Content uploaded by Roberto Ontañón
Author content
All content in this area was uploaded by Roberto Ontañón on Jun 27, 2016
Content may be subject to copyright.
Efficient quantification procedures for data evaluation of portable X-ray fluorescence –
Potential improvements for Palaeolithic cave art knowledge
M. Gay
a
, K. Müller
a
,F.Plassard
b
, J.-J. Cleyet-Merle
c
,P.Arias
d
,R.Ontañón
d
,I.Reiche
a,e,
⁎
a
Sorbonne Universités, Université Paris 6, Laboratoire d'Archéologie Moléculaire et Structurale, UMR 8220 CNRS –Université Pierre et Marie Curie, 4 place Jussieu, 75005 Paris, France
b
Université de Bordeaux, UMR 5199, PACEA, Allée Geoffroy Saint-Hilaire, CS 50023-33615 Pessac Cedex, France
c
Musée National de Préhistoire, 24620 Les Eyzies-de-Tayac, France
d
Instituto Internacional de Investigaciones Prehistóricas de Cantabria, Universidad de Cantabria, av. de los Castros 52, 39005 Santander, Spain
e
Rathgen-Forschungslabor, Staatliche Museen zu Berlin-Stiftung Preußischer Kulturbesitz, Schloßstraße 1a, 14059 Berlin, Germany
abstractarticle info
Article history:
Received 6 February 2016
Received in revised form 26 May 2016
Accepted 2 June 2016
Available online xxxx
Portable x-ray fluorescence spectrometry (pXRF) has become fundamental in prehistoric research since it en-
ables chemical studies that preserve the integrity of rock art or other investigated archaeological objects. This
unique and fragile expression of our ancestors requires the use of non-invasive and non-destructive in situ ana-
lytical techniques. This provides significant sources of physicochemical information for enhancing the compre-
hension of the symbolic and ideological realm of past societies. Thus, XRF data acquired in the field allow
giving more detailed insights into the pigment used by Palaeolithic artists, the rock art organisation inside the
cave and the different frequentation periods of it. However, ifthe qualitative study is now well established and
routinely used, quantitative evaluation encountersdifficulties linked to the context of the study (karsticenviron-
ment in our case) and theheterogeneous nature of the analysed material (nature of the pigments used, presence
of several layers, conservation state of the rock art, type of the rock art support). Moreover, the non-invasive na-
ture of this technique is faced with a large number of data since it offers the acquisition of statistically relevant
data by multiple measurements of different spots on the same figure. The present work struggles with the
issue of filling the gap of well-adapted quantitative procedures devoted to caves or rock-shelters analyses, and
offers efficient tools and methodologies, which take into account the specificities of the studied rock art and its
context. Additionally, the evaluation procedures of the high volume of data have to be effective. The analyses
of drawings, monochrome and polychrome paintings of three Palaeolithic key cave sites, namely Rouffignac
and Font-de-Gaume in Dordogne, Southern-France, and La Garma in Cantabria, Northern Spain, illustrate the
new approaches and procedures developed in this study.
© 2016 Elsevier Ltd. All rights reserved.
Keywords:
Portable X-ray fluorescence analyses
Prehistoric paints
Cave art
Iron and manganese oxides
Quantification procedures
Principal component analysis
Monte Carlo simulations
1. Introduction
The cognitive behaviour of the Prehistoric populations, which sur-
rounds the artistic work in caves or rock-shelters, is still fascinating.
These artistic evidences belonging to the first ones known of human
mankind that reached us from this distant period belong to the longest
timespan of art history. For some of this rock art,Prehistoric artists have
even ventured in closed and dark spaces, sometimes over long dis-
tances. On the basis on this exceptionality, the understanding of these
graphic expressions is one of the key issues of prehistoric research.
Many studies have been carried out since the recognition of the au-
thenticity of Palaeolithic rock art in 1902. It has been described and pre-
cisely surveyed to understand the symbolic and ritual significations.
Physicochemical analyses, according to the instrumental developments
made in the last few years in the portable XRF technique, have been in-
cluded increasingly to the stylistic approach to become nearly automat-
ic. These analyses focus on the determination of the chemical
composition of the pigments, the methods of paint preparation and
the application techniques employed. The research path is providing
significant sources of information to get to the technical skills of prehis-
toric artists and the “chaîne opératoire”of the painting activities
(Cabrera-Garrido, 1978; Ballet et al., 1979; Vandiver, 1983; Clottes et
al., 1990; Lorblanchet et al., 1990; Pepe et al., 1991; Menu and Walter,
1992; Baffier et al., 1999; Chalmin et al., 2002, 2003; Vignaud et al.,
2006). However, many questions remain about the creation, dating
and meaning of these artistic behaviours and research on it is still on-
going.
The study of rock art encounters severe restrictions, which are im-
posed by the necessity of complete conservation of its integrity. The
analysis has to be done on-site, non-destructively and non-invasively,
to strictly preserve the artwork. Portable X-ray fluorescence (pXRF)
Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
⁎Corresponding author at: Sorbonne Universités, Université Paris 6, Laboratoire
d'Archéologie Moléculaire et Structurale, UMR 8220 CNRS –Université Pierre et Marie
Curie, 4 place Jussieu, 75005 Paris, France.
E-mail addresses: ina.reiche@upmc.fr,i.reiche@smb.spk-berlin.de (I. Reiche).
JASREP-00512; No of Pages 9
http://dx.doi.org/10.1016/j.jasrep.2016.06.008
2352-409X/© 2016 Elsevier Ltd. All rights reserved.
Contents lists available at ScienceDirect
Journal of Archaeological Science: Reports
journal homepage: www.elsevier.com/locate/jasrep
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
spectrometry is one of the most suitable non-invasive methods to study
rock art on-site. This now widespread method is easily and routinely
implemented to determine to the elemental composition and to achieve
a qualitative analysis of the pigment (Jenkins et al., 1995; Mantler and
Schreiner, 2000;de Sanoit et al. (2005);Hocquet et al., 2008; Roldán
et al., 2010; Nuevo et al., 2012; Beck et al., 2012, 2014; Olivares et al.,
2013; Gay et al., 2015).
