978.30.3). Denison (1958:507) remarked
that the PrO, and possibly the R, P and PNs, '. . . show the most variable development of any of
the dermal bones of the arctolepid cranial roof, and that this may be 'an indication of the
relative plasticity of the anterior part of the skull'. Miles & Westoll (1968: 390) commented that
in Coccosteus 'the pineal region is by far the most variable part of the dermal bone-pattern',
although they never found a separate postpineal 'in any of the several hundred individuals
of C. cuspidatus examined' or, indeed, in any other coccosteomqrph. They conclude
that 'differential growth rates of the bones from their radiation centres is considered sufficient to
explain all observed conditions . . . '. Species closely related to Phlyctaenius, such as Heighting-
tonaspis anglica (Traquair) (White 1961, 1969) and Aggeraspis heintzi (Gross) (Gross 1962)
sometimes show a postpineal between the PrO and C. I also note considerable variation in the
presence or absence of small roofing bones in Acipenser (Jarvik 1948), Eusthenopteron (Jarvik
1944), dipnoans (Miles 1977) and Osteolepis (Graham-Smith 1978b). It is concluded that the
presence of a postpineal is an individual variation rather than a distinguishing character of
P. atholi.

The figures and description by Pageau (1969) of the skull roof and several thoracic plates of
Gaspeaspis cassivii Pageau from the Battery Point formation, Gaspe Bay, Quebec suggest that
this species should be considered a synonym of P. atholi. I have not included it in the formal
synonymy since I have not examined the material at first hand. Pageau specified two characters
for Gaspeaspis cassivii, the form of the sub-paranuchal depression on the visceral surface of the
skull roof and the superficial ornament. The visceral surface of the PNu of P. atholi is unknown
and this first character cannot be compared. The superficial ornament of G. cassivii is of large,
uniformly distributed tubercles with rounded peaks. This is a character of P. atholi (compare
Pageau 1969: pi. 28, fig. 5, P. atholi, with his pi. 27, fig. 2, G. cassivii). There are also several
other points of similarity from which I infer that these species are conspecific:

1. The breadth of the skull roof at the M is considerably greater than the length (Pageau 1969:
pi. 27, fig. 2; pi. 28, figs 1,2);

2. The posterolateral margin of the PNu is divided by an angle into two sections (Pageau
1969: pi. 28, fig. 2);

3. The postoccipital para-articular process on the PNu is well developed (Pageau 1969: pi. 28,
fig. 2); and

4. The infraorbital sensory canals tend to converge anteriorly (Pageau 1969: pi. 27, fig. 2; pi.
28, fig. 1).

Phlyctaenius stenosus sp. nov.
Figs 5C; 7A, B; 8; 9C; 11C; 12C; 13; 14; 15; 16A, C, F; 17B-F; ISA, C-E

1888 Coccosteus acadicus Whiteaves; Whiteaves: 94; pi. 9, fig. 4.

18906 Phlyctaenius acadicus (Whiteaves) Traquair: pi. 3, fig. 1.

1891 Phlyctaenaspis acadica (Whiteaves) Woodward: 296.

1892fl Phlyctaenaspis acadica (Whiteaves); Woodward: pi. 1, fig. 7.

1907 Phlyctaenaspis acadica (Whiteaves); Whiteaves: pi. 4, figs 1,2.

1925 Phlyctaenaspis acadica (Whiteaves); Stensio: text-fig. 21 .

1933 Phlyctaenaspis acadica (Whiteaves); Heintz: 128; text-figs 1, 2, 3 (in part), 4, 5 (in part), 6 (in

part); pi. 1, figs 1^; pi. 3, figs 2, 3.

1958 Phlyctaenaspis acadica (Whiteaves); Denison: text-figs 105G, 106D, 108B, 109A, 111A, 114A.

1969 Phlyctaenaspis acadica (Whiteaves); Pageau: 814.

1969 Phlyctaenaspis acadica (Whiteaves); Miles: 147; text-fig. 9h.

V. T. YOUNG

10mm

Fig. 8 Phlyctaenius stenosus sp. nov. Restoration of skull roof based on the holotype P6555 and

P56125 and P56116a. Scl - sclerotic plates.

DIAGNOSIS. Species of Phlyctaenius in which the skull roof breadth at the level of the M is equal
to, or a little less than, the PrO-PNu length; 100 B/L index = 90-99 (excluding RP), postero-
lateral margin of PNu marked by an internal angle; anterior and lateral margins of RP gently
concave; sutures evident; ornament tubercles form regular concentric rows parallel to the plate
margins; tubercles generally medium-sized with sharp to rounded peaks; growth centre of PNu
inside posterolateral margin of plate and fairly close to posterior margin; postoccipital para-
articular process on PNu well developed; infraorbital canals subparallel, or converging
anteriorly; anterior point of Nu posterior to growth centres of C; Nu length less than half
PrO-PNu length of skull roof; AL equidimensional.

TAXONOMY OF PHLYCTAENIUS 15

NAME. Stenosus is from the Greek crrevo?, narrow, in reference to the most striking feature of
this species, the narrowness of the skull roof at the level of the M and the PNu.

HOLOTYPE. P6555, a skull roof, Lower or Middle Devonian, Campbellton, New Brunswick,
Canada. Fig. 5C.

HORIZON AND LOCALITY. Latest Lower or earliest Middle Devonian, Campbellton, New Bruns-
wick, Canada.

MATERIAL. Specimens showing skull roofs: holotype P6555, P5475, P5972, P6573h,
P6573i, P56114a,b, P56116a,b, P56117a,b, P56121, P56122, P56123, P56125, RSM
GY 1887.20.44, RSM GY 1897.51. 118, RSM GY 1897. 51. 125, RSM GY 1898. 180.24, RSM
GY 1978.30.5, RSM GY 1978.30.7; and possibly P6573e, P56124.

Specimens showing only thoracic plates: P5973, P6559a,c, P6577b,d, P56126a,b, P56131,
P56134, P56142, P56144, P56146, P56148, P60900, RSM GY 1897.51. 120, 121, 134, 135, 136,
139, 140, RSM GY 1978.30.10, 11.

Comparative description

This section is intended to supplement and amplify the information included in the species
diagnoses and accompanying remarks.

The plates forming the skull rooioiPhlyctaenius are normal for arthrodires. They are strongly
sutured together, except for the PM, of which a fragment is questionably identified on P6577a,
here referred to P. acadicus. As with other primitive (non-brachythoracid) arthrodires the
orbital notches are shallow, the sutures show very little overlap, the Cs are hexagonal and the Nu
is parallel-sided and anteriorly wedged between the Cs. The shape and proportions of the head
shield and constituent plates, particularly the Nu and PNu, vary between the three species.
These differences are best seen by comparing Figs 4, 5, 6 and 8.

The suture lines are clearly visible on the ornamented surface in P. stenosus (Figs 5C, 8),
normally visible on P. atholi (Figs 5B, 6) and are not evident on P. acadicus (Figs 5 A, 4), where
the path of the sutures is deduced from the ornament pattern, and, on the visceral surface, by the
junctions of radiating lines (Figs 5D, 10). The dorsal surface immediately adjacent to the
posterior margin of the skull roof of P. acadicus is unornamented and bevelled and was probably
covered with skin. The skull roof specimens of Phlyctaenius have been flattened in preservation.
A paper model of the thoracic shield was made by restoring to best fit drawings of individual
plates. This suggests that in life the head showed a marked transverse curvature, the highest
point coinciding with the growth centre of the Nu. From here ridges run to the growth centre of
each C thus delimiting a dorsal flattened triangular area. From the sides of this area the skull roof
slopes laterally. The RP (not known in P. atholi) differs in P. acadicus and P. stenosus. In the
former the anterior and lateral margins are convex (Figs 4, 5A) while in the latter these margins
are gently concave (Figs 5C, 8) and where they meet the resulting corners turn downwards - a
feature noted by Heintz. Heintz (1933: 130) also noted indistinct impressions on P6555, here
referred to P. stenosus, as evidence of nasal apertures. However, nasal apertures were not
recognized in the specimens studied. The limit between the R and P components of the RP is
obvious in P. acadicus where there is a clear break in the ornament (RSM GY 1897.5 1 . 129) and a
suture line on the visceral surface (P6554), but in P. stenosus the ornament is continuous.

The shape of the PrO appears highly variable in Phlyctaenius, particularly along the contact
with the C (p. 13) and there may be marked asymmetry in this region of the roof of P. stenosus
(Fig. 7A, B) and P. atholi (Fig. 7C, D). The variation in this region of tne skull roof is discussed
more fully on p. 13. The anterior margin of the PrO of P. atholi is strongly concave (Figs 5B, 6)
and this contrasts with the more gently concave anterior margin of the PrO of P. acadicus (Figs 4,
5 A) and P. stenosus (Figs 5C, 8).

Towards the posterior end of the head shield further differences between the three species are
seen in the shape of the PNu (Figs 4, 5A-D, 6, 8). The posterior margin meets the lateral margin
at varying angles: between 40° and 50° in P. acadicus, between 60° and 77° in P. atholi, 90° in

V. T. YOUNG

Fig. 9 Different development of the para-articular process on the visceral surface of the PNu in the
three species of Phlyctaenius: A. P. acadicus (Whiteaves), P6554. B. P. atholi (Pageau), P6573b. C.
P. stenosus sp. nov., RSM GY 1898.180.24.

P. stenosus. Beneath the surface of the PNu the cranial division of the craniothoracic joint is seen
as the development of a para-articular process and a glenoid fossa which receives the trochlear of
the ADL. The relative sizes of the process differ in the three species and are best seen in Fig. 9.
Of particular diagnostic significance is the margin of the dorsal ornamented surface above the
joint, where it shows a marked inside angle in P. stenosus and P. atholi absent in P. acadicus
(Figs4,5A-C,6,8).

A cucullaris depression was recognized on one specimen (P6554, P. acadicus) showing the
visceral surface, and is similar to that described for Buchanosteus by Young (1979: fig. 2). A
pineal foramen is only recognized in one specimen of P. stenosus (P6555) and the broken margin
suggests that it was quite small. The visceral surface of the skull roof of P. acadicus (P6554,
P6572) suggests that both a pineal fontanelle within the neurocranium and a shallow pineal pit in
the visceral surface of the P are developed.

The external opening for the endolymphatic duct is only recognized in P. atholi (RSM
GY 1887.20.44a); this occurs just mesial to the growth centre of the PNu. Although an opening
for the endolymphatic duct was not recognized in P. acadicus it is possible that the duct ran for
some distance through the PNu. In RSM GY 1897.5 1.129 this is indicated by a slight linear
indentation which runs from a point mesial to the growth centre of the PNu to the C/PNu/Nu
junction (Figs 4, 5A). It may have been formed by the collapse of the roofing bone of the duct
(D. Goujet, personal communication).

Little evidence of the neurocranium is preserved, as with 'Phlyctaenius' sp. (Gross 1937: pi. 2,
fig. 10; text-fig. 21q). Stensio (1925: 165) suggested that this indicates that the neurocranium was
either entirely cartilaginous or only slightly ossified. The outline of the neurocranium in P.
acadicus is evident as the boundary between two types of surface on the underside of the skull
roof and by the surface relief. Specimen P6554 shows the outline most clearly (Fig. 10) but
remnants occur on several other specimens. Where present the outline appears to be generally
similar in all three species of Phlyctaenius. The bone surface which must have lain beyond the
limits of the neurocranium is dark-coloured, smooth and with few canals for blood vessels and
nerves. It is present around the skull roof margins. The surface which originally lay above the
neurocranium is lighter in colour, uneven and marked by numerous grooves. The radiating
growth pattern of the individual dermal plates is very clear (Figs 5D, 10). This surface is covered
in places by a thin layer of lighter-coloured bone with a shiny, uneven surface. This might

TAXONOMY OF PHLYCTAENIUS

pr.so

prapo

gr.scc

pr.ppo

dep.cu

?prsv

10 mm

Fig. 10 Phlyctaenius acadicus (Whiteaves). Visceral surface of skull roof showing area covered by
neurocranium (P6554). The path of the sutures is deduced from the junctions of radiating lines,
dep . cu - cucullaris depression ; fe . hyp - hypophysial fenestra ; gr . sec - groove for semicircular canal ;
pr.ant - antorbital process; pr.apo - anterior postorbital process; pr.ppo - posterior postorbital
process; pr.so - supraorbital process; ?pr.sv - possible supravagal process. Scale bar 10 mm.

represent isolated areas of perichondral ossification. Denison (1978: 59) comments that
'. . . the neurocranium . . . of Phlyctaenius . . . is perichondrally ossified'. The original source
is 0rvig (1951 : 408; pi. 7, fig. 1), where he says that the endocranium of P. acadicus is ossified to a
large extent, and has a lining consisting of thin perichondral bone, and a thin basal layer of
globular calcified cartilage. He goes on to say that the skull roof bone has fused with the
perichondral bone layer of the dorsal endocranial wall beneath the growth centres. However,
perichondral bone was not certainly recognized in the material examined here.

18 V. T. YOUNG

The RP is often detached from the remainder of the skull roof in all three species studied here
(in P. acadicus it is detached in two out of six suitable specimens, in P. stenosus in 13 out of 14
suitable specimens and in P. atholi in all eight suitable specimens). This suggests that the nasal
capsules were units independent of the orbital and occipital region of the neurocranium. This
type of neurocranium would seem to belong to Stensio's (1963: 820) 'type B' group. The dorsal
outline of the postethmo-occipital portion of the neurocranium differs in many respects from the
reconstruction of that of P. acadicus provided by Stensio (1925: text-fig. 20). It is broader, has
more processes and a wider occipital region (Figs 5D, 10). The postethmoidal region is generally
similar in outline to that of Kujdanowiaspis (Stensio 1945, 1963), though it differs in details. It is
broad throughout. The remaining fragments of the occipital region indicate that it appears to be
wider than in Actinolepis magna (Mark-Kurik 1973: fig. 7a,b), and Kujdanowiaspis (Stensio
1945, 1963). The antorbital, supra-orbital, anterior and posterior postorbital and possibly
supravagal processes are developed (Fig. 10). Between the antorbital and supra-orbital pro-
cesses is a notch for the orbital recess. The anterior postorbital process is situated at the PtO/M
suture, and the posterior postorbital process at the M/PNu suture. The latter process does not
bifurcate as it does in Kujdanowiaspis (Stensio 1945, 1963), Actinolepis (Mark-Kurik 1973: fig.
7a, b) and Dicksonosteus (Goujet 1975: fig. 4). However, the impression left by this process ends
abruptly at the lateral margin of the skull roof, between the M and the PNu. It may well bifurcate
beneath the PM as in Actinolepis (Mark-Kurik 1973: fig. 7a,b), although no evidence of the
neurocranium is indicated on the fragment of the PM of specimen P6577a.

Specimen P56120 is here identified as P. atholi; although poorly preserved it has several
interesting features. Impressions are present for the infra-orbital and lateral sensory canals, and
impressions of a number of other canals are preserved. From a comparison with those of
Kujdanowiaspis rectiformis (Brotzen) (Stensio 1945: fig. 1) they can be interpreted: antero-
laterally there are dorsal canals for cutaneous nerves and vessels, lateral to the growth centres of
the Cs there are canals for a dorsal branch of the glossopharyngeal and vessels, and postero-
laterally canals for a supratemporal branch of the vagus and for vessels. Other canals are
indistinct on this specimen.

The general pattern of sensory lines on the skull roof is seen in Figs 4, 5A-C, 6, 8. In P.
stenosus the infraorbital canals run nearly parallel to one another while in the other two species
they converge anteriorly, particularly in P. atholi. Several specimens of P. stenosus show
variations in the sensory canals: paired posterior pit-line grooves, not normally identified in
Phlyctaenius, occur on specimen RSM GY 1897.5 1 . 1 18 (Fig. 7A) . Each occurs as a paired row of
tubercles extending between the growth centres of the C and PNu. In many actinolepids and
phlyctaeniids posterior pit-lines are not evident, or are indicated as short grooves near to the
growth centres of the C and the PNu (e.g. Simblaspis, Sigaspis, Arctolepis; Denison 1978:
text-figs 31, 38). A posterior pit-line is evident on Actinolepis extending between the growth
centres of the C and the PNu (Denison 1978: text-fig. 31). Posterior pit-lines may occur as
discontinuous grooves between the growth centres of the C and the PNu of brachythoracids such
as Holonema, Buchanosteus, Millerosteus, Coccosteus and Dicksonosteus (Denison 1978:
text-figs 45, 49, 57). Specimen P5475, P. stenosus, has a short groove which runs from the growth
centre of the PtO to the PtO/C border, which may be a short duplication of the central canal
(Fig. 7 A). Although not typical of Phlyctaenius, shallow traces of grooves, which may be
profundus pit-line grooves, run between the growth centres of the PrOs and PtOs of P. stenosus
(P5475; P6573i; P56125, Fig. 7A). This is also seen in one specimen of P. atholi (RSM
GY 1978.30.3, Fig. 7C). Profundus grooves occur in some other actinolepids (e.g. Bryantolepis
brachycephalus and Simblaspis cachensis Denison, 1958: 508). Graham-Smith (19780: 26)
suggests that 'profundus' grooves in some dolichothoracids may be produced as a result of an
extension of the suborbital canal becoming anchored ontogenetically to a rudiment of the PtO.

Fig. 11 Phlyctaenius, ornament tubercles: A. P. acadicus (Whiteaves) (RSM GY 1897.51.129) x 22,
SEM SP5/407. B. P. atholi (Pageau) (RSM GY 1897.51. 123) x 25, SEM SP5/428. C. P. stenosus sp.
nov. (P6573h) x 25, SEM SP5/415.

TAXONOMY OF PHLYCTAENIUS

V. T. YOUNG

In some specimens of P. stenosus (P6555, P56125, P5972) the suture line between the PtO and
the M on one side of the infraorbital canal is displaced relative to that on the other side (Figs 7 A,
8). In each case the section of the suture line lateral to the infraorbital canal is more anterior than
the mesial section. Similar displacement at the junctions of sensory canal grooves has been
figured by Gross (1941: text-fig. 7a) for several species of Bothriolepis , and Graham-Smith
(19780: 23-25) has proposed an explanation of how this may have occurred.

The occipital cross-commissural groove is seen in P. acadicus and P. atholi where it runs from
the growth centre of the PNu posteromesially to notch the posterior margin about half-way
along. It is possible that the path of this groove may indicate the presence of extrascapular plates,
as suggested by Miles & Dennis (1979: 45). However, extrascapular plates were not seen in any
species of Phlyctaenius.

The ornament of Phlyctaenius is of individual tubercles, the shape, size and arrangement of
which differ among the three species (Fig. 11A-C). In each species the ornament tubercles of
individual plates are smallest around the growth centre and largest at the plate margins.
Tubercles are uniformly arranged and are close together in P. acadicus, and normally uniformly
arranged in P. atholi. In P. stenosus they are arranged in rows parallel to the plate margins. Thus
it is usually possible to distinguish the approximate outlines of the plates in the three species. In
all three species tubercles lateral to the infraorbital sensory canal are smaller than those mesial.
In several areas of specimen RSM GY 1897. 51. 129 it seems that tubercles have overgrown
earlier tubercles. This feature was noticed in P. acadicus by Gross (1957: pi. 6, fig. 7), and by
0rvig (1957: fig. 9a) who figured a section of P. acadicus which shows two generations of
tubercles separated by a thin layer of laminar bone. In Arctolepis magna Mark-Kurik tubercles
often overgrow other tubercles so that 'the concentric arrangement of the tubercles is often
considerably confused' (Mark-Kurik 1973: 97).

SOs (Fig. 12C, D) occur on P6555 (P. stenosus) and as isolated fragments of Phlyctaenius sp. ,
RSM GY 1897.51 . 118 and 126. Areas missing, due to fracture, from specimen P6555 are present

Fig. 12 Cheek plates of Phlyctaenius. A. P. acadicus (Whiteaves) , fragments of SM and PM,P6577a.
B. P. sp. SM showing groove for hyomandibular, P6573d. C. P. stenosus sp. nov., SO, holotype
P6555. D. P. atholi (Pageau), SO, RSM GY 1897.51.126.

TAXONOMY OF PHL YCTAENIUS 2 1

on the isolated fragments of the SO (Fig. 12D). In section the SO may be divided into two
laminae which lie at an angle to one another (Heintz 1933: fig. 1). The bend between the two
laminae runs horizontally across the 'blade' to the orbital margin. The postorbital branch of the
infraorbital canal has not previously been recorded for Phlyctaenius but it is present in RSM
GY 1897. 51. 118 and 126, and in each it runs in usual arthrodire fashion (Fig. 12D). A post-
suborbital was not found.

The sclerotic plates (Fig. 8, Scl) occur as fragments on specimen P6555, P. stenosus, where
there are apparently four on each side, as is usual for arthrodires (Denison 1978: 2). Each is thick
and strongly arched, similar to those of Arctolepis (Heintz 1962: 36-38). Heintz (1933: 130)
commented that 'the sclerotic ring in Phlyctaenaspis must form something like a more or less
strong capsule protruding from the orbital opening of the head roof. The sclerotic plates of
Placodermi are sometimes ornamented (Denison 1978: 2; Arctolepis, Heintz 1962: 36-38), and
the outer surface of those of Phlyctaenius have an ornament of uniformly arranged tubercles.
Heintz (1933 : 129) commented on the presence of fragments of possible PNs occurring near the
orbit on P6555. However, they are poorly preserved and cannot be positively identified.

The remains of three isolated gnathal elements are recognized (Heintz 1933: pi. 2, figs 2-4). All
three are poorly preserved, fractured and incomplete. Two are believed to be inferognathals,
and one a superognathal, and are believed, by association, to belong to P. stenosus (RSM
GY 1897.51. 125), and P. atholi (RSM GY 1897.51. 126). The inferognathal elements (RSM
GY 1897.51. 125, 126) are small and slender and consist of an anterior tooth-bearing section and
a posterior blade, without teeth. RSM GY 1897.5 1.126 is narrow posteriorly, becoming wider
anteriorly. On the tooth-bearing section of this specimen about 25 to 30 teeth are evident. In
both elements the teeth are of varying sizes, are conical and tend to become larger towards what
is presumed to be the anterior end. Broken sections confirm Heintz' (1933 : 132) comment that
the teeth appear to have no pulp cavity.

Other cranial elements of Phlyctaenius are preserved as fragments. Four poorly-preserved
specimens are identified as submarginals (Fig. 12A, B): P6577a (Heintz 1933: pi. 2, fig. 1),
P6573d - both P. acadicus; RSM GY 1897.51. 125 - P. stenosus- RSM GY 1897.51. 118 -
Phlyctaenius sp. All plates show the visceral surface and show small traces of ornament as
impression. The first two specimens mentioned show a shallow longitudinal groove anteriorly
which probably represents the point of attachment for the hyomandibular.

A small fragment of the visceral surface of a plate, possibly a PM, is attached to the margin of
the skull roof of P6577a, P. acadicus (Fig. 12A herein and Heintz 1933: pi. 2, fig. 1) and an
impression of the visceral surface of the PM is present on P5972, P. stenosus. The outline of the
impression is indistinct.

Specimens of the thoracic shield of Phlyctaenius are represented mostly as isolated plates,
although some occur in association and a few with a skull roof. The ornament of individual plates
may be matched, in most cases, with one or other of the three types of skull roof ornament.

Previous restorations of the trunk shield of Phlyctaenius (Heintz 1933) were based upon
information from more than one species. The present reconstruction of the thoracic shield (Figs
13, 14, 15) is based upon Phlyctaenius stenosus since, of the three species, the specimens of this
show the most complete series of thoracic plates. The reconstruction was difficult because
of variation in size, and the incompleteness of many of the individual plates, and because
associations of trunk plates from one individual are rare and never complete. The reconstruction
was done by making outline tracings of individual plates. The size of the tracings was adjusted
with the Grant Projector to relate to those of one of the more complete associations of plates
belonging to a single individual. The relative sizes of the overlap areas were taken into
consideration. The resulting composite plate outlines were traced onto thin card, cut out
individually, and from these the thoracic shield was reconstructed.

The ventral surface of the thoracic shield is flat with the remaining plates forming an arch.
Sections vertically through the thoracic shield show an anterior triangular cross section and a
posterior seven-sided section, similar to those described and figured by Heintz (1933: 141, fig.
5).

The mutual relationships of constituent plates is that typical of primitive arthrodires (Denison

V. T. YOUNG

Fig. 13 Phlyctaenius stenosus sp. nov. Restoration of skull roof and trunk shield in dorsal view. The
SO is drawn separately since its exact position is unknown. The sensory lines are dotted and the
growth centres starred.

1950, 1978), in which the PL and PVL contact one another to enclose a pectoral fenestra.
However, the extent of the overlap areas of constituent plates suggests that the margin of the
pectoral fenestra is bounded by the AVL, AL and possibly the PVL (Figs 13, 14). There is a
well-developed 'Dorsolateralkante' and 'Ventrolateralkante' (Gross 1932) and a postbrachial
lamina on the AVL and PVL.

The differences between the thoracic shields of the three species are most conspicuous in the
shapes of the MD, AL and ADL. The MD of P. stenosus (Figs 13, 16A) shows a square anterior
end in contrast to the rounded margin in P. atholi (Fig. 16B). In both, there is a low dorsal crest
(more developed in P. stenosus} and the visceral surface shows a small median ridge which might

TAXONOMY OF PHLYCTAEN1US

V. T. YOUNG

Fig. 15 Phlyctaenius stenosus sp. nov. Restoration of trunk shield in ventral view. The growth

centres are starred.

be interpreted as a rudimentary keel. Only one specimen (RSM GY 1897.51.128) of the MD of
P. acadicus was examined and this was too poorly preserved for comment.

The ADL is known in P. atholi (Fig. 17A) and P. stenosus (Figs 14, 17B, C). Both show an
articular condyle below which there is, in P. stenosus, a swollen anterior margin resembling an
obstantic process, although it is not as well developed as that process in Holonema (Miles 1971:
158). Similarly, there is no obvious development of a para-articular face but the shape of the
matching angle of the paranuchal suggests that there must have been some articulation with the
ADL ventrolateral to the articular condyle articulation. Specimens of the visceral surface of the
ADL include: RSM GY 1897.51. 124, 125, 126; P6575; P56143. The specimens are incomplete
and it is not possible to identify them to species by their outline. On these specimens the swollen
process running from below the articular condyle anteroventrally is very prominent as a ridge
along the anterior margin of the lower lamina of the plate, and is similar to the 'ventral ridge' on
the ADL of Holonema (Miles 1971: text-fig. 68). It borders the upper part of the anterior margin
of the AL (Figs 13, 14, 17A-C). There appears to be a small para-articular face of rough-
surfaced bone on the visceral surface of RSM GY 1897.51.125 and P56143. The differences in
shape of the ADL of P. atholi and P. stenosus are seen in Fig. 17A-C. In both, the plate is
divided by a pronounced ridge beneath which the lateral line groove runs to notch the posterior
margin. The main lateral line is continued as a groove as far as the growth centre of the PDL of
both P. stenosus (Figs 14, 17D, E) and P. atholi; thereafter it does not mark the bone (cf. Heintz
1933: fig. 3). The PDL of these two species (unknown for P. acadicus} is very similar except that
the groove for the lateral line is not accompanied by a ridge in P. atholi as it is in P. stenosus. The
PDL of P. stenosus is interesting in showing a small triangular area posteriorly in which the
arrangement of the tubercular ornament differs from (P56126a, b), or is similar to (P6577d, Fig.
17E), that on the rest of the plate.

I thank Dr Gavin Young for drawing my attention to an unusual PDL figured and described by
Pageau (1969: 819; pi. 30, fig. 10; fig. 21M; specimen LTC-29D in Laboratoire Teilhard de
Chardin, Univ. Quebec, Montreal). The specimen is from a horizon yielding specimens of

TAXONOMY OF PHLYCTAENIUS

Fig. 16 Phlyctaenius, isolated thoracic plates to show shape differences between species. MD: A. P.
stenosus sp. nov. (based on RSM GY 1978.30. 10 and P56131); B. P. atholi (Pageau) (RSM
GY 1897.51. 132); Sp: C. ? P. stenosus sp. nov. (RSM GY 1897.51.137); AL: D. P. acadicus
(Whiteaves) (P6576); E. P. atholi (Pageau) (P6577e, h - drawing reversed); F. P. stenosus sp. nov.
(P56144, RSM GY 1897.51.120, 121). Arrow indicates anterior.

Phlyctaenius acadicus and has ornament similar to Phlyctaenius (P. stenosus). Pageau provision-
ally identified the PDL as P. acadicus. However, the specimen is short and is unlike any of the
specimens of PDLs of Phlyctaenius studied here (Figs 14, 17D, E).

The AL is known for all three species and comparative outlines are given in Fig. 16D-F, from
which it may be seen that the AL of P. acadicus is very tall and narrow with a short spinal margin.
This contrasts with the relatively long plate of P. stenosus; P. atholi is intermediate in these
dimensions. The posterodorsal corner of the AL of P. atholi is distinctive in being truncated. As

V. T. YOUNG

Fig. 17 Phlyctaenius, isolated thoracic plates to show shape differences between species. ADL: A. P.
atholi (Pageau) (based on RSM GY 1978.30. 13); B. P. stenosus sp. nov. (based on P56142 and
P6577d); C. P. stenosus sp. nov. (P6577d). PDL: D. P. stenosus sp. nov. (based on P6577d and
P56126a,b); E. P. stenosus sp. nov. (P6577d). PL: F. P. stenosus sp. nov. (P6577d). Arrow
indicates anterior.

is usual in primitive arthrodires the AL is divided into four quadrants by ridges and the pattern of
ornament may vary considerably over the surface of the plate (Heintz 1933: fig. 4).

The outline of the AVL of the three species discussed here is shown in Figs 15, ISA, B, where
the differences in proportion can be seen. Particular points of distinction are the width of the
AVL in P . acadicus associated with the short and divergent spinal margin. The AVL of P. atholi
is relatively narrow, with P. stenosus intermediate in proportion. The ornament on the AVL is
very variable but is most completely seen in specimens of P. stenosus. It is of interest to note that
the side of the plate immediately bordering the subpectoral emargination is heavily ornamented
with tubercles, usually uniformly arranged. This is unlike the AVL of Coccosteus (Miles &
Westoll 1968:434) and Barrydalaspis (Chaloner et al. 1980: 131), where tubercles are absent
immediately adjacent to this margin. Impressions of the scapulocoracoid are indicated on the
visceral surface of several AVLs, and are approximately similar in outline to that of Barry-
dalaspis theroni (Chaloner et al. 1980: fig. 3).

Spinals are generally poorly preserved. The shape, which appears constant for all three
species, may be seen in the restoration of P. stenosus (Figs 13-15) and in Fig. 16C. The spinal of
P. stenosus is the best known. The ornament over the inner and central areas is of small- to
medium-sized tubercles. These may be rounded or elongated, sometimes forming rows parallel,
or slightly inclined, to the longitudinal axis of the plate; occasionally they are uniformly arranged.
The outer margin of the plate is covered by larger tubercles. Although Heintz (1933: 138)
commented that spines were not evident on either the inner or outer margins of the Sp, small
spine-like projections, possibly modified tubercles, are present on the inner margin of P6576b

TAXONOMY OF PHLYCTAENWS

Fig. 18 Phlyctaenius, isolated thoracic plates. AVL: A. ? P. acadicus (Whiteaves) (RSM
GY 1897.51.123); B. ? P. atholi (Pageau) (P56126b). PMV: C. P. stenosus sp. nov. (P56126a, b).
PVL: D. P. acadicus (Whiteaves) (P56131); E. P. stenosus sp. nov. (RSMGY 1897.51.118). AMV:
F. P. acadicus (Whiteaves) (P6577a). Arrow indicates anterior.

and RSM GY 1897.51.134. Specimen P6576 is possibly that figured by Heintz (1933: figs 3, 5, 6).
It is relatively small and short. Small tubercles occupy the centre of the plate, and larger
tubercles the outer margin. The plate is situated close to an AL assigned to P. acadicus and may
belong to this species.

The IL is very incompletely known. Several specimens (RSM GY 1897.51. 134-136) of P.
stenosus displaying this bone show that the ventral surface is narrow (Fig. 15) and has a rounded
recess for the AMV towards the mid-line. A mesial section of the IL is present on RSM
GY 1897.51 . 134 and two laminae of the IL, set at an angle to each other, seem to be indistinctly
indicated. What little of the surface ornament is preserved seems to be of uniformly arranged
tubercles.

28 V. T. YOUNG

Heintz (1933: 142) recognized four kinds of scales:

1 . Rhomboidal or quadrangular scales without evidence of overlap margins, and with typical
Phlyctaenius-like ornament.

2. '. . . more or less oblong triangular scales with clearly overlapping margins along both the
longitudinal sides . . .'. They are strongly arched and thick.

3. '. . . fulcra-like quite big scales . . . bent along the longitudinal axis with a deep incut in
the hind margin'.

4. '. . . quite large, thin roundish scales ... on the outside covered with concentrically
arranged tubercles.'

The rhomboidal or quadrangular scales on RSM GY 1897.51.128 (Heintz 1933: pi. 3, fig. 6a) are
situated close to an MD of P. acadicus, and probably belong to this species. Scales which are
similar, though subtriangular or round, are also present in this area. The long, narrow,
triangular scales figured by Heintz (1933) are associated with plates of P. stenosus (Heintz 1933 :
pi. 3, fig. 5)andanMDof P. acadicus (Heintz 1933: pi. 3, fig. 6b). At each of the longer margins
of the scales on RSM GY 1897.51.125 is a narrow, sloping lateral surface which may be overlap
surface. Heintz comments that the scales are ornamented with large tubercles (Heintz 1933: pi.
2, figs 6,1), though this is not clear on the specimens he described. These scales resemble the
flank scales of Sigaspis (Goujet 1973), which are tall and narrow and overlap one another. It is
possible that the tall narrow scales on RSM GY 1897.51. 125 and 128 are flank scales of P.
stenosus and P. acadicus. The 'fulcra-like' scales, examples of which were described and figured
by Heintz (1933: pi. 3, fig. 4: P6559b, an isolated scale, and fig. 5, lower right corner of his
photograph: RSM GY 1897.51.127, 128) are believed to be dorsal ridge scales. The ornament is
of P. acadicus type. The scale is heart-shaped, conical and deep in section. Its deepest point is at
the growth centre. At its wider end is a V-shaped opening with rounded edges extending from
the growth centre to the scale margins. On specimens RSM GY 1897.51.127 and 128 two dorsal
ridge scales, which are mentioned by Miles (1969: 132), are present, associated with an MD of
P. acadicus and a large cluster of scales. Both specimens are fractured and compressed. The
outlines are indistinct and the V-shaped notch is not seen. It seems that the pointed end of the
scale is anterior. Heintz (1933: pi. 3, fig. 3; P7084) figured a '. . . quite large, thin, roundish
scale' or 'plate . . . covered with concentrically arranged tubercles'. This specimen is believed
to be a C of P. stenosus.

In addition to the scales described above two further varieties are recognized. Small round or
ellipsoidal scales are situated close to the thoracic shield of P6577a, P. acadicus, and possibly
belong to this species. There is no evidence of overlap surfaces or ornament. Also a small, round
scale, about 8mm in diameter and ornamented with concentrically arranged tubercles, is
associated with small, presumed juvenile, thoracic plates of P. stenosus (P6577a), and may
belong to the same individual. Two ridges diverge from the central growth centre to the margin of
the scale. This specimen is believed to be a median ridge scale, and may be a juvenile, or a
different variety of that described above.

Discussion

Traditional classifications of Phlyctaenius (often as Phlyctaenaspis: see pp. 5, 10) have been
provided by Woodward (1891), Zittel (1895, 1932), Fowler (1947), Denison (1958, 1975),
Obruchev (1964, 1967), Miles (1969, 1973) and Stensio (1969). Woodward (1891) and Zittel
(1895) placed Phlyctaenius in the family Coccosteidae; later Woodward in Zittel (1932),
followed by Obruchev (1964, 1967) and Miles (1973), placed the genus in the family Phlyctaena-
spidae; Fowler (1947), followed by Denison (1978), placed it in the family ,Phlyctaeniidae.
Stensio (1969) placed Phlyctaenius in the order Dolichothoraci.

In attempting to reconstruct the phylogeny of arthrodires Miles (1969), Stensio (1969) and
Denison (1975) identified evolutionary trends in order to establish characters by which taxa
could be grouped. Miles (1969) divided the arthrodires into four main groups, which he
recognized as grades of taxa at a particular level of biological organization, each successively

TAXONOMY OF PHL YCTAENIUS 29

more 'biologically efficient' than the last. Phlyctaenius acadicus (Whiteaves), representing the
phlyctaenaspid level of organization, was regarded as intermediate between the presumed more
primitive, actinolepid, and the more advanced, brachythoracid, levels of organization. More
recently placoderm interrelationships have been expressed in the form of cladograms, proposed
by Denison (1978), Miles & Dennis (1979), Dennis & Miles (I979a,b, 1980) and Young (1979,
1980, 1981). The cladogram of Denison (1978) is a general cladogram for the Arthrodira. That
of Miles & Dennis (1979) and Dennis & Miles (1979a,6, 1980) (with slight variations) is
concerned mainly with the brachythoracid arthrodires. The cladogram of Young (1981) is
concerned with phlyctaenioids. The cladograms of Denison (1978) and of Young (1979, 1981)
agree generally in the relationships of the Phlyctaeniidae. In all three cladograms the suborders
Phlyctaeniina, Heterosteina, Coccosteina and Pachyosteina (Denison 1978: fig. 30), or repre-
sentative taxa, are grouped together on the possession of a hinged dermal neck-joint. Young
(1979, 1981) adds the loss of AV plates (if primitive for placoderms), although Denison (1978:
fig. 30) suggests that the presence of paired AVs is a specialization of the Actinolepidae. Miles &
Dennis (1979: 43) and Dennis & Miles (19790: 19, 19796: 308) proposed the presence of a
ventral ridge on the MD as a phlyctaenioid synapomorphy, but Young & Gorter (1981) reject
this since a ventral ridge on the MD is present also in some actinolepids (e.g. Aethaspis,
Sigaspis, Actinolepis). Denison (1978) separated the suborder Phlyctaeniina, including the
families Phlyctaeniidae, Holonematidae and possibly the Williamsaspidae, on the basis of a
long, narrow MD. He also suggested that the elongation of the Sps is a specialization of the
Phlyctaeniidae. However, as Young & Gorter (1981 : 109) remark, neither of these characters is
reliable since the Sps of some actinolepids and phlyctaeniids are of similar length; a long narrow
MD is also present in some brachythoracids, e.g. Gemuendenaspis. They conclude, therefore,
that both characters are plesiomorphous and this leaves the Phlyctaeniina without a character.
They also recognize that the family Phlyctaeniidae can only be defined on the basis of primitive
characters, and cannot therefore be justified as a monophyletic group.

Within the family Phlyctaeniidae Dennis & Miles (19790: 19, 19796: 308, 1980: 47) have
proposed a specialization to separate Phlyctaenius together with more advanced taxa; the occipital
cross-commissure passing off the posterior margin of the PNu. However, this character seems
unacceptable since, as Young & Gorter (1981) have commented, in the brachythoracid
Buchanosteus this sensory groove has the supposed primitive position. As Young & Gorter
(1981) conclude, within the family Phlyctaeniidae subgroups do not seem to be readily defined
and the family may be a paraphyletic group.

Of the phlyctaeniids (listed by Denison, 1978) some genera may be more closely related to
Holonema or Groenlandaspis . The genera Huginaspis, Kolpaspis and Prosphymaspis, each
known only by trunk plates, resemble Groenlandaspis and Tiaraspis in having a high trunk
shield and a convex, high-crested MD. Diadsomaspis, based on trunk plates, resembles Groen-
landaspis and Tiaraspis in the highly arched MD, and it has a holonematid-like ornament of
ridges, though unlike Groenlandaspis the ADL is narrow. Denison (1978) comments that the
head shield oiArctolepis resembles those of the Holonematidae in the wide, fused rostral and
postnasals, the large pineal separating the PrOs, and in the shape of the Nu. This genus was
included by Young (1981: fig. 17) as the sister-group of a group including Holonema. Denison
(1978: 55) remarks that Aggeraspis includes skull roofs and trunk plates which indicate both
phlyctaeniid and actinolepid characters, and may include two genera, or 'might represent a
derivative of Actinolepidae that had attained some of the phlyctaeniid specializations'. Of the
remaining genera of the family Phlyctaeniidae listed by Denison (1978), Phlyctaenius may be
distinguished from Arctaspis and Svalbardaspis in the characteristic shield-shape, and wider
proportions, at the M and PNu, of the skull roof. The trunk shield of Heterogaspis is short and
broad compared with that of Phlyctaenius, while that of Neophlyctaenius sherwoodi Denison is
proportionately longer, and shapes of the plates differ from those of Phlyctaenius (cf. Figs 13-15
herein with Denison 1950: pi. 2, figs 1, 2; pi. 3, figs 1, 2). The Sp of Elegantaspis is very long in
comparison with that of Phlyctaenius.

However, I have found no unique derived character shared by these genera, and so they
cannot form a monophyletic group. The best known species of the Phlyctaeniidae is Dickso-

30 V. T. YOUNG

nosteus arctica Goujet. The general shape of the skull roof and pattern of plates is similar to
those of Phlyctaenius, although D. arcticus is more slender overall. The widths of the skull
roof at the levels of the rostral and of the occipital are about equal in D. arcticus, contrasting
with the condition in Phlyctaenius where the rostral region is narrower. The lateral margin of the
skull roof of Phlyctaenius gently narrows anteriorly at the level of the M and PtO, contrasting
with that of D. arcticus where it is strongly indented. The R of Phlyctaenius is relatively
narrower than that of D. arcticus and the P is shorter. The PNus of Phlyctaenius are wider than
those of D. arcticus, and the PrO/C margins slope posteromesially, unlike those of D. arcticus
where they are more transverse. The main differences in the thoracic plates of Phlyctaenius and
D. arcticus concern the MD, Sps and PVLs. The posteriorly pointed MD of Phlyctaenius differs
from the rounded posterior margin of the MD of D. arcticus. The Sps of D. arcticus are more
slender and curved than those of Phlyctaenius and have distinct spines on their inner surface.
The mutual PVL overlap of Phlyctaenius is simple, while in D. arcticus each PVL overlaps the
other in a complex S-shaped suture.

The other Phlyctaeniidae include Denisonosteus weejasperensis Young & Gorter (1981)
from the Middle Devonian near Wee Jasper, New South Wales, Australia, and genera described
by Pageau (1969) from the Battery Point formation of Gaspe Bay, Quebec. D. weejasperensis
may be distinguished from Phlyctaenius by the shape of the Nu, the convex posterior margin of
the skull roof, and by differences in the shapes of some of the thoracic plates (cf. Figs 4, 5, 6, 8,
13, 14, 16, 17 herein, and Young & Gorter 1981: figs 22, 24). In addition to P. atholi from
Campbellton, Pageau described six new species from beds contemporaneous with material
described here, and some comments are necessary. However, I have not examined the material
at first hand, and my comments are based solely on published information. Three of the species
are based on trunk plates only: Kolpaspis beaudryi, based on an MD and other thoracic plates, is
more reminiscent of Groenlandaspis as discussed above. Batteraspisfulgens, known only by an
AL plate, may well be a separate species of Phlyctaenius (Denison 1978: 60). However, there is
little basis for comparison. Laurentaspis splendida is based on an MD, PVL and AL, and an
isolated C. There is no basis for considering this species to be closely related to Phlyctaenius, and
Denison (1978: 105) placed it in Arthrodira incertae sedis. Quebecaspis russelli (renamed by
Denison (1978: 59) as Pageauaspis) and Cartieraspis nigra, each based on a skull roof and
isolated thoracic plates, differ from Phlyctaenius species in the shapes of some of the skull roof
plates, particularly the long, slender Nu. The posterior margin of the PNu of C. nigra slopes
anterolaterally, unlike that of Phlyctaenius where it slopes very gently posterolaterally. The
ADL of C. nigra is very long and narrow, unlike that of Phlyctaenius. Gaspeaspis cassivii is
based on a skull roof and thoracic plates and, as Denison (1978: 58) remarks, is doubtfully
distinct from Phlyctaenius. It should probably be considered synonymous with P. atholi. The
reasons are discussed under P. atholi (p. 13).

Acknowledgements

I wish to thank Dr G. C. Young of the Bureau of Mineral Resources, Geology and Geophysics,
Canberra, and Drs P. L. Forey, R. S. Miles and C. Patterson of the British Museum (Natural
History), for reading and commenting on the manuscript. Also Dr D. Goujet, Museum
National d'Histoire Naturelle, Paris, and Dr Forey for valuable discussion and comment, and
Ms K. Shaw, British Museum (Natural History), for help with the multivariate analysis. I am
grateful to Dr N. Tebble, Director, and Dr S. M. Andrews of the Royal Scottish Museum,
Edinburgh, for the loan of some of the specimens, and to Dr D . Russell of the National Museum
of Canada, Ottawa, for a cast of the type specimen of P. acadicus. The photographs, except the
scanning electron micrographs, were taken by the Photographic Unit, British Museum (Natural
History).

References

Alcock, F. J. 1935. Geology of the Chaleur Bay region. Mem. geol. Surv. Brch Canada, Ottawa, 183.
iv + 146 pp. , 15 figs, 16 pis.

TAXONOMY OF PHLYCTAENIUS 3 1

Chaloner, W. G., Forey, P. L., Gardiner, B. G., Hill, A. J. & Young, V. T. 1980. Devonian fish and plants

from the Bokkeveld Series of South Africa. Ann. S. Afr. Mus., Cape Town, 81: 127-157, 14 figs,

1 addendum.
Chapman, F. 1916. On the generic position of "Asterolepis ornata var. australis" McCoy with description of

a new variety. Proc. R. Soc. Viet., Melbourne, 28: 211-215, 2 pis.
Dennis, K. & Miles, R. S. 1979a. A second eubrachythoracid arthrodire from Gogo, Western Australia.

Zool. J. Linn. Soc., London, 67: 1-29, 17 figs.
19796. Eubrachythoracid arthrodires with tubular rostral plates from Gogo, Western Australia.

Zool. J. Linn. Soc., London, 67: 297-328, 18 figs.

1980. New durophagous arthrodires from Gogo, Western Australia. Zool. J. Linn. Soc.,

London, 69: 43-85, 22 figs.
Denison, R. H. 1950. A new arthrodire from the New York State Devonian. Am. J. Sci., New Haven, 248:

565-580, 5 figs, 3 pis.
1958. Early Devonian fishes from Utah. Part 3, Arthrodira. Fieldiana, Geoi, Chicago, 11: 461-551,

30 figs.

1975. Evolution and classification of placoderm fishes. Breviora, Cambridge, Mass. , 432: 1-24, 6 figs.

1978. Placodermi. In Schultze, H.-P. (ed.), Handbook of Paleoichthyology , 2. vi + 128 pp., 94 figs.

Stuttgart.

Fowler, H. W. 1947. New taxonomic names of fish-like vertebrates. Notul. Nat., Philadelphia, 187: 1-16.
Gardiner, B. G. 1966. Catalogue of Canadian fossil fishes. Contr. Life Sci. Div. R. Ont. Mus., Toronto, 68:

1-154.
Goujet, D. 1973. Sigaspis, a new arthrodire from the Lower Devonian of Spitsbergen. Palaeontographica,

Stuttgart, (A) 143: 73-88, 3 figs, 1 pi.
1975. Dicksonosteus, un nouvel arthrodire du Devonien du Spitsberg. Remarques sur le squelette

visceral des Dolichothoraci. Colloques int. Cent. natn. Rech. scient., Paris, 218: 81-99, 7 figs, 5 pis.
Gower, J. C. 1971. A general coefficient of similarity and some of its properties. Biometrics, Raleigh,

N.C., 27 (4): 857-871.
Graham-Smith, W. 1978a. On some variations in the latero-sensory lines of the placoderm fish Bothrio-

lepis. Phil. Trans. R. Soc., London, (B) 282: 1-39, 44 figs.
19786. On the lateral lines and dermal bones in the parietal region of some crossopterygian and

dipnoan fishes. Phil. Trans. R. Soc., London, (B) 282: 41-105, 25 figs.
Gross, W. 1932. Die Arthrodira Wildungens. Geol. palaeont. Abh., Jena, 23 (= N. F. 19): 1-61, 26 figs,

2 pis.
1937. Die Wirbeltiere des rheinischen Devons, Teil II. Abh. preuss. geol. Landesanst., Berlin, 176:

1-83, 29 figs, 10 pis.
1941. Die Bothriolepis-arlen der Cellulosa-Mergel Lettlands. K. svenska VetenskAkad. HandL,

Stockholm, (ser. 3) 19 (5): 1-79, 45 figs, 29 pis.
1957. Mundzahne und Hautzahne der Acanthodier und Arthrodiren. Palaeontographica, Stuttgart,

(A) 109: 1-40, 16 figs, 6 pis.

1962. Neununtersuchung der Dolichothoraci aus dem Unterdevon von Overath bei Koln. Palaeont.

Z., Stuttgart, (H. Schmidt-Festband): 45-63, 10 figs.
Heintz, A. 1933. Some remarks about the structure of Phlyctaenaspis acadica Whiteaves. Norsk geol.

Tidsskr., Oslo, 14: 127-144, 6 figs, 3 pis.
1962. New investigations on the structure of Arctolepis from the Devonian of Spitsbergen. Arbok

norsk Polarinst., Oslo, 1961: 23^0, 10 figs, 2 pis.
Hussakof, L. 1938. Structure of the primitive arthrodire Phlyctaenaspis (abstr.). Proc. geol. Soc. Am., New

York, 1937:280-281.
Jarvik, E. 1944. On the dermal bones, sensory canals and pit-lines of the skull in Eusthenopteron foordi

Whiteaves, with some remarks on E. sdve-soderberghi Jarvik. K. svenska VetenskAkad. HandL,

Stockholm, (ser. 3) 21 (3): 1^8, 19 figs.
1948. On the morphology and taxonomy of the Middle Devonian osteolepid fishes of Scotland. K.

svenska VetenskAkad. HandL, Stockholm, (ser. 3) 25 (1): 1-301, 85 figs, 37 pis.
Lehman, J. P. 1964. A propos de quelques arthrodires et ichthyodorulites Sahariens. Mem. Inst. fr. Afr.

noire, Dakar, 68: 193-200, 2 figs, 5 pis.
Logan, W. E. 1846. Report of progress for the year 1844. Rep. geol. Surv. Can., Montreal, 1846: 36-44.

1863. Geology of Canada. Rep. geol. Surv. Can., Montreal, 1863. xxvii + 983 pp., 498 figs.

Mark-Kurik, E. 1973. Actinolepis (Arthrodira) from the Middle Devonian of Estonia. Palaeontographica,

Stuttgart, (A) 143: 89-108, 13 figs, 5 pis.
McGerrigle, H. W. 1950. The geology of eastern Gaspe. Geol. Rep. Dep. Mines Quebec, 35: 1-168, 17 pis,

7 maps col.

32 V. T. YOUNG

McGregor, D. C. 1973. Lower and Middle Devonian spores of eastern Gaspe, Canada. I Systematics.
Palaeontographica, Stuttgart, (B) 142: 1-77, 38 figs, 9 pis.

1977. Lower and Middle Devonian spores of eastern Gaspe, Canada. II. Biostratigraphy. Palaeonto-
graphica, Stuttgart, (B) 163: 111-142, 9 figs, 2 pis.

Miles, R. S. 1962. 'Gemuendenaspis' n. gen., an arthrodiran fish from the Lower Devonian Hunsriich-
schiefer of Germany. Trans. R. Soc. Edinb., 65: 59-77, 3 figs, 1 pi.

1969. Features of placoderm diversification and the evolution of the arthrodire feeding mechanism.

Trans. R. Soc. Edinb., 68: 123-170, 14 figs.

- 1971. The Holonematidae (placoderm fishes), a review based on new specimens of Holonema from
the Upper Devonian of Western Australia. Phil. Trans. R. Soc., London, (B) 263: 101-234, 126 figs.

- 1973. An actinolepid arthrodire from the Lower Devonian Peel Sound Formation, Prince of Wales
Island. Palaeontographica, Stuttgart, (A) 143: 109-118, 6 figs, 3 pis.

1977. Dipnoan (lungfish) skulls and the relationships of the group: a study based on new species from

the Devonian of Australia. Zool. J. Linn. Soc., London, 61: 1-328, 158 figs.

& Dennis, K. 1979. A primitive eubrachythoracid arthrodire from Gogo, Western Australia. Zool. J.

Linn. Soc., London, 66: 31-62, 15 figs.
& Westoll, T. S. 1968. The placoderm fish Coccosteus cuspidatus Miller ex Agassiz from the Middle

Old Red Sandstone of Scotland. Part 1. Descriptive morphology. Trans. R. Soc. Edinb., 67: 373-476, 51

figs, 12 pis.
Obruchev, D. V. 1964. [Agnatha, Fishes]. In Orlov, Y. A. (ed.), Osnovi Paleontologii, 11. 522 pp., 45 pis.

Moscow (Nauka). Engl. transl. 1967. x + 825 pp., 45 pis. Jerusalem (Israel Program for Scientific

Translations).
0rvig, T. 1951. Histologic studies of placoderms and fossil Elasmobranchs. 1. The endoskeleton, with

remarks on the hard tissues of lower vertebrates in general. Ark. Zool., Stockholm, 2: 321^54, 22 figs,

8 pis.
1957. Remarks on the vertebrate fauna of the lower Upper Devonian of Escuminac Bay, P.O.,

Canada, with special reference to the Porolepiform Crossopterygians. Ark. Zool., Stockholm, 10:

367^426, 13 figs.
Pageau, Y. 1968. Nouvelle faune ichthyologique du Devonien Moyen dans les Gres de Gaspe (Quebec). I.

Geologic et ecologie. Naturaliste can., Quebec, 95: 1459-1497, 8 figs, 2 pis.
1969. Nouvelle faune ichthyologique du Devonien Moyen dans les Gres de Gaspe (Quebec). II.

Morphologic et systematique. Deuxieme section: Arthrodires: Dolicothoraci. Naturaliste can.,

Quebec, 96: 805-889, 5 figs.
Stensio, E. A. 1925. On the head of the macropetalichthyids, with certain remarks on the head of the other

arthrodires. Fid Mus. News, Chicago, 4: 85-197, 26 figs, 13 pis.
1945. On the heads of certain arthrodires, II. On the cranium and cervical joint of the Dolicothoraci

(Acanthaspida). K. svenska VetenskAkad Handl. , Stockholm, (ser. 3) 22 (1): 1-70, 14 figs.
1959. On the pectoral fin and shoulder girdle of the arthrodires. K. svenska VetenskAkad. Handl.,

Stockholm, (ser. 4) 8 (1): 1-229, 75 figs, 25 pis.
1963. Anatomical studies on the arthrodiran head. Part I. K. svenska VetenskAkad. Handl.,

Stockholm, (ser. 4) 9 (2): 1-419, 124 figs, 62 pis.

1969. Anatomic des arthrodires dans leur cadre systematique. Annls Paleont., Paris, 57: 151-186.

Traquair, R. H. 1890a. Notes on the Devonian fishes of Scaumenac Bay and Campbelltown in Canada.

Geol. Mag., London, 7: 15-22.
1890/?. On Phlyctaenius, a new genus of Coccosteidae. Geol. Mag., London, 7: 55-60, 1 pi.

- 1890c. Note on Phlyctaenius, a new genus of Coccosteidae. Geol. Mag., London, 7: 144.
1893. Notes on the Devonian fishes of Campbelltown and Scaumenac Bay in Canada. Geol. Mag.,

London, 10: 145-149, 1 fig.

1894. Notes on Palaeozoic fishes. No. 1. Ann. Mag. nat. Hist., London, (6) 14: 368-374, 1 fig., 1 pi.

Westoll, T. S. & Miles, R. S. 1963. On an arctolepid fish from Gemunden. Trans. R. Soc. Edinb., 65:

139-153, 6 figs, 2 pis.
White, E. I. 1961. The Old Red Sandstone of Brown Clee Hill and the adjacent area. II Palaeontology.

Bull. Br. Mus. nat. Hist., London, (Geol.) 5 (7): 243-310, 61 figs, 15 pis.
1969. The deepest vertebrate fossil and other arctolepid fishes. Biol. J. Linn. Soc., London, 1:

293-3 10, 38 figs, 2 pis.
Whiteaves, J. F. 1881. On some fossil fishes, Crustacea and mollusca from the Devonian rocks at Camp-

bellton, N.B., with descriptions of five new species. Can. Nat. & Geol., Montreal, 10: 93-101, 1 fig.

- 1888. Illustrations of the fossil fishes of the Devonian rocks of Canada. Part 2. Proc. Trans. R. Soc.

Can., Ottawa, 6 (IV): 77-96, 2 figs, 6 pis.

TAXONOMY OF PHLYCTAENWS

1907. Illustrations of the fossil fishes of the Devonian rocks of Canada. Part 3. Supplementary notes.

Proc. Trans. R. Soc. Can., Ottawa, (3) 1 (IV): 245-274, 4 pis.
Woodward, A. S. 1891. Catalogue of the fossil fishes in the British Museum (Natural History), 2. xliv + 567

pp., 58 figs, 16 pis. London; Brit. Mus. (Nat. Hist.).
1892a. On the Lower Devonian fish fauna of Campbellton, New Brunswick. Geol. Mag., London, 9:

1-6, 1 pi.

18926. Further contributions to knowledge of the Devonian fish fauna of Canada. Geol. Mag.,

London, 9: 481^85, 2 figs, 1 pi.
Young, G. C. 1979. New information on the structure and relationships of Buchanosteus (Placodermi:

Euarthrodira) from the early Devonian of New South Wales. Zoo/. J. Linn. Soc., London, 66: 309-352,

18 figs, 5 pis.
1980. A new Early Devonian Placoderm from New South Wales, Australia, with a discussion of

Placoderm phylogeny. Palaeontographica, Stuttgart, (A) 167: 10-76, 27 figs, 2 pis.
1981. New early Devonian brachythoracids (placoderm fishes) from the Taemas-Wee Jasper region

of New South Wales. Alcheringa, Adelaide, 5: 245-271, 17 figs.
& Gorter, J. D. 1981. A new fish fauna of Middle Devonian age from the Taemas/Wee Jasper region

of New South Wales. Bull. Bur. Miner. Resour. Geol. Geophys. Aust., Canberra, 209: 83-147, 28 figs,

9 pis.
Zittel, K. A. von 1895. Grundziige der Palaeontologie (Palaeozoologie) . viii + 972 pp. , 2048 figs. Munich &

Leipzig.
1932. Text-book of palaeontology, 2. 2nd Engl. edn (revised by Woodward, A. S.). xvii + 464 pp., 533

figs. London.

Index

The page numbers of the principal references are in bold type. An asterisk (*) denotes a figure.

Acipenser 13
acknowledgements 30
actinolepids 18, 29
Actinolepis 18, 29

magna 18
Aethaspis 29
Aggeraspis 29

heintzi 13

anterior dorsolateral (ADL) 2, 11, 16, 22, 24, 26*
29-30

lateral (AL) 2, 10-11, 14, 22, 24-7, 25*, 30

median ventral (AMV) 2, 27, 27*

ventrolateral (AVL) 2, 22, 26, 27*
anteroventral (AV) 2, 29
antorbital process 17*, 18
'Araldite' 3
Arctaspis 29
ArctolepislS,21,29

magna 20
Arthrodira 5-15, 21-2, 28-30

Barrydalaspis 26

theroni 26

Batteraspis fulgens 10, 30
Battery Point Formation 2, 13, 30
Bothriolepis 20
brachythoracids 18, 29
British Museum (Natural History) 2
Bryantolepis brachycephalus 18
Buchanosteus 16, 18, 29

Campbellton, New Brunswick 1-2, 10-11, 15, 30

Cartieraspis nigra 30

central (C) 2, 4* , 1 1-16, 18, 28, 30

cheek plates 20*

cladograms 29

Coccosteidae 28

Coccosteina 29

coccosteomorphs 13

Coccosteus 13, 18, 26

acadicus2,5, 10, 13

cuspidatus 13
craniothoracic joint 16
cucullaris depression 16, 17*

Denisonosteus weejasperensis 30
dermal neck-joint 29
description, comparative 15-28
Devonian 1-2, 10-11,15,30
Diadsomaspis 29
Dicksonosteus2, 18

arctica 20
dipnoans 13
discussion 28-30
Dolichothoraci 18, 28
dorsolaterals 2

Elegantaspis 29
endocranium5, 17
endolymphatic duct 16
Eusthenopteron 13
extrascapular plates 20

V. T. YOUNG

Gaspe Bay 13, 20

Sandstone 2

Gas peas pis cassivii 13, 30
Gemuendenaspis 29
glenoid fossa 16
glossopharyngeal 18
gnathal elements 21
Goujet, D. 16
Graham-Smith, Dr W. 1-2
Grant projector 3,21
groenlandaspids 2
Groenlandaspis 29-30
growth centres 5, 18, 20,

Heightingtonaspis anglica 13
Heterogaspis 29
Heterosteina 29
Holonema 18, 24, 29
Holonematidae 29
Huginaspis 29
hyomandibular 21
hypophysial fenestra 17*

identification of species 3, 5

inferognatha!21

infraorbital sensory canals 11, 13, 18, 20-1

interolateral (IL) 2, 27

Kolpaspis 29

beaudryi 30
Kujdanowiaspis 18

rectiformis 18

laterals, see anterior, posterior laterals
Laurentaspis splendida 30

marginal (M) 2, 4*, 10-11, 13-15, 18, 20, 29-30
materials 2

median dorsal (MD) 2, 22, 24, 25*, 28-30
median ventrals, see anterior, posterior median

ventral
methods 2-3
Millerosteus 18
Montreal, Univ. Quebec 24
multivariate analysis 2-3, 4*, 5-7, 13

nasal capsules 18

National Museum of Canada, Ottawa 2

Neophlyctaenius shenvoodi 29

neurocranium 16, 17*, 18

New Brunswick, Canada, see Campbellton

nuchal (Nu) 2, 4*, 5, 11-12, 14-16, 29-30

obstantic process 24
occipital 30

groove 20
orbital notches 15

recess 18

ornament, see tubercles
Osteolepis 13

Pachyosteina 29
Pageauaspis russelli 30
para-articular process, face 16*, 16, 24
paranuchal (PNu) 2-3, 4*, 5, 10-11, 13-16, 16*,

18, 20, 24, 29-30
pectoral fenestra 22
perichondral bone 17
Phlyctaenaspidae 28
Phlyctaenaspis 5, 28

acadica 5, 10-11,13

atholill
Phlyctaenii 5-15

Phlyctaeniidae 5-15, 28-30; not monophyletic 29
Phlyctaeniina 29
Phlyctaenioidei 5-15, 29
Phlyctaenium 5

Phlyctaenius 1-30 passim, esp. 5, 6-15; proposal
of name 2, 5

acadicus 1,3,5, 10, 6*, 7*, 8*, 9*, 12-13, 15-18,
16*, 17*, 19*, 20-1, 20*, 24-9, 25*, 27*

atholi 1, 3, 6*, 7*, 9*, 10-13, 11*, 12*, 15-16,
16*, 18, 19*, 20-2, 20*, 24-6, 25*, 26*, 27*, 30

stenosus sp. nov. 1, 3, 6*, 7*, 9*, 12, 12*, 13-15,
14*, 16, 16*, 18, 19*, 20-2, 20*, 22*, 23*,
24-8, 24*, 25*, 26*, 27*
'Phlyctaenius' sp. 16, 20-1, 20*
pineal (P) 2, 13,15-16,30

fontanelle 16

foramen 16

pit 16

pit-line grooves 18
Placodermi21,29
postmarginal (PM) 2, 5, 15, 18, 21
postnasal (PN) 2, 13,21,29
posterior dorsolateral (PDL) 2, 24-5, 26*

lateral (PL) 2, 22, 26*

median ventral (PMV) 2, 27*

venterolateral (PVC) 2, 22, 27*, 30
postorbital (PtO) 2, 4*, 18, 20, 30

processes 17* , 18
postpineal, median 12-13
post-suborbital 21

preorbital (PrO) 2, 4*, 10-15, 18, 29-30
profundus grooves 18
Prosphymaspis 29

Quebecaspis russelli 30

Restigouche River 10
rostral (R) 2, 10, 13,15,29-30
rostralo-pineal (RP) 2, 4*, 10-12, 14-15, 18
Royal Scottish Museum, Edinburgh 2
rubber, silicone or latex 2-3

scales 28

scanning electron microscope 2-3

scapulocoracoid 26

sclerotic plates, ring 14*, 21

semicircular canal 17

TAXONOMY OF PHLYCTAENIUS

sensory canals, lines 18, 20, 22-3, 29
Sigaspis 18, 28-9
Simblaspis 18

cachensis 18

similarity, coefficient of 3
skull roof 1-3, 5, 8*, 9*, 10-11, 11*, 12*, 13, 14*,
15-18, 21, 22*, 29-30

plates, abbreviations 2, 5
spinal (Sp) 2, 25*, 26-7, 29-30
spores 2

submargina!2, 21
suborbital (SO) 2,21-3

canal 18

sub-paranuchal depression 13
superognathal 21
supraorbital process 17*, 18
supravagal process 17*, 18
sutures 7*, 15, 18; see under bones
Svalbardaspis 29

taxonomy 3-15

teeth 21

thoracic plates 1, 10, 12-13, 15, 25*, 26*, 27*, 30

shield 15, 21, 28
Tiaraspis 29
trochlear 16

trunk shield, plates 21, 22*, 23*, 24*, 29-30
tubercles 12-14, 18, 19*, 20, 26-7

vagus 18

ventrals, see anteroventral, median ventrals
ventrolaterals, see anterior, posterior ventro-
laterals

Wee Jasper, N.S.W. 30
Westell, Prof. T.S. 1-2
Wild stereo microscope 3
Williamsaspidae 29

York River Formation 2
Young, DrG. 24

Accepted for publication 8 June 1982

British Museum (Natural History)

An account of the Ordovician rocks of the Shelve Inlier in west Salop
and part of north Powys

By the late W. F. Whittard, F.R.S. (Compiled by W. T. Dean)

Bulletin of the British Museum (Natural History), Geology series
Vol. 33 No. 1. Dec. 1979. 69pp. 38 figs. Large full-colour map

The late Professor W. F. Whittard, F.R.S. , who died in 1966, devoted much of
his life to the study of the Shelve Inlier, and his great monograph on its trilobites
remains fundamental. The area, in west Salop (including a small part of north
Powys), was the scene of famous early geological studies by Murchison, and
Lapworth. By Palaeozoic standards it is in places richly fossiliferous, and exhibits
the best continuous Ordovician succession in Britain, one which is indeed amost
complete. This classic area is of continuing interest, not only to professionals
but also to amateur geologists and students, few of whom complete their
studies without at least one field visit; but amazingly this is the first detailed
map ever to be published. That the work of Whittard, now made available
through the efforts of Professor W. T. Dean of Cardiff, is authoritative there
can be no doubt: for over thirty-five years he studied these rocks, unravelling
their complexities and perfecting his map.

The work complete with map, £10.50 (Post & packing 30p)
Map only, £1.00 (P & p. lOp)

A related work :

Ordovician Brachiopoda from the Shelve District, Shropshire

By A. Williams

Bull. B.M.(N.H.), Geology Supplement 11, 1975. 163pp., 28 plates, 5 tables,
11 text figs. £13.00 (P & p 50p)

All British Museum (Natural History) publications are obtainable from Agents,
Booksellers, the Museum bookshop or by post direct from :
Publication Sales, British Museum (Natural History), Cromwell Road,
London SW7 5BD, England

Titles to be published in Volume 37

Taxonomy of the arthrodire Phlyctaenius from the Lower or Middle
Devonian of Campbellton, New Brunswick, Canada.

By V. T. Young

Ailsacrinus gen.nov.: an aberrant millericrinid from the Middle
Jurassic of Britain. By P. D. Taylor

Miscellanea

Printed by Adlard & Son Ltd, Bartholomew Press, Dorking, Surrey

/"«•«,

Bulletin of the

British Museum (Natural History)

Ailsacrinus gen. nov., an aberrant
millericrinid from the Middle Jurassic of
Britain

P. D. Taylor

Geology series Vol 37 No 2 28 July 1983

Publications Sales,

British Museum (Natural History),
Cromwell Road,

London SW7 5BD,
England.

World List abbreviation: Bull. Br. Mus. nat. Hist. (Geol.)

©Trustees of the British Museum (Natural History), 1983

The Geology Series is edited in the Museum's Department of Palaeontology
Keeper of Palaeontology: Dr H. W. Ball
Editor of the Bulletin: Dr M. K. Howarth

Assistant Editor: Mr D. L. F. Sealy

ISSN 0007-1471 Geology series

Vol37No2PP37-77
British Museum (Natural History)

Cromwell Road

London SW7 5BD Issued 28 July 1983

Ailsacrinus gen. nov., an aberrant millericrinid
from the Middle Jurassic of Britain

P. D. Taylor,

Department of Palaeontology, British Museum (Natural History), Cromwell Road, London
SW75BD

Contents

Synopsis 38

Introduction 38

Localities 38

Ailsacrinus abbreviates 38

Northleach and Eastington 38

Other localities 41

Ailsacrinus prattii 41

Lansdown 41

Kirtlington 42

Systematic descriptions 42

Genus Ailsacrinus nov 42

Ailsacrinus abbreviatus sp. nov 42

Ailsacrinus prattii (Gray) 43

Morphology 45

Stem 45

Dorsal cup 49

Tegmen 52

Arms 52

Pinnules 56

Reconstruction 58

Colour 58

Stem ontogeny 58

Evidence from single crinoids 60

Lenticular columnals 60

Variation in columnal size 61

Distalmost columnal 62

Evidence from crinoid populations 62

Conclusion 64

Palaeoecology 64

Broad environment 64

The Eastington crinoid bed .......... 65

Lithology 65

Sedimentary structures 65

Crinoid preservation 65

Crinoid orientation . . 66

Population density 66

Population variability and structure 67

Tentative model of Crinoid Bed formation 68

Stem function 68

Feeding ecology 69

Evolution 70

Phylogenetic affinities 70

Adaptive evolution 73

Acknowledgements 73

References 73

Index 75

fiw//. Br. Mas. nat. Hist (Geol.) 37 (2): 37-77 Issued 28 July 1983

38 P. D. TAYLOR

Synopsis

The genus Ailsacrinus is proposed for the millericrinids A. abbreviates sp. nov. , from a new Lower-Middle
Bathonian locality at Eastington, near Northleach (Gloucestershire), and the established U. Bathonian
species A. prattii (Gray 1828), known mainly from Lansdown Hill, north of Bath. Ailsacrinus is unusual in
having a short tapering column, highly variable in length (1-70 columnals), with a rounded distal end. The
small calyx contains reduced basals and, in some individuals, an irregular development of accessory plates
often with tubercles. Previously undescribed among millericrinids are syzygies in the arms of Ailsacrinus,
and differentiation of pinnules into an oral series with transversely-ridged pinnulars and a distal series with
cover plates. Well-preserved stereom ultrastructure is described in A. abbreviatus. The palaeoecology of
Ailsacrinus is inferred using functional morphological analysis, comparison with living echinoderms,
preservational evidence, and facies relationships. Following detachment of the proximal part of the column
and crown from the substratum - perhaps quite late in ontogeny - individuals of Ailsacrinus probably led a
free-living existence. There is some evidence for columnal addition after detachment in A. abbreviatus,
although not in A. prattii, and columnals may have been shed occasionally. Eleutherozoic adults of A.
abbreviatus seem to have lived in dense interlocking aggregations or mats which would have provided
individuals with stability in the absence of grasping cirri, and might also have acted as current baffles to aid
suspension feeding. The well-articulated Eastington crinoids were apparently buried catastrophically by
shell sand. Although Ailsacrinus is presumed to have evolved from an attached millericrinid, the evolu-
tionary trend within the genus is in the opposite direction. Morphological similarities with comatulids may
be due to synplesiomorphy or convergence.

Introduction

In 1882 P. Herbert Carpenter published the first full description of the 'Lansdown Encrinite',
Millericrinus prattii (Gray 1828), an unusual crinoid characterized by a short tapering stem
without any obvious means of attachment. Several later authors (Bather 1900, Kirk 191 1 , Gislen
1934, Ubaghs et al. 1978, Roux 1978) used Carpenter's description and reproduced his figures
when discussing M. prattii as an example of a free-living (eleutherozoic) crinoid belonging to a
group otherwise consisting of permanently attached crinoids. It is somewhat surprising, then,
that this interesting crinoid has not been re-studied since the time of Carpenter.

sMost museum specimens of M. prattii were obtained from the Great Oolite of Lansdown Hill,
near Bath. However, Carpenter also mentioned apparently conspecific crinoids from North-
leach, about 60 km north-east of Bath (Fig. 1). A re-examination of this crinoid was prompted by
the discovery of a locality near Northleach which has yielded several hundred well-preserved
individuals from rocks of an earlier age than those present at Lansdown. During the study it
became apparent that the Northleach and Lansdown crinoids were not conspecific. However,
they proved to be sufficiently similar to one another and distinct from established millericrinid
genera to warrant their inclusion in a new genus. Ailsacrinus gen. nov. is proposed to include the
type-species A. abbreviatus sp. nov., from Northleach, and A. prattii (Gray).

Detailed morphological study of Ailsacrinus, as well as corroborating many of Carpenter's
observations, has revealed new information relating to arm and pinnule structure and stereom
ultrastructure. This has enabled the ontogeny, palaeoecology and phylogenetic affinities of
Ailsacrinus to be reconsidered.

Specimens studied are in the collections of the British Museum (Natural History)
(abbreviated BMNH), the Sedgwick Museum, Cambridge (SM), and the Oxford University
Museum (OUM).

Localities

Ailsacrinus abbreviatus

NORTHLEACH AND EASTINGTON. Carpenter (1882) was the first to note the occurrence of the
crinoid herein called A. abbreviatus in the 'Stonesfield Slate' of Northleach, Gloucestershire.
He referred to specimens in the British Museum (Natural History), apparently register numbers
9570, 9572 and 9574 of the Mantell Collection. Other BMNH specimens from Northleach are in

ABERRANT MILLERICRINID AILSACRINUS

ITIIIIIIIIIIIII

ANABACIA LSI.

Fig. 1 Geographical and stratigraphical occurrence of Ailsacrinus gen. nov. in southern England.
Map shows Bathonian outcrop together with localities for A. abbreviates sp. nov. (triangles
numbered 1-6) and A. prattii (Gray) (squares numbered 7 and 8). Locality details: 1, new locality
near Eastington; 2, Isolation Hospital, Northleach; 3, Notgrove; 4, Windrush; 5, Miserden Park; 6,
Corsham; 7, Lansdown Hill; 8, Kirtlington. Stratigraphical details based on Cope et al, (1980)
modified for Northleach.

the Richardson Collection (E14882-5) and are labelled 'Lower Great Oolite, Quarry near
Isolation Hospital'. This is presumably the old quarry described by Richardson (1933: 42) and
located north of the town (loc. 2 of Fig. 1).

The newly-discovered locality is a small exposure on a low river cliff near Eastington, 2km
south-east of Northleach (loc. 1 of Fig. 1). Precise locality details have been lodged with the
Nature Conservancy Council. Facies comparisons and local geological mapping suggest that the
limestone exposed here lies within the Sharps Hill Formation (Sellwood & McKerrow 1974) and
its age is likely to be either early Bathonian tenuiplicatus Zone or mid-Bathonian progracilis
Zone (Cope et al. 1980). An alternative possibility is that the crinoid bed belongs to the Taynton
Limestone Formation (progracilis Zone) which is of a similar facies to the Sharps Hill Formation
around Northleach. However, this seems less likely because the Taynton Limestone Formation
is quartz-deficient, whereas the crinoid bed abounds in quartz. Crinoids occur throughout the
thickness (about 26cm) of the crinoid bed but are most conspicuous on several successive
bedding planes (Fig. 2) and tend to be more abundant and better preserved near to the base of
the bed. The bed overlies and grades into an oobiosparite and in turn is overlain, with a sharp

P. D. TAYLOR

Fig. 2 Ailsacrinus abbreviates gen. et sp. nov. covering a bedding plane from the Eastington crinoid
bed (Bathonian, ? Sharps Hill Fm.), BMNH E67791. Most of the crinoids on this undersurface are
upright but some are prostrate or obliquely orientated. Divisions of scale bar are 1 cm.

ABERRANT MILLERICRINID AILSACRINUS

Fig. 3 Thin section cut from the Eastington Crinoid Bed, BMNH E67832. A, sandy bio-oosparite
composed mostly of molluscan shell fragments subparallel to bedding; PPL, x 17. B, brachial of
Ailsacrinus abbreviatus gen. et sp. nov. ; PPL, x 33.

contact, by a cross-bedded shelly limestone. Lithologically, the crinoid bed is a grain-supported
(grainstone), sandy bio-oosparite composed largely of molluscan shell fragments, 0-1-0-4 mm in
length, which are generally orientated parallel to the bedding (Fig. 3). Isolated crinoid ossicles
are scarce. Subrounded to subangular quartz grains make up about 20% of the rock and impart a
sandy texture to weathered surfaces. Development of the fossils is aided by the presence of thin
layers or drapes of clay which are easily washed away from the crinoid-covered bedding planes.
The clay contains illite, quartz and calcite (determined by X-ray diffractometry). Good preser-
vation of stereom ultrastructure at this locality is probably the result of clay particles penetrating
the outer part of the skeleton and preventing the formation of syntaxial overgrowths within the
porous lattice. Apart from the crinoids, macrofossils are sparse and poorly preserved; a few
abraded brachiopods, epifaunal bivalves and echinoids are present, together with woody
carbonaceous fragments. Preservation of the crinoids is discussed below under Palaeoecology,
p. 65.

OTHER LOCALITIES. Richardson (1904: pi. 19, fig. 1) illustrated crowns of A. abbreviatus
supposedly from the U. Aalenian (murchisonae Zone) Lower Limestone of Andoversford,
10 km west of Northleach. These specimens were not from his personal collection and their
stratigraphical horizon and provenance may be doubted.

The J. Morris Collection at the BMNH contains a specimen (E67) labelled 'Great Oolite,
Corsham, Wiltshire', loc. 6 of Fig. 1.

Carpenter (1882) mentions the occurrence of M. prattii, probably referring to A. abbreviatus,
at Windrush (loc. 4 of Fig. 1), Notgrove (loc. 3) and Miserden Park (loc. 5). Material from
Miserden Park includes BMNH E14661 figured in Carpenter's pi. 1, fig. 9. No exact strati-
graphical details are given but all may be from low in the Bathonian sequence.

Ailsacrinus prattii

LANSDOWN. The great majority of existing A. prattii specimens were obtained over 150 years
ago from Lansdown Hill near Bath (loc. 7 of Fig. 1). As early as 1833 Jelly wrote of the small
chance of ever procuring further specimens. Lansdown Hill is now part of urban Bath. Initially,
Gray (1828) made the mistake of supposing the Lansdown crinoids to have come from the Lias
but Jelly corrected this misapprehension and gave their horizon as Great Oolite. A well record

42 P. D. TAYLOR

from Beckford's Tower at the summit of Lansdown Hill records 30 feet of Great Oolite
(Woodward 1894: 243) which, around Bath, is generally regarded as being of late Bathonian
aspidoides Zone age (Cope et al. 1980). The Lansdown crinoids are contained in a matrix of
coarse oobiosparite. They are reasonably well articulated but arms are often separated from stem
and calyx. Most are preserved in a prostrate attitude, i.e. with the stem lying parallel to bedding.
An associated fauna includes well-preserved echinoids. Both Jelly (1833) and Carpenter (1882)
mention the occurrence of Pentacrinites in association with A. prattii. The Sedgwick Museum
(Cambridge) collections contain two pieces (J33842, J33850) with isocrinids but these are in a
finer-grained rock and are without associated A. prattii.

KIRTLINGTON. A single specimen (J 142 19) in the Oxford University Museum is allegedly from
the Great Oolite of Kirtlington (loc. 8 of Fig. 1). This is also likely to be of late Bathonian age
and the matrix suggests (T. J. Palmer, personal communication 1979) that it may be from the
Lower Cornbrash (discus Zone).

Systematic descriptions

Order MILLERICRINIDA Sieverts-Doreck in Moore et al., 1952
Suborder MILLERICRININA Sieverts-Doreck in Moore etal., 1952
Family MILLERICRINIDAE Jaekel, 1918
Genus AILSACRINUS nov.

DIAGNOSIS. Millericrinidae with reduced column, tapering distally and terminated by a rounded
columnal; incomplete or lenticular columnals may be present; cup bowl-shaped to conical with
small basals and often with irregularly developed, tiny accessory plates, interbasally-placed,
generally tuberculate; arms have frequent syzygies; pinnules differentiated into oral and distal
series, the oral pinnules having high pinnulars with adoral transverse ridges.

TYPE SPECIES. Ailsacrinus abbreviates sp. nov., Lower-Middle Bathonian (M. Jurassic),
Gloucestershire, England. This species is chosen in preference to A. prattii because of the
superior preservation of available material.

REMARKS. This new genus is proposed to accommodate two aberrant millericrinid species in
which the reduced stem with a rounded end is a synapomorphy. They are also distinguished from
previously-described millericrinids by having differentiated oral pinnules and abundant syzygies
in the arms.

The established species now referred to Ailsacrinus, Encrinites (Apiocrinites) prattii Gray
1828, has usually been placed in the genus Millericrinus d'Orbigny, 1841. However, Milleri-
crinus, as defined by the type species Encrinites milleri von Schlotheim 1823, has a flat-
bottomed, five-sided cup with large basals and a stem which does not increase in diameter
towards the cup (see Rasmussen in Ubaghs et al. 1978: fig. 550,1; Roux 1978: fig. 9). E. prattii
was provisionally included by Rasmussen (in Ubaghs et al. 1978: T822) in the genus Liliocrinus
Rollier, 1911, the type species of which was designated by the same authors as Millericrinus
polydactylus d'Orbigny 1841. In this species the basals and radials are of approximately equal
size and the arms divide two or three times, unlike Ailsacrinus in which the basals are consider-
ably smaller than the radials and there is only one division of the arms.

NAME. Ailsacrinus is named in recognition of Miss Ailsa M. Clark of the Department of
Zoology, British Museum (Natural History).

Ailsacrinus abbreviates sp. nov.

1882 Millericrinus Prattii Gray; Carpenter: 29 (partim); pi. 1, fig. 9 only.
1904 Millericrinus Pratti Gray; Richardson: 250; pi. 19, fig. 1.

ABERRANT MILLERICRINID AILSACRINUS 43

DIAGNOSIS. A species of Ailsacrinus with small, bowl-shaped cup and tiny triangular basals
which either fail to touch or barely touch adjacent basals; radials high relative to basals;
accessory plates inconspicuous; column tapering distally and short (1-9 columnals); distal facet
of proximale with petaloid areola.

HOLOTYPE. BMNH E67797 (Fig. 25, p. 52), Bathonian (?Sharps Hill Fm.), Eastington,
Gloucestershire. P. D. Taylor Coll., 1979.

PARATYPES. BMNH 9570, 9572, 9574, Stonesfield Slate, Northleach; Mantell Coll., 1839. E67,
Great Oolite, Corsham, Wiltshire; J. Morris Coll., 1880. E14661 (Carpenter 1882: pi. 1, fig. 9),
Forest Marble?, Miserden, Gloucestershire; J. F. Walker Coll., 1908. E14882-5, Lower Great
Oolite, Quarry near Isolation Hospital, Northleach; L. Richardson Coll., 1910. E67791-6,
E67798-889 and E68070-84, Bathonian (?Sharps Hill Fm.), Eastington, Gloucestershire; P. D.
Taylor Coll., 1979.

OCCURRENCE. Lower-Middle Bathonian of Gloucestershire (Fig. 1). The only accurate strati-
graphical information available is from Eastington where the rocks exposed are probably of the
Sharps Hill Formation, regarded as late early Bathonian (tenuiplicatus Zone) or early mid-
Bathonian (progracilis Zone) in age.

DESCRIPTION. Details of the morphology of A. abbreviatus and A. prattii are considered
together below, p. 45.

REMARKS. This new species has been previously included in A. prattii (Carpenter 1882,
Richardson 1904). However, these smaller crinoids from the Lower-Middle Bathonian can be
distinguished consistently from topotypes of A. prattii which occur higher in the Bathonian at
Lansdown. The stem of A. abbreviatus is less variable and there are none having stems with
more than 10 columnals. The cup is smaller and more bowl-shaped than in A. prattii, basals are
smaller, and accessory plates are less conspicuous, never exceeding one per ray.

NAME. The trivial name abbreviatus alludes to the greatly reduced basals and stem in this
species.

Ailsacrinus prattii (Gray 1828)

1828 Encrinites (Apiocrinites) Prattii Gray: 219.

1831 Apiocrinites obconlcus Goldfuss: 187; pi. 57, figs 5a-n.

1833 A. [Apiocrinites} fusiformis Jelly: 46; pi. 1, figs 1-11.

1840 Millericrinus obconicus (d'Orb.) [sic] d'Orbigny: 80; pi. 14, figs 23-28.

1862 Apiocrinites obconicus nobis [sic]; Goldfuss: 174; pi. 57, figs 5a-n.

non 1881 Millericrinus obconicus d'Orb. -Apiocrinites obconicus Goldf.; Moriere: 85; pi. 1 (—Milleri-
crinus morierei de Loriol 1883).

1882 Millericrinus Prattii (Gray) Carpenter: 29 (partim); pi. 1, figs 1-8, 10-23 (fig. 9 = Ailsacrinus
abbreviatus sp. nov.).

1884 Millericrinus Prattii (Gray) ; de Loriol: 618.

1900 Millericrinus pratti (Gray) ; Bather: 135 ; fig. 52.

1911 Millericrinus prattii (Gray); Kirk: 48; pi. 6, figs 1-6.

1 934 Millericrinus prattii (Gray) ; Gislen : 6 , fig . 7 .

1936 Millericrinus pratti (Gray); Biese: 464.

1 978 Millericrinus prattii (Gray) ; Ubaghs in Ubaghs et al. : T93 ; fig . 70 , 2 .

1978 Liliocrinus prattii (Gray): Rasmussen in Ubaghs etal.: T822; fig. 551, 2a, b, f, g, i, 1.

REVISED DIAGNOSIS. A species of Ailsacrinus with moderately large, bowl-shaped to conical cup;
basals touching adjacent basals and pentagonal in shape; radials similar in height to basals;
accessory plates generally conspicuous, frequently more than one per ray; column tapering
distally and of highly variable length (one to more than 66 columnals); distal facet of proximale
with tuberculate areola, not petaloid.

HOLOTYPE. The single specimen (Fig. 4) described by Gray (1828), BMNH E24663; Great
Oolite (probably Upper Bathonian aspidoides Zone), Lansdown, near Bath.

P. D. TAYLOR

Fig. 4 ^4 ilsacrinus prattii (Gray) . Holotype BMNH E24663 , from the Great Oolite of Lansdown Hill ;

X2.

OTHER MATERIAL. BMNH: 48900, Lansdown; R. Etheridge Coll., 1868. E5722, Great Oolite,
Lansdown. E24664, Great Oolite, ?locality; Bowerbank Coll., 1865. OUM: J14219, Great
Oolite, Kirtlington. SM: J33689-734, J33769-70, J33806-15, J33822, J33834, Great Oolite,
Lansdown; Walton Coll.

DESCRIPTION. See below.

REMARKS. The synonymy lists only those references which give descriptions or figures of A.
prattii. Elsewhere, the species has been cited as an example of an eleutherozoic crinoid or
included in stratigraphical lists of fossils. Pre-1936 references of this type are given by Biese
(1936).

Apiocrinites obconicus was described by Goldfuss (1831) using specimens obtained from the
Great Oolite of Bath. The illustrations of Goldfuss show that these are clearly conspecific with
the earlier Encrinites prattii of Gray. D'Orbigny (1840) acknowledged Goldfuss' species but
appended his own name as author when referring the species to Millericrinus . Crinoids from the
Bathonian of Normandy, considered to be the same species by Moriere (1881), were later
described as Millericrinus morierei by de Loriol (1883). They differ from A. prattii in having
long stems and a more conical, Apiocrinites-like transition between stem and cup.

Jelly (1833) was aware that the Lansdown species had been named Encrinites prattii by Gray
(1828) but, being critical of Gray because he had stated incorrectly that their horizon was Lias
and also employed a specific name derived from a proper name, chose to ignore this name.
Instead, he called the crinoid the 'Lansdown Encrinite' throughout most of his paper before, in a
concluding paragraph (p. 46) stating ' . . . this, perhaps, might not incorrectly be called A.
[Apiocrinites] fusiformis' .

Knowledge of Ailsacrinus prattii (Gray) derives largely from the description and figures of
Carpenter (1882). Although these are based mostly on Lansdown material, one of Carpenter's
figures (pi. 1, fig. 9) is a specimen (BMNH E14661) from Miserden and is here considered to be
A. abbreviatus sp. nov. Bather (1900), Kirk (1911), Gislen (1934) and Ubaghs etal. (1978) all
copied, directly or indirectly, Carpenter's figures.

ABERRANT MILLERICRINID AILSACRINUS

Figs 5-8 Ailsacrinus prattii (Gray), Great Oolite, Lansdown Hill. Specimens coated with ammonium
chloride. Fig. 5, BMNH E5722, crown and proximal columnals of a presumed long-stemmed
individual; x 3-5. Fig. 6, SM J33704, crinoid with a stem composed of a single rounded columnal
attached to the basals; x7. Fig. 7, SM J33700, crinoid with a short conical stem and several
accessory plates in the cup; x 6. Fig. 8, SM J33719, short-stemmed crinoid with irregular lenticular
columnals; x 4.

Morphology

Stem

The feature of Ailsacrinus which has attracted most attention is the short, variable stem tapering
distally towards a rounded terminal columnal.

Carpenter (1882) emphasized the extreme variability in length of the stem among individuals
of A. prattii from Lansdown. The longest-stemmed Lansdown crinoid has an incomplete stem of

P. D. TAYLOR

66 columnals totalling more than 50mm in length (Carpenter 1882: pi. 1, fig. 14). In contrast, the
incomplete stem of another crinoid (pi. 1, fig. 6), although only 27mm long, possesses 58
columnals. At the other extreme is an individual (SM J33704; pi. 1, fig. 18) with a stem
composed of a single, gently convex columnal (Fig. 6). Were it not for the lack of cirri this plate
would be virtually indistinguishable from a comatulid centrodorsal. Three other specimens (SM
J33709, J33712 and J33715) also appear to have stems of one columnal only although these are
higher than that of J33704. Two individuals have stems of two columnals only and ten others
have numbers between 4 and 53; see caption to Fig. 9.

The mean value of stem length in these crinoids is equivalent to 12-5 columnals, though this
figure may be an underestimate of the true population mean because the stems of crinoids with
short stems are more likely to be preserved unbroken. The distribution is very strongly
positively skewed with a mode of 1 columnal per stem.

The stem of A. abbreviatus is generally shorter than A. prattii and there is less variability. Fig.
9 shows the frequency of stems of different length. Again, the mean value of 4-2 columnals may
be an underestimate of the true mean for the same reasons as for A. prattii. The longest stems
consist of 9 columnals, the shortest 1 columnal (Fig. 11), and the modal value for the sample is 2
columnals. The longest stems are about 11 mm long, whereas 3 mm is a typical length.

1 1 -

10-

abbreviatus

N:24

X 7

i 6"

® 5-

^»

2-
1-
ft-

•

123456789

columnals

Fig. 9 Size frequency histogram of stem length (expressed as number of columnals) in 24 individuals
of Ailsacrinus abbreviatus gen. et sp. nov. from Eastington. In contrast, 16 individuals of A. prattii
(Gray) from Lansdown Hill showed the following numbers of columnals: 1 (4 individuals); 2 (2
individuals); 4, 5, 6, 7, 14, 17, 21, 32, 33, 53 (1 individual each).

ABERRANT MILLERICRINID AILSACRINUS 47

Proximal columnals in Ailsacrinus have a greater diameter than distal ones, the stem tapering
away from the cup. The angle of taper tends to be greater in stems with fewer columnals, for
example compare Figs 7 and 20 A. The proximal-distal gradient of decreasing columnal width is
quantified for two long-stemmed individuals of A prattii in Fig. 43 (p. 62) and is discussed below
under Ontogeny (p. 58).

The most distal columnal in the majority of Ailsacrinus specimens has a smooth rounded end
(Figs 6, 8, 11). In some cases the axial lumen is exposed but in others it appears to be plugged, as
in some isocrinids (Ubaghs etal. 1978: T848). Certain specimens of A abbreviates have a distal
columnal which, rather than being rounded, exhibits a corroded petaloid crenularium of a
symplectial articulation facet (Figs 13, 14). Though observed in somewhat weathered material,
this may well have been the condition of the columnal in the living crinoid prior to burial. No
individuals are known with massive encrusting holdfasts, so typical of millericrinids, or the cirri
which typify many other articulates. A small branching structure resembling a stem with some
columnal fusion occurs on a slab (SM J33693) with arms of A. prattii and is depicted by
Carpenter (1882: pi. 1, fig. 5). This was interpreted by both Jelly (1833) and Carpenter (1882) as a
possible 'root' but unfortunately is incompletely preserved and a teratological origin cannot be
discounted.

The stem of Ailsacrinus is homomorphic; nodals are not detectable on the basis of columnal
size. Long-stemmed individuals of A prattii do, however, exhibit a proximal-distal gradient of
changing columnal height. Usually columnals decrease in height away from the cup but in one
specimen (SM J33694) the reverse holds true (Fig. 42, p. 61).

Incomplete or 'lenticular' (Carpenter 1882) columnals are fairly common in both A. prattii
and A abbreviatus. These fail to encompass the entire circumference of the stem (Figs 8, 10,
20 A). When traced around the stem they 'pinch-out' or boudinage and the space they would
have occupied is taken up by thickening of the two contiguous columnals.

A single specimen of A prattii (SM J33707) has an unusual overgrowth extending downwards
from the cup to cover the top of the stem (Carpenter 1882: pi. 1, fig. 21). This irregular structure
consists of two columnal-like plates, one of which is incomplete.

Poorly-preserved stereom on the outer surface of A. abbreviatus columnals appears laby-
rinthic in form with a spacing of 5-10 u,m between elements of the lattice.

Articulations between columnals are symplectial. Externally, the crenellae and culmina are
seen to interlock in a crenulate manner (Fig. 10). The distal facet of the top columnal in A
abbreviatus has a petaloid areola and a quinquelobate lumen (Fig. 14). In A. prattii the
crenularium has crenellae and culmina which are better defined, and the areola is not petaloid

Figs 10-11 Scanning electron micrographs of the stem in Ailsacrinus abbreviatus gen. et sp. nov.,
Bathonian (? Sharps Hill Fm.), Eastington. Fig. 10, BMNH E68070, part of a long stem showing
crenulate symplectial articulations and a small lenticular columnal; x 11. Fig. 11, BMNH E68071,
short stem composed of a rounded columnal corroded in appearance; x 5'5.

P. D. TAYLOR

Figs 12-15 Scanning electron micrographs of the cup in Ailsacrinus abbreviatus gen. et sp. nov.,
Bathonian (? Sharps Hill Fm.), Eastington. Fig. 12, BMNH E68072: A, aboral view showing
broken stem, small basals and radials; x 3-7; B, tuberculate accessory plate located between basals;
x 15-7; C, detail of tubercle; x 83. Fig. 13, BMNH E68073, cup with a solitary columnal attached;
x4-9. Fig. 14, BMNH E68074, corroded petaloid crenularium and axial lumen of top stem
columnal; x 7. Fig. 15, BMNH E68075, stemless specimen showing basal facets; x 9-1.

ABERRANT MILLERICRINID AILSACRINUS

Fig. 16

Height of basals plotted against height of radials in Ailsacrinus abbreviatus gen. et sp. nov.

and A. prattii (Gray).

but has radiating rows of tubercles (Fig. 19). The proximal facet of the top columnal, articulating
with the basals, is a weakly marked symplexy or cryptosymplexy in A. abbreviatus but more
strongly marked in A. prattii. The high pyramidal form of this facet suggests permanent
attachment of the columnal to the cup, i.e. that the topmost columnal is a proximale. However,
the columnal is not united to the cup by a synostosis as in the Recent crinoids discussed by
Breimer in Ubaghs etal. (1978: T25).

Dorsal cup

The cup is small and bowl-shaped (Ubaghs in Ubaghs etal. 1978: fig. 72) in A. abbreviatus (Fig.
46, p. 66), rather larger and more conical in A. prattii (Fig. 5). A depressed base (Fig. 15)
accommodates the pyramidal facet of the top columnal. Syntaxial overgrowths obscure details
of the adoral interior of the cup in all specimens examined. However, external preservation of
plates is good.

Basals are very small and triangular in A. abbreviatus. Externally, adjacent basals -either fail
to touch or barely touch one another (Figs 12-14). They are seen, however, to make contact
internally in specimens whose stems have been removed (Fig. 15). Basals of A. prattii, though

P. D. TAYLOR

Figs 17-19 Ailsacrinus prattii (Gray), Great Oolite, Lansdown Hill. Specimens coated with ammo-
nium chloride. Fig. 17, SM J33707, radial facet; x 11. Fig. 18, SM J33703, basal facet; x 8. Fig. 19,
SM J33708, distal facet of top columnal; x 7.

Figs 20-21 Ailsacrinus pratti (Gray), Great Oolite, Lansdown Hill. Specimens coated with ammo-
nium chloride. Fig. 20, SM J33695: A, long stem with 5 lenticular columnals (arrowed); x 3-9; B,
proximal columnals and tuberculate accessory plate; x 12. Fig. 21, SM J33700, ray containing
several irregular accessory plates, one of which is tuberculate; x 12.

ABERRANT MILLERICRINID AILSACR1NUS

also reduced in size, are larger and touch adjacent basals externally giving them a five-sided
external shape (Figs 5-8). The proximal facet of each basal has a median radial depression and a
marginal crenularium which, like the facet of the adjoining columnal, is well-developed in A.
prattii (Fig. 18) but poorly-developed in A. abbreviates (Fig. 15). The facet between basals and
radials has not been observed.

Radials exceed the height of basals in A. abbreviates but are of about the same height in A.
prattii (Fig. 16). Their distal articulating facet (i.e. that which articulates with the first brachials)
is inclined at a variable angle to the long axis (proximal-distal) of the crinoid. An angle of 15° has
been estimated in A. abbreviates and angles between 30° and 60° in A. prattii. This variability
may relate, at least in part, to the attitude of burial and the nature of plate disarticulation during
compaction. Radial facets have a deep aboral ligamental fossa, small interarticular ligamental
fossae and large muscular fossae (Fig. 17).

In addition to the usual plates of the cup, small accessory plates are a characteristic feature of
Ailsacrinus. These seem to be present in a minority of individuals of A. abbreviates where they
are generally inconspicuous, but are present in most specimens of A. prattii. Those individuals
of A. abbreviates with accessory plates do not usually have them in every ray. Accessory plates
are more numerous in A. prattii, some individuals having up to 3 or 4 plates per ray (Fig. 21).
Their position is perradial, i.e. between basals. Sometimes the accessory plates are in contact
with the top stem columnal (Figs 20B, 22), particularly in A. abbreviates where the basals are
small and not contiguous. In A. prattii accessory plates are often located at a triple junction
between two basals and a radial (e.g. right-hand accessory plate of Carpenter 1882: pi. 1, fig. 3),
or they may be extended distally into embayments within radials (Fig. 21). Large accessory
plates in some A. prattii specimens are located in the basal circlet in contact with both column
and radials (e.g. left-hand accessory plate of Carpenter 1882: pi. 1, fig. 3). Several specimens of
A. prattii have shallow pits in the cup which are in appropriate locations to have accommodated
large accessory plates (Fig. 5). Most are vacant but some are occupied by small accessory plates.
An interesting feature of the accessory plates is the presence of tubercles on some of them.
These resemble the spine-bearing tubercles of echinoids in having a mamelon and a central
foramen (Fig. 12C). The sporadic occurrence of accessory plates led Carpenter (1882: 35) to
suppose that they were 'without any morphological importance'. However, they seem to be
present in too many individuals for Carpenter's opinion to be acceptable. Among living crinoids
they would appear to have no close analogues. Andrew Smith (personal communication 1981)
has suggested a comparison with statocysts, balancing structures developed in several groups of
echinoderms. Kirk (1911) identified accessory plates as infrabasals. Some specimens of A.

Figs 22-24 Ailsacrinus abbreviatus gen. et sp. nov., Bathonian (? Sharps Hill Fm.), Eastington.
Specimens coated with ammonium chloride. Fig. 22, BMNH E68076, unusually large accessory
plate and poorly-preserved radial facet; x 17. Fig. 23, BMNH E68077, facet of the most proximal
syzygy (on the distal side of secundibrach 4) showing numerous culminae; x 18. Fig. 24, BMNH
E68078, oral pinnules lying across the adoral surface of the calyx; x 12.
4*

P. D. TAYLOR

abbreviatus have tabular plates concealed between the basals which resemble the infrabasals of
Liliocrinus polydactylus (d'Orbigny) illustrated by Ubaghs in Ubaghs et al. (1978: fig. 73,5).
However, their relationship with accessory plates is unclear.

Tegmen

Definite tegminal plates have not been identified in Ailsacrinus, although a mass of small plates
originally above the displaced calyx in a specimen of A. prattii (SM J33712) may include
tegminal plates as well as pinnulars. Alternatively, the tegmen in Ailsacrinus may have been like
that of many living comatulids, i.e. naked with the exception of microscopic skeletal elements.

Arms

Arm morphology was discussed little by Carpenter (1882), whose material consisted mostly of
Lansdown crinoids lacking arms or with disarticulated brachials. In contrast, the specimens of
A. abbreviatus from Northleach often display finely-preserved arms (Figs 25-27). In several

Figs 25-27 Ailsacrinus abbreviatus gen. et sp. nov., Bathonian (? Sharps Hill Fm.), Eastington.
Fig. 25, Holotype BMNH E67797, cup and radially-arranged arms (eleven-armed appearance is
due to the presence of an arm from a second individual); xl. Fig. 26, BMNH E67808,
crinoid with regenerated arm arising from the most proximal syzygy (arrowed); x 1-6. Fig. 27,
BMNH E67794, pinnulated arm resting on a bedding plane; x 2- 1 .

ABERRANT MILLERICRINID AILSACRINUS

synarthry

Fig. 28 Diagram showing proximal brachial
articulations in Ailsacrinus gen. nov. With the
exception of the synarthries and the syzygy all
articulations are muscular. R, radial; Br,, 1st
primibrach; Br2, 2nd primibrach.

individuals these seem to be almost completely preserved, interrupted only by minor disarti-
culation and dislocation. However, the tendency for arms to be preserved lying oblique to
bedding means that their entire length is never seen. There would appear to be no significant
difference in arm structure between A. abbreviatus and A. prattii. Both have identical patterns
of articulation and similarly-shaped brachials, though rather larger in A. prattii. The arms divide
only once and hence Ailsacrinus is ten-armed. Arms in A. abbreviatus have been observed to
exceed 19cm in length.

The first primibrach has a muscular articulation with the radial, and the second primibrach is an
axillary (Fig. 28). Therefore the first brachitaxis contains two brachials, a common condition in
articulate crinoids. However, an A. prattii specimen (SM J33709) illustrated by Carpenter (1882:
pi. 1, fig. 23) has two rays containing a third primibrach. A single ray of an A. abbreviatus
individual (BMNH E68072) contains only one primibrach, the axillary, in the first brachitaxis.
These rare variants may be meristic or perhaps pathological.

Articulations between primibrachs 1 and 2, and between secundibrachs 1 and 2, are synarth-
rial (Fig. 28). Synarthrial facets have depressed ligamental fossae on either side of a fulcral ridge
with an adoral-aboral orientation (Fig. 30). Stereom of the ligamental fossae does not show
well-defined galleries in specimens examined by electron microscope (Fig. 31).

A second kind of nonmuscular articulation occurring commonly in the arms of Ailsacrinus is a
syzygy. The first syzygy (Fig. 28) is situated between secundibrachs 4 and 5 (unlike in coma-
tulids, where it generally occurs between secundibrachs 3 and 4; Rasmussen in Ubaghs et al.
1978) and subsequent syzygies are present at frequent intervals along the arm. On average,
about 25% of joints are syzygial, and the intersyzygial interval is 2, 3 or 4 muscular joints. Some,
but not all, lengths of arm of A. abbreviatus display a regular pattern (Fig. 32) of syzygies
separated alternately by 2 and 4 muscular joints.

Syzygial facets have a series of culmina radiating from the axial canal (Figs 23, 33, 34A, B).
The number of culmina per facet is variable, ranging from about 7 (Fig. 34A) to 20 (Fig. 23).
Occasionally, the culmina are reduced to a row of tubercles, in the manner of isocrinid
cryptosyzygies (Breimer in Ubaghs et al. 1978: T38). Culmina of adjoining brachials are clearly
seen to oppose (Figs 35 A, B). Stereom of the culmina is dense and consists of closely-spaced
broad rods, 15-20 u,m in width, linked by small cross-struts (Fig. 34C); orientation of the rods is
approximately concentric about the axial lumen. This results in some of the rods being oblique
to the culmina on which they are situated. Culmina stereom appears to have been strong and
capable of resisting abrasion caused by adjacent brachials rubbing against one another. Though
similarly dense, stereom of the culmina in the living comatulid Nemaster rubiginosa (Macurda,
Meyer & Roux in Ubaghs et al. 1978: fig. 191,4) contrasts with that of Ailsacrinus in having a
knobbly appearance without a directional fabric. Stereom of the crenellae, which is presumed to
have served as an anchorage for ligament, in Ailsacrinus appears to be galleried with a pore
diameter of about 5 |xm (Fig. 34C).

P. D. TAYLOR

29A

Figs 29-31 Scanning electron micrographs of muscular and synarthrial articulations in Ailsacrinus
abbreviates gen. et sp. nov., Bathonian (? Sharps Hill Fm.), Eastington. Fig. 29, BMNH E68079:
A, poorly-preserved, moderately oblique muscular facet; x 10-8; B, fine stereom of the muscular
fossa; x67. Figs 30-31, BMNH E68074: Fig. 30, synarthrial facet; x25; Fig. 31, stereom of the
ligamentary fossa of a second synarthrial facet; x 185. *

Fig. 32 Diagram showing a common pattern of distribution of brachial articulations and pinnules in
the arms (viewed aborally) of Ailsacrinus abbreviatus gen. et sp. nov. Syzygies are beaded and
labelled 'S'; the remaining articulations are muscular.

Brachials situated proximally and distally of a syzygy, hypozygals and epizygals respectively,
are rather narrower than other brachials (Fig. 35A). When arm regeneration is observed, the
regenerated arm always arises from a syzygial joint, in one instance from the most proximal
syzygy in the arm (Fig. 26). Regeneration from a syzygy was illustrated by Jelly (1833: figs 4, 5)
and is a characteristic of most living crinoids (Breimer in Ubaghs et al. 1978: 134).

ABERRANT MILLERICRINID AILSACRINUS

Figs 33-35 Scanning electron micrographs of syzygial articulations in Ailsacrinus gen. nov. Fig. 33,
A. prattii (Gray), BMNH E5722 (fragment), Great Oolite, Lansdown Hill; syzygial facet with
poorly-developed culminae; x 15. Fig. 34, A. abbreviate sp. nov., BMNH E68080, Bathonian (?
Sharps Hill Fm.), Eastington: A, syzygial facet with well-developed culminae; x 18-5; B, culminae
radiating from the axial canal; x52; C, stereom of culmina and crenellae; x 130. Fig. 35, A.
abbreviates sp. nov. , BMNH E68081, Bathonian (? Sharps Hill Fm.), Eastington: A, lateral view of
an arm showing narrow hypozygal and epizygal brachials; x 10-4; B, opposing culminae of
hypozygal and epizygal; x 56.

Muscular facets are rarely observed in A. abbreviatus because pre-fossilization arm breakage
invariably occurred at syzygies and articulated brachials are now firmly bound together with
syntaxial calcite overgrowths. Unfortunately stereom preservation is poor in Lansdown A.
prattii where muscular articulation facets are more commonly visible. Muscular articulations in

P. D. TAYLOR

Ailsacrinus are slightly to moderately oblique. Facets show a large aboral ligamental fossa and
have an adoral region above the fulcral ridge where a poorly-defined break in slope appears to
separate large interarticular ligamental fossae from small muscular fossae (Fig. 29 A). A layer of
fine stereom apparently overlies coarser stereom in the muscular fossa of a poorly-preserved
specimen of A. abbreviatus (Fig. 29B).

Pinnules

Pinnulation in Ailsacrinus is relatively complete. The two primibrachs and secundibrach 1 lack
pinnules, and so the first pinnule arises from secundibrach 2. Thereafter pinnules arise on
alternate sides of the arm from each brachial, with the exception of hypozygals preceding
syzygies. The distribution of syzygies means that along each side of the arm, pinnules are borne
on either every second or on every third brachial (Fig. 32). There are at least two distinct types of
pinnules in Ailsacrinus, an oral series and a distal series.

Oral pinnules occur in the proximal parts of the arms and are generally found lying across the
adoral surface of the calyx (Fig. 24). There seem to be two or three pairs of oral pinnules on each
arm. Although complete oral pinnules have not been seen, their preserved length and taper
suggests that they are shorter than distal pinnules. Oral pinnules lack ambulacral grooves and
associated cover plates, and their pinnulars are short and high. Each pinnular has an adoral

, -*<>n&W'

».cy . -^
>-• ... '••'' •"' *1

* >•** * *' '£H,*,/% .

Figs 36-38 Scanning electron micrographs of pinnulars and cover plates in Ailsacrinus abbreviatus
gen. et sp. nov., Bathonian (? Sharps Hill Fm.), Eastington. Fig. 36, BMNH E68082; A, boot-
shaped cover plates; x 75; B, coarse, irregular cover plate stereom; x 180. Fig. 37, BMNH E68083,
rectangular cover plates collapsing into the ambulacral groove; x62. Fig. 38, BMNH E68084,
fragment of an oral pinnule with one complete pinnular and part of a second pinnular; x 53.

ABERRANT MILLERICRINID AILSACRINUS

median transverse ridge (Fig. 38), broad at the two outer edges of the pinnular and narrow at the
centre. Coarse stereom flanks the ridge. In profile, this ridge forms a triangular-shaped top to
the pinnular. The nature of articulations between oral pinnulars is unknown. The function of
oral pinnules in living crinoids is unclear but they may have roles in defence, manipulation of
food particles, and/or fixation.

Distal pinnules have ambulacral grooves and are composed of comparatively elongate pin-
nulars (Fig. 39 A). The maximum observed length of a distal pinnule in A. abbreviates is over
2 cm in an incomplete pinnule composed of 28 pinnulars. However, another distal pinnule in the
same species was complete and consisted of 21 pinnulars. Distal pinnules are terminated by a
pinnular which tapers to a point. The articulation between brachials and pinnular 1 is muscular
with a fulcral ridge orientated almost parallel to the length of the arm. The articulation between
pinnulars 1 and 2 is also muscular but the articulation between pinnulars 2 and 3 is synarthrial.
The presence of a large aboral ligament is responsible for the gap commonly observed between
the short pinnular 1 and pinnular 2, when viewed from the side (Fig. 39B). Stereom on the outer
sides of distal pinnulars is fascicular (Fig. 39C), rather like that of syzygial culmina but
contrasting with stereom of the brachials (Fig. 39D). Broad rods (15-20 u-m in width) connected
by cross-struts parallel the length of the pinnular and are orientated slightly obliquely to the
surface of the pinnular. The rods have distally directed pointed ends (Fig. 39C). Minute
rectangular or boot-shaped cover plates border the ambulacral groove (Figs 36A, B, 37).

>tf

(•nMMKfc

'&&

•*V,

r «**y* '• - ^
>*^^l

ry'Ty^V '*" ^
^ &**"$** ***

••*f"-*£'*Jp6fA* **J> .31

wm^M

*•*•#

Fig. 39 Scanning electron micrographs of brachials and distal pinnules in Ailsacrinus abbreviates gen.
et sp. nov. , BMNH E68083, Bathonian (? Sharps Hill Fm.), Eastington. A, general view; x 16-4. B,
prominent aboral ligament fossa visible between first and second pinnulars; x 34. C, fascicular
stereom of pinnular; x 175. D, brachial stereom; x 175.

58 P. D. TAYLOR

Depending on their length , each pinnular may have two, three or even four pairs of cover plates.
The stereom of cover plates is irregular and labyrinthic, with a pore diameter between 3 and 20
ixm (Fig. 36B).

In some specimens of A abbreviates, pinnules situated immediately distal to the oral pinnules
do not usually have preserved cover plates although they are otherwise indistinguishable from
typical distal pinnules. By analogy with living comatulids (Breimer in Ubaghs et al. 1978: T43),
the position of these pinnules suggest that they may have been genital pinnules.

Reconstruction

The appearance of a complete A. abbreviatus individual is reconstructed in Fig. 40. Of
particular note is the disproportionately short stem relative to arm length.

Colour

Individuals of both A. prattii from Lansdown and A. abbreviatus from Northleach may be
conspicuously coloured. Their colour varies from pale grey-purple to dusky red-purple and is
usually restricted to the calyx and column. One side of the crinoid is sometimes more deeply
coloured than the other (e.g. BMNH E5722) and distribution of the colour may be patchy. It
seems possible that this colouration is a remnant of an original pigmentation. Living crinoids are
often deeply pigmented (Hyman 1955) and, although pigmentation is fugitive (spirit-preserved
specimens tend to lose their colour), it is known that organic pigments can survive fossilization in
crinoids. Blumer (1960, 1962) extracted hydrocarbon pigments, 'fringelites', from U. Jurassic
Millericrinus . He interpreted their preservation as indicating a strongly reducing environment
beneath the sediment-water interface. Proof that the colouration of Ailsacrinus is due to similar
organic pigments would necessitate time-consuming chemical analysis which has not been
undertaken. However, it may be significant that the largest specimen of A. abbreviatus (BMNH
E67807) from Northleach is also the most intensely coloured; living crinoids concentrate
pigment during life and thus become more deeply coloured as they grow.

Stem ontogeny

Undoubtedly the feature of Ailsacrinus that has attracted most attention is the highly variable
stem. Hypotheses regarding the ecology of Ailsacrinus must take into account stem morphology
and variability. Stem morphology observable in specimens of Ailsacrinus is the outcome of
ontogenetic processes which acted during the life of the crinoids.

It is thought that all crinoids pass through a stage during their development when they are
fixed firmly to the substrate by means of a stem or column. In the cystidean and pentacrinoid
stages of early ontogeny in comatulids (Breimer in Ubaghs et al. 1978), the column may possess
many and well-differentiated columnals. For example, John (1938) describes a pentacrinoid of
the comatulid Notocrinus virilis Mortensen with a crown 2-2 mm long and a 10mm long column
comprised of 45 columnals. Comatulids end their pentacrinoid stage when autotomy causes the
crown to break free of the column. However, in stalked crinoids the crown remains attached to
the column and a pentacrinoid stage may not be readily distinguishable (Breimer in Ubaghs et
al. 1978: T56).

The early ontogeny of Ailsacrinus was most probably like that of comatulids, with a fixed
pentacrinoid stage followed by a free-living adult stage. The duration of the fixed stage may have
been short, as in comatulids, or more protracted. Kirk (191 1: 49) believed that detachment took
place in late ontogeny, not much earlier than crinoids represented in some of Carpenter's (1882)
figures of small Ailsacrinus individuals. If the supposed 'root' of A prattii (Carpenter 1882: pi.
1, fig. 5) is truly a holdfast, then Kirk's belief may be correct because the structure possesses
apparent columnals of a moderately large size. However, in the absence of small individuals
there is no way of confirming or refuting this suspicion.

Stem growth in stalked crinoids is achieved by the formation of new columnals together with
accretionary growth of existing columnals (Ubaghs in Ubaghs etal. 1978: T82). Ailsacrinus has a
homomorphic stem (nodals and internodals are not recognizable), apparently with a fused top

ABERRANT MILLERICRINID AILSACRINUS

Fig. 40 Reconstruction of Ailsacrinus abbreviates gen. et sp. nov. Arms have a total length of about
15-20 cm and are depicted in an arbitrary orientation which was not necessarily their position during
feeding.

60 P. D. TAYLOR

columnal or proximale. In this type of stem, columnal addition is localized to a generating area
immediately beneath the proximale. Continued columnal addition pushes earlier columnals
further down the stem and gives a proximal-distal gradient of increasing columnal age. The
initial width of each new columnal is equivalent to that of the proximale at the time of
columnal formation, i.e. about the same width as the base of the cup. Enlargement of the
generating area during ontogeny causes new columnals to become successively wider. If this
were the only factor controlling columnal width the stem would taper away from the cup.
However, a second factor is the accretionary growth of existing columnals. This factor in
isolation would produce a stem which tapered towards the cup because older columnals are
situated furthest from the cup. Final stem-form is a result of interaction between these two
factors (see Seilacher, Drozdzewski & Haude 1968). Axial growth in columnal height occurs
concurrently with transverse columnal growth. If all new columnals initially had the same
height, then there should be a proximal-distal gradient of increasing columnal height towards
older columnals situated near the base of the stem.

Reduction in length of the stem is a further possibility in crinoids especially pertinent in the
case of Ailsacrinus. This could result from either the shedding of whole columnals (cf . isocrinids;
Rasmussen 1977) by autotomy or accident, or columnal resorption. Bather (1900: 191) said of
A. prattii 'the crown breaks off from the root, the stem is gradually resorbed, and a free-floating
stage attained'. Kirk (1911: 49) believed that the column was shortened by 'the dropping off of
some of the distal columnals accompanied by more or less resorption'.

Consequently, there are three main possibilities for post-detachment stem ontogeny in
Ailsacrinus:

1 . stem lengthening by addition of columnals, and/or growth of existing columnals,

2. no change in stem length, or

3. stem shortening by shedding columnals and/or resorption.

Several lines of enquiry can be followed to decide which of these is the most likely.

Evidence from single crinoids

LENTICULAR COLUMNALS. Some individuals of both species have stems with incomplete or
lenticular (Carpenter 1882) columnals. These columnals, instead of extending all the way
around the circumference of the stem, when traced laterally in either direction are seen to
'pinch-out' or boudinage. They may be of slight lateral extent (Fig. 10) or may encompass most
of the stem (Fig. 20A). A specimen of A. prattii (Fig. 20A), incorrectly drawn by Carpenter
(1882: pi. 1, fig. 7), has four lenticular columnals aligned exactly above one another and
alternating with complete columnals. Lenticular columnals are not exclusive to Ailsacrinus; de
Loriol (1877-9) figured similar structures in Millericrinus and Apiocrinus.

Carpenter (1882: 33) regarded lenticular columnals as columnals in the process of formation,
i.e. columnals fossilized in an early ontogenetic state. If this opinion is correct then the
occurrence of lenticular columnals points to columnal addition during late ontogeny, probably
after detachment.

Little is known of the process of columnal addition in crinoids. Ubaghs (in Ubaghs et al. 1978:
fig. 60, 1,2) illustrates longitudinal sections through juvenile and mature portions of the column
of Silurian Barrandeocrinus. In the juvenile column, new columnals are present as thin discs
which taper away from the column axis and are not visible at the surface. The mature column has
columnals of even thickness, all reaching the surface of the column. Columnals are apparently
introduced in a similar manner in articulates (Rasmussen in Ubaghs et al. 1978: T269), begin-
ning as thin concealed discs. These immature columnals are clearly unlike the lenticular
columnals of Ailsacrinus. Growth of lenticular columnals to completion and uniform thickness
would necessitate transfer of skeletal material from adjacent thickened columnals (Fig. 41).
This complication suggests that lenticular columnals are not columnals in the process of
formation. Further evidence against Carpenter's hypothesis comes from the distribution of
lenticular columnals which are often found distal to the expected site of columnal addition

ABERRANT MILLERICRINID AILSACRINUS

1 2 3

Fig. 41 Three-stage diagram showing that growth to completion of lenticular columnals of the type
depicted in Fig. 20A (p. 50) would occur at the expense of adjoining columnals.

immediately beneath the proximale. Therefore, no significance can be given to lenticular
columnals in stem ontogeny.

VARIATION IN COLUMNAL SIZE. Gradients of change in columnal size in a proximal-distal
(young-old) direction are readily quantifiable in long-stemmed individuals of A. prattii.
Turning first to columnal height, the usual pattern is one of decreasing height in a distal direction
away from the cup (J33695 of Fig. 42), i.e. presumed older columnals are shorter than younger
columnals. However, in at least one specimen (SM number J33694 of Fig. 42) columnal height
increases away from the cup, i.e. presumed older columnals are taller than younger columnals.
The occurrence of this reverse trend means that columnal height cannot be used as a reliable
indicator of columnal age and provides no useful information about stem ontogeny pre- or
post-detachment.

1-0
0-9

I °"

J. 0-7

0-6

0-5

g 0-4-

2 0-3

W 0-2

0-1

A A A A A A A

•• A Af A A

AAA!*** •••••••

• J 33695

* J 33694

0 5 10 15 20 25 30 35 40 45 50 55 60 65

number from cup

Fig. 42 Change in columnal height away from the cup in two long-stemmed individuals (SM J33695

and J33697) of Ailsacrinus prattii (Gray).

P. D. TAYLOR

•J 33695
- J 33694

•

0 5 10 15 20 25 30 35 40 45 50 55 60 65

number from cup

Fig. 43 Change in columnal width away from the cup in two long-stemmed individuals (SM J33695

and J33697) of Ailsacrinus prattii (Gray).

Columnal width always decreases away from the cup (Fig. 43), as is shown by the distal taper
of Ailsacrinus stems. Stems with fewer columnals are generally found to taper more steeply than
stems with a large number of columnals (cf. Figs 7 and 20A). If the width of columnals were
proportional to the size of the generating area, then this proximal-distal size gradient would
reflect addition of successively wider columnals as the crinoid grew. However, the second factor
of accretionary growth after inception must be taken into account. Even if no new columnals
were added after detachment, some amount of accretionary growth is likely to have occurred in
order that stem width should keep pace with increasing cup width. Consequently, columnal
width gradients do not help in resolving the problem of stem ontogeny.

DISTALMOST COLUMNAL. The columnal terminating the stem in Ailsacrinus most typically has a
blunt, rounded end (Figs 8, 1 1). In some short-stemmed individuals of A. abbreviatus, however,
the distalmost columnal has a worn quinquelobate symplectial facet (Fig. 13) which may have
been a life condition rather than a result of preburial stem fracturing. The axial lumen may be
exposed or occluded at the base of this columnal. It has not been possible to identify dense
fabrics indicative of resorption but this may be due to poor preservation of stem stereom.
However, it is clear from the general shape of the distalmost columnal that it has undergone
some sort of modification, though this could be the result of any combination of post-
detachment columnal growth, resorption and mechanical abrasion.

Evidence from crinoid populations

Assuming size to be a reasonable indicator of the age of a crinoid, comparisons of stem length
between individuals of differing size should throw light on post-detachment ontogeny of the
stem in Ailsacrinus. The overall size of the crinoid is impossible to determine even in these
exceptionally well-preserved crinoids. A frequently employed measure of crinoid size, that of
the dimensions of the dorsal cup (e.g. Roux 1978), is not suitable in Ailsacrinus because it is
influenced by burial attitude - the cup is shortened and splayed outwards in crinoids buried
upright relative to crinoids buried prostrate. In order to quantify crinoid size the dimension
chosen was the height of the axillary brachial, an easily defined and measured parameter which
would appear to have no causal correlative link with stem length.

in
<*)

-S

o
n

3
C

0 •£ c
52 •- w

C C/5 U.

5 o D

O CX-C

•s|l

^ c 2
c £-0

11!

Ijl

ill

0-,

=3 C

x 2

CO OB

«.s
5 ts

_, c

-a

> 4) >

O N S
C -^ 4>

ex.
x

a 1 S .8

2 ? ^'5

C ^) Q W

'C !s u u
o 2 " H

Si«

0£

64 P. D. TAYLOR

Number of columnals and height of the axillary brachial were determined in 24 specimens of
A. abbreviates from Eastington and 14 specimens of A. prattii from Lansdown. These para-
meters were found to be positively correlated in the A. abbreviatus sample but not so in the
A. prattii sample where there is a wide scatter of points (Fig. 44). Therefore, the A. abbreviatus
data are consistent with an ontogenetic net increase in columnal number (i.e. columnal addition
exceeding columnal shedding) but a similar hypothesis is not supported by data from the
longer-stemmed A. prattii. But this result is suspect because of the probable existence of high
levels of non-ontogenetic variation (e.g. in accessory plate and lenticular columnal develop-
ment) within populations of Ailsacrinus . This non-ontogenetic 'noise' superimposed over
ontogenetic variability may be responsible for the trend evident in A. abbreviatus and the lack of
trend in A. prattii.

Comparison between individuals in a population is useful in discounting the possibility of
ontogenetic shortening of stems by resorption of stereom more or less equally from each
columnal (as opposed to resorption of the distalmost columnal only). Individuals have short
stems because they have few columnals not because they have columnals of lesser height.

Conclusion

The dynamics of stem ontogeny in Ailsacrinus are equivocal. The presence of lenticular
columnals in some stems cannot be taken as evidence for addition of columnals; proximal-distal
gradients of columnal size-change are variable and can be interpreted in more than one way;
distal columnals with exposed symplectial facets in A. abbreviatus might indicate some shedding
of columnals; and crinoid size: column length comparisons within populations suggest post-
detachment net addition of columnals in A. abbreviatus but not in A. prattii.

Palaeoecology

Broad environment

The Middle Jurassic sediments of southern England are predominantly carbonates. They
accumulated in a shallow shelf sea where conditions were influenced by the presence of the
London-Belgian Island in the east, a probable source of fresh water and terrigenous clastic
material (see Ware & Windle 1981). Comparatively open, marine-shelf environments existed in
the south and west during the Bathonian (for example, around Bath). Environments of the
Northleach area may have often been more stressful for marine biota as a result of closer
proximity to land (see Palmer 1979).

Sellwood & McKerrow (1974) discussed the stratigraphy and depositional environments of
the lower part of the Bathonian in Oxfordshire and north Gloucestershire including the
Northleach region. They recognized three stratigraphical divisions: Chipping Norton For-
mation, overlain by Sharps Hill Formation and then Taynton Limestone Formation. The fauna
of the Sharps Hill Formation, in which A. abbreviatus probably occurs, is predominantly
marine, although the presence ofLiostrea life assemblages may indicate some salinity restriction
(as, for example, some of the present day Florida Keys). Water depth is believed to have been
little more than 3m. The Sharps Hill Formation appears to grade laterally into the upper part of
the Lower Fullers Earth Formation which is well-developed further south and west. This led
Sellwood & McKerrow (1974) to ascribe deposition of the Sharps Hill Formation to a minor
transgression which caused deeper-water sediments of Lower Fullers Earth Formation lithology
to spread onto the carbonate-dominated area of north Gloucestershire and Oxfordshire.

Green & Donovan (1969) described the Great Oolite of the Bath region but did not deal with
outliers north of the River Avon such as Lansdown Hill. They divided the Great Oolite
sequence, from bottom to top, into Combe Down Oolite, Twinhoe Beds, Bath Oolite and
Upper Rags. It is not known where the Lansdown A. prattii locality fits within this succession.
The Bath area was apparently located on the outer part of a stable carbonate shelf. The Combe
Down Oolite is interpreted as a shallow-water deposit formed by oolite deltas which flanked
tidal flats with migrating channels (like the present day Trucial Coast). While the succeeding

ABERRANT MILLERICRINID AILSACRINUS 65

Twinhoe Beds may have been deposited in quieter and rather deeper water, the Bath Oolite
seems to make a return to conditions similar to those inferred for the Combe Down Oolite. The
Upper Rags may represent a more varied depositional regime. Analysis of a single bed exposed
on Bathampton Down (Elliott 1974) suggests that it accumulated on a current-swept, inter-reef
seafloor like some modern environments which exist in water 30-60 m deep off the Capricorn
Islands (Great Barrier Reef) and the Bermudas.

The Eastington crinoid bed

A detailed study has been undertaken of the A. abbreviates bed from the new locality near
Eastington. Within the confines imposed by poor exposure, this has allowed a tentative model to
be proposed for the genesis of the Eastington crinoid bed incorporating sedimentological,
palaeoecological and taphonomic inferences.

LITHOLOGY. The well-sorted bio-oosparite (Fig. 3, p. 41) comprising the bulk of the bed is
clean-washed and was undoubtedly formed, although not necessarily deposited, in a compara-
tively agitated environment. Features indicating a storm deposit ('tempestite' of Ager, 1974),
such as matrix-supported intraclasts, are notably absent. The subparallel orientation of shell
fragments (Fig. 3A) suggests grain by grain deposition rather than the nearly instantaneous
deposition caused by a storm. The thin muddy layers which drape each fossiliferous bedding
plane stand in marked contrast to the bio-oosparite. The major clay mineral present in this mud
is illite; there are no clay minerals which may be interpreted as. having a volcanogenic origin.
This is important because elsewhere in the British Jurassic, Ali (1977) has postulated smothering
by volcanic ash as a source of catastrophic mortality of corals. Survival of discrete muddy layers
without mixing with the clean-washed carbonate sediment suggests that the mud was stabilized
during carbonate deposition. Stabilization of mud in Recent sedimentary environments is
commonly achieved by the presence of an algal mat or other organic film (e.g. Bathurst 1975:
122).

SEDIMENTARY STRUCTURES. Bioturbation is absent from the crinoid bed and this is true for the
Sharps Hill Formation in general (Sellwood & McKerrow 1974). Burrowing animals, whose
activities would probably have disarticulated the buried crinoids, may have been excluded by
rapid deposition (Sellwood & McKerrow 1974) or by unfavourable anaerobic conditions
beneath the sediment surface (cf. Rosenkranz 1971). The existence of anaerobic conditions is
supported by apparent pigment preservation in A. abbreviates; Blumer (1960) ascribes preser-
vation of organic pigments in fossil crinoids to the presence of strongly reducing conditions. This
in turn is consistent with the possibility of an organic film stabilizing the muddy sediment.

Traces of symmetrical ripples occur near the top of the crinoid bed. These wave-generated
bedforms would have been produced in shallow water above wave base.

CRINOID PRESERVATION. Most specimens of A. .abbreviates are exceptionally well preserved,
lacking post-mortem abrasion and with delicate structures (e.g. pinnule cover plates) usually
intact and unbroken. The crinoids are well-articulated, especially near the base of the bed. The
overlying shell lag deposit contains short, articulated fragments, possibly reworked from the
crinoid bed below. Where arms are disarticulated, the amount of dislocation between the
disarticulated portions tends to be small. Similar preservation in other fossil crinoids is generally
attributed to rapid burial in situ or with very minor transportation (e.g. Brett 1978, Brower 1973,
Hess 1972, 1973). Aslin (1968) also suggests rapid burial to account for good preservation of
echinoids in rocks of Middle Jurassic age in Northamptonshire. Working with living crinoids,
Cain (1968: 192) found that, within two days of death, specimens oiAntedon bifida in still water
'collapsed into a mass of arms and cirrus fragments'. Similarly, comatulids studied by Liddell
(1975) were completely disarticulated within two days of death when placed in an agitated
environment but crinoids buried and then exhumed after six days were well-preserved and
retained their colour. However, considerable interspecific variation in the disarticulation rates
of Recent crinoids and ophiuroids was found by Meyer (1971). Scavenging organisms played an
important role in the disarticulation process. These studies on living crinoids provide strong

P. D. TAYLOR

Figs 45-46 Ailsacrinus abbreviates gen. et sp. nov., Bathonian (? Sharps Hill Fm.), Eastington.
Fig. 45, BMNH E67807, short-stemmed crinoid in an upright orientation on the underside of a
bedding plane; x 2-3. Fig. 46, BMNH E67817, long-stemmed crinoid lying prostrate on a bedding
plane; x 1-9.

evidence for rapid burial of individuals of A. abbreviates. This probably occurred while the
crinoids were still alive or, less likely, shortly after death. Transport of dead crinoids over
significant distances can be discounted but it is possible that they were swept to their burial site
before death.

CRINOID ORIENTATION. Over 50% of the crinoids studied are preserved in presumed life-position,
i.e. with their adoral surfaces facing upwards. In these individuals the arms diverge radially from
the cup (Fig. 25, p. 52) and lie parallel or almost parallel to the bedding. The short stem is
orientated perpendicular to the bedding (Fig. 45). About 40% of the crinoids are prostrate,
typically with arms close together and arms and stem parallel or subparallel to the bedding (Fig.
46). Some individuals are obliquely orientated and a few are upside down. There is no
discernible alignment of prostrate crinoids or groups of arms on the bedding planes. Tangling of
arms is rare despite high concentrations of specimens. Many arms are bent, flexed either
aborally or adorally . Some bent arms extend upwards through a few centimetres of sediment.

This orientational evidence shows that at least some of the crinoids were disturbed from their
presumed life positions before or during burial. The arms of partially buried crinoids are
unlikely to have projected above the sea-bed for very long before disarticulating. The occur-
rence of arms orientated at high angles to the bedding thus again suggests rapid deposition, and
their lack of alignment that the crinoids were not buried in a regime of strongly directional
currents.

