Content uploaded by Peter Veth
Author content
All content in this area was uploaded by Peter Veth on Feb 12, 2015
Content may be subject to copyright.
EXCAVATIONS AT PARNKUPIRTI, LAKE GREGORY, GREAT SANDY DESERT: OSL ages for
occupation before the Last Glacial Maximum
Author(s): Peter Veth, Mike Smith, Jim Bowler, Kathryn Fitzsimmons, Alan Williams and
Peter Hiscock
Source:
Australian Archaeology,
No. 69 (December 2009), pp. 1-10
Published by: Australian Archaeological Association
Stable URL: http://www.jstor.org/stable/27821528 .
Accessed: 06/02/2015 00:58
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
Australian Archaeological Association is collaborating with JSTOR to digitize, preserve and extend access to
Australian Archaeology.
http://www.jstor.org
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
EXCAVATIONS AT PARNKUPIRTI,
LAKE GREGORY, GREAT SANDY
DESERT:
OSL ages for
occupation before the Last Glacial Maximum
Peter
Veth1,
Mike Smith2
5,
Jim
Bowler3,
Kathryn Fitzsimmons4,
Alan
Williams5 and Peter Hiscock6
Abstract
We report on early occupation from the Parnkupirti site
on Salt Pan Creek at Lake Gregory, on the edge of the
Great Sandy Desert of northwest Australia. OSL ages
from excavations, and stratigraphie correlations between
dated exposures along Salt Pan Creek, show some stone
artefacts in situ in
sediments dating from greater than 37ka
and most probably on stratigraphie grounds in the range
of ~50-45ka. The deep stratigraphie section at Parnkupirti
also provides a long record of the Quaternary history of
Lake Gregory, which remained a freshwater system during
the Late Quaternary.
Introduction_
The Lake Gregory system in northwestern Australia is the
desert terminus for the
monsoonal catchment of Sturt Creek
draining the
eastern
Kimberley (Figure 1).
Previous studies
have shown that the lakes preserve a palaeomonsoon record
for northwest Australia (Bowler 2006; Bowler et al. 2001).
Relict shorelines and lake deposits - as well as dunefields
truncated by the palaeolake - show that Lake Gregory was
much larger at times during the late Quaternary, and that it
incorporated the smaller basins that make up the Gregory
Paruku system
today (Figure 1).
The juxtaposition
of a rich
freshwater ecosystem - with an abundance of birds, mussels
and fish
- set
within arid dunefields
means that the lake is
likely to have been a focal point for late Pleistocene hunter
gatherer groups in this region. There are obvious parallels with
the
Willandra Lakes system in southeastern Australia (Bowler
1998;
Bowler et
al. 2003) and elsewhere in
northern Australia
at
Lake
Woods (Bowler
et
al 1998;
Hutton et
al 1984;
Smith
1986;
Veth 1980).
One point of difference, however, is that
Lake Gregory is still active.
The challenge for archaeologists has been to identify an
early phase of occupation associated with Pleistocene lake
phases. In
2006-2007, geomorphic
fieldwork
by
Bowler (2006)
identified
a series
of localities
where artefacts
appeared to
be
eroding from strata associated with an expanded lake phase. A
1
National Centre for Indigenous Studies, The Australian National
University, Canberra, ACT 0200, Australia peter.veth@anu.edu.au
2
Centre for Historical Research, National Museum of Australia,
GPO Box 1901, Canberra, ACT 2601, Australia m.smith@nma.gov.au
3
School of Earth Sciences, University of
Melbourne, VIC 3010, Australia
& School of Geography and Environmental Science, Monash
University, Clayton, VIC 3800, Australia jbowler@unimelb.edu.au
4
Research School of Earth Sciences, The Australian National University,
Canberra, ACT 0200, Australia kathryn.fitzsimmons@anu.edu.au
5
Fenner School of Environment and Society, The Australian National
University, Canberra, ACT 0200, Australia alan.williams@anu.edu.au,
m.smith@nma.gov.au
6
School of Archaeology and Anthropology, The Australian National
University, Canberra, ACT 0200, Australia peter.hiscock@anu.edu.au
? SaS Pan Creek
{Pamkupfcli)
Figure 1
Geomorphic map of the Lake Gregory system, showing the
boundaries of the palaeolake. Today, Lake Gregory is made up of a
series of smaller basins, collectively known as the Gregory-Paruku
lakes.
The lower inset (outlined in
bold on the geomorphic map) shows
the Parnkupirti sites along Salt Pan Creek near Mulan community.
preliminary
series
of
optically
stimulated
luminescence
(OSL)
age determinations suggested ages of 20-50ka for these units
(Bowler
2006).
Our work in 2008 followed
up these finds with
an initial series of archaeological excavations to clarify whether
or not there was a late Pleistocene archaeological record. This
paper reports on these excavations and discusses the future
potential of the Lake Gregory system for archaeological and
palaeoenvironmental research.
Regional Context_
The Lake Gregory system represents a rare set of circumstances,
where freshwater flowing into the desert juxtaposes two
distinctive ecosystems: an alkaline freshwater system set
within an
arid environment. The area has an array of past shoreline features
demonstrating periods when the lake area was some 10 times
Number 69,
December 2009 australian ARCHAEOLOGY 1
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Excavations at Parnkupirti, Lake Gregory, Great Sandy Desert
M^^^^^^j^^j ^^^^^^
Figure 2 View of the creek section at Parnkupirti Site 3 (L-R: Jim Bowler, Peter Veth,
Wade Freeman and Mulan community in
frame) (Photograph:
Mike Smith).