Quantitative processing of the XRF data acquired in karstic environ-
ments is challenging and presents difficulties, which are particular to
the analysis of an art performed on a rock support. This rock art, when
regarded as a sample, can be defined as a simplified two-layered
model in the first order: the first layer being the pigment and the sec-
ond, the substrate. It is supposed that there are no deposits on the
paint surface although this is sometimes encountered in caves as for in-
stance in the Large cave of Arcy-sur-Cure, Yonne, France (Chalmin et al.,
2008) where opaque calcite layers obstruct the Palaeolithic paint layers.
Quantifying the concentration of each chemical element of the pigment
is tricky for several reasons. The paint layer is supposedly very thin (the
exact thickness is not known) and does not cover homogeneously the
rock support. In this instance, X-rays coming fromthe source go through
the paint layer and penetrate the substrate. The result is a high contribu-
tion of the XRF substrate signal in the paint signal. Furthermore, a sub-
traction of the substrate contribution in the paint spectra proved to be
incorrect to extract the pigment information, since the fluorescence in-
tensities measured for each analysed point on the paint and on the pure
substrate cannot be compared directly. Indeed, the experimental condi-
tions specific to the karstic environment (non-planarity of the wall)
cause a shift in the position of the pXRF spectrometer facing the wall,
uncontrolled X-ray incidence and detection angles, and consequently
a significant and unquantifiable bias when comparing different mea-
surements. This plays an important role in the varying fluorescence in-
tensity and the uncertainty of the measurement. Considering this, the
evaluation of the data cannot be based on a single procedure reused
for each site, regardless its specificities, even for different measure-
ments within one site. Indeed, we show here, through distinct case
studies, which provide very different paint systems, that it is necessary
to adapt the quantitative evaluation procedure of the XRF data and to
develop working methodologies that consider the specificity of each
site. The final objective is to offer an adequate strategy to study a specific
rock art paint system effectively. To our best knowledge such a system-
atic approach has never been performed before for XRFstudies on rock
art.
2. Adequate quantitative methodologies for distinct contexts
Different aspects have to be considered before studying an artwork
in a karstic context. Each Palaeolithic site has its own specificities,
starting from the paint matter used by prehistoric artists to create the
artwork to the support on which the paint matter is applied. Two
main mineral pigments have been employed, iron and manganese ox-
ides besides carbon black. They enabled the production of the three
principal Palaeolithic colors, red, yellow and black (Aujoulat, 2004).
Used alone or associated, they offered the media to execute drawings,
mono- or polychrome paintings arranged sometimes in impressive
compositions. To study this rock art with the adequate quantitative
methodology, the kind of wall support and the paint preservation
state has to beassessed too. Closelytied to the environmental equilibri-
um of the cave, the wall's state and their evolution over time will impact
the cave art, depending on several factors (morphologic, geologic, hy-
drologic hydrogeological, physicochemical, chemical, climatic and bio-
logical factors) (Shoeller, 1967; Kervazo et al., 2010; Ferrier et al.,
2012, 2014; Lacanette et al., 2013) as well as depending on the process-
es occurring on the surface of the substrate. Some alteration mecha-
nisms could be illustrated by the formation of calcite coverings
(Chalmin et al., 2008; Chalmin and Reiche, 2013), vermiculations
(Hœrlé et al., 2011; Konik et al., 2014) and the development of micro-
organisms (Bastian et al., 2010; Martin-Sanchez et al., 2012). Further-
more, some of the caves are opened to the public and equipped for vis-
itors but few others have never been opened to the public and thus
provide a closed karst system without visitor equipment.
Although every cave art site is a single case, the painted rock art type
in karst environments can be categorised in three classes in a first ap-
proximation: black rock art without any coating, red or yellow rock
art without coating and a complex rock art composed of mixed paints
and presenting various preservation states of the support and the
paint layers. Three contexts allowed us to address these three categories
during our research in three Palaeolithic caves: Rouffignac and Font-de-
Gaume located in the Périgord region in the south-west of France, and
La Garma in the Cantabrian region of Spain. Three quantitative ap-
proaches have been tested in these caves in order to evaluate their ap-
propriateness as a function of the defined category, to enhance the
knowledge of archaeologists on the rock art based on stylistic observa-
tions and to bring new physicochemical insights into the organisation
of the representations and their relationship with the others inside the
cave.
2.1. Black rock art case
First, rock art that is only composed with black pigments assimilated
to manganese oxides is considered. Their chemical composition is suffi-
ciently different from that of the calcareous wall substrate to allow defin-
ing the black pigment with very few characteristic chemical elements. As
shown in de Sanoit et al. (2005) and Beck et al. (2012, 2014),theblack
pigment can be determined from three main oxides, MnO
2
,Fe
2
O
3
and
BaO since the contribution from the support of Mn and Ba can be
neglected and that of Fe has a little effect on the pigment signal (Fig. 1).
The quantitative evaluation to determine the concentration of these
three oxides can be made with Monte Carlo simulations. It allows
obtaining their theoretical relative proportion (de Sanoit et al., 2005).
However, Monte Carlo simulations have provided similar results to
those obtained by extracting their concentration with the fundamental
parameters method (Beck et al., 2012, 2014). This latter one has turned
out to be a more effective method to evaluate XRF data than the Monte
Carlo simulations in this particular case of black pigments, and a much
better adapted method to process large data sets.
Once the concentration of the three oxides is extracted, their sum is
normalized to 100%. This makes measurement results comparable
among each other and independent fromflux variations between differ-
ent analyses. Moreover, this semi-quantification procedure makes the
subtraction of the signal of the support from that of the paint layer un-
necessary. This is essential because a subtraction of the substrate contri-
bution in the paint spectra turns out to be incorrect, for the reasons
previously exposed.
2.2. Physical properties of red or yellow rock art
In contrast to black pigments, the red or yellow ones are assimilated
to iron oxides and have an elementary composition very similar to that
of the substrate (Fig. 2). The absence of specific chemical elements of the
paint does not allow a precise discrimination between paint layer and
support. Thus, the quantification method applied for the black rock art
cannot be used in this case.