POPULATION DENSITY. The mud-draped bedding planes are characterized by a high density of
crinoids, exceeding 200/m2 in some instances. High population density may be attributed to
concentration by currents or burial of a densely aggregated living population. Dense aggre-
gations of fossil crinoids, so-called 'crinoid gardens' (Moore & Teichert in Ubaghs et al. 1978:
T8), are well known among pelmatozoic species (e.g. Brower 1973, Brett 1978). Highly
aggregated populations are also a feature of some Recent comatulids (see Reese 1966; Breimer
in Ubaghs et al. 1978: T328); Marr (1963) for example illustrates an Antarctic sea-bed densely
colonized by comatulids. According to Keegan (1974), Antedon bifida may achieve population
densities of 1200/m2 on the west coast of Ireland. Aggregation may arise from poor larval
dispersal, selection of favourable habitats already populated by conspecifics, or truly gregarious

ABERRANT MILLERICRINID AILSACR1NUS

behaviour (preferential settlement of larvae close to conspecifics). The advantages of aggre-
gation have been considered by Warner (1971, 1979). He notes that it is likely to promote
cross-fertilization, increase stability in current-swept areas, allow the mutual support of arms
raised in suspension feeding and enhance settlement of food particles from suspension because a
mass of individuals forms an effective current baffle. Wilson, Holme & Barratt (1977) suggest
protection from predation as a further advantage of aggregation in echinoderms. It is difficult to
discount conclusively concentration by currents as the cause of high population density in A.
abbreviatus, but the alternative hypothesis of burial of an aggregated living population is more
appealing in view of the analogues which exist among comatulids at the present day.

POPULATION VARIABILITY AND STRUCTURE. Variation in axillary brachial height, used as a con-
venient indicator of crinoid size (see p. 62), shows the variability of A. abbreviatus in the
Eastington crinoid bed (Fig. 47). A sample of 128 crinoids derived from various parts of the bed
has a size frequency distribution which is almost normal. In contrast, a subsample of 37 crinoids
on a single bedding plane has a negatively skewed distribution. Interpretation of these patterns
of distribution can be made either on the premise that they show variation within a population of

35n

30J

X
u

C
Q)

3
<T

height Ax

Fig. 47 Frequency histogram of crinoid size (expressed as height of the axillary brachial in mm) in 128
individuals of Ailsacrinus abbreviatus gen. et sp. nov. from the Bathonian (? Sharps Hill Fm.) of
Eastington. A subsample of 37 individuals from a single bedding plane is unshaded.

68 P. D. TAYLOR

equal-aged individuals, or that size reflects age and the distribution reveals the demographic
structure of the crinoid population. In reality, the distribution is likely to be the result of a
combination of these non-ontogenetic and ontogenetic factors. However, for present purposes
it will be assumed that ontogenetic factors predominate and the data will be analysed
accordingly.

Hallam (1972) reviewed the interpretation of population structure in fossils. He distinguished
between living populations and death assemblages, each of which is likely to produce a different
type of size frequency distribution. The size frequency histogram for a living population is very
often polymodal because recruitment to the population tends to be episodic (e.g. seasonal),
giving distinct age/size classes. For death assemblages a unimodal distribution is more probable
and the shape of the distribution is dependent largely on rates of growth and mortality. Benthic
assemblages usually have positively skewed distributions due to high juvenile mortality, while
normal distributions are unusual, and negatively skewed distributions highly exceptional.

The size frequency distribution of A. abbreviates fits neither that typical of a living popu-
lation (though population structure in crinoids specifically may be unknown) nor that typical of a
death assemblage. If the model proposed below for the formation of the Eastington crinoid bed
is correct, then the crinoids represent a succession of living populations which were catastro-
phically buried. The size frequency distribution of the large sample of crinoids from throughout
the bed may be a mixture of several living populations. However, that of crinoids on a single
bedding plane could reflect the structure of a single population. This is a negatively skewed
distribution; large, presumed old, individuals are over-represented. Such a population structure
is consistent with continuous recruitment of adult crinoids into the population following a period
of attached life elsewhere. Assuming growth rate to have decreased during ontogeny, the
proportion of large crinoids in the population would increase with time, yielding a negatively
skewed size frequency histogram.

TENTATIVE MODEL OF CRINOID BED FORMATION. Evidence for rapid sediment-deposition and
disturbed crinoid burial seems clear. The sediment responsible for crinoid burial appears to have
been the clean-washed carbonate sand rather than the mud of the bedding-plane drapes.
Significant transport of crinoids before burial is thought unlikely and sediment inundation of
densely-aggregated living populations in situ or very locally transported seems more probable.
A multi-event model seems more compatible with the evidence than a single event model. This
model can be summarized as follows:

1 . Colonization of the sea-bed by crinoids and deposition of muddy sediment in fairly quiet
water aided by the baffling action of the crinoid arms. Once deposited, the mud may have been
stabilized by an organic film which also promoted anaerobic conditions within the sediment and
precluded infauna whose activities may otherwise have disarticulated crinoids already buried.
Adult crinoids were recruited into the densely aggregated crinoid population by migration from
sites of attachment located elsewhere.

2. Rapid influx of clean-washed carbonate sand generated in a higher energy environment
and possibly introduced by tidal currents. Some of the crinoids were buried immediately and
retained their upright life-orientation, but others were disturbed from their life-orientation,
locally transported, and buried prostrate or upside down.

3. Return to normal conditions with recolonization by crinoids and recommencement of mud
sedimentation.

4. Repetition of this sequence of events to give the full thickness of the crinoid bed.

5. Change in the sedimentary regime signalled by the deposition of a cross-bedded shell lag
over the crinoid bed, perhaps due to the advance of a dune field.

Stem function

Inference of stem function is important in understanding the mode of life of Ailsacrinus. As
there are no living crinoids of known ecology which have a stem morphology similar to that of
Ailsacrinus, a functional morphological approach has been applied to the problem of stem
function (Rudwick 1961).

ABERRANT MILLERICRINID AILSACRINUS 69

Potential functions of the stem in Ailsacrinus are:

1 . as a column to hold the crown aloft;

2. as a means of attaching the crinoid to the substrate;

3. as ballast to stabilize the crinoid;

4. as a counterpoise to keep the crinoid in an appropriate orientation; or

5. without function, at least during the unattached period of life.

COLUMN. The paradigm for a column functioning to elevate the crown is a stout structure with
little flexibility between columnals, thickened at its base where stresses caused by horizontal
water-movements could be concentrated, and flat-bottomed or firmly attached to the substrate
by some means. This paradigm, approached in articulates such as Apiocrinites (Breimer & Lane
in Ubaghs etal. 1978: T334), is clearly not fulfilled by Ailsacrinus .

ATTACHMENT. Attachment structures or holdfasts are part of the fossilizable skeleton in various
crinoids. Alternative but equally viable solutions to anchoring the crinoid are provided by
different types of holdfast fitted to soft substrates, hard bottoms, substrates with a complex
relief, etc. (see Brett 1981). Paradigms for attachment generally involve structures of expanded
surface area (e.g. cemented bases or divided distal ends of the stem) and/or with the ability to
grasp (e.g. comatulid cirri). It seems clear that the stem of Ailsacrinus lacked any adaptation for
attaching the crinoid.

BALLAST. The function is fulfilled by any structure denser than sea-water. The optimal weight of
ballast required might be expected to vary according to the unknown factors of net crinoid
buoyancy and strength of environmental water currents. Therefore, it is difficult to assess the
possible role as ballast of the stem in Ailsacrinus.

COUNTERPOISE. The paradigm for a counterpoise structure providing stability has a low centre of
gravity close to the substrate. This is well illustrated by the stemless inadunates Agassizocrinus
(Ettensohn 1975) and Paragassizocrinus (Ettensohn 1980), which have heavily-calcified infra-
basal cones giving them a 'roly-poly doll' construction. The counterpoise paradigm may be
approached adequately in some very short-stemmed individuals of Ailsacrinus but it is certainly
not fulfilled in long-stemmed individuals of A. prattii where the centre of gravity is likely to have
been located in the proximal part of the stem some way above the substrate.

FUNCTIONLESS. To argue effectively that a structure is functionless it is necessary to eliminate all
possible functions. This is clearly impossible if only for the reason that not every function may
have been conceived. However, a hypothesis which deserves consideration for Ailsacrinus is
that the stem was functional (e.g. for attachment and crown elevation) during the attached stage
of ontogeny but essentially functionless when the free-living stage was reached. In Recent
isocrinids (Rasmussen 1977) the long cirriferous stem may fracture at the cryptosymplexy
beneath a nodal, leaving the crown and proximal part of the stem to drift away before becoming
re-attached elsewhere. Stem fracturing may represent true autotomy or, as believed by
Rasmussen, breakage in response to water movements or other external forces. A similar
process of stem fracturing, but without subsequent cirral reattachment, may be envisaged for
Ailsacrinus, perhaps at a late stage in ontogeny (see above). If caused by external pressure, stem
breakage may have left individuals with stems of widely varying length. Thereafter, individuals
possibly lacked the ability to shorten the stem and relied on occasional accidental shedding of
columnals.

Feeding ecology

Present interest in the feeding ecology of living and fossil crinoids focuses on an apparent
polarization into current-seeking rheophiles and current-avoiding rheophobes (Breimer 1969,
Breimer & Lane in Ubaghs et al. 1978: T333). Rheophiles, present among Recent stalked
crinoids and comatulids, commonly form brachial filtration fans (Magnus 1967). The arms are
spread in a paraboloid with their adoral surfaces pointing upcurrent. A radial feeding posture is
less common. Rheophobes, possibly a minority of living crinoids, include deep water comatulids

70 P. D. TAYLOR

(Peres 1958, 1959) in which the arms form a collecting bowl for feeding on the plankton rain.
However, there may be considerable variation in feeding behaviour within some species and the
ecological dichotomy into rheophiles and rheophobes may break down. For example, La
Touche (1978) observed that flexibility in the arm movements of Antedon bifida allowed
individuals to feed in diverse current regimes. In slack water, animals most commonly held their
arms in an inverted cone. Animals in current speed of up to 30cm/s held their arms in the shape
of a bent-over, flattened cone.

Breimer & Lane (in Ubaghs et al. 1978) discuss features of the stalk and arms useful as a guide
to inferring the palaeoecology of fossil crinoids. Species of Millericrinus with a rudimentary
stem (evidently referring to Ailsacrinus) they consider (1978: T334) to be benthic rheophobes.
Certainly it is difficult to envisage Ailsacrinus forming a radial brachial filtration fan for
rheophilic feeding; the stem is not long enough to hold the crown high enough aloft. The lack of
anchorage structures seems to be another problem. Individuals lack grasping cirri and also
hooks or spines on the arms and pinnules which are used for attachment in some comatulids (e.g.
Comanthina schlegeli) that secondarily lose their cirri (Meyer & Macurda 1977). However, not
all rheophiles form brachial filtration fans and, it Ailsacrinus lived in dense populations, stability
may not have necessitated direct attachment to the substrate. Ecological analogies may be valid
with the living brittle star Ophiothrix fragilis (see Warner 1979). Like Ailsacrinus , O. fragilis is a
suspension feeding echinoderm living in dense aggregations composed of individuals lacking a
means of direct attachment to the sea bed. Arms of individuals are stretched upwards into the
current and interlock to form a mat stable in velocities exceeding 20cm/s. Aggregations are
probably maintained by preferential settlement of larvae around adults, combined with the
ability of dislodged adults to locate aggregations and walk towards them.

To summarize, Ailsacrinus may have been either a rheophobic or rheophilic suspension
feeder living in dense populations. Recruitment into these crinoid mats seems to have taken
place by immigration of individuals which had attained an adult size during a protracted period
of attached ontogeny spent elsewhere. The active locomotory abilities of Ailsacrinus were
perhaps limited; arms are robust and muscular fossae small. Migration was more likely achieved
by passive drifting or comasterid-like crawling than by antedonid-like swimming. This ecological
model, devised largely using evidence from A. abbreviates, may or may not apply to A. prattii.

Evolution

Phylogenetic affinities

The Articulata are divided into seven orders by Ubaghs (in Ubaghs et al. 1978: T364):
Millericrinida, Cyrtocrinida, Bourgueticrinida, Isocrinida, Comatulida, Uintacrinida and
Roveacrinida. Evolutionary relationships between these orders are poorly understood (see
Rasmussen in Ubaghs etal. 1978: T302-5 ; Pisera & Dzik 1979). Although articulate crinoids are
usually considered to have evolved from the Poteriocrinina, a group of dicyclic Palaeozoic
inadunates, their structural diversity is taken by some to suggest a polyphyletic origin (Ubaghs in
Ubaghs etal. 1978: T281).

Cyrtocrinids, bourgueticrinids, uintacrinids and roveacrinids are well-defined and morpho-
logically distinctive groups which can be eliminated from any discussion of the affinities of
Ailsacrinus. This leaves the millericrinids, isocrinids and comatulids, all of which are known in
deposits of Bathonian age. Millericrinids are characterized by the presence of a proximale and a
lack of cirri. The column typically forms a conical transition to the cup. Nodals are absent,
although the alternately large and small columnals in Angulocrinus (Rasmussen in Ubaghs etal.
1978: fig. 550,2) resemble nodals and internodals respectively. Isocrinids possess nodals and
cirri but lack a proximale. Comatulids have a reduced stem consisting of a single cirriferous
centrodorsal or, in early forms, a few reduced nodals (Hess 1951). Previously, Ailsacrinus (as
Millericrinus prattii) has been assigned to the millericrinids.

Table 1 summarizes the principal morphological characters shared by Ailsacrinus with typical
millericrinids, isocrinids and comatulids. These characters are discussed in turn below.

ABERRANT MILLERICRINID AILSACRINUS 71

Table 1 Morphological characters shared by Ailsacrinus with Millericrinida, Isocrinida and
Comatulida.

Ailsacrinus Millericrinida Isocrinida Comatulida

Stem

reduced

large

reduced

Proximal columnal

fused

free

fused

Cirri

absent

present

Basals

reduced

large

reduced

Oral pinnules

present

absent

present

Syzygies

present

absent?

present

STEM. A reduced stem, shared by Ailsacrinus and comatulids, is presumably an advanced
character but is not a reliable synapomorphy because stem reduction has undoubtedly occurred
several times during crinoid evolution (see Kirk 1911). Furthermore, reduction of the stem to a
single centrodorsal in comatulids seems to have been the result of columnal fusion, a process for
which there is no evidence in Ailsacrinus.

PROXIMAL COLUMNAL. A fused proximal columnal is present as a proximale in Ailsacrinus and
millericrinids, and as a centrodorsal in comatulids. Although fusion may be an advanced
character and is absent in isocrinids, homology between the proximale of Ailsacrinus and the
centrodorsal of comatulids is unlikely because the former is united to the cup by a symplectial
articulation and the latter by a synostosial articulation.

CIRRI. By outgroup comparison with poteriocrininids, cirri may be a primitive character of the
Articulata. If so, absence of cirri is an advanced character shared by Ailsacrinus and milleri-
crinids. Character absences may, however, be unreliable indicators of phylogenetic affinity.

BASALS. Reduced basals are shared by Ailsacrinus, isocrinids and comatulids. Despite the fact
that small basals may be an advanced character state (by outgroup comparison), basal reduction
is a comparatively simple process with a high probability of occurring more than once. If so,
reduced basals are not a good synapomorphy.

ORAL PINNULES. Among living crinoids, oral pinnules are said to be restricted to comatulids
(Breimer in Ubaghs et al. 1978: T43). They appear not to have been described from non-
comatulid articulates prior to this account of Ailsacrinus . The fine structure of oral pinnules in
Ailsacrinus differs somewhat from those of comatulids. Oral pinnules of the antedonid
Promachocrinus are slender, with long, elongate pinnulars (Breimer in Ubaghs et al. 1978:
fig. 28,4), whereas those of comasterids possess distinctive distal pinnulars equipped with teeth
to form terminal combs (Breimer in Ubaghs et al. 1978: fig. 28,3). In Ailsacrinus the pinnulars
are short and high, and lack terminal combs. The phylogenetic significance of oral pinnules in
Ailsacrinus is difficult to assess because of this disparity in detailed structure, and because oral
pinnules may be present in other fossil millericrinids but have not been recognized owing to
inferior preservation. The extant millericrinid suborder Hyocrinina provides no help in
resolving this problem because these living crinoids are devoid of pinnules on their proximal
brachials.

SYZYGIES. Syzygies are well-developed in Ailsacrinus and, as in comatulids, the hypozygals lack
pinnules. The frequency of syzygies along the arms of Ailsacrinus is similar to that of coma-
tulids. However, the first syzygy in Ailsacrinus occurs between secundibrachs 4 and 5 whereas
the usual position for this ligamentary articulation in comatulids is between secundibrachs 3
and 4. There are clear differences in stereom ultrastructure between the syzygial facets of
Ailsacrinus and comatulids, those of Ailsacrinus having a subconcentric fascicular fabric.
Although Rasmussen (in Ubaghs et al. 1978: T817) states that syzygies are absent in milleri-
crinids, facets of disarticulated brachials from another millericrinid, Apiocrinites, often bear a
pattern of radiating ridges suggestive of a syzygial articulation. Furthermore, the occurrence of

P. D. TAYLOR

DC
O

DC
LJJ

CO
HI

H
<

I-
DC

o
oc

o
o

O
Q.

ABERRANT MILLERICRINID AILSACRINUS 73

possible syzygies in certain poteriocrininids (Strimple in Ubaghs etal. 1978: T301) suggests that
they may be a primitive character of articulate crinoids and of no value in deciding relationships
in the group.

The foregoing discussion serves to highlight the acute need for more information on the
morphology of articulates, especially their brachial articulations and pinnule structure. Until
this is forthcoming, relationships within the group will remain obscure. Therefore the favoured
cladogram (Fig. 48) showing the relationships between Ailsacrinus and other articulates is very
tentative. Ailsacrinus is here interpreted as a millericrinid possessing autapomorphies (reduced
stem and basals, oral pinnules, well-developed syzygies) resulting in a morphology convergent
with comatulids.

Adaptive evolution

When discussing the post-Palaeozoic evolution of crinoids, Meyer & Macurda (1977) focused on
the impressive adaptive radiation shown by the Comatulida. They considered articulates to be
pre-adapted for an eleutherozoic existence because they possess muscular arms which are
potentially of value in crawling and swimming, as in comatulids. Active locomotion is used by
comatulids to seek favourable habitats and to avoid stress. Meyer & Macurda (1977) identified
predator stress, imposed by diversifying teleost fish, as an important selective factor during
comatulid evolution.

In view of the similar time of origin of comatulids (known from the Toarcian) and Ailsacrinus,
it is tempting to explain the origin of Ailsacrinus in identical terms. However, eleutherozoic
lifestyles may have characterized many other extinct crinoids and seem to have arisen several
times among Palaeozoic taxa. These Palaeozoic crinoids were neither pre-adapted in the sense
of having muscular arms nor subjected to the predator stress of teleosts. The origin of both
comatulids and Ailsacrinus in Jurassic times may be quite coincidental.

Temporal trends of morphological change apparent within the genus Ailsacrinus are the
opposite of those expected. The older species, A. abbreviates, resembles less the putative
stalked ancestor of the genus than does the later A. prattii, which usually has a longer stem and
altogether more bulky morphology. Early notions of Ailsacrinus were of a crinoid evolving
towards a fully eleutherozoic lifestyle by elimination of the stem. The modest evidence from the
two known species fails to support this hypothesis.

Of Millericrinus prattii, Kirk (1911: 49) said '. . . were Millericrinus to have possessed cirri,
there is small doubt that this very species would have formed the radicle of a line essentially
comatulid in habit, and perhaps of considerable vigor.' This viewpoint remains an appropriate
epitaph for an ecological excursion by the millericrinids into an eleutherozoic lifestyle which
proved unsuccessful in terms of duration and taxonomic fecundity.

Acknowledgements

This study was initiated during the tenure of a N.E.R.C. fellowship at the University College of
Swansea. I am grateful to Frank Cross, John Hicks, Tony Ramsay and Patricia Taylor for
assistance in the field, to Frank Cross for help during specimen preparation, to Miss A. M.
Clark, Mr A. Gale, Dr R. P. S. Jefferies, Mr D. N. Lewis, Dr G. D. Sevastopulo, Dr A. Smith
and Mr H. L. Strimple for their constructive comments on the manuscript, and to Dr G. F.
Elliott and Mr S. Donovan for discussion. Loans of specimens from the Sedgwick Museum were
arranged through the kindness of Dr C. L. Forbes.

References

Ager, D. V. 1974. Storm deposits in the Jurassic of the Moroccan High Atlas. Palaeogeogr. Palaeodirnat.
Palaeoecol., Amsterdam, 15: 83-93.

AH, O. E. 1977. Jurassic hazards to coral growth. Geol. Mag., London, 114: 63-64.

Aslin, C. J. 1968. Echinoid preservation in the Upper Estuarine Limestone of Blisworth, Northampton-
shire. Geol. Mag., London, 105: 506-518.

74 P. D. TAYLOR

Bather, F. A. 1900. Echinoderma. In Lankester, E. R., A Treatise on Zoology, 3. 344 pp. London.

Bathurst, R. G. C. 1975. Carbonate sediments and their diagenesis . 2nd Edition, xix + 658pp. Amsterdam.

Biese, W. 1936. Crinoidea jurassica II. Fossilium Cat. (Pars Animalia), 73: 241-544. Berlin.

Blumer, M. 1960. Pigments of a Fossil Echinoderm. Nature, Land. 188: 1100-1101.

1962. The organic chemistry of a fossil - II. Some rare polynuclear hydrocarbons. Geochim. cos-

mochim. Acta, London, 26: 228-230.
Breimer, A. 1969. A contribution to the paleoecology of Paleozoic stalked crinoids. Proc. K. ned. Akad.

Wet., Amsterdam, (B) 72: 139-150.
Brett, C. E. 1978. Description and paleoecology of a new Lower Silurian camerate crinoid. /. Paleont.,

Chicago, 52: 9 1-103.
1981. Terminology and functional morphology of attachment structures in pelmatozoan echino-

derms. Lethaia, Oslo, 14: 343-370.
Brower, J. C. 1973. Crinoids from the Girardeau Limestone (Ordovician). Palaeontogr. am., Ithaca, 7

(46): 263-499.
Cain, J. D. B. 1968. Aspects of the depositional environment and palaeoecology of crinoidal limestones.

Scott. J. Geol, Edinburgh, 4: 191-208.

Carpenter, P. H. 1882. On some new or little-known Jurassic Crinoids. Q. Jlgeol. Soc. Land. 38: 29-43.
Cope, J. C. W., Duff, K. L., Parsons, C. F., Torrens, H. S., Wimbledon, W. A. & Wright, J. K. 1980. A

correlation of Jurassic rocks in the British Isles. Part Two: Middle and Upper Jurassic. Spec. Rep. geol.

Soc. Lond. 15: 1-109.
Elliott, G. F. 1974. Note on the palaeoecology of a Great Oolite fossil-bed at Bath (English Jurassic). Proc.

Geol. Ass., London, 85: 43-48.
Ettensohn, F. R. 1975. The autecology of Agassizocrinus lobatus. J. Paleont., Chicago, 49: 1044-1061.

1980. Paragassizocrinus: systematics, phylogeny and ecology. /. Paleont., Chicago, 54: 978-1007.

Gislen, T. 1934. Evolutionary series towards death and renewal. Ark. Zool., Stockholm, 26A (16): 1-19.
Goldfuss, A. 1826-33. Petrefacta Germaniae, erster Teil. 252 pp., 71 pis. Dusseldorf.

1862. Petrefacta Germaniae, erster Teil. (2nd edn). 234 pp., 71 pis. Leipzig.

Gray, J. W. 1828. Description of a new kind of Pear-Encrinite found in England. Lond. Edinb. Dubl. phil.

Mag., London, 4: 219-220.
Green, G. W. & Donovan, D. T. 1969. The Great Oolite of the Bath area. Bull. geol. Surv. Gt Br.,

London, 30: 1-63.
Hallam, A. 1972. Models involving population dynamics. In Schopf, T. J. M., Models in Paleobiology:

62-80. San Francisco.
Hess, H. 1951. Ein neuer Crinoide aus dem mittleren Dogger der Nordschweiz (Paracomatula helvetica

n. gen. n. sp.). Eclog. geol. Helv., Basel, 43 (for 1950): 203-216.
1972. Eine Echinodermen-Fauna aus dem mittleren Dogger des Aargauer Juras. Schweiz. palaeont.

Abh., Basel, 92: 1-87.

1973. Neue Echinodermenfunde aus dem mittleren Dogger des Aargauer Juras. Eclog. geol. Helv.,

Basel, 66: 625-656.
Hyman, L. H. 1955. The Invertebrates: Echinodermata. 763 pp. New York.
Jaekel, 0. 1918. Phylogenie und System der Pelmatozoen. Palaeont. Z., Berlin, 3 (1): 1-128.
Jelly, H. 1833. The Lansdown Encrinite. Bath Bristol Mag. 2: 37^6, plate.
John, D. D. 1938. Crinoidea. 'Discovery' Rep., Cambridge, 18: 121-222.
Keegan, B. F. 1974. The macrofauna of maerl substrates on the west coast of Ireland. Cah. Biol. mar.,

Roscoff, 15: 513-530.
Kirk, E. 1911. The structure and relationships of certain eleutherozoic Pelmatozoa. Proc. U.S. natn. Mus.,

Washington, 41: 1-137.
La Touche, R. W. 1978. The feeding behaviour of the featherstar Antedon bifida (Echinodermata:

Crinoidea). J. mar. biol. Ass. U.K., Plymouth, 58: 877-890.
Liddell, W. D. 1975. Recent crinoid biostratinomy. Abstr. Progm geol. Soc. Am., Boulder, Col., 7:

1169.
Loriol, P. de 1877-9. Monographic des Crinoides Fossiles de la Suisse. Abh. schweiz. paldont. Ges., Basel,

4-6: 1-300.
1882-4. Crinoides. Paleontologie franfaise, ou description des fossiles de la France. Ire Ser.,

Animawc invertebres. Terrain jurassique. 11 (1). 627pp., Atlas 121 pis. Paris.
Magnus, D. B. E. 1967. Ecological and ethnological studies and experiments on the echinoderms of the

Red Sea. Stud. trop. Oceanogr., Miami, 5: 635-664.
Marr, J. W. S. 1963. Unstalked crinoids of the Antarctic continental shelf: notes on their history and

distribution. Phil. Trans. R. Soc., London, (B) 256: 327-379.

ABERRANT MILLERICRINID AILSACRINUS 75

Meyer, D. L. 1971. Post mortem disarticulation of Recent crinoids and ophiuroids under natural condi-
tions. Abstr. Progmgeol. Soc. Am., Boulder, Col., 3: 645-646.
& Macurda, D. B. 1977. Adaptive radiation of the comatulid crinoids. Paleobiol., Menlo Park, 3:

74-82.

Moore, R. C., Lalicker, C. G. & Fischer, A. G. 1952. Invertebrate fossils. 766 pp. New York.
Moriere, M. J. 1881. Deux genres de Crino'ides de la Grande Oolithe. Bull. Soc. linn. Normandie, Caen,

(3) 5: 78-87.
Orbigny, A. d' 1840-1. Histoire naturelle, generate et particuliere, des Crinoides, vivants et fossiles,

comprenant la description geologique et zoologique de ces animaux. 98 pp. , 18 pis. Paris.
Palmer, T. J. 1979. The Hampen Marly and White Limestone Formations: Florida-type carbonate lagoons

in the Jurassic of central England. Palaeontology, London, 22: 189-228.
Peres, J. M. 1958. Remarques generates sur un ensemble de quinze plongees effectuees avec le bathyscaphe

FRNS III. Annls Inst. oceanogr. Monaco 35: 254-285.
1959. Deux plongees au large du Japon avec le bathyscaphe francais FRNS III. Bull. Inst. oceanogr.

Monaco 1061: 1-8.
Pisera, A. & Dzik, J. 1979. Tithonian crinoids from Rogoznik (Pieniny Klippen Belt, Poland) and their

evolutionary relationships. Eclog. geol. Helv., Basel, 72/3: 805-849.
Rasmussen, H. W. 1977. Function and attachment of the stem in Isocrinidae and Pentacrinidae: a review

and interpretation. Lethaia, Oslo, 10: 51-57.
Reese, E. S. 1966. The complex behavior of echinoderms. In Boolootian, R. A., Physiology of Echino-

dermata: 157-218. New York.
Richardson, L. 1904. A handbook to the geology of Cheltenham and neighbourhood. 303 pp. Cheltenham.

1933. The country around Cirencester. Mem. geol. Surv. U.K., London. 119 pp.

Rollier, L. 1911. Fossiles nouveaux ou peu connus des Terrains Secondaires (Mesozoiques) du Jura et des

contrees environnantes. Vol. 1: Echinodermes, Anneolides, Spirobranches et Lamellibranches. Abh.

schweiz. paldont. Ges., Basel, 37: 1-32.
Rosenkranz, D. 1971. Zur Sedimentologie und Okologie von Echinodermen - Lagerstatten. Neues Jb.

Geol. Paldont. Abh., Stuttgart, 138: 221-258.
Roux, M. 1978. Ontogenese, variabilite et evolution morphofunctionnelle du pedoncule et du calice chez

les Millericrinida (Echinodermes, Crinoides). Geobios, Lyon, 11: 213-241.

Rudwick, M. J. S. 1961. The feeding mechanisms of the Permian brachiopod Prorichthofenia. Palae-
ontology, London, 3: 450-471.

Schlotheim, E. F. von 1882-3. Nachtrdge zur Petrefactenkunde. 214 pp., 37 pis. Gotha.
Seilacher, A., Drozdzewski, G. & Haude, R. 1968. Form and function of the stem in a pseudoplanktonic

crinoid (Seirocrinus). Palaeontology, London, 11: 275-282.
Sellwood, B. W. & McKerrow, W. S. 1974. Depositional environments in the lower part of the Great Oolite

Group of Oxfordshire and north Gloucestershire. Proc. Geol. Ass., London, 85: 189-210.
Ubaghs, G. etal. 1978. Echinodermata 2. In Moore, R. C. & Teichert, C. (eds), Treatise on Invertebrate

Paleontology, T (1-3). 1027 pp., 619 figs. Lawrence, Kans.
Ware, M. & Windle, T. M. F. 1981 . Micropalaeontological evidence for land near Cirencester, England, in

Forest Marble (Bathonian) times: a preliminary account. Geol. Mag., London, 118: 415^20.
Warner, G. F. 1971 . On the ecology of a dense bed of the brittle-star Ophiothrixfragilis. J. mar. biol. Ass.

U.K., Plymouth, 51: 267-282.

- 1979. Aggregation in Echinoderms. In Larwood, G. & Rosen, B. R., Biology and Systematics of

Colonial Animals: 375-396. London.
Wilson, J. B., Holme, N. A. & Barrett, R. L. 1977. Population dispersal in the brittle-star Ophiocomina

nigra (Abildgaard) (Echinodermata: Ophiuroidea). J. mar. biol. Ass. U.K., Plymouth, 57: 405-439.
Woodward, H. B. 1894. The Jurassic Rocks of Britain. Vol. IV. The Lower Oolitic Rocks of England

(Yorkshire Excepted). Mem. geol. Surv. U.K., London. 628 pp.

Index

The page numbers of the principal references are in bold type. An asterisk ( * ) denotes a figure.

Aalenian 41 aggregation 66-8, 70

aboral ligament 57 Ailsacrinus gen nov. 37-73 passim, esp. 42, 43-4, 53*
accessory plates 42-3, 48*, 50*, 51-2, 51*, 64 abbreviatm sp. nov. 38-9, 40*, 41*, 41 , 42-3, 44,

acknowledgements 73 46-7, 47*, 48*, 49*, 49, 51-3, 51*, 52*,

adaptive evolution 73 54*, 55*, 55, 56*, 57-8, 57*, 59*, 62, 63*,

Agassizocrinus 69 64-5, 66* , 67-8, 67* , 70, 73

P. D. TAYLOR

prattii 38-9, 41-2, 43-A 44*, 45*, 45-7,
49*, 49, 50*, 51-3, 55*, 55, 58, 60-1, 61*, 62*,
63*, 64, 69-70, 73
ambulacral grooves 56-7, 56*
Andoversford 41
Angulocrinus 70, 72
Antedon bifida 65-6, 70
Apiocrinites 44, 69, 71 ; see Encrinites

fusiformis 43-4

obconicus 43-4
Apiocrinus 60
areola 47

arms 38, 42, 52-6, 52*, 55*, 66, 70
Articulata 60, 70-2
articulation facet, symplectial 47
articulations, see muscular, synarthries, syzygies
aspidioides Zone 42-3
attachment 69; see holdfast
autotomy 58, 60, 69

ballast 69

Barrandeocrinus 60

basal 43, 48*, 49*, 49, 51-2, 71-2

facet 48*, 50*
Bath 44, 64; see Lansdown Hill

Oolite 64-5
Bathampton Down 65
Bathonian 38, 39*, 40-4, 47-8, 51-2, 54-7, 64,

66-7, 70

Beckford's Tower 42
Bermudas 65
bivalves 41 ; see Liostrea
Bourgueticrinida 70
brachia!51-7,57*,69,71

articulations 53*, 54*, 72-3

axillary, height of 62-4, 67
brachiopods41
British Museum (Natural History) 38, 42

calyx 58; see cup
Capricorn Islands 65
Carpenter, P. Herbert 38
centrodorsa!71

Chipping Norton Formation 64
cirri 47, 69-73
cladogram 72
Clark, AilsaM. 42
colour 58, 65

columnals, column 42-3, 45-7, 46*, 48*, 49, 50*,
5 1,58, 60-1, 64, 69, 71; see stem

lenticular 47, 50*, 60-1, 61*, 64

number of 62-4

terminal (distalmost) 45-7, 62, 64

variation in size 61-2, 61*, 62*
Comanthina schlegeli 70
comasterids71

Comatulida, comatulids46, 52-3, 65-6, 69-73
Combe Down Oolite 64-5
corals 65

Cornbrash, Lower 42

Corsham39,41,43

counterpoise 69

cover plates 56-8, 56*

crenellae47, 53,55

crenularium 47, 48*, 51

'crinoid gardens' 66

cryptosyzygies 53

culminae47,53,55*,57

cup (dorsal cup, calyx) 42-3, 48*, 49, 51-2, 51*,

52*, 56, 60, 62, 66, 71
Cyrtocrinida 70
cystidean stage 58

death assemblage 68
disarticulation 65
discus Zone 42
distal pinnules 56-7, 57*

Eastington, near Northleach, Glos. 38-9, 43,
46-8, 51-2, 54-7, 63-7

crinoid bed 40*, 41*, 65-8; tentative model 68
echinoids41-2,51,65
Encrinites (Apiocrinites) milleri 42

prattii 42^
environment 64-5
epizygals54,55*
evolution 71-3

feeding ecology 69-70

fish, teleost73

fixation 58

Florida Keys 64

Forest Marble 43

fossae 51

'fringelites' 58

Fuller's Earth Formation, Lower 64

genital pinnules 58

Great Oolite 38-9, 41-5, 50, 55, 64

holdfast 47, 58, 69
Hyocrinina71
hypozygals 55-6, 55*, 71

infrabasals 5 1-2

internodals 58, 70

Ireland 66

Isocrinida, isocrinids 42, 47, 53, 60, 69-72

Jurassic 58, 64-5, 73; see Bathonian, &c.
Kirtlington39,42,44

'Lansdown Encrinite' 38, 41-2, 44

Lansdown Hill, N. of Bath 38-9, 41-6, 50, 52, 55,

58, 63^1
Lias 41, 44
ligamental fossae 53, 54*, 56, 57*

ABERRANT MILLERICRINID AILSACRINUS

Liliocrinus polydactylus 42, 52

prattii 42-3
Liostrea 64
Iithology41,65
localities 38-42, 39*
lumen 47, 48*, 53, 62

mamelon51

Mantell Collection 38, 43

Millericrinida, millericrinids 38, 42-4, 47, 70-3

Millericrinidae 42-4

Millericrinina 42-4

Millericrinus 42, 58, 60, 70

milleri 42

morierei 42-3

obconicus 43

polydactylus 42

prattii (pratti) 38, 41-4, 70, 73
Miserden (Park) 39, 41 , 43-4
morphology 45-58
Morris, J., Collection 41, 43
mud, stabilization of 65, 68
murchisonae Zone 41
muscular articulations 53, 54*, 55, 57

fossae 56

Nature Conservancy Council 39

Nemaster rubiginosa 53

nodals47,58,69,71-2

Normandy 44

Northamptonshire 65

Northleach 38-9, 41, 43, 58, 64; see Eastington

Notgrove 39, 41

Notocrinus virilis 58

ontogeny 38; see stem

Ophiothrix fragilis 70

ophiuroids 65

oral pinnules 56-8, 56*, 71, 73

orientation 66

Oxford University Museum 38, 42

palaeoecology 38, 64-70
Palaeozoic 70, 73; see Silurian
Palmer, T. J. 42
Paragassizocrinus 69
Pentacrinites 42
pentacrinoid stage 58
phylogenetic affinities 38, 70-3
pigmentation, see colour
pinnulars 52, 56-8, 56*, 57*, 71
pinnules 38, 42, 51*, 54*, 56-8, 70-1 , 73
population 62^4

density 66-7

variability and structure 67-8
Poteriocrinina 70-3

predator stress 73
preservation 41 , 65-6
primibrachs53, 56
progracilis Zone 39, 43
Promachocrinus 71
proximale 43, 49, 60-1 , 70-1

radial 43, 48*, 49*, 51

facets 50*, 51
reconstruction 58, 59*
reduction in stem length 60
regeneration of arms 54
resorption, columnal 60, 62, 64
rheophiles 69-70
rheophobes 69-70
Richardson, L. , Collection 39, 41 , 43
'root' 60; see holdfast
Roveacrinida 70

secundibrachs 53, 56, 71

Sedgwick Museum, Cambridge 38, 42

sedimentary structures 65

Sharp's Hill Formation 39-40, 43, 47-8, 51-2,

54-7, 64-7
Silurian 60

stalk, see stem, column
statocysts51

stem, stalk 45-7, 46*, 47*, 48*, 49, 66, 69-71, 73;
see columnals

function 68-9

length 62-4

ontogeny 58, 60-4

reduction of 7 1-3
stereom 38, 41, 47, 53, 54*, 55-8, 56*, 57*, 62, 64,

'Stonesfield Slate' 38, 43
stratigraphy 64-5
synarthries 53, 54*, 57
syzygies 38, 42, 51*, 53, 54*, 55*, 55-6, 71, 73

tabular plates 52

Taynton Limestone Formation 39, 64

legmen 52

tenuiplicatus Zone 39, 43

terminal combs 71

Toarcian 73

Trucial Coast 64

tubercles 48*, 49, 50*, 51

Twinhoe Beds 64-5

Uintacrinida 70
Upper Rags 64—5

Walker, J. F., Collection 43

Windrush39,41

woody fragments 41

Accepted for publication 7 July 1982

The Earth Generated and Anatomized
by William Hobbs

An early eighteenth century theory of the earth
Edited with an introduction by Roy Porter

March 1981. 158pp. 6 plates, 8 text figures

Bulletin of the British Museum (Natural History)

Historical Series Vol. 8 Paper covers, £16.00

Wherein is shewn What the Chaos was: How and when the Oyster-shells,
Cockle-shells, and all the other Marine productions, were brought upon, and
incorporated in the Rocks and Mountains of the Earth. Proving that it was not at,
or by the Deluge, as is Vulgarly Supposed.
Also

Why and When the Said Hills and Mountains were raised. As also shewing not
only the certain Cause of the Ebbing and Flowing of the Tides, But even the Two
places where they are once in XV Dayes originally Moved; and where they
ultimately meet each other. By which, as also by diverse Arguments, the Vulgar
Notion of the Moon's Governing them, is fully confuted.
Together with

Many other Philosophical Doctrines and Discoveries; Suitable to such a Subject;
not before advanced.

(From Hobbs' title page)

The Bulletin of the British Museum (Natural History) is published in five Series,
Zoology, Entomology, Botany, Geology, and Historical. Details and complete lists
are free on request.

Publication Sales British Museum (Natural History)
Cromwell Road London SW7 5BD

Titles to be published in Volume 37

Taxonomy of the arthrodire Phlyctaenius from the Lower or Middle
Devonian of Campbellton, New Brunswick, Canada.

By V. T. Young

Ailsacrinus gen.nov.: an aberrant millericrinid from the Middle
Jurassic of Britain. By P. D. Taylor

Miscellanea

Printed by Adlard & Son Ltd, Bartholomew Press, Dorking, Surrey

Bulletin of the

British Museum (Natural History)

Miscellanea

Geology series Vol 37 No 3 24 November 1983

Publications Sales,

British Museum (Natural History),
Cromwell Road,

London SW75BD,
England.

Wo rid List abbreviation: Bull. Br. Mus. nat. Hist. (Geol.)

The Geology Series is edited in the Museum's Department of Palaeontology
Keeper of Palaeontology: Dr H. W. Ball
Editor of the Bulletin: Dr M. K. Howarth

Assistant Editor: Mr D. L. F. Sealy

ISSN 0007-1471

Geology series

British Museum (Natural History) Vol 37 No 3 pp 79-171

Cromwell Road
London SW7 5BD Issued 24 November 1983

Miscellanea

Contents

Glossopteris anatolica sp. nov. from uppermost Permian strata in south-east
Turkey. By S. Archangelsky & R. H. Wagner

The crocodilian Theriosuchus Owen, 1879 in the Wealden of England.
By E. Buffetaut

A new conifer species from the Wealden beds of Feron-Glageon, France.

By H. L. Fisher and J. Watson

Late Permian plants including Charophytes from the Khuff Formation of

Saudi Arabia. By C. R. Hill & A. A. El-Khayal

British Carboniferous Edrioasteroidea (Echinodermata). By A. B. Smith

A survey of Recent and fossil Cicadas (Insecta, Hemiptera-Homoptera)in

Britain. By P. E.S.Whalley

The Cephalaspids from the Dittonian section at Cwm Mill, near Abergavenny,
Gwent. ByE. I. White &H. A.Toombs

105

113
139

149

Glossopteris anatolica sp. nov. from uppermost
Permian strata in south-east Turkey

S. Archangelsky

Urquiza 1132, Vicente Lopez 1638, Buenos Aires, Argentina

R. H. Wagner

Department of Geology, The University, Beaumont Building, Brookhill, Sheffield S3 7HF

Synopsis

A full description is given of the leaf impressions recorded in 1962 as Glossopteris cf . stricta Bunbury from
the Hazro flora in south-east Anatolia. Comparisons are made with several species from India, South
America (Patagonia) and Antarctica. Glossopteris anatolica is regarded as an immigrant from the Gondwana
Realm which reached the equatorial belt in latest Permian times. The composition of the Hazro flora is
commented on in the light of a current revision.

Introduction

The presence of Gondwana elements in the Late Permian flora of Hazro in south-eastern
Anatolia, Turkey, which is predominantly of Cathaysian affinity, has been reported by Wagner
(1959, 1962). Most important in this respect was a species of Glossopteris which was compared
with the Indian Gondwana species G. stricta Bunbury. Plumstead (in Discussion of Wagner
1962) criticized the identification and hinted strongly at the possibility that homeomorphy might
have given rise to leaf shapes and venations similar to those of Glossopteris from Gondwanaland.
The identification as Glossopteris was supported by Lacey (in Discussion of Wagner 1962) and
by Archangelsky & Arrondo (1970: 81, footnote). Asama (1976: 203), on the other hand,
regarded 'the plant reported from the Hazro flora as Glossopteris ... to have been derived
from the Euramerian plant Lonchopteris by Enlargement.' There is a marked difference
between the fernlike fronds of the pteridosperm Lonchopteris and the Glossopteris type leaves
of the Hazro region. The reference to Enlargement would tend to imply a comparison between
pinnules and entire leaves, a comparison which cannot be sustained on morphological grounds.
If Asama's principle of fusion and reduction is applied, there should be vestiges of scaled down,
fused pinnules in the entire leaves found in the Hazro area. These do not occur. The Hazro
specimens are sufficiently complete to dismiss the notion that large pinnules rather than entire
leaves might be represented.

The original collection from Hazro was made in a single afternoon and in view of the
considerable variety of plant remains obtained on that occasion, it seemed useful to return to the
locality and to gather a larger collection. This aim was finally realized in 1979 and 1980 when R.
H. Wagner had the opportunity to study the stratigraphical succession in the Hazro inlier, at
about 70km ENE of the provincial capital Diyarbakir, and to collect new material including
several specimens of the disputed species of Glossopteris. A short list with partly revised
identifications was given in Fontaine et al. (1980) and a paper providing stratigraphical details is
being prepared in collaboration with E. Demirta§li of the Mineral Research and Exploration
Institute of Turkey. The present paper is the first contribution to a full description of the floral
elements collected from the Upper Permian rocks at Hazro. Foraminiferal data reported by
Fontaine et al. (1980) have allowed dating the plant-bearing succession as Dzhulfian, i.e. the
highest Permian.

Bull. Br. Mus. nal. Hist. (Geol.) 37 (3): 81-91 81 Issued 24 November 1983

82 S. ARCHANGELSKY & R. H. WAGNER

Acknowledgements

R. H. Wagner is grateful to the Maden Tetkik ve Arama Enstitiisii in Ankara for the provision of facilities to
study the Hazro area in the field. The companionship of E. Demirtasji has been appreciated above all and
his considerable efforts in organizing the fieldwork are gratefully acknowledged; he also wishes to record
the assistance given by Ahmet Angih of M.T.A. The Research Fund of the University of Sheffield defrayed
a large part of the travel expenses incurred by R.H.W.

The Royal Society of London and CONICET in Buenos Aires made it possible for S. Archangelsky to
study the material from Hazro in the University of Sheffield, and to compare it with other species of
Glossopteris in the collections of the British Museum (Natural History) and of the Sedgwick Museum in
Cambridge, to which Dr C. R. Hill and Dr N. F. Hughes kindly granted access. Facilities at the Geology
Department, University of Sheffield, are also gratefully acknowledged, and Howard Crossley is thanked
for photographic assistance. The BM(NH) made a photograph available of the type specimen of
Glossopteris stricta Bunbury which is reproduced here.

Systematic description
Order GLOSSOPTERIDALES

Genus GLOSSOPTERIS Brongniart 1822

Glossopteris anatolica sp. nov.
Figs 1-8, 11,13-14

1959 Glossopteris stricta Bunbury; Wagner: 1379-1381 (non Bunbury 1861: 331; pi. IX, fig. 5).

1962 Glossopteris cf. stricta Bunbury; Wagner: 745-746; pi. 24, figs 2, 2a (part), fig. 3; pi. 25, fig. 5 (part),

figs 6, 7, fig. 8 (part).
1980 Glossopteris sp. nov. ; Wagner in Fontaine et al. : 919.

DIAGNOSIS. Leaves extremely variable in size, the longest (incomplete) fragment measuring
10 cm x 3-4 cm at constant width (this specimen lacks both base and apex). Midrib strong, up to
4mm wide, and consisting of several parallel strands; it persists into the leaf apex. Base of leaf
acute, probably cuneate; apex obtuse (c. 90°), slightly emarginate in smaller specimens. Lateral
veins decurrent, strongly arching near midrib and passing to the leaf margins at angles of 65° to
80°, which become slightly less in the apical part (c. 50°). Anastomoses and somewhat less
common pseudo-anastomoses form a compact mesh with short areolae near the midrib
(1-5-2 mm long and 1 mm wide) and passing into more elongate, narrower meshes towards the
margins and in the apical part of the leaf.

HOLOTYPE. British Museum (Natural History) register no. V. 60797.

PARATYPES. BM(NH) V. 60793-6 and V. 60798-801, and additional specimens from the type
locality (Wagner loc. no. 3111) in the Natural History Museum of Maden Tetkik ve Arama
Enstitiisii, Ankara.

TYPE LOCALITY. Coal-bearing succession of the Gomaniimbrik Formation exposed at 750m
SSW of Dada§ village in the western part of the Hazro inlier, c. 70km ENE of Diyarbakirin
south-east Anatolia, Turkey.

NAME. Anatolia, the Asian part of Turkey.

DESCRIPTION. The specimens figured in 1962 are joined by new collections made from different
bands in the same general locality south of Dada§ in the western part of the Hazro inlier.
Unfortunately, no complete leaves have been obtained, the most entire specimen (Fig. 6) being
a rather small leaf, 4cm long and 1-8 cm wide in the middle. It shows an obtuse, slightly
emarginate apex. Larger specimens are up to 10cm long (Fig. 1), despite the lack of preserved
bases and apices. It is assumed that these leaves reached an approximate length of 15 cm or
more. Their observed maximum width is 4cm, and it thus appears that the larger leaves may
have been narrowly oblong, lorate (following the terminology established by Dilcher, 1974, and

PERMIAN GLOSSOPTERIS IN TURKEY

Figs 1-4 Glossopteris anatolica sp. nov. Fig. 1, middle part of a leaf showing a wide midrib and the
characteristic lateral vein meshes, BM(NH)V. 60793, x3. Figs 2-3, middle part of leaf with
characteristic venation, BM(NH) V.60794, x6. See also Fig. 8. Fig. 4, basal part of a leaf,
BM(NH)V. 60795, x3.

S. ARCHANGELSKY & R. H. WAGNER

Figs 5-8 Glossopteris anatolica sp. nov. Fig. 5, apical part of a leaf, BM(NH) V.60796, x3. Fig. 6,
small leaf showing the base as well as a slightly emarginate apex, Holotype BM(NH) V. 60797, x3.
Fig. 7, part of a small, narrow leaf, BM(NH) V. 60798, x3. Fig. 8, middle to near-basal part of leaf
with a strong midrib and characteristic lateral veining pattern, BM(NH) V.60794, x3. (Details of
the same specimen, x6, see Figs 2-3).

PERMIAN GLOSSOPTERIS IN TURKEY 85

adopted for Glossopteris leaves by Chandra & Surange, 1979). Apical and near-basal leaf
fragments were illustrated in 1962, and also appear in the new collections (Figs 4, 5).

The midrib is strong, always persistent to the apex, and rather wide in the basal part of the
leaves. It consists of several (usually 5 to 6) parallel, non-anastomosing strands (Fig. 14). Lateral
veins are decurrent in the basal and medial sectors of the leaves, and slightly less decurrent near
the apex. They are strongly arching quite near the midrib (Figs 2, 7, 8, 11), i.e. within a distance
of 4 to 5 mm, and run a straight, subparallel course in most of the width of the leaves, reaching
the leaf margin generally at angles of 65° to 75° (overall variation is 50° to 80°). The vein pattern
is reticulate throughout, with an apparent predominance of complete anastomoses, but also
showing pseudo-anastomoses. The arching veins near the midrib show a mesh with short and
wide areolae; more elongate, narrower vein meshes occur in the straighter, subparallel course of
the veins towards the leaf margin (Figs 3, 13).

COMPARISONS. The most comparable species is Glossopteris stricta Bunbury, as described from
the Upper Permian Kamthi 'Stage' of India (Bunbury 1861, Chandra & Surange 1979). The
Anatolian species differs mainly in the secondary venation. Although the general pattern of vein
meshes is similar, with short and broad areolae near the midrib and narrower, more elongate
meshes towards the margin, it is noted that the veins of G. stricta are not quite as decurrent as in
G. anatolica. They also meet the leaf margin at almost 90°, whereas the angle varies between 50°
and 80° in G. anatolica (depending on the size of the leaf and the position of the veins within the
leaf). Also, the apex of G. stricta leaves is more acute and does not display the emargination
seen in at least one specimen of G. anatolica (Fig. 6). Although the general shape of the leaves
appears to be similar in both species, it seems that the leaves of G. stricta are relatively longer (as
follows from an examination of the lectotype, BM(NH) V. 19620, which shows a length/breadth
ratio of about 10:1). Making allowance for the incompleteness of the known leaves of G.
anatolica, it appears that these show a length/breadth ratio of up to 6:1. The lectotype of
Glossopteris stricta Bunbury has been refigured by Banerjee (1978: pi. 8, figs 17-18).

Glossopteris leaves described as G. stricta from Patagonia in South America (Archangelsky
1958a,b) are comparable to G. anatolica in the general shape, size and anastomosed vein
pattern. However, the Patagonian leaves are different in having longer and narrower vein
meshes near the midrib, and possessing less decurrent lateral veins. Complete specimens from
Patagonia display a length/breadth ratio of 9:1, as against a probable 6:1 ratio for G. anatolica.

Glossopteris stricta Bunbury, as recorded from Antarctica by Plumstead (1962), shows
somewhat less decurrent lateral veins which display a tendency towards free veining near the
leaf margin (compare Plumstead 1962: pi. X, fig. 1; pi. XI, fig. 1).

Glossopteris indica Schimper, as revised by Chandra & Surange (1979), is a polymorphic
species, with a changing length/breadth ratio as the species is followed up the stratigraphic
column. Ratios of 6:1, 4:1 and 3:1 are recorded for the Barakar, Kamthi and Raniganj 'stages' in
upward succession. The lateral veins of G. indica are not decurrent near the midrib, and they
usually abut onto the leaf margin at angles of c. 45° to 50°, reaching angles of up to 70° in some
medial portions. G. anatolica shows clearly decurrent veins near the midrib in the basal, medial
and apical parts of leaves of all sizes. It is also noted that the apex in G. indica is more acute than
it is in G. anatolica. Also it never appears to be emarginate.

Glossopteris pantii Chandra & Surange (1979) is a leaf of similar shape to that of G. anatolica,
albeit with an obtuse-cuneate base as against an acute-cuneate base in the latter. Moreover, its
veins follow a different pattern, being more horizontal in the medial sector and reaching the leaf
margin at 45° in the basal parts of leaves. This contrasts with the venation of G. anatolica which is
more generally uniform in different parts of the leaf. Also, the areolae near the midrib are
shorter and wider in G. anatolica, which displays a more marked contrast in mesh size and shape
between the central and marginal parts of the leaf.

Glossopteris arberi Srivastava (1956) shows leaves of similar size and shape to those of
G. anatolica, but its lateral veins dichotomize only 2-3 times and the areolae are correspond-
ingly longer than those of the Anatolian species.

Glossopteris tortuosa Zeiller, as figured by Plumstead (1952: pi. 49, fig. 4) from South Africa,

S. ARCHANGELSKY & R. H. WAGNER

•Hi •

Figs 9-10 Glossopteris stricta Bunbury. Fig. 9, lectotype (Bunbury 1861: pi. 9, fig. 5),
BM(NH)V. 19620, xl. Fig. 10, detail of the venation of the lectotype (lower part of the leaf), x3.

Fig. 11 Glossopteris anatolica sp. nov. Detail of the lateral vein meshes (for comparison with
G. stricta), BM(NH) V.60799, x3.

PERMIAN GLOSSOPTERIS IN TURKEY 87

differs from G. anatolica in the shape of its leaves which are broadly ovate. It also shows
narrower vein meshes near the midrib than occur in the latter.

DISCUSSION. Glossopteris leaves have been found in three different bands in the Gomaniimbrik
Formation south of Dada§ village in the Hazro inlier. They are common without being
abundant, about 30 specimens having been found altogether. Every single specimen shows the
characteristic nervation of Glossopteris anatolica, and it is clear that only a single species is
represented. No cuticle has been recovered from any of these specimens and there are no
fructifications assignable to Glossopteris associated with the leaf prints (Professor W. S. Lacey,
who kindly examined some poorly-preserved fructifications from the original collection from
Hazro, did not find convincing evidence of any Glossopteris fructification, although some
specimens seemed to suggest the possibility - Lacey, in litt. 30. XI. 62). On the other hand, the
midrib composed of parallel strands and the characteristic mesh formed by anastomosed and
pseudo-anastomosed lateral veins do not admit of a generic attribution other than to
Glossopteris. The comparisons made with several undisputed Glossopteris species emphasize
this point.

The Hazro flora

It remains to explain the presence of Glossopteris in an assemblage of plants which are mainly
characteristic of the equatorial belt and not of the Gondwana Realm. A revision of the floral
remains recorded in 1962 from Hazro, in conjunction with the new material collected from
different bands in the same locality, has been undertaken by R.H.W. A revised list of taxa,
incorporating additional species discovered most recently, is as follows: Glossopteris anatolica
Archangelsky & Wagner sp. nov., Bicoemplectopteris hallei Asama, Taeniopteris sp.,
Botrychiopsis sp. , Fascipteris hallei (Kawasaki) Gu & Zhi, Pseudomariopteris hallei (Stockmans
& Mathieu) Wagner, Cladophlebis tenuicostata (Halle) comb, nov., Sphenopteris sp.,
Pecopteris (Asterothecal} calcarata Gu & Zhi, Dizeugotheca! sp. nov., Pecopteris nitida
Wagner, Pecopteris pirae Wagner, Sphenophyllum cf. koboense Kobatake, Lobatannularia
heianensis (Kodaira) Kawasaki, Cordaites sp.

This is not the place for a full discussion of the revision which is still in progress. However, a
few brief comments may be in order. Bicoemplectopteris hallei refers to the specimens figured as
Gigantopteris nicotianaefolia in 1962 (see Asama, 1976: pi. XXX, fig. 6). Fascipteris hallei is the
material recorded as 'Validopteris'1 sensu Stockmans & Mathieu (non Bertrand) in 1962.
Cladophlebis tenuicostata has been identified mainly from new material, but incorporates
specimens recorded originally (Wagner 1962) as Pecopteris tenuicostata Halle and Cladophlebis
roylei Arber. The new combination is based on Pecopteris tenuicostata Halle as figured and
described from the Upper Shihhotse of central Shansi, China (Halle 1927: 99-100; pi. 26, figs
1-2). Pseudomariopteris hallei, Sphenopteris sp. and Pecopteris calcarata are new identifications
based on specimens collected most recently. Dizeugothecal sp. refers to a fertile pecopterid
similar to that figured from Saudi Arabia by El-Khayal et al. (1980: fig. 2c). Sterile remains of
this plant were illustrated in 1962 under the name of Pecopteris cf. wongi Halle pars (see also
Lemoigne 1981: pi. 6, fig. 1). Pecopteris nitida Wagner, which is now available in large
specimens showing the variation within the frond, also incorporates the remains identified in
1962 as Pecopteris phegopteroides (Feistmantel) and Pecopteris jongmansi Wagner. Recent
investigations on the Permian flora of Unayzah in Saudi Arabia, undertaken by R.H.W. in
collaboration with Dr A. A. El-Khayal of King Saud University, Riyadh, have shown that
Pecopteris tenuidermis Wagner (1962) represents the cuticular imprints of pinnules of Pecopteris
pirae Wagner. The single leaves of Zamiopterisl sp. figured in 1962 proved to belong to
Sphenophyllum cf. koboense Kobatake, a Late Permian species in which the leaves show the
development of a midvein.

Wagner (1962) claimed that the Hazro flora contained a mixture of Cathaysian and
Gondwana elements. The Cathaysian aspect of this flora has not been seriously disputed, and
the newly discovered additional species have strengthened the comparisons with the East Asian

S. ARCHANGELSKY & R. H. WAGNER

'e~«

k --

Fig. 12 Botrychiopsis sp. Basal portion of a frond showing a double row of pinnules, with totally

fused lamina at the extreme base and more individualized, semicircular pinnules a little higher up.

BM(NH) V.60802, x3. Part of this specimen was figured by Wagner (1962: pi. 26, fig. 12).
Figs 13-14 Glossopteris anatolica sp. nov. Fig. 13, detail of lateral veins in the apical sector of a leaf,

BM(NH) V. 60800, x3. Fig. 14, lower part of a leaf showing the wide, striate midrib formed by

parallel strands, BM(NH)V. 60801, x3.

PERMIAN GLOSSOPTERIS IN TURKEY 89

Cathaysia flora. The Gondwana component has been regarded as more controversial.
Glossopteris anatolica is the most striking representative of the Gondwana flora at Hazro where
it is of common occurrence. Its Late Permian (Dzhulfian) age puts it near the end of the
stratigraphical occurrence of the genus, and this implies that the Anatolian species had sufficient
time to migrate from the main area of the Gondwana Realm where Glossopteris is both
diversified and abundant. The palaeogeographic position of India alongside East Africa and
Madagascar provides the possibility of a direct migration route from either India or eastern
Africa. The upper Mesopotamian region, to which Hazro belongs, can be regarded as the
northernmost part of the Arabian Shield which forms part of the Gondwana Plate.

Another plant of Gondwana affinity in the Hazro flora is that figured as Dicroidiuml vel
Thinnfeldial sp. by Wagner (1962: pi. 26, figs 12-13). Lacey (in Discussion of Wagner 1962)
compared these specimens with the basal parts of the frond of Neuropteridium validum
Feistmantel. He later referred to them as cf. Gondwanidium validum (Feistmantel) Gothan
(Lacey 1975: 129) and quoted Archangelsky (1971, personal communication) as supporting this
identification. Only two specimens are available from the old collection and no further remains
have come to light. The most complete specimen is refigured here as Fig. 12. It was only partially
figured in 1962. This specimen shows a wide rachis with fine longitudinal striations, and two
lateral laminae with little differentiation in the basal part and gradually more individualized
pinnules higher up. The venation of the individual pinnules is decurrent, dichotomous, and
generally odontopteroid. The pinnules are broadly attached, and almost semicircular, being
about twice as wide as they are high. There is a reasonable resemblance to the basal parts of
fronds of Botrychiopsis (= Gondwanidium) as figured by Archangelsky & Arrondo (1971:
pi. I). Three species of Botrychiopsis are known at present: B. weissiana Kurtz, B. plantiana
(Carruthers) Archangelsky & Arrondo, and B. valida (Feistmantel) Archangelsky & Arrondo.
The specimens from Hazro cannot be identified with any of these. Furthermore the Hazro flora
is very Late Permian in age whilst the youngest of these species, Botrychiopsis valida, belongs to
the Early Permian (for a recent description, see Archangelsky & Cuneo, 1981). Although it may
be assumed that the Hazro specimens belong to a new species of Botrychiopsis, the material is
inadequate for a proper description, particularly in view of the fact that Botrychiopsis fronds
show a good deal of variation as a result of lobing. This variation can be brought out only by large
specimens or many different fragments from a single locality. Since most of the Hazro flora is of
Cathaysian affinity, a sustained search was made in the literature for any comparable species
from the contemporaneous equatorial belt. This failed to produce results. The Hazro specimens
are therefore assigned confidently to Botrychiopsis sp., and it is assumed that this is another
immigrant from the Gondwana Realm.

Two other species mentioned in 1962 were also assumed to be indicative of a Gondwana
affinity. One of these, Pecopteris phegopteroides (O. Feistmantel) (Wagner 1962: pi. 25, fig. 8
pars; pi. 28, fig. 26 - with cf.), cannot be retained in the list of species from Hazro. Fairly large
fragments of the frond of Pecopteris nitida Wagner, which have been collected most recently,
have shown that the specimens figured as P. phegopteroides fall within the range of variation of
the latter species. The second putative Gondwana element, Cladophlebis roylei Arber (Wagner
1962: pi. 27, figs 16-17), is here referred to Cladophlebis tenuicostata (Halle) comb. nov.
Additional material from the Hazro flora has shown a range of variation which apparently fits
Halle's species from the East Asian Cathaysia flora. It is noted that Cladophlebis mongolica
Durante, from the Permian of Mongolia, also seems to fit this species which Halle assigned to
Pecopteris. Permian representatives of Cladophlebis are generally uncommon. They appear to
be restricted to Upper Permian strata and it is assumed here that they are elements of the warm,
humid, equatorial belt floras, which are found only occasionally in Upper Permian Gondwana
assemblages.

General considerations on floral distribution

Wagner (1962) presented a map on which the Hazro locality was shown as belonging to both the
Cathaysian (of the palaeoequatorial belt) and Gondwana realms. Chaloner & Lacey (1973) and

90 S. ARCHANGELSKY & R. H. WAGNER

El-Khayal, Chaloner & Hill (1980) modified the northern boundary of the Gondwana Realm,
leaving the Hazro locality well inside the palaeoequatorial belt. This is consistent with the
information obtained most recently, which strongly emphasizes the Cathaysian connection.
Also, the Saudi Arabian flora reported by El-Khayal et al. (1980), and which is currently being
investigated in more detail, shows clear Cathaysian affinities. This flora is at present regarded as
being of mid-Permian age. Lemoigne (1981) even referred to it as belonging to the Upper
Permian. The boundary between palaeoequatorial (sensu lato) and Gondwana floras shown on
the map published by El-Khayal et al. (1980), and which we accept as more nearly correct in the
light of current information, leaves most of the Arabian Peninsula in the palaeoequatorial belt.
The lack of information from southern Arabia and the horn of Africa makes it also possible that
the northern boundary of the Gondwana Realm should be even further south, leaving the entire
Arabian Shield in the palaeoequatorial belt (compare Lemoigne, 1981). Most palaeogeographic
reconstructions place India alongside eastern Africa. These two areas are here regarded as the
likely source for the two plants of Gondwana affinity in the Hazro flora, i.e. Glossopteris
anatolica and Botrychiopsis sp. It is a well-known fact that Permian times saw an appreciable
amelioration of climate which led to substantial mixing of floral elements. This is mainly
recorded as the influx of 'equatorial', presumably more thermophile, elements into the
Gondwana floras which thus became a little less restricted in composition. The reverse
migration, from the Gondwana Realm into the equatorial belt, is less well documented, and it
seems that the Hazro flora provides one of the rare examples of it. It is probably no coincidence
that this migration is found in a flora of the latest Permian age. Glossopteris is almost exclusively
a Gondwana element which lived in a temperate climate. The migration of Glossopteris
anatolica to lower latitudes, and a warmer climate, may have been aided by the fact that the
Hazro locality coincides with the northern edge of the Gondwana Plate in upper Mesopotamia.
There seems to have been a continuous land area from East Africa/India to Arabia and
Mesopotamia.

Another, more spectacular case of migration of the glossopterids is recorded by Zimina (1967 ,
1977), who figured and described three species of Gangamopteris , two of Glossopteris and one
of Palaeovittaria from the region of Vladivostok in the Soviet Far East. These occur in the lower
part of the Upper Permian.

'Gu & Zhi' (1974: pi. 110, figs 3-4) recorded a Glossopteris guizhouensis from the lower part
of the Upper Permian in Guizhou (Kuichow) Province in China, but this species has recently
been transferred to a new genus, Abrotopteris , which may be unrelated to the glossopterids.
This species is currently described as Abrotopteris guizhouensis (Gu & Zhi) Mo (Zhao Xiuhu,
personal communication).

Attention is drawn to Kon'no's (1963) record of Glossopteris cf. angustifolia Brongniart from
the Permian deposits of Phetchabun in Thailand. Asama (1966), who studied the Phetchabun
flora in more detail, accepted Kon'no's record in principle but mentioned that the specimen
figured by Kon'no had an indistinct venation and that further collecting from the Phetchabun
locality failed to turn up additional remains. In fact, he hinted at the possibility that the specimen
might belong to Taeniopteris . The Phetchabun flora is in the East Asian Cathaysia Province.

References

Archangelsky, S. 19580. Estudio geologico y paleontologico del Bajo de la Leona (Santa Cruz). Actageol.
lilloana, Tucuman, 2: 5-133, figs 1-56.

- 1958fc. Las Glossopterideas del Bajo de la Leona. Revta Asoc. geol. argent., Buenos Aires, 12 (3):
135-164, pis 1-12.

— & Arrondo, O. G. 1970. The Permian taphofloras of Argentina, with some considerations about the
presence of 'northern' elements and their possible significance. In Gondwana Stratigraphy: 71-89.
IUGS Symposium, Buenos Aires 1967.

1971. Palaeophytologia Kurtziana. III. 2. Estudio sobre el genero Botrychiopsis Kurtz

(= Gondwanidium Gothan) del Carbonico y Permico Gondwanico. Ameghiniana, Buenos Aires, 8
159-227, pis 1-6.

PERMIAN GLOSSOPTER1S IN TURKEY 91

- & Cuneo, R. 1981. Sobre la presencia del genero Botrychiopsis Kurtz en la Formation Nueva

Lubecka, Permico inferior de Chubut, Argentina. Anais II Congreso Latino-americano Paleontologia

Porto Alegre 1981: 157-167, pis 1-2.
Asama, K. 1966. Permian plants from Phetchabun, Thailand and problems of floral migration from

Gondwanaland. Contributions to the Geology and Paleontology of Southeast Asia, XXV. Bull natn

Sci. Mus. Tokyo, 9 (2): 171-211, pis 1-6.
1976. Gigantopteris flora in southeast Asia and its phytopaleogeographic significance. In Kobayashi,

T. & Hashimoto, W. (eds), Geology and Paleontology of Southeast Asia 17: 191-207, pis 28-30. Tokyo.
Banerjee, M. 1978. Genus Glossopteris Brongniart and its stratigraphic significance in the Palaeozoics of

India. Part 1. A revisional study of some species of the genus Glossopteris. Bull. hot. Soc. Beng

Calcutta, 32: 81-125, pis 1-10.
Bunbury, C. J. F. 1861. Notes on a collection of fossil plants from Nagpur, Central India. Q. Jlgeol. Soc.

Lond., 17: 325-346, pis 8-12.
Chaloner, W. G. & Lacey, W. S. 1973. The distribution of Late Palaeozoic floras. In Hughes, N. F. (ed.),

Organisms and Continents through Time. Spec. Pap. Palaeont., London, 12: 271-289.
Chandra, S. & Surange, K. R. 1979. Revision of the Indian species of Glossopteris. Monogr. Birbal Sahni

Inst. Palaeobot., Lucknow, 2: 1-291, pis 1-47.
Ditcher, D. L. 1974. Approaches to the identification of angiosperm leaf remains. Bot. Rev., New York 40

(1): 1-157.
Durante, M. V. 1971. [On the Late Permian flora from Mongolia and southern boundary of the Angara

realm at that time.] Paleont. Zh., Moscow, 4: 101-112, tab. 13-14. [In Russian; Engl. transl. Paleont. J.,

Washington, 5 (4): 511-522].
El-Khayal, A. A., Chaloner, W. G. & Hill, C. R. 1980. Palaeozoic plants from Saudi Arabia. Nature,

Lond., 285 (5759): 33-34.
Fontaine, J.-M., Brunton, C. H. C., Lys, M., Rauscher, R. & Balkas., 6. 1980. Donnees nouvelles sur la

stratigraphie des formations paleozoi'ques de la plate-forme arabe dans la region d'Hazro (Turquie).

C.R. hebd. Seanc. Acad. Sci., Paris, (D) 291: 917-920.
'Gu & Zhi' (= Li Xingxue, Deng Longhua, Zhou Zhiyan, Xu Ren & Zhu Jiagou) 1974. [An introduction to

Chinese fossils. Chinese plant fossils 1, Chinese Palaeozoic plants], iii + 277pp., 142 figs, 130 pis. Peking

(Scientific Publishing House). [In Chinese].
Halle, T. G. 1927. Palaeozoic Plants from Central Shansi. Palaeont. sin., Peking, (ser. A) 2 (1): 1-316,

64 pis.
Kon'no, E. 1963. Some Permian plants from Thailand. Contributions to the Geology and Palaeontology of

Southeast Asia, V. Jap. J. Geol. Geogr., Tokyo, 34 (2-4): 139-159, pi. 8.
Lacey, W. S. 1975. Some problems of 'Mixed' Floras in the Permian of Gondwanaland. In Campbell,

K. S. W. (ed.), Gondwana Geology: 125-134. Canberra (Papers III Gondwana Symposium).
Lemoigne, Y. 1981. Flore mixte au Permien superieur en Arabic Saoudite. Geobios,Lyon, 14(5): 611-635,

pis 1-6.
Plumstead, E. P. 1952. Description of two new genera and six new species of fructifications borne on

Glossopteris leaves. Trans, geol. Soc. S. Afr., Johannesburg, 55: 281-328, pis 43-52.
1962. Fossil floras of Antarctica (with an Appendix on Antarctic Fossil Wood by R. Krausel). Scient.

rep. transantarct. Exped., London, (Geol.) 9: 1-154, pis 1-28.
Srivastava, P. N. 1956. Studies in the Glossopteris flora of India. 4. Glossopteris, Gangamopteris and

Palaeovittaria from the Raniganj Coalfield. Palaeobotanist, Lucknow, 5 (1): 1-44, pis 1-14.
Wagner, R. H. 1959. Une flore permienne d'affinites cathaysiennes et gondwaniennes en Anatolic

sud-orientale. C.R. hebd. Seanc. Acad. Sci., Paris, 248: 1379-1381.
1962. On a mixed Cathaysia and Gondwana flora from S.E. Anatolia (Turkey). C.R. 4e Congres

Carbonifere, Heerlen 1958, Maastricht, III: 745-752, pis 24-28.
Zimina, V. G. 1967. [On Glossopteris and Gangamopteris found in Permian deposits of Southern

Primoriye.] Paleont. Zh., Moscow, 2: 113-121, tab. 1. [In Russian; Engl. transl. Paleont. J.,

Washington, 1 (2): 98-106].
1977. [The flora of the Lower and early Upper Permian in the southern Primoriye.] 127 pp., 20 pis.

Moscow (Nauka). [In Russian.]

The crocodilian Theriosuchus Owen, 1879 in the
Wealden of England

£. Buffetaut

E.R.A. 963 du C.N.R.S., Laboratoire de Paleontologie des Vertebres, Universite Paris VI,
4 place Jussieu, 75230 Paris Cedex 05

Synopsis

A skull fragment from the Wealden of Brook (Isle of Wight) is described and referred to the genus
Theriosuchus Owen, 1879, previously known from the Purbeck of England. A little-known previous report
of Theriosuchus teeth from the Wealden is discussed. Isolated procoelous vertebrae from the English
Wealden named Heterosuchus valdensis by Seeley and often referred to the enigmatic crocodilian
Hylaeochampsa may actually belong to Theriosuchus.

Introduction

This paper reports the previously overlooked occurrence of the crocodilian Theriosuchus in the
Wealden of England. The genus Theriosuchus was erected by Owen in 1879, with T. pusillus as
type species, for remains of a small crocodilian found by W. H. Deckles in the Purbeck Beds of
Dorset. Theriosuchus may be a genus of great importance in crocodilian evolution. Its
systematic position was interpreted variously before Joffe (1967) showed that it closely
resembled the Atoposauridae from the Upper Jurassic of Europe, and suggested it should be
included in that family. She also noted that Theriosuchus was very progressive in some respects
(palatal structure, procoelous vertebrae) and might have been close to the ancestry of the
Eusuchia, or have evolved in parallel with them. Although basically I agree with Joffe's
conclusions, I think Theriosuchus may be sufficiently divergent from the typical Atoposauridae
to warrant its inclusion in a separate family, as already suggested by Kalin in 1955 (see Buffetaut,
1982, for a more complete discussion).

The fossil reptile collection of the British Museum (Natural History) contains a fairly large
number of remains of Theriosuchus pusillus from the 'Feather Bed' of the Middle Purbeck (see
Joffe, 1967, for more details). However, apart from a very brief report, apparently overlooked
by all later authors, which will be discussed below, there was until now no record of
Theriosuchus from other localities.

In September 1980 Dr Peter Wellnhofer (Bayerische Staatssammlung fur Palaontologie und
historische Geologic, Munich) was examining the Hooley collection of pterosaurs from the
Wealden of the Isle of Wight in the British Museum (Natural History). While doing so he came
across the fragmentary posterior part of the skull of a small crocodilian (reg. no. R.176), and
later he kindly mentioned this to me. Although very incomplete, the specimen turned out to be
identifiable as Theriosuchus.

Previous report of Theriosuchus from the Wealden

At first it was thought that this skull fragment was the first find of Theriosuchus in the Wealden,
but a careful search through the literature revealed that the genus had already been reported
from this formation. In 1912 there appeared in Nature a very short anonymous news item on
fossils recently presented to the British Museum (Natural History), which is quoted here in full:

The Geological Department of the British Museum (Natural History) has also recently received a valuable
gift of Wealden fossils from the Revs P. Teilhard and F. Pelletier, S.J., who made the collection during a
four years' residence near Hastings. A large proportion of the specimens are small teeth from bone-beds

Bull. Br. Mus. nat. Hist. (Geol.) 37 (3): 93-97 93 Issued 24 November 1983

E. BUFFETAUT

which had previously been very little examined, and among them is the unique mammalian tooth described
under the name of Dipriodon valdensis by Dr Smith Woodward in 1911. There are numerous teeth of the
dwarf crocodile Theriosuchus , which has hitherto been known only from the Purbeck beds. The series of
plant-remains is also important and will shortly be described by Prof. A. C. Seward in a communication to
the Geological Society.

In the discussion following the 1911 paper by Woodward on the above-mentioned mammal
tooth, Charles Dawson had mentioned that Teilhard de Chardin and Pelletier (who were then
studying theology at Hastings) had been helping him for two years in his researches on the
Wealden bone beds of the Hastings area.

The report in Nature went unnoticed, and I have been unable to find any later mention of
Theriosuchus in the Wealden. The isolated teeth in question are kept in the British Museum
(Natural History) under the collective numbers R. 4424-31 inclusive. They come mainly from
the Wadhurst Clay and the Ashdown Clay of Fairlight West, although some were collected near
Brede (both localities in the vicinity of Hastings). Similar teeth (R.3697) were presented even
earlier, in 1909, by Charles Dawson.

Some of these teeth are not especially characteristic, being of the usual crocodilian type,
conical and pointed; they cannot really be identified with any accuracy. Others are more
distinctive, being laterally compressed, with a low, rather blunt crown, which is somewhat
leaf-shaped in profile; these are very reminiscent of the posterior teeth of Theriosuchus pusillus .
However, supplementary and more convincing evidence for the occurrence of Theriosuchus in
the Wealden of England is provided by the present skull fragment.

The skull fragment R.176

The skull fragment found by Dr Wellnhofer in the Hooley collection (Fig. 1) comes from the
Wealden of Brook (also spelt Brooke) on the southern coast of the Isle of Wight. A detailed
description of the Wealden section at Brook Bay is given by Osborne White, who mentions

Fig. 1 A, B: Skull fragment of Theriosuchus sp. indet. from the Wealden of Brook, Isle of Wight,
BM(NH) R.176, in dorsal (A) and posterior (B) views. C: posterior part of the skull of
Theriosuchus pusillus from the Purbeck of Dorset, BM(NH) 48216, in dorsal view, for comparison
(after Owen, 1879). All xl. Drawings A and B by Dominique Visset.

WEALDEN THERIOSUCHUS 95

(1921: 8) that 'the Wealden Beds of Brook Bay have yielded the remains of various large
reptiles, including Iguanodon bernissartensis Boul., Hoplosaurus hulkei (Seeley), and
Heterosuchus valdensis Seeley'. Incidentally, Heterosuchus valdensis is by no means a large
reptile, but a small crocodilian, of which more will be said below.

The skull fragment comprises the greater part of the parietal, small medial portions of the
squamosals, and parts of the bones of the occipital region (supraoccipital, exoccipitals,
basioccipital); however, the occipital surface is poorly preserved, being much crushed and
cracked. The lateral surfaces of the braincase are also very poorly preserved.

What is left of the parietal is roughly trapezoidal in outline, the bone becoming increasingly
wider towards the rear. It is definitely narrower (7 mm) between the upper temporal fenestrae
than farther posteriorly (maximum width 22mm). Anteriorly, the parietal is incomplete, but
apparently not much is missing. The dorsal surface of the bone is distinctly concave transversely,
especially in its posterior part, as can readily be seen in posterior view. The anterolateral edges
of the parietal are raised into blunt ridges, which extend forwards and become narrower
between the upper temporal fenestrae. The dorsal surface of the bone is ornamented with small
irregular pits. An interesting feature is the presence of a very well marked, fairly sharp, median
ridge, which extends along the entire length of the bone. Posteriorly, the parietal overhangs the
occipital surface. On this surface, there is a prominent triangular median tuberosity or spine,
which seems to be formed partly by the parietal and partly by the supraoccipital. Only the dorsal
part of the latter bone is preserved; it shows a pair of depressions, one on either side of the
above-mentioned spine. More ventrally, the bones are so crushed that little is discernible; the
site of the foramen magnum is still visible, but the occipital condyle has disappeared. When the
dorsal surface of the parietal is placed in a horizontal plane, the occipital surface is seen to be
sloping forwards at an angle of about 60°, a condition more primitive than in modern
crocodilians (in which the occipital plane is nearly vertical), but not so archaic as in many
mesosuchians. The occipital surface seems to have been high relative to its width. As mentioned
above, little is left of the squamosals. A remarkable feature is the presence of a fairly deep and
very distinct groove between the parietal and the squamosals. The latter seem to have had
relatively well developed posterior expansions; in dorsal view, the posterior edge of the skull
roof is distinctly concave.

Although fragmentary, this specimen exhibits several features present in Theriosuchus
pusillus, suggesting its reference to the same genus. The shape of the parietal is very similar,
although its posterior part is wider in the Wealden specimen than in those from the Purbeck.
The lateral ridges on the parietal also occur in Theriosuchus pusillus , and the sharp median ridge
is a constant feature of all specimens from the Purbeck. The grooves between the parietal and
the squamosals are present both in the fragment from Brook and in Theriosuchus pusillus . The
ornamentation of the skull table of the Wealden specimen is very reminiscent of that of the
Purbeck form. Comparison of the occipital surfaces is hardly possible, since that region,
although not much flattened, is poorly preserved in the Wealden fossil, while all the Purbeck
specimens are strongly compressed dorsoventrally.

The Wealden specimen is somewhat larger than the type of Theriosuchus pusillus, but it still
indicates a small animal. Joffe (1967) suggested that most of the specimens of Theriosuchus
pusillus were juveniles, but this seems doubtful. Her evidence was based partly on a single femur
from the Purbeck referred to Theriosuchus pusillus, larger than other femora of the species.
However, the femur can hardly be called a very diagnostic bone in crocodilians, and the
specimen in question may not belong to Theriosuchus. The allegedly juvenile characters of the
skull of Theriosuchus pusillus listed by Joffe are observable also in the Atoposauridae from
continental Europe, which, according to Wellnhofer (1971), are not juveniles. It should also be
mentioned that the crocodilian obviously related to Theriosuchus briefly described (but not yet
named) by Langston (1974) from the 'Comanchean' (Lower Cretaceous) of Texas is hardly
larger than the type of Theriosuchus pusillus. The Wealden specimen described here also
suggests that Theriosuchus and its allies may never have grown to a large size.

The resemblances to Theriosuchus pusillus listed above indicate that the skull fragment from
the Wealden of Brook should be referred to the genus Theriosuchus. However, the specimen is
2

96 E. BUFFETAUT

too incomplete to warrant a specific identification and I think it better to designate it as
Theriosuchus sp. indet.

Stratigraphical range of Theriosuchus

The skull fragment from the Wealden of Brook provides the first really convincing evidence of
the occurrence of Theriosuchus in the Lower Cretaceous of England, and thus extends its
Stratigraphical range, previously limited to the Purbeckian. This of course is only a small
extension, since the Middle Purbeck beds which have yielded Theriosuchus pusillus are only
slightly below the Jurassic-Cretaceous boundary as defined in Dorset by Casey (1963).
According to Casey, the limit lies within the Purbeck beds, at the base of the 'Cinder Bed'. As
shown by Joffe (1967), Theriosuchus pusillus comes from the 'Feather Bed', about 10ft (3m)
below the 'Cinder Bed'. The time-span separating the Wealden Theriosuchus from the Late
Jurassic Theriosuchus pusillus is certainly not very great.

The occurrence of Theriosuchus in the Wealden beds is not really unexpected, since the
crocodilian faunas of the Purbeck and of the Wealden have several other elements in common
(notably Goniopholis crassidens and G. simus, as well as the genus Pholidosaurus) .
Theriosuchus can now be added to the list of crocodilians known from the Wealden of England,
which already includes the genera Goniopholis, Pholidosaurus, Vectisuchus (described by
Buffetaut & Hutt, 1980) and Bernissartia (reported by Buffetaut & Ford, 1979).

Theriosuchus and Heterosuchus

Seeley (1887) described as Heterosuchus valdensis a series of procoelous vertebrae in a small
nodule from the Hastings Sands of Hastings, which had been collected by Gideon Mantell and is
now in the collections of the British Museum (Natural History) under reg. no. 36555. He also
referred to this species 'a few isolated vertebrae of similar character' from the Wealden of
Tilgate and of Brook, also collected by Mantell and later purchased by the British Museum.
More isolated procoelous vertebrae from the Wealden were later referred to Heterosuchus
valdensis by Lydekker (1888).

Lydekker (1887) suggested that the vertebrae named Heterosuchus by Seeley might actually
belong to Hylaeochampsa vectiana, a peculiar crocodilian described by Owen (1874) from the
Wealden of the Isle of Wight. Hylaeochampsa vectiana is known by a single incomplete skull, in
which the internal nares are in a typical eusuchian position (i.e. totally enclosed by the
pterygoids) and which shows a peculiar construction of the palate, with large openings in the
ectopterygoids. The phylogenetic and systematic positions of Hylaeochampsa are still doubtful
(Buffetaut 1975), as it is uncertain whether it is closely related to modern eusuchians or is the
result of convergent evolution. Many authors have followed Lydekker's suggestion that
Heterosuchus is probably a junior synonym of Hylaeochampsa (Kalin 1955, von Huene 1956,
Romer 1956, 1966, Steel 1973). However, the skull of Hylaeochampsa was not associated with
vertebrae, and the only reason to assume that the vertebrae of Heterosuchus belong with the
skull of Hylaeochampsa is that in the Eusuchia skulls with internal nares in a position similar to
that of Hylaeochampsa are associated with procoelous vertebrae like those of Heterosuchus. It
should nevertheless be remembered that the evolution of a modern type of palate need not have
been synchronous with that of procoelous vertebrae. This is beautifully shown by Theriosuchus
pusillus, in which, as pointed out by Joffe (1967), there are already procoelous vertebrae, while
the palate is still of advanced mesosuchian type. Similarly, the Theriosuchus-like crocodilian
from Texas mentioned by Langston (1974) has an advanced mesosuchian palate and procoelous
vertebrae. Since Theriosuchus is now known to occur in the Wealden of England, one may
reasonably suppose that at least some of the procoelous vertebrae found in the same beds belong
to that genus rather than to Hylaeochampsa. The latter may have had procoelous vertebrae, but
this remains to be demonstrated by the discovery of associated skull and vertebral material.

WEALDEN THERIOSUCHUS 97

Acknowledgements

I thank Dr Peter Wellnhofer for drawing my attention to the skull fragment described in this
paper. I am also grateful to Dr A. J. Charig and to Mr C. A. Walker for making the Theriosuchus
material kept in the British Museum (Natural History) available for study.

References

Anonymous 1912. [Wealden fossils presented to British Museum by Revs P. Teilhard and F. Pelletier, S.J.J

Nature, Lond., 90: 111.
Buffetaut, E. 1975. Sur 1'anatomie et la position systematique de Bernissartia fagesii Dollo, L., 1883,

Crocodilien du Wealdien de Bernissart, Belgique. Bull. Inst. r. Sci. not. Belg., Brussels, (sci. terre) 51

(2): 1-20, 4 pis.
1982. Radiation evolutive, paleoecologie et biogeographie des Crocodiliens mesosuchiens. Mem.

Soc. geol. Fr., Paris, (n.s.) 60 (142): 1-88.
& Ford, R. L. E. 1979. The crocodilian Bernissartia in the Wealden of the Isle of Wight.

Palaeontology, London, 22 (4): 905-912.
& Hutt, S. 1980. Vectisuchus leptognathus , n.g. n.sp., a slender-snouted goniopholid crocodilian

from the Wealden of the Isle of Wight. Neues Jb. Geol Palaont. Mh., Stuttgart, 1980 (7): 385-390.
Casey, R. 1963. The dawn of the Cretaceous period in Britain. Bull. S.-east. Un. scient. Socs., Tunbridge

Wells, 117: 1-15.

Huene, F. von 1956. Palaontologie und Phylogenie der niederen Tetrapoden. 716 pp. Jena.
Joffe, J. 1967. The 'dwarf crocodiles of the Purbeck Formation, Dorset: a reappraisal. Palaeontology,

London, 10 (4): 629-639.

Kalin, J. 1955. Crocodilia. In Piveteau, J. (ed.), Traite de Paleon tologie, 5: 695-784. Paris.
Langston, W. 1974. Nonmammalian Comanchean tetrapods. Geosci. Man, Baton Rouge, 8: 77-102.
Lydekker, R. 1887. Note on Hylaeochampsa. Geol. Mag., London, 4: 512-513.
1888. Catalogue of the fossil Reptilia and Amphibia in the British Museum (Nat. Hist.), I. xxviii +

309 pp. London.
Osborne White, H. J. 1921. A short account of 'the geology of 'the Isle of "Wight. 219pp. London (Mem. geol.

Surv. U.K.).
Owen, R. 1874-79. Monograph on the fossil Reptilia of the Wealden and Purbeck Formations. Supplement

no. VI, Crocodilia (Hylaeochampsa). 1 pp. and pi. II of Suppl. V. Palaeontogr. Soc. (Monogr.),

London (1874). Supplement no. IX. Crocodilia (Goniopholis, Brachydectes, Nannosuchus,

Theriosuchus and Nuthetes}. 19 pp., 4 pis. loc. cit. (1879).
Romer, A. S. 1956. Osteology of the reptiles. 772 pp. Chicago & London.

1966. Vertebrate Paleontology (3rd edn). 468 pp. Chicago & London.

Seeley, H. G. 1887. On Heterosuchus valdensis Seeley, a procoelian crocodile from the Hastings Sand of

Hastings. Q. Jl geol. Soc. Lond., 43: 212-215.
Steel, R. 1973. Crocodylia. In Kuhn, O. (ed.), Handbuch der Paldoherpetologie, 16. 116 pp. Stuttgart &

Portland.
Wellnhofer, P. 1971. Die Atoposauridae (Crocodylia, Mesosuchia) der Oberjura-Plattenkalke Bayerns.

Palaeontographica, Stuttgart, (A) 138 (5-6): 133-165.
Woodward, A. S. 1911. On some mammalian teeth from the Wealden of Hastings. Q. Jlgeol. Soc. Lond.

67: 278-281.

A new conifer species from the Wealden beds of
Feron-Glageon, France

Helen L. Fisher and Joan Watson

Department of Geology, The University, Manchester M13 9PL

Synopsis

Cuticle studies of various Lower Cretaceous conifers revealed a plant with several unusual features, here
described as Br achy phy Hum carpentieri n. sp., known only from the Wealden beds of Feron-Glageon
(Nord), France. Whilst resembling a typical Brachyphyllum Lindley & Hutton ex Brongniart species in
gross morphology, the combination of lobed leaf margin, abaxial glands, minute hypodermal cells and a
complex form of stomatal apparatus distinguish it from any other known conifer, fossil or extant. Silicified
wood from the same locality has been described by Lemoigne & Demarcq (1967) and may belong to the
same plant.

Introduction

During the recent revision of some Lower Cretaceous conifer species several hand specimens in
the foreign Cretaceous collection at the British Museum (Natural History) were re-examined.
More precisely the revision of a supposedly widespread conifer Sphenolepis kurriana (Dunker)
Schenk (Fisher 1981) led to the examination of French material previously thought to belong to
this species (Carpentier 1927, 1939). These French specimens from Feron-Glageon were given
to the Museum by A. Carpentier in exchange for English Wealden specimens. Hand specimens
of the French material bear a close resemblance to known German specimens of Sphenolepis
kurriana (Dunker) Schenk. Several cuticle preparations were made using Schulze's solution for
maceration. When examined microscopically it was immediately apparent that this conifer
displays several unique features which clearly distinguish it from the other revised conifer
species. The combination of a lobed leaf margin with glands opening into the notches between
the lobes and a complex stomatal apparatus, coupled with extensive minute hypodermal cells
which completely obscure the epidermis, could lead one to question whether the cuticle is
indeed coniferous. However, the leaf shape, phyllotaxy and pattern of stomatal distribution
could hardly be more typical of many Brachyphyllum species. It is interesting to note that
Lemoigne & Demarcq (1967) raised a similar question concerning silicified wood described
from the same locality. Evidence of association led them to suggest that the wood and the leafy
shoots belonged to the same conifer although the wood had some characteristics which were not
typically gymnospermous.

Systematic description

Order CONIFERALES
Form-genus BRACHYPHYLLUM Lindley & Hutton ex Brongniart

Brachyphyllum carpentieri sp. nov.
Figs 1-10

1927 Sphenolepidium kurrianum (Dunker); Carpentier: 71; pi. 19, figs 1-7.
1939 Sphenolepidium kurrianum (Dunker); Carpentier: 157; pi. 1, figs 1-11.

DIAGNOSIS. Small shoots up to 3 mm wide. Leaves scale-like, tips free, arising from the centre of
a rhomboidal leaf base cushion; leaf and basal cushion combined up to 3 mm long x 2 mm wide.

Bull. Br. Mus. not. Hist. (Geol.) 37 (3): 99-104 99 Issued 24 November 1983

100

H. L. FISHER & J. WATSON

Free part of leaf up to one-third total length. Leaf margins converging at up to 55° towards
bluntly pointed apex; margins microscopically lobed with glands present in notches between
lobes.

(Adaxial cuticle imperfectly known). Abaxial cuticle up to 7/xm thick. Stomata occurring in
two broad bands on leaf and basal cushion, concentrated on cushion, avoiding mid-line.
Stomatal apparatus circular, guard cells deeply sunken below a ring of papillate, subsidiary cells;
up to 6 papillae around rim of circular stomatal pit. Diameter of stomatal apparatus 28-44 /urn
(n = 32); stomatal orientation irregular. Ordinary epidermal cells rectangular or spindle-
shaped, some bearing small papillae. Hypodermal cells small, oval to circular, strongly
cutinized, randomly arranged in main part of leaf, diverging in lobes; 4-8 /u,m long x 3-5 ^m
wide (n = 100) with straight, unpitted anticlinal walls up to 1-5 ^tm wide.

HOLOTYPE. V. 17064, British Museum (Natural History).

TYPE LOCALITY AND HORIZON. Feron-Glageon (Nord), France. Barremian.

MATERIAL. This species is common in the Wealden beds of Feron-Glageon, the precise locality
of which was poorly understood until Lemoigne & Demarcq (1967) published the following
details. The beds are near the Hirson-Avesnes railway line; 1 km west of Couple voie (parish of
Glageon) and barely 100m from the boundary of the parish of Feron.

NAME. The species is named after Alfred Carpentier.

Description

The description is based on the holotype, V. 17064, from the foreign Cretaceous collection in the
British Museum (Natural History), Fig. 1, together with information from the work of

•I

Fig. 1 Brachyphyllum carpentieri n. sp. Holotype, x4. V. 17064, BM(NH).

WEALDEN CONIFER

101

Figs 2-4 Brachyphyllum carpentieri n. sp. Fig. 2, single leaf showing lobed margin; apex top left,
patch of stomata bottom right, with unusual pattern of hypodermal cells between; x 100. Fig. 3,
margin of leaf showing portions of two lobes with gland in notch between; x 175. Fig. 4, SEM view
of portions of two marginal lobes in foreground with gland in notch between. Spindle-shaped
epidermal cells with pitted walls show clearly on the right; x700. All preparations from holotype,
V. 17064.

Carpentier (1927, 1939). Three of Carpentier's figures (1939: pi. 1, figs 1-3) agree with the
cuticle obtained from the holotype in every detail. The remaining figures show the epidermal
cells and subsidiary cells of the stomata; such detail has not been seen in cuticle preparations of
the holotype. The small amount of cuticle remaining on V. 17064 is thick and easy to prepare but
most of the preparations are from the abaxial surface. Only a small part of the adaxial surface
was seen, thus the stomatal distribution on the adaxial surface is still unknown.

102

H. L. FISHER & J. WATSON

Figs 5-10 Brachyphyllum carpentieri n. sp. Fig. 5, SEM view of outer surface showing papillae, pits
and faint outlines of epidermal cells; x700. Fig. 6, group of stomata showing subsidiary cell
papillae around stomatal pits; x400. Fig. 7, SEM view of outer surface showing probable stomatal
pit; pits and papillae were as ill-defined as this in all SEM preparations. Epidermal cells just
discernible; x 1000. Fig. 8, SEM view of inner surface of cuticle showing a stoma closely invested
with highly cutinized hypodermal cells. Guard cells missing; X2000. Fig. 9, hypodermal cells seen
by light microscope; x700. Fig. 10, hypodermal cells by SEM; x700. All preparations from
holotype, V. 17064.

WEALDEN CONIFER 103

The epidermal details of the abaxial surface, though clearly seen in several of Carpentier's
figures, are extremely difficult to distinguish in the preparations from the holotype because they
are totally obscured by the tiny, thickly cutinized hypodermal cells. Under the light microscope
ordinary epidermal cells are impossible to detect, but when the outside surface of the cuticle is
viewed by scanning electron microscopy (SEM) the outlines of these cells are discernible (Figs 4,
5). They are long and rectangular or spindle-shaped, resembling those figured by Carpentier
(1939: pi. 1, figs 4, 5). In these preparations of Carpentier, showing the epidermal cells clearly,
there is no sign of hypodermal cells. Some epidermal cells bear papillae, often several per cell,
but we can detect no pattern to the distribution of papillate cells. Many cells have pores or pits in
the outer periclinal walls (Fig. 5), a feature exhibited by another Wealden conifer species
(Brachyphyllum obesum Heer), which is to be redescribed in detail elsewhere.

In the greater part of the leaf and cushion the hypodermal cells are arranged in long arcs which
radiate from the centre, swirl around the stomata and then diverge in the marginal lobes (Fig. 2).
When the inside surface of the cuticle is viewed by SEM the nature of these hypodermal cells is
fairly clear (Fig. 10). They are quite thickly cutinized, unpitted and very small, certainly the
smallest of any conifer known to us.

The details of the stomatal apparatus are still imperfectly understood. Carpentier's figures
(1939: pi. 1, figs 4, 6, 9, 10) show the stomatal apparatus to have 4-6 subsidiary cells with
occasional encircling cells present. In V. 17064 the hypodermal cells completely obscure the
subsidiary cells (Fig. 6). Individual subsidiary cells have not been observed but using the light
microscope up to six papillae have been seen around the stomatal pit. The stomata in Fig. 6 show
those papillae quite clearly, yet by SEM the pits and papillae all appear ill-defined, as in Fig. 7.
The guard cells are quite deeply sunken and have only been seen as rather eroded remnants.

The multicellular glands deeply embedded in the leaf tissue are another unusual feature . They
frequently occur below many of the notches separating the marginal lobes and several are
apparent in other parts of the leaf surface. The glands appear conical in shape with the widest
part of the cone opening into the notches (Figs 3,4).

Discussion

The shoots of this species closely resemble those of Sphenolepis kurriana (Dunker) Schenk in
gross morphology and the original attribution by Carpentier is quite understandable. The cuticle
however is unique, bearing no resemblance to S. kurriana or to any other conifer, fossil or
extant, that we have seen. A similar lobed leaf margin has been seen in one other conifer, from
the Lower Cretaceous of China, but the cuticles of the two species differ considerably (Zhou
Zhiyan, personal communication).

The presence of glands on the abaxial surface, whilst not unknown in conifers, is certainly not
common. Many extant members of the Cupressaceae have prominent resin glands but of the
species we have studied none bear any resemblance to Brachyphyllum carpentieri, where the
glands open into the notches of the leaf and involve a complex organization of the epidermal
cells.

Wood described by Lemoigne & Demarcq (1967) as Dadoxylon arduennense may belong to
the same plant as Brachyphyllum carpentieri. The silicified wood indicates a tall arborescent
plant which was at least 20 cm in diameter at the base of the trunk. The wood is of a type limited
to the Jurassic-Cretaceous. It is characterized by septate tracheids which are unknown in
modern gymnosperms. The nature of these septa is unclear and indeed it is difficult to conceive
what function they may have had in the water transport system. Lemoigne & Demarcq stress
that they cannot confirm that the wood is coniferous and comment that the affinity of the wood
may be closer to that of the Caytoniales, which are also represented at Feron-Glageon.

If, however, D. arduennense should prove to be coniferous it seems probable that the unique
structure of the wood combined with the unusual cuticle characters of B. carpentieri imply a
specific adaptation to particular environmental conditions.

104 H. L. FISHER & J. WATSON

Acknowledgement

The research for this paper was undertaken whilst H.L.F. was in receipt of a N.E.R.C.
studentship.

References

Carpentier, A. 1927. La flore Wealdienne de Feron-Glageon (Nord). Mem. Soc. geol. N., Lille, 10: 1-151,

25 pis.
1939. Les cuticules des Gymnospermes Wealdiennes du Nord de la France. Annls Paleont., Paris, 27:

153-179.
Fisher, H. L. [1981]. A Revision of some Lower Cretaceous Conifer Species. Ph.D. Thesis, University of

Manchester (unpubl.).
Lemoigne, Y. & Demarcq, G. 1967. Nouvelle espece de Dadoxylon a tracheides septees provenant du

Wealdien de Feron-Glageon (Nord). Bull. Soc. geol. Fr., Paris, (7) 9: 53-56, 1 pi.

Late Permian plants including Charophytes
from the Khuff Formation of Saudi Arabia

C. R. Hill

Department of Palaeontology, British Museum (Natural History), Cromwell Road, London
SW7 5BD

A. A. El-Khayal

Geology Department, King Saud University, Riyadh, Saudi Arabia

Synopsis

A fossil flora of probable Late Permian age is reported from the Khuff Formation of central Saudi Arabia.
The coniferous element is of typical European, Zechstein composition, whilst other elements provide
hitherto unrecognized similarities between Permian floras of the western and eastern hemispheres. Stems
and reproductive structures of a charophyte - Palaeonitella tarafiyensis sp. nov. - represent an unusually
complete occurrence of this group in the Permian.

Introduction

The discovery of Permian plants in the clastic facies of the Lower Khuff Formation (El-Khayal,
Chaloner & Hill 1980, Lemoigne 19810, b) has stimulated further search of Khuff beds
exposures by A. A. El-Khayal. The Lower Khuff has continued to yield a diverse flora which
will be further described elsewhere. The present note reports for the first time a markedly
different plant assemblage, from the Middle Khuff beds, and one which indicates a later
Permian age than the Lower Khuff flora.

The plant remains were collected from a clay quarry 15 km NNE of Buraydah, the capital city
of Qasim province, at the base of the escarpment named Jal al Watah on the road from
Buraydah to Al Tarafiyah (Fig. 1). Exposures in the quarry are approximately at the base of the
Khartam escarpment, in the Midhnab Shales which outcrop near the top of the Middle Khuff
beds as described by Powers et al. (1966: D31). The rock matrix is a deeply weathered
grey-brown claystone. Plant megafossils occur sparsely; they are limonitized and lack cuticles,
but show some cellular detail of internal structure. Our descriptions are brief, because both the
diversity and quality of preservation of the assemblage are limited.

Systematic descriptions
Division TRACHEOPHYTA

The tracheophytes in the present assemblage range from more or less fragmentary foliar
remains (Figs 2-4, 11), the majority indeterminable, to relatively well preserved coniferous
shoots and cones (Figs 5-8). The indeterminable pinna shown in Fig. 2 is of generalized
cycadalean aspect whilst the scale-like foliar organ of Fig. 3 may be pteridospermous. Two
specimens of such scales are known, one apparently trilobed and with coarse venation
diverging from the assumed base of the scale. At intervals the veins dichotomize and
anastomose. At the centre in both specimens there is a more or less well developed scar,
suggesting attachment of some other organ, possibly an ovule or reproductive branch. In view of
its incomplete preservation we refrain from naming this organ formally, other than as

Bull. Br. Mus. nat. Hist. (Geol.)37(3): 105-112 105 Issued 24 November 1983

106

C. R. HILL & A. A. EL-KHAYAL

Fig. 1 Road map of Buraydah region showing location of present outcrop at X and of site reported

by El-Khayal et al. (1980) at Y.

'Problematicum A', in the hope that determined search may yield better specimens. Clearly it is
of considerable interest as it displays some hints of glossopterid affinity.

The leaf pinna of Fig. 4, although also of obscure botanical affinity, may be securely named as
Wattia texana Mamay (1967), originally described from the Early Permian of North America.
Wattia is possibly of noeggerathialean affinity. Similar fossils occur in the Russian Permian.

These botanically rather obscure remains are associated with well-preserved coniferous leafy
shoots that make up about half of the specimens collected. Shoots with relatively long, narrow
leaves (Fig. 5) are determined as Pseudovoltzia Florin, and those with broad leaves (Fig. 6) as
Culmitzschia Ullrich, both characteristic of the Late Permian Zechstein floras of Europe (Florin
1963, Schweitzer 1968). A well-preserved female cone (Fig. 7) is also referred to Pseudovoltzia.
Fig. 8 illustrates the deeply divided cone scale with five fingerlike lobes and a seed (arrowed), as
in comparably preserved European material of Pseudovoltzia liebeana (Geinitz) Florin
(Schweitzer 1963, 1968). We refer the Saudi material to the same species. In view of its limited
preservation, however, we cannot entirely rule out similarities with female cones of allied
conifers such as Voltzia Brongniart and Glyptolepis Schimper, which range into the Mesozoic.

Three specimens (Fig. 9) resemble Halle's (1927) 'Problematicum', reported from the
Permian of China, though the Saudi material has a more pronounced distal taper to the scars
considered to represent points of attachment of sporangia. Halle's material is now known to
represent small fragments of detached scales from the possibly noeggerathialean cone Discinites
orientalis Li et al., in 'Gu & Zhi'1 (1974), which Li & Yao (1980) assign to an Early Permian age.
Despite differences from Halle's specimens the Arabian ones fall well within the range of form
of D. orientalis as figured by 'Gu & Zhi' and are identified with it. Discinites is reported also
from the Early Permian of North America (Mamay 1954) and species occur widely but sparsely
in the Euramerian Carboniferous (Boureau 1964).

Two fragmentary specimens of Pecopteris (Fig. 11) were collected, showing evidence of basal
fusion of the pinnules and with a simple venation of a kind common in Stephanian and Permian

''Gu & Zhi' represents the contracted and latinized name of an editorial committee called 'Zhongguo Gwshengdai
Zhiwu (literally, 'Chinese Palaeozoic Plants'), and has been used as an author citation by the Chinese ('Gu& Zhi' 1974:
1). The actual authors, who appear to be the members of the committee ('Gu & Zhi' 1974: 2) are Li Xingxue, Deng
Longhua, Zhou Zhiyan, Xu Ren & Zhu Jiagou. If so the full citation given here is redundant and might well be reduced
to Li et al. alone, bearing in mind however that the authors' names do not appear on the title page of the work.

PERMIAN PLANTS OF SAUDI ARABIA

107

Figs 2-12 Plant fossils from the Middle Khuff beds, 15 km NNE of Buraydah, Permian of central Saudi
Arabia. Scale bars each represent 5-0 mm except for Fig. 10 where the bar represents 0-5 mm. Fig. 2,
indeterminable pinna. Fig. 3, Problematicum A. Fig. 4, Wattia texana Mamay. Fig. 5, Pseudovoltzia
liebeana (Gienitz) Florin, foliage. Fig. 6, Culmitzschia sp. Figs 7, 8, Pseudovoltzia liebeana (Geinitz)
Florin: Fig. 7, female cone; Fig. 8, female cone scale at higher magnification, with ovule arrowed.
Fig. 9, Discinites orientalis Li etal. in 'Gu & Zhi'. Figs 10, 12, Palaeonitella tarafiyensis sp. nov.: Fig. 10,
oosporangium (scale bar 0-5 mm), see also Fig. 13; Fig. 12, stems with holotype stem arrowed, BM(NH)
Palaeontology Dept. no. V. 60935. Fig. 11, Pecopteris sp. indet. All specimens except those of Figs 10,
12, are in the Geology Department, King Saud University, Riyadh.

pecopterids. In the absence of pinna terminals and of larger fragments indicating the range of
variation within the frond, we do not attempt specific identification.

All the tracheophyte specimens are housed in the Geology Department, King Saud
University, Riyadh.

Plesion CHAROPHYTA
Genus PALAEONITELLA Pia in Hirmer, 1927

Palaeonitella tarafiyensis sp. nov.
Figs 10, 12-18

DESCRIPTION. Stems (broken during fossilization) exceeding 27mm in length, internodes
smooth and without preserved cortical cells, up to 0-7 mm wide; nodes occurring at intervals of

108

C. R. HILL & A. A. EL-KHAYAL

Figs 13-18 Palaeonitella tarafiyensis sp. nov., drawings on photographs. All from BM(NH)
Palaeontology Dept. no. V. 60935, scale bars each representing 1-0 mm. Figs 13, 14, oosporangia,
Fig. 13 shown also in Fig. 10. Fig. 15, node compressed in plane of the whorl, showing six branches.
Figs 16-18, portions of stem compressed laterally, showing nodes with basal scars of branches.

1-25 mm or more, swollen to about twice the width of the internodes. Nodes bearing a single
ring of branches. (Branches seen when compressed in the plane of the whorl as in Fig. 15, whilst
in denuded stems compressed sideways their points of attachment are represented by a ring of
elliptical basal scars as in Figs 16-18). Obscure indications of additional cells or possibly
stipulodes occur above or below the branch scars (Fig. 17). Branches 12-16 per node (count
based mainly on numbers of basal scars). Reproductive structures (oosporangia) in intimate
association but not seen definitely attached to stems, urn shaped, 1-0 mm long x 0-5 mm wide
at broadest point, striated helically in sinistral direction, about 15-20 such stria per
oosporangium; actual number of coiled cells unknown (Figs 10, 13, 14). (Superficially the
coiling of striae looks dextral, since the surface visible is a mould of the oosporangium).

HOLOTYPE. V. 60935; specimen arrowed in Fig. 12 and portion shown also in Fig. 17.

MATERIAL. Several fragments, as shown in Fig. 12, all on one block only and apparently limited
to one bedding plane.

PERMIAN PLANTS OF SAUDI ARABIA 109

LOCALITY AND HORIZON. As indicated in the Introduction, p. 105.

DIAGNOSIS. Differs from the only other species of the genus, P. cranii (Kidston & Lang) Pia in
Hirmer, in its wider stems and in having branches that apparently lack septae.

PRESERVATION. The botanical as opposed to stratigraphic and palaeobiogeographic interest of
the Middle Khuff flora is considerably enhanced by these charophytes. The material represents
an early and exceptionally complete occurrence of plants of this group, which elsewhere is
known as fossils almost solely from reproductive structures. There are notably few records from
the Permian (Peck & Eyer 1963).

In the Saudi material vegetative remains occur closely association with reproductive organs.
Their mode of preservation as impression fossils yields considerably less detailed information
than is available from the usual preservation of charophytes as petrifactions. However, the
preservation is sufficiently detailed to indicate that the Saudi stems were ecorticate, resembling
those described by Kidston & Lang (1921) from the Devonian and by Ishchenko & Saidakovsky
(1975) from the Silurian.

NOMENCLATURE. Since assignment of the Saudi specimens to a family is problematic we use a
form genus. The name Palaeonitella was applied by Kidston & Lang (1921) to vegetative
remains from the Rhynie Chert, some at least of which clearly resemble those of charophytes but
with no oosporangia. The internodes are ecorticate. Unfortunately Kidston & Lang applied the
name so informally that it cannot be attributed to them - they named their material Algites
(Palaeonitella) cranii - and the name was only validated by Pia in Hirmer (1927). Horn af
Rantzien (1956) discusses the nomenclatural details, with which we agree fully, though we
believe the citation for the generic name should be in the form given here rather than as
Palaeonitella (Kidston & Lang) Pia. This citation more properly belongs to the species name
Palaeonitella cranii.

Although the attribution of Kidston & Lang's material to the Charophyta is open to some
doubt in the absence of reproductive organs (Groves 1933), the name Palaeonitella is used here
in order to limit gratuitous inflation of names (see Harris, 1962-3). We use it for any
uncorticated stems resembling those of Charophyta and of which the oosporangia may be
known - as in P. tarafiyensis - or may not be known - as in P. cranii - but of which the
oosporangia even when known do not permit narrower determination than to the Charophyta as
a whole. (Charaxis Harris is available for corticate stems). We note in passing that only the
largest specimens of P. cranii fall within the range of stem width represented by the slenderest
species of extant Nitella, whereas P. tarafiyensis is typical of the majority of extant species.

Such limited detail of the oosporangia as is preserved in P. tarafiyensis neither refutes nor
supports assignment to the Porocharaceae, which would be the predicted familial determination
based on the age of the material (Grambast 1974). Further specimens are needed.

PALAEOECOLOGY. These Permian stems are remarkably like stems of living species in size and
known morphology, resembling ecorticate species of extant Char a, Nitella and Tolypella.
Charophyta are of considerable palaeoecological interest since all living species are limited to
fresh or brackish water environments, though some earlier Palaeozoic representatives are
believed to have been shallow-marine (Racki 1982). Notably in this present fossil flora the
charophyte remains are associated with fresh-water bivalves, suggesting a fresh-water-perhaps
lacustrine - environment of deposition; presumably they were not transported far from their
place of growth whereas other species in the flora are clearly drifted. There is no evidence to
suggest P. tarafiyensis was lime-encrusted in life, in contrast to most described fossil charophytes
and to many living species. The Middle Khuff matrix is in fact gypsiferous and lacks calcium
carbonate.

Discussion

FLORISTIC AFFINITIES. The affinities of the assemblage as a whole are no less intriguing than
those of the Lower Khuff flora (El-Khayal et al. 1980). In that flora, Pecopteris, Fascipteris,
Lobatannularia, Cordaites and Marattiopsis are conspicuous elements whilst conifers are

110 C. R. HILL & A. A. EL-KHAYAL

lacking. The Middle Khuff flora is in stark contrast: out of the genera just mentioned only
Pecopteris occurs, and that inconspicuously, whereas the strong coniferous element
(Pseudovoltzia, Culmitzschia) is in marked contrast to the earlier assemblage. Such sharp
differences in plant assemblages having rather slight age differences are unusual and surely
signify a strong degree of environmental control, either ecologically or sedimentologically or
both.

Almost all the Middle Khuff species are unknown in floras of roughly comparable age from
the Middle East but closely resemble those found in floras from far distant localities, mainly in
the northern hemisphere Permian. The only hint of Gondwana affinities is provided by the
scale-like foliar organ Problematicum A (p. 106), though that in itself - if its glossopterid
affinities were substantiated - would be of great interest. Wattia is described from the Early
Permian of Texas, U.S.A. , Discinites ranges into the Early Permian of China, Korea and North
America, but Pseudovoltzia and Culmitzschia are characteristic of the European Late Permian
(Florin 1963). Clearly this flora fills a gap between the Permian floras of the Euramerian and
Cathaysian areas of Chaloner & Meyen (1973) and thus represents a mixed flora. Lemoigne
(1981a, b) argues similarly for the Lower Khuff flora reported briefly by El-Khayal etal. (1980).
Lemoigne's perceptions of a strong Cathaysian element in that flora, extending vigorously
along the shores of Tethys, rest however largely on the validity of his determinations and
interpretations. We hope the further studies now in progress by Wagner & El-Khayal may
resolve taxonomic problems raised by El-Khayal et al. (1980) and by Lemoigne's 19816.

STRATIGRAPHY. That the Middle Khuff assemblage reported here is younger than the Lower
Khuff flora is certain from the field relations. If the strong coniferous element is emphasized at
the expense of other taxa, the plant megafossils suggest a probable Late Permian age.
Nevertheless, in such strictly palaeobotanic terms, exact age assignment may be regarded as
problematic for a number of reasons. Firstly, the floras of the Arabian peninsula, as at Hazro
in Turkey, are regionally unique. They fill gaps and thus obscure formerly clearly perceived
boundaries between floral provinces, yet have a character of their own. Secondly, Permian
floras worldwide are rather poorly known. Whilst they therefore represent a challenging and
stimulating research topic, in which a great deal is still to be learnt, stratigraphic ranges of
Permian megafossil plants are as yet poorly documented. The literature, too, is scattered. Thus,
if the present floras in Saudi Arabia are in fact both Late Permian, they considerably extend the
ranges of Wattia and Discinites whilst the ranges of Marattiopsis, Pseudovoltzia and other
species remain as previously understood by Florin (1963), Burago (1977) and others. For these
reasons the initial report by El-Khayal et al. (1980) on the Lower Khuff flora cautiously
suggested a rather broad age range: from not older than Westphalian (Upper Carboniferous) to
not younger than Early Permian, rather than a more precise age assignment. Others, notably
Lemoigne (1981a,b) have felt able to provide a more detailed age range based on the
palaeobotanical evidence, attributing a degree of precision to palaeobotanical dating which in
our view may be premature for such limited floras in a Permian context. It also takes little
account of a third limitation, the likely environmental as opposed to stratigraphic control over
such marked changes as occur between the Lower and Middle Khuff floras. Nonetheless, by the
same argument, we do not consider Lemoigne's specifically Late Permian age assignment for
the Lower Khuff flora unreasonable. If pressed to give a narrower age range for the flora,
however, we now tentatively favour an early Late Permian age for the Lower Khuff flora, based
entirely on the plant data.

The evidently somewhat vexed question of the age of the Lower Khuff flora has been further
discussed recently by Sharief (19820,6) and Ibrahim (1982). In view of the limitations of
palaeobotanic data when considered in isolation, referred to above, we believe Ibrahim
overemphasizes the botanically-based age assignment of El-Khayal et al. (1980). Moreover,
Popper (1959), Lakatos (1970) and others have stressed the importance of using independent
lines of evidence to test and thus critically to evaluate scientific hypotheses. As Sharief (19826)
rightly points out, independent evidence is available from palynology - though regrettably
unpublished ('Aramco' 1975). A Late Permian age is also indicated by the calcareous algae

PERMIAN PLANTS OF SAUDI ARABIA 111

(Rezak 1959). Such evidence, whilst not in itself necessarily superior to megafossil
palaeobotanic evidence, uniformly suggests a Late Permian age for the Khuff Formation.
Present evidence, therefore, all seems to favour a Late Permian age for both the Lower and
Middle Khuff floras.

To clarify the stratigraphic nomenclature used in this discussion we should emphasize that
usage of the term 'Khuff Formation' follows that of Powers et al. (1966), in including the
Unayzah plant bed within the Lower Khuff. El-Khayal & Wagner (in preparation) argue that
the Unayzah beds should be separated off from the Khuff as a separate Formation.

Acknowledgements

Dr R. H. Wagner, Professor W. G. Chaloner and Dr M. Feist kindly criticized an early draft of
the manuscript. Photographs were prepared by BM(NH) photographers.

References

'Aramco' [1975]. Generalized Saudi Arabian stratigraphic section, formational, zonal and nomenclature.

(Unpublished).
Boureau, E. (ed.) 1964. Traite de Paleobotanique 3 (Sphenophyta, Noeggerathiophyta). 544 pp., 436 figs.

Paris.
Burago, V. I. 1977. [Elements of Mesozoic floras in the Late Permian floras of South Primorye]. In

Krassilov, V. A. (ed. ), Palaeobotany in the Far East: 45-51 . Vladivostok (USSR Academy of Sciences).

[In Russian].
Chaloner, W. G. & Meyen, S. V. 1973. Carboniferous and Permian floras of the northern continents. In

Hallam, A. (ed.), Atlas of palaeobiogeography: 169-186. Amsterdam and New York.
El-Khayal, A. A., Chaloner, W. G. & Hill, C. R. 1980. Palaeozoic plants from Saudi Arabia. Nature,

Lond., 285 (5759): 33-34, 2 figs.
Florin, R. 1963. The distribution of Conifer and Taxad genera in time and space. Acta Horti Bergiani,

Stockholm, 20 (4): 121-312, 68 figs.

Grambast, L. J. 1974. Phylogeny of the Charophyta. Taxon, Utrecht, 23 (4): 463-481, 10 figs.
Groves, J. 1933. Charophyta. Fossilium Cat., Berlin (II: Plantae) 19. 74 pp.
'Gu & Zhi' 1974. [An introduction to Chinese fossils. Chinese plant fossils 1, Chinese Palaeozoic plants].

iii + 277pp., 142 figs, 130 pis. Peking (Scientific Publishing House). [In Chinese]. See footnote, p. 106.
Halle, T. G. 1927. Palaeozoic plants from Central Shansi. Palaeont. sin., Peking, (ser. A) 2 (1): 1-316,

64 pis.

Harris, T. M. 1962-3. Presidential address: the inflation of taxonomy. Proc. Linn. Soc. Lond., 175: 1-7.
Hirmer, M. 1927. Handbuch der Paldobotanik, 1. xvi + 708 pp. , 817 figs. Oldenbourg, Munich and Berlin.
Horn af Rantzien, H. 1956. An annotated check-list of genera of fossil Charophyta. Micropaleontology,

New York, 2 (3): 243-256.
Ibrahim, M. W. 1982. Lithofacies distribution of the Permian-Triassic rocks in the Middle East: A

Discussion. /. Petrol. Geol., Beaconsfield, 5 (1): 97-99.
Ishchenko, T. A. & Saidakovsky, L. J. 1975. [Finding of charophytes in the Silurian of Podolia]. Dokl.

Akad. Nauk SSSR, Leningrad, 220 (1): 209-211, pi. 1. [In Russian].
Kidston, R. & Lang, W. H. 1921. On Old Red Sandstone Plants showing Structure, from the Rhynie Chert

Bed, Aberdeenshire. Part V. The Thallophyta occurring in the Peat-Bed; the Succession of the Plants

throughout a Vertical Section of the Bed, and the Conditions of Accumulation and Preservation of the

Deposit. Trans. R. Soc. Edinb., 52 (4): 855-902, 11 figs, 10 pis.
Lakatos, I. 1970. Falsification and the methodology of scientific research programmes. In Lakatos, I. &

Musgrave, A. (eds), Criticism and the growth of Knowledge: 91-196. Cambridge.
Lemoigne, Y. 19810. Presence d'une flore comprenant des elements cathaysiens, dans le centre de 1'Arabie

Saoudite au Permien superieur. C. r. hebd. Seanc. Acad. Sci., Paris, (3) 292 (17): 975-977, 1 fig.

- 19816. Flore mixte au Permien superieur en Arabic Saoudite. Geobios, Lyon, 14 (5): 611-635, 7 figs,

6 pis.
Li Xingxue et al. 1974. See 'Gu & Zhi' 1974 and footnote, p. 106.

— & Yao Zhaoqi. 1980. An outline of recent researches on the Cathay sia flora in Asia. (Paper for the First

Conference of the International Organization of Palaeobotany, London and Reading, 1980). 15 pp.

Nanjing (Institute of Geology and Palaeontology, Academia Sinica).

112 C. R. HILL & A. A. EL-KHAYAL

Mamay, S. H. 1954. A Permian Discinites cone. /. Wash. Acad. 5c/., 44 (1): 7-11, 5 figs.

- 1967. Lower Permian plants from the Arroyo Formation in Baylor County, North-central Texas.
Prof. Pap. U.S. geol. Surv., Washington, 575C: C120-C126, 2 figs.

Peck, R. E. & Eyer, J. A. 1963. Pennsylvanian, Permian, and Triassic Charophyta of North America. J.

Paleont., Tulsa, Okla., 37 (4): 835-844, 1 fig, pis 100-101.
Popper, K. R. 1959. The logic of scientific discovery. 480 pp., 2 figs. London.
Powers, R. W., Ramirez, L. F., Redmond, C. D. & Elberg, E. L. jr 1966. Geology of the Arabian

Peninsula, Sedimentary Geology of Saudi Arabia. Prof. Pap. U.S. geol. Surv., Washington, 560D.

vi + 147 pp., 14 figs, 10 pis.
Racki, G. 1982. Ecology of the primitive charophyte algae; a critical review. Neues Jb. Geol. Palaont.

Abh., Stuttgart, 162 (3): 388-399, 5 figs.
Rezak, R. 1959. Permian algae from Saudi Arabia. /. Paleont., Tulsa, Okla., 33 (4): 531-539, 1 fig., pis

71-72.
Schweitzer, H.-J. 1963. Der weibliche Zapfen von Pseudovoltzia liebeana und seine Bedeutung fiir die

Phylogenie der Koniferen. Palaeontographica, Stuttgart, (B) 113 (1-4): 1-29, 32 figs, 9 pis.

- 1968. Die Flora des Oberen Perms in Mitteleuropa. Naturw. Rdsch. Stutt., 21 (3): 93-102, 13 figs.
Sharief, F. A. 1982a. Lithofacies distribution of the Permian-Triassic rocks in the Middle East. /. Petrol.

Geol., Beaconsfield, 4 (3): 299-310, 5 figs.

- 1982ft. Lithofacies distribution of the Permian-Triassic rocks in the Middle East: A Reply. /. Petrol.
Geol., Beaconsfield, 5 (2): 203-206.

A. B. Smith

Department of Palaeontology, British Museum (Natural History), Cromwell Road, London
SW7 5BD

Synopsis

The entire British fauna of Carboniferous edrioasteroids is revised and redescribed and a newly discovered
hardground, where edrioasteroids are exceedingly abundant, is described. All belong to the family
Agelacrinitidae and three genera and four species are recognized. Two of the genera, Lepidodiscus and
Postibulla, are known from North America but the third, Stalticodiscus (type species Lepidodiscus milleri),
is new. The new species Postibulla neglecta is described. Growth and plate ultrastructure of Stalticodiscus
milleri have been studied using scanning electron microscopy. This species was able to orientate itself in
currents and possibly lived in bimodal (i.e. tidal) current regimes with the anterior-posterior axis at right
angles to the flow of water.

Introduction

It is now more than a hundred years since the first edrioasteroid was described from the
Carboniferous of the British Isles, yet they have remained extremely rare fossils known from
only a few localities in northern England. Three species have been described with varying
accuracy, Lepidodiscus lebouri Sladen 1879, Lepidodiscus milleri Sharman & Newton 1892 and
Lepidodiscus fistulosus Anderson 1939. This last species was thought so distinct that Regnell
(1950) erected the genus Anglidiscus for it.

The only edrioasteroids known from the Carboniferous are isorophids belonging to the family
Agelacrinitidae. These, like all isorophids, were sessile and lived attached to hard substrates.
Only the upper (ventral) surface is calcified and the skeleton consists of imbricate or tesselate
plates set within a soft tissue membrane. Upon death, the theca rapidly dissociates as the soft
tissue decays and so, to be preserved, edrioasteroids must be buried alive or within a very short
time of death. However, isorophids lived on hard substrates in areas of active erosion where
they stood very little chance of being preserved. It is therefore only under exceptional
circumstances that we ever find them in the fossil record. Until now, most specimens found in
the British Carboniferous have been single individuals attached to shells, presumably living at
the limits of tolerance offshore to the main population. The discovery of a new horizon in the
Lower Carboniferous of Cumbria yielding an abundance of well-preserved edrioasteroids is an
unusual and important find and I am extremely grateful to Dr Paul Taylor of the British Museum
(Natural History) who brought this occurrence to my notice.

The discovery prompted a re-examination of the previously-described species, and it soon
became apparent that published descriptions were unsatisfactory. The contemporary American
fauna is now well known from the work of Bassler (1936), Kesling (1960) and especially Bell
(1976a). This paper sets out to revise the British Carboniferous edrioasteroids.

Occurrence

All the British Carboniferous edrioasteroids come from the Early Asbian stage of the
Dinantian. American Carboniferous edrioasteroids have been discovered at various levels
throughout the Mississippian and it is not at all clear why the British fauna should be so restricted
in its occurrence. There are five localities that have yielded edrioasteroids, all of them in

Bull. Br. Mus. nat. Hist. (Geol.)37(3): 113-138 113 Issued 24 November 1983

114 A. B. SMITH

northern England. I am indebted to Dr W. H. C. Ramsbottom of the Institute of Geological
Sciences, Leeds, who supplied me with accurate stratigraphic data on these localities, as follows.

(i) The River Irthing, one mile (1-6 km) east of Waterhead, Northumberland (National Grid
reference (approx.) NY 635685). This occurrence was reported by Sharman & Newton (1892)
and Ramsbottom (1970: 172). Here the Millerhill Limestone outcrops as a strike section for
some distance. This limestone is divided into an upper and a lower unit but it is not known from
which of these the single edrioasteroid came. The Millerhill Limestone lies within the Upper
Border Group and is middle Early Asbian in age. One specimen of Stalticodiscus milleri
(Sharman & Newton) has been collected from here.

(ii) The River Irthing, % of a mile (1 km) south of Lampert, Northumberland at the foot of
Linen Sike (NY 683735). A specimen of Stalticodiscus milleri (Sharman & Newton) has been
found here in a shelly calcareous shale that lies just above the Millerhill Limestone. This is again
middle Early Asbian in age. It was recorded by Sharman & Newton (1892) and Ramsbottom
(1970: 172).

(iii) A horizon 103ft 6 in (31 -5m) down the Hetton House borehole (NU 042296) yielded
some 33 edrioasteroids (Anderson 1939). These come from a siltstone with calcareous bands
situated about half way up the Scremerston Coal Group and Early Asbian in age, probably not
very far from the Millerhouse Limestone horizon. The fauna consists mainly of Lepidodiscus cf .
squamosus Meek & Worthen with subsidiary Postibulla neglecta.

(iv) An impure limestone outcropping in the River Rede, where it forms a low waterfall just
north of the bridge above the village of East Woodburn, Northumberland (NY 901877). Here
was found the only known specimen of Lepidodiscus lebouri Sladen (Sladen 1879; Miller 1887:
41). This horizon is approximately 900ft (275 m) below the Redesdale Limestone (not 700ft as
stated by Miller) and lies close to the top of the Early Asbian.

(v) The road cutting at Penruddock on the north side of the A66 road just at the end of a
stretch of dual carriageway some 6 miles (9-5 km) west of Penrith, Cumbria (NY 438275). The
succession here is given in Fig. 1. Edrioasteroids occur crowded on the upper surfaces of bored
and encrusted micritic concretions which can be several feet in diameter. This horizon had
obviously been exposed for a considerable time before being smothered by a rapid influx of
mud. In general the encrusting bryozoa and inarticulate brachiopods are found on the
undersides of the concretions whereas acrothoracic barnacle borings and edrioasteroids are
found on the upper surfaces. Concretions with abundant borings tend not to have edrioasteroids
and vice versa.

The brachiopods from this locality suggest an Asbian age, according to Dr C. H. C. Brunton
of the BM(NH) (personal communication). The foraminifera, which include Koninckopora
inflata (de Koninck), Eostaffella parastruvei Rauser, Archaeodiscus sp. and Globoendothyra,
were identified by Dr A. R. E. Strank of the I.G.S., Leeds, and indicate a Holkerian or Early
Asbian age. Dr W. H. C. Ramsbottom informs me that the beds at this locality probably belong
to the undivided Sixth/Seventh Limestones, in the lower part of the Early Asbian.

Taxonomy

Order ISOROPHIDA Bell, 1976

Suborder ISOROPHINA Bell, 1976

Family AGELACRINITIDAE Chapman, 1860

Genus LEPIDODISCUS Meek & Worthen, 1868
[= Anglidiscus Regnell, 1950]

Lepidodiscus cf. squamosus Meek & Worthen, 1868
Figs 2-5, 8

1868 Agelacrinites (Lepidodiscus) squamosus Meek & Worthen: 357-358.
1939 Lepidodiscus fistulosus Anderson: 68 (part).

CARBONIFEROUS EDRIOASTEROIDEA

115

cavernous biosparite

Lithostrotion colonies and gigantoproductids
in life position

large broken colonies of Syringopora

shelly biosparite with transported
coral fragments

black clay

greenish clay with botrioidal
micrite concretions
?caliche horizon

cross- laminated biosparite with transported
fauna of large productids, bryozoa, etc.

cross-bedded biosparite

• biosparites with siltstone partings

siltstones with tnin red-weathering crinoidal biosparites

crinoidal biosparite

= <-^>. <=>

1 meter

siltstone with bored and encrusted
micrite concretions at base
calcarenite with conglomeratic top
of rounded micrite intraclasts

bioturbated fine grained calcarenite
with many partings

siltstone

impure fine grained calcarenite with
spiriferids, etc.

Fig. 1 Sedimentary log for succession exposed at Penruddock road cutting, near Penrith (loc. v). A,
succession towards the west end of the road cutting. B, succession in the old quarry immediately
above the road cutting. C, succession towards the east end of the road cutting, east of the an
obvious fault, o - the horizon with edrioasteroids encrusting limestone concretions.

116

A. B. SMITH

3N_ "f 11 ' »m»%

f J-/ ^^ ' ' *

• -w ,sv

CARBONIFEROUS EDRIOASTEROIDEA 117

1950 Anglidiscus fistulosus (Anderson); Regnell: 6 (part).
1966 Anglidiscus fistulosus (Anderson); Regnell: U162 (part).

1976a Lepidodiscus squamosus Meek & Worthen; Bell: 253-257 (q.v. for full bibliography of American
records).

DIAGNOSIS. An agelacrinitid with a clavate theca and long curved ambulacra. Ambulacra I-IV
curve sinistrally, V curves dextrally (rarely all curve sinistrally). Cover plates arranged cyclically
in groups of six or seven. Oral area composed of many cover plates continuous with ambulacral
series. Hydropore rise in posterior right of oral area (type VI of Kesling, 1960), posterior side
formed of many plates. Interambulacral plates squamose, imbricate. Periproct an anal valve of
two cycles of plates. Flooring plates uniserial, imbricate.

MATERIAL. Institute of Geological Sciences no. 60235. Half of a six-inch (150mm) core with
parts of 31 specimens.

LOCATION AND AGE. Hetton House bore-hole, Northumberland (loc. iii, p. 114). Early Asbian,
Dinantian.

DESCRIPTION. The British specimens of this species all come from one piece of bore-hole core.
There are parts of 31 specimens of which only 15 are tolerably complete. A further two
specimens belong to the genus Postibulla and are described later. All the specimens are
preserved upside down, revealing the inner surface of the theca. Latex moulds were made of
three of the better-preserved specimens.

