Evidence for Pleistocene seed grinding at Lake Mungo, south-eastern Australia
RICHARD FULLAGAR, ELSPETH HAYES, BIRGITTA STEPHENSON, JUDITH FIELD,CARNEY MATHESON, NICOLA STERN and KATHRYN FITZSIMMONS
RF, EH: University of Wollongong; BS: In the Groove Analysis Pty Ltd; JF: The University of New South Wales;CM: Lakehead University; NS: La Trobe University; KF: Max Planck Institute for Evolutionary Anthropology
ABSTRACT
Grinding stones and fragments have often been found in archaeological sites at Lake Mungo, south-western New South Wales, and theirfunction has mostly been inferred on the basis of grindstone morphology. Of particular interest has been the antiquity of grass seedgrinding, which is usually associated with deeply grooved, large sandstone dishes. Previous studies of grinding stones from the regionhave found no compelling evidence for seed grinding prior to the Pleistocene/Holocene boundary. One of the problems has been thatthe grinding stones have been found on deflated surfaces and have been difficult to accurately provenance and date. Here, we report afunctional study of 17 sandstone artefacts, recently collected from the central part of the Mungo lunette, where a suite of OSL ages haveprovided bracketing age estimates for the stratigraphic units. Ten artefacts are attributed to Unit E deposited between c.25 and 14 ka.Four artefacts are attributed to Unit F, deposited c.8 ka. Three artefacts from the Golgol lag are of unknown age. Usewear indicates alikely seed grinding function for 14 of the artefacts. Use-related residues include starch, cellulose and collagen. The results of this studyprovide additional support for Pleistocene plant processing and seed grinding activities in Sahul.
Correspondence: Richard Fullagar, Centre for Archaeological Science, University of Wollongong, Northfields Avenue,Wollongong NSW 2522, Australia. Email: [email protected]
INTRODUCTION
Archaeological investigations at Lake Mungo, a dry lakein semi-arid south-eastern Australia (Figure 1), haveidentified some of the oldest human burials, faunalremains, hearths, ochre, flaked artefacts and grindingstones known from Sahul (Pleistocene Australia – NewGuinea) (Bowler et al. 2003; Mulvaney & Bowler 1981).Grinding stones from Sahul were used for many tasks,including the production of ground-edge axes andhatchets; as well as the processing of bone, shell, ochre,small animals, medicines, drugs, poisons and numerousother materials, as witnessed ethnographically (see Gott2002). Key problems in Sahul prehistory have concernedthe antiquity of the grinding technology, the emergence ofseed grinding activities, and correlations with climatechange and consequent shifts in resource availability.Secure evidence for the processing and grinding of specificplant foods (such as seeds and tubers) in a Pleistocenecontext has been reported for only one site, CuddieSprings, in semi-arid south-eastern Sahul (Fullagar et al.2008).
This paper presents the results of a functional study of17 sandstone artefacts (including eight refitted fragments,LMGS2–9), all of which were collected from the centralpart of the Mungo lunette. Optically stimulated
luminescence (OSL) dating has provided bracketing ageestimates for the strata in this part of the lunette(Fitzsimmons et al. 2014), and thereby providesapproximate ages for the associated late Pleistoceneartefacts, including grinding stone fragments, which havebeen documented within the mapped region (Figures 1 &2; Fitzsimmons et al. 2014; Stern et al. 2013).
Allen (1974) has described the ethnographic evidenceand archaeological significance of grinding stones in thelocal and regional Aboriginal economies. There isabundant ethnographic evidence for the processing of grassseeds, which have been identified as a major subsistencebase in the arid regions of mainland Australia (Tindale1977). Smith (1985, 1989) has also documented thearchaeological significance and distinctive morphologies ofCentral Australian seed grinding implements, particularlythe large dished millstones and mullers, used forprocessing seeds. However, when dealing with smallgrinding stone fragments, the most common formrecovered from archaeological contexts, it is not alwayspossible to extrapolate the original shape or size of agrinding stone on the basis of morphology. Furthermore,there is no evidence to suggest that the grinding stonesused in the Pleistocene are necessarily the samemorphologically as those documented ethnographically inCentral Australia. The best indicators of artefact function
bs_bs_banner
Archaeology in Oceania, Vol. 50 Supplement (2015): 3–19DOI: 10.1002/arco.5053
are usewear traces and associated residues (Fullagar et al.1996, 2008).
Stone artefacts with ground surfaces have been found insome of the oldest archaeological sites in Australia(grinding stones – Fullagar & Field 1997; Kamminga &Allen 1973; Schrire 1982; Roberts et al. 1990; Smith 1985,1988: edge-ground hatchets – Balme & O’Connor 2014;Geneste et al. 2012; Jones & Johnson 1985: 217; Jones1985: 297; O’Connor 1999: 75). However, microscopicusewear and use-related residues on these implements haverarely been examined in any detail. In his reviews of seed
grinding in arid Australia and Pleistocene grindstonesreported from the Willandra Lakes and the Darling Basin,Smith (1985, 1986, 1988) could not identify anyPleistocene seed grinding stones, based on comparisonswith gross morphological features of ethnographic wet seedgrinding implements from Central Australia. Gorecki et al.(1997) canvassed a broader typological range and reductionsequences of Pleistocene seed grinding implements,subsequently questioning whether grinding tools used forseed processing necessarily had distinctive macroscopicforms such as grooves.
Figure 1. A map showing the location of the study area in the central Mungo lunette. Lake Mungo is part of a relictoverflow system, the Willandra Lakes, which were fed by waters that flowed westward from the Australian Alps duringperiods of reduced temperatures and evaporation. The white box marks the location of the foot survey work undertakenbetween 2009 and 2011; the black inset marks the location of the mapped area shown in Figure 2. The inset shows thelocation of the Willandra Lakes in the south-west corner of the Murray–Darling Basin and their relationship to theAustralian Alps.
The aim of this paper is to reassess the evidence for latePleistocene and early Holocene seed grinding at LakeMungo by adopting an integrated approach to analysingusewear, optically and biochemically visible residues andmacroscopic technological features of sandstone artefactsassociated with dated contexts.
THE LAKE MUNGO GRINDINGSTONE ASSEMBLAGE
Seventeen sandstone artefacts, thought to be possiblegrinding stones on the basis of macroscopic features, werecollected from exposed sediments in the central portion of
the Mungo lunette (Figure 2). The lunette was the focus ofa systematic archaeological foot survey between 2009 and2011 (Stern et al. 2013). The archaeological survey wascoupled with systematic geological mapping of the samearea between 2010 and 2013 and OSL dating of each of themapped geological units (Fitzsimmons et al. 2014; Sternet al. 2013). The survey targeted archaeological featuresand isolated finds in primary depositional context that wereexposed on the surface of the eroding lunette. Some rarefinds (including ground stone fragments, shell tools andfragments of ochre), whose encasing sediment had beenremoved, were also recorded, but only if the relationshipsbetween stratigraphic boundaries, palaeo-topographic
Figure 2. A geological map showing the location of the sandstone fragments (with field codes and LMGS no.) in relationto mapped and dated stratigraphic units. The location of this geological map with respect to the Mungo lunette is shown inFigure 1.
features and modern topography permitted assessment oftheir stratigraphic unit of origin. The artefacts lay on top orat the base of a residual landform, or on the surfaces oflow-angled slopes, in settings in which there was notopographic feature at higher elevation containingsediments from an older or younger stratigraphic unit.Furthermore, the artefacts lay in settings in which therewere no topographic features (such as a rill or gully) thatcould have transported them to their find locations.
