Keywords

1 Introduction

Once upon a time, early Anatomically Modern Humans drew images on the walls of their rock-shelters and caves. These images told myths that were transmitted by their ancestors. Thus, the present and the past are mirroring infinitely, and the notion of time is a mirage. Nowadays, in our modern world, time is at the center of all human actions and is invading our lives. A clear effect of this trend may be seen in the National Geographic project entitled “In search of Europe’s oldest art.” This indicates that research surrounding the oldest dates has in itself become a goal for prehistoric archaeologists, implying that the ultimate goal of archaeologists is to go back in time further than their colleagues. In recent years, archaeologists have engaged in a race to discover the oldest art in the world. While this is not something new, the competition is now global. Rock art older than 40,000 years has been discovered all around the world and the universality of the phenomenon is now recognized (Moro Abadía and González Morales 2013).

An example of this tendency is a recent paper published by Hoffmann et al. in Science. In this article, the authors suggest an antiquity of ca. 65,000 years old for the cave paintings of three Spanish caves (Hoffmann et al. 2018a). Since Neanderthals were the sole occupants of Europe at that time (Anatomically Modern Humans -AMHs- did not arrive before 42–40 ka) this chronology implies Neanderthal authorship. The announcement, though exclusively based on the measurement of the U/Th ratio in calcite crusts overlying paintings using a sophisticated mass spectroscopy method, was presented as scientific evidence, even if several archaeologists, paleontologists, and specialists in U/Th dating have expressed a number of scientific objections to these dates (Sauvet et al. 2017; Pearce and Bonneau 2018; Slimak et al. 2018; Aubert et al. 2018a; White et al. 2019).

The main problem with Hoffmann et al.’s paper is that the authors do not take into consideration a number of archaeological arguments. For instance, they do not refer to the knowledge that anthropologists and archaeologists have accumulated about Neanderthals for an entire century. Similarly, they do not mention that most of the hand stencils known in France and Spain have been dated by radiocarbon and U/Th, directly or indirectly, to a period between 35 and 25 ka (please see Figs. 9.1 and 9.2). This fact is particularly relevant since the authors discuss a date of 66.7 ka reported for a hand stencil from Maltravieso cave (Cáceres). However, the uniqueness of this date (that is 30,000 years older than the oldest previously known dates) is not even mentioned in the paper. At the very least, the singularity of the date in question should have been critically discussed and argued.

Fig. 9.1
A table. The column headers read site, sample description, dating method, laboratory number, carbon 14 age, age, and reference. The sites are in France and Spain.

Dated hand stencils (directly or indirectly) by radiocarbon and by U/Th (Table)

Full size image
Fig. 9.2
A table. The column headers read site, laboratory, age, age calculated B P, and reference. The sites are Grande Grotte Arcy, Les Fieux, Vilhonneur, Le Moulin de Laguenay, and Fuente del Salin.

Hand stencils found in a dated Gravettian archaeological context in France and Spain (Table)

Full size image

In the same way, Hoffmann et al. avoid recalling the potential sources of error linked with the U/Th method. U/Th dating is based on the hypothesis that calcite behaves as a closed system. However, calcite is open to exchange with the environment and some uranium can be eliminated due to its solubility in water (this phenomenon is called ‘lixiviation’), leading to a large overestimation of the resulting age. This is probably why U/Th appears as the best candidate to find the “oldest art” and fulfill the objective of the National Geographic project.

As this example illustrates, the emphasis on the origins of ‘art’ is not without problems.Footnote 1 Moreover, as I examine in the next section, the race to discover the oldest art in the world is now a global one.

