INSTITUTE OF PHYSICS PUBLISHING

MEASUREMENT SCIENCE AND TECHNOLOGY

Meas. Sci. Technol. 14 (2003) 1487–1492

PII: S0957-0233(03)57428-8

Direct radiocarbon dating of prehistoric cave paintings by accelerator mass spectrometry H´el`ene Valladas Laboratoire des Sciences du Climat et de l’Environnement, Unit´e mixte CEA-CNRS, Bˆatiment 12, Avenue de la Terrasse, 91198 Gif-sur-Yvette Cedex, France

Received 13 December 2002, accepted for publication 26 February 2003 Published 29 July 2003 Online at stacks.iop.org/MST/14/1487 Abstract Advances in radiocarbon dating by accelerator mass spectrometry now make it possible to date prehistoric cave paintings by sampling the pigment itself instead of relying on dates derived from miscellaneous prehistoric remains recovered in the vicinity of the paintings. Presented below are some radiocarbon dates obtained at the ‘Laboratoire des Sciences du Climat et de l’Environnement’ for charcoal used in the execution of prehistoric paintings decorating two French caves: Cosquer and Chauvet. The presentation of the dates will be preceded by a short discussion of the experimental procedure used in our laboratory (pigment sampling, chemical treatment, etc). The ages obtained so far have shown that the art of cave painting appeared early in the Upper Palaeolithic period, much earlier than previously believed. The high artistic quality of the earliest paintings underlines the importance of absolute chronology in any attempt to study the evolution of prehistoric art. Keywords: AMS carbon-14 dating, prehistoric cave paintings, charcoal, Upper Palaeolithic, Cosquer cave, Chauvet cave

1. Introduction Until quite recently cave paintings were dated according to stylistic criteria loosely associated with dates obtained for archaeological remains found in the vicinity of decorated surfaces. About two decades ago radiocarbon dating was revolutionized when accelerator mass spectrometric (AMS) techniques allowed for the dating of organic samples weighing as little as 1 mg. Paintings done in charcoal could now be sampled without visibly damaging the paintings. In addition to wood charcoal, which has received the most attention (Rowe 2001, Valladas et al 1992, Igler et al 1994), beeswax (Nelson et al 1995) and plant residues (Watchman and Cole 1993, Hedges et al 1998) used in the paintings have also been dated. Below we present the approach used at the Laboratoire des Sciences du Climat et de l’Environnement (LSCE) to date Palaeolithic charcoal drawings and paintings and discuss the results obtained in two French caves.

2. Problems peculiar to the dating of prehistoric pigments The first problem, to which there is no simple scientific answer, has to do with the question of the age of the charcoal at the 0957-0233/03/091487+06$30.00 © 2003 IOP Publishing Ltd

time of execution of the painting. Did the artists use freshly made charcoal, leftover material from prior cave occupation (Bednarik 1994) or a mixture of charcoals of several origins? The possibility that fossil charcoal could have been used cannot be excluded either (Bednarik 1994). To compound the problems there is also the possibility that some paintings were retouched by a later generation of artists. Some of these questions can be answered by examining the nature and composition of the pigment under a scanning electron microscope, others require meticulous in situ examination of the pigment layer with a good magnifying glass. The sampling, the first step of the dating process, is done after preliminary analysis has revealed that the black pigment contains charcoal. In some instances the wood could be identified as belonging to the species Pinus. To protect the visual integrity of the drawings, pigment is scraped from rock cracks or from the thickest layers. If the charcoal is well preserved and thick enough, it is best to collect the sample from a limited area of a figure. When possible, two or more samples from different portions of a painting should be taken in order to get several dates and check the age spread. Otherwise, if the pigment layer is too thin and

Printed in the UK

1487

H Valladas

The great majority of contaminants introduce carbon younger than the pigment charcoal and if not eliminated by a proper treatment will produce a date more recent than the true one.