The highly visible dense concentrations ofarchaeological debris that are strewn across the low,undulating surfaces at the toe of the Mungo lunette are acombination of lag and transported assemblages as well asmaterial derived from the stratigraphic unit on which theylie. These complex archaeological assemblages could bederived from any of the stratigraphic units accumulatedover the past 55 ka and have been excluded from analysis.For the same reason, archaeological debris lying on or inthe active gully systems (including the “overbank”sediments they deposit), and material lying on the surfaces
of extensive slope-wash surface with large, upslopecatchments, have also been ignored. Material found inmicro-topographic settings such as incipient rills that areperiodically scoured out and refilled with reworkedsediment has also been ignored, because of the highprobability that it had been re-cemented into thosesediments.
Fourteen of the 17 sandstone fragments described inthis paper come from one of two stratigraphic units(Table 1): Units E and F. Unit E was deposited during theperiod straddling the Last Glacial Maximum (LGM)between 25 ka and 14 ka (Fitzsimmons et al. 2014) – aperiod of lake-level oscillations fluctuating between fulland drying out, and ending with final lake retreat around14 ka. Unit E is equivalent to the Arumpo Unit from theschema by Bowler (1998). Sandstone fragments were alsoattributed to Unit F, which accumulated subsequent to thefinal retreat of the lake, during an episode of locally morearid conditions approximately 8 ka ago (Fitzsimmons et al.2014). Three of the 17 fragments were found on the
Table 1. Details of Mungo grinding stone fragments and morphological characteristics.
Depositionalcontext
Laboratorycode
Supportinginformation(SI) plate number
Grinding stonemorphology
Complete Surface Useconfidence
Surfacemorphology
Unit E LMGS1 1, 2 Recycled millstone/Upper
Fragment 1 3 Convex2 3 Concave
Unit E LMGS2 3, 4, 5 Lower stone Fragment 1 3 Flat2 0 Irregular
Unit E LMGS3 3 Recycled millstone/Muller
Fragment 1 3 Convex2 3 Concave
Unit E LMGS4 3 Lower stone Fragment 1 3 Flat2 0 Irregular
Unit E LMGS5 3, 4, 5 Lower stone Fragment 1 3 Flat2 0 Irregular
Unit E LMGS6 3, 4, 5 Lower stone Fragment 1 3 Flat2 0 Irregular
Unit E LMGS7 3, 4, 5 Lower stone Fragment 1 3 Flat2 0 Irregular
Unit E LMGS8 3, 5 Lower stone Fragment 1 3 Flat2 1 Irregular
Unit E LMGS9 in situ 3 Lower stone Fragment 1 3 Flat2 0 Irregular
Unit F LMGS10 6 Muller Fragment 1 3 Flat Facet2 0 Flat
Unit E LMGS11 7 Upper stone Complete 1 3 Convex2 3 Flat3 3 Convex
Golgol lag LMGS12 8 Lower stone Fragment 1 3 Concave2 2 Irregular
Golgol lag LMGS13 9 Uncertain Fragment 1 1 Irregular2 1 Irregular
Unit F LMGS14 10 Uncertain Fragment 1 3 Flat2 3 Flat
Unit F LMGS15 11 Uncertain Fragment 1 3 Flat2 2 Flat
Unit F LMGS16 12 Lower stone Fragment 1 3 Flat2 3 Irregular
Golgol lag LMGS17 13 Lower stone Fragment 1 3 Concave2 0 Irregular
Use confidence: 0, not used (no traces of use); 1, possible use; 2, probable use; 3, definite use.
surface of the “Golgol lag”, an ancient erosional surfacethe age of formation of which is not known. They wereincluded in this study to establish whether wear traces andresidues survive on ground stone tools that may have beensubject to prolonged exposure on the surface of asemi-arid landscape.
Most of the sandstone fragments are isolated finds.However, one set of three fragments (LMGS14, 15 and 16)originated amongst a scatter of debris surrounding atermite heat-retainer hearth that lies in Unit F. The threefragments were originally part of a set of six fragmentsthat were recorded at different locations on the slopebelow the hearth in 2009. The three uncollected fragmentslay at the greatest distance from the hearth anddisappeared following heavy rains in February 2011,apparently washed into the downslope gully system. Asecond set of refitting fragments belongs to a slab ofdistinctive purplish-coloured sandstone that has a precisestratigraphic provenience. Seven fragments of this slab(LMGS3, 4, 5, 6, 7, 8 and 9) were found initially in ashallow (a few centimetres thick) gully that drains anadjacent topographic high, comprising alternating lenses ofsands and clays (Unit E), which is cutting into underlyingquartz sands (also part of Unit E). One fragment (LMGS2)was later found in situ in the alternating lenses of sand andclay after being exposed by heavy rains. The in situfragment confirms the initial inference made about thelikely stratigraphic origin of the seven fragments from theshallow gully, and the age of these refitting artefacts hasbeen determined by the bracketing age estimates for UnitE as a whole.
The remaining two sandstone fragments in the study arefrom settings in which the micro-topographic features athigher elevation are made up of the same stratigraphic unitas the one on which the artefacts were lying in positionswhere there is no slope wash surface or rill system thatcould have transported those artefacts to their presentlocations. LMGS11 lay on a flat, stable platform of Unit Esediments at the base of a large residual also comprisingUnit E sediments. The platform exposure on which theartefact lies is elevated with respect to the surroundingslopes. LMGS10 lay just below the crest of the lunette ona small, flat ledge of calcium carbonate formed withinUnit F. The upslope catchment is small and consists ofUnit F sediments.
StratigraphyUnit E (the lateral equivalent of Bowler’s Arumpo andZanci units, see Bowler 1998: 125) comprises the greatestvolume of sediment in the central Mungo lunette. Thepredominance of sandy sediments in Unit E corresponds tolake-full phases; these alternate with clay-bearing layersindicating frequent drying events. The dominant sandsindicate that the lake frequently held water throughout theperiod 25–14 ka (Bowler et al. 2012; Fitzsimmons et al.2014), a time slice corresponding to the lead up to, duringand immediately following, the LGM. This time periodwas characterised by increased flows in the Lachlan River
system that fed the Willandra lakes at the time, thought tobe associated with increased run-off and seasonalsnowmelt (Kemp & Rhodes 2010). At the same time,inland Australia, including the Willandra region, was mostlikely relatively arid (Fitzsimmons et al. 2013). The floodpulses responsible for the lake-full phases of Unit E wouldhave enhanced the biological productivity of the overflowsystem, which may account for Unit E containing such anabundant and varied array of traces of past human activity(Stern et al. 2013). Hearths contain fish remains, thegeochemistry of which reveals that they lived in afreshwater lake and died under evaporative conditions(Long et al. 2014); this gives some indication as to thepotential duration of the lake-full and drying phases duringUnit E’s deposition.
Following the LGM, the Willandra Creek, a distributaryof the Lachlan River system responsible for inflow to theWillandra Lakes, became inactive, with the result thatmost of the lakes in the Willandra system had dried up byc.14 ka (Fitzsimmons et al. 2014; Stern et al. 2013).However, aeolian reworking of the older sediments(including Unit E) resulted in the accumulation ofsediment units on the crest and leeward side of the lunette;the major stratigraphic unit for this phase has been termedUnit F (Fitzsimmons et al. 2014). The presence ofnumerous discontinuous and weakly developed soilhorizons within Unit F suggests that local conditionsalternated between relatively arid (aeolian reactivation) torelatively moist (soil formation and dune stabilisation)(Fitzsimmons et al. 2014; Stern et al. 2013).Archaeological traces are less abundant and less varied inUnit F than in Unit E, but include heat-retainer hearths,sets of refitting stone artefacts and shell tools, as well asgrinding stone fragments (Fitzsimmons et al. 2014; Sternet al. 2013).