2 The Race to Discover the Oldest Art in the World: A Global Competition

In the current frame of globalization, the competition to discover the oldest rock art is now worldwide. For instance, the above-mentioned dates from the Spanish caves have found a strong competitor in a fragment of hematite bearing a cross-hatched pattern from Blombos cave (South Africa) and another fragment of silcrete bearing a few red lines (please see Fig. 9.3c). This piece has been claimed to be “the oldest drawing in the world” because it was found in a layer dated to about 73 ka ago by OSL and thermoluminescence (Henshilwood et al. 2018). But the first place in this particular race is probably for a shell bearing some linear marks found at Trinil (Java) attributed to Homo erectus, dated by OSL to 430 ka ago (please see Fig. 9.3a), and triumphantly claimed to be “older than the oldest geometric engravings described so far” (Joordens et al. 2015). Two candidates fight for second place: A colored shell from Los Aviones cave (Spain), found in a Neanderthal context, dated by U/Th to 114 ka ago (Hoffmann et al. 2018c), and a bone bearing parallel lines found at Lingjing (China) in an archaeological layer dated by OSL to 125–105 ka ago and attributed to the Denisovans (please see Fig. 9.3b) (Li et al. 2019). According to these scholars, these elementary graphisms, mainly comprised of straight lines are enough to proclaim that “the cultural adaptations of archaic hominins involved symbolically mediated behavior” (Li et al. 2019: 896) and that “Neanderthal shared symbolic thinking with early modern man” (Hoffmann et al. 2018c).

Fig. 9.3
7 illustrations. 1. A shell of a shellfish and triangular marking on a stone surface. 2. A stone surface with parallel lines. 3, 4, and 5. Fragments of silcrete. 6. A map of Gorham’s cave. 7. A photograph of the La Roche-Cotard cave.

Symbolic artefacts attributed to archaic hominins (a–b), Neanderthal (d–g), and early AMH (c) prior to 40 ka. a: Trinil (Java, 430 ka, OSL) (photograph by J. Joordens and W. Lustenhouwer with permission, Naturalis Biodiversity Center (Joordens et al. 2015); b: Lingjing (China, 105–125 ka, OSL) (photograph by F. d’Errico and L. Doyon with permission in Li et al. 2019); c: Blombos, fragment of silcrete (Southern Africa, ~ 73 ka, OSL) (tracing by F. d’Errico with permission in Henshilwood et al. 2018); d: La Ferrassie (Dordogne, 40–54 ka, OSL) (redrawn from D. Peyrony 1934); e: Temnata (Bulgaria, ~ 50 ka, TL) (redrawn from M. Crémadès in Crémadès et al. 1995); f: Gorham’s cave (Gibraltar, >39 ka cal BP (14C) (tracing by F. d’Errico with permission in Rodríguez-Vidal et al. 2014); g: La Roche-Cotard (Indre-et-Loire, 75 ka, OSL) (photograph by J.-C. Marquet with permission (Marquet et al. 2014)

Full size image

Other dates coming from Indonesia, also obtained by U/Th techniques, are younger and oblige to use more subtle arguments. They are claimed to be the “earliest hand stencils” (Sulawesi Island, ~ 40 ka) (Aubert et al. 2014), the “earliest figurative art” (Borneo, ~ 40 ka) (Aubert et al. 2018b), the “earliest hunting scene” (Sulawesi Island, ~ 44 ka) (Aubert et al. 2019), and lastly “the earliest known representational work of art in the world” (Brumm et al. 2021).

The race to discover the oldest prehistoric art is often framed in terms of a competition between Europe and the rest of the world. The journalistic terms found in a number of websites can illustrate this point: “The art was painted at least 40,000 years ago, debunking the view that cave art solely existed in Europe,” (https://www.dw.com/en/oldest-known-figurative-art-found-in-borneo-cave/a-46202523), “the discovery of 40,000 years-old paintings depicting animals and the outline of hands on Indonesia’s Sulawesi Island suggests Europe was not the birthplace of art as long believed” (https://www.businessinsider.com/afp-asian-cave-paintings-challenge-europe-as-cradle-of-art-2014-10?IR=T), and “the mural dates back at least 44,000 years, which makes it about twice as old as most similar cave-art scenes in Europe” (https://www.nationalgeographic.com/science/article/ancient-cave-art-in-indonesia-may-be-worlds-oldest-hunting-scene). These sensational headlines promoted the false idea among the general public that archaeologists are obsessed with finding the oldest rock art. Needless to say, none of these non-scientific sources mentions that most of the dates are obtained by the U/Th method, which is subject to large errors, particularly in the case of thin layers of calcite which can lead to age overestimation (please see the discussion below).