SAMPLE PREPARATION

Acid (HCl, 0.5M) High - purity H2O

3. Sample treatment

Deposition on pre-cleaned quartz filter

Basic treatment Na4P2O7 , NH3 (aq), NaOH

Charcoal Acid (HCl, 1M) H2O

Humic fraction Precipitation and collection on a quartz filter

Drying

Drying

Heating, 1hr at 300 - 320°C in a stream of oxygen

Heating, 1hr at 280°C in a stream of oxygen

Sample transferred to a combustion tube containing CuO and Ag wire and sealed under vacuum

Oxidation at 850°C for 7 hours Purification and determination of CO2 pressure Catalytic reduction to graphite Compression to a pellet used as target

AMS measurement

Figure 1. A diagram illustrating the experimental procedure.

has to be scraped from several points, sometimes far apart, one gets an average date. The collected samples, usually weighing from 10 to 100 mg, often contain calcite grains or clay from the rock face, in addition to wood charcoal. To-date, calcium oxalate, which can be an important contaminant (Watchman 1990, Russ et al 1996, Hedges et al 1998) of outdoor parietal art in semi-arid regions (particularly in the vicinity of cacti, lichens, for example), has not been detected in paintings inside West European caves. A major problem, inherent in all methods of radiocarbon dating, is the possible presence of extraneous carbon (Hedges et al 1989). Exposure of a paintings renders a cave painting’s pigment particularly vulnerable to contamination. The degree of contamination depends on when and how the caves were discovered. For obvious reasons, caves sealed until recently and not open to the public should give the most reliable dates. In caves frequently visited in the past the most common organic contaminants come from contact with visitors’ hands, cloth fibres, acetylene lamp soot, etc. Moreover, all caves harbour a variety of microorganisms whose growth is stimulated by emanations from the human body (Laiz et al 1999). One must also consider contamination by carbonic, humic or fulvic acids transported by underground waters. 1488

The sample pre-processing used to date the Palaeolithic charcoal drawings and paintings has been described in recent publications (Valladas et al 1999, 2001a). The treatment of charcoal varies in intensity according to the sample size. It involves a succession of ‘acid–base–acid’ treatments which first dissolve the carbonates that may have come from the limestone wall or ground water, then humic acids arising from the transformation of organic matter, and bacteria or other living microorganisms. A schematic representation of the treatment steps is shown in figure 1. The residue from the initial acid bath is retained on a pre-cleaned quartz-frit filter and subjected to the subsequent basic treatment. This treatment, gentle at first, is increased in intensity according to the fragility of the sample. One begins with a dilute solution of sodium pyrophosphate whose concentration is increased progressively. Aqueous ammonia of gradually increased concentration is used next, followed by sodium hydroxide treatment in cases of alkali-resistant pigments. As a rule, the treatment stops when the filtrate becomes highly coloured. The coloration suggests that not only have the outer grain layers been stripped, but that a good fraction of the original charcoal has passed into solution. If the treatment is not interrupted in time, no charcoal might be left for dating. The remaining charcoal grains are washed again with aqueous HCl. After the chemical treatment, the purified charcoal or humic acids collected on another quartz filter are heated in a stream of oxygen for about an hour between 280 and 320 ◦ C to remove some additional organic contaminants. Whatever remains is oxidized to carbon dioxide, then reduced to graphite and compressed into pellets for the accelerator (Arnold et al 1987). The purification process eliminates more than 90% of the original mass leaving us with pellets usually containing from 0.5 to 1 mg of carbon (tables 1 and 2, column 3). This procedure has been tested on a piece of charcoal from an Upper Palaeolithic layer (Solutrean). The piece was broken into several subsamples, of which some were subjected to very strong chemical treatment, others treated in the same way as the pigment samples and still others subjected to chemical but not thermal treatment. We found that a strong chemical treatment did not give significantly different results from the weaker treatment usually reserved for the paintings, and that the thermal treatment did eliminate some additional contamination by more recent carbon, since the samples thus treated gave slightly older ages. The results also confirmed the good reproducibility of our protocol (Valladas et al 2001a). The extent of contamination by modern carbon during sample preparation was determined by subjecting several charcoals over 100 000 years old (‘blank’ sample without 14 C) to the same treatment as our pigment samples. This yielded background contamination that was used to make a suitable correction to the calculated pigment ages.

