Full text of "Records of the Western Australian Museum"

See other formats

ISSN 0312 3162

Records

of the

Western Australian
Museum

Volume 23 Part 1 2006

Records

of the

Western Australian Museum

The Records of the Western Australian Museum publishes the results of research into all branches of natural
sciences, and social and cultural history, primarily based on the collections of the Western Australian
Museum and on research carried out by its staff members.

Collections and research at the Western Australian Museum are centred on Earth and Planetai^ Sciences,
Zoology, Anthropology, Archaeology and History. In particular the following areas are covered: sy.stematics,
ecology, biogeography and evolution of living and fossil organisms; palaeontology; mineralogy; meteoritics;
anthropology and archaeology; history; maritime history and maritime archaeology; and conservation.

Western Australian Museum

Perth Culturai Centre, James Street. Perth. Western Australia 6000

Mail; Locked Bag 49, Welshpool DC, Western Australia 6986

Tel. (08) 9427 2700

Fax. (08) 9427 2881

E-mail reception@museum.wa.gov.au

Minister for Culture

and the Arts The Hon. Sheila M'^Hale MLA

Chair of Trustees Mr Tim Ungar BEc. MAICD, FAIM
Executive Director Dr Dawn Ca.sey PSM FAHS

Editor Dr Kenneth J. McNamara BSc(Hons), PhD, FAAAS

Editorial Board Dr Alex Baynes MA, PhD

Dr Alex Bevan BSc(Hons), PhD
Ms Ann Delroy BA(Hons), MPhil
Dr Paul Doughty BSc(Hons), PhD
Dr Bill Humphreys BSc(Hons), PhD
Dr Bany Hutchins BSc(Hons), PhD
Dr Moya Smith BA(Hons) DipEd. PhD

The Records of the Western Australian Museum i.s publi.shcd approximately three times per year. A serie.s of
Supplements is also produced The Records are available for sale and exchange, the current price being $11 plus
postage per part. Each volume comprises four parts. Subscriptions can be taken out for whole volumes at a rate of $33
plus postage. Supplements can be purchased from the W.A. Museum Bookshop. Prices on request. Prices include GST.

Printed and published by the Western Australian Museum
© Western Australian Museum, April 2006
ISSN 0312 3162

Cover: The leptolepid fish Cavenderichthys talhragarensis from the Late Jurassic of New South Wales.
Drawing by Jill Ruse.

Records of the Western Australian Museum 23: 1-6 (2006).

"’Si-

The fishes of Lake Kununurra, a highly regulated secHohT>£The
Ord River in northern Western Australia __

2006

H.S. Gill’, D.L. Morgan’, R.G. Doupe^ and A.J. Rowland’

' Centre for Fish and Fisheries Research, Murdoch University,

South Street, Murdoch, Western Australia 6150
’ Fish Health Unit, Division of Veterinary and Biomedical Sciences, Murdoch University,
South Street, Murdoch, Western Australia 6150

Abstract - The Ord River, situated in the east Kimberley region of Western
Australia, is regulated by two dams that supply irrigation water for tropical
agriculture. The regulation of water in the 55 km span of river between these
dams has resulted in this section now resembling a lacustrine rather than
riverine environment that is reflected in its name. Lake Kununurra. Utilising
various sampling techniques we captured/observed 4157 fish from 19 species
in 13 families, whilst the presence of a further three species was confirmed by
reliable sources. Nemalaiosa erebi, Craterocephalus stramineus and Melanotaenia
australis were the three most widespread and abundant species, being
encountered throughout the lake and representing -66% of all fish recorded.
Ambassis sp., Anus midgleyi, A. graeffei, Glossamia aprion, Amniataba percoides,
Hephaestus jenkinsi, Glossogobius giurus and Toxotes chatareus were found
throughout the lake but generally at fewer sites and in smaller numbers (-30%
of all fish) than the three dominant species. Tlae remaining species Neosilurus
ater, Strongylura krefftii, Leiopotherapon unicolor, Syncomistes butleri, bates
calcarifer, Mogumda mogumda, Oxyeleotris lineolatus and Ambassis maclea}'i were
generally found at few sites and in low numbers, and accounted for less than
4% of the overall catch. The regulation of the Ord River has apparently altered
the species composition and abundances in Lake Kununurra. For example,
marine/estuarine species, commonly encountered in the freshwaters of large
northern Australian Rivers, are largely absent, while species such as L.
unicolor, which is one of the most abundant species in nearby rivers and
tropical Australian rivers in general, contributed to <1% of the total catch.

INTRODUCTION

The Ord River, situated in the east Kimberley
region of Western Australia, has an arid and semi-
arid monsoonal catchment of over 50 000 km^ that
extends into the western Northern Territory. The
650 km long river, which originates northwest of
Halls Creek and drains into the Cambridge Gulf
near Wyndham, has a mean annual stream flow of
3940 GL (into Lake Argyle), the second largest in
Western Australia (Water and Rivers Commission
2000). The river is regulated by two dams that
supply irrigation water for tropical agriculture.
Completed in 1963, the Kununurra Diversion Dam
was built across the Ord River and created Lake
Kununurra. Approximately 55 km upstream of the
Diversion Dam, the Ord River Dam that forms Lake
Argyle, was completed in 1973.

Operation of the Kununurra Diversion Dam
spillway gates, along with continual water release
from the Ord River Dam, maintains relatively
constant water levels in Lake Kununurra
throughout the year. Thus, these dams have
essentially altered the flow regime of this section of

the river from lotic to lentic conditions. Such
modification often results in changes in fish
communities, such as a reduction in the relative
abundance of species that favour faster flowing
reaches of rivers along with a concomitant increase
in the abundance of those species restricted to
slower flowing sections. Dams also often impede
species migrations within river systems (Bunn and
Arthington 2002).

Constant water levels within the lake have
encouraged the development of profuse riparian
vegetation. The inundation of surrounding
lowlands also has resulted in many fringing
swamps that are characterised by rich aquatic plant
growth, large woody debris and extensive stands of
emergent vegetation, such as the bulrushes Typha
domingensis and Eleocbaris spp. (Jaensch 1993). In
recognition of its significance as a dry-season refuge
for waterbirds, due to the consistency in the water
level of the lake and its varied habitats. Lake
Kununurra is now classified under the Ramsar
convention as a "wetland of international
importance".

Despite its listing as a wetland of international
importance, and noting the likely effects of habitat
change and large dams on the teleost fauna, no
literature exists detailing the lake's fishes. Thus, the
primary aim of this study was to determine the fish
fauna of Lake Kununurra. The distributions and
broad habitat associations of the species in the lake
are also discussed.

MATERIALS AND METHODS
Sample sites

The term 'Lake Kununurra', in the present study,
refers to the 55 km-long body of water between the
Lake Kununurra Diversion Dam and the Ord River

H.S. Gill, D.L. Morgan, R.G. Doupe, A.J. Rowland

Dam, including the tributaries and associated
perennial swamps and lagoons (Figure 1) (Jaensch
1993). In order to capture representatives of each
species within Lake Kununurra extensive sampling
was undertaken at 16 sites throughout its length in
a variety of habitats (Figure 1).

Sampling methods

Sampling was conducted during day-light hours
in November and December 2002. During this time
the fish fauna of the sites were sampled using a
variety of methods, including monofilament gill
nets (50, 100, 125, 150 and 200 mm stretched mesh),
seine nets (5 and 15 m nets of 3 mm woven mesh),
rod and line, and visual surveys (e.g., mask and

Figure 1 The sites sampled and the methods used to survey the fish fauna in Lake Kununurra.

Fishes of Lake Kununurra

snorkel). On capture, fish were identified, the
number of each species recorded and the majority
then released. A global positioning system (GPS)
was used to determine the longitude and latitude of
each site, and these data were then used, in
conjunction with the computer program Mapinfo
(Mapinfo Corporation 1998), to produce a map
showing their location (Figure 1).

RESULTS

A total of 4157 fish from 19 species in 13 families
were captured/observed in Lake Kununurra during
the present study (Table 1). Three species
dominated catches at the majority of sites within
the system and accounted for 66% of the total catch.
Bony Bream Nematalosa erebi (Gunther, 1868)
accounted for 27.3% (1133) of captured fish and was
recorded at 11 of the 16 sites sampled. The
Strawman Craterocephalus stramineus (Whitley , 1950)
and Western Rainbowfish Melanotaenia australis
(Castelnau, 1875) represented -20.1% (837
individuals) and -18.6% (775 individuals),
respectively, and were each recorded from 12 sites
(Table 1). The first of these species was most
common in shallows near deep water where groups
were observed feeding over sand or gravel
substrates whereas the two smaller species were
typically caught along the well-vegetated margins
where they congregated in moderate to large
schools near the surface in areas of low flow.

A further seven species were also widely
distributed, and accounted for just over 30% of the
total catch (Table 1). In decreasing abundance were
Northwest Glassfish Ambassis sp. (-8.5%, 356
individuals, 6 sites), a species associated with areas
of little water flow and copious growths of aquatic
vegetation and was formerly referred to as Ambassis
muelieri Klunzinger, 1880, by Allen and Burgess
(1990); Silver Cobbler or Shovel-nosed Catfish Arius
midgleyi Kailola and Pierce, 1988 (4.8%, 201
individuals) was captured at nine sites, from below
the Ord River Dam downstream to Packsaddle
Creek Lagoon; Mouth Almighty Glossamia aprion
(Richardson, 1842) made up almost 5% of the total
catch and was found throughout Lake Kununurra
(198 individuals, 8 sites), where it was generally
associated with aquatic vegetation; Barred Grunter
Amniataba percofdes (Gunther, 1864) was widespread
and relatively common (-3.2%, 134 individuals, 11
sites), forming aggregations in the shallows over
gravel, sand, or mud in fast (e.g., upper Spillway
Creek) to slow flowing or still waters (e.g., Lilly
Creek Lagoon); Jenkin's or Western Sooty Grunter,
or colloquially Black Bream Hephaestus jenkinsi
(Whitley, 1945) represented -2.8% of the total catch
(117 individuals) and was captured in nine sites
from Packsaddle Creek Lagoon to Carlton Gorge
where it was often associated with deep water and

structure in the form of submerged roots, logs,
rocks or vegetation; Flathead Goby Glossogobius
giurus (Hamilton, 1822) was relatively common and
widespread (-2.4%, 98 individuals, 8 sites) and was
most often found in the shallows on sandy bottoms;
Lesser Salmon or Blue Catfish Arius graeffei Kner
and Steindachner, 1867 was captured at six sites
from Carlton Gorge downstream to Packsaddle
Creek Lagoon (-2.3%, 96 individuals) and was also
common in the lower Ord River and Lake Argyle
(Rowland unpublished data); Seven-spot Archerfish
Toxotes chatareus (Hamilton, 1822) was relatively
common and widespread throughout Lake
Kununurra (-1.6%, 66 individuals, 8 sites), where it
was commonly observed patrolling the surface near
river banks or around riparian vegetation.

The remaining seven species encountered during
this study, i.e., Black Catfish Neosilurus ater
(Perugia, 1894); Freshwater Longtom Strongylura
krefftii (Gunther, 1866); Spangled Perch
Leiopotherapon unicolor (Gunther, 1859); Butler's
Grunter Syncomistes butleri Vari, 1978; Barramundi
hates calcarifer (Bloch, 1790); Northern Trout
Gudgeon Mogumda mogumda (Richardson, 1844);
Sleepy Cod Oxyeleotris lineolatus (Steindachner,
1867) and Macleay's Glassfish Ambassis macleayi
(Castelnau, 1878), were caught in low numbers and/
or at few sites and in total contributed less than 4%
to the overall catch (Table 1).

Three further species not captured during this
study have been reported from Lake Kununurra;
these were Freshwater Sawfish Pristis microdon
Latham, 1794 (Thorburn et al. 2003), Rendahl's
Catfish Porochilus rendahli (Whitley, 1928) (G.
Allen and M. Allen unpublished data) and Giant
Glassfish Ambassis gulliveri (Castelnau, 1878) (S.
McIntosh personal communication).

DISCUSSION

The results of this studv demonstrate that,
although Lake Kununurra is a highly modified
section of the Ord River, the number of freshwater
species inhabiting the lake is comparable to those
found elsewhere in the Kimberley. For example, the
capture of 18 freshwater species, not including the
catadromous L. calcarifer, during this study, in
addition to reports of two other species from
reliable sources, compares with 23 freshwater fish
species found throughout the entire Fitzroy River
(Morgan et al. 2002), 18 species from the Prince
Regent River (Allen 1975), 19 from the Drysdale and
Carson Rivers (Hutchins 1977), and nine from the
Mitchell River (Hutchins 1981).

Of the 4157 fish caught in the present study the
most abundant and commonly encountered species,
Nematalosa erebi, Craterocepbalus stramineus,
Melanotaenia australis, Ambassis muelieri, Arius
midgleyi, Glossamia aprion, Amniataba percoides,

Table 1 The number of each fish species caught at each of the sites sampled in Lake Kununurra. NB. Site ,1 Swim Beach boat ramp; 2, Packsaddle Creek Lagoon; 3, Lilly
Creek Lagoon (boat ramp); 4, Lilly Creek Lagoon; 5, Emu Creek Lagoon; 6, Four Mile Creek Lagoon; 7, main channel (opposite Stonewall Creek); 8, Stonewall Creek
(mouth); 9, Stonewall Creek (Roy's Retreat); 10, Spillway Creek (upper); 11, Spillway Creek (below spillway); 12, main channel (Sandy Beach); 13, Cooliman Creek;
14, main channel (lower Carlton Gorge); 15, main channel (upper Carlton Gorge); 16, main channel (below Ord Dam). * Although no N. erebi were captured at Site 16
during the present study they were seen there on a previous trip. (See Figure 1 for site locations).

H.S. Gill, D.L. Morgan, R.G. Doupe, A.J. Rowland

H .a

'O a^ ro LO

rn 00 cp p

oi ^ ^

OO if) 2

3 ^ -c

t/1 <2

'c ’6

O IN O

I I

o ^

T3 » C

W 41 3
3 <3 '4

I is I

Ci, c 3

E5 <3 A

■b S rS — ^
^ O a; •<

On <N CO 00

(N 00 tN. ro VD

(ri (N o o ^

(N ro

S f- ~

C S 5 3

!lj qj g ^

sill

O O
V V

^ o S

55 ^ o
,2 E E -b

" ^ ^ c s 1

I'.-q .1. r

C/i "a

03 t/3

"o 3

5 .S 2

3 0 ,3

E S

^5 ■§

“ I I

Fishes of Lake Kununurra

Hephaestus jeukiiisi, Glosso(;;obius giiiriis, Arius
graeffci, Toxotes chatnreus, were all recorded from
the lower-most regions of the lake (Swim Beach
Boat Ramp and/or Packsaddle Creek Lagoon) and,
with the exception of Anibassis sp., Arius graeffei
and T. chatnreus, were also found in the upper-most
sites (i.e., upper Carlton Gorge and /or immediately
below the Ord Dam). The last three species were,
however, captured at Cooliman Creek and/or lower
Carlton Gorge, sites that are in the upper third of
the lake.

Neinatnlosa crebi, C. stramiueus and M. australis,
in addition to being widespread throughout the
system, were the most numerous and frequently
encountered fishes caught during this study.
However, whilst N. ercbi was often associated with
the deeper main channel, the smaller species C.
stramiueus and M. australis were usually observed in
the shallow vegetated littoral zones. The fact that
much of the Lake Kununurra environment consists
of the preferred habitat of N. ercbi, i.e. still or slow
flowing vegetated waterbodies with mud/sand
substrate (Bishop et al. 2001) would help explain
why this species is so abundant and widespread
within the lake. In the case of C. stramiueus and M.
australis, the dense emergent and submerged littoral
vegetation would provide not only a refuge for
these small fish from predators, but also a rich
source of terrestrial and aquatic insect prey.

The rarer species encountered in Lake Kununurra
include Ambassis macleai/i, Leiopotherapou uuicolor,
Syucomistes butleri, Moguruda moguruda and
Oxyeleoiris liueolatus. Some are genuinely rare
throughout parts of their range such as A. macleayi
(Merrick and Schmida 1984), whilst others, such as
O. liueolatus, arc generally crypitic sfiecies that hide
amongst snags (Allen et ai. 2002) and are difficult to
catch, and thus are possibly under-represented in
our samples. However, L. uuicolor is usually one of
the most numerous species encountered in
freshwaters throughout the Pilbara and Kimberley
(Morgan et nl. 2004; Morgan and Gill 2004), yet is
very rare in Lake Kununurra. A piossible
explanation for this observation is that in L.
uuicolor the stimulus to sfiawn is apparently
associated with flood events (Llewellyn 1973;
Beumer 1979), which, due to regulation, are
minimised within the main body of Lake
Kununurra. An examination of L. uuicolor capiture
locations lends support to this explanation as the
majority of L uuicolor were cajatured from sites in
Stonewall and Spillway creeks, i.e. watercourses
that resemble naturally flowing Kimberley streams
and are subjected to flood flows during the wet-
season. Such an explanation is also supaported by
the work of Bunn and Arthington (2002) who noted
that some Australian freshwater fish species, as a
result of life history strategies evolved in direct
response to natural flow, are likely to suffer

recruitment failure when subjected to altered flow
regimes and impoundment conditions.

Although the fish fauna of the Lake Kununurra
section of the Ord River is relatively rich, it is likely
that it would have been even more diverse prior to
damming. For example, the 20 m dam wall
provides an insurmountable barrier to the
movement of catadromous species, such as
Bullshark Cnrcharhiuus leucas (Valenciennes, 1839),
Freshwater Sawfish P. microdou. Freshwater
Whipray Himautura chnophrya Monkolprasit and
Roberts, 1990, Tarpon Megalops cypriuoides
(Broussonet, 1782), Barramundi L. calcarifer and
various others, including mullets (Mugilidae), all of
which are paresent either immediately below or
within five kilometres of the dam wall (Thorburn et
nl. 2003; Doupe et al. 2005). The occurrence of four
L. calcarifer above the Diversion Dam is likely the
result of escapes from aquaculture in Lake Argyle
(Doupe and Lymbery 1999) or by their release into
Lake Kununurra by recreational anglers. The
current debate regarding the development of Lake
Kununurra as a barramundi fishery has raised the
issue of the feasibility of building a fishway
between Lake Kununurra and the lower Ord River,
either directly at the Diversion Dam, or indirectly
via Packsaddle Plain and the Dunham River (Doupe
and Bird 1999; Doupe et nl. 2005). If such a .scheme
was to proceed, then it is possible that movement of
catadromous species into the lake will occur. This
may lead to the fish community above the
Diversion Dam returning to a more natural
condition, i.e. one that more closely mirrors that
which would have been present before the dam was
constructed.

ACKNOWLEDGEMENTS

We are grateful for the financial assistance
provided by the Lake Kununurra Fish Stock
Enhancement Committee, the support of its
members and the project management provided by
Dick Pasfield. Mark and Gerry Allen, Steve
McIntosh, Andrew Storey and Dean Thorburn
provided unpublished data. Thanks to Ernest Pucci
and 'Jingles' Brown for access to the Lake Argyle
Spaillway. We also appreciate the general co-
operation and hospaitality of the East Kimberley
community, and in paarticular the general help and
hospaitality extended by Steve McIntosh and Sheree
Lethbridge.

REFERENCES

Allen, G.R. (1975). A preliminary checklist of the
freshwater fishes of the Prince Regent River Reserve
north-west Kimberley, Western Australia. Wildlife
Research Bulletin of Western Australia 3: 1-116.

Allen, G.R. and Burgess, W.E. (1990). A review of the

H.S. Gill, D.L. Morgan, R.G. Doupe, A.J. Rowland

glassfishes (Chandidae) of Australia and New
Guinea. Records of the Western Australian Museum
Supplement No. 34: 139-206.

Allen, G.R., Midgley, S.H. and Allen, M. (2002). Field
Guide to the Freshwater Fishes of Australia. Western
Australia Museum, Perth.

Beumer, J.P. (1979). Reproductive cycles of two
Australian freshwater fishes: the spangled perch,
Therapon unicolor Gunther, 1859 and the East
Queensland rainbowfish, Nematocentris splendida
Peters, 1866. Journal of Fish Biology 15: 111-134.

Bishop, K.A., Allen, S.A., Pollard, D.A. and Cook, M.G.
(2001). Ecological studies on the freshwater fishes of
the Alligator Rivers Region, Northern Territory:
autecology. Supervising Scientist Report 145, Darwin.

Bunn S.E. and Arthington, A.H. (2002). Basic principles
and ecological consequences of altered flow regimes
for aquatic biodiversity. Ecological Management 30:
493-507.

Doupe, R.G. and Bird, C. (1999). Opportunities for
enhancing the recreational fishery of Lake Kununurra
using barramundi (Lates calcarifer): A review.
Proceedings of the Royal Society of Queensland 108: 41-
48.

Doupe, R.G. and Lymbery, A.J. (1999). Escape of cultured
barramundi Lates calcarifer (Bloch) into
impoundments of the Ord River system. Western
Australia. Journal of the Royal Society of Western
Australia 82: 131-136.

Doupe, R.G., Morgan, D.L. and Gill, H.S. (2005).
Prospects for a restorative fishery enhancement of
Lake Kununurra: a high-level tropical impoundment
on the Ord River Western Australia. Pacific
Conservation Biology 11: 136-146.

Hutchins, J.B. (1977). The freshwater fish fauna of the
Drysdale River National Park North Kimberley,
Western Australia. Wildlife Research Bulletin of
Western Australia 6: 1-133.

Hutchins, J.B. (1981). Freshwater fish fauna of the
Mitchell Plateau Area, Kimberley, Western Australia.
In, Biological Sun'ey of Mitchell Plateau and Admiralty
Gulf, Kimberley, Western Australia. Western Australian
Museum, Perth.

Jaensch, R.P. (1993). Lake Kununurra. p.ll5. In S. Usback
and R. James (eds) A Directory of Important
Wetlands in Australia, 115. Australian Nature
Conservation Agency, Canberra.

Llewellyn, L.C. (1973). Spawning, development, and
temperature tolerance of the spangled perch,
Madigania unicolor (Gunther), from inland waters in
Australia. Australian Journal of Marine and Freshwater
Research 24: 73-94.

Maplnfo Corporation (1998). Maplnfo professional-user's
guide. Maplnfo Corporation, New York.

Merrick, J.R., and Schmida, G.E. (1984). Australian
Freshwater Fishes: Biology and Management. Griffin
Press, Netley.

Morgan, D.L., Gill, H.S. and Potter, l.C. (1998).
Distribution, identification and biology of freshwater
fishes in south-western Australia. Records of the
Western Australian Museum Supplement No. 56: 1-97.

Morgan, D.L., Allen, M.G., Bedford, P. and Horstman,
M. (2004). Fish fauna of the Fitzroy River in the
Kimberley region of Western Australia - including
the Bunuba, Gooniyandi, Ngarinyin and Walmajarri
Aboriginal names. Records of the Western Australian
Museum 22: 147-161.

Morgan, D.L. and Gill, H.S. (2004). Fish fauna in inland
waters of the Pilbara (Indian Ocean) Drainage
Division of Western Australia - evidence for three
subprovinces. Zootaxa 636: 1^3.

Thorburn, D.C., Peverell, S., Stevens, J.D., Last, P R. and
Rowland, A.J. (2003). Status of Freshwater and Estuarine
Elasmobranchs in Northern Australia. Report to the
Natural Heritage Trust.

Water and Rivers Commission (2000). Ord River Scientific
Report Recommendation for Estimation of Interim
Ecological Water Requirements of the Ord River. June
2000. Water and Rivers Commission, Perth.

Manuscript received 14 November 2003; accepted 23 March
2005

Records of the Western Australian Museum 23: 7-11 (2006).

Distribution of the spotted minnow {Galaxias maculatus (Jenyns, 1842))
(Teleostei: Galaxiidae) in Western Australia including range extensions

and sympatric species

David L. Morgan, Andrew Chapman, Stephen J. Beatty and Howard S. Gill

Centre for Fish and Fisheries Research, Murdoch University, South Street, Murdoch, Western Australia 6150

Abstract - Galaxias maculatus was captured from a number of rivers outside
its previously known range. In Western Australia, it was formerly only
known from rivers and lakes between the Goodga River (Two People's Bay,
30 km east of Albany) and the Dailey River (50 km east of Esperance), with
additional records from the Boat Harbour Lakes (Kent River). An intensive
survey of the inland fishes in rivers and lakes along the south coast of
Western Australia has extended its distribution east by 50 km (Thomas River),
west by approximately 40 km (Walpole River) and north by 400 km (Harvey
River). The Western Australian Museum also has a specimen from the
Canning River, a further 100 km north. Field salinity tolerance of G.
maculatus was high, with fish found alive in 81 mSem ' (-45 ppt). The
freshwater piscifauna cast of, and including, the Pallinup River is
depauperate, with G. maculatus being the only freshwater species present. All
sympatric teleosts are tolerant of salinity and, with the exception of the
introduced Gambusia holbrooki, are estuarine, including Pseudogobius olorum,
Leptatherina wallacei and Acanthopagrus butcheri.

INTRODUCTION

The Spotted Minnow, Common Galaxias or
Common Jollytail (Galaxias maculatus (Jenyns,
1842)) is a small, elongate, partially translucent,
scaleless, osmeriform fish (Figure 1). It is one of
the most widely distributed freshwater fishes in
the world (Berra et al. 1996; Waters et al. 2000),
with populations occurring in Australia, Lord
Howe Island, New Zealand, Chatham Islands,
Falkland Islands and South America. Both land-
locked (i.e. complete their life-cycle in inland
waters) and catadromous (with a marine larval
phase) populations are recognised, and this
widespread distribution has been attributed to the
presence of catadromous populations throughout
its range in eastern Australia, New Zealand and
South America, with the larvae moving into the
sea after hatching (Benzie 1968; Pollard 1971;
McDowall et al. 1975). The presence of G.
maculatus in Western Australia was first reported
by Coy (1979) and McDowall and Frankenberg
(1981), where the distribution was thought to
extend from the Goodga River in the west to the
streams around Esperance (see Figure 2). jaensch
(1992) later reported a single specimen to the west
of the Goodga River in the Boat Harbour Lakes
(Kent River) (Jaensch 1992) (Figure 2).

This paper presents new data on the distribution
of G. maculatus within W'estern Australia and
extends the range of the species in this State.

MATERIALS AND METHODS

This study focused on sampling the fish fauna in
the rivers and lakes east of, and including, the
Pallinup River on the south coast of Western
Australia, excluding estuaries. Fish were sampled
primarily using a combination of fine mesh seine
nets in 148 sites between the Pallinup River and
Poison Creek (east of the Thomas River, see Figure
2). Data presented in Morgan etal. (1998) were used
for the distribution of G. maculatus to the west of
the Pallinup River, while the unpublished reports
by Morgan and Beatty (2003a, b) were utilised for
the occurrences of G. maculatus in the Walpole and
Harvey rivers, respectively.

Species distribution maps were created using site
latitude and longitude data in the program Mapinfo
(MapInfo Corporation 1998).

Dissolved oxygen, temperature and conductivity
were measured in situ with a WTW Multiline P4
meter. Unless otherwise stated, measurements were
made immediately below the water surface.

RESULTS

Distribution of Galaxias maculatus in Western
Australia

During this study G. maculatus was captured
from the following rivers and lakes within its
previously known range: Goodga and Angove

D.L. Morgan, A. Chapman, S.J. Beatty, H.S. Gill

Figure 1

The spotted minnow {Gnlaxins nmculntiis).

rivers (Morgan et al. 1998), Pallinup, Bremer,
Gairdner, Fitzgerald, Hamersley, Phillips, Steere,
Jerdacuttup, Oldfield, Munglinup, Torradup,
Young, Port and Dalyup rivers; Yallobup, Kateup,
Coramup and Bandy creeks; Moates, Gardner,
Angove, Wheatley, Mullet and Woody lakes (Figure
2). Additionally, G. maculatus was captured from:
Lake Boolenup and the Thomas River,
approximately 100 km east of Esperance; the
Walpole River, approximately 140 km west of the
Goodga River; and in a drainage canal of Bancell
Brook, a tributary of the Harvey River,
approximately 380 km west and north from the
Goodga River by coast (Figure 2).

Sympatric species

The estuarine Swan River Goby {Psciidogobiiis
oloruni) was found in the majority of systems
between the Pallinup and Thomas rivers; the
Western Hardyhead {Lcptatheriiia waUacei) was
found from the Pallinup River to Bandy Creek;
Black Bream {Acanihopagrits butchcri), was
captured with G. maadatus in the Oldfield River
and in Wheatley and Woody lakes; and the
introduced eastern Mosquitofish {Gambusia
holbrooki) was found only in the Pallinup River
(Figure 3).

Environmental variables

Galnxins maculntiis was captured in
conductivities ranging from 0.3 to 94.4 mSem ’, and
temperatures from 12 to 30°C, however, all fish
found in conductivites greater than 88 mSem ’ were
dead, yet in all salinities <81.6 mSem ' (~45 ppt) they
were alive. Field measurements and repeated
observations revealed that when dissolved oxygen
levels were less than 1.5 mgl ’ at 30 cm depth, G.

maciiIntKS approached the surface to respire
aerially. In these instances the surface dissolved
oxygen concentration was always marginally higher
than that at 30 cm. Furthermore, dead fish were
recorded on several occasions when water
temperatures were >30°C and salinity was <2 ppt.

DISCUSSION

During this study the range of G. maculatus was
extended considerably, i.e. east to Lake Boolenup
and the Thomas River, approximately 100 km east
of Esperance; west to the Walpole River,
approximately 140 km west of the Goodga River;
and, in a drainage canal of Bancell Brook, a
tributary of the Harvey River, approximately 580
km west and north from the Goodga River by coast
(see Figure 2). The data presented here highlights
the absence of studies of the inland fish fauna on
the south coast of Western Australia.

There is a marked reduction in the number of
species of native freshwater fish from west to east
along the south coast of Western Australia. Ten
freshwater fi.sh species occur in the region with a
moderate Mediterranean climate to the west of, but
not including, the Pallinup River (Morgan ct al.
1998; Allen ct al. 2002), whereas only one
freshwater species (i.e., G. tiuKiilnliis) occurs in the
region with a dry Mediterranean climate east of the
Pallinup River (this study) (Figure 2). The ranges of
G. iiiaciilatiis and the estuarine P. oloniiii extend to
the eastern most river on the south coast, i.e. the
Thomas River (Figures 2 and 3), while the estuarine
L. -umllncei extends as far east as Bandy Creek. The
species composition of the rivers east of, and
including, the Pallinup River is a reflection of the
higher natural salinities of these lower rainfall

Galaxias maculatus in Western Australia

systems than for those rivers to the west. Thus, only
species that are salt tolerant are found in the rivers
east of the Pallinup. The feral G. holbrooki is able to
tolerate salinities up to -60 ppt (Morgan et al. 2004)
but does not currently extend east of the Pallinup
River however it is found in most of the river
systems west of the Pallinup and north to the Hutt
River, approximately 450 km north of Perth (see
Morgan et al. 1998, 2004). This species is known to
seriously impact other small south-western Western
Australia fishes (Gill et al. 1999), yet it is not clear
what impacts it has on the species in the Pallinup
River. It is probably inevitable that it will eventually
be introduced into other rivers on the south coast of
Western Australia.

During the current study, the number of G.
maculatus captured at those sites within its
previously known range was far greater than those
recorded to the west of its previously known range
(i.e., Walpole and Harvey rivers). For example,
despite extensive sampling, only four and one
individuals were captured in the Walpole and
Harvey rivers, respectively (Morgan and Beatty
2003a, b). Therefore, these individuals, particularly
in the case of those in the Harvey River, are either
indicative of populations with very low abundances

or are simply marine stragglers. It is likely that the
establishment of new, self-maintaining populations
is only achieved when adequate numbers migrate
into these systems and environmental tolerances,
particularly salinity and possibly temperature, are
not exceeded. For example, while G. maculatus is
extremely tolerant to high salinity (acute LD^pOf 45
ppt. Chessman and Williams (1975) and supported
by field observations in this study), water
temperatures experienced on the west coast of
Western Australia (e.g., Harvey and Canning rivers)
may often exceed 30°C during summer, a
temperature that Richardson et al. (1994) suggest is
lethal to at least some New Zealand populations.

Previous workers have faced uncertainty whether
Western Australian G. maculatus were land-locked
or diadromous, though on the basis of an inland
breeding record from the Pallinup River, the former
was suspected (Allen 1982). A recent detailed study
on the biology of G. maculatus in south-western
Australia that included populations in the Goodga,
Phillips, Oldfield and Jerdacuttup rivers confirms
that, over most of its range in Western Australia, G.
maculatus has a self sustaining, land-locked
breeding strategy (Chapman 2004). Further
evidence for this is the presence of fish in the

D.L. Morgan, A. Chapman, SJ. Beatty, H.S. Gill

Figure 3 The distribution of the three sympatric species captured within the main range of Galnxias mnculatus.

Jerdacuttup River, which has probably not had
contact with the sea for 6 000 years (Hodgkin and
Clark 1990). In other rivers, occasional contact with
the sea by either adult or larval fish is a possibility
as raised beach bars can either be breached by
flooding, vigorous winter flow or human
intervention (artificial excavation).

McDowall (1988) considered that diadromous
galaxiids represented the primitive or ancestral
stock and that freshwater limited populations were
a more recently derived phenomenon. Waters et al.
(2000a) proposed that the loss of the primitive
marine juvenile phase may be an important
mechanism of galaxiid speciation. It is very likely
then that on the south coast of Western Australia
the land-locked strategy has developed from
diadromous stock in response to changes to local
coastal geomorphology over recent geological
time, with most of the rivers of the region now
being only intermittently open. For example,
Culham Inlet, previously the Phillips River
estuary, where G. maculatus is confirmed to
undergo a land-locked life-history (Chapman
2004), was permanently open to the sea as recently
as 3 500 years ago but is now better described as a
coastal salt lake (Hodgkin 1997). Most of the
region's rivers are now open to the sea only after
major rainfall events.

Although the presence of diadromous
populations of G. maculatus in rivers on the south
coast was not demonstrated during the present
study, its presence in the Walpole River, which
enters the sea via a permanently open channel at
the mouth of the Nornalup Inlet (Hodgkin and
Clark 1988), suggests that that there is at least the
potential for diadromy in Western Australia.

Recent allozyme (Berra et al. 1996) and
mitochondrial (Waters et al. 2000b) studies have
demonstrated the considerable powers of dispersal,
due to a marine larval stage in some populations, of
G. maculatus. For example Berra et al. (1996)
considered that gene flow continues to occur
between Australia, New Zealand and South
America, and whilst Waters et al. (2000b) confirmed
gene flow between Australia and New Zealand,
they were less convinced that dispersal occurred
across the Pacific. It is likely that the presence of G.
maculatus in the Walpole and Harvey rivers, is a
result of larval drift via ocean currents from either
outside Western Australia or from a south coast
river. The collection of more mitochondrial data is
required to test which of these hypotheses is the
more likely. Such data would also permit
comparisons of the genetic structure of populations
within Western Australia, and between Western
Australia, eastern Australia, New Zealand and

Galaxias maculatus in Western Australia

South American populations, i.e. is there regular
gene flow between systems?

ACKNOWLEDGEMENTS
We would like to thank the Natural Heritage
Trust, Murdoch University, Department of
Fisheries, Water and Rivers Commission, Harvey
River Restoration Trust and Fishcare WA for
providing funds towards this project. Thanks to
CALM for providing permits to sample in National
Parks and A Class Reserves. Thank you to Charlotte
Morgan for help with sampling.

REFERENCES

Allen, G.R. (1982). A Field Guide to Inland Fishes of Western
Australia. Western Australian Museum, Perth.

Allen, G.R., Midgley, S.H. and Allen, M. (2002). Field
Guide to the Freshwater Fishes of Australia. Western
Australia Museum, Perth.

Benzie, V. (1968). Some ecological aspects of the
spawning behaviour and early development of the
common whitebait Galaxias maculatus attenuatus
(Jenyns). Proceedings of the New Zealand Ecological
Society 15; 31-39.

Berra, T.M., Crowley, L.E.L.M., Ivantsoff, W. and Fuerst,
P.A. (1996). Galaxias maculatus: an explanation of its
biogeography. Marine and Freshwater Research 47: 845-
849.

Chapman, A. (2004). Biology of the spotted minnow
Galaxias maculatus (Jenyns 1842) (Pisces: Galaxiidae)
on the south coast of Western Australia. MSc Tliesis.
Murdoch University, Perth, Western Australia.
Chessman, B.C. and Williams, W.D. (1975). Salinity
tolerance and osmoregulatory ability of Galaxias
maculatus (Jenyns) (Pisces, Salmoniformes,
Galaxiidae). Freshwater Biology 5: 135-140.

Coy, N. J. (1979). Freshwater Fishing in South-West
Australia. Jabiru Books, Perth.

Gill, H.S., Hambleton, S.J. and Morgan, D.L. (1999). Is
Gambusia holbrooki a major threat to the native
freshwater fishes of south-western Australia? In B.
Seret and J.-Y. Sire (eds) Proceedings 5th Indo-Pacific
Fish Conference (Noumea, 3-8 November 1997), pp.
79-87, Societe Francaise d'lchtyologie et Institut de
Recherche pour le Development, Paris.

Hodgkin, E.P. (1997). History and management of
Culham Inlet, a coastal salt lake in south-western
Australia. Journal of the Royal Society of Western
Australia 80: 239-247.

Hodgkin, E.P. and Clark, R. (1988). Estuaries and Coastal
Lagoons of south Western Australia - Nornalup and
Walpole inlets and the estuaries of the Deep and Frankland
rivers. Estuarine Studies Series No. 2. Environmental
Protection Authority, Perth.

Hodgkin, E.P. and Clark, R. (1990). Estuaries of the Shire
of Ravensthorpe and the Fitzgerald River National
Park. Estuarine Studies Series No. 7. Environmental
Protection Authority, Perth.

jaensch, R.P. (1992). Fishes in Wetlands on the South
Coast of Western Australia. Unpublished Technical
Paper. Department of Conservation and Land
Management Western Australia, Perth.

Mapinfo Corporation (1998). Mapinfo professional- user's
guide. Mapinfo Corporation, New York.

McDowall, R.M. (1988). Diadromy in Fishes: Migrations
Between Freshwater and Marine Environments. Croom
Helm, London.

McDowall, R.M. and Frankenberg, R.S. (1981). The
galaxiid fishes of Australia (Pisces: Galaxiidae).
Records of the Australian Museum 33: 443-605.

McDowall, R.M., Robertson, D.A. and Saito, R. (1975).
Occurrence of galaxiid larvae and juveniles in the sea.
New Zealand Journal of Marine and Freshwater Research
9: 1-9.

Morgan, D. and Beatty, S. (2003a). Freshwater fishes of the
Walpole River and impact of the weir to fish and lamprey
migrations. Unpublished report to Fisheries Western
Australia.

Morgan, D. and Beatty, S. (2003b). Fish and decapod fauna
of Bancell Brook (Harvey River) and the impacts of
irrigation slot boards on migrations. Unpublished report
to the Southern Peel Partnership Landcare Group.

Morgan, D.L., Gill, H.S., Maddern, M.G. and Beatty, S.J.
(2004). Distribution and impacts of introduced
freshwater fishes in Western Australia. New Zealand
Journal of Marine and Freshwater Research 38: 51 1-523.

Morgan, D.L., Gill, H.S. and Potter, I.C. (1998).
Distribution, identification and biology of freshwater
fishes in south-western Australia. Records of the
Western Australian Museum Supplement No. 56: 97 pp.

Pollard, D.A. (1971). The biology of a land-locked form of
the normally catadromous salmoniform fish Galaxias
maculatus (Jenyns) I. Life cycle and origin. Australian
Journal of Marine and Freshwater Research 22; 91-123.

Richardson, J, Boubee, J.A.T. and West, D.W. (1994).
Thermal tolerance of some native New Zealand
freshwater fish. New Zealand Journal of Marine and
Freshwater Research 28; 399M07,

Waters, J.M., Dijkstra, L.H. and Wallis, G.P. (2000b).
Biogeography of a southern hemisphere freshwater
fish: how important is marine dispersal? Molecular
Ecology 9: 1815-1821.

Waters, J.M., Lopez, J.A. and Wallis, G.P. (2000a).
Molecular phylogenetics and biogeography of
galaxiid Fishes (Osteichthyes: Galaxiidae); dispersal,
vicariance, and the position of Lepidogalaxias
salamandroides. Systematic Biology 49: 777-795.