HHP a
WmMM/^
Figure 3 Site 3. Units A-C and location of OSL samples from the
Parnkupirti section (Photograph: Mike Smith).
larger
than today,
interspersed
with phases of dune building
representing episodes of enhanced aridity. This represents
an important archive of environmental history and of greatly
amplified monsoonal activity. Earlier investigations (Bowler et
al 2001;
Wyrwoll and
Miller 2001) used thermoluminescence
(TL) dating to
provide
a broad chronology
for
the lake
system
spanning the past 200,000 years. Later work, using OSL, allows
some refinement of the Quaternary history of Lake Gregory. At
the present time the lake stands at an intermediate phase with
water level near 280m, well inside the perimeter of a former
mega-lake phase where palaeoshorelines rose to 296-300m
(Figure
1).
Salt Pan Creek Drainage
The northern catchment of Sturt Creek is supplemented by
significant flow in creeks from the east and southeast. Salt Pan
Creek and
Djaluwon Creek drain catchments
rising
in
low
hills
some 10- 15km east of the lake.
Our data are
derived
mainly
from
stratigraphy exposed along
the banks of Salt Pan Creek. Exposures on both the southern
and northern
banks of Salt Pan Creek (Figures 1-3) show a
major sequence
of channel
gravels
with interbedded fluvial
and
slackwater silt
and clay
deposits spanning
the last
lOOka.
Eight
OSL samples were taken from the section and pits on the northern
bank of the creek
(Site
3), and six
OSL samples
were collected
from equivalent stratigraphie units from the southern bank
(Site
2) to refine
the
existing
palaeoenvironmental
record
for
Lake Gregory. Bowler (2006) also mapped ironstone and silcrete
artefacts
within the
gravels
of
Djaluwon Creek (to the
south),
which are heavily rounded and appear to be associated with a
high-energy pluvial phase almost certainly of Pleistocene age.
Stratigraphy along Salt Pan Creek
A sequence
of
wet to
dry
environmental
changes is
defined
by
stratigraphie units from the northern and southern banks of
Salt
Pan Creek. Lying
approximately
5m
below the surface
of
the
plain and incised into older lake sediments,
the
modern
channel carries bedload sands. Lying several metres above
2 aUStralian ARCHAEOLOGY Number
69,
December 2009
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Peter Veth et al.
Figure 4 Correlation of stratigraphie units and location of OSL samples across Parnkupirti Sites 1-3. Vertical scale from top of Unit E to base of Unit
A is c.5m. Not? that OSL sample K2006 (37.2?5.8ka) is
from Unit D overbank sediments inset into Unit C. The PKP4-1 core is located
within the upper
levels of
Unit C pre-dating deposition of Unit D lake
muds.
the channel floor and bearing no relationship to the modern
regime, a lag of pebbles and cobbles remains from an earlier
high discharge phase. By contrast, modern mobile gravels are
absent, suggesting that even during modern monsoonal rains,
discharges are insufficient to carry bedload larger than sand
and occasional pebbles. This stands in
marked contrast to the
presence
of earlier beds of large
pebbles and cobbles,
mainly
of shiny dark brown quartzite, often greater than 10cm in
diameter. Constituting torrent gravels, these represent periods
of greatly increased discharge signaling the onset of former
high lake levels, an observation supported by the presence of
overlying
limestone
plates (Site 1,
Unit ID,
Figure
4).
Site 1_
Located on the
relatively
steep
northern bank (Figure
4), the
lower 2-4m (Unit 1A) consists of cemented and strongly
weathered lake limestone. These relate to an earlier mega-lake
phase around 100-120ka, with basal levels even older (Bowler
et al. 2001). Weathered limestone passes up to a l-2m zone of
karstic weathering defined by red clays and secondary calcrete
nodules (Unit IB). Lying some 4m above the channel
floor,
this
weathering
zone identifies
a long dry period during
which
the lake failed to rise to this level.
This dry interval
occurs
immediately after the 100-120ka major expansion.
The red
clays
are
overlain
abruptly
by
a thin
layer
of brown
silcrete
pebbles (Unit 1C),
evidence
of the first
return
of
water
to this
level
after the
long
dry interval.
These fluvial
bedload
sediments are in turn overlain by a thin plate of
microcrystalline
limestone
(Unit ID), evidence of
quiet shallow
water
deposition
some 5-6m higher than the present lake level. Karstic erosion,
which has produced pitting
and fluting
of its
upper surface,
reflects
a period
of
drying
and
weathering
during
a subsequent
fall
in lake levels.
The final depositional signature involves a thin 5-10cm cover
of
quartz
sand
and tabular
silcrete
pebbles (Unit IE).
These grade
laterally some 100-150m north from the central channel to
merge
into beach sands, a zone commonly associated with artefacts.
This represents a final late phase of lake expansion. Wind action
removing
fines results
in
a lag
deposit
of tabular
pebbles
forming
a distinct desert pavement with an underlying red silt layer, the
product of dust accumulation protected by the surface stone cover.
In summary, a long weathering period after -lOOka was
interrupted
by a phase of strong
fluvial
discharge
depositing
coarse clastic bedload over margins of the lake some 5m above
present water level. Following a period of quiet-water limestone
deposition, lake levels fell,
marking another dry interval. A final
high
lake
phase
deposited
the
uppermost
sand
units (IE) on the
northern bank. This sequence can be correlated with both the
units on the south bank and the upstream sequence.