Direct radiocarbon dating of prehistoric cave paintings by accelerator mass spectrometry

Table 1. Radiocarbon dates for prehistoric paintings at the Cosquer cave. Humic acid dates are written in italics. For pictures of the dated paintings see Clottes and Courtin (1994). Ly = Lyon, France; GifA = Gif-sur-Yvette, France. Reference Horse 1

Feline Bison 1

Megaloceros 1 Horse 7 Deer Star mark Horse 5 Hand 12 Bison 2

Hand 1

Hand 19 Oval mark Soil charcoal Soil charcoal Soil charcoal Soil charcoal

GifA 92416 GifA 92417 GifA 92422 GifA 92418 GifA 92419 GifA 92492 GifA 92423 GifA 95135 GifA 95365 GifA 98186 GifA 98196 GifA 98188 GifA 96075 GifA 96072 GifA 95358 GifA 95372 GifA 96069 GifA 95195 GifA 95308 GifA 92409 GifA 92491 GifA 92424 GifA 96073 GifA 96074 Ly-5558 GifA 92348 GifA 92349 GifA 92350

Dateable carbon (mg)

Date year (BP)

Error (year) 1 sigma

1.56 0.94 1.23 1.52 0.64 1.22 0.26 1.25 0.12 0.84 0.29 0.25 0.87 0.84 0.63 0.26 1.79 2.04 0.23 0.86 1.59 0.44 1.3 2.12

18 840 18 820 18 760 19 200 18 010 18 530 16 390 19 340 13 460 19 720 19 740 19 290 17 800 24 730 24 840 23 150 26 250 27 350 23 080 27 110 27 110 26 180 27 740 28 370 18 440 20 370 26 360 27 870

250 310 220 240 200 190 260 200 330 210 340 340 160 300 340 620 350 430 640 430 400 370 410 440 440 260 440 470

2.39 2.17 2.06

Table 2. Radiocarbon dates for prehistoric paintings at the Chauvet cave (Clottes et al 1995). Humic acid dates are written in italics. (∗ Thirteen other dates have been obtained by the LSCE on charcoal samples collected on the ground of the Megaceros Gallery; 11 of them range between 29 700 and 32 900 and the two other between 25 400 and 26 600 years BP.) Dateable carbon (mg)

Date year (BP)

Error (year) 1 sigma

GifA 95132 GifA 95133 GifA 95126 GifA 96065 GifA 98157 GifA 98160 GifA 95129 GifA 95130 GifA 95158

1.4 1.22 0.8 0.69 0.27 2.3 1.76 0.308

32 410 30 790 30 940 30 230 20 790 29 670 26 980 26 980 25 700

720 600 610 530 340 950 410 420 850

GifA 96063 Ly-6878

0.85 5.000

31 350 29 000

620 400

GifA 95128 GifA 95155

0.83 0.42

30 340 30 800

570 1.500

GifA 95127 GifA 99081

1.22 1.73

26 120 26 230

400 280

GifA 99809 GifA 99810 GifA 99811

2.27 1.12 2.21

32 360 31 390 32 600

490 420 490

Reference Hillaire Chamber Right rhinoceros Left rhinoceros Running cow Horse Torch scraping 1

Megaceros Gallery* Megaloceros Soil charcoal Salle du Fond Bison Cierge Chamber Torch scraping 2 Hearth Crˆane Chamber Under bear skull

1489

H Valladas

Figure 2. Schematic layout of the Gif-sur-Yvette AMS apparatus—‘Tandetron’ (Duplessy and Arnold 1985).

4. Description of tandem mass accelerator Figure 2 shows a schematic diagram of the Gif-sur-Yvette tandem accelerator (UMS 2004, CNRS-CEA) which is used to measure 14 C/12 C and 13 C/12 C isotopic ratios (for a detailed description see Duplessy and Arnold (1985) and Arnold et al (1987, 1989)). The apparatus has three major components: (a) Low-energy ion source: positive caesium ions that are focused on the carbon sample. The caesium ion beam bombards the sample and produces both molecular and atomic negative ions. This step is followed by a first mass separation so that ion beams of mass 12, 13 and 14 are well separated. As a result, the mass 12 consists of only the ions 12 C− , but the mass-13 beam comprises both 13 − C and 12 CH− and the mass-14 beam comprises 14 C− together with unwanted molecules such as 13 CH− , 12 CH2− , which are roughly 109 –1011 times more abundant than the 14 − C ions to be measured. (b) Molecular elimination with the tandem accelerator: the mass-14 ion beam is then injected in the tube of the tandem accelerator where the negative ions are first accelerated. A small stream of argon is injected in the central section of the instrument so that polyatomic molecules are dissociated by gas collisions and transformed into a mixture of positive ions H, 12 C, 13 C and 14 C ions (most probable charge: +3) which are accelerated in the final section of the tandem. Their energy is a function of both mass and charge so that each particle has a unique energy signature. 1490