A different approach is proposed in Gay et al. (2015) using a semi-
quantification based on CaO content. It is assumed that only the wall
support contains this oxide. Performing a semi-quantification based
on CaO content means that the contribution of the substrate in the spec-
tra detected through the paint layer, when analysing the paint, remains
the same from one analysed point to another. Hence, when normalizing
the CaO content, the result of the wall contribution is made comparable
between measurements and its impact negligible, providing a fair com-
parison between the set of studied representations. However, making
comparable the contribution of the substrate through the paint layer
2M. Gay et al. / Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
from one analysed point to another, presumes that the thickness of the
paint layer is adapted at the same time.This approach is likely related to
non-chemical paint properties, such as the density of the painting mat-
ter or the intensity of the pigment, which are the record of paint prac-
tices of the prehistoric artists, for instance the application techniques
employed and the preservation state of the paint layer.
2.3. Chemical composition of red and yellow rock art
An alternative quantitative procedure is required to access correctly
the chemical composition of the pigment from XRF spectral data. Monte
Carlo simulations seem to be the most suitable method to provide the
chemical concentration of each element composing the red or yellow pig-
ment. Thus, it enables us to consider two layers composing the sample,
the paint layer and the substrate, and to determine the chemical informa-
tion of them separately (Schoonjans et al., 2012). Once a fixed composi-
tion for the substrate and an approximate composition for the paint
layer are defined, the Monte Carlo simulations are applied in an ‘inverse’
manner using iterative algorithms. The paint layer composition is adapted
iteratively until the simulated curve of it converges towards the experi-
mental spectrum, regarding the intensity of the peaks. The implementa-
tion of such an algorithm allows getting the inherent chemical
information of the pigment alone, without the contribution of the support.
2.4. Different conservation states of the rock art
Rock art is especially threatened when the equilibrium of its karstic
ecosystem is disrupted. Several parameters could generate different al-
terations of the rock art, involving various preservation states. When
studyinga set of representations in a cave,these differences in the pres-
ervation states could change the perception of representations and
make their interpretation incorrect, especially in the case of a compari-
son based on the physical properties of thepainting matter. In that spe-
cific case, the data evaluation by a method involving semi-
quantification based on CaO content becomes inappropriate. Monte
Carlo simulations seem to be most advised to analyse a red or yellow
rock art while a definition of black pigments from a few characteristic
chemical elements is most recommended. As a consequence, the choice
of the quantitative evaluating methods of the XRF data has to be deter-
mined according to the alteration state of the rock art, too.
2.5. A heterogeneous substrate
The degree of the substrate heterogeneity is a key parameter to be
considered in the data evaluation. The paint is generally a very thin
layer actually composed of scattered particles. Moreover, the particles
do not cover homogeneously the rock support. These two aspects of
the paint layer (scattering, heterogeneity) play an important role espe-
cially when performing the method based on Monte Carlo simulations.
Indeed, Monte Carlo simulations suppose that the pigment layer is dis-
tributed homogeneously on the rock surface, without discontinuity of
the line. This does evidently not reflect the real situation. Thus, the dis-
continuity of the paint line makes the thickness value under-evaluated
in the Monte Carlo model. It should be kept in mind that the spectra
are a simulated result from the spectral information of the pigment
and the substrate. When the thickness of the paint layer goes below a
critical value, its weight in Monte Carlosimulations becomes negligible
and simulations will only take into account the chemical composition of
the substrate and not the information of the pigment. Therefore, simu-
lating the experimental curve only by adapting iteratively the chemical
composition of the pigment does not work anymore. Moreover, the
more discontinuous and/or thinner the paint layer is, the more drastic
the variations of the wall composition will be in the spectrum of the pig-
ment. In the case of a very heterogeneous substrate, it is difficult to de-
termine precisely its real composition underneath the pigment. This
highlights the importance of the thickness and the continuity of the
paint layer under investigation by the X-ray beam,in order to decrease
the weight of the substrate information. Systematic analyses of the wall,
just next to the paint or pigment stroke measurement spot, are essential
to record its potential heterogeneity.
3. Methods
3.1. Portable X-ray fluorescence spectrometer
The XRF analyses have been carried out withan in-house developed
device composed of: a 40 kV MOXTEK X-ray tube with a palladium
Fig. 1. Example of the XRF spectra from a black pigment associated toa manganese oxide (black spectrum) and from a limestone substrate (grey spectrum) (Bison 100 of the “Great
Ceiling”of Rouffignac cave).
Fig. 2. Example of the XRF spectra from a red pigment associated to an iron oxide (red
spectrum) and from a limestone substrate (grey spectrum) (Aurochs IV-6 of the panel
of the southern part of zone IV in the Lower Gallery of La Garma cave). (For
interpretation of the references to color in this figure legend, the reader is referred to
the web version of this article.)
3M. Gay et al. / Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
anode and a collimator to focus the beam spot to a size of approximately
1mm
2
;a7mm
2
Silicon Drift Detector (SDD) with an energy resolution
of 140 eV (FWHM at 5.9 keV) to collect the XRF signal. Both the X-ray
tube and detector are fixed on a positioning system, with micrometric
adjustment, again mounted on a tripod (Fig. 3). This has enabled the
analysis of panels that are difficult to access. The incident angle of the
X-ray is 45° while the detector is perpendicular to the analysed surface.
Two lasers have been used to mark the analysis location and to keep a
fixed 5 mm working distance between the spectrometer and the
ornamented surface. Consequently the thickness of atmosphere layer
between the object surface and the detector is kept constant too from
one measurement to another. This configuration is well suited for the
study of rock art since there is no contact with the decorated wall. The
relatively short time of thesignal acquisition, which is 5 min, has offered
the acquisition of statistically relevant data by multiple measurements
of different spots on the same artistic figure, as illustrated with Fig. 4.
The underlying rock surface has been analysed too, next to the analysis
spot on the pigment. This systematic analysis of the rock substrate, as
stated before, is important because it provides an evaluation of the
wall heterogeneity.