One of the corollaries of this oversimplification is that discussion is often reduced to the chronology of the rock images. An example can illustrate this point. In 2014, archaeologists found a schematic engraving (possibly a reindeer) in the cave of Cathole (South Wales) in an archaeological context spanning from the Gravettian to the Bronze Age (Nash et al. 2012). The representation was attributed to the Upper Paleolithic and claimed to be “the oldest rock art in the British Isles” on the basis of a U/Th date of 14,505 ± 560 years ago even though, stylistically, the image significantly differs from Upper Paleolithic engravings.

3 Numerical Dating Techniques and Rock Art: A Brief Survey

Archaeology was born in the nineteenth century, related to the impact of two major scientific books: On the origin of species by Charles Darwin (1859) and the three volumes of Antiquités celtiques et antédiluviennes by Boucher de Perthes (1847–1864). Since then, archaeologists’ main concern has been to restitute human history in its temporal dimension. At the beginning, geology and stratigraphy served as a model to build an initial chrono-cultural framework.

With the emergence of numerical dating methods in the 1960s, archaeology - first considered as part of anthropology - gradually turned into archaeometry. In the field of rock art research, scientists have developed many dating techniques which are more or less adequate to date rock images, including radiocarbon, Uranium-series, Thermoluminescence (TL), and Optically Stimulated Luminescence (OSL), so that the main question is now: “How old is this representation?”. In this section, I briefly review the scope of each of these techniques, their field of application, and more importantly, some problems associated with them.

3.1 Radiocarbon Dating

In the 1950s, a new technique appeared which allowed researchers to determine the age of ancient objects in wood by calculating the content of residual radiocarbon (Libby 1955). The method was soon applied to determine the age of strata or layers in stratigraphy. Radiocarbon dating was a revolution. Much later, in the 1980s, the development of AMS (Accelerator Mass Spectrometry) allowed researchers to count the ratio of isotopes 12C and 14C directly and it required smaller samples of bones or charcoal. The reduction of the minimum sample size made possible to apply this technique to the dating of rock art starting in the 1980s. This revolution led some investigators to state that we had entered a “post-stylistic era” (Lorblanchet and Bahn 1993) and to proclaim the end of the stylistic dating of Paleolithic rock art (Bednarik 1995).

The major source of error in radiocarbon dating is related to the contamination of the samples. The presence of very old carbon (e.g., residue of carbonates) tends to make the dates appear older, whereas traces of recent organic materials (such as microorganisms and humic acids) make the dates appear younger. The latter effect is much more sensitive and serious than the former one. Laboratories have worked to solve this problem in recent years and, as a result of these efforts, the precision of this technique has significantly improved. Moreover, the pre-processing of samples can much better eliminate organic impurities. The most sophisticated method consists of extracting the collagen from bones by ultrafiltration (Bronk Ramsay et al. 2004) and then obtain an alpha-amino acid such as hydroxyproline (Marom et al. 2012). This led to ages which may be 4000 to 5000 years older than those obtained by the conventional method using a Geiger counter (please see Fig. 9.4). Thanks to these improvements, the chronological limit of this technique has been pushed back up to 50,000 years. The precision is now good, but the accuracy of the resulting age remains to check, because other sources of error need to be taken into consideration (Bednarik 1996). For instance, the radiocarbon age refers to the death of the organic material, not the age of the painting (that may be much younger). This effect, put forward by the detractors of the age of Chauvet cave, is called “old wood effect” (Kim et al. 2019).

Fig. 9.4
A table. The column headers read site, method, age B P, age cal B P, and references. The sites are Geissenklosterle Early Aurignacian and Sungir 3.

Influence of the pretreatment of bones on the result of radiocarbon dating (Table)

Full size image

The discovery that the concentration of radiocarbon in the atmosphere was not constant throughout time (contrary to Libby’s initial assumption) has led scientists to propose calibration curves. Broadly speaking, beyond 30,000 years, the calibrated dates (cal BP) are typically 4000 to 5000 years older than the radiocarbon age (BP). As a consequence, all the dates obtained prior to these improvements are now obsolete, and our chronological benchmark should be revised (Valladas et al. 2013). The fact that new dates are systematically older than the previous ones unfortunately fuels the race to look for the earliest dates.