In life the specimens must have been tall and domal in shape but most are preserved in the
contracted state. In all but one specimen ambulacra I-IV curve sinistrally and ambulacrum V
curves dextrally. There is, however, one (? abnormal) specimen in which all five ambulacra
curve sinistrally (Fig. 8). Distally the ambulacra curve round to become parallel to the
periphery. The arrangement of cover plates is nowhere clear but cycles of three or four large
plates together with small intercalated plates can be seen. This seems comparable with the
cyclical cover plate arrangement seen in the better-preserved American material. Ambulacral
flooring plates are uniserial and imbricate. Contrary to Anderson's (1939) findings, the flooring
plates are totally imperforate. Preservation around the oral area is too poor to permit a detailed
analysis of the cover plate arrangement but it is clear that the oral cover plates are continuous
with the ambulacral cover plates and that no enlarged oral primary cover plates are present.
Oral cover plates are numerous and small. There is a prominent hydropore bulge situated to the
posterior of the oral area adjacent to ambulacrum V. The posterior side of this bulge is formed
by a number of small plates.

Interambulacral areas are composed of numerous squamose, imbricate plates that become
noticeably smaller towards the ambulacra and around the anal cone. The inner surface of these
plates is composed of a coarse-meshed stereom which led Anderson (1939) to believe
mistakenly that they were perforate. The anal cone lies roughly central in interambulacrum 5
and consists of a double circlet of rather elongate triangular plates. The peripheral rim is of
standard appearance.

Figs 2-6 IGS no. 60235, the half core from which Anderson described Lepidodiscus fistulosus. Fig. 2,
the whole specimen, x 0-8; L = lectotype of Anglidiscus fistulosus (= Lepidodiscus cf. squamosus)
(see Fig. 4), P = holotype of Postibulla neglecta sp. nov. (see Fig. 6). Fig. 3, Lepidodiscus cf.
squamosus Meek & Worthen, latex cast of specimen xxix of Anderson (1939: 70) situated to the
lower right of the letter P in Fig. 2, x4. Fig. 4, Lepidodiscus cf. squamosus Meek & Worthen,
natural mould, lectotype of Anglidiscus fistulosus (Anderson) (L in Fig. 2), x4. Fig. 5, Lepido-
discus cf. squamosus Meek & Worthen, latex cast of specimen xxii of Anderson (1939: 70), an
abnormal individual with all five arms curving sinistrally, x4; see Fig. 8. Fig. 6, Postibulla neglecta
sp. nov., holotype (Pin Fig. 2), number xxvi of Anderson (1939: 70), x4; see Fig. 11.

Fig. 7 Lepidodiscus lebouri Sladen, BM(NH) E29330, holotype x2V2 . See Fig. 9.

Figs 2-7 whitened with ammonium chloride sublimate.

A. B. SMITH

Fig. 8 Lepidodiscus cf. squamosus Meek & Worthen. Camera-lucida drawing of Anderson's
specimen xxii (IGS no. 60235), an abnormal individual with all five arms curving sinistrally. See
Fig. 5.

DISCUSSION. One rather unusual feature of this material is that the specimens are preserved in a
siltstone with thin calcareous bands. They are not all preserved on one level but are present at
two levels a few millimetres apart. It is obvious that the sediment was not lithified into a hard
ground at this horizon and that the edrioasteroids were not attached to the sediment, since they
have all separated to reveal their inner surfaces. The most likely explanation for this is that the
edrioasteroids were originally attached to one or more fronds of free-standing alga which
became detached and transported before being rapidly buried. Decay of the alga left the
edrioasteroids buried without trace of the substratum to which they were attached.

CARBONIFEROUS EDRIOASTEROIDEA 119

The fact that the specimens only show the inner surface of their thecal plating, together with
the rather poor state of preservation, makes the interpretation of their structure difficult. This
probably explains why Anderson's (1939) original description contains a number of basic
misconceptions. No holotype of Lepidodiscus fistulosus was designated by Anderson and only a
general reconstruction was given. However, as Anderson's reconstruction was supposedly
based mainly on his specimen (1939: 70) viii, this is here designated the lectotype. Anderson
believed that all the edrioasteroids on this block belonged to L. fistulosus, but although most
individuals, including the lectotype, are here referred to Lepidodiscus cf. squamosus, two
belong to the genus Postibulla. There is no evidence that the flooring plates are pierced by pores,
nor can I find the purported median groove on the flooring plates. More importantly, the oral
plating arrangement shown in Anderson's reconstructoin is incorrect. Anderson (1939: 78-79)
assumed that 'the mouth ... is covered by three peristomal plates as in A(gelacrinites) pileus
Hall, though only the posterior one can be recognized' . I can only think that he mistook the large
hydropore plate seen in one of the specimens of Postibulla for the posterior primary oral cover
plate (Fig. 11, p. 122). The reconstructed arrangement of ambulacral cover plates is also
incorrect.

Lepidodiscus fistulosus was referred to the family Hemicystitidae by Regnell (1950), who
created the new genus Anglidiscus for it. He did this on the strength of Anderson's description
and without having seen the original specimens. As the species is synonymous with Lepidodiscus
squamosus, Anglidiscus is a junior synonym for Lepidodiscus.

In America, Lepidodiscus squamosus is known from Indiana and Pennsylvania in beds of the
Kinderhookian and Osagean Series (Bell 19760) which pre-date the British find (George et al.
1976). There Lepidodiscus is a fairly long-ranged genus occurring throughout the Mississippian.
L. squamosus differs from all other species referred to this genus in having imbricate rather than
abutting flooring plates. Compared with American material, in the British specimens of L. cf.
squamosus the imbrication of the flooring plates is slightly less pronounced and there is
somewhat less of an overlap of the two posterior ambulacra behind the periproct. Otherwise the
two are comparable as far as can be made out.

The one specimen in which all five ambulacra curve in the same direction (Fig. 8) is in all other
respects identical to the remaining specimens of L. cf. squamosus. It is clearly just an abnormal
individual within the population.

Lepidodiscus lebouri Sladen, 1879
Figs 7, 9, 10

1876 Agelacrinites (Lepidodiscus) squamosus Meek & Worthen; Lebour: 22.
1879 Lepidodiscus lebouri Sladen: 745; pi. 37, figs 1-4.
1936 Lepidodiscus lebouri (Sladen); Bassler: 20; pi. 1, fig. 19.

MATERIAL. Holotype and only known specimen British Museum (Natural History)
Palaeontology Dept. no. E29330.

LOCATION AND AGE. From the River Rede near East Woodburn, Northumberland (loc. iv,
p. 114). Early Asbian, towards the top of the sub-stage.

DIAGNOSIS. Large clavate species of Lepidodiscus with long, curved ambulacra: ambulacra I-IV
curve sinistrally, ambulacrum V curves dextrally. Cover plates arranged in cycles of six. Anal
cone lies in interambulacrum 5 close to ambulacrum I and is bordered distally by the tip of
ambulacrum V. Hydropore included to the posterior right of the oral area, bounded posteriorly
by a few small plates. Interambulacra composed of numerous tesselate plates.

DESCRIPTION. There is only one specimen known of this species and it is preserved ventral
surface uppermost. Peripherally the plating curves underneath suggesting that in life the theca
was clavate in shape. It is a large edrioasteroid with a diameter of 25mm. There are six

120

A. B. SMITH

Fig. 9 Lepido discus lebouri Sladen. Camera-lucida drawing of the holotype BM(NH) no. E29330.

See Fig. 7.

ambulacra, an abnormality produced by ambulacrum I bifurcating shortly after it had separated
from ambulacrum II. Ambulacrum V curves dextrally; all other ambulacra curve sinistrally. The
oral area is covered by a large number of cover plates that are continuous with the ambulacral
cover plates. Unfortunately the plating is somewhat disrupted (Fig. 9) and the exact
arrangement cannot be determined. There are no distinctly larger primary cover plates.
Ambulacral cover plates are arranged in cycles of six, three or four larger plates plus two or three
tiny occluded cover plates in each cycle (Fig. 10). The perradial suture is markedly zigzag except
over the oral area and the ambulacra form obvious ridges on the theca. Cover plates are small,
triangular and wedge-shaped in cross section. The larger cover plates in each cycle have
intrathecal extensions. Towards the distal end of the ambulacra the small occluded cover plates
are lost from the cycles. Ambulacral flooring plates are largely covered, but can be seen in cross
section in ambulacrum II. They are uniserial and U-shaped in cross section.

The hydropore belongs to type VI of Kesling (1960). It lies in the right posterior side of the
oral area and is bounded by cover plates anteriorly, and posteriorly by a small number of (?)

CARBONIFEROUS EDRIOASTEROIDEA 121

Fig. 10 Camera-lucida drawing of cover plate
arrangement from ambulacrum II of the
1 mm holotype of Lepidodiscus lebouri (BM(NH)

_._ no. E29330).

interambulacral plates. The posterior slope to the oral area is steep and formed by two rather
large plates together with a number of smaller plates (Fig. 9).

Interambulacral areas are broad and composed of numerous sub-polygonal tesselate plates.
These are relatively thick and imbricate adorally. The interambulacral plates are largest near the
centre of each area but become obviously smaller close to ambulacra and around the periproct.

The periproct is largely disrupted and individual plates of the anal cone lie scattered nearby. It
is situated in the more distal left-hand side of interambulacrum 5, fairly close to ambulacrum I.
The tip of ambulacrum V curves round to lie just posterior to the periproct.

DISCUSSION. There are three species of Lepidodiscus known from North America (Bell 1976a):
L. squamosus Meek & Worthen, L. laudoni (Bassler) and L. sampsoni (Miller). L. lebouri
differs from L. squamosus in having tesselate ventral plating and a clavate body. It differs from
L. sampsoni in having curved ambulacra: the ambulacra in L. sampsoni form long, straight
ridges on the ventral surface. L. lebouri comes closest to the common North American species
L. laudoni, which is found throughout the Mississippian ranging from the Kinderhookian to the
Chesterian. Unfortunately the plating of the pedunculate zone and the internal aspect of the
ambulacral flooring plates are unknown for L. lebouri. In other features the two species are
closely comparable, save for the abnormal sixth ambulacrum in L. lebouri and the presence of
two prominent plates forming the posterior slope to the oral area. Although L. lebouri will
probably prove to be conspecific with L. laudoni, the two species are here retained as distinct
until further British material becomes available for comparison. In uniting the two species L.
laudoni would become a junior synonym, which would have the undesired consequence of
making the holotype of this common species a six-armed abnormality.

Sladen's (1879) original description was comprehensive and for the most part accurate, as was
the accompanying illustration. His interpretation of the arrangement of ambulacral cover plates
is not quite correct, however, as he failed to notice the presence of small occluded plates. Sladen
quite correctly recognized the species' distinctness from Lepidodiscus squamosus and
'Lepidodiscus' (Discocystis) kaskaskiensis .

Genus POST1BULLA Bell, 1976a

Postibulla neglecta sp. nov.
Figs 2, 6, 11

1939 Lepidodiscus fistulosus Anderson: 68 (part).

1950 Anglidiscus fistulosus (Anderson) Regnell: 6 (part).

1966 Anglidiscus fistulosus (Anderson) Regnell: U162 (part).

DIAGNOSIS. Agelacrinitid with a domal theca. Ambulacra tall, narrow; ambulacra I-III curve
sinistrally, ambulacra IV and V curve dextrally. Cover plates arranged in an alternating series
with both large and intercalated plates but precise arrangement not clear. Oral area markedly
elongate. Oral cover plates small, undifferentiated from ambulacral cover plates; anterior and
posterior series equally developed. Hydropore rise large, separated from oral area; includes one
very prominent hydropore plate. Interambulacral plates numerous, squamose and imbricate.
Anal pyramid narrow and prominently elevated. Peripheral skirt unknown.

122

A. B. SMITH

Fig. 11 Postibulla neglecta sp. nov. Camera-lucida drawing of the holotype, on IGS no. 60235 (P in

Fig. 2). See Fig. 6.

CARBONIFEROUS EDRIOASTEROIDEA 123

NAME. 'Overlooked'.

MATERIAL. Holotype, Institute of Geological Sciences no. 60235, specimen xxvi of Anderson
(1939: 70). Paratype, specimen xxx, same slab.

LOCATION AND AGE. Hetton House bore-hole, Northumberland (loc. iii,p. 114). Early Asbian,
Dinantian.

DESCRIPTION. Both specimens are 12 to 13mm in diameter and in life would have been
moderately tall and domal in shape; they are preserved in their retracted state. Ambulacra are
long and narrow and form prominent ridges over the theca. They are fairly straight adorally but
towards the periphery they curve to run parallel with the margin. Ambulacra I-III curve
sinistrally whereas ambulacra IV and V curve dextrally. The cover plates are tall and narrow.
Where they have fallen outwards each can be seen to have a prominent sagittal ridge on its inner
face. The cover plates form a tall steep-sided arch above the ambulacral grooves. Cover plate
arrangement is not clear but there appears to be an irregular biseries of larger plates alternating
with smaller intercalated plates (Fig. 11). The flooring plates cannot be seen.

The oral area is narrow but laterally extensive so that the ambulacra are in a clear 2-1-2
arrangement. The oral cover plates are continuous with ambulacral cover plates and there are
no obviously larger plates. One large plate forms the posterior slope to the oral area opposite
ambulacrum III but this I interpret as the most proximal interambulacral plate. An obvious
mound to the posterior right-hand side of the oral area forms part of the hydropore structure.
There is one large distal hydropore plate clearly separated from the oral area and a smaller
proximal hydropore plate. Some of the proximal cover plates of ambulacrum V also border on
the hydropore slit.

Interambulacral areas are formed of numerous squamous and imbricate plates. In the most
adoral areas of interambulacra 2 and 3 there are crescentic raised areas (Fig. 6, p. 116) which
Anderson (1939) interpreted, wrongly in my opinion, as the sites of gonads. The periproct is
situated centrally in interambulacrum 5 . It is narrow-based and composed of two cycles of rather
elongate plates that are wedge-shaped in cross section. In the holotype the anal cone is
preserved in its open position.

Beyond the ambulacra, interambulacral plates are vertical and stacked together, showing that
the theca was tall in life and has since collapsed. Neither specimen shows the peripheral rim.

DISCUSSION. In erecting the species Lepidodiscus fistulosus Anderson (1939) did not realise that
individuals belonging to two species (Lepidodiscus cf . squamosus and Postibulla neglecta) were
present in his material. His composite reconstruction (1939: fig. 1) bears little resemblance to
either species. The reconstruction of soft tissue anatomy given by Anderson (1939: fig. 6) is
based on the internal appearance of the holotype of Postibulla neglecta. However, apart from
the crescentic ridges adjacent to the oral area which he interpreted as the sites of gonads, other
structural details cannot be verified.

Both the specimens are preserved upside down and reveal the mould of the external surface.
As was discussed for Lepidodiscus cf. squamosus (p. 118) the specimens are not attached to a
recognizable hard ground surface and were probably growing on fronds of a free-standing alga.

The arrangement of ambulacra, the steep-sided, narrow ambulacral ridges and the separation
of the hydropore bulge from the oral area clearly place these specimens in the genus Postibulla.
Five species of Postibulla are known, but only two of them come from the Carboniferous (Bell
19760). These are P. legrandensis (Miller & Gurley) and ? P. jasperensis (Harker), both from
the Kinderhookian, low in the Mississippian. P. legrandensis differs from the British species in
having three plates forming the posterior rim of the hydropore opening and in having a
prominent bulge directly opposite ambulacrum III on the posterior margin of the oral area (Bell
19760: pi. 40). ? P. jasperensis is known only from one small specimen which was placed in this
genus only with reservation by Bell (19760). The cover plates are simple, not irregularly biserial,
and the posterior oral proturberance is hardly developed; it is quite unlike the British
specimens.

Postibulla neglecta is the youngest species of this genus known.

124

A. B. SMITH

CARBONIFEROUS EDRIOASTEROIDEA 125

Genus STALTICODISCUS nov.

DIAGNOSIS. A genus of agelacrinitid with tall domal to subclavate theca. Ambulacra I-IV curve
sinistrally, ambulacrum V curves dextrally . Ambulacral cover plates arranged in cycles of three,
one of which is usually occluded adradially, and with an irregularly zigzag perradial suture.
Hydropore rise included in posterior right of oral area (type VI of Kesling, 1960), posterior
bounded by two or three plates only. Oral cover plates small, numerous, not differentiated from
ambulacra! cover plates. Flooring plates uniserial and strongly imbricate. Valvular anal cone
consisting of two cycles of plates situated centrally in interambulacrum 5 . Interambulacral plates
more or less tesselate ventrally but becoming imbricate laterally where they form a cylindrical
pedunculate zone. Peripheral skirt present, consisting of some five or six series of plates.

NAME. Greek oraA,Tix6?, 'contracting'.

TYPE SPECIES. Lepidodiscus milleri Sharman & Newton, 1892. Monotypic.

DISTRIBUTION AND AGE. From Penruddock, Cumbria and the River Irthing, Northumberland:
Early Asbian.

DISCUSSION. This genus is easily distinguished from Lepidodiscus on the arrangement of the
cover plates. Lepidodiscus has cover plates that are arranged in cycles of six, usually with three
larger plates and three smaller occluded plates (Fig. 10), whereas Stalticodiscus has cover plates
arranged in cycles of three with two larger plates and a smaller, often occluded plate.
Discocystis, like Stalticodiscus, has its cover plates arranged in cycles of three, or occasionally in
cycles of four, but here the perradial edge of the cover plates is obviously serrated, suggesting
that there are 'multiple intra-ambulacral extensions on the ambulacral tunnel surfaces of the
cover plates' (Bell 1976a: 251). Furthermore, Discocystis has polygonal, tesselate plating in the
interambulacral areas and a downwardly constricting pedunculate zone of subrectangular plates
clearly demarcated from the ventral surface. In Stalticodiscus, interambulacral plating is, at
most, sub-tesselate and the pedunculate zone is neither clearly demarcated from the ventral
surface, not downwardly tapering. The flooring plates of Discocystis abut along vertical sutures,
whereas those of Stalticodiscus are strongly imbricate. For these reasons Stalticodiscus and
Discocystis, though clearly closely related, are separated as distinct genera.

Stalticodiscus milleri (Sharman & Newton, 1892)
Figs 12-51

1892 Lepidodiscus milleri Sharman & Newton: 150; pi. 2, figs 1-5.
1936 Lepidodiscus milleri Sharman & Newton; Bassler: 20; pi. 7, fig. 7.

DIAGNOSIS. A subclavate species of Stalticodiscus up to 15 mm in diameter and 30 mm or so in
height. Interambulacral plating sub-tesselate ventrally, imbricate laterally. Periproct situated
centrally in interambulacrum 5. Ambulacra I and V just overlapping behind the periproct in
adults.

MATERIAL. Holotype, Institute of Geological Sciences no. 7662. Other specimens, IGS 25105,
also many hundreds of individuals from Penruddock, British Museum (Natural History)
Palaeontology Dept nos E29878-925.

LOCATION AND AGE. The holotype IGS 7662 and IGS 25105 come from the Millerhill limestone,
Early Asbian, of the River Irthing, Northumberland. The holotype comes from near Waterhead

Figs 12-13 Stalticodiscus milleri (Sharman & Newton), BM(NH) no. E29878. Fig. 12, a limestone
concretion encrusted by adults some in extended posture. Notice that all are aligned with their
anterior towards the right or top right. Fig. 13, an enlarged view of the two fully extended
individuals seen in Fig. 12, showing the imbricate pedunculate zone. On collapse the theca has been
compressed onto its own elevated peripheral rim to produce the circular impression towards the
base. Scale bar = 5 mm. All specimens whitened with ammonium chloride sublimate.

126

A. B. SMITH

Figs 14-19 Stalticodiscus milleri (Sharman & Newton). Figs 14-15, BM(NH) no. E29884: Fig. 14
under xylene, Fig. 15 whitened with ammonium chloride sublimate. Figs 16-17, BM(NH) no.
E29889: Fig. 16 under xylene, Fig. 17 whitened with ammonium chloride sublimate. Fig. 18,
BM(NH) no. E29885. Oral surface with cover plates partially lost to reveal uniserial flooring plate
arrangement. Fig. 19, BM(NH) no. 29888. Posterior half of the specimen showing the peripheral
rim plating and the arrangement of plates forming the periproct. Note the large hydropore bulge to
the posterior right of the oral area. Scale bar 5mm.

CARBONIFEROUS EDRIOASTEROIDEA

127

Fig. 20 Stalticodiscus milleri (Sharman & Newton), BM(NH) no. E29884. Camera-lucida drawing.

(loc. i, p. 114), the other from near Lampert (loc. ii). Many hundreds of individuals encrusting
limestone concretions have also been collected from Penruddock, Cumbria (loc. v) and are also
Early Asbian in age.

DESCRIPTION. Until recently, only two poorly-preserved specimens of this species were known.
The discovery of colonies totalling many hundreds of individuals, ranging in size from less than
0-5 mm to more than 15mm in diameter, at Penruddock has provided an abundance of
well-preserved material on which to base this description.

Mature individuals are subclavate in shape when fully extended (Fig. 13) but are usually
preserved in a contracted posture with the lateral peduncular zone telescoped together just
inside the marginal ring (Fig. 27) . When fully extended the theca is twice as tall as it is broad, and
it tapers slightly towards its attachment base. The ambulacra extend down less than half the
height of the theca. Juveniles are less elevated and generally domal in shape. The pedunculate
zone only starts to become obviously developed once individuals have reached a diameter of
about 8 to 10mm. The peripheral rim is usually circular in outline but may be distorted if the
individual is attached to an uneven surface or a relatively small object such as a shell.

128

A. B. SMITH

Fig. 21 Stalticodiscus milleri (Sharman & Newton), IGS no. 7662, holotype. Camera-lucida

drawing. See Fig. 25.

Figs 22-23 Camera-lucida drawings of ambulac-
ral cover plate arrangement in Stalticodiscus
milleri (Sharman & Newton), approximately
mid-length along the ambulacrum. Fig. 22
BM(NH) no. E29571; Fig. 23 BM(NH) no.
E29885.

CARBONIFEROUS EDRIOASTEROIDEA 129

Ambulacra form prominent ridges over the theca, and are only moderately long. Ambulacra
I-IV curve sinistrally, ambulacrum V curves dextrally. The two posterior ambulacra curve
round to encircle the periproct, but it is only in the largest individuals that the two ambulacra
more or less meet posteriorly behind the periproct. These ambulacra may just overlap at the
anterior-posterior mid-line. Ambulacra maintain a uniform breadth along most of their length,
tapering only at their distal tip. Cover plates are arranged in irregular cycles of three, usually
consisting of two larger plates and a smaller, often occluded plate (Figs 22, 23). Proximally, near
the oral area plating becomes more irregular. Near the distal tip (Fig. 34) the perradial suture is
obviously zigzagged, but over most of the ambulacrum this suture is much more irregular in its
path (Fig. 30, p. 131).

Cover plate arrangement continues across the oral area without obvious distinction from
ambulacral areas (Figs 30-32). A few plates are slightly larger than any found in the ambulacra,
but there are no enlarged primary oral plates. Two shared cover plates situated at the fork
between paired lateral ambulacra are usually conspicuous. The arrangement of plates in the oral
area is not fixed, and in the large number of specimens available much variation exists,
particularly in the presence or absence of small occluded plates.

Ambulacral flooring plates are uniserial and imbricate. Each plate has a large distal tongue
which overlaps the base of the adjacent flooring plate (Figs 38-40). This imbrication is not
immediately apparent from an external view of an ambulacrum stripped of its cover plates (Fig.
18). Flooring plates have a deep, U-shaped channel which shows no signs of muscle attachment
scars or grooves for soft-tissue tracts. Although the floor of this channel is relatively thick, the
lateral walls are thinner. There is a distinct bulge on each side approximately half-way along
(Figs 38-40). The functional significance of this structure is unknown but it is present in at least
some other agelacrinitid edrioasteroids (Bell 1976fl) . Cover plates rest on top of the lateral walls
of the flooring plates but there appears to be no obvious structural modification to accommodate
them. Adradially the cover plates have a small intrathecal extension which overlaps the flooring
plate to extend beneath adjacent interambulacral plating (exposed in areas shown in Figs 31 and
34). The structure of the oral frame is unknown but is presumably like that of other
agelacrinitids.

The hydropore structure forms a distinct bulge to the right posterior of the oral area (Figs
27-31). It is bounded by several cover plates of ambulacrum V anteriorly and by two large
hydropore plates to the posterior. The hydropore bulge matches a type VI hydropore of Kesling
(1960).

Interambulacral areas are broad. On the ventral surface the plating is sub-tesselate (Figs
14-17) but laterally the interambulacral plates imbricate to form a pedunculate zone.
Interambulacral plates are thin with bevelled edges. The outer surface is covered by a dense
stereom layer with a granular surface (Figs 31-36). The bulk of the plate is thickened with a
coarse labyrinthic stereom but there are a couple of retiform stereom layers near the outer
surface (Figs 37, 41). Within each interambulacrum larger plates are found towards the centre
whereas smaller plates border the ambulacra and surround the periproct. The imbricate plating
making up the pedunculate zone is arranged irregularly (Fig. 13). This zone could be expanded
and contracted like a telescope by varying the amount of overlap of the constituent plates.
Presumably there were both meridional and circumferential muscle layers underlying the thecal
plating to bring about this change in shape.

The periproct lies centrally in the posterior interambulacrum. It forms a well-defined anal
cone composed of two cycles of plates. The primary cycle consists of some eight large triangular
plates. Within this cycle, and largely hidden from sight, is a second cycle of smaller plates lying
distally nestled between primary anal cone plates. When the anal cone is undisrupted only the
very tips of these secondary plates can be seen at the apex of the cone (Fig. 30). In large
specimens the primary and secondary plates alternate.

The peripheral rim is circular in outline and forms a cohesive framework. It consists of some
five rows of plates which become progressively smaller towards the margin (Figs 19, 24, 36). The
innermost plates alternate, every other plate being set behind. These plates have an expanded
base which is set firmly on the substratum. The base is marked by a series of radial ridges and

130

A. B. SMITH

CARBONIFEROUS EDRIOASTEROIDEA

131

Fig. 30 Stalticodiscus milleri (Sharman & Newton), BM(NH) no. E29887. Camera-lucida drawing
showing the cover plate arrangement in the oral area. See Fig. 29.

grooves which presumably played some role in adhesion. The more distal plates of the
peripheral rim are tesselate and also have radial ridges on their lower surfaces, but these are less
well developed.

GROWTH. The changes that take place during the growth of isorophid edrioasteriods have been
described by Bell (19766). Bell was able to study ontogeny in all the isorophid families save for
the Agelacrinitidae. The large number of juvenile Stalticodiscus milleri now available permits
the first detailed description of growth in a member of the Agelacrinitidae. In this study, small
individuals were studied using a scanning electron microscope and a representative growth
series is shown in Figs 42-51.

Figs 24-29 Stalticodiscus milleri (Sharman & Newton). Fig. 24, BM(NH) no. E29886, 7mm
diameter juvenile, x8. Fig. 25, IGS no. 7662, holotype from the River Irthing, x6; see Fig. 21. Fig.
26, IGS no. 25105, the other specimen mentioned by Sharman & Newton (1892) as coming from
the River Irthing, x6. Fig. 27, BM(NH) no. E29879, a large specimen; anterior to the right
showing the well-developed zone of imbricate plates, x6. Fig. 28, BM(NH) no. E29883, small
adult from Penruddock with the same diameter as the holotype, x6. Fig. 29,BM(NH)no. E29887,
specimen with a well-preserved oral area, x6; see Fig. 30. All specimens whitened with ammonium
chloride sublimate.

132

A. B. SMITH

CARBONIFEROUS EDRIOASTEROIDEA 133

Fig. 41 Reconstructed cross section through an
interambulacral plate of Stalticodiscus milleri
(Sharman & Newton). A, outer dense, perfo-
A rate stereom layer with granular surface. B,

thin middle layer of laminar stereom. C, thick
inner layer of coarse labyrinthic stereom.

As in all isorophids, the peripheral rim is much more prominent in juveniles than it is in adults.
The central disc is only 55% of the diameter of the theca at about 1 mm diameter, approximately
60% at 2-3 mm diameter, 65-70% at 4-7 mm diameter and reaches a little less than 80% by
12-15 mm diameter. At 1 mm diameter the peripheral rim consists of just two rows of plates but
the number of rows increases progressively to a total of five by about 7-8 mm diameter (Fig. 36).
Bell (1976£) reported that new peripheral rim plate cycles were added by insertion between the
first two cycles of plates. I could find no evidence that this took place in S. milleri, where new
plate cycles appear to have been added at the outer edge. The inner alternating cycle of large
peripheral rim plates do not form an obvious palisade until about 2mm diameter. New plates
continue to be added to the inner cycle of plates until approximately 6 mm diameter, after which
time growth continues by plate enlargement only.

Within the disc, plating is poorly differentiated to start with. At 1 mm diameter the disc is
dominated by an elongate mound (Fig. 42) but at this size it is impossible to distinguish
individual plates. Presumably at this stage ambulacral cover plates exist but are extremely thin
and poorly preserved. By 2mm diameter the ambulacra have started to differentiate and there
are five small but distinct points to the elongate oral area. Cover plate arrangement can be made
out and there are four slightly larger cover plates in the oral area; two anterior ones situated at
the junction of ambulacra II and III, and ambulacra III and IV, and two posterior ones between
ambulacra I and V. The large posterior right oral cover plate eventually becomes the hydropore
plate. At 2 mm diameter the rudimentary ambulacra consist of just two or three cover plates per
column, but by 4 mm diameter the ambulacra are obviously developed and consist of some six to
eight cover plates per column. Ambulacra are still more or less straight. The ambulacra start to
curve gently by about 5 -5 mm diameter (by which time there are 12 to 14 cover plates per
column) and, for the first time, small occluded cover plates become apparent. The cyclical
arrangement of cover plates is fully developed by 7 or 9 mm diameter. At 7 mm diameter the
anterior and two lateral ambulacra are more or less prominently curved, but the posterior two
ambulacra are only weakly curved and their tips only slightly convergent. By 13 mm diameter
the posterior two ambulacra have grown to more or less meet behind the periproct and in larger
specimens the tips of the ambulacra may overlap very slightly. New ambulacral cover plates are
added distally at the tips of growing ambulacra.

Figs 31-40 Scanning electron micrographs of Stalticodiscus milleri (Sharman & Newton) from
Penruddock. All BM(NH) numbers. Fig. 31, E29907, oral area of a 13mm diameter individual
showing a clear hydropore bulge. Fig. 32, E29908, individual 10mm in diameter showing a rather
different arrangement of oral cover plates. Fig. 33, E29915, the distal part of ambulacrum II in an
8 mm diameter individual, showing the early appearance of secondary (occluded) cover plates near
the distal tip. Fig. 34, E29908, the adoral part of ambulacrum V in a 10mm diameter specimen
showing the cyclical arrangement of cover plates and the zigzag perradial suture. Fig. 35, E29897,
periproct in a 5-5 mm diameter individual with both primary and secondary cycles of periproctal
plates; see Fig. 51. Fig. 36, E29895, the fully formed peripheral rim of plates in an individual
6-5 mm in diameter. Fig. 37, E29921, a large interambulacral plate seen from the inside with an
extensive zone of labyrinthic stereom and a more marginal laminar stereom. Figs 38^0, E29918,
ambulacral flooring plate: Fig. 38, external (distal edge to top of photomicrograph); Fig. 39,
lateral (distal edge to left of photomicrograph); Fig. 40, front view of distal face. Scale bar = 0-5
mm.

134

A. B. SMITH

CARBONIFEROUS EDRIOASTEROIDEA

135

1 mm

Figs 49-51 Camera-lucida drawings of juvenile Stalticodiscus milleri (Sharman & Newton). All
BM(NH) numbers. Fig. 49, E29893, see Fig. 48; Fig. 50, E29885; Fig. 51, E29897, shown inverted in
Fig. 35, p. 132.

Figs 42-48 Scanning electron micrographs of juvenile Stalticodiscus milleri (Sharman & Newton),
all to the same scale. All BM(NH) numbers. Fig. 42, E29899; Fig. 43, E29898; Fig. 44, E29911 ; Fig.
45, E29917; Fig. 46, E29901; Fig. 47, E29894; Fig. 48, E29893, see Fig. 49.

136 A. B. SMITH

Interambulacral areas are minimal at 1 mm diameter and no interambulacral plates can be
made out at this size. Interambulacral plates first become apparent at about 2 mm diameter. As
growth proceeds, interambulacral areas become progressively larger relative to the disc, and new
plates are added immediately inside the peripheral rim as well as adjacent to the ambulacra and
periproct. By about 8mm diameter the full complement of ventral interambulacral plates are
present and it is at about this size that the pedunculate zone starts to develop. At 7 mm diameter
the theca is still low and domal in profile but around 8 to 10mm diameter a zone of imbricate
plating starts to form between the edge of the ventral surface and the peripheral rim. This
pedunculate zone is fully developed by 12 to 15 mm diameter, but at what stage it starts to
develop varies. For example, it is poorly developed in the holotype, which is 9 mm in diameter,
but most Penruddock specimens have a fairly well developed zone of imbricate plates by this
diameter. Presumably environmental factors influenced the size at which the pedunculate zone
formed.

The periproct does not appear until about 2mm diameter, at which stage it is a circular area
composed of some four triangular plates set flush with the interambulacral plates. By 3 mm
diameter the periproct has become elevated and forms a conical structure, and by 4mm
diameter the full complement of primary cycle periproctal plates is present. At about 6mm
diameter the first of the secondary cycle of periproctal plates appears.

Plating arrangement over the oral area is in its final form by about 6mm diameter. The
hydropore cover plate appears very early and is present by 2mm diameter. It continues to
enlarge throughout the early growth stages. The hydropore bulge to the posterior right of the
oral area is absent at 4 mm diameter, starts to form at about 5 to 6 mm diameter and is prominent
by 7mm diameter. This probably coincides with the onset of gonadal development, as the
so-called hydropore passageway is probably a combined hydropore/gonopore. The appearance
of the hydropore bulge would then be the best morphological change on which to distinguish
juveniles from sexually mature adults.

The ontogenetic changes that take place in S. milleri are in close agreement with the
development of other isorophids reported by Bell (19766). There are, however, two minor
differences. Firstly, I could find no evidence that plates of the peripheral rim were added by
insertion: here addition took place at the outer edge. Secondly, the cover plates of the oral area
are hardly differentiated from ambulacral cover plates even during the early ontogenetic stages.
This quite obviously reflects the fact that in adult agelacrinitids there are no enlarged oral cover
plates such as are found in other families.

DISCUSSION. The holotype of ' ' Lepidodiscus" milleri is not well preserved and comes from a
slightly higher horizon in the Early Asbian than does the Penruddock material. The holotype
differs slightly from the Penruddock material in that it has a more juvenile appearance than
would be expected for its size. At 9mm diameter it has the appearance of a 7mm diameter
Penruddock specimen, lacking the pedunculate zone development and strong ambulacral
curvature expected by this size. As the only difference between the holotype and the
Penruddock specimens is one of developmental rate, all are placed within the same species. The
rate of ontogenetic development is likely to be influenced by environmental factors such as the
hydrodynamic regime or substratum availability and intraspecific competition.

The Penruddock population of S. milleri was killed by a sudden influx of sediment and the
animals are preserved in various stages of contraction (compare Figs 13 and 27). Specimens that
are not fully contracted are generally preserved lying over to one side (see, for example, Fig.
12), flattened by the influx of sediment that killed them. The direction in which they have fallen
presumably coincides with the prevailing current direction at death. The fully contracted
specimens have collapsed further following compaction and probably had a low domal profile in
life. Presumably contraction and expansion were effected by circumferential and longitudinal
muscle layers immediately beneath the thecal plating. Adults were not permanently cemented
to the substratum but could swivel, much like sea anenomes. Orientation is more or less random
on boulders settled by juveniles. However, once the pedunculate zone had begun to form,
orientation obviously became more important. For example, on BM(NH) nos E29878-82 (one

CARBONIFEROUS EDRIOASTEROIDEA 137

Fig. 52 Rose diagram showing the orientation

(anterior of the anterior-posterior axis) for 55

adults encrusting one limestone concretion.

Current The current arrow shows the direction in which

____— the sediment that smothered the colony was

brought in.

boulder), all adults are orientated in the same direction with their anterior ambulacra facing
more or less at right angles to the direction of sediment input that killed them (Fig. 12). A strong
orientation is found on many other boulders (Fig. 52), though not on all. The direction of
sediment input need not correspond to the direction of the prevailing current. However, Foerste
(1914) found precisely the same orientation in populations of Carney ella pilea. Here, as in S.
milleri, it is ambulacra IV and V that face towards the direction of sediment influx. The fact that
in both species the anteior-posterior axis was orientated at right angles to the direction of the
current that smoothered them suggests that this was indeed the orientation that they adopted in
currents. It also suggests that these isorophids were living in a tidal regime with bimodal
currents, since in unidirectional flow one would expect the anterior to face into the current so
that the periproct was positioned downstream of the mouth.

Acknowledgements

I am greatly indebted to Dr W. H. C. Ramsbottom (IGS) for all his assistance in pin-pointing the
position of the older localities and unravelling their stratigraphic horizons. I also wish to thank
Dr P. D. Taylor (BM(NH)), who first brought the Penruddock locality to my attention and
assisted in many ways. Help in dating the Penruddock horizon was received from Dr C. H. C.
Brunton (BM(NH)), who examined the brachiopods, and Dr A. R. E. Strank (IGS), who
identified the foraminifera, and for this I am extremely grateful. Mr K. C. Veltkamp (University
of Liverpool) assisted with scanning electron microscopy. This research was supported by
N.E.R.C. grant GR3/3473.

References

Anderson, F. W. 1939. Lepldodiscus fistulosus sp. nov. from Lower Carboniferous rocks, Northumber-
land. Bull. geol. Surv. Gt Br., London, 1: 67-81, pi. 5.
Bassler, R. S. 1936. New species of American Edrioasteroidea. Smithson. misc. Collns, Washington, 95

(6). 33 pp., 7 pis.
Bell, B. M. 1976a. A study of North American Edrioasteroidea. Mem. N. Y. St. Mus. Sci. Serv. 21. 446 pp.,

63 pis.
1976ft. Phylogenetic implications of ontogenetic development in the class Edrioasteroidea

(Echinodermata). J. Paleont, Menasha, 50: 1001-1019.
Foerste, A. F. 1914. Notes on Agelacrinitidae and Lepadocystinae, with descriptions of Thresherodiscus

and Brockocystis. Bull, sclent. Labs Denison Univ., Granville, Ohio, 17 (14): 399-487, 6 pis.
George, T. N., Johnson, G. A. L., Mitchell, M., Prentice, J. E., Ramsbottom, W. H. C., Sevastopulo, G. D.

& Wilson, R. B. 1976. A correlation of Dinantian rocks in the British Isles. Spec. Rep. geol. Soc. Lond.

7. 87 pp.
Kesling, R. V. 1960. Hydroporesin edrioasteroids. Contr. Mus. Paleont. Univ. Mich., Ann Arbor, 15 (8):

139-192, 13 pis.

138 A. B. SMITH

Lebour, G. A. 1876. Note sur deux fossiles du Calcaire Carbonifere du Northumberland. Annls Soc. geol.

Belg., Liege, 3: 21-24.
Meek, F. B. & Worthen, A. H. 1868. Remarks on some types of Carboniferous Crinoidea with descriptions

of new genera and species of the same, and one echinoid. Proc. Acad. not. Sci. Philad., 5: 335-359.
Miller, H. 1887. The geology of the country around Otterburn and Elsdon. Mem. geol. Surv. U.K.,

London (Quarter-sheet 108 SE = n.s. Sheet 8). 147 pp.
Ramsbottom, W. H. C. 1970. Lower Carboniferous palaeontology. /nDay, J. B. W. etal., Geology of the

country around Bewcastle: 166-207. Mem. geol. Surv. U.K., London (n.s. Sheet 12).
Regnell, G. 1950. Agelacrinites ephraemovianus (Bogolubov) and Lepidodiscus fistulosus Anderson

(Edrioasteroidea). K. fysiogr. Sallsk. Lund Forh. 20 (20): 218-237.

- 1966. Edrioasteroids. In Moore. R. C. (ed.), Treatise on Invetebrate Paleontology U (Echinodermata

3): U136-U173. Lawrence, Kansas.
Sharman, G. & Newton, E. T. 1892. On a new form of Agelacrinites (Lepidodiscus milleri n. sp.) from the

Lower Carboniferous Limestone of Cumberland. Q. Jl geol. Soc. Lond., 48: 150-152, pi. 2.
Sladen, W. P. 1879. On Lepidodiscus lebouri, a new species of Agelacrinitidae from the Carboniferous

Series of Northumberland. Q. Jl geol. Soc. Lond., 35: 744-751, pi. 37.

A survey of Recent and fossil Cicadas
(Insecta, Hemiptera-Homoptera) in Britain

P. E. S. Whalley

Department of Entomology, British Museum (Natural History), Cromwell Road, London
SW7 5BD

Synopsis

The current status and origin of the single extant British cicada, Cicadetta montana Scopoli, are discussed.
An account of cicadas from the Tertiary and Mesozoic of Britain is given with an analysis of the taxonomy
and morphology of the Mesozoic species. The family Cicadidae is recorded from the Mesozoic for the first
time.

Introduction

This survey was prompted by an examination of specimens of cicadas amongst the British Upper
Triassic collections at the Institute of Geological Sciences and the British Museum (Natural
History) .

Rohdendorf (1962) divided the 'infraorder Cicadomorpha' into two superfamilies,
Palaeontinidea and Cicadidea. In the latter he placed three families, Prosbolidae, Cicadidae and
Tettigarctidae. The Prosbolidae are known only from the Carboniferous to Triassic (Evans
1956: 196-206). The Cicadidae, which include the only British species, Cicadetta montana
Scopoli, are widespread in the warmer parts of the world (Evans 1963) but extend into the
temperate zone where they are represented by fewer species. As fossils the Cicadidae are known
from the Tertiary, Zherikhin (1980) suggesting the Upper Cretaceous as the earliest record. The
Tettigarctidae are common in the fossil record since the early Mesozoic and occur in the
Tertiary. In contrast with the widespread distribution and numerous species of Recent
Cicadidae, the only two living species of Tettigarctidae are restricted to Australia.

Recent cicadas in Britain

Cicadetta montana in Britain is restricted to a very small area in Hampshire (Grant 1972).
Morley (1941) gave an account of the history and distribution of C. montana, stating that it has
been 'in our midst since Britain's severance from the continent in Pliocene times'. He also
pointed out that with its limited powers of flight it could not cross the Channel, although it is
found in northern France (Villiers 1977). Grant (1972) supported the view that C. montana is a
relict species, stating that its history in Britain 'is directly traceable to the old land continuity
with Europe and ancient vegetative spread'. It is evident from recent work (Grant 1970, 1972;
Morley 1941) that C. montana has never been a common insect and is mostly restricted to one
southern county. It is not an easy insect to locate in the woods, in spite of its call, and it has been
suggested that this call is inaudible to some people (Morley 1941: 54).

C. montana was first discovered in Britain in 1812 but Curtis (1832) commented that he and
another well-known entomologist (Dale) searched for it without success for over 20 years before
they finally rediscovered it. Buckton (1890) in his monograph on British Homoptera also
commented on its local and very patchy occurrence. Both Grant and Morley dismissed as
unlikely natural or accidental introduction of the cicada to Britain because of its relatively weak
powers of flight, ephemeral adult life and subterranean early stages. However, the eggs of the
cicada, which are inserted into the stems of woody plants, might well have been brought
(accidentally) into Britain. A modern parallel can be drawn from the homopteran

Bull. Br. Mus. nat. Hist. (Geol.) 37 (3): 139-147 139 Issued 24 November 1983

140

P. E. S. WHALLEY

Graphocephala fennahi Young which was first recorded in Britain in 1936 as an introduction
(under the name C. coccinea Forster). This species lays its eggs in the sepals of rhododendron
(Morcos 1953).

Morley's (1941) suggestion of the Pliocene for the origin of the British cicadas is unacceptable
since it implies that this warm-loving species had survived several glacial periods in Britain.
Grant's (1972) suggestion of an origin during the Boreal age (Flandrian, c. 7000 years B.P.) is
more plausible, but I believe that if cicadas were present prior to the early 1800s then some
folk-lore or published account of this large and relatively noisy insect would have appeared.
However, if the 'little ice-age1 from the 15th-18th century had reduced the population to a very
low level it could well have been overlooked. Thus while it is generally accepted that the cicada
in Britain is a relict species the possibility of its being an introduction should not be ruled out.

Fossil cicadas in Britain

The first fossil cicadas were found in Britain nearly 150 years ago but the inadequate descriptions
and figures that were published led Handlirsch (1906-08) to consider that they were incorrectly
identified. Fossil cicadas are known from the Eocene and Upper Triassic in Britain.

Eocene

The specimen of cicada described from the British Eocene is of considerable palaeogeographic
interest and consists of one incomplete hindwing from the Isle of Mull, Scotland (Zeuner 1941:
88; 1944). It was described as Eotettigarcta scotica by Zeuner (1944: 110) (Fig. 1), who compared
it with Recent Tettigarcta (Tettigarctidae) from Australia. While not congeneric, he regarded it
as 'very closely related'. (Living Tettigarctidae are restricted to Australia, where the species are
associated with an alpine environment, although fossil representatives of the family are much
more widespread; Woodward etal. 1970). I have re-examined the holotype (In. 38883) and have
no reason to doubt Zeuner's classification of the fossil on the evidence available. Species of
Tettigarctidae have been described from the Triassic and Jurassic of Asia.

Triassic

Several specimens from south-west England were described and figured by Brodie (1845) but
only one species, C. murchisoni, was named. I have re-examined Brodie's specimens and have
additional material from the Upper Triassic.

Fig. 1 Eotettigarcta scotica Zeuner, holotype. Isle of Mull. In. 38883, BM(NH).

BRITISH CICADAS

141

The generic classification of Mesozoic cicadas is based entirely on forewing venation
(Rohdendorf 1962), making comparison virtually impossible with the incompletely preserved
wings of British fossils. All previously described cicadas from the Mesozoic have been placed in
the family Tettigarctidae but the character used to separate extant species of this family from the
Cicadidae are rarely well-preserved. Woodward et al. (1970) separated Recent species of the
two families on the presence or absence of tymbals on the dorsal side. However Dr J. P. Duffels
(Amsterdam), after examining the specimens, pointed out that the large and separate pro- and
mesonotum clearly shown in the British fossils are characteristic of the Cicadidae. The small
amount of wing venation preserved also indicates this family rather than the Tettigarctidae. The
Cicadidae have not previously been recorded from the Mesozoic.

The most distinctive feature of the British Triassic specimens is the extremely long rostrum.
From an examination of Recent cicadas in the British Museum (Natural History) collection it is
apparent that the rostrum of the fossils is proportionally longer than the rostrum of most Recent
species. However, in the Recent genus Platypleura Amyot & Serville there is one species,
P. adouma Distant (Fig. 4), where the rostrum is much longer than in others of the same genus.
Even so, this species does not have a rostrum quite as long as in the fossils. In view of the
variation in length of rostrum between species in Recent genera, the description of a new fossil
genus based only on this character seems unwarranted.

Fig. 2 Liassocicada ignotatus Brodie, holotype, cf . Gloucestershire (Forthampton). In. 3539,

BM(NH)
Fig. 3 L. ignotatus. Worcestershire (Strensham). In. 10449, BM(NH). suggested nymphal stage. 1 -

possible emergence of imago. 2 - parallel, narrow sclerotized tergites. 3-nymphal wing-pad.

142

P. E. S. WHALLEY

Bode (1953) based the genus Liassocicada on the fragment of a forewing from the Upper Lias
of Germany, placing it in the Cicadidae. Rohdendorf (1962) more correctly considered it as
Cicadoidea insertae sedis since none of the characters used to define the family are preserved in
the type specimen of Liassocicada. However, I propose to redefine this genus and to place the
British species in it provisionally.

Systematic description

Family CICADIDAE Leach, 1815

Genus LIASSOCICADA Bode, 1953

TYPE SPECIES. Liassocicada antecedens Bode, by monotypy. Jurassic.

Because the definition of this genus is based on a fragment of the forewing it is re-defined here
to include L. ignotatus Brodie (below).

DIAGNOSIS. Cicadas with elongate rostrum reaching well down the abdomen.
RANGE. Triassic-Jurassic.

Liassocicada ignotatus (Brodie) comb. n.
Figs 2-3, 5-11

1845 Asilus (?) ignotatus Brodie: 102 [described in the Order Diptera].
1845 Cicada murchisoni Brodie: 101; syn. n.
1873 Cicada larva, Brodie: 25.
1873 Cicada pupa, Brodie: 25.

Fig. 4 Platypleura adouma Distant, $. Recent Africa. BM(NH).
Fig. 5 L. ignotatus. Locality unknown. In. 59079, BM(NH).

BRITISH CICADAS

143

1906 Asilus (?) ignotatus Brodie; Handlirsch: 503.
1906 Cicada murchisoni Brodie; Handlirsch: 504.
1906 Cicada larva, Handlirsch: 511.
1906 Cicada pupa, Handlirsch: 511.

DIAGNOSIS. As genus.

DESCRIPTION. Head with prominent, ridged frons. Eyes large, oval. Rostrum very long,
reaching to base of ovipositor in female. Pro- and mesonotum large and separate. Fore tarsal
segments rounded, several long spurs on hind tibia. Tympanal organ possibly represented by
sclerotization at ventral side of first abdominal segment. Ovipositor short, curved and strongly
sclerotized, with sclerotized (?) spermatheca preserved in some specimens (Fig. 10). Specimen
In. 3539 (Fig. 2) is probably a male, having a rather truncate tip to the abdomen and more
slender body than the females. Traces of wing venation are also present on this specimen.
Specimen In. 10449 (Fig. 3) is probably a nymph, showing the split along the dorsal side of the
thorax with (?) partially emerged adult (Fig. 3, arrow 1). [Ocelli, most of wings, tymbal organs
not preserved].

HOLOTYPE. In. 3539. Forthampton, Gloucestershire; Brodie coll. in British Museum (Natural
History). Fig. 2.

OTHER MATERIAL. All except the last in British Museum (Natural History) collections.

In. 3537. Hasfield, Gloucestershire. Holotype of C. murchisoni.

In. 10449. Strensham, Worcestershire. Brodie coll. Fig. 3.

In. 10440. Strensham, Worcestershire. 'Cicada larva', Brodie coll.

Figs 6, 7 L. ignotatus. $ [Worcestershire], 'Lower Lias'. IGS GSb 273 (part and counterpart).
Institute of Geological Sciences, Geol. Soc. coll. See also Fig. 11.

144

P. E. S. WHALLEY

Figs 8, 9 L. ignotatus. Worcestershire (Stren-
sham). In. 11244, BM(NH). Fig. 9, head,
stylets and ridged frons, enlarged. See also
Fig. 10.

In. 11113. Strensham, Worcestershire. Brodie coll.

In. 11240. Strensham, Worcestershire. 'Cicada', Brodie coll.

In. 11244. Strensham, Worcestershire. 'Cicada pupa', Brodie coll. Figs 8-10.

In. 59079. 'Cicada pupa', Brodie coll.; locality unknown but similar in preservation and

appearance to the Strensham material. Fig. 5.

IGS GSM GSb 273 [Worcestershire] 'Lower Lias' (no other details); part and counterpart. In

Institute of Geological Sciences. Figs 6, 7, 11.

AGE AND DISTRIBUTION. Upper Triassic, Rhaetian Stage; Penarth Group, Lilstock Formation,
Gotham Member, Pseudomonotis Bed (formerly an 'Insect Limestone'); north-west
Gloucestershire and Worcestershire. 'Insect Limestones' have been described by several
authors from exposures in Somerset, Avon, Gloucestershire, Worcestershire and Warwick-
shire, and they are not all at the same horizon. The Insect Limestone in the Tewkesbury and
Upton-upon-Severn area from which the cicadas described here were obtained is better called
the Pseudomonotis Bed, in order to distinguish it from similar beds in other areas which may
belong to different horizons. Brodie (1845: 100-102), and more recently Richardson (1948:
143-144; 1966: 153), stated that the bed belonged to the Lower Lias, but most other authorities
agree that it was one of the top layers of the 'Rhaetic Beds' (Wright 1878: 14; Richardson 1903:
127-174; 1904: 22, 207-210; Arkell 1933: 107). The confused stratigraphical nomenclature and
doubts about its Triassic or Jurassic age have been superseded by the Geological Society's
detailed correlations of the British Triassic (Warrington et al. 1980) and Jurassic Systems (Cope
et al. 1981). In the latter report the base of the Jurassic is drawn at the horizon of the first
appearance of Psiloceras planorbis , and all lower beds (including the lowest part of the 'Liassic
Series') belong to the Triassic System. So the Pseudomonotis Bed is now firmly established as of
Triassic, Rhaetian Stage, age.

BRITISH CICADAS

145

DIMENSIONS. Body length 20-25 mm, males smaller than females.

DISCUSSION. Although the name murchisoni has page priority over ignotatus, and was
recognized as hemipterous by Brodie while ignotatus was considered dipterous, the holotype of
murchisoni (In. 3537) is not well preserved. Unless the specimen was formerly more complete it
is difficult to see why it was considered a cicada. Nothing on it actually rules it out as a cicada but
equally only the incompletely preserved forelegs suggest that it might be one: murchisoni is here
considered a nomen dubium. The specimen described as A. ignotatus (Fig. 2) by Brodie has the
long rostrum characteristic of the other specimens (Figs 3,6,7) and is chosen in preference to
murchisoni.

Dr J. P. Duffels has suggested that In. 10449 (Fig. 3) is a nymphal stage, possibly with the
emerging adult (arrow 1). There are two parallel sclerites (arrow 2) which are typical of nymphal

I ',

Fig. 10 L. ignotatus. Worcestershire (Strensham). In. 11244, BM(NH). Ovipositor valves, enlarged.

See also Figs 8, 9.
Fig. 11 L. Ignotatus. 9 [Worcestershire], 'Lower Lias'. IGS GSb 273, ovipositor valves, enlarged.

Institute of Geological Sciences, Geol. Soc. coll. See also Figs 6, 7.

146 P. E. S. WHALLEY

cicadas. The separation of the three thoracic segments is also more clearly shown, suggesting a
nymphal instar, and possible wing pads (arrow 3) are indicated. All the other specimens have
traces of wings or ovipositors, indicating that they were adults. Cicada nymphs are subterranean
with the last instar coming to the surface to moult to the adult stage.

The ovipositor and associated structures are well preserved in most specimens, suggesting a
typical strong, slightly curved cicada-type capable of inserting eggs into woody plant tissue. In
two specimens (Figs 8, 10) there are associated structures at the base of the ovipositors which
may represent the spermatheca but could even have been eggs. Probably the most remarkable
structure of the British Mesozoic cicadas is the long rostrum which was at least 14 mm long and in
the female reached the base of the ovipositors. The structures actually preserved are the stylets,
the elongate maxillae and mandibles with only parts of the surrounding rostrum preserved in a
few places. There is no evidence that the stylets were coiled up inside the head capsule, and
comparing it with the Recent species (Fig. 4) where the rostrum is also long, it was probably held
between the legs. With the humped thorax and typical adpressed head, the method by which the
stylets were inserted into the plant tissue is interesting. Aphids with long stylets tend to feed on
fissured bark of tree trunks or large roots (Dr V. F. Eastop, personal communication), but it is
difficult to see how the Triassic cicadas could insert the long stylets into a plant using the
technique of Recent, short-rostrum cicadas. It is possible that its length was important in
probing down packed leaf-buds or scales to get at the tissue these were protecting, for example
to get at the embryo deep between the scales of a Pinus-type cone. It is also possible that the
stylets were inserted into plant tissue, but in the absence of evidence from the feeding behaviour
of Recent species no further light can be thrown on this remarkable structure.

Acknowledgements

I am indebted to Dr H. Ivimy-Cook, Institute of Geological Sciences, for the loan of some of the
specimens and to Dr M. K. Howarth, BM(NH), for advice on the stratigraphy. Dr J. P. Duffels
of Amsterdam, Dr W. J. Knight, BM(NH), and Mr E. A. Jarzembowski, BM(NH), examined
the specimens and made useful comments, and Dr V. F. Eastop, BM(NH), read the manuscript;
to all I offer my thanks.

References

Arkell, W. J. 1933. The Jurassic System in Great Britain. 684 pp. Oxford.

Bode, A. 1953. Die Insektenfauna des ostniedersachsischen Oberen Lias. Palaeontographica, Stuttgart,

(A) 103 (1-4): 1-375.
Brodie, P. B. 1845. A history of the fossil insects in the Secondary rocks of England. 130 pp. , 11 pis. London.

- 1873. The distribution and correlation of fossil insects and supposed occurrence of Lepidoptera and
Arachnidae in British and Foreign strata, chiefly in secondary rocks. Rep. Warwicksh. nat. Hist,
archaeol. Soc., 37: 12-28.

Buckton, G. B. 1890. Monograph of the British Cicadae or Tettigidae, 1. 134 pp., 3 + 38 pis col., London.
Cope, J. C. W. etal. 1981. A correlation of Jurassic rocks in the British Isles. Part I. Spec. Rep. geol. Soc.

Lond. 14. 73 pp.

Curtis, J. 1832. British Entomology (&c.), 9: 386-433. London.
Evans, J. W. 1956. Palaeozoic and Mesozoic Hemiptera (Insecta). Aust. J. Zool. , Melbourne, 4: 165-258.

- 1963. The phylogeny of the Homoptera. A. Rev. Eht., Palo Alto, 8: 77-94.
Grant, P. J. 1970. Search for our insect singers. Countryside, London, (NS) 21: 301-307.

- 1972. Conserving Britain's cicadas. Countryside, London, (NS) 22: 8-11.

Handlirsch, A. 1906-08. Diefossilen Insekten und die Phylogenie der rezenten Formen. 1430 pp. Leipzig.
Morcos, G. 1953. The biology of some Hemiptera-Homoptera (Auchenorryncha). Bull. Soc. Fouad I.

Ent., Cairo, 34: 405-409.
Morley, C. 1941. The history of Cicadetta montana in Britain, 1812-1940. Entomologist's mon. Mag.,

London, 77: 41-56.
Richardson, L. 1903. The Rhaetic rocks of North-west Gloucestershire. Proc. Cotteswold Nat. Fid Club,

London, 14 (2): 127-174.

BRITISH CICADAS 147

— 1904. A handbook to the Geology of Cheltenham and Neighbourhood. 303 pp. Cheltenham.

— 1948. The upper limit of the Rhaetic series and the relationship of the Rhaetic and Liassic series.
Proc. Cotteswold Nat. Fid Club, London, 29: 143-144.

1966. The upper limit of the Rhaetic series and the relationship of the Rhaetic and Liassic series: a

correction. Proc. Cotteswold Nat. Fid Club, London, 34 (3): 153.
Rohdendorf, B. B. 1962. [Arthropoda. Tracheata and Chelicerata.] Osnovy Paleontologii, Moscow, 9.

560 pp. [In Russian].

Villiers, A. 1977. Atlas des Hemipteres (nouv. edn). 301 pp., 24 pis col. Paris.
Warrington, G. etal. 1980. A correlation of Triassic Rocks in the British Isles. Spec. Rep. geol. Soc. Lond.

13. 78 pp.
Woodward, T. E., Evans, J. & Eastop, V. F. 1970. Part 26, Hemiptera. In: Insects of Australia: 387-457.

Canberra, C.S.I.R.O.
Wright, T. 1878-86. Monograph on the Lias Ammonites of the British Islands. 503 pp., 88 pis.

Palaeontogr. Soc. (Monogr.), London.
Zeuner, F. E. 1941. The Eocene insects of the Ardtun Beds, Isle of Mull, Scotland. Ann. Mag. nat. Hist.,

London, (11) 7: 82-100.
1944. Notes on Eocene Homoptera from the Isle of Mull, Scotland. Ann. Mag. nat. Hist., London,

(11) 11: 110-117.
Zherikhin, V. V. 1980. Class Insecta. In Menner, V. V. (ed.), [Development and change of Invertebrates

on the border between the Mesozoic and Caenozoic . . .]: 40-97. Moscow, Paleontologicheskii

Institut, Akademiya Nauk SSSR. [In Russian].

The Cephalaspids from the Dittonian section at
Cwm Mill, near Abergavenny, Gwent

Errol I. White

Department of Geology, University of Reading, Berkshire.

Harry A. Toombs

lately Department of Palaeontology, British Museum (Natural History), Cromwell Road,
London SW7 5BD

Synopsis

An account is given of some fifty articulated specimens of Cephalaspis discovered in the mid 1930s by W. N.
Croft in a stream section near Abergavenny. All are small animals and are mostly referable to C.
cradleyensis Stensio, but three new species C. cwmmillensis, C. abergavenniensis and C. (Cwmaspis)
billcrofti subgen. et sp. nov. are also represented.
The development and means of distribution of cephalaspids are discussed.

Introduction

An interesting discovery was made by W. N. Croft in the Dittonian (Lower Devonian) of the
Anglo-Welsh area some years ago in a stream section in a small tributary of the River Gavenny
at Cwm Mill near Mardy, about one mile (1-6 km) north-east of Abergavenny (National Grid
ref. SO 311156). Croft in his original field notes records the locality as lying '3/4 mile NNE of
Asylum, Abergavenny'.1 Like virtually all the sections and pits in the region that formerly
yielded good cephalaspid material the Cwm Mill locality has now been worked out. Here a thin
bed of grey-green siltstone, apparently not more than 2 or 3 inches (c. 65 mm) thick, yielded a
quantity of articulated specimens of small Cephalaspis. Some fifty specimens were collected and
doubtless all, or nearly all, must have been complete when first buried, but they were massed
together and flattened, lying top-side up, on their backs, even occasionally on their sides, all
close together and very often on top of one another. This, combined with the softness of both
matrix and specimens, made collecting very difficult and the results were often rather
disappointing.

Nevertheless, the collection is of much interest, for articulated specimens from Dittonian
strata of the Anglo- Welsh region are extremely rare - Stensio (1932) recorded only three
specimens with part of the body attached - and at undescribed localities only Wayne Herbert,
10V2 miles (17km) away, has produced articulated cephalaspids in a much more diversified
fauna (Miles 1973), while a single complete specimen was found in a nodule in a stone-breaker's
pile just below Castle Mattock, some seven miles (11-25 km) north of Cwm Mill.

The Cwm Mill section has already been noticed in literature (White 1950: 56; Allen & Tarlo
1963: 145), and although the fauna has never before been described, the cyclothem of which it
forms part has been described in some detail and illustrated by Allen (1964: 184-6, fig. 11).
Unfortunately, the precise relationship of the Cwm Mill section to the levels of the principal
quarries that have yielded the bulk of useful material during the last half century is not known.
While those quarries lie in a stable block dipping gently to the south-east and can be related to
the 'Psammosteus Limestone' (P.L.), Cwm Mill is in a much less stable area where there is no
sign of the P.L. , but Allen in his description of the cyclothem states that it 'lies about the middle
of the Dittonian stage'.

'The field notebooks of W. N. Croft (1915-1953) are housed in the Department of Palaeontology, British Museum
(Natural History).

Bull. Br. Mus. nat. Hist. (Geol.) 37 (3): 149-171 149 Issued 24 November 1983

150 E. I. WHITE & H. A. TOOMBS

It may be convenient at this point to list the principal quarries of the area, with the attendant
form ofPteraspis (White 1935; 1950: 58 footnote): all are now out of use and largely overgrown:

WERN (or GWYN) GENNI. 650 feet (200 m) above P.L. 6 miles (9-6 km) NW of Wayne Herbert.

With Pteraspis stensioi.
POOL QUARRY. 350 feet (110m) above P.L. 3l/2 miles (5-6 km) SSE of Wayne Herbert. With

typical P. crouchi and P. rostrata var. waynensis.
CASTLE MATTOCK (CLODOCK). 240 feet (75 m) above P.L. 3V2 miles (5-6 km) south of Wayne

Herbert. With P. jackana and P. crouchi \ar.mattockensis.
WAYNE HERBERT. 220 feet (67 m) above P.L. lll/2 miles (18-5 km) NNE of Abergavenny. With

P. rostrata var. waynensis, P. rostrata var. virgoi and P. ? jackana above siltstone lenticle and

P. rostrata var. toombsi in it.
CWM MILL. 'About the middle of the Dittonian stage'. 11A miles (2 km) NE of Abergavenny and

10V2 miles (17km) SSW of Wayne Herbert. With P. ? crouchi.

The specimens used in the compilation of this paper belong to the collections of the British
Museum (Natural History), London and are referred to by register number with or without the
prefix P.

The Cwm Mill Fauna

This was a very restricted exposure for the fossil vertebrates; apart from scattered fragments,
they occur 'from a distinct horizon' (Allen 1964: 185, fig. 11) over a distance of not more than a
foot or two (less than 1 m) . Curiously enough Croft in his field notes made very little reference to
the discovery of this remarkable assemblage of ostracoderms. Under the heading 'Cwm Mill -
Cephalaspis Loc. (244)' he gives a section 26 inches (0-66 m) in height in which is shown a bed
2-3 inches (c. 65 mm) thick simply labelled 'Pt. (rare) above. Ceph. below': yet from this small
exposure an unprecedented number of specimens of Cephalaspis, originally complete, were
extracted from a single layer, a grey-green siltstone, which seems to have been lenticular.

All the specimens except three seem to belong to a single species, Cephalaspis cradleyensis
Stensio, while each of the remainder belongs to a different undescribed species. Outside the
siltstone lenticle a darker, harder bed yielded a few fragments of still other species of the genus,
in addition to pieces of cephalic discs of C. cradleyensis.

The fauna of the siltstone lenticle

Family CEPHALASPIDAE Agassiz, 1843

Genus CEPHALASPIS Agassiz, 1835

Cephalaspis cradleyensis Stensio
Figs 1-11

1932 Cephalaspis cradleyensis Stensio: 130, text-fig. 44; pi. 15, fig. 6.
1952 Cephalaspis cradleyensis Stensio; Wangsjo: 255, text-fig. 24; pi. 2.

DIAGNOSIS (emended). A small species of Cephalaspis with total length 100-120 mm. Cephalic
shield about 40mm long and approximately 40mm in maximum breadth measured across the
tips of the cornua. Lateral margins of shield gently convex, narrowing rather rapidly towards a
rounded front without rostral angle. Cornua directed slightly laterally, short, their length
scarcely exceeding one quarter of distance between tips and median point of rostral margin;
inner margins without denticles. Pectoral sinus rather narrow and shallow; interzonal part broad
and short with low but well-defined median crest projecting a little to form a very short posterior
angle. Orbital openings somewhat oval and situated rather nearer front than back of shield.
Dorsal sensory field long and narrow, about three and half times as long as maximum breadth,
and pointed behind. Lateral sensory fields reaching a short distance externally onto surface of

DITTONIAN CEPHALASPIDS

151

C\J

I—
UJ

o
o

(N
X

Q.
<U

a.
E

•a

152 E. I. WHITE & H. A. TOOMBS

cornua. Surface of cephalic shield smooth except for fine denticles around orbits, and with up to
twenty rows of fine pits parallel with margins under brim. About 22 rows of scales in front of
dorsal fin with at least six ridge-scales.

HOLOTYPE. Cephalic shield P. 5375 (Fig. 1); counterpart P. 16960 (Fig. 2); Lower Old Red
Sandstone, Cradley, Herefordshire.

MATERIAL. Some 42 individuals on 18 blocks, all from Middle Dittonian of Cwm Mill. In some
instances a single specimen block contains more than one species and for this reason individuals
are separately numbered. The following specimens are from the principal siltstone band:
P.22973, P.22974a,b, P. 22990-2, P.22993a,b, P.22994, P.22998a,b, P.22999a,b, P.23000a,b,
P.23001a,b, P.23002a,b, P.23003a,b, P.23004a,b, P.23005a,b, P. 23008-9 (part and counter-
part), P.23010a,b, P. 23013, P. 60867-8 (part and counterpart), P. 60869a,b, P.60870-4,
P.60875a,b, P.61033a,b, P.61035a,b, P.61036a,b, P.61037-42, P.61043a,b, P.61044a,b,
P.61045-6, P.61047a,b. The remainder are from the hard darker bed: P. 23006-7 (part and
counterpart), P. 25100, P. 25178-9 (part and counterpart).

Cephalaspis cradleyensis is noteworthy in that it is the only species of the genus so far recorded
as being common to the Anglo- Welsh area and Spitsbergen. The original description was based
on a single specimen without counterpart from Cradley, Hereford and Worcester, consisting of
a small, imperfect and somewhat distorted cephalic shield which certainly did not allow more
than a very restricted diagnosis of the species (Fig. 1). This specimen was purchased by the
Museum with the H. B. Hill Collection in 1887 but is not recorded by Woodward (1891),
possibly on account of its relative insignificance. The counterpart, recorded and figured for the
first time here (Fig. 2), was discovered in 1934 in the Museum at Bootle in Lancashire and was
then generously presented to the British Museum (Natural History) by the Committee of that
Museum. The specimen had been bought originally from the well-known dealer in fossils, J. R.
Gregory of London, as an example of Cephalaspis lyelli.

DESCRIPTION. Most of the specimens from Cwm Mill are in some respects rather disappointing
in spite of their original completeness. The siltstone matrix and the armour of the animals are
relatively soft and are often not easy to develop profitably with either tools or acid, and all too
often magnification does little to clarify details. All from the siltstone are flattened.

The holotype and its newly figured counterpart have the great advantage of being
three-dimensional and show a depth at the back of the skull of 12 mm without the crest and, with
the Cwm Mill specimens, it is possible to correct for distortion in regard to the size and form of
the cornua, so that a reasonable restoration of the cephalic shield may be made (Fig. 6, p. 155).

The finest of the Cwm Mill slabs (with its counterpart) is undoubtedly that shown in Fig. 3, for
on its surface are the remains of at least seven specimens, four of which, labelled in the figure as
B (P.22999a), C (P.61043a), D (P.61044a) and E (P.61045), are very nearly complete and show
the squamation of the body reasonably well; the fins and tail are, however, poorly preserved. All
the specimens except D lie in the usual dorso-ventral position, and as the shields are almost
completely flattened the front margins appear entirely rounded, almost semicircular, whereas
specimen D lies, most exceptionally, on its side with the cephalic shield almost in lateral profile.
A second specimen (P.23003a; Fig. 7, p. 157) is in a similar position.

The total length of a complete specimen with tail would be about 120mm, with the median
length of the head, the body from head to base of tail, and the tail itself all very nearly equal,
about 40mm apiece; and 40mm is the average maximum breadth, measured across the tips of
the cornua, in a flattened cephalic shield. There is one specimen that shows nearly the whole
length of the fish, only lacking the tip of its tail (P.23000a, b) and this has approximately the
proportions given.

A most typical cephalic shield is P. 23008-9 and to this in the counterpart is attached part of the
body with the impression of some of the left paddle. The shield is less crushed than many of the
other specimens and gives a more accurate impression of the shape. The polygons formed by the
inter-areal canals of the mucous canal system are as a rule not visible as the superficial layer of
the exoskeleton is continuous, but they do appear occasionally, as in the interzonal part of this

DITTONIAN CEPHALASPIDS

153

Fig. 3 Cephalaspis cradleyensis Stensio. Siltstone block showing remains of four nearly complete
individuals, lettered B (P.22999a), C (P.61043a), D (P.61044a) and E (P.61045). xl-3.

E. I. WHITE & H. A. TOOMBS

w%'.,'

DITTONIAN CEPHALASPIDS

155

Fig. 6 Cephalaspis cradleyensis Stensio. Res-
toration of cephalic shield based on holotype
(P.5375) and its counterpart (P. 16960), and on
P. 25178-9; in dorsal view (a) and left lateral
view (b).

specimen, not naturally but due to the cracking of the convex surface along such lines of
weakness under pressure. There is a low but very definite ridge or spine medially on the short
but wide interzonal part, with correspondingly narrow and shallow pectoral sinuses.

The best example of the median area of a dorsal shield of C. cradleyensis is shown on a piece of
the succeeding darker shaly bed (P. 25178-9; Fig. 5). The specimen, like the original shield from
Cradley, is almost uncrushed, although both sides have been broken away. The median length
of the shield is 38mm; the pineal plate is 16mm from the rounded anterior margin, while the
orbits are each 14 mm away from its centre point. They are oval and measure 4 mm long by 3 mm
at their widest. The dorsal sensory field is long and narrow, measuring 12x5 mm, and is pointed
behind. The lateral sensory fields are 3-5 mm wide in front and are there separated by 6 mm. It is
partly on this specimen that the restoration in Fig. 6 is based.

The external surface of the visceral exoskeleton is very well seen in impression in P. 60868
(Fig. 10, p. 161): there is the same cover of small scales, numerous and irregular in shape, as
Stensio (1932: 43, fig. 8) illustrated in Hemicydaspis , and there is a similar wide and narrow
mouth. In P. 23010 (Fig. 9, p. 159), in addition to mouth and scales, there are impressions of at
least eight box-like branchial pouches on each side. Curiously enough, each of these two
specimens has superimposed on the details of the ventral surface clear impressions of the central
features of the dorsal exoskeleton; the orbits, the pineal plate, the circumnasal fossa and part of
the dorsal sensory field (Fig. 10)! Branchial pouches, ventral scales and part of the mouth are
also to be seen in P. 22993a.

Ornamentation is absent on the upper surface of the cephalic shield except for small areas of
minute tubercles around the orbits and the circumnasal fossa. Under the brim of the shield there

Figs 4-5 Cephalaspis cradleyensis Stensio. Fig. 4, part of a siltstone block: above, in faint outline a
whole specimen, probably a juvenile with incompletely developed armour (P.23005a): below, a
normally developed cephalic shield (P.23004a); xl-8. Fig. 5, block from darker rock showing, at
top right, an imperfect but well-preserved cephalic shield of C. cradleyensis (P. 25178); also dorsal
and ventral discs of Pteraspis (Belgicaspis) crouchi (P. 61 150), either from a small variety or young
specimens, with fragments of plants; xl-6.

156 E. I. WHITE & H. A. TOOMBS

are many rows of minute pittings running parallel with the margin (P. 60868; Fig. 10). On the
body-scales there are numerous short grooves parallel with the length that vary greatly in
definition.

The orbits were said by Stensio to be 'rather or fairly large', but it must be remembered that
the eye-socket is a truncated cone so that the inner aperture in the basal layer, which Stensio was
in fact seeing in the type-specimen, can be, and indeed was, substantially larger than the external
opening seen in the counterpart, which was rather on the small side (cf. Figs 1 and 2). They are
oval, those in P. 23008-9 measuring approximately 4mm long and 3mm wide in a shield
measuring 40mm along the mid-line.

A similar caution should be observed in regard to the length of the lateral sensory fields and
the extent to which they are supposed to run onto the bases of the cornua, since their cavity
extends beyond the tesselated upper surface; if that is removed the field may appear to go
further than it does in fact. In C. cradleyensis the lateral sensory fields are narrow and long and
do continue on to the short cornua.

The dorsal sensory field is also long and narrow, and in P. 23010b it measures 11 mm in length
and 3-5 mm at its widest.

The arrangement of the squamation is much the same as in C. lyelli (White 19580), except that
the main lateral row appears to be rather deeper and the scales are much subdivided
immediately behind the head-shield. The dorsal median scale marking the position of the lost
anterior dorsal fin is relatively insignificant. There are approximately 22 scale-rows to the level
of the remaining dorsal fin.

The finest tail in the collection is P.22974a, b (Fig. 8, A). It is preserved in a completely lateral
position. Unfortunately all the cephalic shield is lost so that the specific identity of the specimen
is not absolutely sure, but although the body is a shade larger than in other specimens, it does
show similar features. The cut-water scales of the dorsal fin are large and six or seven of them
are very well seen, and there are at least 32 ridge-scales along the upper margin of the tail-fin.
Between the latter and the main squamation of the tail and again between the main squamation
and the ventral fin-rays there are single rows of minute longitudinal scales. There are more than
fifty main rows of scales on the tail. The 'fin-rays' are formed of very small scales in rows that
bifurcate at least twice to form the fringe of the tail-fin.

Remains of the 'horizontal antero-ventral lobes of the tail' (Heintz 1939: 112) or 'ventral axis
of caudal fin' (vhp. in Stensio 1932: pi. 34) may be seen in this and other specimens but nowhere
to advantage.

The pectoral fins or paddles are represented in several specimens, usually by impressions of
the basal part, but in P.23002a there is a faint but complete impression of the left fin (Fig. 7, top).
It is 13 mm in length with a maximum breadth of 6 mm. The shape is rather leaflike with a gently
convex outer margin and a slightly sinuous inner margin, forming a rather broad terminal point.
The scales on the limb are as usual largest at the base, diminishing in size distally and marginally.
This specimen is of the usual size, with the cephalic shield 40mm in median length.

The specimens in the darker bed (P. 23006-7, P. 25100 and P. 25178-9) are all imperfect
isolated cephalic shields, associated with broken dorsal and ventral discs oiPteraspis cf. crouchi.
As noted before, in spite of their imperfections, the specimens from this bed are much better
preserved in detail and less crushed than the more complete specimens from the main bed.

Uncrushed shields of Cephalaspis cradleyensis are very simple in form and at a glance not
unlike small editions of the genotype, C. lyelli, but even so there are no species recorded from
the Anglo- Welsh or Spitsbergen areas with which this species may be confused except that
represented by the unique specimen from the latter province which Wangsjo (1952: 255, fig. 24;
pi. 2) placed in C. cradleyensis itself. However, there do seem to be differences between that
specimen and those from the Anglo- Welsh area. The latter do not show a rostral angle nor
denticles along the inner margins of the cornua (Figs 1-2), as Stensio averred in his original
description of the species (1932: 130, fig. 44; pi. 15, fig. 6). These features do not appear to be
very obvious in either of the figures in Wangsjo's plate. It is probably safer then to label the
rather poorly preserved, unique specimen from Spitsbergen simply as 'Cephalaspis cf.
cradleyensis' and to await further discoveries.

DITTONIAN CEPHALASPIDS

157

Fig. 7 Cephalaspis cradleyensis Stensio. Siltstone block with the remains of four specimens, that on
the left (P.23003a) lying on its side. Other specimens visible are P.61035a and P.23002a; xl-9.

158

E. I. WHITE & H. A. TOOMBS

Fig. 8 Cephalaspis cradleyensis Stensio. Siltstone block showing a complete body and tail lying on
one side, A (P. 22974a); and a much flattened cephalic shield with part of the body, B (P. 22973);
x 1-7.

DITTONIAN CEPHALASPIDS

159

Fig. 9 Siltstone block with two specimens of Cephalaspis. The larger specimen (P.23010a) is an
example of C. cradleyensis and shows the impression of most of the lower surface of the body and
almost all the undersurface of the cephalic shield with mouth, scales and branchial pouches on each
side. The smaller specimen at the top (P. 2301 la) is the hoiotype of C. cwmmillensis sp. nov. which
shows the inner impression of the head and the ventral impression of the right cornu. X2-1.

160 E. I. WHITE & H. A. TOOMBS

Cephalaspis cwmmillensis sp. nov.

Figs 9, 12-14

DIAGNOSIS. A species of Cephalaspis of very small size with maximum breadth of cephalic
shield, at base of cornua, about 1-25 times as great as length of shield in median line. Shield
narrows evenly in front without rostral angle and with sides forming a continuous curve with
cornua. Cornua broad at base but narrowing rapidly to a fine point distally and curving gently
inwards to a level a little behind that of posterior interzonal angle. The distance from the cornual
tips to the centre of rostral margin about 3Vi times as long as the length of the cornua. Inner
margins of cornua without denticles. Pectoral sinuses narrow and deep. Interzonal part short
and broad, its breadth between posterolateral angles being about half maximum breadth of
shield, with low median ridge. Posterior angle of interzonal part approximately a right angle,
reaching well behind posterolateral angles. Dorsal sensory field narrow, nearly four times as
long as broad and blunt behind. Lateral sensory fields narrow and short, not reaching onto
cornua. Orbital openings relatively large, oval in shape, lying considerably nearer pectoral
sinuses than rostral margin of shield. Small independent pineal plate present. Exoskeleton
ornamented with numerous minute thorn-like denticles.

HOLOTYPE. Imperfect cephalic shield in counterpart P.23011a, b: Middle Dittonian, Cwm Mill,
Abergavenny, Gwent. The only specimen.

DESCRIPTION. This shield is reasonably well preserved, its deficiencies being largely due to
difficulties in collecting. Originally doubtless much of the body was attached, but as it is, the left
cornual region and the whole of the body has been lost save for the impressions of a few
fragments of body-scales and of the base of the right paddle on the counterpart. The right cornu
lay under the cephalic rim of a head shield of a specimen of C. cradleyensis described above
(Fig. 9). The specimen of C. cwmmillensis is one of the few specimens from this locality that
show fair impressions of any of the vessels of the head (Fig. 12).

The median length of the shield is approximately 28 mm and the estimated maximum breadth,
across the base of the cornua, is 34mm. The breadth of the interzonal part between the
posterolateral angles was about 16mm; the length of the surviving cornu is 10mm and the
distance of its tip from the centre of the front margin of the shield is 34 mm; the distance of the
pineal foramen from the posterior tip of the shield is 14 mm and about the same from the rostral
end, so that the oval orbits were approximately at the middle of the length of the shield but very
much nearer to the pectoral sinuses than to the rostral margin. They measure 3 x 2-25 mm.
The shield is not especially broad but a notable feature is the almost even, continuous curve of
the sides and cornu, and although the anterior margin does narrow rather quickly, there is
certainly no rostral angle and the inner margin of the broad-based but sharply pointed cornu is
smooth. There is some evidence that the interzonal part bore a low median ridge.

The dorsal sensory field is moderately long but almost evenly narrow, and truncated at the
posterior end. It measures 2-5 x 8mm and is about 7mm from the posterior point of the
interzonal aprt. The lateral sensory fields are also narrow and appear to have been unusually
short, stopping some little distance from the base of the cornua.

The exoskeleton is represented by the impression of the external surface only, and this shows

that the upper surface was covered with very small thorn-like denticles, while under the brim

there were numerous rows of interrupted fine ridges parallel with the outer margin of the shield.

The mucous canal system was apparently entirely enclosed in the exoskeleton.

The internal cast shows the impressions of a number of features rarely seen in this fauna, such

Fig. 10 Impression of cephalic shield of C. cradleyensis Stensio showing decoration of the
submarginal rim and the ventral scales of the anterior half; and also impressions of the eyes, the
circumnasal fossa, the pineal plate and the anterior end of the median sensory field (P. 60868).
x3-l.

Fig. 11 Part of siltstone slab with holotype of Cephalaspis abergavenniensis sp. nov. (P. 610345).
X2-1.

DITTONIAN CEPHALASPIDS

161

x/-" "; "^^£r*

162

E. I. WHITE & H. A. TOOMBS

Figs 12-14 Cephalaspis cwmmillensis sp. nov. Fig. 12. Dorsal view of holotype (P. 23011a). The
anterior part has been destroyed and shows impressions of a number of internal features: nsf,
canals of nerves of lateral sensory field; rv, rostral vein; sof, supra-oral field. Fig. 13. Counterpart
of holotype (P.23011b) showing impression of external surface. Fig. 14. Outline restoration of
cephalic shield.

as the supra-oral field, the aortic groove, the naso-hypophysial openings, canals of nerves of the
lateral sensory fields, and of a rostral vein.

REMARKS. The only other species of Cephalaspis from the Anglo- Welsh area similar to
C. cwmmillensis is C. heightingtonensis Stensio (1932: 97; pi. 14, figs 5-7), but that species is
even smaller and relatively broader, and among other obvious differences, the orbits are further
forward, the cornua have denticles along the inner margins, and the pectoral sinuses are
shallower.

Among the several species comprising small individuals from Spitsbergen described by
Wangs jo (1952) none bears much resemblance to C. cwmmillensis, and the same remark applies
to those from Scotland described by Stensio in 1932.

Cephalaspis abergavenniensis sp. nov.
Figs 11, 15-17

DIAGNOSIS. A small species of Cephalaspis with maximum breadth of headshield, at tips of
cornua, about 1-25 times as great as length of shield in median line. Shield narrows rapidly and
evenly forward without forming rostral angle but with sides in continuous curve with cornua,
which are rather stout and of medium length, pointing almost directly backwards and reaching
somewhat beyond level of interzonal median angle; inner side without denticles. Pectoral
sinuses rather narrow and very deep. Interzonal part broad and long with very obtuse median
posterior angle and clearly comprising at least five rows of body-scales, with low median crest
projecting only slightly behind. Dorsal sensory field elongated oval in shape, three times as long
as maximum breadth. Lateral sensory fields extending well onto cornua. Orbital openings
situated well in front of middle of median length , about equidistant from centre of rostral margin
and pectoral sinuses. Small independent pineal plate present. Circumnasal fossa large with

DITTONIAN CEPHALASPIDS

163

Figs 15-17 Cephalaspis abergavenniensis sp. nov. Fig. 15. Dorsal view of holotype (P.61034a).
Fig. 16. External impression of counterpart of holotype (P.61034a). Fig. 17. Restoration of
holotype.

prominent rim. Outer parts of exoskeleton of shield conspicuously subdivided into polygonal
areas by circum-areal mucous grooves. Ornament of exoskeleton of shield of numerous small
but well separated stellate tubercles, increasing slightly in size and becoming thorn-like towards
back of shield and ridge-like on scales of body.

HOLOTYPE. An imperfect cephalic shield with much of the body attached, in counterpart,
P.61034a,b: Middle Dittonian, Cwm Mill, Abergavenny, Gwent. The only specimen.

DESCRIPTION. This small specimen is reasonably well preserved and was certainly complete
when first collected, but now lacks the caudal half of the body and part of one side of the
head-shield, and even more of the body in the counterpart. It lies on a slab with the remains of at
least three specimens of C. cradleyensis.

The median length of the cephalic shield is 36mm and the maximum breadth, flattened, is
44mm across the middle of the cornua; the breadth of the interzonal part between the
posterolateral angles was about 24 mm. The length of the cornu is 12 mm, and the distance of its
tip from the centre of the rostral margin is about 44 mm. The distance of the pineal plate from the
posterior angle of the shield is 20mm and from the rostral margin 15mm, so that the orbital
openings are much nearer to the front of the shield: they are about 14mm from the pectoral
sinuses and are oval, measuring 3 x 2-5 mm.

A small independent pineal plate is present and the circumnasal fossa is large with a
conspicuous rim.

The shield is rather broad at the level of the cornua, but it narrows fairly quickly towards the
front without forming a rostral angle.

The interzonal part is 24mm broad between the posterior lateral angles and is long, being
clearly made up of five or six body-scales which are incompletely fused at the sides. There is a

164 E. I. WHITE & H. A. TOOMBS

short, low median ridge well separated from the median sensory field in front of it, which
scarcely protrudes beyond the very obtuse posterior angle.

The median sensory field is elongate-oval in shape, measuring approximately 9 x 4mm, and
the lateral sensory fields are long and narrow, well separated in front, where there is a small
expansion, and behind they pass for some distance onto the cornua.

The surface of the shield is divided into moderately-sized polygons by the mucous grooves
which, as usual, are much smaller on the sensory fields.

The surface of the shield is ornamented with small but conspicuous stellate tubercles, which
become somewhat larger and more thorn-like backwards towards the body and become short
ridges on the squamation. There are impressions of parts of about 20 rows of scales to be seen.

REMARKS. Like the other diminutive single-specimen species associated with C. cradleyensis at
Cwm Mill C. abergavenniensis is readily distinguished from any other species recorded from the
Anglo-Welsh areas or from Spitsbergen, in this instance by its simple, rather wide cephalic
shield, the 'craquelure' of the outer surface, the long interzonal part with its marked composite
scale-structure and short, low median ridge, together with its very diminutive size.

Subgenus CWMASPIS nov.

DIAGNOSIS. Small species of Cephalaspis with very wide cephalic shield, almost semicircular,
without rostral angle but with very short cornua not even reaching level of acute posterior angle
of very brief and broad interzonal part with low, small median ridge. Pectoral sinuses very
shallow with no denticles on inner side of cornua. Dorsal sensory field long and narrow, oval,
pointed at rear; lateral sensory fields not extending onto cornua and widely separated in front.
Orbital openings situated well in front of middle of median length of shield and considerably
nearer to rostral margin than to pectoral sinuses.

TYPE SPECIES. Cephalaspis billcrofti sp. nov. (only species).

REMARKS. The great breadth and shortness of the whole shield and the interzonal part with the
extreme brevity of the cornua readily distinguish this species from all others and give it a likeness
superficially to some forms of Benneviaspis, but the shape of the sensory fields and the position
of the orbits are those of a true species of Cephalaspis.

Cephalaspis (Cwmaspis) billcrofti sp. nov.
Figs 18-20

DIAGNOSIS. As for subgenus (only species).

HOLOTYPE. Imperfect head shield P. 23012, Middle Dittonian, Cwm Mill, Abergavenny,
Gwent. The only specimen.

DESCRIPTION. This small cephalic shield has lost nearly all the right side but there is sufficient
remaining for the whole to be reconstructed (Fig. 19).

The shield is extremely wide and short and the surviving cornu is very short indeed, probably
not exceeding 10 mm in length, without any denticles along the inner edge. The front margin of
the shield forms a continuous curve from side to side with no suggestion of a rostral angle and the
maximum breadth, across the tips of the cornua, was approximately 56mm, and the median
length only some 40 mm: even so, the point of the median posterior angle was some 7 mm behind
the level of the tips of the cornua. The interzonal part of the shield was very short but broad,
measuring about 40mm between the posterolateral angles which lie far forward, so that the
pectoral sinuses were very shallow and narrow, and the posterior border is somewhat sigmoidal
between the posterolateral angles and the slightly projecting median point formed by a short
and shallow median ridge.

The dorsal sensory field is 13mm long, in shape an elongated oval, pointed behind, with a
maximum breadth of about 4mm. The lateral sensory field preserved is short and pointed

DITTONIAN CEPHALASPIDS

165

10 mm

Figs 18-19 Cephalspis (Cwmaspis) billcrofti subgen. et sp. nov. Fig. 18. Dorsal view of holotype

(P.23012). Fig. 19. Restoration of holotype.

Fig. 20 Cephalaspis (Cwmaspis) billcrofti subgen. et sp. nov. Dorsal view of holotype

(P.23012); X2-1.

behind; it does not reach onto the cornu. Anteriorly the lateral sensory fields must have been
widely separated.

The orbital openings were situated well in front of the middle of the median length of the
shield and measure approximately 4x3-5 mm.

A separate pineal plate is present, lying some 15 mm from the centre of the anterior margin
and about 22 mm from the tip of the posterior angle of the interzonal part. The circumnasal fossa
is conspicuous.

The exoskeleton has a fine granular surface.

166 E. I. WHITE &H. A. TOOMBS

REMARKS. There is no other species of Cephalaspis with which this form may be confused. It is
named for W. N. (Bill) Croft who discovered this interesting section in the course of a
comprehensive study of the Lower Old Red Sandstone of the area which unhappily he did not
live to complete.

Summary of the fauna of the siltstone lenticle:
Cephalaspis cradleyensis Stensio - About 40 specimens
Cephalaspis cwmmillensis sp. nov. - One specimen
Cephalaspis abergavenniensis sp. nov. - One specimen
C. (Cwmaspis) billcrofti subgen. et sp. nov. - One specimen.

The fauna of the 'Darker Bed'

The so-called 'Darker Bed', of which the siltstone lenticle presumably forms part, is to be found
in Units 2 or 3 of the cyclothem at Abergavenny described by Allen (1964: 184-7, fig. 11). The
fauna is not only different from that of the lenticle but very differently preserved: instead of
being crowded together, flattened and comprising complete animals, the fossils are well
separated, three-dimensinal and broken usually into fairly large pieces, and bodies (with one
exception) and tails are completely absent. Cephalaspis cradleyensis is present but rare in the
material collected (Fig. 2), but dorsal and ventral discs of young or a small form of Pteraspis
(Belgicaspis) crouchi are common, as are plant fragments. There are also three substantial
pieces of a much larger species of Cephalaspis than C. cradleyensis, and of these two may well be
parts of the same cephalic shield (P. 22995-7). One specimen shows a round orbital opening
about 5mm in diameter and lanceolate dorsal sensory field 14mm long with a maximum
breadth of 3-5 mm; the other specimen has a low but distinct dorsal crest on the interzonal part;
in both the outer surface is smooth.

The third piece has a short but powerful cornu 22 mm long directed slightly outwards, a rather
narrow pectoral sinus and part of a narrow lateral sensory field running onto the cornu.

There is yet another specimen from the 'Darker Bed' that is worth a mention for, although
very imperfect, it is clearly different from any of the other specimens recorded from the locality
and part of the body is preserved in impression. The matrix is different; although dark it has red
rustlike markings and is very sandy, so that the specimen (P. 60872) is extremely fragile. It shows
the impression in counterpart of the ventral surface of the left cornu and the left side of the body
as far as the base of the tail, and there is a faint outline of perhaps three-quarters of the cephalic
shield. The total length of the animal preserved was approximately 65 mm, while the maximum
breadth of the shield at the level of the tip of the cornu is estimated at 44 mm. The cornu itself at
15 mm is relatively long and there is no sign of denticles along the inner margin. It is 4 mm broad
at its base. The pectoral sinus is deep and rather narrow. The outlines of the small ventral scales
are in places well shown with fine horizontal ridges. There are just enough characters to make it
clear that it does not belong to the same species as any of the other specimens of Cephalaspis
from Cwm Mill but insufficient to carry identification further. Much the same may be said of the
three larger fragments from the 'Darker Bed' already described, and they cannot be positively
associated with any of the congeners from the type-locality of C. cradleyensis (Stensio 1932:
200).

The occurrence of specimens of Cephalaspis

The wide range in time and space of the cephalaspids (Wangsjo 1952: 9-14; Halstead & Turner
1973: 74, figs 7-9), in particular of the type genus Cephalaspis Agassiz (sensu lato), at one time
promised to provide a valuable means of correlating and dating the strata of the Lower Old Red
Sandstone in its several apparently discrete areas. But as Wangsjo rather sadly remarks in his
admirable monograph on the cephalaspids of Spitsbergen (1952: 585), 'for a safe correlation

DITTONIAN CEPHALASPIDS 167

with other areas . . . the Spitsbergen Cephalaspids are in general of fairly little importance
... as the species apparently were not very widespread'. Only two forms, C. cradleyensis,
then a very rare species form the Anglo-Welsh Borders, and a new variety of the Scottish
C. powriei, each represented in Spitsbergen by a single very imperfect cephalic shield, provide
tenuous links with other areas. Further, we may note that no species as yet has been recorded as
common to the neighbouring Scottish and Anglo-Welsh areas.

Worthwhile specimens of Cephalaspis are not so very common considering the very large
regions and the thickness of Lower Devonian non-marine strata from which specimens have
been collected for many years. The Spitsbergen material, on which Stensio (1927: v-ix) based
his classic anatomical studies, dates from collections made from 1909 to 1926, but these
specimens and those collected in 1939 on which Wangsjo based his researches (1952) were much
less satisfactory from the purely systematic point of view, and thus for precise correlation, owing
to natural imperfection of the specimens and so to the relative scarcity of those that could be
clearly named specifically. Wangsjo could identify satisfactorily only about 160 specimens of the
genus from the combined Red Bay and Wood Bay formations, while more than 20 species were
based on single imperfect cephalic shields. As Wangsjo himself remarked (1952: 249), 'In such
cases the diagnosis is, of course, only preliminary', a comment that is unhappily of almost
universal application.

The specimens from the Midland Valley of Scotland are very different in their state of
preservation, especially in Angus (Tayside) in the north-eastern part, where numerous nearly
complete animals have been found (Stensio 1932: pis 25, 28, 30-40, 43, 46). In some quarries
they are flattened, but in others the shape of head and body has been well maintained, as with
the lectotype of Cephalaspis lyelli Agassiz (White 19580), and isolated cephalic shields also
occur uncrushed.

The specimens of Cephalaspis so far descibed from the Anglo-Welsh Borders resemble those
of Spitsbergen in consisting almost entirely of isolated cephalic shields: indeed, out of a total of
about 65 specimens recorded by Stensio (1932) only three had part of the body attached.
Although on the whole much better preserved in regard to external features they show very little
indeed of the internal anatomy.

The first specimens of Cephalaspis were described by Agassiz (1835), yet after nearly a
century Stensio (1932), in his comprehensive monograph of the cephalaspids of Great Britain,
could muster no more than 141 worthwhile specimens for the record, 77 from Scotland and 64
from England. Doubtless other specimens were unaccounted for, simply because they were
unknown or not available to the author at the time the monograph was written, but even so the
numbers are strikingly small. In the Anglo- Welsh area the relative scarcity of specimens is in
great measure due to the high degree of cultivation of the land and its generally flat topography.
Moreover, the Lower Old Red Sandstone there is in general too soft to occasion large-scale
quarrying for building material. Indeed, until fairly recent times much of the collecting seems to
have been done rather casually by amateurs, largely from the labourers in small temporary pits
opened up for farming or other local purposes. This does perhaps explain the all too frequently
poor condition of the specimens and also the very common separation or loss of valuable
counterparts, with the consequent loss of important information in the description of rare
species: one may note that of the twenty-three species and varieties of the genus Cephal-
aspis described from Great Britain by Stensio fifteen type specimens have no known
counterparts to date. Four have counterparts shared by different institutions, and of the other
four species now with counterparts three had the two sides, or parts thereof, reunited after many
years of separation. The lectotype of C. lyelli Agassiz was reunited after 36 years (20087,
presented by Sir Charles Lyell in 1846; P. 3233, in the Enniskillen collection 1882); the holotype
of C. lankesteri after 58 years (45943 in the Lightbody bequest 1874; P. 16155 originally in the
Hereford Museum and presented to the BM(NH) in 1932); and the holotype of C. cradleyensis
after 47 years (P. 5375, H. B. Hill collection 1887; P. 16960, Bootle Museum presentation 1934-
as related above, p. 152). The only instance of a holotype and its counterpart being kept
together in all the Scottish and Anglo-Welsh material described by Stensio is that of the little
cephalic shield of C. heightingtonensis which was purchased complete in 1864.

168 E. I. WHITE & H. A. TOOMBS

The growth and distribution of Cephalaspis

Questions relating to the individual growth, original source and dispersal of the Agnatha in
general have long been matters of discussion. As regards the cephalaspids and the genus
Cephalaspis in particular, it has long been noticed that no juvenile stages have ever been
recorded, and Westoll (1945, 1946, 1958: 192), Denison (1947, 1951, 1956) and White (19586)
on this and other grounds came to the not unreasonable conclusion that 'at least many
cephalaspids acquired their bony skeleton only when fully grown'. That this was so seems now
to be generally accepted (Wangsjo 1952: 247). Such a late development of the armour fits in
very well, as an adult character, with the idea that the earlier Osteostraci were entirely
soft-bodied, which in turn does away with a major objection to the theory that the Agnatha were
marine or at least coastal in origin by explaining the absence of their remains in salt-water
deposits. Such an environment would readily explain their very wide and relatively swift
distribution. Even so, it is still rather difficult to explain the complete absence of partly grown
specimens, rapid though the development of hard parts may have been. Clearly it is a matter of
habitat and where such a metamorphosis could have taken place - obviously not in the area of
formation of the usual Lower Old Red facies. There is in fact just one specimen in the collection
from Cwm Mill that seems to show incompletely formed hard parts. This specimen (P.23005a, b;
Fig. 4, p. 154) is slightly smaller than average specimens of C. cradleyensis and is worth special
consideration. It is a whole animal with tail and is on the same surface of rock as a
normally-developed example of C. cradleyensis. Like that specimen, it is completely flat and in
counterpart, yet only the outlines of its various features can be seen: it is just a ghost of a
specimen. It cannot have been dissolved away after entombment, for nothing has affected the
adjacent specimen, which is perfectly normal in its condition: it does suggest very strongly a
young animal in the early stages of forming its armour. If this interpretation of the fossil is
correct it indicates that ossification took place evenly more or less throughout the animal.

From the acquisition of the hard parts at full growth stage and the subsequent inability to
expand further, it follows that all specimens of a species should be roughly of the same size.
Therefore size was a specific character, and this seems to have been tacitly understood in the
composition of specific diagnoses. Wangsjo (1952: 247), in writing on the Spitsbergen forms,
remarks that 'the shields preserved are always from full grown individuals . . . the variation on
the length of the shield in a single species seems to be at most about 20% of the mean length.'
However, the range in size of some of the British species as described by Stensio (1932) seems to
have been much greater than was anticipated: that is, if the identifications are accepted as
correct.

The following are the variations in the median length of the cephalic shield in five forms, with
percentage variation:

C. salweyi 80 to 145 mm in 14 specimens 80% variation.

C. whitei 45 to 70 mm in 21 specimens 65% variation.

C. pagei 20 to 60 mm in 32 specimens 200% variation.

C. powriei brevicornis 51 to 78 mm in 8 specimens 53% variation.

C. powriei asper 60 to 80 mm in 6 specimens 33% variation.

The first three species certainly call for further consideration and answers may lie between
preservation and identification: certainly more than one species is covered by C. whitei as
originally described.

Other general questions relating to the species of the genus concern the actual habitat and the
proliferation of species at the same level and locality.

The later Dittonian deposits, in which the remains have been found most abundantly in the
Anglo- Welsh region, have been admirably documented and discussed (e.g. Allen & Tarlo 1963:
398, Allen 1964: 194, Allen 1979): we may note that 'the Dittonian facies of the Welsh
Borderland has been interpreted as the deposit of a floodplain complex' and comparison made
with the modern sediments of the Colorado Delta and River (Allen 1963: 398). The climate was

DITTONIAN CEPHALASPIDS 169

'probably warm to hot' (Allen 1974: 152), at any rate in the not so distant Clee area, and that
according to authors there quoted southern Britain was on 'the borders of a major arid zone' or
'in the southern hemisphere within a few degrees of latitude of the Devonian palaeomagnetic
equator'.

However, there are one or two important points that invite further comment. Allen & Tarlo
(1963: 144-6) state that 'During early Dittonian times ... the vertebrates must all have been
freshwater living' and 'in the main have been transported downstream after death', and again in
the Ditton Group that 'Although the vertebrates show evidence of water sorting and
transportation after death, and some were clearly reworked through the floodplain, there can be
no doubt that as in the "Psammosteus" Limestones Group, the animals inhabited the fresh
waters of rivers.' Lastly, Allen (1979: 67) remarks that in his distal alluvial facies 'The
vertebrates emerge as channel-dwellers for at least part of their lives, their remains tending to
accumulate, after much reworking, in the lag deposits formed on the channel floors.'

It is undoubtedly true that the cephalaspids were inhabitants of fresh waters, not of the rivers
in the floodplains to which their distintegrated remains were commonly carried by stream
action, but as inhabitants of the upper reaches, the head waters, lakes and the like, from which
sometimes the complete animals were swept by storm action, generally dead and decaying but
occasionally still alive and subsequently dying when the resultant pools dried up. The
floodplains and the annectent rivers and channels were not the natural environment of the
cephalaspids, they were their mortuaries and graveyards. The statement (Allen & Tarlo 1963:
146) that 'The majority of the cephalaspids were obviously able to survive considerably longer
[than the pteraspids and a few cephalaspids] in such an unfavourable environment' as had
existed at the time of the Cwm Mill or of any other deposition must be considered a
misinterpretation of the facts.

The species of Cephalaspis were evidently poor swimmers. Their forelimbs were not primarily
paddles to aid in progression but balancers to check the depressing effect of the relatively large
and somewhat incongruous heterocercal tail, virtually the only means of propulsion, and of the
heavy armoured head.2 Active animals would not require so complete a protection as in these
creatures, and the flattened undersides of both head and body as well as the form of the
transverse mouth clearly proclaim them as bottom-living scavengers largely suctorial in their
feeding; as such they would not fare very well in the intermittent torrents and rivers of the
floodplains. The idea of carriage from a considerable distance is clearly supported by the rarity
of the preservation of the body or its elements, so manifest in the Spitsbergen and Anglo-Welsh
provinces.

Comment has already been made on the way in which cephalaspids, along with other
Agnatha, 'generally appear at particular horizons only, and are replaced by entirely different
forms' (Allen & Tarlo 1963: 151). This indeed may be so, but the evidence so far published in
regard to the systematics of the cephalaspids and to the relative levels of the known localities
leaves much to be desired, especially in the two provinces in Great Britain. The further remark
that 'This individuality can best be explained by postulating a series of immigrations to the
Anglo- Welsh Province, ... as the presence of some of the genera and species in such distant
Provinces as Podolia and Spitzbergen can only be accounted for by a faunal interchange via the
sea' conjures up the interesting but unlikely picture of endless waves of naked young
cephalaspids, constantly differing in species, largely local in origin but with an occasional

2The suggestion of Janvier (1978: 22) that 'ces nageoires etaient capables de mouvements latero-mesiaux importants' on
the evidence of supposedly special smooth areas on the dorsal and ventral surfaces of the cornua 'chez presque tout les
Cephalaspidides' is certainly not acceptable as a general rule, at least for the earlier, more typical species of the genus
Cephalaspis (s.l.), for of the twenty species recorded from Great Britain by Stensio (1932) no less than eleven bear spines
or denticles along the inner margins of their cornua, and of these eight are described as having also narrow pectoral
sinuses, a combination that would have made the movement of fins up and down past the level of the cornua impossible
without damage. From the Red Bay Series of Spitsbergen out of thirty-seven recorded species nine had the double
handicap of denticles and narrow sinuses, but in the succeeding Wood Bay Series in only one species are such denticles
said to have been present. That species had a very wide head-shield with very wide pectoral sinuses, like most of the
other species known from the formation , a trend that certainly would have allowed freedom of movement to the pectoral
fins.

170 E. I. WHITE & H. A. TOOMBS

intruder from foreign parts, assembling in estuaries from time to time over millions of years,
waiting to don their armour prior to facing the dangers of fluviatile ascent. Further, the notion
that these seemingly sluggish bottom-feeders should have in any way anticipated the
life-wanderings of either salmon or eel seems an even nearer approach to the realms of fantasy.

Wangsjo (1952: 570-1) gives a good generalized idea of the coming-in and disappearance of
cephalaspids and doubtless provides a pattern for the occurrence of the genus Cephalaspis in
other provinces, that is, of the intermittent appearance of apparently unrelated species, often
several at a time but differing in number and very restricted in time and usually in area, but
generically covering a very great period of time and thickness of strata. The injection of fresh
species at irregular intervals is not acceptable if it is agreed that the cephalaspids were originally
soft-bodied and marine in habitat and only developed hard parts when fully grown and
established in fresh water. An endless succession of such metamorphoses taking place over the
millions of years that the genus spanned seems less than likely, nor is the development of almost
endless congeneric species in the open sea a probability.

If, on the other hand, Cephalaspis was first established in fresh waters, it was likely to have
been much more active in its original unarmoured condition that in its ultimate adult
development, allowing the animals to ascend the rivers to the quieter permanent headwaters in
the 'distant land masses' in the north of 'Euramerica' (Young 1981: 226, fig. 1). These waters
were possibly in the form of large connected lakes or inland seas, rather after the pattern of those
in central Africa (White 1950: 58), and may have been sufficiently extensive to cover the
drainage systems of all the Cephalaspid Province. At this stage adult armour was presumably
developed against newly-encountered earlier resident predators, such as eurypterids and large
acanthodians (Miles 1973), and subsequent diversificaton must have taken place in these
relatively quiet waters after the manner of the living cichlid fishes in Africa (Greenwood 1974).
It is interesting to note that Greenwood (1974: 112) records that in Lake Victoria 'over 150
species [of the genus Haplochromis] have evolved within little more than three-quarters of a
million years, from one or at most a few closely related species', and that in the smaller Lake
Nabugabo five endemic species of the same genus have evolved since the lake 'was isolated from
Lake Victoria by a sand spit formed some 3500 years B.P.' The analogy cannot, of course, be
taken too far, since one is dealing with different animals, and at vastly different periods of time,
and there is one major factor in the speciation of the cichlids that the cephalaspids lack, and that
is variation in dentition and hence in diet. Heintz (1940: 181-2) has indeed indicated some
differences in the external details of the oral area in some cephalaspids but there is no indication
of change in their microphagous diet.

If the major habitat of the cephalaspids was, as has been suggested, in the fresh waters of the
distant uplands, it would explain the spasmodic appearance of their remains in the floodplain
deposits, as the results of overflow following unusual rainfall in the region of the headwaters.
The generally disarticulated condition of the fossils resulting, with very rare exceptions, in
nothing but isolated cephalic shields, was due to the distance corpses were carried and the time
that it took, and the readiness with which the flimsy, lightly attached scales would be dispersed
and carried away once decay had set in.

That very occasionally there should occur very local, usually lenticular deposits in which
complete animals with the bodies and fins intact are to be found, as at Cwm Mill and Wayne
Herbert, is to be expected as a result of the floodwaters from exceptionally violent storms in the
uplands rapidly carrying still living or moribund animals down the rivers to be immediately
entombed in the drying-up pools of the warm floodplains. That the lenticle in the somewhat
younger section at Wayne Herbert should have yielded a much more widely diversified fauna
than at Cwm Mill is merely a reflection of the local circumstances at the time.

Acknowledgements

It is our pleasant duty to thank Professor Percival Allen for his generous hospitality to one of us
in the Geological Department at Reading University and continued by his successor Dr Clive
McCann. We are grateful indeed to Professor J. R. L. Allen for much valuable information on

DITTONIAN CEPHALASPIDS 171

the rather involved stratigraphy of the Lower Old Red Sandstone of South Wales, and to Mr
John Cooper for useful hints on other matters.

To Dr H. W. Ball, the Keeper of the Department of Palaeontology at the British Museum
(Natural History), we are greatly indebted for numerous loans of material from his Department
without which, of course, nothing could have been accomplished; also to Dr P. L. Forey,
immediately in charge of the specimens, who not only arranged the loans but transported them
to and from Reading, and to Miss V. T. Young who prepared them.

References

Agassiz, L. 1835. Recherches sur les Poissons fossiles , 2 (1): 85-200. Neuchatel.

Allen, J. R. L. 1963. Depositional features of Dittonian rocks: Pembrokeshire compared with the Welsh

Borderland. Geol. Mag., London, 100: 385-400, 3 figs, pi. 25.
1964. Studies in fluviatile sedimentation: six cyclothems from the Lower Old Red Sandstone,

Anglo-Welsh Basin. Sedimentology , Amsterdam, 3: 163-198, 13 figs.
1974. Sedimentology of the Old Red Sandstone (Siluro-Devonian) in the Clee Hills area, Shropshire,

England. Sedim. Geol., Amsterdam, 12: 73-167, 33 figs.
1979. Old Red Sandstone facies in external basins, with particular reference to southern Britain. In

House, M. R., Scrutton, C. T. & Bassett, M. G. (eds), The Devonian System. Spec. Pap. Palaeont.,

London, 23: 65-80, 6 figs.
& Tarlo, L. B. 1963. The Downtonian and Dittonian facies of the Welsh Borderland. Geol. Mag.,

London, 100: 129-155, 4 figs.
Denison, D. H. 1947. The exoskeleton of Tremataspis. Am. J. Sci., New Haven, 245: 337-365, 13 figs,

pis 1-3.

1951. The exoskeleton of early Osteostraci. Fieldiana, Geol., Chicago, 11 (4): 197-218, 6 figs.

1956. A review of the habitat of the earliest vertebrates. Fieldiana, Geol., Chicago, 11 (8): 357-457.

Greenwood, P. H. 1974. Cichlid fishes of Lake Victoria, East Africa: the biology and evolution of a species

flock. Bull. Br. Mus. nat. Hist., London, (Zool. Suppl.) 6. 134 pp., 74 figs.
Halstead, L. B. & Turner, S. 1973. Silurian and Devonian ostracoderms. In Hallam, A. (ed.), Atlas of

Palaeobiogeography: 68-79. Amsterdam.
Heintz, A. 1940. Cephalaspida from Downtonian of Norway. Skr. norske Vidensk-Akad mat. -nat. Kl.,

Oslo, 1939(5). 119 pp., 30 pis.
Janvier, P. 1978. Les nageoires paires des Osteostraces et la position systematique des Cephalaspido-

morphes. Annls Paleont., Paris, (Vert.) 64: 113-142, 14 tigs.
Miles, R. S. 1973. Articulated acanthodian fishes from the Old Red Sandstone of England, with a review of

the structure and evolution of the acanthodian shoulder girdle. Bull. Br. mus. nat. Hist., London,

(Geol.) 24: 113-213, 43 figs, pis 1-21.
Stensio, E. A. 1927. The Downtonian and Devonian vertebrates of Spitsbergen. Part 1. Family

Cephalaspidae. Skr. Svalb. Nordishavet, Oslo, 12 (1). xii + 391 pp., 103 figs, pis 1-112.
1932. The Cephalaspids of Great Britain, xiv + 220 pp., 70 figs, 66 pis. London, British Museum

(Natural History).
Wangsjo, G. 1952. The Downtonian and Devonian vertebrates of Spitsbergen, IX. Morphologic and

systematic studies of the Spitsbergen Cephalaspids. Skr. norsk Polarinst. , Oslo, 97: 1-615, 109 figs,

pis 1-111.
Westoll, T. S. 1945. A new cephalaspid fish from the Downtonian of Scotland, with notes on the structure

and classification of Ostracoderms. Trans. R. Soc. Edinb., 61: 341-357, 7 figs, 1 pi.

1946. ('Interchange by sea' in discussion). Q. Jl geol. Soc. Lond., 101: 242.

1958. The origin of Continental vertebrate faunas. Trans, geol. Soc. Glasg., 23: 79-105, 4 figs.

White, E. I. 1935. The ostracoderm Pteraspis Kner and the relationships of the agnathous vertebrates.

Phil. Trans. R. Soc., London, (B) 225: 381-457, 97 figs, pis 25-27.
1950. The vertebrate faunas of the Lower Old Red Sandstone of the Welsh Borders. Bull. Brit. Mus.

nat. Hist., London, (Geol.) 1: 51-67, 2 figs.

1958a. On Cephalaspis lyelli Agassiz. Palaeontology, London, 1: 99-105, 3 figs, pis 18, 19.

1958&. The original environment of the craniates. In Westoll, T. S. (ed.), Studies on Fossil

Vertebrates: 213-233, 8 figs. London.
Woodward, A. S. 1891. Catalogue of the Fossil Fishes in the British Museum (Natural History), 2. xliv +

567 pp., 16 pis. London.
Young, G. C. 1981. Biogeography of Devonian vertebrates. Alcheringa, Adelaide, 5: 225-243.

British Museum (Natural History)

An account of the Ordovician rocks of the Shelve Inlier in west Salop
and part of north Powys

By the late W. F. Whittard, F.R.S. (Compiled by W. T. Dean)

Bulletin of the British Museum (Natural History}, Geology series
Vol. 33 No. 1. Dec. 1979. 69pp. 38 figs. Large full-colour map

The late Professor W. F. Whittard, F.R.S., who died in 1966, devoted much of
his life to the study of the Shelve Inlier, and his great monograph on its trilobites
remains fundamental. The area, in west Salop (including a small part of north
Powys), was the scene of famous early geological studies by Murchison, and
Lapworth. By Palaeozoic standards it is in places richly fossiliferous, and exhibits
the best continuous Ordovician succession in Britain, one which is indeed amost
complete. This classic area is of continuing interest, not only to professionals
but also to amateur geologists and students, few of whom complete their
studies without at least one field visit; but amazingly this is the first detailed
map ever to be published. That the work of Whittard, now made available
through the efforts of Professor W. T. Dean of Cardiff, is authoritative there
can be no doubt: for over thirty-five years he studied these rocks, unravelling
their complexities and perfecting his map.

The work complete with map, £10.50 (Post & packing 30p)
Map only, £1.00(P&p. lOp)

A related work :

Ordovician Brachiopoda from the Shelve District, Shropshire

By A. Williams

Bull. B.M.(N.H.\ Geology Supplement 11, 1975. 163pp., 28 plates, 5 tables
11 text figs. £13.00 (P & p 50p)

All British Museum (Natural History) publications are obtainable from Agents,
Booksellers, the Museum bookshop or by post direct from:
Publication Sales, British Museum (Natural History), Cromwell Road
London SW7 5BD, England

Titles to be published in Volume 37

Taxonomy of the arthrodire Phlyctaenius from the Lower or Middle
Devonian of Campbellton, New Brunswick, Canada.

By V. T. Young

Ailsacrinus gen.nov.: an aberrant millericrinid from the Middle
Jurassic of Britain. By P. D. Taylor

Miscellanea

Devonian Fish from Australia. By B. G. Gardiner

Printed by Adlard & Son Ltd, Bartholomew Press, Dorking, Surrey

Bulletin of the

British Museum (Natural History)

The relationships of the palaeoniscid
fishes, a review based on new specimens
of Mimia and Moythomasia from the
Upper Devonian of Western Australia

ByB. G. Gardiner

Geology series Vol 37 No 4 29 November 1984

Publications Sales,

British Museum (Natural History),
Cromwell Road,

London SW75BD,
England.

World List abbreviation: Bull. Br. Mus. nat. Hist. (Geol.)

©Trustees of the British Museum (Natural History), 1984

The Geology Series is edited in the Museum's Department of Palaeontology
Keeper of Palaeontology: Dr H. W. Ball
Editor of the Bulletin: Dr M. K. Howarth

Assistant Editor: Mr D. L. F. Sealy

ISBN 0565 00967 2
ISSN 0007-1471

Geology series

British Museum (Natural History) Vol 37 No 4 pp 173-428

Cromwell Road
London SW7 5BD Issued 29 November 1984

The relationships of the palaeoniscid fishes, a review
based on new specimens of Mimia and Moythomasia
from the Upper Devonian of Western Australia

B. G. Gardiner

Department of Biology , Queen Elizabeth College , University of London , London W8 7 AH

Contents

Synopsis

Introduction

Lettering used in text figures

Systematic descriptions

Family Stegotrachelidae Gardiner ....
Genus Mimia Gardiner & Bartram

Mimia toombsi Gardiner & Bartram .
Genus Moythomasia Gross .....
Moythomasia durgaringa Gardiner & Bartram

Neurocranium: general features

Occipital region

Mimia toombsi

Moythomasia durgaringa

Occipital region: discussion

1. Posterior dorsal fontanelle ....

2. Occipital fissure

3. Vestibular fontanelle

4. Ventral otic fissure

5. Supraoccipital

6. Aortic canal

7. Canal for abducens nerve

8. Zygal plates

9. Occipital artery

10. Segmental structure of occiput .

11. Longitudinal intervertebral ligament

Otic and orbitotemporal regions

Review of ossification centres

Mimia toombsi .......

Moythomasia durgaringa

Otic and orbitotemporal regions: discussion

1. Parampullary process

2. Articulation of first suprapharyngobranchial

3 . Articulation of first infrapharyngobranchial

4. Lateral commissure and trigeminofacialis chamber

(a) Actinopterygians

(b) Osteichthyans

5. Hyomandibular facet

6. Otico-sphenoid fissure

7. Fossa bridgei and lateral cranial canal

8. Spiracle and spiracular canal ....

9. Origin of dorsal hyoid constrictor muscle .

10. Origin of dorsal mandibular constrictor muscle .

1 1 . Endolymphatic duct

12. Pterosphenoid pedicel

13. Prootic bridge ana posterior myodome

175

176

177

181

182

185

199

201

203

204

206

207

208

210

215

228

231

233

234

235

236

238

239

240

241

243

244

245

246

247

Bull. Br. Mus. not. Hist. (Geol.)37(4): 173-428

Issued 29 November 1984

173

174 B. G. GARDINER

14. Anterior and middle cerebral veins 249

15. Sclerotic bones 251

Ethmoid region and associated dermal bones . . , . .• . . . . 254

Mimia toombsi ' 254

Moythomasia durgaringa . . . .; .' ". • .' _. . • >. " •. • . 260

Ethmoid region: discussion .......... 264

1 . Anterior myodome . . . . . 264

2. Postnasal wall and nasal capsule . 266

3. Dermal bones of the snout 267

Parasphenoid and associated toothplates 271

Mimia toombsi 271

Moythomasia durgaringa .......... 273

Parasphenoid: summary and discussion 273

1 . Parabasal canal 275

2. Internal carotid artery 276

3. Basipterygoid process 276

4. Ascending process 277

5. Parasphenoid teeth 278

6. Bucco-hypophysial canal 279

7. Subcephalic muscles 279

8. Accessory toothplates 280

Palatoquadrate and dermal bones of the cheek 280

Mimia toombsi 280

Moythomasia durgaringa 292

Palatoquadrate: summary and discussion 293

1 . Palatoquadrate commissure and vomer 293

2. Anterior articulation 297

3. Otic process and palatobasal articulation 298

4. Otic process and prespiracular cartilage 301

5. Ossifications of the palatoquadrate 305

(a) Cartilage bones 305

(b) Dermal bones 306

(d) Dermopalatines 307

(e) Entopterygoid 308

(f) Dermometapterygoid 310

Dermal bones of cheek: summary and discussion 310

Sensory canals of cheek: summary and discussion 313

Dermal bones of the skull roof 316

Mimia toombsi 316

Moythomasia durgaringa 318

Dermal bones of skull roof: summary and discussion 320

1 . Homologies of dermal bones of skull roof 320

2. Sensory canals of skull roof 324

Lower jaw 325

Mimia toombsi 325

Moythomasia durgaringa 331

Lower jaw: discussion , 332

1 . Meckelian ossifications 332

2. Dermal bones . 332

Operculogular series 338

Mimia toombsi 338

Moythomasia durgaringa . . 339

Operculogular series: summary and discussion . . . . . . . 339

1. Branchiostegal rays and gular plates ........ 339

2. Opercular cartilages and opercular bones . 342

Hyoid and branchial arches 344

Mimia toombsi 344

Moythomasia durgaringa 349

RELATIONSHIPS OF PALAEONISCIDS 175

Hyoid and branchial arches: discussion 349

l.Hyoidarch 349

2. Basibranchial and branchial arches 359

Axial skeleton 363

Mimiatoombsi 363

Moythomasia durgaringa .......... 364

Axial skeleton: discussion 364

1 . Arcualia 364

2. Centra 365

3. Ribs 366

4. Supraneurals and neural spines 368

Shoulder girdle and pectoral fin 368

Mimiatoombsi 368

Moythomasia durgaringa 374

Shoulder girdle and pectoral fin: discussion 374

1 . Dermal bones of shoulder girdle 374

2. Endoskeletal girdle 378

3. Pectoral fin 381

Pelvic girdle and fin 382

Mimiatoombsi 382

Moythomasia durgaringa 382

Pelvic girdle and fin: discussion 383

Median fins 384

Mimiatoombsi 384

Moythomasia durgaringa 385

Median fins: discussion 386

1. Dorsal and anal fins 386

2. Caudal fin 386

Squamation 386

Mimia toombsi 386

Moythomasia durgaringa 387

Squamation: discussion 388

1 . Scale structure 388

2. Basal fulcra 392

3. Fringing fulcra 394

Phylogenetic results 394

Interrelationships of actinopterygians 394

Classification 399

Relationships of actinopterygians 400

Acknowledgements 406

References 407

Index 418

Synopsis

Two species of palaeoniscids are described from the Frasnian of Gogo, Western Australia: Mimia toombsi
Gardiner & Bartram, and Moythomasia durgaringa Gardiner & Bartram. A detailed account of their head
structure, appendicular and axial skeletons and squamation is given in a series of accounts of regional
anatomy. Each account is accompanied by a discussion of the salient features in which comparison is made
with living and fossil actinopterygians as well as all other major gnathostome groups. In the course of these
comparisons previously-described material of Cheirolepis, Acanthodes, various placoderms, hybodont
sharks and cephalaspids is reinterpreted.

The Western Australian genera are similar in many respects, including the pattern of the dermal bones of
the skull, cheek and pectoral girdle. Moythomasia has a short, blunt snout with small premaxillae
separated by a toothed rostral, and a palatoquadrate with a short anterior ramus. Mimia has a longer,
hooked snout with large premaxillae which meet in the mid-line, excluding the rostral from the jaw margin,
and a palatoquadrate with a long anterior ramus. Moythomasia shows one marked advance over Mimia in
the possession of a rudimentary ascending process on the parasphenoid.

The most striking primitive features of the Gogo palaeoniscids are the incomplete cranial fissure in

176 B. G. GARDINER

which the ventral otic fissure was cartilage-filled and separate from the perichondrally-lined
otico-occipital fissure, the presence of a lateral cranial canal, dermohyal, basal and fringing fulcra, a
perforated propterygium, and the assumed enclosure of the adductor mandibulae muscle by the
palatoquadrate and dermal cheek bones.

The principal anatomical conclusions concern the history of the myodome and trigeminofacial chamber,
and the spiracular canal.

The ossification patterns of the skull roof, cheek, palate, lower jaw and shoulder girdle of osteichthyans
are reviewed critically. It is concluded that there are two distinct dermal roofing bone patterns and that an
operculogular series is a primitive gnathostome attribute whereas submandibulars are a sarcopterygian
synapomorphy.

The principal conclusion about the interrelationships of actinopterygians is that the Palaeonisciformes
are a paraphyletic group. Mimia and Moythomasia are stem-group actinopterans whereas Cheirolepis is a
basal actinopterygian. Most other palaeoniscids are stem-group neopterygians and may be inserted
between the Chondrostei and the Neopterygii.

The principal broader phylogenetic conclusions concern the interrelationships of the sarcopterygians
and the relationships of the chondrichthyans and placoderms. The Porolepiformes are considered to be the
sister-group of the choanates, the placoderms the sister-group of the osteichthyans and the
chondrichthyans the primitive sister-group of other gnathostomes.

Introduction

The first aim of this paper is to make known the palaeonisciform fishes from the Devonian Gogo
Formation of Western Australia.

The specimens were mainly collected in 1967 by a joint expedition from the British Museum
(Natural History), the Western Australian Museum and the Hunterian Museum, Glasgow
(Brunton, Miles & Rolfe 1969; with references) at Gogo Station, a cattle property some 250 km
SE of Derby in the Fitzroy Trough. The Fitzroy Trough lies on the northern flank of the Canning
Basin, with its northern limit faulted against the Kimberley Plateau, a stable Precambrian block
(Playford & Lowry 1966). The Canning Basin was apparently land during the Middle Devonian,
with a southern shore-line near its junction with the Kimberley Plateau. In the late Devonian the
Canning Basin slowly subsided, leaving much of the Fitzroy Trough as a near-shore shelf (the
Lennard Shelf) some 300km in length and several km wide. Both fringing and atoll
stromatoporoid and algal reefs grew upon this shelf, with typical reef development including
reef, back-reef, fore-reef and inter-reef facies. The Gogo Formation is inter-reef and composed
of shales and siltstones with thin bands of limestone and numerous calcareous concretions.
About half of the calcareous siltstone concretions contain fossils, chiefly phyllocarid crustaceans
and fishes (Gardiner & Miles 1975). The formation is well dated on palaeontological grounds as
Frasnian la- to 1/3 (Roberts et al. 1972; with references). Since Devonian times weathering has
removed much of the softer inter-reef deposits of the Fitzroy Trough, leaving the fossil-bearing
concretions lying on the surface. A map of the Gogo Formation localities is given by Miles
(19716: fig. 1). A few specimens were collected on an earlier expedition by H. A. Toombs of the
British Museum (Natural History) in 1963.

The fishes have been prepared by the standard acetic acid techniques and one specimen
(Mimia) has been serially sectioned. The specimens are uncrushed and often almost complete,
which suggests that the concretions themselves developed during an early stage of diagenesis in
still- water conditions.

The second aim of this paper is to discuss certain aspects of actinopterygian comparative
morphology that may have a bearing on the problems of actinopterygian relationships and
interrelationships. Thus the anatomical descriptions are divided into several parts each of which
is followed by a discussion section. The discussions are intended to establish or propose
primitive conditions in various groups (i.e. synapomorphies of those groups). Wherever
possible homologies are established by congruence with other characters, but in some cases the
criterion of commonality is used. I have also employed ontogenetic precedence and outgroup
comparison as well as the stratigraphical succession in helping to establish the polarity of
transformation series.

RELATIONSHIPS OF PALAEONISCIDS

177

The term 'palaeoniscid' is used throughout for those fossil fishes which have traditionally been
included within the extinct Palaeonisciformes. 'Actinopterygian' is used for any member of the
group Actinopterygii, 'actinopteran' for members of the Actinopteri (Rosen et al. 1981) and
'osteichthyan' for bony fishes plus tetrapods. 'Sarcopterygian' refers to members of the
Actinistia plus Choanata, while the term 'rhipidistian' is used for Osteolepiformes,
Porolepiformes and Youngolepididae.

Specimen numbers are prefixed as follows: BMNH, British Museum (Natural History); RSM,
Royal Scottish Museum; GSM, Institute of Geological Sciences, London.

aasc
acv

aesc
af

a.ghy

ahy

aipl

alig

amyd

amyv

ano

anpl

apal

apr
apse
Ar
art.H

art.Hb

ascf
asp

asup 1

Aup

bhc
bhm

bine

Lettering used in text figures

anterior ampullary chamber bpopc
foramen of anterior cerebral

vein bpt

external ampullary chamber br
area of fusion between

occipital arch and otic Bsp

capsule

area of attachment of can

geniohyoideus muscle can.W

groove for afferent hyoidean Cb

artery cao

articular facet for first cdic

infrapharyngobranchial Ch

area of origin of aortic ligament chy

dorsal anterior myodome cla

ventral anterior myodome Clav

angular dm

anterior nasal opening cnc

(external incurrent nostril)

angular pit-line copl

anterior pit-line Cor

articular facet for autopalatine cor

(palate) corf

anterior process cotel
posterior ampullary chamber

articular crd
articulation surface for

hypohyal

articulation surface for csim

hypobranchial c.sp.

anterior scapular foramen ctel
ascending process of

parasphenoid dasc
articular surface for first

suprapharyngobranchial De

autopalatine dend
accessory vomerine tooth plate

Dhy

basibranchial Dmpt

bucco-hypophysial canal dop
foramen for branches of

mandibular nerve to pit Dpi

organs dpi

tube or foramen for dorsal dpsc

branch of infraorbital

sensory canal in premaxilla; Dspo

or pore opening therefrom dt

branch of preopercular sensory

canal

basipterygoid process
bridge of bone separating V

and VII hyomandibular
basisphenoid

canal through propterygium

ascending canals of Williamson

ceratobranchial

aortic canal

capsule housing diencephalon

ceratohyal

hyomandibular canal

canal for lateral aorta

clavicle

cleithrum

cavity occupied by nasal

capsule

capsule housing optic lobes
coronoid

canal from orbit to skull roof
coracoid foramen
canal joining orbit with

telencephalon
canal for ascending branch of

superficial ophthalmic

nerves

cavum sinus imparis
cell spaces
capsule housing telencephalon

dorsal opening of anterior

semicircular canal
dentary
tube or foramen for

endolymphatic duct
dermohyal
dermometapterygoid
depression for opercular

cartilage?
dermopalatine
dentary pit-line
dorsal opening of posterior

semicircular canal
dermosphenotic
dentinal tubules

178

B. G. GARDINER

Ecpt

Enpt

epi

epopc

ethc

Exsc

EXSC!

