Vol.:(0123456789)

Archaeological and Anthropological Sciences (2024) 16:155 
https://doi.org/10.1007/s12520-024-02057-7

RESEARCH

Testing tools: an experimental investigation into technical 
and economic aspects of Levantine rock art production

Neemias Santos da Rosa1,2,3  · Danae Fiore4  · Ramon Viñas5 

Received: 13 June 2024 / Accepted: 13 August 2024 / Published online: 2 September 2024 
© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024

Abstract
For over a century, scholars have presented several proposals regarding the paint application tools used to create Levantine 
rock art. However, such proposals have largely rested on theoretical assumptions and inductive inferences that have not been 
rigorously tested. In this experimental investigation, we compiled these hypotheses and crafted 60 tools that were tested 
through systematic experiments to assess their performance and technical affordances for the creation of images with techno-
visual features equivalent to those of the Levantine paintings. These experiments allowed us to obtain analytical parameters 
with independent validation, which were used as diagnostic criteria to analyse the rock art from nine sites located in eastern 
Spain. Results reveal that out of 60 tools, only 10 –five hair brushes, three plant brushes, and two feather brushes–afforded 
the production of images with techno-visual features highly similar to those of the archaeological paintings. Subsequently, we 
employed these tools to paint 10 full-size replicas of Levantine motifs, aiming to explore their potential role in the toolkit of 
prehistoric painters. Drawing on the data from both the systematic and the replicative experiments, and taking into account 
the level of labour investment required by the technical operations that permeate the production of the paint application 
tools, we suggest that Levantine technology was not based on an economic logic aimed at minimizing the costs associated 
with producing the images, but rather on a perspective where the priority was creating paintings with specific techno-visual 
qualities that reinforced the information transmitted through the graphic code.

Keywords Levantine rock art · Paint application tools · Experimental Archaeology · Technology

Introduction

Since the latter decades of the nineteenth century, archae-
ologists have been fascinated and intrigued by the paint 
application tools used in the production of prehistoric 
images. Marcelino Sanz de Sautuola (1880: 16), for exam-
ple, emphasised the precision of the tools employed to 
create the Palaeolithic bisons of Altamira, as well as the 

outstanding technical skill of the Magdalenian painters. 
Although this and other initial remarks were mainly based 
on general observations of the rock art motifs, the develop-
ment of experimental archaeology from the 1970s onwards 
allowed some scholars to investigate this issue from a deeper 
empirical standpoint (e.g. Couraud 1976; Pérez-Seoane 
1988; Lorblanchet 1991), yielding important insights into 
the tools and techniques employed in the production of the 

 * Neemias Santos da Rosa 
 neemias.sdarosa@gmail.com

 Danae Fiore 
 danae_fiore@yahoo.es

 Ramon Viñas 
 rvinas@iphes.cat

1 UMR 5199 PACEA – University of Bordeaux, Bâtiment 
B2 – Allée Geoffroy Saint Hilaire – CS 50023, 
33615 Pessac Cedex, France

2 School of Geography, Archaeology and Environmental 
Studies, Rock Art Research Institute, University 

of Witwatersrand, Private Bag X3, Wits 2050, Johannesburg, 
South Africa

3 Departament d’Història i Arqueologia, Facultat de Geografia 
I Història, Institut d’Arqueologia, Universitat de Barcelona, 
Carrer de Montalegre 6, 08001 Barcelona, Spain

4 National Scientific and Technical Research Council – 
CONICET, University of Buenos Aires, Asociación de 
Investigaciones Antropológicas, Bartolomé Mitre, 1131 7 
“G” CP (1036), Buenos Aires, Argentina

5 Catalan Institute of Human Paleoecology and Social 
Evolution (IPHES), Rovira i Virgili University, Campus 
Sescelades 4, 43007 Tarragona, Spain

http://crossmark.crossref.org/dialog/?doi=10.1007/s12520-024-02057-7&domain=pdf
http://orcid.org/0000-0001-8800-146X
http://orcid.org/0000-0003-1672-3070
http://orcid.org/0000-0003-2568-4325


 Archaeological and Anthropological Sciences (2024) 16:155155 Page 2 of 22

paintings. However, in Europe, the research on the subject 
has remained restricted to an exploratory level, and there is 
still a lack of systematic experimental studies concerning 
the paint application tools used in Palaeolithic and post-
Palaeolithic periods.

Thus, in this article, we present an experimental research 
on the paint application tools used in the production of 
Levantine rock art (LRA), the visual expression of a post-
Palaeolithic cultural tradition that emerged and flourished 
in eastern Spain. As part of an extensive experimental pro-
gramme aimed at reconstructing the LRA chaîne opératoire 
(Santos da Rosa 2019a; Santos da Rosa et al. 2023), we 
compiled all hypotheses proposed over more than a century 
of studies and crafted 60 experimental tools, which were 
systematically tested to evaluate their performance and to 
assess the level of correlation between the morphology of 
their active zones and the morphology of the experimen-
tal marks produced through their use. These experiments 
allowed us to obtain analytical parameters with independ-
ent validation that were used as diagnostic criteria to study 
the LRA of nine sites and to identify which tools produce 
experimental rock art marks with techno-visual features 
most similar to Levantine paintings. In a subsequent phase 
of this study, we employed the tools classified as the most 
analogous to the prehistoric ones to create 10 full-size repli-
cas of LRA motifs, aiming to investigate their potential role 
in the toolkit of Levantine painters. Drawing on the data 
from both the systematic and the replicative experiments, 
and taking into account the labour investment required by 
the technical operations that permeate the production of the 
paint application tools, we formulate a series of empirically 
supported inferences concerning technical and economic 
aspects of LRA technology.

The Levantine paint application tools: 
over a century of controversy

Identified in nearly a thousand shelters distributed across 
Mediterranean Spain, LRA is renowned for its depictions 
of hunting, food gathering, dances and conflicts, made 
through the use of thin lines and different surface treatments 
in red, black, and white (López-Montalvo 2018; Rubio et al. 
2019; Santos da Rosa et al. 2021a, b; Fernández Macías 
et al. 2023). The chronology of this artistic tradition has 
been a matter of debate since its academic discovery in the 
early twentieth century, and currently, researchers still dis-
cuss whether the origins of LRA date back to the Epipal-
aeolithic/Mesolithic (Viñas et al. 2016; Ruiz et al. 2022) 
or the Neolithic (Hernández Pérez 2012; Villaverde et al. 
2012). In this context, although the technological aspects of 
LRA have been treated as a marginal topic throughout the 
history of research, scholars always exhibited a particular 

interest in the continuous and precise lines used to cre-
ate the images (Fig. 1), and occasionally have engaged in 
meticulous examinations of these elements (Porcar 1943). 
Such interest has led to several proposals concerning the 
paint application tools used by Levantine painters, which 
according to different researchers, could be hair brushes 
(animal or human hair inserted into a haft), plant brushes (a 
branch whose end is crushed and cut to obtain bristles with 
specific morphologies), plant fragments (branches or roots 
with a whittled point), feathers (used in their natural state), 
or feather brushes (an entire feather of a feather fragment 
inserted into a haft) (Table 1; for a detailed synthesis, see 
Santos da Rosa 2019b).

To succinctly illustrate this controversy, certain proposals 
warrant closer scrutiny. Grimal and Alonso (2001: 95), for 
instance, contended that the production of hair brushes is a 
laborious and time-consuming process that would be beyond 
the technical capabilities of prehistoric populations. Hence, 
based on non-systematic experiments –mentioned by Grimal 
(1991: 52), but whose data have never been published– these 
authors proposed that feathers were the only paint applica-
tion tool used in the production of LRA, which led them 
to coin the popular term “trazo de pluma levantino” and to 
defend the idea that the lines observed in the archaeological 
paintings could not have been produced with any other type 
of tool (Grimal and Alonso 2001: 96). On the other hand, 
after producing replicas of two archers and one quadruped, 
Bea (2007: 8) considered plant brushes as effective paint 
application tools and pointed out that they could have been 
used in the same way as feathers or even hair brushes in 
the creation of Levantine images (Bea 2007:7–8). Adopting 
a similar perspective, Ruiz (2012) conducted a replicative 
non-systematic experiment with feathers and argued that this 
tool would have been widely employed in the production of 
LRA. According to this author, the diagnostic information 
could be found on the initial and final tips of the lines, areas 
in which the traces left by the feather would be more dis-
cernible. In this regard, rounded and elliptical shapes would 
frequently appear in the initial tips, while pointed shapes 
would be more common in their final tips (Ruiz 2012: 335). 
Furthermore, analysing photographs of Levantine motifs 
from the Sierra de las Cuerdas (Cuenca), Ruiz observed –in 
addition to traces presumably made with feathers– the pres-
ence of thin parallel lines (~ 0.2 mm wide) that he inter-
preted as marks left by a hair brush (Ruiz 2012: 337).

Building on this background, it is possible to note that the 
aforementioned proposals are mainly based on theoretical 
assumptions, superficial observations, and the results of non-
systematic replicative experiments in which the causal rela-
tionship between different variables was not properly taken 
into account. Concerning the latter, there is a lack of precise 
information about the experimental protocols employed in 
these studies, as well as the methods researchers used to 



Archaeological and Anthropological Sciences (2024) 16:155 Page 3 of 22 155

address the technical variability in tool performance and 
results potentially influenced by chance. This implies that, 
even after more than a century, the hypotheses regarding the 
paint application tools used in the production of LRA still 
need to be rigorously tested.

Theoretical framework: performances, 
affordances and labour investment 
in producing and using paint application 
tools

Over the past five decades, experimental archaeology has 
proved to be an invaluable tool in testing hypotheses con-
cerning the production of rock art paintings (e.g., Couraud 

1988; Paunero 1992; Blanco and Barreto 2016; d’Errico 
et al. 2016; López-Montalvo 2017; Landino et al. 2023; 
Ozán et al. 2023; Santos da Rosa et al. 2023) and engrav-
ings (e.g. Bard and Busby 1974; Bednarik 1998; Keyser and 
Rabiega 1999; Álvarez and Fiore 1995; Álvarez et al. 2001; 
Santos da Rosa 2012; Santos da Rosa et al. 2014; Vergara 
and Troncoso 2015; Zotkina and Davydov 2022) in sites 
from diverse chronological periods and geographic areas.

This experimental approach is particularly significant in 
the context of a theoretical and methodological framework 
that aims to explore rock art’s materiality through the study 
of technological processes and economic factors that per-
meate the creation of the images (Fiore 2007, 2009, 2018; 
Santos da Rosa 2019c). According to this perspective, the 
materiality of rock art is not limited to the components used 

Fig. 1  Example of LRA and 
its technical aspects: a and e) 
Cova dels Rossegadors (Photo: 
A. Rubio); b) Cova del Civil 
(Photo: A. Rubio); c and d) 
Cova dels Cavalls (Photo: N. 
Santos da Rosa); e) Abric del 
Mas d’en Josep (Photo: N. 
Santos da Rosa). All the motifs 
shown in this figure come from 
the sites that comprise our 
archaeological study sample



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 4 of 22

in its production but also encompasses the performance and 
affordances of these elements, which are activated through 
human agency. Thus, in this article, we begin by exploring 
the concept of performance, here defined as the efficiency of 
the paint application tools in producing images with specific 
technical features. Subsequently, we employ the concept of 
techno-visual affordances (sensu Fiore 2020), which in this 
study pertains to the production and perception of the visual 
qualities of rock art motifs painted on a rocky surface.