METHODS
Multiple lines of evidence are important for confidentinterpretations when investigating the function(s) ofarchaeological artefacts, in this case stones with groundsurfaces (Fullagar et al. 1996). Key sources of evidenceinclude tool design and manufacture (technology),experimental studies, tool stone properties, wear traces,residues (both microscopically visible structures andinvisible, adsorbed molecules), archaeological context andany relevant historical and ethnographic data (Fullagar2014; Rowan & Ebling 2008). Secure residueidentification depends on relevant modern comparativereference collections (Field 2006, 2007; Fullagar &Matheson 2013). Experimental studies of grinding stoneshave shown that usewear can be diagnostic of tasks andmaterials processed, and are consistent over a variety ofstone raw materials (e.g. Adams 1989, 2014a,b; de laTorre et al. 2013; Dubreuil 2004; Dubreuil & Savage2013; Gilabert et al. 2012; Hamon 2008; Liu et al.2010a,b; Stephenson 2011; Wright 1993).
Important forms of usewear include variations in themacro- and microtopography of the ground surface ofquartz grains and crystalline matrix. These variationsinclude abrasive smoothing, alignments and directionalityof features, striations, pits, rounding and fracture of grainsand polish. Residue and usewear reference libraries used inthis study include experimental, ethnographic andarchaeological grinding tools made on sandstone,quartzite, granite and other stone types (Fullagar et al.2012; Fullagar & Wallis 2012). A series of grindingexperiments were undertaken to further assist withfunctional analysis of well-cemented and weakly cementedsandstone grinding implements used to process seeds ofvarious hardness and size (kangaroo grass seeds, warregograss seeds, Acacia seeds, kurrajong seeds, wheat seedsand coffee beans). Sandstone slabs were used in theseexperiments as filing tools to grind wood, bone, volcanicstone axes, other sandstone and hematite.
Microscopically visible residues can be indicative ofseed grinding and include starch, phytoliths and cellulose.Invisible adsorbed residues from, but not necessarilyspecific to, seed grinding may include lipids, proteins,fatty acids and carbohydrates. The mechanical forcesassociated with grinding often break the structural bondsof some residues, thereby altering their microscopicdiagnostic features. For example, collagenous residuesassociated with the processing of animal material can bedetected by biochemical staining, which allows a numberof microscopically altered residues to be visualised(Stephenson 2011). All of the residues described above canpotentially be recovered and identified.
The functional study of the Lake Mungo sandstoneartefacts includes four different analyses. The technologyhas been treated separately and is presented in the resultssection.
Usewear analysis by RF & EHThe Mungo artefacts were examined at low magnification(6× to 50×) with Olympus SZ61 and Leica MZ16A
dissecting microscopes and oblique light.High-magnification examination of the artefacts wasperformed with an Olympus BH-2 metallographicmicroscope at magnifications of ×50 to ×500, with verticalincident light (brightfield and darkfield) and polarisingfilters. Images were captured with a Lumenera® Infinity 2digital camera. Multifocal images of the artefact surfaceswere obtained using the Leica MZ16A stereomicroscopewith an automatic Z-stacking function and a DFC320Leica camera, and stitched using Leica LAS V4.4software.
Residue analyses by EH & CMResidue samples were recovered with a solvent mixture ofacetonitrile, ethanol and water in equal concentrations, sothat water-insoluble residues could also be extracted. Thesolvent mixture was applied to several locations on theartefact surface using a variable volume pipette, the nylontip of which was used to agitate the surface beforeremoving approximately 20–40 μL of sample solution.
Residues collected by RF/EH were stained using CongoRed (C32H22N6O6S2Na2) to facilitate the identification ofdegraded or damaged particles such as gelatinised starchand cellulose fibres (Lamb & Loy 2005). Other stains usedin this study included iodine potassium iodide (IKI), whichstains intact starch; methylene blue (C16H18N3SCI) forcellulose; phloroglucinol (C6H6O3) and safranin(C20H19CIN4) for lignin; Orange G (C16H10N2Na2O7S2) andRhodamine B (C28H31CIN2O3) for collagen and keratin;and Sudan IV (C24H20N4O) for lipids (Lillie 1976).Approximately 3–10 μL of each stain was applied toprepared slides (containing up to 5 μL of extract) and leftfor at least 10 min for the stain to develop (Table 2). Forthe temporary stains (Congo Red, phloroglucinol andmethylene blue), rinsing was kept to a minimum so thatonly excess stain was removed, but not rinsed outcompletely. All slides were examined microscopically forany positive colour changes under transmitted light.
Table 2. The staining agents utilised in the residue examination.
Staining agent Chemical formula Stained material Colour change
Congo Red C32H22N6O6S2Na2 Gelatinised starchDamaged starchCellulose
Red
Iodine potassium iodide (IKI) IKI Intact starchCellulose
Blue/black
Methylene blue C16H18N3SCI Cellulose BlueOrange G C16H10N2Na2O7S2 Collagen
KeratinOrange
Phloroglucinol C6H6O3 Lignin Yellow/brownPicrosirius Red C45H26N10Na6O21S6 Collagen Yellow/orange/greenRhodamine B C28H31CIN2O3 Collagen
KeratinPink/purple
Safranin C20H19CIN4 LigninCell wallsCell nuclei
Red
Sudan IV C24H20N4O LipidsTriglycerides lipoproteins
Biochemical assays can potentially detect residues thatmay not be visible microscopically. The followingcolorimetric biochemical tests have been included here asa pilot investigation and were carried out by EH and CM.The Bradford Assay was used to detect protein (Joneset al. 1989; Kruger 1994), diphenylamine and phenol –sulphuric acid (PSA) were used to detect carbohydrates(Kanzaki & Berger 1959; Masuko et al. 2005; Mecozzi2005). Fatty acids were detected by a method (Cu-TEA/CPC) developed by Falholt and Lund (1973). Hemastix®strips were used to detect hemoglobin (Matheson & Veall2014) and iodine potassium iodide (IKI) was used todetect starch (McCready & Hassid 1943). These tests wereselected to canvas the range of plant and animal residues.Positive reactions were registered by a colour change, andwere quantified with an EpochTM Multi-VolumeSpectrophotometer System (Biotek), using Gen 5 softwarewith standard references: blood protein, corn starch (bothheated and non-heated), cooking oil and a combination ofsucrose and glucose (see Supporting Information AppendixS1: Chemical Test Protocols).