Exsc2
fapcv
fbmand.ext.VII

fboca
fb.IX

fb.X

fcb

fendc

fepsa

fexna
ffr
fhm
fhm.VII

fhm.VII + pal

fhy.VII
fia
fica
fica2

fi.Sh
fm

fmand.V

fmand.VII
fmand.int.VII

fmxv.Vbuc.VII

epibranchial

ectopterygoid

entopterygoid

epural

epineural process

entrance of preopercular

sensory canal
ethmoidal commissural

sensory canal or pit opening

therefrom
extrascapular
medial extrascapular
lateral extrascapular

foramen for anterior tributary

of posterior cerebral vein
foramen for branches of

external mandibular branch

of facial nerve
foramen for branch of occipital

artery
foramen for branches of

glossopharyngeal nerve to

pit-line
foramen for branches of vagus

nerve to pit-line
ceratobranchial foramen
fenestra endonarina communis
foramen for efferent

pseudobranchial artery
fenesta exonarina anterior
fringing fulcra
hyomandibular facet
foramen of hyomandibular

trunk of facial nerve
foramen of hyomandibular and

palatine trunks of facial

nerve
foramen for hyoid branch of

facial nerve
foramen for intersegmental

artery
foramen of internal carotid

artery
ascending canal of internal

carotid artery in

basisphenoid pedicel
canals for fibres of Sharpey
foramen magnum
foramen for mandibular

branch of trigeminal nerve
foramen for mandibular

branch of facial nerve
foramen for internal

mandibular branch of facial

nerve
foramen for maxillary branch

of trigeminal and buccal

branch of facial nerve

foa

foca

focn

fona
fopa
for

fos
fotc

fpal
fpa!2

Fr
frd

frla

frlai , frla2
frmx
fst.IX

fv
fvii

gboca

gdend

gf
ghm.VII

gic

gla

gmand.ext.VII

goa

goca

gona

gpl
gpcv

gph-X

foramen of orbital artery
foramen of occipital artery
foramen or notch for occipital

nerve

foramen of orbitonasal artery
foramen for ophthalmic artery
fenestrae linking dorsal

myodomes
otico-sphenoid fissure
otico-occipital fissure
foramen of otic nerve
foramen of palatine branch of

facial nerve
entry foramen of palatine

nerve into parabasal or

palatine canal
frontal
foramen or notch for ascending

branch of superficial

ophthalmic nerves
foramen of ramus lateralis

accessorius
foramen of branch of ramus

lateralis accessorius
foramen for maxillary branch

of trigeminal nerve
foramen of supratemporal

branch of glossopharyngeal

nerve

ventral otic fissure
venous foramen

ganoine

gap in ossification of wall of

olfactory nerve canal
groove for branch of occipital

artery

groove for endolymphatic duct
glenoid fossa
groove for hyomandibular

trunk of facial nerve
groove for internal carotid

artery
lateral gular
groove for lateral aorta
median gular
groove for external mandibular

branch of facial nerve
groove for orbital artery
groove for occipital artery
groove for orbitonasal artery
gular pit-line
groove for posterior cerebral

vein
groove or foramen for

pharyngeal branch of vagus

nerve

RELATIONSHIPS OF PALAEONISCIDS

179

gr
gst.IX

gst.X
g-X

hbpt

hll

hpl

Iclav
iepl

Ih
inc

inw
ios

IP
It
ivl

jg
Ju

Lac
lapf

Ic
Ice
Icom
11

Imc

Impt
Inabc

Men

mcv

mnabc

groove in dermopalatines and Mpt

ectopterygoid and in mr

coronoids and prearticular msc

groove for supratemporal mscp

branch of glossopharyngeal mtp

nerve mvfon

groove for supratemporal MX
branch of vagus nerve

groove for vagus nerve n

first hypural Na

haemal arch na

hypobranchial nabc

hole for basipterygoid process
haemal canal nc

hypohyal nfendc

horizontal longitudinal lamina
horizontal pit-line
hyomandibula nona

not

interclavicle npl

insertion points of

ethmopalatine ligament oahm
interhyal
tube or foramen for

infraorbital sensory canal or oaop

pore opening therefrom
internasalwall

interorbital septum oatm

infrapharyngobranchial
intertemporal oem

area of insertion of oexr

intervertebral ligament

jugular canal

jugular groove oims2

jugal

olab

lachrymal

fossa for levator arcus palatini olap

muscle

cephalic division of lateral line Op

lateral cranial canal orb

lateral commissure ore

foramen through which lateral Ors

line enters supratemporal osubc

or lateral line scale
lower muscle canal

( = supracoracoid foramen) p
lamina of metapterygoid

lateral nasobasal canal Pa

pamp

mentomeckelian Par

mandibular sensory canal pare
foramen of middle cerebral

vein pchl

ossified Meckelian cartilage Pel

medial nasobasal canal pdf

middle pit-line ped

metapterygoid

marginal fin-ray

mesocoracoid arch

mesocoracoid process

metapterygium

margin of vestibular fontanelle

maxilla

notch in margin of

supratemporal
nasal

neural arch
foramen of nasobasal canal in

floor of nasal capsule
nasal capsule
notch for posterolateral part of

fenestra endonarina

communis

notch for orbitonasal artery
notochordal canal
nasal pit-line

area of origin of adductor

hyomandibulae portion of

dorsal constrictor
area of origin of adductor

opercularis portion of

dorsal constrictor
area of origin of anterior trunk

muscles

area of origin of eye muscles
area of origin of external rectus

muscle
area of origin of first

intermuscular septum
area of origin of second

intermuscular septum
area of origin of levator

branchialis muscles
area of origin of levator

palatini muscle
opercular
orbit

orbito-rostral canal
orbitosphenoid
area of origin of subcephalic

muscle

pores in premaxilla, nasal and

lachrymal
parietal

parampullary process
prearticular
opening or course of parabasal

canal

pit for ceratohyal ligament
postcleithrum
posterior dorsal fontanelle
'alisphenoid pedicel'

180

pesc

Pg
pinf
pitf
plcc

Pmx

pno

pnw
Po
po
podp

Pop

pope

por

PP
PP

prepf
prh

Pro

prob
prof

prof2

propt
Prscl
psc

Psp

Pts
pu8
pv

B. G. GARDINER

posterior opening of external

semicircular canal
pelvic girdle
pineal foramen
pituitary fossa
posterior opening of lateral

cranial canal
premaxilla
posterior nasal opening

(external excurrent nostril)
postnasal wall
postorbital
lateral line pore
pharyngeal (parotic) tooth

plate

preopercular

preopercular sensory canal
postorbital process
postparietal
posterior pit-line
prepalatine floor
hyoid process of

branchiostegal
prootic

prootic bridge (dorsum sellae)
foramen or canal for profundus

nerve
foramen or canal for branches

of profundus nerve
propterygium
presupracleithrum
cavity occupied by, or ridge

over, posterior semicircular

canal

parasphenoid
post-temporal
pterosphenoid
eighth pre-ural centrum
foramen or pathway for

pituitary vein

rmet
rmye

ropl

rot

rpl

rsoc

rtel

sacr

San

scf

Scl

sgf

Soc

Sop

spic

spig

spip

ssu

St
stc

sue

svfotc
svr

Tab

recess housing metencephalon
recess housing

myelencephalon
rostral

recess housing optic lobe
otic nerve
radial plate

recess on roof of otic region
recess housing telencephalon

saccular recess

supra-angular

scale

scapular or coracoid foramen

supracleithrum

supraglenoid foramen

supraneural

supraoccipital

subopercular

suprapharyngobranchial

spiracular opening

spiracular bar

spiracular groove

spiracular tooth plate

division of labyrinth cavity for

the sinus superior
supratemporal
supratemporal commissural

sensory canal
tube or foramen for

supraorbital sensory canal

or pore opening therefrom
sub-vagal portion of otico-

occipital fissure
recess housing saccus

vasculosus

tabular
toothplate

Qu quadrate

Quj quadratojugal

qujpl quadratojugal pit-line

r radial

rbuc buccal nerve

Rbr branchiostegal ray

rcor coracoid ridge

rdo ascending branches of

superficial ophthalmic

nerves
rhm+pal hyomandibular trunk and

palatine branch of facial

nerve
rla ramus lateralis accessorius

utr

va
vfon
vnabc
vnabcf

Vo
vpl

utricular recess

ventral arch
vestibular fontanelle
ventral nasobasal canal
opening of ventral nasobasal

canal in roof of mouth
vomer
vertical pit-line

zygal plate

tube or foramen for olfactory
tracts

RELATIONSHIPS OF PALAEONISCIDS 181

II optic fenestra VI^V^ foramina of abducens nerve

III notch, foramen or canal of VII foramen or canal for facial

oculomotor nerve nerve

IV notch or foramen of trochlear VII. lat foramen or canal for lateralis

nerve trunk of facial nerve

V foramen or canal for trigeminal IX foramen of glossopharyngeal

nerve nerve

VI foramen of abducens nerve X foramen of vagus nerve

Systematic descriptions

Family STEGOTRACHELIDAE Gardiner, 1963

Genus MIMIA Gardiner & Bartram, 1977

DIAGNOSIS. Stem-group actinopteran fishes in which the ventral otic fissure passes into the rear
of the orbit; the parasphenoid is broad but without basipterygoid or ascending processes; the
otico-sphenoid fissure is cartilage-filled; a pair of orbitonasal arteries passed into the orbit
immediately lateral to the ventral otic fissure; the spiracular groove is wide, there is a spiracular
slit between intertemporal and dermosphenotic; the neurocranium contains a lateral cranial
canal; the perforated pectoral propterygium is embraced by the bases of the marginal rays; and
basal and fringing fulcra are present.

TYPE SPECIES. Mimia toombsi Gardiner & Bartram, 1977.

REMARKS. Mimia and Moythomasia are closely related Devonian fishes which have been placed
in the family Stegotrachelidae (Gardiner 1963; Gardiner & Bartram 1977). Unfortunately all
the characters they share with Stegotrachelus are primitive, but until we know more of the
internal anatomy of Stegotrachelus and other stem-group actinopterans it is premature to
suggest alternative family groupings. Consequently the genera will be referred to either as
Mimia and Moythomasia or as 'the Gogo palaeoniscids' in the following account.

Mimia toombsi Gardiner & Bartram, 1977

1970 Devonian stegotrachelid; Gardiner: 285; fig. 3.

1971 Stegotrachelid palaeoniscoid; Gardiner in Moy-Thomas & Miles: figs 5, 6.
1973 Gogo palaeoniscid 'A'; Gardiner: 106; figs 1, 2, 6, 8, 9.

1973 Gogo palaeoniscid 'B'; Gardiner: figs 3, 4.

1975 Gogo palaeoniscid; Gardiner & Miles: fig. 2.

19776 Palaeoniscoid; Patterson: fig. IB.

1977 Mimia toombsi Gardiner & Bartram: 228; figs 1-6.

DIAGNOSIS. As for genus.

HOLOTYPE. Western Australian Museum 70.4.245; partly disarticulated specimen wanting fins,
in counterpart, from the Upper Devonian, Gogo Shales, Gogo Station (H. A.T. 67/80, see Miles
19716), Fitzroy Crossing, W. Australia.

SPECIMENS. This study is based on 61 specimens from the following Gogo localities: 21, 25, 27,
30, 36, 37, 42, 47, 54, 55, 56, 63, 73, 80, 84, 87, 89, 91, 92, 302 (for Gogo localities see Miles
19716: fig. 1).

Genus MOYTHOMASIA Gross, 1950
[= Aldingeria Gross 1942:431]

DIAGNOSIS. See Gross (1942: 430) and Jessen (1968: 89). In addition the ventral otic fissure
passes into the rear of the orbit; the parasphenoid is broad with a rudimentary ascending process
but no basipterygoid process; the palatoquadrate has a short anterior process; there is a
spiracular slit between the intertemporal and dermosphenotic and the dermosphenotic is hinged

182 B. G. GARDINER

to the jugal as in Mimia and Cheirolepis; the neurocranium has a lateral cranial canal; the lower
jaw has a supra-angular; the pectoral fin has a perforated propterygium embraced by the bases
of the marginal rays; there are prominent basal fulcra above and below tail; and fringing fulcra
are present on all fins.

TYPE SPECIES. Moythomasia perforata (Gross).

REMARKS. The genus is known from the Devonian (Frasnian) of Kokenhusen, Latvia, Bergisch
Gladbach and Wildungen, Germany and Gogo, Western Australia (Gross 1950, 1953; Jessen
1968; Gardiner & Miles 1975; Gardiner & Bartram 1977).

Moythomasia durgaringa Gardiner & Bartram, 1977

1973 Gogo palaeoniscid 'B'; Gardiner: figs 5 and 7 only (not figs 3, 4).

1977 Moythomasia durgaringa Gardiner & Bartram: 238; fig. 7.

1981 Moythomasia durgaringia Gardiner & Bartram; Forey & Gardiner: 140.

DIAGNOSIS (emended). A Moythomasia with scales with up to 15 serrations posteriorly.

HOLOTYPE. Western Australian Museum, 70.4.244; partly disarticulated head and body in
counterpart from the Upper Devonian, Gogo Shales, Gogo Station (H.A.T. 67, see Miles
19716), Fitzroy Crossing, W. Australia.

SPECIMENS. This study is based on 18 specimens from the following Gogo localities: 36, 37, 72,
78, 80, 84, 86, 89.

Neurocranium: general features

The neurocrania of Mimia and Moythomasia are very similar, more or less completely ossified,
with both external and posterior semicircular canals visible externally.

Comparative measurements show that in both the maximum breadth (between the
postorbital processes) is about 60% of the total length, while the depth, which remains
remarkably constant throughout, is some 34% of the length. The orbital length (postorbital
process to tip of ethmoid) constitutes at least 55% of total braincase length in Mimia, but in
Moythomasia the orbital length is 50% of total braincase length.

Although the neurocrania are, for the most part, both perichondrally and endochondrally
ossified, the degree of ossification is greater posteriorly. Thus the greatest thickness of
endochondral bone occurs around the tip of the notochord (in the area of the prootic bridge) and
there is little endochondral bone in front of the basipterygoid process ventrally or the pineal
foramen dorsally. Well-developed endochondral bone is confined to the postorbital portion of
the braincase and that area around the hypophyseal foramen (including the basipterygoid
process). In this respect it is interesting to note that ossification of the osteichthyan skull
commences around the notochordal plate and proceeds anteriorly. The whole of the preorbital
regions of the neurocrania of Mimia and Moythomasia are only perichondrally ossified, while in
front of the basipterygoid processes up to the anterior limits of the orbits endochondral bone can
only be recognized in thin section as mere wisps of tissue in the more lateral areas of the
neurocranial ossifications. In presumed younger individuals the area anterior to the nasal
capsules is often not even perichondrally ossified; only in a few mature individuals does a
complete layer of perichondral bone occur.

The anterior region of the neurocrania of Mimia and Moythomasia is thus only perichondrally
ossified. In this respect they resemble some of the pholidophorid and leptolepid neurocrania
described by Patterson (1975: 473), of which he remarked 'the ethmoid region is often missing
from the fossils, and when preserved is as a rule less thoroughly ossified'; he added that the
ethmoid area is frequently only perichondrally ossified. Although the ethmoid region is as a rule
less thoroughly ossified in primitive actinopterygians (cf. Pteronisculus Nielsen 1942, Birgeria
Nielsen 1949, Kansasiella Poplin 1974), in the majority of other fossil osteichthyans this region is
invariably endochondrally ossified (Nesides Stensio 19326, Eusthenopteron Jarvik 1954,
Glyptolepis Jarvik 1972, Chirodipterus Save-Soderbergh 1952, Griphognathus Miles 1977).

RELATIONSHIPS OF PALAEONISCIDS 183

From what evidence is available it appears that the neurocranium of chondrichthyans,
acanthodians and placoderms is composed solely of perichondral bone. If the Actinopterygii are
the sister-group of the Sarcopterygii (Rosen et al. 1981) then the absence of endochondral bone
in the snout of early actinopterygians is probably a primitive feature.

The endochondral bone, where it occurs in the neurocrania of Mimia and Moythomasia, is
thick and cancellate with large medullary spaces surrounded by delicate laminae, and with
external and internal surfaces and all canals for vessels or nerves lined with thin, laminate,
perichondral bone which shows few traces of radial structures; consequently it is difficult to
deduce individual ossification centres. In this they resemble many fossil actinopterygians
including palaeoniscids, perleidids, pholidopleurids, parasemionotids (Nielsen 1942, 1949) and
pholidophorids (Patterson 1975: 288).

While separate ossifications are not evident in the neurocrania of many primitive
actinopterygians, there are several notable exceptions. Discrete ossifications have been
described in the palaeoniscids Cosmoptychius (Watson 1928: 49), Pteronisculus magnus (Nielsen
1942) and Birgeria (Stensio 1921, Nielsen 1949), in Perleidus cf. stoschiensis (Patterson 1975:
456), and in the living Polyptems, Acipenser and Polyodon (see Patterson 1975: 463 for
summary). Elsewhere within the actinopterygians ossification centres are also recognizable in
parasemionotids (Lehman 1952: 162, Patterson 1975: 432), caturids such as ' Aspidorhynchus'
(Rayner 1948: 315, Patterson 1975: 436), Macrepistius (Schaeffer 1971) and Caturus furcatus
(Patterson 1975: 441), and in the amiids Enneles (Santos 1960), Sinamia (Stensio 1935) and the
extant Amia. They have also been described in the semionotids Lepidotes (Woodward 1916,
Patterson 1975: 449) and the living Lepisosteus , in the pachycormids Pachycormus (Rayner
1948, Patterson 1975: 443), Hypsocormus (Stensio 1935) and Protosphyraena (Loomis 1900,
Lehman 1949), in the Mesozoic pholidophorids and leptolepids (Patterson 1975) and in most
Recent teleosts.

In all the specimens of Mimia and Moythomasia examined the braincase was fully ossified and
sutureless, yet some specimens of Moythomasia are only a third the size of others. Patterson
(1975: 287) observed a similar size discrepancy in the ossified neurocrania of Pholidophoms
bechei, so presumably as in Pholidophorus ossification in Mimia and Moythomasia must have
set in early in ontogeny and growth have been terminated by fusion of the bones.

However, the absence of separate ossifications in the neurocrania of Devonian
actinopterygians and most of the known primitive fossil actinopterygians, and the presence of
separate bones in later actinopterygians, can be accounted for by two hypotheses. The
neurocranium may have ossified primitively in a single piece and in subsequent evolution
have fragmented into several ossification centres, independently in different lines (Stensio
19326: 297). Alternatively the neurocranium may always have ossified from a discrete set of
centres, in more primitive forms growth being thus terminated by fusion of the constituent
bones, whereas in others the sutures remained open to allow persistent growth, as in living
teleosts (Gardiner 1960: 359, Schaeffer 1971: 21, Patterson 1975: 288). Elsewhere discrete
endocranial bones are found in Acanthodes (Miles 1971a), Ctenurella (Miles & Young 1977), all
post-Devonian actinistians (and possibly even Nesides in which the back of the skull is missing),
Eusthenopteron (Stensio 19326: 297, Jarvik 1972) and tetrapods. Separate bones have also been
described in the Devonian dipnoan Dipnorhynchus (Campbell & Barwick 1982), but with the
exception of the exoccipitals of Neoceratodus no separate ossifications have ever been noted in
Recent dipnoan neurocrania. In shark neurocrania perichondral bone develops at the surface of
the individual tesserae (Kemp & Westrin 1979) and there are therefore numerous ossification
centres.

If the phylogenetic fragmentation hypothesis is correct then fragmentation must have
occurred on at least five separate occasions: within the chondrichthyans, within the
acanthodians, within the placoderms, within the actinopterygians and once again within the
remaining osteichthyans. This hypothesis is thus extravagant, and since the endochondral bone
of the braincase in all Recent gnathostomes grows from a suite of fixed centres it is difficult to
believe that in the braincase of Devonian fishes endochondral (and perichondral for that matter)
bone grew in some as yet undescribed fashion. Consequently, I am forced to continue to support

184 B. G. GARDINER

the non-fragmentation hypothesis of braincase ossification, in which it is proposed that the
neurocranium always ossified from a discrete set of centres.

Unfortunately, this hypothesis has been further complicated in actinopterygians by two
additional, conflicting hypotheses, either that there has been an increase of ossification centres
('fragmentation'), or that there has been an decrease ('fusion', loss). But, as Patterson (1975:
470) points out, most of the evidence in support of a subdivided neurocranium being the derived
condition rests on the assumption that teleosts possess the most highly subdivided neurocrania.
Patterson (1975: 470) has however clearly demonstrated that the braincase of living teleosts
contains fewer endochondral bones than the more primitive pholidophorids, and they certainly
contain fewer bones than early pachycormids (personal observation). From his survey of
braincases Patterson (1975) has shown that 'the dominant process in actinopterygian evolution
has been reduction in the number of endocranial ossifications, not increase . . . examples of loss
being numerous especially in the pholidophorids and teleosts'.

Other evidence for the fragmentation hypothesis rests on the observations that in Polypterus
and in palaeoniscids, where ossification patterns are known (Cosmoptychius and Birgeria),
there are fewer bones than in teleosts and many halecomorphs, while the Permian Acanthodes
has a pattern of neurocranial bones similar to Cosmoptychius. Patterson (1975: 465) has
demonstrated that the conditions in Polypterus and Birgeria are similar and that they both differ
considerably from the palaeoniscid type from which they can most reasonably be derived by
assuming loss of ossifications.

Cosmoptychius (Watson 1928, Schaeffer 1971) on the other hand is known from but a single,
incomplete specimen. It is similar in many respects to some of the smaller skulls of Moythomasia
and, as Patterson (1975: 402) commented, is probably composed of the same series of bones as in
most palaeoniscids. The fact that there is but one large paired ossification, one small median
ossification and two pairs of smaller ossifications in this specimen is not evidence that the
braincase ossified from the same number of centres. Thus, the occipital ossification probably
included basi- and exoccipitals as well as an intercalar , and the upper part of the occiput, which is
missing from the specimen, must have included epioccipitals.

Finally the pattern of ossification in Acanthodes is superficially actinopterygian-like and said
to resemble that in Cosmoptychius (Schaeffer 1971, Miles 1973a). But the braincase of
Acanthodes (Miles 1971a: fig. 4-7) differs from most actinopterygians in not possessing either
epioccipitals or prootics and in not having endochondral ossification. It is also possible (Denison
1979) that the condition in Acanthodes (the last of the acanthodians) is not primitive for
acanthodians and that Acanthodes, like the living chondrosteans and Polypterus, may have lost
several ossification centres.

The conclusion which may be drawn from this discussion is that the hypothesis of loss of
neurocranial ossification centres has more to support it than the conflicting hypothesis of
increase of centres by fragmentation. It follows from the acceptance of this conclusion that the
common ancestor of the gnathostomes possessed a neurocranium which ossified from a number
of centres and that the number of centres in chondrichthyans is far greater than in the rest of the
gnathostomes. Furthermore the pattern of ossification seen in chondrichthyans is different from
that in other gnathostomes. In chondrichthyans the prismatic calcifications remain tesserate
throughout life and the cap zone of the individual tesserae forms a thin veneer of perichondral
bone; growth is accomplished by enlargement of the individual prisms (Kemp & Westrin 1979).
In osteichthyans the ossification centres are far fewer, ossification commences as a disc on the
surface of the cartilage model and the sutures may remain open so that growth is continuous.

Within the various non-chondrichthyan lineages loss of neurocranial ossification centres has
occurred independently, maybe in relation to varying mechanical factors such as neurokinesis
etc., and until the phylogenetic homologies have been worked out for each group we can only
presume that topographically similar bones in the neurocrania of acanthodians, actinoptery-
gians, actinistians, rhipidistians and tetrapods are not necessarily homologous.

Miles (1977) has recently argued that the presence of endochondral bone in the snout of the
Devonian dipnoan Griphognthus is primitive in relation to the condition in other Devonian
dipnoans such as Chirodipterus and Holodipterus , where endochondral bone is restricted to the

RELATIONSHIPS OF PALAEONISCIDS 185

otic and occipital regions. But if, as argued above, endochondral bone is primitively absent from
the snout of actinopterygians, and the actinopterygians are the sister-group of the remaining
osteichthyans (Gardiner 1973, Rosen et al. 1981), then it is more likely that the condition of the
snout of Griphognathus is specialized in respect to Chirodipterus and Holodiptems.

In order to substantiate the argument that endochondral bone is primitively absent from the
snout of actinopterygians it is necessary to re-examine the occurrence of endochondral bone in
early vertebrates.

Some authors (Miles 1977, Schaeffer 1971) consider perichondral and endochondral bone to
be of equal antiquity and also primitive for all bony vertebrates. There is little doubt that
perichondral bone, like dermal bone, is a primitive vertebrate tissue since both occur in
cephalaspids (Stensio 1927, 19320), placoderms (Stensio I963a,b), acanthodians, bony fishes
and tetrapods. On the other hand endochondral bone has a more limited distribution. In the
whole of the Agnatha the only record of endochondral bone is in the cephalaspid Boreaspis
(Wangsjo 1952: fig. 1). Re-examination of Wangsjo's thin section of Boreaspis has convinced
me that this is merely an extensive perichondral ossification similar to that seen in thin sections
of other cephalaspids. It is worth noting that in mammals growth of a perichondral ossification
may result in replacement of the entire cartilage that it envelops, and that in very weakly ossified
neurocrania, such as that of the chondrostean Polyodon, the bones consist merely of scale-like
perichondral ossifications.

Through the kindness of Dr G. Young I have had the opportunity of examining several
neurocrania of Devonian placoderms from Australia including Brindabellaspis stensioi Young,
Buchanosteus confertituberculatus (Chapman) and Wijdeaspis warrooensis Young; all show
extensive perichondral ossifications but none shows any trace of endochondral bone. Recently
Miles & Young (1977: 168; fig. 21) have stated that in one specimen of the ptyctodontid
Ctenurella gardineri Miles & Young endochondral bone is present in the ethmoidal bone.
Re-examination of this specimen (BMNH P. 57665) convinces me that what they termed
endochondral bone is probably calcified cartilage. If placoderms lack endochondral bone and
are more closely related to chondrichthyans than to other gnathostomes (Miles & Young 1977),
or are the sister-group of osteichthyans, then endochondral bone must be considered a
specialization of osteichthyans, as Rosen et al. (1981) presumed, and the absence of
endochondral bone from the snout of Devonian actinopterygians and dipnoans is a primitive
character.

Occipital region

The neurocrania of Mimia and Moythomasia are ossified in a single piece as in the palaeoniscids
Boreosomus (Nielsen 1942), Pteronisculus macropterus (Beltan 1968) and Kansasiella (Poplin
1974); consequently individual bones are not apparent and it is not easy to delimit regions,
except in a very general way. Here the occipital region is taken to be that area of braincase
behind the occipital fissure (fissura otico-occipitalis) and below the vestibular fontanelle and
extending anteriorly as far as the ventral otic fissure (fissura oticalis ventralis).

Mimia toombsi

Dorsally the occipital region is separated from the otic by a small, oval posterior dorsal
fontanelle (PL 1; pdf, Figs 26, 79). This represents the expanded dorsal portion of the occipital
fissure. The posterior dorsal fontanelle is smaller than that in Kansasiella (Poplin 1974: fig. 12),
Pteronisculus (Nielsen 1942: fig. 7), Australosomus (Nielsen 1949: fig. 3) and Kentuckia
(Rayner 1951: fig. 6), and is lined throughout with perichondral bone. Although this fontanelle
is also perichondrally lined in Australosomus (Nielsen 1949: 27) there is no perichondral lining in
Pteronisculus (Nielsen 1942: 41) and presumably it was cartilage-filled. The fontanelle is closed
by bone in some specimens of Boreosomus whereas in others it remains open (Nielsen 1942:
287).

The occipital fissure (fotc, Figs 4, 5, 6, 12, 13, 15, 50), which represents the persistent metotic
fissure between the occipital arch and otic capsule of the embryo, passes anteroventrally from

186

B. G. GARDINER

18 17 16 15

Fig. 1 Mimia toombsi Gardiner & Bartram. Neurocranium and attached dermal bones in posterior
view, basisphenoid displaced ventrally; from BMNH P. 56504. Key (diagram below): 1, ivl; 2, Pa;
3, oimsj; 4, 11; 5, St; 6, oahm + oaop; 7, apse; 8, fhm; 9, ghm.VII; 10, jc; 11, focn; 12, vfon; 13,
oims2; 14, foca; 15, Psp; 16, alig; 17; cao; 18, fica; 19, gona; 20, bpt; 21, aip 1; 22, not 23, goa; 24,
fm; 25, X; 26, por; 27, fotc; 28, oatm; 29, pdf. For explanation of lettering used on text-figures, see
pp. 177-181.

the posterior dorsal fontanelle to terminate in an ovoid vestibular fontanelle. The vestibular
fontanelle (vfon, Figs 1, 5, 13, 14, 15, 50) has no perichondral lining and was presumably
cartilage-filled. It represents an area between adjacent ossifications (opisthotic, basioccipital
and prootic ossifications deduced from Patterson 1975: 461) and is one of the few areas of the
braincase which remains unossified in the adult early osteichthyan.

gboca

fboca

not

RELATIONSHIPS OF PALAEONISCIDS

ivl

187

oatm

foca

crsp

01 ms

focn

oims,

vfon

aip 1

cao

Fig. 2 Mimia toombsi Gardiner & Bartram. Occipital ossification in posterior view, from BMNH

P.53243.

fboca

oims

oatm

gboca

oims

fboca

focn

2mm

Fig. 3 Mimia toombsi Gardiner & Bartram. Dorsal part of the occipital ossification in posterior

view, from BMNH P. 56496.

188

B. G. GARDINER

The vestibular fontanelle of Mimia is somewhat smaller than in Kansasiella (Poplin 1974: fig.
13) and much smaller than in either Kentuckia (Rayner 1951: fig. 7) or Pteronisculus (Nielsen
1942: fig. 4). In size and shape it is quite similar to that in Pholidophorus bechei (Patterson 1975:
fig. 56).

In early, less specialized, palaeoniscids the occipital fissure is complete and perichondrally
lined throughout, except in Moythomasia (see p. 201 and Figs 8, 9), where a small area above
the upper margins of the vagus canal lacks perichondral bone, and in some individuals of
Boreosomus piveteaui (Nielsen 1942: fig. 59) where the mid-dorsal part of the fissure may be
closed by a narrow bridge of bone. In other later, more advanced palaeoniscids such as Birgeria
(Nielsen 1949: 190) the perichondral lining is missing and the fissure is presumably already
cartilage-filled.

Anteriorly the occipital region in Mimia is bounded by another, completely separate fissure,
the ventral otic fissure. This fissure (fv, Figs 13, 14, 15, 16, 22, 26, 50) lies in the floor of the
neurocranium and passes up immediately behind the foramen for the pituitary vein and anterior
to the foramen for the abducens nerve. The fissure is not perichondrally lined and was
cartilage-filled in life; it represents the cartilage remaining between ossifications in the
trabeculae (+ polar cartilages) and parachordals, and must represent the gap between the
chordal and prechordal skeleton of the embryo. The fissure separates the basioccipital from the
basisphenoid (ossifications deduced from Patterson, 1975) in the mid-line and from theprootics

fst IX gst X

oaop

fhm

fotc

gboca

oims

focn

fboca

goa

gphX

foca

Fig. 4 Mimia toombsi Gardiner & Bartram. Otic and occipital regions in right lateral view, from

BMNH P. 56501.

RELATIONSHIPS OF PALAEONISCIDS 189

dorsolaterally. From the floor of the neurocranium the fissure passes up behind the 'prootic'
bridge to open into the front of the notochordal canal (not, Figs 25, 26). Laterally, at the level of
the presumed junction between the basioccipital and prootics, the ventral otic fissure gives way
on either side to a large foramen for the orbitonasal artery (fona, Figs 15, 50).

In the Gogo palaeoniscids the ventral otic fissure is therefore situated below the hind wall of
the orbit and well in front of the vestibular fontanelle , from which it is separated by the bony wall
of the neurocranium (the prootic and basioccipital) . The ventral otic fissure is also separated
from the vestibular fontanelle in other palaeoniscids such as Kansasiella (Poplin 1974: figs 13,
14), Pteronisculus macroptems (Beltan 1968: fig. 2) and Boreosomus (Nielsen 1942: fig. 63).

Although the limits of the individual ossifications making up the occipital region cannot be
made out with certainty because in most cases individual sutures are absent, in some specimens
and in certain areas ossification centres can be determined with considerable confidence. In all
specimens of the Gogo palaeoniscids the floor and roof of the notochordral canal (and
neurocranial floor) are incompletely ossified in the mid-line posteriorly, and in one specimen of
Mimia (BMNH P. 53250) there is a median suture running the whole length of the floor of the
notochordal canal from the ventral otic fissure to the posterior limit of the neurocranium. This
obviously paired ossification in the floor of the braincase must be the basioccipital. Elsewhere in
osteichthyans (including tetrapods) the basioccipital is usually a median ossification (except
perhaps for Polypterus and Pachycormus; Patterson 1975: 445, 448), but its origin must have
been from paired ossification centres in the parachordal cartilages. Interestingly the
basioccipital ofAcanthodes (Miles 1973a: fig. 11) is incised in the mid-line both anteriorly and
posteriorly and this is taken to indicate the paired origin of that perichondral bone. In front of
the foramen magnum the basioccipital forms the floor of the neurocranium, the floor and lateral
walls of the notochordal canal (see Fig. 25) and the ventral and posterior margins of the
vestibular fontanelle. Anterolaterally the basioccipital passes indistinguishably into the
prootics. The basioccipital fails to meet in the mid-line posteriorly beneath the notochordal
canal in several other palaeoniscids, thus the notochordal canal is contiguous with the aortic
canal in Pteronisculus (Nielsen 1942: 32), Kansasiella (Poplin 1974: figs 11, 20), Boreosomus
(Nielsen 1942: fig. 61) and Cosmoptychius (Schaeffer 1971).

The roof of the notochordal canal and the floor of the foramen magnum is made up by another
pair of ossifications, the exoccipitals. This can be recognized because the ossification centres for
these bones in other osteichthyans lie lateral and somewhat ventral to the floor of the foramen
magnum, and because, although the foramen magnum is completely lined with bone dorsally
and laterally, the ossifications do not always meet in the mid-like posteroventrally (see
Moythomasia, Figs 8, 9, 10). A posterior notch in the floor of the foramen magnum in
Pteronisculus (Nielsen 1942: 32) and Kansasiella (Poplin 1974: fig. 14) represents this unossified
area in the Gogo palaeoniscids.

The upper margin of the occipital arch is occasionally produced anteriorly, partly closing the
fontanelle, and this may represent a median supraoccipital. The extent of the supraoccipital is
very uncertain in most specimens of Mimia, but can be more clearly discerned in Moythomasia
(see Fig. 10). While there is no evidence of a supraoccipital in Pteronisculus (Nielsen 1942: fig.
7), Kentuckia (Rayner 1951: fig. 6) and Australosomus (Nielsen 1949: fig. 3), a median
protuberance in Kansasiella (Poplin 1974: fig. 12; av.) and the closure of the posterior dorsal
fontanelle in Boreosomus (Nielsen 1942: 286) support the idea that such a bone existed in these
last-named forms at least. Elsewhere in primitive actinopterygians a supraoccipital is possibly
present in Perleidus (Patterson 1975: 456), where its presence is inferred from the mode of
closure of the uppermost part of the occipital fissure (posterior dorsal fontanelle).

The occipital region occupies less than 15% of the total neurocranial length measured through
the vagus foramen and in this respect Mimia is similar to Pteronisculus (Nielsen 1942: fig. 4) and
Kansasiella (Poplin 1974: fig. 13). The occipital region is much shorter in the pholidopleurid
Australosomus (Nielsen 1949: fig. 2) and in the primitive teleost Pholidophorus (Patterson 1975:
fig. 56). The occipital region is not as deep as the rest of the neurocranium, the greatest depth
being attained immediately in front of the ventral otic fissure, much as in Boreosomus (Nielsen
1942: fig. 62).

190

oaop

B. G. GARDINER

gst x

fotc

fst IX

oahrn +
oaop

fhm

focn

goa

fboca

foca

Fig. 5 Mimia toombsi Gardiner & Bartram. Otic and occipital regions in right lateral view, from

BMNH P.53234.

The posterior face of the occiput (Figs 1, 2, 3) is broader above the foramen magnum than
below, and the dorsal margin is gently rounded. The greatest width is at the level of the dorsal
margin of the formen magnum where the outline is more acutely rounded before the bone turns
increasingly forwards and inwards to merge into the ventral surface. An acutely rounded
dorsolateral prominence (Fig. 2) is assumed to be made up by the intercalar. This lies
immediately posterodorsal to the vagus foramen and has been termed the cranio-spinal process
by Nielsen (1942: 38). It is homologous with the cranio-spinal process in other palaeoniscids and
chondrosteans such as Saurichthys, Acipenser and Polydon. Patterson (1975: 315) has shown
that the intercalar in Pholidophorus is the homologue of the cranio-spinal process of
chondrosteans but here, as in halecostomes, it received the ligament from the ventral limb of the
post-temporal. No such ventral limb existed in palaeoniscids and there is no evidence of a
ligament; there is likewise no such limb in chondrosteans though an extension from the
post-temporal in Acipenser reaches the occipital process posterior to the vagal foramen (Jollie
1980: 240). (A limb does exist in Polypterus but this is inferred to have arisen independently of
that in neopterygians). The cranio-spinal process is also found in acanthodians (Miles 1973a: 86)
and placoderms (Stensio 1969; Young 1980: fig. 8, infravagal process) and may be a primitive
gnathostome attribute. Miles (1977: 55) has suggested that either the transverse process in
dipnoans may be homologous with the cranio-spinal process, or it may be serially homologous

RELATIONSHIPS OF PALAEONISCIDS

frla,

191

gst X

oaop

oahm

fotc

fhm

gboca

focn

goa

f boca

2mm

pamp

vfon

Fig. 6 Mimia toombsi Gardiner & Bartram. Dorsal portion of the otic and occipital regions in

right lateral view, from BMNH P.56496.

with it if the cranio-spinal process represents an epineural. The first suggestion is unlikely since
the intercalar lies in front of the first intermuscular septum in Mimia and forms part of the
posterior margin of the occipital fissure as well as the hind margin of the foramen for the vagus
nerve, whereas in fossil dipnoans the transverse process lies behind the third occipital nerve and
so is quite distant from the vagus foramen and the occipital fissure. Separate epineurals are
synapomorphous for teleosts. However, epineural processes are present on the anterior
vertebrae of Griphognathus where they appear to ossify independently (Rosen et al. 1981: fig.
54 A). Furthermore in Neoceratodus the cartilaginous transverse process serves as a boss for the
articulation of a cranial rib. Thus the dipnoan transverse process seems more likely to be
homologous with an epineural process than with the actinopterygian cranio-spinal process.

The foramen magnum is somewhat higher than it is broad (fm, Figs 1,2,3) and much smaller
than the entrance to the notochordal canal. The notochordal canal (not, Figs 1, 2, 26) is ovoid in
section, a little broader than high and with its long axis extending through the whole of the otic
region and terminating in the top of the ventral otic fissure (Figs 16, 20, 26). The width of the
notochordal canal diminishes rapidly in an anterior direction (Fig. 26) but then remains the same
width up to the fissure. There is no real occipital condyle and the notochordal canal is not lined
with a cone of perichordal tissue as in pholidophorids, leptolepids and Recent teleosts
(Patterson 1975: 318).

Beneath the notochordal canal lies the aortic canal (cao, Figs 1,2, 14, 15, 16, 25, 26, 50) which

192 B. G. GARDINER

is almost circular in section. The aortic canal leads forwards and downwards to open on the
ventral surface on a level with the vestibular fontanelles. The walls of the canal are smooth,
perichondrally lined and formed by the basioccipital. The aortic canal narrows somewhat
anteriorly, then finally widens into a half-funnel shaped opening with two well-marked grooves
diverging from the lateral margins of its mouth. These grooves (gla, Fig. 14) mark the paths of
the lateral aortae. Just anterior to the point of bifurcation of the aorta (in BMNH P. 53259) there
is a small longitudinal crest which marks the point of attachment of the aortic ligament (alig, Figs
14, 50). In other specimens of Mimia and in Moythomasia this crest is replaced by a small
protuberance (Gardiner 1973: fig. 2; sup) in the same position or occasionally placed more
anteriorly, that is more distant from the point of bifurcation. The dorsal aorta bifurcated level
with the anterior end of the vestibular fontanelle.

In one specimen of Mimia (BMNH P. 53259, Fig. 15) the aortic canal itself bifurcates
anteriorly, as in Polypterus, and here the aortic ligament must have originated on the pillar
which divides the canal. In this specimen the point of bifurcation lies behind the level of the
vestibular fontanelle. In Kamasiella (Poplin 1974: fig. 14) the aortic canal is much longer than in
Mimia; it bifurcates anteriorly (as in P. 53259), but additionally has a paired opening in the floor
of the canal for the exit of the second efferent arteries. The aortic canal is also very long in
Boreosomus (Nielsen 1942: fig. 63) but here both the anterior opening and the opening in the
floor of the canal (for the second efferents) are unpaired. In Pteronisculus (Nielsen 1942: fig. 6)
and Kentuckia (Rayner 1951: fig. 4) the canal is long, the anterior opening bifurcated, but the
opening in the floor single. Cosmoptychius (Schaeffer 1971: fig. 8) has a long canal bifurcated
anteriorly with no apparent ventral opening.

In pholidopleurids such as Australosomus the aortic canal is much shorter, but there is still a
well-marked median peg-like process for the attachment of the aortic ligament (Nielsen 1949:
figs 5, 6). A similar process (alig, Fig. 7 and see p. 201) can be seen in Moythomasia although the
aortic canal is much longer.

The wall of the aortic canal is perforated by a large dorsolaterally-dijected canal which
transmitted the occipital artery (foca, Figs 1, 2, 13, 14, 15). In most specimens there is a single
foramen on either side, but in some (BMNH P. 53250 for example) there is a single opening for
the occipital artery on one side and a double opening on the other, while in other specimens the
canal is double on both sides (foca, Figs 4, 5). Allis (1922: 207) described a blind canal in the
basi-exoccipital of Polypterus in the position of the occipital artery in Amia and Mimia, but he
also showed how the succeeding two intervertebral arteries in Polypterus arise as a single artery
from the dorsal aorta, then branch into two, the anterior branch passing up (presumably) in
relation to the intermuscular septum and the more posterior passing into the cranial cavity
through the occipital nerve foramen.

Elsewhere within palaeoniscids and pholidopleurids there is always a dorsolaterally-
directed canal for the occipital artery, single in Pteronisculus (Nielsen 1942: fig. 9), Kansasiella
(Poplin 1974: fig. 20) and Australosomus (Nielsen 1949: fig. 4) but double in Kentuckia (Rayner
1951: figs 4, 7). Although there are two openings on either side for the occipital arteries in
Parasemionotus (Lehman 1952: fig. 3), in the majority of more advanced actinopterygians the
occipital artery, where it occurs, issues through a single opening on either side of the aortic canal
or groove (Lepidotes Patterson 1975: fig. 109; Dapedium Patterson 1975: fig. 113; Amia Allis
1897: 706; Pachycormus Patterson 1975: fig. 106; pholidophorids Patterson 1975: 320).

At the anterior end of the basioccipital, a small paired perichondrally-lined canal passes
anteroventrally down through the bone from a point in the floor of the neurocranium (VI2, Fig.
25) at the level of the anterior end of the zygal plate (see below, p. 194) to open ventrally in the
floor of the orbit dorsolateral to the ventral otic fissure (Figs 16, 20) . As it passes into the floor of
the orbit this canal enters the ventral edge of the prootic (this can be inferred from BMNH
P. 53245 in which the perichondral margin of the prootic just encloses the anterior limit of the
canal). This canal must have transmitted the abducens nerve (VI, Fig. 29). In one specimen
(BMNH P. 53234, Fig. 25) the canal is forked dorsally within the basioccipital and each branch is
continuous with a short canal in the zygal plate. The cranial entrance to these canals is a pair of
oval foramina on the medial surface of each zygal (a similar pair of foramina is also present in

194

B. G. GARDINER

Soc

focn

fboca

gpcv

g X

not

cs im

mvfon

2mm

Fig. 8 Moythomasia durgaringa Gardiner & Bartram. Occipital region in anterodorsal view, looking
into the rear of the vestibular fontanelle from the left side, from BMNH P. 53221.

BMNH P. 53249). In all living actinopterygians the root of the abducens nerve is double, but in
this particular specimen (BMNH P. 53234) of Mimia the two roots remained separate until they
entered the basioccipital. In all other specimens examined the canal for the abducens nerve
opens below the zygal plate. However, when there is a single short canal in the posterior region
of the zygal, it is always in direct line with the internal opening of the abducens canal. Rayner
(1951: fig. 10) suggested that a similar foramen in the zygal plate of Watson's palaeoniscid A
served to transmit the abducens nerve. In Polypterus (Allis 1922: 228) the canal for the abducens
opens in the floor of the cranial cavity immediately posterior to the base of the cartilaginous
prootic bridge and runs anteriorly in the cartilage between the basi-exoccipital and
basisphenoid, to open by a notch in the edge of the basisphenoid immediately below the

RELATIONSHIPS OF PALAEONISCIDS

195

trigeminal foramen. Thus the relationship of the abducens is similar in Mimia and Polypterus
except that in the latter the prootic ossification is absent. In other palaeoniscids and
pholidopleurids the abducens nerve never passes through the basioccipital. Instead it always
leaves the cranial cavity in front of the ventral otic fissure, usually through the prootic bridge
(Pteronisculus Nielsen 1942: figs 9, 10; Kentuckia Rayner 1951: fig. 8, Kansasiella Poplin 1974:
fig. 23;Australosomus Nielsen 1949: fig. 7). In halecostomes such asAmia, pholidophorids and
most Recent teleosts the abducens nerve passes through the prootic bridge (formed by the
prootics into the roof of the myodome. Primitively in actinopterygians the abducens nerve is
deduced to have passed through the basioccipital for part of its course, but with rearward
migration of the ventral otic fissure this relationship was lost.

The zygal plates arise from the anterior ends of the walls of the notochordal canal and are
more or less contiguous with the underlying basioccipital. They rise upwards and outwards at an
angle of perhaps 70° and are occasionally joined in the mid-line ventrally by a thin strut of
endochondral bone (in the roof of the notochordal canal). In BMNH P. 53249, in which much of
the internal, perichondral margin of the prootic has not yet fused with adjacent bones (Pro, Fig.
25), the separate nature of the paired zygals can be recognized. They are delicate ossifications
completely covered in perichondral bone except along their ventral margins where they are
partially fused to be basioccipital. Sometimes there is a small posterior foramen in each zygal,
occasionally two foramina, one anterior and one posterior (VIl5 VI2, Fig. 25; see also Fig. 26).
These foramina transmitted the abducens nerve from the floor of the brain into the adjacent
canal in the basioccipital. Dorsally each zygal has a distinct notch in its margin; this presumably
served for the passage of the auditory nerve from the brain to the otic capsule. That these plates
really are separate ossifications can be inferred from the limits of their perichondral covering
and from the nature of their endochondral core, which is made up of very small units quite

not

mvfon

f boca

Soc

Fig. 9 Moythomasia durgaringa Gardiner & Bartram. Occipital region in antero-dorsolateral
view, looking into the floor and rear of the vestibular fontanelle from BMNH P. 53227.

196

B. G. GARDINER

dissimilar from the large 'bubbles' of bone making up the underlying basioccipital. These plates
form the inner wall of the pocket that housed the sacculus, separating it from the floor of the
brain. In Polypterus a cartilage in a similar position performs the same function. The
relationships of these plates to the sacculus can best be seen in a reconstruction of an endocranial
cast of Kansasiella (Poplin 1974: fig. 23). Zygal plates have previously been recorded in only one
other actinopterygian, Pteronisculus (Nielsen 1942: fig. 9, om; Bjerring 1971: fig. 6), but they
are also recognizable in Kentuckia (Rayner 1951: 70 - median projection) and Kansasiella
(Poplin 1974: fig. 20, t).

In Pteronisculus (Nielsen 1942: 53) the displaced zygal plate is obviously a separate
ossification since it has no connection with any other bone. Presumably it formed above the
notochord and occupied a similar position in life to that in Mimia. It is a small, horizontal,
bilaterally symmetrical plate, devoid of perichondral lining ventrally. In both Kentuckia and
Kansasiella the zygal plate is shown as a median ossification in the roof of the notochordal canal
and is inseparable from the underlying basioccipital, but produced dorsolaterally into
perichondrally-lined wings, very similar to the paired zygal plates in Mimia. From these four
examples it is not possible to decide whether paired or unpaired zygals represent the more
primitive condition, but certainly the presence of zygals is primitive (see p. 207) both for
actinopterygians and osteichthyans.

Externally the surface of the occipital region around the notochordal canal and foramen
magnum is marked by two distinct parallel ridges which run vertically up the lateral walls of the
neurocranium. The more posterior ridge starts on the aortic canal just behind the foramen for
the occipital artery and runs in a more or less uninterrupted line to fade out dorsally on a level
with the middle of the foramen magnum, posterodorsal to the foramen for the occipital nerve.
This ridge marks the origin of the second intermuscular septum (oims2 , Figs 1,2,3,4). The more
anterior of the two ridges also commences on the aortic canal. It is considerably more elevated
than the posterior ridge and continues further dorsally and finally peters out well above the
foramen magnum. The first intermuscular septum presumably originated (oims!, Figs 1, 2, 3, 4)
on this ridge.

In those specimens with only one opening for the occipital artery (BMNH P. 53243, Fig. 2) a
groove leads anterodorsally from that foramen towards the foramen for the occipital nerve. A
little below the occipital nerve foramen the groove bifurcates and the anterior branch proceeds
almost horizontally through a gap in the ridge for the first intermuscular septum to pass

gpcv

focn

Soc

2mm

Fig. 10 Moythomasia durgaringa Gardiner & Bartram. Dorsal part of the occipital ossification in

anterodorsal view, from BMNH P.53221.

RELATIONSHIPS OF PALAEONISCIDS 197

immediately into the mouth of a small foramen (fboca, Fig. 2) which gives direct access to the
floor of the cranial cavity, anterior to the occipital nerve (see Moythomasia, fboca, Figs 8, 9, 10).
This short canal runs longitudinally through the bone, and clearly transmitted a blood vessel
rather than a nerve. From the disposition of the groove leading to it, the canal must have carried
a branch of the occipital artery into the rear of the cranial cavity. The other branch of the groove
passes up and turns anterodorsally just below the occipital nerve foramen, passes through a
more dorsal notch in the ridge for the first intermuscular septum (gboca, Figs 2, 3), and
continues as a groove onto the dorsal surface of the occiput. In one specimen (BMNH P. 56496),
and on one side only, this dorsal branch passes through a distinct foramen in the ridge (fboca,
Fig. 3) for the first intermuscular septum. This dorsal branch must have supplied blood from the
occipital artery to the first trunk muscle.

Where there is a double opening for the occipital artery the more anterior opening always lies
in front of the ridge for the first intermuscular septum (BMNH P. 56501 and P. 53234, Figs 4, 5).
In these specimens a gutter runs directly from this anterior opening to the entrance of the canal
which passes into the cranial cavity (fboca, Figs 4, 5). There can be no doubt here that the canal
which opens into the cranial cavity transmitted a branch of the occipital artery. Moreover in
these specimens, as the canal for the branch of the occipital artery passes horizontally through
the wall of the neurocranium, it gives off another branch dorsally. This branch emerges on the
lateral wall on a level with the foramen for the occipital nerve and from its mouth an even more
distinct gutter (gboca, Figs 4, 6) runs up onto the dorsal occipital surface. This foramen and
gutter carried a branch of the occipital artery to the first trunk muscle, and presumably is
equivalent to the dorsal branch of the occipital artery in those forms where the occipital artery is
single.

In the inner wall of the foramen magnum there are two foramina, the more anterior of which
transmitted the branch of the occipital artery. The more posterior foramen is the larger and the
canal from it runs posterolaterally to open just in front of the ridge for the second intermuscular
septum. This canal transmitted an occipital nerve (focn, Figs 1, 2, 3, 4, 5, 6, 13). In the floor of
the foramen magnum, level with the foramen for the branch of the occipital artery, is a shallow
paired depression (see Moythomasia, csim, Fig. 8) but whether or not this can be regarded as an
incipient cavum sinus imparis depends on interpretation. A more obvious median depression is
seen in the palaeoniscid Pteronisculus (Nielsen 1942: fig. 5), but the true extent of the cavum
sinus imparis is perhaps only seen in teleosts (Patterson 1975: 316). If, however, a vascular
plexus did exist primitively in the floor of the foramen magnum then a direct arterial supply
would have been an advantage.

Remnants of this rather elaborate occipital arterial blood supply can be recognized in several
later forms. Patterson (1975: 292) has described in Pholidophorus a dorsolaterally-directed
canal which originates in the cranial cavity immediately in front of the occipital nerve and opens
on the dorsolateral surface of the exoccipital. This canal anastomoses (within the bone) with an
anteriorly-directed canal which opens in the upper part of the vagus canal. Patterson (ibid.)
suggested that the anterior branches carried a tributary of the posterior cerebral vein while the
dorsolaterally-directed canal carried both a vein and a branch of the occipital nerve. There
seems little doubt that the dorsolaterally-directed canal is homologous with a similar canal in
Mimia which, as shown above, served for a branch of the occipital artery. On the other hand in
Pteronisculus (Nielsen 1942: 38, fig. 5) there is a canal running from the foramen magnum to the
hind wall of the vagus canal. This appears to be homologous with the anteriorly-directed canal in
Pholidophorus. Further, Nielsen (1942: 35) described a possible connection between this canal
and a more ventral canal which opens on the occipital surface near the occipital artery (Nielsen
1942: fig. 4, k). In one specimen of Mimia (BMNH P. 53245) there is, on one side only, a similar
connection within the bone between an anteriorly-directed canal which opens in the upper
part of the vagus canal and the canal carrying a branch of the occipital artery into the cranial
cavity. Thus it is likely that the anteriorly-directed canal in Pteronisculus and Pholidophorus
transmitted yet another branch of the occipital artery. Bjerring's (1971: fig. 6) suggestion that
this canal in Pteronisculus transmitted a hypothetical branch of the abducens nerve which
innervated the subcranial muscle is without foundation (see also Patterson 1975: 294).

198

B. G. GARDINER

dend rsoc

fapcv

gpcv

Fig. 11 Mimia toombsi Gardiner & Bartram. Left half of otic region of the neurocranium and
attached dermal bones in posterior view, from BMNH P. 53245.

A dorsolaterally-directed canal originating in the floor of the foramen magnum and opening
on the occipital roof is also present in parasemionotids (Patterson 1975: fig. 97), Caturus
(Patterson 1975: 319), 'Aspidorhynchus' (Patterson 1975: fig. 100) and Heterolepidotus
(Patterson 1975: fig. 104). Presumably in all these cases it carried a branch of the occipital artery.

Apart from Mimia (and perhaps Pteronisculm] the only other palaeoniscid in which a branch
of the occipital artery passes directly into the cranial cavity is Kansasiella (Poplin 1974: fig. 19,
spi).

A single occipital nerve canal as seen in Mimia is characteristic of most palaeoniscids
(Pteronisculus , Boreosomus, Kentuckia, Kansasiella), Australosomus , parasemionotids,
pholidophorids and leptolepids (Patterson 1975: 319) and is considered to be the primitive
condition for actinopterygians. Although this single occipital nerve corresponds to the first
occipital nerve of Polyptems, Acipenser, Polyodon, Lepisosteus and Amia, all these forms
have incorporated one or more neural arches and corresponding spino-occipital nerves into the
braincase. Similar incorporations are deduced to have occurred in Birgeria, Saurichthys,
Lepidotes and Dapedium (Patterson 1975: 319).

No real evidence for the position of intermuscular septa has previously been produced in
palaeoniscids. In fact the only previous record of septal position in fossil actinopterygians is
from pholidophorids. On the epioccipital of Pholidophorus germanicus (Patterson 1975: 297)
a posteriorly-projecting shelf of membrane bone marks the point of origin of the first
intermuscular septum, while just behind the external opening of the occipital nerve canal a large
trifid projection marks the origin of the second intermuscular septum, much as in Amia (Allis
1897) and Scomber (Allis 1903). These projections in Pholidophorus, Amia and Scomber are
homologous with the more complete ridges in Mimia.

The dorsal face of the occiput above the foramen magnum slopes gently upwards in the
mid-line, then at the level of the external opening of the occipital nerve canal it rises steeply to a
short median crest. The ridge so formed is not as pronounced as the so-called crista occipitalis of

RELATIONSHIPS OF PALAEONISCIDS

199

por

fotn

fst IX

fotc

dpsc

Fig. 12 Mimia toombsi Gardiner & Bartram. Left otic and occipital regions of the neurocranium in
dorsal view, from BMNH P. 53234. The broken line marks the limits of the lateral cranial canal.

Pteronisculus (Nielsen 1942: fig. 3) or Kansasiella (Poplin 1974: fig. 13). Dorsally, just below the
supraoccipital, the crest gives way to a triangular prominence which presumably served for the
insertion of the longitudinal intervertebral ligament (ivl, Figs 1, 2, 3). In one specimen (BMNH
P. 53245) the triangular prominence is missing and instead there is a prominent median ridge.
Elsewhere in fossil actinopterygians, a distinct cartilage-lined pit in the exoccipital region above
the foramen magnum is seen in parasemionotids (Patterson 1975: fig. 98) and in Dapedium
(Patterson 1975: fig. 113); this marks the insertion of the longitudinal ligament in those fishes.
On either side of the median crest in Mimia two distinct depressions mark areas of origin of the
anterior trunk muscle (oatm, Figs 1, 2, 3).

The canal for the vagus and the nervus lineae lateralis is represented by an inflated part of the
occipital fissure. The internal opening of the canal lies some distance above the floor of the
foramen magnum from whence it passes posterolaterally to exit immediately posteriorly to the
parampullary process. The posterior wall of the vagus canal is divided by a narrow ridge into two
roughly equal divisions. The upper division contained the posterior cerebral vein (gpcv, Fig. 11)
and the lower the vagus nerve (g.X, Fig. 11), as in Saurichthys (Stensio 1925: fig. 4),
Pteronisculus (Nielsen 1942: 39) and Pholidophorus (Patterson 1975: 293).

Moythomasia durgaringa

The occipital region in this species is very similar to that of Mimia and only the salient differences
will be noted.

The posterior dorsal fontanelle is perichondrally lined and larger than in Mimia (Gardiner
1973: fig. 7). The occipital fissure is more steeply inclined and the vestibular fontanelle much

RELATIONSHIPS OF PALAEONISCIDS 201

larger than in Mimia, nearer in size to that of Pteronisculus . The occipital fissure is
perichondrally lined except for a large oval area between the vagus canal and supraoccipital
(Figs 8, 9, 10).

The ventral otic fissure occupies a similar position to that in Mimia although it is not quite as
extensive dorsomedially (fv, Fig. 29). The occipital region occupies less than 17% of the total
neurocranial length (measured through the vagus foramen) and, unlike Mimia, is higher than
any other portion. A more clearly definable centre of ossification, presumably representing the
supraoccipital, can be recognized in some specimens (BMNH P. 53221) but the cranio-spinal
process is a little more ventral than in Mimia, level with the bottom of the foramen magnum. The
foramen magnum is marginally wider than high while notches in the hind ventral wall of the
notochordal canal similar to those in Mimia show ossification to be incomplete in these areas.
The aortic canal is similar in shape and size to that of Mimia but the peg-like process for the
attachment of the aortic ligament (alig, Fig. 7) is much more prominent and more posterior in
position, lying immediately in front of the anterior opening of the aortic canal. The wall of the
aortic canal is perforated by a single, dorsolaterally-directed canal for the occipital artery, but
even though the ridge for the insertion of the second intermuscular septum is not as distinct as in
Mimia, the course of the occipital artery is clear. A groove leads up from the foramen for the
occipital artery (BMNH P. 53221) towards the foramen for the occipital nerve, and just below
this latter foramen the groove branches. One branch passes anteriorly, leads through a gap in
the ridge for the first intermuscular septum, and immediately enters a small foramen from which
a canal runs horizontally inwards to open in the anterior floor of the foramen magnum (fboca,
Figs 8, 9, 10). As in Mimia, this canal must have transmitted a branch of the occipital artery into
the floor of the cranial cavity. The other branch of the groove passes anterodorsally through a
more dorsal gap in the ridge for the first intermuscular septum, then up towards the top of the
occiput. Each of these paired grooves terminates in a foramen on either side of the mid-line,
below the insertion of the intervertebral ligament. Each foramen connects, by means of a short
anteroventrally-directed canal, with the cranial cavity above the foramen magnum. In one
specimen (BMNH P. 51380, Gardiner 1973: fig. 7) the two grooves terminate in a single
foramen, set just off centre, below the insertion of the intervertebral ligament. These dorsal
grooves and foramina transmitted a second branch of the occipital artery into the dorsal half of
the cranial cavity. The only other osteichthyans in which similar dorsal openings have been
described are the Devonian dipnoans Griphognathus and Chirodipterus (Miles 1977: figs 12, 16,
nut) but in these forms the canals end blindly in the cranial bone.

In the floor of the foramen magnum, anterior to the opening for the branch of the occipital
artery, there is a shallow paired depression which possibly housed the cavum sinus imparis.
Normally, the floor of the foramen magnum posterior to this depression is unossified in the
mid-line (Fig. 8) but in one specimen (BMNH P. 56502) the floor is complete and raised in a
distinct median ridge which runs from the depression to the hind margin of the foramen
magnum. In some specimens a small foramen in the groove for the posterior cerebral vein
(BMNH P. 56502) could have transmitted another branch of the occipital artery. Dorsally,
below the supraoccipital, a well-marked triangular area with several small protuberances must
have served for the insertion of the intervertebral ligament.

Occipital region: discussion

1 . Posterior dorsal fontanelle. A distinct posterior dorsal fontanelle seems to be confined within
actinopterygians to the palaeoniscids and chondrosteans such as Acipenser (Bridge 1878),
although Patterson (1975: 307) has suggested that an imperfectly ossified area in front of the
supraoccipital in Pholidophorus bechei represents the remains of a small, paired posterior dorsal
fontanelle. Elsewhere a distinct posterior dorsal fontanelle is seen in the rhipidistian
Eusthenopteron (Jarvik 1954: figs 21B, C), where it appears as a crescent-shaped opening in the
roof of the supraotic cavity, and in Acanthodes (Miles 19730: fig. 3), where it is similar in shape
and size to that in Mimia. Whether this is a true dorsal fontanelle in Acanthodes or merely a
cartilaginous area devoid of perichondral ossification could not be determined. Miles (1977: 101)
has convincingly homologized the posterior dorsal fontanelle of Eusthenopteron with openings

202

B. G. GARDINER

acv

frd

fotc

foca

2mm

cao

Fig. 14 Mimia toombsi Gardiner & Bartram. Occipital, otic and orbitotemporal regions of
neurocranium and attached dermal bones in ventral view, from BMNH P. 53259. Parasphenoid and
basisphenoid missing.

of the paired endolymphatic ducts in the dipnoans Griphognathus and Chirodipterus , which
differ from Recent dipnoans in having an external opening for the endolymphatic duct on the
dermal skull roof. It is possible that in gnathostomes the posterior dorsal fontanelle primitively
served for the exit of the endolymphatic ducts to the surface of the chondrocranium, since in
xenacanth sharks (Schaeffer 1981: 22) the unpaired, slit-shaped endolymphatic fossa is
confluent with the occipital fissure, and in Moythomasia the paired endolymphatic ducts open
into the dorsal fontanelle.

RELATIONSHIPS OF PALAEONISCIDS

203

2. Occipital fissure. Apart from palaeoniscids a completely uninterrupted, perichondrally-lined
occipital fissure is only known in the pholidopleurid Australosomus (Nielsen 1949, Beltan 1968)
and in the teleost Pholidophorus (Patterson 1975: 417). However, in the latter there is no
obvious posterior dorsal fontanelle and parts of the fissure may be covered superficially by
membranous outgrowths from the intercalar and supraoccipital. In Perleidus cf. stoschiensis
(Patterson 1975: 460) the fissure is still perichondrally lined from the vagus canal up to the
medial margin of the epioccipital and pterotic.

Not surprisingly perhaps, remnants of this perichondrally-lined fissure can be seen in several
other actinopterygian groups. Within the amioids and in some parasemionotids (Patterson 1975:
434) the perichondral lining persists from the vagus canal to just above the external semicircular
canal, while in some pachycormids (Patterson 1975: 448) a small area opposite the medial part of
the intercalar is perichondrally lined. In the earliest known leptolepid braincase, from the
Sinemurian, the perichondrally-lined portion of the cranial fissure still extends from the vagus
canal up to the lower margin of the epioccipital.

Patterson (1975: 418) has shown how the occipital fissure is closed in pholidophorids,
leptolepids and several other groups of actinopterygians by three distinct processes: obliteration
of parts of the fissure by cartilage (this requires simple ontogenetic fusion of the occipital arch
with the otic capsule prior to ossification), obliteration by forward extension of the occipital
bones into the otic region, or the development of membrane bone outgrowths to bridge the
fissure. In living chondrosteans the occipital fissure is obliterated by cartilage in Acipenser, and
in Polypterus its path is represented by the suture between opisthotic and basi-exoccipital
(Patterson 1975: 463).

From this brief survey of the occipital fissure in actinopterygians it is evident that a
perichondrally-lined fissure is primitive (Gardiner 1973: 106; Patterson 1975: 567). Apart from
actinopterygians an occipital fissure is found in acanthodians (Acanthodes, Miles 19730: 66),
Palaeozoic selachians (Xenacanthus , Tamiobatis, Schaeffer 1981), where it is presumed to have
been cartilage-filled and uncalcified, early dipnoans (Miles 1977), and rhipidistians (Jarvik
1954: fig. 1; 1972: 64). In all described actinistians the occipital fissure is obliterated, and as in
Polypterus represented by sutures between bones. In Laugia membranous outgrowths from the
occipital region further obliterate it (Forey, personal communication).

I have argued elsewhere (Gardiner 1973: 129) that a perichondrally-lined occipital fissure is
primitive for osteichthyans (see also Patterson 1975: 466) and is a synapomorphy they share with
acanthodians. More detailed examination, however, has convinced me that in Acanthodes
bronni the fissure is not perichondrally lined.

3. Vestibular fontanelle. This lies at the anteroventral corner of the occipital fissure, usually
anterior or ventral to the glossopharyngeal foramen, and is open in all known palaeoniscid
braincases; it is large in Pteronisculus (Nielsen 1942: 48) and Kentuckia (Rayner 1951: fig. 7).
Though the fontanelle was cartilage-filled in Mimia (see particularly Figs 14, 15, 25) the upper
part at least is perichondrally lined in Pteronisculus (Nielsen 1942: 48), and in Boreosomus
(Nielsen 1942: 290) the whole fontanelle is perichondrally lined. Apart from these two
palaeoniscids no other actinopterygian has been described in which the vestibular fontanelle has
a perichondral lining. A vestibular fontanelle is still seen in the pholidopleurid Australosomus
(Nielsen 1949: 41) and in Perleidus (Patterson 1975: 460), but in the latter it is often obliterated
in more fully ossified skulls. Within the amioids a large fontanelle persists in some
parasemionotids (Patterson 1975: 434), whereas in others such as Ospia and some individuals of
Watsonulus it is again obliterated. A quite large fontanelle persists in pachycormids (Patterson
1975: fig. 106) while in the primitive teleost Pholidophorus bechei (Patterson 1975: fig. 56) the
fontanelle is as large as in Mimia. In some other species of Pholidophorus the fontanelle is
closed. In Recent teleosts the vestibular fontanelle can often be recognized in the adult
neurocranium as an area of membrane or cartilage at the junction of the prootic, basioccipital
and exoccipital. Thus it would seem that primitively in actinopterygians the vestibular fontanelle
was cartilage-lined and confluent with the cranial fissure.

The vestibular fontanelle clearly corresponds to the basicapsular fenestra of the embryo

204 B. G. GARDINER

(Patterson 1975: 466). In the embryo the basicapsular fenestra is that space between the
auditory capsule and the parachordal cartilage, bounded anteriorly by the embryonic
connection of the capsule to the parachordals, the anterior basicapsular commissure (de Beer
1937: 399). A long metotic fissure separates the otic capsule from the parachordals posteriorly in
both selachians and osteichthyans, and the basicapsular fenestra is eventually closed off
posteriorly by the posterior basicapsular commissure, but this often happens quite late in
ontogeny: not until the 9-5 mm stage in Amia (Pehrson 1922) or the 11 mm stage in Lepisosteus
(Veit 1911, Hammarberg 1937). Thus, in the palaeoniscids and other primitive actinoptery-
gians, the relationship of the vestibular fontanelle to the occipital fissure resembles that seen in
early fish embryos.

A vestibular fontanelle, cartilage-filled in life, is also present in the rhipidistians
Eusthenopteron (Jarvik 1954: fig. 1) and Youngolepis (Chang 1982: fig. 10), and an area of the
braincase wall of Acanthodes (Miles I91la: fig. 4.7, Jarvik 1977: fig. 3) anteroventral to the
occipital fissure and devoid of perichondral bone possibly included the vestibular fontanelle.
Miles (1977: 49) has argued that since the fontanelle is filled with cartilage in most primitive
actinopterygians, and can thus close in bone during ontogeny, it is devoid of phylogenetic
significance. Hence, though dipnoans lack a fontanelle they do not differ significantly from
primitive actinopterygians and Eusthenopteron in this respect. However, from the history and
occurrence of the vestibular fontanelle I regard the absence of the fontanelle in dipnoans as a
derived character, in contrast to its absence in palaeoniscids and Eusthenopteron.

The vestibular fontanelle is obliterated in actinistians but persists in tetrapods where it forms
the fenestra ovalis of the auditory capsule. A vestibular fontanelle has also been recorded in
fossil selachians (Xenacanthus, Tamiobatis, Schaeffer 1981: figs 5, 21).

From the arguments outlined above for the occipital fissure and from the fact that the
vestibular fontanelle is present in selachians, actinopterygians, rhipidistians and tetrapods I
conclude that a cartilage-lined fontanelle is a primitive gnathostome character.

4. Ventral otic fissure. In the Gogo palaeoniscids Kansasiella, Pteronisculus macropterus and
Boreosomus, the ventral otic fissure is separate from the vestibular fontanelle.

In other palaeoniscids such as Pteronisculus stensioei (Nielsen 1942: figs 4, 6), Kentuckia
(Rayner 1951 : fig. 9) and 'Ambipoda' (Beltan 1968: tig. 4) the ventral otic fissure passes through
the base of the otic region and runs into the vestibular fontanelle; thus the endocranium contains
two median ossifications in the adult. However, since the position of the ventral otic fissure in
the Gogo palaeoniscids represents the gap between the trabeculae (+ polar cartilages) and
parachordals in the embryo (Gardiner 1973: 106) and is in an identical position to the ventral
part of the intracranial joint in rhipidistians (Eusthenopteron Jarvik 1954; Glyptolepis Jarvik
1972) and actinistians (Latimeria Millot & Anthony 1958, 1965; Diplocercides Bjerring 1972:
fig. 3), the anterior position of the fissure must be primitive for osteichthyans (Gardiner 1973:
107). Subsequently in actinopterygian evolution the fissure migrated backwards, as the
myodome developed, and became confluent with the vestibular fontanelle and the occipital
fissure (Gardiner 1970, 1973: 106; Patterson 1975: 541; Gardiner & Bartram 1977: fig. 8).
Bjerring (1978) however, has denied the homology of the intracranial joint with the ventral otic
fissure, and Schaeffer & Dalquest (1978) have doubted the migration of the ventral otic fissure
in actinopterygians.

In all other actinopterygians, where it is possible to distinguish the two fissures, the ventral
otic fissure is a ventral continuation of the occipital fissure, through the vestibular fontanelle. In
Polypterus the ventral otic fissure persists as a broad tract of cartilage between basisphenoid
(= sphenoid) and basioccipital, but in Australosomus (Nielsen 1949: fig. 4) it is much narrower
and dearly opens into the vestibular fontanelle. Similarly in lightly ossified specimens of
Perleidus (Patterson 1975: 460) and in Birgeria the ventral otic fissure is represented by a large
tract of cartilage contiguous with the vestibular fontanelle.

In Amia and Lepisosteus the fissure is represented by a broad band of cartilage between the
basioccipital and prootics. In parasemionotids (Patterson 1975: 434), though often obliterated
externally, the ventral otic fissure is visible as a suture on the internal bone surface. The fissure is

RELATIONSHIPS OF PALAEONISCIDS 205

also observable in Lepidotes (Patterson 1975: 450), young individuals of pachycormids, later
pholidophorids and leptolepids (Patterson 1975: 466). In all living teleosts it is represented by
the suture or cartilage between the prootics and basioccipital. From the position of the ventral
otic fissure in Mimia (and perhaps Polypterus, see above) it appears that primitively the fissure
passed between the basioccipital and basisphenoid (ossifications deduced from Patterson 1975)
at least in its most ventral part (Gardiner & Bartram 1977: 230). Subsequently, as a consequence
of myodome formation and accompanying rearward migration of the fissure, the prootics
replace the basisphenoid as the anterior margin of the ventral otic fissure in later
actinopterygians .

Patterson (1975: 466) has shown that in many primitive actinopterygians, such as Saurichthys,
some specimens of Perleidus and parasemionotids, early caturids, Dapedium, and early
pholidophorids and leptolepids, all forms in which sutures do not persist in the fully ossified
braincase, the ventral otic fissure may close completely. This is of interest in relation to the
possibility of neurokinesis in actinopterygians. Schaeffer (1968: 216) has argued that where
confluence of the ventral otic fissure and occipital fissure has occurred there is the possibility of
the two halves of the neurocranium moving relative to one another and that a flexible joint
existed in palaeoniscids. I have argued elsewhere (Gardiner 1970: 286) that in primitive
palaeoniscids, at least, there was never any neurokinesis, whereas Patterson (1975: 418) has
demonstrated that no such kinesis could have occurred in philodophorids. Patterson (1975, and
see above) has also shown how, within almost every actinopterygian group, closure of one part
or another of the two fissures has taken place. It would appear that the only possible candidates
for neurokinesis are those post-Devonian palaeoniscids such as Pteronisculus (Nielsen 1942),
Kentuckia (Rayner 1951), Paramblyptems (Heyler 1969), 'Ambipoda' (Beltan 1968) and the
pholidopleurid Australosomus (Nielsen 1949) in which the ventral otic fissure and occipital
fissure are continuous and the parasphenoid ends at the ventral otic fissure. But in Pteronisculus
there are species (P. macropterus) in which the two fissures are still separate (Beltan 1968: pi. 2),
while in Australosomus the configuration of the dermal skull roof (Nielsen 1949: fig. 21) makes
neurokinesis highly unlikely. Furthermore, no living cladistian or chondrostean shows any
semblance of neurokinesis. Therefore it is highly unlikely that any actinopterygian ever
exhibited neurokinesis (see also Pearson & Westoll 1979: 386).

As stated earlier (p. 204) it is my belief that the ventral otic fissure corresponds to the ventral
part of the intracranial joint in actinistians and rhipidistians (Gardiner 1970: 286; Gardiner 1973:
108). Jarvik (1954, 1960, 1968, 1972) and Bjerring (1967, 1973) have insisted that the
intracranial joint is a persisting vertebral joint and therefore primitive for gnathostomes. This
has led to some disagreement over the exact position of the ventral otic fissure in actinistians and
rhipidistians. Jarvik (1954) believed that the more posterior position of the ventral fissure in
Pteronisculus was primitive and homologized it with a cartilage-filled fissure in an apparently
similar position, linking the vestibular fontanelles in Eusthenopteron. Bjerring (1971) on the
other hand claimed that it was homologous with his 'anterior intraotic joint' and later (Bjerring
1973) insisted that the intracranial joint was not homologous in rhipidistians and actinistians.
These arguments have been critically reviewed at length elsewhere (Miles 1975, 1977; Patterson
1975; Gardiner & Bartram 1977; Wiley 1979) and as Miles (1977: 50) succinctly summed them
up, 'the best reason for rejecting Jarvik's and Bjerring's conclusions is that they lead to
unacceptable phylogenetic results.' No such difficulties arise if we consider a ventral otic fissure
as primitive for osteichthyans and the intracranial joint a shared specialization of actinistians
and some rhipidistians (see also Rosen et al. 1981: 259).

Finally, the condition of the ventral otic fissure in dipnoans is of interest since it closely
parallels that of later actinopterygians (Miles 1977: 50). Although there is no myodome
in dipnoans, the ventral otic fissure has migrated posteriorly and is covered by a long
parasphenoid stem. In the Gogo dipnoans Griphognathus and Chirodipterus (Miles 1977: figs
13, 17) the ventral otic fissure is already continuous with the occipital fissure, there is no
expanded vestibular fontanelle and the median portion of the ventral otic fissure has been
obliterated externally, much as in some parasemionotids. I have suggested elsewhere (Gardiner
1973: 108) that early phylogenetic obliteration of this inherent line of weakness in the braincase

206 B. G. GARDINER

floor of dipnoans is related to their specialized feeding habits and to the concomitant fusion of
the palatoquadrate and neurocranium. The parasphenoid also grew back to close over the
the ventral otic fissure in later actinopterygians and possibly in tetrapods other than
ichthyostegids (but see Rosen et al. 1981: 259).

5. Supraoccipital. This does not occur in Polypterus or Acipenser or in any fossil or living
lepisosteoid, amioid, pachycormid or semionotid (Patterson 1975: 432-450); nevertheless it is
characteristic of pholidophorids and teleosts. However, since a supraoccipital may be present in
palaeoniscids such as Mimia and Moythomasia, its absence in amioids, pachycormids and
semionotids could be a derived condition. A supraoccipital occurs in actinistians and tetrapods,
and the median ossification in the upper margin of the occipital arch of Eusthenopteron (Jarvik
1975: fig. 9), immediately behind the posterior dorsal fontanelle, may also represent a
supraoccipital.

The presence of a supraoccipital is therefore considered a primitive osteichthyan condition.

6. Aortic canal. This is always present in palaeoniscids and is a common feature in other
primitive actinopterygians. Patterson (1975: 320) has demonstrated how the point at which the
dorsal aorta bifurcated and the aortic ligament originated on the braincase migrated backwards
in actinopterygian evolution; he suggested this movement may be correlated with enlargement
of the circulus cephalicus (or lengthening of the lateral aortae). This rearward migration of the
point of bifurcation of the dorsal aorta presumably resulted in shortening or obliteration of the
aortic canal. In other forms where the lateral aortae remain short, such as Acipenser, Polyodon
(Danforth 1912: 442, fig. 15), Amia and Lepisosteus (Goodrich 1930), backward growth of the
parasphenoid may also have caused loss of the aortic canal (Gardiner 1973: 116). A short but
distinct canal still exists in some parasemionotids (Broughia Stensio 19326: 270), the semionotid
Dapedium (Frost 1913: fig. 1; Gardiner 1960: fig. 38, and in some leptolepids (Patterson 1975:
319). In Recent teleosts an aortic canal has been reported in the notopterid Xenomystus
(Taverne 1973) but Patterson & Rosen (1977: 129) have shown this to be a neomorph produced
in relation to the ear/swimbladder connection. An aortic canal is clearly a primitive feature in
actinopterygians .

There is no aortic canal in actinistians, and the canal is also absent in dipnoans and
rhipidistians. That the canal should be missing in dipnoans and rhipidistians is not surprising
since in both groups there is good evidence to show that the dorsal aorta bifurcated behind the
occiput (Ectosteorhachis Romer 1937: fig. 1; Eusthenopteron Jarvik 1954: fig. 7; Chirodipterus
Save-Soderbergh 1952, Miles 1977: fig. 18; Neoceratodus Sewertzoff 1902: 593). Miles (1977: 56)
has suggested that absence of the aortic canal in dipnoans is secondary and correlated with
backward expansion of the parasphenoid, but it seems more likely to be related to the fact that
dipnoans, like primitive tetrapods, have long lateral dorsal aortae (or epibranchial arteries) and
this places the point of bifurcation of the aorta behind the occiput. Moreover, expansion of the
parasphenoid has not occurred in actinistians or rhipidistians, yet they all lack an aortic canal.
Despite there being no aortic canal in dipnoans and no sign of an aortic ligament in Recent
dipnoans, Miles (1977: fig. 13) considered a pit on the hind face of the occiput in Griphognathus ,
between the parasphenoid and cranial centrum, to be the site of origin of the aortic ligament. He
(1977: 56) further suggested that a similar notch in the back of the parasphenoid in Birgeria
(Nielsen 1949: fig. 62) was for the aortic ligament. In actinopterygians where there is any
evidence of an aortic ligament, even in Recent clupeoids, salmonoids and cyprinoids, the point
of attachment is always to the basioccipital and not, as in Griphognathus, to the base of the
cranial centrum. Moreover, it is difficult to believe that Birgeria differs from all other described
palaeoniscids, Polypterus and chondrosteans in having the bifurcation of the dorsal aorta behind
the occiput, particularly since a similar notch in the back of the parasphenoid is to be seen in
Polypterus, Saurichthys (Stensio 1925) and Chondrosteus (RSM 1887.15.2).

Outside osteichthyans an aortic ligament is said to have been present in Acanthodes (Miles
1973a: fig. 5), but the only record of an aortic canal other than in actinopterygians is in the
holocephalan Helodus (Moy-Thomas 1936: fig. 4). In all living selachians the paired lateral

RELATIONSHIPS OF PALAEONISCIDS 207

aortae are comparatively shorter than in Polypterus, Acipenser, Amia and Lepisosteus , and the
dorsal aorta bifurcates behind or just at the level of the occiput. The lateral aortae are often
enclosed in paired canals in primitive selachians such as Cladodus (Gross 1937: fig. 5),
Cladoselache (Harris 1938: 9), Tamiobatis (Romer 1964) and xenacanths (Schaeffer 1981: fig.
6); they are similarly enclosed in the living carcharhinoid Dirrhizodon (Compagno 1973: 19). In
other fossil selachians such as Hybodus (Maisey 1983) the lateral aortae lay in well-marked
grooves beneath the occipital region much as in Megalichthys. In placoderms such as Wijdeaspis
(Young 1978) there are long paired grooves on the occipital region which must have housed the
lateral aortae, but in Brindabellaspis (Young 1980: fig. 7) the lateral aortae pass through
separate canals as in Cladodus.

A median aortic canal is therefore present only in actinopterygians and holocephalans, and it
is not possible to decide whether it has arisen independently in the two groups or is a primitive
gnathostome character.

7. Canal for abducens nerve. Primitively in actinopterygians the abducens nerve passed through
the basioccipital and entered the orbit through the corner of the prootic. In Latimeria (Millot &
Anthony 1958) the abducens is said to pass down through the floor of the saccular cavity without
piercing any ossification. This course is not surprising since the basioccipital is small and
confined to the posterior end of the neurocranium. However, in the rhipidistian Eusthenopteron
Jarvik (1972: fig. 93) has restored the neurocranium with a foramen for the abducens in the edge
of the ossification lateral to the notochord. This foramen is in the prootic (deduced from a
similar ossification in actinopterygians and Latimeria), but in Eusthenopteron, as in Latimeria,
the basioccipital does not extend ventrally beneath the anterior portion of the notochord. The
path of the abducens in dipnoans is difficult to follow; for example, in the development of
Neoceratodus the abducens is intimately connected with the roots of the facial and trigeminal
nerves (Fox 1965: 505). Nevertheless this nerve never passes through the prootic bridge as in
later actinopterygians; in the Devonian Chirodipterus (Miles 1977: figs 17, 21) it is presumed to
pass laterally with V and VII through the prootic area. In the development of amphibians such as
Ambystoma (Goodrich 1911) the abducens nerve still pierces the anterior parachordal (so called
because the parachordal is restricted to the extreme anterior end of the notochord) to emerge on
the ventral surface of the skull.

I conclude that in osteichthyans the abducens nerve primitively passed through the
basioccipital.

8. Zygal plates. In primitive actinopterygians such as Mimia paired zygal plates are found in the
roof of the notochordal canal. Similar plates are also present in Moythomasia, but in other
palaeoniscids such as Pteronisculus the zygal plate is median and unpaired. In actinopterygians
zygal plates are found only in palaeoniscids. They appear to be present in all actinistians. In
Nesides (Bjerring 1971 : 194) the zygals are said to be paired, but in Latimeria (Millot & Anthony
1958: pi. 17) there is a median, bilobed plate. In rhipidistians paired zygals have been reported in
Eusthenopteron (Bjerring 1971: 192) but in Glyptolepis (Jarvik 1972: 68) there is a median plate
as in Latimeria.

From this distribution we may conclude that paired zygal plates are a primitive osteichthyan
feature. It is easy to see how, with increase in size of the myodome and concomitant regression
of the notochordal canal, they have been lost in later actinopterygians. The zygals appear to play
an important part in the intracranial joint in Latimeria and some rhipidistians; they are missing
in dipnoans and tetrapods.

9. Occipital artery. Primitively, this appears to be related to the second permanent myomere,
since in both Mimia and Moythomasia the foramen fof the artery arises between the ridges for
the insertion of the first and second intermuscular septa. This direct relationship, also seen in
Amia, is recognizable in any other actinopterygian, but in the rhipidistian Ectosteorhachis
(Romer 1937: fig. 2) the foramen for the occipital artery opens in the line of the ridge for the
second intermuscular septum and presumably belongs to the second permanent segment, not

208 B. G. GARDINER

the third as suggested by Romer (1937: 8). Support for this view can be obtained from one
specimen of Mimia (BMNH P. 54501) in which that part of the basioccipital around the aortic
canal and notochord projects posteriorly for some distance as it does in Ectosteorachis .
Furthermore, in the development of Polypterus Allis (1922: 208) has demonstrated how the
basioccipital portion of the basi-exoccipital can project posteriorly beyond its exoccipital
portion to a distance equal to about half that of the first free vertebra and still be part of the
segment anterior to it (that incorporating the ventral root of the second occipital nerve).
Elsewhere an occipital artery has been described in Eusthenopteron where it is also presumably
related to the second permanent occipital segment (Jarvik 1975: fig. 8). A similar
dorsolaterally-directed canal has also been described in Devonian dipnoans (Save-Soderbergh
1952: figs 1, 8; Miles 1977: figs 11, 15, 23), but its relationships with neighbouring foramina is not
clear. The course of the occipital artery has been described in Neoceratodus (Spencer 1893: 10)
and an artery in a similar position exists in urodeles (Driiner 1901) and anurans (Gaupp 1899). In
placoderms such as Brindabellaspis (Young 1980: fig. 7) the occipital artery passed into the
cranial cavity in the vicinity of the first occipital nerve, and then appears to have run backwards
through the occiput. Finally in Acanthodes (Miles 1973«: fig. 3) the foramen for the occipital
artery again appears to be related to the second occipital segment since it lies in front of the
opening for the second occipital nerve (first occipital, see below).

The primitive course of the occipital artery after leaving its canal in the basioccipital is less
certain. In Amia (Allis 1897: 706) it runs up over the lateral surface of the basioccipital and
exoccipital onto the dorsal surface of the occiput, where it sends branches to all occipital
myomeres. In Polypterus, although there is no occipital artery as such, the first intervertebral
artery apparently takes over its function. In a 75 mm larva, Allis (1922: 207) has shown how the
intervertebral artery divides into two and the anterior branch passes into a canal in the
basioccipital. From thence one part of it passes into the cranial cavity through the second
occipital nerve canal and the other continues up towards the second intermuscular septum. The
posterior branch also enters the cranial cavity, but through the canal for the ventral root of the
first spinal nerve. Unfortunately, in other living osteichthyans the occipital artery either does
not groove the surface or does not enter the occiput, consequently there is no model on which to
base the distribution of the various branches. Nevertheless, a system of grooves for the branches
of this artery has been recognized in the rhipidistian Eusthenopteron (Bjerring 1971: fig. 18;
Jarvik 1975: fig. 9).

From the evidence given above for both Mimia and Moythomasia the canal which enters the
anterior floor of the foramen magnum after running through the exoccipital at right angles to the
long axis of the neurocranium must have carried a branch of the occipital artery and not an
occipital nerve. If this is so then the canal in a corresponding position in Kansasiella (Poplin
1974: figs 24, 25) must also have served for a branch of the occipital artery. It follows that a canal
with identical relationships and running at right angles to the long axis of the brain in the
rhipidistian Ectosteorhachis (Romer 1937: fig. 9) must also have served for a branch of the
occipital artery, not the first occipital nerve as suggested by Romer (1937: 8). It should be noted
that occipital nerves normally run obliquely backwards to their points of exit. It also seems likely
that in Acanthodes (Miles 19730: fig. 2) the foramen lying in the groove for the occipital artery
transmitted a branch of that artery rather than the first occipital nerve.

10. Segmental structure of occiput. In view of the variable composition of the occipital region in
Recent fishes and the loss of somites from the metotic series during development, determination
of the primitive number of adult occipital myomeres is difficult. For example, among
actinopterygians Polypterus has incorporated one centrum into its braincase, Amia two,
Lepisosteus three, and Acipenser up to eight; among dipnoans Neoceratodus has added three.
In the Gogo palaeoniscids there must have been at least two permanent myomeres, more
probably three, judged from the position of the two intermuscular septa and the single occipital
nerve canal. A similar condition existed in pholidophorids and leptolepids (Patterson 1975:
318). In all these forms the first myomere is characterized by the absence of any canal for an
occipital nerve. From this evidence I conclude that the ancestral condition for actinopterygians

RELATIONSHIPS OF PALAEONISCIDS

209

is a single occipital nerve canal (which carried a ventral root only) related to the second
permanent myomere.

In Latimeria (Millot & Anthony 1958) both the basioccipital and the paired exoccipitals are
much reduced; nevertheless, together with the 'supraoccipital' these ossifications must
represent the first and only occipital sclerotomes, since they are associated with the first trunk
muscle. There is no corresponding occipital nerve; instead the associated nerve leaves the
foramen magnum posterior to this segment. Other Recent bony fishes in which the first occipital
myomere has either no related ventral nerve root, or if it does, whose nerve exists in a more

pitf

fotc

gboca

cao

cla

foca

2mm

Fig. 15 Mimia toombsi Gardiner & Bartram. Occipital, otic and orbitotemporal regions of
neurocranium and attached dermal bones in ventral view, from BMNH P. 56483. Anterior regions
of parasphenoid and basisphenoid missing.

210 B. G. GARDINER

posterior segment, include Amia (Allis 1897: 725), Polypterus (Allis 1922: 207), Lepisosteus
(Schreiner 1902), and Lepidosiren (Bridge 1898). Furthermore, in the development of the
amphibian Ambystoma (Goodrich 1911) the first permanent myomere never has a ventral root.
On the other hand Cryptobranchus (de Beer 1927) is said to be unique among living amphibians
in possessing one pair of occipital foramina , but there is evidence of two segments in this animal .

In Ectosteorhachis (Romer 1937), Rhizodopsis, Eusthenopteron (Jarvik 1975) and
Youngolepis (Chang 1982) there is evidence from the imprint of the intermuscular septa and the
course of the occipital artery of three occipital myomeres, as in Mimia. Two pairs of foramina
have also been described in the occiput of all four genera. The posterior pair (posterior
spino-occipital nerve canals of Jarvik 1975) corresponds to the occipital nerve (ventral root)
foramina of Mimia, while the anterior pair (anterior spino-occipital nerve canals of Jarvik 1975)
occupies a similar position to foramina I attribute to branches of the occipital artery in
Moythomasia and Mimia.

In Griphognathus and Chirodiptems (Miles 1977) the occipital foramen is followed posteri-
orly by three pairs of spino-occipital nerve foramina, and in Holodipterus, Dipterus and
Conchopoma (Schultze 1975) by at least two pairs of spino-occipital nerve foramina. The skull of
Neoceratodus (Fox 1965) includes three occipital arches (one occipital and two spino-occipital),
that of Protopterus (de Beer 1937) two, and Lepidosiren one (Agar 1906). In Protopterus the
occiput is pierced by a single pair of occipital nerves and in Neoceratodus the occiput encloses
two pairs of spino-occipital nerves.

In many selachians the occipital nerves leave either through the vagus canal or behind the
condyles. Nevertheless, in Scyllium (de Beer 1937), where the occiput is believed to comprise
three segments, the occipital arch is pierced by a single pair of occipital nerves. In Xenacanthus,
Tamiobatis (Schaeffer 1981) and Hybodus (BMNH P. 50869) the occipital region is pierced by at
least three (four in Hybodus) pairs of nerve foramina, but foramina are apparently wanting in
Cladoselache, Cobelodus and 'Cladodus' (Schaeffer 1981).

In placoderms there are invariably several pairs of occipital nerve foramina; up to seven pairs
of spino-occipital nerves have been described in Buchanosteus (Young 1979), five pairs in
Brindabellaspis and three pairs in Ctenurella (Miles & Young 1977).

Finally in Acanthodes the basioccipital is almost as extensive as in Mimia and incorporates the
foramen for the occipital artery, whereas the 'lateral occipital' includes the foramen for the
occipital nerve as well as a more anterior foramen (Miles 1973a: fig. 2, onl) for a branch of the
occipital artery.

From this brief survey I conclude that the first myomere of osteichthyans is characterized by
the absence of an occipital nerve.

11. Longitudinal intervertebral ligament. The presence of this is a primitive feature for
gnathostomes (Goodrich 1930: 21), and in later actinopterygians such as parasemionotids and
semionotids the ligament terminated in a deep, cartilage-lined pit in the exoccipital region above
the foramen magnum. A comparable pit in the same region has been described in the Devonian
dipnoans Griphognathus and Chirodipterus (Miles 1977: figs 12, 16), and an unossified area
immediately above the foramen magnum in Eusthenopteron (Jarvik 1975: fig. 10) presumably
served the same function.

Otic and orbitotemporal regions

Review of ossification centres

The otic and orbitotemporal regions of the braincase consist of a single ossification which shows
few sutures, apart from the otico-sphenoid fissure. Several of the more lightly ossified specimens
of Mimia (BMNH P. 56495, P. 56496) and Moythomasia (BMNH P. 56480) do, however, show
partial gaps in the perichondral lining of the internal surface of the neurocranium, which give
some indication of the internal extent of the individual ossifications. The external extent of these
ossifications is occasionally marked by faint sutures and by the more obvious fissures (ventral
otic fissure, otico-sphenoid fissure; fv, fos, Fig. 22). From such specimens it is possible (by

RELATIONSHIPS OF PALAEONISCIDS 211

comparison with Perleidus, parasemionotids and pholidophorids) to estimate the extent of the
individual bones with confidence. Thus in Mimia and Moythomasia it appears that the prootic
ossifies around a centre in the lateral commissure and occupies only a small part of the outer
lateral wall of the braincase and the postero ventral corner of the orbit, but probably includes the
trigeminal and facial foramina. Internally the prootic forms the anterolateral wall of the otolith
chamber and the ampullary chambers of the anterior and external semicircular canals. Dorsally
the prootic forms the the anteromedial portion of the lateral cranial canal. I conclude that the prootic
forms a smaller part of the braincase than in parasemionotids and a considerably smaller part
than in pholidophorids (Patterson 1975). This discrepancy may be directly related to the absence
of a myodome in the Gogo palaeoniscids, since in pholidophorids, leptolepids and Recent
teleosts where the myodome is extensive the prootic is proportionally larger. The prootic is
large in fossil actinistians and Latimeria (Millot & Anthony 1958) where the bone extends
postero ventrally as far as the ventral fissure and posterolaterally to the hyomandibular facet.

The prootic is small in Acipenser and occupies only the posteroventral corner of the orbit. In
Polypterus the prootic is a transient ossification only seen in embryos (Pehrson 1947: 405),
where it forms a small perichondral thickening around the posterior margin of the trigeminal
foramen. Elsewhere within the palaeoniscids a prootic is absent from Birgeria (Nielsen 1949). In
all these forms the ascending process of the parasphenoid is long and complex and covers that
lateral area of the neurocranium normally occupied by the prootic. A prootic is not obviously
present in Acanthodes (see for example Miles 19730: fig. 2; Jarvik 1977: fig. 3). The prootic
appears to have been a small ossification (smaller than the opisthotic) in primitive
actinopterygians. Its increase in size and importance in later halecomorphs and teleosts is
presumed to be related to myodome formation, whereas its large size in actinistians may be
related to the absence of a sphenotic.

A separate stout sphenotic ossifies from a centre in the postorbital process. In Mimia it is
presumed to extend posteriorly as far as the hyomandibular facet which marks the junction
between it and the opisthotic. A wide groove on the face of the postorbital process housed the
spiracle which opened dorsally on the skull roof. An endoskeletal spiracular canal is absent, but
occasionally a pair of processes partially delimit the top of the spiracular groove (Fig. 14;
Gardiner 1973: fig. 5). Similar processes in Moythomasia (Fig. 28) may join to form a complete
bar (spic, Figs 7, 30). An endoskeletal spiracular canal is characteristic of most palaeoniscids
(Pteronisculus, Boreosomus, Kentuckia, Kansasiella, Birgeria etc.), Perleidus (Stensio 19326:
fig. 59), Australosomus (Nielsen 1949), many fossil halecostomes including Pholidophoms
(Patterson 1975: 399) and all extant chondrosteans and holosteans; it is absent in Polypterus.
Within primitive actinopterygians a separate sphenotic has been described in Cosmoptychius
(Watson 1928; Schaeffer 1971), Perleidus (Patterson 1975), Birgeria and Polypterus and is
universally present in halecomorphs and teleosts. A similar ossification of the postorbital
process in Acanthodes (Miles 19730: fig. 1; Jarvik 1977: fig. 3) may also be a sphenotic since, as
in primitive actinopterygians, it forms the anterior margin of the hyomandibular facet; the
suggestion by Jarvik (1977: 207) that the head of the hyomandibula articulated with the middle
of the otic ossification is considered unlikely since in ossified neurocrania the hyomandibular
facet always lies at the junction of two or more ossifications. An ossified postorbital process is
also present in placoderms (Buchanosteus Young 1980) and a tesserate postorbital process is
a prominent feature of most chondrichthyans. The sphenotic is apparently missing from those
actinistians in which separate ossifications have been described (Wimania Stensio 1921, 1925;
Macropoma, Whiteia Beltan 1968: 114; Rhabdoderma Forey 1981; Latimeria Millot & Anthony
1958) and a single ossification occupies the area of the basisphenoid, pterosphenoid and
sphenotic. This ossification also includes the postorbital process (antotic process of other
authors). The postorbital process is missing in Ectosteorhachis and Rhizodopsis whereas in
Eusthenopteron, Holoptychius and Glyptolepis (Jarvik 1972: figs 20, 21) it is represented by the
suprapterygoid process.

The anterior internal limits of the opisthotic can be estimated from Mimia (BMNH P. 56496).
Its external limits can be confirmed partly from Cosmoptychius (Schaeffer 1971: fig. 8, ot),
where a separate opisthotic forms the anterior wall of the vagus canal and includes the canal for

212 B. G. GARDINER

the glossopharyngeal nerve and a groove for its supratemporal branch (Watson 1928: 49), and
partly from Polypterus where the bone is very large.

The opisthotic in Mimia, Cosmoptychius (Schaeffer 1971: fig. 8A), Pteronisculus (Nielsen
1942: 16; fig. 3) and Perleidus (Patterson 1975: fig. 115) has a posteroventral tongue and includes
the lower portion of the posterior semicircular canal as well as the external semicircular canal.
The centre of ossification in Mimia (and Moythomasia) is around the ampulla of the posterior
semicircular canal and is indicated by the downwardly projecting parampullary process (pamp,
Fig. 6) but it extends anteriorly to meet the sphenotic, thereby forming the posterior margin of
the hyomandibular facet. The hyomandibular facet also marks the junction between the
opisthotic and sphenotic in Pteronisculus and Perleidus. Anterodorsally the opisthotic provides
a large facet for the origin of much of the constrictor hyoideus dorsalis (oahm + oaop, Figs 1,4,
5, 13) muscle. In the 75mm Polypterus described by Allis (1922: 219) the opisthotic begins to
form around the projecting hind end of the otic capsule; that is, around the base of the posterior
semicircular canal. In the large specimen oiPolyodon described by Bridge (1878) the opisthotic
is represented by a thin perichondral ossification superficial to the ampulla of the posterior
semicircular canal. In Acipenser it forms in a comparable position between the foramina of the
glossopharyngeal and vagus nerves (Holmgren & Stensio 1936: fig. 336). The opisthotic is
absent in semionotids, Amia and Lepisosteus as well as leptolepids and more advanced teleosts.

There is a well-marked parampullary process in actinistians (Nesides, Bjerring 1977: fig. 23;
Latimeria, Millot & Anthony 1958) and in several forms (Macropoma, Laugia, Wimania, Stensio
1921) this is borne on a discrete, small ossification which, by comparison with actinopterygians,
must be an opisthotic. The probable centre of an opisthotic in rhipidistians is indicated by the
postotical process in Eusthenopteron (Jarvik 1954: figs 1, 21) and the 'paroccipital process' in
Ectosteorhachis (Romer 1937: fig. 1). In both cases the centre of ossification is inferred to lie
lateral to the base of the posterior semicircular canal. In actinistians the centre of ossification
appears to lie somewhat more anteriorly, around an opening in the wall of the labyrinth cavity
(Jarvik 1954: fig. 4, fplab) just in front of the parampullary process. The opisthotic seems to be
large in early actinistians and rhipidistians. In Nesides (Jarvik 1954: fig. 4; Bjerring 1977: fig. 23)
and Rhabdoderma (Forey 1981: fig. 1) the opisthotic forms most of the lateral wall of the
braincase from the vagus foramen to the hyomandibular facet (both dorsal and ventral parts).

The remaining ossification in the otic region is the pterotic. The extent of this can be inferred
partly by comparison with Perleidus, where the pterotic is quite small and occupies the
posterodorsal corner of the otic capsule (Patterson 1975: fig. 115), and partly from the radiating
structure of the bone which is recognizable in several specimens. From this evidence the centre
of ossification of the pterotic may be estimated as lying within a dorsolateral prominence in the
posterior otic region which presumably served for the origin of the posterior part of the
constrictor hyoideus dorsalis (oaop, Figs 4, 5, 6, 13). The constrictor hyoideus dorsalis also
originates on the upper outer corner of the otic region of the skull in Recent chondrichthyans
such as Heptanchus, Squalus and Galeus. The pterotic is missing in Birgeria, Polypterus and
Acipenser; in Polypterus the posterior part of the constrictor hyoideus (the adductor
opercularis) originates on the opisthotic. The pterotic is proportionally much larger in
Pholidophorus than in Mimia and forms the greater part of the subtemporal and post-temporal
fossae but, as in Mimia, the centre of ossification is at the posterolateral shoulder of the otic
capsule (Patterson 1975: fig. 75), just posterior to the foramen for the supratemporal branch of
the glossopharyngeal nerve. In leptolepids and advanced teleosts the pterotic is much reduced,
probably as a result of the closure of the cranial fissure (Patterson 1975: 380), its centre of
ossification being located along the external semicircular canal.

In Amia, Lepisosteus and Lepidotes there is only one bone in the posterodorsal region of the
skull; this may be interpreted as either an epioccipital or a pterotic. Patterson (1975: 443)
believes that in Amia this so-called 'epiotic' probably represents an epioccipital which has
extended forwards following the loss of the pterotic. Since the pterotic primitively appears to
have served for the origin of the posterior part of the constrictor hyoideus dorsalis and this
muscle (adductor opercularis) partly originates on the intercalar in Amia it is reasonable to
suppose that the remaining ossification is the epioccipital (associated as it is only with trunk

RELATIONSHIPS OF PALAEONISCIDS

ctel P'nf

213

acv

frla

prof

spig
jc

fhm VI l+pal
Icom

nona

fos

aip 1

not

cao

2mm

Fig. 16 Mimia toombsi Gardiner & Bartram. Otic and orbitotemporal regions of neurocranium in
anterior view, as if cut through the middle of the orbit, from BMNH P. 53259.

musculature). Patterson (1975: 453) further concluded that in Lepidotes this bone represents the
pterotic because it includes a blind pit which resembles a dorsolateral expansion of the cranial
cavity in the palaeoniscid Boreosomus (Nielsen 1942: fig. 66, Iv). In Lepisosteus it is also likely to
be the pterotic since the adductor opercularis is attached to the lateral surface of this bone.

In Acanthodes (Miles 1973a: fig. 2, swpamp) a distinct boss, on the braincase wall below the
jugular canal and the foramen for the glossopharyngeal nerve, housed the posterior ampulla
(but see Jarvik 1977: fig. 3 where a similar swelling lies above the jugular canal). The ridge above
this ampullary boss delimits the jugular groove and appears to be the centre of ossification of the
otic capsule and presumably also served for the origin of the constrictor hyoideus dorsalis. The
centre of ossification of the capsule is more dorsal than the opisthotic in Mimia and this together
with the presumed muscle origin suggests that the ossification is better homologized with the
pterotic than with the opisthotic.

The whole orbital region of the Gogo palaeoniscids, apart from the basipterygoid process, is
ossified as a perichondral shell penetrated by perichondrally-lined canals for nerves and blood
vessels. Presumably there were three pairs of ossifications, basisphenoid, pterosphenoid and

214 B. G. GARDINER

orbitosphenoid, as in Amia and many teleosts. A separate pterosphenoid has been described in
Pteronisculus (Nielsen 1942: 90). The basisphenoid is an extensive ossification and its paired
nature can be seen in Mimia (BMNH P. 56483, Fig. 15). Posteriorly it consists of a vertical pillar
which flares dorsally into a pair of dorsolaterally-directed arms which join the orbital surface
just beneath the oculomotor foramen to form the dorsum sellae (Gardiner & Bartram 1977:
230). Beneath the bridge the junction of the basioccipital with the basisphenoid is marked by the
ventral otic fissure. The basisphenoid extends anteriorly for a short distance beneath the orbit to
the level of the optic fenestra. The basisphenoid also appears to be paired in Pteronisculus
(Nielsen 1942: fig. 2) and its centres of ossification are inferred to lie on either side of the vertical
pillar, as in Mimia. It has been argued elsewhere (Gardiner & Bartram 1977: 237) that the
cup-shaped depressions on the basisphenoid pillar in Mimia were the points of origin of at least
three of the recti muscles. In Polypterus (Allis 1922: 252) three of the recti muscles originate (by
a short tendinous stalk) on the basisphenoid near its ventral edge and immediately posterior to
the optic foramen (as in selachians). The fourth muscle (internal rectus) has its origin slightly
more anteriorly, still on the basisphenoid, but anterior to the optic foramen. During
development the basisphenoid of Polypterus arises from paired perichondral lamellae between
the optic and oculomotor foramina. Thus its centre of ossification lies at the point of insertion of
the rectus muscles. In other primitive fossil actinopterygians where separate ossifications have
been recognized the basisphenoid is only clearly delimited in Perleidus (Patterson 1975: 457),
and even here it is fused with the prootics.

The basisphenoid in Pholidophorus bechei (Patterson 1975: 381) not only forms the pillar but
also extends ventrally to form the endochondral floor of the orbit. In later pholidophorids this
vental part is less thoroughly ossified while in leptolepids and advanced teleosts all that remains
is a slender pedicel consisting mainly of membrane bone. In most halecomorphs the basi-
sphenoid is a median bone forming little more than the pedicel, as in later pholidophorids. In
Amia, however, the basisphenoid is a small paired ossification which ossifies late in the
transverse 'bolster' in front of the floor of the myodome. Thus, the centre of ossification of the
basisphenoid is ventrolateral to that in Mimia, Polypterus and teleosts. Since three of the rectus
muscles originate on this transverse 'bolster' in Amia the concomitant shift in ossification centre
is hardly surprising. The basisphenoid in pachycormids is not very different from that in
Pholidophorus, while within the semionotids the basisphenoid of Lepidotes is stout and median,
and confined to the posteroventral corner of the orbit. In Lepisosteus there is no basisphenoid,
the rectus muscles originating on the floor of the orbit, lateral to the interorbital septum.
Although the shape of the 'basisphenoid' in actinistians suggests that it arises from paired
centres, it is never paired either in fossil material or in the embryo (Forey, personal
communication). The basisphenoid in Acanthodes (Miles 19730: figs 8, 9) is also a large,
unpaired ossification, but is similar in size and shape to that seen in Mimia. Its centre of
ossification is inferred to lie at the base of the basisphenoid pillar.

That a separate pterosphenoid was present in the orbit of Mimia can be deduced from the
presence of a distinct ridge running upwards from the trigeminal foramen to the roof of the orbit
(Figs 16, 17, 20), and from the presence of a pedicel over the trigeminal and facial foramina in
Moythomasia (Figs 29, 30). The ridge presumably represents the centre of ossification of the
pterosphenoid. The limits of the pterosphenoid can be determined with a fair degree of
confidence from estimates of distances between neighbouring ossification centres. From this
type of analysis the pterosphenoid appears to occupy well over half of the posterior orbital
surface. It extends laterally to just beyond the foramen for the otic nerve, where it meets the
sphenotic (junction often marked by a series of fenestrae), and ventrally to just above
the trigeminal and facial foramina, where it meets the prootic. It extends anteroventrally to
just below the oculomotor foramen where it meets the basisphenoid. Together with the
orbitosphenoid the pterosphenoid forms a complete interorbital septum (Fig. 13). Thus the
pterosphenoid is possibly a large ossification as in other palaeoniscids and contributes to much
of the posterior orbital surface.

A small paired pterosphenoid is found in Acipenser but it is absent in Polypterus. The
pterosphenoid is large in halecomorphs (' Aspidorhynchus', Patterson 1975: figs 99, 101;

RELATIONSHIPS OF PALAEONISCIDS

frd

215

V+VI I lat+mcv

com

fos fhm VI l+pal

2mm

Fig. 17 Mimia toombsi Gardiner & Bartram. Left orbitotemporal region of neurocranium in

anterior view, from BMNH P. 56504.

Macrepistius , Schaeffer 1971: figs 3, 4), semionotids (Lepidotes, Patterson 1975: figs 108, 109),
pachycormids (P achy cor mus, Patterson 1975: fig. 106) and pholidophorids (Pholidophorus,
Patterson 1975: 382). Therefore it seems likely that primitively in actinopterygians the
pterosphenoid was an important constituent of the posterodorsal orbital surface. A similar,
large dorsal perichondral ossification in the orbit ofAcanthodes (Miles 1973a: fig. 4; Jarvik 1977:
fig. 2) may have included an ossification centre homologous with that of actinopterygians.

In actinistians (Rhabdoderma Forey 1981: fig. 1; Latimeria Millot & Anthony 1958) there is a
single ossification centre in front of the prootic. This ossification fills the area occupied by the
basisphenoid, pterosphenoid and possibly the sphenotic in actinopterygians, and also bears the
postorbital process (=antotic process). The postorbital process is stout and provides an
articulation for the dorsal surf ace of the palate. In Acanthodes (Miles 1973a: figs 2, 15) and some
rhipidistians such as Eusthenopteron and Holoptychius (Jarvik 1954, 1972) the postorbital
ossification also bears an articulatory facet for the palate.

Mimia toombsi

The otic region of the neurocranium is separated from the occipital ossification by the posterior
dorsal fontanelle dorsally and the occipital fissure laterally, but ventrally the two regions pass
into one another without any distinct boundary. The posterior face of the otic region is lined with
perichondral bone from the vestibular fontanelle upwards, thus the subvagal portion of the
fissure is open (Fig. 25) as in Pteronisculus, Kansasiella and some individuals of Pholidophorus
bechei (Patterson 1975: 232). The perichondral lining is interrupted at the level of the lateral
cranial canal (plcc, Figs 11, 12) but otherwise extends dorsomedially to the posterior dorsal
fontanelle. Two notches near the posterior margin of the opening of the cranial cavity lead into a
pair of shallow, ventrolaterally-directed grooves. The more dorsal groove is the broader and
contained the posterior cerebral vein (gpcv, Fig. 11) and a small foramen within the groove

216

B. G. GARDINER

aasc

prof

fhra VI l+pal

1mm

Fig. 18 Mimia toombsi Gardiner & Bartram.
Internal, medial view of left utricular recess,
trigeminal and facial foramina, from BMNH
P. 56504. Intramural passages indicated by
broken lines.

(fapcv, Fig. 11) must have transmitted a small vein into the ampullary cavity of the posterior
semicircular canal as in Pholidophorus bechei (Patterson 1975: fig. 59). The lower groove was
occupied by the vagus nerve, while a large foramen near its lateral limit (gph.X, Fig. 11) which
leads out anterolaterally onto the surface of the otic region served for the passage of the
pharyngeal branch of the vagus. In other specimens of Mimia (cf. BMNH P. 56501, Fig. 4) the
pharyngeal branch merely notched the end of the vagal canal. A further notch (gst.X, Fig. 11) in
the posterolateral margin of the dorsal groove marks the passage of the supratemporal branch of
the vagus onto the lateral otic wall.

Below the vagus groove the posterior face of the otic region turns forwards and the
perichondral lining gives way to a cartilage-filled vestibular fontanelle (vfon, Fig. 13) as
in Pteronisculus and other palaeoniscids. The dorsal surface of the otic (Fig. 12) and
orbitotemporal regions (Figs 33, 34) is complete and the only opening is the pineal foramen (PI.
1; pinf, Fig. 33). There is no anterior dorsal fontanelle, in contrast to Polypterus, Pteronisculus
(Nielsen 1942: fig. 7), Kansasiella (Poplin 1974: fig. 12), Kentuckia (Rayner 1951: fig. 6) and
Pholidophorus (Patterson 1975: fig. 60). There is also no fossa bridgei and the recurrent lateralis
branch of the facial nerve appears to have emerged onto the roof of the otic region just behind
the hyomandibular facet and beneath the rim of the intertemporal (frla2, Figs 6, 11). The dorsal
limit of the spiracular groove (spig, Fig. 13) lies in front of the hyomandibular facet, posterior to
the postorbital process (por) and the otic nerve (fotn, Figs 12, 13) emerged through its medial
wall. In Kansasiella (Poplin 1974: fig. 12) and Pteronisculus (Nielsen 1942: fig. 12) the otic nerve
passed into the spiracular canal. There are two further foramina in the posterior face of the otic
region. The more lateral, smaller foramen leads into the lateral cranial canal (fv, Fig. 11); the
more medial foramen (dend, Fig. 11) housed the blind-ending endolymphatic duct. From it a
gutter runs down towards the cavity occupied by the sinus superior (Fig. 26). There is a recess in
the roof of the otic region (rsoc, Fig. 11) which marks the anterior limit of the posterior
fontanelle. In Pholidophorus bechei (Patterson 1975: fig. 65) the membranous extension of the
supraoccipital bone enters this recess.

The lateral face of the otic region has a complex relief (Figs 4, 5, 6, 13). Anterodorsally there is
a prominent postorbital process which forms the anterior boundary of the wide spiracular
groove (spig, Fig. 13). This groove passes ventromedially, crosses the otico-sphenoid fissure
(fos), continues on the basisphenoid behind the basipterygoid process and fades out on the
parasphenoid at the level of the bucco-hypophysial canal (bhc, Fig. 50). There is no
post-temporal fossa and this is considered primitive for actinopterygians. The post-temporal
fossa is also missing in all other palaeoniscids, Polypterus and Lepisosteus, but occurs in caturids,
semionotids, pycnodonts, Amia, pachycormids, pholidophorids and most other teleosts
(Patterson 1975: 395).

The hyomandibular facet (fhm, Fig. 13) lies obliquely across the lateral commissure and is not
lined by perichondral bone. Ventrally the facet extends onto the roof of the jugular canal (Figs 4,
5, 6). Behind the hyomandibular facet is an extensive raised area of bone, triangular in outline

fotn

RELATIONSHIPS OF PALAEONISCIDS

frla

217

mcv

IV III acv

V+VII lat

fotn

prof
V+VIIlat

por

2 mm

1
fhmVII+pal

fos

nona

VI fv

Fig. 19 Mimia toombsi Gardiner & Bartram. Otic and orbitotemporal regions of neurocranium in
anterior view, as if cut through the middle of the orbit, from BMNH P. 53234. (A), right side of the
rear of the orbit; (B), left side of orbit.

with a groove or gutter dissecting it posteriorly. This area, which marks the point of origin of the
dorsal hyoid constrictor muscle (oahm, oaop, Figs 4, 5, 6 ), stretches from the hyomandibular
facet to the occipital fissure. Presumably the posterior portion of this constrictor muscle served
for the adduction of the operculum (oaop, Figs 4,5,6) while the anteriormost region served for
the adduction of the hyomandibula (oahm, Fig. 6).

Below the area of origin of the dorsal hyoid constrictor muscle a well-marked jugular groove
(jg, Fig. 6) runs horizontally across the lateral face of the otic region. Behind the parampullary
process (pamp, Fig. 6) the groove turns dorsolaterally in front of the vagus canal where it
received the posterior cerebral vein from the upper division of that canal (gpcv, Fig. 11). The
supratemporal branch of the vagus nerve (gst.X, Fig. 4) passed forward beneath the
parampullary process, in the posterior portion of the jugular groove, then turned upwards and
ran in a short groove through the area of origin of the dorsal hyoid constrictor muscle (gst.X, Fig.
5) and out onto the dorsal surface. Immediately beneath the posterior portion of the jugular
groove there is often a further groove which soon fades out anteriorly. This groove transmitted
the pharyngeal branch of the vagus nerve (gph.X, Fig. 4). The glossopharyngeal foramen (IX,
Figs 4, 5, 6, 7, 13) lies either in the jugular groove or a little below it. The supratemporal branch
of the glossopharyngeal nerve passed upwards from this foramen, through a distinct channel
(gst.IX, Fig. 14) in the area of origin of the dorsal hyoid constrictor muscle to enter a foramen
(fst.IX, Figs 4,5,6) immediately beneath the dermal skull roof. In some specimens this channel
is confluent with that for the supratemporal branch of the vagus nerve (BMNH P. 56501,
P. 56496, Figs 4, 6, 14), whereas in others (BMNH P. 53234, Fig. 5) it is separate.

Below and in front of the glossopharyngeal foramen the wall of the saccular recess is inflated.
This inflation terminates in the vestibular fontanelle (vfon, Figs 14, 15). More dorsally the
ampulla of the posterior semicircular canal causes in the lateral wall a distinct swelling which is

218 B. G. GARDINER

often drawn out into a prominent, ventrally-facing, parampullary process (pamp, Fig. 6) to
which the first suprapharyngobranchial was presumably ligamentously attached. In earlier
reconstructions (Gardiner 1973: fig. 1) I erroneously assumed that the first suprapharyngo-
branchial articulated at the level of the glossopharyngeal foramen.

The lateral commissure is short and broad and composed entirely of cartilage bone. A
well-marked groove (goa, Fig. 15) runs dorsolaterally from the region of the orbitonasal artery
foramen up into the jugular canal. Occasionally a narrow strut of bone encloses the top of this
groove (BMNH P. 53234, Fig. 5). In life this groove housed the orbital artery. The orbital artery
passed into the jugular canal then out into the orbit by way of one or more dorsolateral foramina
(foa, Figs 16, 17, 19, 20). Immediately behind the ventral portion of this orbital artery groove
there is an area devoid of perichondral bone (aipl,Figs 13, 14, 15, 20). This area, which is
directed anteroventrally, was the articulation for the first infrapharyngobranchial.

The foramen for the orbitonasal artery (fona, Figs 15, 50) is formed by two notches within the
basisphenoid (nona, Figs 23, 24) and basioccipital (nona, Figs 16, 19, 20). This foramen
transmitted the orbitonasal artery up into the floor of the orbit (Gardiner & Bartram 1977: 230).
Lateral to this foramen the junction between the presumed prootics and basisphenoid remained
cartilage-filled, as the otico-sphenoid fissure, which is found in most specimens (fos, Figs 13,
15). In one specimen (BMNH P. 56483; Gardiner & Bartram 1977: fig. 3) and on one side only
this fissure (Gardiner 1973: 106) has been obliterated by bone.

The jugular canal is a short longitudinal canal whose posterior opening transmitted the
jugular vein, orbital artery and hyomandibular trunk of the facial nerve. The medial wall of the
jugular canal (prefacial commissure and pila antotica) is ossified and there are separate facial,
lateralis, trigeminal and profundus foramina. The geniculate ganglion lay in the funnel-like
opening of the facial canal (fhm. VII + pal, Figs 16,17,19, 20, 21 , 22) in the floor of the jugular
canal and was clearly extracranial. From the geniculate ganglion the palatine nerve passed down
into the back of the orbit and through the palatine fenestra (fpal, Fig. 20) while the
hyomandibular trunk passed back laterally in the floor of the jugular canal. Dorsal to the facial
canal and posterolateral to the trigeminal canal is a separate foramen (VII. lat, Figs 18, 22) which
is presumed to have transmitted the lateralis branches of the facial nerve. The corresponding
lateralis ganglion would have lain alongside the geniculate in the extramural chamber which
opens in front of the jugular canal (V + VII. lat, Fig. 19). In some specimens distinct grooves
pass from the mouth of the lateralis canal to the otic nerve foramina (fotn, Fig. 20) and to the
foramen of the ramus lateralis accessorius (frla, Figs 20, 22), whereas in others (BMNH
P. 53234) bridges of bone convert parts of these grooves into canals (cf. Figs 17, 19). The internal
opening of the lateralis canal (VII. lat, Fig. 18), which lies in the anterior opening of the utricular
recess, is smaller in diameter than the internal openings of the facial and trigeminal canals. The
facial canal (fhm. VII + pal, Figs 18, 26) originates outside the utricular recess and below the
bridge of bone which separates the trigeminal from the lateralis root.

The external opening of the trigeminal canal (V, Fig. 22) is medial to the lateralis canal and
dorsal to the facial canal. The opening is anteriorly-directed and lies in front of the jugular canal.
The internal opening is twice as large as that of the facial canal and originates in front of the
utricular recess. A bridge of bone internally (br, Fig. 18) separates the trigeminal root from the
lateralis root. There is a separate canal for the profundus nerve (prof, Fig. 18) which originates
in the anterior wall of the trigeminal canal and opens medial to the trigeminal foramen (prof,
Figs 16, 19, 20, 22). Medial to the profundus foramen are two further foramina lying one above
the other (III, Figs 16, 19, 20, 21); these presumably served for the two main branches of the
oculomotor nerve. In two specimens only (BMNH P. 56485, Fig. 22; BMNH P.53249, Fig. 25)
there is a single, large oculomotor foramen as in most other actinopterygians.

The basisphenoid region (Figs 22, 23, 24) consists of a hollow vertical pillar which flares
dorsally to join the orbital surface at the level of the oculomotor foramen. The basisphenoid
forms the lateral and posteroventral margins of the pituitary fossa (pitf, Fig. 26) and the large
hypophysial recess is open both dorsally and anteriorly. Ventrally, in the foot of the pillar, this
recess leads to a narrow bucco-hypophysial canal (bhc, Fig. 26) which passes through the
parasphenoid into the roof of the mouth (bhc, Fig. 50). Immediately behind the hypophysial

RELATIONSHIPS OF PALAEONISCIDS

219

frd

frla

fos

cao

aip 1

Fig. 20 Mimia toombsi Gardiner & Bartram. Braincase in anterodorsal view, looking up into the
rear of the orbit from the left side, from BMNH P.53259. Basisphenoid missing. The arrows depict
the courses of the nerves and vessels as they passed into the orbit.

220

B. G. GARDINER

RELATIONSHIPS OF PALAEONISCIDS

221

222

B. G. GARDINER

recess, in the foot of the basisphenoid pillar, ran the pituitary vein (pv, Figs 22, 23, 26).
Anteriorly the space for the pituitary vein is confluent with the hypohysial recess (Fig. 23). The
dorsum sellae (prob , Figs 25 , 26) , which forms the roof of the pituitary vein canal , is presumed to
be ossified by the basisphenoid. The posterior wall of the pituitary canal is expanded into a
short, stout pillar (cf. Fig. 23) which flares dorsally into the dorsum sellae. Anterodorsal to the
pituitary vein canal, in the anterior surface of the dorsum sellae, is a cup-shaped depression (svr,
Fig. 26). This housed the saccus vasculosus, and a median canal connecting the depression with
the pituitary vein canal presumably served for the passage of the saccus vasculosus vein.

Immediately in front of the ventral otic fissure and after giving off the orbitonasal artery the
internal carotid arteries entered the parabasal canal (fica, Fig. 1) in the floor of the
basisphenoid. Although the parabasal canal (pare, Fig. 22) runs the whole length of the
basisphenoid (between it and the parasphenoid) and opens anteriorly into the roof of the mouth,
only the posterior, enlarged portion, between the ventral otic fissure and the basipterygoid
process, housed the internal carotid artery. The internal carotid arteries (Gardiner & Bartram
1977: figs 5, 6), after passing through this enlarged posterior portion of the parabasal canal,
turned upwards and ran in a vertical canal in the anterior portion of the basisphenoid pillar
(fica2, Fig. 23) to enter the cranial cavity through the pituitary fossa (fica2, Figs 24, 26). In some
specimens (BMNH P. 56501; Gardiner & Bartram 1977: fig. 6) the internal carotid briefly ran in
a groove in the lateral wall of the basisphenoid pillar, and gave rise to an anterior branch which
ran anteroventrally towards the snout in a short groove on the dorsal surface of the basisphenoid
before passing down into the palatine (parabasal) canal. The palatine nerve entered the
parabasal canal through an anteroventrally-directed foramen (fpa!2, Figs 23, 24), immediately in
front of the ventral otic fissure, and presumably ran the entire length of the parabasal canal
before emerging in the roof of the mouth at the level of the vomers. It was probably
accompanied anteriorly by the palatine artery and vein.

Just in front of the basipterygoid process a short canal runs laterally from the parabasal canal
to open above the edge of the parasphenoid (fepsa, Figs 13, 22, 50). This carried the efferent
pseudobranchial artery which, after its anastomosis with the internal carotid (Gardiner &
Bartram 1977: figs 5, 6), turned upwards and forwards through a distinct foramen (fopa, Figs 13,
22, 23, 24) into the floor of the orbit as the ophthalmic artery.

f ica2

pitf

fos

:mm

Fig. 23 Mimia toombsi Gardiner & Bartram. Posterior basisphenoid region of braincase cut
horizontally at level of pituitary vein, in dorsal view, from BMNH P. 56504.

RELATIONSHIPS OF PALAEONISCIDS

223

fpal

nona

Fig. 24 Mimia toombsi Gardiner & Bartram. Preserved parts of basisphenoid in dorsal view, from

BMNH P.53225.

On the lateral wall of the basisphenoid pillar, dorsal to the ophthalmic artery foramen, is a
pronounced cup-shaped depression, divided into three components by prominent ridges (oem,
Fig. 22; see also Gardiner & Bartram 1977: fig. 6). This depression must have housed at least
three of the rectus muscles, but since there is not even a hint of a myodome the origin of the
fourth (external) rectus muscle can only be guessed at. Perhaps it also was attached to the lateral
wall of the pillar.

224

B. G. GARDINER

The posteroventral floor of the orbit behind the ventral otic fissure has smoothly curved walls
in the area of the facial foramen, but in front of the anteroventrally-directed canal for the
abducens nerve (VI, Figs 16, 19, 20, 21) it is often incompletely ossified and frequently
fenestrated (cf. Fig. 22). This is the area where the external rectus muscle might be expected
to originate. In the floor of this area there is often a short canal for the palatine nerve (fpal,
Fig. 20).

Above the jugular canal the walls of the orbital face consist of smooth bone only interrupted
by a distinct ridge running up towards the roof of the orbit from the trigeminal foramen. This
ridge passes posterior to the foramen for the trochlear nerve (IV, Figs 13, 16, 17, 19, 20, 21, 22).
Laterally the walls flare out to meet the postorbital processes and dorsally they meet the

dend

aesc

Soc

mcv

dpsc

fotc

I I

Pro

prob

not

vfon

Fig. 25 Mimia toombsi Gardiner & Bartram. Preserved post-ethmoid part of neurocranium in left
anterolateral view, from BMNH P. 53249. The right otico-orbitotemporal wall is missing; dotted
lines indicate broken surfaces.

RELATIONSHIPS OF PALAEONISCIDS

225

frontals. The two walls come very close together in the mid-line (cf. Fig. 16) and are scarcely
separated by the small, median optic fenestra (II, Fig. 13).

A further foramen (mcv, Figs 18, 20) opens into the roof of the recess for the lateralis and
geniculate ganglia. From this foramen a canal passes anterodorsally to open on the internal
surface above the saccular recess and behind the trochlear foramen (mcv, Figs 18, 25, 26). This
canal must have transmitted the middle cerebral vein. In one specimen (BMNH P. 53234, Fig.
19) the canal opens above the ganglion recess. Anterior to the ramus lateralis accessorius
foramen there is a series of up to four foramina (frd, Figs 20, 21, 22), which transmitted branches
of the superficial ophthalmic nerves to the neuromasts of the supraorbital canal (crd, Figs 33,

26 25 oL \ \ \
' 24 23 22 21

Fig. 26 Mimla toombsi Gardiner & Bartram. Post-ethmoid portion of neurocranium and
parasphenoid in sagittal section, from the left side, based on BMNH P. 53234. From Gardiner &
Bartram (1977). Key (diagram below): 1, dasc; 2, dpsc; 3, ssu; 4, pesc; 5, pdf; 6, plcc; 7, rmye; 8,
fotc; 9, X; 10, apse; 11, fm; 12, focn; 13, fboca; 14, not; 15, IX; 16, foca; 17, cao; 18, sacr; 19, Z; 20,
fhm.VII; 21, alig; 22, fv; 23, Psp; 24, prob; 25, pv; 26, pitf; 27, bhc; 28, fica2; 29, svr; 30, III; 31,
VII. lat; 32, V; 33, IV; 34, rtel; 35, ropl; 36, mcv; 37, acv; 38, rmet; 39, utr; 40, aasc; 41, aesc.

226

Ors

B. G. GARDINER

rsoc Pa

dend

gdend

fst

psc

plcc

fhm

foa

gpcv

9 X

gphX

svfotc

Fig. 27 Moythomasia durgaringa Gardiner & Bartram. Dorsal portion of otic and orbitotemporal
regions of neurocranium and attached dermal bones in posterior view, from BMNH P. 53227.

34). In some specimens (BMNH P. 56504, Fig. 17) these branches were contained for part of
their orbital course within a short canal, while in others (BMNH P. 53234, P. 53245, Figs 19, 21 A)
these branches passed up through the rim of the lateralis ganglion recess before passing out over
the orbital surface and into the four dorsal foramina.

The foramen for the anterior cerebral vein (acv, Figs 13, 14, 16, 19) opens into the roof of the
orbit just behind the dorsal anterior myodome (amyd, Fig. 13). It passes medially and originates
in the recess housing the telencephalon (Fig. 26). Occasionally this vein was developed on the
left side only (cotel, Figs 33, 34) as in the specimen of Kansasiella described by Poplin (1974: fig.
22) and in the Latimeria dissected by Robineau (1975: fig. 1A). The olfactory nerves were
sheathed by perichondral bone (I, Figs 16, 33, 34, 35), but a small gap in the roof of the olfactory
canal (gl, Fig. 13), where it passes beneath the floor of the dorsal anterior myodome, affords
communication with the orbit.

The brain is assumed to have been closely enveloped by bone, rather more completely than in
other palaeoniscids, and if so its shape and size may be accurately deduced. The relief of the
brain cavity is shown in sagittal section (Fig. 26), and the dorsal extent of the cranial cavity in
Figures 33, 34. The anterior dorsal fontanelle is reduced to the pineal foramen (pinf , Figs 16, 33,
34) which opens into that part of the cranial cavity which accommodated the diencephalon; the
complete closure of the anterior dorsal fontanelle in adults is considered a primitive
osteichthyan feature. Anterior to the diencephalon the cranial cavity is less broad where the
telencephalon was housed (ctel, Figs 33, 34; rtel, Fig. 26). Anteriorly this telencephalic cavity
may be overlain by inpushing of that part of the orbital wall (PI. 1 ; amyd, Fig. 33) which delimits
the dorsal anterior myodomes.

Posterior to the pineal foramen there is a marked increase in the breadth of the cranial cavity
(copl, Fig. 33). In sagittal section this appears as a marked, rounded depression (ropl, Fig. 26)
from which the trochlear nerve (IV) passed anteriorly into the orbit. This depression, which is
medial to the postorbital process, contained the large optic lobe from which the optic nerves

RELATIONSHIPS OF PALAEONISCIDS

fhm frla

227

fotn-,

frla

fst IX

frla

fvi

gst X

2mm

Fig. 28 Moythomasia durgaringa Gardiner & Bartram. Dorsal portion of otic and orbitotemporal
regions of neurocranium and attached dermal bones in right lateral view, from BMNH P. 53227.

passed out anteroventrally through the optic fenestra (II). The cranial cavity beneath the optic
lobes decreases rapidly in breadth and is floored by the 'prootic' bridge (dorsum sellae, prob,
Fig. 26). The walls in this region are perforated by the oculomotor foramen (III) and the floor in
front of the 'prootic' bridge by the pituitary fossa (pitf, Fig. 26). A further strong depression
(rmet, Fig. 26) behind the optic lobes and above the anterior ampullary chamber (aasc, Fig. 26)
and utricular recess (utr, Fig. 26) accommodated the cerebellum. The middle cerebral vein
(mcv) left the antero ventral corner of this depression on its way to the orbit. Below the
cerebellum the walls of the cranial cavity decrease in breadth much as beneath the optic lobes
and are perforated by the foramen for the trigeminal plus profundus nerves (V, Figs 18, 26) and
the foramen for the hyomandibular and palatine trunk of the facial nerve (fhm. VII + pal, Fig.
18). The lateralis branch of the facial nerve (VII. lat, Figs 18, 26), however, passed out through
the front of the recess for the utriculus (utr, Figs 18, 26).

There is a wide communication between the cranial and labyrinth cavities, as in other
palaeoniscids, Polypterus, Acipenser, Lepisosteus andAmia. Nevertheless the labyrinth cavity
is mostly enclosed within the bony walls of the otic region. The saccular recess is extensive and in
the form of an almost square pocket similar in size and shape to that of Pteronisculus (Nielsen
1942: fig. 14) but deeper than in Pholidophorus (Patterson 1975: fig. 66). Although this pocket is
in wide communication with the cranial cavity above, the zygal plates (Z, Fig. 26) form an inner,
dorsal wall to the pocket separating it (and the contained sacculus) from the floor of the brain.

From the serial sections it can be seen that a single otolith is present in each saccular recess. It
appears to be longer than deep and relatively compact. The glossopharyngeal nerve passed
through the dorsoposterior part of the saccular recess. Immediately above the glossopharyngeal
foramen (IX, Fig. 26) and below the opening of the lateral cranial canal there is a deep recess
which housed the ampulla of the posterior semicircular canal (apse, Figs 25, 26). The sinus
superior lay in front of and medial to the opening of the lateral cranial canal in a distinct concavity
(ssu , Fig . 26) in the cranial wall . Ventrally , in the region of the recess for the posterior ampullary

228 B. G. GARDINER

chamber, a flange of bone forms a partial posterior boundary to the sinus superior recess. A
similar flange of bone is found in Pholidophorus (Patterson 1975: fig. 65). A dorsal opening at
the front of the ampullary recess marks the exit of the external semicircular canal. Thus this
canal must have passed laterally through the anterior part of the ampullary recess as in
Kansasiella (Poplin 1974: fig. 20), Perleidus, Ospia, Caturus and Pholidophorus, whereas in
other palaeoniscids such as Pteronisculus, Kentuckia and Boreosomus the openings of the
posterior ampullary recess and the external canal are separated by a small pillar of bone. At the
top of the groove for the sinus superior lie the dorsal openings of the anterior (dasc, Fig. 26) and
posterior (dpsc) semicircular canals. The anterior semicircular canal is considerably longer than
the other two canals and anteriorly enters its ampulla (aasc, Figs 18, 26) in the anterodorsal
portion of the utricular recess. The external ampullary chamber (aesc, Figs 18, 26) lay in a
posterior diverticulum of the utricular recess behind and below a well-marked projection on the
posterior margin of the entrance to this recess. A similar projection has been described in
Pholidophorus (Patterson 1975: fig. 65) and in the dipnoan Griphognathus (Miles 1977: fig. 10).