Manuscript received 19 Januanj 2004; accepted 23 March 2005

Records of the Westerrj Australian Museum 23: 13-18 (2006).

A new species of Lechytia from eastern Australia
(Pseudoscorpiones: Lechytiidae)

Mark S. Harvey

Department of Terrestrial Invertebrates, Western Australian Museum,
Locked Bag 49, Welshpool DC, Western Australia 6986, Australia

Abstract - The first Australian representative of the chthonioid family
Lechytiidae, Lechytia libita new species, is described from Queensland where
it appears to favour tree bark microhabitats in rainforest habitats.

INTRODUCTION

Members of the family Lechytiidae have a
sporadic distribution around the world with ten
species recorded from the Americas, six species
from Africa, one species from Turkey and five
species from Asia and the Pacific region (Harvey
1991), extending as far east as Hawaii (Muchmore
2000). The sole genus Lechytia Balzan has usually
been placed in its own tribe (e.g., Beier 1932;
Chamberlin 1929; Muchmore 1975) or subfamily
(Morikawa 1960; Murthy and Ananthakrishnan
1977) within the Chthoniidae, but Harvey (1992)
proposed the Lechytiidae as a family distinct from
the remainder of the extant Chthonioidea, the
Chthoniidae and Tridenchthoniidae. Muchmore
(1975, 2000) divided the genus into two species-
group, the L. arborea species-group tor L. arborea
Muchmore, L. sini Muchmore and L. sakagamii
Morikawa, and the L. hoffi species-group for L. hoffi
Muchmore. The type species L. chthoniitormis
(Balzan 1887) was recognised as a member of the L.
arborea group by Mahnert (2001), but the remaining
species of the genus have not yet been placed. A
new species of Lechytia has been recently detected
in eastern Australia which extends the distribution
of the family for the first time into the Australian
region. This new species is the subject of this paper.

The specimens examined as part of this study are
lodged in the Queensland Museum, Brisbane (QM)
and the Western Australian Museum, Perth
(WAM). Specimens were examined using dilute
lactic acid under a compound microscope, and all
have been returned to ethanol. Terminology largely
follows Chamberlin (1931) and Harvey (1992). In
parficular, I have followed the naming conventions
applied by Harvey (1992) based upon perceived
homologies in the trichobothria. In this case, this
affects the names of fhe trichobofhria of fhe
movable finger which are here termed st, sb, b and t
(from the most basal to the most distal). In previous
systems they are termed b, sb, st and t.

SYSTEMATICS

Family Lechytiidae Chamberlin, 1929

Genus Lechytia Balzan, 1892

Lechytia Balzan 1892: 499; Harvey 1991: 186 (full
synonymy).

Type species

Roncus chthoniiformis Balzan, 1887, by original
designafion.

Remarks

Members of the family Lechytiidae share a
number of unusual features, the most peculiar of
which is the arrangement of the trichobothria where
eb and esb are situated on the dorsum of the chelal
hand. In all other chthonioids, these trichobothria
are situated at the base of the chelal finger. This
feature will serve to distinguish them from all other
chthonioids.

Species of Lechytia have been recorded from many
parts of the world (Harvey 1991), including South
and central America [L. ch&ioniiformis (Balzan, 1887),
L. chilensis Beier, 1964, L. defamarei Vitali-di Castri,
1984, L, kuscheli Beier, 1957, L. rnartiniquensisVitaM-
di Castri, 1984 and L. trinitatis Beier, 1970], North
America (L. arborea Muchmore, 1975, L. cavicola
Muchmore, 1973, L. hoffi Muchmore, 1975 and L.
sini Muchmore, 1975), Asia (L. anatolica Beier, 1965,
L. asiatica Redikorzev, 1938, L. himalayana Beier,
1974, L. indica Murthy and Ananthakrishnan, 1977
and L. madras/ca Sivaraman, 1980), Africa [L. dentata
Mahnert, 1978, L. garambica Beier, 1972, L. leleupi
Beier, 1959, L. maxima Beier, 1955b, L. natalensis
(Tullgren, 1907) and L. serrulata Beier, 1955a] and
the Pacific region (L sakagamii Morikawa, 1952). A
single species, L. tertiaria Schawaller, 1980, has been
described from Tertiary Amber deposits in the
Dominican Republic (Schawaller, 1980). The species
described below is the first to be found in Australia.

M.S. Harvey

Lechytia libita sp. nov.

Figures 1-14

Material examined

Ho/otypc

d, Windsor Tableland, Queensland,
AUSTRALIA, 16°16'S, 145°08'E, 1160 m, site 2,
pyrethrum, 27 December 1988, E. Schmidt and
ANZSES (QM SI 7220).

Parah/pes

AUSTRALIA: Queensland: 1 $, same data as
holotype (QM S67680); 1 $, Oakview State Forest,
summit, 26°10'S, 152°20'E, 600 m, pyrethrum on
trees, rainforest, 26 May 2002, G. Monteith (QM
S64848); 2 $, Peeramon Scrub, 17°19'S, 145°37'E, 750
m, pyrethrum on trees, 9 December 1995, G.
Monteith (QM S45776); 1 d, Mt Fort William, 6 km
NE. of Kalpovver, 24°39'S, 151°20'E, 700 m,
pyrethrum, rainforest, 18 September 1989, G.B.
Monteith (QM S41073); 3 d, 3 9, Bauple State
Forest, 25°52’54"S, 152°37T2"E, tree hollow, 12
January 2001, M. Shaw (WAM T62601).

Diagnosis

Trichobothria sb and b (formerly sb and st)
separated by about 1 areolar diameter; chelal teeth
reduced to a lamina for most of finger length; chela
0.419-0.435 (d), 0.462 ($) mm in length, and 4.11-
4.22 (d), 3.76 ($) times longer than broad; chelal
hand 0.200 (d), 0.210 ( $ ) mm in length.

Description

Adult Colour pale brown. Pedipalp (Figures 1-3):
trochanter 1.64-1.66 (d), 1.55 ($), femur 3.29-3.74
(d), 3.17-3.90 (9), patella 1.65-1.86 (d), 1.60-1.90
(9) and chela 4.08^.36 (d), 3.75-3.94 (9) times
longer than broad; chelal hand 1.86-196 (d), 1.71-
1.88 (9) times longer than broad; movable chelal
finger 1.24-1.35 (d), 1.20-1.35 (9) times longer than
hand; femur with anterior face flattened so that a
faint keel is present on the antero-dorsal and antero-
ventral margins; chelal fingers with approximately

7 distal teeth, remaining teeth obsolete, fused into a
lamina (Figure 1); fixed chelal finger and hand with

8 trichobothria, movable chelal finger with 4
trichobothria (Figure 1); ib, isb, eband esb on dorsum
of hand; ib and isb situated basally; eb and esb
situated medially; b situated slightly closer to sb
than to h b and sb only about one areolar diameter
apart; xs situated slightly distal to et near tip of
fixed finger, each hair shorter than those of other
trichobothria; venom apparatus absent. Chelicera
(Figures 6, 7) with 5 setae on hand and 1 medial
seta on movable finger; fixed finger with 3 small
teeth, the distal-most tooth largest; movable finger
much shorter than hand, with 2 small teeth;
flagellum (Figure 8) consisting of 7 blades, the

subdistal blade strongly recumbent, others straight;
galea of S absent, that of 9 a short rounded nubbin.
Carapace (Figure 4) with 2 small corneate eyes;
anterior margin finely denticulate (Figure 5); with
18 setae arranged 6: 4: 4: 2: 2; the pre-ocular seta
about 50% length of other setae in anterior row;
with 4 pairs of lyrifissures, one pair situated antero-
medially, the second pair situated interno-lateral to
the eyes, the third pair situated slightly interior to
the sole pair of setae of the intermediate row, and
the fourth pair situated exterior to the sole pair of
setae of the posterior row. Fergites and sternites
undivided; tergal chaetotaxy 6, 6: 5-6: 5-6: 6: 6: 6:
6: 6: 6: 4: 1T2T1: 0; 9, 6: 6: 6: 6: 6: 6: 6: 6: 6: 4: 1T2T1:
0; sternal chaetotaxy 6, 10-11: (3)32-33(3): (3)8-
10[4+4|(3): 10: 7-8: 8: 8: 8: 2TT2: -: 2; 9, 6: (3)6(3):
(3)8(3): 10: 8: 8: 8: 8: 8: -: 2. All setae bordering male
sternite III bifurcate or trifurcate (Figure 13).
Genitalia of male not studied in detail; of female
weakly sclerotized with U-shaped frame. Pleural
membrane smoothly plicate. Coxal chaetotaxy
(Figure 10): d, 2+3: 4:4-5: 4-7: 7; 9, 2+3: 4: 5: 6-7: 7;
manducatory process with 2 distal setae, about
equal in length (Figure 10), the distal seta terminally
bifurcate (Figure 11), plus an additional 3 setae
situated close to trochanteral foramen; coxal spines
and intercoxal tubercle absent (Figure 10); coxa I
with small, triangular apical projection with single
seta situated at base (Figure 12); other setae on coxa
I situated near trochanteral foramen. Legs robust,
femur+patella IV 2.03 (d), 1.96 (9) times longer
than deep; heterotarsate; tarsi with two elongate
gland openings along dorsal .surface (see Muchmore
2000) each with crenulate margins (Figure 9);
arolium slightly shorter than claws (Figure 9), claws
simple.

Dimensions (mm)

Males: Holotype (QM SI 7220) followed by other
males (where applicable): Body length 1.17 (0.88-
0.98). Pedipalps: trochanter 0.123/0.074, femur
0.288/0.077 (0.265-0.28/0.073-0.082), patella 0.160/
0.092 (0.138-0.160/0.080-0.097), chela 0.435/0.103
(0.395-0.419/0.092-0.102), hand length 0.200 (0.175-
0.200), movable finger length 0.256 (0.235-0.260).
Chelicera 0.211/0.115, movable finger length 0.112.
Carapace 0.315/0.307 (0.29-0.31/0.253-0.286); eye
diameter 0.031). Leg 1: femur 0.178/0.046, patella
0.090/0.053, tibia 0.109/0.035, tarsus 0.186/0.028. Leg
IV: femur + patella 0.320/0.157, tibia 0.208/0.093,
metatarsus 0.112/0.051, tarsus 0.186/0.029.

Females: Paratype (QM S67680) followed by other
females (where applicable): Body length 1.22 (1.06-
1.20). Pedipalps: trochanter 0.129/0.083, femur
0.304/0.096 (0.285-0.320/0.085-0.090), patella 0.166/
0.104 (0.159/0.091-0.100), chela 0.462/0.123 (0.415-
0.455/0.110-0.120), hand length 0.210 (0.198-0.225),
movable finger length 0.262 (0.240-0.275). Chelicera
0.223/0.122, movable finger length 0.127. Carapace

A new Australian Lechytia

Figures 1-9 Lechytia libita, sp. nov., holotype male unless stated otherwise; 1, left chela; 2, detail of distal portion of
fingers; 3, right pedipalp; 4, carapace; 5, epistome; 6, chelicera; 7, movable cheliceral finger (paratype
female); 8, flagellum; 9, right tarsus 1 (paratype female). Scale lines; 0.05 mm (Figures 6, 7); 0.1 mm
(Figures I, 4, 5, 9); 0.2 mm (Figure 3).

M.S. Harvey

0.352/0.328 (0.290-0.340/0.285—0.345); eye diameter
0.023. Leg I: femur 0.188/0.051, patella 0.103/0.458,
tibia 0.109/0.036, tarsus 0.186/0.026. Leg IV; femur +
patella 0.315/0.161, tibia 0.224/0.070, metatarsus
0.122/0.056, tarsus 0.198/0.269.

Remarks

Lechytia libita belongs to the ‘L. arborea' species-
group as defined by Muchmore (1975), as the distal
seta of the pedipalpal coxa is bifurcate (Figure 11),
the chelal teeth are strongly reduced (Figure 1),
tergite XI has a chaetotaxy of 1T2T1, the male galea
is much reduced in comparison with the female
(Figures 6, 7) and the tarsal gland openings are
enlarged and possess crenulate margins (Figure 9)
(Judson 1992; Muchmore 2000). It is most similar to
those species of Lechytia in which trichobothria sb
and b are slightly separated from each other, usually

by about one areolar diameter, but it differs from
all of these species by its smaller size: L. chilensis
from Chile [e.g., chelal hand 0.24 mm ($)], L.
kuscheh from Juan Fernandez Island [e.g. chelal
hand 0.25-0.26 mm (d), 0.285 mm ($)], L. serrulata
from Zaire [e.g., chelal hand 0.24 mm (d)[, L.
maxima from Kenya [e.g., chelal hand 0.28 mm (d,
$)], and L. cavicola from Mexico [e.g., chela length
0.51-0.52 mm (d), 0.51 mm (?), chelal hand 0.235-
0.25 mm (d), 0.24 mm (9)]. In addition, the chelal
teeth are reduced to a thin lamina in L. libita, L.
chilensis, L. kuscheh, and L. cavicola, but are well-
defined in L. serrulata and L. maxima. Lechytia
libita also differs from some of the lamina-bearing
species by the relative dimensions of the chela: L.
chilensis has a chela which is 4.8 ($) times longer
than broad, while the chela of L. kuscheh is 4.8 (d),
4.3 ( $ ) times longer than broad; L. libita has a chela

A new Australian Lechytia

that is 4.11-4.22 (S), 3.76 ($) times longer than
broad. Geographically, L. libita is closest to L.
sakagamii from the Pacific region and L. asiatica
Redikorzev from Vietnam. It differs from L.
sakagamii by the relative positions of trichobothria
sb and b, which are separated by about one areolar
diameter in the Australian species and by about half
an areolar diameter in L. sakagamii (Beier 1957:
figure 3c; Morikawa 1960: plate 7, figure 10;
Muchmore 2000: figure 5b). Lechytia libita also
differs from L. asiatica in the positions of sb and b,
which are contiguous in L. asiatica [Dr Mark
Judson, in litt., has kindly examined the two
syntypes (1 d, 1 $) of L. asiatica which are lodged
in Museum national d'Histoire Naturelle, Paris].

Distribution

Lechytia libita is the first member of the genus to
be found in Australia and has been recorded from
rainforest habitats in eastern Queensland (Figure
14). All available records suggest that L. libita is a
corticolous species as all known specimens were
taken from tree habitats. Whilst a corticolous habitat
is generally unusual for chthonioid pseudo-
scorpions - they are more abundant in leaf litter
and soil - a corticolous environment is not
unknown amongst lechytiids. Lechytia arborea has
been collected from "base of leaf of cabbage palm"
and "beating foliage" from south-eastern U.S.A.,
whilst some nymphs that were tentatively assigned
to the species were collected from beneath the "bark
of a pine tree" at two different locations in Florida

(Muchmore 1975). Some of the type material of L.
anatolica from Turkey was taken under bark of a
rotting pine (Beier 1965), and Lechytia delamarei was
collected under bark on Guadeloupe (Vitali-di
Castri 1984). Lechytia himalayana has been taken
from the bark of Rhododendron and Abies
(Schawaller 1987).

Etymology

The specific epithet is a Latin noun alluding to
the pleasure of finding the first species of Lechytia
in Australia {libitus, Latin, pleasing, agreeable).

ACKNOWLEDGEMENTS
I am very grateful to Robert Raven and Owen
Seeman who kindly provided access to the
collections of the Queensland Museum, Matthew
Shaw for the donation of the Bauple State Forest
specimens, to Mark Judson (Museum national
d'Histoire naturelle, Paris) for providing
information regarding the syntypes of L. asiatica,
and to Mark Judson and Volker Mahnert for their
constructive comments on a draft of the manuscript.

REFERENCES

Balzan, L. (1892). Voyage de M. E. Simon au Venezuela
(Decembre 1887 - Avril 1888). Arachnides. Chernetes
(Pseudoscorpiones). Annales dela Sodete Entomologique
de France 60: 497-552.

Figure 14 Distribution of Lechytia libita, sp. nov.

M.S. Harvey

Beier, M. (1932). Pseudoscorpionidea I. Subord.
Chthoniinea et Neobisiinea. Tierreich 57: i-xx, 1-258.

Beier, M. (1957). Pseudoscorpionida. Insects of
Micronesia 3: 1-64.

Beier, M. (1965). Anadolu'nun Pseudoscorpion faunasi.
Die Pseudoscorpioniden-Fauna Anatoliens. Istanbul
Universitesi Fen Fakiiltesi Mecmuasi 29B: 81-105.

Chamberlin, J.C. (1929). A synoptic classification of the
false scorpions or chela-spinners, with a report on a
cosmopolitan collection of the same. Part 1. The
Heterosphyronida (Chthoniidae) (Arachnida-
Chelonethida). Annals and Magazine of Natural History
(10) 4: 50-80.

Chamberlin, J.C. (1931). The arachnid order
Chelonethida. Stanford University Publications,
Biological Sdences 7(1): 1-284.

Harvey, M.S. (1991). Catalogue of the Pseudoscorpionida.
Manchester University Press: Manchester.

Harvey, M.S. (1992). The phylogeny and systematics of
the Pseudoscorpionida (Chelicerata: Arachnida).
Invertebrate Taxonomy 6: 1373-1435.

Judson, M.L.l. (1992). African Chelonethi. Studies on the
systematics, biogeography and natural history of
African pseudoscorpions (Arachnida), Ph.D. thesis.
Department of Pure and Applied Biology, University
of Leeds: Leeds.

Mahnert, V. (2001). Cave-dwelling pseudoscorpions
(Arachnida, Pseudoscorpiones) from Brazil. Revue
Suisse de Zoologie 108: 95-148.

Morikawa, K. (1960). Systematic studies of Japanese
pseudoscorpions. Memoirs of Ehime University (2B) 4:
85-172.

Muchmore, W.B. (1975). The genus Lechytia in the
United States (Pseudoscorpionida, Chthoniidae).
Southwestern Naturalist 20: 13-27.

Muchmore, W.B. (2000). The Pseudoscorpionida of
Hawaii Part 1. Introduction and Chthonioidea.
Proceedings of the Entomological Society of Hawaii 34:
147-162.

Murthy, V.A. and Ananthakrishnan, T.N. (1977). Indian
Chelonethi. Oriental Insects Monograph 4: 1-210.
Schawaller, W. (1980). Fossile Chthoniidae in
Dominikanischem Bernstein, mit phylogenetischen
Anmerkungen (Stuttgarter Bernsteinsammlung:
Arachnida, Pseudoscorpionidea). Stuttgarter Beitrage
zur Naturkimde (B) 63: 1-19.

Schawaller, W. (1987). Ncue Pseudoskorpion-Funde aus
dem Nepal-Himalaya, II (Arachnida: Pseudo-
scorpiones). Senckenbergiana Biologica 68: 199-221.
Vitali-di Castri, V. (1984). Chthoniidae et Cheiridiidae
(Pseudoscorpionida, Arachnida) des Petites Antilles.
Bulletin du Museum National d'Histoire Naturelle, Paris
(4) 5: 1059-1078.

Manuscript received 11 November 2004; accepted 26 May
2005

Records of the Western Australian Museum 23: 19-41 (2006).

Mulka's Cave Aboriginal rock art site: its context and content

R. G. Gunn

329 Mt Dryden Road,

Lake Lonsdale, Victoria 3380, Australia
email: gunnb@netconnect.com.au

Abstract - The Mulka's Cave Aboriginal site, within "The Humps" Nature
Reserve near Hyden, Western Australia, was recorded in detail prior to an
overall tourist-orientated development of the Reserve. The site features 452
motifs, an extremely high number for the region where most sites have fewer
than 30 motifs. The artwork is dominated by 275 handstencils, with 40
sprayed areas, 23 handprints, 23 paintings, 3 drawings and a single object
stencil produced with a wide range of colours. The high diversity of art
attributes is unusual in a region where the rock art is dominated by red
handstencils. Tire site appears to have been of considerable importance to the
Noongar people in the past and remains significant to them today. Its
significance to the broader Western Australian community is evidenced by
the high number of tourists it currently receives.

INTRODUCTION

The Mulka's Cave Aboriginal site is a decorated
rock shelter with an occupation deposit within "The
Humps" Nature Reserve near Hyden, Western
Australia. The site's management is to be upgraded
as part of an overall tourist-orientated development
of the Reserve. In line with this upgrade, the site's
rock art has been recorded in detail for research,
interpretation and conservation needs. The name
"Mulka's Cave" derives from the associated
Aboriginal myth (see below). The site is currently
promoted to the general public and is estimated to
receive around 80,000 visitors per year, a very high
level of visitation for an Australian rock art site
(excluding the exceptionally popular tourist sites at
Uluru and Kakadu; Gale and Jacobs 1987). The
recording was undertaken as part of the Wheatbelt
Heritage Management Strategy of the Department
of Indigenous Affairs. The strategy aims to develop
a procedure of consultation with local Noongar
people with respect to the management of
Aboriginal sites within the Noongar Native Title
area.

THE SITE AND ITS CONTEXT

Mulka's Cave is 18 km north-east of Hyden
(Figure 1) and 300 km inland from Perth. It lies 15
km north of Wave Rock, the region's, and one of the
nation's, premier tourist attractions. While Mulka's
Cave is a promoted tourist attraction, it is largely
"riding on the back" of the very successful Wave
Rock tourism development.

Mulka's Cave has formed beneath a large boulder
at the base of an outcrop of Archaean granite now

known as The Humps (Figures 1 and 2). The
outcrop is some 2 km x 1.5 km in area and rises to
80 m above the surrounding plain (Twidale and
Bourne 2003). It is the highest of several granite
inselbergs in the region that form notable
landmarks in an otherwise flat landscape. These
inselberg were important for the Aboriginal people
as both navigational guides and water reserves
(Bindon 1997). The rock consists of "variably
textured, medium and coarse grained seriate granite
and adamellite" (Geological Survey of Western
Australia 2003). Small, deeply weathered areas of
the outcrop on the eastern side form linear ridges
and tors, and it is beneath one of these tors that
Mulka's Cave has formed.

The mean annual rainfall for the region is 345 mm
(Bureau of Meteorology web site for Hyden).
Rainfall tends to be heavier in winter (June 52 mm)
and lightest in December (mean 14 mm), but can
occur throughout the year. The temperature
extremes range from 48°C in summer to 5°C in
winter. The interior of Mulka's Cave does not
receive any direct sunlight and hence it is not
subject to the extreme temperature changes that are
a common factor in rock deterioration (Twidale
1980: 134). Run-off water flows through the cave
making the interior floor damp and slippery for
prolonged periods after rain. This run-off has
doubtless also affected the integrity of the
archaeological deposits at the main entrance. Water,
both flowing through the shelter and percolating
through the rock matrix, is probably the principal
agent for the massive exfoliation occurring within
the shelter that threatens the existing artwork (cf.
Twidale 1980: 142; Thorn 2001).

R.G. Gunn

Figure 1 Mulka's Cave, The Humps, near Hydon. Main entrance arrowed.

Figure 2 Interior of Mulka's Cave looking down to the main entrance from the main chamber.

The vegetation surrounding The Humps is
essentially a mallee scrub with localised stands of
salmon gums {Eucalyptus salmonophloia)
(Bowdler et al. 1989), with pockets of relic

vegetation on The Humps themselves (Twidale
and Bourne 2003). Yams and wattle seeds are the
only plant foods whose use by Indigenous
hunter-gatherers is documented in the region

Mulka's Cave rock art site

(Bowdler et al. 1989). Early explorers (e.g.. Roe
1852) found water and feed for their horses scarce
throughout the region around Hyden, and
sandalwood cutters relied upon the water they
found in "granite rockholes" (Bowdler et al.
1989). As such they would have been of
paramount importance to the Aborigines utilising
the area (Bindon 1997).

The few early references to Aborigines around
Hyden are scanty at best and have been
interpreted as suggesting that the area was little
visited because of its poor natural resources
(Bowdler et al. 1989). However, the number and
distribution of Aboriginal sites associated with
gnammas across the region (Webb and Gunn 2004)
suggests that the area was regularly visited
following suitable rains.

Tindale (1974) placed the area around Hyden
within the tribal area of the Njaki Njaki people who
shared "overall cultural similarities" with the
people of south-west Western Australia (Bowdler et
al. 1989). Consequently, they were part of the south-
western cultural group known as the Goreng
(Tindale 1974) or, more usually, Nyungar or
Noongar (Berndt 1980). The Njaki Njaki were also
known by the names Kokor and Kikkar (Tindale
1974). Today the site is within the area of the South-
West Aboriginal Land and Sea Council (SWALSC)
which represent the various Noongar groups of the
South-West.

Methods

The Western Australian Department of
Indigenous Affairs (DIA) contracted the author to
produce a detailed recording of the site. The
Department undertook initial consultation for the
project and received permission and approval from
the local and regional Noongar representatives for
it to proceed. DIA subsequently organised an on-
site meeting between the consultant and the local
Noongar representatives when the recording aims
and methods were fully explained. The Department
also gave their permission for the publication of this
paper.

A plan of the site was prepared using handheld
GPS, with finer detail mapped by tape and
compass. Arbitrary key points were placed on the
art panels and mapped onto the plan. These were
later used to integrate the photographic record into
a photomosaic.

The shelter interior, which is poorly lit on even
the brightest days, was illuminated using generator-
powered arc lights. The artwork was recorded by
freehand sketching and photography, following
published procedures (Gunn 1995a, 1995b) and a
photomosaic subsequently prepared from which
detailed drawings were made. Using the freehand
sketches and the photographs, each image and area
of pigment was allocated a discrete motif number.

Following Maynard (1977, 1979), a list detailing the
attributes of technique, colour, form, type,
condition and size for each motif was then
compiled (Gunn 2004). Colour was generalised to
orange-red, purple-red, brown-red, orange, yellow,
cream and white. For hand stencils and prints, the
knuckle size was measured on-site as an indicator
of the size/age of the hand's owner. The maximum
length of other distinct motifs was measured to the
nearest half centimetre.

The Mulka Dreaming

Two versions of the story of Mulka have been
published. The earliest is very brief:

The outcast Mulka, driven from the tribe because it
was feared that his crossed eyes would bring a curse
to those he looked upon, took refuge in the cave at
the Humps (Meeking 1979).

The second comes from a brochure of the
Department of Aboriginal Sites (DAS 1989):

Mulka was the illegitimate son of a woman who fell
in love with a man to whom marriage was
forbidden. As a result, Mulka was bom with
crossed eyes. Even though he grew up to be an
outstandingly strong man of colossal height, his
crossed eyes prevented him from aiming a spear
accurately and becoming a successful hunter. Out
of frustration, Mulka turned to catching and eating
human children, and he became the terror of the
district. He lived in Mulka's Cave where the
impressions of his hands can still be seen much
higher than those of an ordinary man. His mother
became increasingly concerned with Mulka and
when she scolded him for his anti-sodal behaviour,
he turned on his own mother and killed her. This
disgraced him even more and he fled the cave,
heading south. Aboriginal people were outraged by
Mulka's behaviour and set out to track down the
man who had flouted all the rules. They finally
caught him near Dumbleyung, 156 km south-west
of Hyden, where they speared him. Because he did
not deserve a proper ritual burial, they left his body
for the ants; a grim wanting to those who break the
law.

Bowdler et al. (1989) consider this myth to be
atypical of Aboriginal myths and consequently they
dispute its origin. The author, however, has
recorded similar myths in Central and Northern
Australia, and considers this to be quite typical of
local (non-travelling) myths, although even the
longer version given here is likely to be a very much
abridged rendition of a longer and more complex
story. A myth with close similarities to the Mulka
myth was also recorded by Meggitt (1962: 261-262)
from the Warlpiri of Central Australia. [The
Warlpiri are a Western Desert language group who
share cultural links with the people to the east of
the Noongar (see Gould 1969a; O'Connor et al. 1998,
for maps)].

R.G. Gunn

Mulka's Cave

Mulka's Cave is the only known rock art site
within The Humps Reserve. In common with most
Aboriginal rock art sites, Mulka's Cave is part of a
larger site complex encompassing a range of other
site types (cf. Gunn 1997). Here the associated sites
consist of an adjacent open artefact scatter, a "lizard
trap", five gnammas and a rockhole.

The Cave site lies at the base of the eastern slopes
of the steep sided outcrop that forms The Humps
hill (Figures 1, 3). The lizard trap, rockhole and four
of the gnammas occur on flat granite pavements
300 m to the north of the cave. The fifth gnamma

lies on the northern mid-slope of the hill. This
distribution of sites suggests that the principle area
of Aboriginal interesf was to the eastern side of the
Reserve. While no further art sites or water reserves
have been located on The Humps, no survey for
stone artefacts has been undertaken to the west of
the hill.

The first published account of the site was by R.
B. Day (1951; quoted at length in Serventy 1952).
Day mentioned that the site contained "hundreds
of hand-marks done in red ochre. ..and the
remnants of a large native drawing" but he did not
provide any illustrations. Davidson had previouslv

Mulka's Cave rock art site

recorded the site in 1938-1939 but he did not
publish his findings until the following year
(Davidson 1952: 113). He identified seven paintings,
92 stencilled hands, and 37 printed or drawn hands.

These are in red, white or yellow pigments.
Davidson's study is the most detailed to date and is
referred to further below. Serventy (1952), Wolfe-
Okongwu (1978) and Flood (1990) also briefly

Figure 4 Mulka's Cave plan and section.

R.G. Gunn

discuss the site but incorporate few new details into
their descriptions. Internal reports of the then
Department of Aboriginal Sites (DAS; now DIA)
offer further descriptions of the site including note
of "a recent-looking painting in orange (?ochre)
which is probably not Aboriginal" that has not been
mentioned by previous researchers. This large and
imposing work of graffiti was almost certainly
produced between 1952 and 1971 and is excluded

from the following description. A number of rock
art conservation and management studies have
been undertaken at Mulka's Cave (Randolph 1973;
Clarke 1976; Haydock and Rodda 1986; Gunn
2003a) but these need not be discussed here.

The Shelter

Mulka's Cave is a tafone cavern underneath a
large granite tor. A tafone {pi. tafoni) is a hollow or

Figure 5 Front chamber artwork.

Mulka's Cave rock art site

cavern that develops on the underside of a boulder
(Twidale and Bourne 1975: 481), while a tor as used
here, is an Australian term for a "boulder of
weathering" (Twidale 1980: 95-105). The tor here is
25 m long by 18 m wide and 5 m high and irregular
in plan and section (Figure 4). The top of the tor
slopes steeply (20°) downhill such that rainwater
run-off is directed down over the shelter mouth,
contributing to the erosion of the deposits in front
of the shelter. The cave is a steeply inclined tunnel
beneath the tor, with low entrances on both the
downslope and upslope. The ground in front of the
shelter slopes very gently down to the east, forming
a level area that radiates for about 20 m out from
the downslope (and main) entrance. Behind the
shelter. The Humps hill rises rapidly in a series of
vegetated rock benches. A number of large water-
retaining pans are at the southern end of one of
these benches, some 50 m south of, and 20 m above,
the shelter.

Mulka's Cave is 15 m deep by 9 m wide, with a

maximum height of 2.5 m above the floor. The
interior is dark due to the low entrances and, while
light penetrates through both entries, artificial light
is required to see most of the artwork. For
discussion and description, the cave is divided into
five units on the basis of its interior morphology:
main entrance, front chamber, main chamber, upper
shelter and side tunnel (Figure 4).

The "front chamber" is immediately inside the
main entrance (Figures 2, 4). It is 6 m wide and
deep, with a domed ceiling up to 2 m high. Artwork
covers most of the ceiling, with the exception of the
actively eroding southern wall. The rear of the
chamber is taken up with a large sloping block of
granite. The chamber is divided from the main
chamber by this block and a corresponding drop in
the ceiling height. The floor consists of a flat deposit
of grey sand in a narrow strip that runs diagonally
back and into the side tunnel chamber. This was
found to be rich in archaeological material (Bowdler
etal. 1989; discussed below).

R.G. Gunn

The "main chamber" is 6 m wide by 9 m deep,
with a domed and undulating ceiling that rises to a
maximum height of 2.5 m above the floor. The floor
slopes steeply (c.20°) towards the entrance and at
its lowest point is 1 .5 m above the floor of the front
chamber. Boulders cover most of the floor but there
is a small area of flat soil at the lowest point. These
boulders are convenient to sit on but there is little
flat area suitable for sleeping. Towards the back
entrance, the floor is covered by smaller, loose rocks
(c.0.5 m diameter). Artwork covers most areas of
the ceiling but with a denser concentration on the
downhill side (facing uphill and away from the
entrance; figure 2). This area of ceiling is the most
obvious to decorate for people sitting in the cave.

The "upper shelter" is on the uphill side of the
tor. Although it is 10 m wide, 5 m deep and 2 m
high, the floor is both steep and rock strewn,
making it unsuitable for occupation. A small panel
of artwork occurs directly above this entry.

The "side tunnel" is about 12 m long. Although
1.5 m wide and high at its mouth, it decreases
rapidly in width such that despite its flat soil floor,
only the first six meters is comfortable to sit in.
There is no artwork within the tunnel, whose
ceiling and walls are actively deteriorating.

The Artwork

A total of 452 motifs were recorded from three
alcoves within the cave (Figures 5 and 6). [The full
list of motifs and their attributes is given in Gunn
(2004).) The front chamber has 131 motifs, the main
chamber 318, and the rear shelter three. The densest
area is on the downhill face of the main chamber
ceiling. The northern sidewall is most easily viewed

without artificial lighting. The quantity of artwork
on the eastern face suggests that the Aboriginal
artists would have worked with artificial lighting.

The absence of motifs within the side tunnel and
the south wall of the front chamber is probably
attributable to moisture and consequent rock
deterioration. The small number of motifs within
the rear shelter most likely reflects its unsuitability
for occupation as well as its water-washed wail
surface, while the total lack of extant artwork on the
outside wall at the main entrance can be attributed
to its water-washed surface.

A lack of artwork on the dry wall of an adjacent
shelter to the immediate north of the cave, however,
is less readily explained. This small shelter has a
flat floor with a light scatter of stone artefacts,
indicating that it was used in the past, presumably
by the same people who decorated Mulka's Cave.

Motif types

Eighteen distinct motif types were recorded in
Mulka's Cave (Table 1). Visually and numerically
the artwork is dominated by hands (69%); these are
primarily stencils but include a few handprints
(Figures 5 and 6). Left hands outnumber right hands
108 to 72 (a ratio of 1.5:1). Similar ratios occur
throughout Australia and are thought to reflect the
proportion of left and right handed people in the
population, as right-handed people tend to stencil
their left hands and vice versa (Gunn in prep).
While most of the hand stencils and prints are of
the standard type with splayed fingers, a small
number of the stencils are variants (cf. Walsh 1979),
with the second (middle) and third (ring) fingers
held together (Figure 7). Among the handprints, a

Table 1 Motif types by technique.

Motif Type

Spray

Paint

Draw

Total

Unknown object

Spray area

Variant hand

Right hand

Unknown hand

113

Left hand

103

108

Band

Other apex element

Arc set

Bar set

Line single

Other element

Arc pair

V shape

Line pair

Simple design kla

Simple design klb

Simple design k3

Total

314

375

Fragments

100

Mulka's Cave rock art site

Figure 7 Group of red "variant" handstencils.

Figure 8 Group of red "variant" handprints, possibly
imitating a trail of macropod tracks.

few have all their fingers closed together but with
the thumb splayed, giving the shape of a kangaroo
or other macropod track (Figure 8; and see
McDonald 1983), but whether this resemblance was
intentional or fortuitous is unknown.

Solid areas of pigment sprayed onto the wall (in
the manner of a stencil but without stencilled object)
are a feature of the artwork. Several are unusually
large, up to 40 cm, and involve the use of a
considerable amount of pigment. No ochre source is
known in the immediate vicinity and hence it is
assumed that the ochres used here would have to
have been brought in from some distance away.
Consequently the use of large quantities of ochre
implies some particular value in the artwork
produced even though no recognisable image is
apparent. It is probable then that the importance was
in the application of pigment to the wall rather than
the production of any particular image. This would
then fall into one of Ross's "associated rock art
traditions" that include other non-image works such
as abraded areas, scribble areas, etc. (Ross 2003; 89).

Two large, linear, simple designs in the front
chamber are the most visually impressive motifs in
the cave (Figure 9). Other motifs consist of small
"geometric element" motifs: bars, lines and arcs, in
either sets or other various combinations (Figure
10). This graphically simple repertoire is consistent
with the pattern of rock art throughout much of
southern Australia (Wolfe-Okongwu 1978; Gunn
1981, 1983, 2002).

Small numbers of large linear designs within a
suite of numerous handstencils is now becoming
apparent throughout south-west Western Australia
(Hallam 1972; Gunn and Webb 2003). White all of
the design motifs are similar in structure (linear),
they are unique in their individual designs.

Techniques

Four art production techniques were recorded in
Mulka's Cave (Table 2); spraying (86%), printing
(8%), painting (5%) and drawing (<1%).

Table 2 Motif colour by technique.

Colour Paint

Spray

Draw

Total

Monochrome

Red (brown-red)

Yellow

Red (purple-red)

199

225

White

106

111

Red (orange-red)

Orange

Cream

Bichrome

White+red

Yellow+red

White+yellow

Total

389

452

R.G. Gunn

Figure 9 The major paintings in the front chamber of Mulka's Cave. Motif 68 is in purple-red and overlies motif 67 in
an orange-red. Motif 67 is 1270 mm and motif 68 is 830 mm. The colour scale is 10 cm.

Of the sprayed motifs, 38 are solid areas of
sprayed pigment, while the others are sprayed over
an object, most commonly a hand, that was then
removed leaving a negative stencil image. Although

most of the stencils were produced on an area of
bare rock wall, 35 were produced by placing the
hand on a pre-pigmented surface (Figure 11).
Although artworks produced by this technique are
known elsewhere in Australia in small numbers, the
number of examples here make the technique a
feature of the Mulka's Cave art. In some instances
this appears to be incidental overlap, but in a
number of cases the placement is central to the
pigment area and therefore probably deliberate.
While it is not possible to determine whether or not
the original pigment was applied immediately
before the stencilling (and hence as part of the
stencilled motif), or whether a previously sprayed
area was deliberately chosen, a number of examples
suggest the former. In these cases, red pigment was
used as the underlying colour and white as the
stencilling colour. Due to some unusual character
of the pigment or the wall surface itself, the white
pigment has dropped off the wall taking with it the
underlying red pigment. What remains is a positive
red image, of the negative stencilled hand, within a
field devoid of pigment. Superficially these
resemble handprints, yet the character of the hand
is distinctly in the form of a stencil. Other examples
in the cave demonstrate the intermediary stages of
this deterioration process and clearly show the
white overlying the earlier red layer.

In contrast to stencils which produce a negative
image, printing produces a positive image by
imprinting a pigment-laden object, again usually a
hand, directly onto the rock surface. Here only
handprints occur, both standard splayed hands and

Mulka's Cave rock art site

Figure 11 White handstendls over red pigment areas in Mulka's Cave. The white pigment and the underlying red
have eroded away leaving a red positive hand on a bare rock surface. The motifs in the upper two photo-
graphs retain fragments of white pigment around the rod hand.

R.G. Gunn

some "v'ariant" hands with non-evenly splayed
fingers (Figure 8).

The paintings, due to the lack of any brush
striations, appear to have been executed with a
daubed finger rather than any form of brush. The
drawings were produced using a dry ochre nodule,
in the manner of a crayon.

The motifs tend to form two groups (Table 1):

• Paintings and drawings : Geometric elements

and simple designs

• Sprays and prints : Hands, objects and

spray areas

This is a common division in pictogrammic art
(e.g., the "art" and "associated art traditions"
divisions of Ross 2003) where "freeform" images
(where the shape of the object is determined by the
artist's mental image) are distinct from preform"
images (where the shape of the image is derived
from an actual object or by the technique itself)(see
below).

Colour

Seven distinct colours were recorded (Table 2):
purple-red (50%), white (25%), orange-red (10%),
and brown-red, yellow, cream, and orange in
smaller numbers. The three reds appear to be
distinctly different hues, but analysis may show
them to be the same colour in various stages of
preservation. Purple-red is the most common colour
among the sprays, prints and paintings.

Three bichrome colour combinations were also
recorded: white+red, white+yellow, and

yellow-i-red. These are all handstencils produced
over an underlying colour. Cream is restricted to
printing. However, as the number of motifs
involved in these two groups is small, the

distinctions are not considered significant at this
stage. On the other hand, the lack of yellow
handprints and the very small number of white
prints (two, cf 106 white stencils) appears culturally
motivated, as the pigments were obviously readily
available.