(c) The final separation, detection and counting of 14 C ions in the high-energy section: the ion beam which leaves the tandem accelerator is then re-focused and enters an electrostatic deflector then another magnetic spectrometer which allows the ions to be separated according to their energy, their mass and their charge. Finally the beam which is mainly made of 14 C ions with a small fraction of 13 C and 12 C ions enters a gas-filled (argon–methane) ion chamber, where carbon ions lose their energy by gas collision; the 14 C ions are easily detected and separated from the unwanted particles. To get the sample age, as in the case of conventional radiation-counter radiocarbon dating, one compares the calculated 14 C/12 C and 13 C/12 C isotope ratios with those obtained for reference standards of known age. Two types of operation are needed to determine the carbon isotopic composition of an unknown sample: (i) measurement of the number of 14 C ions; (ii) successive introduction of mass 12 and 13 beams into the accelerator without changing its tuning and measurements of the 12 C3+ and 13 C3+ beams collected by Faraday cups. The two sets of measurements are repeated until sufficient counts have been obtained for the required statistics and the apparent 14 C/12 C and 13 C/12 C isotopic ratios to be determined. Without changing the accelerator tuning, the apparent isotopic ratios are measured for the reference sample and also for the blank.

Direct radiocarbon dating of prehistoric cave paintings by accelerator mass spectrometry

5. Discussion Whenever enough material is available, multiple datings are done on the same drawing to test the reproducibility and coherence of the results, and on the humic acid fraction obtained during the basic treatment to see how much the initial pigment sample might have been contaminated (Batten et al 1986). In real life situations one encounters three types of cases, illustrated by the dates obtained for the Cosquer and Chauvet cave drawings listed in tables 1 and 2 (the ages obtained with the humic acid fractions are given in italics). In case 1 the purified charcoal and the humic acid fraction yield similar results (horse 1, hand R7 in the Cosquer cave, bison and torch scraping 1 in the Chauvet cave). While good agreement between the two sets of dates generally increases one’s confidence in the reliability of the dates, one can never exclude a remote possibility that both fractions may have somehow been similarly contaminated. When the two fractions yield different dates the humic acid fraction, which one expects to contain more contaminants, tends to give a lower figure—case 2 (see bisons 1 and 2, Megaloceros 1 at Cosquer cave). In such cases the age of the purified charcoal is more trustworthy. The less common case 3 refers to examples where the humic acid fraction yields an age greater than the purified charcoal (see the horse in the Chauvet cave and some Altamira bisons (Valladas et al 2001a)). In general, we have found older dates for a given sample to be more reliable after noting how much more frequent was contamination by recent carbon and consequent age-reduction. Exposed pigments can be polluted by organic materials, some of which can resist the chemical treatment meant to eliminate them. Some samples are so small and fragile that if the solid component is not to dissolve completely the purification has to be less rigorous. In such cases the humic acid fraction, which consists of parts of original charcoal that were dissolved in an alkaline environment and re-precipitated, will give a more correct greater age.

6. Results The tabulated ages obtained for drawings in the Cosquer and Chauvet caves show what type of important information can nowadays be obtained by the use of AMS radiocarbon dating. These caves, which are currently being studied by multidisciplinary teams, provided optimum sampling conditions, so that more than one sample of certain figures could be dated. It was also possible to compare the ages of different fractions of a given scraping to see if the dates are coherent. Moreover, by dating some of the abundant charcoal fragments found on the ground near the drawings we were able to determine the periods of human presence in the cave, a presence that may have been related to artistic activities. The Cosquer cave, whose entrance is now 40 m below sea level, is richly decorated in rock paintings and carvings (Clottes et al 1992, 1997). About 24 dates (table 1) were obtained for 13 charcoal drawings including animal figures, negative hands and geometric signs. Some pigment samples scraped from several points of a figure were divided in two and the two halves were treated and dated separately (horse 1, bisons 1 and 2); they yielded compatible ages, suggesting that these paintings were done within a relatively short time period