Between the posterior semicircular canal and the recess for the sinus superior is an intramural
lateral cranial canal (Ice, Fig. 12). Posteriorly this canal opens into the cranial cavity (plcc, Figs
12, 25, 26) above the posterior ampullary recess and in front of the occipital fissure. This lateral
cranial canal (Jarvik 1980) is reduced to a small pocket in many specimens, where it is similar in
extent to that described in Kentuckia (Rayner 1951: fig. 9, X) and Kansasiella (Poplin 1974: fig.
20, elmy). In other less well ossified specimens the canal is more extensive (Fig. 12) and may
even communicate with the cranial cavity anteriorly (in front of the sinus superior) by several
small foramina. In other individuals the roof of the canal is fenestrated and widely open dorsally .
A lateral cranial canal is present in most palaeoniscids, Recent chondrosteans, Perleidus,
Lepisosteus, most halecomorphs, pholidophorids and leptolepids.

The sclerotic ring is well preserved and usually consists of either a complete ring or two
segments, each comprising three separate layers of bone. The outer layer is dermal and
ornamented with ridges of ganoine which run more or less concentrically round the ring (BMNH
P. 56483), whereas the underlying layers (inner and outer) are made up of very thin perichondral
bone (BMNH P. 56496). Where there are two segments the two halves are disposed dorsally and
ventrally, not fore and aft of the eyeball as in teleosts and fossil halecomorphs. In one specimen
of Mimia, however, there are three separate segments and in another, presumably a juvenile
(BMNH P. 53258), there are four separate plates. The innermost part of the sclerotic is ossified
as a thin, perichondral, basal sclerotic bone around the entry of the optic nerve and vessels
(BMNH P. 53228). This basal sclerotic bone is cup-shaped and like the outer perichondral layer
which underlies the dermal ring, has a corresponding inner layer of perichondral bone. Thus, bone is
developed on both surfaces of the sclerotic cartilage as in other primitive actinopterygians (Patterson
1975: 415), placoderms and agnathans (seep. 253).

Moythomasia durgaringa

The posterior face of the otic region is only partially lined with perichondral bone. A large ovoid
area (af, Fig. 27), stretching from just above the vagus canal almost to the opening of the
endolymphatic duct, has no perichondral lining and must have been cartilage-filled during life.
A similar loss of the perichondral lining is presumed to have occurred early in the phylogeny of
dipnoans and osteolepiforms (Gardiner 1973: 111). A notch (gph.X, Fig. 27) in the groove for the
vagus nerve served for the passage of the pharyngeal branch of that nerve as in some specimens of
Mimia, but the foramen for the endolymphatic organ (dend) is considerably larger than the
corresponding foramen in Mimia. The lateral face of the otic region (Fig. 28) only differs from that of
Mimia in the more ventral position of the raised areas for the origin of the dorsal hyoid constrictor
muscle (oahm + oaop).

The jugular canal is a trifle longer than in Mimia and the orbital artery entered posteriorly by a
separate vertical canal (goa, Fig. 7). In one specimen (BMNH P. 53227, Fig. 31), in which the
orbital region is broken open, the individual canals for the various nerves opening into the
jugular canal can be recognized as distinct tubes of perichondral bone.

The facial canal (VII, Figv 31), which transmitted the palatine and hyomandibular trunk of the

RELATIONSHIPS OF PALAEONISCIDS

229

230

B. G. GARDINER

facial nerve, opens into the ventromedial corner of the orbital opening of the jugular canal . This
external opening is confluent with a palatine fenestra (fpal, Fig. 29) in which the geniculate
ganglion must have lain. Thus Moythomasia is unique amongst palaeoniscids in possessing a
prepalatine (or prefacial) floor (prepf, Fig. 29) to the jugular canal.

Lying immediately above the facial canal the lateralis canal (VII. lat, Fig. 31) opens together
with the trigeminal canal (V, Fig. 31) into a large pocket in the mouth of the jugular canal, dorsal
to the palatine fenestra. This pocket must have housed both the lateralis and gasserian ganglia.
Two grooves in the medial wall of this pocket run upwards from the mouth of the lateralis canal.
One of the grooves passes up towards the foramen for the otic nerve and must have transmitted
the otic branch (rot, Fig. 31), while the more medial groove served for the dorsal branch of the
superficial ophthalmic nerves and the recurrent branch of the facial nerve (rdo + rla). From this
medial groove the ophthalmic and recurrent branches entered a short canal to re-emerge in the
back of the orbit (frd, Fig. 29) before finally entering separate foramina in the orbital roof (frla,
frd, Fig. 30). The trigeminal canal runs dorsal to the lateralis canal with the canal for the
profundus (prof) running anterior to both. The profundus canal has a separate internal opening,
whereas externally it opens in front of the pocket for the gasserian and lateralis ganglia (prof,
Figs 29, 30). A pedicel of bone (ped, Figs 29, 30) often covers this pocket and is homologous with
the pterosphenoid pedicel in Amia and Pholidophorus . In some specimens (BMNH P. 56480,
Fig. 30) it is complete and spans the mouth of the jugular canal. A further foramen beneath the
posterior margin of the pterosphenoid pedicel is presumed to have transmitted the middle
cerebral vein (mcv, Fig. 31).

The internal carotid artery, after passing through the enlarged posterior portion of the
parabasal canal, turned upwards to emerge in the floor of the orbit in the foot of the basi-
sphenoid pillar (fopa, Figs 7, 32), from whence the ophthalmic artery ran forwards along the

frd

frla

spic

frd
V+VI I lat+mcv

ped
prof

foa

fhm VI I + pal

Fig. 30 Moythomasia durgaringa Gardiner &Bartram. Preserved part of the rear of the right orbit in
oblique anterolateral view, from BMNH P. 56480.

RELATIONSHIPS OF PALAEONISCIDS

frla

231

mcv

fotn

Fig. 31 Moythomasia durgaringa Gardiner &
Bartram . Sketch of the rear of the right orbit in
anterior view, from BMNH P. 53227. Mouth of
the trigeminofacial chamber is drawn as if cut
away, and the cut surface cross-hatched. The
passage of individual nerves is represented by
arrows.

rot' ~~l

foa

floor of the orbit while the internal carotid ran upwards in a groove in the anterolateral wall
of the basisphenoid pillar to enter the cranial cavity through the pituitary fossa.

Muscle scars, in the form of two distinct cups, one above the other, are present on the
basisphenoid bolster dorsal to the foramen for the internal carotid artery, in an identical position
to those in Mimia. Whether or not the external rectus muscle originated here or in the back of
the orbit ventrolateral to the abducens foramen (oexr, Fig. 29) could not be determined with
certainty. However, the otico-sphenoid fissure is closed by bone anteriorly and laterally, where
the ascending process of the parasphenoid bridges it, but posteriorly it is open for a short
distance (fos, Fig. 7).

The sclerotic ring consists of two segments of dermal bone as in some specimens of Mimia, but
no evidence of perichondral ossifications could be found. In the closely allied Moythomasia
nitida four dermal plates have been described (Jessen 1968: fig. 12).

Otic and orbitotemporal region: discussion

1. Parampullary process. This is a prominent feature of the opisthotic in Mimia and
Moythomasia, and is also prominent in Kentuckia (Rayner 1951: fig. 7), Birgeria (Nielsen 1949:
fig. 60), 'Ambipoda' (Beltan 1968: pi. 6) andAustralosomus (Nielsen 1949: fig. 7). In Polypterus
the opisthotic ossifies late in ontogeny from a centre over the base of the posterior semicircular
canal and in the adult it is a large ossification with extensive membrane bone components. It has
a strongly-developed ridge extending dorsoposteriorly along its length. Under this ridge the
hyomandibular and opercular adductor muscles take origin and the branchial levator muscles
originate more posteriorly. The posterior portion of this opisthotic ridge in Polypterus is taken
to be the homologue of the parampullary process in palaeoniscids, and the parampullary process
is taken to have developed primitively in relation to the branchial levator muscles. No
parampullary process as such is present on the opisthotic of Perleidus or parasemionotids,
though it seems likely that the branchial levator muscles must have been attached to this bone,
since the intercalar is still small and has not grown over the cranial fissure. In more advanced
actinopterygians the parampullary process is often difficult to recognize because that region of
the opisthotic on which the branchial levator muscles originate has been captured by
anteroventrally-directed membrane outgrowths from the intercalar. Thus inAmia the branchial
levator muscles originate entirely on the membranous intercalar. Presumably most of these

232

B. G. GARDINER

pare

bhc

pi tf

nona

fpal

Fig. 32 Moythomasia durgaringa Gardiner & Bartram. Preserved parts of basisphenoid in dorsal

view, from BMNH P.53219.

muscles originated on the intercalar in caturids also, but in ' Aspidorhynchus' (Patterson 1975:
fig. 99, prim), Macrepisteus (Schaeffer 1971: fig. 5) and Heterolepidotus (Gardiner 1960: fig. 29)
a knob on the surface of the prootic must have served for the origin of at least the anteriormost
branchial levators.

In Devonian dipnoans the adotic process, a knob which arises from the ventral edge of the
jugular groove behind the glossopharyngeal foramen, homologous with the adotic eminence of
Devonian actinistians (Nesides Bjerring 1977: fig. 23A) and the 'process for the attachment of
adductor muscles of hyomandibula' of Eusthenopteron (Bjerring 1971: fig. 8), is said by Miles
(1977: 79) to be a similar outgrowth to the caturid prootic knob. But since the adotic process
lies partly below the jugular canal it seems more likely that it received either the ventral portion
of the first branchial levator muscle, which in Polyptems is attached to the parasphenoid, or the
ceratobranchial ligament (cf. Polypterus, Allis 1922: 234). In pholidophorids, where the
intercalar has extensive membrane bone outgrowths covering the adjacent otic bones, the
parampullary process of the opisthotic can still be recognized (Patterson 1975: fig. 61, ampp). In
pholidophorids and Upper Jurassic leptolepids struts of bone from the prootic and intercalar
unite to form a bridge over the subtemporal fossa. Elops and Osteoglossum have a similar
bridge, and here the branchial levator muscles originate, as well as the ligamentous attachment

RELATIONSHIPS OF PALAEONISCIDS 233

of the first suprapharyngobranchial. There is a distinct parampullary process in actinistians
which is borne on the opisthotic in Macropoma, Laugia and Wimania. In the Recent Latimeria a
cartilaginous process in an homologous position is the origin of both the branchial levator
muscles and the ligament of the suprapharyngobranchial. A parampullary process may also be
recognized in rhipidistians such as Ectosteorhachis (Romer 1937: figs 2,4,5, popcp ?) where its
relationships to the foramina for the glossopharyngeal and vagus nerves (particularly the
supratemporal branches) are exactly as in Mimia and Moythomasia. A similar process in
Eusthenopteron (Jarvik 1954: fig. 1, prpo) also served as a point of articulation for the first
suprapharyngobranchial. In the Devonian dipnoan Griphognathus (Miles 1977: 79) a small
ventral outgrowth from the upper margin of the jugular groove, between the foramina for the
vagus and glossopharyngeal nerves, may also have given origin to branchial levator muscles. No
such process has been described in acanthodians, placoderms or chondrichthyans. The presence
of a parampullary process on the opisthotic is presumed to be a primitive osteichthyan character.

2. Articulation of first suprapharyngobranchial. In the Gogo palaeoniscids there is no obvious
facet for the articulation of the spatulate first suprapharyngobranchial. It rested against or
articulated with the opisthotic region of the braincase in front of the parampullary process and
below the jugular canal (Fig. 119). In Polypterus, while there is no suprapharyngobranchial as
such, the first epibranchial articulates with the opisthotic below the jugular canal and in front of
the glossopharyngeal foramen (Devillers 1958: 665). In sturgeons the first suprapharyngobran-
chial articulates with the opisthotic (when present) below the jugular canal, but the second
suprapharyngobranchial articulates with the braincase above the jugular canal (Bertmar 1959:
305, 329). The first suprapharyngobranchial also articulates with the otic region in Polyodon
(Bridge 1878). In Amia, as in Polypterus, there is no suprapharyngobranchial. In Lepisosteus
the cartilaginous first suprapharyngobranchial does not articulate with the braincase: it lies
behind the glossopharyngeal nerve. In teleosts which have retained an ossified first
suprapharyngobranchial such as Elops, and other members of the families Elopidae and
Alepocephalidae, this element inserts by a ligament together with the branchial levator muscles
on the intercalar strut. The first pharyngobranchial in Latimeria is also in ligamentous
attachment to the parampullary process (Millot & Anthony 1958) and in Eusthenopteron (Jarvik
1954: fig. 23) the pharyngobranchial is said to articulate directly with that process.

Elsewhere in actinopterygians the pattern is variable. In Birgeria (Nielsen 1949: fig. 60) the
suprapharyngobranchial articulated with a distinct facet, lacking perichondral lining, on the
parampullary process of the opisthotic, as in Eusthenopteron. On the other hand in
Pteronisculus (Nielsen 1942: 196) the first suprapharyngobranchial articulated with a large facet
lacking perichondral lining on the posteroventral portion of the opisthotic below the
glossopharyngeal foramen. This articulation is below the jugular groove and is in a similar
position in Mimia, Moythomasia and Acipenser. A smaller, paired articulatory surface, lying
rather more anteriorly but still below the jugular canal, served for the articulation of the first
suprapharyngobranchial in Kansasiella (Poplin 1974: fig. 13). In the pholidopleurid
Australosomus (Nielsen 1949: fig. 37) the articulatory facet is in an identical position to that in
Pteronisculus. In caturids (Caturus Gardiner 1960: fig. 36; 'Aspidorhynchus' Patterson 1975: fig.
99, asup.l; Heterolepidotus Patterson 1975: fig. 102, asup.l; Osteorachis Patterson 1975: 397)
the articular area is equally distinct, and as in Pteronisculus, Kansasiella and Australosomus lies
just below the glossopharyngeal foramen, in a notch in the margin of the intercalar. Elsewhere
within amioids an articulatory facet is not recognizable nor is one to be seen in semionotids,
pachycormids or leptolepids. The only other recorded occurrence of a distinct facet for the
articulation of the first suprapharyngobranchial is in pholidophorids, where it is said to lie below
the jugular groove on the prootic, midway between the glossopharyngeal and facial foramina
(Patterson 1975: 397). This position is considerably more anterior and more ventral than in any
other actinopterygian, rhipidistian or actinistian. It seems likely that this facet on the prootic was
not for the first suprapharyngobranchial (which was probably in ligamentous contact with the
intercalar strut as in Elops) but for the second infrapharyngobranchial. The second
infrapharyngobranchial in Elops lies in close proximity to the posterior portion of the prootic,

234 B. G. GARDINER

with its head in a similar position to the facet described by Patterson (1975: fig. 56, asup.l) in
Pholidophorus . The second infrapharyngobranchial also articulates with the braincase in
Acipenser and Polyodon (in front of the vagus foramen) and in Eusthenopteron where it
articulates with the basioccipital region (as does the first infrapharyngobranchial). The
condition in Eusthenopteron is similar to Australosomus (Nielsen 1949: 122) except that in the
latter the second infrapharyngobranchial merely lies adjacent to the underside of the
basioccipital.

There are no known pharyngobranchials in dipnoans (Miles 1977: 287) but cartilaginous
nodules are said to underlie the medial ends of the epibranchials in Neoceratodus (Nelson 1968:
fig. 5D). Since no suprapharyngobranchials are known in placoderms, chondrichthyans or
acanthodians they are presumed to be a derived feature of osteichthyans (Rosen etal. 1981; see
also under branchial arches, p. 362).

3 ^Articulation of first infrapharyngobranchial. In Mimia and Moythomasia the articulation of
the first infrapharyngobranchial is represented by an area devoid of perichondral lining at the
posteroventral corner of the prootic , posterior to the ventral otic fissure and immediately behind
the groove for the orbital artery. Elsewhere in osteichthyans the first infrapharyngobranchial
(=pharyngobranchial) articulates with the braincase posterior to the ventral otic fissure in the
actinopterygian Cosmoptychius (Schaeffer 1971: fig. 8), and in the rhipidistians Eusthenopteron
(Jarvik 1954: fig. 1) and Ectosteorhachis (Romer 1937: fig. 2). In most actinopterygians the
articulation lies anterior to the fissure owing to the presumed posterior migration of the ventral otic
fissure (Gardiner 1970; Gardiner & Bartram 1977). Possible exceptions to this are Polypterus,
Polyodon and Acipenser, in which the limits of the fissure are not precisely determinable, and
Australosomus (Nielsen 1949: 122) in which the articulation straddles the ventral otic fissure. As a
result of the rearward growth of the parasphenoid in Polypterus, later palaeoniscids and higher
actinopterygians (Gardiner 1973: 115) the first infrapharyngobranchial has often become
secondarily associated with it. Thus the first infrapharyngobranchial articulates with the
parasphenoid in Pteronisculus (Nielsen 1942: fig. 45), Polypterus, Acipenser, Amia, Lepisosteus,
Upper Jurassic leptolepids (Patterson 1975: 398) and many other teleosts, whereas in Polyodon it
articulates in the notch between the ascending and posterior processes of the parasphenoid. In
pholidophorids (Patterson 1975: 398), parasemionotids, most caturids (Caturus, Patterson 1975:
398; Heterolepidotus, Patterson 1975: fig. 102; 'Aspidorhynchus', Patterson 1975: fig. 99),
pachycormids (Pachycormus , Patterson 1975: fig. 106) and semionotids (Dapedium, Patterson 1975:
fig. 112; Lepidotes, Patterson 1975: fig. 108) the situation is as in Polyodon except that there is a
well-marked facet on the prootic and a notch in the overlying parasphenoid. In all these
actinopterygians the articulation remains (as far as can be deduced in the fossil forms) approximately
on the level at which the lateral aortae give rise to the orbital arteries. Primitively then in
osteichthyans the first infrapharyngobranchial articulated with the prootic behind the ventral otic
fissure and orbital artery.

The pharyngobranchials of acanthodians are homologous with those of selachians and with
osteichthyan supra + infrapharyngobranchial (see p. 362). The pharyngobranchials of
selachians and acanthodians project posteromedially (Nelson 1968; Miles 1973a: 96; Jarvik 1977: fig.
8) and this is primitive for gnathostomes. Pharyngobranchials in chondrichthyans and placoderms
are usually located posterior to the neurocranium, but in Heterodontus and several other sharks as
well as holocephalans the first pharyngobranchial lies close to the underside of the occiput though
never articulating with it. This latter condition is considered to be derived, as suggested by Miles
(1971a).

According to Miles (1973«: 88) the second pharyngobranchial in Acanthodes articulated with
the ventral occipital ossification by a facet just behind a groove for an efferent branchial artery.
This would be impossible if the pharyngobranchial were backwardly projecting; I suggest the
facet may have served for the articulation of the first epibranchial.

The change in branchial arch suspension in osteichthyans, with the development of the
forwardly-directed first infrapharyngobranchial (articulating with the braincase) and the
development of the suspensory first suprapharyngobranchial, is presumed to be related to the

RELATIONSHIPS OF PALAEONISCIDS 235

increasing importance of the levator arcus palatini muscles and to the hyoid bar pump in
expanding the orobranchial chamber.

4. Lateral commissure and trigeminofacialis chamber. The lateral commissure is penetrated by
the jugular canal and forms the side-wall to the trigeminofacialis chamber in osteichthyans. It
was first described in actinopterygians (Amia, Lepisosteus, Salmo], where it is formed (de Beer
1926: 332; 1937: 391) by the junction of the prootic process (developed from the otic capsule)
with the basitrabecular and postpalatine processes (developed from the edge of the basal plate).

There has been much discussion as to whether the lateral commissure has a neurocranial or
visceral origin. In actinopterygians there is little doubt that the commissure is entirely
neurocranial in origin according to the work of Swinnerton (1902) on Gasterosteus, de Beer
(1926, 1937) on Amia and Salmo, Hammarberg (1937) on Lepisosteus, Hubendick (1943) on
Leuciscus, Daget & d'Aubenton (1957) on Heterotis and Bertmar (1959) on Hepsetus. Only
Holmgren (1943: 33, 37, 42) suggested that the lateral commissure in actinopterygians is a
visceral structure, because in Acipenser, Amia and Lepisosteus he found a membranous basal
connection between the palatoquadrate and trabecular region, which he homologized with the
spiracular cartilages in sharks. Since there is no suggestion of a transfer of cartilage from the
palatoquadrate (or any other visceral source) to the neurocranium Holmgren's assumptions
seem ill-founded. Bertmar (1959: 339) reinvestigated Holmgren's (1943) material and
concluded that the lateral commissure in Acipenser, Amia and Lepisosteus is a primary
neurocranial structure, as de Beer (1926: 332) had originally said.

The initial suggestion that the lateral commissure was of visceral origin was made by Allis
(1914a), who maintained that in Neoceratodus it came from the mandibular arch. Holmgren
(1940; 1943: 43) later claimed to have furnished complete evidence that the lateral commissure
in sharks was derived from the mandibular arch. This evidence consists of a membranous
connection between the postorbital process and the basiotic lamina in embryo Squalus
(Holmgren 1940: fig. 67), a membranous connection between the same process and the
hyomandibula in Etmopterus (Holmgren 1940: figs 81, 89) and between the postorbital process
and the spiracular cartilages in Raja (Holmgren 1940: 184). In none of Holmgren's descriptions
is there any indication of chondrification within this membrane, other than the formation of
spiracular cartilages ventrally. Jollie (1971: 37) confirmed the mandibular arch origin of the
lateral commissure in Squalus, even though he regarded the structure as a mandibular
commissure, not as the lateral commissure proper, which he equated with part of the otic
capsule unrelated to the jugular vein. Bertmar (1959: 314, 339) pointed out that while Allis's
(19140) deductions were founded on imperfect information in Neoceratodus, Holmgren (1940,
1943) had confused the lateral commissure of sharks with mandibular arch structures (see also
Bjerring 1967: 262). However, Bertmar (1959: 314; 1963: 337) went on to suggest that the lateral
commissure in Neoceratodus is derived from the hyoid arch and represents fused infra- and
suprapharyngohyals, and thus added some credence to the theory of the hyal origin of the
commissure in Eusthenopteron proposed by Jarvik (1954: 75).

The only possible confirmation of Bertmar's (1959) theory would be to show that the lateral
commissure in selachians is of hyoid arch origin, which is exactly what Jollie (1971: 37)
proposed. Unfortunately, Jollie homologized the lateral commissure of actinopterygians with
the otical shelf of sharks, despite the fact that a lateral commissure exists in many selachians
(Oxynotus, Scymnodon, Centrophorus, Cladodus; Holmgren 1940, 1941) and is massive in
Squatina and many fossil sharks (Xenacanthus, Tamiobatis, Hybodus). Thus from the evidence
presented by Holmgren (1940), El-Toubi (1949) and Jollie (1971) for Squalus, and Holmgren
(1940) for Etmopterus, there is no reason to suppose that the lateral commissure in selachians
forms in a manner significantly different from that in actinopterygians. Furthermore, although
the lateral commissure is missing in hexanchoids, galeomorphs, many rays, torpedoes and
chimaeroids, its presence in other selachians, osteichthyans, placoderms (Young 1980) and
acanthodians (Miles 1973a: fig. 4) suggests it is a primitive feature of gnathostomes.

Allis (19146, 1919) first introduced the term 'trigeminofacialis chamber' for the space in the
side wall of the braincase of actinopterygians immediately in front of the auditory capsule. This

236 B. G. GARDINER

chamber is made up by the pars ganglionaris and the pars jugularis. Allis pointed out that the
trigeminofacial chamber is single in Amia and Lepisosteus because the pars ganglionaris and
pars jugularis are confluent, but that in Scomber and the scorpaenoid teleosts the chamber is
divided. Goodrich (1930: 277), knowing that in selachians the trigeminal and facial ganglia are
intramural, decided that the condition in Amia and Lepisosteus, in which an intramural recess
(pars ganglionaris) is confluent with an extramural recess (pars jugularis), represents the basic
configuration for actinopterygians. He further believed that in teleosts the trigeminofacial
chamber is secondarily divided by a bony wall. This view was also held by de Beer (1937: 56,
428), who qualified it by pointing out that the trigeminofacial chamber of Amia could be derived
from the condition in Squalus if the acustico-trigeminofacialis recess and the jugular canal of the
latter 'were thrown into one'. Earlier (de Beer 1926), however, he had shown that the
trigeminofacial chamber is separate from the jugular canal in the development of Acipenser and

Amia.

In chondrichthyans the pars jugularis is separated from the pars ganglionaris by the lateral wall of
the neurocranium (prefacial commissure), whereas the medial wall of the trigeminofacialis chamber
(prefacial commissure + pila antotica) is invariably complete and ossified in palaeoniscids, fossil
halecostomes, halecomorphs and all the major groups of teleosts. From this it is clear that the
condition in Amia, Lepisosteus and certain advanced teleosts in which the prefacial commissure and
pila antotica fail to ossify, thereby allowing the chamber to communicate widely with the cranial
cavity, is specialized.

Unlike chondrichthyans the facial and trigeminal ganglia are primitively extracranial in
actinopterygians. This is certainly the case in palaeoniscids, Polypterus, Acipenser, caturids,
Lepidotes, pholidophorids and early leptolepids, whereas in Ichthyokentema, Upper Jurassic
leptolepids and primitive living teleosts the geniculate and gasserian ganglia are partly or wholly
intracranial (Patterson 1975: 401). Schaeffer (1971: 7) attempted to reconcile the varying
locations of the two ganglia in the jugular canal of different actinopterygian groups, by
suggesting that the term 'trigeminofacialis chamber' be restricted to the extramural cavity
between the lateral cranial wall and the lateral commissure. If we accept this simple definition
then it is obvious that the trigeminofacialis chamber of primitive actinopterygians (palaeoniscids,
Polypterus, Acipenser, etc.) is not very different from that of selachians such as Oxynotus, Squatina
and Squalus. Since such a chamber is also found in placoderms (Brindabellaspis , Young 1980: fig. 10)
it must be considered a primitive gnathostome character. Nevertheless, because the geniculate and
gasserian ganglia have varied relationships to this extramural cavity (trigeminofacialis chamber) and
to the mouth of the jugular canal in actinopterygians and selachians, it is necessary to establish the
primitive condition in actinopterygians, osteichthyans and gnathostomes.

(a) ACTINOPTERYGIANS. In most palaeoniscids the lateral commissure is a massive endochondral
structure formed by the prootic and penetrated by a long jugular canal. The commissure is also
massive in Polyodon but is less extensive in Polypterus, Acipenser, Mimia and Moythomasia and
is considerably reduced in Amia, Lepisosteus, parasemionotids, caturids, semionotids,
pachycormids and pholidophorids. The lateral wall of the jugular canal in many palaeoniscids is
perforated by several other canals; a ventral one which transmitted the orbital artery, and one or
more dorsal foramina which transmitted the hyomandibular trunk or its branches, or both.
Actinopterygians with three separate posterior openings (for the jugular vein, orbital artery and
hyomandibular trunk) include Pteronisculus, Kentuckia, Kansasiella, Perleidus, leptolepids and
other primitive teleosts. Patterson (1975: 400), however, has demonstrated that the condition in
teleosts is secondary and developed as a result of extensive membrane bone outgrowths.

The jugular canal may open posteriorly by a single foramen transmitting the jugular vein,
orbital artery and hyomandibular trunk as in Mimia, Acipenser, parasemionotids, caturids,
pachycormids, Lepidotes, Lepisosteus, Dapedium and pholidophorids; or the orbital artery may
enter by a separate vertical canal as in Kansasiella, Moythomasia, Boreosomus, Australosomus,
Pteronisculus cicatrosus, Saurichthys, Polyodon and Amia; or the orbital artery may pass
outside the commissure as in Polypterus.

The facial canal, which primitively transmitted only the palatine and hyomandibular trunk of the

RELATIONSHIPS OF PALAEONISCIDS 237

facial nerve, opens into the ventromedial corner of the orbital opening of the jugular canal in
Pteronisculus, Mimia, Moythomasia, Kentuckia, Boreosomus, Saurichthys, Perleidus and
parasemionotids and the geniculate ganglion lay in the floor of the jugular canal. In Acipenser the
facial canal emerges in the orbit and the hyomandibular trunk turns posteriorly to traverse the
jugular canal.

In Polyptems and Kansasiella the facial canal opens into the middle of the jugular canal and
the geniculate ganglion lay within the jugular canal.

In caturids, Lepidotes, pachycormids and pholidophorids, where the lateral commissure is
reduced, the facial canal opens just behind the jugular canal.

The lateralis branches of the facial nerve issue through the trigeminal foramen in
palaeoniscids, pholidophorids, pachycormids, leptolepids and other teleosts, and in most of
these fishes the external opening of the trigeminal canal lies anterodorsal to the facial foramen, in the
upper part of the orbital opening of the jugular canal. In Polypterus there is a separate lateralis
ganglion (lateralis -communis of Allis 1922: 274) dorsal to the trigeminal ganglion. This lateralis
ganglion is partly extracranial and partly intracranial and is continuous with the intracranial portion
of the facialis ganglion. In Amia a similar lateralis ganglion lies above the gasserian ganglion. In
Polyodon and Acipenser the lateralis branches of the facial nerve emerge into the orbit through
separate foramina (superficial ophthalmic and otic branches). This is considered to be a
specialization following loss of the prefacial commissure in Acipenser. Both caturids and Amia have
an intramural canal for the superficial ophthalmic nerves, and in Ospia and some parasemionotids
the trigeminal canal is divided by a horizontal partition, with the superficial ophthalmic nerves
passing out separately through the dorsal part (Patterson 1975: 405).

In the Gogo palaeoniscids there is a separate foramen dorsal to the facial canal and
posterolateral to the trigeminal canal. Although no such foramen or canal has been reported in
any other actinopterygian, by comparison with Polypterus and Amia this foramen must have
transmitted the lateralis branches of the facial nerve.

The facial and trigeminal canals originate in front of the recess for the utriculus in Mimia,
Moythomasia, Pteronisculus, Australosomus , and pholidophorids, but in Mimia a bridge of
bone separates the facial canal from the lateralis canal and the latter originates in the most
anterior part of the utricular recess. In many other palaeoniscids such as Kansasiella, Kentuckia
and Boreosomus, and in Perleidus and fossil and living neopterygians (parasemionotids,
Caturus, Lepidotes, Dapedium, Amia, Lepisosteus), the facial and trigeminal canals originate in
the utricular recess. This is considered a specialization (Patterson 1975: 408).

In Mimia and all previously described palaeoniscids, chondrosteans, perleidids and
pholidopleurids (Pteronisculus, Kentuckia, Kansasiella, Acipenser, Polyodon, Boreosomus,
Birgeria, Perleidus, Australosomus) the facial canal opens into the orbital opening of the jugular
canal (but see Lehman 1969), the geniculate ganglion lay in the floor of the canal, the palatine
nerve passed down into the parabasal canal along the hind wall of the orbit, and there is no
prefacial (prepalatine) floor to the jugular canal. This is also the condition in parasemionotids,
Lepisosteus and Amia except that in some of the more fully ossified specimens of
parasemionotids (Patterson 1975: 405) there are rudiments of a prepalatine strut.

In Moythomasia, however, the external opening of the facial canal is confluent with a palatine
fenestra, a rather large opening in the floor of the jugular canal. Thus Moythomasia is the only
palaeoniscid so far described with a prefacial floor to the jugular canal. This prefacial floor has
the same proportions as in Pholidophorus and Pachycormus (Patterson 1975: 404; fig. 64) and
the geniculate ganglion must have lain within the palatine fenestra. In Lepidotes, Dapedium,
Leptolepis and many living teleosts the opening in the floor of the jugular canal decreases in size
so that only a palatine foramen remains. The palatine nerve passed through this opening into the
myodome and the geniculate ganglion must have lain in the extramural space in the floor of the
jugular groove.

In most caturids (Heterolepidotus, ' Aspidorhynchus' , Caturus, Macrepistius) the prepalatine
floor is represented by a slender strut, but in the Caturus described by Rayner (1948: fig. 5) the
floor is more complete and similar to that in Moythomasia and Pholidophorus.

The otic nerve canal originates in the wall of the orbit above the opening of the jugular canal in

238 B. G. GARDINER

Mimia, Moythomasia, Kansasiella, Perleidus, Polypterus, parasemionotids, Dapedium, Lepi-
dotes, Pachycormus and Pholidophoms , but in Pteronisculus, Kentuckia, Boreosomus,
Australosomus, Heterolepidotus, Caturus, Macrepistius, Amia and Lepisosteus it originates in
the roof of the mouth of the canal . The otic nerve canal passes through the postorbital process to
join the spiracular canal in Boreosomus, Australosomus, Lepidotes, Pachycormus, and
Pholidophorus , while in Kansasiella (Poplin 1974: fig. 12) it is only just excluded from the top of
the spiracular canal. But in Pteronisculus, Kentuckia, Saurichthys, Acipenser, parasemionotids,
Dapedium, Heterolepidotus, Lepisosteus and Amia the otic canal opens into the floor of the fossa
bridgei, medial to the spiracular canal. In larval Amia (Goodrich 1930: fig. 733) the otic nerve passes
into the top half of the spiracular canal and not into the fossa bridgei, and in Polypterus the otic nerve
perforates the postorbital process, passes beneath the ampulla of the anterior semicircular canal and
emerges on the roof of the neurocranium medial to the most anterior portion of the spiracle. In
Mimia and Moythomasia the otic nerve canal similarly passes through the postorbital process medial
to the spiracular groove.

The superficial ophthalmic nerves are believed to have emerged through the trigeminal
foramen in palaeoniscids, Perleidus, Dapedium, Lepidotes, Pachycormus and Pholidophorus.
In Moythomasia there is a short canal leading upwards from the roof of the recess for the
trigeminal and lateralis ganglia. This canal passes up through the wall of the orbit
(pterosphenoid pedicel) and must have served for the superficial ophthalmic nerves. Loss of the
separate lateralis canal and enlargement of the external opening of the trigeminal canal would
lead to the condition seen in Kentuckia, Ospia, caturids and Amia.

There is a separate profundus foramen in Mimia, Moythomasia, Kentuckia, Kansasiella,
Pteronisculus, Perleidus, parasemionotids, caturids, Lepidotus, Lepisosteus, Dapedium,
Pachycormus, pholidophorids and the Sinemurian Leptolepis, but in all other leptolepids there
is no separate profundus foramen. In some living teleosts there is a separate profundus foramen,
but in many others the nerve enters the orbit through the oculomotor foramen, as in Polypterus.
There is no sign of a profundus foramen in Boreosomus, Saurichthys or Australosomus and there
is no separate foramen in Amia. From this evidence it is not clear whether a separate profundus
foramen is primitive. In Mimia the root of the profundus, together with the trigeminal nerve,
passed into the base of the trigeminal canal. The trigeminal canal immediately divides and the
profundus passed forwards and downwards to enter the orbit through a separate foramen. The
condition in Boreosomus, Saurichthys and Australosomus could be derived from that in Mimia
merely by enlargement of the trigeminal canal, so that the profundus emerged with the
trigeminal nerve (see also Nielsen 1949: 58), as in selachians.

From this survey I conclude (as did Patterson 1975: 408) that primitively in actinopterygians
the lateral commissure was massive and that the posterior lateral wall of the jugular canal was
perforated by the hyomandibular trunk dorsally and the orbital artery ventrally. The
extracranial gasserian and geniculate ganglia lay in the mouth of the orbital opening of the
jugular canal and the facial and trigeminal canals originated in front of the utricular recess. The
facial canal transmitted the palatine nerve and hyomandibular trunk, with the latter turning
posteriorly to traverse the jugular canal and with the palatine nerve passing into the parabasal
canal. The trigeminal canal opened into the orbit dorsomedial to the facial canal and the otic
nerve canal originated in the wall of the orbit above the jugular canal and passed up onto the roof
of the neurocranium medial to the spiracle.

(b) OSTEICHTHYANS. In actinistians the lateral commissure is well developed (Nesides, Bjerring
1977: fig. 23a; Macropoma, Latimeria, Millot & Anthony 1958) and is similar to that in the Gogo
palaeoniscids. The jugular canal is long and transmits the hyomandibular trunk, orbital artery
and jugular vein in Latimeria. In rhipidistians such as Eusthenopteron (Jarvik 1954: fig. 1),
Ectosteorhachis (Romer 1937: fig. 2), Porolepis (Bjerring 1967: 224) and Glyptolepis (Jarvik
1972: fig. 21) the lateral commissure is also massive and the jugular canal is similar in length to
that in palaeoniscids and actinistians. The lateral commissure is not easily recognizable in adult
dipnoans owing to fusion of that region with the palatoquadrate, but Bertmar (1963: 337)
believed it can readily be distinguished in embryos of Neoceratodus . A long jugular canal is

RELATIONSHIPS OF PALAEONISCIDS 239

present and transmits the hyomandibular trunk, jugular vein and orbital artery, as in many
actinopterygians. In the Devonian dipnoans Griphognathus and Holodipterus there are
separate posterior openings (Miles 1977: figs 14, 44) for these three structures, as in primitive
actinopterygians, but in Chirodipterus there are two, one for the jugular canal and the other for
the hyomandibular trunk and the orbital artery, while in the Recent Neoceratodus there is a
single posterior opening for all three structures.

In some actinistians (Nesides, Latimerid) the opening of the facial canal lies in front of the
jugular canal, but in Rhabdoderma and the rhipidistians Ectosteorhachis, Eusthenopteron,
Glyptolepis (Jarvik 1972: tig. 21) and Youngolepis (Chang 1982: 51) the facial canal opens into
the mouth of the jugular canal much as in palaeoniscids.

In the Devonian dipnoan Griphognathus (Miles 1977: fig. 53) the root of the facial nerve
opens into the centre of the jugular canal, but the anterior part of the jugular canal is considered
a dipnoan specialization, formed by fusion of the ascending process of the palate to the pila
antotica. The facial and trigeminal ganglia were extracranial as in Neoceratodus.

In Latimeria (Millot & Anthony 1958) the gasserian and geniculate ganglia are separate and
intramural, the geniculate ganglion lies between the two moieties of the braincase, and there is
no separate lateralis canal. In fossil actinistians such as Diplocercides they may also be assumed
to have been intracranial.

In rhipidistians such as Ectosteorhachis (Romer 1937: fig. 2) and Eusthenopteron (Jarvik 1954:
fig. 1A; Bjerring 1971: fig. 9) the facial and trigeminal ganglia appear to have been intracranial.

The trigeminal, facial and lateralis canals arise in front of the recess for the utriculus in the
dipnoans Griphognathus and Chirodipterus (Miles 1977: figs 10, 17) and all the branches of the
trigeminal and facial nerves appear to originate well in front of the utricular recess in
Eusthenopteron (Jarvik 1975: fig. 14) and Ectosteorhachis (Romer 1937: figs 8, 13). The otic canal
originates above the jugular canal in the actinistians Latimeria, Nesides, Rhabdoderma and
Laugia and the rhipidistians Eusthenopteron, Ectosteorhachis, Rhizodopsis and Glyptolepis. The
otic nerve canal originates behind the spiracular groove in Eusthenopteron (Jarvik 1954: fig. 1,
spic) and opens into the anterior portion of the post-temporal fossa. In Youngolepis (Chang
1982: fig. 15A, r.ot.l) the otic nerve ran in a groove (anterodorsal to the lateral commissure)
which terminates below the temporal sensory canal. In Latimeria the otic nerve passes up behind
the spiracle.

A separate canal for the superficial ophthalmic nerves is found in Griphognathus (Miles 1977:
figs 10, 33, 55, VIIs?) and Chirodipterus (Miles 1977: figs 21, 35, 38, VI?). In Diplocercides
(Bjerring 1971; 1977: fig. 23A) a somewhat smaller foramen above the facial canal must have
also served for the lateralis branches of the facial nerve. The lateralis root in Latimeria exits
through the intracranial joint.

There is a separate profundus canal in actinistians and Devonian dipnoans. In actinistians
(Latimeria, Rhabdoderma, Macropoma, Nesides) it is within the 'basisphenoid', but in
Chirodipterus (Miles 1977: figs 17, 47) its origin is just in front of the trigeminal canal and above
the prootic bridge (as in palaeoniscids).

Thus from this brief summary I conclude that the osteichthyan morphotype must have been
similar to the actinopterygian one outlined above, apart from the path of the otic nerve canal.

(c) GNATHOSTOMES. In primitive sharks such as Hybodus (BMNH P. 50869, Maisey 1983)
and Xenacanthus (Schaeffer 1981: fig. 6) the lateral commissure is a massive structure of
calcified cartilage penetrated by a long jugular canal much as in Polyodon, but only the jugular
vein and hyomandibular trunk passed through it. In Xenacanthus, Cladodus and Squalus, the
hyomandibular trunk exits through a separate foramen. In many other selachians such as
Scymnodon and Oxynotus the hyomandibular trunk exits from the skull posterior to the jugular
canal and does not pass through it; this is considered specialized and related to the posterior
position of the hyomandibula. A lateral commissure is also clearly recognizable in
pristiophorids, some rhinobatoids (Compagno 1977: 310, 316) and Squatina (Holmgren 1941:
36). The orbital artery never passes through the jugular canal or the lateral commissure; instead

240 B. G. GARDINER

it often pierces the subotical shelf (cf. Squalus). The palatine nerve, on the other hand, may
pierce the basal part of the lateral commissure (Tamiobatis, Xenacanthus, Squalus).

The facial canal is variable in position in selachians. In Scymnorhinus the facial canal opens
into the middle of the jugular canal as in Polypterus (except that the ganglion is intramural), but
in Rhinobatus, where the lateral commissure is reduced, the facial canal opens in front of the
jugular canal and in Etmopterus and Squatina it opens into the ventromedial corner of the orbital
opening of the jugular canal, as in many palaeoniscids and parasemionotids.

In most selachians the lateralis branches of the facial nerve issue through the trigeminal
foramen, as in actinopterygians, and the trigeminal foramen often lies in the orbital opening of
the jugular canal (Oxynotus, Scymnodon, Xenacanthus, Tamiobatis, Cladodus, Hybodus). In
other selachians (Etmopterus, Squatina, Rhinobatus) the trigeminal foramen opens in front of
the jugular canal. The gasserian and geniculate ganglia are always intracranial. The otic nerve
canal passes from the orbit up through the postorbital process to open on the roof of the
neurocranium medial to the spiracle in Chlamydoselachus and Heterodontus , and there is a
separate foramen for the profundus.

A massive lateral commissure is also found in Devonian placoderms such as Brindabellaspis
and Buchanosteus (Young 1979; 1980: figs 8, 9) and the jugular canal is longer proportionally
than in primitive actinopterygians. The hyomandibular trunk passed out through the anterior
opening of the jugular canal in Brindabellaspis, but passed across the canal (at right angles to it)
in Buchanosteus. The orbital artery entered the jugular canal by a separate foramen in
Kujdanowiaspis (Young 1979: 329), but in Buchanosteus (Young 1979: fig. 9) it passed lateral to
the jugular canal and in Brindabellaspis (Young 1980: 41) medial to it.

The facial canal seems to be less variable in position in placoderms than selachians. In
Brindabellaspis and Wijdeaspis (Young 1978, 1980) it opens into the mouth of the jugular canal,
but in Buchanosteus (Young 1979) the facial canal opens into the anterior half of the jugular
canal. There is a separate profundus canal in Jagorina, Brindabellaspis and Macropetalichthys
and a separate canal for the superficial ophthalmic nerves in Buchanosteus (Young 1979: fig. 5) .

InAcanthodes the lateral commissure is not really discernible because of lack of ossification in
that region, but a short perichondral commissure is present as Miles (1973«: fig. 4) presumed
(BMNH specimens).

From this brief survey I conclude that the primitive gnathostome possessed a long lateral
commissure with separate openings for the jugular vein and hyomandibular trunk; the facial
canal emerged in the orbital opening of the jugular canal and the trigeminal canal dorsomedial
to it; the otic nerve canal originated in the posterior wall of the orbit and passed up to the
neurocranial roof medial to the spiracle.

5. Hyomandibular facet. In the Gogo palaeoniscids this lies obliquely between the sphenotic and
opisthotic with the prootic forming a small portion of the ventral margin. A similar oblique facet
above the jugular canal is characteristic of most palaeoniscids, Polyodon, Acipenser,
Polypterus, etc.

In later actinopterygians such as amioids, semionotids, pachycormids, pholidophorids and

Recent teleosts the facet has become more or less horizontal. In amioids, pholidophorids and

pachycormids it lies between the sphenotic and prootic anteriorly and the pterotic and

opisthotic posteriorly (except in Amia and semionotids), while in leptolepids and many other

teleosts the posterior end lies entirely within the pterotic. There is a single hyomandibular

articulation in actinopterygians and the oblique facet in palaeoniscids is considered primitive.

In actinistians the hyomandibular articulation is a very large, bilobed, cartilage-capped area

(Nesides, Bjerring 1977: fig. 23; Rhabdoderma, Forey 1981: fig. \\Latimeria, Millot & Anthony

1958) which straddles the jugular canal obliquely and lies between the prootic and opisthotic. In

rhipidistians (Eusthenopteron, Jarvik 1954: fig. 1; Ectosteorhachis , Romer 1937: fig. 2; Porolepis,

Bjerring 1967: 224; Youngolepis, Chang 1982: fig. 15A) the articulation is double, oblique and

crosses the jugular canal. I assume it lies between the prootic and opisthotic in Eusthenopteron.

The hyomandibular facet in Devonian dipnoans straddles the jugular canal obliquely, as in

actinistians and rhipidistians (Miles 1977: 90), and the articulation is divided into dorsal and

RELATIONSHIPS OF PALAEONISCIDS 241

ventral areas by a thin tract of perichondral bone. Despite the arguments of Miles (1977: 91) I
see no reason to change my previous view (Gardiner 1973: 109, 122) that the hyomandibula has a
double articulation with the neurocranium in Devonian dipnoans, much as in rhipidistians. In
Recent dipnoans the facet is lost.

In Acanthodes (Miles 1964: fig. IB) the head of the hyomandibula is presumed to have lain
obliquely between the perichondral lateral commissure and otic ossification and dorsal to the
jugular canal, as in actinopterygians.

In chondrichthyans the condition is variable. In Heptanchus, Squatina (Holmgren 1941: 30)
and Cobelodus (Zangerl & Case 1976) the hyomandibular facet is set obliquely across the
posterior part of the auditory region and in Mustelus, Carcharinus and Heterodontus (Holmgren
1941: 45) the articulation is on the anterior half of the auditory region. In Chlamydoselachus
and Orectolobus the hyomandibular facet is a long, broad groove dorsal to the auditory bulla,
whereas in Squalus it is below and behind the bulla. In Rhinobatus, Raja, Discobatus, Pristis,
Pristiophoms, Dasybatus (Holmgren 1941) and all other rays the articulation is divided into two
facets which are situated in the posterior part of the auditory region. In all these Recent sharks
and rays without exception the hyomandibula articulates with the endocranium ventral to the
jugular vein, and usually below the ridge for the horizontal semicircular canal, and is
suspensory. This ventral position of the hyomandibula with respect to the jugular vein in
selachians has generated much argument as to whether the condition is primitive (Holmgren
1943: 102, 104) or advanced (Gardiner 1973: 122), while its posterior position has been
responsible for the development of elaborate theories of the incorporation of hyoid arch
material (pharyngohyal) into the auditory capsule (Holmgren 1940; 1943: 43, 68), to form the
otical shelf and hyomandibular facet in rays. Jarvik (1954: 75) has used this supposed evidence of
incorporation of the hyoid arch material into the auditory capsule of selachians to support a
similar presumed incorporation of hyoid arch material into the lateral commissure and 'otical
shelf region of Eusthenopteron. However, the supposed incorporation of hyoid material in
sharks and rays described by Holmgren (1940, 1943) always occurs posterior to the lateral
commissure (Holmgren 1940: fig. 130), whereas the 'otical shelf of Eusthenopteron (Jarvik
1954: fig. 1A) lies anterior to the lateral commissure. But the discovery in Tristychius (Dick
1978) and Hybodus (Maisey 1982, 1983) of a hyomandibula which articulated with the braincase
above the jugular canal as in actinopterygians, actinistians and rhipidistians suggests that it is no
longer worth while to invoke theories of hyoid arch incorporation to explain the 'otical shelf in
selachians.

In those placoderms in which the hyomandibular articulation has been described in detail
(Young 1980, Stensio 19630, Goujet 1975) it is at the base of the lateral commissure
(Brindabellaspis Young 1980) and in front of the foramen for the hyomandibular trunk.

I conclude that primitively in gnathostomes the hyomandibula must have articulated with the
braincase in the region of the lateral commissure and was suspensory. Whether the articulatory
facet is above or below the jugular canal seems of little significance and there is no reason to
suppose that the hyomandibular-neurocranial attachment developed more than once.

6. Otico-sphenoid fissure. This has so far only been reported in the Gogo palaeoniscids. It was
cartilage-filled and separates the anteroventral corner of the prootic from the basisphenoid. It
appears to correspond with the gap between the lateral commissure and the polar cartilage -
trabecular bar of the embryo actinopterygian (Gardiner & Bartram 1977: 241). Together with
the ventral otic fissure it is homologous with the ventral part of the intracranial joint of
actinistians and rhipidistians (Gardiner & Bartram 1977: 242). However, this homology has
been denied by Bjerring (1978).

7. Fossa bridgei and lateral cranial canal. The Gogo palaeoniscids resemble Boreosomus
(Nielsen 1942) in the considerable variation shown in the degree of ossification of the dorsal otic
region. In some specimens ofMimia (Fig. 12) there is large lateral cranial canal in the roof of the
otic region occupying the whole of the area between the posterior and anterior semicircular

242 B. G. GARDINER

canals. This chamber is roofed by a sutureless skin of perichondral bone and connects with the
cranial cavity posteriorly by a large opening through the loop of the posterior semicircular canal
(Fig. 26). Anteriorly it may occasionally connect with the cranial cavity in front of the sinus
superior by various small foramina. The supratemporal branch of the glossopharyngeal nerve
enters the canal laterally. The spiracular groove lies anterolateral to the canal and is not included
within it. In other specimens of Mimia, the lateral cranial canal is little more than a pocket in
front of the posterior semicircular canal, but still maintains its posterior connection with the
cranial cavity by a large foramen through the loop of the semicircular canal much as in
Kansasiella (Poplin 1974: fig. 20, e.l.m, y) and Kentuckia (Rayner 1951: fig. 9, x).

The presence of a lateral cranial canal is considered to be a specialization of actinopterygians
(Gardiner 1973: 113). Jarvik (1980) has shown that in Recent chondrosteans and holosteans
there is a hemopoietic organ dorsal to the medulla which in Lepisosteus has earlike lobes passing
forwards and upwards into the lateral cranial canal.

The Gogo palaeoniscids have no fossa bridgei, which if present would have overlain (in part)
the lateral cranial canal. A lateral cranial canal is well developed in Kentuckia but the fossa
bridgei is poorly defined and consists of several irregular cavities (Rayner 1951: figs 6, 9). Most
palaeoniscids possess both a lateral cranial canal and a fossa bridgei, but the absence of the latter
in the Gogo palaeoniscids is considered primitive.

In Polypterus there is neither a fossa bridgei nor a lateral cranial canal and this is considered
primitive for osteichthyans. Actinopterygians in which the fossa bridgei and lateral cranial canal
are separate include Kansasiella, Boreosomus, Pholidophoms (Patterson 1975: 336), Leptolepis
(Patterson 1975: fig. 73), Caturus (Rayner 1948: fig. 9), Heterolepidotus, Dapedium, Lepidotes,
parasemionotids and Lepisosteus (Gardiner 1973: 13). In many other actinopterygians there is a
connection between the deeper, posterior part of the fossa bridgei and the cranial cavity
(Nielsen 1942: 294; Patterson 1975: 392, 414). These connections are topographic homologues
of parts of the lateral cranial canal. Forms in which the fossa bridgei communicates posteriorly
with the cranial cavity by way of the lateral cranial canal include Pteronisculus , some specimens
of Boreosomus, and Polyodon (cf. Boreosomus Nielsen 1942: fig. 59, where in one individual
the fossa bridgei is separated by bone from the lateral cranial canal on one side, but on the other
the two cavities are connected).

In several other actinopterygians (Ospia, Acipenser} where the lateral cranial canal and fossa
bridgei have become confluent the lateral cranial canal portion has lost its connection with the
cranial cavity; this is also considered to be a specialization. The lateral cranial canal
communicates with the cranial cavity posteriorly through the loop of the posterior semicircular
canal in the Gogo palaeoniscids, Kansasiella (Poplin 1974: fig. 22), Kentuckia, Boreosomus and
Polyodon, but in caturids (Caturus, Heterolepidotus} and pholidophorids the canal has both
anterior and posterior openings into the cranial cavity and in Perleidus there is only an anterior
opening (Patterson 1975: 414). In leptolepids, Lepisosteus and Lepidotes the lateral cranial
canal has lost the inner wall and thus has the form of a posterodorsal diverticulum of the cranial
cavity (Patterson 1975: 413). In Polyodon the connection between the fossa bridgei and the
cranial cavity, by way of the lateral cranial canal, is fat-filled.

The spiracular groove in the Gogo palaeoniscids and the spiracular canal in Kentuckia lie
lateral to the area occupied by the fossa bridgei in other forms, and in Polypterus the spiracle lies
alongside the braincase in the adult. The spiracle also lies freely alongside the intracranial joint
in Latimeria. In Eusthenopteron Jarvik (1954: fig. 1A, spic) has claimed that a spiracular canal
enters the fossa bridgei. However, the 'fossa bridgei' in Eusthenopteron is more reasonably
interpreted as a post-temporal fossa and the 'spiracular canal' as the otic nerve canal.
Subsequent evolution within actinopterygians brought about the formation of the spiracular
recess (often called 'fossa bridgei anterior', Nielsen 1942: fig. 11) containing a blind-ending
spiracular diverticulum in Acipenser, Polyodon and Amia. The recess is confluent with the fossa
bridgei (Gardiner 1973: 114) in these fishes. In the Gogo palaeoniscids the spiracular groove is
entirely outside the sphenotic, but in halecomorphs and pholidophorids the spiracular canal
penetrates the postorbital process at the junction between the sphenotic and prootic (Patterson
1975: 399). The condition in which the fossa bridgei includes the spiracular recess but is still

RELATIONSHIPS OF PALAEONISCIDS 243

roofed by dermal bone is met with in Polyodon, Boreosomus, Perleidus, Acipenser, Ospia,
parasemionotids, caturids and pholidophorids.

In Amia and Lepisosteus the fossa bridgei is represented by an elongate pocket or groove,
containing no musculature, which extends anteriorly to the opening of the spiracular canal. An
area medial to the anterior semicircular canal in Pholidophorus , referred to as the antero-medial
portion of the fossa bridgei by Patterson (1975), has no homologue in the Gogo palaeoniscids and is
presumably a specialization of later actinopterygians.

In parasemionotids and primitive pholidophorids, behind the fossa bridgei and separated
from it by a wall of bone, there is a small post-temporal fossa which probably contained trunk
musculature (Patterson 1975: 392). In Upper Jurassic pholidophorids the post-temporal fossa
and fossa bridgei have become confluent, by the breakdown of the intervening wall, allowing the
axial muscles to extend into the fossa bridgei, as in Recent teleosts (Patterson 1975: 392). There
is no homologue of the post-temporal fossa in palaeoniscids, but a rudimentary post-temporal
fossa appears to be present in Birgeria, Perleidus, Saurichthys, and sturgeons (Griffith & Patterson
1963: 35; Patterson 1975: 392). The post-temporal fossa is presumed to be a specialization of later
actinopterygians, and Patterson (1975: 393) has shown how the pre-epiotic pocket of primitive
teleosts is the topographic homologue of the posteromedial portion of the fossa bridgei of
pholidophorids and primitive leptolepids. A further small blind-ending fossa (wrongly referred to as
a pre-epiotic fossa by Gardiner, 1973: 113) which runs down under the posterior semicircular canals
in the clupeid Alosa (Todd 1973) is possibly the homologue of the lateral cranial canal of
Pholidophorus.

There is no evidence of a fossa bridgei in Latimeria or any fossil actimstian (including
Nesides). A small post-temporal fossa is present in some later actinistians and receives trunk
muscles in Latimeria; this must have developed in parallel with that in later actinopterygians and
rhipidistians (see below).

A well-developed post-temporal fossa is also present in many rhipidistians, and in
Eusthenopteron (Jarvik 1954; 1975: figs 8,9,11, 13; fossa bridgei) it is roofed by dermal bone. It
lies between the posterior and anterior semicircular canals and is limited laterally by the parotic crista
which encloses the external semicircular canal. It is in a similar position to the posterior part of the
fossa bridgei and the lateral cranial canal in palaeoniscids. Presumably it was invaded by axial
muscles much as in later actinopterygians, actinistians and tetrapods. A similar post-temporal fossa is
found in Ectosteorhachis, Rhizodopsis, Osteolepis and Glyptolepis.

In Devonian dipnoans, according to Miles (1977: 74), there is a fossa comparable to that in
rhipidistians, the masseter fossa. It is limited laterally by the parotic crista (which encloses the
external semicircular canal), roofed by dermal bone, and extends medial to the posterior
vertical semicircular canal; the fossa is open posteriorly. However, since the fossa in Devonian
dipnoans is presumed to have housed only the adductor mandibulae (cf . Neoceratodus) it may
not be homologous with the rhipidistian post-temporal fossa as Miles assumed. Moreover,
although a post-temporal fossa is present in many amphibians (loxommatids) it is missing in
primitive temnospondyls, anthracosaurs and other primitive amniotes.

The fossa bridgei is recognizable only in later actinopterygians and its absence is considered a
primitive osteichthyan feature. There is no indication of a fossa bridgei or a post-temporal fossa
in acanthodians, placoderms or chondrichthyans.

8. Spiracle and spiracular canal. In the Gogo palaeoniscids the spiracular groove is long and the
spiracle was open. However, in some specimens of Moythomasia the groove is embraced
dorsally by a thin ring of bone (Fig. 83) similar to that described in Eusthenopteron (Jarvik 1954:
fig. IB, b.al). The spiracle opened dorsally through a slit in the dermal roof. In Polypterus the
anterior edge of the spiracular pouch lies in the fossa spiracularis (Allis 1922: 193) and the
spiracle opens through a slit between the 'parietal' and two spiracular ossicles.

In Acipenser, Polyodon and Lepisosteus the spiracular pouch divides into two tubes, one of
which opens on the skull roof (spiracle) and the other forms the blind-ending spiracular canal. In
Amia only the blind-ending spiracular canal remains. A homologous spiracular canal penetrates
the postorbital process of most palaeoniscids and other primitive actinopterygians. The

244 B. G. GARDINER

spiracular canal is still relatively short in Pteronisculus, Kansasiella and Kentuckia and is
presumed to be contained within the sphenotic. The canal is proportionally longer in
Boreosomus, Birgeria, Perleidus, Australosomus, Acipenserand Polyodon, but in these last two
living chondrosteans it is housed in cartilage.

A spiracular canal (or dorsal diverticulum) is found in many selachians where it is contained
within the ventrolateral wall of the auditory capsule, below the external semicircular canal.
Another more ventromedial diverticulum houses a spiracular sense organ. This dorsal or
auditory diverticulum of sharks is directed towards the ventral wall of the auditory 'capsule'
whereas that of Adpenser, Lepisosteus and Amia passes up along side the trigeminofacialis
chamber before piercing the postorbital process. From this evidence and from the fact that in
Latimeria the spiracle was also unenclosed and without a diverticulum, the condition in the
Gogo palaeoniscids, Polypterus and Latimeria is considered primitive for osteichthyans.
Subsequent evolution within the actinopterygians led towards elongation of the enclosed
portion of the spiracular canal at the expense of the spiracular groove (Patterson 1975: 399) and
the spiracular canal came to open within the fossa bridgei (Gardiner 1973: 113). A long
spiracular canal is found in parasemionotids, caturids (Caturus, Rayner 1948: fig. 5;
Heterolepidotus , Patterson 1975: fig. 102; 'Aspidorphynchus' ', Patterson 1975: fig. 99), amiids
(Sinamia, Stensio 1935; Enneles, Santos 1960), semionotids (Dapedium, Patterson 1975: fig.
112; Lepidotes, Patterson 1975: fig. 108), pachycormids (Pachycormus, Patterson 1975: fig.
106) , pholidophorids and Lower Jurassic leptolepids, but it is obliterated in later leptolepids and
living teleosts (Gardiner 1973: 111; Patterson 1975: 398).

Living chondrosteans and holosteans have a neuromast organ lodged near the top of the
spiracular canal, supplied by the otic branch of the facial nerve. A similar neuromast sense organ
occurs in selachians (Wright 1885), but here it is lodged in a closed vesicle which has become
pinched off from the base of the spiracular cleft.

In Latimeria the spiracle lies free alongside the intracranial joint. It is closed dorsally and
terminates just beneath the dermal roof. As in Polypterus (and holocephalans) there is no
spiracular sense organ, but the overlying neuromast of the temporal canal is greatly enlarged
(Forey, personal communication).

Eusthenopteron is the only rhipidistian in which there is clear evidence of a spiracular groove
(Jarvik 1954: fig. 22, gr. psp.). The spiracle in lungfishes is rudimentary and no trace of it has
been described in fossil dipnoans (Gardiner 1973: 113; Miles 1977). A spiracular sense organ
(Pinkus' organ), however, is found in Recent dipnoans where it is lodged in the otic process of
the palatoquadrate , and a deep fissure in front of the hyomandibular facet in Devonian dipnoans
(Miles 1977: 79) may have housed a similar sense organ.

It is thus apparent that a bone-enclosed spiracular canal and sense organ is a synapomorphy of
advanced actinopterygians, whereas the presence of the sense organ may be more general
(selachians, later actinopterygians and dipnoans).

9. Origin of dorsal hyoid constrictor muscle. In living chondrosteans there is an undifferentiated
dorsal hyoid constrictor which forms a continuous sheet of muscle and originates along the
dorsolateral margin of the braincase from hyomandibular facet to auditory capsule. It inserts
ventrally on the hyomandibula and opercular.

A similar constrictor hyoideus dorsalis is found in selachians and its dorsal portion inserts on
the hyomandibula (Edgeworth 1935: 98). From this evidence we can assume that primitively in
gnathostomes the hyomandibula was levated from above by the anterodorsal portion of the
hyoid constrictor.

In the development of Polypterus, Amia and Lepisosteus the anterior edge of the hyoid
constrictor spreads forwards internal to the hyomandibula to form the adductor hyomandib-
ulae, and the posterior portion of the constrictor differentiates into the adductor operculi. In
Amia and teleosts the constrictor hyoideus is still more elaborately differentiated. If Polypterus
is the sister-group of the chondrosteans plus neopterygians then the adductor hyomandibulae
must have evolved separately in at least two lines of Recent fishes. Latimeria (Millot & Anthony

RELATIONSHIPS OF PALAEONISCIDS 245

1965), in which the adductor operculi lies dorsal to the adductor hyomandibulae, suggests yet
another separate origin.

In the Gogo palaeoniscids there is a continuous dorsal muscle scar stretching from just behind
the point of articulation of the hyomandibula to the occipital fissure, only interrupted by a
groove for branches of the glossopharyngeal and vagus nerves. It is presumed that this elongate
area gave rise to an undifferentiated dorsal hyoid constrictor muscle as in living chondrosteans.
It is unlikely that any part of the dorsal hyoid constrictor had its origin in the conspicuous
'subtemporal fossa' (Gardiner 1973: fig. 3) as suggested by Patterson (1975: 395), since the
whole of this fossa consists of fenestrated bone, totally unsuitable for muscle origin. This
so-called 'subtemporal fossa' in the Gogo palaeoniscid results from the lack of ossification of the
side wall of the braincase in the neighbourhood of the jugular vein and is not such a distinct
feature as in pholidophorids and other teleosts. There is no obvious subtemporal fossa in any
other palaeoniscid, living chondrostean or Polypterus, and in the latter fish the adductores
hyomandibulae and operculi form a continuous muscle (Lauder 1980) which originates on the
dorsolateral surface of the opisthotic.

In pholidophorids, leptolepids and other primitive teleosts the subtemporal fossa presumably
served principally for the origin of the adductor operculi and there is usually an anterior, less
well-marked depression for the origin of the adductor hyomandibulae. The first recognizable
depression which may be considered to be the homologue of the subtemporal fossa of
pholidophorids and teleosts is found in fossil neopterygians (parasemionotids, Heterolepidotes,
'Aspidorhynchus', Dapedium, Lepidotes and Pachycormus; Patterson 1975: figs 97-9, 102,
106, 108, 112) although in Amia and Lepisosteus there is no obvious subtemporal fossa.

In Perleidus (Patterson 1975: fig. 115) and Australosomus (Nielsen 1949: figs 2, 29) there is a
large subtemporal fossa, but this is presumed to be the homologue of both the anterior
depression and subtemporal fossa of fossil halecomorphs, pholidophorids and teleosts.

In summary, the subtemporal fossa appears to be a specialization of later actinopterygians
which in halecomorphs and teleosts serves for the origin of the adductor operculi.

In Latimeria (Millot & Anthony 1965: fig. 31) the dorsal hyoid constrictor has differentiated
into an adductor hyomandibulae and an adductor operculi. The origin of the adductor
hyomandibulae is remote from the adductor operculi and below the jugular canal; by
comparison with selachians and chondrosteans this condition is presumed to be derived, as is the
double articulation of the head of the hyomandibula. In the Devonian Nesides a distinct fossa
(Bjerring 1977: fig. 22, ar.m.ad.hy) below the jugular canal must have served for the origin of
the adductor hyomandibulae and a more dorsal fossa behind the dorsal hyomandibular
articulation contained the adductor operculi. Identical fossae are recognizable in the
rhipidistians Ectosteorhachis (Romer 1937: fig. 2) and Rhizodopsis. In Eusthenopteron (Jarvik
1954: fig. 21A, B; 1975: fig. 13A, B) the so-called 'process for the attachment of adductor
muscles of hyomandibula' (Bjerring 1971: fig. 8) forms the posterior boundary of a depression
which, by analogy with Nesides and Latimeria, also served for the origin of the adductor
hyomandibulae.

There is no adductor hyomandibulae in Recent dipnoans and it is likely that the loss of this
muscle is correlated with the reduction of the functional importance of the hyomandibula
following fusion of the palate with the braincase.

10. Origin of dorsal mandibular constrictor muscle. The constrictor dorsalis is undifferentiated
in living chondrosteans (Edgeworth 1935: 48), where it forms a large and powerful protractor
hyomandibularis which originates dorsally on the postorbital process. A similar constrictor
dorsalis is also characteristic of selachians where it has its origin on the postorbital process and
extends downwards to insert on the inner (medial) surface of the palatoquadrate. In selachians it
is called the levator maxillae.

In Polypterus the constrictor dorsalis has differentiated into four distinct muscles, the levator
arcus palatini, protractor hyomandibularis, dilatator operculi and musculus spiracularis (Allis
1922: 254), but in all living neopterygians it differentiates into only two muscles (Lauder 1980),

246 B. G. GARDINER

the levator arcus palatini and the dilatator operculi, with the posterior portion of the levator
arcus palatini inserting on the hyomandibula (= protractor hyomandibularis).

From this evidence it is clear that primitively in actinopterygians the posterior portion of the
constrictor dorsalis (levator arcus palatini) was attached to the hyomandibula. The condition in
living chondrosteans, where the anterior portion of the levator arcus palatini is missing, must be
considered derived and related to the reduction of the palatoquadrate and its loss of contact with
the braincase. In Polypterus the levator arcus palatini and protractor hyomandibularis originate
together, the levator arcus palatini inserting on both the dorsal edge of the entopterygoid and
the hyomandibula while the protractor hyomandibularis also inserts (via a membrane) on the
dorsal edge of the palatoquadrate and the anterior edge of the hyomandibula. Both of these
muscles, in being inserted to the suspensorium, may be equated with the single levator arcus
palatini of neopterygians. The presence of separate levator arcus palatini and protractor
hyomandibularis muscles in Polypterus is considered derived and related to the reduction of the
posterodorsal corner of the palatoquadrate (otic process), which has allowed part of the levator
arcus palatini to insert on the hyomandibula and then eventually to separate off as a distinct
muscle. A separate protractor hyomandibularis is not seen in any other living fish (other than
chondrosteans: see above) and it is doubtful that one existed in rhipidistians as postulated by
Jarvik (1954: fig. 26).

There is no evidence of separate areas for the origins of dilatator operculi and levator arcus
palatini muscles in any palaeoniscid and it is unlikely that these muscles differentiated. The
constrictor dorsalis is postulated to have originated on the postorbital process, much as in
Polypterus and selachians. With the elongation of the spiracular canal in more advanced
actinopterygians a new area was made available for the hyoid constrictor, anterior to and above
the hyomandibular facet. This area, first recognizable in Perleidus and parasemionotids, is
termed the dilatator fossa. In lepisosteids, halecomorphs and pholidophorids the anterior area
of the dilatator fossa housed the levator arcus palatini and the posterior area the dilatator
operculi. In leptolepids (Patterson 1975: 387) the dilatator fossa only housed the dilatator
operculi as in Recent teleosts, and is separated by a crest of bone from the area of origin of the
levator arcus palatini.

In Latimeria (Millot & Anthony 1958) the dorsal constrictor has apparently differentiated
into two muscles much as in neopterygians. The so-called 'levator arcus palatini' runs from in
front of the dorsal hyomandibular facet to insert on the hind edge and inner (medial) surface of
the palatoquadrate. The partial insertion of this muscle on the inner palatoquadrate surface is
considered primitive for osteichthyans and gnathostomes, while the condition in actinoptery-
gians, where the levator arcus palatini inserts on the outer surface of the palate, is derived. The
effect this different origin has on elevation of the palate has been commented on elsewhere
(Gardiner 1973: 121). It suffices to add that the action of the 'levator arcus palatini' in Latimeria
merely pulls the palatoquadrate inwards and backwards. Anteriorly the 'levator arcus palatini'
originates on the prootic behind the spiracle, not on the postorbital process as in selachians and
actinopterygians. The dilatator (elevator) operculi in Latimeria is a small muscle lying just
behind the 'levator arcus palatini' and which has presumably evolved in parallel with that in
actinopterygians. It is difficult to see where the dorsal constrictor would have inserted in
rhipidistians, but by comprison with actinistians, it must have originated behind the spiracular
groove on the anterodorsal face of the lateral commissure. In Eusthenopteron (Jarvik 1954: fig.
23) the constrictor dorsalis could only have originated on the lateral commissure just anterior to
the post-temporal fossa; in Ectosteorhachis (Romer 1937: figs 2, 6) and Rhizodopsis there is a
well-marked depression in this area.

In dipnoans the constrictor dorsalis has not separated from the adductor mandibulae (cf.
holocephalans) and this is considered a specialization related to autostyly (Miles 1977: 121).

11. Endolymphatic duct. In Acipenser the endolymphatic organ is club-shaped and extends into
the cranial cavity, but in Polypterus, Lepisosteus, Amia and most teleosts it is shorter and
confined to the otic capsule. In Mimia there is a gutter running from near the junction of the
posterior semicircular canal with the sinus superior to the dorsal surface of the cranial cavity in

RELATIONSHIPS OF PALAEONISCIDS 247

front of the dorsal fontanelle. This must have housed the blind-ending endolymphatic organ. In
Moythomasia, on the other hand, the gutter for the endolymphatic duct opens into the posterior
dorsal fontanelle and the duct was probably still open, as in chondrichthyans, placoderms and
Devonian dipnoans. Unfortunately in one specimen ofMimia figured by Gardiner (1973: fig. 4,
mcc) the area of the braincase around the end of the endolymphatic ducts had not completely
ossified and this, together with the report of a median intramural chamber in Dapedium
(Patterson 1975: 413), led Miles (1977: 102) to postulate the presence of a median supraotic
cavity in actinopterygians. However, there is no such cavity in any living chondrichthyan,
actinopterygian or Latimeria and it is better regarded as a specialization confined to some
dipnoans, rhipidistians and tetrapods.

The endolymphatic duct is primitively open in gnathostomes and its closure in
actinopterygians is considered a specialization. There is no evidence of a hypertrophied
endolymphatic sac in actinopterygians.

Miles (1977: 103) has argued that the anterior division of the so-called supraortic cavity of
primitive dipnoans is a synapomorphy shared with choanates. However, Young (1980: fig. 10)
has described what he considers to be a posterior 'endolymphatic diverticulum' in the placoderm
Brindabellaspis and states that this shows some resemblance to the supraortic cavity of
Devonian dipnoans. But the supraortic cavity of rhipidistians and choanates is related to the
endolymphatic duct which passes through it onto the roof of the neurocranium, and is
primitively divided into a posterior median division and an anterior paired division. The
'endolymphatic diverticulum' of Brindabellaspis has none of these features and can in no way be
homologized with the supraortic cavity.

12. Pterosphenoid pedicel. Patterson (1975: 409) proposed that this was a primitive
actinopterygian feature, mainly because he recognized the relationship of this structure to the
intramural chamber for the superficial ophthalmic nerves in Kentuckia, Kansasiella and
caturids. From the condition in Moythomasia (see p. 230) it is now more certain that he was
correct in homologizing the upper part of the pterosphenoid pedicel with that intramural
chamber.

In Moythomasia (Fig. 29) the pedicel is often only partially developed, much as in
Pteronisculus magnus (Nielsen 1942: 90) and Kansas palaeoniscid B (Watson 1925: 843), but in
some specimens of Moythomasia (Fig. 30) it is complete and the lower portion is believed to
have been formed by the prootic, as in Macrepistius (Schaeffer 1971: fig. 4). A similar pedicel is
seen in Amia and its fossil relatives but here it is formed entirely by the pterosphenoid and meets
the ascending process of the parasphenoid ventrally . In other caturids only the dorsal portion of
the pedicel is present ('Aspidorhynchus', Patterson 1975: fig. 101) or sometimes only that part
which envelops the superficial ophthalmic nerves (Caturus, Heterolepidotus, Osteorachis}. In
the larger pholidophorids (Patterson 1975: 409) the upper pterosphenoid portion is often
present while in the Sinemurian Leptolepis only the lower prootic portion is developed. As in
Recent teleosts, however, where the pedicel is frequently well developed, it consists mainly if
not entirely of membrane bone.

From this we may conclude that the presence of a pterosphenoid pedicel is primitive, at least
for actinopterygians, whereas enclosure of the superficial ophthalmic nerves in a separate bony
canal is possibly a primitive osteichthyan feature (cf. Youngolepis) .

13. Prootic bridge and posterior myodome. In vertebrate embryos the relationships of the
orbital cartilage are surprisingly constant. Posteriorly it is invariably attached to the polar
cartilage (or trabecula) by a pila antotica. The bases of the pilae antoticae are joined by a
transverse connective cartilage, the acrochordal, which eventually forms the crista sellaris,
dorsum sellae or prootic bridge.

In selachians (Pristiurus, Matveiev 1925: fig. 2; Scyllium, de Beer 1937: 69; Squalus,
Holmgren 1940: 94; Torpedo, de Beer 1937: 69), dipnoans (Neoceratodus, Sewertzoff 1902: fig.
1) and amphibians (Ambystoma, Stohr 1879) the orbital cartilage and pila antotica develop
early, whereas in actinopterygians (Polyptems, Moy-Thomas 1934; Acipenser, Sewertzoff 1928;

248 B. G. GARDINER

Lepisosteus, Hammarberg 1937; Amia, Pehrson 1922; Salmo, de Beer 1927) these structures
develop later and lag behind the auditory capsule, polar cartilage and trabecula. In selachians
the dorsum sellae appears early, but in actinopterygians its development is considerably
delayed, often until after the lateral walls of the braincase are completed. This delay in
actinopterygians is often attributed to the development of the myodome, but the dorsum sellae
is equally delayed in Polyptems and Acipenser, fishes without a myodome.

We now know that there are palaeoniscids without a posterior myodome (Mimia,
Moythomasia) and that this was probably the primitive condition for actinopterygians (Gardiner
1973; Gardiner & Bartram 1977). We also know that the relationships of the profundus are quite
varied (see above), but that bone in the position of the pila antotica always separates the
trochlear, optic and oculomotor nerves from the trigeminal foramen. It is difficult then to
concede that, because of the varying relationships of the profundus, the pila antotica is a
secondary structure in Polypterus, Acipenser, and Salmo as de Beer (1937) and Schaeffer (1971:
6) assumed. It is even more difficult to accept its absence in Lepisosteus (de Beer 1937), where
the difference is that the pila antotica lies somewhat more anteriorly than in other forms.

The dorsum sellae is assumed to have been ossified by the basisphenoid in Mimia (Gardiner &
Bartram 1977: 230) and Moythomasia, as in Polypterus, but in all other actinopterygians it is
formed by the prootics. The abducens nerve passes out below the dorsum sellae in Mimia,
Moythomasia and Polypterus, behind it in Acipenser, in front and above in Lepisosteus (where
the myodome is assumed to have been secondarily lost - see below), whereas in all other
actinopterygians where a myodome is present the abducens pierces the prootic bridge.

An ossified prootic bridge (dorsum sellae) is also present in the anterior moiety of the
braincase of actinistians and rhipidistians (Romer 1937, Thomson 1967, Schaeffer 1968) and is
not missing as erroneously suggested by Gardiner (1973: 108). The lateral wall of the
neurocranium of Latimeria is presumed to exhibit the primitive gnathostome condition, with the
profundus, trochlear, optic and oculomotor nerves all passing through separate canals. The
dorsum sellae is ossified by the basisphenoid and this bone also contains the foramina of the
oculomotor and profundus nerves, as in Polypterus, whereas the abducens nerve leaves the
braincase behind the dorsum sellae as in Acipenser.

In the rhipidistians Ectosteorhachis (Romer 1937: fig. 2) and Eusthemopteron (Jarvik 1954: fig.
Ic) it may be inferred that the dorsum sellae, as in Latimeria and Nesides, is ossified by the
basisphenoid and that the abducens nerve did not pass through it.

The dorsum sellae is cartilaginous in Recent dipnoans but ossified in the Devonian forms. In
Griphognathus the profundus is supposed to pass out below and behind the dorsum sellae (Miles
1977: fig. 10), but this canal (Vt) could be more convincingly interpreted as transmitting the
abducens nerve. In Chirodipterus the dorsum sellae lies further posteriorly than in
Griphognathus, the profundus passed out above it, and the canal for the abducens (Miles 1977:
figs 17, 35) is said possibly to originate between the facial and trigeminal canals, above the
dorsum sellae.

In chondrichthyans such as Heterodontus, Torpedo and Hydrolagus the dorsum sellae is a
stout bar of cartilage and the abducens nerve exits with the trigeminal above and slightly behind
it. In Squalus, however, the abducens leaves through a separate canal below and posterior to the
dorsum sellae (Jollie 1962: fig. 5.10).

In the placoderms Buchanosteus (Young 1979: figs 5, 6) and Kujdanowiaspis (Stensio 19636:
fig. 26) the abducens has a separate canal which leaves the brain behind the root of the
trigeminal, much as in Squalus.

I conclude that primitively in osteichthyans the dorsum sellae was ossified by the basisphenoid
bone and in the primitive gnathostome the abducens nerve passed out below and behind it.

Posterior myodomes are not present in the primitive actinopterygian braincase (Polypterus,
living chondrosteans) and are also absent in the Gogo palaeoniscids. There is little doubt that at
least three of the rectus muscles originated on the basisphenoid pillar in Mimia and
Moythomasia (Gardiner & Bartram 1977: 237), but whether or not the external rectus muscle
originated here or in the back of the orbit ventrolateral to the abducens canal could not be
determined, though it seems unlikely that it originated posteriorly. In Polypterus three of the

RELATIONSHIPS OF PALAEONISCIDS 249

muscles (superior, inferior and external recti) arise from a short tendinous stalk with its origin on
the basisphenoid immediately posterior to the optic foramen in a homologous position to those
in selachians, while the fourth (internal rectus) takes origin anterior to the optic foramen.
Similarly in Acipenser the four recti muscles attach to the cartilaginous interorbital region. The
rectus muscles also originate on the basisphenoid in Latimeria, and a distinct pit in front of the
basipterygoid process in Holoptychius (Jarvik 1972: fig. 20B) and Youngolepis (Chang 1982: fig.
15B) must also have served for the origin of all four muscles. In Acanthodes (Miles 19730: fig. 9;
Jarvik 1977: fig. 3A) the perichondrally ossified basisphenoid pillar may have been related to
these muscles. Primitively in gnathostomes the rectus muscles originated on the interorbital
septum posterior to the optic nerve. In Lepisosteus the origins of all four rectus muscles have
moved onto the basipterygoid process (which is made up of the prootic; the basisphenoid is
absent) and this is considered to be a specialization, as are the path of the abducens nerve and the
loss of the myodome. In Amia, on the other hand, the superior, internal and inferior recti
muscles originate on the transverse bolster of the basisphenoid and the external rectus enters the
myodome. Myodomes only occur in more advanced actinopterygians (Amia and teleosts among
living fishes) , presumably as a result of the backward growth of the external rectus muscles onto
that area of the basisphenoid immediately below and lateral to the pituitary.

In Kentuckia the myodome is represented by small paired depressions lateral to the pituitary
vein foramen and ventral to the abducens canal. The ventral otic fissure lies more posteriorly
than in Mimia or Moythomasia and I have argued elsewhere (Gardiner 1973: 106; Gardiner &
Bartram 1977: fig. 8) that this is a consequence of the enlargement of the myodome (but see
Schaeffer & Dalquest 1978); the larger the muscle canal is, the further back lies the fissure (see
also Patterson 1975: 543). With this increase in size of the myodome the ventral otic fissure
migrated further posteriorly until it became confluent with the vestibular fontanelles, and the
enlarged dorsum sellae, which in Mimia and Moythomasia is ossified by the basisphenoid, is now
ossified by the prootic. Consequently the ventral otic fissure is represented by the
prootic/basioccipital suture, not the basisphenoid/basioccipital suture as in the Gogo
palaeoniscids and Polypterus. Alternative explanations are provided by Schaeffer & Dalquest
(1978) and Bjerring (1978).

14. Anterior and middle cerebral veins. Primitively, both these veins were well developed in
actinopterygians, the anterior cerebral vein leaving the telencephalon recess high up through a
transversely-directed canal in front of the epiphysial crest and trochlear nerve, whereas the
middle cerebral vein left the cranial cavity through the metencephalic recess (Mimia,
Polypterus) or the recess for the optic lobe (Pteronisculus, Moythomasia) but always behind the
trochlear foramen.

Canals for the anterior and middle cerebral veins are present in Mimia, Moythomasia,
Kansasiella (Poplin 1974: fig. 20, v, v.cer.ant), Pteronisculus (Nielsen 1942: fig. 6, v, v.cer.ant)
and Polypterus (Allis 1922: 228), and in Polypterus both veins run into the supraorbital vein as in
larval Lepisosteus (Hammarberg 1937: figs 25-27).

Despite the fact that the middle cerebral vein fails to develop in Amia (Bertmar 1965) and the
anterior cerebral vein is absent in adult Lepisosteus, both veins were present in para-
semionotids, Caturus, Heterolepidotus, 'Aspidorhynchus', Macrepistius, Dapedium, Lepi-
dotes, Pachycormus, pholidophorids and leptolepids (Patterson 1975: 412). Both veins occur in
most living teleosts but according to Bertmar (1965: 122) and Patterson (1975: 411) have
migrated to a more posterior position during the ontogeny and phylogeny of the group; the
middle cerebral vein now falls into the jugular vein.

The position of the anterior cerebral vein in non-teleostean actinopterygians is remarkably
constant but the same cannot be said for the middle cerebral vein. Primitively the canal for the
middle cerebral vein originated in the optic recess above the trigeminal canal and behind the
trochlear foramen. It passed down through the pterosphenoid to emerge in the orbit below and
behind the pterosphenoid pedicel in the roof of the ganglion recess (for the gasserian and
lateralis ganglia), close to the opening of the superficial ophthalmic canal. This is the condition
seen in Moythomasia and most caturids; also in pholidophorids and leptolepids except that in

250

B. G. GARDINER

prof

cotel

ctel

copl

ano

sue

amyd

crd

Fig. 33 Mimia toombsi Gardiner & Bartram. Preserved orbitotemporal and ethmoid parts of the
neurocranium and attached dermal bones, in dorsal view, from BMNH P. 56476.

RELATIONSHIPS OF PALAEONISCIDS 251

the latter two groups the ophthalmic canal is missing and the foramen for the middle cerebral
vein opens in line with the groove for the superficial ophthalmic nerves.

Thus primitively the trigeminal and middle cerebral vein canals lay close together (see
particularly Mimia) and consequently it is not surprising to find the former capturing the latter in
many teleosts.

In Latimeria the anterior cerebral vein leaves the cranial cavity through a separate foramen in
the anterior wall of the orbit and falls into the supraorbital vein (Millot & Anthony 1958;
Robineau 1975: 46) much as in Polypterus. There is, however, doubt as to whether or not a true
middle cerebral vein occurs. Millot & Anthony (1958: fig. 5) indicated a middle cerebral vein
passing behind the dorsal prefacial process of the prootic and running into the jugular vein
immediately in front of the jugular canal, whereas Robineau (1975) said it was absent, adding
that the ventrolateral sinus which lies in the posterior cranial cavity between the dura mater and
cranial wall is its likely homologue. If a true middle cerebral vein is present in Latimeria then by
comparison with actinopterygians it would pass out between the two parts of the cranium with
the trigeminal nerve, as reported by Millot & Anthony (1958).

In Neoceratodus the anterior cerebral vein is much as in Latimeria and actinopterygians,
passing through a separate foramen in the orbital wall and falling into the supraorbital vein
(Spencer 1893: fig. 15). In the Devonian dipnoan Griphognathus Miles (1977: fig. 61) has
demonstrated an extensive epiphysial plexus similar to that reported in Latimeria (Robineau
1975: 46) and in one specimen found a pair of foramina in the orbitotemporal for the exit of two
branches of the anterior cerebral vein (cf . Torpedo, Holmgren 1940: fig. 154). In Chirodipterus,
however, Miles (1977: figs 17, 47; see also Stensio 19636: fig. 43c) has also demonstrated the
presence of a perichondrally-lined canal passing posterolaterally under the utricular recess and
opening into the jugular canal, which undoubtedly housed the middle cerebral vein in life.
Although other workers (Chirodipterus, Save-Soderbergh 1952: fig. 6; Dipnorhynchus ,
Thomson & Campbell 1971: fig. 32) have restored the canal for the middle cerebral vein to a
more dorsal position, all agree that it lies behind the facial and trigeminal canals. This must be a
dipnoan specialization since primitively in osteichthyans the middle cerebral vein leaves
the cranium in front of the trigeminal nerve or between the trigeminal and facial nerves (Mimia,
Ectosteorhachis).

In rhipidistians a canal for the anterior cerebral vein has been described in Rhizodopsis
(Save-Soderbergh 1936), Ectosteorhachis (Romer 1937: fig. 8, unlabelled) and Eusthenopteron
(Stensio 19636: fig. 50A). In Ectosteorhachis (Romer 1937: fig. 8) there was also an anastomosis
between the anterior cerebral veins reminiscent of the epiphysial blood plexus of Latimeria,
Griphognathus and the embryonic stages of actinopterygians and selachians. The path of the
middle cerebral vein in Ectosteorhachis is represented by a large canal which lies below and
between the trigeminal and facial canals and opens into the jugular trough. A narrower canal in
a similar position has also been dscribed in Glyptolepis (Jarvik 1972: fig. 21B). On the other
hand in Eusthenopteron Bjerring (1971: fig. 17) has suggested that the middle cerebral vein
passed dorsal to the trigeminal canal as in actinopterygians (see also Jarvik 1980: 186).

In selachians the anterior cerebral vein is as in osteichthyans whereas the middle cerebral
vein, which in development passes in front of the root of the trigeminal nerve (Holmgren 1943:
fig. 57, v.c.int; Bertmar 1965: fig. 9), passes through the trigeminal foramen with the trigeminal
nerve in the adult and falls into the jugular vein (cf. teleosts).

15. Sclerotic bones. In Mimia the sclerotic ring may be comprised of two, three or four segments,
or may be a complete ring. The only possible explanation for these varying conditions is to
assume that there were always four ossification centres and that ontogenetic fusion occurred.
Similarly in Pteronisculus , where there are normally four segments, three have been recorded in
some instances (Nielsen 1942), and in Moythomasia there may be two segments (M. durgaringd)
or four (M. nitida, lessen 1968). The majority of palaeoniscids also have four segments
(Cheirolepis, Watsonichthys, Mesonichthys, Nematopychius, Amblypterus, Gonatodus, Cornu-
boniscus, Paramblypterus, Commentrya, Birgeria). There are also four sclerotics in
Australosomus, Lawnia, Chondrosteus, Acipenser, Dorypterus, Bobasatrania and some

B. G. GARDINER

crd

amyv

amyd

sue

cotel

ctel

crd

3 mm

pinf

Fig. 34 Mimia toombsi Gardiner & Bartram. Preserved ethmoid region of the neurocranium and
attached dermal bones, in dorsal view, from BMNH P. 56505.

pholidophorids, but in Pholidophorus macrocephalus , leptolepids and other teleosts there are
only two (Patterson 1975: 414). There are similarly two segments in fossil actinopterans such as
Lepidotes and Dapedium (Edinger 1929) and in Aspidorhynchus, Pachycormus, Mesturus and
Macromesodon. From this evidence we may conclude that a sclerotic ring composed of four
segments is primitive for actinopterygians.
Elsewhere within the osteichthyans the number of segments is much greater. Thus in

RELATIONSHIPS OF PALAEONISCIDS

253

amyd

cor

fendc

ore

amyv

2mm

Fig. 35 Mimia toombsi Gardiner & Bartram. Ethmoid region in anterior view, from BMNH
P. 53240. Broken lines denote limits of nasal capsules.

actinistians there are 18 to 20 in Latimeria (Millot & Anthony 1965). There are approximately 18
in the onychodont Strunius (Jessen 1966), 17 in the rhipidistian Osteolepis, and in
Eusthenopteron as many as 35 segments (Jarvik 1944«). Within the dipnoans similar large
numbers have been recorded in Rhinodipterus and Dipterus (more than 20 in each; Schultze
1970), but in the Devonian Griphognathus Miles (1977: 249) has described a single undivided
ring. Miles decided this was 'a specialized feature brought about by the late ontogenetic fusion of
numerous segments'.

Elsewhere in gnathostomes a sclerotic ring occurs in placoderms and acanthodians. In the
latter group it always appears to comprise five segments (Acanthodes, Protogonacanthus.,
Homalacanthus, Triazeugacanthus , Miles 1965, 1966; Cheiracanthus, Mesacanthus, Watson
1937) but in placoderms the number is usually four (Phlyctaenaspis, Arctolepis Heintz 1962: 36;
Coccosteus, Holonema Miles 1971b: 141). Only three have been recorded in Bothriolepis
(Stensio 1948) but in the rhenanids Gemuendina and Jagorina there are as many as 10 and 12
segments respectively (Stensio 1950). In Goodradigbeeon (White 1978: 196) there is a circlet of
four dermal plates but a further dermal plate articulates with part of the inner margin of this ring
and an almost complete layer of perichondral bone covered both surfaces of the calcified
sclerotic cartilage (see BMNH P. 33734, P. 50455). Although the so-called sclerotic ossifications

254 B. G. GARDINER

recorded in the agnathans Jamoytius and Lasanius (Ritchie 1968: 21) appear to be no more than
the remains of the pigmented retina, and in other anaspids are probably circumorbital scutes
(Janvier 1981: 137), distinct sclerotic bones occur in a number of celphalaspids (Janvier 1981)
and both dermal and perichondral elements can be recognized in specimens of Atelaspis
(Aceraspis} robusta (BMNH P.2138) and Hemicyclaspis (BMNH P. 8809, P.8801). These
apparently number four segments.

A sclerotic ring is thus a synapomorphy of advanced 'agnathans' and gnathostomes and which
in gnathostomes and cephalaspids primitively consisted of four segments.

A sclerotic ring of more than 12 segments is a synapomorphy of osteolepids, actinistians,
onychodonts, porolepids and choanates.

Ethmoid region and associated dermal bones

The ethmoid region is well preserved in Mimia and the cartilage was entirely enclosed in a thin
layer of perichondral bone. Such a layer also lines the cavities of the nasal capsules and the many
canals. There was no endochondral ossification in the ethmoid region. The perichondral bone is
closely applied to the overlying dermal bones except in the region beneath the premaxillae and
the anterior part of the rostral.

Mimia toombsi

The orbital face of the postnasal wall is shown in Fig. 36; it is divided sagittally by the very thin
interorbital septum. This septum widens in the dorsal region where the paired, diverging
perichondrally-lined canals of the olfactory nerves pass into the ethmoid region. There is an
elongated dorsolateral gap in the ossification of this canal (gl, Figs 13, 36), as in Boreosomus
(Nielsen 1942: fig. 62).

Above and below the olfactory nerves are pairs of depressions in the postnasal wall. The lower
pair (amyv, Figs 13, 36, 40) have a greater rostral extent than the upper pair (amyd, Figs 13, 36,
40). A similar set of depressions is present in Boreosomus (Nielsen 1942: fig. 65) and
Pteronisculus (Nielsen 1942: fig. 17). These depressions must be anterior myodomes as Nielsen
suggested, the dorsal pair for the superior oblique muscles and the ventral pair for the inferior
oblique muscles.

The wall separating the dorsal myodomes in Mimia is pierced by a few minute holes. Much
larger apertures are found in a corresponding position in other primitive actinopterygians. Thus
in Boreosomus (Nielsen 1942: fig. 62) there is a large fenestra connecting each pair of
myodomes. In Saurichthys (Stensio 1925: fig. 10B), where the anterior myodomes are very
shallow, there are fenestrae in a similar position to those of Boreosomus, and the interorbital
septum dividing the single anterior myodome of Australosomus (Nielsen 1949: fig. 2) is also
pierced by a large fenestra.

Another conspicuous feature of the postnasal wall is a pair of marked lateral notches slightly
below the level of the olfactory nerve canals. Such notches occur in Pteronisculus and
Kansasiella, and mark the lateral limits of the fenestra endonarina communis, described below.

The paired articulation for the palatoquadrate cartilage faces posterolaterally at the junction
of the postnasal wall and suborbital shelf.

The anterodorsal corner of the orbit and the postnasal wall are pierced by a large number of
pores, many more than have been recorded in other primitive osteichthyans, and they form
three groups. The first consists of canals leading from the anterodorsal corner of the orbit and
dorsolateral parts of the dorsal anterior myodomes to the skull roof (nasals), the second of
canals leading from the upper part of the postnasal wall to the cavity of the nasal capsule , and the
third of canals leading from the lower part of the postnasal wall to the nasal cavity, to the
premaxillae and rostral bones and to the nasobasal canals (see below).

The canals of the first group are of variable calibre and occur medial and lateral to the
supraorbital sensory canal (crd, Figs 34, 36; cor, Figs 35, 36, 40). There are at least nine such
canals on the left side and fewer on the right in BMNH P. 56505, an acid-prepared specimen on
which most of the following description is based. The most posterior of these canals leads to the

RELATIONSHIPS OF PALAEONISCIDS

255

amyd

cor

crd

prof

sue

pnw

nfendc

Pmx

amyv

fmxV,

bucVII

i nc

IOS

apal

3 mm

Fig. 36 Mimia toombsi Gardiner & Bartram. Braincase and associated dermal bones as if cut
transversely through anterior region of orbit, to show the postnasal wall in posterior view; cut
'surfaces cross-hatched. From BMNH P. 56505.

supraorbital sensory canal. This sensory canal lay in an open groove, bounded by raised walls,
beneath the nasal bone. The raised walls fitted in turn into a groove, lined with perichondral
bone, on the dorsal surface of the neurocranium. The canal connected with this groove and
presumably conveyed branches of the superficial ophthalmic nerve. The remaining canals
presumably carried other branches of the superficial ophthalmic nerves to the skull roof.

The canals of the second group are also asymmetrical in number in BMNH P. 56505. On the
left side one canal passes through into the posterior wall of the nasal capsule. Below this are four
canals which pass slightly dorsally into the postero ventral corner of the cavity. Finally, two fine
canals pass into the nasal capsule close to the lateral notch in the postnasal wall. On the right
side the dorsal canal is symmetrical in position to that on the left side, but there is only one lateral
canal and two ventral canals. Of the last two the lateral canal is the largest in the group. These
orbitonasal canals are assumed to have carried branches of the profundus nerve.

256

B. G. GARDINER

i nw

prof,

sue

ano

cnc

-bi nc

amyv

pnw

Pmx

i nc

ore

vnabc

Fig. 37 Mimia toombsi Gardiner & Bartram. Braincase and associated dermal bones as if cut
transversely through the nasal capsules, and viewed from the rear, from BMNH P. 56505. Cut
surfaces cross-hatched.

The pores of the ventral group are also asymmetrical in distribution, and in the directions
which their corresponding canals take. Although these canals are visible for most of their length,
their exact anterior limit and their minute terminal branches are not always clear. To investigate
them more fully would involve further damage to the specimen: the specimen which was serially
sectioned was deficient in this region. On the right side is an arc of six pores and two pores
respectively dorsal and lateral to these. The lowermost pore of the arc passes anteromedially
into a groove on the neurocranium running in the same direction, just above the tooth row of the
premaxilla. Canals from the three following dorsal pores in the arc run into this groove, which
appears to end near the tip of the rostral bone. The groove is remote from the more lateral
infraorbital sensory canal in the premaxilla, but nerve endings could have crossed the gap
between the neurocranium and the dermal bone. The next pore in the series leads to a
dorsally-directed canal which connects with the central of the three anterior nasobasal canals
described below. This canal sends fine branches to the infraorbital sensory canal. The canal
associated with the most dorsal pore of the arc is short and leads into the same sensory canal. The
pore dorsal to the group just described leads to a dorsally-directed canal which anastomoses with
the lateral anterior nasobasal canal. Finally, the lateral pore forms a short canal leading to the
infraorbital sensory canal.

On the left side is an arc of five pores and one lateral to these. The lowermost leads to a canal

RELATIONSHIPS OF PALAEONISCIDS

257

prof

sue

vnabc

Fig. 38 Mimia toombsi Gardiner & Bartram. Braincase and associated dermal bones as if cut
transversely through the nasal capsules, and viewed from the front, from BMNH P. 56505. Cut
surfaces cross-hatched.

and groove corresponding to that described on the right side. The next dorsal canal appears to
turn into this groove while the third canal in the series runs parallel to the second and also joins
the groove, after giving off a branch which anastomoses with the fourth canal. The fourth canal
joins the most posterior of the three left side ventral nasobasal canals. The most dorsal pore of
the left side arc leads to a highly-branched canal; two of these branches go to the rostral area,
and two pass dorsally to open into the floor of the nasal capsule. At several points along this
canal there are short branches to the infraorbital sensory canal. Finally, the pore lateral to the
arc also leads to the sensory canal. The complicated system of canals described above is not
found in any living fish but by comparison with living and fossil dipnoans it may be suggested that
they transmitted either maxillary branches of the trigeminal or buccal branches of the facial
nerve, or in some cases both.

The bulbous nasal capsules are more completely enclosed in bone than in other
actinopterygians. The walls enclosed the ventral, posterior and the medial half of the anterior
surface of the capsule, leaving a large fenestra endonarina communis. The latter faces antero-
laterally and slightly dorsally, and is covered, except for the fenestra exonarina anterior, by the
nasal and narrow strips of the rostral and premaxillary bones. The posterior nasal tube opened
laterally between the nasal and the notch in the postorbital wall. The canal for the olfactory
nerve opened into the posteromedial wall of the nasal cavity.

258

prof

B. G. GARDINER

ano

cor

sue

ore

IDS

i nc

Pmx

vnabc

3 mm

Fig. 39 Mimia toombsi Gardiner & Bartram. Braincase and associated dermal bones as if cut
transversely through anterior region of orbit; showing postnasal wall in posterior view. Cut
surfaces cross-hatched. The intramural canals and cavities lined with perichondral bone are
stippled. From BMNH P.56505.

Apart from the pores piercing the posterior wall and floor of the nasal capsule there are two
other groups of pores. On the left side two canals, the posterior one branched, run from the roof
of the cavity to the rostral above. On the right side there is only one such canal. These canals
(prof2, Figs 37, 38, 40) presumably contained twigs of the upper profundus branch which crossed
the rear wall of the nasal cavity, as in Griphognathus (Miles 1977: fig. 63) and Porolepis (Jarvik
1942: fig. 42).

The pores of the second group are larger and pierce the anteroventral corner of the nasal
cavity in a row of three (nabc, Fig. 37; mnabc, Inabc, Figs 39, 40), medial to the opening of the
dorsal branch of the infraorbital sensory canal (bine, Fig. 37) in the premaxilla. They lead to
pores opening beneath the rostral (but not piercing it) and appear to be homologous to the two
nasobasal canals (in a similar position) in Eusthenopteron (Jarvik 1942: 470). Communicating
with the anterior ends of the medial and lateral nasobasal canals in Mimia is a further set of
canals (three on the left, two on the right) which pass posteroventrally to open by a pair of pores
in the roof of the mouth, one on either side of the midline (vnabcf, Figs 38, 41). These are the
homologues of the ventral nasobasal canals of Eusthenopteron (Jarvik 1942: 470) and
presumably transmitted branches of the palatine nerve.

RELATIONSHIPS OF PALAEONISCIDS

cor

259

Prof.

ethc

mnabc+
Inabc

vnabc

for

amyd

amyv

105

Pmx

Fig. 40 Mimia toombsi Gardiner & Bartram. Ethmoid region of braincase and associated dermal
bones in sagittal section, from BMNH P.56505; showing perichondrally lined canals. Sectioned
bone cross-hatched.

The ethmoid region of the neurocranium is covered dorsally and anteriorly by a median
rostral and paired nasals and premaxillae.

The rostral extends a little over 25% of the length of the skull roof (Figs 41 , 101). The rostral is
bounded posteriorly by the f rentals, which it joins in a serrated suture. The lateral edges of the
bone are straight, forming sutures with the nasals except where these two bones are emarginated
to form the anterior nostril. At the level of the anterior nostril, the rostral curves sharply
ventrally (Fig. 40) and less sharply laterally, thus forming a rounded protruding tip to the snout.
The antero ventral edge of the rostral is V-shaped, forming sinuous sutures with the dorsomedial
edges of the premaxillae (Fig. 41), which exclude the rostral from the border of the mouth. The
external surface of the rostral bears the usual vermiculate pattern of ganoine ridges. The centre
of radiation of the bone appears to occur slightly anterior to the mid-point of the line joining the
anterior nostrils, at the point of greatest curvature of the bone. The ganoine ridges which extend
posteriorly from this point to the f rentals are long and parallel (Fig. 42). Anteriorly, the ganoine
ridges form an irregular, maze-like radiation. The passage of the ethmoidal commissure is
indicated on the external surface of the rostral by an arc of pores (four in BMNH P. 56483),
which extends through the centre of radiation to points on the lateral part of the suture between

260 B. G. GARDINER

the rostral and premaxillae. Internally this sensory canal lay in an open groove limited by two
raised ridges of bone (Fig. 42).

The nasals are paired, narrow elongated bones flanking the rostral, and slightly shorter than
this bone. The lateral border of the nasal forms a gentle, regular curve limiting the anterodorsal
edge of the orbit; this border is not emarginated by the posterior nostril, as it is in many other
palaeoniscids. Posteriorly, the nasal forms an oblique suture with the frontal and dermo-
sphenotic. The anterior suture with the premaxilla is transverse and slightly curved. The
external surface of the nasal bears long ridges of ganoine orientated along the length of the bone
(Fig. 43). These are particularly narrow and closely-spaced near the orbital edge. The ganoine
rugae become shorter anteriorly, close to the anterior nostril. The supraorbital sensory canal
occupied a groove on the inner surface of the nasal for the first half of its course, but anteriorly
was enclosed in a raised tube of bone which ends blindly at the level of the anterior nostril. The
passage of the canal is indicated externally by a row of fine pores (five in BMNH P. 56483). In
many specimens (e.g. BMNH P. 56483) the nasal bears a short pit-line anterior to, and in line
with, the supraorbital canal; this pit-line probably represents the anterior continuation of the
sensory canal, as in Polypterus.

The paired premaxillae are flat bones facing anteriorly and slightly laterally and ventrally, and
form the anterior edge of the mouth. This edge bears two kinds of teeth. The larger kind forms a
row of about ten inner teeth. These are deciduous, leaving large circular scars on the bone; they
do not differentiate a sharply-defined enamel (acrodin) cap and match the major teeth on the
dentary and maxilla. The smaller teeth, whose equivalents are also found on the last two bones,
are exterior to the major teeth, and occur close to the edge of the mouth (Fig. 44). The lateral
edge of the premaxillary tooth row gives way to a small area of overlap for the first infraorbital
bone (lachyrmal). The infraorbital sensory canal entered this area of the premaxilla. The lateral
edge of the premaxilla lies alongside the lateral edge of the postnasal wall of the ethmoid region
of the braincase below the posterior nostril; the edge of the premaxilla completes, with the nasal
and dermosphenotic, the curve of the dorsal and anterior edge of the orbit. The two premaxillae
form a short, straight median suture with each other. The sutures with the rostral and nasal
bones are described above.

The ornament of the premaxilla is complex. Along the orbital edge it consists of the usual
straight, narrow, parallel rugae and along the oral edge, close to the teeth, it is in the form of
ganoine tubercles. Between these two regions extends an area of irregular, vermiculate rugae
orientated mainly towards the tip of the snout. The pattern of sensory canal pores is also
complex. An irregular line of pores passes from the ventrolateral corner of the bone to the
suture with the rostral. This line belongs to the anterior part of the infraorbital sensory canal.
Internally this canal was housed in a tube which becomes an open groove in the upper part of the
bone (Fig. 44); the tube is pierced by fine pores which transmitted nerves from the buccal and
maxillary rami of the facial and trigeminal nerves. A short, dorsal branch of the infraorbital
sensory canal, enclosed in bone, opens internally into the nasal cavity before reaching the dorsal
edge of the premaxilla. Apart from the pores belonging to the infraorbital sensory canal, there
are others which vary in position and distribution from specimen to specimen. Thus in the right
premaxilla of BMNH P. 56505 these pores are distributed above and below the middle region of
the infraorbital canal. The left premaxilla of the same individual has only the ventral pores.

The ethmoid region is partly covered ventrally by an anterior projection of the parasphenoid
and by a pair of vomers (Vo, Figs 41, 50, 55). The vomers are small, triangular-shaped
ossifications which lie medial to the articular facets for the palatoquadrates and in front of the
postnasal wall. A median tongue of the parasphenoid passes between them (Psp, Fig. 50). The
vomers are covered by sharply-pointed teeth, larger than those on the parasphenoid.

Moythomasia durgaringa

Unfortunately, no specimen of this species has been found so far in which the ethmoid region is
well preserved. Nevertheless the orbital face of the postnasal wall is visible in several specimens.
It is very similar to that described in Mimia; there is the same dorsolateral gap in the canals for
the olfactory nerves (Fig. 7) and there are two anterior myodomes with the lower pair having the

RELATIONSHIPS OF PALAEONISCIDS

261

262

B. G. GARDINER

ano

ethc

Fig. 42 Mimia toombsi Gardiner & Bartram. Rostral in posterior (left) and anterior views, from

BMNH P.56483.

sue

ano

I mm

Fig. 43 Mimia toombsi Gardiner & Bartram. Right nasal in lateral (right) and medial views, from

BMNH P.56483.

RELATIONSHIPS OF PALAEONISCIDS

263

i nc

Fig. 44 Mimia toombsi Gardiner & Bartram.
Right premaxilla in lateral (above) and medial
views, from BMNH P.56483.

bine

i nc

greater rostral extension. Similarly the articulations for the palatoquadrates occur at the
junction of the postnasal wall and suborbital shelf, and the postnasal wall is pierced by a large
number of pores. The nasal capsules are bulbous and are enclosed to the same extent as in
Mimia.

The same dermal bones cover the ethmoid region as in Mimia, but their shape and extent are
quite different. The rostral extends for over 25% of the length of the skull roof and forms a
highly digitate suture with the frontals posteriorly. Behind the anterior nostrils the lateral edges
of the rostral are straight and parallel, but at the level of the anterior nostrils, where the bone
curves sharply ventrally, the lateral edges flare outwards and then converge, so that the
antero ventral edge of the rostral is V-shaped (Figs 45, 48). Unlike Mimia, the rostral in
Moythomasia separates the two premaxillae and takes part in the border of the mouth. The
ornamentation of the rostral consists of very stout tubercles and ridges of ganoine, which for the
most part have fused into an irregular maze-like configuration. The centre of radiation is
considerably anterior to the anterior nostrils and the ethmoid commissural sensory canal passed
through it. Internally this canal was housed in a tube, pierced by two fine pores (Fig. 45) which
presumably transmitted branches of the buccal and maxillary nerves. On the ventral margin of
the bone the ganoine tubercles are pointed and give way to one (Fig. 48) or two (Fig. 45) teeth
which correspond to the smaller teeth on the premaxillae, maxillae and dentaries, external to the
major, replaceable teeth.

The nasals are paired, narrow bones and the sutures between them and the rostral are straight
except where they are interrupted by the notch for the anterior nostril. The lateral border of the
nasal is emarginated by a much larger notch for the posterior nostril. Posteriorly, the nasal forms
an oblique suture with the frontal and dermosphenotic, and ventrally it overlaps the premaxilla
for a short distance (Fig. 48). Antero ventrally, a little beyond the notch for the anterior nostril,

264

B. G. GARDINER

the nasal loses contact with the rostral altogether and a large foramen is formed between the
rostral, nasal and premaxilla. The external surface of the nasal bears massive fused tubercles and
ridges of ganoine which latererally, near the orbital edge, are orientated along the length of the
bone. The supraorbital sensory canal extended in a groove on the inner surface of the nasal and
terminated at the level of the dorsal margin of the anterior nostril. The passage of the canal is
indicated externally by a row of eight pores in BMNH P. 53255 (Fig. 46). Lateral to and parallel
with this sensory canal is an elongate foramen (p, Fig. 48). In other specimens (Fig. 46) this
foramen is the same size as the sensory canal pores.

The flat, stout premaxillae face anterolaterally, and ventrally bear two kinds of teeth. There
are never more than four of the deciduous larger kind, which have sharply-defined enamel caps
(acrodin). The smaller, pointed teeth form an external row of some 15-20 (Fig. 48). Anteriorly
the premaxilla is in sutural contact with the rostral but dorsally the suture gives way to a small
area of overlap for the nasal. Posteriorly, above the tooth row, the posterior margin has a more
extensive overlap area for the first infraorbital (lachrymal). The lateral edge of the premaxilla
lies alongside the lateral edge of the postnasal wall and forms the lower margin to the posterior
nostril.

The ornamentation on the premaxilla is similar to that on the rostral, with stout striae of
ganoine dorsally and rounded tubercles ventrally. Internally the infraorbital sensory canal was
housed for the most part in an open gutter, though the dorsal branch which opens short of the
dorsal edge of the premaxilla was enclosed in a tube. The course of the infraorbital sensory canal
is marked externally by a line of up to twelve pores which in some specimens are in the form of
slits. Below the middle region of the infraorbital canal is a further series of four pores, which lead
into a short canal within the bone. Internally this canal is pierced by three pores (p, Fig. 47).

The paired vomers are covered by sharply-pointed teeth, the posterior row of which is much
enlarged (Gardiner & Bartram 1977: fig. 7). The vomers are widely separated, irregular in
outline and lie medial to the articulation facets for the palatoquadrates (Fig. 7).

Ethmoid region: discussion

1. Anterior myodome. Two pairs of anterior myodomes appear to be present in most
palaeoniscids (Mimia, Moythomasia, Kentuckia, Pteronisculus, Boreosomus) . In the palaeo-
niscid Kansasiella (Poplin 1974: fig. 18), however, the ventral anterior myodome is median. In
the pholidopleurid Australosomus there is a single, large paired anterior myodome ventral

a no

ethc

Fig. 45 Moythomasia durgaringa Gardiner & Bartram. Rostral in posterior (left) and anterior

views, from BMNH P.53255.

RELATIONSHIPS OF PALAEONISCIDS

265

sue

ano

1mm

pno

Fig. 46 Moythomasia durgaringa Gardiner & Bartram. Left nasal in lateral (right) and medial views,

from BMNH P. 53255.

to the olfactory nerve canal (Nielsen 1949: fig. 13). Two pairs of rather shallower anterior myo-
domes are present in Saurichthys, parasemionotids, Macrepistus (Schaeffer 1971: fig. 6) and
Dapedium, and as in palaeoniscids the upper pair lie dorsal to the olfactory nerve canal and often
communicate with one another through an interorbital fenestra. In some caturids (Caturus,
Heterolepidotus, 'Aspidorhynchus'}, however, the dorsal myodome is obscured by an enlarged
interorbital fenestra (Patterson 1975: fig. 102). InAmia both oblique muscles, together with a
vein, pass through a large orbitonasal canal into the olfactory nerve canal and are attached to its
floor.

In Pachycormus (Patterson 1975: 516) there is no ventral anterior myodome, and in this it
resembles pholidophorids and Recent teleosts where there is a median, dorsal anterior

i nc

Fig. 47 Moythomasia durgaringa Gardiner &
Bartram. Right premaxilla in medial (above)
and lateral views, from BMNH P. 53255.

266

B. G. GARDINER

myodome and where both pairs of oblique muscles enter the foramen olfactorium evehens; this
is a specialization.

There is no anterior myodome in living chondrosteans or Lepisosteus; the oblique muscles
originate separately in the former but together in the latter. In Polypterus the two muscles are
separated as in chondrosteans and dipnoans, but the superior oblique muscle enters the
funnel-like opening of the profundus canal where it attaches to the medial wall.

In Nesides and Euporosteus (Jarvik 1942: fig. 75A, B) there is said to be a single small
myodome ventral to the olfactory nerve, similar to that in Australosomus , except that through it
also ran the orbitonasal artery. In Diplocercides (specimen figured by Stensio, 1922: pi. 3, fig. 1)
there are two pairs of shallow ventral myodomes, and these depressions are in a position
homologous to the origins of the superior and inferior oblique muscles in Latimeria, although in
Latimeria myodomes are absent. The condition in dipnoans is less easy to interpret, but in larval
Neoceratodus the superior oblique muscle originates in the upper anteromedial corner of the
orbit, whereas the inferior oblique arises near the bottom corner. In adult Neoceratodus there
are no anterior myodomes and Miles (1977) also failed to find any such structures in the Gogo
dipnoans.

In the rhipidistian Eusthenopteron (Jarvik 1942: figs 49, 50, fo.m.obl), very shallow paired
ventral pits have been described in the postnasal wall, which according to Jarvik (1942: 437)
were probably the place of origin of the oblique muscles. No such depressions have so far been
described in any other rhipidistian (Ectosteorhachis, Rhizodopsis, Porolepis, Holoptychius,
Youngolepis); in Glyptolepis sp. (BMNH P. 47838), however, there are two pairs of shallow
ventral myodomes below the profundus canal, in a position homologous to the paired
depressions in Diplocercides. No myodomes have been described in chondrichthyans but
depressions have been noted in the perichondral lining of the front of the orbit in certain
placoderms (Brindabellaspis, Young 1980: fig. 10), which presumably served as points of origin
for individual muscles. A corresponding anteroventral depression in the floor of the orbit of
Macropetalichthys (Stensio I963b: fig. 32) has been attributed to the inferior oblique muscle.

From this evidence we may conclude that anterior myodomes, like posterior myodomes, were
primitively absent in gnathostomes and osteichthyans. Anterior myodomes have been
independently acquired in actinopterygians, actinistians and some rhipidistians. In actinoptery-
gians the anterior myodome typically consists of two pairs of shallow depressions, one pair
above and the other below the olfactory nerve canal, whereas in actinistians and rhipidistians the
two pairs of depressions are both below the olfactory nerve canal, in the floor of the postnasal
wall.

2. Postnasal wall and nasal capsule. The ethmoid region of Mimia toombsi is traversed by a
greater number of canals than in any other osteichthyan and the number of pores piercing the
postnasal wall is correspondingly high. The dorsal group of canals (4-7) in the postnasal wall are
presumed to have served for branches of the profundus nerve.

The ventral group of canals (5-8) pierce the ventrolateral corner of the postnasal wall (ore,
Figs 37-40). The most lateral of these canals communicates with the infraorbital sensory canal
and thus must have transmitted the buccal branch of the facial nerve. It is impossible to decide
whether the remainder of these canals transmitted maxillary or buccal nerve branches or both.
Several of these dorsal and ventral canals (ore, Fig. 39) communicate with the nasobasal canals
(mnabc, Inabc, Fig. 39), which run from the nasal cavity to the rostral area and there
communicate with a set of ventral nasobasal canals (vnabc, Figs 39, 40). Similar nasobasal
canals are found in the floor and anterior wall of the nasal capsule of Eusthenopteron (Jarvik
1942: fig. 57), Glyptolepis (Jarvik 1972), Youngolepis (Chang 1982: fig. 14) and Griphognathus
(Miles 1977: fig. 62(a), prog.V, V2) and must be presumed to be a primitive feature of
osteichthyans. In Eusthenopteron and Glyptolepis Jarvik (1942, 1972) considered that these
nasobasal canals transmitted branches of the profundus nerve and that the palatonasal canal of
Eusthenopteron served for a branch of the maxillary nerve. Rosen et al. (1981: 192), however,
considered that the nasobasal canals in Eusthenopteron transmitted the truncus infraorbitalis
and accompanying vessels. Miles (1977: 130) decided that in Griphognathus the dorsomedial of

RELATIONSHIPS OF PALAEONISCIDS

267

the two nasobasal canals carried a ventral division of the profundus nerve, whereas the lateral
canal transmitted the maxillary nerve.

Thus we may conclude that the nasobasal canals transmitted maxillary and buccal nerve
branches as well as branches of the profundus nerve.

The olfactory organ in fishes is a blind sac, with a single anterolateral or ventrolateral
aperture. This aperture is varyingly subdivided by flaps or more extensive tissue barriers into
incurrent and excurrent openings. Each olfactory organ is surrounded by cartilage or bone apart
from a single anterolateral aperture which serves for both incurrent and excurrent streams. This
nasal capsule is developed in continuity with the front end of the trabeculae; a median septum
nasi separates the right from the left capsule, the floor of the capsule is termed the solum nasi
and the roof the tectum. This is essentially the condition in Chlamydoselachus and Polyptems.

Primitively in osteichthyans the postnasal wall is ossified by the lateral ethmoids (Polyptems,
Latimeria, Acipenser, Amia and teleosts), but the only living fishes with any endoskeletal
ossification anterior to the postnasal wall are Amia, teleosts and Latimeria. However, the nasal
capsule is more or less completely perichondrally ossified in Mimia, Moythomasia and
Europorosteus (Jarvik 1942: 556) and in Eusthenopteron (Jarvik 1972: fig. 66D) and Gripho-
gnathus there is also some endochondral bone present. In Mimia and Moythomasia the
ethmoid region is a shell of thin perichondral bone with all the canals for nerves and blood
vessels also surrounded by tubes of perichondral bone. Consequently it is difficult to recognize
individual ossification centres in this region. Nevertheless, it would appear that paired lateral
ethmoids were primitively developed in the postnasal wall of osteichthyans because paired
lateral ethmoids are found in Polypterus, Acipenser, Birgeria, Perleidus, Lepidotes, Caturus,
Macrepistius, Amia, pachycormids, Recent teleosts, Macropoma and Latimeria (Patterson
1975: 499).

In Amia, pholidophorids, Hypsocormus and Recent teleosts there are additional endo-
skeletal ossifications. Paired pre-ethmoids occur in Amia and Hypsocormus, a median
supraethmoid, ventral ethmoid and anterior myodome bones are found in pholidophorids and
up to five ossifications are present in some teleosts, including a median supraethmoid, ventral
ethmoid and anterior myodome bones and paired capsular ethmoid bones. As Patterson (1975:
502) concluded, there is good evidence that the region anterior to the lateral ethmoids is a new
formation in advanced actinopterygians. In other words there has been an increase in the
number of ossification centres. From this I conclude that the primitive osteichthyan possessed
one pair of ossifications in the ethmoid region, the lateral ethmoids, and that each olfactory
organ was completely surrounded by bone apart from a lateral aperture.

3. Dermal bones of the snout. Primitively the roof of the nasal capsule in actinopterygians is
covered by a broad, shield-like rostral, bordered by the frontals behind and the nasals laterally.
The rostral contains a portion of the ethmoid commissure. This is essentially the condition in
Polypterus and palaeoniscids. In most previous descriptions of palaeoniscids (e.g. Aldinger
1937; Moy-Thomas & Dyne 1938; Nielsen 1942, 1949; Gardiner 1963, 1967; Jessen 1968) the
rostral is mistakenly called the postrostral, but it is now clear that it is penetrated by the ethmoid
commissure in almost all instances. Re-examination of Cheirolepis (Fig. 49) has convinced me
that the rostral is a single ossification (e.g. BMNH P.60533, P.4051a; GSM 88873; RSM
1973 . 12. 150) and there is little evidence for considering it to be comprised of several ossifications
as Gardiner (1963) and Pearson & Westoll (1979) have suggested. The evidence for a lateral
postrostral in Cheirolepis canadensis (Pearson & Westoll 1979: fig. 3d, h) rests on the
interpretation of a single specimen (BMNH P. 6815). In my estimation this lateral postrostral is
more reasonably interpreted as the rostral.

A similar shield-like rostral is also met with in pholidophorids (Patterson 1975: 497),
perleidids, redfieldiids, ptycholepids and Dapedium, but in Pachycormus (Patterson 1975: 511)
the large median rostral has fused with the underlying, toothed, lateral dermethmoids. In
parasemionotids (Patterson 1975: fig. 137), caturids (Caturus, Furo, Heterolepidotus,
Osteorachis), amiids, many semionotids (Acentrophorus, Lepidotes} and Lepisosteus (Patter-
son 1975: figs 135, 136) the median rostral is reduced to little more than a tube around the

268

B. G. GARDINER

ano

Pmx

ethc

Fig. 48. Moythomasia durgaringa Gardiner &
Bartram. Restoration of dermal bones of
snout, drawn as if folded out in one plane. Inset
at lower left is a restoration of sensory canals
and pores, indicated by broken lines.

RELATIONSHIPS OF PALAEONISCIDS

269

\ \v

X- V\

X \\

\ V\

\ ^

[

\ x

270 B. G. GARDINER

ethmoid commissure with the nasals meeting in the mid-line behind it. In teleosts, where the
ethmoid commissure remains bone-enclosed (e.g. leptolepids, Megalops, Elops), the tube-like
rostral is fused with the underlying dermethmoid as in Pachycormus (Patterson 1975: 511).

A median rostral is also found in Holoptychius (Jarvik 1972: figs 35, 36) where it is so small
that it only just embraces the ethmoid commissure. The rostral of actinopterygians may have
captured the middle part of the ethmoid commissure as Patterson (1975: 512) suggests, or
conversely it may have been primitively associated with it. Support for the latter point of view is
afforded by its presence in the rostrals of actinopterygians and Holoptychius, and for the former
by the condition in actinisitians, where in fossil forms (Laugia, Rhabdoderma, Macropomd)
the ethmoid commissure is contained within the premaxillae as in some rhipidistians (Eustheno-
pteron Jarvik 1942; Osteolepis Jarvik 1948; Megalichthys Thomson 1964). In Latimeria the
commissure is separate from the underlying premaxillae and bears four short bones along its
length. An ethmoid commissure wholly contained within the premaxilla is found in early
actinistians and osteolepiforms, and this is believed to be primitive. Conversely a shield-like
rostral is considered synapomorphous for actinopterygians.

The remaining portion of the snout was primitively covered in actinopterygians by the paired,
toothed premaxillae which contain the characteristic triradiate anterior portion of the
infraorbital sensory canal and the greater part of the ethmoid commissure. The premaxilla in
early actinopterygians has often been termed the rostro-premaxillo-antorbital because the
premaxilla and antorbital of higher actinopterygians are believed to result from its subdivision
(Gardiner 1963). However, it is better to regard this bone in early actinopterygians as the
premaxilla as it is homologous with that bone in rhipidistians where it also bears replacement
teeth and the triradiate portion of the infraorbital sensory canal.

A canal-bearing premaxilla is found in all Devonian palaeoniscids, ptycholepids and
Polypterus, but in perleidids, parasemionotids, caturids, amiids, semionotids, pachycormids,
pholidophorids and Recent teleosts the premaxillary and antorbital (canal-bearing) compo-
nents are separate.

The nasal is single in actinopterygians with the exception of Polyodon and Polypterus where
there are three. The nasal in Cheirolepis (BMNH P. 65527, P. 65528, P. 4050) is very similar in
shape to that of Mimia and there is little evidence to suppose it is composed of more than one
ossification.

Tetrapods, like most actinopterygians, have a single pair of nasals, but elsewhere in
osteichthyans the number is higher. In primitive actinistians (Diplocercides Stensio 1937)
there are five pairs of nasals, and some dipnoans (Soederberghia Lehman 1959) have as many as
eight. In osteolepids, Eusthenopteron (Jarvik 1948: fig. 16) has three, Megalichthys and
Osteolepis six, while the porolepid Holoptychius (Jarvik 1972: figs 35, 36) usually has five nasals.
Primitively in osteichthyans there were probably four or more nasals. In primitive
actinopterygians there was one.

In a previous publication (Gardiner 1963) I assumed that the antorbital branch of the
infraorbital sensory canal primitively anastomosed with the terminal portion of the supraorbital
canal between the nostrils in actinopterygians (see however Jollie 1969). The supraorbital canal
does anastomose with the infraorbital canal in Lepisosteus, Amia, Latimeria, porolepids and
osteolepids, but in dipnoans, Cheirolepis, Mimia, Moythomasia, Polypterus, chondrosteans and
teleosts (Nybelin 1967) it does not.

The primitive osteichthyan condition probably lacks such an anastomosis. In the primitive
actinopterygian the supraorbital canal passed between the nostrils as in chondrosteans,
Polypterus, Lepisosteus and Amia. In Latimeria the supraorbital canal joins the infraorbital
both in front of and between the nostrils, whereas in osteolepids and porolepids the supraorbital
and infraorbital canals anastomose in front of the nostrils. In chondrichthyans and some
placoderms the anastomosis of the two canals is behind the nostrils. In fossil dipnoans such as
Holodipterus (BMNH P. 52566) the ethmoid commissure connects the two supraorbital canals
rather than the two infraorbital canals as it does in actinopterygians, chondrichthyans and some
placoderms. The former condition must be regarded as secondary and in part due to the
interruption of the infraorbital sensory canal by the choana in dipnoans and primitive tetrapods.

RELATIONSHIPS OF PALAEONISCIDS 271

In osteichthyans there is a further series of bones associated with the roof of the nasal region,
the internasals or postrostrals. These are anamestic bones, variable in number and
arrangement. In actinopterygians these elements occur only in long-snouted forms such as
Acipenser (where there are more than twelve internasals), Polyodon, Chondrosteus and
Saurichthys (where there is a single pair of extremely long bones). In actinistians the internasals
form a median series of five or more in Diplocercides (Stensio 1937), two or more in Latimeria,
and one in Rhabdoderma (Forey 1981). In dipnoans there may be a single median internasal, as
in Fleurantia (Graham-Smith & Westoll 1937), Ceratodus and Neoceratodus, long paired
internasals, as in Griphognathus (Miles 1977: fig. Ill), Holodipterus and Scaumenacia
(Holmgren & Stensio 1936), or an irregular series, as in Dipnorhynchus, Uranolophus (Denison
1968, 1969) and Chirodipterus (Miles 1977). In osteolepids, Megalichthys (Thomson 1964) has a
single internasal, Osteolepis (Jarvik 1948) two median internasals, Panderichthys (Vorobjeva
1973) four and Eusthenopteron one large median internasal and five smaller anterior ones. The
porolepid Holoptychius (Jarvik 1972: fig. 35) has an irregular series of eight to fourteen
internasals. A small median internasal is present in Ichthyostega (Jarvik 1952) and loxommatids,
but in no other tetrapod. From this we may conclude that internasals are primitively present in
sarcopterygians but absent in actinopterygians.

Finally, there is one further series of bones associated with the nasal region in osteichthyans,
the tectals, which are not canal-bearing and form a series lateral to the nasals and dorsal to
the nostrils. Tectals are present in osteolepids (Eusthenopteron, Osteolepis), actinistians
(Rhabdoderma, Latimeria), porolepids (Holoptychius) and dipnoans (Dipnorhynchus,
Chirodipterus), but absent in actinopterygians. Thus I conclude that tectals are a synapomorphy
of sarcopterygians.

Parasphenoid and associated toothplates

The parasphenoid is without a stalk (posterior elongation below the otic region) in the Gogo
palaeoniscids and this is considered primitive for osteichthyans and placoderms.

Mimia toombsi

The parasphenoid is shown in lateral view in Fig. 13 and in ventral view in Fig. 50. It is a broad
bone, of roughly rectangular shape, with a pectinate anterior margin. A median tongue of bone
extends from this anterior margin to terminate between the vomers posterior to the paired
openings of the ventral nasobasal canals. Behind the basipterygoid process the parasphenoid is
produced into a short, posterolaterally-directed arm which passes up towards the oticosphenoid
fissure. Occasionally, as in BMNH P. 53247 (Fig. 50), one of these arms may project back below
and behind the spiracular canal, but this projection is not an ascending process in the strict sense
(Gardiner & Bartram 1977: 231). A short median posterior extension ends at the level of the
ventral fissure or occasionally a short distance in front of it. Posteriorly this extension is not
applied to the basisphenoid and in this respect it is similar to that of Australosomus (Nielsen
1949: fig. 26). The buccal surface is completely covered by teeth, apart from two notches which
delimit the posterolateral arms from the medial portion. The smooth areas around these notches
(osubc, Fig. 50) probably mark the points of insertion of the subcephalic muscles (Nelson 1970a:
468). On the dorsal surface at the level of the basipterygoid process the parasphenoid is
produced into a small cup around the wide bucco-hypophysial canal (see Moythomasia, bhc,
Fig. 52). The bucco-hypophysial canal passes through the centre of ossification. From the
ventral opening of the bucco-hypophysial canal paired spiracular grooves (Fig. 50; see also
Moythomasia, spig, Fig. 51) pass back towards the oticosphenoid fissure.

There is no dermal basipterygoid process (anterior ascending process) and in this respect
Mimia resembles Cheirolepis. The efferent pseudobranchial artery passed in between the
parasphenoid and the basisphenoid (fepsa, Fig. 50).

A paired toothplate is associated with the parasphenoid in Mimia. It is an elongate, ovoid
plate, rounded anteriorly but more pointed posteriorly, which fits between the edge of the
parasphenoid and the entopterygoid, overlapping the latter. This paired bone (Av, Fig. 53) is

272

B. G. GARDINER

apa

Psp

fepsa

bhc

fos

fhm

vfon

fotc

cao

Fig. 50 Mimia toombsi Gardiner & Bartram. Restoration of braincase in ventral view, based on
BMNH P. 56496 and P.53247. From Gardiner & Bartram (1977).

RELATIONSHIPS OF PALAEONISCIDS 273

shallowly concave dorsally and its entire oral surface is covered with closely-set, small teeth
similar to those on the parasphenoid. It covers the toothless zone between the lateral edge of the
parasphenoid and the adjoining part of the entopterygoid. It stretches from the foramen for the
efferent pseudobranchial artery to just behind the vomer (Vo, Fig. 50). I will call it an accessory
vomerine toothplate. An almost identical pair of toothplates occurs in Moythomasia and
Pteronisculus (Nielsen 1942: fig. 34, Vo); a narrower, smaller pair has been described in
Australosomus (Nielsen 1949: fig. 26). It is probable that these toothplates are widespread in the
palaeoniscids.

Moythomasia durgaringa

The parasphenoid (Figs 7, 51, 52) is more extensive than in Mimia. Anteriorly it reaches the
lateral margins of the basisphenoid and posteriorly not only extends to the level of the ventral
fissures but also bears rudimentary ascending processes (asp, Fig. 7). Its toothed area, however,
is less extensive than in Mimia and in front of the bucco-hypophysial canal only the central third
of the bone is toothed. Anteriorly the median tongue of bone is little longer than the paired
lateral ones and terminates at the level of the posterior margins of the vomers.

Behind the foramen for the efferent pseudobranchial artery the parasphenoid of one
specimen (BMNH P. 53221) continues out for a short distance onto the basipterygoid process on
one side only, but this can hardly be considered to be a true dermal basipterygoid process
(anterior ascending process). Posteriorly the posterolateral arm extends up onto the corner of
the prootic forming an ascending process (or posterior ascending process) which terminates
level with the bottom of the jugular canal. The ascending process, like the rest of the area behind
the basipterygoid processes, is covered with teeth which are never as large as those in front of the
bucco-hypophysial canal. The spiracular groove (spig, Fig. 51) which passes back on the oral
surface from the bucco-hypophysial canal extends onto the ascending process. Posteriorly the
parasphenoid is spear-shaped, with a shallow, smooth ledge forming the hindmost margin.
Between this spear-shaped portion and the ascending process is a distinct notch (gic, Fig. 51).
The outer border of this notch is smooth (osubc) and presumably served for the insertion of the
subcephalic muscle.