A matrix of colour and motif types shows that
whereas hands and sprayed areas are made with a
wide range of colours, geometric elements and
designs use a limited range basically restricted to
the three reds (Table 3). Contrary to Davidson
(1952), who found that left and right hands occur in
equal numbers with all colours, the author found
that when all hands are included, left hands
predominate over right in all colours in a ratio
ranging from 1.5:1 to 2:1.

Motif forms

Six motif form types were identified (Table 4):
bandstencil (73%), solid (11%), handprint (10%),
linear (6%), and single examples of an object stencil
and an outline-t-infill painted motif. Motif form can
be subdivided into preforms, where the form is
dictated by the technique used, for example
handstencils, and freeforms, where the form is
independent of the technique, such as paintings.
The preforms recorded here account for 93% of the
motifs, and comprise hand and object stencils,
handprints and sprayed areas. The freeform motifs
are dominated by linear types, the most complex of
which is a simple outline+infill design. Apart from
the solid sprayed areas, and a single painted "hand"
motif, there is a distinct separation of the motif
types between freeforms and preforms (Table 4).

The single yellow object stencil is a narrow shaft
about a metre long and 25 mm thick. Its simple

Table 3 Motif type by colour.

Motif Type

w+r

y+r

w+y

Total

V shape

Line pair

Arc pair

Other element

Line set

Simple design klb

Simple design k3

Arc set

Band

Other apex element

Bar set

Left hand

108

Unknown hand

113

Right hand

Spray area

Variant hand

Simple design kla

Unknown

Total

163

108

375

Fragments

Mulka's Cave rock art site

Table 4 Motif typos by form.

Motif type

Preforms

Object

stencil

Hand

stencil

Hand

Freeforms

Linear

Solid

Outline

+infill

Total

Unknown

Variant hand

Right hand

Unknown hand

113

Left hand

103

108

V shape

Other apex element

Arc set

Arc pair

Bar set

Fragments

Line single

Line pair

Simple design kla

Simple design klb

Other element

Band

Spray area

Simple design k3

Total

275

377

Table 5 Knuckle size statistics for stencils and prints
(mm).

Stencils

Prints

min

max

mean

median

7.0

106

shape offers few clues to its identification. It could
be a spear, a piece of ceremonial paraphernalia, a
thin digging stick, or simply a straight stick.

Motif size

The width at the knuckles of stencilled and
printed hands was recorded as an indication of the
stature and possible age of the hand's owner (Table
5, Figures 12 and 13). Hand size can therefore
indicate the population range that decorated the
site. The stencils range in size from 50 mm to 90
mm with a mean of 80 mm. The prints range from
70 mm to 90 mm, but also with a mean of 80 mm. In
general, the knuckle measurements of stencils are 5
mm larger than the knuckles of the hand stencilled,
and printed knuckles 5 mm smaller. Hence, the two
techniques represented in Mulka's Cave seem to
represent the same population group. The presence
of two very small hands (knuckle sizes 50 mm and
55 mm) amongst the stencils suggests that infants

Print knuckle size (mm) Stencil knuckle size (mm)

Figure 12 Hand stencil knuckle size frequencies

(n=106). Figure 13 Hand print knuckle size frequencies (n=ll).

R.G. Gunn

Table 6 Technique and colour by knuckle size (mm).

Technique Colour

Stencils

Red-brown

Red-orange

Red-purple

White+red

White

Yellow

Whlte-ryellow

Yellow+red

(n)

Prints

Orange
Cream
Red (or)
Red (pr)

2 1

13 11

1 1 1

were also present at the site (cf. McDonald 1995;
Gunn in prep). At present the age and sex of the
other stencillors can only be attributed to a group
that may include adult women, adult men and/or
adolescents. No large stencilled hands (knuckle >95
mm), which are exclusively indicative of adult
males, occur at the site (Gunn in prep). None of the
printed hands are of infants or large adult males.
The lack of very large hands contrasts with the
reported size of Mulka in the story cited above,
although the myth may refer to the height of some
of the stencils above the floor (>2 m) and hence to
Mulka's stature rather than the presence of his
"large" hands.

There is no preference for particular ages to use
particular colours (Table 6). Purple-red is the most
commonly used colour amongst the hand stencils
and has the widest spread of knuckle sizes. The
range of hand sizes represented in the other colours
covaries with the number made. A similar trend
appears to occur with the handprints but, with the
smaller numbers, this pattern is not as clear.

Sixteen freeform motifs could be measured. These
ranged in overall size from 130 mm to 1270 mm
(Table 7). There is no distinction in size between the
painted and drawn motifs. Freeforms are mostly in
red, with two in yellow. None are made in white or
cream. The largest motif is a "simple design"
prominently placed just inside the main entrance. It
is 1270 mm long and 360 mm wide.

Composition

Compositions are defined as a group of
apparently contemporary motifs with one or more
common traits whose visual impact is greater than
that of the individual motifs of which they are
composed (Clegg 1978; Gunn 1997: 55). In rock art,
their recognition is necessarily subjective (efic)
unless an informed (ewic) explanation can be given
by knowledgeable informants. Aggregates of motifs
can also be seen as a form of composition.

There are no distinct compositions at Mulka's
Cave, although there is a tendency for hand stencils
to occur in rows and to follow the irregularities in
the rock surface. One group of distinctive red
handstencils in the front chamber does, however,
form an aggregate. These are stencilled with the
second and third fingers held together, while the
other fingers and thumb are splayed (Figure 7). Tlais
group appears to have been produced at the one
occasion and is most probably the work of a single
individual. This form of variant hands is similar to
the "3mf" handstencils from Arnhem Land
(Chaloupka 1993: 232), where the three central
digits are held together. These Arnhem Land
handstencils are one element of an art phase that
was produced more than 9000 years ago
(Chaloupka 1984; Lewis 1988). Those at Mulka's
Cave, in contrast, are likely to be much younger,
given the apparent faster rate of the granite
exfoliation. Although isolated examples of this

Table 7 Technique and colour by motif size (mm) (freeform motifs).

Technique Colour 130 140 150 160 170 180 190 220 330 390 420 560 780 830 900 1270

Paint Purple-red 1 1

Orange-red 11 1

Brown-red 11 1

Yellow

Draw Orange-red 11 1

Stencil Yellow

Mulka's Cave rock art site

variant form will doubtless be found in the very
varied stencil art of the Carnarvon Range in central
Queensland (Walsh 1979), no other suites of this
type are known from elsewhere in Australia.

Another variant hand in Mulka's Cave is a white
handstencil in which the thumb and index finger
are splayed and the other three fingers held
together. This stencil is amongst a group of other
white handstencils with standard splayed fingers
and is likely to be the idiosyncratic work of a
particular individual.

A second loose cluster of ten motifs, fhis time of
handprints, occurs on the cusp of the ceiling
between the front and main chambers (Figure 8). In
nine of these, the fingers are held together and the
thumb splayed. The resultant image is suggestive of
a trail of macropod tracks. These motifs appear to
have been made quite recently and
contemporaneously, and are also most probably the
work of a single individual.

Three examples of horizontally placed
handstencils are noteworthy as they occur within
clusters of otherwise vertically orientated hands at
widely spaced locales within both chambers.

The yellow stencil placed centrally within the
prominent purple nested "C" shapes is the younger
of the two. Its placement was clearly deliberate,
although whether the association was culturally
meaningful is unknown.

The two most distinctive motifs within the shelter
are the painted "C" shape just mentioned and an
adjacent rambling design in orange-red. They are
two of the largest motifs within the cave (830 mm
and 1270 mm respectively). Their large size and
prominent location immediately inside the front
chamber suggests they had considerable
significance (cf. Gunn 2003c). The "C" shaped
design overlies the rambling design and hence its
placement was clearly calculated, to either add to
the significance of the earlier motif or, alternatively,
to over-ride it. The rambling design is visually
comparable to other large maze designs at several
shelters elsewhere in the region (Webb and Gunn
2004). They appear to form a distinct stylistic set
whose nature has yet to be explored.

Condition

The condition of pigment art is assessed on the
basis of the amount of pigment adhering to the

Table 8 Motif condition by technique.

Technique

good

fair

poor

very p

(n)

(%)

Spray

389

Nos

Paint

Draw

Table 9 Motif condition by colour.

Colour

good

fair

poor

very p

(n)

(%)

White

111

Red-orange

Red-purple

225

White+red

Nos

Cream

Red-brown

Orange

Yellow+red

White+yellow

Yellow

rockface. Motifs with a dense coating of pigment
are said to be in excellent condition, while those
with the merest of traces are allocated a very poor
condition. The motifs in Mulka's Cave ranged from
good to very poor. No motifs were in excellent
condition, although prints and paintings tend to be
better preserved than the sprayed motifs (Table 8).
This suggests that printing and painting were more
favoured over the recent period of the shelter's use.

The data for condition by colour is limited for
most colours due to low numbers (Table 9). It
appears that cream, brown-red, and white
pigments, being among the best preserved colours,
represent the most recent colours used at the site. In
contrast, purple-red and orange-red have the
highest proportion of very poor motifs, suggesting
that they were more commonly used in the earlier
period of the shelter's use.

Superimposition

Superimposition (the overlying of one motif by
another) provides a sequence that can be a guide to
chronology, if not age. Underlying motifs must be
older than overlying ones. Eighty-seven examples
of superimposition were interpreted. Many other
examples of overlapping motifs also occur but their
sequences could not be reliably elucidated. Not all
of the aspects of the sequence (colour, technique
and motif type) were clear and consequently the
tallies in Tables 9-11 are not the same.

It appears that purple-red was nrore frequently
used during the earlier period of art production,
with white, cream and orange used during the latest
period (Table 10). Given that white preserves very
poorly (Clark 1978), it is again stressed that white
may have been used during the earlier period but
that the motifs have since deteriorated beyond
recognition. The same may be said for cream
pigments, which tend to have a white base. The lack
of purple-red in the recent layers, however, does
seem to be evidence of a real change in pigment
preference.

R.G. Gunn

Table 10 Colour supcrimpositioning.

OVERLAYER

UNDERLAYER

White

Red-

purple

White

+red

Cream

Red-

orange

Orange

Yellow

Red-

brown

Yellow

+red

Total

Purple-red

White

Yellow

Orange-red

White+red

Total

Table 11 Technique superimpositioning.

UNDERLAYER

Spray

OVERLAYER

Print Paint

Draw

Spray

Paint

Table 12

Hand superimpositioning.

OVERLAYER

UNDERLAYER Left

Right

Unknown

Fragment

Left

Right

Unknown

Fragment

Variant

Spray area

Age and chronologx/

Condition by technique and colour suggest that
cream, brown-red and white prints, paintings and
sprayed motifs are more recent than purple-red
sprayed motifs. However, as there is considerable
overlap of the two groups, no clear division can be
made on the basis of preservation alone. Further,
red pigments (ochres) usually preserve better in
rock art than white (kaolin or other pipeclays) (e.g.,
Clarke 1978). Also, it is possible that white pigment
was used during the earlier period of shelter use
but that these motifs have completely deteriorated,
leaving no archaeological evidence of their earlier
presence. The picture is further complicated by the
unusual situation of the apparent placement of
white stencils over a prepared or incidental red
ground. On those few of the motifs that retain white
pigment, the white appears to be relatively recent.
This suggests that some particularly rapid
deterioration process is occurring to these motifs,
masking their recent production. Consequently,
condition here is not a reliable indicator of motif
age.

Similarly, superimpositioning of techniques
suggests that, of the surviving techniques, spraying
was the first used at the site, and that printing,
painting and drawing were introduced into the

repertoire at a later time (Table 11). Also, there was
no change in the preference for hands over time,
with right and left hands occurring in similar
proportions in both the earlier and the more recent
periods of artwork (Table 12). Taken together, motif
condition and superimposition indicate a bipartite
art sequence in which purple-red sprayed motifs
predate sprays, prints and paintings in cream,
brown-red, and white.

The presence of many superimposed motifs
indicates that the site has been used for a
considerable period of time, most likely a far
greater period than is represented by the 500 year
old carbon date (Bowdler et al. 1989). In common
with many other rock art sites throughout
Australia, it is likely that the artwork here was
produced over the past two or three thousand years
(Mulvaney and Kamminga 1999, David 2002).
Handstencils are amongst the earliest rock art
throughout Australia (Morwood 2002) and
excavations around the rim of the Wheatbelt have
demonstrated Pleistocene occupation of south-
western Australia (Bordes et al. 1983; C. Dortch
1979; J. Dortch 1996). However, apart from
walganha, the few decorated shelters that have been
excavated are in broad agreement with the younger
age. For example. Frieze Cave, an art site some 200
km to the west of Mulka's Cave and 100 km inland
from Perth, produced an occupation phase dating
from <3000 BP to the ethnographic present (Hallam
1972). In the Murchison District, 600 km to the north
of Mulka's Cave, most decorated shelters have
occupation similarly limited to the last 4500 years
(Webb in prep.).

The floor deposits and surface scatter

A team from the University of Western Australia
excavated a test pit in the floor of the lower shelter
in 1989 (Bowler et al. 1989). The pit, one metre
square and excavated to bedrock at a depth of 1.10
m, was placed about half a metre inside the dripline
(Figure 4). A limited stone artefact assemblage was
recovered, along with faunal remains representative
of several genera native to the region (principally
small mammals such as bettong and possums, but
no large kangaroos). The artefacts were dominated
by rock types of quartz, with small numbers of

Mulka's Cave rock art site

dolerite, sedimentary rock, and granite. Red and
yellow ochres were also present. Apart from the
few dolerite, sedimentary artefacts and the ochres,
which would have to have been imported into the
site from some distance, the granite and
predominant quartz could be obtained locally from
outcrops on the hill. The lowest artefact was
recovered at 0.7 m, from a pit-like feature
interpreted as a drainage channel. Charcoal
recovered from a depth of 0.65 m was dated to
420+50 BP.

The excavators concluded that the site had been
only briefly or periodically occupied and suggested
that the late occupation of the site (c.500 BP) was
consistent with the "intensification phenomena"
associated with the late Holocene use of marginal
resource zones. However, given the disturbed and
water-washed nature of the excavated deposits,
their conclusions are unlikely to be representative
of the site's overall occupation. Consequently, their
interpretation of brief or periodic occupation is not
seen as definitive. Likewise their c.500 BP date,
coming from a refilled drainage line, cannot be
taken to represent the initial period of shelter use.
Clearly, further excavation at the site would be
required if the period of use of the shelter and its
outside deposits are to be more precisely dated.

Run-off has eroded a gutter out from the front of
the main entrance to the shelter, 100 mm deep,
exposing numerous stone artefacts. Artefacts are
lightly scattered over an area 100 m N-S and 20 m
E-W, following around the base of the rocky
outcrop. This distribution suggests that while the
cave was a focal point for people coming to the site,
most of the occupation occurred on the flat open
ground outside the cave; a situation commonly
observed in arid areas but also noted in southern
woodland areas (cf. Gunn 1997, 2003b).

In a sample of 132 artefacts from the gully
analysed during the present project, quartz

Table 13 Artefact stone types.

Stone type

Milky qz

Clear (non-crystal) qz

Silcrete

Banded ironstone

Ochre (red)

Other siliceous

Total

132

100

accounted for 96% of the artefacts, with silcrete,
banded ironstone, an unidentified siliceous stone
and ochre making up the remainder (Table 13).
Quartz is readily available from a major outcrop
near Captain Roe Rocks, 25 km south-west of The
Humps, and in smaller seams on The Humps itself
and other granite domes in the region. Banded
ironstone outcrops occur in the rises of the Southern
Mineral Field north of Lake Cronin, 70 km to the
east (Geological Survey of Western Australia 2003).
A source for the silcrete is unknown. One of the red
ochre nodules is an orange-red and has a flat facet
with striation marks, consistent with being ground
for use. The other piece is brown-red and
unmarked. Both colours are represented in the
artwork.

The artefacts were mostly (95%) small: <25 mm.
They consisted of debitage fragments (without
platform or bulb; 110 or 83%), flakes (with platform
and/or bulb; 16 or 12%), backed flakes (3 or 2%),
one retouched flake (unifacial and uni-marginal
scraper) and two nodules of red ochre (Table 14).
Four geometric microliths were located near the
main entrance, away from the gutter (Figure 14).
These tool types are characteristic of the Australian
Small Tool Tradition and are therefore likely to be
less than 6000 years old (Mulvaney and Kamminga
1999).

Figure 14 Field drawings of select geometric microliths.

R.G. Gunn

Table 14 Artefact summaries.

Artefact type

Size (mm)
Stone ty'pe

6-10

11-15

16-20

21-25 26-30 31-35 81-S5 broken

Total

Chips

Milky quartz

15 1

Clear quartz

Silcrete

Flakes

Milky quartz

Clear quartz

Banded iron

Other

Backed flake

Clear quartz

(geometric

Silcrete

microliths)

Retouched flake

Banded iron

Ochre nodule

Total

17 1 1 1 3

132

OTHER ARCHAEOLOGICAL SITES

17 cm). It is raised 13 cm off the pavement at its

upslope end by two small rocks (Figure 15). How,

Lizard traps

indeed if, they were used is unknown. Given the

Although little documented

in the literature (see

adeptness of traditional desert Abor

igines in

Webb and Gunn 2004), "lizard traps"

are a widely

catching lizards (cf. Gould 1969b, Tonkinson 1978),

distributed and well-known Aboriginal site-type in

the raised slab was probably an artificial crevice

south-western

Western Australia. Generally they

beneath which the lizards could retreat

to when

consist of a thin granite slab, around 100 cm square

disturbed; only then to be caught with nowhere else

and 10 cm thick (Webb and Gunn 2004), propped
on one end to a height of 10 cm, using one or more
smaller stones (usually around 10 cm diameter).
The trap at The Humps is quite large (240 by 120 by

to run. There would seem to be no point in
dropping the rock onto and crushing any hiding
animals, as has been suggested in colloquial
discussions.

Figure 15 The Humps lizard trap site with support rock arrowed.

Mulka's Cave rock art site

Water reserves

Gnammas, rockholes and pans are different types
of depressions in impermeable bedrock that can
catch and hold water after rain. Although gnammas
and pans are formed by the same geological process
(Twidale and Corbin 1963), archaeologically a
distinction is made between pit gnammas
(gnammas) and pan gnammas (pans) as the former
have a good depth relative to their width and hence
hold water for a good deal longer than pans, which
are broad and relatively shallow (up to 20 cm) and
are readily evaporated. Gnammas are essentially
rain-fed. In contrast to gnammas, rockholes form
along creek lines, usually as plunge pools, and are
fed by the creek-flow.

Throughout the Murchison/Wheatbelt area,
gnammas were a well-known, well-utilised and
highly valued water source for Aborigines and the
early settlers (Bowdler et al. 1989; Bindon 1997;
Webb and Gunn 2004). They provided a reliable
reserve that could be targeted by people familiar
with the country in which they occur (e.g., Gould
1969a, Myers 1991). Rockholes tend to develop at
the base of escarpments, whereas pans develop on
top of suitable rock exposures such as granite
domes and duricrust surfaces on top of
escarpments. Many gnammas, in contrast, occur as
nondescript features in a superficially nondescript
landscape and would be impossible to intentionally

locate without detailed knowledge of the country
(e.g., Gunn and Webb 2003). Many also had granite
slab "lids" to prevent animals draining the supply
and to reduce evaporation (e.g., Webb and Gunn
2004). Traditionally, rockholes and gnammas would
have had Aboriginal names and some associated
mythology. Pans, on the other hand, were not
regarded as highly because of the more limited life
of their reserves due to evaporation.

Five gnammas have been located on and around
The Humps (Figure 3). Four occur on the granite
pediment to the north of the Mulka's Cave. Two are
quite large (e.g.. Figure 16); whereas the two
adjacent to the lizard trap are very small (Table 15).
The fifth occurs on the northern mid-slope of the
hill.

A sizeable rockhole has formed within a shallow
drainage line that flows off the hill to the north of
Mulka's Cave. The ephemeral stream that feeds this
rockhole is a prominent feature of the hill. It
supports a number of well-vegetated soil patches
on an otherwise steep and barren rock slope. The
stream flows readily after rain, and continues to
trickle for some days after rain (personal
observations). The rockhole measures 8.5 by 2.0 by
0.8 m. It has a "V" profile and contains some 650
litres of water when full. As the stream leaves the
hill it winds into granite sands that quickly become
boggy. It is likely that small sub-surface soaks

Figure 16 The largest gnamma at The Humps.

R.G. Gunn

Table 15 Gnamma dimensions and approximate volumes.

Site

Length

(m)

Width

(m)

Depth

(m)

Capacity

(1)

Humps 1

1.40

0.85

0.65

Humps 2

4.00

2.50

0.60

300

Humps 3

0.45

0.20

0.30

Humps 4

0.20

0.15

0.20

Humps 5 upper pan

0.85

0.60

0.20

Humps 5 lower gnamma

0.25

0.20

0.35

would be available along the stream's path for some
time after the rockhole had dried.

The pans on top of the hill are numerous and
large (around 4.0 by 2.0 by 0.1 m) and together,
after rain, would probably hold around 1000 litres
of water. With another 1200 litres held over in the
gnammas and rockhole, and the possibility of sub-
surface soaks along the stream bed, a good
minimum supply of around 2000-3000 litres could
be ensured; more than enough for an extended
family group's transitory stopover or a larger
gathering for brief periods. The presence of these
reserves may well explain why Mulka's Cave was
chosen for decoration.

AN INTERPRETATION OF MULKA'S CAVE

In common with other sites in the Wheatbelt
(Davidson 1952; Wolfe-Okongwu 1978; Webb and
Gunn 2004; Dept. Indigenous Affairs, Perth,
records), the art of Mulka's Cave is dominated by
handstencils, but it also includes a small number of
large and visually impressive linear paintings. The
site also houses a small number of less impressive
handprints, small paintings and drawings. Mulka's
Cave, with a total of 452 motifs, has a far higher
quantity of rock art than most other Wheatbelt rock
art sites, which tend to have fewer than 30 motifs.
The artwork also utilises a wider range of colours
than other regional sites.

As McDonald (1995) inferred for rock arf sites in
the Sydney Basin, the presence of infant
handstencils and others that most likely are those of
women and adolescence suggests that at some times
family groups camped by and used the cave. The
unsuitability of the cave's floor for sleeping suggest
that the cave was used as a wet weather retreat and
that most camping occurred on the level ground in
front of the cave. This is supported by the pattern of
surface artefact distribution.

On the other hand, the painting of large geometric
designs are almost certainly the work of senior
males and most likely refer to the Dreaming tracks
of totemic ancestors, as they do in the Western
Desert and Central Australia (Spencer and Gillen
1899; Gould 1969b; Bardon 1979; Myers 1991).
These paintings and the associated Mulka story

suggest that, at times, the cave was associated with
local nearby rituals, at which times it would have
been off-limits to women and children. A similar
dual-use has been proposed for the densely
decorated shelter at walganha (Walga Rock), which
was known to have been used for ritual in the
recent past but was apparently open to women and
children at non-ceremonial times (Gunn etaJ. 1997).

Overall then, it is suggested that Mulka's Cave
was used by the full age-range of the population,
although whether at the same time or at different
times for different purposes, is yet to be
determined. This view does not support the
suggestion that the cave was used as a store-house
for ceremonial objects (Serventy 1952), as such
store-houses were, and in many areas still are,
invariably accessed only by senior males and are
always off-limits to women and children (e.g.,
Gould 1969b, Mountford 1976).

The large area of the campsite in front of the
shelter along with the higher number of stone
artefacts relative to other art sites in the region
suggests that occupation here was indeed recurrent;
but whether through short-term, large gatherings or
prolonged occupation by small family groups is
likewise unknown. The artefacts were made from a
limited array of raw materials, all available locally,
suggesting that Mulka's Cave was not a focus for
large inter-regional gatherings when materials from
well outside the region would be brought in by the
visitors.

Most water sources and granite outcrops within
the region, particularly those with rock art or stone
arrangements, probably had associated myths
similar to the Mulka story recorded for this site.
This was certainly the case elsewhere in Australia
where such information has been collected (e.g.,
Hallam 1972; Berndt and Berndt 1977: 243-256;
Strehlow 1978; Myers 1991). Hence, while it is
tempting to give some special credence to Mulka's
Cave because of the survival of the Mulka myth,
this should not be seen as particularly unusual. On
the other hand, some myths were more significant
than others, possibly due to the strengths of
particular clans or individuals (see Spencer and
Gillen 1899). Whether the Mulka story was of
regional significance or of only local importance is

Mulka's Cave rock art site

unknown. The high quantity of artwork, the wide
range of colours used, the size of the campsite and
the extensive area encompassed by the myth,
suggest that Mulka's Cave had at least a regional
significance. In contrast to this, the restricted range
of motif types used suggests that the site was not a
place of inter-tribal gatherings or "aggregation site"
as defined by Conkey (1980) and Gait-Smith (1997).
At most, the Mulka's Cave site complex would
seem likely to have been a focus for local group
ceremonies, such as initiation or maintenance
rituals. From what we know of such rituals, they
were invariably focused on open areas away from
general campsites and that rock art use was
generally a tangential, albeit important, component
of the main ceremony. It is therefore unlikely that
Mulka's Cave itself was the principal focus for these
ceremonies. If ceremonies did take place at The
Humps, they would have been held nearby, either
on the dome or out on the adjacent plains, with
visits to the cave just one aspect of the ceremony. A
likely location for the performance of the main
ceremony has not been located.

CONCLUSION

In keeping with the regional profile, the rock art
preserved within Mulka's Cave is dominated by
handstencils but with a small number of geometric
designs. Typically however, one or two of the latter
are the largest and most visually impressive motifs
within the site. The suite at Mulka's Cave is the
densest concentration of rock art yet recorded
within south-western Western Australia and
consequently must be seen to be one of the most
significant archaeological sites within the region.

Similarly, the artefact scatter in front of the shelter
is far denser than those noted at most other rock art
complexes in south-western Australia (Webb and
Gunn 2004), suggesting that the site was important
as a recurrent focus of occupation. The
archaeological potential of the deposits here have
been tested but have yet to be explored to their full
depth. The importance of this potential must be
acknowledged in the future management of the site.

The interpretation of the evidence to date,
although based on analogies from other areas,
suggests that the site was used both for ritual
purposes related to broader ceremonies held nearby
and, probably at different times, for general
occupation.

Today, Mulka's Cave is of great significance to
the local Noongar community and also, but for
different reasons, to non-Indigenous residents of
the region and the local tourist industry. It is one of
the most visited rock art sites in Western Australia
and hence has a broad potential for positive
educational uses.

The main focus of this study was to document the

site and illustrate the importance of its' rock art. It
is hoped that this will ensure that Mulka's Cave,
and The Humps Reserve in which it lies, continues
to be managed and interpreted, in a culturally
sensitive manner, to the highest standards are
available.

ACKNOWLEDGEMENTS
The recording project was funded jointly by the
Department of Indigenous Affairs (Southern
Region) and the Hyden-Karlgarin Landcare
Council, and supervised by Robert Reynolds (DIA
Midlands) and Anthony Galante (DIA Southern
Region). It would not have occurred without the
support of the native title claimants. Robert
Reynolds and Brian Blurton (DIA), Monica Jimenez-
Lozano, Alana Rossi and Jodee Smith (Honours
graduates from the University of Western Australia)
also assisted in the field. Jane Mouritz (Hyden)
organised the essential generator and floodlights. I
also thank Esmee Webb (Edith Cowan University)
for her patience, as the project was undertaken
midway through fieldwork with her on another
study . I also profited from discussions with Rowle
Twidale regarding geological terminology. Sylvia
Hallam and Charlie Dortch reviewed the paper and
made me rethink several of my propositions, while
Esmee Webb, Leigh Douglas and Sarah Turnbull
also offered suggestions for improving my English.
The author however, takes full responsibility for the
final presentation and the conclusions presented.

REFERENCES

Bardon, G. (1979). Aboriginal Art of the Western Desert.
Rigby, Adelaide.

Berndt, R. M. (1980). Aborigines of the south-west. In R.
M. Berndt and C. M. Berndt (eds) Aborigines of the
West, pp. 81-89. University of Western Australia
Press, Perth.

Berndt, R. M. and Berndt, C. H. (1977). The World of the
First Australians. Ure Smith, Sydney.

Bindon, P. (1997). Aboriginal people and granite domes.
Journal of the Royal Society of Western Australia 80: 173-
179.

Bordes, F., Dortch, C.E., Thibault, C., Raynal, J-P. and
Bindon, P. (1983). Walga Rock and Billibilong Spring:
two archaeological sequences from the Murchison
Basin, Western Australia. Australian Archaeology 17:
1-26.

Bowdler, S., Harris, J., Murphy, A., Nayton, G. and
Pocock, C. (1989). Test excavations at Mulka's (Bate's)
Cave near Hyden. Unpublished report to the
Department of Aboriginal Sites, Western Australian
Museum, Perth.

Chaloupka, G. (1984). From Paleo Art to Casual Paintings.
Northern Territory Museum of Arts and Sciences
Monograph Series 1, Darwin.

Chaloupka, G. (1993). Journey in Time: the World's
Longest Continuing Art Tradition. Reed, Chatswood.

R.G. Gunn

Clarke, J. (1976). Graffiti removal. Bate's Cave, Hydcn,
Western Australia. Unpublished report to the
Department of Aboriginal Sites, Western Australian
Museum, Perth.

Clarke, J. (1978). Deterioration analysis of rock art sites.
In C. Pearson (ed.) Conservation in rock art, pp. 54-63.
Institute for the Conservation of Cultural Material,
Perth.

J- (1978). Mathesis W'ords, mathesis pictures.
Unpublished MA thesis. Dept. Anthropology,
University of Sydney.

^legg, J. (1987). Human picturing behaviour and the
study of prehistoric pictures. Rock Art Research 4: 29-
35.

Conkey, M. (1980). The identification of prehistoric
hunter-gatherer aggregation sites: the case of
Altamira. Current Anthropology IV. 609-630.

David, B. (2002). Landscapes, Rock Art and the Dreaming: an
Archaeology of Preunderstanding. Leicester University
Press, Leicester.

Davidson, D. S. (1952). Notes on the pictographs and
petroglyphs of Western Australia and a discussion of
their affinities with appearances elsewhere on the
continent. Proceedings of the American Philosophical
Society 9b: 76-117.

Day, R. B. (1951). Western Australian, 27 January 1951.

Department of Aboriginal Sites (1989). Mulka’s Cave (also
known as Bates Cave), Hyden, IVesfern Australia.
Brochure, Dept. Aboriginal Sites, West Perth.

Dortch, C. (1979). Devil's Lair, an example of prolonged
cave use in Western Australia. World Archaeology 10-
258-279.

Dortch, J. (1996). Late Pleistocene and recent occupation
of Tunnel Cave and Witchcliffe Rock Shelter, south-
western Australia. Australian Aboriginal Studies 1996/
11: 51-60.

Flood, ]. (1990). The Riches of Ancient Australia. University
of Queensland Press, St. Lucia.

Gale, F. and Jacobs, J. M. (1987). Tourism and the
National Estate. Australian Government Publishing
Service, Canberra.

Gait-Smith, B. (1997). Motives for motifs. Unpublished
BA Honours thesis. University of New England,
Armidale.

Geological Survey of Western Australia (2003).
Geological Atlas of Western Australia. CD-ROM
version. Geological Survey of Western Australia,
Department of Industry and Resources, Perth.

Gould, R. A. (1969a). Subsistence behaviour among the
Western Desert Aborigines of Western Australia.
Oceania 39: 253-274.

Gould, R. A. (1969b). Yiwara: Foragers of the Australian
Desert Charles Scribner's Sons, New York.

Gunn, R. G. (1981). The Aboriginal rock art of the Mt
Lofty Ranges, SA. Uncommissioned report to the
Dept, of the Environment, Adelaide.

Gunn, R. G. (1983). Aboriginal rock art in the Grampians.
Records of the Victorian Archaeological Survey No
16. Ministry for Conservation, Melbourne.

Gunn, R. G. (1995a). Recording rock images for
management purposes. In G. K. Ward and L. A. Ward
(eds) Management of rock imagery, pp. 93-96.

Occasional AURA Publication No. 9. Australian Rock
Art Research Association, Melbourne.

Gunn, R. G. (1995b). Guidelines for recording Australian
Aboriginal rock imagery. In G. K. Ward and L. A.
Ward (eds) Management of rock imagery, pp. 124-
127. Occasional AURA Publication No. 9. Australian
Rock Art Research A.ssociation, Melbourne.

Gunn, R. G. (1997). Arrernte rock art, occupation and
myth: the correspondence of symbolic and
archaeological sites at rock art complexes in Central
Australia. Rock Art Research 14: 124-136.

Gunn, R. G. (2002). Mudgegonga-2 and the rock art of
north-east Victoria. Rock Art Research 19: 117-132.

Gunn, R. G. (2003a). Mulka's Cave Aboriginal rock art
site: A discussion of management options.
Unpublished report to the Department of Indigenous
Affairs, Midlands.

Gunn, R. G. (2003b). Three more pieces to the puzzle:
Aboriginal occupation of Gariwerd (Grampians),
Western Victoria. The Artefact 26: 32-50

Gunn, R. G. (2003c). Arrernte rock-art: interpreting
physical permanence in a changing social landscape.
Australian Aboriginal Studies 2003: 52-73.

Gunn, R, G. (2004). Mulka's Cave Aboriginal rock art
site: an archaeological recording. Unpublished report
to the Department of Indigenous Affairs, Albany, and
the Hyden- Karlgarin and District Landcare Group,
Hyden.

Gunn, R. G. (in prep) Hand sizes in rock art: Interpreting
the measurements of hand stencils and prints in
central Australia. Rock Art Research.

Gunn, R. G. and Webb, R. E. (2003). Art and archaeology
on Coodardy, Austin Downs and Noondie pastoral
leases, west of Cue. Unpublished report to the Thoo
Thoo Warninha Aboriginal Cooperative, Cue, and the
Australian Institute of Aboriginal and Torres Strait
Islander Studies, Canberra.

Gunn, R. G., Webb, E. and Marmion, D. (1997). Walganha
(Walga rock): a Wajarri rock art and Dreaming site.
Report to the Yamaji Language Centre, Geraldton,
and the Australian Heritage Commission, Canberra.

Hallam, S. J. (1972). An archaeological study of the Perth
area. Western Australia: a progress report on art and
artefacts, dates and demography. Australian Institute
of Aboriginal Studies Newsletter 3{5): 11-19.

Haydock, P. and Rodda, J. (1986). A survey of rock art
conservation in the Murchison/Wheatbelt area of WA:
a study of past treatments and new methods of
measurement and site management. Unpublished
report of the Department of Aboriginal Sites, Western
Australian Museum, Perth.

Lewis, D. (1988). The Rock Paintings of Arnhem Land,
Australia. BAR International Series 415, Oxford.

McDonald, J. (1983). The identification of species in a
Panaramitee style engraving site. In M. Smith (ed.)
Archaeology at Anzaas 1983, pp. 236-272. Western
Australian Museum, Perth.

McDonald, J. (1995). Looking for a woman's touch:
indications of gender in shelter sites in the Sydney
Basin. In J. Balme and W. Beck (eds) Gendered
archaeology: the second Australian Women in
Archaeology Conference, pp. 92-96. ANH

Mulka's Cave rock art site

Publications, Australian National University,
Canberra.

Maynard, L. (1977). Classification and terminology in
Australian rock art. In P. J. Ucko (ed.) Form in
indigenous art, pp. 387^02. Australian Institute of
Aboriginal Studies, Canberra.

Maynard, L. (1979). The archaeology of Australian
Aboriginal art. In S. Mead (ed.) Exploring the visual
arts of Oceania, pp. 83-111. University Press, Hawaii.

Meeking, J. (1979). The History ofHyden. Mid-West Print,
Northam.

Meggitt, M. J. (1962). Desert People. Angus and Robinson,
Sydney. (1974 edition).

Morwood, M. J. (2002). Visions from the Past: the
Archaeology of Australian Aboriginal Art. Allen and
Unwin, Sydney.

Mountford, C. P. (1976). Nomads of the Australian Desert.
Rigby, Adelaide.

Mulvaney, D. J. and Kamminga, J, (1999). Prehistory of
Australia. Allen and Unwin, Sydney.

Myers, F. R. (1991). Pintubi Country, Pintubi Self.
University of California Press, Berkeley.

O'Conner, S., Veth, P. and Campbell, C. (1998). Serpent's
Glen rockshelter: report of the first Pleistocene-aged
occupation sequence from the Western Desert.
Australian Archaeology 46: 12-22.

Randolph, P. (1973). Bates Cave, Hyden. Unpublished
report of the Department of Aboriginal Sites, Western
Australian Museum, Perth.

Roe, J. S. (1852). Report of an expedition under the
Surveyor General, Mr J. S. Roe, to the south-eastward
of Perth, in Western Australia, between the months of
September, 1848, and February, 1849, to the Hon. the
Colonial Secretary. Journal of the Royal Geographic
Society 22: 1-57.

Ross, J. (2003). Rock art, ritual and relationships: an
archaeological analysis of rock art from the central
Australian arid zone. Unpublished PhD thesis.
University of New England, Armidale.

Serventy, V. (1952). Cave paintings near York and
Hyden. Western Australian Naturalist 13(5): 121-130.

Spencer, B. and Gillen, F. J. (1899). The Native Tribes of
Central Australia. Macmillan and Company, London.

Strehlow, T. G. H. (1978). Central Australian Religion:
Personal Monototemism in a Polytotemic
Community. Australian Association for the Study of
Religions, Bedford Park.

Tonkinson, R. (1978). The Mardudjara Aborigines. Holt,
Rinehart and Winston, New York.

Thorn, A. (2001). Conservation of Mt Pilot sites.
Unpublished report to Aboriginal Affairs Victoria,
Melbourne.

Tindale, N. B. (1974). Aboriginal Tribes of Australia: their
Terrain, Environmental Controls, Distribution, Limits,
and Proper Names. ANU Press, Canberra.

Twidale, C. R. (1980). Geomorphology. Nelson, Melbourne.

Twidale, C. R. and Bourne, J. A. (1975). The subsurface
initiation of some minor granite landforms. journal of
the Geological Society of Australia, 22: 477-484.

Twidale, C. R. and Bourne, J. A. (2003). Proposed survey
of the Humps. Unpublished MS on file with the
Tourism Officer, Shire of Kondinin.

Twidale, C. R. and Corbin, E. M. (1963). Gnammas. Revue
de Geomorphologie Dymamique 14: 1-20.

Walsh, G. L. (1979). Mutilated hand or signal stencils?
Australian Archaeology 9: 33-41.

Webb, R. E. (in prep.). Results of test excavating three
decorated rock shelters on Wutha country, east of
Cue, Western Australia. Report to the Australian
Institute of Aboriginal and Torres Strait Islander
Studies, Canberra, the Wutha native title claimants
and Thoo Thoo Warninha Aboriginal Corporation,
Cue.

Webb, R. E. and Gunn, R. G. (2004). Re-recording
culturally significant sites in south-western Australia
as a guide to Noongar usage of the region in the past.
Rock Art Research 21 : 93-97.

Wolfe-Okongwu, W. (1978). Aboriginal art of south-west
of Western Australia. Unpublished BA (Honours)
thesis. Department of Anthropology, University of
Western Australia, Perth.

Manuscript received 6 October 2004; accepted 26 ]uly 2005

Records of the Western Australian Museum 23: 43-76 (2006).

The leptolepid fish Cavenderichthys talbragarensis (Woodward, 1895) from
the Talbragar Fish Bed (Late Jurassic) near Gulgong, New South Wales

L. B. Bean

Dept of Earth and Marine Sciences, The Australian National University,
Canberra, ACT 0200, Australia
e-mail: Lynne.Bean@ems.anu.edu.au

Abstract - "Leptolepis" talbragarensis Woodward, 1895, is the most common
fish species in the Talbragar Fish Bed near Gulgong, New South Wales. The
genus Cavenderichthys Arratia, 1997, has this species as its type. The three
species originally proposed by Woodward (1895) for "Leptolepis" are a single
species. A detailed comparison of Cavenderichthys talbragarensis with members
of the genus Leptolepis, and also with the Late Jurassic forms Tharsis dubius
and Leptolepides sprattiformis, indicates that Cavenderichthys talbragarensis is
most closely related to Late Jurassic members of the Family Leptolepididae.
Analysis of zircons for geochronology showed that the sediment just below
the richest fish layer has a youngest component of 151.55 ± 4.27 Ma,
corresponding to the Kimmeridgian Stage of the Late Jurassic. Thin sections
of the upper prolific fish layer show preservation in tuffaceous sediments,
indicating that the fish population was killed by ash falls of felsic tuff that
filled the pond they inhabited.