In this context, we consider that the visual qualities of 
a rock art painting result from the use of tools, pigments, 
supports, and techniques, and from the combination of their 
performances and affordances via human labour to create 
the forms, colours, and textures of the artwork (Fiore 2020). 
Afforded by the technical means employed to produce the 
image, these techno-visual qualities emerge from the paint-
ing’s materiality to evoke specific visual perceptions in those 
who observe the motifs and to complement the information 
transmitted through the graphic code. In the words of British 
anthropologist Alfred Gell, this is where the “enchantment of 
technology” resides: it allows people to “make what is out of 
what is not” (Gell 1994).

Furthermore, given that the tool’s performance and 
techno-visual affordances are activated through a work pro-
cess, we also adopt the concept of labour investment (sensu 
Fiore 2007) in an attempt to qualitatively assess the role 
of aspects such as time, energy, and skill in the production 
of paint application tools and their use in creating Levan-
tine images. Shedding light on the making process within 
and beyond the practical operations, the use of this con-
cept provides us with a better understanding of the cost and 
complexity of certain activities carried out by prehistoric 
painters, and contributes to deconstructing misconceptions 

about the technical and economic dimensions of technologi-
cal processes developed in the past.

Materials and methods

In order to systematically study the performances and 
affordances of the experimental paint application tools, we 
designed an experimental protocol comprising seven stages: 
1) procurement of raw materials in the study area; 2) pro-
duction of experimental paint application tools; 3) system-
atic testing of the experimental tools; 4) evaluation of the 
experimental tools’ efficiency; 5) analysis of the correlation 
between tools’ active zones and the morphology of the line 
tips; 6) application of the experimental criteria to a LRA 
sample; and 7) creation of full-size replicas of Levantine 
motifs.

Procurement of raw materials

To craft the experimental tools, we selected raw materials 
that were available in the study area during the Mesolithic 
and Neolithic periods. To produce the hair brushes, we 
gathered guard hairs of bull (Bos taurus), goat (Capra hir-
cus), deer (Cervus elaphus), rabbit (Oryctolagus cunicu-
lus), boar (Sus scrofa), and fox (Vulpes vulpes), as well as 
human scalp hair. This choice is justified by the presence 
of these animals in the zooarchaeological record of Cova 
Fosca (Olària 1988), a Mesolithic/Neolithic site located 
at the heart of the study area and just 1 km from Cova 
Remígia and Cingle de la Mola Remígia. Furthermore, 

Table 1  Proposals concerning the paint application tools used by the Levantine painters

Paint application tool Bibliographic reference Criteria

Hair brushes Hernández-Pacheco (1918: 67); Obermaier and Breuil (1927: 514); Obermaier (1938: 115); 
Almagro Basch (1949: 103); Beltrán Martínez (1968: 27); Utrilla (2000: 34); Sanchidrián 
Torti (2001: 382); Domingo Sanz (2012: 121); Viñas (2014: 690)

theoretical

Bea (2007: 7–8) non-systematic experiments
Gavira et al. (2008: 54); Ruiz (2012: 338) analysis of photographs

Plant brushes Domingo Sanz (2012: 121); Viñas (2014: 690) theoretical
Bea (2007: 7–8) non-systematic experiments

Plant fragments Obermaier and Breuil (1927: 514); Utrilla (2000: 34) theoretical
Bea (2007: 7–8) non-systematic experiments

Feathers Beltrán (1968: 27); Alonso and Grimal (1989: 18); Utrilla (2000: 34); Sanchidrián Torti 
(2001: 382)

theoretical

Porcar et al. (1935: 66); Grimal (1991: 52); Grimal and  Alonso (2001: 95); Bea (2007: 7); 
Ruiz (2012: 333)

non-systematic experiments

Gavira et al. (2008: 54); Ruiz (2012: 333) analysis of photographs
Feather brushes Obermaier and Breuil (1927: 514); Ripoll (1990: 72); Viñas (2014: 690) theoretical

Obermaier (1938: 115–116) non-systematic experiments



Archaeological and Anthropological Sciences (2024) 16:155 Page 5 of 22 155

this selection allowed us to examine the influence of hair 
texture and flexibility on the tool performance, given that 
bull, rabbit, fox and human hairs tend to be smooth and 
flexible, while goat, deer and boar hairs are coarse and 
more rigid.

To create paint application tools from plants, we col-
lected fresh branches of hazel (Corylus avellana), lentisk 
(Pistacia lentiscus), lesser bulrush (Typha angustifolia), 
rush (Juncus subulatus), kermes oak (Quercus coccifera), 
and Mediterranean spurge (Euphorbia characias), as well 
as a dry branch of reed (Phragmites communis) and a root 
of mallow (Malva silvestris). To carry out this procedure, 
we relied on information from the archaeological context 
of the study area (Antolín et al. 2010) and the collabora-
tion of Prof. Francesc Burjachs (IPHES – Rovira i Virgili 
University), a leading expert on the paleoenvironment of 
Mediterranean Spain. Employing such resources, it was 
possible to test whether tools crafted from woody (hazel, 
lentisk, kermes oak, and dry reed), fibrous (rush, Mediter-
ranean spurge, and mallow) and spongy plants (lesser bul-
rush) exhibit significant variations in terms of efficiency.

We also gathered flight feathers and tail feathers of 
crow (Corvus corax), rock dove (Columba livia), and 
Spanish sparrow (Passer hispaniolensis), which allowed 
us to test the performance of feathers from birds of large, 
medium, and small sizes. In this sense, it should be noted 
that remains of Corvus sp. and Columba sp. were iden-
tified in the aforementioned zooarchaeological record of 
Cova Fosca (Olària 1988), and that the Spanish sparrow 
is an autochthonous bird of the Mediterranean region that 
was present in the study area in prehistoric times (Ericson 
et al. 1997).

Finally, we collected dry reed branches to be used as 
hafts, Aleppo pine resin (Pinus halepensis) to be employed 
as an adhesive, grasses to tie the hairs used in the brushes 
(Brachypodium phoenicoides), flint flakes for cutting and 
shaping some of the raw materials, hammerstones to crush 
the active zone of plant brushes and to grind mineral pig-
ments, red ochre and binders (water and honey) to pro-
duce red paint, as well as natural limestone slabs with 
geological and topographic features equivalent to the walls 
of Levantine shelters. As for the latter, special attention 
was given to selecting flat supports without pronounced 
concavities or convexities and as similar as possible to 
each other, so that any micro-topographic differences 
between them would not generate biases in the experimen-
tal results. This implies that in all the experiments con-
ducted, the interaction between paint and support in terms 
of fixation and absorption remained constant, while the 
interaction between tool and support varied based on the 
level of adaptability to the rocky surface exhibited by each 
tool – such variations were part of the results observed and 
recorded in our experimental protocol.

Production of experimental paint application tools

Employing the raw materials described in the previous sec-
tion, we crafted 21 hair brushes, 12 plant brushes, 8 plant 
fragments, and 6 feather brushes. The experimental set was 
completed with 6 feathers and 7 feather fragments, culmi-
nating in a total of 60 paint application tools (Fig. 2; for 
details on the production of each tool, see Supplementary 
Material – Sect. “1”). To test a wider range of technical pos-
sibilities, the bristles of hair and plant brushes were shaped 
into three distinct active zone morphologies: pointed (p.a.z), 
convex (c.a.z), and flat (f.a.z). Conversely, feather brushes 
and feather fragments were crafted only with pointed and 
convex active zones, as the rachis precludes the use of a 
flat end.

Systematic testing of the experimental paint 
application tools

To evaluate the efficiency of the experimental tools, we 
painted with each tool five straight lines (6 cm long) and five 
curved lines (8 cm long) on vertically positioned limestone 
slabs, which resulted in a total of 600 lines (300 straight and 
300 curved) (Fig. 3; see all lines in Santos da Rosa 2019a). 
The production of such experimental marks was carried out 
through single and continuous gestures from top to bottom, 
and aimed at creating a set of highly comparable images to 
assess the efficiency of the tools. Also, this procedure ena-
bled the application of the resultant experimental parameters 
to the archaeological sample, as both straight and curved 
lines feature in all Levantine images. In this sense, the crea-
tion of five straight lines and five curved lines with each 
tool was intended to detect potential variability in their 
technical performance and to minimise the occurrence of 
results attributable to chance. In all cases, the experimental 
marks were produced using a red paint composed of 1 g of 
red ochre, plus 1 ml of water and 1 g of honey, a pictorial 
recipe experimentally tested and classified as highly similar 
to those used in the creation of LRA (Santos da Rosa et al. 
2023). After their production, the experimental marks were 
stored vertically and left to dry for six months in a sheltered 
open-air space. This drying process was crucial to achieving 
a realistic view of the morphology and structural aspect of 
the lines, since the paint can exhibit slight shrinkage after 
being applied to the support due to evaporation of the water 
used as one of the binders.

In this context, it is noteworthy that to ensure the mor-
phology of the experimental marks was not biased by the 
technical skill of the experimenter (NSR; 14 years of prac-
tical experience in the experimental study of rock art paint-
ings and engravings), we conducted a pilot experiment 
where all tools were also used by one of the co-authors of 
this article (RV), who is not only an archaeologist but also 



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 6 of 22

a professional painter with more than 50 years of experi-
ence in producing rock art replicas. Therefore, the official 
experiments were conducted only when the lines painted 
by both the experimenter and the professional painter 
exhibited identical morphologies, reflecting the maximum 
performance of the paint application tools.

Evaluation of experimental tools' efficiency

To assess the efficiency of the paint application tools, we 
analysed the causal relationship between three independent 
variables –tool category, tool raw material and morphol-
ogy of the active zone– and three dependent variables: a) 

Fig. 2  The 60 experimental 
paint application tools produced 
and used in the present study



Archaeological and Anthropological Sciences (2024) 16:155 Page 7 of 22 155

precision –the tool’s ability to produce experimental marks 
with linear lateral edges; b) adaptability to the support sur-
face –the tool’s ability to adapt to the irregularities of the 
support and produce a uniform paint layer; and c) paint load 
capacity –the tool’s ability to retain paint on its active zone 
and deposit it on the support allowing for the creation of 
continuous lines. The performance of each tool concerning 
these attributes was scored on a scale from 0 (poorest per-
formance) to 10 (optimal performance) (see Supplementary 
Material – Sects. “2, 3, 4 and 5”). Subsequently, the scores 
obtained for each dependent variable were used to calculate 
the Efficiency Index (EI) of each experimental tool accord-
ing to the following weighted average formula:

where:

• Ei represents the EI of the experimental tool i.
• WP is the weight attributed to the precision.
• Pi is the precision score of the tool i.
• WA is the weight attributed to the adaptability to the 

support surface.
• Ai is the adaptability to the support surface score of the 

tool i.
• WPL is the weight attributed to the paint load capacity.
• PLi is the paint load capacity score of the tool i.