Immediately behind the parasphenoid in the roof of the mouth there is a pair of parotic
toothplates (Gardiner & Bartram 1977: 240). These plates fill the area between the back of the
parasphenoid and the aortic canal and underlie the ventral otic fissure. They are approximately
rectangular in outline, with a more rounded anterior margin where they fit onto the ledge at the
back of the parasphenoid. The left-hand plate has a narrow smooth ledge along its medial
margin where it is overlapped by its partner. Both plates are covered in tubercular teeth, similar
to those on the hind end of the parasphenoid. There is also a pair of accessory vomerine
toothplates identical to those described in Mimia.

Parasphenoid: summary and discussion

The simplest form of actinopterygian parasphenoid is that seen in Mimia and Cheirolepis
(Pearson & Westoll 1979), where it consists of a short, relatively broad toothed plate without a
posterior stem or stalk and with no basipterygoid process (anterior ascending process), and
in which the ascending process (posterior ascending process) does not extend across the
oticosphenoid fissure onto the lateral commissure. It is further characterized by the presence of
a spiracular groove on its buccal surface which terminates at or near the bucco-hypophysial
canal.

A similar type of parasphenoid (without posterior stem or basipterygoid process, and where
the ascending process does not reach the lateral commissure) is found in the rhipidistians
Eusthenopteron (Jarvik 1954: fig. 18) and Ectosteorhachis (Romer 1937: fig. 4), the actinisitians
Nesides (Stensio 19636: fig. 45), Wimania (Bjerring 1967: pi. 2B) and Latimeria (Millot &
Anthony 1958), the dipnoans Uranolophus (Denison 1968: fig. 8) and Dipnorhynchus
(Thomson & Campbell 1971: fig. 25), porolepids such as Glyptolepis and Holoptychius (Jarvik
1954: figs 19, 20), and the youngolepidids (Chang 1982). The structure called basipterygoid
process in porolepids by Miles (1977: 159) is little more than the edge of the notch made by the

274

B. G. GARDINER

pare

fepsa

spig

gic

podp

Fig. 51 Moythomasia durgaringa Gardiner & Bartram. Parasphenoid and associated structures in
ventral view, from BMNH P.53221. After Gardiner & Bartram (1977).

passage of the efferent pseudobranchial artery. The spiracular groove (seen in Moythomasia,
pqrplepids, Youngolepis and Eusthenopteron) is not present in actinistians or dipnoans, but its
4i$tfib;Ution in other osteichthyans and its presence on the neurocranium ofAcanthodes (Miles
1973a) suggests it is a primitive osteichthyan feature. Thus the parasphenoid of Moythomasia
and Cheirolepis must exemplify both the primitive actinopterygian and osteichthyan conditions.
The posterior elongation of the parasphenoid in post-Devonian actinopterygians has been

RELATIONSHIPS OF PALAEONISCIDS

275

pare

Fig. 52 Moythomasia durgaringa Gardiner &
Bartram. Parasphenoid in dorsal view, from
BMNH P.53217.

bhc

dealt with in detail by Gardiner (1973: 116), Patterson (1975: 527) and Miles (1977: 158), all of
whom agree that elongation is the result of differential growth in phylogeny and that such a
process occurred independently within the actinopterygians and dipnoans. In most post-
Devonian palaeoniscids and in pholidopleurids this posterior extension never crossed the
ventral otic fissure, but with the increase in size of the myodome the fissure came to lie further
posteriorly (Gardiner & Bartram 1977: fig. 8) and consequently a short rounded stem developed
behind the ascending process (Pteronisculus , Nielsen 1942; Kentuckia, Rayner 1951;
Boreosomus, Nielsen 1942; 'Ambodipia', Beltan 1968; Kansasiella, Poplin 1974; Coccolepis,
BMNH P. 50822; Australosomus , Nielsen 1949). In other actinopterygians the stem of the
parasphenoid is longer and extends across the fissure onto the ventral surface of the basioccipital
(Perleidus, Pachycormus, parasemionotids, pholidophorids, Patterson 1975: 528). In Poly-
pterus, living chondrosteans, Paleopsephurus, Errolichthys (Lehman 1952), Chondrosteus (RSM
1887.15.2), Saurichthys (Stensio 1925), Birgeria (Nielsen 1949), pycnodonts, semionotids, caturids,
Amia, Lepisosteus and most extant teleosts (Patterson 1975: 528) the parasphenoid floors the entire
basioccipital and terminates beneath the occipital condyle. Similarly in Bobasatrania (Nielsen 1952)
the parasphenoid extends back almost to the hind end of the neurocranium. The interrelationships of
these actinopterygians possessing long-stemmed parasphenoids imply that the condition has been
independently acquired on at least four occasions: in Polyp terns, in Birgeria, in Chondrosteus,
Paleopsephurus and living chondrosteans, and in neopterygians. A long stem has also been acquired
on perhaps two other occasions: once in dipnoans (no stem in the Lower Devonian
Uranolophus, Denison 1968) and possibly once within tetrapods (there is no stem in
Ichthyostega according to Jarvik 1952; 1955: fig. 8).

1 . Parabasal canal. Primitively in actinopterygians (and placoderms) neither the internal carotid
nor the efferent pseudobranchial arteries pierced the parasphenoid. Instead the internal carotid
artery entered the neurocranium behind the parasphenoid (in the short-stemmed forms) and
passed forward between the parasphenoid and the basisphenoid in a short parabasal canal
(Cheirolepis, Mimia, Moythomasia, Pteronisculus, Australosomus, etc.). The efferent pseudo-
branchial artery passed in above the parasphenoid and occasionally notched its lateral margin
(Mimia, Moythomasia). In Mimia and Moythomasia the parabasal canal commences just behind
the spiracular groove and a canal with precisely the same morphological relationships is also
found in Polypterus, in the porolepid Glyptolepis (Jarvik 1972: fig. 19, cd.a.pal) and in
Youngolepis (Chang 1982: fig. 7, c.a.ci). In Glyptolepis, however, the internal carotids passed

276 B. G. GARDINER

through the short stem of the parasphenoid into the parabasal canal through a foramen
interpreted by Jarvik (1972: fig. 31) as having transmitted a medial branch of the internal carotid
artery (see Gardiner 1973: 118). Anterior to the spiracular groove in Glyptolepis (prespiracular
groove of Jarvik 1954, 1972; Bjerring 1971; but see Patterson 1975: 534; Gardiner & Bartram
1977: 243) and Youngolepis (Chang 1982: fig. 8B) the parasphenoid is notched by the passage of
the efferent pseudobranchial artery (Gross 1936: 10A), though in some species of Glyptolepis
(Jarvik 1972: fig. 31) this artery passed through the edge of the bone. A parabasal canal is also
present in Neoceratodus (Holmgren & Stensio 1936: fig. 288), Griphognathus (Miles 1977: fig.
50, c.i.?; fig. 56) and tetrapods (Cryptobranchus), and in dipnoans the efferent pseudobranchial
artery enters the canal after passing in above the edge of the parasphenoid. A parabasal canal must
therefore have been primitively present in osteichthyans and its subsequent loss in living
chondrosteans, Saurichthys, Latimeria and Eusthenopteron (Jarvik 1954: fig. 6A, c.a.c.i) may be
correlated with the narrowness of the parasphenoid in each of these fishes. A parabasal canal is not
found in placoderms and is presumed to be a synapomorphy of osteichthyans.

2. Internal carotid artery. The backward growth of the parasphenoid in more advanced
actinopterygians had a marked effect on the course of the internal carotids and associated
arteries and at least four different topographies may be recognized (Gardiner 1973: 116).

In Polypterus the posterior stem of the parasphenoid is not only below the carotid arteries, but
also below the dorsal arterial system and this condition is unique. In Birgeria, Saurichthys,
Saurorhynchus, Errolichthys, Chondrosteus and Paleopsephurus a parabasal canal is missing
and the stem of the parasphenoid as well as the ascending processes must have been above the
carotid arteries as in Acipenser. In saurichthyids (Stensio 1925; Gardiner 1960: fig. 21) and
Chondrosteus (RSM 1887.15.2) there is a large, paired foramen in the parasphenoid beneath the
posterior margin of the ascending process; this presumably transmitted the orbital artery (as
originally suggested by Stensio) and not the common carotid as Gardiner (1973: ,116) and
Patterson (1975: 331) have supposed. In Lepisosteus andAmia the efferent pseudobranchial and
internal carotid arteries pass through notches in the lateral edges of the parasphenoid, into the
parabasal canal (as in Pteronisculus, Perleidus, parasemionotids, caturids and Lepidotes) and the
stem of the parasphenoid lies between the internal carotids. However, the internal carotid artery
in Boreosomus, unlike all other described palaeoniscids, passed through a foramen in the base of
the ascending process. This is a variation of the last condition. In Dapedium, pholidophorids,
leptolepids and other primitive teleosts (Patterson 1975: 532) there are foramina in the
parasphenoid for both the internal carotid and efferent pseudobranchial arteries and it appears
that in these fishes the notches seen in more primitive forms such as Perleidus, parasemionotids,
caturids and Lepidotes have become foramina, probably as a result of growth of the lateral edges
of the parasphenoid. Conditions are similar in the porolepid Glyptolepis (Jarvik 1972: fig. 31). In
advanced dipnoans the stem of the parasphenoid lies between the internal carotid artery and
above the efferent pseudobranchial artery, much as in Amia and Lepisosteus. A similar
relationship has been achieved in such tetrapods as Cryptobranchus, although here the carotid
artery passes through a foramen in the lateral edge of the parasphenoid.

3. Basipterygoid process. An endoskeletal basipterygoid process is present in chondrichthyans,
acanthodians and osteichthyans and is considered primitive for gnathostomes.

Primitively the osteichthyan skull possessed a well-developed endoskeletal basipterygoid
process (Mimia, Cheirolepis, Nesides, Glyptolepis, Eusthenopteron, Youngolepis} but only in
later, more advanced actinopterygians did it acquire support from the parasphenoid. Although
the lateral angle of the parasphenoid in dipnoans may be in a similar topographic position to the
actinopterygian dermal basipterygoid process, there is no reason to consider it homologous as
Miles (1977: 159) has done. A dermal basipterygoid process is an actinopterygian specialization,
possibly developed to reinforce the ventral wall of the myodome as originally suggested by
Rayner (1951: 81). In the more primitive palaeoniscids and chondrosteans a dermal
basipterygoid process is lacking (Cheirolepis, Mimia, Moythomasia, Polypterus, Acipenser,

RELATIONSHIPS OF PALAEONISCIDS 277

Polyodon, Chondrosteus, Saurichthys, Errolichthys, Birgeria), while in Polypterus, Birgeria and
Saurichthys the endoskeletal basipterygoid process has also been reduced or lost.

A dermal basipterygoid process is first observed in some of the more advanced palaeoniscids
and their relatives. Though small in Boreosomus, Platysomus and Perleidus, it is extensive in
Pteronisculus, Cosmoptychius and Kansasiella (Poplin 1974: fig. 8) where the endoskeletal and
dermal components are about equal. A similarly well developed dermal basipterygoid process,
with a stout endoskeletal component above, is also present in Dapedium (Patterson 1975: fig.
112) and Lepisosteus, but in some caturids the process is very small (Heterolepidotus, Catums
chirotes, Rayner 1948: fig. 7) and in others it is absent (Catums furcatus, ' Aspidorhynchus' ,
Macrepistius). The basipterygoid process (dermal and endoskeletal) has also been lost in
amioids, pachycormids and most Recent teleosts. By contrast in Pholidophorus bechei and
Lepidotes (Patterson 1975: 529) the dermal basipterygoid process is very large and the
endoskeletal part almost vestigial. In other pholidophorids and leptolepids the basipterygoid process
is entirely dermal; a similar massive dermal process is found in a few other teleosts such as
Diplomystus, osteoglossoids and ichthyodectids.

In summary, the endoskeletal basipterygoid process has been lost on several occasions; in
Polypterus, in Birgeria and Saurichthys, in Australosomus, in caturids and amiids, in
pachycormids and in later teleosts. The dermal basipterygoid process must also have been lost
on more than one occasion (amiids, pachycormids and several teleosts groups).

4. Ascending process. This is another important outgrowth of the parasphenoid in actino-
pterygians, developed primarily in relation to the spiracular diverticulum and which reinforces
the outer wall of the myodome. Such an ascending process (or posterior ascending process) is a
specialized feature developed within the group. Primitively in osteichthyans it was very short
and did not extend across the ventral otic fissure in Mimia and Cheirolepis, porolepids
(Glyptolepis, Gross 1936: pi. 8), youngolepidids (Chang 1982: fig. 7), osteolepids (Eustheno-
pteron, Jarvik 1954: fig. 18) and Devonian actinistians (Diplocercides , Bjerring 1972: fig. 3A).
Only in later actinopterygians and Polypterus is a long ascending process developed.

There has been considerable confusion over the identification of the ascending process.
Stensio (1925: 85) called the ascending process on the parasphenoid of sturgeons and
Saurichthys the 'processus ascendens posterior', but homologized the same process in Amia and
Lepisosteus with the dermal basipterygoid process of palaeoniscids and so called it the
'processus ascendens anterior'. The view that the ascending process in Amia is not comparable
to that in Acipenser received support from Pehrson (1940: 38) and Holmgren (1943: 39), but
Nielsen (1942: 106; 1949: 84) and Rayner (1951: 81) demonstrated that the two are strictly
homologous. Jarvik (1954: 51) considered the ascending process of the rhipidistian and
actinistian parasphenoid to be the homologue of the basipterygoid process of the parasphenoid
in Pteronisculus, referring to it as the anterior ascending process. But as Patterson (1975: 533)
and Gardiner & Bartram (1977: 243) have shown, it is homologous with the ascending process of
primitive actinopterygians.

A true ascending process (one which crosses the oticosphenoid fissure) is first met with in
Moythomasia and Kentuckia, where the slender process extends onto the lateral commissure but
terminates level with the bottom of the jugular canal. In more advanced palaeoniscids such as
Kansasiella, Pteronisculus, Boreosomus and in Australosomus the ascending process is more
extensive and its tip approaches or enters the lower opening of the spiracular canal. In Polypterus,
where the spiracle is unconstricted, the ascending process is large and very complicated (Jarvik 1954)
and may be assumed to have had a different history from that of Birgeria, Saurichthys and living
chondrosteans, where it covers a major part of the otic and orbitotemporal walls.

In more advanced actinopterans such as Perleidus, parasemionotids, semionotids and caturids
(Patterson 1975: 533) the ascending process, though as extensive as in most palaeoniscids, is less
stout and often more distinctly grooved by the spiracular diverticulum. The ascending process in
pholidophorids, leptolepids (Patterson 1975: 519), pachycormids (Patterson 1975: fig. 106) and
Ichthyokentema (Patterson 1975: fig. 150) on the other hand is shorter and terminates below the
spiracular canal when the latter is present. The ascending process in most Recent teleosts is also

278 B. G. GARDINER

short, but occasionally it is enlarged and meets the frontal as in Gasterosteus and Arapaima.
Patterson (1975: 534) has suggested that the ascending process may be reduced in height in
teleosts as a result of the reduction in the spiracular diverticulum. Whether or not this is true,
there is little doubt that the ascending process has developed on at least two occasions within
actinopterygians, once in Polypterus and once in the actinopterans.

The spiracular grooves on the ascending processes of Moythomasia are continued onto the
ventral surface of the parasphenoid (Fig. 51), where they join around the lower opening of the
bucco-hypophysial canal on a level with the efferent pseudobranchial foramina. The spiracular
grooves are linked by a transverse groove in Pholidophorus bechei and the Sinemurian
Leptolepis (Patterson 1975: figs 62, 143). A corresponding groove is found in the porolepids
Holoptychius (Gross 1936: fig. 10A), Glyptolepis (Jarvik 1972: fig. 31) and Porolepis (Jarvik
1972: pi. 9), and in Youngolepis (Chang 1982: fig. 8). Teeth occur in the spiracular groove of
Moythomasia (Fig. 51), some specimens of Pteronisculus, Perleidus (Patterson 1975: fig. 115)
and on the ventral plate of the ascending process in the mouth of the spiracular cleft in
Polypterus. Teeth are also present in the transverse groove in Moythomasia, Glyptolepis and
Youngolepis. Some confusion has resulted from the misidentification of this transverse groove
in rhipidistians, where it has been designated 'prespiracular' by Jarvik (1954, 1972) and Bjerring
(1971, 1977), but as Patterson (1975: 534) and Gardiner & Bartram (1977: 243) have
demonstrated, the groove is homologous with that in actinopterygians and is therefore a
spiracular groove. Although it is absent in dipnoans and actinistians, the occurrence of this
groove in actinopterygians, porolepiforms and youngolepidids suggests that it is a primitive
osteichthyan character.

A similar transverse groove is present on the parasphenoid of placoderms (Kulkzycki 1956:
pi. 1G; Miles & Westoll 1968: fig. 18a; White & Toombs 1972: figs 5, 6, 7; Young 1979: pi. 5).
This has been variously interpreted as having housed blood vessels (Kulkzycki 1956: 107; Stensio
1963a: 122) or outgrowths of the hypophysial stalk (White & Toombs 1972: 388). Nevertheless, it
bears the same relationship to the bucco-hypophysial canal as in many osteichthyans and the
presence of teeth along its borders (White & Toombs 1972: fig. 5) suggests it is a spiracular groove.

5. Parasphenoid teeth. The entire oral surface of the parasphenoid is toothed in Mimia (Fig. 50)
and Cheirolepis. In Moythomasia, however, although the posterior region is completely
toothed, anterior to the lower opening of the bucco-hypophysial canal the teeth are confined to a
narrow, central band (Fig. 51). In Polypterus teeth occupy the full width of the parasphenoid
anteriorly, but beneath the orbit they are restricted to a narrow band similar in extent to that in
Moythomasia. Posteriorly this band separates into two curved bands of teeth which run along
the anterior edges of the ventral plate of the ascending process (Allis 1922: fig. 9). In most other
palaeoniscids, parasemionotids and the majority of caturids the parasphenoid is toothed from
the level of the ascending process forwards. Reduction and loss of parasphenoid teeth has
occurred on several occasions within the actinopterygians. Thus the teeth are reduced to a small
tooth patch around the lower opening of the bucco-hypophysial canal in Saurichthys and are
completely wanting in Birgeria, Chondrosteus and living chondrosteans. Parasphenoid teeth are
also missing in Australosomus, Bobasatrania, pycnodonts, Macrepistius, Ichthyokentema,
Catervariolus and most teleosts (Patterson 1975: 529).

In Devonian and Carboniferous actinistians and in Eusthenopteron the parasphenoid has
a tooth-bearing median ridge which extends back to enclose the lower opening of the
bucco-hypophysial canal. In later actinistians and in some osteolepids (Ectosteorhachis,
Megalichthys) the tooth-bearing ridge is confined to the area in front of the bucco-hypophysial
canal. In Youngolepis (Chang 1982: fig. 8) the parasphenoid is completely toothed but in
porolepids such as Glyptolepis (Jarvik 1972: fig. 31) the tooth patch is restricted to the area
around the opening of the bucco-hypophysial canal and, as in Elops, there are in addition
numerous small toothplates in the roof of the oral cavity. In Porolepis (Jarvik 1972: fig. 65),
however, the teeth are reduced to a median row of some eight large teeth borne on a raised area;
parallel enlargement and specialization of parasphenoid teeth is seen in various teleosts (e.g.
albulids, hiodontids, osteoglossids).

RELATIONSHIPS OF PALAEONISCIDS 279

In the Lower Devonian dipnoan Uranolophus (Denison 1968: fig. 8) the buccal surface of the
parasphenoid is toothed except posteriorly where there is a small area of smooth bone, whereas
in the other Lower Devonian form (Dipnorhynchus} the buccal surface has a continuous dentine
covering (Campbell & Barwick 1982). In many dipnoans with a stalked parasphenoid the buccal
surface is covered with teeth whereas the posterior stem is smooth (Griphognathus,
Holodipterus Miles 1977: 151, 154). In Dipterus valenciennesi (White 1965) the bucco-
hypophysial duct opens in the anterior portion of this toothed area. In Conchopoma the toothed
anterior portion is expanded and opposed to the median basibranchial toothplate (Denison
1969: 199), but in Chirodiptems (Miles 1977: 153) the parasphenoid teeth have been entirely
lost.

In placoderms the parasphenoid may be completely covered by numerous small teeth as in
Buchanosteus (White & Toombs 1972: fig. 5; Young 1979: pi. 5) and Kujdanowiaspis (Jarvik
1954: fig. 34).

Primitively, therefore, in osteichthyans and placoderms the oral surface of the parasphenoid
was toothed.

6. Bucco-hypophysial canal. The bucco-hypophysial duct opens through the parasphenoid into the
roof of the mouth in primitive actinopterygians (Mimia, Moythomasia, Cheirolepis, Polypterus,
Saurichthys, Chondrosteus, Perleidus), and this opening is retained in more advanced forms such as
'Aspidorhynchus', Dapedium, Lepidotes, Caturus, Heterolepidotus and Elops (Patterson 1975: 530).

An open bucco-hypophysial duct is also characteristic of most actinistians (Nesides,
Latimeria), all described porolepids (Jarvik 1972), and Youngolepis (Chang 1982),
Eusthenopteron (Jarvik 1954), Dipterus and Ichthyostega (Save-Soderbergh 1932). In
Glyptolepis (Jarvik 1972: fig. 19D) the opening is paired.

The parasphenoid also contains an open bucco-hypophysial canal is placoderms. White &
Toombs (1972: fig. 6) figured a paired opening for the bucco-hypophysial canal in
Buchanosteus, but Young (1979: fig. 17; pi. 5) has shown, in better preserved material, that the
opening, although bilobed, is single.

The striking similarity between the placoderm and osteichthyan parasphenoids (teeth,
spiracular grooves, and foramen for bucco-hypophysial canal) suggests that a parasphenoid is a
synapomorphy of a group containing osteichthyans and placoderms.

7. Subcephalic muscles. Nelson (1970a: 468) has suggested that the subcephalic muscles of
Latimeria, which extend beneath the intracranial joint from the otico-occipital to the parasphenoid,
may be derived from one or more of the anterior body myomeres such as occur in
Polypterus. In Polypterus the body musculature extends forwards beneath the occipital and otic
regions, to be inserted on the parasphenoid immediately behind the ascending process. It thus
spans the ventral otic fissure which lies above the parasphenoid. The area of muscle insertion lies
between the posterior margin of the ascending process and the stem of the parasphenoid.

A distinct notch is present in the posterior margin of the parasphenoid in Mimia and
Moythomasia, in the latter immediately behind the ascending process, and below and anterior to
the articulation for the first infrapharyngobranchial; it may be inferred that subcephalic muscles
of the type found in Polypterus were inserted in this region (osubc, Figs 50, 51). This notch may
also be identified by the same topographic and morphological criteria as have been used to
recognize the insertion of the subcephalic muscles in rhipidistians and actinistians (Bjerring
1967, 1971; Jarvik 1966, 1972).

Bjerring (1971: fig. 6), however, taking Latimeria as his model, has restored subcephalic
muscles in Pteronisculus inserting on the hind end of the parasphenoid, but originating on the
underside of the basioccipital in the triangular area circumscribed by the grooves for the lateral
dorsal aortae (and thereby spanning the ventral otic fissure as in Latimeria). Patterson (1975:
538) suggested that although such an area on the basioccipital could well have served for muscle
attachment it does not follow that the muscles were directed forwards and proposed that it was
more likely that it served for the attachment of anterior trunk muscles, serially homologous with
those attaching to the back of the parasphenoid in Polypterus. In my opinion this triangular area

280 B. G. GARDINER

of the basioccipital (primitive for actinopterygians; Patterson 1973: 558) never served for muscle
attachment in actinopterygians. It lies above the general level of the floor of the occipital region
in Mimia (Fig. 13), Moythomasia (Fig. 7) and Kentuckia (Rayner 1951: fig. 9); it never shows
any sign of muscle scars, is frequently fenestrated (Figs 14, 15, 50) and in Moythomasia (Fig. 51)
the paired parotic toothplates neatly fill in the area between the back of the parasphenoid and
the aortic canal. Furthermore this triangular area, although domed in Mimia and Moythomasia,
is depressed and smooth in Pteronisculus (Nielsen 1942: fig. 6) and Kentuckia (Rayner 1951: fig.
9). Thus it appears that a subcephalic muscle of rhipidistian type originating on the underside of
the occipital ossification did not exist in actinopterygians, and I suggest that the foremost trunk
myomere inserted on the posterior part of the parasphenoid only, as in Polypterus, and not on
the underside of the occipital ossification.

Distinct pits in a similar position to the notches in Mimia and Moythomasia occur on the
underside of the parasphenoid of Cosmoptychius (Schaeffer 1971: fig. 8A) and Coccolepis
(BMNH P. 50822), immediately behind the bucco-hypophysial canal and medial to the internal
carotid foramen. There is a pair of pits in this position in several more advanced actinopterans
including Heterolepidotus, Caturus chirotes (Gardiner 1960: fig. 36, eff. pseud.), Lepidotes
latifrons, Pholidophorus bechei and Ichthyokentema (Patterson 1975: 535; figs 62, 150). In other
pholidophorids, leptolepids and pachycormids the pits are more posteriorly placed. Thus I
conclude that subcephalic muscles of the type found in Polypterus were primitively present in
actinopterygians and inserted on the posterior part of the parasphenoid. Identically-placed pits
to those seen in Cosmoptychius and Pholidophorus occur in the osteolepids Megalichthys,
Ectosteorhachis (Romer 1937: fig. 4, ci), Eusthenodon (Bjerring 1967: pi. 2C) and Eus-
thenopteron (Bjerring 1967: pi. 2 A), but there is no reason to assume that the muscles inserting
in these pits were more like those of Latimeria than of Polypterus, particularly since the area of
muscle insertion in Latimeria is far in advance of the bucco-hypophysial canal.

In other actinopterygians such as Perleidus and parasemionotids there is a much wider,
irregular recess (Patterson 1975: 536; figs 98, 116) for the subcephalic muscles; this condition is
paralleled by the porolepids Glyptolepis and Porolepis (Jarvik 1972: 86) and by Youngolepis
(Chang 1982: figs 7, 8), where there is a similar recess behind the spiracular groove, implying a
broad insertion. This has prompted Patterson (1975: 538) to postulate that a broad insertion may
be the primitive condition.

8. Accessory toothplates. In Mimia (Av, Fig. 53), Moythomasia, Elonichthys (Watson 1925: fig.
22, D.Pti), Pteronisculus (Nielsen 1942: fig. 34) and Australosomus (Nielsen 1949: fig. 26) there
is a large toothplate between the entopterygoid and parasphenoid (see p. 271), which I have
termed the 'accessory vomerine toothplate'. The occurrence of such a toothplate is considered
primitive for actinopterygians.

In chondrichthyans there is a shagreen of small denticles in the skin of the roof of the mouth
(Nelson 19706: 2) and in the porolepid Glyptolepis (Jarvik 1972: figs 8C, D, 16, 22, 30) there are
numerous small dental plates. Similarly in Elops (Nybelin 1968) there is a small patch of
toothplates lying free in the mucous membrane in the region of the bucco-hypophysial canal.
Thus primitively in gnathostomes there must have been numerous small dental plates in the skin
lining the roof of the mouth. In early actinopterygians some of these are presumed to have been
replaced by a single accessory vomerine toothplate.

Paired toothplates also occur in the roof of the mouth immediately behind the short
parasphenoid of primitive osteichthyans. These paired parotic plates are found in Moythomasia
(Fig. 51) and Eusthenopteron (Jarvik 1954: fig. 22).

Palatoquadrate and dermal bones of the cheek

Mimia toombsi

In Mimia the palatoquadrate is very long; anteriorly it articulates with the lateral ethmoid (apal,
Fig. 50) while its posterior part reaches beyond the occiput (Fig. 55). In the majority of
specimens the palatoquadrate is ossified throughout as one bone, as in the larger specimens of

RELATIONSHIPS OF PALAEONISCIDS

281

282

B. G. GARDINER

Pteronisculus and Australosomus (Nielsen 1942: 143; 1949: 99) and as in specimens of
Eusthenopteron and Glyptolepis (Jarvik 1954: 27; 1972: 70). In all these specimens it is
impossible to detect ossification centres. Two smaller specimens of Mimia, however, show three
separate ossifications: one in the quadrate region (Fig. 57), one in the metapterygoid region and
one in the autopalatine region; these three ossifications were separated in life by a large area of
cartilage. In one or two other specimens the junctions between these bones may be inferred.

The quadrate appears to have been the most prominent ossification in the palatoquadrate
cartilage, with its centre of ossification in the condyle region. The quadrate forms the inner
margin of the adductor mandibulae fossa, and reaches anterodorsally to the level of the hole for
the basipterygoid process. Here it meets the much smaller autopalatine ossification, the
junction being marked by a prominent ridge. The centre of ossification of the metapterygoid lies
dorsally around the fossa for the levator arcus palatini muscle and ventrally this bone extends to
just below the spiracular groove (spig, Fig. 56). Anteriorly the metapterygoid forms the
posterior margin of the hole for the basipterygoid process, the anterior margin being formed by
the autopalatine. Thus the quadrate is the largest endoskeletal ossification, the metapterygoid is
somewhat smaller and the autopalatine considerably less extensive. In this respect Mimia
resembles Acanthodes (Miles 1965: fig. 1).

The palatoquadrate ossification is roughly triangular in its outline with a circular hole (hbpt,
Fig. 56) in the anterodorsal margin. During life the basipterygoid process slid through this hole
which functioned as a guide during lateral movements of the palatoquadrate. A similar hole has
been described in the palate of certain species of Pteronisculus and Boreosomus by Lehman
(1952: figs 35, 57) and in the palate of Kentuckia by Rayner (1951: fig. 3), but unlike
Pteronisculus (Nielsen 1942: fig. 35) and Kentuckia there is no additional process on the palate in
this region.

The posterior division of the palatoquadrate forms a high, nearly vertical plate which is curved
dorsally so that its anterolateral face is concave in a dorsoventral and a rostrocaudal direction
and its medial face is correspondingly convex. The concavity in the anterior face of the

Ecpt

Enpt

Fig. 54 Mimia toombsi Gardiner & Bartram. Dermal bones of the right palate in medial view. (A),
from BMNH P. 56490; (B), from BMNH P.56486.

RELATIONSHIPS OF PALAEONISCIDS 283

metapterygoid (lapf, Fig. 56) presumably served for the insertion of the levator arcus palatini
muscle. A similar concavity is recognizable in Pteronisculus and Australosomus (Nielsen 1942:
144; 1949: 102). The whole of the posterior margin of the palatoquadrate is in contact with the
medial face of the preopercular and ventrally with the medial face of the quadrate jugal (Fig. 57).
This junction is often so complete that the preopercular and quadratojugal are immovably fixed
to the palatoquadrate (Fig. 60). Ventrally the palatoquadrate forms the medial boundary of the
opening for the adductor mandibulae, but anterior to this fossa it is in contact with the medial
face of the ventral part of the maxilla. As with the preopercular, the junction is often so complete
that the maxilla is fused to the palatoquadrate. This junction is further strengthened by fusion of
the dermopalatines with both the overlying palatoquadrate and the maxilla (Figs 56, 60).

In the metapterygoid region there is a gradual transition between the dorsal, laterally bent
part and the vertical or quadrate part of the palatoquadrate ossification, but there is no distinct
angle between these two regions as there is in Pteronisculus (Nielsen 1942: 145). On the medial
side of the metapterygoid region there is a conspicuous groove running back from the hole for
the basipterygoid process down onto the quadrate (spig, Figs 56, 57). A similar groove has been
figured in Pteronisculus (Nielsen 1942: figs 35, 36). Gardiner (1973: fig. 8, icg) originally
attributed this groove to the internal carotid artery, but from its position just dorsal to the
tooth-bearing sheet of dermal bone it is more probable that it represents the lateral surface of the
ventral end of the spiracular diverticulum. A similar groove on the parasphenoid (spig, Fig. 50)
is apparently a continuation of the groove on the metapterygoid. In Porolepis and
Eusthenopteron (Jarvik 1954: fig. 16, resh) there is a similar groove on the palate which has been
called the spiraculo-hyomandibular recess by Jarvik, and as in Mimia and Pteronisculus this
groove is in the same position as the posterior division of the spiracular slit in Polypterus.

The posterodorsally-facing margin of the palatoquadrate ossification is pierced by several
short transverse canals. In the metapterygoid and in the dorsal region of the quadrate there are
two such canals (fmand.int. VII, Figs 55, 56, 57), while ventrally in the quadrate there is either a
single canal or a pair of canals immediately above the condyles (fmand.int. VII, Fig. 57). These
canals presumably transmitted the internal mandibular branch of the facial nerve as in
Polypterus (Allis 1922: 282), the dorsal pair of foramina serving for the entrance of that nerve
and the ventral pair for its exit. A single short transverse canal and groove has also been
described in Pygopterus (Aldinger 1937: 145) and Pteronisculus (Nielsen 1942: 145). Lateral to
the posterior end of the spiracular groove and immediately above these transverse canals lay the
hyomandibula with the interhyal beneath it (Hy, Ih, Fig. 55); ventrally the quadrate articulated
with the lower jaw by a double-headed joint, the facets or condyles of which lie lateral to one
another as in Pteronisculus (Nielsen 1942: fig. 33) and Eusthenopteron (Jarvik 1954: fig. 25).

That part of the palatoquadrate ossification in front of the point of junction with the
basipterygoid process is mostly formed by the autopalatine. This forms a thin plate of bone,
concave dorsolaterally and convex ventromedially. Posteriorly the limit of the autopalatine is
marked by a distinct ridge behind which there is a somewhat deeper concavity in the lateral
surface. This concavity immediately in front of the hole for the basipterygoid process marks the
most anterior insertion point of the adductor mandibulae muscle. Anteriorly the autopalatine
turns inwards to articulate with the lateral ethmoid by a cartilaginous interface.

Like the cartilage bones, in most specimens the dermal toothplates on the medial surface of
the palatoquadrate are ossified throughout as one bone, and closely resemble those of
Eusthenopteron (Jarvik 1954: fig. 16), except that in the latter there are separate
dermopalatines. In two specimens, however, individual bones are apparent: in one there are
nine and in the other ten (Fig. 53), four dermometapterygoids, an entopterygoid, an
ectopterygoid and three or four dermopalatines.

The entopterygoid (Enpt, Figs 53, 54) is the largest of the dermal bones on the oral face of the
palatoquadrate. It is approximately triangular in outline and tapers to a point anteriorly where it
rests against the anterior end of the autopalatine. Like the pterygoid of Eusthenopteron and
Glyptolepis (Jarvik 1972: fig. 25), in the adult it fuses indistinguishably with the underlying
autopalatine and metapterygoid. However, neither the entopterygoid nor the underlying
dermopalatine reach the anterior limit of the autopalatine. The entopterygoid is broadest

284

CL
i/l

3 ffl

O ^
>. ed

« 3

03 ty

cfl O

c *
'co CQ

CQ £

. O

CQ .g

•5 <u

CO '>

5 ^H

-S

RELATIONSHIPS OF PALAEONISCIDS

285

S
pa

•o

OJ
M
•5

cr
o

JS
d.
*
^

286

B. G. GARDINER

posteriorly where it interdigitates with the ectopterygoid and the anteriormost dermometa-
pterygoid. Medially it is covered by closely-set, small teeth, similar to those on the
parasphenoid, except for a shelf-like margin of thin bone dorsally. Posteriorly this untoothed
margin terminates in a small process (Fig. 53) which fits beneath the anterior margin of the hole
for the basipterygoid process. A similar anterior process on the anteriormost dermometaptery-
goid forms the hind margin of this hole. The accessory vomerine toothplate (Av, Figs 53, 54) fits
loosely on this dorsal margin of the entopterygoid, spanning the gap between it and the
parasphenoid. The radiation centre of the entopterygoid lies near the dorsomedial margin, but
more posteriorly than in Pteronisculus (Nielsen 1942: fig. 37). Ventrally the entopterygoid
adjoins the dermopalatines and covers a marginal zone of the oral face of these bones, but leaves
a well-marked groove (gr, Figs 53, 54) between it and the dermopalatine tooth row.

There are four dermometapterygoids (Dmpt, Figs 53, 54) which rest against the medial face of
the metapterygoid and quadrate, and in adult fish fuse indistinguishably with them and one
another and with the ectopterygoid. The two anterior dermometapterygoids lie well below the
spiracular groove, but the two posterior bones form the lateral edge of the spiracular gill slit. All
four bones interdigitate with one another; the anteriormost also interdigitates with the
entopterygoid while the three more posterior elements suture with the ectopterygoid ventrally.
All three posterior bones are completely covered by small, closely set teeth.

The ectopterygoid (Ecpt, Figs 53, 54) covers the postero ventral part of the oral face of the
palatoquadrate. Caudally the bone extends almost as far as the posterior dermometapterygoid,
but rostrally it terminates before the basipterygoid articulation and the anteriormost
dermometapterygoid. Posteriorly the ectopterygoid lies against the vertical oral face of that part
of the quadrate which bounds the adductor fossa. Immediately in front of the adductor opening
the ventral part of the ectopterygoid is turned under and outwards to form an almost horizontal
external lamina. This lamina sutures with the maxilla, overlapping a narrow marginal zone of the

spig

Ecpt

vpl

Pop

epopc
fmand. int VI

2mm

Fig. 57 Mimia toombsi Gardiner & Bartram. Quadrate region of palatoquadrate and associated
dermal bones in posterior (left) and lateral views, from BMNH P. 53254.

RELATIONSHIPS OF PALAEONISCIDS 287

internal horizontal lamina. Anteriorly the ectopterygoid sutures with the posterior
dermopalatine and the entopterygoid. The dorsal margin is sutured to the three posterior
dermometapterygoids while the anterodorsal corner of the ectopterygoid just reaches the
anteriormost dermometapterygoid. The entire oral face of the bone is covered by closely set,
pointed teeth, apart from a narrow groove anteriorly (gr, Fig. 53). This groove is continuous
with a similar groove in the dorsal margin of the dermopalatines (Fig. 54). The centre of
ossification lies at the postero ventral corner, in front of the adductor opening where the
ectopterygoid contacts the maxilla.

The dermopalatines form a series of three or four interdigitating bones which provide an
anterior continuation of the horizontal external lamina of the ectopterygoid. The lateral margins
of the dermopalatines overlap a narrow marginal zone of the internal horizontal lamina of the
maxilla and their medial margins are overlapped orally by the entopterygoid. Laterally the
dermopalatines are covered by small, closely set pointed teeth, medial to which is a series of
much larger pointed teeth. The internal lamina beyond the tooth row is devoid of teeth and
forms a well-marked groove (gr, Fig. 54) between the toothed area and the overlapping
entopterygoid. The centre of ossification of each dermopalatine lies more or less at the middle of
the bone.

The maxilla is of the usual palaeoniscid type (also seen in onychodonts, lessen 1966) in which
there is a narrow suborbital part and a high posterior expansion (Fig. 60). The suborbital portion
is longer than in Cheirolepis, Moythomasia and Pteronisculus . The dorsal and posterior margins
of the posterior expansion overlap a considerable portion of the lower margin of the
quadratojugal (Fig. 63) so that only a small part of the quadratojugal is visible. The suborbital
extension of the maxilla is overlapped by the postero ventral margin of the jugal and by the
lachrymal. A horizontal longitudinal lamina stretches along the medial, ventral margin of the
maxilla from the anterior margin of the adductor fossa to the anterior limit of the bone. The
lamina increases in breadth as it passes anteriorly but narrows again from the level of the
basipterygoid process anteriorly. The maxilla is connected by its horizontal lamina with the
ectopterygoid and dermopalatines, as in Pteronisculus (Nielsen 1942: fig. 35). The ventral edge
of the palatoquadrate sits above this horizontal, longitudinal lamina and in the quadrate region
is produced dorsally into a flattened flange (Impt, Fig. 55) which is attached to the inner surface
of the maxilla. This quadrate flange stretches from the adductor fossa to the posterior limit of the
autopalatine and is intimately connected with the maxilla. Ventrally the free margin of the
maxilla bears teeth; the largest teeth occur at the anterior end of the posterior expansion and the
smallest posteriorly. An outer series of much smaller teeth grades almost imperceptibly into the
surface ornamentation. The radiation centre of the maxilla lies above the horizontal lamina and
just anterior to the expanded posterior region (see Fig. 66, Moythomasia).

The preopercular is a long, acutely angled bone which extends forwards above the posterior
expansion of the maxilla and carries the preopercular canal which is directed towards the otic
portion of the infraorbital canal (temporal canal). Its anteroventral margin overlaps the
quadratojugal (Figs 60, 61, 63) to such an extent that this latter bone is difficult to recognize in
lateral view. The preopercular canal does not run the whole length of the bone as it does in
Pteronisculus (Nielsen 1942: fig. 27), but instead exits dorsally before reaching the anterodorsal
margin (epopc, Figs 56, 63). The canal runs parallel to the posterodorsal margin, piercing the
radiation centre. From close to the radiation centre the horizontal pit-line passes anteriorly
towards the anterior margin of the preopercular. Internally, it is marked by a small but distinct
branch of the preopercular canal (bpopc, Fig. 63). In some specimens a short vertical pit-line
(vpl, Fig. 61) joins the horizontal pit-line posteriorly, but in others the two lines are separate
(Fig. 62). The preopercular is thickened internally along the route of the preopercular canal and
it is this region which is in intimate contact with the palatoquadrate. Both the metapterygoid and
quadrate areas of the palatoquadrate contribute to this dorsoposterior flange (Fig. 55) which
fuses with the inner surfaces of the preopercular and quadratojugal. This intimate contact of the
palatoquadrate with the preopercular and quadratojugal dorsally and the maxilla ventrally,
together with the strongly overlapping sutures between the maxilla, preopercular and
quadratojugal, produces a rigid cheek unit (Gardiner 1967).

288

289

290

B. G. GARDINER

Z
S

c
o
.0

73
OJ

•a

J2
3

RELATIONSHIPS OF PALAEONISCIDS

291

The quadrate jugal is overlapped dorsally by the preopercular and anteriorly by the maxilla
and only a small portion of its surface is ornamented (Fig. 63). The radiation centre lies
immediately anterior to the exposed, ornamented area beneath the pit-line. The quadratojugal
pit-line, although short, is about the same length as the vertical pit-line and has at least three
nerve foramina serving it.

The infraorbital series comprises three bones, the dermosphenotic, jugal and lachrymal. The
dermosphenotic is considered here with the other cheek bones because it is in series with the
infraorbital bones and is loosely attached or hinged to the skull roof.

The dermosphenotic is triangular in shape, sits over the top of the postorbital process and
forms the anterolateral margin of the spiracular opening. The radiation centre is in the posterior
third of the bone near its dorsal margin. The centre is pierced by the infraorbital canal which
traverses it in a vertical direction. The infraorbital canal passes into the intertemporal anterior to
the spiracular opening. A branch of the infraorbital canal passes anteriorly within the
dermosphenotic to terminate blindly before reaching the anterior end of the bone. Ventrally the
dermosphenotic is overlapped to a small degree by the anterodorsal margin of the jugal;
anteriorly it may just make contact with the nasal and dorsally it contacts the frontal and
intertemporal, resting on a ledge formed by the neurocranium. In some specimens it may also
meet the anterior end of the supratemporal.

The jugal is the largest element of the infraorbital series. It has a convex posteroventral
margin which overlaps the concave anterodorsal margins of the preopercular and maxilla. The
course of the infraorbital canal changes from nearly vertical to nearly horizontal within the
bone, the change of direction marking the radiation centre.

epopc

pope

qujpl

Fig. 61 Mimia toombsi Gardiner & Bartram. Sketch restoration of left preopercular and
quadratojugal in lateral view, to show pit-lines.

292

B. G. GARDINER

pope

hpl

epopc

Fig. 62 Mimia toombsi Gardiner & Bartram. Left preopercular in lateral view, from BMNH

P.56473.

The lachrymal is a short, thin ossification tapering to a point anteriorly. Posteriorly it overlaps
the jugal while anteriorly it is overlapped by the premaxilla. The infraorbital sensory canals runs
the length of the bone and the centre of radiation lies near the posterior margin. The external
openings of the sensory canal lie along its ventral margin.

Moythomasia durgaringa

The palatoquadrate in Moythomasia differs in proportions from that of Mimia. It has a much
longer post-basipterygoid portion and a correspondingly shorter and stouter anterior section. In
shape it more nearly resembles that of Cheirolepis (Pearson & Westoll 1979: figs 7, 8) than any
other described palaeoniscid. In all specimens the palatoquadrate is a single ossification and it is
difficult to detect individual ossification centres. The anterior end, which turns inwards to
articulate with the lateral ethmoid, is much broader than in Mimia, as is the facet on the lateral
ethmoid. The whole of the ventral margin in front of the adductor fossa is produced laterally into
a flattened flange (Fig. 59) which is intimately attached to the inner surface of the maxilla above
the horizontal longitudinal lamina (hll, Fig. 67). Anteriorly this flange is pierced by one or more
nerve foramina (frmx, Fig. 59) for branches of the maxillary nerve. As in Mimia there is a
circular hole in the anterior dorsal margin (hbpt, Fig. 59) to accommodate the basipterygoid
process. Beneath the hole on the medial surface of the palatoquadrate are several
dorsally-directed pits (iepl, Fig. 58). By analogy with Polypterus these are presumed to have
been insertion points for the ethmopalatine ligament (Allis 1922: 244). The posterolateral face
of the palatoquadrate forms a broad flange which is in intimate contact with the preopercular
and quadratojugal. This flange is pierced in its ventral half by four foramina. The dorsalmost of
these foramina served for the entrance of the internal mandibular branch of the facial nerve and
the ventralmost for its exit.

Like the cartilage bones, the dermal toothplates on the medial surface are ossified throughout
as one bone. The toothplate is more extensive than in Mimia and forms much of the margin to
the spiracular groove.

Laterally the dermopalatines are covered by small rounded teeth similar to those on the
ventral ectopterygoid region. Medial to these is a series of much larger, pointed teeth, which
show a similar replacement sequence to those on the maxilla and dentary . A well-marked groove
separates this tooth row from the entopterygoid.

The maxilla has a shorter but much stouter postorbital portion than in Mimia. The teeth
likewise are much stouter and less needle-like. At the centre of radiation of the maxilla several

RELATIONSHIPS OF PALAEONISCIDS

epopc-

293

qujpl

imm

bhm

Fig. 63

Mimia toombsi Gardiner & Bartram. Left preopercular and quadratojugal in lateral (left)
and medial views, from BMNH P. 56484.

nerve foramina pass into the medial surface (bhm, Fig. 67), whereas externally a corresponding
series of pits presumably housed the neuromasts of the anterior part of the horizontal pit-line
(hpl, Fig. 66). An anterior continuation of the horizontal pit-line is also found on the maxilla of
Polypterus (Jarvik 1947: fig. 1) and Pteronisculus (Nielsen 1942: pi. 9, fig. 1).

The preopercular is not so expanded dorsally as that of Mimia and the preopercular canal
exits two-thirds of the way along the posterodorsal margin (epopc, Fig. 58).

The quadratojugal is stout and triangular and the pit-line makes a long slanting groove on its
surface. Medially three nerve foramina transmitted fine branches of the mandibular nerve to the
line.

The dermosphenotic is less extensive both anteriorly and ventrally than in Mimia. Anteriorly
it tapers to a point and scarcely contacts the nasal. Posteroventrally it is overlapped by the jugal.

The jugal is more strongly convex posteriorly than in Mimia and the infraorbital canal opens
by two sets of pores ventrally rather than a dorsal suite of pores as in Mimia.

The lachrymal is very different in shape from that of Mimia; much broader, and bifurcated
anteriorly at the point of the exit of the infraorbital canal. Posteriorly the lachrymal overlaps the
jugal, but anteriorly it overlaps the premaxilla. Anteroventrally the ornament closely resembles
that on the ventral edge of the maxilla. A similar ornamentation is found on the posteroventral
margin of the premaxilla and rostral. The infraorbital sensory canal enters and leaves the
lachrymal through dorsally-directed pores (inc, Fig. 74). Beneath the canal several pores pass
right through the bone (p, Fig. 74), as they do through the premaxilla.

Palatoquadrate: summary and discussion

1. Palatoquadrate commissure and vomer. In the ontogenetic development of actinopterygians
and dipnoans the anterior ends of the palatoquadrates are joined by a blastema both to one
another and to the overlying trabeculae (Holmgren 1943, Bertmar 1966). This blastema

294

B. G. GARDINER

E
•D

Ofl

"o"

8. .

O (N

« O

§ CQ

1 I
c«fc

'•B

RELATIONSHIPS OF PALAEONISCIDS 295

2mm

Fig. 65 Mimia toombsi Gardiner & Bartram. Right maxilla in medial view, from BMNH P. 56486.

(intermediating body or symphysial portion) then chondrifies to form the palatoquadrate
commissure and is incorporated in the floor of the ethmoid region.

In Amia, Lepisosteus and teleosts (Holmgren 1943, Bertmar 1959) the primary
palatoquadrate commissure chondrifies as a single unit which later fuses into the ethmoid plate.
In Acipenser (Holmgren 1943: figs 26, 28) this primary commissure is possibly represented by
the so-called tentacle blastemas.

In selachians (Holmgren 1940), although there is always an early, mesenchymatic, frontal
connection between the palatoquadrates and the trabeculae, there is never an intermediating
body or symphysial portion joining the palatoquadrates to the floor of the ethmoid region.
Instead, somewhat later in development, after the formation of the basal (orbital) processes, the
anterior ends of the palatoquadrates grow forwards and inwards to meet in the mid-line, forming
a symphysis (see under anterior articulation, p. 297) which ventrally supports teeth. This
symphysis is found in all extant selachians. A similar shark-like palatoquadrate commissure is
seen in Acipenser (where it must be considered to be secondary, not primitive; see below under
anterior articulation), and in the acanthodian Ptomacanthus (Miles I973b: pi. 6) and possibly in
the placoderm Jagorina (Stensio 1969: 71).

In osteichthyans the vomer develops beneath the primary palatoquadrate commissure
and therefore lies in sequence with the dermopalatines. From this it follows that the
dermopalatine-vomer sequence is homologous throughout the osteichthyans.

The vomer is paired in Mimia, Moythomasia, Boreosomus, saurichthyids (Gardiner 1960: fig.
21), ' Aspidorhynchus' , Lepisosteus, caturids (Gardiner 1960: fig. 36), parasemionotids
(Patterson 1975: figs 30, 41), Amia, pachycormids (Lehman 1949: fig. 4; Patterson 1975: 513),
and the teleosts Hiodon and Osmerus (Patterson 1975: 513).

There is a median vomer in living chondrosteans (toothless in Polyodon, Acipenser and
Scaphirhynchus , Sewertzoff 1926: figs 3, 4, 39, the so-called median basirostral), Bobasa-
trania (Nielsen 1952: 199), in the semionotids Dapedium and Lepidotes (Gardiner 1960: 322), in
pycnodonts, leptolepids and the majority of teleosts (Patterson 1975: 515). In the last group the
vomer fuses with the ventral ethmoid during ontogeny. There is usually good evidence in
teleostean embryology for the paired origin of the vomer (de Beer 1937: 126, 130, 159).

A vomer is absent in Pteronisculus, Australosomus and adult specimens of Polypterus.
Nevertheless a binary primordium has been described in the 30 mm stage of Polypterus bichir by
Holmgren & Stensio (1936: 397) , and a median vomer in the 24 mm, 32 mm and 125 mm stages of
Polypterus by Pehrson (1947: 448).

The vomer is paired in actinistians (Whiteia, Latimeria, Millot & Anthony 1958), porolepids
(Porolepis, Glyptolepis, Holoptychius, Jarvik 1972: pis 3, 17, 25) and osteolepids (Megalichthys,
Ectosteorhachis, Jarvik 1966; Eusthenopteron, Jarvik 1942: fig. 56). In dipnoans it may be paired
or median. The vomer is paired in Uranolophus (Denison 1968), Dipnorhynchus (anterior

296

B. G. GARDINER

•5

<N
<N

£
Z

E
Jc

I
S

O
a

RELATIONSHIPS OF PALAEONISCIDS

297

pterygoids of Thomson & Campbell 1971), Griphognathus (dermopalatinum of Schultze 1969:
fig. 4; dermopalatine 1 of Miles 1977: fig. 57), Ceratodus, Sagenodus, Uronemus (Miles 1977:
175), Conchopoma (Schultze 1975), Gnathorhiza (Herman 1968), Monongahela (Lund 1970),
Neoceratodus, Lepidosiren and Protopterus (Miles 1977: 175). There is a median vomer in
Chirodiptems, Holodipterus (Miles 1977: figs 67, 87) and Scaumenacia (Jarvik 1967a: pi. 6).

The vomer is paired in Recent Amphibia, but median in Recent chelonians, lacertilians, birds
and monotremes. However, in the development of these choanates with a median vomer there is
often evidence of paired origin (de Beer 1937: 434).

In summary, the osteichthyan vomer is primitively a paired bone which fuses into a median
element in actinopterygians, dipnoans, lacertilians, chelonians, birds and monotremes. In
actinopterygians this fusion has occurred independently on at least five occasions, in Polypterus,
in Recent chondrosteans, in Bobasatrania, in semionotids and pycnodonts, and in teleosts.

2. Anterior articulation. In actinopterygian ontogeny, after the separation of the intermediating
body, the anterior ends of the palatoquadrates (the so-called pterygoid processes) come into
close contact with the ethmoid region (postnasal wall). Subsequently an articulation develops
between the anterior end of the palatoquadrate (autopalatine) and the lateral ethmoid, the
rostro-palatine articulation. An anterior articulation is found in most actinopterygians, with the
exception of Recent and fossil chondrosteans where in Acipenser and Polyodon the two
palatoquadrate bars meet in the mid-line forming a symphysis well below the ethmoid region. In '
Acipenser, however, they are still connected with this region by a ligament (Holmgren 1943: fig.
27), much as in Carcharhinus. The palatoquadrates are separate and distinct in Chondrosteus
(BMNH P. 2048) and meet in the mid-line. The rostro-palatine articulation is single in primitive
actinopterygians (Mimia, Moythomasia, Polypterus), Amia and halecomorphs, but in teleosts it
is often double (Salmo, etc.; Gardiner 1973: 119).

i nc

Fig. 68 Mimia toombsi Gardiner & Bartram. Right dermosphenotic in lateral (above) and medial

views, from BMNH P.56483.

298

B. G. GARDINER

i nc

2mm

Fig. 69 Moythomasia durgaringa Gardiner & Bartram. Dermosphenotics in lateral (above) and
medial views, from BMNH P. 53255. The right dermosphenotic is on the left.

A similar, single rostro-palatine articulation is found in actinistians (Latimeria, Rhab-
doderma, Macropoma, Undina), but in porolepids (Glyptolepis Jarvik 1972: fig. 25) and
Youngolepis (Chang 1982) this articulation is supported by additional articulatory facets or
points of fusion (Jarvik 1972: 71) between the autopalatine and suborbital shelf. In
Eusthenopteron the rostro-palatine articulation is said to be double (Jarvik 1942: figs 48, 50, 54,
art m, art 1) but in an acid-prepared specimen (BMNH P. 60310, Rosen et al. 1981) it is single.
This articulation is supported by a more medial articulation between a dorsomedial process of
the autopalatine and the orbital wall (suborbital shelf) (Jarvik 1954: figs 24, 40, pr.dm) and by
the head of the dermopalatine which articulates with the vomer (Rosen et al. 1981: figs 13, 14).
In Polypterus the primary rostro-palatine articulation is also supported by articulations of the
dermopalatines.

In placoderms a single rostro-palatine articulation has been recorded in Holonema (Miles
19716) and Dicksonosteus (Goujet 1975: fig. 4), and a double articulation in Ctenurella (0rvig
1960: pi. 29, fig. 1; Miles & Young 1977: fig. 27D) and Buchanosteus (Young 1979: fig. 15). In
acanthodians an anterior articulation is unknown and in the Permian Acanthodes (Miles 1965:
fig. 1) the palatoquadrates extend forward only as far as the posterior rim of the orbit. Here
there is no possibility of an anterior articulation (rostro-palatine), or of an anterior symphysis as
in Ptomacanthus (Miles 19736: pi. 6), because the palatine ossification has an unbroken
covering of perichondral bone (Reis 1896). It cannot have had an unossified palatine process as
proposed by Holmgren (1942: 138).

The palatoquadrate always meets its fellow in the mid-line in Recent selachians and may have
a ligamentous attachment with the ethmoid region, as in Carcharhinus , or a sliding articulation,
as in Chlamydoselachus.

On this evidence an anterior articulation cannot be primitive for gnathostomes. However,
there is little doubt that a single anterior articulation is the primitive osteichthyan condition,
with the autopalatine meeting the lateral ethmoid. The articulation of the autopalatine with the
postnasal wall is considered synapomorphous for placoderms plus osteichthyans.

3. Otic process and palatobasal articulation. The posterodorsal expansion of the palatoquadrate
is referred to as the otic process. This process is frequently single as in many chondrichthyans
(Cladoselache, Cladodus, Xenacanthus, Squalus, Heterodontus] , placoderms (Ctenurella,
Buchanosteus, Young 1979: figs 2, 12), acanthodians (Climatius, Miles 19736: fig. 8),
actinopterygians (Polypterus, Polyodon) and actinistians (Rhabdoderma, Wimania, Latimeria}.
The otic process may be notched for the maxillary and mandibular branches of the Vth nerve

RELATIONSHIPS OF PALAEONISCIDS 299

(Chlamydoselachus, Mustelus, Acanthodes, Amia, Eusthenopteron, Porolepis). When notched
the anterior division of the otic process is referred to either as the basal (orbital) process
(selachians, actinopterygians) or as the ascending process (actinistians, rhipidistians). The
single dorsal process of tetrapods is also referred to as the ascending process. The connection
which develops between the anteroventral region of the otic process (so-called basal process)
and the basipterygoid process is referred to as the palatobasal articulation.

The basipterygoid process is typically developed in osteichthyans, but it is also recognizable in
many selachians where it forms part of the subocular shelf as in Heptranchias , and in Squalus
(Jollie 1971) where it arises from the site of the polar cartilages.

In living actinopterygians the contact between the palatoquadrate and basipterygoid
processes is only to be seen in Lepisosteus (Hammarberg 1937: fig. 9) andAcipenser (Holmgren
1943: 30). In the latter a distinct blastema joins the palatoquadrate with the trabecula
(Holmgren 1943: figs 26, 27), and this later transforms into a ligament which according to
Bugajew (1929) may chondrify. However, no such chondrification was observed by de Beer
(1937: 91), Edgeworth (1935) or Holmgren (1943: fig. 27).

In Polypterus (Budgett 1901), Amia and Salmo (Holmgren 1943: 37, 40) the basipterygoid
process is not developed. Nevertheless in the development of Amia and Salmo the
palatoquadrate is connected with the trabecula by means of a thin membrane along its upper
border. In Pteronisculus there is either a tongue-like process of the metapterygoid (in part
supported by a corresponding tongue on the entopterygoid; Nielsen 1942: figs 34-37), which
articulates with the basipterygoid process, or there is a hole in the metapterygoid at the base of
the otic process (Lehman 1952: fig. 53), as in Mimia, Moythomasia and Kentuckia (Rayner 1951:
fig. 3), through which the basipterygoid process presumably slid during lateral movements of the
palatoquadrate. There is an articular fossa at the base of the otic process in Boreosomus (Nielsen
1942: fig. 70), while in Australosomus (Nielsen 1949: fig. 28) the anterior edge of the otic process
is produced into a prominent, medially-directed process. The loss of the palatobasal articulation
in Polypterus, Amia and teleosts may be directly correlated with the loss of the endochondral
basipterygoid process in these fishes and in Polypterus with the articulation of the entopterygoid
with the parasphenoid.

In Devonian actinistians (Nesides, Diplocercides] the basipterygoid process articulates with
the inside of the metapterygoid near the base of the otic process, where there is a marked
articular fossa (Bjerring 1977: fig. 28F), but as in later actinopterygians the basipterygoid
process is missing in Macropoma, Rhabdoderma and Latimeria.

In the porolepid Glyptolepis (Jarvik 1972: 58) and in the osteolepids Megalichthys and
Eusthenopteron (Jarvik 1954: figs 16, 22B) the condition is much as in Nesides, with a
well-developed basipterygoid process articulating with the base of the otic process; the latter
extends almost to the underside of the fronto-ethmoidal shield and there articulates with the
neurocranium.

In later acanthodians such as Acanthodes (Miles 1965: 238) there is a distinct notch in the
anterior margin of the otic process and the anterior process (basal process) articulated with the
basipterygoid process. The only difference from the condition in osteichthyans is that the
anterior process is ossified by the autopalatine and not the metapterygoid.

In ptyctodont placoderms the otic process has a grooved medial surface for a presumed
articulation with a basipterygoid process. A palatobasal articulation has also been described at
the base of the otic process in arthrodires (Buchanosteus , Young 1979: figs 2, 12).

In dipnoans the base of the otic process fuses with the trabecula (de Beer 1937: 172) and its top
fuses with the orbital wall during development. A similar fusion occurs in holocephalans,
anurans and urodeles.

In the fossil amphibian Palaeoherpeton the otic process is prominent and is ossified by the
epipterygoid. Ventrally on its inner surface there is a distinct roughened recess (Panchen 1964:
fig. 5), where it was probably in cartilaginous (immovable) contact with the basipterygoid
process. In monotremes the otic process is ossified by the alisphenoid (Presley & Steel 1976) and
consequently this bone may be homologous with the osteichthyan metapterygoid and the
amphibian and 'reptilian' epipterygoid (Broom 1914).

300

B. G. GARDINER

RELATIONSHIPS OF PALAEONISCIDS 301

The palatoquadrate and Meckel's cartilages are generally regarded as the epimandibular and
ceratomandibular elements respectively, with the anterior division of the otic process (basal
process) often being regarded as dorsalmost elements of a branchial arch (Huxley 1876; de Beer
1937). Sewertzoff & Disler (1924) suggested that the basal process (anterior otic process) is
serially homologous with the pharyngobranchials of the succeeding visceral arches; Holmgren
(1943: 64) and Bertmar (1959) homologized it with the actinopterygian suprapharyngo-
branchial. The evidence for considering the anterior division of the otic process a
pharyngomandibular rests on the claims of Sewertzoff & Disler (1924) that there is an
independent nodule of cartilage (or prochondral rudiment) which fuses to the medial surface of
the basal process in Squalus, Scy Ilium, Mustelus and Somniosus, and on the presence of separate
cartilages in the anterior region of the palatoquadrate in Scaphirhynchus and Acipenser
(Bugajew 1929). However, de Beer found no evidence of this cartilage in any of his specimens of
Squalus (1937: pis 11, 12) or Scyllium (1937: pis 13, 14, 15), nor did Holmgren (1940) find any
trace of a separate nodule in his exhaustive study of the embryology of Squalus and Etmopterus.
Holmgren concluded (1943: 64) that 'in Squaloid sharks the orbital process is formed in
continuum with the palatoquadrate proper'. Nevertheless he found that in Scyllium (Holmgren
1940: 153; fig. 104) and Mustelus (1943: 53) the anterior division of the otic process (basal
process) had a separate blastemic origin, although it subsequently chondrified in conjunction
with the remainder of the palatoquadrate (1940: 164). The cartilages described by Ivanzoff
(1887) in Scaphirhynchus and by Bugajew (1929: 98) in Acipenser lie in the connective tissue
covering the basitrabecular process and thus do not appear to belong to the palatoquadrate.
There is therefore no evidence in the development of the anterior division of the otic process
(basal process) to support the theory that it is a pharyngomandibular. Moreover since
suprapharyngobranchials appear to be an osteichthyan specialization (see p. 362) there is even
less evidence for considering the anterior otic process to be a suprapharyngomandibular.

Unfortunately the situation is further complicated by the assumption of several workers
(Holmgren 1943) that the anterior division of the otic process of actinopterygians is not
homologous with the selachian anterior otic process and that neither is homologous with the
choanate otic (ascending) process (Goodrich 1930: 413; de Beer 1937: 419). But the so-called
ascending process which is presumed by Goodrich (1930: 413) and de Beer (1937: 420) to be
present only in Dipnoi and tetrapods is no more than a dorsally extended otic process, much as in
Nesides (Jarvik 1954: fig. 15), Glyptolepis (Jarvik 1972: fig. 25) and Eusthenopteron (Jarvik
1954: fig. 23B). In actinistians, porolepids and Eusthenopteron the top edge of the otic process
articulates with the postorbital process (antotic process). In dipnoans and Recent amphibians
the top of the process fuses with the neurocranium (orbital cartilage). From these comparisons it
follows that both the single otic process and the palatobasal articulation may be regarded as
primitive gnathosome characters. The palatoquadrates of Ctenurella andJagorina suggest that
the omega-shaped (Schaeffer 1975) palatoquadrates of other placoderms are derived.

4. Otic process and prespiracular cartilage. In both fossil and Recent selachians the otic process
is prominent and in 'Cladodus', Xenacanthus, Tamiobatis (Romer 1964), Hybodus (Maisey
1982, 1983), Heptranchias (Daniel 1934) and Pseudocarcharias (Compagno 1973: 20) it
articulates with the postorbital process as in Acanthodes, In the majority of sharks the process
does not articulate with the braincase (e.g., Chlamydoselachus, Etmopterus, Isurus, Oxynotus,
Scyliorhinus, Squalus), and the otic process is missing in rays (e.g., Raja) and in Urolophus
(Holmgren 1940: fig. 181).

Both de Beer (1937: 420) and Holmgren (1943: 61) believed that in many selachians (Scyllium,
Squalus, etc.), the otic process becomes detached as the prespiracular cartilage (or spiracular
rudiment). In the development of Heterodontus, Raja, Urolophus and Etmopterus Holmgren
(1940; 1943: 60) has shown how the spiracular rudiment arises as a mesenchymatic lamella,
partly attached to the margin of the palatoquadrate anterior to the spiracular canal. It
subsequently contacts the postorbital process and fuses with it dorsally. Later that part of the
rudiment in front of the spiracle chondrifies independently of the palatoquadrate, with which it
loses contact, as the prespiracular cartilage. In Etmopterus Holmgren (1940: fig. 95) described

302

B. G. GARDINER

RELATIONSHIPS OF PALAEONISCIDS

303

304

B. G. GARDINER

2mm

i nc

Fig. 73 Moythomasia durgaringa Gardiner & Bartram. Right jugal in lateral (left) and medial views,

from BMNH P.53221.

both an otic process and a prespiracular cartilage and it became increasingly difficult to see how
the prespiracular cartilage could be the homologue of the otic process. In order to resolve this
apparent dilemma, Holmgren (1940: 140, 144; 1943: 56, 62) suggested that in Etmopterus the
otic process was not homologous with that in some other selachians (notidanids), dipnoans and
amphibians and that there were two distinct otic processes in sharks and rays, the processus
oticus externus and the processus oticus internus. He showed that the processus oticus externus
(1940: 140; 1943: 62) developed from the fusion of an extra palatoquadrate blastema with the
palatoquadrate blastema, after which chondrification took place, while the processus oticus
internus (1940: 131; 1943: 62) formed simultaneously as the lateral commissure. In 1940
Holmgren concluded that the otic process in Etmopterus was homologous with that in
Heptranchias , but in 1943 he was so confused that he at first considered them not the same, then
homologous and finally non-homologous all on the same page (p. 62). It is very difficult to
comprehend the movements and fusions of the various cell masses associated with the
development of the palatoquadrate in selachians as described by Holmgren (1940, 1943), but if
we confine our attention to the cartilage it is remarkably similar to that in osteichthyans.
Further, since the palatoquadrate and prespiracular cartilages always chondrify as separate
structures there is no reason to believe the latter to be a detached otic process, especially when

RELATIONSHIPS OF PALAEONISCIDS

305

i nc

Fig. 74 Moythomasia durgaringa Gardiner & Bartram. Left lachrymal in lateral (above) and medial

views, from BMNH P. 53221.

both an otic process and a prespiracular cartilage are to be found in Etmopterus and
Chlamydoselachus. There is even less evidence to support the view that the lateral commissure is
yet another detached otic process, since a lateral commissure also occurs in Etmopterus.

5. Ossifications of the palatoquadrate . (a) CARTILAGE BONES. In primitive actinopterygians such
as Mimia, Moythomasia, Polypterus (Allis 1922: 244), Pteronisculus (Nielsen 1942: 143) and
Acropholis (Aldinger 1937: 43) the palatoquadrate ossifies from three centres, the autopalatine,
metapterygoid and quadrate. Conditions are similar in more advanced forms such as Amia,
Pholidophorus and teleosts (e.g. Salmo, Gasterosteus, Cydoptems, de Beer 1937). However, in
Lepisosteus, Macromesodon (Nursall 1966) and some teleosts there is no autopalatine
(Patterson 1973: 246). In larger specimens of Mimia, Moythomasia, Pteronisculus, other

306 B. G. GARDINER

palaeoniscids (Elonichthys, Boreolepis, Aldinger 1937: 126) and Ospia (Stensio 1932ft: 252)
fusion of individual bones during ontogeny must have occurred because the palatoquadrate is
ossified throughout as one bone. In other specimens of Pteronisculus (Nielsen 1942: 143),
Saurichthys (Stensio 1925: 97), Acropholis (Aldinger 1937: 43) and Boreosomus (Stensio 1921:
211) there are only two ossifications. Actinistians have the same three ossifications as
actinopterygians (Nesides, Jarvik 1954: fig. 15; Macropoma, Undina, Rhabdoderma, Forey
1981: fig. 4; Latimeria, Millot & Anthony 1958) and this is undoubtedly the primitive
osteichthyan condition.

In all described porolepids (Jarvik 1972: 72) and in Eusthenopteron (Jarvik 1954: fig. 16) the
palatoquadrate shows no signs of subdivision, but is a single ossification, as in larger specimens
ofMimia and Pteronisculus. In Megalichthys Watson (1926: 247; fig. 33) described a continuous
series of endochondral suprapterygoid bones, but as in his descriptions of palaeoniscid palates
Watson (1925: 852; 1928: 52) misinterpreted the material. In his figured specimen of
Megalichthys the epipterygoid, suprapterygoids and quadrate are all one ossification. In the
palaeoniscids Elonichthys pectinatus (Watson 1925: 852; fig. 21), Elonichthys binneyi (Watson
1925: fig. 22) and Elonichthys aitkeni (Watson 1925: fig. 23) the so-called suprapterygoids are all
dermal bones (see under dermometapterygoid, p. 310), but in Nematoptychius greenocki
(Watson 1925: fig. 26; 1928) and Gonatodus (Watson 1925: fig. 27) the anterior suprapterygoid
is part of the autopalatine and the posterior suprapterygoid is part of the dermometaptery-
goid. Thus there is no evidence of more than three ossifications in the palatoquadrate
of osteichthyans.

The palatoquadrate in Dipnoi is attached to the auditory capsule and to the orbitotemporal
region of the neurocranium by the otic and antorbital processes (Sewertzoff 1902: 593), and in
Recent forms its only ossification is the quadrate in Neoceratodus .

The palatoquadrate in fossil dipnoans is large, rigidly fused to the neurocranium, and ossified
throughout as one bone. From its size and complexity in Dipnorhynchus (Thomson &
Campbell 1971: fig. 27), Griphognathus, Holodipterus, Chirodipterus (Miles 1977: figs 14, 22,
35, 53) and Stomiahykus (Bernacsek 1977: fig. 8) it is difficult to believe that it is ossified entirely
by the quadrate and it is likely that at least a metapterygoid was also present in its ontogeny.

In urodeles and apodans (Triton, Cryptobranchus, Ichthyophis) only the quadrate ossifies in
the palatoquadrate, as in Neoceratodus. But in fossil amphibians such as Palaeoherpeton there is
both a quadrate and a metapterygoid (the epipterygoid of Watson, 1926). Similarly in
lacertilians, chelonians and Sphenodon both quadrate and metapterygoid (epipterygoid) are
present.

In Acanthodes (Miles 1965: fig. 1) three perichondral ossifications exist in the palatoquadrate
cartilage, a large quadrate, smaller metapterygoid and much smaller autopalatine. The
quadrate and metapterygoid are separated by a large unossified portion, much as in smaller
specimens of Pteronisculus (Nielsen 1942: 143).

In arthrodiran placoderms the palatoquadrate sometimes contains two perichondral
ossifications, an anterior autopalatine and a posterior quadrate which is fused to the inner
surface of the postsuborbital (Miles 1971ft: figs 8, 9B). The two ossifications are separated by a
large unossified area as in Acanthodes and some specimens of Pteronisculus. In Dicksonosteus
(Goujet 1975: fig. 2) the palatoquadrate is perichondrally ossified as one bone, as it is in
Jagorina (Stensio 1969). In the ptyctodont Ctenurella (0rvig 1960, 1962) the palatoquadrate is
perichondrally ossified in three separate ossifications, autopalatine, metapterygoid and
quadrate, which are of approximately the same size (Miles & Young 1977: fig. 23).

Three ossifications in the palatoquadrate cartilage must be the primitive osteichthyan
condition, considered synapomorphous for a group containing acanthodians, placoderms and
osteichthyans.

(b) DERMAL BONES. Lining the roof of the mouth of osteichthyans is an extensive sheet of
tooth-bearing dermal bones associated with the oral face of the palatoquadrate. In
actinopterygians these bones form two series: an outer or ventral arcade comprising the
ectopterygoid and dermopalatines and an inner or dorsal series which primitively included

RELATIONSHIPS OF PALAEONISCIDS 307

dermometapterygoids and an entopterygoid (Fig. 75A, B). In all other osteichthyans the inner
or dorsal series (dermometapterygoids and entopterygoid) is absent (Figs 75C, D, 76).

As Rosen et al. (1981) suggest, cladistic relationships of sarcopterygians require reduction
rather than increase in palatal bones. Thus the presence of demometapterygoid and
entopterygoid bones is regarded as primitive rather than a synapomorphy of actinopterygians.

(c) ECTOPTERYGOID. Embryologically the ectopterygoid forms in sequence with the
dermopalatines in Polypterus (Pehrson 1947: fig. 28), Amia (Pehrson 1922) and many teleosts
(Salmo, Clupea, etc.). In juvenile specimens of Mimia (Figs 53, 54), Pteronisculus , Boreosomus
(Nielsen 1942: figs 34, 71), Birgeria (Nielsen 1949: fig. 71) and Elonichthys (Watson 1925: fig.
21), the ectopterygoid and dermopalatine clearly belong to the same series.

Primitively the ectopterygoid covers the posteroventral part of the oral face of the
palatoquadrate and overlies that part of the palatoquadrate which bounds the adductor opening
(Pteronisculus, Boreosomus, Birgeria, Elonichthys, Mimia, Polypterus). It also joins the maxilla
anteroventrally and the quadrate posteriorly (Mimia, Fig. 56; Pteronisculus, Boreosomus,
Nielsen 1942: 151; fig. 71; Australosomus, Birgeria, Nielsen 1949: figs 26, 71; Elonichthys,
Watson 1925: fig. 22; Polypterus, Allis 1922: fig. 25). Further, the centre of ossification of the
ectopterygoid is on a level with the most anterior part of the adductor opening, just where the
ectopterygoid meets the maxilla (Pteronisculus, Nielsen 1942: fig. 37; Australosomus, Birgeria,
Nielsen 1949: 105, 232; Mimia, Fig. 53; Polypterus, Pehrson 1947: fig. 28).

Since the ectopterygoid of actinopterygians reaches back to the quadrate and is the posterior
member of a series with the dermopalatines, it is clear that this bone is not homologous with the
bone cabled the ectopterygoid in actinistians (Macropoma, Watson 1921: fig. 4, ecpt),
poro\Qpias\Glyptolepis, Jarvik 1972: fig. 31, Ecpt), osteolepids (Eusthenopteron, Jarvik 1954:
fig. 16, Ecpt) and tetrapods (Presley & Steel 1978: fig. 2, ect). The ectopterygoid of
non-actinopterygians compares more favourably with the actinopterygian dermopalatine (see
below). On positional and other anatomical and developmental evidence the actinopterygian
ectopterygoid is better homologized with the entopterygoid of actinistians, lungfishes and the
pterygoid of tetrapods (Figs 75, 76).

(d) DERMOPALATINES. The dermopalatines form a series of interdigitating bones associated with
the anteroventral region of the palatoquadrate. Within the palaeoniscids a varying number of
dermopalatines has been recorded: four in Watsonichthys (Watson 1925: fig. 21, pal. 1-4),
three or four in Mimia (Figs 53, 54), three in Mesonichthys (Watson 1925: fig. 23), two in
Elonichthys (Watson 1925: fig. 22), Nematoptychius (Watson 1925: fig. 26), Pteronisculus,
Boreosomus and Birgeria (Nielsen 1942: figs 34, 71; 1949: fig. 71). There is one in Gonatodus
(Watson 1925: fig. 27), Namaichthys (Gardiner 1962: fig. 3) and Polypterus. In Amia there are
two dermopalatines, but in most higher actinopterygians there is only one (Lepidotes, Gardiner
1960: fig. 47; Lepisosteus; Ospia; Elops, Nybelin 1968: fig. 1). In many teleosts (e.g. Salmo, de
Beer 1937: 126) this single dermopalatine fuses with the autopalatine to form a composite bone.
Dermopalatines are missing in Australosomus (Nielsen 1949: fig. 30) and pycnodonts
(Macromesodon, Nursall 1966). In actinistians the so-called ectopterygoid (Millot & Anthony
1958) is undoubtedly a member of the dermopalatine series; in many fossil forms the
corresponding bone is indistinguishable in size and shape from the preceding dermopalatine
with which it is closely sutured (e.g. Macropoma, P. L. Forey, personal communication). Thus
the ectopterygoid of actinistians is better regarded as a posterior dermopalatine, in which case
all actinistians possess three dermopalatines as in Mimia and Watsonichthys.

In porolepids (Glyptolepis, Jarvik 1972: fig. 31) the ectopterygoid is again in sequence with an
anterior dermopalatine which it resembles in shape, size and disposition of teeth; and similarly
in the osteolepids Eusthenopteron (Jarvik 1954: fig. 16) and Glyptopoma (Jarvik 1950: fig. 6).

Thus in porolepids, osteolepids and onychodonts (Andrews 1973: 146) there are always two
dermopalatines (not three as in actinistians), and in this respect these fishes resemble Amia and
tetrapods. Primitively in tetrapods there are two palatines, an anterior dermopalatine and a
more posterior transpalatine (ectopterygoid of Presley & Steel 1978) and their homology with
the two dermopalatines in porolepids and osteolepids can be established by positional evidence.

308

B. G. GARDINER

ENDOPTERYGOID

Fig. 75 Palates in ventral view and suggested homology of dermal bones and their relation to
palatoquadrate. A, Mimia toombsi Gardiner & Bartram; B, Polypterus bichir Saint-Hilaire; C,
Eusthenopteron foordi Whiteaves (from Jarvik 1954); D, Glyptolepis groenlandica Jarvik (from
Jarvik 1972). Compare Fig. 76. From Rosen et al. (1981).

In Recent dipnoans both dermopalatine and ectopterygoid are missing but in the fossil
Griphognathus (Miles 1977: fig. 57, Dpl2) one bone of this series remains, which from its
position bordering the medial edge of the fenestra ex ochoanalis, must be homologous with the
dermopalatine of tetrapods. A similar parallel loss of the transpalatine has occurred in the
Lissamphibia; in the urodeles Salamandra and Cryptobranchus the palatine also disappears at
metamorphosis (Wintrebert 1922: 239). The transpalatine is also missing in chelonians and
several fossil 'reptiles' (Placodus, Ichthyosaurus, etc).

1 Re-examination of the specimen of Cheirolepis (BMNH P. 36061) described by Pearson & Westoll (1979: fig. 8) has
convinced me that what they call lines of individual entopterygoids are no more than fragments of a broken dermal cheek
bone.

RELATIONSHIPS OF PALAEONISCIDS

309

2 - "

Fig. 76 Palates in ventral view and suggested homology of dermal bones and their relation to
palatoquadrate. A, Griphognathus whitei Miles (from Miles 1977); B, Ichthyostega sp. (from
Romer 1966); C, Eogyrinus attheyi Watson (from Panchen 1972); D, Benthosuchus sushkini
Efremov (from Bystrov & Efremov 1940); E, Tylototriton verrucosus Riese (from Noble 1931).
Compare Fig. 75. From Rosen et al. (1981).

actinopterygians with the exception of pycnodonts (Macromesodon, Nursall 1966). It is not
found in any other osteichthyan. In Polypterus the entopterygoid arises in series with the
dermometapterygoid (Pehrson 1947: 448), somewhat later in ontogeny than the ectopterygoid
and dermopalatine. Although in Mimia (Enpt, Figs 53, 54) the entopterygoid could be regarded
as being in series with either the ectopterygoid or dermometapterygoids, in most palaeoniscids
(and Polypterus) it is clearly in series with the dermometapterygoid. Further, in Pteronisculus
(Nielsen 1942: fig. 37) and Polypterus the centres of ossification of the entopterygoid and
dermometapterygoid lie near the middle of the dorsal margin of those bones whereas the centres
of ossification of the ectopterygoid and dermopalatines are at or near their ventral margins. The
centre of ossification of the entopterygoid in Mimia also lies near its dorsal margin and similarly
in Amia and Elops.

310

B. G. GARDINER

SPIRACULARS

POSTORBITAL
LACRIMAL

SQUAMOSAL I

7SUBOPERCULAR
7PRESPIR ACULAR

OPERCULAR

SUBOPERC ULAR

INFRAORBITO-
MAXI LLA

DERMOSPHENOTIC PREOPERCULAR

OERMOHYAL

ACCESSORY
OPERCULAR

SQUAMOSAL 2

SUBOPERCULAR

QUADRATOJUGAL
PPREOPERCULAR

Fig. 77 Cheek bones and operculum. A, Rhabdoderma elegans (Newberry) (after Forey 1981); B,
Strunius walteri lessen (after Jessen 1966); C, Polypterus bichir Saint-Hilaire (after Daget 1950);
D, Cheirolepis trailli Agassiz (after Pearson & Westoll 1979). Bone names are those used by the
authors cited. From Rosen et al. (1981).

The entopterygoid is primitively excluded from the jaw margin by the ectopterygoid and
dermopalatines, but in Australosomus (Nielsen 1949: fig. 30), where the dermopalatines are
absent, the entopterygoid contacts the maxilla anteriorly.

(f) DERMOMETAPTERYGOID. The dermometapterygoids form a series of interdigitating bones
associated with the posterodorsal region of the palatoquadrate, in particular the metapterygoid.
Polypterus and Amia are the only living fish with a dermometapterygoid. Within the
palaeoniscids a varying number of dermometapterygoids have been recorded. There are five
dermometapterygoids in Watsonichthys (Watson 1925: fig. 21, spt) and Elonichthys binneyi
(Watson 1925: fig. 22, spt 2-6), four in Mimia (Dmpt, Figs 53, 54) and Elonichthys aitkeni
(Watson 1925: fig. 23, the four small bones behind the accessory vomer, spt 1), and one in
Elonichthys caudalis, Elonichthys semistriatus, Gonatodus, Nematoptychius (Watson 1925: figs
24, 25, 27, metpt; 1928: fig. 7), Pteronisculus (Nielsen 1942: fig. 37), Birgeria (Nielsen 1949: fig.
71), Polypterus and Amia. Both entopterygoid and dermometapterygoid bones are primitively
present in actinopterygians but neither occurs in other osteichthyans.

We may also conclude that the primitive osteichthyan possessed a further row of
tooth-bearing dermal bones consisting of a pterygoid (= ectopterygoid) and several
dermopalatines (at least two).

Dermal bones of the cheek: summary and discussion

The bones of the cheek are fairly uniform throughout the osteichthyans (Figs 77, 78). Perhaps
the one exception is the osteichthyan dermosphenotic (= postorbital of tetrapods). The
dermosphenotic carries the infraorbital canal; it is lateral to the intertemporal in primitive
actinopterygians and ventral to the so-called 'dermosphenotic' (Jarvik 1954, 1972) in
osteolepids, and is in sequence with the infraorbital bones (jugal and lachrymal). It also forms
the dorsolateral border to the spiracle in Mimia, Moythomasia and the spiracular pouch in

RELATIONSHIPS OF PALAEONISCIDS

311

POSTORBITAL

PO5T5PIRACULAR

PREOPERCULAR

POSTORBITAL

SQUAMOSAL

PREOPERCULAR

5UBOPERCULAR
QUADRATOJUG AL

QUADRATO JUG AL

Fig. 78 Cheek bones and operculum. A, Eusthenopteron foordi Whiteaves (after Jarvik 1944a); B,
Porolepis brevis Jarvik (after Jarvik 1972); C, Ichthyostega sp. (after Jarvik 1952); D,
Griphognathus whitei Miles (after Miles 1977). From Rosen et al. (1981).

Latimeria; in Polypterus and Acipenser it lies anterior to the spiracle. It is in contact with the
frontals in primitive actinopterygians (Moythomasia, Mimia, Cheirolepis, Polypterus) and
porolepids (Holoptychius, Porolepis, Glyptolepis, Jarvik 1972: figs 43, 44, 45); with the
'dermosphenotic' /X and intertemporal /Yt in osteolepids (Eusthenopteron, Osteolepis, Jarvik
1972: fig. 61), onychodonts (Onychodus) and primitive dipnoans (Uranolophus, Gripho-
gnathus, Chirodipterus, Miles 1977: fig. Ill); with the parietals in actinistians and with the
parietals and supratemporals in Ichthyostega. In other primitive tetrapods (loxommatids,
temnospondyls) it contacts the postfrontal and supratemporal.

Most workers, however, have failed to recognize that the actinopterygian dermosphenotic is
homologous with the tetrapod postorbital, because they have been too concerned in trying to
find a one-to-one relationship between the cheek and roofing bones of osteolepids and
tetrapods. In primitive actinopterygians (Cheirolepis, Mimia, Pteronisculus; Fig. 88) there are
two bones along the otic portion of the temporal sensory canal (supratemporal ana
intertemporal). There is one in actinistians and porolepids (supratemporal), but in Powichthys,
Youngolepis, osteolepids, onychodonts and dipnoans there are three (supratemporal,
intertemporal, and 'dermosphenotic'; Jarvik 1972: fig. 61; Miles 1977: fig. Ill, Y2, Yt, X).
Thus, those authors who have accepted Westell's (1938) theory that the rhipidistian
(osteolepid) frontal is homologous with the tetrapod parietal (see p. 320) consider the
osteolepid supratemporal to the homologous with the tetrapod tabular, the intertemporal with
the supratemporal, and the 'dermosphenotic' with the tetrapod intertemporal (Panchen 1964:
fig. 18; Andrews 1973: fig. 3; Vorobjeva 19770: fig. 2). They are then able to homologize the
osteolepid postorbital with the tetrapod postorbital and so achieve a one-to-one relationship.
Save-Soderbergh (1932: fig. 15), on the other hand, considered that in Ichthyostega the
dermosphenotic had fused with the postorbital, but that in other amphibians (Palaeoherpeton,
Save-Soderbergh 1935: fig. 41;Aphaneramma, Save-Soderbergh 1936: fig. 31A-D) it has fused
with the supraorbital. Stensio (1947: 93; fig. 26) maintained that in the majority of fossil
amphibians, including Palaeoherpeton, the dermosphenotic had fused with the postorbital and
possibly 'one or a couple of adjoining dermopterotic elements too'. Jarvik (19676: figs 10, 13)
supported Save-Soderbergh as far as Palaeoherpeton was concerned, considering the
dermosphenotic to have fused with the supraorbital in this form and in 'reptiles', but in
temnospondyls he believed it had fused with the intertemporal.

312 B. G. GARDINER

Westell's (1938) theory concludes that in osteolepids (Fig. 88F) the most anterior element on
the main lateral-line (temporal) canal, before the latter turns down onto the cheek
(dermosphenotic of Jarvik), is the homologue of the tetrapod intertemporal. Since the
intertemporal in tetrapods never has a groove or sulcus for the main lateral-line canal (see for
example Eogyrinus, Watson 1940: fig. 12), but instead is associated with the supraorbital canal
(Dendrerpeton, Steen 1934; Edops, Romer & Witter 1942; Trimerorhachis , Romer 1947: 247;
Palaeoherpeton , Panchen 1964: 221; fig. 11), it is difficult to homologize the tetrapod
intertemporal with the osteolepid 'dermosphenotic'. Similarly, since the main lateral-line canal
always passes through the intertemporal in actinopterygians and osteolepids (Moythomasia,
Jessen 1968: fig. 1, Dsph; Eusthenopteron, Osteolepis, Jarvik 1955: fig. 4), where this bone is
present, the actinopterygian and osteolepid intertemporal cannot be homologous with the
tetrapod intertemporal. This confusion stems from the failure of most authors to recognize that
the three bones on the otic portion of the temporal canal of osteolepids are not matched by the
three bones in tetrapods. In tetrapods only the supratemporal is associated with the otic portion
of the temporal canal. The tetrapod intertemporal, where present, is associated with the
supraorbital canal and the tabular with the supratemporal commissure (see below under dermal
bones of skull roof, p. 320). In dipnoans there is an additional series of bones between the
temporal series and the parietal and median postparietal.

I can now turn to the fusion theory of Save-Soderbergh (1932: fig. 15; 1935: fig. 41), Stensio
(1947: 93), and Jarvik (19676: figs 10, 13). As pointed out by Jardine (1970: 345), Nelson (19690)
and Miles (1977: 221), the terms loss and fusion have no clear meaning when applied to
phylogeny so that it is not possible to choose objectively between loss and fusion hypotheses.
However, there is no need to infer either in this situation because the homologue of the
actinopterygian dermosphenotic is clearly recognizable in tetrapods. Many fossil amphibians
have been described in which the temporal canal leaves a well-marked lateral groove on the
supratemporal before running onto the postorbital, where it then turns down onto the jugal
(Lyrocephalus, Metoposaurus, Aphaneramma, Save-Soderbergh 1937: figs 4A, 12, 31, etc.;
Trimerorachis , Case 1935). Thus the tetrapod postorbital is the homologue of the
actinopterygian dermosphenotic.

The dermosphenotic/postorbital is loosely attached or hinged to the skull roof in many
actinopterygians (Mimia, Stegotrachelus, Lepisosteus, teleosts), Gyroptychius (BMNH
50104), osteolepids, actinistians and porolepids, and this is presumed to be the primitive
osteichthyan condition.

The 'dermosphenotic' of Eusthenopteron (Jarvik 1972: fig. 61) and Onychodus is not
homologous with the actinopterygian dermosphenotic; instead it is considered topographically
homologous with the 'dermosphenotic' (Fig. 89C) or bone X of dipnoans (Miles 1977: fig. 111).

The remaining cheek bones are less contentious. The jugal canal joins the infraorbital canal
below the eye on the suborbital portion of the cheek in actinisitians, onychodonts, porolepids,
osteolepids, dipnoans and tetrapods as it does in some selachians and acanthodians. In
actinopterygians the jugal canal is wanting except in Polyodon and there is usually a single
ossification on the preopercular canal, the preopercular. Exceptions include Boreosomus,
Bobasatrania and Luganoia with two ossifications, and Polyodon and Macromesodon with up to
eight tubular bones.

In actinistians, osteolepids and the tetrapods Ichthyostega and Acanthostega (Save-
Soderbergh 1932: fig. 15; Jarvik 1952: fig. 33a), there are normally two bones on the
preopercular- jugal canal, whereas in porolepids (Jarvik 1972: figs 43, 44) there may be three, in
dipnoans six (Neoceratodus) or seven (Griphognathus , Miles 1977: fig. 112), and in tetrapods
other than Ichthyostega and Acanthostega one, the squamosal. By comparing Ichthyostega and
Acanthostega with other osteichthyans, the squamosal is presumed to be homologous in
actinistians, tetrapods and osteolepids and the condition in dipnoans, where there are numerous
elements, is derived. The actinopterygian condition, with the preopercular extending forwards
above the posterior expansion of the maxilla, is likewise derived.

In primitive actinopterygians (Cheirolepis , Mimia, Moythomasia), osteolepids (Eustheno-
pteron, Osteolepis, Eusthenodon) and some tetrapods (Ichthyostega, Acanthostega, Palaeoher-

RELATIONSHIPS OF PALAEONISCIDS 313

peton) the lower margin of the cheek is formed by the quadratojugal and the toothed maxilla.
Porolepids and Polypterus are similar except that here the preopercular also contributes to the
lower margin. In actinistians the quadratojugal is wanting and possibly the maxilla also, and in
this latter respect actinistians parallel later dipnoans. Thus a separate quadratojugal (bone 10) is
still recognizable in many early dipnoans such as Dipnorhynchus (Thomson & Campbell 1971:
fig. 7), Griphognathus and Chirodipterus (Miles 1977: figs 112, 117, bone 10), where it is
associated with the cheek pit-line as in actinopterygians, osteolepids, porolepids and some
tetrapods (Palaeoherpeton, Panchen 1964: fig. 12). Furthermore the bone described as an
ectopterygoid in Griphognathus by Miles (1977: fig. 57), and which bites outside the lower jaw,
is most probably a maxilla (Rosen et al. 1981: fig. 7).

The remaining bones of the cheek constitute the infraorbital series which together form the
hind and lower borders of the orbit. In primitive actinopterygians (Mimia, Moythomasia) and in
actinistians (Macropoma, Rhabdoderma, Latimeria), osteolepids and porolepids there are only
two bones in this series, as in all tetrapods. It seems likely therefore that they are homologous
with the jugal and lachrymal. In later actinopterygians the number of bones in the infraorbital
series is greatly increased and may be as high as seven in many teleosts (Nelson 1969a: 4). In
osteolepids and porolepids the condition is often obscured by ontogenetic fusion of the bones
anteriorly. Nevertheless in most described cases at least two bones are recognizable, the jugal
and lachrymal of Jarvik (1980). In Panderichthys (Vorobjeva I917b: fig. 2), however, there are
as many as four elements. In dipnoans the number of infraorbitals is more variable and in this
respect they parallel later actinopterygians. Thus there are four infraorbitals in Neoceratodus,
five in Dipnorhynchus, Scaumenacia, Griphognathus and Sagenodus, and six in Chirodipterus
and Dipterus.

A summary of the dermal bone homologies outlined above is presented in Table 1, p. 323.

Sensory canals of the cheek: summary and discussion

The preopercular canal joins the infraorbital canal behind the orbit and beneath the spiracle in
actinistians, porolepids, osteolepids, onychodonts, dipnoans and amphibians. The preopercular
canal also joins the infraorbital canal ventral to the spiracle in Polyodon but in this instance (as in
Macromesodon) the cheek plates are reduced and the canal runs in a series of tubular ossicles, a
condition which is assumed to be secondary (Stensio 1947). This connection between the
preopercular and infraorbital sensory canals is generally referred to as the jugal line or canal
(hyomandibular line; angular-jugal line; supramaxillary line) since in sharks it often
differentiates as an independent line (Rudd 1920). In sharks the jugal line is invariably
connected to the infraorbital canal from which it may also develop (Holmgren 1940: 85). Usually
the jugal canal is not joined posteriorly to the preopercular canal but in some specimens of
Chlamydoselachus and in Torpedo it links the infraorbital and preopercular canals (Holmgren
1942: fig. 19). The jugal canal also grows out from the postorbital portion of the infraorbital
canal in Neoceratodus (Allis 1934: 369), but in amphibians (Platt 1896, Stone 1922) this
connection is achieved by a branch of the preopercular canal growing forwards and downwards
across the cheek to meet the infraorbital canal. The preopercular canal is also joined to the
infraorbital canal by the jugal canal in many acanthodians and placoderms and this is presumed
to be the primitive gnathostome condition.

In actinopterygians, where there is no connection between the two canals, there is a
horizontal pit-line in the position of the jugal canal which is similarly innervated by a branch of
the mandibular nerve (hyoideo-mandibularis of Pehrson 1947). This pit-line is not found in
other groups and it is also missing in Polyodon; it is generally considered to be the homologue of
the jugal canal of other forms (Stensio 1947). The horizontal pit-line arises from the upper part
of the preopercular canal in Polypterus and Amia (Allis 1889: Pehrson 1947) and remains
intimately connected with it in both embryo and adult. Traces of the anterior limits of the
horizontal pit-line are found on the maxilla of Polypterus (Jarvik 1947: fig. 1A) and
Moythomasia.

A further pit-line is present in many osteichthyans, the vertical pit-line (or postmaxillary line,
Stensio 1947). This line is usually in two parts and the dorsal component meets the horizontal

314

B. G. GARDINER

ano

pinf

por

spig

Fig. 79

fotc

Mimia toombsi Gardiner & Bartram. Neurocranium and attached dermal bones in dorsal
view, from BMNH P. 53243.

pit-line to give a > -shaped structure in primitive actinopterygians (Mimia, Polypterus). The
ventral part of this line in actinopterygians crosses the quadrate jugal when this bone lies near the
surface (Mimia, Polypterus, Pteronisculus) . In more advanced actinopterygians (Lepisosteus,
Amid) the vertical pit-line is in one piece. The vertical pit-line is also in two parts in some
actinistians (Rhab do derma), porolepids (Holoptychius), osteolepids (Eusthenopterori) , some
dipnoans (Dipterus) and a few primitive amphibians (Palaeoherpeton), with the ventral portion
crossing the quadratojugal where this bone is present. The line is single in Griphognathus,
Neoceratodus and Protopterus. The vertical pit-line meets or crosses the jugal canal in sharks
(Chlamydoselachus), actinistians (Nesides, Rhabdodermd) , osteolepids (Eusthenopteron) and

RELATIONSHIPS OF PALAEONISCIDS

315

Plate 1 Mimia toombsi Gardiner & Bartram. Braincase in dorsal view, from P. 56505, xll!/2.

316

B. G. GARDINER

i nc

sue

2mm

Fig. 80 Mimia toombsi Gardiner & Bartram. Dermal bones of the skull roof in dorsal view
(intertemporal missing), from BMNH P. 56473.

dipnoans (Griphognathus, Protopterus) . The absence of a squamosal bone and a jugal sensory
canal are considered autapomorphous for actinopterygians.

Dermal bones of the skull roof

Mimia toombsi

The dermal bones on the dorsal surface of the neurocranium anterior to the occipital fissure are
closely applied to the dorsal neurocranial surface. Even the nasals may have a fragile attachment
at points where the delicate perichondral nerve canals join the underside of the supraorbital
sensory canal (PI. 1). The only areas where this attachment is less than secure is where the
perichondral lining of the neurocranial roof is interrupted posteriorly around the lateral cranial
canal and anteriorly in front of the pineal foramen.

The parietal is roughly rectangular in outline, somewhat longer than broad and with its
radiation centre beneath the middle pit-line. It has a zigzag suture anteriorly with the frontal.
Ventrally two sets of nerve foramina presumably served for branches of the glossopharyngeal
nerve (fb.IX, Fig. 81) to the middle pit-line and for branches of the vagus (fb.X, Fig. 81) to the

RELATIONSHIPS OF PALAEONISCIDS

317

fblX

i nc

sue

Fig. 81 Mimia toombsi Gardiner & Bartram. Dermal bones of the skull roof in ventral view
(intertemporal missing), from BMNH P. 56473.

posterior pit-line. The anterior pit-line is continuous with the supraorbital canal which leaves the
parietal through a tongue-shaped projection of the anterior margin.

The largest element of the dermal roof is the paired frontal which tapers to a point anteriorly.
Posteriorly it forms a slight overlap with the parietal. The radiation centre is nearer the posterior
than anterior margin, at the level of the rear of the pineal foramen. The supraorbital canal
pierces the radiation centre. The anteromedial margin of the frontal has a zigzag suture with the
rostral while the anterolateral corner sutures with the nasal. Ventrally the passage of the
supraorbital sensory canal is marked by a rounded ridge which may have a slit-like opening
anteroventrally (Fig. 81).

The supratemporal is a long, narrow bone which is sutured medially to the parietal and
anteriorly to the intertemporal. There is a small notch in its lateral margin (n, Figs 80, 81) dorsal
to the head of the hyomandibula. This notch, present in the lateral margin of the intertemporal
of many actinopterygians (Cheirolepis, Pteronisculus, Elonichthys, Moythomasia) , was said by
Aldinger (1937: 249) to be related to the underlying fossa bridgei and by Jessen (1968: fig. 1) to
be the spiracular opening. But there is no fossa bridgei in Mimia or Moythomasia and the
spiracular opening is between the intertemporal, supratemporal and dermosphenotic. Instead I
postulate that this notch allowed the head of the hyomandibula greater flexibility in respiratory
movements. The supratemporal is traversed by the otic part of the main lateral-line canal which
passes through the radiation centre. This centre lies just behind the notch in the lateral margin.
Anterolaterally a flange of the supratemporal forms the posterolateral margin of the spiracular
opening (Fig. 82).

318

B. G. GARDINER

i nc

por

sue

2mm

Fig. 82 Mimia toombsi Gardiner & Bartram. Otic and orbitotemporal regions of neurocranium and
attached roofing bones in dorsal view, from BMNH P. 53259.

The intertemporal is a small triangular bone sitting above the postorbital process. It sutures
with the parietal and supratemporal posteriorly and with the frontal medially. Its radiation
centre is pierced by the main lateral-line canal and lies near the posterior margin. The passage of
the sensory canal through both temporal bones is marked ventrally by a rounded ridge.

The extrascapular series consist of a single pair of bones which just meet in the midline.
Anteriorly the extrascapular sits on the slightly bevelled transverse margin of the parietals.
Posteriorly and laterally the extrascapular overlies the occipital region of the neurocranium,
covering the posterior dorsal fontanelle and part of the occipital fissure. The supratemporal
commissure pierces the bone in a transverse direction and the cephalic division of the main
lateral-line runs in a longitudinal direction. The radiation centre is situated at the confluence of
these two canals. A posterolateral peg-like projection (Figs 84, 85) of the extrascapular passes
under the post-temporal, while its posterior margin rests on an anterior flange of the
post-temporal.

Moythomasia durgaringa

The parietals and frontals are very similar to those of Mimia (Fig. 83). However, the
posteroventral margin of the parietal is more shelf-like than in Mimia and more intimately fused
with the underlying neurocranium. The posterior corner of the supratemporal is also more
markedly pointed. The extrascapular series consists of two pairs of bones, a smaller medial pair
and a much larger lateral pair (Fig. 87).

The lateral extrascapular contacts the parietal anteriorly, but laterally sits on an
inwardly-directed flange of the supratemporal. The medial extrascapular comprises two
components: an anterior plate of bone and a posterior tubular portion for the supratemporal
commissure.

319

Fig. 83 Moythomasia durgaringa Gardiner &
Bartram. Left otic and orbitotemporal regions
of neurocranium and attached roofing bones in
dorsal view, from BMNH P. 53221.

i nc

pinf

320

B. G. GARDINER

StC

Fig. 84 Mimia toombsi Gardiner & Bartram.
Right extrascapular in dorsal (above) and
ventral views, from BMNH P.54498.

stc

Dermal bones of skull roof: summary and discussion

1 . Homologies of dermal bones of skull roof

The osteichthyan dermal roofing bones form two distinct patterns (Rosen etal. 1981). The more
primitive is believed to be that found in actinopterygians, osteolepiforms, porolepiforms and
actinistians, in which the paired frontals and parietals form the major constituents of the skull
roof and the parietals reach the posterior limits of the otic region of the underlying
neurocranium. The pineal foramen is invariably situated between the frontals. The alternative,
derived pattern is seen in dipnoans and tetrapods (Fig. 89) , where there is a cluster of at least two
pairs of bones behind the parietals and the pineal foramen lies either between the parietals or
just anterior to them (Dipnorhynchus , later tetrapods).

The two pairs of bones covering the dorsal side of the otic region in actinopterygians,
osteolepiforms, porolepiforms and actinistians were originally called frontals and parietals
because they appeared to be homologous with those bones in mammals. The cluster of six bones
behind the parietals in primitive tetrapods were called postparietals, tabulars and supra-
temporals. Lying behind these skull roofing bones and therefore not attached to the under-

RELATIONSHIPS OF PALAEONISCIDS

321

StC

1mm

stc

Fig. 85 Mimia toombsi Gardiner & Bartram. Left extrascapular in dorsal (above) and ventral views,

from BMNH P. 56497.

lying neurocranium is a series of scale bones or extrascapulars (Jollie 1981) . This series is missing
in tetrapods.

In the search for tetrapod origins it was necessary to reconcile these two distinct dermal
roofing bone patterns, because osteolepiforms were thought to include the tetrapod ancestor.
Thus Save-Soderbergh (1932) proposed that the ancestral dermal roof of crossopterygians and
tetrapods must have contained two pairs of frontals and parietals and that the tetrapod parietal
and postparietal should be regarded as fronto-parietal and parieto-extrascapular respectively.
The assumption that the extrascapular series of scale bones could in some way become
intimately associated with the otic region of the neurocranium had already been proposed by
Watson & Day (1916) when they homologized the crossopterygian medial extrascapular with
the tetrapod postparietal and the lateral extrascapular with the tabular. This theory,
subsequently modified by Save-Soderburgh (1935, 1936) and championed by Jarvik (1967b:
205), demands that a transverse series of scale bones moved forward onto the neurocranium. As
Rosen et al. (1981: 222) have pointed out, the dermal roofing bones in osteichthyans (and
placoderms - see Jarvik 1967 b: fig. 3) are frequently attached to the underlying neurocranium
by descending laminae of membrane bone. Where laminae are missing the dermal bones may be
equally tightly attached to the underlying perichondral bone as in Mimia, Moythomasia and
Eusthenopteron (Jarvik 1975: fig. 13). It therefore appears unlikely that these dermal roofing
bones would have been able to move forward on the otic region to make room for the
extrascapular series. Moreover, in many dipnoans the cluster of five roofing bones (behind the
parietals) is followed by a series of extrascapular scale bones which lie loosely behind them
(Miles 1977: figs 111, 116, 118), free of the underlying otic region of the neurocranium.

An alternative theory to that of Watson & Day (1916) was Westell's (1936, 1938) proposition
that the crossopterygian frontal was homologous with the tetrapod parietal. Westell's theory
demands the reverse of Watson & Day's: that is, it assumes that there was a backward
movement of the parietals, again without regard to the underlying neurocranium. To accept
Westoll's theory is to deny the presence of frontals in all bony fishes with the exception of the
osteolepids, Elpistostege and Panderichthys and certain dipnoans (Diptems, Uronemus,

322

B. G. GARDINER

Fig. 86 Mimia toombsi Gardiner & Bartram. Right post-temporal in dorsal (left) and ventral views,

from BMNH P. 56498.

Rhinodipterus, Scaumenacia, Ctenodus). I have suggested (Gardiner 1980) that an extra pair of
bones in Panderichthys be called postparietals, but they are associated with transverse pit-lines
(Vorobjeva 19776: fig. 2B) and are better homologized with the parietals of Eusthenopteron.

The temporal bones (intertemporal, supratemporal /Y^) have already been discussed (see
p. 311), but it is worth noting that in later actinopterygians (e.g. many palaeoniscids,
Lepisosteus, Amia, teleosts) the two temporal bones are replaced by a single dermopterotic. A
single bone similarly occupies this position in Ichthyostega, most temnospondyls, and primitive
amniotes where it is called the supratemporal. Two temporal bones occur in some loxommatids,
a few temnospondyls, anthracosaurs and Seymour ia.

From the evidence given above it seems that the tetrapod supra- and intertemporal are not
homologous with similarly-named bones in other osteichthyans. If this conjecture is correct,
then those bones associated with the otic portion of the infraorbital sensory canal in primitive
osteichthyans are absent in tetrapods. Support for this suggestion is afforded by living
amphibians where the otic part of the infraorbital line is reduced to a single organ (Platt 1896;
Stone 1922) and by primitive fossil amphibians where the infraorbital line ends blindly in the
postorbital (Ichthyostega, Loxomma, Crassigyrinus, etc.).

RELATIONSHIPS OF PALAEONISCIDS

323

<u "O
> o

13
_o>

13 I-H

1 &

§• £

c m

"e3

CO
0

03
00

•-C ex

03 5

i-i

<u %

'C <u

OH H

C.«
0 i-

•— 03
U- DH

ti
« ex

£ oo

3 T2 r^

-0 00"

cd -^ cd

E- ^J

3
0"

"co

I-H

—

Ichthyosteg

Frontal
Parietal

Median
postpari*

Postorbital

Supratempi

Tabular

Jugal
Lachrymal

Squamosal
Preopercul

Quadratojl

Maxilla

Premaxilla

Dipnoans

Frontal/E
Parietal/C

Median
postparietal/B

Postorbital/4

Dermosphenotic

Intertemporal/Yl ,X

Supratemporal/Y2

Anterior lateral
parietal/J

Posterior lateral
parietal/I

Infraorbitals/5 ,6,7,1

l-H

Maxilla

Premaxilla fused
into snout

4-10 Nasals

3 Extrascapulars

i-i

13 3

oo o

CO
0
to
03

Frontal

Parietal

Postorbi

£
1

£
i-i
13 -=
w> y

5 J

° Jr:

E(U
„ cx
2 °

S-fi

OO OH

Premaxi

3 Nasals

3-7Exti

•o

I-H 1 _C

£ «-

I-H

^ i , i *5

^ 3

03 3 3 c

£ 0

'ex

0
OH

Frontal

Parietal

Postorbit

__ i_

CO -f-

oo w

>— > 1— 1

£u w £
„ ex cxxi
2 o o 3

3 <U <U oo
O" i- >-i
OO OH OH

Quadrate

Maxilla

Premaxil
into sn

4-5 Nasa

3 Extrasc

Eusthenopte
1

Frontal

Parietal

Postorbital

Dermospher

Intertempor

Supratempo

13
£

13 J2
c* y

Squamosal
Preopercula

Quadratojuj

Maxilla

Premaxilla

3 Nasals

3 Extrascapi

» 5

l-l
CO

C 0

.2 "^

Primitive
Actinopteryg
(Mimia, Moy

Frontal

Parietal

Dermospheni

Intertempora

Supratempor

13
|

_ I-H

CO -f-

oo y

3
*— > i— 1

Preopercular

Quadratojug

Maxilla

Premaxilla

oo
CO

2-4 Extrasca]

324 B. G. GARDINER

Behind the parietals and postparietal of actinopterygians, actinistians, rhipidistians and
dipnoans lie the transverse series of extrascapular bones. In actinopterygians this series
frequently consists of a single pair of bones (e.g. Cheirolepis, Mimia, Amia, Elops). In other
actinopterygians it comprises two pairs of bones (Polypterus, Moythomasia, Lepisosteus),bu\. in
Acipenser it is made up of a lateral pair with a much larger median element. A pattern similar to
that of Acipenser is found in most osteolepids, porolepids, onychodonts and primitive
actinistians (Diplocercides, Rhabdoderma) . In later actinistians the number of lateral
extrascapulars increases; thus in Diplums there are three lateral pairs and in Latimeria there are
four pairs as well as the median element. The condition in dipnoans is less clear, although there
appears to be a median element and at least one pair of lateral bones in primitive forms
(Griphognathus, Chirodiptems, Dipterus, Scaumenacia, Jarvik 1968; Miles 1977). An
additional pair of scale bones which are not canal-bearing occurs in Griphognathus and
Chirodipterus , whereas in Dipnorhynchus there are three pairs of canal-bearing lateral bones
and a median element (Thomson & Campbell 1971). Outside the osteichthyans a single pair of
scale-like extrascapulars (or postnuchals) has been recorded in a few advanced placoderms such
as the coccosteids Miller osteus and Dicksonosteus , and a median extrascapular is said to be
present in the actinolepid Sigaspis (Goujet 1973). In the coccosteids the extrascapulars are
traversed by the supratemporal commissure. Although the placoderms show a third pattern of
dermal roofing bones, little of this pattern matches either of those seen in osteichthyans.
However, the presence of large dermal roofing bones with descending laminae and contained
tubular sensory canals in at least the presumed primitive placoderms (Miles & Young 1977) is
considered synapomorphous for a group including placoderms and osteichthyans.

2. Sensory canals of skull roof

The supraorbital canal joins the infraorbital canal above the eye in most living actinopterygians
(Polypterus, Acipenser, Polyodon, Amia, Lepisosteus, many teleosts). Nevertheless the two
canals arise separately and in Amia the supraorbital canal anastomoses with the infraorbital by
the penultimate primary pore and then continues back onto the parietal (Allis 1889). The two
canals remain separate in many primitive actinopterygians such as Cheirolepis, Mimia,
Moythomasia and Elonichthys, as well as in more advanced forms like caturids, pachycormids
and Leptolepis.

The canals join behind the eye in actinistians (e.g. Nesides, Rhabdoderma, Whiteia, Diplurus,
Latimeria) and in all described osteolepiforms (e.g. Eusthenopteron, Osteolepis) , porolepiforms
(e.g. Holoptychius, Glyptolepis, Porolepis) and Powichthys. In dipnoans the canals are separate
in Uranolophus and Dipnorhynchus, but join behind the eye in almost all other forms
(Griphognathus, Chirodipterus, Dipterus, Fleurantia, Neoceratodus, Protopterus) . Again in
many fossil tetrapods (Trematosaurus, Lyrocephalus, Batrachosuchus) the canals remain
separate, but in others like Trimerorhachis, Metoposaurus and in living genera such as Pelobates
they join behind the orbit. In most selachians (e.g. Chlamydoselachus, Mustelus, Torpedo) and
holocephalans (e.g. Callorhynchus, Chimaera), although the two canals invariably develop
independently (Rudd 1920, Holmgren 1940), they join behind the eye in the adult (Garman
1888). In Laemargus (Somniosus), however, Garman (1888) concluded that the two canals
remained separate, but Ewart (1895) showed that the infraorbital and supraorbital canals open
to the exterior by a common pore. They agreed that the remaining lines terminated
independently on the top of the head in Laemargus. In placoderms the infra- and supraorbital
canals or lines remain separate in such diverse forms as Lunaspis, Holopetalichthys,
Romundina, Arctolepis and Leiosteus, whereas they may join in a somewhat unusual fashion in
Coccosteus and Ctenurella (Miles & Young 1977). The pattern of the pit-lines on the head of
arthrodires and phyllolepids, where three sets of lines converge (supraorbitals, infraorbitals/
central canal, posterior pit-lines) is very similar to that in the selachian Laemargus. This
similarity may be regarded either as a synapomorphy of placoderms and Laemargus or as the
retention of the primitive gnathostome condition. The presence of three pairs of converging
pit-lines on the parietals of actinopterygians (and their innervation, Allis 1922, Pehrson 1947)
and on the posterior parietals of many fossil dipnoans (Griphognathus, Chirodipterus, Miles

RELATIONSHIPS OF PALAEONISCIDS

325

Exsc

stc

Fig. 87 Moythomasia durgaringa Gardiner & Bartram. Extrascapulars and post-temporals in dorsal

view, from BMNH P.53221.

1977: figs 113, 116; Rhinodipterus, Scaumenacia) convinces me that the latter view is the more
likely.

In acanthodians the infra- and supraorbital canals always remain separate (e.g. Euthacanthus,
Ischnacanthus, Diplacanthus, Homalacanthus, Acanthodes, Watson 1937), and in Diplacanthus
the infraorbital canal is continued up on the top of the head as the central sensory line. In this
respect Diplacanthus resembles placoderms and Laemargus and this must be the primitive
condition. Thus we may conclude that primitively in gnathostomes the infraorbital and
supraorbital canals were separate.

Lower jaw

Mimia toombsi

The Meckelian cartilage is ossified throughout its length in presumed older individuals. In less
well ossified specimens there are two perichondral ossifications, one anteriorly and one
posteriorly. In others the perichondral covering is complete apart from the glenoid fossa and
there are endochondral cores anteriorly and posteriorly. These ossifications are the
mentomeckelian and articular bones. Separate mentomeckelian and articular bones are only
distinguishable in a few specimens (cf. BMNH P. 56473) and even so the mentomeckelian
usually has the two anterior coronoids closely applied to its medial surface.

Where ossification is complete all the exposed surfaces of the Meckelian bone are
perichondrally ossified except in the glenoid fossa. The bulk of the articular region is formed of
dense endochondral bone. Posterodorsally the glenoid fossa is represented by two distinct
depressions which match the double condyle of the quadrate. Posteriorly and ventrally the

326

B. G. GARDINER

Fig. 88 Skull roofs. A, Kentuckia deani (Eastman) (from Rayner 1951); B, Moythomasia nitida
Gross (from Moy-Thomas & Miles 1971); C, Pteronisculus magnus (Nielsen) (from Nielsen 1942);
D, Polypterus sp. (from Schmalhausen 1968); E, Porolepis brevis Jarvik (from Jarvik 1972); F,
Eusthenopteron foordi Whiteaves (from Jarvik 19676); G, Eusthenopteron, composite, showing
variations in bone patterns (from Jarvik 19676); H, composite showing maximum number of
separate bones observed in Osteolepis, Thursius, and Gyroptychius (from Jarvik 1948); I,
Holoptychius sp. (from Jarvik 1972). Scale bones at back of braincase omitted. From Rosen et al.
(1981).

articular is free of any dermal bone investment, as is the greater part of the ventromedial face of
the Meckelian bone.

There is a distinct groove (gmand.ext.VII, Fig. 91) behind the lateral corner of the glenoid
fossa. The groove continues anteriorly on the medial surface of the dentary. Several foramina on
the dentary open into this groove, and presumably served for the innervation of that section of
the mandibular sensory canal. An identical groove has been recorded in Pteronisculus (Nielsen
1942: figs 38, 40, sm). By comparison with Polypterus (Allis 1922) the groove is presumed to

RELATIONSHIPS OF PALAEONISCIDS

327

Fig. 89 Skull roofs. A, Ichthyostega sp. (from Romer 1966); B, Scaumenacia sp. ; C, Griphognathus

whitei Miles (from Miles 1977).

have carried the external mandibular branch of the facial nerve. Three or four large foramina
(fmand. V, Fig. 91) in the ventral margin of the Meckelian bone, anterior to the junction of the
angular and dentary , presumably served for the passage of branches of the trigeminal nerve from
the adductor fossa into this groove, as in Polypterus.

On the medial face of the Meckelian bone and a little way in front of the top of the previously
mentioned groove is a distinct foramen (fmand. int. VII, Fig. 91). This foramen leads into a canal
which passes anterodorsally between the prearticular and the perichondral covering of the
Meckelian bone and then between the posterior coronoid and the perichondral covering of the
Meckelian bone. The canal finally opens into the groove between the coronoids and the
dentigerous edge of the dentary, as in Polypterus. The canal therefore must have transmitted
the internal mandibular branch of the facial nerve. A similar canal has been described in
Pteronisculus, Birgeria (Nielsen 1942, 1949) and other palaeoniscids (Poplin 1974: fig. 40).

Anteriorly the Meckelian bone is densely ossified in the region of the mentomeckelian
ossification. The mentomeckelian bone extends back beneath the coronoids where it merges
indistinguishably with the anterior end of the articular. Beneath the coronoids the
mentomeckelian bone is ridged in an anteroposterior direction. This ridging is presumed to
represent the area of origin of the geniohyoideus muscle, this being precisely its area of origin in
Polypterus (Allis 1922: 255). The same ridges may also be for the intermandibular muscles. In
the same region of Pteronisculus (Nielsen 1942: 165) there is a series of shallow depressions.

The external surface of the mandible is composed of two dermal bones, the dentary and the
angular. Together they form the outer boundary of the adductor fossa. The angular forms the
hind margin of the mandible and is roughly triangular in outline . The mandibular canal pierces it
from end to end, passing through the radiation centre which is marked by a slot-like pit-line. In
its posterior margin the angular has a small depression (pchl, Fig. 90). This is presumed to have
been the origin of the ceratohyal ligament, as in Polypterus (Allis 1922: 246). A similar ligament
in Lepisosteus (Wiley 1976) has its origin on the retroarticular. Anterodorsally the angular is
joined to the dentary by an interdigitating suture. The angular and dentary in this region are

328

ffi
z
S

o
4:

1
CQ

I
s

—

329

-i 6
o

TJ
<U

330

B. G. GARDINER

•a
u

oft-

c c
« 2

o
£ o

OJD

- t-l

, o

£ £

03 C

<U U

.£ '>

*£

RELATIONSHIPS OF PALAEONISCIDS 331

devoid of ornamentation and the triangular area of smooth bone represents exactly that part of
the mandible permanently overlapped by the maxilla. The radiation centre of the angular lies
near the posterodorsal corner.

The dentary is very long and its anterior end curves medially. It bears on its dorsal edge a row
of sharp, stout teeth, and outside this row there are numerous smaller teeth. The larger teeth
possess an apical cap of acrodin (PI. 4, c). The dentary overlaps the angular posteroventrally and
is traversed for almost its full length by the mandibular sensory canal. The canal enters the
dentary at its most ventral point of contact with the angular and from there the canal rises at a
low angle up through the dentary, changing direction somewhat anteriorly. Where the canal
changes direction there is a group of five pores and this represents the radiation centre. Two of
these pores form a pit-line and in other specimens a distinct anterodorsal slot is developed in this
region. A dentary pit-line has only been recorded elsewhere in Polypterus (Jarvik 1947: fig. 1 A).
An oral sensory canal is found in the surangular of dipnoans.

By far the largest dermal bone on the medial side of the mandible is the prearticular, which
covers the dorsal and much of the medial face of the Meckelian bone in the region of the
adductor fossa. Posteriorly this bone ends on the lateral face of the Meckelian bone just behind
the glenoid fossa. It extends anteriorly for half the total jaw length, much as in Pteronisculus .
The prearticular is gently rounded dorsoventrally, with its dorsal surface forming a horizontal
lamina in front of the adductor fossa. This lamina contacts the horizontal lamina of the dentary
laterally, while anteriorly it bears a groove which is continuous with that on the coronoids (gr,
Fig. 91). The groove is a mirror image of a similar groove on the dermopalatines. The whole of
the outer surface of the prearticular is covered by a shagreen of small rounded teeth similar to
those on the palate. The radiation centre lies dorsally, immediately anterior to the adductor
fossa. Anteriorly the prearticular is joined to the fourth coronoid by a deeply interlocking zigzag
suture. Similar sutures are present between successive coronoids. The coronoid series consists
of four bones of which the posterior is the largest. The third and fourth coronoids are of the same
general shape with a gently rounded medial lamina and a grooved, stouter horizontal lamina.
Their radiation centres lie in the middle of the bones in the anteroposterior groove. They are
covered with a similar shagreen of rounded teeth as is found on the prearticular. The two
anterior coronoids are invariably closely applied to the underlying mentomeckelian bone and
the first coronoid does not possess a medial lamina. Instead the tuberculated lamina of the
second coronoid (Fig. 92) is produced anteriorly and overlies the posteromedial surface of the
first coronoid. A similarly-situated bone in the anterior part of the mandible of Pteronisculus
(Nielsen 1942: figs 38, 39, 40, Mmd), said by Nielsen to contain both an endochondral and a
dermal component, may be reinterpreted as the medial lamina of a coronoid fused to the
underlying mentomeckelian, as in Mimia.

The two anterior coronoids have, in addition to their small teeth, a row of larger, acutely
pointed teeth along the outside of the aforementioned groove.

Moythomasia durgaringa

The mandible of Moythomasia is similar to that of Mimia but differs in the presence of a
supra-angular.

Par

2mm

Fig. 93 Mimia toombsi Gardiner & Bartram. Dermal tooth plates of the right lower jaw of an
incompletely ossified individual in medial view, from BMNH P. 56473.

332 B. G. GARDINER

The prearticular consists of two ossifications: there is a separate, much smaller, posterior
ossification overlying the entrance to the canal for the internal mandibular branch of the facial
nerve. The prearticular teeth are confined to the posterior and dorsal margins of the first
prearticular and the medial laminae of the coronoids are mostly toothless. The supra-angular is a
stout ossification which forms the lateral border of the adductor fossa. It overlaps the dentary
and prearticular anteriorly and the angular ventrally. Posterodorsally it is attached to the
articular. All four coronoids bear a row of much larger, acutely pointed teeth. These are
continuous with a few similar teeth on the anterior portion of the first prearticular.

Lower jaw: discussion

1. Meckelian ossifications

In presumed juveniles of Mimia and Moythomasia a thin perichondral sheath is present round
the anterior and posterior ends of Meckel's cartilage, much as in Acanthodes (Miles 1973a,
Jarvik 1977). Later an endochondral core forms in the articular posteriorly and in the
mentomeckelian bone anteriorly. Eventually these two ossifications meet and the whole
cartilage is endochondrally and perichondrally ossified. It is then usually referred to as the
Meckelian bone. A fully-developed Meckelian bone is characteristic of most adult
palaeoniscids, including Mimia, Moythomasia, Pteronisculus and Boreosomus, of Australoso-
mus, of osteolepids such as Eusthenopteron and Panderichthys , porolepids such as Glyptolepis,
and primitive dipnoans (Griphognathus, Chirodipterus, Holodipterus, Melanognathus,
Dipterus). Two discrete ossifications separated by cartilage are characteristic of Polypterus,
parasemionotids, Pholidophorus germanicus, fossil actinistians such as Rhabdoderma,
Diplocercides, Coelacanthus and Coccoderma, and tetrapods such as Ichthyophis, Cryptobran-
chus and Lacerta. But in Lepisosteus, most teleosts and Latimeria the posterior end of Meckel's
cartilage possesses two ossifications, the articular proper and the retroarticular, whereas in
Amia there are three (two articulars and a retroarticular). There are also two posterior
ossifications (in tandem) in several fossil actinistians (Macropoma, Laugia, Whiteia), but in
larger specimens of Whiteia and in Rhabdoderma there is only a single ossification (P. L. Forey,
personal communication). Nelson (1973) concluded that the presence of a discrete articular and
retroarticular was a plesiomorphic character of actinopterygians, whereas Patterson & Rosen
(1977: 129, character 19) considered an independent retroarticular the derived condition. Yet
even if we allow that the articular bone in fossil actinistians is developed from two ossification
centres (articular, retroarticular) which may fuse during ontogeny, its single nature in
Polypterus and Recent amphibians suggests that this is the primitive adult osteichthyan
condition.

In placoderms the Meckelian cartilage ossified perichondrally in two regions, as in
acanthodians and presumed juvenile Mimia. In placoderms and acanthodians the cartilage
frequently calcifies, in placoderms invariably as globular calcified cartilage. They are referred to
as the mentomandibular and articular ossifications, and are characteristic of many arthrodires
(Stensio 1963a, Miles 197 Ib) and of Ctenurella. Thus two principal ossification centres in
Meckel's cartilage, one anterior and one posterior, are probably synapomorphous for a group
containing acanthodians, placoderms and osteichthyans, and furthermore these two centres
presumably correspond to the hypobranchial and ceratobranchial ossification centres.

2. Dermal bones

The dermal bones of the lower jaw are more numerous in primitive actinopterygians than in
later teleosts. The outer surface of the jaw is composed of two dermal bones in Polypterus,
Cheirolepis and Mimia. These bones, in sequence the angular and the dentary, are
canal-bearing. In many palaeoniscids and primitive neopterygians there is a third bone, the
supra-angular, lying dorsal to the angular, sandwiched between it and the back of the dentary.
On the inner surface of the jaw there is a further coronoid series, which in Mimia and
Moythomasia comprises a prearticular (double in Moythomasia) and four coronoids, and in
Polypterus a prearticular (=splenial) and two coronoids. A similar dermal bone pattern is
encountered in most palaeoniscids (Pteronisculus, prearticular, three coronoids), Ospia

RELATIONSHIPS OF PALAEONISCIDS 333

(prearticular, two coronoids), Lepisosteus (prearticular, two coronoids), Atractosteus
(prearticular, three coronoids) and Amia (prearticular, five coronoids). Pearson & Westoll
(1979) record a supra-angular in Cheirolepis based on a single specimen (RSM 1877.30.5, figs
lOb, lie), but the bone in question is more probably a branchiostegal ray.

Neither a supra-angular nor a coronoid series is present in any living teleost (Nelson 1973); in
sturgeons the only dermal ossifications are a prearticular and a dentary whereas in Polyodon
only the dentary is present. The mandibular canal does not penetrate the dentary in Polyodon
and is missing altogether in sturgeons.

Coronoids are present in halecomorphs, Pachycormus, Ichthyokentema and Pholidophorus
higginsi, but absent in other pholidophorids and in leptolepids (Patterson 1977«). A
supra-angular and a prearticular are present in all of these apart from leptolepids other than
Proleptolepis (Patterson & Rosen 1977: fig. 32A), which retains a supra-angular. Both bones are
missing in Recent teleosts. In advanced teleosts the angular and articular bones are co-ossified
(Patterson I917b), but in some leptolepids they fuse during ontogeny, as do the mento-
meckelian bone and coronoids of Mimia and Moythmasia.

From this analysis I conclude, like Patterson (1982), that the supra-angular has been acquired
within the actinopterygians and is primitively absent in Cheirolepis, Polyptems and Mimia.

The presence of a mandibular sensory canal within the dentary bone is unique to
actinopterygians (Stensio 1947). In other osteichthyans this canal runs through an independent
splenial series. The splenial series consists of several bones, two in actinistians, four in
osteolepiforms, porolepiforms, onychodonts, primitive dipnoans and early tetrapods. The
single canal bone (angular) in actinopterygians must be part of this series. The splenial series,
from the back forwards, are termed surangular, angular, splenial (postsplenial) and presplenial.
The surangular is a canal-bearing bone and is therefore unlikely to be homologous with the
actinopterygian supra-angular (Nelson 1973). It carries the mandibular canal and part of the oral
canal in osteolepiforms (Eusthenopteron, Jarvik 1947) and porolepiforms (Holoptychius, Jarvik
1972). In Devonian dipnoans it carries the oral canal and in primitive tetrapods such as
temnospondyls (Nilsson 1943, 1944) and anthracosaurs (Panchen 1972, 1977) it is grooved by
both the mandibular and oral canals. In actinistians such as Rhabdoderma (Forey 1981) and
Latimeria the posterior bone, which embraces the articular, is called the angular, yet it contains
both the mandibular sensory canal and the oral pit-line, much as does the angular in Mimia
and Polypterus. Thus I conclude that the angular of actinopterygians and actinistians is
topographically homologous with the surangular of Eusthenopteron, Holoptychius, dipnoans
and tetrapods. Furthermore, the large bone at the back of the lower jaw in later dipnoans (cf .
Neoceratodus) is better interpreted as a surangular (in the traditional sense) rather than an
angular as Thomson & Campbell (1971) and Miles (1977) have regarded it.

In actinistians the large 'angular' is followed by a small splenial and a very small dentary. The
dentary, in all but the Devonian forms, bears separate toothplates. There are five coronoids and
a large prearticular in Latimeria; three of the coronoids lie above the prearticular. In
Macropoma and Whiteia (P. L. Forey, personal communication) there are four coronoids, a
large one above the prearticular and three anterior to it. In onychodonts there is a single
coronoid with a whorl of teeth much as on the anterior coronoid of Holoptychius.

The lower jaws of Eusthenopteron and Holoptychius (Jarvik 1972) are very similar, and the
dentary reaches the articular to form the dorsal margin of the adductor fossa, as in Cheirolepis.
There are four coronoids, three lying above the prearticular and one anterior to it (para-
symphysial plate of Jarvik 1972).

The dipnoan dentary is reduced in size, as in actinistians (Miles 1977: 217) and there is a single
median toothplate at the symphysis (adsymphysial plate, Miles 1977). A median adsymphysial
plate is an apomorphy of dipnoans. In Neoceratodus larvae there is also a separate coronoid on
either side of the adsymphysial plate (Semon 1899: pi. 20). The surangular usually forms the
margin of the adductor fossa, as in actinistians and tetrapods, and the oral canal passes through
the surangular into the dentary in Chirodipterus, Holodipterus and Dipnorhynchus . A full oral
line is found in acanthodians, chondrichthyans such as Chlamydoselachus, dipnoans and
amphibians (Stensio 1947) and is probably a primitive gnathostome character. But only in

335

60 o
•S ^

5S >
1° -

ve
es

336 B. G. GARDINER

dipnoans and tetrapods (Nilsson 1943, 1944, Panchen 1972) is it ever associated with, the
dentary. The lower jaws of temnospondyls and anthracosaurs possess three coronoids, but those
of Lissamphibia are characterized by the absence of the splenials and angular (= surangular of
most authors). The lower jaw of anthracosaurs is characterized by two large meckelian
fenestrae, separated by the postsplenial bone.

In acanthodians the ventral margin of Meckel's cartilage frequently sits in a groove along the
dorsal margin of the so-called mandibular bone (styliform process of Hancock & Atthey 1869;
extramandibular spine of Reis 1890, 1895). The lateral surface of this bone is often ornamented
(Miles 1966), confirming its dermal origin. The medial surface of the dentary of Polypterus is
similarly deeply grooved along its entire length. The mandibular bone stretches the whole length
of Meckel's cartilage in Acanthodes (and projects anteriorly beyond it in many specimens) but in
Mesacanthus it is relatively shorter. It has been homologized with the splenials of osteichthyans
(Jaekel 1899, Dean 1907) and temnospondyls (Stensio 1947). A mandibular bone is present in
many members of the Diplacanthidae, Ischnacanthidae and Acanthodidae, but its absence in
certain Devonian Acanthodidae and the Lower Devonian Ischnacanthus has prompted Denison
(1979) to consider this the primitive condition. The mandibular canal runs ventral to the
mandibular bone; nevertheless from its topographic position, stretching as it does from the
articular to the end of the mentomeckelian, it is better homologized with the splenial of
osteichthyans than with the dentary. From its distribution with the acanthodians I consider it
synapomorphous for a group containing acanthodians and osteichthyans. Both Miles (19710)
and Denison (1979) considered the function of the mandibular bone was to stiffen the Meckelian
cartilage.