In situ measurements have been performed in theatmosphere with-
out helium flux. Thus, the detection of the elements lighter than Al is not
possible because of the absorption of their X-ray florescence radiation
by air components. The semi-quantification of the data can start from
P. Below the mass of P, the emittedX-rays of the elements are attenuat-
ed and semi-quantification cannot be performed because of the uncer-
tainty of the X-ray attenuation.
3.2. Tools for quantification and implementation
Semi-quantitative data acquired by XRF have been extracted by
means of two distinct open source softwares:
-Fromthefit results of the spectra using the fundamental parameter
method and implemented in the PyMca software developed by Solé
et al. (2007).
- From the Monte Carlo simulation method and implemented in XMI-
MSIM software developed by Schoonjans et al. (2012).
For Monte Carlo simulations, the sample is defined as a simplified
two-layered model in the first order: the first layer being the pigment
and the second, the substrate. A third layer is inserted between the sam-
ple and the spectrometer, representing the atmosphere. Layers are sup-
posed to be homogeneous. The definition of multilayers is possible with
PyMca software but the chemical elements of each layer have to bedif-
ferent to quantify them. In the case of red or yellow paint on limestone
rock, the pigment and the support layers are composed of basically the
same chemical elements. Whenusing XMI-MSIM software, thefirst step
is to define the composition of the substrate next to the measured spot
on the paint mark. The substrate concentrations are calculated by
means of PyMca and then put intoXMI-MSIM. This step is essential es-
pecially when the substrate composition is heterogeneous. Once the
substrate layer is well defined, the paint layer is added to the model.
Its composition is first roughly estimated with PyMca by subtracting
Fig. 3. Detailed illustration of the pXRF device developed at LAMS (a) and picture of its implementation in La Garma cave(b).
4M. Gay et al. / Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
the wall contribution from the spectrum. This estimation, even if inex-
act, is a first basis for the Monte Carlo simulation of the quantitative
composition of the paint layer. The experimental curve is then simulat-
ed by iteratively adapting the estimated chemical composition of the
pigment (Fig. 5). The procedure is not automated yet, making it a time
consumingapproach. To give an order of idea, the simulation of one pig-
ment spectrum took around 3 h.
The value of the paintlayer thickness in the used Monte Carlo model
is under-estimated (less than one micrometre) and does not reflect the
real situation of the presence of scattered particles. Nevertheless, this
enables to compensate for the discontinuity of the paint layer. Despite
this, the model stays correct in a first approximation and the results
are comparable from one spectrum to each other.
4. Three Palaeolithic caves, three adapted strategies
4.1. The case of Rouffignac
The cave of Rouffignac is located in the Rouffignac-Saint-Cernin
commune, in the Dordogneregion (France). Theprincipal pictorialtech-
niques used at Rouffignac by prehistoric artists are drawings, without
exception black drawings, and engravings. Correctly named “the hun-
dred mammoths' cave”,Rouffignac is famous for its 160 graphical mam-
moths, which accounts for two-thirds of the total ornamentation of the
cave. It confers to the cave its significantplace in the Palaeolithic rock art
landscape (Barrière, 1982; Plassard, 2005; Plassard and Plassard, 2000,
2016). The black pigments were studied in-depth for two major panels:
the “ten mammoths Frieze”and the “Great Ceiling”(de Sanoit et al.
(2005);Beck et al., 2012, 2014).
The “ten mammoths Frieze”, located in the “Henri Breuil Gallery”,
gathers four mammoths facing six others. The ten representations
were executed at a constant scale and on a symmetrical axis along the
wall. Whereas the frieze is structured with a stylistic unit, the “Great
Ceiling”is organisedwithout apparent structuration. The “Great Ceiling”
assembles 25% of the total rock art of the cave on a surface of less than
40 m
2
, and the five principal animal species represented in the cave.
They are juxtaposed as single elements and as groups as well as
superimposed leading to an irregular organisation.
The Mn oxides, composing the black pigment of these two panels,
were characterised by XRF andX-ray diffraction as well as Ramanspec-
troscopy (Beck et al.,2012, 2014; Lahlil et al., 2012). The chemical com-
position of the black pigment was extracted from the three different
oxides, MnO
2
,Fe
2
O
3
and BaO, following the methodology discussed
above for the first study case of black rock art. Actually, this method of
processing has proven its efficiency for the analytical evaluation of
such data (Beck et al., 2012, 2014). Thus, results enabled to differentiate
the existence of homogeneous (“ten mammoth Frieze”) and heteroge-
neous (“Great Ceiling”) pictorial groups, supporting the stylistic obser-
vations. One type of Mn oxide paint was found to be used for
executing the homogeneous ten mammoths, while at least two differ-
ent Mn oxide paints (Ba rich or not) were identified in the representa-
tions of the “Great Ceiling”, underlining its heterogeneity (Fig. 5). In
addition, X-ray diffraction (XRD) analyses confirm the presence of two
distinct Mn oxides, clearly identified as pyrolusite (MnO
2
)and
romanechite (Ba
2
Mn
5
O
10
)(Beck et al., 2012, 2014; Lahlil et al., 2012).
Actually, the “Great Ceiling”gathers 65 black animal drawings. The
number of the analysed figures amounts to less than one quarter of
the total figures, givingonly an incomplete understandingof the general
organisation of this panel, too poor for its interpretation. To offset the
current lack, a systematic characterisation of the Mn oxides is engaged
in Rouffignac by the LAMS (Gay et al., forthcoming). Twelve additional
drawings were already analysed by portable XRF (Fig. 6).
4.2. The case of La Garma
The red rock art of La Garma can be assimilated with the second
study case. This cave contains more than five hundred exceptional
Palaeolithic graphical units, some of them linked to the Magdalenian
floors. La Garma cave, located in the Cantabria region, has never been
opened to the public and thus provides a closed karst system with un-
touched archaeological surfaces, conferring to it an exceptional position
in the study of the Upper Palaeolithic in the North of Spain (Arias et al.,
1996, 2000, 2011; Ontañón, 2003; Arias and Ontañón, 2012). Two zones
were analysed: one decorated panel of the southern part of zone IV in
the Lower Gallery and one in the Intermediate Gallery (Gay et al.,
2015). The first panel is a singular and especially interesting one. It
Fig. 4. Multiple measurements of different spotson the same artistic figure(white points)
and on the substrate (crosses) performed to acquire statisticallyrelevant XRF data and to
evaluatethe wall heterogeneity. The example of mammoths89 and 94, and the rhinoceros
93 of the “Great Ceiling”of Rouffignac cave.