Each weight reflects the relative importance of the cor-
responding dependent variable in determining the overall 
efficiency of the tool. In this regard, given that precision 
is paramount for producing well-defined lines and painting 

Ei = (WP ⋅ Pi) + (WA ⋅ Ai) + (WPL ⋅ PLi)

small details, it was assigned a weight of 0.5. This higher 
weight, compared to the other two technical features, is 
justified by the fact that issues related to the tool's inherent 
precision cannot be compensated by the technical skill of 
the painter or by the repeated execution of the technical 
operation, since attempts to retouch parts of the line’s lat-
eral edges to correct errors tend to alter its thickness, mak-
ing it necessary to repaint the entire element to maintain 
the correct proportions. Adaptability, in turn, is directly 
related to the appearance of the paint layer and the inter-
nal structure of the lines. Despite its significance, the low 
adaptability of a tool can be compensated by repainting 
flaws in the structure of the painted elements. However, 
such correction demands a very high level of technical 
skill and labour investment and tends to be applicable only 
to lines that are 2 mm wide, or thicker. For this reason, a 
weight of 0.3 has been attributed to this variable. Finally, 
the paint load capacity determines how much a tool can be 
used before it runs out of paint in its active zone. A subop-
timal performance in this aspect can be mitigated by the 
successive repetition of the technical operation, namely, 
the creation of several short linear strokes to form a longer 
continuous line. Given that this process does not require 
a high level of technical expertise or significant labour 
investment, a weight of 0.2 has been assigned to this tech-
nical feature. Thus, taking into account these different 
weights, the formula for calculating the EI was devised 
as follows:

Assessment of the correlation 
between the morphology of the tool's active zone 
and the morphology of the line tips

To test the assumption that the morphology of the line tips 
tends to reflect the morphology of the tool’s active zone, 
we examined 1200 tips of the 600 straight and curved lines 
and classified them as pointed, convex, flat or irregular. The 
data derived from this examination were used to ascertain, 
for each tool, a Correlation Index (CI) that spans from 0 to 
5 and reflects the total number of initial or final tips whose 
morphology corresponds to that of the tool active zone (see 
Supplementary Material – Section “6”). Based on this clas-
sification, the level of correlation was rated as high (CI: 5), 
medium (CI: 3 – 4), low (CI: 1 – 2), or non-existent (CI: 
0). Subsequently, we used these data to determine how fre-
quently these different levels of correlation manifest in each 
tool category and in relation to the different types of active 
zones.

Ei = (0.5 ⋅ Pi) + (0.3 ⋅ Ai) + (0.2 ⋅ PLi)

Fig. 3  Systematic testing of experimental paint application tools



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 8 of 22

Application of the experimental criteria to the LRA 
sample

To identify the paint application tools potentially used in 
the production of the Levantine paintings, we applied the 
diagnostic criteria obtained through the development of 
the experimental protocol to an archaeological sample of 
nine sites located in the Maestrazgo region and surround-
ings (Fig. 4; see Supplementary Material – Sect. “7”). This 
selection was based on the fact that: the nine sites have been 
extensively studied, and there is a broad range of data about 
their paintings; b) the approximately 2.500 images docu-
mented in these shelters form a representative sample of 
the LRA, reflecting its technical, typological and thematic 
diversity; and c) several authors argue that the rock art repre-
sented in these sites is the result of a common technological 
tradition, socially transmitted over many generations (Ober-
maier 1938; Porcar 1945; Viñas 1982; Domingo Sanz 2012).

This application of the experimental criteria to the LRA 
sample implied a detailed comparison between each of the 
600 experimental lines and the lines present in about 1,000 
well-preserved rock art motifs from the selected sites (see Sup-
plementary Material – Sect. “8”). The infeasibility of trans-
porting dozens of heavy limestone slabs to the field for in-situ 
comparison led us to carry out this procedure through the use 
of high-resolution photographs. Hence, the lateral edges, the 
internal structure, and the initial and final tips of the experi-
mental lines were compared to those of the lines used in out-
lining the archaeological images and depicting details such as 

bows, arrows, headdresses, body ornaments, and anatomical 
features in human and animal figure (Fig. 5), which gener-
ally exhibit the following characteristics: a) lateral edges that 
are either completely linear or slightly wavy in some of their 
sections, which never appear irregular or fragmented (except 
for taphonomic degradation) and never exhibit smudging or 
fine lines adjacent to the main one due to the splay of the 
brush bristles; b) a solid and essentially homogeneous inter-
nal structure, with a very low incidence of flaws in the paint 
layer (once again, except when the paintings are impacted by 
taphonomic processes); c) a width that tends to remain stable 
throughout the entire line; and d) initial and final tips with 
well-defined morphology. Within the scope of this compari-
son, each experimental line was evaluated and assigned a score 
reflecting its degree of similarity to the lines observed in the 
Levantine images. The aggregation of scores for the five lines 
painted with a particular tool culminated in the calculation of 
a Similarity Index (SI), which ranges from 0 (dissimilar) to 10 
(highly similar) and indicates the overall degree of similar-
ity between the experimental marks produced with the tool 
and the archaeological sample (see Supplementary Material 
– Sect. “9”).

Production of full‑size replicas of LRA motifs

After identifying the experimental tools that produce 
straight and curved lines with techno-visual features most 
similar to those of the Levantine paintings, we used them 
to create 10 full-size replicas of LRA motifs that are part 

Fig. 4  Location of the nine rock 
art sites that comprise the study 
sample



Archaeological and Anthropological Sciences (2024) 16:155 Page 9 of 22 155

of the study sample and representative of this artistic tra-
dition (Fig. 6; see Supplementary Material – Sect. “10”). 
This procedure aimed to ascertain the techno-visual affor-
dance of these tools in outlining figures, infilling the bod-
ies of human and animal motifs, and representing delicate 
anatomical details (antlers, hairs, hooves, etc.) as well as 

objects (bows, arrows, ropes, etc.). Furthermore, creating 
these replicas enabled us to observe the wear processes 
that occur when the different types of tools come into con-
tact with the support surface, and to identify which of 
them could have their lifespan extended through mainte-
nance and/or recycling procedures.

Fig. 5  Example of comparison between experimental and archaeological lines



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 10 of 22

Experimental results

Assessing the efficiency of the experimental paint 
application tools

To assess the efficiency of the experimental paint applica-
tion tools, their EIs were sorted from highest to lowest and 
divided into three performance levels (PL): a) High (EI: 8.0 
– 10.0; ovr-EI: 16.0 – 20.0); Medium (EI: 5.0 – 7.9; ovr-
EI: 10.0 – 15.9); and Low (EI: 0 – 4.9; ovr-EI: 0.0 – 9.9) 
(Tables 2, 3, 4, 5, 6, and 7). Based on this classification 
system, in the following pages we analyse the results of each 
tool category focusing on the tool’s overall performance and 
their respective ovr-EIs, which correspond to the sum of the 
EIs obtained for producing straight and curved lines. None-
theless, the performance of the tools when used for painting 
each of these two types of experimental marks can also be 
observed in the tables that summarize the data.

Thus, out of the 21 hair brushes, three exhibited a high 
PL (ovr-EI: 16.3 – 19), while the PL of 12 was rated 
as medium (ovr-EI: 10.3 – 15.6) and that of the remain-
ing six as low (ovr-EI: 6.4 – 9.7) (Table 2). Bull, boar 
and human hair brushes proved to be the most suitable 
for the production of brushes, with bull hair yielding the 
best result due to its notable balance between texture and 
flexibility (Fig. 7a). Deer and goat hair brushes, in turn, 
exhibited intermediate performances, attributed to the 

more pronounced roughness of their textures (Fig. 7b). 
Finally, rabbit and fox hair brushes had lower efficiency 
due to their excessive softness and susceptibility to defor-
mation when the tool encounters the irregularities of the 
support (Fig. 7c). Also, tools with pointed and flat active 
zones performed better than those with convex bristles. 
In this sense, it is worth noting that the efficiency of hair 
brushes is strongly dependent on the correct combination 
between the hair type and the morphology of the active 
zone. A clear example of this dependency can be found 
among the bull hair brushes. In this case, the tool with 
a p.a.z. reached an ovr-EI of 19.0, whereas those crafted 
with the same raw material but with convex and flat active 
zones exhibited ovr-EIs of 9.7 and 10.7, respectively.

Regarding the 12 plant brushes, eight exhibited a medium 
PL (ovr-EI: 10.0 – 15.5) and four a low PL (ovr-EI: 4.6 
– 9.4) (Table 3). Among these tools, those crafted from 
rush and lesser bulrush (fibrous and spongy plants) showed 
a better performance and ranked in the upper half of the 
classification table (Fig. 7d). Conversely, those made from 
hazel and lentisk (woody plants) proved less efficient and 
fell within the lower half, mainly because the harder bristles 
of these brushes cause them to exhibit a more pronounced 
drop in performance when used to paint curves (Fig. 7e). 
Furthermore, it was observed that, as in the case of hair 
brushes, the efficiency of the tools is contingent upon the 
selection of an active zone morphology that is compatible 

Fig. 6  Example of the LRA 
full-size replicas. Upside-down 
red deer from Cova Remígia 
(Digital tracing of the original 
rock art motif: López-Montalvo 
2018: 213)



Archaeological and Anthropological Sciences (2024) 16:155 Page 11 of 22 155

Table 2  Results of the efficiency analysis of hair brushes. Green: 
tools whose performance level was classified as High (EI: 8.0 – 10.0; 
ovr-EI: 16.0 – 20.0); Yellow: tools whose performance level was 

classified as Medium (EI: 5.0 – 7.9; ovr-EI: 10.0 – 15.9); Red: tools 
whose performance level was classified as Low (EI: 0 – 4.9; ovr-EI: 
0.0 – 9.9)

Straight lines Curved lines Straight and curved lines

Tool EI Tool EI Tool Ovr
EI

Bull hair brush – p.a.z. 9.5 Bull hair brush – p.a.z. 9.5 Bull hair brush – p.a.z. 19.0

Boar hair brush – f.a.z. 9.5 Boar hair brush – f.a.z. 7.8 Boar hair brush – f.a.z. 17.3

Human hair brush – p.a.z. 9.0 Human hair brush – p.a.z. 7.3 Human hair brush – p.a.z. 16.3

Deer hair brush – c.a.z. 9.0 Goat hair brush – f.a.z. 7.3 Deer hair brush – c.a.z. 15.6

Human hair brush – f.a.z. 8.5 Deer hair brush – c.a.z. 6.6 Goat hair brush – f.a.z. 14.4

Deer hair brush – f.a.z. 7.3 Rabbit hair brush – c.a.z. 6.6 Human hair brush – f.a.z. 12.9

Goat hair brush – f.a.z. 7.1 Human hair brush – c.a.z. 6.0 Boar hair brush – p.a.z. 12.3

Boar hair brush – p.a.z. 7.0 Fox hair brush – f.a.z. 6.0 Rabbit hair brush – c.a.z. 12.1

Goat hair brush – p.a.z. 6.7 Deer hair brush – p.a.z. 5.5 Deer hair brush – f.a.z. 11.9

Bull hair brush – f.a.z. 6.7 Boar hair brush – p.a.z. 5.3 Goat hair brush – p.a.z. 11.8

Bull hair brush – c.a.z. 6.4 Fox hair brush – p.a.z. 5.2 Fox hair brush – p.a.z. 11.2

Fox hair brush – p.a.z. 6.0 Goat hair brush – p.a.z. 5.1 Human hair brush – c.a.z. 11.1

Boar hair brush – c.a.z. 6.0 Rabbit hair brush – f.a.z. 4.9 Rabbit hair brush – f.a.z. 10.9

Rabbit hair brush – f.a.z. 6.0 Deer hair brush – f.a.z. 4.6 Bull hair brush – f.a.z. 10.7

Rabbit hair brush – c.a.z. 5.5 Human hair brush – f.a.z. 4.4 Fox hair brush – f.a.z. 10.3

Human hair brush – c.a.z. 5.1 Bull hair brush – f.a.z. 4.0 Bull hair brush – c.a.z. 9.7

Fox hair brush – f.a.z. 4.3 Goat hair brush – c.a.z. 3.8 Deer hair brush – p.a.z. 9.6

Rabbit hair brush – p.a.z. 4.2 Bull hair brush – c.a.z. 3.3 Boar hair brush – c.a.z. 8.3

Deer hair brush – p.a.z. 4.1 Fox hair brush – c.a.z. 2.5 Goat hair brush – c.a.z. 7.8

Fox hair brush – c.a.z. 4.1 Boar hair brush – c.a.z. 2.3 Fox hair brush – c.a.z. 6.6

Goat hair brush – c.a.z. 4.0 Rabbit hair brush – p.a.z. 2.2 Rabbit hair brush – p.a.z. 6.4