INTRODUCTION

Fossil fishes were first discovered at Talbragar
about 30 km northeast of Gulgong by Arthur Lowe
of Wilbertree, NSW in 1889 (Woodward 1895).
Later, many specimens were collected by Charles
Cullen, the collector of fossils for the NSW Mines
Department. This material is now in the Australian
Museum, Sydney, and the NSW Department of
Mineral Resources. Associated with the fishes is
abundant plant material, first described by Walkom
(1921), then re-examined and classified by White
(1981). Some undescribed insect remains are also
housed in the Australian Museum, Sydney.

Woodward (1895) described a representative
selection of different fossil fishes that had been sent
to London in 1890. He considered that the
assemblage was of Jurassic age, despite an original
field assessment of the age as Triassic, made by W.
Anderson of the Geological Survey of NSW. The
vast majority of the fishes in the material belong to
Leptolepis talbragarensis Woodward, 1895, which was
cited by Long (1991) as "the first appearance of the
teleosteans in the Australian fossil record". Other
fishes include one species of palaeoniscid, Coccolepis
australis (Woodward, 1895), and the holosteans
Archaeomaene tenui Woodward, 1895, Madariscus
robustus Wade, 1941, Aphnelepis australis
Woodward, 1895, Aetheolepis mirabilisYloodward,
1895, and Uarbryichthys latusWade, 1941 . Interest in
Leptolepis talbragarensis is due largely to its early
teleostean features. Nybelin (1974) suggested that L.
talbragarensis should be excluded from the family
Leptolepididae Agassiz, 1833-44. This was based

partly on his own observations, but also on the
work of Cavender (1970) who compared
coregonines and other salmonids with some of the
earliest known teleosts, including L. talbragarensis.
Arratia (1997) erected a new genus, Cavenderichthys,
with talbragarensis as the type species, on the basis
of material from the Natural History Museum,
London, the Field Museum of Natural History,
Chicago, and the Swedish Museum of Natural
History, but she did not have access to the vast
amount of material available in Australia.

The Talbragar site has revealed the best-preserved
Jurassic fish in Australia. The outcrop now is very
poor, as so much material has been removed in the
past and it is now in a paddock used for grazing.
The age has been difficult to confirm because there
is no control of stratigraphy as the relationship to
surrounding rocks is unclear. Previously, no
volcanic rocks had been identified to be dated, and
palynology is impossible because of the highly
oxidised nature of the rocks.

The assemblage of fossil fishes has been thought
to indicate an Upper Jurassic age (Long 1991), but
early workers suggested a Middle Jurassic age, for
example Hind and Helby (1969) who suggested
Early to Middle Jurassic based on palynology of the
Purlawaugh Formation, within which the Fossil
Fish Bed occurs. The site is interpreted as a mass
kill site with a longitudinal extent of possibly 200
metres. The upper layer contains a high
concentration of extremely well preserved fossil
fish, while the layer below, probably less that one
metre thick, has scattered fish throughout.

L.B. Bean

indicating a lacustrine environment. Until now
evidence for the cause of death has been sparse,
although Percival (1979), and White (1981), have
made suggestions.

The purpose of this paper is to reassess the
description and classification of Cavenderichthys
talbragarensis, and to discuss the environment of
deposition, the age of the fossil bed, and the nature
of preservation. To do this the type material in the
Australian Museum, as well as about 250 other
specimens from the Australian Museum, the N.S.W.
Geological Survey and the Australian National
University have been examined. The sediment has
been studied in thin section, as has its geochemistry,
and plant content. Zircon dating was carried out
using the SHRIMP method.

GEOLOGY

The Talbragar Fossil Fish Bed is the informal
name given by Dulhunty and Eadie (1969) to the
outcrop found on the northeastern side of Farrs
Hill, about 5 km south of the Talbragar River. The
location is GR 753090 6437910, Dubbo 1:250 000
Geological Sheet (Pogson and Cameron 1999). The
site is now a geological reserve administered by the
National Parks and Wildlife Service, Mudgee
Office. The strike is generally north-south, and the
dip of adjacent beds is about 10° west. The Fish Bed
is thin, forming part of a non-marine sequence, just
below the fossil bed are layers that contain
tuffaceous sections. Unweathered samples are grey,
very fine grained, and contain angular fragments of
minerals such as quartz, some of which is detrital
and some of which appears to be igneous in origin.
There is no evidence of sedimentary flow structure.

Stratigraphically below this unit are quartz
sandstones of the Purlawaugh Formation, which
show sedimentary structures such as cross-bedding,
pebble layers and washouts. This sandstone unit is
comparable to the nearest units of the Purlawaugh
Formation that outcrop about 50 km away. The Fish
Bed is probably the upper unit of the Purlawaugh
Formation, but no equivalent outcrop to the Fish
Beds is exposed in New South Wales.

SHRIMP (Sensitive High mass Resolution Ion
MicroProbe) analysis of zircons was carried out
using the SHRIMP RG machine in the ANU
Research School of Earth Sciences. The age of the
youngest population was 151.55 ± 4.27 Ma,
corresponding to the Late Jurassic (Veevers 2000),
indicating that the sediment must be this age, or
younger if the zircons were all of sedimentary
origin. The morphology of the youngest grains does
not show any evidence of transportation by water.
Examination of the zircons shows that the rock
contains a small tuffaceous component. The range
of different types of zircons was quite large, and
many of them showed clear evidence of a

sedimentary history (Dr 1 Williams, personal
communication).

Dulhunty and Eadie (1969) described the "Fish
Bed Chert" as a hard, fine limonitic cherty-shale,
and Pogson and Cameron (1999) stated "In thin
section the unit is a red-brown silty mudstone with
compaction bedding features and chips of
?tuffaceous quartz, clayey patches after feldspar
and/or lithic fragments, magnetite, ankeritic cement
and manganese oxide dendrites." Thin section and
chemical analysis shows that the fossil-bearing
rocks are largely tuff and sediments derived from
the underlying sandstones, some of the tuffs
representing one or more very fine-grained ash falls
(Prof. R. Arculus, personal communication). EDXA
(Energy Dispersive X-ray Analysis) has not shown
any evidence of carbonate or calcium ions being
present, excluding an ankerite [FeCa(CO,), ] cement
(Dr A Christy, personal communication). There is
evidence of hne bedding and subsequent
compaction. The red-brown colour is post
depositional because each block has concentric
bands of varying intensity of colour as the iron
oxide has penetrated from the joint block
boundaries. Manganese dioxide is often found
infilling the fossil fish cavities and is generally close
to the edge of a block, forming dendrites. Many of
the fish and most of the plant fossils are white,
having not taken up the red iron oxides. EDXA
shows the composition of the infilling of plants and
animals is not the same. The plants have been
replaced by very fine-grained opalised quartz,
whereas kaolin is present with the opalised quartz
in the infilling of the fish (Dr A Christy, personal
communication).

Tlie Talbragar Fossil Fish Bed is probably no more
than 60 cm thick (Percival 1979). The current state
of the outcrop is poor as the bed occurs as small
blocks of fossil-bearing rock scattered through the
soil of a paddock. It is impossible to measure the
precise thickness or the boundaries of the bed
without excavation. The layers of fossil-bearing rock
vary from about 2 cm to 4 cm thick, but within
these layers the fish are scattered in overlapping
layers, rather than all being at the top or all at the
bottom of the layer. The exception is some large
blocks covered with vast numbers of small fish,
some available in part and counterpart. Tliere is no
evidence of the original location of these blocks, but
it is assumed that this very fossiliferous layer is the
upper layer of the deposit, and thus represents one
mass-kill event. There is no evidence of desiccation
in the sediment, such as mud cracks or aerially
exposed surfaces, so this is not a mound spring
deposit. It is not an overbank deposit either, as
these usually have cyclic layers including sands and
coarse-grained layers from flooding, interspersed
with soil developments from dry times. The layers
that contain an abundance of small fish, which are

Late Jurassic leptolepid fish Cavenderichthys talbragarcnsis

thought to occur at the top of the bed, are very fine
grained and represent a period of slow deposition,
or a time when the pond was still and suddenly
became anoxic. Lower layers have occasional
scattered fish that have been deposited along with
sediment. Percival (1979) recorded that "it is now
thought to represent the erosional remnant of the
margin of a freshwater lake bed deposit". Evidence
now points to the destruction of the lake by several
eruptions of volcanic ash.

The fossils show no preferred orientation,
although most are laterally flattened. Only one of
several hundred specimens is dorso-ventrally
flattened. Some of the smaller individuals show
dorsal flexion. This flexion of the spine could
represent greater flexibility of the juvenile
individuals, or could possibly be a result of them
dying in suddenly anoxic water as the result of an
ash fall. The dorsal flexion of the small specimens
was also commented upon by Waldman (1971) in
his description of the fish in the Cretaceous
Koonwarra beds in Victoria. He considered that
particular assemblage, which includes a large
number of the closely related species, Leptolepis
koonwarri Waldman, 1971, was due to winterkill and
claimed the flexion is due to asphyxia of the
individuals when the pond was covered by ice. As
is noted later in this paper, many of the fish are
preserved with their mouths open, which could
support the idea of anoxia.

Plant fossils are commonly associated with the
fish. None of these represent plants growing in situ,
and there is no evidence for any water dwelling
plants. The plant material consists of twigs,
individual leaves, occasional cones, and very small
fragments. Some beds have masses of very finely
shredded plant material.

The area surrounding the lake was heavily
forested with an araucarian pine, Agathis jurassica
(White 1981). The fine detail of plants and fishes
preserved implies an anaerobic burial environment.
Most of the fish are intact with very few examples
of disarticulated bones, indicating a lack of post-
mortem turbulence, predation and decay. The plant
fragments show venation and coll structure, thus
showing no signs of decay or transportation.

Etheridge and Olliff (1890) described one example
of a cicada named Cicada? lowei found in the fish
beds, and further examples of insects have since been
found. The Australian Museum houses a collection
of Talbragar insects that has not been studied in
detail. These insects are the only preserved evidence
of a food source for the fish. The insects are
apparently found in the upper layer where the fish
fossils are most concentrated (R. Beattie, personal
communication). Since none have been found in the
lower layers where the fish are more scattered it may
indicate that the insects were trapped by the ash fall
that finally filled in the pond.

Materials

The specimens described in this study come from
three sources; Australian Museum, Sydney, prefix
AMF (30 specimens); NSW Geological Survey,
Sydney, prefix MMF (107 specimens); Australian
National University, Canberra, prefix ANU (106
specimens).

Where more than one fossil appears on a
numbered specimen, the individual fossils have
been allocated a letter suffix to distinguish them, eg
MMF36743b.

Some material mentioned in text and figures
relates to specimens in the Natural History
Museum, London, prefix BMNH.

SYSTEMATICS
Subclass Teleostei Muller, 1844
Family Leptolepididae (Agassiz, 1833-44)
Genus Cavenderichthys Arratia, 1997

Synonymy

See Arratia (1997:19).

Diagnosis

Small teleosts ranging from about 4 cm to 12 cm;
head with short snout; lower jaw projecting
anteriorly; fusiform body. Frontal bone short
anteriorly. Suborbital bone absent. Quadrate-
mandibular articulation below anterior half of
orbit. Elongated symplectic and hyomandibular, as
well as ventral limb of preoperculum. Lower jaw
with deep coronoid process and wide leptolepid
notch. Hyomandibular with a preopercular
process. No suprapreopercular bone. Infraorbital
sensory canal with very few tubules; generally
four broad tubules on lower limb of
preoperculum, one at the angle, and one on
vertical limb. Anterior ceratohyal short and
usually not fenestrate; with six thin arcinaciform
branchiostegal rays, and three or four spathiform
branchiostegal rays associated with the posterior
ceratohyal. 35-45 vertebrae with autogenous
neural arches in abdominal region and fused
neural and haemal arches in caudal region, with
20-26 pairs of ribs. Midcaudal autocentra thin,
ring-like, with or without a longitudinal crest on
lateral surface. 12 pectoral rays, 12 pelvic rays, 12
dorsal rays + 3 procurrent dorsal rays, and 10 anal
rays. Pelvic, dorsal and anal rays branching
distally into 4 lepidotrichs. Preural centrum 1 with
short neural spine. Three or rarely four epurals;
seven hypurals and five uroneurals; 10+9 principal
caudal rays. Well-developed dorsal processes on
bases of innermost principal caudal rays of dorsal
lobe of caudal fin absent. Two "urodermals". Six
basal fulcra on upper lobe of caudal fin.

L.B. Bean

Remarks

This diagnosis is based on Arratia (1997:19).
However changes have been made where
examination of new material has added information
that contradicts the original diagnosis. For example,
Arratia cited a deep body, the hyomandibular
lacking a preopercular process, the lower jaw
lacking a leptolepid notch, 12 or 13 branchiostegal
rays, 43-45 vertebrae, 25-27 ribs, nine hypurals,
seven uroneurals, and a lack of epipleural bones.
These features are discussed later.

Cavenderichthys talbragarensis (Woodward, 1895)
Figures 1-22

Leptolepis talbragarensis Woodward, 1895: pp. 21,
22, pi. 6, figs 1-8.

Leptolepis lowei Woodward, 1895: pp. 22, 23, pi. 6,
figs 9, 10.

Leptolepis gregarius 'Woodward, 1895: pp 23-24, pi.
4, figs 8-10, pi. 5, fig. 5, pi. 6, figs 11, 12.

Material Examined

Holotype

MMF81. This specimen has been examined, but
not used in the current description. It was named as
the type specimen by Woodward (1895) and
appeared as Plate 6, figure 4.

Paratypes

The following specimens, housed in the

Australian Museum, Sydney, are paratypes,
described by Woodward (1895). The old numbers
have prefix MF, and have been replaced by new
numbers with prefix AMF. AMF120525 (MF276),
AMF120509 (MF276), AMF120505 (MF276),
AMF120498 (MF276), AMF120512 (MF276),
AMF120497 (MF276).

Other material

The following specimens have been examined,
provided latex casts in most cases, and are quoted
as examples in the text.

AMF27069, AMF4133, AMF51899, ANU54916,
ANU54940, ANU54946, ANU54956, ANU54962,
ANU54968, ANU54970, ANU54975, ANU54976,
ANU54977, ANU54980, ANU54982, ANU54983,
MMF13555, MMF13561a, MMF13564, MMF13569,
MMF13599b, MMF13603k, MMF13606a,
MMF13734a, MMF36716, MMF36718, MMF36721,
MMF36728, MMF36729, MMF36730, MMF36732,
MMF36732a, MMF36733, MMF36735, MMF36737,
MMF36743, MMF36743a, MMF36743b, MMF36746,
MMF36753a, MMF36758a, MMF36759,

MMF36761b, MMF36773, MMF36778.

Description

Olfactory region

The rostral is evident on several specimens,
including ANU54956 (Figure 2A) and MMF13555
(Figure 2B), and is small and almost shaped like an
isosceles triangle, with the apex anterior. The two
lateral margins are slightly concave, and the posterior

supraorbital
nasal
rostral

supraethmoid,
lachrymal
(infraorbital 1)
premaxilla

parasphenoid

maxilla

dentary
supramaxilla 1

supramaxilla 2
infraorbital 2
gular plate
quadrate
hypohyal 1
hypohyal 2
retroarticular
ceratohyal

frontal

parietal

extrascapulars

,dermosphenotic

,pterotic

supracleithrum

infraorbital 5
infraorbital 4
infraorbital 3
operculum

preoperculum

suboperculum
cleithrum

branchiostegal rays urohyal interoperculum

Figure 1 Cavenderichthys talbragarensis. Reconstruction of head, lateral view, based on MMF36716.

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

1mm

supraorbital

sclerotic ring
parasphenoid

supramaxilla 2

rostral

supraethmoid

premaxilla

nasal (R?)

supraethmoid

premaxilla

? nasal L
(see text)

lachrymal
supramaxilla 2

maxilla
with teeth
dentary

infraorbital 4

preoperculum

infraorbital 3
infraorbital 2
quadrate

ceratohyal

hypohyal 1
hypohyal 2

Figure 2 Cavenderichthys talbragarensis. Details of anterior head. A, this view of the front of the head of ANU54956
shows unusual detail of the anterior bones. B, the anterior of MMF13555 shows the premaxilla and nasal
bones. On the supermaxilla 2 it is possible to see traces of the path of nerves. The preservation of teeth on the
maxilla is unusually clear. The posterior of the parasphenoid can be see to be enlarged.

margin is crenulated where it articulates with the
frontal. In ANU54956 it is disarticulated from the
frontal, but the posterior margin is clear (Figure 2A).
This bone articulates with the maxilla and premaxilla
and presumably with the supraethmoid, although this
articulation has not been observed. The ethmoid has
not been observed in this material.

The supraethmoid is a small median bone that is

under the anterior part of the frontal-rostral and
probably is the ossified covering of the ethmoid
cartilage. Two specimens, ANU54956 (Figure 2A)
and MMF1355 (Figure 2B), show the supraethmoid
to be roughly Y-shaped, with the two branches
being posterior. The articulation of these branches
with other bones cannot be determined in these
specimens.

L.B. Bean

Otic region

The generally poor preservation of the otic region
results from it overlying the back of the braincase
causing it to be usually crushed, and is only known
from one specimen (Figure 7A). The pterotic is
roughly rectangular, and located dorsal to the
preoperculum and posterior to the dermosphenotic.
It carries the sensory canal where it branches off to
the preoperculum. Tire parietals are medial to the
pterotic. Only one specimen has an identifiable
pterotic (Figure 7A).

Dorsal roofing bones

A small cylindrical bone on MMF13555 (Figure
2B), just ventral to the supraethmoid, has unclear
relationships with other bones. It appears to carrv a
sensory canal, and in the flattened fossil it appears
to be adjacent to the anterior end of the
parasphenoid. This bone may be the left nasal. The
right nasal is well exposed on MMF13555 (Figure
2B), having been detached during fossilization. It is
identified on its tubular form and containing the
supraorbital sensory canal. It sits adjacent to the
anterior part of the frontal, where the sensory canal
emerges from under the ridge of covering that
protects it in the region above the supraorbital.

Frontals are the largest skull roof bones, almost
the same size as the dentary bones. They are narrow
at the front and become wider behind the eye. The

suture between them is straight in the narrow
region, and then bends back and forth in the wider
region. The frontals carry the sensory canal, with
pores occurring at the front, above the centre of the
supraorbital bone, and at the end of a branch where
the canal curves down around the eye. The canal is
close to the surface of the bone and is covered by a
ridge in some specimens in the anterior narrow part
of the frontal, e.g., MMF13564 (Figure 3B). In other
cases the delicate ridge has been removed and a
canal is visible, e.g., ANU54916, MMF36781,
MMF36735, MMF36753a (not figured). About two
thirds of the distance from the anterior of the
frontals is a prominent pore, behind which the
sensory canal branches, with one branch leading
into the parietals where it terminates at a pore
(MMF36753a, Figure 6A). The other branch turns
ventrally and passes into the dermosphenotic.
Where the frontal broadens out, it forms the margin
of the orbit between the supraorbital and the
dermosphenotic.

The parietals are generally rectangular in form
and meet medially by an irregular suture. The
sensory canal crosses them from the frontals, but
does not emerge posteriorly. The parietals occur
directly behind the frontals and anterior to the
extrascapulars.

The extrascapulars (Figure 6A) are smaller than
the parietals and are posterior to them. They carry a

supraorbital

covering
ridges

frontals

dermosphenotic
parietals

Figure 3 Cavenderichthys talbragarensis. Skull roofs. A, this photograph of a latex peel of the skull roof of MMF36728
is a rare example of this view. B, on this peel of MMF13564 it is possible to see even more detail including the
mid-line suture and several pores for the emergence of nerves.

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

ANU54916

Figure 4 Cavenderichthys talbragarensis. A, sketches of the differences in arrangements of bones in the upper jaw.

This is natural intra-species variation, and is as much a function of differences in preservation as differences
between individuals. B, sketches of the variations in arrangement of the preopercular canal on the
preoperculum. Scale bars = 1 mm.

L.B. Bean

sensory canal with several pores. They are the
posterior bones of the roof of the skull, but in many
specimens they are crushed and difficult to
interpret. Figures 3A and 3B are photos of the skull
roofs of MMF36728 and MMF13564 respectively.

Lateral skull bones

The premaxilla (Figures 2A, 2B, 4A and 5A) is
small and mobile, and because of this it is often lost
due to poor preservation. It fits into a concavity on
the front of the maxilla (see below). There are about
6 small teeth on the premaxilla, which lies adjacent
to the rostral bone and the supraethmoid,
(ANU54956, Figure 2A) but this region is not
usually seen clearly. Other non-figured specimens
showing the premaxilla include MMF36732a,
MMF36778, MMF13555, ANU54916, and

ANU54968.

The maxilla has along its ventral margin a row of
small, even teeth, which can be seen on specimens
ANU54956 (Figure 4A), ANU54976 (Figure 5A),
and MMF36729 (not figured). This margin is a
smooth gentle convex curve ventrally, with a small
arcuate toothless concavity at the front to
accommodate the premaxilla. The maxilla and the
premaxilla are certainly not fused in any way, as
the premaxilla is often detached from the maxilla,
but their articulation is not obvious. The posterior
end of the maxilla is a smooth semicircular curve
and is connected to the coronoid process of the
mandible by a flat maxillomandibular ligament
(Lauder 1980). The anterior end of the bone has a
peg-like process, which enables it to articulate
probably with the vomer, ethmoid and palatine in a
similar manner to Amia (Lauder 1980). This means
the maxilla is fixed at the anterior end and free to
swing forward and backward from the posterior.
When the mouth is agape, the maxilla swings
forward, and is often found preserved in this
position. Teeth form the margin of the gape when
the maxilla is fully protracted. When the mouth is
closed the maxilla is pulled backward, and passes
outside the dentary, coming to rest on the ridge on
the dentary formed by the heavy ossification
around the Meckelian cartilage. The outside surface
of the maxilla has a ridge running along the middle
from the anterior end. Of 142 specimens in which it
is possible to distinguish the state of the maxilla, 83
have the maxilla wide open.

There are two supramaxillae, the anterior being a
small smooth oval bone that on-laps the anterior
part of the dorsal margin of the maxilla
(MMF36732a, Figure 6B). The posterior
supramaxilla has a generally oval shaped base on-
lapping the maxilla, but it also has a long thin
slightly curved process extending anteriorly under
and above the anterior supramaxilla. This process
has a thin ridge and groove that extend from the tip
down into the body of the bone. The surface of the

body of the posterior supramaxilla has a radiating
pattern of grooves and small ridges, which appear
to represent a point of ligament attachment. The
maxilla and the two supramaxillae move as a unit
and are generally found joined together. Figures 5,
6 and 7 illustrate the various structures of the upper
and lower jaws.

Circumorbital series

The system of naming all lower circumorbitals as
infraorbitals, as used by Cavender (1970), Nybelin
(1974), Patterson (1977) and Arratia (1997) has been
used here. Confusion can arise when some bones
are not preserved, e.g., infraorbital 1 or lachrymal,
and the terminology used by Norden (1961), based
on living fish, which clearly identifies specific
bones, was used initially by the author to establish
relationships. However, the modern terminology
has been used to be consistent with contemporary
publications.

A single supraorbital bone is the anterodorsal
bone of the circumorbital series. It is a long thin
oval with a slight upwards curve to follow the
dorsal margin of the eye. The ends of the bone are
rounded and there is no sensory canal. It forms the
anterodorsal margin of the orbit (ANU54956, Figure
2A and MMF36728, Figure 3A).

The infraorbital 1, also called the lachrymal,
(MMF13555 Figure 2B) is small, forming the lower
anterior rim of the orbit, and contains the terminus
of the infraorbital canals (Norden 1961). It is a
dermal bone external to the ectopterygoid, fitting
into the series around the eye. It is narrow
anteriorly and broadens posteriorly and is almost
triangular in nature with the apex towards the front.
It carries the sensory canal, but due to the state of
preservation it is impossible to determine if it is the
site of the terminus of the infraorbital canal.

Specimens MMF36728 (Figure 5C), AMF51899
(Figure 7A) and MMF36735 (not figured) show the
continuation of the infraorbital series. There are
four bones all about the same size and depth
around the ventral and posterior part of the eye.
They all carry the sensory canal. The first one is the
infraorbital 2. In Cavenderichthys talbragarensis it has
a roughly trapezoid shape. A branch off the canal is
directed ventrally. The infraorbital 2 is not as deep
as the subsequent ifraorbitals. Its ventral margin is
level with the ventral margin of the infraorbital 1 in
front, but at the back it is about half as deep as the
infraorbital 3.

Infraorbitals 3-5, which are all about the same
depth, form the posterior rim of the eye. A suture
between infraorbitals 3 and 4 always appears in
compressed forms to be adjacent to the posterior
end of the parasphenoid. In all the specimens
illustrated in this article there is an easily identified
bone, infraorbital 3, which occurs anterior to the
bend of the preoperculum. Dorsal to this bone is

Late Jurassic leptolepid fish Cavenderichthys lalbragarensis

premaxilla

maxilla

supramaAiiici i
lateral dentary
medial dentary

displaced
supramaxilla 2

ectopterygoid

maxilla open

medial
angular

suboperculum
interoperculum
preopercular canal

ceratohyal (not
fenestrate)

urohyal

supramaxilla 1
lachrymal ?

infraorbital 2
infraorbital 3

infraorbital 2

ceratohyal

hypohyal

acinacifori
branchiosLc^di
rays 0

preoperculum
interoperculum
cleithrum

spathiform
branchiostegal rays

Figure 5 Cavenderichthys talbragarensis. A, detail of the jaws of ANU54976. Left and right dentaries are both visible,
and well preserved teeth on the maxilla. Supramaxilla 2 has been displaced but its medial ridge is clear. B,
medial view of MMF36732a, showing the ectopterygoid and bones of the circumorbital series. In this
specimen the ceratohyal is clearly not fenestrate, and the delicate nature of the urohyal is obvious. C,
MMF36728 has well preserved bones of the circumorbital series, including the lachrymal. D, in this detail of
the posterior ventral region of the head of MMF13555, the ceratohyal is partly covered by the preoperculum,
but the relationship of the two forms of branchiostegal rays to the ceratohyal is clear. Two hypohyals are
visible at the anterior of the ceratohyal.

L.B. Bean

maxilla

quadrate

ceratohyal

supramaxilla 2

angular

Imm

urohyal

ral fin
cleithrum
suboperculum
interoperculum

branchiostegal rays

parietal with terminating end of soc
hyomandibular

frontal with pores on
supraorbital canal

extrascapular

supracleithrum

angular

supramaxilla 2

maxilla with
teeth

dentary

1mm

ectopterygoid

quadrate

Meckelian

groove

angular
gular plate

Figure 6 Cavendehchthys talbragarensis. Heads showing the maxilla in the forward open position. A, this photo-
graph of a peel of MMF36753a includes a ceratohyal that is not fenestrate, and very clear pores on the
supraorbital canal. The characteristic position of the parasphenoid appearing to bisect the orbit is well
demonstrated. B, the detail of the jaws of MMF36732a is taken directly from the specimen. The location of the
Meckelian cartilage present along the interior surface of the dentary is obvious.

infraorbital 4, a squarish bone that in
Cavenderichthys is as deep as infraorbital 3. The most
dorsal infraorbital 5 (Figures 7A, 7B, 8A) is not
squarish but roughly triangular with rounded
corners, the apex pointing dorsally. These bones lie
anterior to the preoperculum and overlap it slightly.
Infraorbitals 3 and 4 have approximately
rectangular shapes with curved margins. There are
no branches of the sensory canal in infraorbitals 4
and 5 (see MMF36761b, Figure 7B, and
MMF13599b, Figure 8A).

Dorsal to infraorbital 5 is the dermosphenotic, or
infraorbital 6 (AMF51899, Figure 7A). Norden
classifies the dermosphenotic as a "small, dermal
postorbital bone, which bears a triradiate sensory
canal." In many specimens this bone is crushed and
difficult to identify, but it is possible to see in
several specimens that it carries a junction of the
sensory canal where it descends from the frontal,
continues into the highest infraorbital, and branches
posteriorly towards the pterotic. It makes part of
the posterior rim of the orbit.

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

post temporal

maxilla

basipterygium

parietal

infraorbital 5 /

interneural rods

neural spine

dermosphenotic
supraorbital

haemal spine

preoperculum

1cm

infraorbital 5

infraorbital 4
infraorbital 3

operculum

sub-

operculum

interoperculum
ceratohyal

tmm

branchiostegal rays

pectoral fin

Figure 7 Cavenderichthys talbragarensis. Heads showing the jaws closed. A, F51799, infraorbitals 3-5 can be identified
as well as the pterotic and dermosphenotic. In the pleural region the vertebral column is well preserved and
interneural bones are visible. The basipterygium can be seen supporting the pelvic fin. B, on the specimen
MMF36761b the bones of the opercular series and the infraorbital bones can be easily identified. The
spathiform branchiostegal rays are seen posterior to the ceratohyal. The path of the preopercular canal
clearly shows five branches.

Preopercidum

A commonly preserved bone is the
preoperculum, well seen on MMF36730,
MMF36759, ANU54916, ANU54968 and ANU54940
(see Figure 4B). It lies behind the infraorbital series,
and is not part of the opercular series. The bone is
arcuate, with an angle of approximately 110°
between the dorsal and ventral limbs, and it carries
a prominent branch of the sensory canal. The dorsal
margin of the upper limb is often crushed. The
postero-ventral margin is smoothly curved, making

a more acute angle than the anterior margin. The
anterior margin of the ventral limb extends to a
point lateral and just posterior to the articulation of
the lower jaw. The sensory canal is carried close to
the outer surface of the bone, and continues onto
the mandible. When the external surface is
preserved it is apparent that the canal makes a ridge
on the surface, with small pits at the ends of side
branches opening to the surface. Often the canal is
preserved by infilling with fine-grained white
material, indicating that it was closed at the time of

L.B. Bean

death and has subsequently been filled with
material that is different from the typical matrix.
The canal is located closer to the anterior margin
than the posterior, with a series of branches running
down towards the posterior margin, on average six
branches, sometimes seven and occasionally five.
There is usually only one branch on the dorsal limb,
about halfway up, with the others evenly spaced,
three along the ventral limb and two adjacent to the
bend. The branches are broad and tend to widen
away from the main canal. Illustrated examples of
the preoperculum are Figure 4B, MMF13555 (Figure
5D), MMF36732 (Figure 5B), and MMF13599b
(Figure 8A).

Suprapreoperculum and suborbital bones are
lacking.

Opercular series

The shape of the operculum is roughly triangular
with the most acute angle dorsal. The top is
sometimes involved in the crushing at the back of
the braincase, but when well preserved it can be
seen to be smoothly rounded. The anterior margin
is approximately perpendicular to the line of the
vertebral column. The posterior margin is convex
posteriorly. The ventral margin is straight and
inclined down towards the front, and overlaps the
suboperculum.

The suboperculum has the appearance of being
an isolated continuation of the operculum (Figures
6A and 7B). The anterior and posterior margins
follow the same lines. The ventral margin of the
suboperculum follows the line of the ventral surface
of the fish and overlaps the branchiostegal rays. It is
adjacent to the cleithrum.

The interoperculum is a triangular bone, ventral
to and partly hidden by the preoperculum. It often
overlaps the branchiostegal rays, and is adjacent to
the suboperculum. It can be seen in specimens
MMF13555 (Figure 5D), MMF36732a, MMF36728,
ANU54976, ANU54983, MMF36761b (Figure 7B),
MMF36753a (Figure 6A), and MMF36735.

Six thin acinaciform branchiostegal rays are
regularly associated with the ceratohyal,
presumably with a ligamentous connection between
them. They leave the ceratohyal ventrally then
curve around to point posteriorly (Figure 5D). They
extend toward the cleithrum and end as thin points,
floating freely. The left and right rays are separated
by the Y-shaped urohyal.

There are three or four spathiform
branchiostegal rays that also seem to be connected
to the epihyal (MMF13555, Figure 5D and
MMF36761b, Figure 7B). They are dorsal to the
thin rays but curve around in a similar fashion and
parallel the base of the suboperculum. This
maximum number of 10 branchiostegal rays is less
than the 12 or 13 counted by Arratia (1997) in her
diagnosis of Cavenderichthys. This may be because

sometimes the rays from both the left and right
sides are visible.

Palatal bones

The generally poor preservation of the front of
the head means that it has not been possible to
identify the vomer, which should be a median
toothed bone between the two premaxillae. In
comparison, the appearance of the parasphenoid
bisecting the orbit is one of the typical features of
Cavenderichthys talbragarensis, e.g.. Figures 2A,
2B, 5B, 5C, 6A, 7 A, 8A, 8B. The parasphenoid is a
median dermal bone that forms the roof of the
mouth, so when the fish is laterally compressed it
appears across the large eye socket. This bone is
thin dorso-ventrally, and it is slightly concave
dorsally. The rear of the parasphenoid is generally
hidden behind the postorbitals, so details of its
relationship with the brain case cannot be
determined. One figure of MMF13555 (Figure 2B)
shows an expanded posterior end. The anterior end
extends to the front of the head and probably
articulates with the vomer. The line of the
parasphenoid is inclined upwards towards the
front, making an angle of about 120° with the line
of the spine.

Mandible

The dentary is the major bone of the lower jaw;
see Figures 2B, 5A, 9A, 9B and MMF36730,
MMF36733, MMF13734a (unfigured). The

Meckelian canal which carries the Meckelian
cartilage lies close to the internal surface, and
adjacent to this on the external surface there is a
prominent ossification of the dentary covering the
sensory canal, making a ridge along the external
surface. This ridge is sometimes removed or
damaged, leaving a deep canal. If the surface is
preserved there are seen to be two or three pores
along the path of the canal, as shown by Arratia
(1997: 21). The front of the dentary always
protrudes beyond the line of the maxilla, whether
the gape is open or closed. The ventral margin of
the dentary forms a smooth gentle curve, passing
posteriorly into the angular. The anterior margin of
the dentary is tightly rounded. The anterior-dorsal
margin is slightly inclined and sometimes carries
about 6-10 small teeth. These are not commonly
seen, even when the maxilla teeth are clearly
preserved. They may be just very small, or possibly
they are not always present. Posterior to the toothed
surface the margin of the dentary forms a
downward notch before rising steeply to the
coronoid process (Figures 9A, 9B). This notch is
equivalent to the "leptolepid notch" referred to by
Nybelin (1974), Patterson (1977), and Arratia (1997)
(see later comparison with other species). The notch
is smoothly curved, is higher on the internal surface
and makes a channel down towards the outer

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

pores on suprorbital canal

infraorbital 5

infraorbital 4
infraorbital 3

preoperculum

frontal covering over

supraorbital canal

parasphenoid

vertebral column

pores on

supraorbital canal

angular

maxilla open
supramaxilla 2
dentary

gular

ceratohyal

hypohyal

pectoral fins

Figure 8 Cavenderichthys talbragarensis. A, MMF13599b, some detail of the roof of the skull is seen, including pores
on the supraorbital canal. Infraorbitals 3-5 are clear. B, this photograph of a peel of MMF36732a shows
uncommon detail of the pectoral fins, the structure and degree of ossification of the anterior vertebrae, and
the pores on the supraorbital canal. The gular plate is preserved below the dentary, and the ceratohyal not
fenestrate.

surface. It is much wider that the notch illustrated
by Nybelin in Leptolepis coryphaenoides, but it occurs
in the same location. The dorsal margin of the
coronoid process is formed anteriorly by the
dentary and posteriorly by the angular. The angular
and dentary are always found closely associated,
together with the retroarticular.

The angular fits into a V-shaped notch on the back
of the dentary and can be seen on MMF13734a

(Figures 9A, 9B). It is thin dorsally and thickens
ventrally adjacent to the Meckelian canal. The
postero-ventral corner, which incorporates the
fused articular process, articulates with the
quadrate. The angular makes up about the posterior
third of the ventral margin of the mandible. There
is not sufficient information to determine a true
relationship between the angular and articular due
to the lack of exposure of this area.

L.B. Bean

dentary

notch

coronoid process angular

\ articulation with

quadrate

dentary | retro-articular

Meckelian
canal

Figure 9 Cavenderichthys talbragarensis. A, lateral view of the left mandible of MMF137a, drawn from Figure 21E. B,
medial view of the right mandible of MMF13734a, drawn from Figure 21 D. C, medial view of the left
hyomandibular of MMF13734a, drawn from Figure 21B. D, lateral view of right hyomandibular of
MMF13734a, drawn from Figure 21C. E, F, G, outlines of the hyomandibulars of Leptolepis normandica. FI, I,
Leptolepis coryphaenoides (Nybelin (1974) figure 3] included to provide a comparison with the structure of
the hyomandibular of Cavenderichthys talbragarensis as seen above in D. The presence and location of the
opercular process (pr.op) and the preopercular process (pr.pop) are very similar.

The retroarticular is a small bone, possibly fused
to the posteroventral margin of the articular and
forming part of the articulation of the lower jaw.

Palatoquadrate arch

The ectopterygoid, which can be seen on
MMF36732a (Figure 6B) and MMF36732 (Figure
5B), forms the anterior sidewall of the mouth. It is
boomerang shaped, thin anteriorly and thickens
posteriorly where it articulates with the quadrate.

The dorsal margin is convex while the ventral
margin of the ectopterygoid is concave, does not
articulate with anything and is usually covered by
the upper jaw.

The quadrate can be seen on MMF36753a (Figure
6B), MMF36732a, ANU54962 and ANU54977 and
is a fan shaped bone that forms part of the
articulation with the mandible. The dorsal part is
thin and spreads out like a quarter of a circle. The
apex of the bone, which is ventral, thickens into an

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

articulating post which articulates with the back of
the angular-articular and the retroarticular
process.

Hyoid arch

Of the hyoid arch, the hyomandibular bone is
only rarely visible as it is generally covered by the
preoperculum. In ANU54976 the dorsal end of the
bone appears in the orbit, but in the dorso-ventrally
flattened MMF13734a, both hyomandibulars are
visible, see Figures 9C, 9D, and later in text the
photographs of Figure 21. The bone is long with a
central shaft, a thickened dorsal end and a thin
expanded ventral end. The antero-dorsal margin
projects anteriorly from the shaft, then curves
smoothly to a dorsal point, resembling the shape of
an axe head. The postero-dorsal margin curves
concavely down to an opercular process. The
postero-ventral margin thins out behind the base of
the shaft into a thin flattened preopercular process
with a curved margin. The ventral margin of this
process extends in a straight diagonal line down to
a point anterior of the shaft, and then curves back
up to meet the shaft just below where the posterior
part of the process intersects the shaft. The
preopercular process is very thin and not likely to
be preserved (or visible) in many cases. It is only
due to the dorso-ventral flattening of just one
specimen (MMFl 3734a) that the details of both the
left and right hyomandibulars can be seen. A clearly
visible foramen on the outer surface of the upper
part of the bone allows passage of a branch of the
facial nerve (Norden 1961), see Figures 9C, 9D.

It has not been possible to find sufficient evidence
to describe the symplectic.

The epihyal, seen on ANU54976 (not figured), is
usually covered by the preoperculum. It is a flat,
roughly rectangular bone with a slightly concave
anterior margin that articulates with the slightly
convex posterior margin of the ceratohyal. The
ventral margin is slightly convex. It appears to be
associated with 3 or 4 spathiform branchiostegal
rays.

The ceratohyal is often preserved and quite
clearly visible, for example on MMF13555 (Figure
2B), MMF36732a (Figure 5B), MMF36761b (Figure
7B), MMF36732a peel (Figure 8B) and MMF36735
(not figured). The ends of the bone are flattened
and splayed out with curved margins. The ventral
margin is smoothly concave, and bears the
attachments of the six thin branchiostegal rays. The
dorsal margin also has a smooth, deep, sometimes
semicircular curve. In this species the ceratohyal
does not appear to generally be fenestrate. In
several specimens (ANU54980, MMF36743,
MMF13606a, MMF36721,) there may be a thin rod
connecting the anterior and posterior dorsal
margins, but in each case it has been broken. More
commonly the ceratohyal resembles the shape of the

archetypal dog's bone, e.g., MMF36732a (Figure
5B), MMF36753a (Figure 6A), MMF36761b (Figure
7B), MMF36781, MMF36743a and MMF36728.
Cavender (1970) suggested that the rod-like
connection between the dorsal ends of the
ceratohyal may be present on the larger specimens.
Many small specimens show that the ceratohyal is
definitely not fenestrate {contra Arratia 1997) as the
dorsal margins can be clearly seen and there is no
sign of a rod connecting anterior to posterior.