Site 2_
The southern bank
of
Salt
Pan
Creek is
marked
by
wider channel
excavation, in contrast to the steeper cliff
on the north. Although
deposited
on a lateral
tributary,
the
units
at this site
(Figure
4)
bear a strong stratigraphie correlation with those at Site 1. In
this context, shallow water limestone in Unit ID is equivalent to
deeper water, muddy sediments at Site 2. This zone is frequently
coated by a white saline efflorescence containing sodium
sulphate, residual salts from alkaline lake waters. OSL ages for
two adjacent sections - Sites 2.1 and 2.2 - are given in Table 1
and shown in Figure 4.
The stratigraphie
sequence
at
Site
2 (Figure
4) shows
gravels
and cobbles over a basal weathering horizon (Unit 2.IB) passing
laterally
into
a thick
l-2m zone of hard red
sandy clays (Unit
2.1C). These grade upwards into a well-developed soil with
prismatic and polygonal domal structure (Unit 2.ID) overlain
by 20-30cm of red A-horizon clayey sands (Unit 2.IE). Some
30-40cm below the
surface,
a
band of
single pisolithic
or
pelletai
ironstones defines a line, rising to the south to
merge with beach
sands near the surface (the corollary of the surface pelletai layer
at Site 1). It is also associated with late Holocene artefacts such
as tulas and tula slugs.
At Site 2, a major erosion surface (equivalent to the B-C
disconformity at Site 1), truncates basal sediments and is
overlain by channel gravels providing direct correlation between
north
and south
bank sites
(Figure
4).
An
OSL sample (K1528)
from immediately below the disconformable contact provides
an age near 52ka, giving an estimate for the end of a dry phase
and initiation of burial. On the south bank, red clayey sands
were deposited in slack water as
muddy lakeshore environments
(equivalent
to
limestone Unit ID on the north
bank) and
provide
basal ages
near
45ka grading
up to
37ka and 21ka.
Ages of 18ka
Number
69,
December 2009 auStralian ARCHAEOLOGY 3
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Excavations at Parnkupirti, Lake Gregory, Great Sandy Desert
and 5ka
are
from
samples likely
to
be subject
to
disturbance
and
bioturbation from a level near the surface soil - and we treat
these as
minimum ages only.
This sequence is consistent
with the onset of fluvial
deposition near 50ka leading to lake expansion, and deposition
of near-shore muddy sediments by 45ka. While these sediments
correlate
closely
with the cobbles and limestone units (1C
and ID) at Site 1,
the actual age of lake
retreat
(equivalent
to
karstic
development
of
Site 1
limestone)
is
not readily
apparent
at Site 2. That still-stand has been obscured here by the return
and deposition
of
equivalent muddy facies which
have all
been
oxidised and homogenised
by pedogenic
modification. In this
context, deposition beginning near 50ka and lasting until at least
30ka is consistent with all evidence currently available.
Site 3 _
Located some 2km upstream from Site 1
on the north bank of
Salt
Pan
Creek,
a cutting
(Figures
3-5) exposes
3m
of
gravels
and
cobbles overlain by 2m of surface red clayey sands grading onto
the surrounding plain. Our main archaeological investigations
lie
within the
upper
2m
of
this section.
The cutting
displays three
distinctive units separated by
stratigraphie breaks (Figure 4). In the lowermost Unit A, indurated
bands
of
sandy
limestone with
reworked
cobbles of the
underlying
limestone are associated with bands of lithic sands and gravels
heavily cemented by secondary carbonate. Well-defined traces of
near-horizontal bedding relate this unit to a coarse textured lake
marginal deposit. In its uppermost 20-30cm, vertical fractures,
showing mottling
with carbonate
leaching, identify
a substantial
time break before
deposition
of the
overlying
Unit
B.
Unit B contains well-rounded gravels and cobbles of brown
Palaeozoic quartzite, and cobbles of lacustrine limestone.
Cobbles are frequently
>5-10cm in diameter and reflect
high
velocity
discharge events involving large
quantities of
water
discharging into the lake. Frequent carbonate encrustations
on cobble surfaces point towards deposition of lacustrine
limestone in the gravel interstices consistent with submergence
for
a relatively long
period
beneath
expanded lake
waters.
This
is further
supported
by a stone-free
upper 20-30cm of
highly
weathered yellowish-red
sandy clays
with prismatic jointing,
evidence of a substantial time break before deposition of Unit C.
Unit C is a 1-1.3m thick layer of well-rounded gravels and
cobbles of dark brown silcrete, platy siltstone and associated
Palaeozoic bedrock fragments. With an imbricate depositional
fabric,
these
gravels
represent
a phase of
major, high
velocity
discharge similar to the fluvial
system
responsible for the
underlying Unit B. However, Unit C cobbles bear no trace of
subsequent limestone deposition but are associated with a
matrix of brick-red clayey sands. The red sands that fill
voids
in the coarse cobble fabric are incompatible with synchronous
cobble transport, and were deposited from the overlying
lacustrine muds.
Unit C is
separated
from
underlying
Unit B by
a well-defined disconformity representing a long time interval.
The lensing gravels
of
Unit C grade into
the
basal sector
of
the
overlying
hard red
sandy clays
(Unit D). The latter include
polygonal vertisol structures enclosing a pelletai, often pisolithic
band of ironstone, both features identical with equivalent levels
on the
south
bank (Site
2,
Figure
4).
Excavations at Parnkupirti Site 3_
Site 3
was chosen for a pilot excavation because it
was one of the
few
points along Salt
Pan Creek where a deep (3.5m) vertical
section was exposed. A narrow elevated terrace led back from
this face, representing a residual bench of Unit D palaeolake
sediments
30m
wide (Figures
5-6).