Table 3  Results of the efficiency analysis of plant brushes. Green: 
tools whose performance level was classified as High (EI: 8.0 – 10.0; 
ovr-EI: 16.0 – 20.0); Yellow: tools whose performance level was 

classified as Medium (EI: 5.0 – 7.9; ovr-EI: 10.0 – 15.9); Red: tools 
whose performance level was classified as Low (EI: 0 – 4.9; ovr-EI: 
0.0 – 9.9)

Straight lines Curved lines Straight and curved lines

Tool EI Tool EI Tool Ovr
EI

Hazel brush – c.a.z. 9.2 Rush brush – f.a.z. 7.5 Rush brush – f.a.z. 15.5

Rush brush – f.a.z. 8.0 Rush brush – c.a.z. 7.2 Rush brush – c.a.z. 14.7

L. bulrush brush – p.a.z. 7.5 L. bulrush brush – p.a.z. 6.0 L. bulrush brush – p.a.z. 13.5

Rush brush – c.a.z. 7.5 Rush brush – p.a.z. 5.5 Rush brush – p.a.z. 11.5

Hazel brush – p.a.z. 6.3 L. bulrush brush – f.a.z. 5.1 L. bulrush brush – c.a.z. 11.1

L. bulrush brush – c.a.z. 6.1 L. bulrush brush – c.a.z. 5.0 L. bulrush brush – f.a.z. 10.6

Rush brush – p.a.z. 6.0 Lentisk brush – f.a.z. 5.0 Hazel brush – p.a.z. 10.0

Lentisk brush – c.a.z. 5.8 Lentisk brush – c.a.z. 4.2 Lentisk brush – c.a.z. 10.0

L. bulrush brush – f.a.z. 5.5 Hazel brush – p.a.z. 3.7 Hazel brush – c.a.z. 9.4

Lentisk brush – f.a.z. 3.8 Lentisk brush – p.a.z. 3.4 Lentisk brush – f.a.z. 8.8

Hazel brush – f.a.z. 3.0 Hazel brush – c.a.z. 2.0 Lentisk brush – p.a.z. 5.2

Lentisk brush – p.a.z. 1.8 Hazel brush – f.a.z. 1.6 Hazel brush – f.a.z. 4.6



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 12 of 22

with the intrinsic properties of their raw material. In this 
context, brushes made from fibrous, spongy and woody 
plants reached their peak performances when crafted with 
flat, pointed, and both pointed and convex bristles, respec-
tively. The importance of this proper combination between 
raw material and active zone also becomes evident when 
comparing plant brushes with plant fragments with a whit-
tled point. Out of the eight tools that comprise this latter 

category, two exhibited a medium PL (ovr-EI: 10.4 – 11.5), 
while the remaining six tools were rated with low PLs (ovr-
EI: 4.8 – 9.7) (Table 4). Also, the highest ovr-EIs of the 
tools made from rush, lesser bulrush, hazel and lentisk were 
considerably lower than those presented by the brushes pro-
duced with these plants, suggesting that tools whose active 
zone is formed by bristles tend to be more efficient than 
those with a whittled end (Fig. 7f). It should also be noted 

Table 4  Results of the efficiency analysis of plant fragments. Yellow: tools whose performance level was classified as Medium (EI: 5.0 – 7.9; 
ovr-EI: 10.0 – 15.9); Red: tools whose performance level was classified as Low (EI: 0 – 4.9; ovr-EI: 0.0 – 9.9)

Straight lines Curved lines Straight and curved lines

Tool EI Tool EI Tool Ovr
EI

Kermes oak branch – p.a.z. 6.4 Rush branch – p.a.z. 6.5 Reed branch – p.a.z. 11.5

Reed branch – p.a.z. 5.5 Reed branch – p.a.z. 6.0 Rush branch – p.a.z. 10.4

Hazel branch – p.a.z. 4.4 Lentisk branch – p.a.z. 5.8 Kermes oak branch – p.a.z. 9.7

Rush branch – p.a.z. 3.9 L. bulrush branch – p.a.z. 4.8 Hazel branch – p.a.z. 8.4

Mallow root – p.a.z. 3.5 Hazel branch – p.a.z. 4.0 Lentisk branch – p.a.z. 8.2

L. bulrush branch – p.a.z. 2.5 Kermes oak branch – p.a.z. 3.3 L. bulrush branch – p.a.z. 7.3

Lentisk branch – p.a.z. 2.4 M. spurge branch – p.a.z. 3.2 Mallow root – p.a.z. 6.1

M. spurge branch – p.a.z. 1.6 Mallow root – p.a.z. 2.6 M. spurge branch – p.a.z. 4.8

Table 5  Results of the efficiency analysis of feathers. Green: tools 
whose performance level was classified as High (EI: 8.0 – 10.0; ovr-
EI: 16.0 – 20.0); Yellow: tools whose performance level was classi-

fied as Medium (EI: 5.0 – 7.9; ovr-EI: 10.0 – 15.9); Red: tools whose 
performance level was classified as Low (EI: 0 – 4.9; ovr-EI: 0.0 – 
9.9)

Straight lines Curved lines Straight and curved lines

Tool EI Tool EI Tool Ovr
EI

Crow feather – p.a.z. 9.5 Rock dove feather – p.a.z. 5.5 Crow feather – p.a.z. 14.0

Rock dove feather – c.a.z. 6.6 Crow feather – p.a.z. 4.5 Rock dove feather – p.a.z. 10.6

S. sparrow feather – p.a.z. 6.3 S. sparrow feather – c.a.z. 4.2 S. sparrow feather – p.a.z. 10.3

Crow feather – c.a.z. 6.0 Crow feather – c.a.z. 4.1 S. sparrow feather – c.a.z. 10.2

S. sparrow feather – c.a.z. 6.0 S. sparrow feather – p.a.z. 4.0 Crow feather – c.a.z. 10.1

Rock dove feather – p.a.z. 5.1 Rock dove feather – c.a.z. 3.5 Rock dove feather – c.a.z. 10.1

Table 6  Results of the efficiency analysis of feather fragments. Yellow: tools whose performance level was classified as Medium (EI: 5.0 – 7.9; 
ovr-EI: 10.0 – 15.9); Red: tools whose performance level was classified as Low (EI: 0 – 4.9; ovr-EI: 0.0 – 9.9)

Straight lines Curved lines Straight and curved lines

Tool EI Tool EI Tool Ovr
EI

P.p. rock dove feather – c.a.z. 6.3 L.p. rock dove feather – c.a.z. 4.5 P.p. rock dove feather – p.a.z. 8.5

P.p. rock dove feather – p.a.z. 5.5 P.p. rock dove feather – p.a.z. 3.0 L.p. rock dove feather – c.a.z. 8.0

L.p. rock dove feather – c.a.z. 3.5 L.p. rock dove feather – p.a.z. 2.0 P.p. rock dove feather – c.a.z. 8.0

L.p. rock dove feather – p.a.z. 2.0 P.p. rock dove feather – c.a.z. 1.7 L.p. rock dove feather – p.a.z. 4.0

Cal. rock dove feather – p.a.z. 1.8 Cal. rock dove feather – p.a.z. 1.6 Cal. rock dove feather – p.a.z. 3.4

Cal. rock dove feather – c.a.z. 0.4 Cal. rock dove feather – c.a.z. 1.0 Cal. rock dove feather – c.a.z. 1.4

Rac. rock dove feather – p.a.z. 0.0 Rac. rock dove feather – p.a.z. 0.0 Rac. rock dove feather – p.a.z. 0.0



Archaeological and Anthropological Sciences (2024) 16:155 Page 13 of 22 155

that, in this category of tools, no predominance of a specific 
type of raw material in terms of performance was observed, 
as fibrous (rush, mallow, and Mediterranean spurge), spongy 

(Lesser bulrush), and woody plants (hazel, lentisk, kermes 
oak, and dry reed) appear distributed throughout the entire 
classification table.

Table 7  Results of the efficiency analysis of feather brushes. Yellow: tools whose performance level was classified as Medium (EI: 5.0 – 7.9; 
ovr-EI: 10.0 – 15.9); Red: tools whose performance level was classified as Low (EI: 0 – 4.9; ovr-EI: 0.0 – 9.9)

Straight lines Curved lines Straight and curved lines

Tool EI Tool EI Tool Ovr
EI

R. dove brush (pp) – p.a.z. 7.7 R. dove brush (pp) – p.a.z. 6.3 R. dove brush (pp) – p.a.z. 14.0

S. sparrow brush (r) – p.a.z. 7.7 S. sparrow brush (r) – p.a.z. 6.0 S. sparrow brush (r) – p.a.z. 13.7

S. sparrow brush – p.a.z. 6.6 R. dove brush (lp) – c.a.z. 5.7 S. sparrow brush – p.a.z. 10.5

R. dove brush (pp) – c.a.z. 5.8 R. dove brush (pp) – c.a.z. 4.1 R. dove brush (pp) – c.a.z. 9.9

R. dove brush (lp) – c.a.z. 3.9 S. sparrow brush – p.a.z. 3.9 R. dove brush (lp) – c.a.z. 9.6

R. dove brush (lp) – p.a.z. 3.2 R. dove brush (lp) – p.a.z. 2.1 R. dove brush (lp) – p.a.z. 5.3

Fig. 7  Example of the straight and curved lines produced with the 
different types of experimental paint application tools: a) bull hair 
brush – p.a.z; b) deer hair brush – c.a.z; c) rabbit hair brush – p.a.z; 

d) rush brush – c.a.z.; e) hazel brush – c.a.z.; f) rush fragment – 
p.a.z.; g) crow feather – p.a.z; h) rock dove feather fragment (lp) – 
p.a.z; i) rock dove feather brush (pp) – p.a.z



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 14 of 22

Concerning the feathers, all six tools exhibited a medium 
PL (ovr-EI: 10.1 –14.0) (Table 5). The best results were 
achieved through the use of a crow feather, apparently 
because its behaviour is more stable than that of smaller 
feathers during the execution of technical gestures (Fig. 7g). 
Even so, a significant reduction in performance was noted 
when this tool –as well as the other feathers– was employed 
to paint curved lines. Furthermore, it was observed that for 
the three types of feathers tested, the highest ovr-EIs were 
obtained using pointed tools, which highlights, once again, 
the importance of compatibility between the properties of 
the raw material and the morphology of the active zone.

On the other hand, the seven feather fragments exhib-
ited a low PL (ovr-EI: 0.0 – 8.5) (Table 6). Besides indicat-
ing that hard parts of the feather, such as the calamus and 
rachis, are not suitable paint application tools, the data sug-
gest that modifications in the feather structure tend to make 
their technical behaviour more unstable and unpredictable, 
resulting in a limited efficiency that manifests in tools with 
both pointed and convex active zones (Fig. 7h). However, 
this instability in the technical behaviour of the feather frag-
ments can be minimized by hafting them to create feather 
brushes. Out of the six tools in this category, three exhib-
ited a medium PL (ovr-EI: 10.5 – 14) (Table 7) and three 
a low PL (ovr-EI: 5.3 – 9.9). These data suggest that the 
hafting of both feather fragments and small feathers tends 
to substantially improve their efficiency by facilitating the 
prehension of the tool, thereby providing enhanced control 
and precision in executing technical gestures (Fig. 7i). In 
this category, tools with a pointed active zone outperformed 
those with convex bristles, as observed in the case of raw 
feathers. Additionally, it was observed that using pine resin 
to attach the Spanish sparrow feather to a reed haft increased 
the stability of the active zone and contributed to the tool 
exhibiting better performance.