Teeth when present in acanthodians occur in three forms; as single teeth, spirals or whorls, or
fused to dermal jaw bones (Denison 1979). Thus numerous small, single teeth were situated in
the lining of the mouth of many ischnacanthids and there was a well-formed, large, lower
median symphysial tooth whorl (cf. Chlamydoselachus). Mandibular toothplates are also found
in many ischnacanthids and Denison (1979) proposed that these are a unique derived character
of Ischnacanthidae. Many attempts have been made to elucidate the structure of these
toothplates (0rvig 1957, 19670, 1973; Gross 1957; Miles 1966) and it is now generally agreed
that they consist of tooth-cusps anchylosed to a supporting bony base. There is no clear
boundary between the dentinous tissue of the teeth and the basal bone tissue. The teeth are
described as stephanodont (Jaekel 1919) and said to be composed of dentine or dentinous tissue
(0rvig 1973). The underlying bone may be cell-bearing or acellular, fairly dense with vascular
canals or cancellous. The teeth are arranged in one or more longitudinal rows and are frequently
worn by use. In specimens of Nostolepis there are worn tooth-cusps alternating with shearing
edges formed by the abraded rows of side-cusps (0rvig 1973). As in placoderms these worn
teeth are never replaced by subsequent teeth in the same positions. The principal tooth-cusps in
Nostolepis are said to be made up of an external layer of pallial dentine (Gross 1957) or pallial
mesodentine (0rvig 1973). The side cusps of the multicuspidate teeth of Nostolepis also consist
of mesodentine. The cores of the teeth are of osteodentine and osteomesodentine. Mesodentine
is also said to be present in osteostracans (0rvig 19670).

In rhenanid placoderms the outer face of the palatoquadrate and Meckelian cartilage is
covered by small tesserae with stellate tubercles, in stensioellids by small denticles, and in
pseudopetalichthyids by small scale-like plates. Dermal jaw bones are unknown in
Acanthothoraci, Petalichthyida and Phyllolepida, but Ptyctodontida, Arthrodira and Antiarcha
have well-developed dermal jaw bones (Denison 1978). In arthrodires there is a large dermal
toothplate or inferognathal associated with the Meckelian cartilage. Anteriorly in Eastman-
osteus (Gardiner & Miles 1975) the inferognathal sits over the perichondral mentomandibular,
but posteriorly it rests on the medial surface of the articular. The inferognathals in
branchythoracids carry teeth (Zahntuberkeln of Gross 1957) which are continuous basally with
the adjoining bone tissue and may therefore be considered stephanodont as in acanthodians
(Jaekel 1919). These teeth are simply cusps on the bone surface which become worn away during
life (Miles 19710). New teeth are said to be added to one end of a tooth row as the gnathals grow
at their bases. The teeth consist externally of semidentine surrounding an osteosemidentine

RELATIONSHIPS OF PALAEONISCIDS

337

core. In some arthrodires (Sedowichthys, Mylostoma, 0rvig 1967 'a, 1973) and the
acanthothoracid Romundina, the thickened ridges consist of mesodentine, as in many
acanthodians. On abraded plates the osteosemidentine is said to have worn away and the biting
area to consist of dense bone. The most prominent teeth in arthrodires are seen on the
superognathal of the phlyctaeniid Dicksonosteus , where they are very markedly tubercular. In
antiarchs the inferognathals have a broad biting anterior portion and a slender posterolateral
ramus.

In ptyctodonts there is a large crushing or sectorial toothplate. This plate may have a shearing
edge, as in Rhamphodopsis , or a large central tritural area, as in Ptyctodus, or separate tritural
cusps as in Palaeomylus (Miles 1971a). These toothplates all have dense bone beneath and are
strengthened by inwardly-growing hypermineralized columnar tissue (secondary dentine of
Gross 1957; osteosemidentine of 0rvig 1973) which in Ptyctodus forms the tritural areas as the
outer layers of normal dentinal tissue become won away. A similar columnar tissue is developed
in brachythoracid arthrodires, in holocephalans and in dipnoans (Miles 19710, 0rvig 1973). In
arthrodires and ptyctodonts the osteosemidentine is surrounded by semidentine and
occasionally by mesodentine. In some ptyctodonts the gnathals have an outer layer of
orthodentine and an inner mass of trabecular dentine. In this latter respect and in the similar
presumed growth pattern of the stephanodont teeth, placoderms resemble acanthodians

Prscl

Rbr

Fig. 97

2mm

Mimia toombsi Gardiner & Bartram. Opercular, branchiostegal and gular bones and
presupracleithrum, drawn as if folded out in one plane, from BMNH P. 56495.

338

B. G. GARDINER

(0rvig 1973). In holocephalans the dermal toothplates arise as whole units in the dermis of the
jaws, but in Neoceratodus (Kemp 1977) the plates develop from simple groups of isolated cusps
which eventually fuse in ridges. All this evidence suggests either that dermal toothplates
associated with the dorsal surface of the Meckelian cartilage are a primitive feature of
gnathostomes, or that they have been independently developed in holocephalans and in
acanthodians, placoderms and osteichthyans. The latter view is the more parsimonious.

The only other dermal bones said to be associated with the lateral and ventral faces of the
Meckelian cartilage are the infraprelateral and mandibular plate (s) found in Bothriolepis
(Stensio 1931, 1948, Denison 1978). In a specimen of Bothriolepis (BMNH P. 50898)
demonstrated to me by R. S. Miles, the infraprelateral is sutured to the prelateral and is clearly
part of the cheek, and the remaining mandibular plate(s) described by Stensio (1931, 1948) are
too incompletely known to comment on.

Operculogular series

Mimia toombsi

The opercular is four-sided with a convex posterior margin. Ventrally it overlaps the dorsal
margin of the subopercular. The centre of radiation of the opercular lies in the anterodorsal
corner. Internally, just below this centre, there is a small cup-shaped depression (see
Moythomasia, dop, Fig. 99) which as in Polypterus (Allis 1922: pi. 11; fig. 33) is presumed to
have housed an opercular cartilage.

The subopercular is rectangular and its radiation centre is near the anterior margin. The
anterior margin fitted beneath the posterior edge of the preopercular and did not articulate with
the hyomandibula as it is said to have done in Pteronisculus (Nielsen 1942).

There are twelve branchiostegal rays which diminish in size anteriorly, each ray overlapping
the one posterior to it. Several of the branchiostegal rays have a pronounced anterior projection

fbmand.ext .VI I

2mm

Fig. 98 Momia toombsi Gardiner & Bartram. Gular plates in dorsal view, from BMNH P. 56495.

RELATIONSHIPS OF PALAEONISCIDS 339

(hyoid process) devoid of ornament (prh, Fig. 97), which is also seen in Cheirolepis (Pearson &
Westoll 1979: 363) and may have been inserted along the ceratohyal as in Recent forms. The first
branchiostegal ray, although somewhat larger than the second, neither overlapped nor underlay
the lateral gular. The lateral gulars are much larger than the preceding branchiostegal rays.
Posteriorly the left lateral gular overlaps the right. The rhombic median gular in turn overlaps
the anterior medial edges of the lateral gulars. The radiation centres of the three gulars lie near
the centres of the bones, immediately beneath the pit-line; those of the branchiostegals lie much
nearer their mandibular margins. Each gular has a slot-shaped pit-line (gpl, Fig. 97) in the form
of an arc. On the internal surface there is a corresponding cluster of foramina for branches of the
external mandibular nerve (fbmand.ext. VII, Fig. 98). Similar pit-lines are present on the lateral
gulars of Polypterus (Pehrson 1947, Jarvik 1947) and Pteronisculus (Nielsen 1942), and on the
median gular oi Amia.

It will be convenient to deal with the presupracleithrum (Nybelin 1976) at this point. This
bone lies behind the posterodorsal corner of the opercular, posterolateral to the supracleithrum.
It is overlapped anteriorly by the opercular. In Polypterus a similar bone, the posterior
postspiracular, overlaps the dorsal edge of the opercular, which is slightly grooved for it. The
same bone is present in many palaeoniscids (Cheirolepis, Pteronisculus, Boreosomus) and other
fossil actinopterans, where it has also been called the postspiracular bone (Nielsen 1942: 182).
The presence of a presupracleithrum is considered a synapomorphy of actinopterygians.

Moythomasia durgaringa

Moythomasia differs little from Mimia. The opercular is more lozenge-shaped and the
subopercular is deeper than wide. The median gular is kite-shaped and has a V-shaped pit-line
similar to that of Amia.

Operculogular series: summary and discussion

1. Branchiostegal rays and gular plates

Numerous long branchiostegal rays are found in many primitive actinopterygians. There are 12
in Cheirolepis, Mimia and Moythomasia, 20 in Cosmoptychius and more than 30 in Tegeolepis.
In Polypterus there are no branchiostegal rays and Polyodon has only a single pair. Lepisosteus
has three rays, Amia 11 and in higher teleosts there are rarely more than eight. The lateral gulars
are in series with the branchiostegal rays and together with the median gular fill the area
between the jaw rami. All three gulars occur in the most primitive actinopterygians, including
Cheirolepis, Mimia, Pteronisculus and Haplolepis. Boreosomus has only median gulars (one or
two, Nielsen 1942: 349) and Polypterus only lateral gulars. Gulars are missing altogether in
sturgeons, paddlefishes, Lepisosteus and most teleosts, but a median gular is present in Amia,
pachycormids, pholidophorids, leptolepids, elopids and albulids.

Actinistians (cf . Diplocercides, Rhabdoderma, Latimeria) resemble cladistians in possessing a
single pair of lateral gulars and having no branchiostegal rays.

In osteolepiforms, porolepiforms, onychodonts and dipnoans there is an operculogular series
with lateral and median gulars as well as a so-called submandibular series. Miles (1977: 258) and
Patterson (1982) consider the submandibular series to be a specialization of dipnoans and
rhipidistians; Patterson (1982) regards the submandibulars as shortened branchiostegal rays.
Jarvik (1963), however, considered them to be pa"rt of a mandibular gill-cover and therefore not
homologous with branchiostegal rays. Jarvik (1968, 1972) further argued that the primitive
gnathostome possessed both series of bones, submandibular and branchiostegal.

In osteolepiforms such as Eusthenopteron there are eight submandibular bones with the
anterior members of the series intercalated between the gular and the mandible. There are
paired lateral gulars and a small median gular similar in size to those in primitive
actinopterygians.

In porolepiforms (Porolepis, Glyptolepis, Holoptychius, Jarvik 1972) there are 9-10
submandibulars and very large lateral gulars. The lateral gulars are large in actinistians
(Diplocercides, Rhabdoderma) and cladistians. There may be two median gulars (Porolepis) or
none (Holoptychius). The anterior median gular in Porolepis does not carry a pit-line and is best

340

B. G. GARDINER

E
Z

</T

I
I

«
<U

•O
1

I
I

RELATIONSHIPS OF PALAEONISCIDS

341

Fig. 100 Moythomasia durgaringa Gardiner &
Bartram. Gular plates in ventral view, from
BMNH P.56480.

gpi

homologized with the median submandibular of dipnoans (Jarvik 1967a) and the anterior
median gular of Boreosomus (Nielsen 1942: fig. 73). The posterior median gular carries a
pit-line in both Porolepis and Chirodipterus (Miles 1977: figs. 105, 129, sbm.m).

In fossil dipnoans the submandibulars vary greatly in size and shape. The most posterior
member (subopercular plate 2 of Griphognathus, Miles 1977; posterolateral gular of
Scaumenacia, Jarvik 1967a) looks like a branchiostegal ray, whereas the most anterior (principal
submandibular) resembles the lateral gulars. Between them is a short, triangular lateral
submandibular (Jarvik 1967a: fig. 5B; Miles 1977: fig. 126). There are usually three
submandibulars in dipnoans (Griphognathus, Chirodipterus, Scaumenacia}, counting the
anterior gular-like pair, but Rhinodipterus may have more. If submandibulars are shortened
branchiostegal rays then the anterior (principal) submandibulars are more parsimoniously
interpreted as anterior lateral gulars and as a synapomorphy of dipnoans. Neither an
operculogular series nor branchiostegal rays is present in tetrapods.

In the placoderm Bothriolepis , Stensio (1931, 1948) described a fragmentary series of dermal
bones which he suggested were associated with the lower jaw and the hyoid gill cover. These
have not been described in any other placoderm; nevertheless Jarvik (1968, 1972) has suggested
that some of these elements may belong to the submandibular series. The bones in question do
not in my estimation form a recognizable series. Elsewhere in placoderms there is a distinct
hyoid gill cover in the form of a dermohyal (submarginal, middle preoperculum, extralateral),
which is fused to the head of the hyomandibula in phlyctaeniids (Goujet 1975: fig. 4) and
coccosteids (Miles 1971&: fig. 111). The inner surface of the dermohyal is also attached to the
opercular cartilage in Holonema (Miles 19716). In some Pachyosteina (Pachyosteus,
Brachyosteus) the dermohyal is incorporated into the head shield. A large dermohyal is found
in rhenanids (Gemuendina, Gross 1963; Jagorina, Stensio 1969), pseudopetalichthyids
(Pseudopetalichthys, Paraplesiobatis, Gross 1962), petalichthyids (Lunaspis, Gross 1961) and
arthrodires.

The association of the dermohyal with the hyomandibula in placoderms and actinopterygians
forces me to regard the dermohyal as a synapomorphy of a group including placoderms and
actinopterygians.

In acanthodians, branchiostegal rays are widespread with up to 25 overlapping rays
in Euthacanthus and Mesacanthus (Watson 1937), and 10 or fewer in Climatius and

342

B. G. GARDINER

Exsc

Prscl

Dspo

Scl

ano

Pel

Clm

An Clav
Fig. 101 Mimia toombsi Gardiner & Bartram. Restoration of skull in lateral view.

Br achy acanthus. In Homalacanthus (Miles 1966) there are some 17 widely-spaced rays behind
the 'preopercular' and in Acanthodes there are more than 20 very weak rays associated with the
ceratohyal (Miles 1973a: pi. 6). In this latter genus they are too short to have covered the gills
completely.

In summary, long branchiostegal rays are found in the hyoid operculum of acanthodians and
actinopterygians and, as Patterson (1982) has argued, must be synapomorphous for a group
containing acanthodians and actinopterygians. The anterior branchiostegal rays are in series
with the lateral and median gulars in primitive actinopterygians, and gular plates are considered
to be a synapomorphy of osteichthyans. Shortened branchiostegal rays (submandibulars) which
have lost all contact with the certohyal are regarded s synapomorphous for carcopterygians.

2. Opercular cartilages and opercular bones

An opercular and subopercular are found in almost all non-tetrapod osteichthyans, where they

are generally believed to be enlarged branchiostegal rays.

Opercular cartilages are found in chondrichthyans, placoderms and osteichthyans. In
selachians there are numerous hyoid ray cartilages and in Scy Ilium, for example, three are
associated with the hyomandibula. In chimaeroids a single large opercular cartilage supports the
opercular rays. A single large opercular cartilage is also characteristic of such diverse
placoderms as Jagorina, Brindabellaspis and Bothriolepis , and in Holonema it helps support a
dermohyal.

Opercular cartilages are spasmodically distributed throughout actinopterygians. Thus, a
small opercular cartilage is seen in larval Polypterus and Anguilla, and in the adult Polypterus it
lines the articular facet of the opercular bone as in Heterotis and Elops (Patterson 19776: 90). In

RELATIONSHIPS OF PALAEONISCIDS

343

ano

Dspo

Lac

Pmx

Rbr

Clm

Clav

2mm

Fig. 102 Mimia toombsi Gardiner & Bartram. Restoration of skull in anterior view.

adult Anguilla (Norman 1926: 298), on the other hand, the opercular cartilage only contacts the
dorsal margin of the opercular. A large opercular cartilage occurs in several teleosts including
saccopharyngoids (Harrisson 1966: 451).

Latimeria also possesses a large opercular cartilage (Millot & Anthony 1958) which contacts
the ventroposterior margin of the opercular. In Lepidosiren and Neoceratodus (Bridge 1898)
there are two opercular cartilages, one attached to the posterior inner surface of the opercular
and the other to the posterior inner surface of the subopercular.

Opercular cartilages are unknown in acanthodians. An opercular and a subopercular are
believed to be synapomorphous for osteichthyans.

In halecostomes (Amia, teleosts) there is an additional bone in the operculogular series, the
interopercular. Although out of line with this series the interoperocular is probably a modified
branchiostegal ray and is considered synapomorphous for halecostomes (Patterson 1973).
(Re-examination of Platysiagum has convinced me that the bone identified as an incipient
interopercular by Brough (1939) and an interopercular by me (Gardiner 1960) is no more than a
displaced branchiostegal ray.)

A further ossification, structurally part of the operculum, is often found in the anterodorsal
corner. This bone is referred to either as the antopercular or as the postspiracular (Nielsen 1936:
42). A single antopercular is present in Pteronisculus, Commentrya, Brachydegma, Pygopterus,
Perleidus and Redfieldius and in the osteolepiform Eusthenopteron. Two antoperculars are
found in Boreosomus, Platysiagum and Polypterus.

B. G. GARDINER

Pel

Sop

1 San

Clav

Clm

Fig. 103 Moythomasia durgaringa Gardiner & Bartram. Restoration of skull in lateral view.

A few primitive actinopterygians also possess accessory operculars in the anteroventral
corner of the operculum: these include Cheirolepis, Watsonichthys, Cosmoptychius, Kentuckia
and Gonatodus.

Hyoid and branchial arches

Mimia toombsi

The hyoid arch includes a hyomandibula, dermohyal, interhyal, ceratohyal and hypohyal.

The hyomandibula is a large, stout, gently curved bone which articulates dorsally with the
otic region of the neurocranium. Distally it articulates with the interhyal. The hyomandibula is
a solid structure of endochondral bone with a small triangular dermohyal intimately fused with
its dorsal shank (PI. 1; Fig. 104). In cross section it is rounded laterally, but medially a
well-marked gutter runs across the bottom of the dorsal shank and continues ventrally to the
posterior margin. Toothplates clothe the anteromedial surface in front of the gutter, while much
smaller plates occur on the anterior margin of the dorsal shank. The disposition of these
toothplates is very similar to that described for Eusthenopteron by Jarvik (1954: fig. 16A, B). A
perichondrally-lined canal (chy, Fig. 104) passes obliquely down through the hyomandibula in
the ventrolateral direction. Medially a wide, shallow groove runs down into the mouth of this
canal and laterally the canal opens into a shallow groove on the surface of the shaft, just below
the dermohyal. This canal presumably transmitted the hyomandibular trunk of the facial nerve,
as in Amia and Lepisosteus. Other surface features of the hyomandibula include a small

RELATIONSHIPS OF PALAEONISCIDS

345

346 B. G. GARDINER

projection or notch in the posterior margin near the ventral corner of the dermohyal. This notch
may have transmitted the hyoid branch of the facial nerve, as in Polyptems (Allis 1922: pi. 17),
and as postulated in other palaeoniscids by Stensio (1925: 169) and in Eusthenopteron by Jarvik
(1954: fig. 16A, B). However, the condition in Polypterus is unique; in all other osteichthyans
where the nerve passes through the bone (Amia, Lepisosteus, generalized teleosts, Latimeria,
larval Neoceratodus) it never divides into the mandibular and hyoid branches until it has pierced
the hyomandibula (see also Moythomasia, Fig. 105). It therefore seems unlikely that it divided
into its constituent branches prior to its passage through the hyomandibula in either Mimia or
Eusthenopteron .

Both the interhyal and ceratohyal are perichondral shells lacking any endochondral
ossification. The interhyal is a small, somewhat cylindrical bone, open at each end where it
articulated with the lateral portion of the proximal end of the ceratohyal and the distal end of the
hyomandibula (Fig. 108), much as in Polypterus. It does not articulate with either the
palatoquadrate or the Meckelian cartilage, as it is said to do in Pteronisculus (Nielsen 1942:
175; fig. 42).

The ceratohyal is a stout, flat, slightly curved bone, expanded ventrally in its posterior third.
On its lateral face there is a broad longitudinal groove (see Moythomasia, ahy, Fig. 106) for the
afferent hyoidean artery. A similar groove has been reported in Pteronisculus, Australosomus
(Nielsen 1949: fig. 37), Plegmolepis, Pygoptems (Aldinger 1937: figs 19, 41B) and
Eusthenopteron (Jarvik 1954: 22; fig. 8A). A continuous toothplate runs along its dorsomedial
margin (BMNH P.53245), as in Elops.

The hypohyal, like the ceratohyal, is a perichondral ossification, open at both ends. It is a
flat, strongly curved bone which lies in the vertical plane, with its distal edge directed
posteromedially to articulate with the anterior end of the basibranchial. A few small toothplates
are found on the dorsomedial surface (Fig. 107). A stout ventral projection of the hypohyal
presumably served for the insertion of the sternohyoideus muscle (cf . Polypterus, Lepisosteus) .

Five branchial arches are present, each component of which consists of perichondral
ossifications with a weak endochondral core. The individual ossifications are presumed to have
articulated with one another by cartilaginous epiphyses. The elements are usually scattered, but
there are enough specimens with parts of the arches in position to enable precise reconstructions
to be made.

The first branchial arch consists of hypobranchial, ceratobranchial, epibranchial, infrapharyn-
gobranchial and suprapharyngobranchial. The hypobranchial is a long, slender bone with an
enlarged, inturned distal end which articulates with the anterior end of the basibranchial,
immediately behind the hypophyal (Fig. 108). Thus the first hypobranchial and hypohyal share
the same articular facet, as they do in Pteronisculus (Nielsen 1942: fig. 45) and Australosomus
(Nielsen 1949: 121), and as they partly do in Polypterus (Allis 1922: pi. 8). Ceratobranchial 1
(Fig. 108) is a very long rod, strongly arched dorsoventrally. It is semicircular in section,
with a deep, longitudinal groove on its ventral face (see Moythomasia , Fig. 115). The groove is
presumed to have carried the afferent artery and branchial nerve, as in Polypterus, Amia etc.
Proximally the edges of the ceratobranchial groove have closed up to form a foramen for the
artery and nerve as in Moythomasia (fcb, Fig. 115). A similar foramen has been described in the
Gogo dipnoan Griphognathus (Miles 1977: fig. 135). Two rows of toothplates cover the
dorsomedial surface of both the hypobranchial and ceratobranchial. The ventromedial edge of
the ceratobranchial is often regularly scalloped. The first epibranchial is about half the length of
the ceratobranchial and has a deep longitudinal arterial groove in its posterior margin. Ventrally
the lateral wall of the groove is produced posteriorly to form a partially covered canal. Dorsally
the medial wall is developed as a dorsally-directed process for the articulation of
suprapharyngobranchial 1. Toothplates clothe the lateral edge of the epibranchial and appear to
be arranged in at least two rows (Fig. 116B). The first infrapharyngobranchial is a short,
elbowed bone, which articulates with the anteriorly-directed proximal end of epibranchial 1 and
with a ventral facet on the otic portion of the neurocranium. Ventrally it is covered by a row of
small toothplates. The first suprapharyngobranchial *s a flat bone, considerably larger than the
infrapharyngobranchial. Posteroventrally it has a short process projecting beyond the articular

RELATIONSHIPS OF PALAEONISCIDS

347

*2
•5

•o
ea

348

B. G. GARDINER

2mm

Fig. 106 Moythomasia durgaringa Gardiner &
Bartram. Right ceratohyal in lateral (above)
and medial views, from BMNH P.56475.

facet. Its medial face is slightly concave where it fitted against the lateral face of the otic region of
the neurocranium, above the vestibular fontanelle and jugular canal and anterior to the exit of
the vagus nerve, as in Pteronisculus (Nielsen 1942: fig. 47). In Adpenser the first
suprapharyngobranchial articulates below the jugular vein, but the second articulates above it
(Bertmar 1959: 305).

The second hypobranchial (Figs 111, 113) is stouter and a little longer than the first. Its distal
end is broad and the articulatory facet directed dorsally . It articulated about one-third of the way
along the basibranchial by a nearly vertical facet. Toothplates clothe the dorsal surface of the
shank behind the expanded distal head. The second ceratobranchial is similar to the first, but
epibranchial 2 is two-thirds of the length of epibranchial 1 . Infrapharyngobranchial 2 is a little
smaller than the first but is similarly shaped, with articular facets at either end and small
toothplates ventrally. No articular facet for its anterior end has been observed on the braincase,

Fig. 107 Mimia toombsi Gardiner & Bartram.
Left hypohyal in (A) medial, (B) dorsal and
(C) lateral views.

1mm

RELATIONSHIPS OF PALAEONISCIDS 349

but by comparison with Acipenser it is likely that it articulated with the ventral margin of the
occiput. The second suprapharyngobranchial is a smaller version of the first. It is presumed to
have articulated with (or pressed against) the occipital region, as in Acipenser.

The third arch has a very short hypobranchial and epibranchial and does not contain either a
supra- or infrapharyngobranchial. The head of the hypobranchial is tapered and inturned where
it articulated with an oblique facet on the basibranchial. Ventrally it has a process similar to that
on the hypohyal, except that it lacks a perichondral covering. The third ceratobranchial is the
same length and shape as the second, whereas the third epibranchial is about two-thirds the
length of the second and has a characteristic broad flange projecting back from its posteromedial
edge. A distinct notch in the ventral margin of this flange gives a beaked appearance to the
postero ventral edge. Toothplates are confined to the lateral edge of the epibranchial.

The fourth arch, like that of Eusthenopteron, lacks an ossified epibranchial. This may be
missing altogether, or may have been cartilaginous, as in Latimeria. (I erroneously attributed
both a third infrapharyngobranchial and a fourth epibranchial to Mimia; Gardiner 1973: fig. 9).
The fourth hypobranchial (Figs 111, 112) is a short, squat ossification, shorter than the third
hypobranchial, which articulated by way of a broad horizontal facet with the extreme hind end
of the basibranchial. The fourth ceratobranchial differs fom the third only in being somewhat
shorter.

The fifth branchial arch is only represented by a ceratobranchial which is a very slender
ossification, circular in cross-section and devoid of a longitudinal groove.

The basibranchial is a sturdy ossification, very similar in shape to that of Polyp terns. The entire
anterior end forms an articular facet for the hypohyals and first hypobranchials. There are
three other oval facets for the remaining hypobranchials: that for hypobranchial 2 is almost
vertical, that for hypobranchial 3 more oblique, and the facet for hypobranchial 4 is horizontal.
The basibranchial is thickened dorsoventrally in the plane of the long axes of the facets for the
second and third hypobranchials, and is also produced ventrally beneath the facets for the
fourth. The dorsal surface is flat and supports a series of irregularly-arranged toothplates similar
to those on the epibranchials.

Although the basibranchial is frequently a single ossification there is a clear evidence of three
ossification centres in several specimens. Thus in BMNH P. 53237 there are two separate
ossifications with the junction between them lying behind the articulation of the third
hypobranchial. The same specimen also shows a break in the perichondral covering behind the
articulation facet for the second hypobranchial and this marks the junction of the first two
ossification centres.

Moythomasia durgaringa

The hyomandibula of Moythomasia is more distinctly elbowed than that of Mimia and the
canal for the hyomandibular trunk of the facial nerve divides as it passes obliquely down through
the hyomandibula, to open by two foramina on the lateral surface of the shaft below the
dermohyal. The more anterior foramen is thought to have transmitted the mandibular branch,
while the posterior served for the passage of the hyoid branch (fhy.VII, Fig. 105). The
remainder of the hyoid arch, the branchial arches and basibranchial agree with the
corresponding structures in Mimia.

Hyoid and branchial arches: discussion

1. Hyoid arch

The hyomandibula is a clearly recognizable element in most gnathostomes and in osteichthyans
where it ossifies it does so as a single bone. Most authors have agreed that it represents the
epihyal in sharks (Luther 1909, Allis 1915, de Beer 1937), but some workers (Gegenbaur 1872,
Holmgren 1940) have suggested that it incorporates pharyngohyal and epihyal elements;
Holmgren (1943) and Jarvik (1977) proposed that in selachians, in addition to the
pharyngoepihyal, there was also a ventral component from a mandibular ray. Holmgren's
evidence (1940, 1943) for both selachians and actinopterygians was based on embryological
12

350

B. G. GARDINER

Fig. 108 Mimia toombsi Gardiner & Bartram. Restoration of ventral gill-arch elements in dorsal
view. Paired elements shown on right side only.

RELATIONSHIPS OF PALAEONISCIDS

Hb,

351

Hb.

art.Hb

Fig. 109 Moythomasia durgaringa Gardiner & Bartram. Basibranchial and hypobranchials of the

right side, from BMNH P.51380.

art.Hb

art.Hb,

art. H
art.Hb

art.Hb.

zmm

art.Hb,

Fig. 110 Moythomasia durgaringa Gardiner & Bartram. Basibranchial in (A), lateral and (B),
ventral views, from BMNH P.53221. (C), basibranchial in ventral view from Western Australian
Museum no. 20.4.244 (holotype).

352

B. G. GARDINER

stages prior to chondrification and his observed blastematic rudiments were merely
mesenchymatous cell masses. Moreover his observations lead to unacceptable conclusions,
such as the inference that the composition of the upper end of the hyomandibula differs in sharks
and most actinopterygians from that in rays and Acipenserl

The presence of the interhyal in osteichthyans and of a ceratohyal ossified in two sections
in neopterygians has further complicated the problem. Thus Allis (1915) believed that the
hyomandibula in teleosts was essentially a pharyngohyal and the interhyal the epihyal element,
while the posterior ceratohyal in neopterygians is often called an epihyal. In Acanthodes (Reis
1896, Miles 1964) the hyomandibula is perichondrally ossified in two sections. Miles (1964) has
compared the dorsal and ventral ossifications respectively with the laterohyal and epihyal of
osteichthyans, whereas Nelson (1968) regarded the dorsal element as the pharyngohyal.
However, the two ossifications of the hyomandibula of Acanthodes are clearly in the same
cartilage and this is consonant with the structure of acanthodian epibranchials which also ossify
from dorsal and ventral centres (Miles 19730: 93). The two ossifications of the hyomandibula of
acanthodians are considered a synapomorphy of that group (compare for example the single
perichondral ossification of the hyomandibula of Jagorina, Coccosteus and Dicksonosteus) .

In Neoceratodus the cartilaginous hyomandibula is said to exhibit great variability (Ridewood
1894: 637), but if we disregard the chondrification in the hyosuspensory ligament, there are two
chondrifications (Bertmar 1959, Fox 1965). The more dorsal is the hyomandibula, and the
ventral the interhyal. In Devonian dipnoans these ossify separately (Miles 1977).

Finally, the hyomandibula of recent holocephalans is unique in being non-suspensory and in
possessing an additional cartilage dorsally, called the pharyngohyal (Devillers 1958: 584). I
consider this latter element to be a new formation and do not agree with de Beer & Moy-Thomas
(1935) that the holocephalans are the only living fishes with a complete hyoid arch. A
pharyngohyal is therefore regarded as an autapomorphy.

An unusual feature of the hyomandibula of Polyptents, many palaeoniscids and some
placoderms (see p. 341) is the presence of a dermohyal. This bone, firmly bound to the
dorsolateral corner, has generally been regarded as an accessory hyomandibula in
actinopterygians (van Wijhe 1882, Bridge 1888) and to be serially homologous with the
spiracular ossicles of Polypterus. Despite its obvious dermal ornamentation the dermohyal of
Polypterus has a complex ontogenetic history. Allis (1922) said that during development it
contained a piece of cartilage and he therefore regarded it as a transformed hyal ray, a view
supported by Holmgren (1943) and Bertmar (1959). Holmgren (1943: fig. 45) and Daget (1950)
showed that the dermohyal first appears as a perichondral bone, with Holmgren (1943: 95)
adding that the subsequent bone lamella could easily be mistaken for dermal bone. I conclude
that the dermohyal is a dermal element fused with the head of the hyomandibula and
synapomorphous for actinopterygians and placoderms.

A peculiar feature of the hyomandibula of Polypterus, advanced palaeoniscids and other
actinopterans is an opercular process. Although it is possible to regard the posterodorsal angle
of the expanded region of the hyomandibula viNesides, Latimeria and Griphognathus , and the
posterior corner of the elbow in Megalichthys, as incipient opercular processes, only in
actinopterygians is this process distinct. This, however, did not prevent Jarvik (1954: 32) from
postulating the presence of a cartilaginous opercular process on the posteroventral margin of the
hyomandibula of Eusthenopteron. The formation of the process itself has also caused
considerable speculation; Allis (1915) for example thought it was a hyoid ray fused on to the
hyomandibula, but his view was criticized by Edgeworth (1926) on the grounds that he (Allis)
had not demonstrated the existence of separate cartilaginous primordia. Holmgren (1943: 86),
on the other hand, maintained that the hyomandibula oiAmia (and Salmo) resulted from the
fusion of two blastematic rudiments, the epihyal and suprapharyngohyal (= opercular process).
But these rudiments coalesce prior to chondrification and Edgeworth's criticism of Allis equally
applies to Holmgren's (1940, 1943) work. In Polypterus the opercular process may be related to
the possession of protractor hyomandibularis and dilatator opercularis muscles and in
actinopterans to the dilatator opercularis. Apart from Polypterus an opercular process is found
in Boreosomus, Acrorhabdus, Lepisosteus and halecostomes. An opercular process is not found

RELATIONSHIPS OF PALAEONISCIDS

353

2mm

Fig. Ill Moythomasia durgaringa Gardiner & Bartram. Right hypobranchials in lateral (left) and

medial views, from BMNH P.53256.

in Cheirolepis, Mimia, sturgeons and many palaeoniscids (Elonichthys, Cheirodus, etc.). From
this evidence , and if our phylogeny is correct (Rosen et al. 1981 , Patterson 1982) , I conclude that
the opercular process has arisen on at least two occasions, once in Polypterus and once within the
actinopterans.

The hyomandibular trunk of the facial nerve passes medial to the hyomandibula in
actinopterygians and turns outwards either to pass round in front of it, as in Polypterus, behind it
as in Acipenser and Polyodon, or to penetrate the bone as in palaeoniscids and most
actinopterans. In adult Acipenser fluvescens the hyomandibular nerve passes through a
cartilaginous extension of the hyomandibula (Jollie 1980). Except in Polypterus the
hyomandibular nerve does not divide into its mandibular and hyoid branches until it has
penetrated or passed the hyomandibula. In Polypterus it branches prior to crossing the
hyomandibula, the mandibular branch passing round the anterior face and the hyoid branch

354

B. G. GARDINER

Fig. 112 Mimia toombsi Gardiner & Bartram.
(A), right first hypobranchial in dorsal (above)
and ventral views; (B), right fourth hypo-
branchial in dorsal (above) and lateral views;
all from P. 56498.

behind the posterior face, above the opercular process. The relationship of the mandibular
branch in Polypterus is unique. In Cheirolepis there is no nerve foramen and the hyomandibular
nerve is thought to have passed round behind the hyomandibula, as in Acipenser. In
Eusthenopteron, Ectosteorhachis (Romer 1937) and Porolepis (Jarvik 1972) the hyomandibular
nerve pierced the hyomandibula, as in most actinopterans. In Latimeria (Millot & Anthony
1958: fig. 20) and Nesides (Bjerring 1977: fig. 25) the mandibular and hyoid branches fork after

Fig. 113 Mimia toombsi Gardiner & Bartram.
(A), second hypobranchials in ventral view;
(B), right second hypobranchial in dorsal view;
both from BMNH P.56498.

RELATIONSHIPS OF PALAEONISCIDS

355

Fig. 114 Moythomasia durgaringa Gardiner &
Bartram. Right first hypobranchial in dorsal
(top), lateral (centre) and ventral (bottom)
views, from BMNH P.53255.

1mm

penetrating the hyomandibula, much as in Moythomasia, and the hyoid branch then passes
through a further small bridge of cartilage or bone before leaving the posterior margin. The
hyomandibular nerve likewise passed through the hyomandibula in the fossil dipnoans
Griphognathus and Chirodipterus (Miles 1977), but in Neoceratodus it passes anterior to it, as in
selachians, before dividing into the hyoid and mandibular branches. In larval Neoceratodus
(Bertmar 1963), in contrast, it passes through the hyomandibula.

There is no nerve foramen in the hyomandibula of placoderms (Jagorina, Coccosteus,

2mm

Fig. 115 Moythomasia durgaringa Gardiner & Bartram. (A), first left ceratobranchial in lateral,
medial (centre) and ventral views, from BMNH P. 53218. (B), posterior end of first right
ceratobranchial in ventrolateral view, from Western Australian Museum no. 20.4.244 (holotype).

356

B. G. GARDINER

zmm

Fig. 116 Mimia toombsi Gardiner & Bartram. First epibranchials in (A), dorsal and (B), medial
views; second epibranchials in (C), dorsal and (D), medial views; (E), third left epibranchial in
lateral view. All from BMNH P.53245.

phlyctaeniids) or Acanthodes (Miles 1973a: pi. 4; Jarvik 1977: 207) and this is considered the
primitive gnathostome condition.

The hyomandibula in Mimia, Moythomasia and Eusthenopteron bears a single, medial row of
toothplates along its anterior margin. Jarvik (1954: 46) has homologized the toothplates of
Eusthenopteron with the accessory hymandibula (= dermohyal) of Polypterus. But in Mimia
and Moythomasia there are both lateral hyomandibular toothplates and a dermohyal. A medial
patch of toothplates has been recorded on the hyomandibula of Nesides (Bjerring 1977) and
Latimeria, and in Elops there is a row, much as in Mimia. Nybelin (1968: 441) has suggested that
this series in Elops may be serially homologous to the medial epibranchial toothplates on the
gill-arches. In Elops the epibranchial toothplates alternate with the gill-rakers, and both
Nybelin (1968) and Nelson (1969ft, 1970ft) have postulated that gill-rakers are modified
toothplates. Gill-rakers occur on the hyomandibula of Acanthodes where their presence has
been taken as indicative of an open hyoid gill-slit (Watson 1937, Nelson 1968). However, 0rvig
(1973: 146) and Jarvik (1977: 210) have suggested that these gill-rakers in acanthodians are
endoskeletal and are therefore homologous with the cartilaginous rods which support the
so-called 'gill-rakers' in sharks and dipnoans. In Acanthodes the hyomandibular gill-rakers form
a single series which projects ventromedially into the pharynx (Miles 1973«). A similar single
series of hyomandibular gill-rakers has been recorded in the selachian Cetorhinus, the

RELATIONSHIPS OF PALAEONISCIDS

357

Fig. 117 Mimia toombsi Gardiner & Bartram.
Third epibranchials in dorsal (above) and
ventromedial views, from P. 56474.

1mm

holocephalan Callorhynchus and the teleost Neonesthes (Holmgren 1942, Miles 1973a). There is
therefore no need to invoke the aphetohyoid hypothesis to explain their occurrence in
acanthodians. Whether or not hyomandibular gill-rakers are a primitive gnathostome feature is
uncertain; from their distribution within the phylogeny they appear to have arisen on more than
one occasion.

The remainder of the hyoid arch may consist of one further cartilage (= ceratohyal) as in
chondrichthyans, or several as in osteichthyans.

In osteichthyan fishes there is invariably a separate ossification or cartilage linking the
hyomandibula with the ceratohyal, the interhyal, and this is considered synapomorphous for the
group. In Recent chondrosteans the cartilage is hypertrophied and has often been incorrectly
referred to as a symplectic (Gardiner 1973, Patterson 1982). An accessory element between the
hyomandibula and the ceratohyal has been described in one specimen of Acanthodes (Miles
1973a: fig. 15; pL 7) but the evidence is equivocal (see BMNH P.4990).

In osteichthyans there is a further separate ossification or cartilage in front of the ceratohyal,
the hypohyal. This, like the interhyal, is synapomorphous for that group. In teleosts other than
pholidophorids and some osteoglossomorphs the hypohyal contains two ossifications (Patterson
1982).

The single ventral hyoid cartilage (called ceratohyal) of chondrichthyans is matched in
acanthodians. In Acanthodes (Fig. 120) it is perichondrally ossified in two sections (one at either
end), like the hyomandibula, epibranchials and ceratobranchials (Miles 1973a). This has
allowed the anterior ossification to be mistaken for a hypohyal (Watson 1937: fig. ISA; Nelson
1968: fig. 3A, B), but in presumed mature individuals of Acanthodes the ceratohyal is all one
ossification (BMNH P. 49977). It is therefore simpler to regard this mode of ossification as
synapomorphous for acanthodians.

In many osteichthyans there is either a single ceratohyal ossification (Cheirolepis, Mimia,
Polypterus, Glyptolepis, Latimeria) or a single cartilage (Neoceratodus, Necturus). In
neopterygians (Pteronisculus, Lepisosteus, Amia, teleosts), however, the ceratohyal is ossified

358

B. G. GARDINER

2mm

Fig. 118 Infrapharyngobranchials and suprapharyngobranchials. A-C, E, G, Mimia toombsl
Gardiner & Bartram. (A) first infrapharyngobranchials in dorsal and (B) ventral views; (C) second
left infrapharyngobranchial in dorsal view, second right in ventral view; (E) first right
suprapharyngobranchial in medial (top) and lateral views; all from P. 53245. (G) second left
suprapharyngobranchial in lateral (left) and medial views, from BMNH P. 56474. D, F,
Moythomasia durgaringa Gardiner & Bartram. (D) first right suprapharyngobranchial in medial
(top) and lateral views; (F) second right suprapharyngobranchial in lateral (left) and medial views;
both from BMNH P.53256.

in two sections which parallel those in Acanthodes, while in recent chondrosteans there are two
separate cartilages (ceratohyal and interhyal of previous authors, but see Patterson 1982). The
posterior ceratohyal cartilage in Polyodon carries a branchiostegal ray, as does the posterior
ceratohyal in Lepisosteus . Two ceratohyal cartilages are considered to be an unique feature of
chondrosteans and two ceratohyal ossifications a synapomorphy of neopterygians, where they
always remain separate and do not fuse as they do in acanthodians.
The remaining element in the hyoid arch is the symplectic. There are three kinds of symplectic

RELATIONSHIPS OF PALAEONISCIDS

359

recorded in fishes (Gardiner 1973, Patterson 1973, 1982) ; that in Recent chondrosteans, which is
probably a hypertrophied interhyal, that in neopterygians, which is confluent with the
hyomandibula, and that in actinistians. The neopterygian symplectic develops as an
anteroventral outgrowth of the hyomandibular cartilage which ossifies independently. It
therefore remains intimately attached to the hyomandibula and is applied to the quadrate in
teleosts or to the quadrate and lower jaw in halecomorphs (Patterson 1982). This type of
symplectic is considered synapomorphous for neopterygians (Patterson 1973: 262).

In Latimeria the symplectic is a large, partly ossified, independent cartilage which is
connected by a ligament to the hyomandibula. It articulates with the interhyal and ceratohyal
dorsally and with the lower jaw ventrally. The articulation with the lower jaw is posterior to and
separate from that between the quadrate and lower jaw. These articulations are therefore in
tandem (Forey 1981) and not side by side as in halecomorphs. Like Forey (1981), I consider this
tandem double jaw articulation synapomorphous for actinistians, and like Patterson (1982) I
regard the actinistian symplectic as characteristic of that group and non-homologous with the
neopterygian symplectic.

2. Basibranchial and branchial arches

Normally in gnathostome fishes there is a series of five arches (Nelson 19696). Dorsally they

consist of paired epi- and pharyngobranchials and ventrally of paired hypo- and cerato-

branchials.

There are hypobranchials in the first four arches in many osteichthyans, though in Acipenser
and most teleosts (Patterson 1977 a) only the first three are developed. Four hypobranchials are
also present in most chondrichthyans, and this is believed to be the condition in acanthodians
(Nelson 1968, Miles 1973a). However, the structures which Miles (1973a) identified as
hypobranchials inAcanthodes have been alternatively interpreted as basibranchials by Rosen et
al. (1981), and in turn the hypobranchials of Watson (1937) and Nelson (1968, 19696) have been
interpreted as anterior ceratobranchials by Miles (1973a). Re-examination of Watson's and
Miles' material has persuaded me that the ceratobranchials are perichondrally ossified in two
portions and that the ossifications lying anterior to the second, third and fourth are
hypobranchials (Miles 1973a: fig. 18, hy.br 2-4) and not basibranchials as Rosen et al. (1981)
presumed. The ossifications in question are inturned anteriorly and a single basibranchial (Fig.
120) against which the ceratohyal and first ceratobranchial articulated can be seen in several
specimens (e.g. BMNH P.49977; Miles 19730: pi. 5B, b.br). Often, only the anterior end of the

Fig. 119 Mimia toombsi Gardiner & Bartram. Reconstruction of posterior part of the neurocranium
and dorsal gill-arch elements in lateral view.

360 B. G. GARDINER

basibranchial is ossified (basihyal of Watson 1937, and see BMNH P. 49933, P. 44934). There is
no hypobranchial in the first arch in Acanthodes.

The basibranchial in chondrichthyans usually consists of two large, distinctly separated
cartilages or copulae, but subdivision of these may occur, particularly in holocephalans. The
ceratohyal and first branchial arch often articulate with the anterior cartilage (Heterodontus,
Heptanchus, Squatina) and the first arch slants obliquely forwards and downwards. The
succeeding hypobranchials point backwards to meet the anterior border of the posterior
cartilage, and this is considered synapomorphous for chondrichthyans.

In actinopterygians a single basibranchial ossification has been reported in Polypterus, Mimia
and Moythomasia. In the Gogo palaeoniscids the basibranchial is composed of three ossification
centres which presumably fused during ontogeny, and which were obviously in the same
cartilage. They correspond to the three basibranchials of Pteronisculus and Birgeria and to the
three minute ossifications recorded in the cartilaginous basibranchial ofAcipenser by van Wijhe
(1882). In Polyodon and teleosts there are three separate cartilages, but in Lepisosteus and Amia
there are four. The first three copulae in Lepisosteus and the second in Amia are perichondrally
ossified. The segmentation seen in the basibranchials of Amia and Lepisosteus must have
occurred independently in each form and four copulae cannot be regarded as synapomorphous
for neopterygians (cf. Wiley 1976). The basibranchial in Polypterus comprises a single
ossification which possibly corresponds to the first or anterior ossification in Mimia and
Pteronisculus (Patterson 1982).

Elsewhere in osteichthyans a basibranchial consisting of more than one ossification has been
described only in Eusthenopteron (Jarvik 1954). The division between the two ossifications is in
exactly the same position as that between the two anterior ossifications of Mimia. Whether these
two ossifications are in a single cartilage, as in Acipenser and Mimia, or represent separate
cartilages, as in Polyodon, we have no means of knowing.

In Latimeria, Glyptolepis and Laccognathus the copula resembles a greatly foreshortened
Polypterus basibranchial. The basibranchial of Neoceratodus and larval urodeles is similar
though considerably reduced. A small, triangular basibranchial has also been described in the
Devonian dipnoans Griphognathus and Chirodipterus . However, Miles (1977) has suggested
that the basihyal in these fossils is in fact a co-ossified anterior basibranchial and basihyal and
accordingly restored the bases of the first two arches on this ossification. Rosen et al. (1981)
suggest that if all the arches were restored entirely on the triangular basibranchial then the
arrangement of the gill-arch bases would be similar to Neoceratodus. Since cartilages rarely fuse
during development (although they may divide), and a separate cartilaginous basihyal is present
in Neoceratodus, I abide by our solution (Rosen et al. 1981) and thus a single broad, triangular
basibranchial is a synapomorphy of porolepiforms, actinistians and dipnoans (non-homologous
with that of Polypterus). Other gill-arch synapomorphies of that group include the reduction or
loss of hypobranchials in actinistians, dipnoans and larval urodeles and the articulation of the
last gill-arch with the base of the preceding arch in porolepiforms, actinisitians, dipnoans and
larval urodeles (Rosen etal. 1981). I further conclude that the primitive osteichthyan possessed
a single, cartilaginous basibranchial with at least two ossification centres. A similar
basibranchial may have been present in acanthodians (see Fig. 120). In acanthodians and
chondrichthyans the hyoid and first gill-arch articulate with the anterior basibranchial and in
osteichthyans at least the second gill-arch also articulates with it. This latter condition is
synapomorphous for osteichthyans (Rosen et al. 1981).

Anterior to the basibranchial a further ossification occurs in advanced actinopterans,
Eusthenopteron and dipnoans. It is usually referred to as the basihyal (Nelson 1969£), though
Jarvik (1954) regarded it as a member of the sub-branchial series and accordingly referred to it as
the 'sublingual rod'. A cartilaginous basihyal is also found in larval apodans, lizards and
Sphenodon (de Beer 1937). This should not be confused with the paired processus lingualis of
chelonians and birds (also called paraglossum or entoglossum), which appears to be
homologous with the anterior horns (or hyoid arch or lesser cornu) of Echidna (Goodrich
1930: fig. 474). The basihyal in actinopterans is confined to teleosts where it occurs in

RELATIONSHIPS OF PALAEONISCIDS

361

Fig. 120 Acanthodes bronni Agassiz. Reconstruction of ventral part of gill-arch skeleton in dorsal

view. Based on BMNH specimens.

osteoglossomorphs and those cladistically more derived groups (Patterson 1977 'a}. It appears to
chondrify separately from the basibranchial in Salmo (de Beer 1937) and Gasterosteus
(Swinnerton 1902), and usually supports a median dermal toothplate which may bear large
teeth. The basihyal of Eusthenopteron is a long, narrow bone with a strong median ridge dorsally
(Jarvik 1954). A similarly-enlarged basihyal is present in Griphognathus where it supports two
pairs of toothplates. In Conchopoma there is a single, median toothplate. Miles (1977: 286) has
suggested that the elongation of the basihyal in Griphognathus may be correlated with
secondary elongation of the lower jaw (the basihyal cartilage in Neoceratodus remains small)
and that the similar basihyals in Eusthenopteron and Griphognathus are the result of parallel
evolution. The absence of basihyals in actinistians, porolepiforms and primitive actinoptery-
gians and the lack of other plausible synapomorphies between Eusthenopteron and dipnoans
(Rosen et al. 1981) suggest that the basihyal has arisen on at least three occasions (teleosts,
Eusthenopteron, dipnoans and tetrapods).

Associated with the dorsal surface of the basibranchial is a series of toothplates. Nelson
(19696) hypothesized that the plesiomorphous gill-arch dentition of osteichthyans consisted of
six rows of dermal elements, and that paired basibranchial toothplates were primitive for
osteichthyans. Paired basibranchial toothplates are found in fossil actinistians where they are
arranged in three pairs lying opposite the first three gill-arches (Forey 1981). In Latimeria the

362

B. G. GARDINER

toothplates are insignificant and according to Nelson (19696) are an adventitious formation. If
Forey's (1981) phylogeny of the actinistians is correct this would seem to be the most
parsimonious hypothesis. In Eusthenopteron (Jarvik 1954) there are two pairs of toothplates on
each basibranchial, and also a lateral row of smaller plates related to the hypobranchials. In
Mimia and Moythomasia there are numerous, irregularly-arranged small toothplates over the
whole dorsal surface of the basibranchials and in Polypterus two widely-separated rows of
asymmetrical plates which come together posteriorly. In Acipenser (Jollie 1980) two transverse
bands of teeth cross the copula at the level of hypobranchials 1 and 2. In Bobasatrania,
Errolichthys and in teleosts other than Pachycormiformes a median toothplate covers the
basibranchials. These observations suggest that primitively the osteichthyan basibranchial was
covered by numerous toothplates which have been aligned into four rows in Eusthenopteron,
and into two rows in actinistians (and Griphognathus) . There is a median plate in teleosts (and
Conchopomd).

Situated beneath the basibranchial and joined anteriorly by ligaments to the hypohyals is a
median bone, the urohyal (in Polypterus the ligaments are ossified). In actinopterygians a
urohyal occurs in Polypterus, Australosomus (Patterson 19776) and teleosts. In Australosomus
and pholidophorids it is presumed to ossify in cartilage bone, but in Polypterus and extant
teleosts it ossifies in membrane bone. Patterson (19776) has convincingly demonstrated the dual
nature of the teleostean urohyal and has hypothesized that it represents the fused interclavicle
and urohyal, the interclavicle having sunk beneath the surface. Interestingly the interclavicle
has also sunk beneath the surface in many actinistians (Forey 1981). An endochondral urohyal
also occurs in Devonian and Carboniferous actinistians, Latimeria, porolepiforms, osteolepi-
forms (Jarvik 1963), Devonian dipnoans (Miles 1977), urodeles and anurans (Jarvik 1963,
Patterson 19776). The urohyal is presumed to have remained cartilaginous in the Gogo
palaeoniscids and Pteronisculus . The presence of a urohyal is therefore hypothesized as
plesiomorphous for osteichthyans.

Turning to the dorsal parts of the arches we find that in some actinopterygians and
sarcopterygians there are two types of pharyngobranchials, infra- and suprapharyngobran-
chials. The presence of the latter has been regarded as a synapomorphy of osteichthyans by
Rosen et al. (1981).

Allis (1925) and Holmgren (1942) suggested, mainly on evidence of orientation, that the
osteichthyan suprapharyngobranchial was homologous with the chondrichthyan pharyngo-
branchial. By contrast Nelson (1968) homologized the infrapharyngobranchial with the
chondrichthyan (and acanthodian) pharyngobranchial because in both groups they support the
pharyngeal roof and denticles. Nelson (1968) also suggested that the condition in the first arch of
Eusthenopteron, where there is a presumed compound supra-infrapharyngobranchial, could be
primitive for osteichthyans and might be the equivalent of the pharyngobranchial of
Acanthodes. He further imagined that the dorsal process of an Acanthodes pharyngobranchial
might be the homologue of the suprapharyngobranchial. Miles' (I913a: fig. 16) redescription of
the pharyngobranchials of Acanthodes has firmly persuaded me of the efficacy of Nelson's
(1968) hypothesis that the form of the pharyngobranchials of Acanthodes is primitive for
gnathostomes. Thus I believe that the acanthodian pharyngobranchial is homologous with that
of chondrichthyans and both the infra- and suprapharyngobranchials of the anterior arches of
osteichthyans and with the pharyngobranchial of the third arch in actinopterygians.

In Latimeria, where the first pharyngobranchial is a single, slender, dorsally-directed rod,
both Millot & Anthony (1958) and Nelson (1968) have interpreted it in the light of
Eusthenopteron and considered it to represent a compound supra-infrapharyngobranchial. But
this rod appears serially homologous with the second suprapharyngobranchial, and furthermore
the contact between the first infrapharyngobranchial and the braincase seen in primitive
actinopterygians and Eusthenopteron is replaced in Rhabdoderma by a direct contact between
the first epibranchial and the braincase (Forey 1981). Both the first and second epibranchials
contact the braincase in Latimeria. An analogous articulation occurs between the first
epibranchial and the auditory capsule in Polypterus, but here the infrapharyngobranchial
remains and articulates in the angle of the parasphenoid, as in Amia and some teleosts. In most

RELATIONSHIPS OF PALAEONISCIDS 363

teleosts the infrapharyngobranchial articulates with the parasphenoid rather than with the
prootic (Patterson & Rosen 1977: 129).

Suprapharyngobranchials occur in several actinopterygians, including Mimia, Moythomasia,
Polyodon, Acipenser, Pteronisculus, Australosomus, Lepisosteus and teleostean fishes of the
families Elopidae and Alepocephalidae (Nelson 19696); in actinistians (Latimerid) and in
Eusthenopteron. Usually they are two in number. Their absence from the first arch in Polypterus
is considered derived and related to the contact between the first epibranchial and auditory
capsule. Infrapharyngobranchials occur on the first three arches in many actinopterans,
including Pteronisculus, Australosomus, Acipenser, Lepisosteus and Amia, though it is possible
to regard this third infrapharyngobranchial as an undifferentiated pharyngobranchial. There are
only two infrapharyngobranchials in Eusthenopteron and Mimia and one in Latimeria (2nd) and
Polypterus (1st). In Recent teleosts there are four pharyngobranchials.

From this survey I hypothesize that the primitive osteichthyan possess two Suprapharyngo-
branchials (on the first two arches) and that three, not four, pharyngobranchials is plesio-
morphic for osteichthyans (cf. Nelson 19696). Furthermore, the loss of pharyngobranchials
(including infra- and supra-) is a synapomorphy of choanates (Rosen et al. 1981).

There are usually four pharyngobranchials in chondrichthyans and four have also been
inferred in acanthodians (Nelson 1968, 19696). However, careful examination of the numerous
casts ofAcanthodes made by Dr Roger Miles of material from the Humboldt University, Berlin,
the Palaeontological Institute of the University of Bonn and the University Museum of Zoology,
Cambridge, failed to reveal more than three pharyngobranchials. Despite some controversy
(Miles 1964, 1965 , Nelson 1968, 19696) it is now generally agreed that the pharyngobranchials of
Acanthodes projected posteromedially into the roof of the pharynx (Miles 1973a). Thus the
posterior position of the gill-arches and the posterior orientation of the pharyngobranchials in
chondrichthyans and acanthodians are considered to be shared primitive characters.

Axial skeleton

Mimia toombsi

The axial skeleton is represented by dorsal and ventral arcual elements throughout its length and
by supraneurals and ribs in the abdominal region. The individual ossifications are perichondral
shells. The notochord was persistent and there is no trace of centra.

The dorsal arcual ossifications are always paired and are presumed to be basidorsals. No
separate interdorsals have been observed and in this respect Mimia resembles Polypterus,
Lepisosteus, teleosts etc. (Gardiner 1983). In the abdominal region each basidorsal ossification
(na, Figs 121, 123) consists of a proximal plate or neural arch and a long distal process or neural
spine. On the lateral face of the neural arch there is a well-developed, laterally-directed
epineural process (epi, Fig. 121). Similar processes in the diapophysial position are found on the
anterior vertebrae of Australosomus, Boreosomus (Nielsen 1949: figs 41, 42, 47, 48), Caturus
(Gardiner 1960: fig. 33, Patterson 1973: 237) pholidophorids, Elops etc. and the dipnoan
Griphognathus (Rosen etal. 1981: fig. 54A). These epineural processes are found on each of the
first 14 vertebrae in Mimia. That on the first is very stout and terminates in several finger-like
projections (Fig. 121D). The bases of all the epineural processes are perforated or notched for
the passage of the intersegmental artery. The second basidorsal of Boreosomus (Nielsen 1949:
fig. 48) likewise has a foramen in its epineural process. The medial faces of the neural arches are
devoid of perichondral lining where they embraced the spinal cord. Posterior to the dorsal fin
the neural spines decrease in length caudally and towards the end of the tail some of the
basidorsals are incompletely segmented from one another and the junction between them is
marked by foramina for the spinal nerves. This is particularly marked in Moythomasia (BMNH
P.53255).

Above the basidorsals of the abdominal region there is a series of unpaired supraneurals. This
series terminates a short distance in front of the dorsal fin (Fig. 124).

The ventral arcual element? are unpaired and are presumed to represent basiventrals. In the
abdominal region these elements are semicircular in cross section with a posteriorly-directed

364

B. G. GARDINER

fia

epi

2mm

Fig. 121 Mimia toombsi Gardiner & Bartram. Neural arches and spines from anterior part of
vertebral column in (A) lateral and (B), (C), (D) anteromedial views. (A) from Western
Australian Museum no. 70.4.245 (holotype), (B) from BMNH P.56500, (C) and (D) from BMNH
P.53228.

median keel similar to that in Pteronisculus (Nielsen 1942: fig. 49). Ossified ribs articulated with
the lateral face of the abdominal basiventrals. In the region of the anal fin the ventral arcual
elements or haemal arches have well-developed median haemal spines (ha, Fig. 122A, B, C, D,
E, G) whose bases enclose the aortic or haemal canal. The lateral faces of the haemal arches in
this region are perforated by foramina for the intersegmental arteries (fia, Fig. 122C), as in
Caturus (Rosen etal. 1981: fig. 59A, B). Occasionally the roof of the haemal canal may also have
a median perforation (Fig. 122E).

In the caudal region there is a series of 22 stout hypurals which decrease in length posteriorly
(Fig. 124). The caudal lepidotrichia embrace the ends of these hypurals.

Moythomasia durgaringa

In several specimens remnants of the axial skeleton are visible and these appear to be identical
with those of Mimia. However, in the caudal region the neural arches and their spines are fused
and the neural spine is median, as in the caudal region of Birgeria.

Axial skeleton: discussion

1. Arcualia

Four pairs of arcualia were primitively present in each segment of gnathostomes (Gardiner
1983), and in all osteichthyans they are represented by ossified or cartilaginous elements in at
least part of the vertebral column. In actinopterygians they occur throughout the whole column
in Acipenser and Polyodon, in Pteronisculus they are restricted to the abdominal region and in
Australosomus, Caturus and Pholidophorus to the caudal region. Separate interdorsals and

RELATIONSHIPS OF PALAEONISCIDS

365

fia

2mm

Fig. 122 Mimia toombsi Gardiner & Bartram. Haemal arches from the anterior part of the caudal
region in (A), (B) lateral, (C(, (G) dorsolateral, (E), (F) dorsal, and (D) ventral views. (A), (B),
(C) from Western Australian Museum no. 70.2.245 (holotype), (D) from BMNH P.53232, (E)
from BMNH P. 56494, (F) and (G) from BMNH P.56500.

interventrals are missing in Polypterus and Lepisosteus, and from the development of most
teleosts; separate interventrals are missing from the trunk of developing Amia.

In actinistians (Latimerid) the full complement of arcualia is confined to the caudal region; in
Neoceratodus the two dorsal pairs are present only in the caudal area whereas the two ventral
pairs are only found in the anterior trunk. In Eusthenopteron, Osteolepis, Glyptolepis and most
temnospondyls the two dorsal pairs are always present although ventrally there is only a single
element. However, as in Latimeria, the full complement is seen in the tail region of several
tetrapods (Archegosaurus, Chelydosaurus).

2. Centra

The earliest recorded actinopterygian centra are the thin, ring centra of Haplolepis (Baum &

Lund 1974) from the Upper Carboniferous. These ring centra comprise thin calcifications in the

notochordal sheath and are developed from dorsal and ventral hemicentra. Similar ring centra

occur in the tail of Pygopterus, but in Turseodus there are separate dorsal and ventral

hemicentra.

Chordacentra appear to have been independently acquired in the pholidopleurids, and in
Australosomus (Nielsen 1949) they are overlain by endochondrally ossified neural and haemal
arches and interdorsals. By the end of the Jurassic many actinopterygians possessed
calcifications in the notochordal sheath including hemichordacentra in Furo philpotae (Agassiz)
and Caturus and complete chordacentra in pholidophorids, archaeomaenids, some pachycor-
mids, pleuropholids, catervariolids, Galkinia and Ichthyokentema (Patterson 1973). Today the
notochordal sheath is still involved in the early ontogenetic stages of centrum formation in
13

366

B. G. GARDINER

2mm

Fig. 123 Mimia toombsi Gardiner & Bartram. Restoration of part of the vertebral column from the
anterior abdominal region: (a), lateral view, (b), anterior view.

primitive living teleosts and if our phylogenies are correct (Patterson 1973, 19770, Rosen et al.
1981) then chordacentra must have arisen on at least three occasions within the actinopterygians
(haplolepids, pholidopleurids and halecostomes).

By the Cretaceous several groups of actinopterygians had acquired much more substantial
centra in the form of perichordal cylinders of membrane bone. Such groups include caturids
(Neorhombolepis, Macrepistius) , macrosemiids (Macrosemius, Ophiopsis), aspidorhynchids
(Belonostomus) and oligopleurids (lonoscopus, Callopterus) . Perichordal, membrane bone
centra are also developed in Polypterus, Lepisosteus, Amia and Recent teleosts (Gardiner
1983). From this we may conclude that membrane bone centra have developed at least five times
within the actinopterygians (Polypterus, Lepisosteus, Amia, aspidorhynchids and teleosts) and
possibly as many as seven (oligopleurids and macrosemiids).

Simple, ring-shaped centra resembling those of Amia are found in several rhipidistians
including Rhizodopsis, Megalichthys, Ectosteorhachis and Strepsodus. But the earliest ossified
centra belong to the Devonian dipnoans Griphognathus, Rhynchodipterus, Soederberghia and
Chirodipterus, where they are spool-shaped and presumed to be made up of cartilage bone, as in
amniotes.

3. Ribs

In some actinopterygians there is a single series of ribs in the wall of the coelom. These are the
ventral or pleural ribs and they develop centrifugally from cartilaginous anlagen close to the
vertebra. In cladistians and teleosts there are in addition dorsal ribs in the horizontal septum.
These develop from cartilaginous anlagen beneath the lateral line, at the outer junction of the
horizontal and transverse septa. Many teleosts have yet a further series of ribs, the so-called
epineurals in the epaxial musclature.

The dorsal ribs in Polypterus are confined to the middle part of the trunk where they are borne
on hypertrophied parapophyses. They are peculiar in that they are firmly tied distally to the
lateral line scales (Pearson 1981). The dorsal (epipleural) ribs of teleosts, on the other hand, are
usually attached by ligaments to the centrum or in the posterior region to the ventral ribs
themselves. They are believed to have arisen within the teleosts and to characterize
elopocephalans and certain osteoglossomorphs (Rosen et al. 1981). Dorsal ribs are a primitive
attribute of neither actinopterygians nor osteichthyans. Thus it follows that the ribs of

RELATIONSHIPS OF PALAEONISCIDS

367

I
I

•I

368 B. G. GARDINER

chondrichthyans and osteichthyans (other than cladistians and teleosts) are homologous and of
the pleural (ventral) type.

Although separate epineural intermuscular (epineural) ribs are synapomorphous for teleosts
the primitive condition seems to be an outgrowth or process of the neural arch (Patterson 1973:
237; Rosen et al. 1981: 244). They are in the diapophysial position; that is, they are where the
bicipital rib of a tetrapod articulates with the base of the neural arch. As Rosen etal. (1981) have
pointed out, these diapophysial outgrowths in actinopterygians and dipnoans show a
morphogenetic gradient which decreases with distance from the occiput; the ribs of primitive
tetrapods show a similar gradient.

4. Supraneurals and neural spines

Supraneurals are median, cartilaginous or bony structures that lie above the neural spine and the
dorsal ligament. They are often confused with median neural spines which may also project
above the dorsal ligament but which are formed by the fusion of the right and left halves of the
neural arch. The Supraneurals may rest on the dorsal ligament (Lepisosteus) or articulate with
the neural spines (Adpenser, Protopterus) , or sit on or between the tips of the neural spines
(Leptolepis, chondrichthyans such as Notorhynchus). Primitively they form an extensive series
and in Pteronisculus, Phanerosteon, Hypsocormus and Leptolepis they extend from occiput to
dorsal fin. In the porolepid Glyptolepis (Andrews & Westoll 1970) they are even more
extensive. In many actinopterygians they are confined to the anterior few segments (Amia,
Salmo) and in others they are missing altogether (Polypterus, many teleosts).

Supraneurals are also found in many Recent chondrichthyans, where they are usually
confined to the anterior few segments and to the caudal region (= epurals). Occasionally they
extend back as far as the anterior dorsal fin and beyond (Rhina). Supraneurals are absent in
acanthodians.

The neural spines are paired in primitive actinopterygians and median neural spines are
considered to be a halecostome characteristic (Patterson 1973: 296). Median neural spines,
however, also occur in the caudal region of Moythomasia, Australosomus, Birgeria (Nielsen
1949) and Haplolepis (Baum & Lund 1974: fig. la), as well as of Polypterus where they are
formed from membrane bone (Gardiner 1983).

Paired neural spines are found in Acanthodes, ptyctodonts (personal observation) and many
chondrichthyans (Mustelus, Rhina, Chimaera), and this is considered to be the primitive
gnathostome condition. In other Recent chondrichthyans the neural arches join below the
dorsal ligament (Lamna, Cetorhinus, Squalus), but in halecostomes, Eusthenopteron,
Griphognathus and tetrapods they join above it. Median neural spines are also found in the
placoderm Jagorina (Stensio 1959: figs 61, 63), the rhipidistians Thursius, Megalichthys and
Osteolepis (caudal region only, Andrews & Westoll 1970: figs 6, 7c, d), Latimeria, dipnoans and
tetrapods.

Shoulder girdle and pectoral fin

Mimia toombsi

The girdle consists of four paired dermal bones arranged in an overlapping series,
post-temporal, supracleithrum, cleithrum and clavicle, with a median interclavicle ventrally.
The endoskeletal shoulder girdle is attached to the ventromedial surface of the cleithrum.

The post-temporal (Fig. 125) is a four-sided rectilinear bone which overlaps the supra-
cleithrum posteroventrally. The bone is slightly curved and its anterolateral margin has an
unornamented ledge where it is overlapped by the extrascapular. The main lateral line passes
through the centre of radiation which lies near the anteroventral corner of the post-temporal.

The supracleithrum is a long, narrow bone which tapers ventrally where it fits over the dorsal
tip of the cleithrum. Its centre of radiation lies near the anterodorsal corner and like the
post-temporal is pierced by the main lateral line which passes through the bone two-thirds of the
way down its posterior margin.

The cleithrum is a large, high bone with a strongly concave anterior margin which forms the

RELATIONSHIPS OF PALAEONISCIDS

369

Exsc

Scl

2mm

Fig. 125 Mimia toombsi Gardiner & Bartram. Left supracleithrum in lateral (left, with outline of
post-temporal and extrascapular) and medial views. From BMNH P. 56498.

posterior boundary to the branchial cavity. Posteriorly the strongly convex margin is deeply
notched ventrally at the point of insertion of the pectoral fin. Ventrally the cleithrum is curved in
a medial direction where it is overlapped by the clavicle. The centre of ossification lies anterior
and slightly dorsal to the notch for the pectoral fin. The cleithrum is ornamented with stout
ridges of ganoine apart from the overlapped areas and the inner edge of the dorsal division.
The clavicle is also a large bone (Figs 126, 130) which curves strongly inwards ventrally,
almost to meet its fellow of the opposite side. The clavicle consists essentially of two parts, a
large, flat, ventral expanse and a dorsal vertical portion which wraps round the cleithrum. In
lateral view the dorsal division is triangular in outline with a long dorsal process. The clavicle is
ornamented in the same way as the cleithrum, with long ridges of ganoine. The centre of
radiation lies at the junction between the dorsal and ventral divisions. There is no evidence of

370

B. G. GARDINER

Fig. 126 Mimia toombsi Gardiner & Bartram.
Right cleithrum, clavicle and endoskeletal
shoulder girdle in lateral view, from BMNH
P.56495.

mscp

Clav

corf

2mm

the medial process seen in Acipenser (lessen 1972: pi. 15) and said to be present in Acrorhabdus
(Stensio 1921: 229).

The interclavicle lies between the clavicles. It is a small ovoid ossification with a few blobs of
ganoine along its mid-line.

In cladistians and chondrosteans the two clavicles meet in the mid-line and exclude the
interclavicle from the ventral surface (Acipenser, Scaphirhynchus Jollie 1980, Polypterus Fuchs
1929). This is considered a derived character because an interclavicle similar to that of Mimia
separates the cleithra in osteolepiforms, porolepiforms, primitive actinistians and tetrapods.

Behind the girdle there is a small postcleithrum (Fig. 127). This bone is little more than a
magnified flank scale with a much enlarged peg and a ventral projection. The peg is overlapped
by the supracleithrum (Rosen et al. 1981: fig. 39B). In Pteronisculus the connection with the
cleithrum is less intimate and the postcleithrum may only be distinguished from the surrounding
flank scales by its slightly larger size. In Acipenser (Jollie 1980) the postcleithrum lies largely
behind the margin of the supracleithrum and in Lepisosteus it is the most dorsal member of a row
of modified scales.

RELATIONSHIPS OF PALAEONISCIDS

371

2mm

Fig. 127 Mimia toombsi Gardiner & Bartram. Postcleithra. (A), right postcleithrum in medial (left)
and lateral views, from BMNH P. 56484. (B), right postcleithrum in lateral view, from BMNH
P.56491.

The endoskeletal shoulder girdle is a single ossification, despite its complex shape. The
mesocoracoid arch is well developed and dorsally is directed towards the posterior margin of the
cleithrum. In other actinopterygians (except Moythomasia and Acipenser} the mesocoracoid
arch is directed towards the anterior cleithral margin. The dorsal junction of the mesocoracoid
arch with the scapular portion is drawn out posteriorly beyond the hind margin of the cleithrum
(mscp, Fig. 126), as in Moythomasia (Fig. 131). This is the mesocoracoid process and it is also
found in Pteronisculus (lessen 1972: fig. 9).

The horizontal middle region of the girdle is produced anteriorly (apr, Fig. 129), as in many
actinopterans, and is perforated by two main foramina. The larger, hinder of these is the
scapular foramen, that nearer the leading edge of the middle region the anterior scapular
foramen. Both foramina occur in Birgeria, Palaeoniscus, Acipenser, Amia, Elops and Salmo
(Jessen 1972). In addition a large coracoid foramen occurs at the junction of the mesocoracoid
arch and the coracoid portion of the girdle; this is restricted to fossil actinopterygians.

The radial endoskeleton of the pectoral fin consists of a propterygium, three radials and a
metapterygium. These articulate with an almost horizontally-orientated, elongate, glenoid
fossa (gf , Fig. 128). The anterior or leading edge of the fin is slightly higher than the trailing edge
and the fin is expanded in a horizontal plane much as in Acipenser, Lepisosteus, Elops and
Salmo.

The propterygium is a short, ovoid ossification perforated by a canal. It is embraced by the
bases of the marginal rays. The three radials (r, Fig. 137) are a little longer than the
propterygium; they are hourglass-shaped, perichondral ossifications. The metapterygium is
over twice as long as the radials and supports three short preaxial radials. No distal radials have
been observed, but these may have been cartilaginous, as in Polypterus, or covered by the
proximal ends of the lepidotrichia, as in Pteronisculus (Nielsen 1942: 235).

There appear to be 18-20 pectoral rays, the principal of which are only articulated distally.
Along the leading edge the terminations of the lepidotrichia alternate with fringing fulcra, as in
Canobius, Mesopoma and Rhadinichthys .

372

B. G. GARDINER

scf

Clm

ascf

corf

Fig. 128 Mimia toombsi Gardiner & Bartram. Left cleithrum and endoskeletal shoulder girdle in

posterior view, from BMNH P. 53245.

RELATIONSHIPS OF PALAEONISCIDS

373

mscp

scf

apr

ascf

corf

2mm

Fig. 129 Mimia toombsi Gardiner & Bartram. Left cleithrum and endoskeletal shoulder girdle in

medial view, from BMNH P.53245.

374 B. G. GARDINER

Moythomasia durgaringa

The pectoral girdle is very similar to that of Mimia. The few differences include the size and
shape of the postcleithrum and interclavicle, and minor differences in the endoskeletal girdle
and fin construction.

The postcleithrum has a much longer, pointed, dorsal peg with an anteriorly-directed process
(Fig. 133). The interclavicle is proportionally larger than in Mimia and ventrally has two distinct
ornamented areas (Figs 134, 135).

In several specimens there is a prominent rostrocaudally-running ridge on the ventral surface
of the coracoid region, which marks the subdivision of the ventral fin musculature. A similar
ridge is found in Acipenser, Acrorhabdus, Pteronisculus (Aldinger 1937, Nielsen 1942),
Perleidus, Pachycormus and Elops (lessen 1972).

The pectoral fin has a propterygium to which the first three lepidotrichia are attached (Fig.
136) and the first ray is ornamented with longitudinal ridges of ganoine. The remainder of the fin
is supported by four radials and a metapterygium. There are 19-24 lepidotrichia, and fringing
fulcra occur along the anterior margin, as in M. nitida (personal observation).

Shoulder girdle and pectoral fin: discussion
1 . Dermal bones of shoulder girdle

Primitive actinopterygians show a series of four paired bones plus a median interclavicle,
whereas sarcopterygians have five paired bones and an interclavicle. The actinopterygian
shoulder girdle differs from that of other osteichthyans in having the cleithrum overlapped
dorsally by the supracleithrum. Rosen et al. (1981) have argued that this is the primitive
osteichthyan condition. In osteolepiforms, actinistians, porolepiforms and dipnoans an extra
element, the anocleithrum, separates the supracleithrum from the cleithrum; Jarvik (19446) has
hypothesized this to be the primitive osteichthyan condition. Jarvik further suggested that the
actinopterygian condition had been arrived at by regression of the anocleithrum. Rosen et al.
(1981), in contrast, suggested that the anocleithrum developed from the scale-like
actinopterygian postcleithrum, and proposed that the incorporation of the postcleithrum
(= anocleithrum) as a functional unit in the girdle is a synapomorphy for Eusthenopteron and
other sarcopterygians.

There is little doubt that the actinopterygian postcleithrum has been derived from the scale
row immediately behind the girdle. In Mimia, for example, its articulation with the
supracleithrum is by an expanded peg, homologous with that on other scales. Further, the
postcleithrum in Polypterus (Jarvik 19446), Pygopterus and Boreosomus is little more than the
most dorsal member of a modified scale row. A pair of scale-like postcleithra occurs in Scomber,
but primitive teleosts have three. A postcleithrum is absent in some teleosts and palaeoniscids
(Stegotrachelus, Cornuboniscus, Watsonichthys) , including Cheirolepis, where the scales are
very small and without peg-and-socket articulations. The absence in Cheirolepis is thought to be
primitive, as are the small scales devoid of peg-and-socket articulations. Pearson & Westoll
(1979) described a postcleithrum in Cheirolepis, but re-examination of their material (including
BMNH P. 41310, P. 36061) has failed to convince me of its presence.

A large scale-like extracleithrum is found in fossil actinistians (cf. Rhabdoderma, Forey 1981:
fig. 7) and is regarded as synapomorphous for that group.

Elsewhere a dermal shoulder girdle is found in acanthodians and placoderms and many
attempts have been made to homologize their various dermal elements with those of living forms
(Jaekel 1899, 1906; Dean 1907; Jarvik 19446; Stensio 1944, 1947, 1959). Despite these attempts
a separate nomenclature is usually employed for the dermal bones of acanthodians (Miles
19736, Denison 1979) and another for placoderms (Denison 1978).

In acanthodians the shoulder girdle is strengthened by dermal plates and spines both ventrally
and laterally. These are best developed in the Climatiidae and Diplacanthidae but are also
present in the Gyracanthidae. Dermal plates are wanting in the Ischnacanthidae and
Acanthodidae (Miles 19736). The dermal plates, where present, are in two series. Ventrally
there is a median unpaired plate (median lorical plate) in Br achy acanthus, Parexus,
Vernicomacanthus and Lupopsyrus, and two median unpaired plates in Climatius (Miles

RELATIONSHIPS OF PALAEONISCIDS

375

Iclav

Clav

1mm

Fig. 130 Mimia toombsi Gardiner & Bartram. Interclavicle in (A), (C) dorsal and (B) ventral views.
(B) Left clavicle in dorsal view. (A) from BMNH P.56473; (B), (C) from BMNH P.56484.

19736). Ventrolaterally, on either side of the median plate, is a pair of so-called pinnal plates in
Erriwacanthus, Vernicomacanthus, Parexus, Lupopsyrus, Sabrinacanthus, Ptomacanthus,
Euthacanthus and Gyracanthus. There are two such pairs in Brachy acanthus and Gyracanthides
and three in Climatius (Miles 19736). The pinnals of Erriwacanthus, Vernicomacanthus and
Sabrinacanthus all have extensive ascending laminae.

Dean (1907) suggested that these plates in acanthodians are homologous with the
osteichthyan interclavicle and clavicle, whereas Jaekel (1899) homologized them with the
cleithrum. Miles (19736: 205) maintained that, although it was possible to compare the ventral
bones in climatiiforms with those of the osteichthyan girdle, he found such comparisons
imprecise; he concluded that the ventral shoulder-girdle plates of acanthodians and
osteichthyans had been independently acquired and that the similarities between them were
fortuitous. He also concluded (19736: 162) that there was little possibility of the dermal plates of
acanthodians and placoderms being homologous, and so found it necessary to introduce a new
terminology for the plates in acanthodians. He came to these conclusions because he assumed
that the generalized climatioid pattern comprised two lorical plates and three or four paired
pinnal plates, as in Climatius. If, however, a single median plate and one pair of lateral plates, as
in Vernicomacanthus, is the primitive acanthodian condition then the correspondence with the
osteichthyan interclavicle and clavicles is far more exact. Moreover, the posterior pinnal plate in
Brachy acanthus and Climatius may be homologous with the spinal plate in ptyctodonts and
other placoderms.

Placoderms resemble osteichthyans in having several lateral dermal plates associated with the
shoulder girdle and it is probable that it is the short girdle that is primitive for placoderms
(Denison 1975), not the elongated thoracic shield of early arthrodires as argued by Gross (1954).
The former point of view is supported by the cladograms of placoderm interrelationships put

376

B. G. GARDINER

Fig. 131 Moythomasia durgaringa Gardiner &
Bartram. Left cleithrum, clavicle and endo-
skeletal shoulder girdle in lateral view, from
BMNH P.53221.

Clm

2mm

forward by Miles & Young (1977) and Young (1980), in which ptyctodonts are considered the
sister-group of all other placoderms.

Miles & Young (1977) proposed that the primitive placoderm possessed median dorsal,
anterior dorsolateral, anterior lateral, interolateral, spinal, anterior ventrolateral and anterior
median ventral plates. Ptyctodonts conform to this pattern apart from the absence of the
interolateral.

Jaekel (1906) and Stensio (1959) homologized the anterior median ventral, anterior
ventrolaterals and anterior laterals of arthrodires with the interclavicle, clavicles and cleithra of
osteichthyans. Their comparisons would, however, have been more exact had they substituted
ptyctodonts for arthrodires. If ptyctodonts are the most primitive placoderms and this sort of
outgroup comparison is meaningful, it follows that additional plates must have been added to
the thoracic shield within the placoderms (see Young 1980: 69). Furthermore, the anterior edge
of the cleithrum (anterior lateral) turns inwards to form a postbranchial lamina in both
ptyctodonts and Romundina (0rvig 1975: fig. 2A), much as in osteichthyans.

Stensio (1959) further homologized the anterior dorsolateral with the osteichthyan
post-temporal because of the contact that both were supposed to have made with the

RELATIONSHIPS OF PALAEONISCIDS

377

Clm

apr

mscp

msc

scf

corf

2mm

Fig. 132 Moythomasia durgaringa Gardiner & Bartram. Right cleithrum and endoskeletal shoulder
girdle in medial view, from BMNH P.53218.

378

B. G. GARDINER

Fig. 133 Moythomasia durgaringa Gardiner &
Bartram. Right postcleithrum in lateral view,
from BMNH P.53256.

neurocranium. But the anterior dorsolateral separates the anterior lateral (cleithrum) and
median dorsal and is overlapped by both of these elements.

From this survey I conclude that the presence of ventral dermal plates on the shoulder girdle
is synapomorphous for a group including acanthodians, placoderms and osteichthyans, that the
cleithrum is a synapomorphy of placoderms and osteichthyans and the anocleithrum a
synapomorphy of sarcopterygians.

2. Endoskeletal girdle

The actinopterygian shoulder girdle is tripartite and characterized by a middle region with an
anterior process (lessen 1972). The dorsal process is termed the mesocoracoid arch and the
dorsomedial muscles of the fin pass beneath it. The anterior diazonal nerves enter the canal
formed by this arch and the middle region from in front, and the ventral branches pass through a
foramen in its floor. This foramen is called the scapular foramen or posterior canal (Jessen
1972). A separate coracoid foramen for at least one of the ventral branches of a diazonal nerve
passes through the medial surface of the scapulocoracoid near the base of the mesocoracoid
arch. A separate anterior scapular foramen transmits a branch of the pectoral vein (Jessen
1972). A well-developed anterior process is found in actinopterygians, including Mimia,
Moythomasia, Palaeoniscus, Lepisosteus, Amia, Caturus, Hypsocormus (Jessen 1972) and
teleosts. Pearson & Westoll (1979) have suggested that a middle region is present in Cheirolepis,
but I am unable to confirm this; however, I have seen remnants of what I interpret as a
mesocoracoid arch in BMNH 19428 and P. 4345. The shoulder girdle is ossified in one piece
(including the mesocoracoid arch) in several actinopterygians, including Mimia, Pachycormus,
Pholidophorus and Leptolepis. In primitive living teleosts, however, it is ossified in three parts,
a ventral coracoid, a middle region or scapula and a dorsal mesocoracoid (e.g. Cyprinus,
Mormyrus, Salmo, Elops). In more advanced teleosts the mesocoracoid arch is lost.

RELATIONSHIPS OF PALAEONISCIDS

379

Iclav

2mm

Fig. 134 Moythomasia durgaringa Gardiner & Bartram. Clavicles and interclavicle in ventral view,

from BMNH P. 53219.

Ventrally the tripartite girdle is attached to the clavicle by the anterior coracoid process which
forms a canal for the marginal and ventral fin muscles, the supracoracoid foramen. This process
and canal are missing in Amia and several teleosts (e.g. Anguilld).

Both the coracoid process and mesocoracoid arch are missing in Polypterus and in common
with some teleosts there are two ossifications, the scapula and coracoid. The canal through the
girdle of Polypterus is thought to be homologous with the coracoid foramen (= posterior canal
of Jessen 1972) of other actinopterygians, since it transmits ventral branches of the fourth spinal
nerve and posterior arteries and veins of the fin. This canal or foramen passes between the
coracoid and scapula in Polypterus but through the scapula in teleosts.

A tripartite girdle similar to that of actinopterygians is also found in osteolepiforms (Janvier

1980) and fossil dipnoans (Rosen et al. 1981), where the supraglenoid foramen is homologous
with the upper muscle canal of Jessen, the supracoracoid foramen with the lower muscle canal
and the dorsal buttress with the mesocoracoid arch. The characteristic actinopterygian middle
region is represented by the small posterior buttress (Patterson 1982), but unlike
actinopterygians there are no apparent nerve foramina. Actinistians on the other hand resemble
Recent dipnoans in having an unfenestrated scapulocoracoid which in Rhabdoderma (Forey

1981) is represented by a triangular bone resting against the inner surface of the cleithrum.

380

B. G. GARDINER

Fig. 135 Moythomasia durgaringa Gardiner &
Bartram. Interclavicle in ventral view, from
BMNH P.56502.

In primitive tetrapods the scapulocoracoid is also ossified in a single piece (e.g. Eogyrinus,
Eryops, Cacops) and the supraglenoid buttress is homologous with the osteolepiform and
dipnoan dorsal buttress and the actinopterygian mesocoracoid arch (Goodrich 1930: 174). In
frogs and many amniotes there are two ossifications, referred to as the scapula and coracoid. In
yet other amniotes (Pareiasaurus, Procolophon, dicynodonts, cynodonts, monotremes) there
are three ossifications, with the additional element being referred to as the procoracoid. Thus
extant tetrapods and many fossil amniotes resemble Polypterus and living teleosts in the
retention of sutures between the cartilage-bones in the adult pectoral girdle.

Large size and anterior extent of the middle region of the shoulder girdle characterize
actinopterans and the scapular foramen and anterior scapular foramen are synapomorphous for
actinopterans. The tripartite girdle with its supraglenoid foramina is synapomorphous for
osteichthyans. We may also infer that the primitive number of ossifications in the osteichthyan
endoskeletal girdle is three.

In chondrichthyans and acanthodians (Ptomacanthus Miles 1973b) the two halves of the girdle
are connected by cartilage or fibrous tissue, and this may be the primitive gnathostome
condition.

In Acanthodes (Miles 19736) the scapulocoracoid is perichondrally ossified in three pieces, a
large scapulocoracoid (with a hollowed coracoid plate), a small suprascapula and a procoracoid
which articulated with the scapulocoracoid and supported the pectoral fin spine. A well-defined
canal passes through the scapulocoracoid and opens externally beneath the glenoid fossa via the
coracoid foramen. This canal is presumed to have transmitted diazonal nerves and vessels.

RELATIONSHIPS OF PALAEONISCIDS

381

There are two such canals in Diplacanthus , whereas in Sabrinacanthus (Miles 19736) there are
numerous fine foramina ramifying through the postero ventral region of the scapulocoracoid, as
in some placoderms.

In chondrichthyans (e.g. Hexanchus, Pristiurus, Chimaera) there is usually a single, internal
coracoid foramen for the diazonal nerves, but this soon divides and opens externally above and
below the glenoid fossa. The coracoid foramen is also seen inAcipenser and other actinopterans,
as well as many tetrapods, and is consequently considered to be a primitive gnathostome
feature. The external upper opening is called the supraglenoid foramen, but this is not
homologous with the similarly-named canal in osteichthyans (see p. 378). The coracoid region
expands immediately below the pericardium and there is a distinct coracoid fossa beneath the
glenoid fossa, as in Acanthodes, for the ventral fin musculature.

There is considerable variation in the pectoral endoskeleton of placoderms and in the
presumed advanced forms the scapulocoracoid is a low structure lacking a prominent scapular
blade (arthrodires). Pseudopetalichthyids, rhenanids and arthrodires (Broili 1933, Stensio
1959, Young 1980) have an extensive anteriorly-directed coracoid process and arthrodires have
an elongated glenoid fossa. The long, low scapulorocacoid of arthrodires is crossed by a series of
diazonal nerves and segmented vessels. An extensive coracoid process is seen elsewhere in
acanthodians (Sabrinacanthus), chondrichthyans and actinopterans and is probably primitive,
as Miles (19736) suggested.

A scapular process is found in Pseudopetalichthys, certain rhenanids (Brindabellaspis,
Jagorina) and ptyctodonts. This is regarded as a primitive gnathostome feature as suggested by
Young (1980: 51).

In ptyctodonts there is a canal for the ventral fin musculature, bounded externally by the
clavicle (anterior ventrolateral plate, Miles & Young 1977). This canal is homologous with the
supracoracoid foramen of osteichthyans and is inferred to have been formed by a plate of dermal
bone bridging the coracoid fossa. In osteichthyans it is bounded by the cleithrum, except in
Acipenser where the clavicle forms its anterior margin. In Rhynchodus (Stensio 1959: fig. 75) the
identical canal has a rim of perichondral bone posteriorly. A similar ventral fossa also occurs in
the palaeacanthaspids Romundina (0rvig 1975: pi. 5) and Palaeacanthaspis (Stensio 1944: fig.
9), but here it is closed anteriorly by perichondral bone. This canal or fossa is absent in other
placoderms but nevertheless is regarded as a synapomorphy of a group containing placoderms
and osteichthyans. In ptyctodonts the endoskeletal girdle and glenoid fossa project posteriorly
beyond the dermal girdle, much as in actinopterygians, and the pectoral fin is similarly
orientated in both groups.

A coracoid foramen has been recorded in Brindabellaspis (Young 1980) and Romundina
(0rvig 1975: pi. 4, dzv).

3. Pectoral fin

The pectoral fin of actinopterygians is characterized by having a propterygium (Rosen et al.
1981), a first radial which is short, broad and strongly articulated. In Mimia, Moythomasia,
Palaeoniscus, Pteronisculus, Acipenser, Lepisosteus and most actinopterygians (Patterson 1982)
the propterygium is perforated by a canal which in living forms conducts nerves and vessels
(Jessen 1972). A perforated propterygium is an actinopteran character (Patterson 1982).

A well-developed propterygium is also found in Cheirolepis (Ra2, Ra3 of Pearson & Westoll
1979: fig. 13) where it is clasped by four lepidotrichia (BMNH P.6096a). There is no evidence of
a propterygial canal and in this respect Cheirolepis resembles Polypterus.

Rosen et al. (1981) have interpreted actinopterygian paired fin structure as a transformation
of a metapterygial fin into a propterygial type. Certainly, pectoral metapterygial elements are
still to be found in such primitive members as Cheirolepis (personal observation), Mimia,
Acipenser, Pteronisculus and Palaeoniscus (Jessen 1972), and from a study of the development
of Polypterus (Budgett 1902) much of the fin in this fish also appears to be metapterygial.

Two sets of radials usually occur in the pectoral fins of actinopterygians. Posteriorly in
sturgeons and on the second radial of Elops there are three sets. The distal radials are triangular
in shape and sit between the tips of the proximal radials in primitive forms.

382

B. G. GARDINER

ffr

propt

can

Zmm

Fig. 136 Moythomasia durgaringa Gardiner & Bartram. Propterygium and leading fin-rays in
external (left) and internal views, from Western Australian Museum no. 70.4.244 (holotype).

The bases of the marginal rays embrace the propterygium in actinopterygians and in
Moythomasia and teleosts the first ray, at least, is attached to it. The first condition is considered
synapomorphous for actinopterygians.

Pelvic girdle and fin

Mimia toombsi

The fin is small and situated about midway between the pectoral and anal fins. Only one

specimen contains traces of the pelvic girdle. The pelvic plate is smaller than in Moythomasia

and is without an anteromedial process. No radials were observed. The fin consists of around 20

lepidotrichia.

Moythomasia durgaringa

The pelvic plate (Pg, Fig. 138) is a thin, triangular, perichondral ossification with a slender
anteromedial process. Seven radials articulate with its posterior margin. The six anterior radials
are of approximately the same size, but the seventh, the most posterior, is much larger and is
presumed to represent the metapterygial axis. Seven radials have been recorded in Acipenser
(Rosen etal. 1981: fig. 28B), eight in Boreosomus (Nielsen 1942), nine in Scaphirhynchus and as
many as 14 in Polyodon. Polypterus has only four.

The pelvic fin is nearer to the anal than to the pectoral (cf. M. nitida Jessen 1968: fig. 4) and
contains 18-20 jointed lepidotrichia. It bears fringing fulcra along its anterior edge.

RELATIONSHIPS OF PALAEONISCIDS

r
mtp

383

propt

Fig. 137 Mimia toombsi Gardiner & Bartram.
Reconstruction of the base of the right pectoral
fin in dorsal view, from several specimens.

ffr

Pelvic girdle and tin: discussion

The pelvic fin in actinopterygians appears to be constructed differently from the pectoral
(Goodrich 1930), whereas in all other osteichthyans the two are similar in structure. Davidoff
(1880) reasoned that this was because in actinopterygians the metapterygial skeleton of the
pelvic fin had shifted inwards and been incorporated in the pelvic girdle. Rosen et al. (1981)
revived Davidoffs theory as it offered an explanation of the observed similarity between the
pelvic endoskeleton of chondrosteans and the metapterygium plus preaxial radials of
chondrichthyans and because the alternative, that the chondrosteans possess the most primitive
paired fins, violated the monophyly of osteichthyans and actinopterans. They further reasoned
that actinopterygians might not have a primary pelvic girdle.

In Polyodon, Scaphirhynchus, Acipenser and some palaeoniscids (Lehman 1966) there is a
series of rod-like basal cartilages which are presumed to represent the segmental metapterygial
skeleton. These fuse in most adults but may remain separate in some palaeoniscids and partially
separate in Scaphirhynchus. Two rows of cartilages are joined to the outer surface of these
internal cartilages, an inner row of elongate radials and a distal row of smaller triangular radials.

Fig. 138 Moythomasia durgaringa Gardiner & Bartram. Pelvic girdle and radials of the right side in

dorsolateral view, from BMNH P. 53236.

384

B. G. GARDINER

Two rows of radials also occur in Polypterus, Pteronisculus, Boreosomus, Pygoptems and many
other palaeoniscids, and also in Lepisosteus andAmia, but in teleosts there is only a single series.
There are some 14 proximal radials in Polyodon, ten in Pteronisculus, nine in Scaphirhynchus ,
eight in Boreosomus, seven or eight in Acipenser and four in Polypterus. In Lepisosteus, Amia
and primitive teleosts the radials are reduced to never more than three small nubbins, and in
many teleosts there is no trace of radials.

I conclude, like Patterson (1982), that a pelvic plate and two series of radials are
synapomorphous for actinopterygians.

Median fins

Mirnia toombsi

The dorsal fin lies in the posterior half of the body opposite the anal fin (Fig. 145). Remnants of
the endoskeleton of the median fins are found in several specimens, but owing to the
post-mortem folding and twisting of the body exact relationships of individual ossifications are
difficult to determine. Nevertheless, both dorsal and anal fins appear to be supported by a single
series of radials.

In the dorsal fin the few anterior radials are rod-like; posteriorly the fin is supported by three
compound radial plates (rpl, Fig. 124). These posterior plates correspond in part to the single
axonost plates described in Pygopterus, Pteronisculus, Birgeria and Australosomus (Aldinger
1937, Nielsen 1942, 1949). The dorsal fin has 28-36 branched, segmented lepidotrichia and on
the leading edge fringing fulcra alternate with the lepidotrichial endings.

The anal fin is supported by seven long radials which are expanded distally and there is a single
complex radial plate posteriorly. This radial plate (rpl, Fig. 124) has lateral wing-like extensions
proximally and these are perforated by a small foramen. A much larger foramen passes through
the centre of the radial plate. The anal fin is made up of 30-40 branched, segmented
lepidotrichia.

2mm

Fig. 139 ( A) , anal radials and anal radial plate of Mimia toombsi Gardiner & Bartram in right lateral
view, from Western Australian Museum no. 70.4.245 (holotype). (B) Moythomasia durgaringa
Gardiner & Bartram, anal radial plate in dorsal view, anal radial (right) in lateral view, from
BMNH P.53218.

RELATIONSHIPS OF PALAEONISCIDS

385

Fig. 140 Mimia toombsi Gardiner & Bartram. (A), two lateral line scales in lateral and medial views
(arrow marks anterior), from BMNH P.56497. (B), two articulated, anterior trunk scales, from
BMNH P.56495.

The caudal fin is deeply cleft and inequilobate and the lepidotrichia are closely set, branched
and jointed. There are some 50 or more lepidotrichia and along the ventral margin the
lepidotrichial endings alternate with fringing fulcra.

Moythomasia durgaringa

The dorsal and anal fins are opposite to one another, nearer to the tail than the head. The dorsal
fin is slightly longer than that of Mimia with around 40 lepidotrichia, but the anal fin is of
comparable size (30-35). A similar radial plate is present at the rear of the anal fin (rpl, Fig.
139); it only differs from that of Mimia in the possession of a posteriorly-directed distal flange.
The caudal fin has around 80 lepidotrichia and the ventral margin supports fringing fulcra.

386 B. G. GARDINER

Median fins: discussion

1. Dorsal and anal fins

Most actinopterygians have a single dorsal and anal fin, but many teleosts also have an adipose
dorsal fin which may be supported by actinotrichia. Other teleosts may have as many as three
dorsal and two anal fins (Gadus) and a continuous dorsal fin is seen in many palaeoniscids (e.g.
Tarrasius) and teleosts (gymnotids, anguillids).

The dorsal and anal fins are supported by a series of parallel radials which are often arranged
in three series (each radial is three-segmented). The radials beneath the dorsal and anal fins are
three-segmented in chondrosteans, many palaeoniscids (Birgeria, Pteronisculus, Boreosomus),
Lepisosteus, Amia and several teleosts (e.g. osteoglossids, cyprinids, salmonids, esocids), and
are two-segmented in Tarrasius, Australosomus and many other teleosts (gadids, characinids,
clupeids; Bridge 1896). In Polypterus the dorsal fin is supported by a single series but in the anal
fin all the radials, apart from the first, are in two parts. The lepidotrichia clasp the ends of the
radials forming modified ball-and-socket joints in neopterygians, where the distal radials are
spherical. Posteriorly some of the distal dorsal and anal radials are compounded into an axonost
or radial plate in Mimia, Pteronisculus , Birgeria, Pygopterus and Australosomus (Nielsen 1942,
1949).

Primitively the fin-rays far exceed the radials in number, but in neopterygians, haplolepids,
perleidids, Bobasatrania, Luganoia and Platysiagum (Patterson 1973) the number of dorsal and
anal fin-rays is equal to that of their supports.

2. Caudal fin

The caudal fin in actinopterygians is primitively heterocercal, although a small epaxial lobe
occurs near the tip of the tail in several adult palaeoniscids (Cheirolepis, Palaeoniscus,
Paramblypterus) and is present in the ontogeny of Recent forms where it is supported by
actinotrichia. In palaeoniscids it is supported by small lepidotrichia. Large epaxial fin-rays are
developed in saurichthyids, amiids, pholidopleurids, pycnodonts, pachycormids and other
teleosts (excluding pholidophorids). If the published phylogenies are correct (Patterson 1973,
1977fl) they must have been independently developed on at least six occasions.

In living chondrosteans there is a series of median epurals lying above the neural arches of the
tail; these are believed to be serial homologues of the supraneurals (Patterson 1973).

In advanced actinopterygians the tail is homocercal and in halecomorphs and teleosts the
epaxial fin-rays are supported by a few epurals. There are seven epurals in pachycormids, six in
Australosomus, four or five in Amia and three or fewer in living teleosts (Patterson 1973). In
pycnodonts the epaxial fin rays are apparently supported by the neural arches (Gardiner 1970).

In Polypterus there is no sure way of distinguishing between radials and epurals or between
dorsal fin-rays and epaxial caudal rays; nevertheless in development there appear to be three
epurals.

The hypaxial lobe in actinopterygians is supported mainly by the expanded haemal spines and
hypurals. The hypaxial radials are reduced to small nubbins of bone or cartilage at the tips of
these spines. Reduced hypaxial radials are found in palaeoniscids (Pteronisculus), chondros-
teans, Lepisosteus and primitive teleosts. Radials are absent in Polypterus and the hypaxial lobe
is supported solely by expanded haemal spines and hypurals. In actinopterygians the
lepidotrichia clasp the spines and hypurals, as well as the radials where present.

Squamation

Mimia toombsi

The body is entirely covered with scales, which have a transverse course up to the inversion line
at the base of the tail. There are approximately 75 transverse scale rows between the cleithrum
and the line of inversion. Ridge scales completely clothe the dorsal margin and the area between
the anal fin and the tail. There are ten ridge scales in front of the dorsal fin, two in front of the
anal fin and eight between the anal and the tail. Those on the caudal fin form a rigid cutwater and
extend right to the tip of the tail. All the ridge scales are median structures.

RELATIONSHIPS OF PALAEONISCIDS

387

Fig. 141 (A), Mimia toombsi Gardiner & Bar-
tram, posterior abdominal scales, from holo-
type, Western Australian Museum no.
70.4.245. (B), Moythomasia durgaringa Gar-
diner & Bartram, scales from the abdominal
region, from BMNH P. 56479.

The lobe of the pectoral fin is not scaly (cf . Polypterus, Cheirolepis) , but the points of insertion
of all the fins are marked by bands of much narrower scales.

The scales themselves have a deeply incised hinder margin and are ornamented with short
ridges of ganoine which terminate posteriorly in sharp points. On the anterior trunk scales there
are 7-10 stout ridges of ganoine, but posteriorly the number is reduced to four or five and near
the base of the tail to one. They have well-marked peg-and-socket articulations, with the peg
exhibiting growth lines (PI. 2c). The lateral line scales are higher than broad and the lateral line
canal enters the scale anteriorly at the junction of the peg with the anterodorsal ornamentation
(Fig. 140A).

The scales have a diagonal long axis (see Gross 1966) and a reduced bony base (Fig. 143).
Both dentinal tubules and dentine are apparently wanting and the cell spaces are exceptionally
large. The ganoine layer is relatively thick and quite unlike that of any other actinopterygian.
The ganoine is in the form of superposed generations which appear to have grown in an aberrant
'onion-skin' fashion without an accompanying layer of dentine. Judged by presumed younger
individuals the ganoine first forms longitudinal ridges or blisters, completely separate from one
another. These later fuse by the addition of further superficial layers of ganoine. The bony base
consists of horizontal layers with contained cell spaces, canals of Williamson and canals for fibres
of Sharpey.

Moythomasia durgaringa

There are far fewer transverse scale rows than in Mimia, with a scale count of between 44 and 48.

The scales are ornamented by ridges of ganoine which branch and anastomose. The ridges
terminate posteriorly in up to 14 serrations on the more anterodorsal flank scales, but
anteroventrally the scales have fewer serrations (eight or nine). Posteriorly the scales have five
or fewer serrations. Behind the pelvic fin there is a pair of elongate, almost oval cloacal scales,
equal in area to approximately three normal scales (BMNH P. 53217).

Initially there are separate ridges of dentine and ganoine (Fig. 141B) and new ganoine
and dentine are added between the ridges until the whole external exposed surface is
ganoine-covered. The ganoine is single-layered over most of the scale, and this is presumed to be
primitive (Schultze 1977). Superposed generations of buried ganoine are confined to the scale
margins (Fig. 144) and to the edges of the primary ganoine ridges. The scales have a diagonal
axis and a reduced bony base. The bony base contains cell spaces and canals for fibres of

388

B. G. GARDINER

1mm

Fig. 142 Moythomasia durgaringa Gardiner & Bartram. (A), three scales in articulation, including
the lateral line scale, medial view, from BMNH P.53221. (B), lateral line scale in lateral view, from
BMNH P.53221.

Sharpey. Above the base is a horizontal vascular canal system from which the dentinal tubules
pass upwards through the dentine layer.

The lateral line scales have surface pores and the ventral medial surface of the scales has a
distinct depression for the articulatory peg of the scale immediately below it. The ornamentation
is very variable and corresponds closely with that of M. perforata (Gross 1953: fig. 5).

Squamation: discussion

1. Scale structure

The scales of generalized actinopterygians are of the ganoid type in which the whole scale
typically has an 'onion-skin' mode of growth, with new material being added concentrically to
both the outer and inner surfaces. Bone is added to the base of the scale and ganoine to the
surface, and the scale becomes thick and shiny. In advanced actinopterans the ganoine is
pseudoprismatic (0rvig 19670, 1978) and in many teleosts is lost altogether.

In Cheirolepis (Gross 1967) the growth lines indicate that the layers of ganoine are added
mainly to the anterior margin.

In Polypterus the superposed generations of buried ganoine are confined to the circum-
ference (Meinke 1982: 371), but as in Cheirolepis they are most prominent on the anterior

RELATIONSHIPS OF PALAEONISCIDS

389

fi .Sh

0. 1mm

can .W

Fig. 143 Mimia toombsi Gardiner & Bartram. Vertical longitudinal sections through two flank

scales. From BMNH P.56503.

fi.Sh

c.sp,

0 . 1 mm

Fig. 144 Moythomasia durgaringa Gardiner & Bartram. Vertical longitudinal section through flank

scale. From BMNH P.53221.

390

B. G. GARDINER

Plate 2 Mimia toombsi Gardiner & Bartram. (a) anterior end of dentary, from BMNH P. 53252, x 30.

(b) posterior abdominal scale, from Western Australian Museum no. 70.4.245 (holotype), x35.

Moythomasia durgaringa Gardiner & Bartram. (d) anterior flank scale, from Western Australian
Museum no. 70.4.244, x 25. (e) anterior scale, from BMNH P. 53221, x 18. (f) enlarged view of
scale peg, same specimen, x 42. Scanning electron micrographs.

391

392 B. G. GARDINER

margin. The scales of Moythomasia (Jessen 1968), like those of Polyptems, have a very thin
layer of ganoine, with superposed generations of buried ganoine mostly confined to the scale
margins. Thus the primitive actinopterygian scale seems to have grown by the addition of
ganoine to the circumference rather than in the concentric 'onion-skin' fashion so typical of
Lepisosteus. All primitive actinopterygian scales possess a superficial layer of ganoine which is
said to be characteristic of actinopterygians (Schultze 1977, Patterson 1982). Slender
peg-and-socket articulations between scales are also characteristic of actinopterygians but their
absence from the scales of Cheirolepis is considered primitive. Nevertheless, the scales of
Cheirolepis have a diagonal long axis and an anterodorsal process. A similar diagonal long axis
and anterodorsal process characterize the scales of most actinopterygians (palaeoniscids,
Polypterus, Lepisosteus) and is therefore considered synapomorphous for the group (Patterson
1982).

The scales of sarcopterygians never grow in the concentric 'onion-skin' fashion seen in
actinopterygians. Instead they are of the cosmoid type with a superficial layer of enamel
(Schultze 1977). Cosmine (dentine -I- enamel) is a hard tissue which encloses a complex
pore-canal system. No living fish has either cosmine or a pore-canal system in the scales but the
presence of both is regarded as a synapomorphy of rhipidistians and dipnoans (Rosen et al.
1981). The mesh canals linking the flask-shaped pore-canals have a horizontal partition in some
osteolepiforms (Gross 1956) and this appears to be a specialization. Cosmine is missing from the
scales of actinistians and tetrapods, but the scales of Latimeria do possess superficial tubercles
which fuse to the scale surface. Separate tubercles occur above the scales in Polypterus,
Lepisosteus and silurids. The scales of the dipnoan Uranolophus (Denison 1968, 1969) are
characterized by Westoll-lines and superposed generations of tubercles.

The scales of euselachian chondrichthyans are typically placoid. Composite scales, however,
occur in a Permian edestid holocephalian and the dorsal scales of 'Orodus' (0rvig 1966) grew in
'onion-skin' fashion by the presumed concentric addition of dentine to the crown and fibrous
bone to the base.

Placoderms are usually devoid of scales, but where they occur they are often small and
rhomboidal (Miles & Westoll 1968).

The scales of acanthodians are small, closely fitting and made of concentric layers of bone and
dentinal tissue. They had an 'onion-skin' mode of growth, similar to that in actinopterygians
(Gross 1966, 1973). In the Nostolepis type the crown is mesodentine (0rvig 19676) and in the
Acanthodes type it is dentine. In Poracanthodes there is a pore-canal system, but its architecture
is quite unlike that of rhipidistians and dipnoans.

Within the gnathostomes this 'onion-skin' mode of scale growth is considered primitive.

2. Basal fulcra

The large paired or unpaired scale-like structures, preceding the bases of the median fins in
primitive actinopterygians (Patterson 1982), are termed basal fulcra; they appear to be modified
ridge scales. They are particularly well developed on the dorsal border of the tail in those
primitive actinopterygians without elongate upper caudal fin-rays (most palaeoniscids, Mimia
and chondrosteans). They are also present in Dapedium, Lepisosteus, Eurycormus, Caturus,
lonoscopus (Patterson 1973), parasemionotids and primitive teleosts. Basal fulcra are absent in
Polypterus but this is presumed to be secondary and related to the acquisition of a diphycercal
tail.

Basal fulcra may be found in front of all the unpaired fins, including both lobes of the caudal
(Cheirolepis, Elonichthys, Cornuboniscus, Mesopoma, sturgeons, Dapedium etc.), or they may
be missing from the anal (Moythomasia, Phanerosteon) , or from both anal and dorsal

Plate 3 Moythomasia durgaringa Gardiner & Bartram, body scales, (a) mid-flank, x27. (b) basal
fulcral, x 16. (c) posterior flank, x 27. (d) anterior flank, x 27. (e) from base of tail, x 27. (f) from
tail, x27. (g) from base of anal fin, x34. (h) from base of pelvic fin, x27. Scanning electron
micrographs, (a), (b), (d), (e), (f), (g) and (h) from BMNH P.56502, (c) from BMNH P.56475.

RELATIONSHIPS OF PALAEONISCIDS

393

394

B. G. GARDINER

Fig. 146 Character phylogeny of some of the
better-known actinopterygian genera. Charac-
ters listed in text. (For Birgeria read
Aetheretmon; for Australosomus read Cos-
mopty chins.)

(Carboveles, Platysomus, Pygopterus, Polyodon, Lepisosteus, generalized teleosts etc.), or
from anal, dorsal and lower caudal (Cryphiolepis , haplolepids), or missing altogether
(Polypterus , many teleosts). The basal fulcra are unpaired except above the tail in Cheirolepis
where the first two scales are unpaired and the remaining 40-50 paired.

A single unpaired structure occurs in front of the median fins of Osteolepis and at least the
dorsal fins of Uranolophus. However, only actinopterygians possess the long row of basal fulcra
on the dorsal margin of the tail and these must be considered synapomorphous for
actinopterygians (Patterson 1982).

3. Fringing fulcra

These are paired, leaf-like structures attached to the leading fin-rays which in fossils are difficult
to distinguish from the dichotomously branched ends of the lepidotrichia. Fringing fulcra are
unique to actinopterans but their distribution is very spasmodic. They are found in Mimia,
Moythomasia, Pteronisculus, Perleidus, Ptycholepis, redfieldiids, Lepisosteus, many fossil
neopterygians and primitive teleosts (upper margin of tail in Megalops and Tarpon), but are
absent in Tegeolepis, Cheirolepis, Styracopterus, Amia, Polypterus, chondrosteans, pycnodonts
and most living teleosts.

In Elonichthys the fringing fulcra are delicate with up to two pairs per lepidotrichial segment
in E. robisoni. In Lepisosteus and Meidiichthys (BMNH P. 1607) the basal fulcra grade into
fringing fulcra and in many fossil actinopterans (parasemionotids, Caturus, Semionotus,
Dapedium, Ophiopsis) the fringing fulcra are particularly stout and form a sturdy 'cutwater'.

Phylogenetic results

Interrelationships of actinopterygians

Although most recent authors have regarded the Cladistia as actinopterygians (Goodrich 1930,
Daget 1950, Gardiner 1973, Rosen et al. 1981, Forey & Gardiner 1981), some authorities have
considered them a separate group of osteichthyans (Jarvik 1942, 1980, Nelson 19696). Patterson
(1982), in a critical examination of the evidence, concluded that they are the sister-group of

RELATIONSHIPS OF PALAEONISCIDS

395

Fig. 147 Character phylogeny of major groups
of gnathostomes. Numbered characters refer
to synapomorphy scheme in text.

Actinopteri and this view is supported by the numerous synapomorphies listed by him. These
include teeth with an apical cap of acrodin, the reduction of the jugal canal to a horizontal
pit-line and the absence of a squamosal, presence of a valvula, otoliths formed of vaterite,
gill-arch and jaw musculature, a protractor hyomandibularis muscle and pectoral propterygium.

Cladistia also share with Actinopteri buried layers of ganoine on the surface of dermal bones
and ganoid scales with an anterodorsal process and peg-and-socket articulations. These
characters are missing from living Chondrostei; however, there is an anterodorsal process on the
scales of Chondrosteus (BMNH P. 41615) and I have also observed small pustules of ganoine on
some of the head bones of Chondrosteus. Since Chondrosteus is the sister-group of the
Acipenseridae (Patterson 1982: 253) the absence of these characters in Chondrostei must be
rated as secondary.

Other features common to cladistians and generalized actinopterygians, but absent in
Chondrostei, include a dentary with an enclosed sensory canal, a dermohyal, presupra-
cleithrum, a shield-like rostral with an enclosed ethmoid commissure, a dilatator operculi
muscle and a levator arcus palatini muscle inserting on the dorsolateral surface of the palate.

Finally, cladistians share with Actinopteri, including chondrosteans, a parasphenoid with an
ascending process and a posterior stem. I have argued elsewhere (Gardiner 1973: 116, and
above) that the parasphenoid has grown back independently on several occasions, and conclude
that the posterior stem in cladistians in non-homologous with that of chondrosteans which in
turn is non-homologous with that of other Actinopteri. I also regard the ascending processes as
non-homologous in cladistians and actinopterans. In actinopterans the ascending process is
developed in the spiracular groove, whereas that in cladistians is large and complicated with
medial, lateral and ventral components and has a different phylogenetic history (Patterson
1982).

Within the Actinopteri the Chondrostei and Neopterygii are generally regarded as
sister-groups (Nelson 1969b; Patterson 1973, 1982). Shared synapomorphies include a
spiracular canal which opens into the fossa bridgei, an ascending process of the parasphenoid
which reaches or enters the spiracular canal, a supramedullary hemopoietic organ (presumed
to occupy the lateral cranial canal of fossil forms), three ossifications or cartilages in the hyoid

396 B. G. GARDINER

bar, a swimbladder and nasal rosette, a perforated pectoral propterygium embraced by marginal
rays, a middle region to the endoskeletal pectoral girdle, and a dermopterotic.

In summary, if only Recent fishes are considered the most economical distribution of
character states places Cladistia as the sister-group of Chondrostei plus Neopterygii.

The Neopterygii may further be divided into Ginglymodi and Halecostomi (Patterson 1973).
Neopterygian synapomorphies include:

symplectic developed as an outgrowth of the hymandibular cartilage;

quadratojugal which braces the quadrate;

premaxilla with an internal process lining the nasal pit;

reduction of body lobe of tail;

symmetrical caudal fin in which the outer principal rays of the upper lobe approximately equal

those of the lower lobe in length;

dorsal and anal fin-rays equal in number to their endoskeletal supports;
antorbitals;

articular with a coronoid process;
dermal basipterygoid process;
hyomandibula with opercular process;

palatoquadrate disconnected from dermal cheek bones posteriorly and dorsally;
preopercular with narrow dorsal limb no longer in contact with the maxilla.

Primitively in actinopterygians the palatoquadrate is assumed to have been tied to the maxilla
anteriorly, the propercular dorsally and the quadratojugal posteriorly, thereby entirely
enclosing the adductor mandibulae muscle in a tube of bone. Although the preopercular is
missing in chondrosteans the adductor mandibulae muscle is still inserted on the outer surface of
the palatoquadrate as in selachians. However, in neopterygians, where both dorsal and
posterior contacts between the dermal cheek bones and the palatoquadrate have been lost, the
adductor mandibulae is inserted on the neurocranium as well as the palatoquadrate. In
cladistians, although contact is maintained between the palatoquadrate and preopercular
posteriorly, the palatoquadrate turns inwards dorsally, not outwards as in acanthodians,
selachians and chondrosteans, to lie in a nearly horizontal position with its edge fitting into a
groove on the parasphenoid (a unique condition). The adductor mandibulae muscle is inserted
on the preopercular, palatoquadrate, hyomandibula and neurocranium. I consider the loss of
the dorsal connection between the palatoquadrate and preopercular and the concomitant
insertion of the adductor muscle on the neurocranium to have occurred independently in
cladistians and in neopterygians.

In halecostomes the anterior connection between the palatoquadrate and maxilla is also lost
and the group is characterized by a mobile maxilla with a peg-like internal head. Other
synapomorphies include:

enlarged posterior myodome occupying at least half the distance between the pituitary fossa

and the vagus foramen;
pre-ethmoids;

large post-temporal fossa without an endoskeletal roof;
supramaxilla and interopercular;
epibranchials with uncinate processes;

intercalar with membranous outgrowths over the surface of the otic region;
loss of the quadratojugal as an independent element;
development of a post-temporal process.

A post-temporal process is also said to be present in Polypterus (Allis 1922: 206) and
Lepisosteus (Jessen 1972: pi. 9, fig. 4), but these processes bear little resemblance to those in
halecostomes and are therefore regarded as convergent.

Having outlined a phylogeny of Recent actinopterygians it is now possible to establish the
approximate order of origin of specializations within the group.
The order Palaeonisciformes is generally considered to include the most primitive

RELATIONSHIPS OF PALAEONISCIDS 397

actinopterygians (Kasantseva-Seleznevak 1981) and Mimia and Moythomasia have both been
assigned to that Order (Gardiner 1973). In the past most of the propositions of relationships of
palaeoniscids have been couched in ancestor-descendent sequences (e.g. Gardiner 1963, 1967),
but a cladistic examination (Patterson 1982) has revealed that the palaeoniscids constitute a
paraphyletic group. It further showed that Cheirolepis is the sister-group of the Actinopterygii,
Mimia is the sister-group of the Actinopteri + Moythomasia, but that the vast majority of the
palaeoniscids are more closely related to the Neopterygii. It also revealed one major anomaly,
the absence of a posterior myodome in Lepisosteus. A posterior myodome is present in all
advanced palaeoniscids and in halecostomes (apart from a few teleosts where it is assumed to
have been secondarily lost). However, from the congruence of other features, its absence in
Lepisosteus is rated as secondary.

The advanced features are now listed in their approximate order of origin in a synapomorphy
scheme which is summarized in Fig. 147 and in the classification.

A. Cheirolepis shares with other actinopterygians:

1. An anterodorsal angle or process to the scale

2. Ganoine

3. Dentary with enclosed mandibular sensory canal

4. Jugal canal not joined to infraorbital canal (except in Polyodon); instead it is reduced to a
horizontal pit-line

5. Squamosal absent

6. Palatoquadrate joined to preopercular dorsally, and to quadratojugal posteriorly

7. Dermohyal covering head of hyomandibula only

8. Presupracleithrum

9. Pectoral propterygium

10. Otoliths formed of vaterite

11. A shield-like rostral with enclosed sensory canal commissure (except Chondrostei)

12. Basal fulcra on upper border of tail.

B. Polypterus has the foregoing characters (except 12) and shares with actinopterans:

13. Acrodin caps on all teeth

14. Peg-and-socket articulations between scales

15. Postcleithrum.

C. Mimia has all the foregoing characters (except 13: some teeth without caps) and shares the
following derived features with Recent actinopterans:

16. Perforated propterygium

17. Bases of marginal rays embrace propterygium

18. A middle region to pectoral girdle

19. Lateral cranial canal.

D. Moythomasia has all the foregoing characters and shares the following derived characters
with Recent actinopterans:

20. An ascending process on the parasphenoid which lies in the spiracular groove

21. Supra-angular.

E. Chondrostei have the foregoing characters apart from 3, 4, 6, 7, 8, 11, 14, 15 and 19, and
share with neopterygians:

22. A spiracular canal

23. An ascending process which reaches or enters spiracular canal

24. A fossa bridgei

25. A dermopterotic

26. Three ossifications or cartilages in the hyoid bar.

It is worth noting that Patterson (1982) has recorded acrodin caps on the teeth of large

398 B. G. GARDINER

Polyodon, a supra-angular in Chondrosteus and I have observed ganoine on the dermal bones of
Chondrosteus and scales with an anterodorsal angle.

F. Kentuckia has the characters of A-E (except perhaps 24, but characters 8, 9, 10, 12, 15, 16,
17, 18, 21, 26 not known) and shares the following derived character with primitive
neopterygians:

27. Myodome.

G. Aetheretmon has the foregoing characters apart from 8 and 15 (19 not known) and shares the
following derived character with neopterygians:

28. Suborbitals.

H. Cosmoptychius has the foregoing characters apart from 8 and shares the following derived
character with neopterygians:

29. Unpaired myodome.

I. Pteronisculus has the foregoing characters and shares the following derived characters with
neopterygians:

30. Dermal basipterygoid process

31. 'Prismatic' ganoine.

The term 'prismatic ganoin' has been used by 0rvig (1967a, 1978) to describe the appearance,
when viewed by polarized light, of the superposed generations of ganoine on the scales and
dermal bones of many palaeoniscids and neopterygians. I have seen this type of ganoine on the
scales of many palaeoniscids including Elonichthys, Rhadinichthys, Gonatodus, Palaeoniscus,
Acrolepis, Gyrolepis, Pygopterus, Centrolepis, Eurynotus and Pteronisculus. It was not seen in
Cheirolepis, Polypterus, Mimia, Moythomasia, Stegotrachelus or Australosomus .

Many Carboniferous palaeoniscids appear to be either interchangeable with or to fit
somewhere near Pteronisculus in this cladogram; these include Elonichthys, Rhadinichthys,
Gonatodus and Kansasiella. Unfortunately, the characters for which information is available do
not provide sufficient evidence for establishing hypotheses of relationships.

J. Platysomus has the foregoing characters apart from 6, 7, 8, 11 and 28 (9, 10, 16, 17, 18, 19, 22,
24, 26, 29 not known), and shares the following derived characters with neopterygians:

32. Palatoquadrate detached from preopercular and maxilla dorsally and from the
preopercular posteriorly

33. Preoperculum and hyomandibula almost vertical.

K. Boreosomus has all the foregoing characters with the exception of 6, 7, 8 and 15. It shares the
following derived features with neopterygians:

34. Hyomandibula with an opercular process

35. Fringing fulcra on upper border of tail.

The tail of Boreosomus is abbreviate heteroceral, with the body axis not quite reaching the end
of the dorsal lobe. On the upper border basal fulcra appear to grade into fringing fulcra, much as
in Lepisosteus. An opercular process also occurs in Polypterus but this is rated as convergent.
Boreosomus marks the end of the grade group (stem-group) Palaeonisciformes, but between
it and the neopterygians are several more advanced groups formerly designated 'subholosteans' .
These include the Perleididae. Other 'subholosteans' such as the Redfieldiidae are either
interchangeable with or fit somewhere near Boreosomus in the phylogeny.

L. Perleididae have all the foregoing characters apart from 6, 7 and 8 (29 not known) and share
the following derived characters with neopterygians:

36. Elongate upper caudal fin-rays

37. Dorsal and anal fin-rays equal in number to their supports

38. Antorbitals and premaxillae

RELATIONSHIPS OF PALAEONISCIDS 399

39. Premaxilla with an internal (nasal) process which lines the nasal pit

40. A dilatator fossa.

The dorsal and anal fin-rays also equal their supports in haplolepids, Bobasatrania, Luganoia
and Platysiagum.

M. Ginglymodi have all the foregoing characters except 6, 7, 27 and 29 and uniquely share with
halecostomes:

41. A sympletic developed as an outgrowth of the hyomandibular cartilage and a
quadratojugal which braces or supports the quadrate

42. Body lobe of tail reduced, symmetrical caudal fin in which outer principal rays of upper
lobe approximately equal in length those of lower

43. Maxilla and preopercular lose contact with posterior margin of palatoquadrate

44. Articular with coronoid process

45. Preopercular with a narrow dorsal limb.

A coronoid process is developed in Polypterus, where it is composed solely of the prearticular.
This is considered convergent with that of neopterygians which always incorporates lateral
investing bones. The coronoid process of Luganoia is considered synapomorphous with that of
neopterygians.

N. Pycnodontiformes have all the foregoing characters apart from 6, 7, 8, 11, 12, 15, 21, 28 and
35 (18, 19, 22, 24, 30, 31, 40 and 41 not known) and uniquely share with halecostomes:

46. A mobile maxilla with peg-like internal head

47. A large posterior myodome

48. Large post-temporal fossa.

In Macromesodon (Nursall 1966) the preopercular canal appears to have joined the infraorbital,
much as in paddlefishes; this is assumed to be secondary. Although the jaw articulation is not
known with certainty, there is a rod-like bone buttressing the quadrate in Microdon which has
the appearance of a quadratojugal.

O. Lepidotes has the foregoing characters apart from 6, 7, 9 and 11 (16 not known) and uniquely
shares with halecostomes:

49. A supramaxilla

50. An interopercular

51. Epibranchials with uncinate processes

52. A post-temporal with an internal process of halecostome type

53. Post-temporal fossa confluent with fossa bridgei.
Epibranchials with uncinate processes also occur in Australosomus .

P. Halecostomi have all of the foregoing characters except 6, 7 and 11, and in addition possess:

54. An intercalar with membranous outgrowths over the surface of the otic region

55. A quadratojugal which no longer remains as an independent element.

Classification

The broad phylogenetic results based on the synapomorphies cited in the text are summarized in
the following outline classification, which follows the conventions of Patterson & Rosen (1977).
't' indicates an extinct group.

SUPERCLASS Gnathostomata
CLASS Chondrichthyes

SUBCLASS Selachii

SUBCLASS Holocephali
Plesion tAcanthodii
Plesion tPlacodermi
CLASS Osteichthyes

400 B. G. GARDINER

SUBCLASS Actinopterygii
INFRACLASS Cladistia
INFRACLASS Actinopteri
SUPERDIVISION Chondrostei
SUPERDIVISION Neopterygii
DIVISION Ginglymodi
DIVISION Halecostomi

SUBDIVISION Halecomorphi
SUBDIVISION Teleostei
SUBCLASS Sarcopterygii
Plesion t Eusthenopteron
INFRACLASS Actinistia
Plesion tAkinetia

ORDER tPorolepiformes
ORDER tYoungolepiformes
INFRACLASS Choanata
SUPERDIVISION Dipnoi
SUPERDIVISION Tetrapoda

The more particular results concerning the interrelationships of actinopterygians are
embodied in the classification below.

SUBCLASS Actinopterygii
Plesion ^Cheirolepis
INFRACLASS Cladistia
INFRACLASS Actinopteri
Plesion ^Mimia
Plesion t Moythomasia
SUPERDIVISION Chondrostei
SUPERDIVISION Neopterygii
Plesion ^Kentuckia
Plesion t Aetheretmon
Plesion "fCosmoptychius
Plesion ^Pteronisculus
Plesion ^Platysomus
Plesion ^Boreosomus
Plesion tPerleididae
DIVISION Ginglymodi
Plesion tPycnodontiformes
Plesion ^Lepidotes
DIVISION Halecostomi

Relationships of actinopterygians

A character phylogeny of Recent and fossil gnathostomes has been presented by Rosen et al.
(1981), in which they have suggested that the acanthodians are the most plesiomorphous
gnathostomes, lungfishes are the sister-group of tetrapods and that rhipidistians are
paraphyletic and form a stem-group series between actinopterygians and lungfishes. The
phylogeny given below is essentially that of Rosen et al. except that the chondrichthyans are
considered the most plesiomorphic gnathostomes and the placoderms the sister-group of
osteichthyans. In addition the Porolepiformes + Youngolepis are considered the sister-group of
the dipnoans and tetrapods (see Fig. 147).

A. Chondrichthyans share with other gnathostomes:

1. A lower jaw supported by a palatoquadrate and hyomandibula. Hyomandibula which
contacts neurocranium

RELATIONSHIPS OF PALAEONISCIDS

401

Plate 4 Mimia toombsi Gardiner & Bartram. (a), (b) coronoid teeth, x 150. (c) anterior end of
palatine with accompanying maxillary tooth row, x 30. (d-f) palatine teeth, (d) x 100, (e) x 200,
(f) x500. Scanning electron micrographs, all from BMNH P.53252

402 B. G. GARDINER

2. A hyoid bar connecting the branchial apparatus with the hyomandibula

3. Anterior branchial arches consisting of hypobranchial which articulates with a
basibranchial, ceratobranchial, epibranchial and pharyngobranchial elements

4. A cephalic lateral-line system that includes the following canal sections: supraorbital,
infraorbital, supratemporal, mandibulo-preopercular, and jugal that joins the infra-
orbital and preopercular

5. Paired pectoral and pelvic appendages with internal supporting girdles and radials

6. Three semicircular canals.

B. Acanthodians also have the following derived features which they share with placoderms and
osteichthyans:

7. Shoulder-girdle with ventral dermal plates and three perichondral ossifications

8. Operculogular series of dermal plates

9. Three ossfications in palatoquadrate

10. Two ossifications in Meckelian cartilage

11. Two ossifications in basibranchial cartilage

12. Dentigerous dermal plates on dorsal surface of Meckelian cartilage

13. Splenial bone.

The hypothesis that acanthodians and osteichthyans are sister-groups (Miles 1964, 1971a, 1973a)
is supported by the presence of branchiostegal rays (8 above) and ventral dermal plates on
the shoulder girdles (7 above), while the suggestion that they are the sister-group of
chondrichthyans (Nelson 1968) is sustained by the posterior position of the gill skeleton and the
posterior orientation of the pharyngobranchials. The double mandibular joint oi Acanthodes is
also said to be strikingly similar to that of amphistylic sharks (Miles 19730: 71). The posterior
position of the gill skeleton is a character shared only by selachians and acanthodians whereas
the Operculogular series and ventral plates of the shoulder girdle are found in acanthodians,
placoderms and osteichthyans. The double jaw joint is seen in selachians, acanthodians and
actinopterygians. However, since acanthodians uniquely share with placoderms and
osteichthyans three ossifications in the palatoquadrate cartilage and two in the Meckelian
cartilage, and have only a few ossification centres in the neurocranium, it seems more
economical to interpret them as the sister-group of the osteichthyans + placoderms.

C. Placoderms have the foregoing characters apart from 13 (3, 11 not known) and share with
osteichthyans:

14. Neurocranium protected by a series of large, interlocking dermal plates, some of which
possess descending laminae of membrane bone

15. True dermal shoulder girdle with lateral plates to which scapulocoracoid is attached (viz.
clavicle + cleithrum)

16. Supracoracoid foramen

17. Autopalatine articulates with postnasal wall

18. Dermal bone associated with the head of the hyomandibula (and completely covers the
hyomandibula)

19. Parasphenoid with teeth, spiracular groove and foramen for buccohypophysial canal.
Evidence for the association of placoderms with osteichthyans has been presented by Forey
(1980), who also cited the shared presence of endochondral bone. In my opinion (see p. 185)
placoderms possess calcified cartilage, never endochondral bone. The competing hypothesis
that placoderms and chondrichthyans are sister-groups (Stensio 1963a, Miles & Young 1977)
rests on two characters, the pelvic clasper and eye stalk.