Two very small hypohyal bones can be seen on
MMF13555 (Figure 2B), MMF36732a (Figure 8B),
MMF36729, and ANU54983, fitting into the slightly
concave anterior surface of the ceratohyal. They
occur one above the other, and would have been
connected to the ceratohyal by cartilage. They
appear as if they may be conical, with the upper
one fitting over the apex of the lower one. Other
forms of fish have only one hypohyal per
ceratohyal, but in leptolepids, having two
hypohyals is not uncommon. It is also the usual
situation in modern Salmonidae.

Branchial arches

No observable articulated branchial arches have
been preserved in these specimens, but isolated
bones are identified below.

The urohyal is a very delicate median bone (see
ANU54982 ,"aNU 54983 and MMF13603k) with a Y-
shape, where the branches are pointing posteriorly.
The urohyal is located between the two sets of
branchiostegal rays (MMF36732, Figure 5B and
MMF36753a, Figure 6A). Its relationship with other
bones, apart from the branchiostegal rays, cannot be
determined.

The gular plate, which is seen on MMM36732a
(Figures 6B, 8B), ANU54982 and ANU54983, is a
small narrow median plate found between the two
lower jaws. It is quite often visible with its anterior
margin hidden behind the dentary, but its posterior
margin pointing downwards. This is the case when
the ceratohyal is also visible below the jaw-line. The
gular plate has continuous growth lines outlining
the elongated oval shape. It probably increases in
length, but not greatly in width, as the individual
grows.

Vertebral column

The number of centra is highly variable. Of the 48
peels with centra that can be counted reliably, there
is a range from 35 to 45 centra (cf. 43-45 recorded by
Arratia 1997). The mean value is 40.4 with a
standard deviation of 3.5. There is also no
relationship between length of the individual and
number of centra. Each centrum is a thinly ossified
cylinder, constricted in the middle. The centra in
the abdominal region have a small process on each
side for the attachment of the ribs.

The last two upturned centra in the tail are called

L.B. Bean

centrum showing process
for attachment of ribs

pore on parietal

pore on frontal

covering of supraorbital
canal

parasphenoid

spine on
supramaxilla 2

ceratohyal

operculum preopercular
canal

two halves of
dorsal fin ray

pterygiophore

ineurals

epipleural

Figure 10 Cavenderichthys talbragarensis. A, pectoral fins of ANU54938, ventral view. B, peel of F27069 showing
detail of the vertebral column. Vertebra turned on its side shows two small processes for the attachment of
the ribs. Just ventral to the spine in the pleural region are small epipleural bones. The epineurals are visible
detached from the dorsal side of the spine. On the pelvic fins distal branching of the rays can be seen.

Ural centra. Anterior to this the centra that carry
haemal arches are called preural centra.

Epineural bones come off the neural arches that
carry ribs, anterior to the dorsal fin, see AMF27069
(Figure JOB), MMF36743b (Figure 16B), and
AMAMF4133. They point backwards at an angle of
about 20° above the column. They occur on each
side of the column because they are inter-muscular
bones that support the flesh of the fish (Gosline
1971).

Anterior to the dorsal fin are intemeural bones,

which are "a series of median supporting rods"
(Norden 1961: 689) extending from just below the
dorsal surface of the fish to the level of the neural
arches, see MMF36743b, AMF4133. They are thin
rods whose proximal tips fit between the tips of the
bifurcating neural arches (AMF51899, Figure 7A).

The neural arches are paired bones that articulate
with the centra and form an arch through which the
spinal cord passes. They extend dorsally
approximately half way to the dorsal body margin.
In the abdominal region it is often possible to see

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

the two separate bones that form the arch, but
behind the dorsal fin the bones of fhe neural arches
are fused, forming neural spines. They are well
ossified.

The haemal arches (ANU54970, Figure 16A) are
composed of a pair of bones thaf fuse together to
form haemal spines. The juncfion of the ural and
preural centra is marked when the haemal artery is
no longer passing through a haemal arch.

The number of ribs is variable as is the number of
centra. Of the 49 specimens where ribs can be
counted, the mean number is 23 with a standard
deviation of 2.3. The range is from 20 to 26 pairs of
ribs.

In a couple of specimens there appear to be
epipleurals (AMF27069, Figure lOB), which are
small fine bony projections originating near the
point of attachment of the ribs and extending
backwards to the level of the next rib. This has only
been seen in the pleural region. Arratia (1997) stated
that Cavenderichthys lacks epipleurals.

Caudal fin

The caudal centra are 3 upturned vertebrae in the
caudal region, comprising preural centrum 1 and
the ural centra, which is a characteristic of
coregonines, salmonines, and thymallines (Norden
1961). The preural centra are well ossified, each
consisting of a ring with a large circular canal in the
middle. The haemal and neural arches always
appear to be attached to the centra. The first preural
centrum is slightly smaller than the subsequent
ones. The first ural centrum appears to be fused
with the second, as it is longer than other centra but
has a smaller diameter. The first and second
hypurals, which articulate with Ul-2, appear to be
fused, or at least closely associated.

The principal rays of the caudal fin form a double
series of rays, which sandwich the ends of the
hypurals, to which they are attached (Figure 12C).
The two outer principal rays are segmented, but do
not divide. The inner rays branch progressively
earlier toward the centre of the tail. As the rays

Figure 11 Cavenderichthys talbragarensis. A, F4133 used for the reconstruction of the skeleton of the whole fish in B.

L.B. Bean

Figure 12 Cavenderichthys talbragarensis. Comparison of reconstructions of the caudal skeleton. A, Patterson and
Rosen (1977) figure 46 reconstructed from BMNH P37973. B, Patterson and Rosen (1977) figure 46
reconstructed from BMNH P12439. C, my reconstruction of MMF13561a, showing species defining
similarities, but some small inter-species variation.

branch they become more and more delicate, so the
complete fin is preserved in relatively few
specimens. They branch up to four times, such that
the middle section of the tail, where branching
occurs early, is very thin, delicate and flexible. The
principal rays articulate with the hypurals in the
upper section, and in the lower section PR 19
articulates with pre-hypural haemal spine 2
(parhypurals), PRs 18 and 17 articulate with pre-
hypural haemal spine 1, then PRs 16, 15, 14, 13 and
12 articulate with hypural 1-2 (combined unit), and
PR 11 articulates with hypural 3. The proximal ends
of several PRs are broad where they articulate with
the hypurals, e.g., on MMF36746 (Figure 14A) the
ends of PRs 8, 9 and 10 are enlarged. In contrast, the
proximal ends of PRs 1-6 taper to a point.

The neural spines on the preural centra are
narrower than the haemal spines. They also become
progressively shorter posteriorly. The neural spines
on preural centra 2, 3 and 4 bend back strongly, and
they overlie the short neural spine on preural
centrum 1.

Tliere are three epurals, rarely four, which extend
from the ends of the long neural spines to the bottom
of the basal fulcra, covering the dorsal surface
between the caudal scute and the first principal ray
(Figure 12C and MMF36773, Figure 13B).

Dorsal to the epurals lie the epaxial basal fulcra
(MMF36773, Figure 13B), usually about six, which
form a series with the three procurrent ravs
between the principal rays and the anterior caudal
scute. Sometimes it is not easy to distinguish

Late Jurassic leptolepid fish Cavenderichthys lalbragarensis

Figure 13 Cavenderichthys talbragarensis. Photographs of the original specimens showing the layered nature of the
caudal skeleton. Peels were not made due to the risk of destroying some of the delicate overlying structures,
especially where the hypurals overlie the rays. A, MMF36718. B, MMF36773.

between the basal fulcra and the unsegmented
procurrent rays. There are also several ventral basal
fulcra.

The caudal scute marks the junction between the
dorsal surface of the fish and its tail (Figure 12C
and MMF36773, Figure 13B). It is similar in size to
the six dorsal basal fulcra, but slightly sinuous in
shape, and broader. The ventral caudal scute is
similarly a sinuously shaped bone.

Generally there are three uroneurals that lie down
either side of the ural centra. They act as bracing for
the thinning end of the vertebral column. They
project anteriorly beyond ural centrum 1 and
overlap preural centrum 1 and 2, effectively
crossing the "chondrostean hinge". Usually one,
sometimes two more, short uroneurals project
backwards from the posterior of ural centrum 1-2.

These five uroneurals make a functional continuous
series from the epurals through uroneurals to the
principal rays in the upper lobe. These can be seen
on MMF36773 (Figure 13B), MMF36746 (Figure
14A), MMF36732a (Figure 15B), and ANU54975
(Figure 15A).

There are seven hypurals (MMF13561a, Figure
12C). The lowest two hypurals articulate with the
first ural centrum 1-2. They often appear to be
fused. Hypural 3, which articulates with ural
centrum 3, carries the inner rays of the upper lobe,
i.e. principal rays 8, 9, and 10. These rays often have
thickened ends where they attach to the hypural.
This probably gives strength to the fragile but
flexible inner rays. In most cases the upper four, or
possibly five, hypurals are covered by the
overlapping ends of the upper principal rays.

L.B. Bean

urodermal

preural haemal arch 1
(parhypural)

Imm

hypural 6

ural centrum 1-2

broad end
PR10

1mm

caudal scute

uroneurals

Figure 14 Cavenderichthys talbragarensis. These photographs of peels support the caudal skeleton reconstruction in
Figure 12 C. A, MMF36746. B, MMF36748.

Urodermal bones are not generally preserved.
However in three specimens, AMF4133, MMF13569
and ANU54975 (Figure 15A), there appear to be
two small fine bones that are not part of the
segmented principal rays in the dorsal part of the
fin. They overlap two principal rays, either 2 and 3
or 3 and 4, and may provide extra bracing at the
end of the vertebral column. According to Cavender
(1970), "they are clearly ossifications at the distal
ends of two of the tendons that originate from the
superficial epaxial muscle mass in the caudal region
and which insert proximally on the dorsal edges of
the first and second branched rays respectively."
Maybe they are not always ossified, or possibly they
are not always preserved because they lie outside
the general caudal fin structure (see also
MMF36746, Figure 14A).

Median fins

The dorsal fin is positioned over the mid point of
the body, see AMF27069 (Figure lOB),
MMAMF4133, ANU54946 and MMF36743. The
anterior margin of the dorsal fin is about level with
the mid-point of the vertebral column. The
posterior margin is level with the anterior margin
of the anal fin. The length of the fin, measured
along the line of the articulation, is 10-12mm. This
position in the mid-portion of the fish indicates a
role as keel (Gosline 1971).

The dorsal fin consists of 12 to 15 rays, the
anterior three being short procurrent rays that do
not branch, the first principal ray being full length
and unbranched, and the rest being full-length rays
that branch into lepidotrichs about halfway along
their length. The rays are supported by internal

Late Jurassic leptolepid fish Cavenderichthys talbragarensis
series of 5 uroneurals

two urodermals

1mm

hypurals

8 basal fulcra

5 uroneurals

6 basal fulcra

parhypural on PU1

haemal spine on PU2

Figure 15 Cavenderichthys talbragarensis. A, the peel of ANU54975 shows two urodermals and the series of
uroneurals. B, on MMF36732a the dorsal and ventral basal fulcra are clearly preserved.

bones called pterygiophores (MMF36718, Figure
13A), also called radials, which are described by
Norden (1961) as being "composed of 3 segments, a
small, rounded distal bone, a short, horizontal
middle bone and a long, pointed proximal bone." It
is possible to see both the proximal and horizontal
pterygiophores in AMF27069 (Figure lOB). The
anterior proximal pterygiophore points forw'ard,
lying at an acute angle to the dorsal surface, and is

forked. It supports the anterior procurrent rays. The
pterygiophores supporting the middle of the fin are
perpendicular to the dorsal surface. They are fine,
pointed rods, but they broaden dorsally where they
articulate with the horizontal elements. The
pterygiophores are single structures, which appear
in the mid-line, dorsal to the neural arches.

The fin rays of the dorsal fin, like the
pterygiophores, are fine, emerging from the broader

L.B. Bean

anal fin

haemal arches

pelvic rays

pelvic fin rays

dorsal lepidotrichs

pelvic fin rays

dorsal pterygiophores

epineural bones

basipterygia

Figure 16 Cavenderichthys talbragarensis. Pelvic fins. A, in this detail of ANU54970 12 pelvic fin rays can be seen, and
the structure of the hemal arches. B, on MMF36743b the pelvic fin is in situ, directly under the position of
the dorsal fin.

ball-like articulating surface that fits over the middle
horizontal pterygiophore. They are composed of two
half cylinders which are held tightly together along
their length. The rays gradually expand slightly
along their length until they bifurcate about half way
along the total length of the fin. Beyond this
bifurcation, the lepidotrichs are jointed, and bifurcate
once more, so that at the back of the fin there are four
lepidotrichs for every ray.

The anal fin is located half way between the
pelvic fins and the caudal fin, see Figure llA and B
and MMF36737, MMF36743b (unfigured). Its
anterior margin is level with the posterior margin
of the dorsal fin. It is always supported by nine
pterygiophores, or radials, which are single bones.

They are narrow bones that broaden toward the
proximal end. The ten anal fin rays are paired bones
that broaden where they articulate with the
pterygiophores. With the exception of PRl, which is
unbranched, the principal rays also branch into
lepidotrichs, which are sometimes preserved,
resulting in four lepidotrichs for every original ray.
A short unbranched ray is at the anterior margin of
the anal fin.

Pectoral girdle

The post-temporal is often not preserved. It has
been seen as a squarish bone that comprises the area
between the extrascapulars and the scales (Figure
7A). It carries the lateral line canal onto the head.

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

and in F51799 (not figured) it appears to have a
canal running diagonally across the bone. The bone
is posterior to the extrascapulars, and dorsal to the
supracleithrum.

The supracleithrum is the dorsal bone of the
pectoral girdle (Figure 6A). It is a narrow bone that,
when fossilized, crosses the position where the
vertebral column emerges from the back of the
braincase. The dorsal margin is rounded, while the
ventral margin abuts or slightly overlaps the top of
the cleithrum.

The cleithrum (Figures 6A, lOA) is the largest
bone of the pectoral girdle. It extends from just
below the level of the vertebral column in a curve,
to end at the level of the branchiostegal rays. The
bone is quite deep, having a strong ridge on the
external surface. The bottom of the operculum and
suboperculum are sometimes crushed over the
cleithrum, making a deep hollow and a high ridge
at the postero-ventral margin of the head.

In the 25 specimens examined for the pectoral
girdle no bone could be described as a post-
cleithrum.

The scapula is a short stout bone that articulates
with the coracoid anteriorly and with the first
principal ray of the pectoral fin posteriorly. It is
often covered by the ventral margin of the
cleithrum.

The coracoid occurs at the antero-ventral margin
of the cleithrum. It appears to be a flattish bone,
slightly pointed anteriorly and broader posteriorly,
which articulates with the cleithrum and the
scapula. The articulation with the cleithrum has not
been observed clearly, although the coracoid bone
is always adjacent to the cleithrum. The articulation
with the scapula can be seen in specimen
ANU54940 (not figured).

Pectoral fills

There are 12 rays in each of the pectoral fins
(Figures 6A, 7B, 8B, lOA). The first ray is more
robust than the others and has a broad head that
articulates directly with the scapula. The other 11
rays are bent over nearly at right angles at the
proximal ends where they form delicate points to
enable them to articulate with the three radials. The
radials are two or three short straight rods that start
near the scapula and extend just far enough to allow
the rays to articulate with them. The direction of
the radials is the same as the direction of the rays,
which explains why the rays have to turn at right
angles to be able to attach to the radials. The radials
are visible in specimen MMF36758a, but are
generally not preserved. The rays themselves are
unsegmented for about half their length, after
which they become segmented then divide into
lepidotrichs. The two pectoral fins originate very
close together on the ventral margin of the body,
just behind the head. This position is much lower

than some more recent fish, but is similar to other
leptolepids.

Pelvic girdle

Norden (1961) described the basipterygia as "a
pair of triangular endochondral bones which form
the pelvic girdle." In Cavenderichthys
talbragarensis, each basipterygium seems to widen
posteriorly where it articulates with the rays of the
pelvic fins (Figure 16B).

Each pelvic fin has 12 rays (Figure 16A). They
also turn around at right angles at the proximal end.
The rays are unsegmented for the first half of their
length, then they become segmented and also
divide into lepidotrichs.

Scales are cycloid with concentric growth rings.
They are generally only visible near dorsal and
ventral body margins, their delicate nature,
combined with the well ossified spinal structures,
leading to their poor preservation.

DISCUSSION

The early work

Woodward (1895) classified the most common
fish in the Talbragar assemblage as a member of the
Family Leptolepididae, genus Leptolepis. He noted
its general characteristics, but commented that no
detailed osteological synopsis had been published
for the genus. He described many of the features
already listed in this paper, such as the narrow roof
of skull between the eyes, the wavy suture
separating the frontals, the small parietals, the
minute premaxilla, and the two elongated
supramaxillary bones. He did say that the sclerotic
ring is ossified, but this has only been observed in
ANU54956 (Figure 2A) and MMF13555 (Figure 2B).
In the majority of specimens there is no evidence
for a sclerotic ring. He did not describe a gular
plate, but it has been now observed in 20 specimens.
When describing the ceratohyal he said it has "the
ordinary hour-glass form, but is noteworthy for the
extension of a supplementary, delicate, straight rod
of bone between its extremities on the upper
side."(Woodward 1895: 20) This comment is very
similar to the description in this paper, and less
aligned with the comment by Arratia that the
ceratohyal is always fenestrate (Arratia 1997: 19).

Woodward (1895) also described three species of
Leptolepis: L. talbragarensis, L. lowei and L. gregarius.
Leptolepis talbragarensis is by far the most common
species. Wade (1941:83) wrote "Previously three
species have been ascribed to the genus from this
locality, but in the view of the uniformity of the
structure of the head the writer is convinced that
only one species, Leptolepis talbragarensis, need be
recognised, the specific differences previously
relied on being due to individual peculiarities or

L.B. Bean

mode of preservation or differences in maturity."
Further examination during the current study of the
bones of L. gregarius demonstrates it is just a
schooling, juvenile form of talbragarensis.
Woodward describes L. gregarius as having a
different proportion of the head, and the anal fin
being slightly more anterior, but these
characteristics are not supported by a statistical
analysis of the dimensions. The two specimens
called L. lowei are slightly more elongated in the
head region than L. talbragarensis, but the rest of the
skeleton, including the tail, is the same as
talbragarensis. It is here considered that they belong
to the same species, but are differently preserved.

Wade (1941) did not add any detail to the
description of Leptolepis talbragarensis, but he did
reclassify the other genera present in the
assemblage. Cavender (1970) used Leptolepis?
talbragarensis as an example of an early teleosf and
compared it with genera from the coregonines and
other salmonids. His description of L? talbragarensis
is quite detailed and it has been the reference
description for subsequent workers including
Arratia (1997), but it formed part of a comparison
and was not a systematic description.

Nybelin (1974) reassessed the classification of L.
talbragarensis noting that it possesses some
characteristics "unfamiliar in the true leptolepids.
The deep body and the configuration of the head
with its short snout are rather unlike the
leptolepids. The anterior part of fhe fronfal, anterior
to the exit of the supraorbital sensory canal, is
strikingly short, much shorter than in the
leptolepids, and seemingly ending at the same level
as the nasal."(Nybelin 1974: 170) The description of
the body as deep is very imprecise, and when
compared with truly deep-bodied fish such as
Aethiolepis mirabilis it is seen as being totally
inappropriate. Both Arratia (1997) and Cavender
(1970) continued the pattern of describing
Cavenderichtbys talbragarensis as deep bodied, but
such a description is misleading, as even the adults
are fusiform. The vast majority of examples are
small, thin, fusiform fish, with the dorsal and
ventral margins being approximately parallel. The
fish has a short snout with a protruding lower jaw,
a centrally placed short dorsal fin, pectoral fins fhat
are placed low, a centrally placed low pair of pelvic
fins, short anal fin positioned just posterior to level
of dorsal fin, and a delicate, symmetrical, bifurcated
tail.

Nybelin (1974) considered the most significant
osteological features of fhe frue lepfolepids are the
presence of a preopercular process on the
hyomandibular and the notch in the ascending
anterior margin of the dentary. Neither he nor
Cavender (1970) saw these features on L.
talbragarensis. Nybelin (1974) considered that the
most significant difference between L. talbragarensis

and the true leptolepids is in the sensory canal
system - few branches from the infraorbital sensory
canal, and the 6 branches on the preopercular canal
is smaller than the number of branches in other
leptolepids. There are similarities between L.
talbragarensis and Leptolepides sprattiformis, but
the differences in the caudal skeleton were
considered to be the most striking difference. "The
presence of four epurals, only a single neural arch
on Ul, all uroneurals mutually free, and two
urodermals, separate talbragarensis and
sprattiformis to such a degree that a close
relationship between them seems definitely
excluded." (Nybelin 1974: 171) Thus the species
"L." talbragarensis was removed from the family
Leptolepididae.

Points of Contention in Understanding the
Morphology

The illustrations of Cavenderichtbys
talbragarensis by Arratia (1997) are refinements of
those published by Cavender (1970), and I agree
with the changes she has made to his description by
including a gular plate and an interoperculum.
Arratia maintained the pattern developed by
Nybelin (1974) and endorsed by Patterson
(Patterson and Rosen 1977) of there being no
connection between the suborbital canal and the
supraorbital canal (contra Cavender 1970). This
interpretation is disputed for several reasons.
Firstly, this part of the skull is notoriously poorly
preserved, and few specimens show clear
identification of the dermosphenotic. Secondly, the
branch of the supraorbital canal that heads ventrally
on the frontal seems to be headed directly for the
position where the dermosphenotic ought to be in
most specimens. Thirdly, in modern teleosts the
connection between the supraorbital canal and the
suborbital canal is the normal condition. In no
specimen in which both canals are visible, are they
nof connected.

Arratia (1997) stated that the ceratohyal is always
fenestrate, whereas many specimens studied herein
show no sign of a connection between the dorsal
projections of the ends of the ceratohyal, neither
bone nor cartilage. However, there are also several
examples showing evidence of such a connection,
such as ANU 54980 and ANU 54976, which have an
incomplete link across the dorsal margin of the
bone. Thus I presume it is possible for this bone to
be fenestrate, as is the common condition in related
species, but it is definitely not present in all
samples.

Observations of the hyomandibular and the
dentary in the dorso-ventrally flattened specimen
(MMF13734a) suggest that C. talbragarensis does
have a preopercular process on the hyomandibular
and also has a significant dent on the rising anterior
surface of the dentary that corresponds with the

Late Jurassic leptolepid fish Cavcnderichthys talbragarensis

leptolepid notch. The lack of these two characters
has been considered significant by other workers,
and led Nybelin (1974) to argue that C talbragarensis
could not be a member of the family
Leptolepididae. The absence of a suborbital bone,
which is present in the genus Leptolepis but not in
later leptolepids, is here considered more
significant than the absence of a small process on
one bone. On the basis that it has been found that
Cavenderichthys talbragarensis does have these
characteristics, it is now felt that it should be
included in the family Leptolepididae.

Description of the caudal skeleton largely agrees
with the illustrations provided by Arratia. She
showed the intra-species variation in the number of
epurals (three and four), and uroneurals (six and
seven), but the eighth hypural is doubtful (only
seven hypurals have been observed in this study).

The dorsal fin has a range of the number of fin
rays and pterygiophores. Considering fin rays, five
specimens have 10 principal rays, four have 11, six
have 12, and one has 13. Considering
pterygiophores, one has 10, one has 11, twelve have
12, and two have 13. Many specimens do not have
all rays readily visible. Arratia's diagnosis of
Cavenderichthys noted 15 dorsal rays, corresponding
to 12 principal rays plus three unsegmented rays.
This is certainly the most common case in my
specimens, but not the only possibility.

The numbers of rays in the fins agree with other
descriptions, but wide variation is found in the
number of vertebrae (range 34 to 45 centra.

excluding the last two ural centra). The number of
vertebrae is not related to the length of the
specimen, but it is closely related to the number of
pairs of ribs. Thus there is variation both within the
abdominal vertebrae and the precaudal vertebrae.

An analysis was made of the structure of the
dorsal fin. The angle of elevation of the dorsal fins
can vary from 70° to 15°, without any sign of
disarticulation or destruction of the joints. Of 43
specimens for which dorsal fins were measured,
three had an angle of elevation of >60°, 25 were
between 60° and 30°, while 15 were collapsed to
less than 30°. This indicates that the fin could be
collapsed when travelling fast, and raised like a keel
when cornering. According to Gosline (1971) this is
a characteristic of teleosts, that "the whole fin can
be raised or lowered, more or less like a partially
collapsible fan. The soft rays may also be swung
from side to side by the contraction of muscles
attached to the base." (Gosline 1971: 23). The size of
the fin is comparatively small and takes up less than
1/5 the total length of the fish.

Arratia (1997: 19) argued that C. talbragarensis is a
"deep-bodied" fish. In order to test this,
measurements were made of its depth to establish
the ratio of depth to length. The ratio of the average
depth to average length is about 1:4.8. The results are
summarised in the following graph (Figure 17). The
straight-line relationship between length and depth
suggests that it is determined only by the stage of
growth. The graph is based on data from 86
specimens. The regression equation is y = 0.25x - 0.3.

Length vs depth

Figure 17 Graph of relationship between length and depth.

L.B. Bean

The concentration of data in the lower left hand
corner represents the large number of small
specimens collected on the slabs representing the
mass kill. The smaller number of larger fish is due
to at least two factors. Firstly fewer individuals
survive to be fossilised in their maturity. These
individuals occur singularly rather than on mass
kill slabs. The other factor is related to the habits of
collectors. Large, well-preserved specimens are well
regarded in collections, so isolated examples of
medium sized fish may be overlooked when faced
with the prolific numbers of small individuals on
large slabs, and the few excellent large specimens.
Thus the distribution on the graph does not
necessarily represent the natural distribution in the
population.

Comparisons between Cavenderichthys and
Leptolepis

The Head

The type species of Leptolepis, which is Leptolepis
coryphaenoides, and L. normandica described by
Nybelin (1974), have several significant differences
from C talbragarensis, as well as the obvious
similarities (see Figure 18). The head of C.
talbragarensis is shorter that L. normandica and L.
coryphaenoides. Nybelin (1974) suggested that this
might be due to the shortness of the frontal bones,
but Nybelin's figures 1 and 4 (Figure 18A, 18B) show
the rostral is the bone that is significantly shorter in
C. talbragarensis. Also in these text figures there is a
large gap behind the upper jaw, resulting from the
large size of the infraorbital 3 bone, which in both L.
coryphaenoides and /.. normandica is considerably
wider than the other bones in the infraorbital series.
In C. talbragarensis the posterior infraorbitals are all
about the same width, allowing the preopercular to
come closer to the orbit. This arrangement is very
similar to Tharsis dubius, an Upper Jurassic
(Kimmeridgian) member of the Family
Leptolepididae (Figure 18D). The other significant
difference is that C. talbragarensis lacks a suborbital
bone between the infraorbital series and the
preoperculum, and once again this is a point of
similarity with T. dubius. This is also tied in with the
consistent width of the posterior infraorbital bones,
leading to an overall shortening of the head in
comparison with L. normandica. In C. talbragarensis
the size of the orbit is relatively large, which means
that the dermal bones occupy a restricted space, even
if the underlying structures of the braincase are very
similar in size to other species.

Nybelin (1974) placed a great deal of emphasis
for classification on the sensory canal systems,
especially on the preoperculum and the infraorbital
series. C. talbragarensis certainly has fewer branches
on both these canal systems. Considering the
preopercular canal, C talbragarensis has a maximum

of six undivided branches of this canal. This is far
less than L. coryphaenoides (19), and less than L.
normandica (10), but quite similar to Leptolepides
sprattiformis (5). The latter (Figure 18C) also has a
very similar jaw arrangement to C. talbragarensis.
The suborbital canal system on C talbragarensis has
very few branches, possibly a few on the lachrymal,
definitely one on the infraorbital, but none or very
few on the postorbitals. This is very different to L.
coryphaenoides, but quite similar to Leptolepides
sprattiformis.

The supraorbital sensory canal in C.
talbragarensis also has very few branches. However,
Nybelin's (1974) figures IB and 4B show that on L.
coryphaenoides and C. talbragarensis (Figure 18A,
18B) the presence of pores along the canal is similar.
There may be more pores present on C.
talbragarensis than figured herein, but they are not
always consistently placed. The back of the skull is
not included in the diagram because it has not been
possible to clarify the arrangement of extrascapulars
and suprascapulars due to poor preservation. The
presence of a branch in the supraorbital canal just
above the position of the dermosphenotic is likely
since the posterior branch of the canal terminates in
the parietal, as shown by Nybelin (1974), and thus
the ventrally directed branch should be connected
to the infraorbital canal on the dermosphenotic.
This is the logical connection, as the lateral line
canal system has to be interlinked, and it is the
general situation in similar extant forms. Possibly
there is no connection between these canals in
members of the Leptolepididae, as figured by
Nybelin (1974), or possibly it is present but has just
not been observed or described.

Comparison of the text-figures from Nybelin
(1974) and those from Patterson and Rosen (1977)
with the new reconstruction of the head of C.
talbragarensis indicates that the parasphenoid is a
valid point of comparison. Nybelin did not draw
the parasphenoid on either L. normandica or L.
coryphaenoides (his figures 1 and 4), however the
photographs in his plate 2 figure 2 and plate 4
figure 2 (L. normandica) and plate 6 figure 1, plate
7 figure 1 and plate 9 figure 4 (L. coryphaenoides)
the parasphenoid is quite clearly visible. This
omission in the drawings may have led to some
confusion because Patterson and Rosen (1977) have
included the parasphenoid in diagrams of
Leptolepides sprattiformis and Tharsis dubius. It is
certainly an obvious characteristic of C.
talbragarensis as it bisects the orbit, thus comparing
very closely with these other species.

The operculum in C. talbragarensis is narrower at
the top with its anterior and posterior margins
diverging ventrally to form an approximately
triangular shape, whereas in L. coryphaenoides the
anterior and posterior margins are almost parallel.

The skull roof is another region that requires

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

Figure 18 Comparative reconstructions of heads. A, Leptolepis normandica from Nybelin (1974, text fig. lA). B,
Leptolepis coryphaenoides from Nybelin (1974, figure 4A). C, Leptolepides sprattiformis from Patterson and
Rosen (1977, figure 49). D, Tharsis dubius from Patterson and Rosen (1977, figure 34).

L.B. Bean

Figure 19 Comparative reconstructions of skull roof. A, Leptolepis coryphaenoides from Nybelin (1974, figure 4B). B,
Leptolepis normandica from Nybelin (1974, figure IB). C, Cavenderichthys talbragarensis, reconstruction
based on MMF13564 and MMF36728.

comparison. Neither Woodward (1895) nor
Cavender (1970) mention the existence of nasal
bones on C talbragarensis, but they are preserved on
several of the studied specimens. The nasal bone
provides a protective tube for the sensory canal as it
leaves the frontal bone. In C. talbragarensis there is
often a quite prominent protective ridge on the

frontal, above the supraorbital, covering the sensory
canal. When this ridge is destroyed during
preservation the canal appears as a deep groove.
Just behind this covering ridge is a large pore
marking the position of the canal. Beyond this pore
the canal curves ventrally, sending a branch
onwards towards the parietal. It seems to end at the

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

Figure 20 Nybelin (1974, figure 37). Comparison of skull roofs, illustrating the difference between families
Pholidophoridae s. str. and Leptolepidae s. str., exemplified by differences in relative size of the two
supramaxillae and by the nasal relation to the anterior end of the frontal. This figure shows Nybelin's ideas
about evolutionary trends in the form of the skull roof. Cavenderichthys talbragarensis would fit on the right
hand end of this series.

back of the parietal where it surfaces through a
pore. The branch of the canal heading in a ventral
direction passes into the dermosphenotic. Cavender
also noted this connection and went on to comment;
"This is a significant point of difference in the
cephalic sensory system of C. talbragarensis, since
the junction of the infraorbital and supraorbital
canals is known to be absent in Leptolepis (Patterson
1967)." (Cavender 1970: 15)

Another bone of contention is the hyomandibular.
Woodward (1895) did not mention it, but Cavender
(1970) describes the bone in C. talbragarensis in
medial aspect. "The upper portion is expanded into
a single broad, articulating head that shows a very
slight emargination toward the middle of the dorsal
margin. The basal half of the opercular arm is
partially differentiated from the expanded portion
of the hyomandibular and produces a convex
posterior margin. The distal (condylar) part of the
opercular arm is not ossified. A large opening is
visible near the centre of the expanded portion,
which is the foramen for the hyomandibular trunk
of the VII nerve. The ventral portion of the
hyomandibular is constructed like a slender pillar."
(Cavender 1970: 21)

He went on to say that he did not find "an
anterior laminar expansion from the upper part of
the hyomandibular which contacts the
metapterygoid, or an adductor ridge along the
postero-lateral margin where it contacts the vertical
limb of the preopercule." (Cavender 1970: 21)
Basing her diagnosis on her own observations and
Cavender's description, Arratia (1997) described a
"hyomandibular without preopercular process, but

with a well developed levator arcus palatini crest"
(Arratia 1997: 19). The latter feature has not been
observed by the author. It is found that this bone is
generally hidden in C. talbragarensis. On seeing the
dorso-ventrally flattened specimen (MMF13734a) in
the NSW Geological Survey collection it became
possible to interpret their descriptions. Nybelin's
(1974, figure 3) is very instructive (Figures 9C-E).

It seems clear from these comparisons that C
talbragarensis does indeed have a preopercular
process on the hyomandibular, that it is not pointed
as in the Leptolepis specimens, but that it is
obviously serving the same function. The
articulation with the preoperculum will cause that
bone to be involved when the gill covering is
opened. This occurs when the lower jaw is
depressed and the opercular series (operculum,
suboperculum and interoperculum) are rotated
dorsally (Lauder 1982).

The same dorso-ventrally flattened specimen that
shows the hyomandibular (MMF13734a) also
exposes two complete lower jaws (Figure 21). They
can be seen to have a deep, wide dent on the
ascending portion of the anterior margin. This dent
is not as constricted as the leptolepid notch shown
in Nybelin (1974, plate 5, figure 9) for L.
normandica, but it occurs in the same position. In
many of his plates, dentaries are shown that do not
exhibit a leptolepid notch, but Nybelin's (1974, plate
14, figure 2), Proleptolepis furcata shows a dentary
with a notch in the same place as in C.
talbragarensis. His description stated "a detached
dentary of Proleptolepis sp. shows, however, a
rather deep notch in its ascending anterior margin

L.B. Bean

left mandible
(lateral)

left

hyomandibular

1mm

right mandible
(medial)

Figure 21 Cavenderichthys talbragarensis. Dorso-ventrally flattened head of MMF13734a. A, ventral view of flattened
head. B, medial view of left hyomandibular. C, medial view of right mandible. D, lateral view of right
mandible. E, lateral view of right hyomandibular.

and a similar notch is obviously present in the
holotype." (Nybelin 1974). This confirms that the
notch in the dentary of C. talbragarensis is
equivalent to a leptolepid notch.

The Caudal Skeleton

A caudal reconstruction (Figure 12C) corresponds
closely with those of Patterson and Rosen (1977,
figures 12A, 12B). The condition of the caudal
skeletons of L. normandica and L. coryphaenoides
is not clear, as there were not any well-preserved
specimens for Nybelin to describe. However,
Patterson and Rosen (1977, figure 22A) figured the
caudal skeleton of L. coryphaenoides. This shows a
caudal fin with three epurals, seven uroneurals, and
short neural spines on both preural centra 1 and 2.
It looks more similar to C. talbragarensis than the
caudal region of Tharsis dubius (Patterson and

Rosen 1977, figure 22B), despite the numbers of
bones being the same in T. dubius and C.
talbragarensis. The other "leptolepid" with a similar
caudal arrangement is Leptolepides sprattiformis,
with three epurals, five uroneurals, seven hypurals
and one urodermals, see Figure 22C.

Patterson was also able to examine some
specimens of C talbragarensis kept in the Natural
History Museum and comments (Patterson and
Rosen 1977: 144) that they only have three epurals,
not four as noted by Cavender. He saw six
uroneurals arranged as four long strap-like bones
and a posterior group of three shorter ones, as well
as two urodermals (Figures 12A, 12B).

EVOLUTION

The rise of the teleosts began in the Triassic but
they flourished in the Early Jurassic when they

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

UN5 eNS,?

Figure 22 Comparative reconstructions of the caudal skeleton. A, Leptolepis coryphaenoides from Patterson and Rosen
(1977, figure 33B), based on BMNH 32456, 32467, P42857 and P7622. B, Tharsis dubius from Patterson and
Rosen (1977, figure 35), based on BMNH P.927. C, Leptolepides sprattiformis from Patterson and Rosen
(1977, figure 50), based on BMNH P926.

L.B. Bean

rapidly achieved a worldwide distribution. Today
the Teleostei is the most abundant and diverse
group of vertebrate animal (about 29,000 species.).
Tlie earliest teleost has been a matter of conjecture
over the years, with the leptolepids long being
considered one of the oldest representative groups
of teleosts. The pholidophorids arose in the Triassic,
a group that has more in common with the
holosteans than do the leptolepids. Gardiner (1960)
described a possible leptolepid from the Upper
Triassic of Tanzania, which he called Leptolepis
africana. This would have been the oldest leptolepid,
except that Nybelin's (1974: 168) re-examination of
the specimen led to its inclusion in the genus
Pholidolepis, a member of the Pholidophoridae. This
family was considered to be more primitive than
the Leptolepididae, but it is still a member of the
teleosts (Patterson 1982).

The teleosts arose from the holosteans grade
group, and replaced them as the dominant
actinopterygians during the Jurassic. Some
characters that are considered primitive in teleosts
by Arratia (1997) include the following: the
parietals suturing with each other; dentition on the
parasphenoid; the possession of a suborbital bone;
the absence of a preopercular process on the
hyomandibular; premaxilla being a slightly
triangular bone; absence of supramaxillae; position
of quadrate-mandibular articulation located below
the posterior of the orbit; having a leptolepid notch;
having epineural bones but not epipleural bones;
having four epurals; having a high number of
hypurals e.g., 10; 10 or more principal caudal rays
in the lower lobe; ganoid scales.

Cavenderichtliys talbragarensis certainlv has many
of the characteristics of an early teleost. It has a
small, mobile premaxilla, which is free from the
maxilla. This is an early teleost characteristic (Rosen
1982), as in later teleosts the premaxilla becomes
the dominant bone of the upper jaw, bearing all the
teeth. The maxilla is hinged in the ethmoid region
and swings anteriorly as the mouth opens. This
helps to prevent water spilling from the corners of
the mouth so encouraging food to be drawn into
the mouth. As the mandible is depressed the
operculum rises and rotates outwards, also
encouraging the through flow of water and food
(Lauder 1982).

The roof of the skull has a continuous suture
between the frontals and the parietals. This is
characteristic of early teleosts (Arratia 1985), but the
later trend is for separation of the parietals to allow
insertion of muscles. There also seems to be a trend
to a reduced number of tubules branching from the
sensory canals, certainly within the leptolepids if
not within all teleosts. In this respect C.
talbragarensis shows an advanced feature.

Actinopterygian locomotion involves two styles,
firstly caudal propulsion which is used for cruising

and sprint swimming, acceleration and fast turns,
and secondly median and paired fin propulsion
used for slow swimming and in precise manoeuvres
(Webb 1982). C. talbragarensis makes use of both
these modes of swimming, w’ith an emphasis on
caudal propulsion. The well-ossified axial skeleton
gives it strength, its homocercal tail outline gives it
a balanced force from the caudal area, while the
scooped out centre of the tail improves flexibility
and steady swimming. The low position of the
pectoral fin and the middle position of the pelvic
fins are early teleost characteristics (Gosline 1971)
as the trend is for the pectoral fins to rise while the
pelvic fins become placed in a forward position.
The collapsible dorsal fin can reduce drag during
fast swimming but can also be raised to act like a
keel, especially during tight turns. The light scales
have reduced resistance in unsteady swimming,
and they help to reduce the mass of the body (Webb
1982).

C. talbragarensis has a primitive location of the
pectoral and pelvic fins, and yet the arrangement of
the premaxilla and maxilla is one stage advanced
from primitive. Rosen (1982) commented, "changes
in jaw mechanics first arose, and in some cases
proliferated, in teleosts with a primitive fin
arrangement and morphology" (Rosen 1982: 269).
There is no inherent reason why characters should
evolve at the same pace. It is possible that the
advanced placement of paired fins did not become
a selective advantage until the jaw structure had
been modified significantly. The change in feeding
style allowed by the more advanced mouth may
have been aided by the forward placement of the
paired pelvic fins, closer to the centre of gravity of
the fish.