Fig. 5. Experimented(purple) and simulated (blackdotted line)spectra of the red painting
bison 4 of Font-de-Gaume cave. (For interpretation of thereferences to colorin this figure
legend, the reader is referred to the web version of this article.)
5M. Gay et al. / Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
includes five red animal drawings (one aurochs, two quadrupeds, one
horse and one megaloceros), which are correlated with two successive
phases of decoration. This interpretation is based on the superimposi-
tion of the figures and the types of paint one on each other. Indeed, dif-
ferent line widths and colors of the paint layers are observed. The
second panel corresponds to red pigment deposits on a stalagmitic
column.
The chosen strategy to evaluate the XRF data had to be adapted to
the analysis of the red pigment, as described above. The quantification
method used for the analysis of the black rock art in Rouffignac is inap-
propriate in this case. A semi-quantification based on fixing the CaO
content was performed to determine the concentration of the chemical
elements composing the paint. It was shown that the physical proper-
ties (the density and/or the intensity of the pigment) have a significant
impact on the distinction of groups, one particular to each gallery, when
coupling with a principal component analysis (PCA) procedure (Fig. 7)
as described in Gay et al., 2015. According to the PCA, the division into
two groups depended on the types of iron oxides and, thus, can be
linked to different preparation methods of the paint palette. A difference
of the paint density could also explain the distinction and would indi-
cate different methods of applying paint by prehistoric artists. Such a
processing method provides a fair comparison between the representa-
tions, while neglecting the contribution of the support. This is clearly
shown in the multivariate analysis, especially in the loading plot
where the CaO content has no influence on the formation of the two
groups visualised in the score plot.
Fig. 7. Results obtainedby XRF analysis at La Garma cave. Score plot (PC1 vs. PC2: topleft) of the analysed red rock art inthe Intermediate Gallery (stalagmiticcolumn 1, photo: top right)
and in the Lower Gallery (aurochs IV-6, megaloceros IV-7, quadrupeds IV-8and IV-9, horse IV-11, photo: bottom right); and the loading plot to specify themain oxides influencing the
structuration of the dataset (bottom left) (Gay et al., 2015). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 6. Ternary diagram of the concentration of the three main oxides (MnO
2
,Fe
2
O
3
and BaO)
characterising black pigments of the “ten mammoths Frieze”and the “Great Ceiling”of
Rouffignac cave according to de Sanoit et al. (2005);Beck et al., 2012, 2014, and our study
(Gay et al., forthcoming). The dispersion of the Fe
2
O
3
values is due likely to a
heterogeneous substrate under the black paint layer. (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of this article.)
6M. Gay et al. / Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
4.3. The case of Font-de-Gaume
The analytical study of the rock art in the Font-de-Gaume cave is
more complex and corresponds to the situations described in 2.3 to
2.5. The Font-de-Gaume cave is located at the entrance of the village
of Les Eyzies-de-Tayac, in the Dordogne region. It is one of the rare
caves with monochrome drawings and polychromic painting compara-
ble to those of the Lascaux cave discovered in France and still open to
the public. A remarkable rock art abounds in the second part of the
cave quite far from the entrance. By contrast, only the vestiges of this
Palaeolithic rock art remain in the first part closer to the entrance. Dif-
ferent techniques are gathered and mixed in the cave, paintings, draw-
ings and engravings, to form large friezes and rich compositions of
artworks (Capitan et al., 1910; Plassard, 2005; Cleyet-Merle, 2014). To
this day, there is hardly any physicochemical study available that con-
tributes to the general understanding of this cave, except some very
few chemical analyses dating from 1902. In the absence of a direct dat-
ing possibility of the prehistoric figures, a new vision of the chemical
composition would offer further basis for reflection to group represen-
tations that were created in a similar period of time and can support
the stylistic knowledge of the site. In the frame of our research project,
systematic physicochemical investigations have been conducted since
2012 to characterise the painting matter used in Font-de-Gaumeby pre-
historic artists and to compare the chemical fingerprint of the pigments.
In Font-de-Gaume several paint types are coexisting. The black rock art
can be studied using the same procedure as in thecase of Rouffignac.
The study of the red rock art of Font-de-Gaume has required a new
methodological approach, different from the other cases described
above because the support is very heterogeneous and its state of preser-
vation is very variable. At Font-de-Gaume cave, thepigments have been
subjected to various leaching processes resulting in differential preser-
vation states. Therefore, the paint layers of different representations
cannot be compared based on physical properties of the paint matter,
as proceeded at La Garma (Fig 7). In order to determine the chemical
composition of the red pigment, an evaluation method based on
Monte Carlo simulations was adopted.
From an archaeological perspective, the preliminary results of the
work carried out in the “principal gallery”at Font-de-Gaume cave
showed that the red painting matter used for the analysed bisons is rel-
atively homogeneous. This is consistent with the stylistic study of the
Fig. 8. Preliminary results obtained for Font-de-Gaume cave. Top: Illustration of therepresentations analysed by XRF in the first part of the “principal gallery”(bison 38,39 and 43 at the
right sidewall). Middle: Scoreplots (PC1 vs. PC2) of the redand black paint matter.Bottom: Respectiveloading plots to specifythe main oxides influencing the structurationof the dataset.
(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
7M. Gay et al. / Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
figures. Two distinct manganese oxides (Ba rich ornot) were identified,
which means that prehistoric artists have used two distinct black pig-
ments in Font-de-Gaume (Fig. 8). This fact tends to indicate that differ-
ent hands at varying periods might have created the black figures.
Therefore there is potential to differentiate different creation phases
through pigment analyses.
From a methodological point of view, the most important point re-
vealed through the Font-de-Gaume case study is thesignificant impact
of the heterogeneouswall composition on the experimental results (Fig.