Residue analyses by BSResidue samples were collected by BS from differentlocations to those sampled by RF and EH, using ultra-purewater that had been triple distilled and passed through aMillipore filter. In this case, 60 mL aliquots were applied toselected surface locations and allowed to soak beforeextraction. Depending on the porosity of the individualfragment, further applications of ultra-pure water wereapplied to the same area, to ensure that the matrix wassaturated. A clean pipette tip was used to agitate the area,draw back and remove the (water + residue) sample. Thesamples were stored in clean 1.5 mL micro-centrifuge tubesand transported to Brisbane, where they were mounted onslides and stained. The prepared samples were examinedusing a polarising Leitz Dialux 22 microscope, andphotographed using a Tucsen ISH 500 camera. A modifiedPicrosirius Red (PSR) staining protocol was applied to themounted samples to highlight any collagenous residues.Collagenous residues were recognised by colorimetricchanges in cross-polarised light. PSR is specific forcollagen protein (in particular, Types I and III), and thisspecificity means that other residues such as plant fibres,starches, amorphous cellulose, phytoliths, minerals, hair,feather barbules and carbonised material are not affectedand can be observed without additional staining aids(Stephenson 2011; Wright et al. 2014).
Ancient starch analysis by JFRecovery of starch residues involved placing grindingstone fragments in distilled water in an ultrasonic bath for1 min to dislodge material. The grinding stones weresampled by submerging either half or the whole artefact inthe bath. The residue samples were sieved with a 125 μmmesh Endecott Sieve, followed by centrifugation (3 min at3000 r.p.m., supernatant discarded). Starch and phytolithswere isolated with a heavy liquid separation method using
sodium polytungstate, specific gravity 2.35 (centrifuged for15 min at 1000 r.p.m.). After further rinsing with waterand centrifugation, the samples were washed in acetoneand allowed to air dry. Samples were mounted on glassslides in 50% glycerol:water. Samples were also stainedwith Congo Red to detect damaged or gelatinised grainsnot detected in the first scan. Slides had total scans using aZeiss Axioskop2 brightfield transmitted light microscopefitted with Nomarksi optics. All starch grains werephotographed using a Zeiss HRc digital camera andarchived using Zeiss Axiovision software.
RESULTS
TechnologyAll of the 17 stones are made of fine-grained,well-cemented sandstone (Table 1, Plates S1–S13). Twelvespecimens were identified as lower stones (displaying aconcave surface or a thin portion of a very worn flatstone). Four specimens, including LMGS1 (Plate S1),LMGS3 (Plate S3b), LMGS10 (Plate S6) and LMGS11(Plate S7), are identified as upper stones, two of which(LMGS3 and 10) have distinct facets (cf. mullers; seeSmith 1985). Two specimens (LMGS1 and 3) are concaveon one surface, and appear to be recycled lower(millstone) fragments, subsequently used as upper stones(Plates S1a–d, S3b–c). One small specimen appears to be acomplete upper stone (LMGS12, Plate S8), with wear onthree surfaces suggesting use as a filing stone. Threespecimens (LMGS13, Plate S9; LMGS14, Plate S10; andLMGS15, Plate S11) have relatively flat or irregularsurfaces, and it is uncertain whether they functionedprimarily as upper or lower grinding stones. The concave,convex and facetted morphological features are consistentwith seed grinding activities, as identified in otherarchaeological and ethnographic artefact collections(Smith 1985). No diagnostic peck marks or traces ofsurface rejuvenation were identified on any of thegrinding stone fragments.
UsewearThe primary features of usewear traces – polish, striations,grain rounding and scarring – as well as their appearanceand extent of coverage, are summarised in Table 3, withdetails and images of usewear in Plates S1–S13.Importantly, the wear pattern is broadly similar acrossmost specimens.
Under low magnification, the sandstone surfaces appearto be abrasively smoothed and erosion of the crystallinematrix was observed. Individual quartz grains are clearlylevelled, with scratches often visible under point source,low-angled light. At higher magnification, themicrotopography appears undulating to flat with highreflectivity, most clearly visible on individual quartzgrains. Higher elevated zones on the polished quartz grainsdisplay bright reticular polish, compared with the lowerzones in the interstices between quartz grains, whichdisplay low or no polish development. The distribution and
degree of wear development varies across each artefact,but generally the wear is extensive and the degree ofpolish development is very high. Micro-striations arepresent on the polished surfaces, and typically occur inmultiple orientations of varying widths and depths. Broad,bright alignments of polish are probably the result ofabrasion, occurring from the friction of stone-on-stoneactivities. Narrow, short, shallow striations are appear to bethe result of stone and grit that was incorporated into thecrushed seed mixture during grinding. A higher degree ofrounding and surface undulation on some quartz grainssuggests that other tissues were processed, includinglarger, softer substances, possibly tubers.
Two specimens from the Golgol erosional surface(LMGS12 and 13) have sustained smoothing and scatteredpatches of use-polish that is not diagnostic of a particularprocessed material but is suggestive of seed grinding. Weinterpret plant processing as a possible/likely function due,in part, to the presence of plant residues on thesespecimens (see below).
ResiduesThe residues identified in the study are presented inTables 4 and 5. Note that specimens LMGS2–9 are broken
pieces from a single, larger grinding stone fragment, onepiece (LMGS9) of which was found in situ (Unit E).Residues are low in abundance, and are dominated bycellulose fibres and amorphous organic tissue. Gelatinisedand intact starch grains are present. The taxonomic originof the starch grains could not be identified due to the verylow numbers recovered in the analysis and the absence ofany diagnostic grains. Nonetheless, it is evident from the
Table 4. A list of the residues identified by four analysts: EH & CM, BS and JF.
Laboratory code Analyst Collagen Plant Starch Carbonised tissue Mineral Other
dimensions and morphology that these starch grains arenot consistent with those from known grasses. The damagedocumented on some grains is consistent with grindingactions.
Collagen was identified on three artefacts: by EH,following Orange G staining on artefacts LMGS1 andLMGS16; and by BS, following the application of PSRstaining protocol on artefacts LMGS1 and LMGS17.Feather barbules were identified on three grindingfragments: LMGS2 (Plate S4a), LMGS5 (Plates S4e–f)and LMGS10.
In addition, PSR staining provided a contrastbackground that facilitated observations of hyphae onLMGS10. Lichen was observed in extractions from
LMGS1 and LMGS3. Along the toe of the lunette,exposures of the unconsolidated sandy sediments that lieat the base of Unit E are held in place by a thin crust oflichen, creating flat, relatively stable exposures. When thecrust is broken (e.g. by foot traffic), the sediments arevulnerable to erosion through wind and water action. Anygrindstones embedded in or lying on these surfaces, suchas LMGS1 and LMGS3, may therefore also be covered inlichen and contain fungal elements.
Biochemical testing of adsorbed biomolecules selectedto detect protein, carbohydrates and starch, was positive onsome grinding stones (Figure 3), though their specificorigin could not be determined. Fatty acids were notconfirmed on any of the grinding stones, although the
Figure 3. Examples of starch grains recovered from grinding stones. Black arrows indicate damage that may beattributable to grinding.
breakdown products of fatty acids (e.g. azelaic acid) wouldnot be detected by the Cu-TEA/DPC test.
DISCUSSION
Grinding stones most often appear in the archaeologicalrecord as fragments, as they are often discarded at the endof their use-life. Sixteen of the 17 grinding stones werefragments. Although LMGS11 is a complete uppergrinding stone with three discrete utilised surfacemorphologies (flat, concave and convex), it may have beenrecycled from a broken, lower grinding stone fragment.
For 14 of the 17 Mungo specimens, usewear appearstypical of that observed on experimental and ethnographicseed processing tools. Usewear variations occur in thedegree of grain rounding (identified at low magnificationusing an external low-angled light source) and the extentand development of polish (observed under highmagnification using vertical incident light).