3.2 Indirect Radiocarbon Dating of Overlying Crusts

When direct radiocarbon dating of charcoal paintings is attempted, specialist need to make sure that they eliminate all organic compounds originating from other sources such as calcium oxalate or other pollutants. For instance, a special acid-base-acid pre-treatment has been recommended to eliminate calcium oxalate (Bonneau et al. 2011). But inversely, calcium oxalate can be specially selected for dating red paintings or even engravings. Calcium oxalate or whewellite is a biofilm resulting from the presence of lichens, fungi, and bacteria in open-air sites. The dating of calcium oxalates has been particularly useful for the dating of Spanish Levantine art (Viñas et al. 2016). A careful sampling of the crust of calcium oxalate over and below the paintings allows researchers to obtain terminus ante quem and post quem, respectively. It was thus possible to obtain a date of 7190 ± 120 BP (8024 ± 150 cal BP) at Les Ermites rockshelter (Tarragona) for the layer overlying the paintings (Viñas et al. 2016). This date is at least 400 years prior to the arrival of Neolithic breeders and farmers commonly dated at ~7600 cal BP (García Puchol et al. 2015) and it constitutes the oldest chronometric date for Levantine art.

It is also possible to determine the time of crystallization of a calcite deposit by dating the calcium carbonate itself. An error inherent to this method occurs due to the inclusion of calcium carbonate from the enclosing rock in the newly formed calcite. The age should then be corrected to account for the ‘dead carbon fraction’ (dcf). This method has been used to determine the age of engravings in India (Taçon et al. 2013). Radiocarbon dating of calcite may also be used to check the reliability of U/Th measurements (see below).

Another technique consists of the radiocarbon AMS dating of the organic matter (diatoms, bacteria, fungi, etc.) trapped in the patina of amorphous silica formed on schist rock by the movement of water. This technique was tentatively used in the case of the Foz Côa site (Portugal) immediately after its discovery. Although specialists agree that a large part of the engravings date back to the Paleolithic period, the results of the chronometric dating fell between 6870 BP and 2170 BP, which caused considerable controversy (Watchman 1995).

3.3 Uranium Series Disequilibrium

Uranium-series disequilibrium (U/Th) is becoming an important dating method, because it allows scientists to date red paintings and/or engravings. The method is based on the natural radioactive disintegration of 234U into 230Th. When calcite forms (by precipitation of calcium carbonate), it contains a small quantity of U (which is soluble in water), but no Th (which is insoluble). With time, the 230Th/234U ratio increases and allows the date of the crystallization of calcite to be determined. Two conditions are required for proper application of this technique: First, it is required that there is no thorium in the beginning (i.e., the absence of particles of detrital thorium). Second, calcite needs to behave as a ‘closed system’ (i.e., no exchange of materials with the external environment). These conditions are often fulfilled when the samples are collected in the axis of growth of stalagmites of large diameter where the dripping water does not penetrate. This is why U/Th is often used for the calibration of radiocarbon dates. That said, it is important to note that the above-mentioned conditions are much more difficult to prove in the case of thin crusts of calcite accumulated over paintings. Within the literature, numerous examples are seen wherein calcite behaves as an open system, leading to serious errors in the age estimation of the calcite, because the 230Th/234U ratio is artificially overestimated by uranium leaching (Sanchidrián et al. 2018; Pons-Branchu et al. 2020). It should be pointed out that pollution by older or younger materials is also a case of open system in radiocarbon dating but the word is not often used.

If the system remains closed, the 230Th/234U ratio tends toward the value ‘1’ asymptotically after an exceedingly long time (secular equilibrium). However, simply obtaining a value lower than 1 does not prove that the system was closed, because even a small loss of uranium is enough to overestimate the 230Th/234U ratio and can lead to important mistakes. For instance, 33,000-year-old calcite that has lost 20% of its uranium will have an apparent age of 43,000 years old, and such a loss is undetectable.