9). This fact demonstrates the need to choose properly the measure-
ment point to be analysed as a function of the thickness and the
amounts of pigment at the analysed spot in order to minimise the
weight of the substrate contribution in the detected spectrum. In that
way, the difficulties linked to the heterogeneity of the wall can be re-
duced substantially and can in favourable cases even be removed.
5. Discussion and conclusion
Through the study of differentkinds of rock art in three Palaeolithic
key caves by X-ray fluorescence analysis, the study presents and dis-
cusses a methodological framework and new analytical data in an ar-
chaeological perspective. Focused on the analytical complexity behind
in situ and non-invasive study of rock art, it guided our thinking about
the development of appropriate and effective setting up of adequate
quantitative procedures of data processing, in order to adjust the evalu-
ation procedures to the unique factors at each rock art site.
Challenges of this kind of study are the separation ofthe chemical in-
formation inherent to the pigment from that of the rock support and
that linked to various preservation states of the decorated wall. Thus,
the complexity of the evaluation of XRF spectra results from the fact
that paint layers are generally very thin and formed by scattered pig-
ment particles that do not cover the rock surface uniformly. The rock
support can additionally be very heterogeneous on a small scale. For
these reasons, the proportion of the physicochemical information spe-
cific to the substrate might be very high in the resulting paint spectrum.
The red or yellow rock art, realised using iron oxide pigments, pro-
vides an additional difficulty. Contrary to black manganese oxides,
which can be discriminated from the support due to some of their
main chemical constituents, iron oxides contain similar main constitu-
ents as the limestone rock. In the case of black pigments, it is relatively
straightforward to extract the characteristic elements and to semi-
quantify their concentration. In the case of red and yellow paints, the
similarity of the composition of the paint and its support raises signifi-
cant methodological problems with respect to the data processing. A
simple subtraction of the wall contribution is inappropriate due to the
varying contribution of the support in the spectra linked to non-repro-
ducible experimental conditions in karstic environments (especially
the non-planarity of the wall), which makes measurements incompara-
ble from one analysis point to another. More complex spectrum evalua-
tion procedures need to be adopted for red and yellow paints. These
limitations, different according to the study case, will induce varying
data processing times according to the necessary data evaluation meth-
odology, a semi-quantification based on fixing the CaO content or
Monte Carlo simulations.
The preservation state of the rock art (leaching of the pigment or de-
positionof other surface layers) is an important aspect to consider since
it impacts on the strategy choice and on the data processing time.
Finally, the heterogeneity of the substrate has to be considered too.
More measurement points are necessary when the substrate is more
heterogeneous to obtain representative analytical data. The analysis
and spectrum evaluation time will be consequently longer.
The consideration of all theseparameters before startinga study will
be useful and will help to estimate the scope of work and time (work in
the field and data evaluation), which will be required to study a rock art
site. Moreover, taking into account these considerations will enable de-
velopment of adequate methodologies. The data processing has to be
performed in accordance with the specificities of the studied site. We
are now capable of proposing an adequate strategy to study a cave
and its rock art in an appropriate and effective way.
Through the study of distinct Palaeolithic sites (Rouffignac, Font-de-
Gaume and LaGarma caves), three strategieswere developed and test-
ed to analyse rock art. The archaeological issue was to improve the gen-
eral knowledge and understanding of these ornamented sites without
destructively sampling the rock art. Indeed, the strong potential of
these adapted methods has been demonstrated to understand different
kinds of cave art and to give more detailed insights into the characteri-
sation of the paints and pigments used by the prehistoric artists and to
answer archaeological questions related to the organisation of the rep-
resentations, possible periods of ornamentation inside the cave, etc. If
addressed correctly taking into account the multiple factors influencing
Fig. 9. TheXRF spectra of the heterogeneoussubstrate acquired in a smallarea (bison 39) and thehigh impact of the wallheterogeneityon three very closeXRF analyses of thered pigment
(bison 4) (Font-de-Gaume cave). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
8M. Gay et al. / Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
the studies, much further information can be expected from in situ non-
invasive studies using quantitative X-ray fluorescence analyses of rock
art.
Acknowledgements
The authors also acknowledge the support of theLAMS for providing
the portable XRF spectrometer as well as the Region Ile de France that
provided financial support through the the DIM Analytics programme
to Ina Reiche that allowed funding of a PhD grant to Marine Gay atthe
doctoralschool (ED388) of UPMC. The following colleagues are thanked
through their supportto access the caves: Jean and Marie-Odile Plassard
and Frédéric Goursolle at Rouffignac cave, Georges Levy, Jean-Marie
Pelltant, Jean-Pierre Vanzo and their colleagues at Font-de-Gaume cave.
References
Arias, P., Ontañón, R., 2012. Chapter 8. La Garma (Spain): long-term human activity in a
karst system. In: Bergsvik, K.A., Skeates, R. (Eds.), Caves in Context: The Cultural Sig-
nificance of Caves and Rockshelters in Europe. Oxbow, Oxford, pp. 101–117.
Arias, P., González, S.C., Moure, A., Ontañón, R., 1996. L'art pariétal Paléolithique du
complexe archéologique de La Garma (Omoño, Cantabria, Espagne). Approche
préliminaire. International Newsletter On Rock Art. N.14, pp. 1–4.
Arias, P., González, S.C., Moure, A., Ontañón, R., 2000. La zona arqueológica de la Garma
(Cantabria): investigación, conservación y uso social. Trab. Prehist. 57 (2), 41–56.
Arias, P., Laval, E., Menu, M., González, S.C., Ontañón, R., 2011. Les colorants dans l'art pa-
riétal et mobilier paléolithique de La Garma (Cantabrie, Espagne). l'Anthropologie
115, 425–445.
Aujoulat, N., 2004. Lascaux. Le geste, l’espace et le temps. Édition du Seuil, Paris, p. 274.
Baffier, D., Girard, M., Menu, M., Vignaud, C., 1999. La couleur à la Grande Grotte d'Arcy-
sur-cure (Yonne). l'Anthropologie 103, 1–21.