All the grinding stones were collected from the surfaceof an eroding landform and were exposed to wind andwater erosion. Some variation in wear traces may be theresult of differential weathering on the artefact surfaces.“Sandblasting” can result in the obliteration of face-uppolished surfaces. Nonetheless, weathering appears to below, because polish has been preserved in sufficient detailto enable comparisons with experimental and ethnographictool surfaces.
Starch grain counts were low and are considered to bethe result of poor preservation in an open site context(Table 5). The scarcity or absence of starch is probably theresult of taphonomic processes, particularly the effects ofwind and/or water over time. The few starch grainspresent, especially those with modification, derive from arange of unknown taxa (Figure 3). In the absence of highercounts and secure identification, the starch grain evidenceis best interpreted as an indicator of possible starchy plantprocessing.
The paucity of residues is most likely the result ofweathering processes. Only those residues protected withinthe deepest interstices of the sandstone survived. Variationin range and types of residues as observed by differentanalysts may be the result of sampling – each analystsampled from a different location, using different solvents,extraction protocols and staining analytical techniques.Generally, all analysts described a higher occurrence ofplant material with only scant evidence of animal material.
There is a broad agreement for the three differentapproaches to residue recovery and testing for starch onLMGS1, 11, 16 and 17. Ultrasonication recovered starchin ten instances (LMGS1, 3, 10, 11, 12, 13, 14, 15, 16 and17), whereas the micropipette methods recovered starch insix instances (LMGS1, 7, 9, 11, 16 and 17). LMGS13 hadno distinctive usewear and yet starch was recovered withultrasonication. Variation in recovery of starch may be duein part to the variation in yields from ultrasonication andpipettes, and also the differential preservation of starch atdifferent locations across artefact surfaces. The latter
differential preservation is supported by various tests. TheIKI test for starch (Figure 4) indicated reactions above thedetection threshold (y = 0.079) for starch on lifts fromLMGS1, 3, 5, 11, 12, 16 and 17. Positive reactions forLMGS3, 16 and 17 were above the detection threshold(y = 0.113) for carbohydrates (diphenylamine; Figure 4).
Plant tissue was found on all 14 grinding stonefragments that displayed usewear indicative of seedprocessing (Table 4). Collagen was identified on threeartefacts (LMGS1, 16 and 17). Collagen has previouslybeen identified on seed grinding stones from open-sitesettings, indicating that they were used for multipletasks (Stephenson 2011). The feather barbules notedon three of the fragments were not identical. Thoseassociated with LMGS2 were morphologicallysimilar to the order Anseriformes (ducks, geese andswans). However, the projections at each node of thebarbules noted on LMGS5 and LMGS10 could belong toany one of three different orders (including Anseriformes,Galliformes and Falconiformes). Collagen, bone and/orblood were not observed in combination with thesefeather residues. The low number observed, and theabsence of accompanying collagenous residues,suggests that the feather barbules are not relatedto use. They have, however, been noted here forcompleteness.
The primary secure basis of functional interpretationin the study of the Lake Mungo grinding stones is theusewear (Figure 5), which is similar to that onethnographic and experimental stones used to grind seeds.Although weathering has clearly reduced their abundance,the identified residues lend support to the hypothesis thatthe grinding stone fragments were used to process plantand animal tissue.
Pounding and grinding technology have considerableantiquity, as proposed by de Beaune (2004), who tracedthe transformation of pounding (found in the LowerPalaeolithic) to grinding and polishing motions in theMiddle Palaeolithic. Several studies of tool residues havedocumented the grinding and consumption of starchy plantfoods including grass seeds during the late Pleistocene,during and prior to Marine Isotope Stage 2, when resourcestress may have increased the importance of plant foods inhuman diet. These include the following:
• c.23 ka at Ohalo II in Israel (Piperno et al. 2004; Weisset al. 2008);
• c.23–19.5 ka at Shizitan Locality 14 in China (Liu et al.2013); and
• c.30 ka at Bilancino II in Italy, Kostenki 16 – Uglyankain Russia, and Pavlov VI in the Czech Republic(Revedin et al. 2010).
Earlier evidence of plant consumption including seedshas also been documented in a study of Neanderthal teethfrom Shanidar III, in Iraq, and Spy I and II, in Belgium(Henry et al. 2011). Even earlier evidence for plantprocessing on grinding stones has been documented from
site 8-B-11, Sai Island, in Sudan, with an estimated age of220–150 ka (van Peer et al. 2003).
The earliest reported Australian grinding stone comesfrom near the base of Madjedbebe, the site formerlyknown as Malakunanja II (Lowe et al. 2014), and is
associated with thermoluminescence (TL) ages of about 50ka (Roberts et al. 1990). The identification of Pleistoceneplant utilisation, processing and consumption is becomingmore frequent in the Australian region, as elsewhere in theworld, with detailed studies of tool residues and plant
Figure 4. Biochemical tests. All tests above the horizontal line (the threshold of detection) are considered positive. (a)Bradford Assay for the detection of protein. The horizontal line (y = 0.121) indicates the measurement of a standardsample (blood). (b) Cu-TEA/CPC test for the detection of fatty acids. The horizontal line (y = 0.856) indicates themeasurement of a standard sample (cooking oil). (c) Diphenylamine test for the detection of carbohydrates. The horizontalline (y = 0.113) indicates the measurement of a standard sample (glucose + sucrose). (d) IKI test for the detection ofstarch. The horizontal line (y = 0.079) indicates the measurement of a standard sample (corn starch). (e) PSA test for thedetection of carbohydrates. The horizontal line (y = 0.094) indicates the measurement of a standard sample (glucose +sucrose).
remains from sediments. For example, non-ground stoneartefacts from the Ivane Valley in the New Guineahighlands had starchy plant residues consistent with yamprocessing at two sites: Joe’s Garden, from layer 3A withage estimates of c.27–30 ka; and South Kov Ridge, layer4, with an age estimate of c.41 ka (Summerhayes et al.2010). The late Pleistocene context for seed grinding andother plant processing has been reported for CuddieSprings, in Australia’s semi-arid south-east (Fullagar &Field 1997).
CONCLUSION
The multidisciplinary study of the Lake Mungo grindingstones presented here has demonstrated that interpretablewear traces and residues are preserved on small groundstone fragments recovered from exposed surfaces ofstratified, eroding landforms in semi-arid Australia. Theoriginal grinding stone morphologies are difficult toreconstruct from small fragments, but include concave,convex, facetted and flat surfaces, many of which bearsimilarities with ethnographic seed grinding implementsfrom Central Australia. The usewear on 14 of the 17artefacts is consistent with experimental and ethnographicdata that indicates seed grinding (Figure 5). All thesandstone pieces examined in this study must have beencarried into this landscape and almost certainly derivefrom larger, broken grinding stones. One specimen haspoorly developed wear with uncertain function
(LMGS12); one specimen has no distinct usewear(LMGS13); and one specimen (LMGS11) has evidencefor plant processing that may or may not include seeds.Residues were recovered in low abundance but could bederived from use. Poor residue recovery is probably dueto poor preservation following erosion from thedepositional matrix. Variation in residue observations mayin part be attributed to the differences between workedsurface areas sampled by the researchers andmethodological approaches. The majority of the plant andanimal tissues recovered have not been identified withtaxonomic precision.