Drawing on these experimental findings, it was possible 
to identify the most efficient paint application tools at both 
individual and category levels. At the individual level, the 
bull hair brush with a pointed active zone, the boar hair 
brush with a flat active zone, and the crow feather with a 
pointed active zone proved to be the most efficient tools for 
painting straight lines, each displaying an EI of 9.5. Con-
cerning the production of curved lines, the bull hair brush 
with a pointed active zone stood alone at the top of the rank-
ing, maintaining an EI of 9.5 while the other tools showed 
lower values. This remarkable performance when used to 
paint both types of experimental marks resulted in an ovr. 
EI of 19.0, making this bull hair brush the most efficient of 
all the tools tested.

At the tool category level, the results exhibited some vari-
ations. Based on their average performance, feathers were 
found to be the most suitable tools for painting straight lines, 
with an avg. EI of 6.58. They were closely followed by hair 

brushes (avg. EI of 6.48), and then by plant brushes (avg. 
EI of 5.88), feather brushes (avg. EI of 5.82), plant frag-
ments (avg. EI of 3.78), and feather fragments (EI of 2.79). 
In the production of curved lines, hair brushes demonstrated 
superior efficiency, achieving an avg. EI of 5.25. They were 
followed by plant brushes and feather brushes (both with an 
avg. EI of 4.68), plant fragments (avg. EI of 4.53), feathers 
(avg. EI of 4.30), and feather fragments (avg. EI of 1.97) 
(Fig. 8).

As can be seen in Fig. 8, nearly all tool categories exhib-
ited functional variability and a significant reduction in 
performance when used to paint curved lines. This is attrib-
uted to the fact that the bristles that form the active zone 
of several tools tended to splay –with varying degrees of 
extent– during the execution of curvilinear movements, 
which leads to a decrease in precision and adaptability to 
the support surface (Fig. 9). This phenomenon is especially 
notable in the case of feathers, due to the high fragility and 
flexibility of their barbs. Exception to this general trend are 
the plant fragments, whose active zone is not composed of 
bristles. In the case of these tools, the curvilinear technical 
gestures allow for a more efficient discharge of paint onto the 
wall, since the contact against the support involves a wider 
portion of the active zone. In addition, the more pronounced 
diagonal angle between the tool and the wall –essential for 
creating curved lines– tends to facilitate their adaptability 
to the support surface, which in most cases also contrib-
utes to an improvement in precision. Therefore, taking into 
account this functional variability exhibited by the experi-
mental paint application tools, the most efficient for painting 
both straight and curved lines were, in this order, the hair 
brushes (ovr. EI: 11.73), the feathers (ovr. EI: 10.88), the 
plant brushes (ovr. EI: 10.56), the feather brushes (ovr. EI: 
10.50), the plant fragments (ovr. EI: 8.31), and the feather 
fragments (ovr. EI: 4.76).

Determining the level of correlation 
between the morphology of the tool's active zone 
and the morphology of the line tips

The data summarized in Table 8 and detailed further in the 
Supplementary Material (Section “6”), enabled the deter-
mination of the level of correlation between the morphol-
ogy of the tool’s active zone and the morphology of the 
1200 line tips. Thus, regarding the 30 paint application tools 
with a pointed active zone, the level of correlation was pre-
dominantly low (CI: 1 – 2) for both the initial (N = 13) and 
final tips (N = 14) of straight lines. In the curved lines, the 
correlation primarily ranged between medium (CI: 3 – 4) 
and non-existent (CI: 0) for the initial tips (both levels with 
N = 9), and remained predominantly low for the final tips 
(N = 13). In the case of the 19 tools with a convex active 
zone, the level of correlation was mainly medium for the 



Archaeological and Anthropological Sciences (2024) 16:155 Page 15 of 22 155

initial tips (N = 8) and non-existent for the final tips (N = 9) 
of straight lines. These levels of correlation were also pre-
dominant in curved lines, with a majority of cases classified 
as medium for the initial tips (N = 8) and non-existent for the 
final tips (N = 13). Finally, concerning the 11 tools with a 
flat active zone, the level of correlation in straight lines was 
predominantly between medium and low for the initial tips 
(both levels with N = 5), and mainly non-existent for the final 
tips (N = 5). In the case of curved lines, it was essentially 
non-existent in the initial tips (N = 6) and low in the final 
tips (N = 7).

These data demonstrate that the assumption that the mor-
phology of the line tips tends to reflect the morphology of 
the tool’s active zone lacks empirical support. In this sense, 
in straight lines produced with tools having a pointed active 
zone, only four cases of high correlation (CI: 5) were identi-
fied in the initial tips, while in curved lines, six cases were 
detected in the initial tips and one in the final tips. This 
implies that even when a high correlation was observed, 
it accounted for no more than 20% (maximum N = 6) of 
cases analysed. Furthermore, concerning the experimental 
marks painted with tools with a convex active zone, two 

Fig. 8  Efficiency Index (EI) of the six categories of experimental paint application tools

Fig. 9  Examples of precision 
and adaptability problems 
caused by the splaying of the 
tool's bristles during the produc-
tion of curved lines: a) boar 
hair brush – c.a.z; b) rock dove 
feather – c.a.z



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 16 of 22

examples of high correlation were identified in the initial tips 
of straight lines and another two in the initial tips of curved 
lines, which in each case account for only 10.5% (maximum 
N = 2) of the sample. Among the lines painted using tools 
with a flat active zone, no instances of high correlation were 
identified.

It is worth noting that, in the case of pointed tools, the 
correlation is stronger in the curved lines when compared 
to the straight lines. For both the initial and final tips of the 
curvilinear marks, there is a higher number of cases where 
the level of correlation is high or medium and a lower num-
ber of cases classified as low or non-existent. This trend 
was reversed in the case of tools with convex bristles, where 
higher levels of correlation were observed in both the initial 
and final tips of straight lines. In the case of the tools with 
flat active zones, the initial tips showed higher levels of cor-
relation in straight lines and the final tips in curved lines. 
Thus, with the exception of these instances in which tools 
with flat bristles were used to paint curves, high and medium 
correlation levels were more frequently observed in the ini-
tial tips of both straight and curved lines. This suggests that, 
compared to the final tips, the initial tips are more likely to 
yield diagnostic information regarding the active zone of the 
paint application tools.

Discussion: integrating experimental 
and archaeological data

Identifying the experimental paint application tools 
most similar to the Levantine ones

The comparison between the archaeological and the exper-
imental samples enabled the determination of the degree 
of similarity between them and the identification of the 
paint application tools that produced straight and curved 
lines with features analogous to the prehistoric ones 
(Table 9). To provide a comprehensive visualization of 
the comparison results, the similarity indices assigned to 
the lines produced with each experimental tool were plot-
ted in the graph presented in Fig. 10, in which they were 
also contrasted with the tool’s efficiency indices.

Regarding the straight lines, 10 out of 60 tested tools 
(16,67%) were able to produce experimental marks with 
the highest degree of similarity (SI: 10) to the archaeo-
logical sample. Among these, hair brushes were the most 
prevalent (N = 5), followed by plant brushes (N = 3) and 
feather brushes (N = 2). Also, tools with a pointed active 
zone predominated over those with convex (N = 3) and 
flat bristles (N = 2). When it comes to curved lines, four 
tools (6,67%) were capable of producing paintings highly 

Table 8  Classification of the level of correlation between the morphology of the active zone and the morphology of the line tips in each tool cat-
egory. H = High (CI: 5), M = Medium (CI: 3 – 4), L = Low (CI: 1 – 2), NE = Non-existent (CI: 0)

Experimental paint application tools N Straight lines Curved lines

Initial tip Final tip Initial tip Final tip

N of tools for each 
level of correlation

N of tools for each 
level of correlation

N of tools for each 
level of correlation

N of tools for each 
level of correlation

H M L NE H M L NE H M L NE H M L NE

Hair brushes – p.a.z 7 2 0 2 3 0 0 4 3 1 2 4 0 0 2 3 2
Plant brushes – p.a.z 4 1 0 2 1 0 0 3 1 1 1 0 2 0 1 2 1
Plant fragments – p.a.z 8 0 0 3 5 0 0 2 6 0 2 1 5 0 5 2 1
Feathers – p.a.z 3 1 1 1 0 0 1 2 0 1 2 0 0 0 0 3 0
Feather fragments – p.a.z 4 0 1 2 1 0 0 2 2 1 2 0 1 1 1 1 1
Feather brushes – p.a.z 4 0 1 3 0 0 3 1 0 2 0 1 1 0 2 2 0
Total 30 4 3 13 10 0 4 14 12 6 9 6 9 1 11 13 5
Hair brushes – c.a.z 7 0 2 3 2 0 1 2 4 0 4 2 1 0 0 3 4
Plant brushes – c.a.z 4 1 2 1 0 0 1 1 2 0 2 2 0 0 0 1 3
Feathers – c.a.z 3 0 0 3 0 0 0 2 1 0 0 1 2 0 0 1 2
Feather fragments – c.a.z 3 1 2 0 0 0 0 1 2 1 2 0 0 0 0 0 3
Feather brushes – c.a.z 2 0 2 0 0 0 0 2 0 1 0 0 1 0 0 1 1
Total 19 2 8 7 2 0 2 8 9 2 8 5 4 0 0 6 13
Hair brushes – f.a.z 7 0 2 3 2 0 2 1 4 0 0 2 5 0 0 5 2
Plant brushes – f.a.z 4 0 2 1 1 0 0 3 1 0 1 2 1 0 2 2 0
Total 11 0 4 4 3 0 2 4 5 0 1 4 6 0 2 7 2



Archaeological and Anthropological Sciences (2024) 16:155 Page 17 of 22 155

similar to the Levantine images. In this case, hair brushes 
(N = 2) and plant brushes (N = 2) with pointed (N = 3) and 
convex active zones (N = 1), which also rank among the 

10 tools that produced highly similar straight lines. This 
implies that only four tools afforded the production of 
both straight and curved lines with techno-visual features 

Table 9  The paint experimental 
paint application tools that 
produce lines highly similar to 
those found in the nine rock art 
sites addressed in this study

Tool category Straight lines highly similar
to the archaeological sample

Curved lines highly similar 
to the archaeological sample

Hair brushes Bull hair brush – p.a.z Bull hair brush – p.a.z
Human hair brush – p.a.zHuman hair brush – p.a.z

Deer hair brush – c.a.z
Boar hair brush – f.a.z
Human hair brush – f.a.z

Plant brushes Lesser bulrush brush – p.a.z Lesser bulrush brush – p.a.z
Rush brush – c.a.zHazel brush – c.a.z