Considering Arratia's (1997) list of primitive
characters, C. talbragarensis has parietals with a
suture between them, the quadrate-mandibular
articulation appears to be below the middle of the
orbit, the leptolepid notch is wide, and it has
epineural bones. In these characteristics it is
primitive. However, many of its features are more
advanced than primitive as it has no dentition on
the parasphenoid, does not have a suborbital bone,
does possess a preopercular process on the
hyomandibular, has two supramaxillae, has
epipleural bones, has commonly three epurals,
seven or eight hypurals, nine principal rays in the
lower lobe of caudal fin, and has cycloid scales.

One of the major advantages of this shidy over
previous work is the large amount of material
included, namely approximately 240 specimens.
Thus it has been possible to assess the internal
variation and whether there is more than one
species in the population. The material includes a
range of sizes of the specimens and a range of states
of preservation, including at least one good
example of all the cephalic dermal bones and many

Late Jurassic leptolepid fish Cavenderichthys talbragarensis

examples of well-preserved fins and axial skeleton.
The major conclusion is that there is only one
species of fish originally called Leptolepis by
Woodward, not three as he proposed. This is
supported by a statistical analysis of the range of
dimensions of the specimens. There are certainly
other genera present in the population, but they will
be described in another paper. The leptolepids
represent a population with a preponderance of
young individuals, but also with a representative
sample of older individuals. Woodward's Leptolepis
gregarius is the juvenile form of C. talbragarensis,
while L. lowei, in which the head appears to be
elongated, is an artefact of preservation.

ACKNOWLEDGEMENTS
I owe a deep debt of gratitude to many people,
including the following; Professor K.S.W.
Campbell, my supervisor, for his patience and
inspiration; Dr R.E. Barwick for teaching me the
intricacies of Photoshop and Illustrator; Dr I.S.
Williams, RSES, for his expertise and help with the
SHRIMP analysis; Dr A. Christy, ANU, for carrying
out the EXDA analysis; Dr M.V.H. Wilson,
University of Alberta for his very helpful review;
Professor R.J. Arculus, ANU, for help with the thin
sections; Dr G.C. Young, ANU and Dr R.K. Jones,
Australian Museum for access to specimens. Dr
G.D. Edgecombe, Australian Museum, for
preliminary editing; Dr J.A. Long, Museum
Victoria, for help with publication; Dr LG. Percival,
NSW Geological Survey and G. Dargan, NSW
Geological Survey for showing me the dorso-
ventrally flattened specimen.

REEERENCES

Agassiz, L. (1833-44). Recherches sur les Poissons
Fossiles. Neuchatel.

Arratia, G. (1985). Late Jurassic tolcosts (Acinopterygii,
Pisces) from northern Chile and Cuba.
Palaeontographica Abt. A 189: 29-61.

Arratia, G. (1997). Basal teleosts and teleostean
phylogeny. Verlag Dr Friedrich Pfeil, Munchen.

Cavender, T. M. (1970). A comparison of Coregonines
and other salmonids with the earliest known
teleostean fishes. In C. C. W. Lindsey (ed.). Biology of
Coregonid Fishes. University of Manitoba Press,
Winnipeg

Dulhunty, ]. A. and Eadic, J. (1969). Geology of the
Talbragar fossil fish bed area. Journal and Proceedings
of the Royal Society of New South Wales 102: 1-4.

Etheridge, R. J. and Olliff, A. S. (1890). The Mesozoic and
Tertiary insects of New South Wales. Memoirs of the
Geological Survey of New South Wales,
Palaeontology 7: 1-12.

Gardiner, B. G. (1960). A revision of certain
actinopterygian and coelacanth fishes, chiefly from

the Lower Lias. Bulletin of the British Museum of
Natural History, Geology 241-384.

Gosline, W. A. (1971). Functional Morphology and
Classification of Teleostean Fishes. The University
Press of Hawaii, Honolulu.

Hind, M. C. and Helby, R. J. (1969). The Great Artesian
Basin in New South Wales. Geology of New South
Wales. G. H. Packham. Journal of the Geological
Society of Australia. 16, 1: 490.

Lauder, G. V. J. (1980). Evolution of the feeding
mechanism in primitive actinopterygian fishes: a
functional anatomical analysis of Polypterus,
Lepisosteus, and Amia. Journal of Morphology 163:
283-317.

Lauder, G. V. J. (1982). Patterns of evolution in the
feeding mechanisms of actinopterygian fishes.
American Zoology 22; 275-285.

Long, J. (1991). The long history of Australian fossil
fishes. In P. Vickers-Rich (ed.). Vertebrate
Palaeontology of Australasia pp. 337^28. Monash
University Publications, Melbourne.

Norden, C. R. (1961). Comparative osteology of
representative salmonid fishes, with particular
reference to the grayling (Thymallus arcticus) and its
phylogeny. Journal Fisheries Research Board of
Canada 18: 679-791.

Nybelin, O. (1974). A revision of the leptolepid fishes.
Acta Regiae Societatis Scientiarum et Litterarum
Gothoburgensis, Zoologica 9: 1-201 .

Patterson, C. (1967). Are the teleosts a polyphyletic
group? Collogues Internationaux du Centre National
de la Recherche Scientifique 163: 93-109.

Patterson, C. (1977). The contribution of palaeontology to
teleostean phylogeny. In M. K. Hecht, P. C. Goody
and B. M. Hecht (eds). Major patterns in vertebrate
evolution pp. 579-643. Plenum Press, New York.

Patterson, C. (1982). Morphology and interrelationships
of primitive actinopterygian fishes. American
Zoologist 22: 241-259.

Patterson, C. and Rosen, D. E. (1977). Review of
Ichthyodectiform and other Mesozoic teleost fishes
and the theory and practice of classifying fossils.
Bulletin of the American Museum of Natural History
158; 81-172.

Percival, I. G. (1979). The Geological Heritage of New
South Wales. Sydney, Australian Heritage
Commission, and the Planning and Environment
Commission of New South Wales, pp. 237-244.

Pogson, D. J. and Cameron, R. G. (1999). Surat Basin.
Explanatory Notes, Dubbo Geological Sheet 1:250 000
SI/55-4, Geological Survey of New South Wales,
Mineral Resources of NSW, pp. 330-332.

Rosen, D. E. (1982). Teleostean interrelationships,
morphological function and evolutionary inference.
American Zoologist 22: 261-273.

Veevers, J. J. (ed.) (2000). Billion-year earth history of
Australia and neighbours in Gondwanaland. GEMOC
Press, Sydney.

Wade, R. T. (1941). The Jurassic fishes of New South
Wales. Journal and Proceedings of the Royal Society
of NSW 75: 71-84.

Waldman, M. (1971). Fish from the freshwater lower
Cretaceous of Victoria, Australia, with comments on

L.B. Bean

the palaeo-environment. Special Paper.'i in
Palaeontology. No 9.

Walkom, A. B. (1921). Mesozoic floras of New South
Wales. Part 1. Fossil plants from Cockabutta
Mountain and Talbragar. Memoirs of the Geological
Survey of New South Wales, Palaeontology'll: 1-21.

Webb, P. W. (1982). Locomotion patterns in the evolution
of actinopterygian fishes. American Zoology 22: 329-
342.

White, M. (1981). Fish beds reveal lush fossil forest.
Australian Natural History 1Q(7): 227-230.

Woodward, A. S. (1895). The Fossil Fishes of the
Talbragar Beds (Jurassic?). Memoirs of the Geological
Survey of New South Wales, Palaeontology' 9: 1-27.

Manuscript received 1 February 2005; accepted 17 July 2005.

Records of the Western Australian Musewn 23: 77-90 (2006).

A bioarchaeological investigation of a multiple burial
associated with the Batavia mutiny of 1629

D. Franklin and L. Freedman

School of Anatomy anci Human Biology, University of Western Australia,
Crawley, Western Australia 6009, Australia

Abstract - On 29 October 1628, the Verenigde Oostindische Compagnie
(VOC) Retoiirschip Batavia embarked on a voyage into infamy. Originally
sailing as part of a fleet of six other ships, the Batavia was subsequently
separated, and wrecked on Morning Reef in the Houtman Abrolhos on 4 June
1629. Tile ship's Commander, Francisco Pelsaert, had survivors landed on
nearby Beacon Island, and then embarked on a rescue voyage to Batavia
(modern day Jakarta). During Pelsaert's absence, an ultimately unsuccessful
mutiny attempt resulted in the murder of at least 125 people.

Human skeletal material has been recovered from excavations of the
Batavia land .sites since the 1960s. Four individual burials were discovered
between 1960 and 1964. A further six individuals were recovered from a
multiple burial between 1994 and 2001. Characteristics of the multiple burial,
such as the age, sex, positioning of individuals interred and evidence of
trauma are analysed and compared for any similarity to individuals listed,
and events outlined and historically recorded. The results of this analysis
suggest that four of the interred are probably the sick individuals who were
amongst the massacre's early victims; two sub-adults were also included in
the burial, at least one of which can also be directly accounted for.

INTRODUCTION

The Verenigde Oostindische Compagnie (VOC)
ship Retoiirschip Batavia was one of tlie largest and
finest armed vessels of the time. Carrying a
complement of approximately 316 people, the
Batavia embarked from Amsterdam on 29 October
1628, destined for Batavia (modern day Jakarta).
Cramped on board were men, women and children
of various socio-economic backgrounds and
nationalities, including VOC officers and crew, in
addition to naval cadets, passengers and soldiers
(Drake-Brockman 1963; Tyler 1970). Originally
sailing alongside a fleet of six other ships, the
Batavia was subsequently separated, and wrecked
on Morning Reef in the Houtman Abrolhos off
Australia's west coast on 4 June 1629 (Figure 1)
(Drake-Brockman 1963; van Huystee 1998).
According to historical records, the exact events
leading to the separation of the Batavia from the
fleet and her subsequent grounding remain
somewhat speculative.

Unable to free the Batavia from Morning Reef, the
ship's Commander, Francisco Pelsaert had 180
survivors landed on nearby Beacon Island, a small
coral island lacking freshwater (Figure 1). About
another 40 people (Pelsaert included) were landed
on one of the smaller islands early the following
morning, leaving approximafely 70 to 80 survivors
on the ship (Tyler 1970). Of those people still
aboard, approximately 40 were reported to have

drowned attempting to swim from the wreck to
land (van Huystee 1998). The exact figures are
unclear, but Drake-Brockman (1963: 50) puts the
total number finally landed at 268. Knowing that
their situation was dire, Pelsaert decided to take a
group in search of water on nearby islands and the
main 'Southland' (Drake-Brockman 1963; 126-127).
Failing in their search, Pelsaert resolved to attempt
the hazardous voyage of more than 1 900 kilometres
to Batavia (Drake-Brockman 1963).

Prior to the wrecking of the Batavia, trouble
between the ship's skipper, Adriaen Jacobsz, the
undermerchant, Jeronimus Cornelisz and Pelsaert
had sowed the seeds of dissent, which escalated on
the island into a plan for mutiny (see Drake-
Brockman 1963: 35-60). So, while Pelsaert
attempted the perilous rescue voyage to Batavia,
Cornelisz remained on Beacon Island and managed
to establish his own 'ruling council', and with the
aid of his followers began to murder all who
opposed him. Cornelisz planned to reduce the total
number of survivors to 40, with whom he planned
to hijack the anticipated rescue ship. Before
Pelsaert's return, Cornelisz and his accomplices had
murdered at least 125 men, women and children.
Upon Pelsaert's return on September 17"', Cornelisz
and his accomplices were captured, tried and most
were duly executed on purpose-built gallows
erected on Seals Island (present day Long Island)
(Drake-Brockman 1963).

D. Franklin, L. Freedman

Figure 1 Wallabi Group, Houtman Abrolhos Islands, showing locations of East and West Wallabi Islands, Beacon
Island (A) and the Batavia shipwreck (+) location (adapted from Green 1989).

So far, the skeletal remains of ten individuals have
been found on Beacon Island. Four individual
burials were discovered between 1960 and 1964
(Stanbury 1998). A further six individuals were
recovered from a multiple grave between 1994 and
2001. All are believed to be victims of the slaughter.
Some brief descriptions of the multiple grave
material have been made by Hunneybun (1995),
Pasveer et al. (1998) and Pasveer (2000). The
purpose of this paper is to present a description
and interpretation (using forensic, archaeological
and historical sources) of the human skeletal
material recovered from the multiple burial. Age,
sex, stature, general state of health and trauma will
be assessed and an attempt made to identify the
recovered individuals.

The primary source of information documenting
the epic voyage of the Batavia, the wrecking of the
ship and the trials of the mutineers is the
manuscript 'Droevige daghaenteyckeningh int
verliesen van ons schip Batavia', usually known as
the 'Pelsaert Journal' (van Huystee 1998).

THE SITE

The Houtman Abrolhos lie approximately 65 km
off the west coast of Australia, between latitudes
28° and 29° south (Figure 1). The Abrolhos are four
well-distinguished geographical units: North Island
and the Wallabi, Easter and Pelsaert Island Groups
(Teichert 1946). The Wallabi Group is the
northernmost in the Abrolhos, consisting of 32
islands scattered over an extensive system of coral
reefs. The elevation of the islands is generally low,
mostly approximately 2.5 metres above sea level
(Dakin 1919). The eastern islands, such as Beacon
and Long, are considerably smaller and consist of
accumulations of coral boulders and shingles (Storr
1965). Beacon Island, with an approximate area of
5.25 hectares, is sparsely vegetated and without a
source of fresh water (Bevacqua 1974; Green and
Stanbury 1988).

EXCAVATION

A review of the research directly concerning the

Bioarchaeological investigation of Batavia mutiny burials

discovery of the multiple burials on Beacon Island
follows.

1994 Field Season

In 1993, Philippe Godard reported John Gliddon's
discovery of a skeleton while digging a hole for a
'septic tank' near his house in 1990 (Figure 2)
(Godard 1993: 237). The concern for further damage
by souvenir hunters, due to the relatively precise
location provided by Godard, supported by
Gliddon's declaration under the Commonwealth
Historic Shipwrecks Amnesty (1993-1994),
provided the impetus for the 1994 field season by
the Western Australian Maritime Museum. This
season aimed to investigate the nature and extent of
the disturbance to the burial site based upon
interviews with residents of Beacon Island and to

make a physical investigation of the area (Gibbs
1994).

Gibbs established that the trench dug for piping
associated with a nearby toilet (leach drain) had
been constructed possibly six years earlier (around
1988) and that at least two or three skeletons were
found. One of the workers removed at least one
skull and mandible (Gibbs 1994). New sites were
identified along the southern edge of the house and
excavation in 1 m squares followed (Figure 2). In
the absence of detectable stratigraphic changes,
arbitrary levels of 5 or 10 cm were removed. Bulk
samples were collected, pH and Munsell soil
colours recorded and all material sieved through
three and five millimetre nested screens (Gibbs
1994).

The excavation area was extended in the hope of

Figure 2 Site map showing the excavated area (from Pasveer 2000). The house currently occupied by the Ashplant
family was formerly occupied by John Gliddon.

D. Franklin, L. Freedman

finding more bone fragments, leading to the
location of the remaining skeletal remains.

The two squares chosen for detailed examination
(E5 and D5 - Figure 2) were founcf to contain two
skulls (one of which was significantly damaged)
situated side by side with several other large adult
bone fragments (Gibbs 1994). The skeletons
appeared to extend south from the region of the
skulls. During excavation there were no discernable
stratigraphic changes and the soil pH ranged
between 8.5 and 10. Bone fragments were evident
from the first level and a maximum depth of 35 cm
was reached during excavation (Gibbs 1994;
Hunneybun 1995).

It was concluded that two skeletons were present
in the site and were disturbed during the toilet
sump and leach drain construction (Pasveer 2000;
5). Two adult crania (SK5 and SK6) were identified;
fragments of SK5 were recovered for analysis and
SK6 was left in situ (Stanbury 1998; Pasveer 2000).
The stratigraphy of the test pits was highly
disturbed as a result of both the trenching and
subsequent nesting activities of burrowing birds.
Gibbs (1994) made recommendations for future
systematic investigation on Beacon Island because
the sites had been seriously disturbed and their
locations were then well known.

1999 and 2001 Field Seasons

The degree of prior human and ongoing
disturbance by burrowing birds, together with the
risk of future vandalism, led to the decision to fully
excavate the burial site in 1999 (Pasveer 2000: 5). An
excavation grid was set up and Gibbs' 1994 squares
were reopened and extended 0.5 m .south (Figure 2)
(Pasveer 2000). Squares were excavated in 5 cm
levels and all material was sieved with nested 3 and
5 mm screens. Bulk soil samples were collected and
Munsell soil colour and pH readings were taken for
most levels. The pH levels recorded in the field
ranged between 5 and 6, but these measures were
subsequently found to be erroneous; reanalysis of
the bulk soil samples showed that the pH actually
ranged between 8.5 and 10 (Franklin 2001). The
soil's high calcium carbonate content favours bone
preservation and reflects this level of alkalinity
(Hunt and Gilkes 1992).

The excavated area was now found to Include five
individuals (three adults and two children) who
had been laid against each other within a circular
pit (Figure 2). These individuals were designated
SK7, SK8, SK9, SKIO and SKll (Figure 3). The bones
of the five individuals were in various stages of
preservation and some (particularly SKIO) were in
very poor condition. The crania identified in 1994

Figure 3 Four of the individuals uncovered in situ during the 1999 excavations; the bones of SK9 were removed at an
earlier stage of excavation. (Photo courtesy of the Western Australian Maritime Museum).

Bioarchaeological

investigation of Batavia mutiny burials

Table 1 Outlines and descriptions of the Batavia multiple burial skeletal material.

Individual

General Description

'SKS/SKll

•SK6/SK10

SK7

SK8

SK9

SK12

Adult: damaged cranium + partially incomplete postcranial skeleton

Adult: reconstructed skull and fragmentary postcranial skeleton

Child: reconstructed skull and largely incomplete postcranial skeleton

Infant: deciduous and permanent teeth only

■ Crania and postcranial skeletons recovered and catalogued separately

(SK5 and SK6) were believed to be associated with
these individuals, bringing the total sample to five
skeletons (Pasveer 2000). The skeletal remains were
found over, under, or in, a large deposit of black
dense matter possibly of organic origin. The bones
embedded in this deposit were left i)i situ and those
removed were poorly preserved (Pasveer 2000).
Later in 2001, the black deposit was excavated (see
Paterson and Franklin 2004) and 16 deciduous and
2 permanent teeth were discovered underneath.
This, the sixth individual recovered from the
multiple burial, was designated SKI 2.

MATERIALS AND METHODS

The present study examines the human skeletal
material recovered from the single multiple burial
excavated during expeditions to Beacon Island in
1994, 1999 and 2001. One of the objects of this study
has been to try and refer particular materials to
known individuals in the burial. To this end age,
sex and stature have been computed even when the
material has been limited. The material is briefly
listed in Table 1; the methods used for aging,
sexing, stature estimation and pathology are
described below.

A multifactorial technique of documenting age
changes in dental development, eruption (Ubelaker
1999) and attrition, and also epiphyseal union and
suture closure, was used for assigning individuals
info one of the following categories: Child
(deciduous teeth present), Sub-Adult (< 20 years -
aged by dentition). Young Adult (20 - 34 years).
Middle Adult (35 - 49 years) and Old Adult (50 +
years). Sex determination was largely based on
Giles and Elliot (1963), the 'Workshop of European
Anthropologists' (WEA 1980), and Buikstra and
Ubelaker (1994). Both cranial and postcranial
metrical and non-metrical sexual characteristics
were examined; however the paucity of os coxae
bones meant that the cranium was the skeletal
element mainly used for sex and age diagnoses.

The stature of adult individuals was assessed
using the tables devised for 'Whites' by Trotter
(1970). Due to the broken and incomplete condition
of many of the long bones, original lengths of the
tibia, humerus and radius are reconstructed and

estimated according to the method outlined by
Muller (1935). Statures are calculated from all
available long bones and only the most reliable
estimates are listed (e.g., lower before the upper
limb bones). Stature ranges are estimated after
consideration of variafion in both reconstructions of
fragmentary bones and stature regression equations.
Palaeopathological diagnoses were made according
to the criteria outlined in Ortner and Putscher
(1981), Iscan and Kennedy (1989), and Buikstra and
Ubelaker (1994). Each bone from the Batavia sample
was examined macroscopically, with a hand lens
and from radiographs of selected bones.

RESULTS

The basic description of the bones, estimated
personal age, assigned sex, estimated stature, and
any observed pathology and trauma of the skeletal
material recovered from the multiple burial are
outlined in this .section.

The crania and postcranial skeletons of SK5/SK11,
and SK6/SK10 were recovered at different times.
The crania and postcranial skeletons were thus only
able to be tentatively associated on the basis of
morphological assessment (size and development of
muscle attachments and metrical dimensions). On
this basis, the cranium of SK5, which has weak
development of muscle attachments, has been
associated with the slender postcranial skeleton of
SKll; the cranium of SK6, which has large areas of
muscle attachment (mastoid processes and nuchal
region of occipital bone), and is metrically larger
than SK5, has been associated with the larger, more
robust postcranial skeleton of SKIO.

SK5 / SKll

This cranium (SK5) and incomplete postcranial
skeleton (SKll) is significantly damaged. Two of
the larger parts of the cranium are the frontal and
parietal bones and there are fragments of the
occipital and temporal bones. Most of the sphenoid
bone is missing, as is the basilar part of the occipital
region. The facial bones are significantly damaged
and fragmented. There is no palate and only two
small pieces of the maxilla are present. Fragments
of both clavicles and scapulae were recovered. The

D. Franklin, L. Freedman

right humerus, radius and ulna are missing. The left
humerus, radius and ulna are in good condition
with only slight damage to the proximal and distal
ends. Ribs 1-11 of the left side are relatively well
preserv'ed. Parts of most vertebrae were recovered
although their bodies are missing. The left femur,
tibia and fibula are in good condition but have some
damage to the proximal and distal ends. Only a few
small pieces of the right tibia were recovered.
Nearly all of the hand and foot bones were
recovered.

Dental attrition and sutural closure suggest a
middle adult, c. 35 to 49 years of age. The form and
size of the nuchal crest, mastoid process and the
glabella indicate indeterminate sex (Buikstra and
Ubelaker 1994). Although damaged, discriminant
function .sexing of the cranium was possible, and
implied male sex (Giles and Elliot 1963). The radial
head diameter (25 mm) is characteristically male
(Berrizbeitla 1989), and the long bones are large but
with relatively weak muscle attachments, overall
they are more masculine than feminine. On balance,
cranial and postcranial morphology indicate a
probable male sex. Stature was assessed as
approximately 1.74 m (range 1.67 m - 1.82 m) from
the reconstructed length of the left tibia.

Tire dentition presents no apparent macroscopic
evidence of enamel hypoplasia or caries, but there
appears to be some abnormal loss of the alveolar
bone of the right maxilla. This resorption may have
resulted from periapical abscesses or tooth loss
(Ortner and Putschar 1981). The postcranial
skeleton presents evidence of traumatic episodes.
The shaft of the left ninth rib was broken near the
costal angle. This injury appears to have occurred a
significant time before death as the rib has healed,
but at an obviously abnormally acute angle. The
shaft of the left ulna has an area of abnormal bone
thickening (periostitis) just superior to the level of
the nutrient foramen. The abnormality was initially
thought to be a healing or healed mal-aligned
fracture, but radiographs show no evidence of a
fracture healed or otherwise. The damage to the
ulna could be the result of an injury not sufficient to
actually break the bone, but enough to have caused
blood vessel haemorrhaging, resulting in a
subperiosteal haematoma (inter-membrane
bleeding and abnormal calcification between the
periosteum and endosteum).

On the basis of the osteological assessment, there
is no evidence of trauma, which could have
contributed to death.

SK6/SK10

This skeleton is also significantly damaged and
largely incomplete. The largest pieces of the
cranium (SK6) include most of the facial region, the
zygomatic and sphenoid bones, the mastoid and
parts of the temporal bone of the left side (Figure

4). Other cranial fragments include parts of the right
orbit, occipital and zygomatic bones. The maxilla
and palate are complete, but there is no mandible.
Only a small proportion of the fragmented
postcranial skeleton (SKIO) was recovered. The left
and right humeri are both fragmented and
incomplete. The right radius and ulna are well
preserved and mostly complete. There is some
damage to the distal regions of the left radius and
ulna. A few ribs, lumbar vertebrae and sacral
fragments were recovered, but there are no pelvic
bones. The lower limb bones are largely fragmented
and incomplete. Most of the hand and foot bones
were recovered.

Dental attrition and sutural closure suggest a
middle adult, c. 35-49 years of age. The form of the
mastoid process and the glabella indicate male sex
(Buikstra and Ubelaker 1994) and discriminant
function sexing of the cranium also suggests male
sex (Giles and Elliot 1963). The radial head diameter
(23.25 mm) is characteristically male (Berrizbeitia
1989), and the right radius and ulna are large and
robust with well-developed muscle attachments,
traits that typify a male individual. Cranial and
postcranial morphology imply male sex. Stature
was assessed as approximately 1.79 m (range 1.75
m - 1.83 m) from the length of the right radius.

The dentition presents no apparent macroscopic
evidence of enamel hypoplasia, caries or dental
disease. However, apparently before death, the
upper right central incisor appears to have been

Figure 4 The damaged cranium of SK6; arrow
indicates impacted upper right central incisor.
(Photo by Patrick Baker, Western Australian
Maritime Museum).

Bioarchaeological investigation of Batavia mutiny burials

forced through the alveolar process into the nasal
cavity (Figure 4). Other than the localised area of
intrusion, neither the incisor nor the surrounding
maxillary bone appears to be significantly damaged.
There is no apparent evidence of postcranial
trauma, but there is .some disparity between the left
and right ulnae and radii, particularly in areas of
muscle or ligament attachments and shaft
circumference and tuberosity diameters. The right
radial tuberosity is better developed and larger than
on the opposing side. This asymmetrical
development may be the result of adapfations
associated with the favoured use of the right side.
Tlie terminal and middle phalanges of the fourth
digit of the right foot are fused. The aetiology of
this is uncertain, but it could be congenital, or the
result of trauma or occupational activities.

Although the osteological assessment indicates
some evidence of trauma in the cranium, it appears
unlikely to have contributed to death.

SK7

This is the most complete skeleton recovered from
the 1999 excavation (Figure 5). The skull is largely
complete and well preserved. The maxillary region
inferior to the right infraorbital foramen is damaged
and there is no palatine bone. There is also some
damage to the left sphenoid bone. Although
fragmentary, most postcranial skeletal elements are
represented, but the proximal and distal ends of
most of the long bones are damaged to some
degree. Of the pelvis only small fragments of the

left acetabulum and greater sciatic notch were
found. The bodies of all vertebrae are missing and
only the arches were found.

Dental development, attrition and sutural closure
suggest a young adult, c. 20-34 years of age. The
form of the nuchal crest, mastoid process and
mental eminence all indicate male sex (Buikstra and
Ubelaker 1994). Discriminant function sexing of the
cranium also indicated male sex (Giles and Elliot
1963). The left femoral head is mostly complete and
the diameter (49.75 mm) is characteristically male
(Stewart 1979). The shape of the greater sciatic
notch indicates a probable male morphology
(Buikstra and Ubelaker 1994). Cranial and
postcranial morphology clearly indicate male sex.
Stature was assessed as approximately 1.76 m
(range 1.71 m - 1.83 m) from the reconstructed
length of the right tibia. The dentition presents no
apparent macroscopic evidence of enamel
hypoplasia or dental disease and the only carious
tooth appears to be the lower right second
premolar, which is also broken midway through the
mesial cusp.

Evidence of trauma is not apparent, but there is
some disparity present between the left and right
clavicle and humeri, particularly in areas of muscle
or ligament attachment and shaft circumference and
diameters. The conoid tubercule of the left clavicle
is more developed and larger than the opposing
side. The mid-shaft diameter and least
circumference of the left humerus are larger than
the right side. This asymmetrical development

Figure 5 Reconstructed skull and postcranial skeleton of SK7. (Skull photo by Patrick Baker, Western Australian
Maritime Museum).

D. Franklin, L. Freedman

might be the result of mechanical changes
associated with the favoured use of the left side.
Radiographs of the right tibia revealed the presence
of at least three Harris lines approximately 20 to 25
mm from the distal articular surface. One of the
three lines is faint and very short, approximately 15
mm. The remaining two lines are clearly visible and
extend at least halfway across the tibial shaft.

Approximately one-quarter of the distal end of
the left tibial shaft is markedly roughened and
porous, a pattern indicating a deficiency in osteoid
production and calcium deposition (Figure 6)
(Collins 1966; Aufderheide and Rodriguez-Martin
1998). The shell of lamellar new bone on the surface
appears to be attached to the original cortex by a
web of more porous bone, which has completely
obliterated the nutrient foramen. This implies some
metabolic disturbance affecting the individual in the
immediate antemortem period and this has been
tentatively identified as possibly being scurvv, a
condition caused by a deficiency of vitamin C
(Aufderheide and Rodriguez-Martin 1998).

Scurvy was common in sailors of the period, and
is suggested because scorbutic individuals are
susceptible to subperiosteal haemorrhaging due to
aberrant blood capillary formation. When the
haemorrhage is of a substantial size the bone often
has a more porous structure (Aufderheide and
Rodriguez-Martin 1998: 311; Ortner ct al. 1999;
Buckley 2000). Since bleeding and the resulting
changes to surrounding bone tissue tend to occur in

regions under mechanical stress (Ortner el nl. 1999:
322), the obliteration of the tibial nutrient foramen
is probably not unusual. Similar scorbutic changes
in the lower long bones have been reported in
individuals from a 17“'’ century Dutch whaling
station at Spitsbergen (Maat 1981) and the 14"’
century Crow Creek massacre site in South Dakota
(Willey and Emerson 1993).

The osteological assessment does not reveal any
evidence of trauma, which could have contributed
to death.

SK8

The skull and some of the postcranial skeleton of
this individual were recovered. The skull is
significantly damaged, although largely complete,
and was reconstructed by Dr Stephen PCnott. Tliere
is some damage to the right facial bones,
particularly the right vomer and maxilla. There is
no palate. Most internal bones are missing and parts
of the right lateral body and condyle of the
mandible are damaged. Fragments of the atlas and
cervical vertebrae are pre.sent. Fragments of both
clavicles, parts of the right scapula and many rib
fragments are present. Both the proximal and distal
end regions of the left and right humeri are missing
and only the shaft remains. The distal regions of the
left radius and ulna are also damaged, Some of the
phalanges of the left hand were recovered. No
postcranial skeletal elements below the pelvis were
examined as they are embedded in a dense soil

Figure 6 Distal left tibia of SK7 displaying marked porosity.

Bioarchaeological investigation of Batavia mutiny burials

feature which has only recently been excavateci in a
solid block (Paterson and Franklin 2004).

Dental cievelopment and sutural closure suggest a
sub-adult, < 20 years (c. 15-16 years) of age. It was
not possible to conclusively determine the sex of
this immature and fragmented skeleton, but the
base of the symphysis and the body shape of the
mandible display some features noted by Loth and
Henneberg (2001) to be male characteristics. The
chin region of SK8 extends abruptly downwards
and squares off at the base of the symphysis and the
transition to the lateral body is sharply angled.
These features are typical male characteristics, but
more definitive sex determination would require
contemporary metrical data from a juvenile Dutch
population or DNA analysis. Stature was assessed
as approximately 1.51 m (range 1.49 m - 1.53 m)
from the reconstructed length of the left radius. The
dentition presents no apparent macroscopic
evidence of enamel hypoplasia, caries or dental
disease.

No apparent evidence of abnormalities was
detected in the skeleton. On the basis of the
osteological assessment there is no evidence of
trauma, which might have contributed to death.

SK9

The skull of this juvenile individual was crushed
by the weight of overlying soil and the postcranial
skeleton is largely incomplete (Figure 7). The skull,
reconstructed by Dr Stephen Knott, is fairly
complete, but parts of the right orbit and nasal
cavity, lacrimal bone and maxilla are missing or
damaged. Most internal bones are also missing and
the mandible is fractured through the middle of the
symphysis. Dental development suggest a child; c.
5-6 years of age. It was not possible to conclusively
determine the sex or reconstruct stature from the

immature skeleton. The dentition presents no
apparent macroscopic evidence of enamel
hypoplasia, caries or dental disease.

There was no evidence of trauma in the skeletal
remains, which might have contributed to death.

SKI 2

The only remains of this individual that were
recovered are the permanent left first molars and all
of the deciduous teeth, except for the lower incisors
and left canine. Dental development is of a child, c.
8-9 months of age, notably because initial cusp
formation of the first permanent molars has
occurred and their coalescence is nearly complete.
Sex determination was not possible. There is no
evidence of enamel hypoplasia in any of the teeth,
although there is some noticeable developmental
disparity with the left side slightly less developed
than the right. This is within a normal range of
variation and is not pathological. There is no
evidence of occlusal function.

DISCUSSION

The demographics and health status of the
individuals recovered from the multiple burial are
summarised (Table 2) and discussed below.

Demographics

All of the adults recovered from the multiple
burial were assessed as probable males. The sub-
adult individual (SK8) is also tentatively classified
as male, but the sex of the one child (SK9) and
infant (SK12) remain unknown. The bias tow'ards
males is expected, given that approximately five
men for every woman were murdered by Cornelisz
and his accomplices (Drake-Brockman 1963; 50).
The Batavia's complement comprised mostly males,

Figure 7 Reconstructed skull and postcranial skeleton of SK9. (Skull photo by Patrick Baker, Western Australian
Maritime Museum).

D. Franklin, L. Freedman

Table 2 Main features of the Batavia skeletal material including proposed associations (/) of cranial and postcranial
skeletons.

Individual

Description

Assigned

Sex

Age

Range'

Mean

Stature

Trauma/Abnormalities

SK5/SK11

Damaged cranium +
postcranial skeleton

Male

35-49 yrs

1.74 m

Healed rib fracture and
subperiosteal haematoma (ulna)

SK6 / SKIO

Damaged cranium +
postcranial skeleton

Male

35M9 yrs

1.79 m

Facial damage and skeletal
asymmetry (upper limb)

SK7

Skull + postcranial
skeleton

Male

20-34 yrs

1 .76 m

Harris lines and
pathological tibia (scurvy?)

SK8

Skull + postcranial
skeleton

Male

15-16 yrs

1.51 m

No evidence of trauma
or abnormalities

SK9

Skull + postcranial
skeleton

5-6 yrs

N/A

No evidence of trauma
or abnormalities

SK12

Deciduous +
permanent teeth

8-9 mo

N/A

Some left/right disparity
in dental development

*yrs = years; mo = months

who were regarded by the mutineers to be the
greatest threat. As such, males were the primary
targets in the initial murders. In contrast, women
were perceived as a lesser threat and many were
kept alive to serve as unwilling concubines for the
mutineers (van Huystee 1998). The age at death of
the sample appears to range from approximately 8
months to no older than 49 years of age. Even
though only a small sample was available for study,
the age demographics, in particular the apparent
lack of older adults (50+ years of age), is not
unusual given the shorter life expectancy in 17"’
century Europe (Jacobs 1991). Furthermore, the
chances of a career sailor serving aboard a VOC
vessel past 50 years of age would be slim, especially
considering shipboard conditions of the early 17*"
century (see Bruijn et al. 1987).

Dental health

Even taking the small sample size into
consideration, the dental health of the Batavia
sample as a whole is assessed as generally good.
There were relatively few examples of dental
diseases such as caries, periodontal disease or
calculus accumulation. Some of the older adults in
the Batavia sample displayed evidence of
antemortem tooth loss, often of the first molar. This
may have been the result of caries or severe wear
exposing the pulp cavity, leading to an infection of
the supporting tissue (Ortner and Putschar 1981).

Trauma

Of the multiple burial individuals, the postcranial
skeleton of SKll displays evidence of a healed
fracture of the left ninth rib and a subperiosteal
haematoma of the left ulna. The relatively poor
alignment of the rib fracture may attest to low
quality or no medical attention to the injury,
although the bone has healed well, which is a

general indicator of good health (Webb 1989). The
subperiosteal haematoma of the left ulna most
probably resulted from a fall or a blow to that arm.
The interpretation of traumatic markers often
affords some insight into certain factors concerning
the lifestyle of the afflicted individuals (Roberts and
Manchester 1995). For SK5/SK11 it was only
possible to infer that this individual may have led a
physical lifestyle where occupational injuries were
common. Other samples of 17'*' century mariners
also reputedly display similar evidence of having
led physical lifestyles (cf. Maat 1981: 169). The
upper right central incisor of fhe cranium of SK6
has been forced through the alveolar process into
the nasal cavity. This likely resulted from a heavy
blow against the teeth, but it is not possible to
ascertain whether the trauma is the result of
violence or accident. It is important to note that the
prevalence of trauma in the Batavia sample as a
whole may be under-represented due to the small
and fragmented nature of the skeletal assemblage.

Nutritional related deficiencies
The tentative diagnosis of scurvy in SK7 is fhe
only possible nutrition-related deficiency in the
sample. One of the primary causes of a high
mortality rate aboard VOC vessels was disease,
particularly scurvy, which could take a high toll on
a crew already undernourished on departure
(Gaastra and Bruijn 1993: 203). Efforts were made to
combat the illness by sending along less perishable
fruits or fruit extracts, and the Cape of Good Hope
was officially founded as a supply base for fresh
food and water in 1652 (Bruijn et al. 1987). The little
or no evidence of scurvy in this sample may
indicate that the provisioning of less perishable
foods and the stop over at the Cape (which the
Batavia made) was helping to combat the disease,
although the sample is only a small fraction of the

Bioarchaeological investigation of Batavia mutiny burials

ship's complement, and much of the recovered
material is poorly preserved.

Non-specific stress indicators

No enamel hypoplasia was detected in the
dentition of the sample. Relatively small and faint
Harris lines were present on SK7, which may
indicate a period of childhood nutritional
deficiency, stress or illness (Rathbun 1987; Hughes
et al. 1996). This was not uncommon in 17*'’ century
Europe (see Maat's 1984 analysis of 17"' to 18"’
century Dutch whalers). The relatively fragmented,
incomplete and poor condition of a large
proportion of the postcranial skeletons in the
multiple burial may be a factor in the possible
under representation of fhe frequency of Harris
lines. A lack of Harris lines does not indicate an
absence of disease; the aetiology and multiple
factors contributing to the appearance of these
skeletal markers is unclear and should not be
regarded as the sole indication of a population's
health (Hughes etal. 1996: 129).

Occupational stressors

Modern clinical data has established that certain
activities can cause osteoarthritis and other skeletal
changes (Roberts and Manchester 1995). Mechanical
factors are obviously important in the production
of pathological skeletal changes, but to assume that
these changes are indicative of occupational
stressors is probably an oversimplified view
(Waldron 1993: 73). On SK7 and SKIO some
asymmetry was observed in shaft diameters and the
size of muscle or ligament attachment sites between
the left and right sides of the upper skeleton. Any
habitual activity performed with a side-preference
can place extra stress on that specific proportion of
the anatomy and can contribute to asymmetrical
skeletal development (Wienlar and Wood 1988;
Capasso et al 1999). Some bending of the spinous
processes of the vertebrae of SK7 and SKll was
observed, but this is not considered to be
pathological and might be an occupational change
related to side-preference.

Interpretation of the multiple burial

From Pelsaert's Journal it is apparent that there
are at least two groups of murder victims buried in
multiple graves on Beacon Island; the first was a
group of sick individuals, the second the
Predicant's family (Drake-Brockman 1963: 175, 186).
There are, however, many other instances where
multiple burials might have been made, among
them the large number of people reported to have
drowned attempting to swim from the wreck to
land (van Huystee 1998). Below we describe two
possible interpretations of the Beacon Island
multiple burial. The first is that they were drowning

victims, and second that they were the family of the
Predicant (the Batavia's official minister) (Pasveer
2000). We conclude that both interpretations appear
to be improbable from the evidence of this research.

We then outline another more probable theory
backed by what we offer as supportive evidence
(Franklin and Freedman 2003).

Drowning victims follozving the wrecking
Could this multiple grave hold the remains of
individuals who had died by drowning in the
immediate period following the wrecking of the
Batavia? The skeletal remains in the multiple burial
were clearly 'roughly' thrown into the burial pit
(Figure 3). It would seem most unlikely that the
Batavia survivors would have abandoned their
religious or social values so soon after the wrecking.
While the situation was dire, it was not without
hope of rescue. The VOC would soon become aware
of their non-appearance and there was also the
option to sail for help in the salvaged sloop (as
Pelsaert did).