Miles & Young (1977) have argued that pelvic and prepelvic claspers are most parsimoniously
explained as a unique specialization of chondrichthyans plus placoderms. But only the
ptyctodonts and holocephalans have both these structures. All other placoderms ai~ devoid of
claspers and selachians possess only pelvic claspers. Moreover the pelvic claspers in
chondrichthyans consist of a varied number of articulated, cartilaginous segments supported by
the distal end of the metapterygium, whereas that of ptyctodonts does not appear to have an

RELATIONSHIPS OF PALAEONISCIDS 403

endoskeleton, projects ventrally from the root of the fin, and is covered by a laterally-toothed
dermal plate. Distally the pelvic claspers are covered with dermal denticles in holocephalans,
but the apex is usually naked in selachians except in the squaloids etc., where it is provided with
one or more movable spines. The prepelvic claspers in ptyctodonts, like the pelvics, are
supported by dermal bone only. They consist of a pair of flat plates which in Ctenurella bear
spines. The corresponding structure in holocephalans is represented by a cartilaginous plate
(grooved in Callorhynchus) covered with dermal denticles. Because of their different
construction the pelvic and prepelvic claspers are accordingly rated as non-homologous in
chondrichthyans and ptyctodonts. Eye stalks, or their scars, have been recorded in three genera
of rhenanids, Radotina (Gross 1958: fig. 5A; Stensio 1969: fig. 51), Romundina (0rvig 1975: pi.
2, figs 1, 2) and Brindabellaspis (Young 1980: pi. 1, fig. 5; pi. 2, fig. 6), in arthrodires
(Buchanosteus Young 1979) and in ptyctodonts, and in many selachians (Oxynotus, Scyllium,
various myliobatids; Holmgren 1941). There are never any scars in xenacanths or hybodonts.
The eye stalk chondrifies independently in selachians (Holmgren 1940: 109) and often leaves no
trace of a scar in the wall of the orbit (e.g. Chlamydoselachus). The non-congruence with all
other characters suggests that the presence of eye stalks is either a primitive gnathostome
attribute or a chance similarity.

Finally Schaeffer (1975) considered placoderms to be the most primitive gnathostomes, based
on the structure of the palatoquadrate. He suggested that their 'omega-shaped' palatoquadrate,
in direct contact with the dermal cheek bones (viz. without a lateral cavity for the insertion of
the adductor mandibulae muscles), was the primitive condition. Forey (1980), in contrast,
considered it a synapomorphy of placoderms. However, the palatoquadrate of ptyctodonts and
gemuendinids can by no stretch of the imagination be considered either 'omega-shaped' or in
direct contact with the dermal cheek bones. Furthermore the adductor mandibulae muscle must
have been inserted on the lateral face of the palatoquadrate in both Ctenurella (Miles & Young
1977: figs 24-28) andJagorina (Stensio 1959: figs 61-64), much as in generalized gnathostomes.
An 'omega-shaped' palatoquadrate is therefore rated as a specialization of later placoderms,
setting them apart from the more primitive ptyctodonts and gemuendinids.

D. Actinopterygians have all the foregoing characters (but the dermohyal only covers the head
of the hyomandibula) and share with other osteichthyans:

20. Endochondral bone

21. Marginal teeth associated with premaxilla, maxilla and dentary (dental arcades), some of
which undergo successional replacement

22. Premaxilla canal-bearing

23. Lepidotrichia in the fins

24. Suprapharyngobranchials on the first two gill arches

25. Radials of fins never extending to the fin margin (except in tetrapods)

26. Interhyal

27. Hypohyal

28. Gular plates

29. Subopercular

30. Basibranchial with consolidated toothplates

31. Anteriorly-directed pharyngobranchials

32. Gill arches 1 and 2 articulating on the same basibranchial

33. Separate branchial levator muscles, interarcual muscles and transversi ventrali muscles

34. Lung or swimbladder.

Rosen et al. (1981) listed a dermal sclerotic ring as a synapomorphy at this level. However, a
sclerotic ring also occurs in placoderms (four plates), acanthodians (five plates) and
cephalaspids (four plates); see p. 253.

E. Eusthenopteron has the foregoing characters apart from 18 (but 33, 34 not known) and shares
the following derived features with actinistians, porolepiforms, dipnoans and tetrapods:

35. Exclusively metapterygial pectoral and pelvic fins, supported by a single basal

404 B. G. GARDINER

36. Teeth with enamel

37. Sclerotic ring of more than 12 segments

38. Enlarged otic or ascending process of palatoquadrate which articulates or fuses with
neurocranium above the basitrabecular process

39. Submandibulars

40. Hyomandibular facet bilobed or double.

The palatoquadrate also articulates with the neurocranium in Acanthodes, but the articulation
point is behind the postorbital process and is therefore rated as convergent, as is the fusion of the
palatoquadrate in holocephalans. Rosen et al. (1981) also cite the presence of an anocleithrum
as a synapomorphy at this level.

F. Actinistians have the characters of A, B, C, D and E apart from 3, 16, 18, 27 and 39, and share
with Porolepiformes, dipnoans and tetrapods:

41. An unornamented anocleithrum

42. Clavicle large relative to cleithrum, and high pectoral appendage insertion

43. Pectoral and pelvic appendages with long muscular lobes and structurally similar
endoskeletal supports

44. Preaxial side of pectoral fin endoskeleton rotated to postaxial position

45. A series of bones (the supraorbital-tectal series) lateral to the frontals and nasals which
carry the supraorbital sensory canal

46. Presence of a rostral organ or labial cavity

47. A single, broad basibranchial

48. Last gill arch articulates with base of preceding arch

49. Reduction or loss of hypobranchials

50. An inferior vena cava and pulmonary vein.

G-l. Porolepiformes have the foregoing characters (24, 33, 34, 43, 44, 46, 50 not known) apart
from 16, 18, 45 and 49, and share with dipnoans and tetrapods:

51. The immobilization of the intracranial joint

52. Cosmine pore-canal system in which the mesh canals are without a horizontal partition
and the pore canals are enamel-lined.

That the intracranial joint was immobilized in porolepiforms is deduced from Glyptolepis
(Jarvik 1972), in which the palate is fused to the postnasal wall anteriorly, into the 'fossa
autopalatina' medially and to the basipterygoid process posteromedially, and where the
articulation between the ascending process and the neurocranium is absent. This deduction is
strenthened by the suggestion (below; see Jessen 1975: 213) that the Youngolepididae
(Youngolepis and Powichthys) is the sister-group of the porolepiforms, since a suture often
exists between the two shields of the skull roof in Youngolepis (Chang 1982) despite the fact that
the underlying endocranium is ossified as a single piece. Moreover most of the specimens of
Youngolepis that have been collected (Chang 1982: 7) are separate anterior cranial portions,
much as in porolepids.

G-2. Youngolepididae (Youngolepis + Powichthys) also uniquely share with porolepiforms:

53. A 'fossa autopalatina' (Chang 1982: pi. 15A)

54. Foramen for the pituitary vein anterodorsal to the basipterygoid process (Chang 1982: 77)

55. Vomers widely separated by internasal pits and parasphenoid; internasal ridge

56. Much enlarged, downwardly pointing basipterygoid process.

Powichthys also shares with actinistians and dipnoans a series of bones lateral to the frontals
which carry the supraorbital canal. Youngolepids may be distinguished from porolepids by the
non-dendrodont form of their teeth.

A phylogeny of Recent gnathostomes rates the intracranial joint as a unique feature of
Latimeria. Nevertheless there are several hypotheses about this structure, some of which have
been generated to satisfy the assumption that rhipidistians rather than lungfishes are closer to
tetrapods.

RELATIONSHIPS OF PALAEONISCIDS

405

Plate 5 Mimia toombsi Gardiner & Bartram. (a) dentary tooth, x 30. (b) dentary tooth, x 150. (c)
branchiostegalray, x20. (d) posterior tip of branchiostegal ray, x 60. (e) ornamentation of clavicle
base, x40. (f) dorsal spine of clavicle, x 100. Scanning electron micrographs, (a), (b) from BMNH
P.53252, (c), (d) from BMNH P.56489, (e), (f) from BMNH P. 56484.

406 B. G. GARDINER

The hypothesis that the intracranial joint is a primitive gnathostome character (Jarvik 1972)
has been rejected because it leads to unacceptable phylogenetic conclusions (Miles 1977: 312;
Forey 1980: 382). Another hypothesis, that it is a shared specialization of actinistians and
choanates, separating this group from dipnoans (Miles 1977: 51), is also rejected, since no extant
choanate possesses such a joint and its occurrence in fossil choanates is dubious (Rosen et al.
1981: 259; Gardiner 1983). The suggestion by Bjerring (1973) that the intracranial joint is not
homologous in choanates and actinistians, which was arrived at by comparing the neurocrania of
Latimeria and Eusthenopteron with the embryological condition in other gnathostomes, cannot
be checked and is therefore regarded as speculation. The phylogeny outlined here interprets the
intracranial joint as homologous in rhipidistians and actinistians, and therefore as a primitive
feature of sarcopterygians. Further, it assumes that the joint has been lost once in porolepiforms
(including youngolepidids) and choanates (dipnoans and tetrapods).

H. Dipnoans have all the foregoing characters except 2, 3, 18, 21, 22, 24, 26, 31 and 36, and
those of G-2, and uniquely share with primitive tetrapods:

57. A choana

58. A labial cavity

59. Second metapterygial segment of paired appendages composed of paired, subequal
elements that are functionally joined distally

60. Two primary joints in each paired appendage, between the endoskeletal girdle and the
unpaired basal element, and between the basal element and the paired elements of the
second segment. In the pectoral appendage, the preaxial member of paired elements with
a ball-and-socket joint with the basal element and the postaxial member articulating on
dorsal (postaxial) margin of basal element

61. Reduction in ratio of dermal fin-rays to supports in paired appendages

62. Muscles in paired fins segmented

63. Fusion of right and left pelvic girdles to form pubic and ischial processes. Presence of
prepubic processes

64. Tetrapodous locomotion

65. Hyomandibula non-suspensory, reduced and associated with otic recess

66. Interhyal absent

67. Pharyngobranchials absent

68. Pterygoids joined in mid-line anteriorly, excluding parasphenoid from roof of mouth

69. Autopalatine absent

70. Elongation of snout region

71 . Two pairs of dermal bones attached to the otico-occipital region of braincase posterior to
parietals

72. Dentary with an oral pit-line

73. In soft anatomy, structure of lung, pulmonary circulation, two-chambered auricle, ventral
aorta as a truncus, glottis and epiglottis, telolecithal jelly-coated bipolar egg, ciliation of
the larva, pituitary structure including neurohypophysial hormone, lens proteins and bile
salts, and gill-arch muscles.

Fusion of right and left pelvic girdles and prepubic processes also occur in many selachians, but
this is considered convergent.

This phylogeny is summarized in the classification above (p. 399).

Acknowledgements

I especially would like to thank Colin Patterson and Peter Forey for the numerous discussions
we have had over the years on the interrelationships of fishes. I also wish to thank Dr H. W. Ball,
Keeper of Palaeontology, British Museum (Natural History), for the loan of specimens and for
making facilities available to me.

RELATIONSHIPS OF PALAEONISCIDS 407

References

Agar, W. E. 1906. The development of the skull and visceral arches in Lepidosiren and Protopterus. Trans.

R. Soc. Edinb., 45: 49-64, pis 1-3.
Aldinger, H. 1937. Permische Ganoidfische aus Ostgronland. Meddr Gr0nland, Copenhagen, 102: 1-392,

pis 1-44.
Allis, E. P. 1889. The anatomy and development of the lateral line system in Amia calva. J. Morph.,

Boston, 2: 463-568, pis 30-42.
1897. The cranial muscles and cranial and first spinal nerves in Amia calva. J. Morph., Boston, 12:

487-808, pis 20-38.
1903. The skull, and the cranial and first spinal muscles and nerves in Scomber scomber. J. Morph.,

Boston, 18: 45-328, pis 3-12.

1914a. Certain homologies of the palatoquadrate of selachians. Anat. Anz., Jena, 45: 353-373.

1914ft. The pituitary fossa and trigeminofacialis chamber in selachians. Anat. Anz., Jena, 46:

225-253, 1 fig.
1915. The homologies of the hyomandibula of the gnathostome fishes. /. Morph., Boston, 26:

563-624.
1919. The myodome and trigemino-facialis chamber of fishes and the corresponding cavities in higher

vertebrates. J. Morph., Boston, 32: 207-326, pis 1-4.
1922. The cranial anatomy of Polypterus, with special reference to Polypterus bichir. J. Anat.,

London, 56: 189-294, pis 3-24.
1925. On the origin of the V-shaped branchial arch in the Teleostomi. Proc. zool. Soc. Lond., 1925:

75-77.
1934. Concerning the course of the latero-sensory canals in recent fishes, prefishes and Necturus. J.

Anat., London, 68: 361-412, 12 figs.
Andrews, S. M. 1973. Interrelationships of crossopterygians. In Greenwood, P. H., Miles, R. S. &

Patterson, C. (eds), Interrelationships of Fishes: 137-177, 5 figs. London.
& Westoll, T. S. 1970. The postcranial skeleton of rhipidistian fishes excluding Eusthenopteron.

Trans. R. Soc. Edinb., 68: 391-489, pis 1-15.
Hauin, K. A. & Lund, R. 1974. Vertebral centra in Haplolepis (Haplolepidae, Paleonisciformes) from the

Allegheny group, Pennsylvanian. J. Paleont., Tulsa, 48: 199-200, 1 fig.
Beer, G. R. de 1926. Studies on the vertebrate head. II. The orbito-temporal region of the skull. Q. Jl

microsc. Sci., London, 70: 263-370, 133 figs.
1927. The early development of the chondrocranium ofSalmofario. Q. Jl microsc. Sci., London, 71:

259-312, 52 figs.
1937. The development of the vertebrate skull, xxiv + 552 pp., 143 pis. Oxford.

- & Moy-Thomas, J. A. 1935. On the skull of Holocephali. Phil. Trans. R. Soc., London, (B) 224:

287-312, 19 figs.
Beltan, L. 1968. Lafaune ichthyologique de I'Eotrias du N.W. de Madagascar: le neurocrane. 135 pp., 55

pis. Paris, C.N.R.S.
Berman, D. S. 1968. Lungfish from the Lueders formation (Lower Permian, Texas) and the Gnathorhiza-

Lepidosirenid ancestry questioned. /. Paleont., Tulsa, 42: 827-835, 4 figs.
Bernacsek, G. M. 1977. A lungfish cranium from the Middle Devonian of the Yukon territory, Canada.

Palaeontographica, Stuttgart, (A) 157: 175-200, pis 1^.
Bertmar, G. 1959. On the ontogeny of the chondral skull in Characidae, with a discussion of the

chondrocranial base and the visceral chondrocranium in fishes. Acta zool., Stockh., 40: 203-364, 85

figs.
1963. The trigemino-facialis chamber, the cavum epiptericum and the cavum orbitonasale, three

serially homologous extracranial spaces in fishes. Acta zool., Stockh., 44: 329-344, 12 figs.
1965. On the development of the jugular and cerebral veins in fishes. Proc. zool. Soc. Lond., 144:

87-130, 25 figs.

1966. The development of the skeleton, blood vessels and nerves in the dipnoan snout, with a

discussion on the homology of the dipnoan posterior nostrils. Acta zool., Stockh., 47: 81-150, 37 figs.
Bjerring, H. C. 1967. Does a homology exist between the basicranial muscle and the polar cartilage?

Colloques int. Cent. natn. Rech. scient., Paris, 163: 223-267, 4 pis.

- 1971. The nerve supply to the second metamere basicranial muscle in osteolepiform vertebrates, with

some remarks on the basic composition of the endocranium. Acta zool., Stockh., 52: 189-225, 22 figs.

1972. The nervus rarus in coelacanthiform phylogeny. Zooligica Scr., Stockholm, 1: 57-68, 7 figs.

1973. Relationships of Coelacanthiformes. In Greenwood, P. H., Miles, R. S. & Patterson, C. (eds),

Interrelationships of fishes: 179-205, 2 pis. London.

B. G. GARDINER

- 1977. A contribution to structural analysis of the head of craniate animals. Zoologica Scr.,
Stockholm, 6: 127-183, 35 figs.

1978. The 'intracranial joint' versus the ventral otic fissure. Acta zool., Stockh., 59: 203-214, 8 figs.

Bridge, T. W. 1878. On the osteology of Polyodon folium. Phil. Trans. R. Soc., London, 169: 683-733,
pis 55-57.

- 1888. Some points in the cranial anatomy of Polypterus. Proc. Bghamphil. Soc., 6: 118-130, pis 1-2.

- 1896. The mesial fins of ganoids and teleosts. J. Linn. Soc., London, (Zool.) 25: 530-602, pis 21-23.

- 1898. On the morphology of the skull in the Paraguayan Lepidosiren and in other dipnoids. Trans,
zool. Soc. Lond., 14: 325-376, pis 28-29.

Broili, F. 1933. Ein Macrapetalichthyidae ausden Hunriickschiefern. Sber. bayer. Akad. Wiss., Munich, 6:

6: 417-437, 7 figs, 1 pi.
Broom, R. 1914. Croonian Lecture: On the origin of mammals. Phil. Trans. Roy. Soc., London, (B) 206:

1-48, pis 1-7.
Brough, J. 1939. The Triassic fishes of Besano, Lombardy. ix + 117 pp., 7 pis. London, Brit. Mus. (Nat.

Hist.).
Brunton, C. H. C., Miles, R. S. & Rolfe, W. D. I. 1969. Gogo expedition 1967. Proc. geol. Soc., London,

1655: 80-83.
Budgett, J. S. 1901. On the breeding-habits of some West- African fishes, with an account of the external

features in development of Protopterus annectens, and a description of the larva of Polypterus lapradei.

Trans, zool. Soc. Lond., 16: 115-136, pis 10-11.

1902. On the structure of the larval Polypterus. Trans, zool. Soc. Lond., 16: 315-346.

Bugajew, J. 1929. Uber das Pharyngomandibulare der Knorpelganoiden. Anat. Anz., Jena, 67: 98-110.
Bystrov, A. P. & Efremov, J. A. 1940. [Benthosuchus sushkini Efr. , a labyrinthodont from the Eotriassic of

Sharzhenga River.] Trudy paleont. Inst., Moscow, 10: 1-152, 88 figs. [In Russian, Engl. summary].
Campbell, K. S. W. & Barwick, R. E. 1982. The neurocranium of the primitive dipnoan Dipnorhynchus

sussmilchi (Etheridge). J. Vert. Paleont., Norman, Okla., 2: 286-327, 26 figs.
Case, E. C. 1935. Description of a collection of associated skeletons of Trimerorhachis . Contr. Mus.

Paleont. Univ. Mich., Ann Arbor, 4: 227-274, pis 1-11.
Chang Mee-Mann 1982. The braincase of Youngolepis, a Lower Devonian crossopterygian from Yunnan,

south-western China. 113 pp., 13 pis. Stockholm.
Compagno, L. J. V. 1973. Interrelationships of living elasmobranchs. In Greenwood, P. H., Miles, R. S. &

Patterson, C. (eds), Interrelationships of fishes: 15-61, pis 1-2. London.

- 1977. Phyletic relationships of living sharks and rays. Am. Zool., Utica, 17: 303-322, 13 figs.
Daget, J. 1950. Revision des affinites phylogenetiques des Polypterides. Mem. Inst. fr. Afr. noire, Dakar,

11: 1-178, 55 figs.

- & d'Aubenton, F. 1957. Developpement et morphologic du crane d'Heterotis niloticus Ehr. Bull.
Inst. fr. Afr. noire, Dakar, (A) 19: 881-936, 31 figs.

Danforth, C. H. 1912. The heart and arteries of Polyodon. J. Morph., Boston, 23: 409-454, 19 figs.

Daniel, J. F. 1934. The elasmobranch fishes (3rd edn). 232 pp., 270 figs. Berkeley.

Davidoff, M. von 1880. Beitrage zur vergleichenden Anatomic der hinteren Gliedmassen der Fische, ii.

Ganoidei Holostei. Morph. Jb., Leipzig, 6: 433^68, pis 21-23.

Dean, B. 1907. Notes on acanthodian sharks. Am. J. Anat., Baltimore, 7: 209-222, 36 figs.
Denison, R. H. 1968. The evolutionary significance of the earliest known lungfish, Uranolophus. In 0rvig,

T. (ed.), Current problems of lower vertebrate phylogeny. Nobel Symposium, Stockholm, 4: 247-257,

9 figs.

- 1969. New Pennsylvanian lungfishes from Illinois. Fieldiana, Geol., Chicago, 12: 193-211, 8 figs.
1975. Evolution and classification of placoderm fishes. Breviora, Cambridge, Mass., 432: 1-24, 6 figs.

- 1978-79. Placodermi. In Schultze, H. P. (ed.), Handbook ofPaleoichthyology, 2. vi + 128 pp., 94 figs
(1978). Acanthodii. Loc. cit., 5. vi + 62 pp., 35 figs (1979). Stuttgart.

Devillers, C. 1958. Le Crane des Poissons. In Grasse, P. P. (ed.), Traite de Zoologie, 13 (1): 551-687, figs

345-443. Paris.
Dick, J. R. R. 1978. On the Carboniferous shark Tristychius arcuatus Agassiz from Scotland. Trans. R.

Soc. Edinb., 70: 63-109, pis 1-2.
Driiner, L. 1901. Studien zur Anatomic der Zungenbein-, Kiemenbogen- und Kehlkopfmuskeln der

Urodelen. 1 Teil. Zool. Jb., Jena, (Anat.) 15: 435-622, pis 25-31.
Edgeworth, F. H. 1926. On the hyomandibula of Selachii, Teleostomi and Ceratodus. J. Anat., London,

60: 173-193, 23 figs.

- 1935. The cranial muscles of vertebrates. \ + 493 pp., 841 figs. Cambridge.

Edinger, T. 1929. Uber Knocherne Scleralringe. Zool. Jb., Jena, (Anat.) 51: 163-226, 61 figs.

RELATIONSHIPS OF PALAEONISCIDS 409

El-Toubi, M. R. 1949. The development of the chondrocranium of the spiny dogfish, Acanthias vulgaris

(Squalus acanthias). J. Morph., Boston, 84: 227-279, 16 figs.
Ewart, J. C. 1895. The lateral sense organs of Elasmobranchs. I. The sensory canals ofLaemargus. Trans.

R. Soc. Edinb., 37: 59-85, pis 1-2.
Forey, P. L. 1980. Latimeria: a paradoxical fish. Proc. R. Soc., London, (B) 208: 369-384, 2 figs.

- 1981. The coelacanth Rhabdoderma in the Carboniferous of the British Isles. Palaeontology,

London, 24: 203-229, 15 figs.

& Gardiner, B. G. 1981. J. A. Moy-Thomas and his association with the British Museum (Natural

History). Bull. Br. Mus. Nat. Hist., London, (Geol.) 35: 131-144, 3 figs.
Fox, H. 1965. Early development of the head and pharynx of Neoceratodus with a consideration of its

phylogeny. J. Zool. Lond., 146: 470-554, pis 1-3.
Frost, G. A. 1913. The internal cranial elements and foramina of Dapedius granulatus, from a specimen

recently found in the Lias at Charmouth. Q. Jl geol. Soc. Lond., 69: 219-222, 2 figs.
Fuchs, H. 1929. Uber das Os parahyoideum der anuren Amphibien und der Crossopterygier; nebst

Bemerkungen uber phylogenetische Wanderungen der Haut und Deckknochen. Morph. Jb., Leipzig,

63: 408-453, 34 figs.
Gardiner, B. G. 1960. A revision of certain actinopterygian and coelacanth fishes, chiefly from the Lower

Lias. Bull. Br. Mus. nat. Hist., London, (Geol.) 4: 239-384, pis 36-43.

- 1962. Namaichthys schroederi Giirich and other Palaeozoic fishes from South Africa. Palaeontology,
London, 5: 9-21, pi. 6.

1963. Certain palaeoniscoid fishes and the evolution of the actinopterygian snout. Bull. Br. Mus. nat.

Hist., London, (Geol.) 8: 255-325, pis 1-2.

- 1967. Further notes on palaeoniscoid fishes with a classification of the Chondrostei. Bull. Br. Mus.
nat. Hist., London, (Geol.) 14: 143-206, 24 figs.

- 1970. Osteichthyes. McGraw-Hill Yb. Sci. Technol., New York, 1970: 284-286, 3 figs.

- 1973. Interrelationships of teleostomes. In Greenwood, P. H., Miles, R. S. & Patterson, C. (eds),
Interrelationships of fishes: 105-135, 10 figs. London.

- 1980. Tetrapod ancestry: a reappraisal. /nPanchen, A. L. (ed.), The Terrestrial Environment and the
Origin of Land Vertebrates: 175-185, 1 fig. London.

- 1983. Gnathostome vertebrae and the classification of the Amphibia. Zool. J. Linn. Soc., London,
79 (1): 1-59, 16 figs.

— & Bartram, A. W. H. 1977. The homologies of ventral cranial fissures in osteichthyans. In Andrews,
S. M., Miles, R. S. & Walker, A. D. (eds), Problems in Vertebrate Evolution: 227-245, 8 figs. London.
& Miles, R. S. 1975. Devonian fishes of the Gogo formation, Western Australia. Colloques int. Cent.

natn. Rech. sclent., Paris, 218: 73-79, 2 figs.
Garman, S. 1888. On the lateral canal system of the Selachia and Holocephalia. Bull. Mus. comp. Zool.

Harv., Cambridge, Mass., 17: 57-119, pis 1-53.
Gaupp, E. 1899. Lehre vom Nerven- und Gefassystem. In Ecker, A. & Wiedersheim, R. (eds), Anatomie

des Frosches, 2: 1-548, 146 figs. Brunswick.
Gegenbaur, C. 1872. Untersuchungen zur vergleichenden Anatomie der Wirbeltiere, III. Das Kopfskelett

der Selachier, ein Beitrag zur Erkenntniss der Genese des Kopfskeletes der Wirbelthiere. x + 316 pp, 22

pis. Leipzig.
Goodrich, E. S. 1911. On the segmentation of the occipital region of the head in the Batrachia Urodela.

Proc. zool. Soc. Lond., 1911: 101-120, 23 figs.

1930. Studies on the structure and development of vertebrates, xxx + 837 pp., 754 figs. London.

Goujet, D. 1973. Sigaspis, un nouvel arthrodire du Devonien inferieur du Spitsberg. Palaeontographica,

Stuttgart, (A) 143: 73-88, pi. 7.

- 1975. Dicksonosteus, un nouvel Arthrodire du Devonien du Spitsberg. Remarques sur le squelette
visceral des Dolichothoraci. Colloques int. Cent. natn. Rech. scient., Paris, 218: 81-99, pis 1-5.

Graham-Smith, W. & Westoll, T. S. 1937. On a new long-headed dipnoan fish from the Upper Devonian of
Scaumenac Bay, P.Q., Canada. Trans. R. Soc. Edinb., 59: 241-266, pis 1-2.

Griffith, J. & Patterson, C. 1963. The structure and relationships of the Jurassic fish Ichthyokentema
purbeckensis. Bull. Br. Mus. nat. Hist., London, (Geol.) 8: 1-43, pis 1-4.

Gross, W. 1936. Beitrage zur Osteologie baltischer und rheinischer Devon-Crossopterygier. Palaeont. Z.,
Berlin, 18: 129-154, pis 7-8.

- 1937. Das kopfskelett von Cladodus wildungensis, I. Endocranium und Palatoquadratum.
Senckenbergiana, Frankfurt, 19: 80-107, 6 figs.

1942. Die Fischfaunen des baltischem Devons und ihre biostratigraphische Bedeutung. KorrespBl.

NaturfVer. Riga, 64: 373-436, 19 figs.

410 B. G. GARDINER

1950. Umbenennung von Aldingeria Gross (Palaeoniscidae; Oberdevon) in Moythomasia n. nom.

NeuesJb. Geol. Palaont. Mh., Stuttgart, 1950: 145.
1953. Devonische Palaeonisciden-Reste in Mittel- und Osteuropa. Palaeont. Z., Berlin, 27: 85-112,

pis 4-7.

- 1954. Zur Phylogenie der Schultergiirtels. Palaeont. Z., Berlin, 28: 20-40, 10 figs.

1956. Uber Crossopterygier und Dipnoer aus dem baltischen Oberdevon im Zusammenhang einer

vergleichenden Untersuchung des Porenkanalsystems palaozoischer Agnathen und Fische. K. svenska

VetenskAkad. Handl., Stockholm, (4) 5: 1-140, 124 figs, pis 1-16.
1957. Mundzahne und Hautzahne der Acanthodier und Arthrodiren. Palaeontographica, Stuttgart,

(A) 109: 1^0, pis 1-6.
1958. Uber die alteste Arthrodiren-Gattung. Notizbl. Hess. Landesamt. Bodenforsch. Wiesbaden, 86:

7-30, 5 figs.
1961. Lunaspis broilii und Lunaspis heroldi aus dem Hunsriickschiefer (Unterdevon, Rheinland).

Notizbl. hess. Landesamt. Bodenforsch. Wiesbaden, 89: 17^13, pis 2-7.
1962. Neuuntersuchung der Stensioellida (Arthrodira, Unterdevon). Notizbl. hess. Landesamt.

Bodenforsch. Wiesbaden, 90: 48-96, pis 1-2.
1963. Gemuendina stuertzi Traquair. Neuuntersuchung. Notizbl. hess. Landesamt. Bodenforsch.

Wiesbaden, 91: 36-73, pis 2-8.
1966. Kleine Schuppenkunde. Neues Jb. Geol. Palaont. Abh., Stuttgart, 125: 29-48, 7 figs.

- 1967. Uber Thelodontier-Schuppen. Palaeontographica, Stuttgart, (A) 127: 1-67, pis 1-7.

1973. Kleinschuppen, Flossenstacheln und Zahne von Fischen aus europaischen und nordamerika-

nischen Bonebeds des Devons. Palaeontographica, Stuttgart, (A) 142: 51-155, pis 26-36.
Hammarberg, F. 1937. Zur kentniss der ontogenetischen Entwicklung des Schadels von Lepidosteus

platostomus. Ada zool., Stockh., 18: 209-337, 65 figs.
Hancock, A. & Atthey, T. 1869. Notes on the remains of some reptiles and fishes from the shales of the

Northumberland Coal Field. Trans, nat. Hist. Soc. Northumb., Newcastle, 3: 14-68, pis 1-3.
Harris, J. E. 1938. The neurocranium and jaws of Cladoselache. Scient. Publs Cleveland Mus. nat. Hist. , 8:

8-12, 2 figs.
Harrisson, C. H. H. 1966. On the first halosaur leptocephalus, from Madeira. Bull. Br. Mus. nat. Hist.,

London, (Zool) 14 (8): 441-486, 1 pi.
Heintz, A. 1962. New investigation on the structure oiArctolepis from the Devonian of Spitsbergen. Arbok

norsk Polarinst., Oslo, 1961: 23-40, pis 1-2.

Heyler, D. 1969. Vertebres de I'Autunien de France. 259 pp., 52 pis. Paris, C.N.R.S.
Holmgren, N. 1940-43. Studies on the head in fishes. I. Development of the skull in sharks and rays. Acta

zool., Stockh., 21: 1-267, 185 figs (1940). II. Comparative anatomy of the adult selachian skull, with

remarks on the dorsal fins in sharks. Loc. cit., 22: 1-100, 74 figs (1941). III. The phylogeny of

elasmobranch fishes. Loc. cit., 23: 129-261, 54 figs (1942). IV. General morphology of the head in

fishes. Loc. cit., 24: 1-188, 85 figs (1943).

— & Stensio, E. A. 1936. Kranium und Visceralskelett der Akranier, Cyclostomen und Fische. In

Bolk, L. et al. (eds), Handbuch der Vergleichenden Anatomie, 4: 233-500, figs 203-373. Berlin &

Vienna.
Hubendick, B. 1943. Zur Kenntnis der Entwicklung des Primordialcraniums bei Leuciscus rutilus. Ark.

Zool., Stockholm, 34 (7): 1-35, 18 figs.
Huxley, T. H. 1876. Contributions to morphology. Ichthyopsida. No. 1. On Ceratodus forsteri, with

observations on the classification of fishes. Proc. zool. Soc. Lond., 1876: 24-59, 11 figs.
Ivanzoff, N. A. 1887. Der Scaphirhynchus. Bull. Soc. Nat. Moscou, (n.s.) 1: 1-41, pis 1-2.
Jaekel, 0. 1899. Uber die Zusammensetzung des Kiefers und Schiiltergurteles von Acanthodes. Z. dt. geol.

Ges., Berlin, 51: 56-60, 2 figs.
1906. Einige Beitrage zur Morphologic der altesten Wirbeltiere. Sber. Ges. naturf. Freunde Berl,

1906: 180-189, 7 figs.

1919. Die Mundbildung der Placodermen. Sber. Ges. naturf. Freunde Berl., 1919: 73-110, 17 figs.

Janvier, P. 1980. Osteolepid remains from the Devonian of the Middle East, with particular reference to

the endoskeletal shoulder girdle. In Panchen, A. L. (ed.), The terrestrial environment and the origin of

land vertebrates: 223-254, 12 figs. London.
— 1981. The phylogeny of the Craniata, with particular reference to the significance of fossil

'agnathans'. /. Vert. Paleont., Norman, Okla., 1: 121-159, 16 figs.
Jardine, N. 1970. The observational and theoretical components of homology; a study based on the

morphology of the dermal skull-roofs of rhipidistian fishes. J. Linn. Soc., London, (Biol.) 1: 327-361,

5 figs.

RELATIONSHIPS OF PALAEONISCIDS 411

Jarvik, E. 1942. On the structure of the snout of crossopterygians and lower gnathostomes in general.

Zool. Bidr. Upps., Stockholm, 21: 235-675, pis 1-17.
1944a. On the dermal bones, sensory canals and pit-lines of the skull in Eusthenopteron foordi

Whiteaves, with some remarks on E. sdve-soderberghi Jarvik. K. svenska VentenskAkad. Handl.,

Stockholm, (3) 21 (3): 1-48, 19 figs.
1944b. On the exoskeletal shoulder girdle of teleostomian fishes with special reference to

Eusthenopteron foordi Whiteaves. K. svenska VetenskAkad. Handl., Stockholm, (3) 21 (7): 1-32,

9 figs.

1947. Notes on the pit-lines and dermal bones of the head in Polypterus. Zool. Bidr. Upps.,

Stockholm, 25: 60-78, 6 figs.
1948. On the morphology and taxonomy of the Middle Devonian osteolepid fishes of Scotland. K.

svenska VetenskAkad. Handl., Stockholm, (3) 25: 1-301, pis 1-37.
1950. On some osteolepiform crossopterygians from the Upper Old Red Sandstone of Scotland. K.

svenska VetenskAkad. Handl., Stockholm, (4) 2: 1-35, 1 pi.
1952. On the fish-like tail in the ichthyostegid stegocephalians with descriptions of a new

stegocephalian and a new crossopterygian from the Upper Devonian of East Greenland. Meddr

Gr0nland, Copenhagen, 114: 1-90, pis 1-21.
1954. On the visceral skeleton in Euthenopteron with a discussion of the parasphenoid and

palatoquadrate in fishes. K. svenska VetenskAkad. Handl., Stockholm, (4) 5: 1-104, 47 figs.

1955. The oldest tetrapods and their forerunners. Sci. Mon., New York, 80: 141-154, 12 figs.

1960. Theories de revolution des vertebres reconsiderees a la lumiere des recentes decouvertes sur les

vertebres inferieurs. 104 pp., 30 figs. Paris.

- 1963. The composition of the intermandibular division of the head in fish and tetrapods and the
diphyletic origin of the tetrapod tongue. K. svenska VetenskAkad. Handl., Stockholm, (4) 9: 1-74,
26 figs.

1966. Remarks on the structure of the snout in Megalichthys and certain other rhipidistid

crossopterygians. Ark. Zool., Stockholm, (2) 19: 41-98, pis 1-5.

- 1967o. On the structure of the lower jaw in dipnoans: with a description of an early Devonian dipnoan
from Canada, Melanognathus canadensis gen. et sp. nov. /. Linn. Soc., London, (Zool.) 47: 155-183,
pis 1-6.

19676. The homologies of frontal and parietal bones in fishes and tetrapods. Colloques int. Cent.

natn. Rech. sclent., Paris, 163: 181-213, pis 1-4.
1968. The systematic position of the Dipnoi. In 0rvig, T. (ed.), Current problems of lower vertebrate

phylogeny. Nobel Symposium, Stockholm, 4: 223-245, 6 figs.
1972. Middle and Upper Devonian Porolepiformes from East Greenland with special reference to

Glyptolepis groenlandica n. sp. Meddr Gr0nland, Copenhagen, 187: 1-295, 107 figs.
1975. On the saccus endolymphaticus and adjacent structures in osteolepiforms, anurans and

urodeles. Colloques int. Cent. natn. Rech. sclent., Paris, 218: 191-211, 15 figs.
1977. The systematic position of acanthodian fishes. In Andrews, S. M., Miles, R. S. & Walker,

A. D. (eds), Problems In Vertebrate Evolution: 199-224, 16 figs. London.

1980. Basic structure and evolution of vertebrates. 1, xvi + 575 pp., 385 figs. 2, iii + 337 pp., 142 figs.

London.
Jessen, H. L. 1966. Die Crossopterygier des Oberen Plattenkalkes (Devon) der Bergisch-Gladbach-

Paffrather Mulde (Rheinisches Schiefergebirge) unter Beriicksichtigung von amerikanischem und

europaischem Onychodus-Material. Ark. Zool., Stockholm, (2) 18: 305-389, 12 figs.
1968. Moythomasia nitida Gross und M. cf. striata Gross, Devonische Palaeonisciden aus dem

Oberen Plattenkalk der Bergisch-Gladbach-Paffrather Mulde (Rheinisches Schiefergebirge). Palaeon-

tographica, Stuttgart, (A) 128: 87-114, pis 11-17.

1972. Schultergurtel und Pectoralflosse bei Actinopterygiern. Fossils Strata, Oslo, 1: 1-101, pis 1-25.

1975. A new choanate fish, Powichthys thorsteinssoni n.g., n.sp., from the early Lower Devonian of

the Canadian arctic archipelago. Colloques int. Cent. natn. Rech. sclent., Paris, 218: 213-222, 4 figs.
Jollie, M. 1962. Chordate morphology. 478 pp., 14 figs. New York.
1969. Sensory canals of the snout of actinopterygian fishes. Trans. III. St. Acad. Sci., Springfield, 59:

203-214, 3 figs.

1971. Some developmental aspects of the head skeleton of the 35-37 mm Squalus acanthias foetus.

J. Morph., Boston, 133: 17^0, 10 figs.
1980. Development of head and pectoral girdle skeleton and scales in Acipenser. Copeia, New York,

1980: 226-249, 10 figs.

412

B. G. GARDINER

- 1981. Segment theory and the homologizing of cranial bones. Am. Nat., Salem, Mass., 118:783-802
5 figs.

Kasantseva-Seleznevak, A. A. 1981. [Late Palaeozoic palaeoniscids of E. Kazahkstan (Systematics and

Phytogeny)]. Trudy Paleont. Inst., Moscow, 180: 1-139, 29 pis. [In Russian].
Kemp, A. 1977. The pattern of tooth plate formation in the Australian lungfish Neoceratodusfosteri Krefft.

Zool. J. Linn. Soc., London, 60: 223-258, pis 1-7.
Kemp, N. E. & Westrin, S. K. 1979. Ultrastructure of calcified cartilage in the endoskeletal tesserae of

sharks. /. Morph., Boston, 160: 75-101, 7 pis.
Kulkzycki, J. 1956. On the parasphenoid of the Brachythoraci. Acta palaeont. pol, Warsaw, 1 (2):

104-111, pis 1-2.
Lauder, G. V. 1980. Evolution of the feeding mechanism in primitive actinopterygian fishes: a functional

anatomical analysis of Polypterus, Lepisosteus, and Amia. J. Morph., Boston, 163: 283-317, 19 figs.
Lehman, J.-P. 1949. Etude d'un Pachycormus du Lias de Normandie. K. svenska VetenskAkad. Handl.,

Stockholm, (4) 1: 1-44, pis 1-9.

- 1952. Etude complementaire des poissons de 1'Eotrias de Madagascar. K. svenska VentenskAkad.
Handl., Stockholm, (4) 2: 1-201, pis 1-48.

1959. Les Dipneustes du Devonien superieur du Greenland. Meddr Gr0nland, Copenhagen, 160:

1-58, pis 1-21.

- 1966. Actinopterygii. In Piveteau, J. (ed.), Traite de Paleontologie, 4 (3): 1-242. Paris.

1969. Apropos de 1'anatomie de 1'endocrane des Actinopterygiens. Bull. Mus. natn. Hist. nat. Paris,

(4) 1: 367-370, 1 fig.
Loomis, F. B. 1900. Die Anatomic und der Verwandschaft der Ganoid- und Knochenfische aus der

Kreide-Formation von Kansas, U.S.A. Palaeontographica, Stuttgart, 46: 213-283, pis 19-27.
Lund, R. 1970. Fossil fishes from southwestern Pennsylvania. Part 1: Fishes from the Duquesne

Limestones (Conemaugh, Pennsylvanian). Ann. Carneg. Mus., Pittsburgh, 41: 231-261, 17 figs.
Luther, A. 1909. Beitrage zur Kenntnis von Muskulatur und Skelett des Kopfes des Haies Stegostoma

tigrinum Gm. und der Holocephalen, mit einem Anhang iiber die Nasenrinne. Acta Soc. Sclent, fenn.,

Helsingfors, 37 (6): 1-60, 36 figs.
Maisey, J. G. 1982. The anatomy and interrelationships of Mesozoic hybodont sharks. Am. Mus. Novit.,

New York, 2724: 1-48, 17 figs.

- 1983. Cranial anatomy of Hybodus basanus Egerton from the Lower Cretaceous of England. Am.
Mus. Novit., New York, 2758: 1-64, 26 figs.

Matveiev, B. 1925. [The structure of the embryonal axial skull of the lower fishes.] Byull. Mosk. Obshch.

Ispyt. Prir., (Otdel. Biol.) 34: 416-475, 15 figs. [In Russian, Engl. summary].
Meinke, D. K. 1982. A light and scanning electron microscope study of microstructure, growth and

development of the dermal skeleton of Polypterus (Pisces: Actinopterygii). J. Zool., Lond., 197:

355-382, 8 pis.
Miles, R. S. 1964. A reinterpretation of the visceral skeleton of Acanthodes. Nature, Lond., 204: 457^59,

2 figs.

- 1965. Some features of the cranial morphology of acanthodians and the relationships of the
Acanthodii. Acta zool., Stockh., 46: 233-255, 2 figs.

- 1966. The acanthodian fishes of the Devonian Plattenkalk of the Paffrath Trough in the Rhineland,
with an appendix containing a classification of the Acanthodii and a revision of the genus
Homalacanthus. Ark. Zool., Stockholm, (2) 18: 147-194, pis 1-10.

- 1967. Observations on the ptyctodont fish, Rhamphodopsis Watson. J. Linn. Soc., London (Zool )
47: 99-120, pis 1-6.

- 1971a. In Moy-Thomas, J. A. & Miles, R. S., Palaeozoic fishes. 2nd edn. vii + 259 pp., 159 figs.
London.

- 1971ft. The Holonematidae (placoderm fishes), a review based on new specimens oiHolonema from
the Upper Devonian of Western Australia. Phil. Trans. R. Soc., London, (B) 263: 101-234, 126 figs.

- 1973a. Relationships of acanthodians. In Greenwood, P. H., Miles, R. S. & Patterson, C. (eds),
Interrelationships of Fishes: 63-103, 7 pis. London.

- 1973ft. Articulated acanthodian fishes from the Old Red Sandstone of England, with a review of the
structure and evolution of the acanthodian shoulder-girdle. Bull. Br. Mus. nat. Hist., London, (Geol.)
24: 111-213, pis 1-21.

- 1975. The relationships of the Dipnoi. Colloques int. Cent. natn. Rech. scient., Paris, 218: 133-148,

3 figs.

1977. Dipnoan (lungfish) skulls and the relationships of the group: a study based on new species from

the Devonian of Australia. Zool. J. Linn. Soc., London, 61: 1-328, 158 figs.

RELATIONSHIPS OF PALAEONISCIDS 413

— & Westoll, T. S. 1968. The placoderm fish Coccosteus cuspidatus Miller ex Agassiz from the Middle
Old Red Sandstone of Scotland. Part 1. Descriptive morphology. Trans. R. Soc. Edinb., 67: 373-476,
51 figs, pis 1-12.

& Young, G. C. 1977. Placoderm interrelationships reconsidered in the light of new ptyctodontids

from Gogo, Western Australia. In Andrews, S. M., Miles, R. S. & Walker, A. D. (eds), Problems in

Vertebrate Evolution: 123-198, 5 pis. London.
Millot, J. & Anthony, J. 1958-65. Anatomic de Latimeria chalumnae. 1, Squelette, muscles et formations de

soutien. 122 pp., 82 pis (1958). 2, Systeme nerveux et organes des sens. 131 pp., 76 pis (1965). Paris,

C.N.R.S.
Moy-Thomas, J. A. 1934. Notes on the development of the chondrocranium oiPolypterussenegalus. Q. Jl

microsc. Sci., London, 76: 209-230, 6 figs.
1936. On the structure and affinities of the Carboniferous cochliodont Helodus simplex. Geol. Mag.,

London, 73: 488-503, pis 1-2.
& Dyne, M. B. 1938. The actinopterygian fishes from the Lower Carboniferous of Glencartholm,

Eskdale, Dumfriesshire. Trans. R. Soc. Edinb., 59: 437-480, 40 figs.
& Miles, R. S. 1971. See Miles 1971a.

Nelson, G. J. 1968. Gill-arch structure in Acanthodes. In 0rvig, T. (ed.), Current problems in lower

vertebrate phylogeny. Nobel Symposium, Stockholm, 4: 129-143, 6 figs.
1969a. Infraorbital bones and their bearing on the phylogeny and geography of osteoglossomorph

fishes. Am. Mus. Novit., New York, 2394: 1-37, 22 figs.
19696. Gill arches and the phylogeny of fishes, with notes on the classification of vertebrates. Bull.

Am. Mus. not. Hist., New York, 141: 475-552, pis 79-92.
1970a. Subcephalic muscles and intracranial joints of sarcopterygian and other fishes. Copeia, New

York, 1970: 468-471, 1 fig.
19706. Pharyngeal denticles (placoid scales) of sharks with a note on the dermal skeleton of

vertebrates. Am. Mus. Novit., New York, 2415: 1-26, 19 figs.

1973. Relationships of clupeomorphs, with remarks on the structure of the lower jaw in fishes. In

Greenwood, P. H., Miles, R. S. & Patterson, C. (eds), Interrelationships of Fishes: 333-349, 8 figs.

London.
Nielsen, E. 1936. Some few preliminary remarks on Triassic fishes from East Greenland. Meddr Gr0nland,

Copenhagen, 112: 1-55, 19 figs.
1942^9. Studies on Triassic fishes from East Greenland. I. Glaucolepis and Boreosomus. Meddr

Gr0nland, Copenhagen, 138: 1-403, 30 pis (1942). II. Australosomus and Birgeria. Loc. cit., 146:

1-309, 20 pis (1949).
1952. A preliminary note on Bobasatrania groenlandica. Meddr dans kgeol. Foren., Copenhagen, 12:

197-204, 2 figs.
Nilsson, T. 1943-44. On the morphology of the lower jaw of Stegocephalia with special reference to

Eotriassic stegocephalians from Spitsbergen. I. Descriptive part. K. svenska VetenskAkad. Handl.,

Stockholm, (3) 20: 3-46, 9 pis (1943). II. General part. Loc. cit., (3) 21: 3-70, 34 figs (1944).
Noble, G. K. 1931. The biology of the Amphibia. 577 pp. London.
Norman, J. R. 1926. The development of the chondrocranium of the eel (Anguilla vulgaris), with

observations on the comparative morphology and development of the chondrocranium in bony fishes.

Phil. Trans. R. Soc., London, (B) 214: 369-464, 56 figs.

Nursall, J. R. 1966. Macromesodon. In Piveteau, J. (ed.), Traite de Paleontologie, 4 (3): fig. 169. Paris.
Nybelin, 0. 1967. Notes on the reduction of the sensory canal system and of the canal-bearing bones in the

snout of higher actinopterygian fishes. Ark. Zool, Stockholm, (2) 19: 235-246, 4 figs.
1968. The dentition in the mouth cavity of Elops. In 0rvig, T. (ed.), Current problems in lower

vertebrate phylogeny. Nobel Symposium, Stockholm, 4: 439-443, 3 figs.

1976. On the so-called postspiracular bones in crossopterygians, brachiopterygians and actinoptery-

gians. Acta R. Soc. sclent, litt. gothoburg., (Zool.) 10: 1-31, 1 pi.
0rvig, T. 1957. Notes on some Paleozoic lower vertebrates from Spitsbergen and North America. Norsk

geol. Tiddskr., Stavanger, 37: 285-353, 16 figs.
1960. New finds of acanthodians, arthrodires, crossopterygians, ganoids and dipnoans in the Upper

Middle Devonian calcareous flags (Oberer Plattenkalk) of the Bergisch-Paffrath Trough. 1. Palaeont.

Z., Berlin, 34: 295-335, pis 26-29.
1962. Y a-t-il une relation directe entre les arthrodires ptyctodontides et les holocephales? Colloques

int. Cent. natn. Rech. sclent., Paris, 104: 49-61, 1 pi.
1966. Histologic studies of ostracoderms, placoderms and fossil elasmobranchs. 2. On the dermal

skeleton of two late Palaeozoic elasmobranchs. Ark. Zool, Stockholm, (2) 19: 1-39, 6 figs, 5 pis.

414 B. G. GARDINER

- 1967o. Phylogeny of tooth tissues: evolution of some calcified tissues in early Vertebrates. In Miles,

A. E. W. (ed.), Structural and Chemical organization of teeth, 1: 45-110, 53 figs. London.
19676. Some new acanthodian material from the Lower Devonian of Europe. J. Linn. Soc., London,

(Zool.) 47: 131-153, pis 1-4.
1973. Acanthodian dentition and its bearing on the relationships of the group. Palaeontographica,

Stuttgart, (A) 143: 119-150, pis 16-18.
1975. Description, with special reference to the dermal skeleton, of a new radotinid arthrodire from

the Gedinnian of arctic Canada. Colloques int. Cent. natn. Rech. sclent., Paris, 218: 41-71, 8 pis.
1978. Microstructure and growth of the dermal skeleton in fossil actinopterygian fishes: Boreosomus,

Plegmolepis and Gyrolepis. Zoologica Scr., Stockholm, 7: 125-144, 29 figs.
Panchen, A. L. 1964. The cranial anatomy of two Coal Measure anthracosaurs. Phil. Trans. R. Soc.,
London, (B) 247: 593-637, 19 figs.

- 1972. The skull and skeleton of Eogyrinus attheyi Watson (Amphibia: Labyrinthodontia). Phil.
Trans. R. Soc., London, (B) 263: 279-326, 16 figs.
1977. On Anthracosaurus russelli Huxley (Amphibia: Labyrinthodontia) and the family

Anthracosauridae. Phil. Trans. R. Soc., London, (B) 279: 447-512, 17 figs.
Patterson, C. 1973. Interrelationships of holosteans. In Greenwood, P. H., Miles, R. S. & Patterson, C.

(eds), Interrelationships of Fishes: 233-305, 27 figs. London.
1975. The braincase of pholidophorid and leptolepid fishes, with a review of the actinopterygian

braincase. Phil. Trans. R. Soc., London, (B) 269: 275-579, pis 8-20.
1977a. The contribution of paleontology to teleostean phylogeny. In Hecht, M. K., Goody, P. C. &

Hecht, B. M. (eds), Major Patterns in Vertebrate evolution: 579-643, 19 figs. New York.
1977&. Cartilage bones, dermal bones and membrane bones, or the exoskeleton versus the

endoskeleton. In Andrews, S. M., Miles, R. S. & Walker, A. D. (eds), Problems in Vertebrate

Evolution: 77-121, 9 figs. London.
1982. Morphology and interrelationships of primitive actinopterygian fishes. Am. Zool., Utica, 22:

241-259, 3 figs.
& Rosen, D. E. 1977. Review of ichthyodectiform and other Mesozoic teleost fishes and the theory

and practice of classifying fossils. Bull. Am. Mus. nat. Hist., New York, 158: 81-172, 54 figs.
Pearson, D. M. 1981. Functional aspects of the integument in polypterid fishes. Zool. J. Linn. Soc.,

London, 72: 93-106, 9 figs.
& Westoll, T. S. 1979. The Devonian actinopterygian Cheirolepis Agassiz. Trans. R. Soc. Edinb., 70:

337-399, 22 figs.
Pehrson, T. 1922. Some points in the cranial development of teleostomian fishes. Acta zool., Stockh., 3:

1-63, 24 figs.

- 1940. The development of the dermal bones in the skull oiAmia calva. Acta zool., Stockh., 21: 1-50,

50 figs.
1947. Some new interpretations of the skull in Polypterus. Acta zool, Stockh., 28: 399-455, 28 figs.

Platt, J. B. 1896. Ontogenetic differentiation of the ectoderm in Necturus. Study II - On the development

of the peripheral nervous system. Q. Jl microsc. Sci., London, 38: 485-545, pis 36-38.
Playford, P. E. & Lowry, D. C. 1966. Devonian reef complexes of the Canning Basin, Western Australia.

Bull. geol. Surv. West. Aust., Perth, 118: 1-150, 48 figs.
Poplin, C. 1974. Etude de quelques paleoniscides pennsylvaniens du Kansas. 151 pp., 40 pis. Paris,

C.N.R.S.
Presley, R. & Steel, F. L. D. 1976. On the homology of the alisphenoid. J. Anat., London, 121: 441-459,

9 figs.
1978. The pterygoid and ectopterygoid in mammals. Anat. Embryol., New York, 154: 95-110,

3 figs.
Rayner, D. H. 1948. The structure of certain Jurassic holostean fishes, with special reference to their

neurocrania. Phil. Trans. R. Soc., London, (B) 233: 287-345, pis 19-22.

- 1951. On the cranial structure of an early palaeoniscid, Kentuckia gen. nov. Trans. R. Soc. Edinb.,
62: 58-83, 12 figs.

Reis, O. M. 1890. Zur Kenntnis des Skelets der Acanthodinen. Geogn. Jh., Munich, 3: 1^13, 8 figs.

- 1895. Illustrationen zur Kenntnis des skeletts von Acanthodes bronni Agassiz. Abh. senckenb.
naturforsch. Ges., Frankfurt, 19: 49-64, pis 1-6.

1896. Uber Acanthodes bronni Agassiz. Morph. Arb., Jena, 6: 143-220.

Ridewood, W. G. 1894. On the hyoid arch of Ceratodus. Proc. zool. Soc. Lond., 1894: 632-640, 3 figs.
Ritchie, A. 1968. New evidence on Jamoytius kerwoodi White, an important ostracoderm from the
Silurian of Lanarkshire, Scotland. Palaeontology, London, 2: 21-36, pis 3-6.

RELATIONSHIPS OF PALAEONISCIDS 415

Roberts, J., Jones, P. J., Jell, J. S., Jenkins, T. B. H., Marsden, M. A. H., McKellar, R. G., McKelvey,

B. C. & Seddon, G. 1972. Correlation of the Upper Devonian rocks of Australia. J. geol. Soc. Aust.,

Adelaide, 18: 467^90. 2 figs.
Robineau, D. 1975. Le systeme de la veine jugulaire et ses homologies chez Latimeria chalumnae (Pisces,

Crossopterygii, Coelacanthidae). C.r. hebd. Seanc. Acad. Sci., Paris, 281 (D): 45-48, 1 fig.
Romer, A. S. 1937. The braincase of the Carboniferous crossopterygian Megalichthys nitidus. Bull. Mas.

comp. Zool. Harv., Cambridge, Mass., 82: 1-73, 16 figs.
1947. Review of the Labyrinthodontia. Bull. Mus. comp. Zool. Harv., Cambridge, Mass., 99: 3-368,

48 figs.
1964. The braincase of the Paleozoic elasmobranch Tamiobatis. Bull. Mus. comp. Zool. Harv.,

Cambridge, Mass., 131: 89-105, 1 pi.

1966. Vertebrate paleontology (3rd edn). ix + 468 pp., 443 figs. Chicago.

& Witter, R. V. 1942. Edops, a primitive rhachitomous amphibian from the Texas red beds. J. Geol. ,

Chicago, 50: 925-960, 14 figs.
Rosen, D. E., Forey, P. L., Gardiner, B. G. & Patterson, C. 1981. Lungfishes, tetrapods, paleontology, and

plesiomorphy. .Bull. Am. Mus. nat. Hist., New York, 167: 159-276, 62 figs.
Rudd, G. 1920. Uber hautsinnesorgane bei Spinax niger Bon. Zool. Jb., Jena, (Anat.) 41: 459-546, pis

34-37.
Santos, R. da Silva 1960. A posi?ao sistematica de Enneles audax Jordan e Branner da Chapado do

Araripe, Brasil. Monografias Div. geol. miner. Bras., Rio de Janeiro, 17. xi + 25 pp., 5 pis.
Save-Soderbergh, G. 1932. Preliminary note on Devonian stegocephalians from East Greenland. Meddr

Gr0nland, Copenhagen, 94: 1-107, pis 1-22.
1935. On the dermal bones of the head in labyrinthodont stegocephalians and primitive Reptilia with

special reference to Eotriassic stegocephalians from East Greenland. Meddr Gr0nland, Copenhagen,

98: 1-211, pis 1-15.
1936. On the morphology of the Triassic stegocephalians from Spitzbergen, and the interpretation of

the endocranium in the Labyrinthodontia. K. svenska VetenskAkad. Handl, Stockholm, (3) 16: 1-181,

pis 1-22.
1937. On the dermal skulls oiLyrocephalus, Aphaneramma and Benthosaurus , labyrinthodonts from

the Triassic of Spitsbergen and N. Russia. Bull. geol. Instn Univ. Upsala, 27: 189-208, 12 figs.
1952. On the skull of Chirodipterus wildungensis, an Upper Devonian dipnoan from Wildungen. K.

svenska VetenskAkad. Handl., Stockholm, (4) 3: 1-29, 7 pis.
Schaeffer, B. 1968. The origin and basic radiation of the Osteichthyes. In 0rvig, T. (ed.), Current

problems in lower vertebrate phylogeny. Nobel Symposium, Stockholm, 4: 207-222, 4 figs.
1971 . The braincase of the holostean fish Macrepistius, with comments on neurocranial ossification in

the Actinopterygii. Am. Mus. Novit., New York, 2459: 1-34, 11 figs.
1975. Comments on the origin and basic radiation of the gnathostome fishes with particular reference

to the feeding mechanism. Colloques int. Cent. natn. Rech. scient., Paris, 218: 101-109, 2 figs.
1981. The xenacanth shark neurocranium, with comments on elasmobranch monophyly. Bull. Am.

Mus. nat. Hist., New York, 169: 1-66, 26 figs.
& Dalquest, W. W. 1978. A palaeonisciform braincase from the Permian of Texas, with comments on

the cranial fissure and the posterior myodome. Am. Mus. Novit., New York, 2658: 1-15, 9 figs.
Schmalhausen, I. I. 1968. The origin of terrestrial vertebrates, xxi + 314 pp., 165 figs. New York.
Schreiner, K. E. 1902. Einige Ergebnisse iiber den Bau und die Entwicklung der Occipital-region von

Amia and Lepidosteus. Z. wiss. Zool., Leipzig, 72: 467-524, pis 28-29.
Schultze, H.-P. 1969. Griphognathus Gross, ein langschnauziger Dipnoer aus dem Oberdevon von

Bergisch-Gladbach (Rheinisches Schiefergebirge) und von Lettland. Geologica Palaeont., Marburg, 3:

21-79, 9 pis.
1970. Die Histologie der Wirbelkorper der Dipnoer. NeuesJb. Geol. Palaont. Abh., Stuttgart, 135:

311-336, pis 38-42.
1975. Die Lungenfisch-Gattung Conchopoma (Pisces, Dipnoi). Senckenberg. leth., Frankfurt, 56:

191-231, pis 1-4.
1977. Ausgangsform und Entwicklung der rhombischen Schuppen der Osteichthyes (Pisces).

Palaeont. Z., Stuttgart, 51: 152-168, pis 13-14.
Semon, R. 1899. Die Zahnentwickelung des Ceratodus forsteri. Denkschr. med.-naturw. Ges. Jena, 4:

115-135, pis 28-30.
Sewertzoff, A. N. 1902. Zur Entwickelungsgeschichte des Ceratodus forsteri. Anat. Anz., Jena, 21:

593-608, 5 figs.
1926. Studies on the bony skull of fishes. Q. Jl microsc. Sci., London, 70: 451-540, 40 figs.

416 B. G. GARDINER

- 1928. The head skeleton and muscles oiAcipenser ruthenus. Acta zool. , Stockh. , 9: 193-319, pis 1-9.

& Disler, N. N. 1924. Das Pharyngomandibulare der Selachier. Anat. Anz., Jena, 58: 345-349, 4 figs.

Spencer, W. P. 1893. Contributions to our knowledge of Ceratodus. Part 1. The blood vessels. In Fletcher,

J. J. (ed.), Macleay Memorial Volume: 1-35, pis 1-5. Sydney (Linn. Soc. N.S.W.).
Steen, M. C. 1934. The amphibian fauna from South Joggins, Nova Scotia. Proc. zool. Soc. Land., 1934:

465-504, pis 1-5.

Stensio, E. A. 1921. Triassic fishes from Spitzbergen, 1. xxviii + 307 pp., 35 pis. Vienna.
1922. liber zwei Coelacanthiden aus dem Oberdevon von Wildungen. Palaeont. Z., Berlin, 4:

167-210, pis 3-5.
1925. Triassic fishes from Spitzbergen. Part II. K. svenska VetenskAkad. Handl., Stockholm, (3) 2:

1-261, 34 pis.
1927. The Downtonian and Devonian vertebrates of Spitsbergen, 1. Family Cephalaspidae. Skr.

Svalbard Nordishavet, Oslo, 12: 1-391, 112 pis.
1931. Upper Devonian vertebrates from East Greenland collected by the Danish Greenland

Expeditions in 1929 and 1930. Meddr Gr0nland, Copenhagen, 86: 1-212, 36 pis.

1932a. The cephalaspids of Great Britain, xiv + 220 pp. , 66 pis. London, British Museum (Nat. Hist.).

1932ft. Triassic fishes from East Greenland collected by the Danish expeditions in 1929-1931. Meddr

Gr0nland, Copenhagen, 83 (3): 1-305, 39 pis.
1935. Sinamia zdanskyi, a new amiid from the Lower Cretaceous of Shantung, China. Palaeont. sin.,

Peking, 3 (c): 1-48, 18 pis.
1937. On the Devonian coelacanthids of Germany with special reference to the dermal skeleton. K.'

svenska VetensAkad. Handl. , Stockholm, (3) 16: 1-56, 23 figs.
1944. Contributions to the knowledge of the vertebrate fauna of the Silurian and Devonian of

Western Podolia, II. Notes on two arthrodires from the Downtonian of Podolia. Ark. Zool.,

Stockholm, 35: 1-83, 14 pis.
1947. The sensory lines and dermal bones of the cheek in fishes and amphibians. K. svenska

VetenskAkad. Handl, Stockholm, (3) 24: 1-195, 38 figs.
1948. On the Placodermi of the Upper Devonian of East Greenland, II. Antiarchi subfamily

Bothriolepinae. With an attempt at a revision of the previously described species of that family.

Palaeozool. Greenland., Copenhangen, 2: 1-622, 75 pis.
1950. La cavite labyrinthique, 1'ossification sclerotique et 1'orbite de Jagorina. Colloques int. Cent.

natn. Rech. sclent., Paris, 21: 9-41, 8 figs.
1959. On the pectoral fin and shoulder girdle of the arthrodires. K. svenska VentenskAkad. Handl.,

Stockholm, (4) 8: 5-229, pis 1-25.
1963a. Anatomical studies on the arthrodiran head. 1. Preface, geological and geographical

distribution, the organisation of the arthrodires, the anatomy of the head in the Dolichothoraci,

Coccosteomorphi and Pachyosteomorphi. Taxonomic appendix. K. svenska VetenskAkad. Handl.,

Stockholm, (4) 9: 1-419, pis 1-62.
1963ft. The brain and the cranial nerves in fossil, lower craniate vertebrates. Skr. norske

Vidensk-Akad. mat.-nat. KL, Oslo, (N.S.) 13: 3-120, 54 figs.
1969. Elasmobranchiomorphi; Placodermata; Arthrodires. In Piveteau, J. (ed.), Traite de

Paleontologie, 4 (2): 71-692. Paris.
Stohr, P. 1879. Zur Entwicklungsgeschichte der Urodelenschadels. Z. wiss. Zool., Leipzig, 33: 477-526,

pis 29-30.
Stone, L. S. 1922. Experiments on the development of the cranial ganglia and the lateral line sense organs

in Amblystoma punctatum. J. exp. Zool., Philadelphia, 35: 421-496, 87 figs.
Swinnerton, H. H. 1902. A contribution to the morphology of the teleostomian head skeleton, based upon

a study of the developing skull of the Three-spined Stickleback (Gasterosteus aculeatus). Q. Jl microsc.

Sci., London, 45: 503-593, pis 28-31.
Taverne, L. 1973. Sur la conservation d'une structure paleonisciforme, le canal aortique basioccipital, chez

le notoptere africain Xenomystus Giinther (Pisces Osteoglossiformes). Revue Zool. Bot. afr., Brussels,

87: 501-504, 2 figs.
Thomson, K. S. 1964. The comparative anatomy of the snout in rhipidistian fishes. Bull. Mus. comp. Zool.

Harv., Cambridge, Mass., 131: 313-357, 10 figs.
1967. Mechanics of intercranial kinetics in fossil rhipidistian fishes (Crossopterygii) and their

relatives. /. Linn. Soc., London, (Zool.) 46: 223-253, 17 figs.
& Campbell, K. S. W. 1971. The structure and relationships of the primitive Devonian lungfish

Dipnorhynchus sussmilchi (Etheridge). Bull. Peabody Mus. nat. Hist., New Haven, Conn., 38: 1-109,

95 figs.

RELATIONSHIPS OF PALAEONISCIDS 417

Todd, E. A. (1973). A revision of certain clupeoids with special reference to the double-armoured herrings.

Ph.D. thesis, University of London (unpubl.).
Veit, O. 1911. Beitrage zur Kenntnis des Kopfes der Wirbeltiere, 1. Die Entwicklung des Prim-

ordialcranium von Lepidosteus osseus. Arb. anat. Inst., Weisbaden, (anat. Hefte) 44 (1): 93-225, pis

A-E (col., some with overlays).
Vorobjeva, E. 1973. Einige Besonderheiten im Schadelbau von Panderichthys rhombolepis (Gross) (Pisces

Crossopterygii). Palaeontographica, Stuttgart, (A) 143: 221-229, 4 figs, 36 pis.
1917 a. Filogeneticheski svazi osteolepiformnykh Crossopterygii i ikh polozhenie v sisteme. In

Menner, V. V. (ed.), Ocherki po filogenii i sistematike iskopaemykh ryb i beschelyusttnykh: 71-88, 8

figs. Moscow, Paleont. Inst. Akad. Nauk SSSR.

19776. [Morphology and features of evolution of crossopterygian fishes.] Trudy Paleont. Inst.,

Moscow, 163: 1-239. [In Russian].
Wangsjo, G. 1952. The Downtonian and Devonian vertebrates of Spitsbergen, IX. Morphologic and

systemic studies of the Spitsbergen cephalaspids. Skr. norsk Polarinst., Oslo, 97: 1-612

+ supplementary note, 118 pis.

Watson, D. M. S. 1921. On the coelacanth fish. Ann. Mag. not. Hist., London, (9) 8: 320-337, 5 figs.
1925. The structure of certain palaeoniscids and the relationship of that group with other bony fish.

Proc. zool. Soc. Lond., 1925: 815-870, 2 pis.
1926. The evolution and origin of the Amphibia. Phil. Trans. R. Soc., London, (B) 214: 189-257,

39 figs.
1928. On some points in the structure of palaeoniscid and allied fish. Proc. zool. Soc. Lond., 1928:

49^70, 15 figs.
1937. The acanthodian fishes. Phil. Trans. R. Soc., London, (B) 228: 49-146, pis 5-14.

- 1940. The origin of frogs. Trans. R. Soc. Edinb., 60: 195-231, 23 figs.
& Day, H. 1916. Notes on some Palaeozoic fishes. Mem. Proc. Manchrlit. phil. Soc., 60: 1-52, 3 pis.

Westoll, T. S. 1936. On the structure of the dermal ethmoid shield of Osteolepis. Geol. Mag., London, 73:

157-171, 2 pis.

1938. Ancestry of the tetrapods. Nature, Lond., 141: 127-128, 2 figs.

White, E. 1. 1965. The head of Dipterus valenciennesi Sedgwick and Murchison. Bull. Br. Mus. not. Hist.,

London, (Geol.) 11: 1-45, 3 pis.
1978. The larger arthrodiran fishes from the area of the Burrinjuck Dam, N.S.W. Trans, zool. Soc.

Lond., 34: 149-262, 12 pis.
& Toombs, H. A. 1972. The buchanosteid arthrodires of Australia. Bull. Br. Mus. nat. Hist., London,

(Geol.) 22: 377-419, 25 figs, 9 pis.
Wijhe, J. W. van 1882. Visceralskelett und die Nerven des Kopfes der Ganoiden und von Ceratodus.

Niederl. Arch. Zool., Haarlem, 5: 207-320, pis 15-16.
Wiley, E. O. 1976. The phylogeny and biogeography of Fossil and Recent gars (Actinopterygii:

Lepisosteidae). Univ. Kans. Publs Mus. nat. Hist., Lawrence, 64: 1-111, 72 figs.
1979. Ventral gill arch muscles and the interrelationships of gnathostomes, with a new classification

of the Vertebrata. Zool. J. Linn. Soc., London, 67: 149-179, 8 figs.
Wintrebert, P. 1922. Embryogenie. Le mode d'edification du vomer definitif au cours de la metamorphose

chez les Salamandridae. C.r. hebd. Seanc. Acad. Sci., Paris, 175: 239-241.
Woodward, A. S. 1916-19. The fossil of the English Wealden and Purbeck formations, vi + 148 pp. , 26 pis.

Palaeontogr. Soc. (Monogr.), London. (1916, pp. 1-48, pis 1-10; 1918, pp. 49-104, pis 11-20; 1919,

pp. 105-148, pis 21-26, title page and index).
Wright, R. R. 1885. On the hyomandibular clefts and pseudobranchs of Lepidosteus and Amia. J. Anat.

Physiol, Lond., 19: 476-499, pi. 24.
Young, G. C. 1978. A new early Devonian petalichthyid fish from Taemas/Wee Jasper region of New

South Wales. Alcheringa, Adelaide, 2: 103-116, 8 figs.
1979. New information on the structure and relationships of Buchanosteus (Placodermi:

Euarthrodira) from the Early Devonian of New South Wales. Zool. J. Linn. Soc., London, 66:

307-352, pis 1-5.

1980. A new early Devonian placoderm from New South Wales, Australia, with a discussion of

placoderm phylogeny. Palaeontographica, Stuttgart, (A) 167: 10-76, pis 1-2.
Zangerl, R. 1982. Chondrichthyes I (Paleozoic Elasmobranchii). In Schultze, H.-P. (ed.), Handbook of

Paleoichthyology, 3A. 115 pp., 115 figs. Stuttgart.
& Case, G. R. 1976. Cobelodus aculeatus (Cope), an anacanthus shark from Pennsylvanian Black

Shales of North America. Palaeontographica, Stuttgart, (A) 154: 107-157, 39 figs.

418

B. G. GARDINER

Index

Numbers in italics are figure numbers.

abducens nerve 188, 192, 194-5,

197, 207, 224,231, 248-9
canal 224, 249

Acanthodes 183-4, 189, 201, 203-4,
206, 208, 210-11, 213-15,
234, 240-1, 249, 253, 274,
282, 298-9, 301, 306, 332,
336, 342, 352, 356-60, 362-

3, 368, 380-1, 392, 402-3
bronni 120, 203

acanthodians 183-5, 190, 203, 233-
5, 243, 253, 276, 295, 298-9,
306, 312-13, 325, 332-3,
336-8, 341-3, 352, 356-60,
362-3, 368, 374-5, 378, 380-
1, 392, 396, 400, 402-3

Acanthodidae 336, 374

Acanthodii 399

Acanthostega 312

Acanthothoraci 336-7

accessory hyomandibula 352, 356
opercular 344
toothplates 280

vomerine 53, 271, 273, 280,
310

Acentrophorus 267

Adpenser 183, 190, 198, 201, 203,
206-8, 211-12, 214, 227,
233-8, 240, 242-4, 246-9,
251, 267, 271, 276-7, 295,
297, 299, 301, 311, 324, 348-
9, 352-4, 359-60, 362-4,
368, 370-1, 374, 381-4
fluvescens 352

Acipenseridae 395

acrochordal 247

acrodin cap 260, 264, 331, 395, 397

Acrolepis 398

Acropholis 305-6

Acrorhabdus 352, 370, 374

Actinistia 177, 183, 400

actinistians 203-7, 211-12, 214-15,
232-3, 238-41, 243, 246,
248, 253-4, 266, 270-1, 273-

4, 277-9, 295, 298-9, 301,
306-7, 311-14, 320, 324,
332-3, 339, 359-63, 365,
370, 374, 379, 392, 403-4,
406

actinopteran 177, 181, 252, 277-8,
280, 339, 352-4, 360, 363,
380-1, 383, 388, 394, 397

Actinopteri 395-7, 400

actiriopterygians 177, 183-5, 189,
191-2, 194, 196, 198-211,
214-16, 218, 228, 231, 233-
49, 251-2, 254, 257, 266-7,
270-1, 273-80, 293, 297,
299, 301, 305-7, 309-14,
316, 320-1, 324, 332-3, 339,
341-2, 344, 349, 352-3, 360,

362-6, 368, 372, 374, 378-9,
381-4, 386, 388, 392, 394,
396-7, 400, 402-3
Actinopterygii 177, 183, 397, 400
actinotrichia 386-7
adductor hyomandibulae 245
mandibulae 243-6, 283, 396, 403
fossa 282-3, 286-7, 292, 307,

327, 331-3

opercularis 212-13, 244-5
adotic process 232
adsymphysial plate 333
Aetheretmon 394, 398
afferent hyoidean artery 346
Agnatha 185, 228, 254
Akinetia 400
albulids 278, 339
Alepocephalidae 233, 363
alisphenoid 299

(pterosphenoid) pedicel 230
Alosa 243
Ambylpterus 251
'Ambipoda' 204-5, 231, 275
Ambystoma 207, 210, 247
Amia 183, 192, 195, 198, 204, 206-
8, 210, 212, 214, 227, 230-1,
233-8, 242-9, 265, 267, 270,
275-7, 295, 297, 299, 305-7,
309-10, 313-14, 322, 324,
332-3, 339, 343-4, 346, 352,
357, 360, 362-3, 365-6, 368,
371, 378-9, 384, 386, 394
amiids 183, 244, 267, 270, 277
amioids 203, 206, 233, 240, 277
amniotes 243, 322, 366, 380
amphibians 207, 210, 243, 247, 297,
299, 301, 304, 311-14, 322,
332-3

ampulla 212-13, 228
ampullary chamber 211-12
anterior 227-8, 238
external 228

posterior 213, 216-17, 227-8
recess 228
anal fin 364, 382, 384, 386, 392,

394, 396, 398-9
anamestic bones 270
anaspids 254
anastomosis 270
Anguilla 342-3, 379
anguillids 386
angular 327, 331-3, 336
anocleithrum 374, 378, 404
anterior articulation 295, 297-8,
ascending process 271, 273, 277
basicapsular commissure 204
cerebral vein 226, 249-51

foramina 226

dorsolateral plate 376, 378
intraotic joint 205
lateral plate 376, 378

median ventral plate 376
dorsal fontanelle 216, 226
myodome 254, 260, 264-6
bone 267
dorsal 254
ventral 254, 264-6
nasal opening 259-60, 263-4
otic process 301
pit-line 317
process 181, 299, 378
semicircular canal 211, 228, 238,

241, 243

ventrolateral plate 376, 381
anterodorsal scale process 395,

397-8

anthracosaurs 243, 322, 333, 336
Antiarcha 336-7
antopercular 343
antorbital 270, 396, 398

process 306

antotic process 211, 215, 301, 306
anurans 208, 299, 362, 380
aorta 192, 406
aortic canal 189, 191-2, 196, 201,

206-8, 273, 280
ligament 192, 201, 206
Aphanerama 311-12
aphetohyoid hypothesis 357
apodans 306, 360
Arapaima 278
archaeomaenids 365
Archegosaurus 365
Arctolepis 253, 324
arcualia 363-5
Arthrodira 336, 403
arthrodires 299, 306, 324, 332, 337,

341, 375-6, 381
articular 325-6, 332-3, 336, 396,

399

fossa 299
articulatory facet for palate 215,

260, 263-4
ascending process of palate 239,

299, 301, 404

of parasphenoid 181, 211, 231,
234, 247, 271, 273, 275-9,
395-6

aspidorhynchids 366
'Aspidorhynchus' 183, 198, 214,
232-4, 237, 244-5, 247, 249,
252, 265, 277, 279, 295
Atelaspis (Aceraspis) robusta 254
Atractosteus 333
auditory bulla 241
capsule 204, 235, 244, 248, 306,

363

diverticulum 244
nerve 195
region 241
auricle 406
Australosomus 185, 189, 192, 195,

RELATIONSHIPS OF PALAEONISCIDS

419

198, 203-5, 211, 231, 233-4,
236-8, 244-5, 251, 254, 264,
266, 271, 273, 275, 277-8,
280, 282-3, 295, 299, 307,
310, 332, 346, 362-5, 368,
384, 386, 398, 400

autopalatine 282-3, 297-9, 305-8,
402, 406

autostyly 246

axial muscles 243
skeleton 124, 363-8

axonost 384, 386

ball-and-socket joint 406

basal connection 235
fulcra 181-2, 392, 394, 397
process 299, 301

basibranchial 109-10, 120, 346,

348-9, 359-62, 402-4
tooth plate 279

basicapsular fenestra 203-4

basidorsals 363

basi-exoccipital 192, 194, 203, 208

basihyal 360, 361

basioccipital 184, 186, 188-9, 192,
194-6, 203-10, 214, 218,
234, 249, 275, 279-80

basiotic lamina 235

basipterygoid process 182, 213, 216,
222, 249, 271, 273, 276-7,
282-3, 286-7, 292, 299, 404
dermal 276-7, 396, 398
endoskeletal 276-7

basisphenoid 22-24, 32, 188, 194,
204-5, 211, 213-16, 218,
222, 239, 241, 248-9, 271,
273, 275
pedicel 214

pillar 214, 218, 222-3, 230-1,
248-9

basitrabecular 235
process 301, 404

basiventrals 363-4

Batrachosuchus 324

Belonostomus 366

Benthosuchus sushkini 76D

Bergisch Gladbach 182

bile salts 406

birds 297, 360

Birgeria 182-4, 188, 198, 204, 206,
211-12, 231, 233, 237, 243-
4, 251, 267, 275-8, 307, 310,
327, 360, 364, 368, 371, 384,
386, 398, 400

Bobasatrania 251, 275, 278, 295,
297, 312, 362, 386, 399

bones X, Y2, Yj 311-12, 322

bony fishes 185
vertebrates 185

Boreaspis 185

Boreolepis 306

Boreosomus 185, 189, 192, 198,
203-4, 211, 213, 228, 236-8,
241-4, 254, 264, 275-7, 282,
295, 299, 306-7, 312, 332,
339, 341, 343, 352, 363, 374,

382, 384, 386, 398, 400
piveteaui 188

Bothriolepis 253, 338, 341-2
Br achy acanthus 342, 374-5
Brachydegma 343
Brachyosteus 341
brachythoracid 336-7
branchial arches 108, 111-20, 234,

301, 344, 346, 349, 359-63
levator muscles 231-3, 403
nerve 346
branchiostegal rays 97, 333, 338-9,

341-3, 348, 402; pi. 5c, d
Brindabellaspis 207-8, 210, 236,
240-1, 247, 266, 342, 381,
403

stensioi 185
Broughia 206
bucco-hypophysial canal 216, 218,

271, 273, 278-80, 402
duct 279
Buchanosteus 210-11, 240, 248,

279, 298-9, 403
confertituberculatus 185

Cacops 380

calcified cartilage 185, 239, 332, 402

Callopterus 366

Callorhynchus 324, 357, 403

canals for Sharpey fibres 387

of Williamson 387
Canning Basin 176
Canobius 371

capsular ethmoid bones 267
Carboniferous 278, 362, 365, 394
Carboveles 394
carcharhinoid 207
Carcharinus 241, 297-8
cartilage bones of palate 305-6
catervariolids 365
Catervariolus 278

caturids 183, 205, 216, 232-4, 236-
8, 242-4, 247, 249, 265, 267,
270, 275-8, 295, 324, 366,
378

Caturus 198, 228, 233-4, 237-8,
242, 244, 247, 249, 265, 267,
279, 363-5, 392, 394

chirotes 277, 280

furcatus 211, 280
caudal fin 385-6, 392, 394, 396,

398-9

cavum sinus imparis 197, 201
centra 208, 365-6
central canal 324-5
Centrolepis 398
Centrophorus 235
cephalaspids 185, 254, 403
ceratobranchial 775, 720, 332, 346,
357, 359, 402

fifth 349

first 346

fourth 349

ligament 232

second 348

third 349

Ceratodus 271, 297

ceratohyal 706, 720, 339, 342, 344,

346, 352, 357-60
ligament 327

ceratomandibular 301

cerebellum 227

Cetorhinus 356, 368

characinids 386

Chelracanthus 253

Cheirodus 353

Cheirolepis 49, 182, 251, 267, 270-
1, 273-9, 287, 292, 308, 311-
12, 317, 324, 332-3, 339,
344, 352, 354, 357, 374, 378,
381, 386-8, 392, 394, 397-8,
400
trailli 77D

chelonians 297, 306, 308, 360

Chelydosaurus 365

Chimaera 324, 368, 381

chimaeroids 235, 342

Chirodipterus 182, 184-5, 201-2,
205-7, 210, 239, 248, 251,
271, 279, 297, 306, 311, 313,
324, 332-3, 341, 355, 360,
366

Chlamydoselachus 240-1, 267, 298-
9, 301, 305, 313-14, 324,
333, 336, 403

choana 270, 406

Choanata 177, 301, 400

choanates 247, 253, 297, 363, 406

chondrichthyans 183-5, 211-12,
233-4, 236, 241, 243, 247-8,
266, 270, 276, 280, 298, 333,
342, 357, 359-60, 362-3,
368, 380-1, 383, 392, 400,
402-3

chondrification 235

chondrosteans 184, 190, 201, 203,
205-6, 211, 228, 237, 242,
244-6, 248, 266, 270, 275-8,
295, 297, 357-9, 370, 383,
386, 392, 394, 396

Chondrostei 395-7, 400

Chondrosteus 206, 251, 271, 275-9,

297, 395, 398
chordacentra 365-6
circulus cephalicus 206
circumorbital scutes 254
Cladistia 395-6, 400

cladistian 205, 339, 366, 368, 370,

394-5
Cladodus 207, 210, 235, 239-40,

298, 301

Cladoselache 207, 210, 298
classification 399

clavicle 726, 730-7, 134, 368-70,

375-6, 379, 381, 402, 404;

pi. 5e, f
cleithrum 726, 728-9, 737-2, 368-

71, 374, 376, 378-9, 381,

386, 402, 404
Climatiidae 374-5
Climatius 298, 341, 374-5
cloacal scales 387

420

B. G. GARDINER

Clupea 307

clupeid 243, 386

clupeoids 206

Cobelodus 210, 241

Coccoderma 332

Coccolepis 275, 280

coccosteids 324, 341

Coccosteus 253, 324, 352, 355

Coelacanthus 332

Commentrya 251, 343

common carotid artery 276

Conchopoma 210, 279, 297, 361-2

condyles of quadrate 282-3

constrictor hyoideus dorsalis 212-13

posterior part 212
coracoid 726, 378-81

anterior process 378-9, 381

foramen 371, 378-81

fossa 381

portion 371, 374
Cornuboniscus 251, 374, 392
coronoid 325, 327, 331-3, 336

process 396, 399

teeth pi. 4a, b

cosmine pore-canal system 404
cosmoid scale 392
Cosmoptychius 183-4, 189, 192,
211-12, 234, 277, 280, 339,
344, 394, 398

cranial cavity 26, 201, 208, 213,
215, 226-8, 231, 236, 242,
246, 249, 251

centrum 206

fissure 203, 212, 231

rib 191

cranio-spinal process 190, 201
cranium 234, 246, 251
Crassigyrinus 322
Cretaceous 366
crista occipitalis 198

sellaris 247
crossopterygians 321
Cryphiolepis 394
Cryptobranchus 210, 276, 306, 308,

332

Ctenodus 322

Ctenurella 183, 210, 298, 301, 306,
324, 332, 403

gar diner i 185
Cydopterus 305
cynodonts 380
cyprinoids 206, 386
Cyprinus 378

Dapedium 192, 198-9, 205-6, 234,
236-8, 242, 244-5, 247, 249,
252, 265, 267, 276-7, 279,
295, 392, 394

Dasybatus 241

Dendrerpeton 312

dental arcades 403

dentary 260, 263, 292, 326-7, 331-
3, 336, 395-6, 403, 406;
pi. 5a

dentinal tubules 387-8

dentine 336-7, 387-8, 392

dermal basipterygoid process 273
bones 185, 243
cheek 77-8, 280, 287-92, 310-

lower jaw 332-8
palatoquadrate 306-10
shoulder girdle 726, 374
skull roof 205, 217, 220, 244,

316-24, 406

snout 41, 48-9, 102, 254-71
dermethmoid 270
dermohyal 341-2, 344, 346, 349,

352, 356, 395, 397, 403
dermometapterygoid 283, 286-7,

306-7, 309-10
dermopalatine 283, 286-7, 292,

295, 297-8, 306-10, 331
dermopterotic 322, 396-7
dermosphenotic 68-9, 181, 260,

263, 291, 293, 310-12, 317
descending laminae 324
Devonian 176, 181-2, 185, 201,
207-8, 210, 232, 239-41,
243^, 247-8, 253, 270, 277-
9, 299, 333, 352, 360, 362,
366

stegotrachelid 181
diapophysial outgrowths 368
diazonal nerve 378, 380-1
Dicksonosteus 298, 306, 324, 337,

352

dicynodonts 380
diencephalon 226
dilatator fossa 246, 399

operculi 245-6, 352, 395
Diplacanthidae 336, 374
Diplacanthus 325, 381
Diplocercides 204, 239, 266, 270-1,

277, 299, 324, 332, 339
Diplomystus 211
Diplurus 324

dipnoans 183, 185, 190-1, 201-8,
210, 228, 232, 234, 238-41,
243-5, 247-8, 251, 253, 257,
266, 270-1, 273-6, 278-9,
293, 295, 297, 299, 301, 304,
306, 308, 311-14, 316, 320-
1, 324, 331-3, 336-7, 339,
341, 352, 355-6, 360-3, 366,
368, 374, 379-80, 392, 400,
403-4, 406
Dipnoi 400

Dipnorhynchus 183, 251, 271, 273,
279, 295, 306, 313, 320, 324,
333
Dipterus 210, 253, 279, 313-14,

321, 324, 332
valenciennesi 279
Dirrhizodon 207
Discobolus 241

dorsal anterior myodome 226, 244
aorta 192, 206-7
arterial system 276
diverticulum 244

fin 124, 363, 368, 384-6, 392,
394, 396, 398-9

hyoid constrictor 217, 228, 244-5

ligament 368

mandibular constrictor 245-6
dorsomedial fin muscles 378
dorsum sellae 214, 222, 247-9
Dorypterus 251
dura mater 251

Eastmanosteus 336

Echidna 360

ectopterygoid 283, 286-7, 292, 306-

10, 313

Ectosteorhachis 206-8, 210-12,
233-4, 238-40, 243, 245-6,
248, 251, 266, 273, 278, 280,
295, 354, 366
edestid 392
Edops 312
efferent arteries 234

second 192
pseudobranchial artery 271, 273-

6,278
Elonichthys 280, 306-7, 317, 324,

353, 392, 394, 398
aitkeni 306, 310
binneyi 306, 310
caudalis 310
pectinatus 306
robisoni 394
semistriatus 310
elopocephalus 366
Elopidae 233, 239, 363
Elops 232-3, 270, 278-80, 307, 309,
324, 342, 346, 356, 363, 371,
374, 381
Elpistostega 321
enamel 404

(acrodin) cap 260, 264
endochondral ossification 182-5,

195, 214, 218, 236, 403
endolymphatic diverticulum 247
ducts 202, 216, 228, 246-7
fossa 202
organ 246-7
sac 247
endoskeletal shoulder girdle 128-9,

132, 371, 374, 378
Enneles 183, 244
entopterygoid 246, 271, 273, 283,

286-7, 292, 299, 307-10
Eogyrinus 312, 380

attheyi 76C
epaxial fin-rays 386

lobe 386
epibranchial 776-77, 234, 346, 349,

352, 356-7, 359, 399, 402
first 776, 233-4, 346, 362-3
second 776, 348, 362
third 777, 349
epiglottis 406
epihyal 349, 352
epimandibular 301
epineural 191, 366, 368

processes 191

epioccipitals 184, 198, 203, 212
'epiotic' 212

RELATIONSHIPS OF PALAEONISCIDS

421

epiphysial crest 249
plexus 251

epipterygoid 299, 306

epurals 368, 386

Erriwacanthus 375

Errolichthys 275-7, 362

Eryops 380

esocids 386

ethmoid region 33-41, 254-71, 295,
297

ethmoidal bone 185
commissure 259-60, 263, 267,
270, 395

ethmopalatine ligament 292

Etmopterus 235, 240, 301, 304-5

Euporosteus 266-7

Eurycormus 392

Eurynotus 398

euselachian 392

Eusthenodon 380, 312

Eusthenopteron 88G, 182-3, 201,
204-8, 210-12, 215, 232-5,
238-40, 242-6, 248, 251,
253, 258, 266-7, 270-1, 273-
4, 276-80, 282-3, 295, 298-
9, 301, 306-7, 311-12, 314,
321-2, 324, 332-3, 339, 343-
4, 346, 349, 352, 354, 356,
360-3, 365, 368, 374, 400,
403, 406
foordi 75C, 78A, 88F

Euthacanthus 325, 341, 375

excurrent opening 267

exoccipitals 184, 189, 197, 199, 203,
208-10

external rectus muscle 223-4, 231,
248-9

external semicircular canals 182,
203, 210, 212, 228, 243-4

extracleithrum 374

extracranial 218, 237-8

extralateral 341

extramural chamber 218, 236

extrascapular 84, 85-7, 125, 318,
324, 368

eye-stalk 402-3

facial nerve 207, 227, 230, 239, 251
buccal branch 257, 260, 263, 267
canal 218, 228, 230, 236-40, 248,

251

external mandibular branch 327
foramen 211, 214, 218, 224, 233
ganglion 236-8
hyoid branch 346, 349, 352
hyomandibular trunk 218, 227-8,

236-40, 344, 349, 353-5
internal mandibular branch 283,

292-3, 327, 332
lateralis branch 227, 237-40
mandibular branch 346, 349,

353-5

otic branch 214, 216, 237, 244
palatine trunk 227-8, 236
recurrent lateralis branch 216,

218, 230

fenestra endonarina communis 254,

257

exochoanalis 308
exonarina anterior 257
ovalis 204
first infraorbital bone (lachrymal)

260,264

fissura oticalis ventralis 185
Fitzroy Crossing 181-2
Fitzroy trough 176
FleurantiaTll, 324
foramen magnum 189-91, 196-9,

201, 208-10

foramen olfactorium evehens 266
fossa autopalatina 404
bridgei 216, 238, 241-3, 317, 395,

397, 399
spiracularis 243

fragmentation hypothesis 183-4
Frasnian 176
fringing fulcra 181-2, 371, 374, 382,

384-5, 394, 398

frontals 80-3, 225, 259-60, 263,
267, 278, 291, 311, 316-8,
320-1, 404
Furo 267

philpotae 365
fusion hypothesis 184, 312

Gadus 386

galeomorphs 235

Galeus 212

Galkinia 365

ganoine 228, 369-70, 387-8, 392,

395-6, 398

ridges 259-60, 263-4, 374
scales 395

tubercles 260, 263-4
gasserian ganglion 230, 236-40, 249
Gasteros teus 235, 278, 305, 361
Gemuendina 253, 341
gemuendinids 403
geniculate ganglion 218, 225, 230,

236-40

geniohyoideus muscle 327
gill-arch muscles 406
gill-rakers 356
Ginglymodi 396, 399-400
glenoid fossa 325-6, 331, 371, 380-1
glossopharyngeal nerve 227
foramen 203, 212-3, 218, 227,

232-3, 316
supratemporal branch 212, 217,

242, 245
glottis 406
Glyptolepis 182, 204, 207, 211, 238-

9, 243, 251, 266, 273, 275-

80, 282-3, 295, 298-9, 301,

307, 311, 324, 332, 339, 357,

360, 365, 368, 404
groenlandica 75D
Glyptopoma 307
Gnathorhiza 297
gnathostomes 184-5, 190, 202, 204-

5, 207, 210, 234-6, 239-41,

244, 246-9, 253-4, 266, 276,

280, 298, 301, 313, 324-5,
333, 338-9, 349, 356-7, 359,
362, 364, 368, 380-1, 392,
400, 403, 406

Gogo 176, 181-2
Formation 176, 181
palaeoniscid 'A' 181
palaeoniscid 'B' 181-2
palaeoniscids 189, 204, 208, 211,
233, 237-8, 240-5, 248-9,
271, 350, 362; see Mimia,
Moythomasia

Gonatodus 251, 307, 310, 344, 398

Goodradigbeeon 253

Griphognathus 182, 184-5, 191,
201-2, 205-6, 210, 228, 233,
239, 248, 251, 253, 258, 266-
7, 271, 276, 279, 297, 306,
308, 311-14, 316, 324, 332,
341, 346, 352, 355, 360-3,
366, 368
whitei 76A, 78D, 89C

gular plates 97, 100, 339, 341-2,
403

gymnotids 386

Gyracanthidae 374

Gyracanthides 375

Gyracanthus 375

Gyrolepis 398

Gyroptychius 312

haemal arches 122, 364-5

canal 364

spines 364, 386
Halecomorphi 400
halecomorphs 184, 211, 214, 228,
236, 242, 245-6, 297, 333,

359, 386

halecostomes 190, 195, 211, 236,

343, 352, 366, 368, 396-7,

399

Halecostomi 396, 399-400
haplolepids 386, 394, 399
Haplolepis 339, 365-6, 368
Helodus 206
hemicentra 365
hemichordacentra 365
Hemicydaspis 254
hemopoietic organ 242, 395
Hepsetus 235
Heptanchus 212, 241, 360
Heptranchias 299, 301, 304
Heterodontus 234, 240-1, 248, 298,

301, 360
Heterolepidotus 198, 232-4, 237-8,

242, 244-5, 247, 249, 265,

267, 277, 279, 280
Heterotis 235, 342
hexanchoids 235
Hexanchus 381
Hiodon 295
hiodontids 278
holocephalan 206-7, 234, 244, 246,

299, 324, 337-8, 352, 357,

360, 392, 402-4
Holocephali 399

422

B. G. GARDINER

Holodipterus 184-5, 210, 239, 270-

1, 279, 297, 306, 332-3
Holonema 253, 298, 341-2
Holopetalichthys 324
Holoptychius 881, 211, 215, 249,
266, 270-1, 273, 278, 295,
311, 314, 324, 333, 339
holosteans 21 1,242, 244
Homalacanthus 253, 325, 342
horizontal pit-line 287, 293, 313,

322, 395, 397
semicircular canal 241
hyal origin 235

ray 352

hybodont sharks 403
Hybodus 207, 210, 235, 239-41, 301
Hydrolagus 248
hyoid arch 235, 344-6, 349-60
bar 235, 395, 397, 402
gill cover 341 , 356
operculum 342
process 339
ray cartilages 342, 352
hyoideo-mandibularis nerve 313
hyomandibula 104-5, 217, 235, 239,
241-2, 244-6, 283, 317, 338,
341-2, 344, 346, 349, 352-4,
356-7, 359, 396-8, 400, 402,
406

hyomandibular adductor 231, 244-5
cartilage 359
facet 211-12, 216-17, 240-1, 244-

6,404

gill-rakers 356-7
protractor 395
trunk 218, 227-8, 236-40
hypaxial lobe 386
hypobranchial 111-14, 332, 346,

349, 359-60, 362, 402, 404
first 346, 349
fourth 348
second 348-9
third 349
hypohyal 107, 344, 346, 349, 357,

362, 403

hypophyseal foramen 182
recess 218, 222
stalk 278

Hypsocormus 183, 267, 368, 378
hypurals 364, 386

ichthyodectids 277
Ichthyokentema 236, 277-8, 280,

333, 365

Ichthyophis 306, 332
Ichthyosaurus 308
Ichthyostega 76B, 78C, 89 A, 271,

279, 311-12, 322
ichthyostegids 206
incurrent opening 267
inferior oblique 254, 265-6

rectus 249

vena cava 403
inferognathal 336-7
infraorbital bones 291, 310, 313

sensory canal 256-8, 260, 264,

266, 270, 287, 291-3, 310,
312-13, 322, 324-5, 399
infrapharyngobranchial 118, 235,

346, 362-3

articulation 234-5, 279
first 218, 234-5, 346
second 233-4, 348
infraprelateral 338
infravagal process 190
interarcual muscles 403
intercalar 184, 190-1, 203, 212,

231-3, 396, 399
strut 233
interclavicle 130, 134-5, 362, 368,

370, 374-6
interdorsals 264, 365
interhyal 283, 344, 346, 352, 357-9,

403, 406

intermandibular muscle 327
intermediating body 295, 297
intermuscular septum 191, 198, 210
first 191, 196-8, 201, 207
second 196-8, 201,207-8
internal carotid artery 222, 230-1,

275-6, 283
foramen 280
rectus 214, 249
internasals 271
interolateral plate 376
interopercular 343, 396, 399
interorbital fenestra 265
region 249

septum 214, 249, 254
interrelationships of actinoptery-

gians 394

intersegmental artery 363-4
intertemporal 79, 82-3, 181, 216,

291, 310-12, 317-18, 322
interventrals 365
intervertebral arteries 192, 208
intracranial 239-40
joint 204-5, 207, 239, 241-2, 244,

279, 404, 406
intramural 239-40
chamber 247
recess 236

lonoscopus 366, 392
Ischnacanthidae 336, 374
Ischnacanthus 324, 336
Isurus 301

Jagorina 240, 253, 295, 301, 306,
341-2, 352, 355, 368, 381,
403

Jamoytius 254

jugal 70-3, 181, 287, 291-3, 310,

312-13
canal 312-14, 316, 395, 397

jugular canal 213, 216, 218, 224,
228, 230, 232, 235-41, 245,
251, 273, 277, 348
groove 216, 232-3, 237
vein 218, 235-6, 238-9, 241, 245,
249, 251

Jurassic 244, 365; see Upper Juras-
sic

Kansas palaeoniscid B 247

Kansasiella 182, 185, 188-9, 192,
195-6, 198-9, 204, 208, 211,
215-16, 228, 233, 236-8,
242, 244, 247, 249, 254, 264,

275, 277, 398

Kentuckia 185, 188-9, 192, 195-6,
198, 203-5, 211, 216, 228,
231, 236-8, 242, 244, 247,
249, 264, 275, 277, 280, 282,
299, 344, 398, 400
deani 88A

Kimberley Plateau 176

Kokenhusen 182

Kujdanowiaspis 240, 248, 279

labial cavity 406

Laccognathus 360

Lacerta 332

lacertilians 297, 306, 360

lachrymal 71-2, 74, 260, 264, 287,

291-3, 310, 313
Laemargus 324-5
Lamna 368
Lasanius 254
labyrinth cavity 212, 227
lateral aortae 192, 206-7, 234
lateral commissure 211, 216, 218,
235-41, 246, 273, 277, 304-5
lateral cranial canal 181-2, 211,
215-16, 227-8, 241-3, 316,
395-6

dermethmoids 267, 270
dorsal aortae 206, 279
ethmoid 267, 280, 283, 292,297-8
line 318, 360, 368, 387, 402

scales 387-8
occipital 210
postrostral 267
later alis branch 227, 237
canal 218, 230, 238
ganglion 218, 225-6, 230, 238,

249

root 218
laterohyal 352

Latimeria 204, 207, 209, 211-12,
215, 233, 238-40, 242-9,
251, 253, 266-7, 270-1, 273,

276, 279-80, 295, 298-9,

306, 311, 313, 324, 332-3,
339, 343, 346, 349, 352, 354,
356-7, 359-63, 365, 368,
392, 404, 406

Laugia 203, 212, 233, 239, 270, 332

Lawnia 251

Leiosteus 324

lens proteins 406

Lepidosiren 210, 297, 343

Lepidotes 183, 192, 198, 205, 212-
15, 234, 236-8, 242, 244-5,
249, 252, 267, 277, 279, 295,

307, 399-400
latifrons 280

lepidotrichia 364, 371, 374, 382,

384-6, 394, 403, 406
lepisosteids 246

RELATIONSHIPS OF PALAEONISCIDS

423

lepisosteoid 206

Lepisosteus 183, 198, 204, 206-8,
210, 212-14, 216, 227-8,
233-8, 242-6, 248-9, 266-7,
270, 275-7, 295, 299, 305,
307, 312, 314, 322, 324, 327,
332-3, 339, 344, 346, 352,
357-8, 360, 363, 365-6, 368,
370-1, 378, 381, 384, 386,
392, 394, 396-8

leptolepids 183, 191, 198, 203, 205-
6, 208, 211-12, 214, 228,
232-4, 236-8, 240, 242-6,
249, 252, 270, 276-7, 280,
295, 333, 339

Leptolepis 237-8, 242, 247, 278,
324, 368, 378

Leuciscus 235

levator arcus palatini 235, 245-6,
282-3, 395

levator maxillae 245

Lissamphibia 308, 336

longitudinal intervertebral ligament
199, 201, 210

lorical plates 374-5

lower jaw 90-6, 325-38, 359, 361

Loxomma 322

loxommatids 271, 311, 322

Luganoia 312, 386, 398

Lunaspis 324, 341

lungfishes 244

lungs 406

Lupopsyrus 374

Lyrocephalus 312, 324

Macrepistius 183, 215, 232, 237-8,
247, 249, 265, 267, 277-8,
366
Macromesodon 252, 305, 307, 309,

312-13, 399

Macropetalichthys 240, 266
Macropoma 211-12, 233, 238-9,
267, 270, 298-9, 306-7, 313,
332-3

Macrosemius, macrosemiids 366
mammals 185, 320
mandibular arch 235
bone 236
commissure 235
constrictor muscle 245
gill-cover 339
joint 402
pit-line 327, 331
plate 338
ray 349
sensory canal 326-7, 331, 333,

336-7, 397

marginal rays 181, 397, 382
masseter fossa 243
maxilla 60, 65-7, 260, 263, 283,
286-7, 291-3, 307, 310, 312-
13, 331, 396, 398-9, 403
maxillary teeth pi. 4c
Meckel's cartilage 301, 325, 332,

336, 338, 346, 402
Meckelian bone 325-7, 331

fenestrae 336
ossification 332
median basirostral 295
dorsal plate 376, 378
fins 145, 384-6
intramural chamber 247
lorical plate 374
postparietal 312
supraotic cavity 247
medulla 242

Megalichthys 207, 270-1, 276, 280,
295, 299, 306, 352, 366, 368
Megalops 270, 394
Meidiichthys 394
Melanognathus 332
membrane bone 203, 214, 231-2,
236, 247, 321, 362, 366, 368,
402

mentomandibular 332, 336
mentomeckelian 325, 327, 331-3,

336

Mesacanthus 253, 336, 341
mesocoracoid 378
arch 371, 378-80
process 371

mesodentine 336-7, 392
Mesonichthys 251, 307
Mesopoma 371, 392
Mesturus 252
metapterygium 371, 374, 381-3,

402-3, 406
metapterygoid 282-3, 286-7, 299,

305-6, 310

metencephalic recess 249
Metoposaurus 312, 324
metotic fissure 185, 204
Microdon 399

middle cerebral vein 227, 230, 249-
51

canal 225
pit-line 316
region of pectoral girdle 378-80,

396-7

Mlllerosteus 324

Mimia 176-400 passim, esp. 181.
toombsi 1-6, 11-26, 33-44, 50,
53-57, 60-63, 65, 68, 70-2,
75A, 79-82, 84-5, 90-3, 97,
101, 104, 107-8, 181, 185-8,
215, 254, 266, 271, 280, 316,
325, 338, 363, 368, 382, 384,
386, pis 1, 2a-c, 4-5
Monongahela 297
monotremes 297, 299, 380
Mormyrus 378
Moythomasia 181-400 passim, esp.

181-2

durgaringa 7-10, 27-32, 45-8,
58-9, 64, 66-7, 69, 73-4, 83,
87, 94-6, 99-100, 103, 105-
6, 109-11, 114-15, 118D, F,
131-6, 138, 139B, 140, 141 B,
142, 144, 182, 199, 228, 251,
260, 273, 292, 318, 331, 339,
349, 364, 374, 382, 385, 387,
pis 2d-f, 3

nitida 88B, 231, 251, 374, 382

perforata 388
musculus spiracularis 245
Mustelus 241, 299, 301, 324, 368
myliobatids 403
Mylostoma 337

myodome 195, 204-5, 207, 211,
214, 223, 237, 247-9, 266,
275-7, 396-9

anterior 279-80
myomere, first 208, 210, 280

second 207-9

Namaichthys 307

nasal 43, 46, 48-9, 254, 257, 259-
60, 263-4, 267, 270-1, 293,
316-17, 404

capsule 182, 254-5, 257-8, 266-7
cavity 260, 266
rosette 396
nasobasal canals 254, 256-8, 266-1,

271

Necturus 357
Nematoptychius 251, 307, 310

greenocki 306

Neoceratodus 183, 191, 206-8, 210,
234^5; 238-9, 243, 247, 251,
266, 271, 276, 297, 306, 312-
14, 324, 333, 338, 343, 346,
352, 355, 357, 360-1, 365
Neonesthes 357

neopterygians 190, 237, 244-6, 275,
332, 352, 357-60, 386, 394,
397, 399

Neopterygii 395-7, 400
Neorhombolepis 366
nervus lineae lateralis 199
Nesides 182-3, 207, 212, 232, 238-
40, 243, 245, 248, 266, 273,
276, 279, 299, 301, 306, 314,
324, 352, 354, 356
neural arches 121, 123, 198, 363,

365, 368, 386
spines 363, 368

neurocranium 7, 13, 50, 182-280,
291, 299, 306, 316, 318, 320-
1, 378, 396, 402, 403, 406,
pi. 1

acanthodian 183-4
actinistian 184
chondrichthyan 183
leptolepid 182-3, 203
pachycormid 183-4
pholidophorid 182, 185
placoderm 183
rhipidistian 184
shark 183
teleost 183-4
tetrapod 184

neurohypophysial hormone 406
neurokinesis 184, 205
neuromasts 225, 244, 293
Nostolepis 336, 392
nostrils 270-1
notidanids 304
notochord 182, 196, 207-8, 363

424

B. G. GARDINER

notochordal canal 189, 191, 195-6,
201, 207

plate 182

sheath 365
notopterid 206
Notorhynchus 368

occipital arch 185, 189, 203, 206,

210

arterial supply 197
artery 192, 196-8, 201, 207-8,

210

condyle 191, 210, 275
fissure 185, 188-9, 191, 199, 201-
5, 215, 217, 228, 245, 316,
318

myomeres 208-10
nerve canal 198
first 198, 208
foramen 192, 196-7, 201, 208-

second 208
third 191
ossification 2-4, 184, 215, 234,

280

process 190

region 4-6, 8-10, 12, 14-15, 185,
188-9, 199, 203, 207-8, 279-
80, 318, 349
discussion 201-10
segment 208
occiput 184, 190, 197-8, 201, 206-8,

210, 234, 280

segmental structure 208-10
oculomotor foramen 214, 218, 227,

238

nerve 248
olfactory nerves 226, 254, 266

canal 226, 254, 257, 260, 265-6
organ 267
oligopleurids 366
onion-skin growth 387, 392
onychodont 253-4, 287, 307, 311-

13, 324, 333, 339
Onychodus 311-12
opercular 99, 217, 244, 338-9, 342-4
adductor 231
bones 338, 342-4
cartilages 338, 341,342-3
process 352-4, 398
rays 342
operculogular series 338-9, 341,

402

Ophiopsis 366, 394
ophthalmic artery 222, 230
canal 251
foramen 223
opisthotic 186, 203, 211-13, 231-3,

240, 245

optic fenestra 214, 225, 227
foramen 214, 249
lobe 226-7, 249
nerves 226, 228, 248-9
oral sensory canal 331, 333, 406
orbit 20, 30-1, 215, 218, 224, 226,
230, 238, 249, 254, 260, 266,

298-9, 313
orbital artery 218, 228, 234, 236,

238-40, 276
cartilage 247, 301
foramina 218
groove 218, 234
region 213, 215
surface 20, 218
orbitonasal artery 181, 189, 218,

222,266
foramen 218
canals 255, 265
orbitosphenoid 214
orbitotemporal region 14-17, 19-
22, 27-31, 33, 210, 216, 231,
247-54, 277, 306
Orectolobus 241
orobranchial chamber 235
'Orodus* 392
Osmerus 295
Ospia 203, 228, 237-8, 242-3, 306-

7,332

ossification centres 183-4, 210
osteichthyans 177-403 passim
Osteichthyes 399
osteodentine 336
osteoglossoids 277-8, 357, 360, 366,

386

Osteoglossum 232

osteolepids 254, 270-1, 277-8, 280,
295, 299, 307, 310-12, 314,
321, 324, 332
Osteolepiformes 177, 288
osteolepiforms 270, 320-1, 324,
333, 339, 343, 362, 370, 374,
379-80, 392

Osteorhachis 233, 247, 267
osteostracans 336
otic bones 232
capsule 185, 195, 203-4, 212-13,

235, 246

nerve 214, 216, 237
canal 237-40, 242
foramina 218, 230
process 246, 298-306

of palatoquadrate 244
recess 406

region 1, 4-6, 11-12, 14-15, 27-
8, 185, 191, 203, 210-18,
227-8, 346, 348, 396
shelf 235, 241
otico-occipital 279
otico-sphenoid fissure 181, 210,

216, 218, 231, 241
otolith 227, 395, 396

chamber 211
Oxynotus 235-6, 239-40, 301, 403

pachycormids 183-4, 203, 205-6,
214-16, 233-4, 236-7, 240,
244, 267, 270, 277, 280, 295,
324, 339, 362, 365, 386

Pachycormus 183, 189, 192, 215,
234, 237-8, 244-5, 249, 252,
265, 267, 270, 275, 333, 374,
378

Pachyosteina 341

Pachyosteus 341

Palaeacanthaspis, palaeacanthaspids
381

Palaeoherpeton 312-14

Palaeomylus 337

palaeoniscids 177, 183-5, 188-90,
192, 195, 198, 201, 203-7,
211, 213-16, 226-8, 230-1,

234, 236-40, 242-3, 245-6,
248, 251, 260, 264-5, 267,
270, 273, 275-8, 287, 292,
306-7, 309-10, 322, 327,
332, 339, 346, 352-3, 374,
383-4, 386, 392, 397-8

Palaeonisciformes 177, 396, 398

palaeoniscoid 181

Palaeoniscus 371, 378, 381, 386,

398

Palaeozoic selachians 203
palate 75-6, 331, 404
palatine artery 222

canal 222

fenestra 218, 230, 237

nerve 218, 224, 227-8, 236-8,
240, 258

ossification 298, 307-8

process 298

vein 222

palatobasal articulation 298-301
palatoquadrate 53-6, 58-60, 206,

235, 245-6, 280-7, 293-310,
336, 346, 396-8, 400, 402-4

basal process 295
cartilage 254, 282
commissure 293-5
ossifications 305-10
otic process 244
pterygoid process 297
symphysis 295

Palaeoherpeton 299, 306, 311
Paleopsephurus 275-6
Panderichthys 271, 313, 321-2, 332
parabasal canal 222, 230, 237-8,

275-6

parachordals 188-9, 204, 207
Paramblypterus 205, 251, 386
parampullary process 199, 212,

217-18, 231-3
Paraplesiobatis 341
parapophyses 366

parasemionotids 183, 198-9, 203-6,
210-11, 231, 234, 236-8,
240, 242-6, 249, 265, 267,
270, 275-8, 280, 295, 332,
392, 394

Parasemionotus 192
parasphenoid 7, 13, 50-2, 181, 205-
6, 210, 216, 218, 222, 231-2,
234, 260, 271-80, 283, 286,
299, 362-3, 395-6, 402, 404,
406

teeth 278-9, 402
parasymphysial plate 333
Pareiasaurus 380
Parexus 374-5

parietals 80-3, 311-12, 316-18,
320-2, 324

paroccipital process 212

parotic crista 243
toothplates57, 273, 280

pars ganglionaris 236

pars jugularis 236

pectoral fin 737, 369, 371, 374,

381-3, 387, 404
spine 380
girdle 725-9, 752, 368, 274-381,

402

propterygium 181, 371, 381, 395-
7

peg-and-socket articulation 392,
395, 397

Pelobates 324

pelvic claspers 402-3
fin 382-3, 387, 404
girdle 138, 382-4, 402, 406

perichondral ossification 182-5,
188-9, 192, 195-6, 199, 201,
203, 210-11, 213-16, 226,
228, 234, 241-2, 249, 251,
253-4, 266-7, 298, 306, 316,
321, 325, 327, 332, 344, 357,
359, 363, 371

perichordal commissure 240, 360
tissue 191

Perleididae 398, 400

perleidids 183, 237, 267, 270, 386

Perleidus 189, 203-5, 211-12, 214,
228, 231, 236-8, 242-6, 267,
275-80, 343, 374, 394
cf. stoschiensis 183, 203

Permian 392

Petalichthyida 336

petalichthyids 341

Phanerosteon 368, 392

pharyngobranchial 233-4, 301, 359,
362-3, 402-3, 406

pharyngoepihyal 349

pharyngohyal 241, 349, 352

pharyngomandibular 301

Phlyctaenaspis 253

phlyctaeniid 337, 341, 356

pholidophorids 182-3, 191-2, 195,
198, 203, 205-6, 208, 210,
214-16, 228, 232-4, 236-8,
240, 242-7, 249, 252, 265,
267, 270, 275-7, 280, 283,
339, 357, 362-3, 365, 386

Pholidophorus 189-90, 197-9, 203,

211-12, 214-16, 227-8, 230,

234, 237-8, 242-3, 267, 305,

364, 378

bechei 183, 188, 201, 203, 214-

16, 277-8, 280
germanicus 198, 332
higginsi 333
macrocephalus 252

pholidopleurids 183, 189, 192, 203,
205, 233, 237, 264, 275, 365-
6,386

phyllolepids 324

phylogenetic results 394

RELATIONSHIPS OF PALAEONISCIDS

pila antotica 218, 236, 239, 247-8
pineal foramen 182, 216, 226, 316-

17, 320
pituitary 249, 406

canal 222

fossa 218, 227, 231,396, 404

vein 188, 222, 249
Pinkus' ogan 244
pinnal plates 375
pit-line 260, 313, 322, 324, 327,

331, 339, 341
Placodermi 399

placoderms 183, 185, 190, 207-8,
210-11, 228, 233-6, 240-1,
243, 247-8, 253, 266, 270-1,
275-6, 278-9, 295, 298-9,
301, 306, 313, 321, 324-5,

332, 336-7, 341-2, 352, 355,
368, 374-6, 378, 381, 392,
400, 402-3

Placodus 308

placoid 392

Platysiagum 343, 386, 399

Platysomus 277, 394, 398, 400

Plegmolepis 346

polar cartilages 188, 241, 247-8,
299

Polyodon 183, 185, 190, 198, 206,
212, 233-4, 236-7, 239-40,
242-4, 270-1, 277, 295, 297-
8, 312-13, 324, 333, 339,
353, 358, 360, 363-4, 382-4,
394, 397-9

Polypterus 88D, 183-4, 189-90,
192, 194-6, 198, 203-8, 210-
12, 214, 216, 227, 231-4,
236-8, 240, 242-9, 251, 260,
266-7, 270, 275-80, 28?,
292-3, 295, 297-9, 305-6,
309-11, 313-14, 324, 327,
331-3, 336, 339, 342-3, 346,
349, 352-4, 356-7, 360, 362-
4, 366, 368, 370-1, 374, 379-
82, 384, 386-8, 392, 394,
396-9
bichir 75B, 77C, 295

Poracanthus 392

pore canal 392

porolepids 254, 270-1, 273-8, 280,
295, 298-9, 301, 306-7, 311-
14, 324, 332, 368, 374, 404,
406

Porolepiformes 177, 400, 404

porolepiforms 278, 320, 324, 333,
339, 360-2, 370, 403-4

Porolepis 238, 240, 258, 266, 273,
278, 280, 283, 295, 299, 324,
339, 341
brevis 78B, 88E

postbranchial lamina 376

postcleithrum 727, 733, 370, 374,
397

posterior ascending process 273,

277

basicapsular commissure 204
cerebral vein 197, 199, 201, 215,

425

217

dorsal fontanelle 185-6, 189, 199,
201-3, 206, 215-16, 247, 318

myodome 247-9, 266, 396-9

nasal tube 257

nostril 260, 263-4

pit-line 317, 324

semicircular canals 182, 212, 216,
227-8, 231, 241-3, 246

stem of parasphenoid 273, 395
post-ethmoid portion of neurocra-

nium 25-6
postfrontal 311
postnasal wall 254, 257, 260, 263-4,

266-7, 298, 402, 404
postorbital 310-12, 322

process 211, 215-16, 224, 226,
235, 238, 240, 242-3, 245-6,
291, 301, 318
postotical process 212
postpalatine process 235
postparietals 320-2, 324
postrostral 267, 271
postspiracular 339, 343
postsplenial 336
postsuborbital 306
post-temporal 86-7, 725, 190, 318,
368, 376, 396, 399

fossa 212, 239, 242-3, 246, 399

process 396

Powichthys 311, 324, 404
prearticular 327, 331-3, 399
pre-ethmoids 267, 396
pre-epiotic pocket 243
prefacial commissure 218, 236-7

floor 237
prelateral 338

premaxilla 44, 47-9, 254, 256-60,
263-4, 270, 292-3, 396, 398-
9,403

preopercular 59-63, 283, 287, 291-
3, 312-13, 338, 342, 396-9

sensory canal 287, 293, 312-13,

399

preoperculo-jugal canal 312
prepalatine floor 230, 237

strut 237

prepubic process 406
prespiracular cartilage 301, 304-5

groove 276, 278
presplenial 333

presupracleithrum 339, 395, 397
prismatic calcifications 184

ganoine 398

profundus canal 218, 230, 239-40,
266

foramen 218, 238, 240

nerve 227, 248, 255, 258, 266-7

root 238

pristiophorids 239
Pristiophorus 241
Pristis 241
Pristiurus 247, 381
processus lingualis 360
Procolophon 380
procoracoid 380

426

B. G. GARDINER

Proleptolepis 333

prootic 184, 186, 188-9, 192, 195,

203-5, 207, 211, 214, 216,

218, 232-4, 236, 240-2, 246-

9, 251, 273, 363

bridge 182, 188, 194-5, 207, 239,

247-9
knob 232
process 235

propterygial canal 381

propterygium 136, 371, 374, 382

Protogonacanthus 253

Protopterus 210, 297, 314, 316, 324,
368

Protosphyraena 183

protractor hyomandibularis 245-6,
352

Pseudocarcharias 301

pseudopetalichthyids 336, 341, 381

Pseudopetalichthys 341, 381

pseudoprismatic 388

Pteronisculus 182, 185, 188-9, 192,
195-9, 201, 203, 205, 207,
211-12, 214-6, 227-8, 233-4,
236-8, 242, 244, 249, 251,
254, 264, 273, 275-80, 282-
3, 286-7, 293, 295, 299, 305-
7, 310, 314, 317, 326-7, 331-
2, 338-9, 343, 346, 348, 357,
360, 362-4, 368, 370-1, 374,
381,384,386,394,398,400
cicatrosus 236

macropterus 185, 189, 204-5
magnus 88C, 183, 247
stensioei 204

pterosphenoid 211, 213-15, 247,

249
pedicel 230, 238, 247, 249

pterotic 203, 212-13, 240

pterygoid 283, 297, 307, 310, 406

Ptomacanthus 295, 298, 375

Ptyctodontida 336

ptyctodonts 185, 299, 306, 337, 368,
375-6, 381, 402-3

Ptyctodus 337

ptycholepids 267, 270

Ptycholepis 394

pulmonary vein 403, 406

Pycnodontiformes 399-400

pycnodonts 216, 275, 278, 295, 297,
307, 309, 386

Pygopterus 283, 343, 346, 365, 374,
384, 386, 394, 398

quadrate 57, 282-3, 286-7, 305-7,

325, 359, 396, 399
quadratojugal 57, 63-4, 283, 287,

291-3, 313-14, 396-7, 399
pit-line 291, 293, 314

radial plate 739, 384-6
radials371, 381-4, 386,403
Radotina 403
Raja 235, 241, 301
ramus lateralis accessorius foramen
218, 225

rays 235, 241, 301, 304, 352

recess for telencephalon 226, 249

recti muscles 214, 223, 248-9

Redfieldiidae 398

redfieldiids 267, 394

Redfieldius 343

relationships of actinopterygians
400

'reptiles' 299, 308, 311

restoration of fish 145
skull 101-3

retina 254

retroarticular 327, 332

Rhabdoderma 211-12, 215, 239-40,
270-1, 298-9, 306, 313-14,
324, 332-3, 339, 362, 374,
379
elegans 77 A

Rhadinichthys 371, 398

Rhamphodopsls 337

rhenanids 253, 336, 341, 381, 403

Rhina 368

rhinobatoids 239

Rhinobatus 240-1

Rhinodipterus 253, 322, 325, 341

rhipidistian 177, 201, 203-8, 212,
215, 233-4, 238-41, 243,
246-8, 251, 253, 266, 270,
273, 277-80, 299, 311, 324,
339, 366, 368, 392, 400, 404,
406

Rhizodopsis 210-11, 239, 243, 245-
6, 251, 266, 366

Rhynchodipterus 366

Rhynchodus 381

ribs 363-4, 366-8
dorsal 366
pleural 366, 368
ventral 366, 368

ridge scales 386

Romundina 324, 337, 376, 381, 403

rostral 42, 45, 48-9, 254, 256, 258-
60, 263-4, 267, 270, 293,
317, 395, 397
organ 404

rostro-palatine articulation 297-8

rostro-premaxillo-antorbital 270

Sabrinacanthus 375, 381
saccopharyngoids 343
saccular cavity 207

recess 217, 225, 227
sacculus 196
saccus vasculosus 222
Sagenodus 297, 313
Salamandra 308
Salmo 235, 248, 297, 299, 305-7,

352, 361, 368, 371
salmonoids 206, 386
sarcopterygian 177, 271, 307, 342,

362, 374, 378, 392, 406
Sarcopterygii 183, 400
saurichthyids 276, 295, 386
Saurichthys 190, 198-9, 205-6, 236-

8, 243, 254, 265, 271, 275-9,

306

Saurorhynchus 276

scales 140-4, 182, 370, 386-8, 392,

398, pis 2, 3
structure 143-4, 388
Scaphirhynchus 295, 301, 370, 382-4
scapula 378-80
scapular foramen 371, 378, 380

anterior 378, 380
portion 371
process 381

scapulocoracoid 378-81, 402
Scaumenacia 89B, 271, 297, 313,

322, 324-5, 341
sclerotic bones 228, 251-4

ring 228, 231,251,403-4
Scomber 198, 236, 374
scorpaenoids 236
Scyllorhinus 301

Scyllium 210, 247, 301, 342, 402
Scymnodon 235, 239, 240
Scymnorhinus 240
second segment 207
Sedowichthys 337

selachians 183, 203-4, 206-7, 210,
214, 234-6, 238-41, 244-9,
251, 295, 298-9, 301, 304,
312-14, 324, 342, 349, 352,
355-6, 396, 402-3, 406
Selachii 399
semicircular canals 402
semionotids 183, 206, 210, 212,
214-16, 233-4, 236, 240,
244, 267, 270, 275, 277, 295,
297

Semionotus 394

sensory canals of cheek 313-16
skull roof 324-5
snout 267-71
Seymourla 322
Sigaspis 324
silurids 392
Sinamia 183, 244
Sinemurian 238, 247, 278
sinus superior 216, 227-8, 242, 246
snout 185, 259
sphenotic 211-12, 214-15, 240, 242,

244

Soederberghia 270, 366
solum nasi 267
Somniosus 301
Sphenodon 306, 360 -
spinal nerve 363
first 208
fourth 379
plate 375-6

spino-occipital nerves 198, 210
spiracle 211, 238-40, 242-4, 246,

277,291, 301, 310-11, 313
spiracular canal 211, 216, 238, 242-
4, 246, 271, 277, 301, 395,
397

cartilage 235
cleft 244, 278, 291, 317
diverticulum 242, 277-8, 283
groove 181, 216, 238-9, 242-4,
246, 271, 273-9, 283, 286,

RELATIONSHIPS OF PALAEONISCIDS

427

292, 395-6, 402
ossicles 243, 352
pouch 243, 310
recess 242
rudiment 301
sense organ 244
slit 181, 283, 286, 317
spiraculo-hyomandibular recess 283
splenial 333, 336, 402
squaloids 403
Squalus 212, 235-6, 239-41, 247-8,

298-9, 301, 368
squamation 386, 388-92
squamosal 312, 316, 395-6
Squatina 235-6, 239-41, 360
stegotrachelid palaeoniscid 181
Stegotrachelidae 181
Stegotrachelus 312, 374, 398
stensioellids 336
stephanodont teeth 336-7
sternohyoideus muscle 346
Stomiahykus 306
Strepsodus 366
stromatoporoid reefs 176
Strunius 253

walteri 77B
sturgeons 233, 243, 277, 333, 339,

353, 381, 392
Styracopterus 394
subcephalic muscles 271, 273, 279-

subcranial muscle 197
'subholosteans' 398
sublingual rod 360
submandibular 339, 341-2, 404
submarginal 341
subopercular 338-9, 342-3, 403
suborbital 312, 398

shelf 254, 263, 298-9
subotical shelf 240
subtemporal fossa 212, 232, 245
superficial ophthalmic nerve 237-8,

247, 251, 255
foramina 225-6, 230, 239-40,

249
superior oblique 254, 265-6

rectus 249
superognathal 337
supra-angular 94, 182, 331-3, 397-8
supracleithrum 725, 339, 368, 370,

374
supracoracoid foramen 379, 381,

402

supraethmoid 267
supraglenoid buttress 380

foramen 378, 380-1
supramaxilla 396, 399
supraneurals 268, 363, 368, 386
supraoccipital 189, 199, 201, 203,

206-8, 216
supraorbital 311, 404
sensory canal 225, 254-5, 260,

263, 270, 312, 316-7, 324-5,

404

vein 249, 251
supraotic cavity 201, 247

suprapharyngobranchial 118, 218,
233-5, 301, 362-3, 403

articulation 233-4, 348

first 218, 233-4, 346, 348

second 233, 348-9
suprapharyngohyal 352
suprapharyngomandibular 301
suprapterygoid 306

process 211

supratemporal 80-3, 291, 311, 317-
18, 320, 322

commissure 312, 318, 324
suprascapula 380
surangular 331, 333, 336
swimbladder 396, 403
symplectic 357-9, 396, 399
synapomorphies summarized 395-
406

tabular 311-12, 320-1

Tamiobatis 203-4, 207, 210, 235,
240, 301

Tarpon 394

Tarrasius 386

tectals 271, 404

tectum 267

teeth 260, 263-4, 271, 273, 278-9,
286-7, 292, 295, 307, 310,
331-2, 336, 403, pi. 4

Tegeolepis 339, 394

telencephalon recess 226, 249

Teleostei 400

teleosts 183-4, 189, 191, 195, 197,
203, 205-6, 211-12, 214,
216, 228, 233-4, 236-8, 240,
243-7, 249, 251-2, 265, 267,
270, 275-8, 295, 297, 299,
305-7, 312-13, 322, 324,
332-3, 339, 343, 346, 352,
357, 359-63, 365-6, 368,
374, 378-80, 382, 384, 386,
388, 392, 394, 397

temnospondyls 243, 311, 322, 333,
336, 365

temporal sensory canal 239, 244,

287, 311-12, 317-18
series 312

tentacle blastemas 295

tesserae 183-4, 336

Tetrapoda 400

tetrapods 183, 185, 204, 206-7, 243,
247, 270-1, 275-6, 299, 301,
307-8, 310-13, 320-2, 324,
332-3, 336, 341, 361, 365,
368, 370, 380-1, 392, 400,
403-4, 406

Thursius 368

tooth plates 271, 273, 278-80, 283,
292, 333, 336-8, 344, 346,
348-9, 356, 361-2

Torpedo 247-8, 251, 313, 324

torpedoes 235

trabeculae 188, 204, 235, 241, 247-
8, 267, 293, 295, 299

transpalatine 307-8

transverse bolster of basisphenoid

214, 249
process 190-1
transversi ventrali 403
Trematosaurus 324
Triazeugacanthus 253
trigeminal canal 218, 230, 237-8,

248-9, 251
foramen 18, 195, 211, 214, 218,

224, 227, 237-8, 248
nerve 207, 227, 238, 248, 251,

327

ganglion 236-9
mandibular branch 298, 313,

327

maxillary branch 298
root 218, 251
trigeminofacialis chamber 235-40,

244

Trimerorhachis 312, 324
Tristychius 241
Triton 306
trochlear nerve 224, 226, 248-9

foramen 225, 249
truncus infraorbitalis 266
trunk muscle 197, 199, 209
Turseodus 365
Tylototriton verrucosus 76E

uncinate processes 398-9

Undina 298, 306

Upper Carboniferous 365

Jurassic 232, 234, 236, 243
Uranolophus 271, 273, 275, 279,

295, 311, 324, 392, 394
urodeles 208, 299, 306, 308, 360,

362
utricular recess 18, 218, 227-8, 237-

9, 251
urohyal 362
Urolophus 301
Uronemus 297, 321

vagus canal 188, 197, 199, 201, 203,

210-11, 217, 228
foramen 189-91, 201, 212, 233-4,

396

nerve 199, 216-17, 245, 316, 348
pharyngeal branch 216-17
supratemporal branch 216-17,

233

valvula 395
vaterite 395, 397
ventral ethmoid 267, 295
ventral + marginal fin muscles 379,

381, 383

ventral otic fissure 181, 185, 188-9,
191, 195, 199, 204-6, 210-
11, 214, 222, 224, 234, 241,
249, 271, 273, 275, 277, 279
ventrolateral sinus 251
Vermicomacanthus 374-5
vertebrae 191, 366

first 208

vertebral joint 205
vertical pit-line 57, 61-2, 287, 291,
313-14

428 B. G. GARDINER

vestibular fontanelle 185-6, 188-9, Watsonulus 203 Xenacanthus 203-4, 210, 235, 239-

192, 199, 203-5, 215-17, Western Australia 176, 182 40, 298, 301

249, 348 Australian Museum 181 Xenomystus 206

visceral arches 301 Westoll-lines 392 v . ,. , .._,. .„,

• • ->••>*• ii/i •.. • in ">r>r TH -IT-, i Youneolepididae 177. 404

origin 235 Whiteia 211, 295, 324, 332-3 , ,,_- ™ o TH AHA

*n TV> o/:n i^^ 0-71 o-ri ii/'-j • -in-r o^n voungolepidids 273, 277-8, 311 , 404

vomer 50, 222, 260, 264, 271, 273, Wijdeaspis 207, 240 L .t ' ^

_n ' 0 '„ . . !. oc Youngolepiformes 400

293-8, 404 warrooensis 185 „ / • ™* ii« i™ ^r, ^,^-r

„,., , 100 Youngolepis 204, 210, 239-40, 247,

Wildungen 182 O-IQ OAA 97/i_A 97«_sn

Watson's palaeoniscid A 194 Wimania 211-12, 233, 273, 298 f£? „, inn iu' Z/8"8U'

251, 307, 310, 344, Zy8' "*• 4UU' 4l

374 xenacanth sharks 202, 207, 403 zygal plates 192, 194-6, 207, 227

Accepted for publication 1 November 1983

British Museum (Natural History)

An account of the Ordovician rocks of the Shelve Inlier in west Salop
and part of north Powys

By the late W. F. Whittard, F.R.S. (Compiled by W. T. Dean)

Bulletin of the British Museum (Natural History), Geology series
Vol. 33 No. 1. Dec. 1979. 69pp. 38 figs. Large full-colour map

The work complete with map, £10.50 (Post & packing 30p)
Map only, £1.00 (P & p. lOp)

A related work :

Ordovician Brachiopoda from the Shelve District, Shropshire

By A. Williams

Bull. B.M.(N.H.}, Geology Supplement 11, 1975. 163pp., 28 plates, 5 tables,
11 text figs. £13.00 (P & p 50p)

All British Museum (Natural History) publications are obtainable from Agents,
Booksellers, the Museum bookshop or by post direct from:
Publication Sales, British Museum (Natural History), Cromwell Road,
London SW7 5BD, England

Titles to be published in Volume 37

Taxonomy of the arthrodire Phlyctaenius from the Lower or Middle
Devonian of Campbellton, New Brunswick, Canada.

By V. T. Young

Ailsacrinus gen.nov.: an aberrant millericrinid from the Middle
Jurassic of Britain. By P. D. Taylor

Miscellanea

The relationships of the palaeoniscid fishes, a review based on new
specimens of Mimia and Moythomasia from the Upper Devonian of
Western Australia.

ByB. G. Gardiner

Printed by Adlard & Son Ltd, Bartholomew Press, Dorking, Surrey

BOUND

20JW.WM

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Bulletin of the British Museum (Natural History)"

See other formats