Rush brush – c.a.z
Feather brushes Rock dove feather brush (pp) – p.a.z –––––

S. sparrow feather brush (r) – p.a.z

Fig. 10  Analysis of the relationship between the Efficiency Index (EI) 
of the experimental tools and the Similarity Index (SI) assigned to the 
straight lines produced with them: 1) bull hair brush – p.a.z.; 2) goat 
hair brush – p.a.z.; 3) deer hair brush – p.a.z.; 4) rabbit hair brush – 
p.a.z.; 5) boar hair brush – p.a.z.; 6) fox hair brush – p.a.z.; 7) human 
hair brush – p.a.z.; 8) bull hair brush – c.a.z.; 9) goat hair brush – 
c.a.z.; 10) deer hair brush – c.a.z.; 11) rabbit hair brush – c.a.z.; 12) 
boar hair brush – c.a.z.; 13) fox hair brush – c.a.z.; 14) human hair 
brush – c.a.z.; 15) bull hair brush – f.a.z.; 16) goat hair brush – f.a.z.; 
17) deer hair brush – f.a.z.; 18) rabbit hair brush – f.a.z.; 19) boar 
hair brush – f.a.z.; 20) fox hair brush – f.a.z.; 21) human hair brush 
– f.a.z.; 22) hazel brush – p.a.z.; 23) lentisk brush – p.a.z.; 24) lesser 
bulrush brush – p.a.z.; 25) rush brush – p.a.z.; 26) hazel brush – 
c.a.z.; 27) lentisk brush – c.a.z.; 28) lesser bulrush brush – c.a.z.; 29) 
rush brush – c.a.z.; 30) hazel brush – f.a.z.; 31) lentisk brush – f.a.z.; 
32) lesser bulrush brush – f.a.z.; 33) rush brush – f.a.z.; 34) hazel 

branch – p.a.z.; 35) lentisk branch – p.a.z.; 36) lesser bulrush branch 
– p.a.z.; 37) rush branch – p.a.z.; 38) kermes oak brush – p.a.z.; 39) 
Mediterranean spurge branch – p.a.z.; 40) reed branch – p.a.z.; 41) 
mallow root – p.a.z.; 42) crow feather – p.a.z.; 43) rock dove feather – 
p.a.z.; 44) Spanish sparrow feather – p.a.z.; 45) crow feather – c.a.z.; 
46) rock dove feather – c.a.z.; 47) Spanish sparrow feather – c.a.z.; 
48) lateral part of rock dove feather – p.a.z.; 49) proximal part of rock 
dove feather – p.a.z.; 50) calamus of rock feather – p.a.z.; 51) rachis 
of rock dove feather – p.a.z.; 52) lateral part of rock dove feather – 
c.a.z.; 53) proximal part of rock dove feather – c.a.z.; 54) calamus 
of rock feather – c.a.z.; 55) brush of rock dove feather (lateral part) 
– p.a.z.; 56) brush of rock dove feather (proximal part) – p.a.z.; 57; 
brush of Spanish sparrow feather – p.a.z.; 58) brush of Spanish spar-
row feather fixed with resin – p.a.z.; 59) brush of rock dove feather 
(lateral part) – c.a.z.; 60) brush of rock dove feather (proximal part) 
– c.a.z



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 18 of 22

equivalent to those of the Levantine paintings – two hair 
brushes and two plant brushes.

A closer examination of these results also allows us 
to investigate the relationship between tool efficiency 
and similarity to the archaeological sample (Fig. 10 and 
Fig. 11). In both straight and curved lines a considerable 
number of tools whose PL was classified as medium (EI: 
5 – 7.9) rank among those that produced lines equivalent 
to the depictions found at the rock art sites. This observa-
tion is important insofar as it implies that the production 
of rock art motifs with technical characteristics analogous 
to the Levantine paintings does not necessarily require the 
use of tools with the highest level of efficiency, but rather 
the use of tools that can afford the creation of lines with a 
specific set of techno-visual features. However, unlike tools 
with a high PL (EI: 8–10), which consistently allow for the 
production of paintings that reflect their high precision, 

adaptability to the support surface, and paint load capacity, 
those with a medium PL exhibit more unstable and some-
times unpredictable technical behaviour. Consequently, 
their use requires greater amounts of time, energy, knowl-
edge and technical skill, which together converge to result 
in a higher labour investment.

Another noteworthy point is that no tool with an EI lower 
than 6.0 ranks among those that produced lines highly simi-
lar to the study sample. This suggests that although tools 
with a medium PL could afford the production of Levantine 
motifs, they must possess a level of efficiency that is at least 
sufficient to enable the technical operation and the creation 
of images with the desired techno-visual features. This evi-
dence demonstrates that, contrary to common assumptions, 
the mere fact that a tool allows the application of paint onto 
a wall does not make it automatically suitable for use as a 
tool for painting rock art. This entails that the production of 

Fig. 11  Analysis of the relationship between the Efficiency Index (EI) 
of the experimental tools and the Similarity Index (SI) assigned to the 
curved lines produced with them: 1) bull hair brush – p.a.z.; 2) goat 
hair brush – p.a.z.; 3) deer hair brush – p.a.z.; 4) rabbit hair brush – 
p.a.z.; 5) boar hair brush – p.a.z.; 6) fox hair brush – p.a.z.; 7) human 
hair brush – p.a.z.; 8) bull hair brush – c.a.z.; 9) goat hair brush – 
c.a.z.; 10) deer hair brush – c.a.z.; 11) rabbit hair brush – c.a.z.; 12) 
boar hair brush – c.a.z.; 13) fox hair brush – c.a.z.; 14) human hair 
brush – c.a.z.; 15) bull hair brush – f.a.z.; 16) goat hair brush – f.a.z.; 
17) deer hair brush – f.a.z.; 18) rabbit hair brush – f.a.z.; 19) boar 
hair brush – f.a.z.; 20) fox hair brush – f.a.z.; 21) human hair brush 
– f.a.z.; 22) hazel brush – p.a.z.; 23) lentisk brush – p.a.z.; 24) lesser 
bulrush brush – p.a.z.; 25) rush brush – p.a.z.; 26) hazel brush – 
c.a.z.; 27) lentisk brush – c.a.z.; 28) lesser bulrush brush – c.a.z.; 29) 
rush brush – c.a.z.; 30) hazel brush – f.a.z.; 31) lentisk brush – f.a.z.; 
32) lesser bulrush brush – f.a.z.; 33) rush brush – f.a.z.; 34) hazel 

branch – p.a.z.; 35) lentisk branch – p.a.z.; 36) lesser bulrush branch 
– p.a.z.; 37) rush branch – p.a.z.; 38) kermes oak brush – p.a.z.; 39) 
Mediterranean spurge branch – p.a.z.; 40) reed branch – p.a.z.; 41) 
mallow root – p.a.z.; 42) crow feather – p.a.z.; 43) rock dove feather – 
p.a.z.; 44) Spanish sparrow feather – p.a.z.; 45) crow feather – c.a.z.; 
46) rock dove feather – c.a.z.; 47) Spanish sparrow feather – c.a.z.; 
48) lateral part of rock dove feather – p.a.z.; 49) proximal part of rock 
dove feather – p.a.z.; 50) calamus of rock feather – p.a.z.; 51) rachis 
of rock dove feather – p.a.z.; 52) lateral part of rock dove feather – 
c.a.z.; 53) proximal part of rock dove feather – c.a.z.; 54) calamus 
of rock feather – c.a.z.; 55) brush of rock dove feather (lateral part) 
– p.a.z.; 56) brush of rock dove feather (proximal part) – p.a.z.; 57; 
brush of Spanish sparrow feather – p.a.z.; 58) brush of Spanish spar-
row feather fixed with resin – p.a.z.; 59) brush of rock dove feather 
(lateral part) – c.a.z.; 60) brush of rock dove feather (proximal part) 
– c.a.z



Archaeological and Anthropological Sciences (2024) 16:155 Page 19 of 22 155

LRA is a technically complex process that required the use 
of tools with specific characteristics, which is corroborated 
by the observation that only lines produced with artefacts 
crafted as brushes achieved the highest level of similarity 
with the archaeological images.

These results provide empirical support for the propos-
als according to which LRA was produced through the use 
of hair brushes, plant brushes, and feather brushes (see 
Table 1). In particular, the data endorse Bea’s (2007) first 
impressions concerning the employment of these catego-
ries of tools and are also consistent with the observations 
made by Ruiz (2012) regarding the potential use of hair 
brushes. However, the assumptions regarding the identi-
fication of lines painted with feathers, based solely on the 
convex morphology of the initial tips and the pointed form 
of the final tips (Ruiz 2012), should be approached with 
great caution. As observed in our analysis of the correla-
tion between the tool's active zone and the morphology of 
the line tips it produces, out of 30 initial tips and 30 final 
tips experimentally produced with pointed feathers, only 
six initial tips exhibited a convex morphology and 12 final 
tips presented a pointed shape (see Supplementary Material 
– Section “6” – Tables 85 and 98). In the remaining cases, 
the tips showed different forms, and in only two of the 30 
lines analysed did a convex initial tip appear associated 
with a pointed final tip. Furthermore, this analysis revealed 
that all tool categories and all types of active zones tested 
are capable of producing lines with convex initial tips and 
pointed final tips, suggesting that, in most instances, the 
shape of the tips is more influenced by the mechanics of 
the technical gesture and the amount of paint in the active 
zone than by the morphology of the latter. This implies 
that lines whose tips exhibit the characteristics previously 
mentioned as indicative of the use of feathers could have 
been produced through the use of other tools, such as hair 
or plant brushes.

Moreover, the experimental evidence refutes the widely 
disseminated proposal according to which feathers were the 
only paint application tool used in the production of Levan-
tine images (Grimal and Alonso 2001). In this regard, the 
results demonstrate that lines highly similar to those found 
in LRA can be produced with other types of tools and that, 
actually, lines painted with raw feathers do not even rank 
among the most similar to the archaeological sample. Also, 
our data and practical experience challenge the idea that 
the crafting of brushes is a laborious and time-consuming 
process that would be beyond the technical capabilities of 
Levantine groups (Grimal and Alonso 2001). The experi-
mental production of 39 brushes using hairs, plants, and 
feathers allowed us to establish that this is a technically sim-
ple procedure which, after the necessary raw materials for 
tool production have already been gathered, requires less 
than five minutes.

Unveiling the Levantine painters’ toolkit

The insights derived from the integration of experimen-
tal and archaeological data enable us to draw a series of 
empirically-based inferences regarding the toolkit used by 
Levantine painters. Considering that the tools that produced 
experimental marks most similar to LRA (see Table 9) 
exhibit the highest probabilities of having been used to 
create the archaeological images, we can suggest that this 
toolkit likely consisted of hair brushes with pointed, convex, 
and flat active zones, plant brushes with pointed and con-
vex active zones, and feather brushes with pointed active 
zones. Thus, for the production of hair brushes, prehistoric 
painters could have used hairs from humans and large ani-
mals such as bulls, boars, and deer, which are also widely 
represented in LRA and the archaeological contexts associ-
ated with the decorated shelters. To craft the plant brushes, 
they could have employed fibrous, spongy, and woody plants 
such as lesser bulrush, rush, and hazel, which are typical of 
the Levantine region and, in the case of the first two, are 
readily found in wet areas and around water courses close 
to the rock art sites. As for the feather brushes, they could 
have been made by hafting flight feathers from small birds, 
such as the Spanish sparrow, or fragments of feathers from 
medium-sized birds, like the rock dove, both of which are 
native species to the study area.

The number of tools comprising this toolkit would 
depend on its level of specialization. If the Levantine 
painters had opted for a generalist toolkit, they might 
have prioritized the use of brushes that exhibit high per-
formance in producing both straight and curved lines, such 
as a bull hair brush with a pointed active zone or a rush 
brush with a convex active zone. On the other hand, if they 
opted for a more specialized toolkit, it might have included 
a greater number of tools with specific functions, such as 
brushes that are highly efficient at painting straight lines 
but less suitable for painting curves, such as a human hair 
brush with a flat active zone or a Spanish sparrow feather 
brush with a pointed active zone. However, the produc-
tion of the 10 full-size replicas of rock art motifs allowed 
us to observe that to create images with the techno-visual 
features of LRA, even a generalist toolkit would need to 
include at least some specialized tools. In this sense, we 
found that brushes with a fine pointed active zone are 
necessary to accurately depict the millimetric anatomical 
features of human figures or the antlers and hooves of deer 
images, while it is more convenient to use brushes with 
convex or flat bristles and a larger diameter for painting the 
inner part of the motifs. Also, we noted that hair and plant 
brushes are more versatile because they can be crafted 
with different diameters, whereas feather brushes are lim-
ited by the reduced thickness of this raw material. This 
limitation implies that feather brushes are more suitable 



 Archaeological and Anthropological Sciences (2024) 16:155155 Page 20 of 22

for painting the outlines and small details of the figures, 
whereas hair and plant brushes can perform the same task 
and also be used for filling the images.