If the survivors were to bury their dead at that
time in a multiple grave, one would certainly expect
more care to have been taken in the burials rather
than the dearly hasty and disrespectful interments
they received. For example, a traditional Christian
burial style would at the very least have the bodies
orientated east-west, which the individuals buried
in the multiple grave are not (Figure 3) (Gerven et
al 1981). Even when people were faced wifh lethal
epidemics such as the 'Black Death', there are
examples of burials that were organised
methodically and with care (Royal Mint site,
London; Margerison and Knusel 2002). A modem
equivalent to the multiple burial on Beacon Island
(but on a larger scale) are crude, hasty multiple
graves associated with the murder of civilians in
the former Yugoslavia (see Stover and Ryan 2001).
The conclusion is that it would seem most unlikely
that the multiple burial represents a deliberate
interment of those who died by drowning or illness
following soon after the wrecking of the Batavia.

The Predicant's family

The Predicant's family included his wife, maid
and six of their seven children; two girls, three boys
and a baby. The historical accounts of the murder of
the Predicant's family describe 'the beating in of the
skull of the wife and that of one of the children'
(Drake-Brockman 1963: 174-186). The individual
burials discovered in the 1960s all display evidence
of cranial damage (sharp weapon trauma and
depressed fractures). On the other hand, in the
Beacon Island multiple burial there is no apparent
evidence of these types of trauma. Also, the number
of individuals in the Predicant's family and their
stated ages and sexes clearly do not match those
individuals from the burial pit (see Table 2)

D. Franklin, L. Freedman

A likeli/ iuterpretntion: early 'sick' zncthus of the
massacre

On July 10"’ or 12"’ (the dates are inconsistent in
Pelsaert's Journal) the early 'sick' murders occurred.
The victims were Passchier van den Enden (a
gunner), Jacob (Jacop) Hendricxsz (a carpenter), Jan
Pinten (an English soldier) and a cabin boy, all of
whom were ill and could offer little resistance. They
had their throats cut by Cornelisz's accomplices. At
least one body (Jacob Hendricxsz) was dragged into
a pit 'which had been made ready' (Drake-
Brockman 1963: 183-199). It would be reasonable to
assume that all four bodies were disposed of in the
ready-made pit simultaneously.

The three adult individuals killed by Jan
Hendricx.sz (the gunner, soldier and carpenter)
would all have been adult males. There are what
appear to be three male adults (SK5/SKT1, SK6/
SKIO and SK7) interred in the burial pit. The
apparent lack of violent trauma on all of the
skeletons is consistent with historical accounts of
the throats of fhese victims being cut rather than
having their heads beaten in, but does not rule out
other sorts of injuries undetectable in the absence of
soft tissue. The poor preservation of the skeletal
material recovered from the Beacon Island multiple
burial could explain the absence of markers
indicative of someone having their throat cut.
Interestingly, the mutineers asked to spare the
carpenter's life, buf Cornelisz reportedly replied,
"...that he was half lame and that he must go"
(Drake-Brockman 1963: 183). If by 'lame' Cornelisz
meant that the carpenter was physically disabled in
some manner, it is worth considering that SK7 very
likely had .some degree of pain and/or movement
impediment due to the pathological condition of
his left tibia (Maat 1981; Aufderheide and
Rodriguez-Martin 1998).

Skeleton SK8 was estimated to be less than 20
years old (probably between 15 and 16 years of age),
close to the typical age of a cabin boy. The sex of
this individual is uncertain, but the skeleton does
display .some male characteristics that offer some
credibility to assigning male sex. After reconciling
historical accounts with biological evidence, it
w’ould .seem reasonable to postulate that the four
individuals described above were those recovered
from the multiple burial pit.

There are two other individuals (SK9 and SKI 2)
represented in the multiple burial. SK9 is of a child
of unknown sex, approximately 5 to 6 years of age.
Approximately two days earlier than the murder of
the gunner, carpenter and soldier (July 8"’), Jan
Hendricxsz strangled the .six-year-old daughter of
Hans Hardens (Drake-Brockman 1963: 183). SK9 is
approximately the same age and does not display
any visible evidence of trauma. There is no mention
of how the body was disposed, but with the pit
available the body might well have been buried

there as well. This murder took place before the
first 'public' murder on July M'" (Drake-Brockman
1963: 115), so the body would have to have been
concealed until it could be permanently disposed.
SKI 2 is an infant approximately 8 to 9 months of
age. This individual cannof be directly accounted
for from historical accounts, but there were several
children of unspecified ages murdered and how
their bodies were disposed of is not always
recorded. Il is possible that the infant could be the
'suckling' child of Maijken Cardoes, who was
strangled by Salomon Deschamps on 20 July; but
this seems unlikely because the infant was
recovered from the bottom of the burial, and we
would have expected it to have died before 10-12
July (cf van Huystee 2000; Paterson and Franklin
2004).

CONCLUSIONS

This study describes the age, sex, stature, trauma
and pathology of the six individuals recovered from
a multiple burial related to the Batavia mutiny on
Beacon Island. The burial of those individuals was
not systematic, and instead repre.sents a hasty
interment in the early part of the mutiny. We
postulate that the remains of four of the interred
can be reconciled with historical accounts and could
be those of a group of sick individuals who were
amongst the early victims of the massacre. In
addition, two sub-adult skeletons were recovered.
The six year old may be the child strangled two
days before the other individuals were killed. The
infant could not be directly reconciled with
historical accounts, but a number of infants are
known to have been killed at various times during
the mutiny.

ACKNOWLEDGEMENTS

The authors would like to thank the staff (past
and present) of the Western Australian Maritime
Museum, including Myra Stanbury, Juliette
Pasveer, Corioli Souter and Patrick Baker, for access
to the Batavia material and for permission to
reproduce photographs. Specialised technical
advice was provided by Dr Alanah Buck, Dr
Stephen Knott (both of the QE II PathCentre), and
Professor Sandra Bowdler (of the Centre for
Archaeology, The University of Western Australia),
and was much appreciated. Special thanks go to Dr
Bruce Stock of Sir Charles Gairdner Hospital who
kindly donated his time and expertise to take the
radiographs. We would also like to thank Profes.sor
George Maat, Anatomy Department, The University
of Leiden, for the helpful correspondence and
informative research publications. The authors
thank Dr Alanah Buck and Dr Shane Burke for
commenting on drafts of this paper. We would also

Bioarchaeological investigation of Batavia mutiny burials

like to thank Myra Stanbury for her helpful review

of this paper. The preparation of this manuscript

was supported financially by ARC Discovery

Project Grant number DP0557157.

REFERENCES

Aufderhcide, A.C. and Rodriguez-Martin C. (1998). The
Cambridge Encyclopedia of Human Paleopathology.
Cambridge University Press, Cambridge.

Bcrrizbcitia, E.L. (1989). Sex determination with the head
of the radius. Journal of Forensic Sciences 34: 1206-1213.

Bevacqua, R. (1974). Archaeological survey of sites
relating to the Batavia shipwreck. Early Days (Journal
and Proceedings of the Western Australian Historical
Society) 7 (6): 50-78.

Bruijn, J.R., Gaastra, F.S. and Schoffer, 1. (1987). Dutch-
Asiatic Shipping in the 17''' and 18"' Centuries.
Volume 1. Introductory Volume. Rijks
Geschiedkundige Publicatien, Grote Serie 165.
Martinus Nijhoff, The Hague.

Buckley, H.R. (2000). Subadult health and disease in
prehistoric Tonga, Polynesia. American Journal of
Physical Anthropology 113: 481-505.

Buikstra, J.E. and Ubelaker, D.H. (1994). Standards for
Data Collection From Human Skeletal Remains.
Arkansas Archaeological Survey, Arkansas.

Capasso, L., Kennedy, K.A.R. and Wilczak, C.A. (1999).
Atlas of Occupational Markers on Human Remains.
Edigrafital S.P.A., Teramo.

Collins, D.H. (1966). Pathology of Bone. Buttcrworth and
Co., London.

Dakin, W. (1919). The Percy Sladen Trust Expeditions to
the Abrolhos Islands (Indian Ocean). Journal of the
Linnean Society London 34: 127-180.

Drake-Brockman, H. (1963). Voyage to Disaster.
University of Western Australia Press, Nedlands.

Franklin, D. (2001). A Bioarchaeological Investigation of
Beacon Island Land Sites and the Victims of the
Batavia Mutiny. Unpublished B.Sc. Honours Thesis,
School of Anatomy and Human Biology, Centre for
Archaeology. University of Western Australia, Perth.

Franklin, D. and Freedman, L. (2003). A
bioarchaeological investigation of Beacon Island Land
sites and the victims of the Batavia Mutiny.
Proceedings of the Australasian Society for Human
Biology, Perth, Western Australia, 9-11 December
2002. Homo 54: 81.

Gaastra, F.S. and Bruijn, J.R. (1993). The Dutch East India
Company's Shipping, 1602-1795, in a Comparative
Perspective. In J.R. Bruijn and F.S. Gasstra (eds). Ships,
Sailors and Spices: East India Companies and Their
Shipping the 16^, 17"' and 18"' Century, pp. 177-208.
CIP-Gegevens Koninklijkc Bibliotheck, The Hague.

Gerven, van D.P., Sandford, M.K. and Hummert, J.R.
(1981). Mortality and culture change in Nubia's Batn
el Hajar. Journal of Human Evolution 10; 395-408.

Gibbs, M. (1994). Report on the Excavation of Skeleton
SK5, A Victim of the Batavia Massacre of 1629,
Beacon Island, Western Australia. Department of
Maritime Archaeology Report No. 112. Western
Australian Maritime Museum, Fremantle.

Giles, E. and Elliot, O. (1963). Sex determination by
discriminant function analysis of crania. American
Journal of Physical Anthropology 21 : 53-68.

Godard, P. (1993). The First and Last Voyage of the
Batavia. Abrolhos Publishing, Perth.

Green, J. (1989). The Loss of the Verenigde Oo.stindische
Companie retourschip Batavia, Western Australia
1629: An Excavation Report and Catalogue of
Artefacts. British Archaeological Reports,
International Series No. 489, Oxford.

Green, J. and Stanbury, M. (1988). Report and
Recommendations on Archaeological Land Sites in
the Houtman Abrolhos. Report - Department of
Maritime Archaeology, Western Australian Maritime
Museum, No. 29.

Hughes, C., Heylings, D.J.A. and Power, C. (1996).
Transverse (Harris) lines in Irish archaeological
remains. American Journal of Physical Anthropology
101: 115-131.

Hunneybun, B. (1995). Skullduggery at Beacon Island.
Unpublished B.Sc. Honours Thesis, Centre for
Archaeology. University of Western Australia, Perth.

Hunt, N. and Gilkes, B. (1992). Farm Monitoring
Handbook: A Practical Down-to-Earth Manual for
Farmers and Other Land Users. University of
Western Australia: Nedlands.

Huystee, van M. (ed.) (1998). The Batavia Journal of
Francisco Pelsaert. Trans, van Huystee M. Report -
Department of Maritime Archaeology, Western
Australian Maritime Museum, No. 136.

Iscan, M.Y. and Kennedy, K.A.R. (eds) (1989).
Reconstruction of Life From the Skeleton. Allen R Liss
Inc, New York.

Jacobs, E.M. (1991). In Pursuit of Pepper and Tea: The
Story of the Dutch-East India Company. Netherlands
Maritime Museum, Amsterdam.

LoOi, S.R. and Henneberg, M. (2001). Sexually dimorphic
mandibular morphology in the first few' years of life.
American Journal of Physical Anthropology 115 ; 179-186.

Maat, G.J.R. (1981). Human remains at the Dutch
whaling station on Spitsbergen. A physical
anthropological study. Proceedings of the
International Symposium on Early European
Exploitation of the Northern Atlantic 800 - 1700.
Arctic Centre, Groningen.

Maat, G.J.R. (1984). Dating and rating of Harris's lines.
American Journal of Physical Anthropology 63: 291-299.

Margerison, B.J. and Knusel, C.J. (2002).
Palcodemographic comparison of a catastrophic and
attritional death assemblage. American Journal of
Physical Anthropology 119 : 134-144.

Muller, G. (1935). Zur bestimmung der lange
beschadigter extremitatenknochen. Anthropology
Anzeiger 12 : 70-72.

Ortner, D.J. and Putschar, W.G.J. (1981). Identification of
Pathological Conditions in Human Skeletal Remains.
Smithsonian Institution Press, Washington.

Ortner, D.J., Kimmerle, E.H. and Diez, M. (1999).
Probable evidence of scurvy in subadults from
archaeological sites in Peru. American Journal of
Physical Anthropology 108 : 321-331.

Pasveer, J. (2000). Archaeology. In M. Stanbury (ed.).

D. Franklin, L. Freedman

Abwlhos Isltinds Archaeological Sites: Interim Report
5-10, Special Publication No. 5. Australian National
Centre of Excellence for Maritime Archaeology,
Fremantle.

Pasveer, J., Buck, A. and van Huystee, M. (1998). Victims
of the Batavia mutiny: physical anthropological and
forensic studies on the Beacon Island skeletons.
Bulletin of the Australian Institute for Maritime
Archaeology 22: 45-50.

Paterson, A. and Franklin, D. (2004). The 1629 mass
grave for Batavia victims, Beacon Island, Houtman
Abrolhos Islands, Western Australia. Australasian
Society for Historical Archaeology 22: 71-78.

Rathbun, T.A. (1987). Health and disease at a South
Carolina plantation: 1840-1870. American Journal of
Physical Anthropology 74: 239-253.

Roberts, C. and Manchester, K. (1995). The Archaeology
of Disease. Alan Sutton Publishing, New York.

Stanbur}', M. (1998). Land archaeology in the Houtman
Abrolhos. In ). Green, M. Stanbury and F. Gaastra
(cds). The ANCODS Colloquium: Papers Presented at
the Australia-Netherlands Colloquium on Maritime
Archaeology and Maritime History, pp 101-107.
Special Publication No. 3. Australian National Centre
of Excellence for Maritime Archaeology', Fremantle.

Stewart, T.D. (1979). Essentials of Physical Anthropology:
Especially as Developed in the United States. Charles
C. Tltomas, Springfield.

Storr, G. (1965). The physiography, vegetation and
vertebrate fauna of the Wallabi Group, Houtman
Abrolhos. Journal of the Royal Society of Western
Australia 48: 1-14.

Stover, E. and Ryan, M. (2001). Breaking bread with the
dead. Historical Archaeology 35: 7-25.

Teichert, C. (1946). Contributions to the geology of
Houtman's Abrolhos, Western Australia. Proceedings
of the Linnean Society of New South Wales 71: 145-
196.

Trotter, M. (1970). Estimation of stature from intact limb
bones. In T.D. Stewart (ed.). Personal Identification in
Mass Disasters, pp. 71-83. Smithsonian Institution,
Washington D.C.

Tyler, P. (1970). The Batavia mutineers: evidence of an
anabaptist 'Fifth Column' within 17"' Century
colonialism? Westerly ‘i, December: 33-45.

Ubelaker, D.H. (1999). Human Skeletal Remains:
Excavation, Analysis, Interpretation. 3'" Ed.
Taraxacum Press, Washington.

Waldron, H.A. (1993). The health of the adults. In T.
Molleson and M. Cox (eds). The Spital fields Project:
Volume 2 - The Anthropology, pp. 67-89. Council for
British Archaeology Report Number 86, London.
Webb, S.G. (1989). The Willandra Lakes Hominids.
Department of Prehistory, Research School of Pacific
Studies, Canberra.

Weinlar, C.W. and Wood, J.E. (1988). Osteological
individuality indicative of migrant citrus laboring.
Journal of Forensic Sciences 33: 562-567.

Willey, P. and Emerson, T. (1993). The osteology and
archaeology of the Crow Creek massacre. Plains
Anthropologist 38: 227-269.

Workshop of European Anthropologists (1980).
Recommendations for age and sex diagnosis of
skeletons. Journal of Human Evolution 9: 517-554.

Manuscript received 3 Mn\j 2005; accepted 13 September 2005

Records of the Western Australian Museum 23: 91-113 (2005).

Structure and function of the tooth plates of the Devonian lungfish
Dipterus valenciennesi from Caithness and the Orkney Islands

Jan L. den Blaauwen^ , Richard E. Barwick^ and Kenton S. W. CampbelF

' Swammerdam Institute for Life Sciences, University of Amsterdam, Kruislaan 406, 1098 5m, Amsterdam,
The Netherlands, e-mail jdblaauw @ science.uva.nl

^School of Earth and Marine Sciences, Australian National University, Canberra, ACT 0200, Australia,
e-mail ken.campbell@anu.edu.au, richard.barwick@anu.edu.au

Abstract - The teeth of the Middle Devonian Dipnoan Dipterus valenciennesi
are described from new material from Caithness and the Orkney Islands,
Scotland. The biostratigraphy of the Old Red Sandstone in these two areas is
described on the basis of new information. The pallial dentine is made up of
groups of hard clusters of material. The core dentine in the tooth plates is
now understood in terms of the development of individual elements making
up the structure of the dentine. The first deposited material is interstitial
dentine, and the second is transparent dentine which is deposited from the
pulp canals against the interstitial dentine. All the core dentine is perforate.
Denteons continue to the tip of the tooth, and dentine tubules run from the
pulp canals through the transparent dentine to the pallial dentine. The
structure is not that of petrodentine. The difficulties of using living material
for the understanding of dentine in remote structures in time are outlined.
The relationships of organisms after the development of new palatal biting in
gnathostomes is discussed.

INTRODUCTION

Dipterus valenciennesi Sedgwick and Murchison,
1829 from the Middle Old Red Sandstone of
Scotland was studied by White (1965). References
to previous work can be obtained from his paper.
Later work by Schultze (1975) and Ahlberg and
Trewin (1995) is available.

The solid-snouted, cosmine-coated specimens
from the Thurso Flagstones, described by Agassiz
in 1844 as Polyphractus platycephalus, was one of
the reasons for Pander (1858) and Watson and Day
(1916) to use the specific name Dipterus
platycephalus. More recently Westoll (1949)
described the skull roof patterns of specimens from
Banniskirk (Caithness) and found them sufficiently
aberrant to separate them into a new species,
Dipterus braebypygopterus, and he revived
Agassiz's specific name platycephalus for all other
Scottish specimens of Dipterus. He advocated the
abandonment of the name Dipterus valenciennesi.
White (1965) concluded that the braebypygopterus
pattern was a variation on other specimens that
occurred at Banniskirk and elsewhere. The only
other genus comparable with Dipterus is a new
genus to be described by Newman and den
Blaauwen from the Middle Old Red Sandstone of
Caithness and Sutherland, formerly included in
either Pentlandia or Dipterus. It has a different
postcranial morphology and skull-roof pattern.

Following White (1965) we consider Dipterus
valenciennesi as a valid name.

STRATIGRAPHY

A stratigraphic table showing the distribution of
the Middle Old Red Sandstone is attached (Figure
1 )-

D. valenciennesi is well known from Eifelian or
Givetian cyclic sequences in the Orcadian Basin.
The species is common in the fish-bearing laminites
of Achanarras (Forster-Cooper 1937; Trewin 1986)
and the equivalent laminites on Orkney, the
Sandwick fishbed (Trewin 1976). These laminites
were deposited in deep water in an extended lake
in the Orcadian Basin. Small specimens lack
cosmine on the scales and dermal plates, but
specimens 20 cm or over in length, have cosmine on
part of the scales on the ventral side and part of the
dermal bones. Fully grown ones have a complete
cosmine cover. Dipterus is also found in the calcitic
nodules from the Moray Firth area where the
sediments show fluvial domination in a southward
extension of the Achanarras fishbed (Trewin and
Thirlwall 2002).

Research in museum collections in the U.K. has
shown that it is not possible to identify D.
valenciennesi positively in sediments older than
the Achanarras-Sandwick fishbed horizon.

].L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

CAITHNESS/

SUTHERLAND

Fauna

ost— arthr — dipn.

ORKNEY

Fauna

ost— arthr — dipn.

John o' Groat

Subgr.

-1

Eday Flags

1 '"]

Mey Subgr.

Mm.

Rousay Flags

Mm.

Latheron Subgr./
Ham-Scarfsk.Subg.

Tp.

Dt.

Upper Stromness

Flags

" 1

Dt.

Achanarras Horizon

1 Gm.

Sandwick Fishbed

Robbery Head

Subgr.

Oni.

Dv.

Lower Stromness

Flags

Ga.=

Om. ■

■

Dv.

Lybster Subgr.

Cc.

Tni.

ost. Cc.

Figure 1 Biostratigraphic table of Caithness, Sutherland and Orkney indicating faunal elements which are of
importance for correlation. Bars indicate the approximate range. Abbrcv.: arthr. arthrodire; Cc. Coccosteus
cuspidatus; dipn. dipnoan; Dt. Dickosteus threiplandi; Dv. Dipterus valenciennesi; Ga. Gvroptychius
agassizi; Gm. Gvroptychius milleri; Mm. Millerosteus minor, Pm. Pentlandia macroptera; Ps. New Dipnoan
Genus; Om. Osteolepis macrolepidotus', Osteolepis panderi; ost. osteolepid ; Ta. Tristichopterus alatus; Tm.
Thursius macrolepidotus', Wf. IVafsonosfeus.

Specimens from the Lybster Subgroup which
belong to the osteolepid Thursius
macrolepidotus are often misidentified as D.
valenciennesi. The specimens from the Robbery
Head Subgroup include a new genus being
described by Newman and den Blaauwen, and
those from the John o' Groat Subgroup belong to
Pentlandia macroptera.

The lacustrine sediments above the Achanarras-
Sandwick fishbed show climatically controlled
cycles resulting from long-term rise and fall of lake
levels in an enclosed basin (Crampton and
Carruthers 1914; Donovan et al. 1974; Donovan
1980; Trewin and Thirlwall 2002). These beds show
playa-lake conditions, though in places the water
may have been sufficiently deep to allow
articulated fish skeletons to accumulate. The
cyclicity of the sediments probably results from
Milankovitch periodicities. Some of the sediments
deposited in shallow water have polygonal
mudcracks and shrinkage cracks. Many secliments
show structures the shape of gypsum crystals or
pseudomorphs showing gypsum crystal solution.
These sedimentary structures are often preserved
by sand infill, introduced by wind transport across

the dried up lake floors (Astin and Rogers 1991;
Rogers and Astin 1991). Sediments indicating very
shallow lake deposits produce only disarticulated
fish remains, sometimes locally concentrated in
'bonebeds'.

From extensive field work and mapping of fish
remains a biostratigraphic pattern has been
distilled. Naturally there are some difficulties in
correlation of fresh wafer sequences in the Orcadian
Basin, where drying of parts of the lake and the
prevalence of desiccation features occur in contrast
with widespread lake extension conditions in which
laminites were deposited. Details of the issues will
be discussed elsewhere by den Blaauwen et al., but
from the point of view of the dipnoans, D.
valenciennesi has been identified from the base of
the Achanarras Horizon to the top of the Mey
Subgroup on the mainland, and the equivalent
Sandwick fishbed to the top of the Rousay Flags in
Orkney.

Specimens used in this study come from the units
indicated in the Figure 1, above the Achanarras and
Sandwick fishbeds. They are common in the
Latheron Subgroup and the Mey Subgroup in
Caithness and in the Upper Stromness Flags and

Tooth plates of Devonian lungfish Dipterus

the Rousay Flags of Orkney. Also specimens from
Tynet Burn, one of the fishbeds from the nodule
localities in the Moray Firth area, have been studied.
Specimens of D. valenciennesi sampled in
sediments indicating shallow lake conditions, are
disarticulated and are mostly mature or even fully
grown. Most specimens possess a well developed
cosmine coating.

SPECIMENS EXAMINED
All the specimens examined have come from
Caithness and the Orkneys. They have been taken
from the collections of den Blaauwen, Michael
Newman and Jack Saxon, and they have been
placed in the National Museum of Scotland (NMS)
collections. The new numbers are as follows:
G2004.10.1 From Clardon Haven, Caithness.
Latheron Subgroup. Posterior end eroded in situ.
Palatal tooth plates well developed.

G2004.10.2 From Clardon Haven, Caithness.
Latheron Subgroup. Palate sectioned to show the
'cosmine' between the teeth.

G2004.10.3 From Clardon Haven, Caithness.
Latheron Subgroup. Mandible with tooth plates.
G2004.10.4 From Thurso East, (the slates),
Caithness. Latheron Subgroup. Mandible with left
tooth plate lost.

G2004.10.5 From Thurso East, Caithness. Latheron
Subgroup. Right palatal tooth plate. Sectioned
horizontally, and vertically.

G2004.10.6 From Clardon Haven, Caithness.
Latheron Subgroup. Pectoral girdle.

G2004. 10.7 From Thurso East, Caithness, Latheron
Subgroup.

G2004.10.8 From Clardon Haven, Caithness,
Latheron Subgroup.

G2004.10.9 From Buckquoy west of Aikerness,
Mainland Orkney, Rousay Flags.

G2004.10.10 Same as 2004.10.9.

G2004.10.il From Thurso East, Caithness, Latheron
Subgroup.

G2004.10.12 to G2004.10.16 From Clardon Haven,
Caithness, Latheron Subgroup.

GROSS FEATURES OF THE DENTAL SYSTEM

The Palatal Tooth Plates
New rows of teeth are introduced between the
anterior sets of rows as spaces become available
(Figures 2, 8A). Some specimens show symmetrical
insertions of the two plates of the one specimen, but
others do not. The specimen figured by White
(1965, plate 1, figure 1) shows small teeth inserted
between the first and second rows, and in places
these teeth are more closely spaced. Specimen
G2004.10.8 (Figure 2A) is remarkable in that it has a
new irregular row of teeth anteromedially inserted.

and part of the original median row resorbed. The
same specimen shows gaps for the occlusion with
the mandibular teeth. Irregularity of rows is shown
by G2004.10.5 which leaves spaces for the insertion
of new rows on the mediolateral parts of the teeth.
Obviously the new rows were formed wherever a
space exists because of irregular growth in old
rows, and we conclude that genetic control on the
precise position of new teeth was limited. New
rows occupy only a small part of the length of the
head.

The parasphenoid is well defined, is up to three
times the length of the tooth plates, and has a well-
defined buccohyphophysial foramen. The nasal
capsules occupy about two thirds of the length of
the plates. Most of the posterior buccal cavity is
therefore not roofed by the dental plates. This point
is emphasized by the mandible in which the dental
plates are relatively small in relation to the whole
structure.

Structure of the Mandible
The best specimens we have are of individuals
which are a little above half grown, and show
features which we consider significant. The tooth
plates are 0.33-0.40 the length of the jaw (Figures 3
A,B), and the distance between the two tooth
plates is large in comparison with the Early
Devonian genera Dipnorhynchus and
Speonesydrion (Campbell and Barwick 1984). The
ratio of the median length to the total length of the
mandible, is only about one third. Note also that
the mandibular dental plates have a short median
length in comparison with the posterior length.
This is different from the shape of the palatal
dental plates, suggesting that the contact between
the mandibular and palatal plates was not one-to-
one. This interpretation is supported by the fact
that the inner face of the mandibular plate is
turned ventrally, and could not have met the
palatal plate on full closure of the jaw. This is
standard for the assembled Late Devonian species
(Barwick and Campbell 1996; Campbell and
Barwick 1998).

A second point is that the anterior gap between
the two mandibular plates is very large, and the
unencumbered space for the tongue pad would be
not only wide but also long and deep. The point of
origin of the tooth rows have been resorbed, and on
G2004.10.3 restorative dentine has been added to
the anteromedian side of the tooth plate. The tooth
plates of Dipterus platycephalus from a Scottish
specimen in the Manchester Museum, and figured
by Watson and Gill (1923, figure 34), also shows the
reduced tooth plates similar to those described
above. The cavities for the cartilage forming the
articulation with the quadrate are deep and slightly
doubled (Figure 3C), thus limiting the lateral
movement of the mandible.

10mm

J.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

Figure 2 A, palate of G4004.10.8. Specimen antero-posteriorly compressed, 'cosminc' removed; bone on the posterior
of the tooth plates. B, G2004.10.], 'cosmine' on the palate; first row of teeth partly covered by 'cosmine'. C,
palatal view of the specimen G2004.10.2; 'cosmine'scctioned from right palatal tooth plate. D, the specimen
from which Figure 2B was prepared; squashed antero-posteriorly. Scale = 10mm.

Tooth plates of Devonian lungfish Dipterus

Figure 3 A-C, two small mandibles G2004.10.3 and
G2004.10.4; A and B are dorsal views show-
ing tooth-plates. C is a more posterior view of
B showing the articulation cavities and the
position of the adductor fossae. D, G2004.10.7
showing both tooth plates.

'Cosmine' on the Medial Parts of the Palatal
Tooth

White (1965, plate 2) and Denison (1974, figure 4)
figured a thin layer of tissue occupying the space
between the two palatal tooth plates. It was termed
'cosmine' by both the above authors, because this
layer has a shiny surface, it often contains a large
number of pores, and superficially it has a
resemblance to cosmine. But pores are not always
present, and where pores are present, no pore
canals can be found beneath them. This point and

the reality of Westoll lines will be considered later
in this paper. The presence of enamel on the surface
of this tissue is the most important point to be
considered here.

Significance of These Gross Features
The functional significance of these features is
largely related to air breathing. We consider these
points under the following headings; extant air
breathers; palatal plates and parasphenoid; tongue
pad space; and brachial laminae

Extant air breathers

Extant dipnoans fall into two groups -
Lepidosiren and Protopterus which are obligate air
breathers, and Neoceratodus which is a facultative
air breather. These two groups have been discussed
by Thomson (1969), who has also compared them
with the Middle Devonian Dipterus.

In Lepidosiren and Protopterus (Bishop and
Foxon 1968) the tongue fits between the pterygoid
tooth plates and makes a closing valve when air is
depressed into the lungs. This is done by the
anterior rotation of both the ceratohyal and the
pectoral girdles. The air is stored in the
parabranchial cavity partly roofed by the elongate
parasphenoid, the teeth are small with respect to
the size of the head and they are separated to leave
a space for the tongue to close off the buccal cavity
when air is forced into the lungs. Associated with
this procedure is the increase in space betw^een the
mandibular tooth plates which allows the tongue
pad to expand forwards. The ventral surface of the
head is also able to expand the buccal cavity to
permit more space for the retention of air.
Neoceratodus has a different arrangement based on
an opercular pump, and Thomson (1969, figure 5)
shows the movement of the opercular fold during a
breathing phase. In addition Neoceratodus has a
massive ceratohyal which takes part in breathing
movements.

Palatal Plates and Parasphenoid

The palatal plates in Dipterus are situated well
anterior in the mouth, and they are well separated
from each other. In comparison with such Early
Devonian genera as Dipnorhynchus or the Late
Devonian Chirodipterus the plates are very short.
The parasphenoid has a long posterior projection,
and extends back over a long distance behind the
pterygoids. Both these features make for long
orobranchial and parabranchial cavities.

Space for the Tongue Pad

The gap left for the tongue between the
prearticulars is large and deep in comparison with
that of Early Devonian genera Dipnorhynchus and
Speonesydrion. The gap between the palatal tooth

plates is covered with 'cosmine', and this shows
that the large tongue pad had ample room to lie
between these plates when the mouth is closed. The
enamel surface on the 'cosmine' shows that the
epidermis was in contact with this surface. This is
the ideal arrangement for the stop valve when the
air was being forced into the lungs from the
orobranchial and parabranchial cavities.

Pectoral Girdle of D. valenciennesi

We have access to several specimens of D.
valenciennesi which are better preserved than any
specimens previously described. The four
specimens are now labeled G2004.10.6,
G2004.10.13, G2004.10.14 and G2004.10.15. We will
describe this material in a separate paper. The
pattern of the branchial laminae and the
scapulocoracoid are very similar to those on
Chirodipterus australis (Campbell and Barwick,
1987) a marine form from the Late Devonian Gogo
Formation, Western Australia. The branchial
laminae would have operated in the same way in
the two species.

Summary

Dipterus has a large opercular plate as well as
small suboperculars, the movement of which would
have produced a large expansion and contraction of
the parabranchial chamber as detailed by Thomson
(1969). We note that Neoceratodus uses a opercular
parabranchial pump when breathing air, and uses
the very large ceratohyals to push the air from the
orobranchial and parabranchial chambers into the
lung. On the other hand, the branchial laminae are
so large and the ceratohyal so short, that Dipterus
could not have used the methods of breathing
adopted by Lepidosiren and Protopterus. We have
concluded that Dipterus was a facultative air
breather, though the structures could not have been
as efficient as those of Neoceratodus.

We note that Schultze and Chorn (1997) consider
that lungs were a feature of primitive osteichthyans,
quoting the fact that lungs are present in primitive
actinopterygians (e.g., Polypterus), actinistians,
lungfish, and tetrapods. Campbell and Barwick
(1999: 137-138) have commented on their
arguments, and these will not be repeated here.
Incidentally they offer no mechanism supporting
their views. Neoceratodus is also gill breathing, and
spends most of its life submerged. Comparison with
Dipterus valenciennesi suggests that this Devonian
form also had the capacity to use gill respiration as
well as aerial respiration.

Our arguments are based on morphology of
Dipterus and the extant dipnoans, and not on
cladistics or the range of air breathing in some
extant animals. Contrary to the argument of
Schultze and Chorn (1997), we still maintain the
marine Devonian dipnoans lacked the

J.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

morphological features which would indicate that
they were air breathers.

DESCRIBED HISTOLOGY OF THE TOOTH
PLATES

The histological structure of D. valenciennesi was
not described from Scottish material until recently,
because like all the bones at the fossiliferous
localities, the teeth were deeply stained by organic
carbon. White (1966, plate 1, figure 2) published a
figure of a section of a tooth plate in which the end
tooth was sectioned medially and showed a
translucent core. The figured adjacent teeth
apparently showed a bony core, presumably
because they were from marginal sections of the
teeth. Denison (1974: 39) commented on these
structures, but his work has not been confirmed.

Smith (1984) described teeth from the specimens
described by White (BMNH P44691), and later
another specimen, BMNH P53537, from Caithness
(Smith 1989). These papers give no details of the
pallial dentine, and their structure of the tooth core
is obscure. Comments will be made on this work
later in this paper.

Kemp (2001) in her paper on petrodentine does
not describe the histological structure of D.
valenciennesi, and most ancient forms dealt with
are of Carboniferous age. Because of this lack of
direct analysis of Dipterus valenciennesi tooth
plates, the discovery of well-preserved specimens
now gives us an opportunity to place these plates in
the primitive position which their stratigraphic
position accords them.

INTERPRETATION OF DENTINE IN NEW
MATERIAL

There is nothing more contentious than the
terminology of dentine in dipnoan teeth. For
present purposes Smith (1984, 1985, Table 1) has
provided the basis on which subsequent work has
developed. Further work on this topic can be found
in Lison (1941), Barwick et al. (1997), Campbell and
Smith (1987), Lund et al. (1992), Kemp (2001) and
Reisz etal. (2004).

Vertical Sections of Teeth
The considerable advantage we have is the
availability of growth stages of the teeth. Growth of
the layers in the dentine can be outlined by a
number of specimens. In the first instance we
describe a number of teeth from a single section.

Sections through sedhnent with placoderm plates
Some specimens have structures sufficiently well
preserved to show histological detail throughout
the teeth. Section G2004.10.9 shows the best vertical

Tooth plates of Devonian lungfish Dipterus

Figure 4 G2004.10.9. Vertical section in single

polarized light; specimen was incompletely
grown; enamel is largely destroyed;
interstitial dentine grey in colour; translucent
dentine around pulp canals.

tooth sections we have seen (Figures 4-6; 14E). The
largest tooth in the section has a large basal pulp
cavity, and the smaller teeth are in process of
formation. These show the development of the
histological structures.

The largest tooth has lost its apex, but on the same
section two other small teeth, one at the lateral edge
of the plate (referred to as the lateral tooth below)
and the other being cut tangentially to the axis of
the tooth (referred to as the tangential tooth below.
Figures 6; 14E). The enamel is present around the
margin of the lateral tooth, but it is partly destroyed
by decomposition on all sides of the larger tooth.
The core dentine consists of two different types of
structure, clear translucent columns, and
interstitial dentine columns (Figure 14E).

As Figure 4 shows, the pallial dentine is thickest
towards the apex but fades away somewhat

towards the base. The edges contain some branched
tubules which arise directly from the pulp canals,
and these are either simple or branched. At higher
magnifications, the tubules subdivide extensively
towards their outer edges (Figure 5A) making a
meshwork-like pattern. In the lower half of the
tooth the pallial dentine shows very fine tubules,
but at the base of the tooth, the pallial dentine turns
inwards (Figure 5B) and has crenellated dendriform
pattern. This is best known as pedestal dentine.
Pallial dentine is well shown on the lateral tooth
and the tangential tooth (Figure 6). It forms a dark
layer which is also penetrated by tubules. In places
the boundary between the pallial dentine and the
core dentine is sharp, but in other areas the
boundary is gradational.

On G2004.10.9, interstitial dentine is clearly
exposed (Figures 4, 6). Dark columns of interstitial
dentine extend into the basal pulp cavity where
their outlines are clear. In the basal pulp cavity at
high magnification the interstitial dentine shows an
open-work structure, which is seen on all three
teeth (Figures 5B; 6B; 14E). The implication is that
the growing edge of the material in the basal pulp
cavity is made of crenellated dendriform material,
which has the same appearance as the pedestal
dentine, and laterally this material joins the pallial
dentine (Figure 5B). In the central part of Figure 4,
and the distal parts of Figures 6A,C, the interstitial
dentine becomes vaguely outlined, not because the
section is marginal to the dentine layer but the
dentine is partly transformed into translucent
dentine.

The marginal tooth is very informative with
respect to the formation of the interstitial dentine.
As shown on Figure 6 A, in the core of the tooth the
interstitial dentine becomes rough in its outline
towards the basal pulp cavity. On the left
ventrolateral margin the interstitial dentine grades
into the pallial dentine and ventrally into the
pedestal dentine as shown on Figure 6B. The
tangential tooth (Figure 6C) also shows the
interstitial dentine extending to the basal pulp
cavity.

The translucent columns are the most striking
part of the core, and show up in plain light as a
clear translucent structure. Near the crest of the
large tooth the structure is clear (Figure 5A) and the
translucent layer carries many tubules which are
derived from the pulp canals, and in places these
run through the layer into the pallial dentine. This
is in an early stage of the tooth formation. In the
central part of the tooth, the fine structure of the
translucent layers is not clear, but it does contain
small openings and vague lines. The mode of
formation of this tissue is clearly demonstrated by
this section (Figures 4; 5B; 14E). It was deposited
against the interstitial dentine by cells in the pulp
canals. The base of the large tooth shows a layer of

J.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

pulp canals

pallial

dentine

interstitial

dentine

tubules

translucent

dentine

translucent

dentine

interstitial

dentine

crenellated

dendriform

dentine

pedestal

dentine

Figxu-e 5 A, apex of Figure 4; tubules penetrating the translucent dentine and connecting with the pallial dentine;

translucent dentine has replaced most of the interstitial dentine. B, enlargement of the bottom right of Figure
4; interstitial dentine around the pulp canals and translucent dentine deposited from the pulp canals on the
interstitial dentine; crenellated edge of the pallial dentine joined by a column of vertical interstitial dentine,
and basally joining the pedestal dentine.

Tooth plates of Devonian lungfish Dipterus

Figure 6 G2004.10.9 as on Figure 14E. A, the lateral tooth rotated; interstitial dentine runs from pulp to pallial dentine;

pallial dentine with tubules; translucent dentine around pulp canals. B, bottom left of Figure 6A with pallial
dentine joining with the pedestal dentine. C, the tangential tooth. Figure 14E; contact between interstitial and
translucent dentine intergrading; translucent dentine often bulbous.

100

J.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

10mm

dentinl

carbonate
in pulp
' canals.

gulp I
canals

Figure 7 A, G2004.10.5C, slightly etched in acetic acid; core of each tooth extends into the basal bone. B and C, optical
sections cut from the opposite face of Figure 7A; base of pulp canals with CaCOj; interstitial dentine near
margins but replaced medially by translucent dentine. D, enlargement of the area outlined in A; pallial
dentine on left joined with pedestal dentine of right tooth.

dentine

0.1mm-- •interstitwi

.* j .'I .V i

0.1 mm

Tooth plates of Devonian lungfish Dipterus

101

translucent material deposited on both sides of the
interstitial dentine, which grew down into the basal
pulp cavity.