A low
density
but extensive
scatter of stone artefacts (averaging <1 artefact/m2) occurs on
the surface of this terrace as well as in a heavily scoured and
eroded area immediately to the
west. Our excavations in
August
Table 1
OSL data and age estimates for Parnkupirti Creek North and South banks (Sites 3 and 2, respectively). Samples K1528, K2006 and K2007
represent the
most reliable age estimates, yielding Gaussian or dominant age peaks (with
varying degrees of saturation or
mixing present in other
samples - consistent with dynamism of lake
muds).
Unit Lab No. Depth De K U Th j
Cosmic f
Water Total Dose Age
{m> (Gy) {%) (ppm) (ppm) j Dose Content Rate (ka}
I Rate (%) <Gy/ka)
(Gy/ka?
North Bank (Site
3)
K2056 2.6?0.1 239?42 0.73?0.04 1.36?0.07 19.5?0.98 0.1 ?0.01 4+2 2.45?0.12
K2059 1.4+0.1 240?8 0.79?0.04 1.43?0.07 15.8?0.8 0.19?0.02 6?3 2.26+0.12
K2061 1.8+0.1 250?23 0.56?0.03 1.32?0.07 21.5+1.1 0.17+0.02 4?2 2.44?0.12
K2062 1.1+0.1 264?17 0.87?0.04 1.77?0.09 15.6+0.8 0.20+0.02 4?2 2.4?0.11
K2063 1.1+0.1 126?7 0.85?0.04 1.7?0.09 16.2?0.8 0.20?0.02 4?2 2.48?0.11
K2064 0.1+0.1 9.6?0.2 0.48?0.02 1.50?0.08 11.6?0.58 0.22+0.02 3?2 1.8?0.24
K2065 1.0+0.1 105+1 0.57?0.03 1.63?0.08 12.2?0.61 0.2?0.03 5?2 1.90?0.10
K2066 0.5?0.1 75.8?0.9 0.63?0.03 1.64?0.08 11.5?0.57 0.21 ?0.05 5?2 1.92?0.10
Site 2
K1528 2.0?0.2 84.2?3.5 0.63?0.03 1.4?0.07 8.4?0.40 0.17?0.02 5?2 1.61 ?0.07
K2007 2.0?0.2 64.8?2.4 0.49?0.03 1.29+0.07 7.99?0.40 0.17?0.02 3?2 1.44?0.07
K1529 0.9?0.1 28.2?3.5 0.40?0.02 1.08?0.05 7.53?0.38 0.20?0.03 5?2 1.29?0.06
K2006 0.9?0.1 75.3?10.5 0.98?0.05 1.55?0.08 12.65?0.68 0.19?0.03 5?2 2.02?0.13
K1530 0.5?0.1 23.1 ?2.0 0.40?0.02 1.02?0.05 7.38?0.37 0.21 ?0.04 5?2 1.28?0.07
K2005 0.5?0.1 8.2?0.7 0.44?0.02 1.31?0.07 9.01?0.45 0.21 ?0.05 15?5 1.54?0.08
4 australian archaeology Number 69, December 2009
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Peter Veth et al.
Figure 5
Archaeological excavations at Parnkupirti Site 3, showing stratigraphie correlations between the creek section and the excavation pits. OSL
samples from Unit C were taken from a silty horizon interleaved within the fluvial cobbles (shown as grey lens). For Pits PKP1 to PKP3, brackets
indicate the vertical distribution of artefacts in
Unit D.
Figure 6 Plan of Parnkupirti Site 3, showing the location of excavated pits and the distribution of artefacts on the eroded area to the
west. Relative
heights show local topography.
Number 69,
December 2009 aUStralian ARCHAEOLOGY 5
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Excavations at Parnkupirti, Lake Gregory, Great Sandy Desert
Figure 7 Site 3. Pit PKP2 showing the Unit C cobbles, pisolith horizon
and Unit D sediments.
2008 sought to clarify
the stratigraphie
provenance of these
artefacts, following Bowler's earlier reconnaissance, which had
suggested that at least some were eroding from Pleistocene
sediments
(Bowler
2006).
Three Im x 2m excavation trenches (PKP1-PKP3) were laid out
in a transect running 50m back from the creek section (Figure 6).
These were excavated with picks, trowels and mattocks and all
sediments were screened for artefacts. The heavily indurated nature
of
Unit
D lake sediments
meant that
the
maximum
depth
reached
in any of these pits was 70cm. Therefore, a fourth and deeper trench
(PKP4)
was excavated
with
a
backhoe,
providing
a 1.8m
section ito
the lake deposits at a point 37m back from the creek section.
Collectively these excavations revealed an upper unit of
heavily
indurated lake
muds (yellowish-red,
5YR 5/8-6/8,
clayey
silts) up to 127cm thick,
with a well-developed polygonal
structure and vertical cracks extending to 1
m below surface. This
unit
- labelled
Unit
D - is divided into
upper
and lower sections
by
a thin
band
of
pisoliths
at
70cm
below surface.
Unit
D overlies
Unit C - a bed of large subangular and rounded cobbles (to
150mm)
in
a
matrix
of
pisoliths
and red
(2.5YR 5/8)
clayey
sand.
The surface of Unit C slopes down at 1? from the creek section
(where it is exposed at the surface) and so was only reached in
pits
PKP3 and
4 (Figure
7).
The excavations recovered a small number of stone artefacts
(N=14) in
Unit
D and all from the
top
25cm of the unit.