The results of this integrated approach to functionalanalysis provide compelling evidence that 14 of the 17grinding stone fragments were used for processing seeds.Other plant tissue and animal residues including collagensuggest multiple functions for at least three artefacts. Onlyone of the grinding stone fragments has a preciseprovenance, although all can be directly linked withparticular strata in the central Mungo lunette. The recentgeomorphological study combined with the chronologicalsequence established for units within the lunette(Fitzsimmons et al. 2014; Stern et al. 2013) have providedstrong evidence of a Pleistocene context (25–14 ka) forgrinding stone fragments from Lake Mungo. The usewearand residue traces indicating Pleistocene seed grinding atLake Mungo also provide further support for thetheoretical argument that the development of seed grindingtechnology may be linked with environmental stress
Figure 5. A comparison of use-polish on experimental sandstone grinding stones and Lake Mungo artefacts: (a,b)use-polish on experimental artefact used to process warrego grass seeds for three hours; (c) use-polish on LM GS 11; (d)use-polish LM GS 14.
associated with the LGM (Edwards & O’Connell 1995;Fullagar & Field 1997: 302).
CONFLICT OF INTEREST
Authors declare they have no conflict of interests for thisarticle.
ACKNOWLEDGEMENTS
This research began with permission from the Elders’Council of the Three Traditional Groups and the Technicaland Scientific Advisory Committee of the Willandra LakesRegion World Heritage Area (WLRWHA) and continuedunder the stewardship of the Elders’ Council of the TwoTraditional Tribal Groups of the WLRWHA. We areindebted to the Elders of the Paakantyi, Ngiyampaa andMutthi Mutthi tribes for their ongoing support of thiswork. The research was funded by an Australian ResearchCouncil Linkage grant (LP0775058) and an AustralianResearch Council Discovery grant (DP1092966); andsupported by La Trobe University, the University ofWollongong and the University of New South Wales. Wegratefully acknowledge the field assistance provided byDaryl Pappin, the project’s Cultural Heritage Officer, RudyFrank and Paul Kajewski, Technical Officers from LaTrobe University, as well as the student volunteers andMungo National Park Discovery Rangers who helpedlocate and document the grindstones. We alsoacknowledge support and facilities offered by the SouthAustralian Museum and the School of Biological, Earthand Environmental Sciences, The University of New SouthWales. Jessica Roe is thanked for processing the grindingstones for starch analysis at UNSW. We thank twoanonymous referees and the AO editors for helpfulcomments.
REFERENCES
Adams, J.L. 1989. Experimental replication of the use of groundstone tools. KIVA 54: 261–271.
Adams, J.L. 2014a. Ground Stone Tools: A TechnologicalApproach. University of Utah Press, Salt Lake City, UT (1stedn 2002).
Adams, J.L. 2014b. Ground-stone use-wear analysis: A review ofterminology and experimental methods. Journal ofArchaeological Science 48: 129–138.
Allen, H. 1974. The Bagundji of the Darling Basin: Cerealgatherers in an uncertain environment. World Archaeology 5:309–322.
Balme, J. and O’Connor, S. 2014. Early modern humans inisland Southeast Asia and Sahul: Adaptive and creativesocieties with simple lithic industries. In R. Dennell andM. Porr (eds), Southern Asia, Australia and the Search forHuman Origins, pp. 164–174. Cambridge University Press,New York.
de Beaune, S.A. 2004. The invention of technology. CurrentAnthropology 45 (2): 139–162.
Bowler, J.M. 1998. Willandra Lakes revisited: Environmentalframework for human occupation. Archaeology in Oceania33: 120–155.
Bowler, J.M., Gillespie, R., Johnston, H. and Boljkovac,K. 2012. Wind v water: Glacial maximum records fromthe Willandra Lakes. In S.G. Haberle and B. David (eds),Peopled Landscapes: Archaeological and BiographicApproaches to Landscapes, pp. 271–296. TerraAustralis 34. Australian National University Press,Canberra.
Bowler, J.M., Johnston, H., Olley, J.M., Prescott, J.R., Roberts,R.G., Shawcross, W. and Spooner, N.A. 2003. New ages forhuman occupation and climatic change at Lake Mungo,Australia. Nature 421: 837–840.
de la Torre, I., Benito-Calvo, A., Arroyo, A., Zupancich, A. andProffitt, T. 2013. Experimental protocols for the study ofbattered stone anvils from Olduvai Gorge (Tanzania). Journalof Archaeological Science 40: 313–332.
Dubreuil, L. 2004. Long-term trends in Natufian subsistence: Ause-wear analysis of ground stone tools. Journal ofArchaeological Science 31: 1613–1629.
Dubreuil, L. and Savage, D. 2013. Ground stones: A synthesis ofthe use-wear approach. Journal of Archaeological Science 48:139–153.
Falholt, K., Lund, B. and Falholt, W. 1973. An easy colorimetricmicromethod for routine determination of free fatty acids inplasma. Clinica Chimica Acta 46: 105–211.
Field, J. 2006. Reference collections for starch studies. InR. Torrence and H. Barton (eds), Ancient Starch Analysis,pp. 95–113. Left Coast Press, Walnut Creek, CA.
Field, J. 2007. Analysis of botanical remains: Starch grainanalysis. In D. Pearsall (ed.), Encyclopedia of Archaeology,vol. 3, pp. 2078–2082. Elsevier, Oxford.
Fitzsimmons, K., Stern, N. and Murray-Wallace, C. 2014.Depositional history and archaeology of the central Mungolunette, Willandra Lakes, southeast Australia. Journal ofArchaeological Science 41: 349–364.
Fitzsimmons, K.E., Cohen, T.J., Hesse, P.P. Jansen, J., Nanson,G.C., May, J.-H., Barrows, T.T., Haberlah, D., Hilgers, A.,Kelly, T., Larsen, J., Lomax, J. and Treble, P. 2013. LateQuaternary palaeoenvironmental change in the Australiandrylands: A synthesis. Quaternary Science Reviews 74:78–96.
Fullagar, R. 2014. Residues and usewear. In J. Balme and A.Paterson (eds), Archaeology in Practice: A Student Guide toArchaeological Analyses, pp. 232–263. Blackwell, Maldon,MA (1st edn 2006).
Fullagar, R. and Field, J. 1997. Pleistocene seed grindingimplements from the Australian arid zone. Antiquity 71 (272):300–307.
Fullagar, R. and Matheson, C. 2013. Stone tool usewear andresidue analysis. In C. Smith (ed.), Encyclopedia of GlobalArchaeology, pp. 7062–7065. Springer, New York.
Fullagar, R. and Wallis, L. 2012. Usewear and phytoliths onbedrock grinding patches in north-western Australia. In L.Russell (ed.), Papers in Honour of Beth Gott. The Artefact 35:69–81.
Fullagar, R., Field, J. and Kealhofer, L. 2008. Grinding stonesand seeds of change: Starch and phytoliths as evidence ofplant food processing. In Y.M. Rowan and J.R. Ebling (eds),New Approaches to Old Stones: Recent Studies of GroundStone Artifacts, pp. 159–172. Equinox, London.
Fullagar, R., Furby, J. and Hardy, B. 1996. Residues on stoneartefacts: State of a scientific art. Antiquity 70: 740–745.
Fullagar, R., Liu, L., Bestel, S., Jones, D., Ge, W., Wilson, A.and Zhai, S. 2012. Stone tool-use experiments to determinethe function of grinding stones and denticulate sickles.Bulletin of the Indo-Pacific Prehistory Association 32:29–44.
Geneste, J.M., David, B., Plisson, H., Delannoy, J.-J. andPetchey, F. 2012. The origins of ground-edge axes: Newfindings from Nawarla Gabarnmang, Arnhem Land (Australia)and global implications for the evolution of fully modernhumans. Cambridge Archaeological Journal 22: 1–17.