with charcoal coming from the same tree (or contemporaneous trees). The paintings can be grouped into two time periods about 10 000 years apart (table 1). This fact is in agreement with the conclusions based on the observation of the decorated wall (Clottes and Courtin 1994). The first group consisting of negative hands, a bison and an oval sign were dated to between 28 000 and 27 000 years ago, during the Gravettian period. Except for one horse, the other animals and the starlike sign were dated to between 19 700 and 18 500 years BP, during the Solutrean period. Taking into account the amplitude of the errors, it is not possible to conclude if each of the two painting phases lasted a brief period of time or stretched over centuries. On the other hand, until more drawings are sampled we cannot tell if a horse and one stencilled hand, which seem to date to about 25 000 BP represent an intermediate period of decoration or are the result of more extensive contamination. The time span that separates the two bison (1 and 2) which are similar and depicted on the same wall is rather surprising. This fact can be interpreted in at least two ways: either the stylistic conventions were maintained over extremely long time periods, or the older one was not done with fresh charcoal (Clottes et al 1997). To help us choose between these alternatives additional dates will be needed. It is noteworthy that charcoal fragments collected on the ground nearby also fall within two distinct time intervals: 18 000–20 000 and 26 000– 28 000 years, respectively. Chauvet cave was discovered in Ard`eche in December, 1994 (Clottes 2001). So far about 40 dates have been obtained (Clottes et al 1995, Valladas et al 2001b): twelve on pigments from six drawings from different sections of the cave, two for charcoal scrapings left by visitors who rubbed their torches against the wall and the rest for the charcoal found in abundance on the ground (table 2). The great majority of dates can be grouped into two tight clusters representing two timeperiods thousands of years apart (29 000–32 500 and 26 000 and 27 000 years BP respectively). The animal representations were dated to between about 32 000 and 30 500 years BP, within the Aurignacian period. The torch scrapings were about 27 000 years old, a date not surprising if one notes that in one case the torch was scraped against a layer of calcite deposited on top of a drawing! So far, there is no drawing dated to this second period of human occupation. Most of the ages obtained for charcoal collected on the ground surface ranged from 26 000 to 32 000 years BP, suggesting the existence of at least two major episodes of human intrusion before the cave was sealed off by a rock-fall. The coherence of the dates obtained for the drawings of the Cosquer and Chauvet caves suggests that the samples were not seriously contaminated. Such satisfactory results can be attributed to the great number of samples available for dating and to the fact that the cave was sealed by a natural phenomenon during the Pleistocene period.

7. Conclusion Even though the direct dating of cave paintings is still in its infancy, the few dates reported so far have convinced art historians of the need to revise prior ideas on the evolution of prehistoric art. The Chauvet cave, in particular, indicates that theories assuming a linear progression from simple to more complex composition have to be discarded and that, as early 1491

H Valladas

as the Aurignacian period, some artists had mastered design and composition (Clottes et al 1995). The AMS radiocarbon dating also makes it possible to establish distinct periods of artistic activity within any one cave. At the present time the AMS technique does not allow for reliable dating of Palaeolithic drawings done in media other than charcoal. Unfortunately, the media most commonly used in the execution of prehistoric paintings are iron oxide and manganese oxide (Menu 2000). There is hope that some of these may be dateable in the future, since chemical analyses have revealed that organic binders of plant or animal origin were occasionally used with mineral pigments (Pepe et al 1991). The quantities are usually tiny, but improved chemical techniques will undoubtedly allow us one day to separate, purify and date such binders.