The
larger pieces (>40mm) may be in situ. However, some are small
enough to have been introduced into these levels via vertical
cracks in these lake sediments - and their provenance is therefore
not secure.
Table 2 provides a summary description of artefacts
excavated from Unit D. These artefacts are made on local chert
and quartzite - the latter probably sourced in streambed cobbles
within a few hundred metres of the excavation site. None of the
flakes was retouched and the assemblage generally reflects core
reduction to produce large flakes. Our field observations at Site 3
indicate that the
surface of
Unit
D has been stripped
and eroded,
and that artefacts on the scoured and eroded areas west of the
excavations have been reworked from the upper part of this unit
(Figure
6).
Artefact
typology
points to
a late
Holocene age for
much of this material as the scatter includes tulas and tula slugs
Table 2 Description of artefacts excavated from Unit D.
Artefact Excavation Provenance
PKP1 Spit 1
Artefact Description
Medial fragment
of
a flake,
with bulb and portion
of eraillure
scar
preserved
on the
ventral face and negative scar terminations on the dorsal face._
PKP1 Spit 3 Flaked
piece containing
one large negative step-terminated
scar (8mm long)
truncated by one of several breaks._
PKP1 Upper 25cm Complete chert
flake
with intact
platform,
clear ring
crack
and bulb
on ventral
surface, and negative scars on the dorsal face. Length=9.4mm._
PKP1 Upper 25cm Proximal fragment of quartzite flake. Ring crack and bulb preserved. It
was
removed from a
water-smoothed cobble and contains the cortex on the dorsal
face. Length=25.5mm._
PKP1 Upper 25cm Complete quartzite
flake with cone and bulb visible on the
ventral face
and one
negative scar on the dorsal face. Length=43.6mm._
PKP1 Upper 25cm Chert core with four negative scars originating from a single conchoidal platform.
Weight=U5g._
PKP2 Spit 1 Quartzite flake from a water-rounded cobble. Cortex .covers the dorsal face. Ring
crack and bulb visible on the ventral surface. Hinge terminated. Length=22.7mm.
PKP2 Spit 1 Chert flaked piece. Negative scars originate from the same platform. No ventral
surface is
discernable, instead there is a flat surface which may reflect the piece
has split along a pre-existing plane. Length=10.9mm._
PKP2 Spit 1 Chert
flake with distinct
point
of impact
and cone. Lengths
16.5mm.
10 PKP2 Spit 2 Chert flake with visible ring
crack and bulb on the ventral
surface. On the dorsal
face there are seven negative scars. Length=41.5mm._
11 PKP2 Spit 2 Quartz flaked piece. It
contains five negative scars deriving from the same platform
surface. Weight=4.2g._
12 PKP3 Spit 2 Complete chert flake.
A distinctive
ring
crack and bulb is
present
on the
hinge
terminated ventral surface. On the dorsal face there are seven negative scars from
the same platform surface. Lengths7.9mm. _
13 PKP3 Spit 2 Proximal end of
a chert
flake. Bulb with eraillure
scar is
preserved on the
ventral
face. Two negative scars from the same platform are visible on the dorsal face.
Length=11.3mm._
14 PKP3 Spit 6 Flake from a quartz cobble. It has a clear fracture initiation point. Length=19.4mm.
6 australian archaeology Number 69, December 2009
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Peter Veth et al.
Figure 8 View of the flaked face of PKP4-1 showing scars free of
encrustations (Photograph: Peter Hiscock).
and is dominated by use of cryptocrystalline stone such as desert
chalcedony (chalcedonic chert). A mantle of cobbles forms an
apron in places along the
margin of the scoured area and shows
that erosion -has also cut into Unit C. Some weathered silcrete
artefacts are present in the artefact scatters. These may be the
counterpart of the PKP4-1 core excavated from Unit C, but as
yet
we cannot securely establish their provenance.
In PKP4 a large silcrete core (PKP4-1) was recovered in
situ
within the Unit C cobble bed (Figures
8-9).
As this
was
the most significant find during the 2008 excavations some
comments on its provenance are warranted. The artefact was
discovered
whilst cleaning
the
PKP4 section.
Only its
edge
was
exposed after the section was cleaned back and it took some
time to extricate the artefact from the tightly-bound cobble
bed.
The core lay
platform
down in
the
deposits at a depth
of
149.5cm below ground surface - 23cm beneath the surface of
Unit C and sealed within the cobble bed.
A thorough
search of
other
exposures
of
Unit C showed
that flaked
or
angular
pieces
are otherwise absent from this cobble bed. In the absence of
evidence for rolling and abrasion, the presence of PKP4-1
within the
cobble layer
suggests
that
Unit C was laid down in
several lenses. People appear to have collected a silcrete cobble
on the lake
margin, then flaked
and discarded the
core,
which
was later covered by more torrent gravels.
Artefact Description: PKP4-1_
PKP4-1 is
a single
platform
core (159.7g)
made on a rounded
cobble of local silcrete
(Figures
8-9). As this artefact is late
Pleistocene in age a detailed description is
given below.
Condition of the Core
Carbonate encrustations have built up on the rounded surface
of the cobble
but do not cover
the
flake
scars,
indicating
that
the
knapping
took
place after
transport
of the cobbles in
Unit
C.
There are none of the impact points or small scars that natural
mechanical damage might produce. The flake surfaces are
uniformly
weathered and scar
edges
show
rounding
(especially
on the
platform edge), consistent
with in situ
weathering (but
not rolling/abrasion).