The production of motif replicas also allowed us to 
explore the dynamics of the wear processes undergone by 
the paint application tools. In this regard, we observed that 
while plant brushes tend to show considerable wear due to 
the friction against the support surface (see Supplemen-
tary Material – Sect. “11”), hair and feather brushes did not 
exhibit any significant damage in their active zones. This 
suggests that plant brushes would likely have been consid-
ered expedient tools, as they not only wear out quickly but 
also have a limited capacity for maintenance and recycling 
–plant brushes tend to dry out within a few days, rendering 
them unusable. Conversely, hair brushes could be main-
tained for much longer and be easily recycled by replacing 
the worn bristles with new hair or feathers. Thus, we pro-
pose that hair and feather brushes fit into the category of 
curated technologies (Binford 1979: 261–266), and that they 
might have been used by the Levantine painters as personal 
gear –tools that would be crafted, maintained, transported, 
and recycled in anticipation of future painting activities. 
In contrast, plant brushes fit into the category of expedi-
ent technologies (ibid.) and might have been employed as 
situational gear –tools crafted to accomplish a specific task 
and in response to an immediate painting activity. There-
fore, while the hair and feather brushes are likely to have 
been produced and curated to be reused, the plant brushes 
are likely to have been discarded after use and replaced by 
new tools, crafted from fresh plants just before painting the 
images.

As previously mentioned, scholars assert that the LRA 
from the nine sites constituting the study sample is the result 
of a common tradition, socially transmitted across genera-
tions (Obermaier 1938; Porcar 1945; Viñas 1982; Domingo 
Sanz 2012). Accordingly, it is pertinent to assume that the 
Levantine people passed down not only their graphic codes 
and knowledge about the production of the rock art motifs, but 
also the chaîne opératoire through which they crafted their 
tools, building true communities of practice and establish-
ing an inextricable link between art production and knowl-
edge transmission (Conkey 2009). Through this process, the 
brushes that comprised the Levantine toolkit transitioned from 
being mere artefacts used to apply paint to the walls of shel-
ters to becoming the medium through which those prehistoric 
painters materialized their culture and identities into images.

Conclusion

After more than a century of controversies, the experimen-
tal approach developed in this study allowed us to test the 
validity of the hypotheses concerning the use of six different 

categories of paint application tools to produce LRA. The 
results show that out of 60 tools systematically tested 
through the production of 600 experimental marks, only 10 
produced straight lines, four produced curved lines, and four 
produced both straight and curved lines, with techno-visual 
features equivalent to those of the archaeological paint-
ings. This limited set of artifacts consists of hair brushes 
(N = 5), plant brushes (N = 3), and feather brushes (N = 2), 
which, configured with active zone morphologies suited to 
the intrinsic properties of the raw materials forming their 
bristles, could have composed the toolkit used by the Levan-
tine painters. Employing these tools to produce 10 full-size 
replicas of Levantine motifs provided us with a better under-
standing of their potential role within the mentioned toolkit 
and allowed us to observe that hair and feather brushes could 
be curated and used as personal gear, while plant brushes 
would be employed as expedient situational gear. In this con-
text, it is worth noting that five of the 10 tools that produced 
experimental marks highly similar to LRA exhibit a per-
formance level classified as medium, which implies that, in 
certain cases, prehistoric painters might have opted for tools 
whose use required a greater labour investment but simul-
taneously offered high techno-visual affordances. Based on 
these data, we can suggest that LRA technology was not 
strictly based on an economic logic aimed at minimizing the 
costs associated with producing the images, but rather on a 
perspective where the priority was creating paintings with 
specific techno-visual qualities that reinforced the informa-
tion transmitted through the graphic code.

The results also made it possible to refute the wide-
spread hypothesis that feathers were the only paint applica-
tion tool used in the production of LRA, as well as to dem-
onstrate that crafting brushes with the resources available 
to prehistoric populations is a technically simple operation 
that requires low levels of labour investment. Additionally, 
the analysis of the experimental evidence revealed that the 
assumption according to which the morphology of line tips 
tends to reflect the morphology of the tool’s active zone 
lacks empirical support, and that the shape of such tips is 
highly influenced by the mechanics of the technical ges-
tures executed during their creation. Therefore, these data 
provide a much deeper understanding of the technical and 
economic aspects that permeated LRA technology, a topic 
that has long been marginalized in the scholarly debates 
surrounding this prehistoric tradition.

Furthermore, these findings underscore the value of 
experimental archaeology as a rigorous method for testing 
the validity of our hypotheses about the past. In this sense, 
although the scope of the results generated by this investiga-
tion is focused on the post-Palaeolithic context of the LRA, 
extrapolation to similar cases is possible if undertaken with 
appropriate methodological caution. Moreover, the experi-
mental protocol presented here can be widely adapted and 



Archaeological and Anthropological Sciences (2024) 16:155 Page 21 of 22 155

employed as an analytical tool for systematically studying 
the production of prehistoric paintings from diverse chro-
nologies and locations around the world. As the protocol 
is adopted by other scholars, it can be expanded by adding 
new variables, refined by creating new ways of measuring 
them, and applied to new contexts and research questions. 
We envision that this process will culminate in the collective 
construction of a robust approach to explore not only the 
technical and economic dimensions of rock art production 
but also the social and symbolic aspects that permeate the 
creation of these meaningful images.

Supplementary Information The online version contains supplemen-
tary material available at https:// doi. org/ 10. 1007/ s12520- 024- 02057-7.

Acknowledgements We express our gratitude to the two anonymous 
reviewers for their detailed comments, constructive feedback, and 
enthusiastic support of our work.

Author contribution Conceptualization: Neemias Santos da Rosa 
Methodology: Neemias Santos da Rosa, Danae Fiore, Ramon Viñas 
Formal analysis and investigation: Neemias Santos da Rosa, Ramon 
Viñas Writing—original draft preparation: Neemias Santos da Rosa 
Writing—review and editing: Neemias Santos da Rosa, Danae Fiore.

Funding No funding was received for conducting this study.

Data availability Data used in this study are available to the public at 
the following repository: http:// hdl. handle. net/ 10803/ 668379.

Declarations 

Competing interests The authors declare no competing interests.

References

Almagro Basch M (1949) Un nuevo grupo de pinturas rupestres en 
Albarracín: “La Cueva de Doña Clotilde.” Teruel 1:91–116

Alonso A, Grimal A (1989) Els pintors prehistòrics de Vandellós. 
Ajuntament de Vandellós, Vandellós

Alvarez M, Fiore D, Favret E, Guerra RC (2001) The Use of Lithic 
Artefacts for Making Rock Art Engravings: Observation and 
Analysis of Use-Wear Traces in Experimental Tools Through 
Optical Microscopy and SEM. J Archaeol Sci 28:457–464. https:// 
doi. org/ 10. 1006/ jasc. 2000. 0571

Álvarez M, Fiore D (1995) Recreando imágenes: diseño de experimen-
tación acerca de las técnicas y los artefactos para realizar grabados 
rupestres. Cuadernos del Instituto Nacional de Antropología y 
Pensamiento Latino Americano 16:215–240

Antolín, F., Caruso, et al (2010) Forest resources exploitation in the 
Late Mesolithic and Early Neolithic site of Cova Fosca (Ares del 
Maestre, Castelló, Spain). In: Delhon C, Théry-Parisot I, Thiébaut 
S (eds) Des hommes et des plantes. Exploitation du milieu et ges-
tion des ressources végétales de la préhistoire á nos jours. XXXe 
Rencontres Internationales d’Archéologie et d’Historie d’Antibes. 
APDCA, Antibes, pp 317–327

Bard J, Busby C (1974) The manufacture of petroglyphs: a replicative 
experiment. Contrib Univ Calif Archaeol Res Facil 20:83–102

Bea M (2007) Aproximación experimental a la pintura levantina. Bol 
Arqueol Exp 7:4–9

Bednarik R (1998) The technology of petroglyphs. Rock Art Research 
15:23–35

Beltrán Martínez A (1968) Arte rupestre levantino. Universidad de 
Zaragoza, Zaragoza

Binford L (1979) Organization and formation processes: looking at 
curated technologies. J Anthropol Res 35:255–273

Blanco R, Barreto N (2016) Experimental rock art studies. Replication 
of pictographs from La Primavera Locality (Santa Cruz, Argen-
tina). In: Bednarik R, Fiore D, Basile M et al (eds) Paleoart and 
Materiality: the scientific study of rock art. Archaeopress, Oxford, 
pp 113–127

Conkey M (2009) Materiality and meaning-making in the understand-
ing of the Palaeolithic arts. In: Renfrew C, Morley I (eds) Becom-
ing Human: Innovation in prehistoric material and spiritual cul-
ture. Cambridge University Press, Cambridge, pp 179–194

Couraud C (1988) Pigments utilisés en préhistoire. Provenance, prépa-
ration, mode d’utilisation. Anthropologie 82:17–28

Couraud C (1976) Étude des techniques de peintures et gravures préhis-
toriques. Master Thesis, École des Hautes Etudes en Sciences 
Sociales

d’Errico F, Dayet Bouillot L, García-Diez M et al (2016) The technol-
ogy of the earliest European cave paintings: El Castillo Cave, 
Spain. J Archaeol Sci 70:48–65. https:// doi. org/ 10. 1016/j. jas. 
2016. 03. 007

Domingo Sanz I (2012) Figura humana, técnicas y territorios: hacia 
una redefinición técnica del arte rupestre levantino. In: García 
Arranz JJ, Collado H, Nash G (eds) The Levantine Question: 
the development of Spanish Levantine Rock-art. Archaeolingua, 
Budapest, pp 117–144

Ericson PGP, Tyrberg T, Kjellberg AS et al (1997) The Earliest Record 
of House Sparrows (Passer domesticus) in Northern Europe. J 
Archaeol Sci 24:183–190. https:// doi. org/ 10. 1006/ jasc. 1996. 0102

Fernández Macías L, Santos da Rosa N, Díaz-Andreu M (2023) Las 
representaciones femeninas en el arte levantino del Bajo Aragón, 
Maestrazgo y Bajo Ebro. Saguntum 55:45–68

Fiore D (2007) The economic side of rock art: concepts on the produc-
tion of visual images. Rock Art Research 24:149–160

Fiore D (2018) The materiality of rock art: image-making technology 
and economy viewed from Patagonia. In: Troncoso A, Armstrong 
F, Nash G (eds) Archaeologies of Rock Art: South American Per-
spectives. Routledge, New York, pp 23–57

Fiore D (2009) La materialidad del arte. Modelos económicos, tec-
nológicos y cognitivo-visuales. In: Barberena R, Borrazzo K, Bor-
rero LA (eds) Perspectivas Actuales en Arqueología Argentina. 
IMHICIHU, Buenos Aires, pp 123–154

Fiore F (2020) The art of making images: Technological affordance 
design variability and labour organization in the production of 
engraved artefacts and body paintings in Tierra del Fuego (South-
ern South America). J Archaeol Method Theory 27(3):481–510. 
https:// doi. org/ 10. 1007/ s10816- 020- 09474-7

Gavira JM, Hernanz A, Ruiz JF (2008) Técnica y tecnologia del arte 
rupestre em el arco mediterráneo. Memoria 10:51–54

Gell A (1994) The technology of enchantment and the enchantment of 
technology. In: Coote J, Shelton A (eds) Anthropology, Art and 
Aesthetics. Clarendon Press, Oxford, pp 40–63

Grimal A, Alonso A (2001) Acerca del estudio del Arte Levantino. 
Millars: Espai i Historia 24:87–110