Gilabert, X.R., Martínez-Moreno, J. and Torcal, R.M. 2012.Pitted stone cobbles in the Mesolithic site Font del Ros(southeastern pre-Pyrenees, Spain): Some experimentalremarks and a controversial tool type. Journal ofArchaeological Science 39: 1587–1598.
Gorecki, P., Grant, M., O’Connor, S. and Veth, P. 1997. Themorphology, function and antiquity of Australian grindingimplements. Archaeology of Oceania 32: 141–150.
Gott, B. 2002. NSWUSE database (disks 1 and 2) [Filemaker Proversion 4.1] Unpublished. copies lodged with AIATSIS andNSW NPWS.
Hamon, C. 2008. Functional analysis of stone grindingand polishing tools from the earliest Neolithic ofnorth-western Europe. Journal of Archaeological Science 35:1502–1520.
Henry, A.G., Brooks, A.S. and Piperno, D.R. 2011. Microfossilsin calculus demonstrate consumption of plants and cookedfoods in Neanderthal diets (Shanidar III, Iraq; Spy I and II,Belgium). Proceedings of the National Academy of Sciences108: 486–491.
Jones, C.J., Hare, D. and Compton, S.J. 1989. Measuring plantprotein with the Bradford Assay: Evaluation and standardmethod. Journal of Chemical Ecology 15: 979–992.
Jones, R. 1985. Archaeological conclusions. In R. Jones (ed.),Archaeological Research Inkakadu National Park, pp.291–299. Australian National Parks and Wildlife Service,Canberra.
Jones, R. and Johnson, I. 1985. Deaf Adder Gorge: Lindner Site,Nauwalabila I. In R. Jones (ed.), Archaeological ResearchInkakadu National Park, pp. 165–227. Australian NationalParks and Wildlife Service, Canberra.
Kamminga, J. and Allen, H. 1973. Alligator RiversEnvironmental Fact Finding Study: Report of theArchaeological Survey. Australian Government, Canberra.
Kanzaki, G. and Berger, E.Y. 1959. Colorimetric determinationof methylcellulose with diphenylamine. Analytical Chemistry31: 1383–1385.
Kemp, J. and Rhodes, E.J. 2010. Episodic fluvial activity ofinland rivers in southeastern Australia: Palaeochannel systemsand terraces of the Lachlan River. Quaternary ScienceReviews 29: 732–752.
Kruger, N.J. 1994. The Bradford methods for proteinquantification. In J.M. Walker (ed.), Methods in MolecularBiology, Volume 32, Basic Protein and Peptide Protocols, pp.9–15. Humana Press, Totowa, NJ.
Lamb, J. and Loy, T. 2005. Seeing red: The use of Congo Reddye to identify cooked and damaged starch grains inarchaeological residues. Journal of Archaeological Science32: 1433–1440.
Liu, L., Bestel, S., Shi, J., Song, Y. and Chen, X. 2013.Paleolithic human exploitation of plant foods during the LastGlacial Maximum in North China. Proceedings of theNational Academy of Sciences of the United States of America110 (14): 5380–5385.
Liu, L., Field, J., Fullagar, R., Bestel, S., Chen, X. and Ma, X.2010a. What did grinding stones grind? New light on EarlyNeolithic subsistence economy in the Middle Yellow RiverValley, China. Antiquity 84: 816–833.
Liu, L., Field, J., Fullagar, R., Zhao, C., Chen, X. and Yu, J.2010b. A functional analysis of grinding stones from an earlyHolocene site at Donghulin, North China. Journal ofArchaeological Science 37: 2630–2639.
Long, K., Stern, N., Williams, I.S. Kinsley, L., Wood, R.,Sporcic, K., Smith, T., Fallon, S., Kokkonen, H., Moffat, I.and Grün, R. 2014. Fish otolith geochemistry, environmentalconditions and human occupation at Lake Mungo, Australia.Quaternary Science Reviews 88: 82–95.
Lowe, K.M., Wallis, L.A., Pardoe, C., Marwick, B., Clarkson, C.,Manne, T., Smith, M.A. and Fullagar, R. 2014.Ground-penetrating radar and burial practices in westernArnhem Land, Australia. Archaeology in Oceania 49 (3):148–157.
McCready, R.M. and Hassid, W.Z. 1943. Separation andquantitative estimation of amylose and amylopectin inpotato starch. Journal of American Chemical Society 65:1154–1157.
Masuko, T., Minami, A., Iwasaki, N., Majima, T., Nishimura,S.-I. and Lee, Y. 2005. Carbohydrate analysis by a phenol –sulfuric acid method in microplate format. AnalyticalBiochemistry 339: 69–72.
Matheson, C.M. and Veall, M. 2014. Presumptive blood testusing Hemastix® with EDTA in archaeology. Journal ofArchaeological Science 41: 230–241.
Mecozzi, M. 2005. Estimation of the total carbohydrate amountin environmental samples by the phenol – sulphuric acidmethod assisted by multivariate calibration. Chemometricsand Intelligent Laboratory Systems 79: 84–90.
Mulvaney, D.J. and Bowler, J.M. 1981. Lake Mungo and theWillandra Lakes. In The Heritage of Australia: The IllustratedRegister of the National Estate, pp. 180–183. Macmillan,Sydney.
O’Connor, S. 1999. 30,000 Years of Aboriginal Occupation:Kimberley, North West Australia. Terra Australis 29. ANHPublications, Centre for Archaeological Research, AustralianNational University, Canberra.
Piperno, D., Weiss, E. and Holst, I. 2004. Processing of wildcereal grains in the Upper Palaeolithic revealed by starchgrain analysis. Nature 430: 670–673.
Revedin, A., Aranguren, B., Becattini, R., Longo, L., Marconi,E., Lippi, M.M., Skakun, N., Sinitsyn, A., Spiridonova, E. andSvoboda, J. 2010. Thirty thousand-year-old evidence of plantfood processing. Proceedings of the National Academy ofSciences of the United States of America 107 (44):18815–18819.
Roberts, R.G., Jones, R. and Smith, M.A. 1990.Thermoluminescence dating of a 50,000-year-old humanoccupation site in northern Australia. Nature 345:153–156.
Rowan, Y.M. and Ebling, J.R. (eds) 2008. New Approaches toOld Stones: Recent Studies of Ground Stone Artifacts.Equinox, London.
Schrire, C. 1982. The Alligator Rivers: Prehistory and Ecology inWestern Arnhem Land. Terra Australis 7. Department ofPrehistory, Research School of Pacific Studies, The AustralianNational University, Canberra.
Smith, M.A. 1985. A morphological comparison of CentralAustralian seed grinding implements and AustralianPleistocene-aged grindstones. The Beagle 2: 23–38.
Smith, M.A. 1986. The antiquity of seed grinding in aridAustralia. Archaeology in Oceania 21: 29–39.
Smith, M.A. 1988. Central Australian seed grinding implementsand Pleistocene grindstones. In B. Meehan and R. Jones(eds), Archaeology and Ethnography: An AustralianPerspective, pp. 94–108. Australian National University,Canberra.
Smith, M.A. 1989. Seed grinding in inland Australia: Currentevidence from seed-grinders on the antiquity of theethnohistorical pattern of exploitation. In D. Harris and G.C.Hillman (eds), Foraging and Farming, pp. 305–317. UnwinHyman, London.