This eliminates
any
possibility
that
the
scars are recent and were produced during the excavation.
The Pattern of Flaking
PKP4-1 has six negative conchoidal scars. All appear to have
hertzian initiations and are positioned on one part of the piece.
The pattern of flake removals is clear and is distinctive of human
knapping. The cobble is an irregular, angular hemisphere with a
^^^^^^^^^^
Figure 9 The order of flaking of the PKP4-1 core (Photograph: Peter
Hiscock).
relatively flat surface on one side. This surface is
flatter at one end
and it is here that the scars are located, a situation more consistent
with human
knapping
choices than
mechanical damage (which
would easily have occurred at the thinner, angular end with a
raised surface). The first scar runs across this flat surface and has
removed a localised rise, thereby making the surface flatter and
more suited as a platform. Subsequently five blows were struck to
this prepared platform surface, removing scars from an adjoining
face. All five blows were orientated in a similar direction, and the
order
of flake removal
was (looking
at the
flaked
face)
right
to
left,
with the
exception
of
the final small flake
removal.
This
was
struck so that it ran down a ridge created by the intersection
of the
4th
and 5th flakes.
The 5th flake
was removed
by
a blow
struck on the edge of a conchoidal platform surface - this being
preferable as a platform rather than the smooth natural exterior.
This pattern, of platform construction then regular flake
removal, restricted to the conchoidal platform and moving
progressively
from
right
to left
for
four
flakes,
is
undoubtedly
patterned in a
way that is consistent with human knapping and
inconsistent with natural processes.
OSL dating at Parnkupirti Sites 2 and 3_
Eight
OSL samples
were collected from
Site
3,
and six
samples
were taken from Site 2 (Table 1). Samples
were collected
by
hammering light-tight steel tubes into cleaned surfaces. All
samples were capped immediately upon extraction to
minimise
exposure to light.
Methods
Samples were processed under low intensity red light to extract
the 125-180um modal size quartz fraction. Carbonates, organic
materials and heavy minerals \vere removed by hydrochloric
acid and hydrogen peroxide reaction, and sodium polytungstate
density
separation,
respectively. Etching by hydrofluoric
acid
removed the outer parts of the grains exposed to
a-radiation and
any feldspars present in the sediments. Samples were cleaned in
hydrochloric
acid then loaded
onto stainless steel
discs as small
aliquots containing fewer than 50 grains each. Between 18 and
24 aliquots were measured for each sample, with an additional
three
aliquots
for each sample initially
measured to identify
the
age range. OSL measurements for equivalent doses were made
using the Single
Aliquot Regenerative (SAR) dose protocol
(Murray
and Wintle 2000,
2003) on automated
Riso TL-DA-12
and TL-DA-15 readers with OSL attachments (Botter-Jensen
1997; Botter-Jensen et al. 2000). Following a preheat plateau
test on sample K2066 using six different preheat temperatures
Number 69,
December 2009 australian ARCHAEOLOGY 7
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Excavations at Parnkupirti, Lake Gregory, Great Sandy Desert
Figure 10 Dose distributions of OSL samples. (A) Natural OSL decay of sample K2056. (B) Indicative dose-response curve for sample K2056. (C)
and (D) Radial plots showing dose distributions for small aliquots for samples (C) K2063 (Unit C) and (D) K2065 (Unit D). Note the
mixed dose
populations for
both samples, indicating that these particular samples cannot be relied upon for
age determinations.
ranging from 180?-280?C, the maximum preheat temperature
on the
plateau (240?C)
was selected
and applied to
all samples.
The SAR protocol used incorporated four regenerative dose steps,
plus
an
additional
repeated
dose step
to
check
recycling
(Murray
and
Wintle 2003).
Test
doses
of
approximately
5
Gy
were applied
following each regenerative dose step, and measurement of the
natural OSL signal.
All samples
exhibited
bright
luminescence
signals
that
rapidly
decayed
within the
first
few
measurement
channels
(Figure
10A),
indicating
dominance of the
fast
component
OSL signal.
The
samples
exhibited luminescence
behaviour indicating
suitability
to the
SAR protocol, including
recycling
ratios
close to
unity,
negligible thermal transfer, and minimal IRSL signals. However,
equivalent
dose distributions
yielded
either luminescence
signal
saturation
(Figure
10B)
or
high
degrees
of
mixing (Figure
10D)
in all
samples
except
K1528,
K2007 and
K2006. The ages
as
given
in this
paper either used the
weighted
mean or
minimum age
for
D calculation.
e
Dosimetry
Dose rates were calculated from radioactive element
concentrations analysed using inductively-coupled plasma
mass spectrometry (for uranium and thorium) and atomic
emission spectrometry (for potassium), undertaken at
Genalysis
Laboratories, Perth. Radionuclide concentrations were converted
to beta and gamma dose rate components using the conversion
factors of
Adamiec and Aitken (1998). The present average water
content was used to calculate dose rate attenuation by moisture
(e.g.
Aitken 1998),
with significant
uncertainties of >33%
incorporated into the calculations to allow for variations in
moisture content through time. Cosmic dose rate
was calculated
using the formulae
of Prescott
and
Hutton (1994),
with an
additional soft component to the dose rate added to account for
the shallow
depth
of
sample
K2064.
The shallow
depth (0.1m)
of
this sample was also used to correct for the lack of 4jt geometry
in the gamma component of the dose rate. Internal alpha dose
rates were considered negligible in these quartz sediments. The
OSL data are presented in
Table 1
and correlations between Sites
1-3 are summarised in Figure 4.