Grimal A (1991) Consideracions técniques pictòriques de la pintura 
rupestre post-paleolíticas i la seva relació amb la cronologia. In: 
IX Col.loqui Internacional d’Arqueologia de Puigcerdà. Puig-
cerda-Andorra, pp 52–54

Hernández Pérez M (2012) Definiendo un arte neolítico. Artes levan-
tino, macroesquemático y esquemático en el arco mediterráneo 
peninsular. In: García Arranz JJ, Collado H, Nash G (eds) The 
Levantine Question: the development of Spanish Levantine Rock-
art. Archaeolingua, Budapest, pp 145–166

https://doi.org/10.1007/s12520-024-02057-7
http://hdl.handle.net/10803/668379
https://doi.org/10.1006/jasc.2000.0571
https://doi.org/10.1006/jasc.2000.0571
https://doi.org/10.1016/j.jas.2016.03.007
https://doi.org/10.1016/j.jas.2016.03.007
https://doi.org/10.1006/jasc.1996.0102
https://doi.org/10.1007/s10816-020-09474-7


 Archaeological and Anthropological Sciences (2024) 16:155155 Page 22 of 22

Hernández-Pacheco E (1918) Estudios de Arte Prehistórico. Revista 
de la Real Academia de Ciencias Exactas, Físicas y Naturales 
XVI:62–84

Keyser J, Rabiega G (1999) Petroglyph manufacture by indirect 
percution: the potential occurrence of tools and debitage in 
datable context. J Calif Gt Basin Anthropol 21:124–136

Landino M, Ahets Etcheberry E, Gheco L et al (2023) De las huellas a 
las técnicas: un abordaje experimental de las formas de aplicación 
de las pinturas rupestres de La Candelaria (Catamarca). Rela-
ciones 48:e058. https:// doi. org/ 10. 24215/ 18521 479e0 58

López-Montalvo E (2018) Hunting scenes in Spanish Levantine rock 
art: An unequivocal chrono-cultural marker of Epipalaeolithic and 
Mesolithic Iberian societies? Quater Inter 472:205–220

López-Montalvo E, Roldán C, Badal E et al (2017) Identification of plant 
cells in black pigments of prehistoric Spanish Levantine rock art by 
means of a multi-analytical approach. A new method for social iden-
tity materialization using chaîne opératoire. PLoS One 12:e0172225. 
https:// doi. org/ 10. 1371/ journ al. pone. 01722 25

Lorblanchet M (1991) Spitting images: replicating the spotted horses 
of Pech Merle. Archaeology 44:24–31

Obermaier H (1938) Probleme der Palaolitischen Malerei Ostspaniens. 
Quartär I:111–119

Obermaier H, Breuil H (1927) Las pinturas rupestres de los alrededores 
de Tormón (Teruel). Bol Acad Hist 90:511–531

Olària C (1988) Cova Fosca: un asentamiento meso-neolítico de caza-
dores y pastores en la serranía del Alto Maestrazgo. Diputación 
de Castellón, Castellón

Ozán IL, Oriolo S, Gutiérrez L et al (2023) Rock Art Painting Taphonomy: 
the Role of Environmental and Technological Factors. J Archaeol 
Method Theory. https:// doi. org/ 10. 1007/ s10816- 023- 09619-4

Paunero RS (1992) Manos pintadas en negativo: un ensayo de experi-
mentación. Rev Estud Reg 9:47–68

Pérez-Seoane MM (1988) Análisis artístico de las pinturas rupestres 
del gran techo de la Cueva de Altamira: materiales y ténicas - 
comparación con otras muestras de arte rupestre. PhD Thesis, 
Universidad Complutense de Madrid

Porcar JB (1945) Interpretaciones sobre el arte rupestre. Archivo de Pre-
historia Levantina 2:31–37

Porcar JB, Breuil H, Obermaier H (1935) Excavaciones en Cueva 
Remigia (Castellón). Junta Superior del Tesoro Artístico, Madrid

Porcar JB (1943) El trazo por impresión directa y el trazo caligráfico 
en el arte rupestre de Ares del Maestre. Boletín de la Sociedad 
Castellonense de Cultura XVIII:262–266

Ripoll E (1990) Acerca del algunos problemas del arte rupestre postpale-
olítico en la Península Ibérica. Espac Tiempo Forma 3:71–104

Rubio A, Viñas R, Santos da Rosa N (2019) Representacions bèl·liques 
de l’art llevantí. In: Viñas i Vallverdú R (ed) I Jornades Internac-
ionals d’Art Rupestre de l’Arc Mediterrani de la Península Ibèrica. 
XXè Aniversari de la Declaració de Patrimoni Mundial, Mont-
blanc, 25–27 d’octubre, 2019. Centre d’Interpretació de l’Art Rup-
estre de les Muntanyes de Prades (CIAR), Montblanc, pp 227–244

Ruiz JF, Royo-Lasarte J, Royo-Guillén JI, Rivero O (2022) Filling the 
Void: Rock-art Continuity Over the Pleistocene-Holocene Bound-
ary in Eastern Iberia. Camb Archaeol J 32:661–687. https:// doi. 
org/ 10. 1017/ S0959 77432 20001 05

Ruiz JF (2012) Del macro-estilo al micro-estilo. Análisis de la técnica del 
arte levantino como factor discriminante estilístico. In: García Arranz 
JJ, Collado H, Nash G (eds) The Levantine Question: the development 
of Spanish Levantine Rock-art. Archaeolingua, Budapest, pp 323–344

Sanchidrián Torti JL (2001) Manual de arte prehistórico. Ariel, 
Barcelona

Santos da Rosa N (2012) Contribuição para o estudo do Complexo de 
Arte Rupestre do Vale do Tejo: o sítio Cachão do Algarve. Master 
Thesis, University of Trás-os-Montes e Alto Douro

Santos da Rosa N (2019a) La Tecnología del Arte Rupestre Levantino: 
aproximación experimental para el estudio de sus cadenas opera-
tivas. PhD Thesis, Universitat Rovira i Virgili

Santos da Rosa N (2019b) La tecnologia del Arte Rupestre Levantino. 
Cuad Arte Prehist 7:120–146

Santos da Rosa N (2019c) Tecnología rupestre: una perspectiva 
teórico-metodológica para el estudio del arte levantino. In: Viñas 
i Vallverdú R (ed) I Jornades Internacionals d’Art Rupestre de 
l’Arc Mediterrani de la Península Ibèrica. XXè Aniversari de la 
Declaració de Patrimoni Mundial, Montblanc, 25–27 d’octubre, 
2019. Centre d’Interpretació de l’Art Rupestre de les Muntanyes 
de Prades (CIAR), Montblanc, pp 481–496

Santos da Rosa N, Cura S, Garcês S, Cura P (2014) Between tools and 
engravings: technology and experimental archaeology to the study 
of Cachão do Algarve Rock Art. In: Cura S, Cerezer J, Gurova 
M et al (eds) Technology and Experimentation in Archaeology. 
Archaeopress, Oxford, pp 87–96

Santos da Rosa N, Fernández Macías L, Díaz-Andreu M (2021a) Las 
escenas de danza en el arte rupestre levantino del Bajo Aragón y 
Maestrazgo: una síntesis crítica. Zephyrvs 87:15–31. https:// doi. 
org/ 10. 14201/ zephy rus20 21871 531

Santos da Rosa N, Fernández-Macías L, Mattioli T, Díaz-Andreu M 
(2021b) Dance Scenes in Levantine Rock Art (Spain): a Criti-
cal Review. Oxf J Archaeol 40:342–366. https:// doi. org/ 10. 1111/ 
ojoa. 12228

Santos da Rosa N, Fiore D, Viñas R (2023) Testing recipes: an experi-
mental approach to paint production processes in Levantine rock 
art (Spain). Archaeometry 65:816–832. https:// doi. org/ 10. 1111/ 
arcm. 12856

Sanz de Sautuola M (1880) Breves apuntes sobre algunos objetos pre-
históricos de la provincia de Santander. Imp. y lit. de Telesforo 
Martínez, Santander

Utrilla P (2000) El Arte Rupestre en Aragón. CAI100, Zaragoza
Vergara F, Troncoso A (2015) Rock art, technique and technology: na 

exploratory study of hunter-gatherer and agrarian communities in 
pré-hispanic Chile (500 to 1450 CE). Rock Art Research 32:31–45

Villaverde V, Martínez R, Guillem Calatayud P, et al (2012) ¿Qué 
entendemos por Arte Levantino? In: García Arranz JJ, Collado H, 
Nash G (eds) El problema “Levantino.” Archaeolingua, Budapest 
Cáceres, pp 81–115

Viñas R (2014) Arte Rupestre Levantino: El testimonio gráfico de los 
últimos cazadores-recolectores de Europa Occidental. In: Sala 
Ramon R (ed) Los cazadores-recolectores del Pleistoceno y del 
Holoceno en Iberia y el Estrecho de Gibraltar. Universidad de 
Burgos Fundación Atapuerca, Burgos, pp 681–695

Viñas R, Rubio A, Ruiz JF et al (2016) Investigación cronoestratigrá-
fica en el conjunto rupestre de la Sierra de la Pietat: abrigos de 
Ermites I y IV (Ulldecona, Tarragona, Catalunya). Revista Cuad-
ernos de Arte Prehistórico 2:70–85

Viñas R (1982) La Valltorta. Arte rupestre del levante español. Edi-
ciones Castell, Barcelona

Zotkina LV, Davydov RV (2022) Tools Used in Tagar Rock Art: Find-
ings of an Experimental Traceological Study. Archaeol Ethnol 
Anthropol Eurasia 50:60–71. https:// doi. org/ 10. 17746/ 1563- 0110. 
2022. 50.3. 060- 071

Publisher's Note Springer Nature remains neutral with regard to 
jurisdictional claims in published maps and institutional affiliations.

Springer Nature or its licensor (e.g. a society or other partner) holds 
exclusive rights to this article under a publishing agreement with the 
author(s) or other rightsholder(s); author self-archiving of the accepted 
manuscript version of this article is solely governed by the terms of 
such publishing agreement and applicable law.

https://doi.org/10.24215/18521479e058
https://doi.org/10.1371/journal.pone.0172225
https://doi.org/10.1007/s10816-023-09619-4
https://doi.org/10.1017/S0959774322000105
https://doi.org/10.1017/S0959774322000105
https://doi.org/10.14201/zephyrus2021871531
https://doi.org/10.14201/zephyrus2021871531
https://doi.org/10.1111/ojoa.12228
https://doi.org/10.1111/ojoa.12228
https://doi.org/10.1111/arcm.12856
https://doi.org/10.1111/arcm.12856
https://doi.org/10.17746/1563-0110.2022.50.3.060-071
https://doi.org/10.17746/1563-0110.2022.50.3.060-071

	Testing tools: an experimental investigation into technical and economic aspects of Levantine rock art production
	Abstract
	Introduction
	The Levantine paint application tools: over a century of controversy
	Theoretical framework: performances, affordances and labour investment in producing and using paint application tools
	Materials and methods
	Procurement of raw materials
	Production of experimental paint application tools
	Systematic testing of the experimental paint application tools
	Evaluation of experimental tools' efficiency
	Assessment of the correlation between the morphology of the tool's active zone and the morphology of the line tips
	Application of the experimental criteria to the LRA sample
	Production of full-size replicas of LRA motifs

	Experimental results
	Assessing the efficiency of the experimental paint application tools
	Determining the level of correlation between the morphology of the tool's active zone and the morphology of the line tips

	Discussion: integrating experimental and archaeological data
	Identifying the experimental paint application tools most similar to the Levantine ones
	Unveiling the Levantine painters’ toolkit

	Conclusion
	Acknowledgements 
	References