References Arnold A, Bard E, Maurice P, Valladas H and Duplessy J C 1989 14 C dating with the Gif-sur-Yvette Tandetron accelerator: status report and study of isotopic fractionations in the sputter ion source Radiocarbon 31 284–9 Arnold M, Bard E, Maurice P and Duplessy J C 1987 14 C dating with the Gif-sur-Yvette Tandetron accelerator: status report Nucl. Instrum. Methods B 29 120–3 Batten R J, Gillespie R, Gowlett J A J and Hedges R E M 1986 The AMS dating of separate fractions in archaeology Radiocarbon 28 698–701 Bednarik R 1994 About rock art dating Int. Newsletter Rock Art 7 16–18 Clottes J (ed) 2001 La Grotte Chauvet. L’art des Origines (Paris: Seuil) Clottes J, Chauvet J M, Brunel-Deschamps E, Hillaire C, Daugas J P, Arnold M, Cachier H, Evin J, Fortin P, Oberlin C, Tisnerat N and Valladas H 1995 Les peintures pal´eolithiques de la grotte Chauvet-Pont d’Arc, a` Vallon-Pont-d’Arc (Ard`eche, France): datations directes et indirectes par la m´ethode du radiocarbone C. R. Acad. Sci., Paris II 320 1133–40 Clottes J and Courtin J 1994 La Grotte Cosquer (Paris: Seuil) Clottes J, Courtin J, Collina-Girard J, Arnold M and Valladas H 1997 News from Cosquer cave; climatic studies, recording, sampling, dates Antiquity 71 321–6 Clottes J, Courtin J, Valladas H, Cachier H, Mercier N and Arnold M 1992 La grotte Cosquer dat´ee Bull. Soc. Pr´ehist. Fran¸caise 89 230–4 Duplessy J C and Arnold M 1985 Radiocarbon dating by accelerator mass spectrometry Nuclear Methods of Dating ed E Roth and B Poty (Dordrecht: Kluwer) pp 437–53

1492

Hedges R E M, Bronk C R, Van Klinken G J, Pettitt P B, Nielsen-Marsh C, Etchegoyen A, Fernandez Niello J O, Boschin M T and Llamazares A M 1998 Methodological issues in the radiocarbon dating of rock paintings Radiocarbon 40 35–44 Hedges R E M, Law I A, Bronk C R and Housley R A 1989 The Oxford accelerator mass spectrometry facility: technical developments in routine dating Archaeometry 31 99–113 Igler W, Dauvois M, Hyman M, Menu M, Rowe M, Vezian J and Walter P 1994 Datation radiocarbone de deux figures pari´etales de la grotte du Portel (Commune de Loubens, Ari`ege) Bull. Soc. Pr´ehist. Ari`ege-Pyr´en´ees XLIX 231–6 Laiz L, Groth I, Gonzales I and Saiz-Jimenez C 1999 Microbiological study of the dripping waters in Altamira cave (Santillana del Mar, Spain) J. Microbiol. Methods 36 129–38 Menu M 2000 Le savoir faire des premiers peintres La Recherche, Hors-S´erie 4 56–8 Nelson D E, Chaloupka G, Chippindale C, Alderson M S and Southon J R 1995 Radiocarbon dates for beeswax figures in the prehistoric rock art of Northern Australia Archaeometry 37 151–6 Pepe C, Clottes J, Menu M and Walter P 1991 Le liant des peintures pr´ehistoriques ari´egeoises C. R. Acad. Sci. Paris II 312 929–34 Rowe M W 2001 Dating by AMS analysis Handbook of Rock Art Research ed D S Whitley (Walnut Creek, CA: AltaMira) pp 139–66 Russ J, Palma R L, Loyd D H, Boutton T W and Coy M A 1996 Origin of the whewellite-rich rock crust in the lower Pecos region of Southwest Texas and its significance to paleoclimate reconstructions Quat. Res. 46 27–36 Valladas H, Cachier H, Maurice P, Bernaldo De Quiros F, Clottes J, Cabrera-Valdes V, Uzquiano P and Arnold M 1992 Direct radiocarbon dates for prehistoric paintings at the Altamira, El Castillo and Niaux caves Nature 357 68–70 Valladas H, Tisnerat N, Cachier H and Arnold M 1999 Datation directe des peintures pr´ehistoriques par la m´ethode du carbone 14 en spectrom´etrie de masse par acc´el´erateur Rev. Arch´eom´etrie Suppl. 1999 39–44 Valladas H, Tisnerat-Laborde N, Cachier H, Arnold M, Bernaldo De Quiros F, Cabrera-Valdes V, Clottes J, Courtin J, Fortea-Perez J, Gonzales-Sainz C and Moure-Romanillo A 2001a Radiocarbon AMS dates for Paleolithic cave paintings Radiocarbon 43 977–86 Valladas H, Clottes J, Geneste J M, Garcia M, Arnold M, Cachier H and Tisnerat-Laborde N 2001b Evolution of prehistoric cave art Nature 413 479 Watchman A 1990 A summary of occurrences of oxalates-rich crusts in Australia Rock Art Res. 7 44–50 Watchman A and Cole N 1993 Accelerator radiocarbon dating of plant-fibre binders in rock paintings from northeastern Australia Antiquity 67 355–8