Ballet, O., Bocquet, A., Bouchez, R., Coey, J.M.D., Cornu, A., 1979. Étude technique des
poudres colorées de Lascaux. In: Leroi-Gourhan, A., Allain, J. (Eds.), Lascaux
InconnuXIIème Supplément à Gallia Préhistoire. Éditions du CNRS, Paris,
pp. 171–174.
Barrière, Cl, 1982. L’art pariétal de Rouffignac. Picard, Fondation Singer-Polignac, Paris,
p. 208.
Bastian, F., Jurado, V., Novakova, A., Alabouvette, C., Saiz-Jimenez, C., 2010. The microbiol-
ogy of Lascaux cave. Microbiology 156, 644–652.
Beck, L., Rousselière, H., Castaing, J., Duran, A., Lebon, M., Lahlil, S., Plassard, F., 2012. An-
alyse in situ des dessins préhistoriques de la grotte de Rouffignac par fluorescence
X et diffraction X portable. ArchéoSciences 36, 139–152.
Beck, L., Rousselière, H., Castaing, J., Duran, A., Lebon, M., Moignard, B., Plassard, F., 2014.
First use of portable system coupling X-ray diffraction and X-ray fluorescence for in
situ analysis of prehistoric rock art. Talanta 129, 459–464.
Cabrera-Garrido, J.M., 1978. Les matériaux des peintures dela grotte d'Altamira.Actes de
la 5
ème
réunion internationale de l'ICOM. Zagreb, pp. 1–9.
Capitan, L., Breuil, H., Peyrony, D., 1910. La caverne de Font-de-Game aux Eyzies (Dor-
dogne). Imprimerie Veuve A, Chêne, Monaco 271 pp.
Chalmin,E., Reiche, I., 2013. Synchrotron X-ray microanalysis andimaging of synthetic bi-
ological calcium carbonate in comparison with archaeological samples originating
from the large cave of Arcy-sur-Cure (28000-24500 BP, Yonne, France). Microsc.
Microanal. 19 (06), 1523–1534.
Chalmin, E., Menu, M., Altuna, J., 2002. Les matières picturales de la grotte d'Ekain (Pays
Basque). Munibe 54, 35–51.
Chalmin, E., Menu, M., Vignaud, C., 2003. Analysis of rock art paintings and technology of
Palaeolithic painters. Meas. Sci. Technol. 14, 1590–1597.
Chalmin, E., Sansot, E., Orial, G., Bousta, F., Reiche, I., 2008. Microanalysis and synthesis of
calcite.Growth mechanisms on prehistoric paintings in the LargeCave, Arcy-sur-Cure
(Yonne, France). X-Ray Spectrom. 37, 424–434.
Cleyet-Merle, J.-J., 2014. La Grotte de Font-de-Gaume. Paris. Edition du patrimoine, Centre
des Monuments Historique 64 pp.
Clottes, J., Menu, M., Walter, P., 1990. La préparation des peintures magdaléniennes des
cavernes ariégeoises. Bulletin de la Société préhistorique française 87(N.6), 170–192.
de Sanoit, J., Chambellan, D., Plassard, F., 2005. Caractérisation in situ du pigment noir de
quelques œuvres pariétales de la Grotte de Rouffignac à l'aise d'un système portable
d'analyse par fluorescence X (XRF). ArchéoSciences 29, 61–68.
Ferrier, C., Debard, E., Kervazo, B., Aujoulat, N., Baffier, D., Denis, A., Feruglio, V., Fritz, C.,
Gély, B., Geneste, J.-M., Konik, S., Lacanette, D., Lastennet, R., Maksud, F., Malaurent,
P., Plassard, F., Tosello, G., 2012. Approche taphonomique des parois des grottes
ornées dir. In: J., C. (Ed.), L'art pléistocène dans le monde, Actes du Congrès IFRAO,
Tarascon-sur-Ariège, septembre 2010, Symposium “Datation et taphonomie de l'art
pléistocène”, No spécial de Préhistoire, Art et Sociétés, Bulletin de la Société
Préhistorique Ariège-Pyrénées, LXV–LXVI, 2010–2011, CD, pp. 1071–1093.
Ferrier, C., Aujoulat, N., Denis, A., Kervazo, B., Konik, S., Lacanette, D., Large, D., Lastennet,
R., Malaurent, P., Paillet, P., 2014. Une grotte-laboratoire pour l'étude taphonomique
des parois des grottes ornées: la grotte de Leye à Marquay (Dordogne, France) dir.
Les arts de la Préhistoire: micro-analyses, mises en contextes et conservation. Actes
du colloque “Micro-analyses et datations de l'art préhistorique dans son contexte
archéologique”, MADAPCA, Paris, No spécial PALEO, pp. 331–338.
Gay, M., Alfeld, M., Menu, M., Laval, E., Arias, P., Ontañón, R., Reiche, I., 2015. Palaeolithic
paint palettes used at La Garma Cave (Cantabria, Spain) investigated by means of
combined in situ and synchrotron X-ray analytical methods. J. Anal. At. Spectrom.
30, 767–776.
Gay, M., Müller, K., Plassard, F., Reiche, I., 2016. Significant Contribution of in situ X-ray
Fluorescence Analysis to the Reading of the Great Ceiling Composition of Rouffignac
Cave (Dordogne, France) forthcoming.
Hocquet, F.-P., Garnir, H.-P.,Marchal, A., Clar, M.,Oger, C., Strivay, D., 2008. A remote con-
trolled XRF system for field analysis of cultural heritage objects. X-Ray Spectrom. 37,
304–308.
Hœrlé, S., Konik, S., Chalmin, E., 2011. Les vermiculations de la grotte de Lascaux: des
matériaux mobilisables par microanalyses physico-chimiques. Karstologia 58, 29–40.
Jenkins, R., Gould, R.W., Gedcke, D., 1995. Quantitative X-ray Spectrometry. second ed.
Marcel Dekker, Inc. 491 pp.
Kervazo, B., Feruglio, V., Baffier, D., Debard, E., Ferrier, C., Perroux, A.-S., Aujoulat, N.,
Delannoy, J.-J., Yves, P., 2010. Parois et art pariétal: approche taphonomique.