Interpretation of the OSL Chronology
The samples taken from the two lowest stratigraphie units at
Parnkupirti Site 3,
Units A and B, were saturated with respect to
the OSL signal
and therefore
yield
minimum ages
only.
It
is
not
8 aUStralian ARCHAEOLOGY Number
69,
December 2009
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Peter Veth et al.
possible to interpret these ages any further than suggesting that
the shallow lacustrine and older fluvial sediments corresponding
to these units were deposited at least 100,000 years ago.
Despite the
apparent
suitability
of the OSL samples
for
dating
using the SAR protocol, the dose distributions of those samples
not saturated with respect to the luminescence signal were widely
distributed, with signs of post-depositional mixing or incomplete
bleaching (Figure
10).
There
were three
exceptions
to
this
(K1528,
K2007,
K2006) - all
from
Site
2.
These
yielded
dose distributions
in the form of Gaussian populations or dominant age peaks, and
therefore more reliable age estimates were calculated. This is
attributed to the
more homogeneous nature of the sediments at
this location.
K1528 and
K2007 lie within
2 standard deviations of
one another, and are therefore statistically the same age, ~50-45ka.
Although
K2063 from Site
3 gave
a similar
age (52ka)
using
the
weighted mean of all aliquots, three age peaks were identified
in
different
aliquots of this
sample (at
c.25ka,
50ka and 75ka).
Therefore, this age estimate is only an average. Nevertheless,
stratigraphie correlations between Sites 2 and 3 constrain the age
of Unit C to ~40-50ka. By association, the PKP4-1 core is older
than 37ka and
given
its location well within the Unit
C gravels
is
most likely to be in the 50-45ka range.
The remaining OSL samples are subject to a number of
limitations. The mixed dose distributions (observed for all
samples except K2056, K2061, K2062, K1528, K2007 and
K2006) may be a consequence of several factors. In Unit C, the
sediment
heterogeneity
(with
a
mixed deposit of cobbles and
pebbles in a fine-grained matrix) contributed to variability in
dose rate estimates, resulting in a
wide age distribution between
measured aliquots. Moreover the luminescence signal may not
have been completely reset within these sediments, since rapid
deposition took place under turbulent subaqueous conditions
with insufficient exposure to sunlight.
Within Unit D, evidence of bioturbation (in particular, termite
activity) and swelling and cracking of lake muds indicates some
post-depositional mixing. Only one sample (K2006) yielded
a reliable age (37.2?5.8ka). This shows that little time elapsed
between deposition of Units C and D, consistent with deposition
of lake
muds (Unit
D) during
an expanded lake
phase initiated
by the torrent gravels in Unit C.
Stratigraphie correlation between Sites 1,2 and 3 underpins
the chronology presented in this paper. The major disconformity
(B-C) underlying
the last
phase of torrent
gravels
at all sites
represents a crucial benchmark. From a relatively horizontal
expression
at Site 1,
Unit C rises to Site 2 (Figure
4). Here
torrent gravels grade laterally into, and are in part overlain
by, red clayey sands, sedimentologically and stratigraphically
equivalent to Units C and D at Site 3 respectively. At Site 2
the cobbles and upper 1.5m of clayey sands are constrained
by an age of ~50-45ka at the disconformity, with the upper
unit (equivalent to
D at Site 3) deposited around 37ka.
A
depositional hiatus was followed by a later
period of lake
expansion
depositing the final
lake mud facies (Unit IE, Site
1, Figure 4). Pedogenesis followed, with substantial termite
modification in upper levels. In summary, the evidence is
consistent with the onset of rapid torrent gravel deposition
near 50ka. The rapid nature of deposition suggests that the age
of the upper gravels will be close to that of the onset of lake
mud deposition, and some time well before 37ka.
Lake Gregory: A Northern Analogue for Lake
Mungo?
The archaeology and geomorphology of the Lake Gregory
system have great potential for reconstructing prehistoric
human-environment relationships in arid northwest Australia.
The promise, of course, is that Lake Gregory will eventually
prove to be a northern analogue for the
Willandra Lakes, with
a richly detailed record of late Pleistocene subsistence and
occupation around freshwater palaeolakes on the margins of
the arid
zone.
The difficulty
is that Lake
Gregory is
a
much
more
challenging geomorphic landscape for this kind of research.
The size of the lake system
makes it
difficult
to identify
focal
points for archaeological research. Expansion and contraction
of the palaeolake has left a spatially diffuse suite of Quaternary
landforms and sediments, where the correlation and relative
sequence of geomorphic units is not straightforward and
rests largely on luminescence dating. Lake Gregory does not
have the deep stratigraphie and archaeological windows that
are provided by extensive natural erosion in the
Willandra
Lakes system.
A multiyear and staged approach is needed to disentangle the
complex environmental and human signals that are present in
this highly active monsoonal area. In this context, our find of a
core stratified within sediments dating to ~50-45ka, is
a first step.
This in itself is significant. For the first time, it indicates human
activity within the arid northwest of the continent from an open
context dating before the Last Glacial Maximum.
The trench
at
Parnkupirti
Site
3 will be reopened
in 2009 and
further samples taken at the base of Unit D to better constrain
the age of the PKP4-1 core. Background gamma spectrometry
will be undertaken to provide better estimates of in situ dose
rates and further excavations across the interface of C-D will test
for further in situ artefacts.
Lake Gregory has been a permanent freshwater body on the
edge of the desert since humans first occupied the northwest. This
landscape has enormous potential for reconstructing deep-time
human and environmental histories for northwestern Australia.