The tangential tooth shows new translucent layers
deposited on pulp canals on the left side and open
pulp canals with new translucent material on the
right. The inner face of the translucent layers form
slightly bulbous surfaces against the pulp canals
(Figure 6C). The bone below the tooth is clearly
distinguishable from the other layers because of the
osteocyte spaces. Under high magnification, the
pedestal dentine is not solid, but consists of a
dendriform mass of material with numerous
openings (Figure 6C), and this passes towards the
tooth tip as interstitial dentine. The tangential tooth
also shows the formation of the translucent layers.
It is clearly laid down on the interstitial layers, has
a bulbous surface in the pulp canals, and fades
away into the basal pulp canal.

Other vertical sections

Two fully developed teeth on G2004.10.5 are
joined together (Figure 7A). The translucent
columns extending to the basal pulp cavity lying on
the pterygoid bone. The pallial and pedestal dentine
occupy a large part of the space between the
adjacent teeth (Figure 7A,D). The enlarged
illustration of this space shows the base of the left
tooth and the broad base of the right tooth with the
pallial, plus pedestal dentine also, preserved. The
perforate nature of the translucent dentine as seen
in single polarised light, and this is apparent over
the whole length of the tooth. Only translucent
dentine is visible in most of this section, as the
interstitial dentine has been converted to
translucent dentine. This change can be seen in both
Figures 7B and C. The interstitial dentine is more
obvious marginally, and this is what the
illustrations on Figure 8 also show. The pallial
dentine and the interstitial dentine merge at their
junction on the sides of the tooth, and above the
basal pulp cavity the two tissues also join. SEM
examination shows that this material is composed
of clumps of material which are roughly joined
together.

This specimen also shows how new material is
added to the margins of the tooth during growth.
Figure 7B,C, shows the central part of the tooth
formed from pulp canals that extend vertically from
the basal pulp canal. Laterally some of these pulp
canals branch and open to the margins of the tooth.
In addition, as the width of the basal pulp cavity
expands laterally during growth, new vertical pulp
canals are developed and these also produce canals
that open to the pallial dentine on the lower surface
of the tooth.

On the lateral margins of Figures 7B,C, interstitial
dentine is visible. These open to the pallial dentine
marginally, but deeper into the tooth they

disappear and are transformed into the translucent
dentine. Interpretation of these structures is best
understood from the transverse sections of the
teeth.

Transverse Sections of Teeth

Sections through C2004.10.5
A palatal plate, showing a number of teeth, has
been cut transversely (Figure 8A), and this is one of
the most illuminating sections we have seen. In the
second row from the left, the second tooth from the
front has the greatest number of pulp cavities, and
we regard this as the most fully developed tooth in
the section (Figure 8B). It shows the large central
pulp cavities and new marginally added pulp
cavities introduced to increase the diameter of the
tooth. The second tooth in the third row from the
left has fewer pulp canals and we use this as a tooth
which has been cut closer towards its tip (Figure
10A,B).

Each tooth section shows that the margins have
the most recently added dentine as shown on the
vertical sections described above. Internally to this
tissue is the dentine added earlier on in the tooth
growth. By examining the sections from the
margins to the central part of the core, the sequence
of changes that took place can be observed.

The core dentine meets the pallial dentine in a
clear edge. Some of the arcs of the crystalline
material open outwards to the pallial dentine and
are partly filled with pallial dentine (Figure 9C).
Others are closed, and the spaces between the arcs
are filled with a shiny substance which must be
interstitial dentine (Figures 8E,F). This should be
compared with Figures 7B,D. We also note that the
columns of crystalline material are asymetrical in
that the pulp canals are displaced towards the
lateral margins. And finally towards the pallial
dentine, the crystalline arcs were open (Figures
8C-F).

What is the nature of these crystalline rings?
Under crossed polars transverse sections through
the teeth in the central part of the tooth, show that
they have a band of dark coloured material around
the pulp canals (translucent dentine) and lateral to
that is a ring of the light coloured appearance
(interstitial dentine). Towards the margin of the
tooth, the central ring (translucent dentine)
decreases in size and its lateral face is the most
reduced; finally only an arc-shaped, incomplete
crystalline ring remains. Surrounding these rings,
both complete and incomplete, is dark coloured
material which we infer from the vertical sections is
interstitial dentine. The structure is illustrated in
Figures 9A-C; IOC; 13E.

Tracing these dark layers towards the centre of
the tooth, they join with shiny layers which separate
the crystalline layers, and even further internally

102

J.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

Figure 8 G2004.10.5D. A, SEM of polished surface of a palatal tooth plate; right row with branching. B, clockwise

rotation of tooth second from the front in second row in A: large pulp canals central, smaller canals margin-
ally; small marginal units surrounded by dark interstitial dentine; outlines of Figures C-F marked in white
lines. C and D, several crystalline layers open to the pallial dentine; outer crystalline layers separated by
dark material (interstitial dentine); dark layers become thinner medially, gradually replaced by transforma-
tion of the interstitial dentine to modified translucent dentine; circles mark the junction between crystals
around adjacent pulp canals; arrows mark the transition of the interstitial dentine incomplete; E and F, dark
layers between the crystalline material with a shiny surface; white circles mark junction between adjacent
crystalline areas.

103

translucent denti

Tooth plates of Devonian lungfish Dipterus

500m

0.1mm

200Mm

figure y A, rotated oDlique cross section ot the anterior tootn in top lett row in figure »A. B, enlargement ot the lower
area marked on Figure 9A; black arrow marks a pulp canal opening into the space in the open translucent
dentine. C, enlargement and rotated area marked on the top right of Figure 9A. Crystalline material passing
into pallial dentine; pallial dentine and interstitial dentine occupying spaces in translucent material. D, part
of the flat dentine occupying the marginal edge of the tooth plate in G2004.10.10.

104

j.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

200|jm

Figure 10 A, tooth second from the end in the third
row of Figure 8A. B, enlargement of the area
marked in A, with fine divisions in dentine.
C, left side of the third tooth in the third row
from left in Figure 8A; double edged
crystalline layers, beginning to form around
the new pulp canals.

these shiny layers become narrower and finally
disappear. This is well shown on Figures 8D-F. In
Figure 8F the junction between the three crystalline
layers around the pulp canals is clear, and are
occupied by very narrow' bands of shiny material.

From these data we infer that the interstitial
dentine is transformed to have a crystalline
structure. To do this, histogenetic fluids must pass
through the translucent dentine, and produce
crystalline dentine from layers which have no
internal structure when it is first deposited. The
core of the tooth is therefore a composite structure,
and shows different features at different levels of
growth. This is what we saw in the vertical sections.
The whole core structure of the tooth is a
dynamically evolving tissue, composite in structure,
and containing abundant pores. It does not have
most of the features used for defining petrodentine.

These sections also show features of importance
with regard to the pallial dentine. In several places
the pallial dentine appears as a mass of clusters
w'ithin which no detail can be seen. The best
illustration comes from the margin of the first tooth
on the left side, where the pallial dentine is
obliquely cut (Figure 9A-C). Clusters of small
patches of tissue make up the pallial dentine. Small
gaps separate each patch and in places these are
joined together to make a narrow canal. Compare
this section with the vertical section on Figure 7D.

Isolated teeth from G2004.20.5

Slightly oblique sections show the internal
structures admirably (Figures 13A,D,E). The
presence of newdy added marginal pulp canals in
these sections together with the small number of
total pulp canals, indicate that the teeth were either
juveniles or cut half way along their adult length.
The core material is crammed with perforations.

200Mm

areas filled
with palhal
dentine

Figure 11 Reconstruction of a tooth drawn from Figure
8B rotated anticlockwise about 90°. Features
marked. Scale applies to horizontal axis only.

Tooth plates of Devonian lungfish Dipterus

105

Presumably the interstitial dentine was converted
to translucent material, because the perforations
were connected by microscopic canals along which
histogenic substances were transmitted. Under
crossed polars there is a band of dark-coloured
material, and around that is a band of lighter
coloured interstitial material. The size of the
translucent material is not as large as one would
expect from the other figured specimens.

The cross sections examined optically all have
very similar structure. This is very important
because it is necessary to interpret the section given
by Smith (1984, figure 51).

Figures 13B,C,F from G2004.10.12 show a larger
tooth which has been cut from near its base. The
central part of Figure 13B shows the pattern where
around each pulp canal is light coloured material
(translucent dentine). Marginal to that the dark
brown material (interstitial dentine) shows up well
near the margins of the white material, but becomes
more obvious marginally where the white material
becomes narrower. The central part of the tooth is
the oldest section and newer elements were added
to the margins. Figures 13C and 13D show up well
the different arrangement of the crystals in the
translucent and interstitial dentines. In these figures
the pallial dentine is very narrow, and gives a better
impression of the fully grown tooth.

Compare Figures 13B,C, F with those of 13A,D,E,
which is a smaller specimen that has lost its outer
edges.

CRYSTALLINE ARRANGEMENT OF THE
DENTINE

In this section we examine the crystallographic
features of the tissue described above (see Figures
13-14). The dark coloured zone around the pulp
canals in the central part of the core as shown up
under crossed polars, could be the interpreted in
two w'ays. It may consist of fluorapatite crystals
with the Z-axis approximately vertical; or
alternately a random array of fluorapatite crystals,
appreciably smaller than the thickness of the
section, so that their net interference colour is close
to zero. SEM images of the structure show an array
of crystals and perforations which suggest that the
second interpretation is correct.

Under crossed polars, there is another ring of
light coloured material around this central ring.
This seems to be made of crystals more or less
parallel with the surface of the section, and fibrous
bundles have weak rough ends on the light bands.
This band represents the interstitial dentine.
Examination under crossed polars and a gypsum
plate, but with two quarter wave plates set at right
angles forming a Benford Plate (Craig 1961), has
been used, because this plate eliminates the
quadrate effect produced by the extinction position.

The results of this are best shown on Figures 13E,F
and 14D where the circularity of the interstitial
dentine is more obvious than on the sections under
crossed polars with a gypsum plate. In the comers
between the separate pulp canal units there is often
a small gap between the circular interstitial dentine.
This would have been expected from the cross
sections on Figure 8D where the arrows show gaps
where the interstitial dentine has not been
transformed. In the areas where the interstitial
dentine has been altered, these small patches show
up with a slightly different pattern.

In the marginal areas of the tooth each pulp canal
has its own discrete, sometimes incomplete ring,
which have been described above, and which is not
in contact with the adjacent rings. The rings are
separated by a dark coloured layer which is
connected with the interstitial layer deeper in the
tooth. Under crossed polars, with the gypsum plate
and with the Benford plate inserted, the
interference colours on the edge of these rings
shows that the tissue represents the translucent
dentine. This observation should be compared with
the diagram of Smith (1984, figure 51).

COMPARISON WITH THE INTERPRETATION
GIVEN BY SMITH (1984)

Smith (1984, figure 51) commented that her cross
section of a tooth showed "birefringent bands of
opposite sign in tissue between the dentine adjacent
to the pulp canals". The central core of the tooth
shows the dark array of crystallites formed from
the translucent columns around the pulp canals.
Note that this dark band of crystallites becomes
narrower in the pulp canals closer to the tooth
margin, and then opens out into the marginal pallial
dentine. This is as we have described above for our
material. The light bands in the figure are the
interstitial dentine which has been transformed in
the central part of the tooth, but marginally it has
the appearance of isolated interstitial dentine. As
we have shown from our specimens (Figures 8D-F;
13A,D,E) this marginal tissue is crystalline and is
surrounded by dark layers which we interpret as
interstitial dentine. Figure 51 of Smith's paper
matches our interpretation exactly.

The statement by Smith (1984, figure 51) on the
figure is not very meaningful, and the comment on
page 394 that the presence of petrodentine is
indicated by "bands of birefringence of opposite
signs produce a woven appearance in polarized
light (figure 50)" requires clarification. The inner
layer around the pulp canals is translucent dentine
and the outer layer is interstitial dentine which has
an independent origin from the translucent dentine.
The composite nature of the core dentine, made up
of interstitial and translucent dentine, make it
difficult to interpret her statements.

106

The other two of her figures of D. valenciennesi
(figures 49-50) are also difficult to interpret. The
plane polarized illustration (figure 49) has many
pores and the core shows little differentiation into
columns. Smith's figure 50, photographed under
crossed polars, has patterns difficult to interpret
compared with our Figure 4. Despite this, there is
no doubt that it is the same tissue.

COMMENTS ON PETRODENTINE

The reader should examine the definition of
petrodentine, a term first used by Lison (1941). In
his original paper Lison makes the following points
- petrodentine is light in colour; looks roughly
homogeneous except near the pulp cavity where it
contains cellular prolongations that are petroblasts;
does not take up biological stains; much less
birefringent than osteodentine in polarized light;
calcified bands similar to the collagen bands; and
contains little organic matter. Subsequent work by
authors working on Protopterus and Lepidosiren
which were used by Lison (Smith 1985; Kemp 2001)
shows that petrodentine continues to grow from the
from the core dentine in the earliest formed teeth;
and petrodentine contains no denteons. Smith
(1984) listed in a Table the criteria for the
recognition of petrodentine, and this has been a
valuable guide.

In the light of our current observations, we note
the following characteristics of the core dentine in
Dipterus valenciennesi.

(a) It was deposited at several levels in the tooth
core, and was not deposited only in the early
growth stages of the tooth, (b) Translucent dentine
was first deposited from the pulp canals, in some
instances well away from the basal pulp cavity, (c)
Pulp canals are present even to the apical core of
the tooth, (d) Tubules appear in the translucent
dentine near the apex of the tooth, and extend into
the pallial dentine, (e) Thin sections under crossed
polars show that the growing translucent material
was deposited in layers around the pulp canals, (f)
Interstitial dentine occurs between the translucent
dentine, and was first formed in the basal pulp
cavity and also in the margins of newly formed
additions to the tooth as it increases its width, (g)
As the tooth grew, the interstitial dentine gradually
converted to what appears to be translucent
dentine. Additions to the tooth margin during
growth also shows similar modifications to the
interstitial dentine, (h) Both translucent and
modified interstitial dentine contain large numbers
of canals throughout their structure.

The next question arises - is there any evidence
that the translucent material was deposited from a
special layer of cells known as petroblasts?
Alternatively could it have been deposited from
some other kinds of cells spread more widely

J.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

through the pulp canals? Naturally one cannot
observe petroblasts in fossils and so one has to
observe features which indicate the presence of
special cells from the position and distribution of
certain hard tissues.

Firstly, the cells depositing the translucent
dentine must have been very widely distributed
and they were active during much of the life of the
tooth. Secondly, the translucent material was not
deposited to make a hard surface on which the wear
of the teeth could be curtailed. It is even deposited
in the core of the basal pulp cavity well away from
the wear surface. Thirdly, the translucent material
in the most exposed core of the tooth contains large
numbers of tubules which would not have
strengthened the tooth against wear. These points
indicate that the cells used to deposit the
translucent material were not petroblasts. The
layering of the translucent material and its
distribution show that the cells forming the material
must have formed on the surface of the pulp canals
throughout history of the tooth.

HISTOLOGY OF 'COSMINE' ON THE MEDIAN
PALATE

Depositional Sequence of the 'Cosmine'

As we have indicated above, the distribution and
histology of the materials forming the 'cosmine' is a
matter of concern. The presence of an enamel (see
below) layer on the surface indicates that an
epithelial layer must have been present. As the
tooth plates grew anteriorly and medially, the new
sequences of 'cosmine' were added. Each 'cosmine'
unit has a down-turned edge, or an edge against
which the new layer of dentine was formed. This
means that the sequence was not the result of
resorption and redeposition, but rather a sequence
of successive depositional layers. The same
conclusion was reached by Denison (1974: 41).

The pattern of deposition figured by Denison
(1974) is different from what we observe in our
specimens, and from specimens figured by White
(1965, plate 2, figures 2-4). The surface of specimen
G2004.10.1 shows the central parts of the palate
(Figure 2B). The right palatal plate has five layers of
shiny smooth substance laid down in sequence. The
oldest layer, labeled 4, lies up against the first
formed teeth in the median tooth row. On the left
plate the sequence is not so clear, but the pattern is
the same. Some of the increments show fine
perforations, but the others do not. Some show
perforations along the line of increments, and some
of these are up to 0.5mm in diameter. In addition
the innermost layer covers up the posterior part of
this tooth row, and on the left plate it extends
posterior to the tooth plate. Presumably this means
that soft tissue covered the posterior part of the

Tooth plates of Devonian lungfish Dipterus

107

200|jm

Figure 12 SEM cross sections of the cosmine on the palate. A-C, G2004.10.12; tooth on the left; single arrow marks
overlap of completed unit; double arrow another unit completed on left side. B, right side of Figure 12A;
white layer around pulp canals; bone at base. C, laminar character of the bone. D and E, G2004.10.2, largo
arrows indicate the mid-line; small arrows are pores in surface. F, G2004.10.2, cross section of cosmine, bone
with osteocytes deeper. G, enlargement of the surface fine pores in the enamel; dentine tubules clear.

108

plate in the later periods of growth, and this would
have been possible because this part of the palatal
plate was not in contact with the mandibular plates
in the later stage of growth. From the study of this
specimen and the illustrations given by White
(1965), we conclude that the distribution of the
layers and the arrangement of the pores is highly
variable, not only between the specimens but also
on the two sides of the one specimen.

This still leaves unsolved the significance of the
pores in the 'cosmine'. It is unlikely that they
contained sensory tissue, but they must have
allowed contacts between the soft tissue over the
surface, and soft tissue in the canals. Enamel was
deposited from an epithelium and this would have
been served by nutrients carried through the pores
from the canals. This would also account for the
lack of symmetry in the distribution of the pores,
and also for the concentration of pores along the
junction between successive layers in some
specimens, where the growing edge would have
needed a supply of nutrients. But there are further
tests that can be carried out before we reach a
conclusion about the use of the term 'cosmine'.

Internal Structure of the 'Cosmine'

One test of the 'cosmine' hypothesis would be the
presence of pore-canals beneath the pores in the
surface. We note that the specimen figured by
White (1965, plate 2 figure 1) does have a surface
with many pores, and pores are found in places on
some of our specimens. The shiny surface of the
'cosmine' is formed of radially arranged crystals as
is normal for enamel. The enamel layer is
perforated by abundant pores in Figure 12G. The
pores open into the underlying layer with
triangular pores (Figure 12D,E), but they have no
indication of any internal structure. No section we
have examined shows any sign of pore-canal
systems, and in this respect it shows no similarity
to most dermal cosmine. The large pores penetrate
beneath the dentine pores into the underlying bone,
but they have no lined connections to the
surrounding tissues. The pores are just the means
by which the nutrients were transferred to the
epithelium which covered the palatal surface of the
'cosmine' during development.

Under the layer of tissue containing the dentine
tubules is a layer with pulp canals surrounded by a
light coloured material (Figure 12A,B), and around
these are layers of banded material. The banded
material therefore extends deeply into the light
layer (Figure 12B). Beneath that is a layer with
complex folding but without cytoplasmic spaces
(Figure 12C). This layer of tissue lies directly on
bone (Figure 12A).

Most of the above description is derived from
G2004.10.12, but similar features can be seen on
G2004.10.2.

J.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

COMPARISON WITH OTHER PRIMITIVE
DIPNOANS

The structure of the dentine in the cores of the
teeth of the tooth plates of Dipterus valendennesi
has structures different from other Early and
Middle Devonian tooth plates whose details now
have to be investigated. The following types of
plates have been described - Dipnorhynchus and
Speonesydrion from the Emsian of New South
Wales (Campbell and Barwick 1984, 2000);
Tarachomylax from the Early Devonian of
Severnaya Zemlya (Barwick etal. 1997); Ichnomylax
from the Taimir Peninsula (Reisz et al. 2004); and
WestoIIrhynchus, from the Hunsriick Mountains in
Germany (Schultze 2001). We do not include the
Canadian Early Devonian genus Melanognathus
here, despite Schultze's (2001) comments. It is a
denticulated form with marginal teeth.
WestoIIrhynchus is based on a single specimen
from which no histology has been obtained. One
can only consider this genus as having dubious
validity.

Tarachomylax is the only genus which has plates
comparable with those of Dipterus. Histologically,
the translucent layers are separated from one
another by layers of material described as
interstitial dentine. No internal structure of the
interstitial dentine was observed. These layers
extend into the basal pulp cavity (Barwick et al.
1997, figure 9: 1-4; figure 10: 1-2; figure 13).
Deposition of the translucent material took place
from cells in the pulp canals. As shown by Barwick
et al. (1997, figure 11:2), the translucent dentine is
porous as is the translucent material in Dipterus.
Because of the unusual features of this dentine, the
authors could not refer to it as petrodentine, and
we used the term 'compact dentine'. This view was
queried by two of the reviewers of the paper who
complained that there was no need for a new term.
Barwick et al. persisted with the new name, but
they did not suggest that it should have a formal
status. They did indicate that this dentine did not
have the characters of normal petrodentine.

This brings us to a position where we have to
discuss why there are so many types of dental
structures appearing in the Early Devonian
(Campbell and Barwick 1990). Palatal biting first
appeared in the Early Devonian, and this is a major
change from marginal biting. In Dipnorhynchus the
margins are added to by small enamel-covered
excresences and the plate thickens by deposition of
dentine at the bone dentine boundary. In
Speonesydrion the conical teeth are added
marginally to the tooth plate, and thickening takes
place by the deposition of new dentine at various
points at the bone-dentine boundary. Uranolophus
has marginal enamel covered ridges around the
plate margins and small denticles covering the main
mass of the plates (Campbell and Barwick 1988).

Tooth plates of Devonian lungfish Diptems

109

Figure 13 G2004.10.5B. A, in single polarized light; D, under crossed polars; and E with a gypsum plate and a Benford

Plate. A, core dentine is finely perforate. D, pulp canals surrounded by dark layers of translucent dentine;
bands around the translucent dentine are modified interstitial dentine, grey in colour and forming quadrate
bands. E, zones around the pulp canals wider, and the interstitial dentine narrower than in D; circular
arrangement of the interstitial dentine bands clearer; from middle of tooth to base, translucent dentine
becomes narrower; in F. rings of coloured bands formed of altered transitional dentine surrounded by dark
interstitial dentine. B,C,F, G2004.10.12. 'Fhree similar photos of a single tooth; significance of the different
colours explained in the text; F, emphasizes the small new dentines on the lower side, and the larger more
open pulp canals at the top where new translucent dentine was being added.

110

].L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

Figure 14 G2004.10.5A. Single tootln. A, in single polarized light; B, under crossed polars; C, under gypsum plate; D,

gypsum plate and a Benford Plate. D, slightly rotated, and interstitial dentine shows a more circular pattern.
E, G2004.10.9. Three teeth on a single slide; also illustrated on Figures 4-6.

Tooth plates of Devonian lungfish Dipterus

111

And Tarachoiin/lax has teeth as described above.
That is, there are four types of plate formation
introduced in the Early Devonian. Only teeth of
the Tamchomylax type are successful in the later
history of the Dipnoi. It has been shown that teeth
are added marginally in later dipnoans and have a
360 Myr history (Krupina 1995, Reisz and Smith
(2001) and Smith and Krupina (2001).

All of these types are discrete; they cannot be
transformed from one type to the other. This then
raises the question of the origin of new structures at
a time when a major new development takes place
in evolution - in this case palatal rather than
marginal biting. This matter has been di.scussed in
works by Raff (1996), Shubin and Marshall (2000)
and Minelli (2003). It is becoming apparent that the
introduction of new major features in the phenotype
is probably the result of production by gene
regulation. As Shubin and Marshall (2000, p. 331)
report... 'major evolutionary changes may not be
due to changes in the number or structure of genes
per se, but may be due to changes in their regulation
(Carroll, 1995, 2000). Indeed the changes in the
spatial pattern and timing of the gene activity play
an important role in generating variation at both
small and large phylogenetic scales.' This article was
reviewed by Russell (2001). The new designs have a
genetic basis, and there is no W'ay these designs can
be changed from one type to another.

If gene regulation is the controlling factor in
producing new designs in organisms with palatal
biting, as the above quotation indicates, similar
regulation factors may have operated on each of
the basic designs later in their history.
Consequently each of the tooth patterns found in
the Early Devonian could have developed new
structures, and these would be the basis for
outlining changes in the Middle and Upper
Devonian. So the use of cladistic methods to
recognize relationships in these later forms will
involve comparison between genera which have
already separated into groups which have
separated by gene regulation. Hence comparison
of so-called synapomorphies will involve
comparisons which are convergences carried over
from the primitive forms which gave rise to the
original dispersion. For this reason the cladistic
analysis of the kind given by Schultze (2001),
which uses the statistical methods to develop
character-state optimization, will be valueless.
Attention must be paid to such a possibility in
outlining the evolutionary pattern in late
Palaeozoic dipnoans A more complete discussion
is given in work on Speoiiesi/drioii (Campbell and
Barwick in press).

ACKNOWLEDGEMENTS

We are indebted to several people who have

helped us during the preparation of this work.
Mr Jack Saxon of Thurso has had many years
surveying the Old Red Sandstone sequences of
Caithness and the Orkneys, and has made useful
collections of dipnoans and other fishes. He has
made some dipnoans available to us. Michael
Newman working with den Blaauwen on the
biostratigraphy of the Old Red Sandstone of
Scotland, has given helpful discussions on the
subject, and has contributed specimens. He has
read the manuscript. Professor R.A. Eggleton of
the Australian National University, has
examined the material from a crystallographical
point of view. The Scanning Electron Images
were made in conjunction with Dr Roger Heady
of the Scanning Electron Unit at A.N.U. Some
optical thin sections were made by Mr Tony
Phimphisane. Thin cuts were made by Mr Harri
Kokkonen of the Research School of Earth
Sciences, A.N.U. Professor Robert Reisz kindly
provided us with a proof copy of the paper on
Idnioiuyhix. Dr Robert Saint discussed the issue
of c/s regulatory structures with us.

REFERENCES

Agassiz, J.L.R. (1833-44). Rechcrchcs sur les poisson
foiisih. Text (5 vols) and Atlas (5 vols). Petitpierre,
Neuchatel.

Agassiz, J.L.R. (1844-45). Monographic ties Poisson
Fossiles dll Vieux Gres Rouge on Systeine Devonien
(Old Red Sandstone) des lies Britanniijues et de
Riissie. Text and Atlas. Jent and Gassman, Neuchatel.
Ahlberg, P.E. and I'rewin, N.H. (1995). Tlie postcranial
skeleton of the Middle Devonian lungfish Dipterus
valeneiennesi. Transactions of the Royal Society of
Edinburgh: Earth Sciences 85: 159 -175.

Astin, T.R. and Rogers, D.A. (1991). Subaqueous
shrinkage cracks in the Devonian of Scotland re-
interpreted. journal of Sedimentary Petrology 61:
850-859.

Barwick, R.E. and Campbell, K.S.W. (1996) A Late
Devonian Dipnoan, Pillararhynchus, from Gogo,
Western Australia, and its relationships.
Palaeontographica A 239: 1-42.

Barwick, R.E., Campbell, K.S.W. and Mark-Kurik, E.
(1997). Tarachomylax: a new Early Devonian
dipnoan from Severnaya Zemblya, and its place in
the evolution of the Dipnoi. Gcobios 30: 45-73.

Bishop, I.R. and Foxon, G.E.H. (1968). The mechanism of
breathing in the South African lungfish, Lepidosiren
paradoxa; a radiological study, lournal of the
Zoological Society of London 154: 263-271.
Campbell, K.S.W. and Barwick, R.E. (1984).
Speonesydrion, an Early Devonian dipnoan with
primitive tooth plates. Palaeolchthyologica 2: 1M8.
Campbell, K.S.W. and Barwick, R.E. (1987). Paleozoic
lungfishes - a review. In W.E. Bemis, W.W. Burgren
and N.E. Kemp (eds). The biology and evolution of
lungfishes. journal of Morphology , Supplement 1:
93-131.

112

Campbell, K.S.W. and Barwick, R.E. (1988)
Uranolophus: a reappraisal »f the a primitive
dipnoan. Memoirs of the Assoeintion of Austrninsinii
Palaeontologists 7: 87-144.

Campbell, K.S.W. and Barwick, R.E. (1990). Paleozoic
dipnoan phylogeny; functional complexes and
evolution without parsimony. Paleobiologi/ 16: 143-
169.

Campbell, K.S.W. and Barwick, R.E. (1998). A new
tooth-plated dipnoan from the Upper Devonian
Gogo Formation and its relationships. Memoirs of
the Queensland Museum 42: 403-437.

Campbell, K.S.W. and Barwick, R.E. (1999). Dipnoan
fishes from the Late Devonian Gogo Formation of
Western .Australia. Records of the Western
Australia}! Museum Supplement 57: 107-138.

Campbell, K.S.W. and Barwick, R.E. (2000). The
braincasc, mandible and dental structures of the
Early Devonian lungfish Dipuorin/nchus kurikae
trom Wee Jasper. New South Wales. Records of the
Australian Museum 52: 103-128.

Campbell, K.S.W. and Barwick, R.E. (in press). The
structure and stratigraphy of Speonesydrkvt, and the
dentition of primitive dipnoans.

Campbell, K.S.W, and Smith, M.M. (1987). The
Devonian dipnoan Holodipterus: dental form and
variation and remodelling growth mechanisms.
Records of the Australian Museum 39: 131-169.

Carroll, S.B. (1993). Homeotic genes and the evolution of
arfhropods and chordates. Nature 376: 479-485.

Carroll, S.B. (2000). Endless forms: the evolution of gene
regulation and morphological diversity. Cell 101:
577-580

Craig, D.B. (1961). The Benford Plate. The American
Minei-alogist 46: 757-758.

Crampton, P. and Carruthers, R.E. (1914). The Geology
of Caithness (Sheet 110 and 116 loith Parts of 109,
115 and 117). Geological Survey of Scotland Memoir.

Denison, R.H. (1974). The structure and evolution of
teeth in lungfishes. Fieldiana Geology 33: 31-58.

Donovan, R.N. (1980). Lacustrine cycles, fish ecology
and strafigraphic zonation in the Middle Devonian
of Caithness. Scottish journal of Geology 16: 35-50.

Donovan, R.N., Forster, R.J. and Westoll, T.S. (1974). A
stratigraphical revision of the Old Red Sandstone of
north-eastern Caithness. Transactions of the Royal
Society of Edinburgh 69: 167-201.

Forster-Cooper, C. (1937). The Middle Devonian fish
fauna of Achanarras. Transactions of the Royal
Society of Edinbingh 59: 223-239.

Kemp, A. (2001). Petrodenline in derived tooth plates.
journal of Vertebrate Paleontology 21: 422-437.

Krupina, N.l. (1995). Comparison of larval dentition
developmental patterns in Devonian and Recent
dipnoans. Ichthyolith Issues Special Publication 1:
35-38

Lison, L. (1941). Rechcrches sur la structure et
I'histogenese des dents des Poissons Dipneustes.
A)-chivcs dll Biologic 52: 279-320.

Lund, R., Bartholomew, P. and Kemp, A. (1992). The
composition and dental hard tissues of fishes. In P.
Smith and E. Thernov (eds). Structure, Function and

J.L. den Blaauwen, R.E. Barwick, K.S.W. Campbell

Evolution of Teeth. Freund Publishing House, Tel
Aviv, pp. 35-71 .

Minelli, A. (2003). The development of animal form,
ontogeny, morphology, and evolution. Cambridge
University Press, Cambridge.

Pander, C.H. (1858). liber die Ctenopterydien des
Dei’onischen Systems. St Petersberg.

Raff, R.A. (1996). The shape of life: genes, development,
and the evolution of animal form. Chicago
University Press, Chicago.

Reisz, R.R., Krupina, N.l. and Smith M.M. (2004). Dental
histology in Ichnomyla.x karatai sp. nov., an Early
Devonian dipnoan from the Taimyr Peninsula,
Siberia, with a discussion on petrodentine. journal of
Vertebrate Paleontology 24: 18-25.

Reisz, R.R. and Smith, M.M. (2001). Lungfish dental
pattern observed for 360 Myr. Nature 411: 548.

Rogers, D.A. and Astin, T.R. (1991). Ephemeral lakes,
mud pellef dunes and wind-blown sand and silt:
re-interpretations of Devonian lacustrine cycles in
north Scotland. In P. Anadon, L. Cabrera and K,
Kelts (eds), Lacustrine Facies Analysis.
International Association of Sedimentologists
Special Publications 12: 201-223.

Russell, D.A. (2001). Review of: Deep time:
paleobiology's perspective. American Scientist 5:
475-476.

Schultze, H.-P. (1975). Das Axialskelet der Dipnoer aus
dem Oberdevon von Bergisch-Gladbach
(Westdeutchland). Problemes actuel de
Paleontologie-Evolution des Vertebres. Colloque
International C.N.R.S. 218: 149-159.

Schultze, H.-P. (2001). Melanognath us, a primitive
dipnoan from the Lower Devonian of the Canadian
Arctic, and the interrelationship of Devonian
dipnoans. journal of Vertebrate Paleontology 21:
781-794.

Schultze,H.-P. and Chorn, J. (1997). The Permo-
Carboniferous genus Sagenodus and the beginning of
modem lungfish. Contributions to Zoology 67: 9-70

Sedgwick, A. and Murchison, R.l. (1829). On the
structure and relations of the deposits contained
between the Primary Rocks and the Oolitic series in
the north of Scotland. Ti-ansactions of the Geological
Society of London, scries 2, 3: 125-160.

Shubin, N.H. and Marshall, C.R. (2000). Fossils, genes,
and the origin of novelty. In D.H. Erwin and S.L.
Wing (eds). Deep Time: Paleobiology's Perspective.
Paleobiology 26(4), Supplement; 324-340.

Smith, M.M. (1984). Petrodentine in extant and fossil
dipnoan dentifions; microstructure, histogenesis and
growth. P}'oceedings of the Linncan Society of Nezo
South Wales 107: 367-M07.

Smith, M.M. (1985). The pattern of histogenesis and
growth of tooth plates in the larval stages of extant
lungfish. journal of Anatomy 140: 627-643.

Smith, M.M. (1989). Distribution and variation of enamel
structure in the oral teeth of sarcopterygians: its
significance for the evolution of a profoprismatic
enamel. FUstorical Geology 3: 97-126.

Smith, M.M. and Krupina N.L (2001). Conserved
developmental processes constrain evolution of
lungfish dentitions, journal of Anatomy 199: 161-168.

Tooth plates of Devonian lungfish Dipterus

113

Thomson, K.S. (1969). Gill and lung function in the
evolution of lungfishes (Dipnoi): an hypothesis.
Forma et Functio 1: 250-262.

Trewin, N.H. (1976). Correlation of the Achanarras and
Sandwich fish beds. Middle Old Red Sandstone,
Scotland. Scottish Journal of Geology 12: 205-208.

Trewin, N.H. (1986). Palaeoecology and sedimentology
of the Achanarras fish bed of the Middle Old Red
Sandstone, Scotland. Transactions of the Royal Society
of Edinburgh: Earth Sciences 77: 21^6.

Trewin, N.H. and Thirlwall, M.F. (2002). Old Red
Sandstone. In N.H. Trewin (ed.). The Geology of
Scotland, 4"’ Edition. Geological Society of London,
London, pp. 213-249.

Watson, D.M.S. and Day, H. (1916). Notes on some
Palaeozoic Fishes. Memoirs and Proceedings of the
Manchester Literary and Philosophical Society 60:
1-52.

Watson, D.M.S. and Gill, E.L. (1923). The structure of
certain Palaeozoic Dipnoi. Journal of the Linnean
Society, Zoology 25: 163-261.

Westoll, T.S. (1949), On the evolution of the Dipnoi. In
G.L. Jepson, E. Mayr and G.G. Simpson (eds).
Genetics, Palaeontology and Evolution. Princeton
University Press, Princeton, pp. 121-184.

White, E.L (1965). The head of Dipterus valenciennesi
Sedgwick and Murchison. Bulletin of the British
Museum (Natural History), Geology 11: 3-45.

White, E.L (1966). Presidential Address: a little on lung-
fishes. Proceedings of the Linnean Society of London
177: 1-10.

Manuscript received 10 December 2004; accepted 26 July 2005

mmmmm

Guide to Authors

Subject Matter:

Reviews, observations and results of research into
all branches of natural science and human studies
will be considered for publication. However,
emphasis is placed on studies pertaining to Western
Australia. Longer papers will be considered for
publication as a Supplement to the Records of the
Western Australian Museum. Short communications
should not normally exceed three typed pages and
this category of paper is intended to accommodate
observations, results or new records of significance,
that otherwise might not get into the literature, or
for which there is a particular urgency for
publication. All material must be original and not
have been published elsewhere.

Presentation:

Authors are advised to follow the layout and style
in the most recent issue of the Records of the
Western Australian Museum including headings,
tables, illustrations and references.

The title should be concise, informative and
contain key words necessary for retrieval by
modem searching techniques. An abridged title (not
exceeding 50 letter spaces) should be included for
use as a mnning head.

An abstract must be given in full length papers
but not short communications, summarizing the
scope of the work and principal findings. It should
normally not exceed 2% of the paper and should be
suitable for reprinting in reference periodicals.

The International System of units should be used.
Numbers should be spelled out from one to nine
in descriptive text; figures used for 10 or more. For
associated groups, figures should be used
consistently, e.g., 5 to 10, not five to 10.

Spelling should follow the Concise Oxford
Dictionary.

Systematic papers must conform with the
International Codes of Botanical and Zoological
Nomenclature and, as far as possible, with their
recommendations.

Synonymies should be given in the short form
(taxon, author, date, page) and the full reference
cited at the end of the paper. All citations, including
those associated with .scientific names, must be
included in the references.

Manuscripts:

The original and two copies of manuscripts and
figures should be submitted to the Editor, c/-
Publications Department, Western Australian
Museum, Locked Bag 49, Welshpool DC. Western
Australia 6986. They must be in double-spaced
typescript on A4 sheets. All margins should be at
least 30 mm wide. Tables plus heading and legends
to illustrations should be typed on separate pages.
The desired position for insertion of tables and
illustrations in the text should be indicated in
pencil. Tables should be numbered consecutively,
have headings which make them understandable
without reference to the text, and be referred to in
the text.

High quality illustrations are required to size
(16.8 cm X 25.2 cm) or no larger than 32 cm x 40
cm with sans serif lettering .suitable for reduction to
size. Photographs must be good quality black and
white prints, not exceeding 16.8 cm x 25.2 cm.
Scale mu.st be indicated on illustrations. All maps,
line drawings, photographs and graphs, should be
numbered in sequence and referred to as Figure/s in
the text and captions. Each must have a brief, fully
explanatory caption. On acceptance a computer disk
containing all corrections should be sent with
amended manuscript. The disk should be marked
with program (e.g. Word, WordPerfect, etc).

In papers dealing with historical subjects
references may be cited as footnotes. In all other
papers references must be cited in the text by author
and date and all must be listed alphabetically at the
end of the paper. The names of journals are to be
given in full.

Processing:

Papers and short communications are reviewed
by at least two referees and acceptance or rejection
is then decided by the editor.

The senior author is sent one set of page proofs
which must be returned promptly.

The senior author will receive fifty free offprints
of the paper. Additional offprints can be ordered at
page proof stage.

Records of the Western Australian Museum
Volume 23 Part 1 2006

CONTENTS

H.S. Gill, D.L. Morgan, R.G. Doupe and A.J. Rowland

The fishes of Lake Kununurra, a highly regulated section of the Ord

River in northern Western Australia 1

David L. Morgan, Andrew Chapman, Stephen J. Beatty and Howard S. Gill

Distribution of the spotted minnow (Galaxias maculatus (Jenyns, 1842))

(Teleostei: Galaxiidae) in Western Australia including range extensions

and sympatric species 7

Mark S. Harvey

A new species of Lechytia from eastern Australia

(Pseudoscorpiones: Lechytiidae) 13

R.G. Gunn

Mulka's Cave Aboriginal rock art site; its context and content 19

L.B. Bean

The leptolepid fish Cavenderichtbys fa/bra^arensfs (Woodward, 1895)

from the Talbragar Fish Bed (Late Jurassic) near Gulgong, New South Wales 43

D. Franklin and L. Freedman

A bioarchaeological investigation of a multiple burial associated with the

Batavia mutiny of 1629 77

Jan L. den Blaauwen , Richard E. Barwick and Kenton S.W. Campbell
Structure and function of the tooth plates of the Devonian lungfish

Dipterus valendermesi from Caithness and the Orkney Islands 91

Dates of Publication

Records of the Western

Australian Museum

Volume 22, Part 1

18 December 2003

Volume 22, Part 2

15 April 2004

Volume 22, Part 3

14 October 2004

Volume 22, Part 4

20 July 2005

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 "Records of the Western Australian Museum"

See other formats