L'exemple de la grotte Chauvet-Pont d'Arc (Ardèche), pp. 43–52 PALEO, supplément
no 3.
Konik, S., Lafon-Pham,D., Riss, J., Aujoulat, N., Ferrier, C., Kervazo, B., Plassard, F., Reiche, I.,
2014. Étude des vermiculations par caractérisation morphologique, chromatique et
chimique. L'exemple des grottes de Rouffignac et de Font-de-Gaume (Dordogne,
France) dir. In: Paillet, P. (Ed.), Les arts de la Préhistoire: micro-analyses, mises en
contextes et conservation. Actes du colloque “Micro-analyses et datations de l'art
préhistorique dans son contexte archéologique”,MADAPCA,Paris,Nospécial
PALEO, pp. 311–321.
Lacanette, D., Large, D., Ferrier,C., Aujoulat, N., Bastian, F., Denis, A., Jurado, V., Kervazo, B.,
Konik, S., Lastennet, R., Malaurent, P., Saiz-Jimenez, C., 2013. A laboratory cave for the
study of wall degradation in rock art caves: an implementation in the Vézère area.
J. Archaeol. Sci. 40, 894–903.
Lahlil, S., Lebon, M., Beck, L., Rousselière, H., Vignaud, C., Reiche, I., Menu, M., Paillet, P.,
Plassard, F., 2012. The first in situ micro-Raman spectroscopic analysis of prehistoric
cave art of Rouffignac St-Cernin, France. J. Raman Spectrosc. 43, 1637–1643.
Lorblanchet, M., Labeau, M., Vernet, J.-L., Fitte, P., 1990. Etude des pigments des grottes
ornées paléolithiques du Quercy. Bulletin de la société des études littéraires,
scientifiques et artistiques du Lot 2, 93–143.
Mantler, M., Schreiner, M., 2000. X-ray fluorescence spectrometry in artand archaeology.
X-Ray Spectrom. 29, 3–17.
Martin-Sanchez, P.M., Novakova, A., Bastian, F., Alabouvette, C., Saiz-Jimenez, C., 2012.
Two new species of the genus Ochroconis,O. lascauxensis and O. anomala isolated
from black stains in LascauxCave, France. Fungal Biol. 116, 574–589.
Menu, M., Walter, P., 1992. Prehistoric cave painting PIXE analysis for the identification of
paint “pots”. Nucl. Instrum. Methods Phys. Res. 64, 547–552.
Nuevo, M.J., Martín, S.A., Oliveira, C., Oliveira de, J., 2012. In situ energy dispersive X-ray
fluorescence analysis of rock art pigments from the ‘Abrigo dos Gaivões’and Igreja
dos Mouros' caves (Portugal). X-Ray Spectrom. 41, 1–5.
Olivares, M., Castro, K., Corchón, M.-S., Gárate, D., Murelaga,X., Sarmiento, A., Etxebarria,
N., 2013. Non-invasive portable instrumentation to study Palaeolithic rockpaintings:
the case of La Peña Cave in San Roman de Candamo (Asturias, Spain). J. Archaeol. Sci.
40, 1354–1360.
Ontañón, R., 2003. Sols et structures d'habitat du Paléolithique supérieur, nouvelles
donnéesdepuis les Cantabres:la Galerie Inférieure de La Garma (Cantabrie,Espagne).
l'Anthropologie 107, 333–363.
Pepe, C., Clottes, J., Menu, M., Walter, P., 1991. Le liant des peintures préhistoriques
ariégeoises. Comptes Rendus de l'Académie des Sciences de Paris 312, 926–934.
Plassard, F., 2005. Les grottes ornées de Combarelles, Font-de-Gaume, Bernifal et
Rouffignac. Contexte archéologique, thèmes et style des représentations. Thèse de
Doctorat de l'Université Bordeaux 1 413 pp.
Plassard,F., Plassard, J., 2000. Figures inédites de la grotte de Rouffignac. Gallia Préhist oire
42, 85–106.
Plassard, F., Plassard, J., 2016. Le grand plafond de Rouffignac. De nouveaux indices sur
l’organisation des imagesPaléo Hors-série. Hommage à Norbert Aujoulat.
Roldán, C., Murcia-Mascarós, S., Ferrero, J., Villaverde, V., López, E., Domingo, I.,Martínez,
R., Guillem, P.M., 2010. Application of field portable EDXRF spectrometry to analysis
of pigments of Levantine rock art. X-Ray Spectrom. 39, 243–250.
Schoonjans, T., Vincze, L., Solé, V.A., del Rio, M.S., Brondeel, P., Silversmit, G., Appel, K.,
Ferrero, C., 2012. A general Monte Carlo simulation of energy dispersive X-ray fluo-
rescence spectrometers –part 5. Polarized radiation, stratified samples, cascade ef-
fects, M-lines. Spectrochim. Acta B 70, 10–23.
Shoeller, H., 1967. Conduite de l'étude hydrogéologique et climatologique des grottes
descendantes. Spélunca mémoire 5, 76–93.
Solé, V.A., Papillon, E., Cotte, M., Walter, P., Susini, J., 2007. A multiplatform code for the
analysis of energy-dispersive X-ray fluorescence spectra. Spectrochim. Acta B 62,
63–68.
Vandiver, P., 1983. Palaeolithic pigments and processing. Master Science Thesis, Depart-
ment of Material Science and Engineering, MIT university (USA).
Vignaud, C., Salomon, H., Chalmin, E., Geneste, J.-M., Menu, M., 2006. Le groupe des “bi-
sons adossés”de Lascaux. Etude de la technique de l'artiste par analyse des pigments.
l'Anthropologie 110, 482–499.
9M. Gay et al. / Journal of Archaeological Science: Reports xxx (2016) xxx–xxx
Please cite this article as: Gay, M., et al., Efficient quantification procedures for data evaluation of portable X-ray fluorescence –Potential
improvements for Palaeolithic..., Journal of Archaeological Science: Reports (2016), http://dx.doi.org/10.1016/j.jasrep.2016.06.008