Stephenson, B. 2011. In the Groove: An Integrated FunctionalAnalysis of Arid Zone Millstones from Queensland.Unpublished BA (Hons) thesis, School of Social Sciences,University of Queensland, St Lucia.
Stern, N., Tumney, J., Fitzsimmons, K. and Kajewski, P. 2013.Strategies for investigating human responses to changes inenvironment at Lake Mungo in the Willandra Lakes, southeastAustralia. In D. Frankel, J. Webb and S. Lawrence (eds),Archaeology in Technology and Environment, pp. 31–50.Routledge, London.
Summerhayes, G.R., Leavesley, M., Fairbairn, A.S., Mandui, H.,Field, J., Ford, A. and Fullagar, R. 2010. Human adaptationand plant use in Highland New Guinea 49,000 to 44,000 yearsago. Science 330: 78–81.
Tindale, N.B. 1977. Adaptive significance of the Panara or grassseed culture of Australia. In R.V.S. Wright (ed.), Stone Toolsas Cultural Markers, pp. 340–349. Australian Institute ofAboriginal Studies, Canberra.
Van Peer, P., Fullagar, R., Stokes, S., Bailey, R.M., Moeyersons,J., Steenhoudt, F., Geerts, A., Vanderbeken, T., De Dapper,M., and Geus, F. 2003. The Early to Middle Stone Agetransition and the emergence of modern human behaviour atsite 8-B-11, Sai Island, Sudan. Journal of Human Evolution45: 187–193.
Weiss, E., Kislev, M.E., Simchoni, O., Nadel, D. and Tschauner,H. 2008. Plant-food preparation area on an Upper Paleolithicbrush hut floor at Ohalo II, Israel. Journal of ArchaeologicalScience 35: 2400–2414.
Wright, D., May, S., Tacon, P. and Stephenson, B. 2014. Ascientific study of a new cupule site in Jabiluka, WesternArnhem Land. Rock Art Research 31: 92–100.
Wright, K. 1993. Ground-stone tools and hunter–gatherersubsistence in Southwest Asia: Implications for the transitionto farming. American Antiquity 59 (2): 238–263.
SUPPORTING INFORMATION
Additional Supporting Information may be found in theonline version of this article:
Plate S1. LMGS1: (a) Grinding Surface 1; (b) Surface 1 atlow magnification, showing well-rounded quartz grains andlevelled plateaux; (c) Surface 2 at low magnification,showing a levelled but weathered surface; (d) GrindingSurface 2; (e–f) Surface 1 at high magnification, showing abright, reticulated use-polish (cf. seed grinding) with arrowsindicating orientation of the striations; (g–h) Surface 2 athigh magnification showing use-polish.
Plate S2. LMGS1 residues: (a–e) Grinding Surface 1: (a) astarch granule photographed at 400× in part-polarised (left)and cross-polarised (right) light; (b) gelatinised starch andplant material stained with Congo Red; (c–d) collagen fibrestained with Picrosirius Red (PSR), photographed at 400× in
(c) part-polarised and (d) cross-polarised light; (e)amorphous collagen and collagen fibres stained with PSR,photographed at 400× in cross-polarised light; (f–h)Grinding Surface 2: (f) amorphous organic materialphotographed at 200×; (g) plant material stained with CongoRed; (h) amorphous cellulose photographed at 400× inpart-polarised light.
Plate S4. LMGS2, 5, 6 & 7 residues: (a) LMGS2 Surface 1feather barbule, photographed at 400× in (left) part-polarisedand (right) cross-polarised light; (b) LMGS2 Surface 2damaged starch stained with Congo Red; (c) LMGS2Grinding Surface 1 amorphous cellulose, photographed at200×; (d) LMGS2 Grinding Surface 2 plant tissue,photographed at 200×; (e–f) LMGS5, feather barbules in (e)part-polarised light and (f) cross-polarised light; (g–h)LMGS5, amorphous plant material stained with Congo Red;(i) LMGS7, plant tissue, photographed at 400×,part-polarised light; (j) LMGS5, minerals and plant fibres,photographed at 400×.
Plate S5. LMGS2, 5, 6, 7 and 8, showing polish on thehighest points of quartz grains, creating a reticularmorphology, cf. seed grinding, with arrows indicatingorientation of the striations: (a) LMGS2 Grinding Surface 1;(b) LMGS2 Grinding Surface 2; (c) LMGS5; (d–e) LMGS6;(f) LMGS7; (g) LMGS8.
Plate S6. LMGS10: (a) Grinding Surface; (b–f) use-polish athigh magnification on the highest points of quartz grains,showing a reticular morphology cf. seed grinding, witharrows indicating orientation of the striations; (g) a starchgranule in (left) part-polarised and (right) cross-polarisedlight; (h) plant tissue stained with Congo Red incross-polarised light.
Plate S7. LMGS11: (a) Grinding Surface 1; (b) Surface 1 atlow magnification, displaying well-rounded quartz grainsand levelled plateaux; (c) Surface 2 at low magnification,showing a levelled surface with deep, interstitial spaces fromwhere quartz grains have been plucked during use; (d)Grinding Surface 2; (e) Surface 1 at high magnification,displaying a bright, reticulated polish cf. seed grinding; (f–g)Grinding Surface 2 at high magnification showinguse-polish, with arrows indicating orientation of striations;(h) Grinding Surface 2, plant fibre; (i) Grinding Surface 2,solidified plant tissue cf. exudate.
Plate S8. LMGS12: (a) Grinding Surface: (b) GrindingSurface 2 at low magnification, displaying a levelled butweathered surface; (c–f) uniform levelled use-polish at highmagnification cf. file or abrading stone; (g) plant materialstained with Congo Red; (h) mineral crystals.
Plate S9. LMGS13: (a) possible grinding surface; (b)possible grinding surface at low magnification, showing anuneven weathered surface; (c–d) possible grinding surface at
high magnification showing quartz grains with relativelyuniform low wear development (cf. weathering) and nodiagnostic traces of use-polish.
Plate S10. LMGS14: (a) Grinding Surface 2; (b) Surface 2 atlow magnification showing a levelled surface with interstitialspacing, from where quartz grains have been plucked duringuse; (c–g) bright, reticulated and very smooth use-polish athigh magnification, showing pitting and arrows indicatingorientation of the striations; (g) amorphous organic material(most likely of plant origin).
Plate S11. LMGS15: (a) Grinding Surface 2; (b) Surface 2 atlow magnification showing a levelled surface and interstitialspacing from where grains have been plucked duringuse;(c–f) use-polish at high magnification. The polish ismost developed on the highest zones of the quartz grains,giving it a bright and reticulated appearance.
grains; (c–f) Surface 2 at high magnification showinguse-polish on the highest zones of the quartz grains, giving ita bright, reticulated appearance cf. seed grinding; arrowsindicate orientation of the striations; (g–h) plant andcellulose stained with Congo Red, in cross-polarised light;(i) solidified, brittle mass of organic material, cf. plantexudate.
Plate S13. LMGS17: (a) Grinding Surface 1; (b) Surface 1 atlow magnification showing levelled and rounded quartzgrains; (c–f) Surface 1 at high magnification showingweathering (evenly smoothed microtopography) anduse-polish on the highest zones of quartz grains show-ing slight reticular morphology; (g) a starch granulestained with IKI in (left) part-polarised and (right)cross-polarised light; (h) collagen fibres and amorphouscollagen stained with PSR, photographed at 400×, incross-polarised light.