Acknowledgements
We are indebted to the community of Mulan and resident
traditional owners for work on the Parnkupirti site, a major
and 'open Two Dog Dreaming mythological site. It
was noted
by several senior owners that this creation story represented the
formation of Lake Gregory and surrounds (known in
Walmajarri
as 'Paruku). That this should be the earliest site from the region
was seen to be congruent with their origin narratives for the
lake and its people. Some 300 people from Mulan, Billiluna
and Balgo communities visited the site in various capacities -
from the stage of gaining traditional owner consent, through to
excavation and processing finds with IPA Rangers, structured
visits and participation by school children and post-excavation
presentations to the community and schools.
This work
was funded
by
ARC Linkage
Grant LP0776332
'Rock Art and Jukurrpa Sites of the Canning Stock Route'; the
Kimberley Foundation of Australia; the Paruku Indigenous
Protected Area Program and Jo
McDonald Cultural Heritage
Management Pty Ltd. Partners in this Linkage Grant (of
relevance here) include the Land and Sea Management Unit
of the Kimberley Land Council; Department of Environment,
Number 69,
December 2009 auStralian ARCHAEOLOGY 9
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
Excavations at Parnkupirti, Lake Gregory, Great Sandy Desert
Water, Heritage and Arts; the Department of Indigenous
Affairs (DIA); Western Desert Lands Aboriginal Corporation
and Kanyirninpa Jukurrpa; the Department of Environment
and Conservation (WA); and JMCD
CHM. Special thanks
go
to
KLC IPA staff Wade Freeman and
Gillian Harvey for field
support; Chair of Mulan Community (Steven Yumari) who
operated the backhoe for excavation of PKP4; Sandra Wallace,
Sam Higgs, Kathryn Przywolnik, Robert Reynolds, Kevin
Dan, John Carty for excavation, processing and traditional
owner liaison in the field. Alan Williams drew Figures 1, 4-6
for
publication. Finally,
thanks
to
Kim
Mahood for
linking
the
artists of Mulan to the project. The KLC Research and Ethics
Committee contact is Jenny Bolton (jenny.bolton@klc.org.au).
References
Adamiec, G. and
MJ. Aitken 1998
Dose-rate conversion
factors:
Update. Ancient
TL 16:37-50.
Aitken,
M. J.
1998
An Introduction
to
Optical
Dating.
Oxford: Oxford
University
Press.
Botter-Jensen,
L. 1997
Luminescence techniques: Instrumentation
and methods.
Radiation
Measurements 27:749-768.
Botter-Jensen,
L., E. Bulur,
G.A.T. Duller and A.S. Murray 2000 Advances in
luminescence instrument
systems.
Radiation
Measurements 32:523-528.
Bowler,
J.M.
1998
Willandra Lakes revisited:
Environmental
framework for human
occupation.
Archaeology in Oceania 33(3):120-155.
Bowler,
J.M.
2006
Mulan Community
Cultural
Resource Surveys.
Unpublished report
to
Australian Institute of
Aboriginal
and Torres Strait
Islander
Studies,
Canberra.
Bowler,
J.M.,G.A.T.Duller,N.Perret,J.R.PrescottandK-H.Wyrwoll
1998
Hydrologie
changes in
monsoonal climates of the last
glacial cycle: Stratigraphy
and
luminescence dating of Lake Woods, N.T., Australia. Palaeoclimates 3(1
3):179-207.
Bowler,
J.M.,
K-H.
Wyrwoll and
Y. Lu 2001
Variations of the
northwest Australian
summer
monsoon over the past 300,000 years: The paleohydrological
record of the
Gregory (Mulan) Lakes System. Quaternary International
83-85:63-80.
Bowler,
J.M.,
H. Johnston,
J.M.
Olley,
J.R.
Prescott,
R.G.
Roberts,
W. Shawcross and
N.A. Spooner 2003 New ages for
human occupation and climatic change at
Lake
Mungo, Australia.
Nature 421:837-840.
Hutton, J.T.,
J.R.
Prescott
and
CR. Twidale 1984
Thermoluminescence dating of
coastal dune sand related
to
a higher
stand
of Lake
Woods, Northern
Territory,
Australia.
Australian
Journal
of
Soil
Research 22( 1
):
15-21.
Murray,
A.S. and A.G.
Wintle 2000 Luminescence dating of quartz using an
improved
single-aliquot
regenerative-dose
protocol.
Radiation
Measurements
32:57-73.
Murray,
A.S. and
A.G. Wintle 2003 The single
aliquot regenerative
dose protocol:
Potential for
improvements
in
reliability.
Radiation
Measurements 37:377-381.
Prescott,
J.R.
and J.T.
Hutton 1994 Cosmic ray
contributions to
dose rates for
luminescence and ESR dating: Large depths and long term variations.
Radiation
Measurements 23:497-500.
Smith,
M.A. 1986
An investigation
of possible Pleistocene occupation at Lake
Woods, Northern
Territory.
Australian
Archaeology
22:60-7'4.
Veth,
P. 1980 Sturt
Creek Survey.
Unpublished report
to
Australian Institute of
Aboriginal and Torres Strait
Islander
Studies,
Canberra.
Wyrwoll,
K-H. and
G.H.Miller 2001 Initiation of the Australian summer monsoon
14,000
years
ago.
Quaternary
International
83-5:119-128.
10 auStralian ARCHAEOLOGY Number 69,
December 2009
This content downloaded from 130.95.186.147 on Fri, 6 Feb 2015 00:58:48 AM
All use subject to JSTOR Terms and Conditions