In some cases, the loss of Uranium can be so relevant that the 230Th/234U ratio becomes larger than 1, leading to virtually infinite ages (please see Fig. 9.5). We do not know the geomorphological conditions favoring such a situation, but cases of infinite ages were found in one third of the 89 stalagmites analyzed in various Italian caves (Borsato et al. 2003). In a case cited by Scholz and Hoffmann (2008), a flowstone in an Austrian cave displayed a brief temporary behavior during which the calcite growth slowed down drastically. As a consequence, the concentration of uranium became 10 times lower, which led temporarily to an infinite age (called ‘out of range’). The authors admit that uranium loss is the best explanation. In line with this observation, Pons-Branchu et al. (2020) have recently noted that the oldest ages found in the Spanish caves studied by Hoffmann et al. (2018a) correspond to particularly low uranium concentration, which may indicate uranium loss.

Fig. 9.5
A line graph. It plots an increasing trend and a horizontal trend. The horizontal trend illustrates the highest peak labeled secular equilibrium. A marking on the top reads a proportion 230 thorium upon 234 uranium is greater than 1 for open systems.

In a closed system, the 230 Th / 234U ratio tends toward a limit of 1 after a very long time (secular equilibrium). If the values of the 230 Th / 234U ratio are greater than 1, this demonstrates that the system was open. However, the system may be open while the 230Th / 234U ratio remains lower than 1. In that case, the calculated ages are overestimated

Full size image

Another problem is that, in an open system, thorium may be incorporated from the beginning in the form of detrital particles. This supply of thorium contains both isotopes 232Th and 230Th, so the age should be corrected for the exogenous 230Th. In most cases, the correction is only approximative because the proportion of both isotopes is unknown (initial 230Th/232Th in bulk earth at secular equilibrium varying, according to various authors, between 0.8 and 1.7). When the sample is very ‘dirty,’ the needed corrections are significant, and it is wise not to take such samples into consideration. As Pettit and Pike mentioned in 2007, “at high levels, it is legitimate to reject a U-series sample as unsuitable” (Pettitt and Pike 2007, 40). Unfortunately, these authors did not follow their own advice at La Pasiega cave, where the oldest date—the only one falling within a Neanderthal time-range (please see Fig. 9.6) –shows an abnormal level of detrital particles (see Hoffmann et al. 2018a).

Fig. 9.6
A bar graph with an increasing trend. P A S 34 presents the highest peak followed by P A S 28, O 72, P A S 36, P A S 32, P A S 03, O 97, P A S 31, P A S 30, and O 100. P A S 34 presents the highest age.

Ages of calcite samples dated by U/Th in La Pasiega cave (Pike et al. 2012; Hoffmann et al. 2018a). The age of sample PAS-34 (79.66 ± 14.9 ka, minimum age 64.76 ka), is probably an outlier, overestimated due to an open system (see location of the sample on Fig. 9.8)

Full size image

Several examples of overestimated ages have been reported in the case of calcite covering prehistoric paintings (Plagnes et al. 2003; Valladas et al. 2017; Pons-Branchu et al. 2020). The best way to check the validity of U/Th dates is by crossdating, a practice consisting of dating carbonates from the same concretion by both U/Th and 14C. The precision of radiocarbon dating is low because the dead carbon fraction (dcf) is generally unknown. Nevertheless, the error arising due to the dcf is much smaller than the one that is due to the leaching of uranium. Crossdating was systematically achieved in Nerja cave, Málaga (Valladas et al. 2017). In some cases, the agreement between both methods is considered as acceptable. For instance, radiocarbon dates varying between 25,374 and 22,716 cal BP, according to the value taken for the dead carbon fraction (dcf) varying between 0% and 20%, can be compared to 26,462 ± 371 cal BP for U/Th. However, for other samples, a large discrepancy has been observed. For instance, in one case radiocarbon values are between 33,769 and 31,030 cal BP for dcf in the range of 0–20%, whereas the U/Th age is 60,276 ± 1300 cal BP. In this case, the overestimation of the U/Th age is much larger than the radiocarbon age, even considering the imprecision due to the unknown dcf. Thus, crossdating seems to be an adequate way to check if the U/Th method is overestimated because of an open system.

As the phenomenon of lixiviation is difficult to demonstrate, the ages determined by U/Th dating of thin layers of calcite overlying prehistoric paintings should always be suspected to be older than their true ages, until confirmed by other means. This is probably the case of the three Paleolithic Spanish caves discussed in the introduction of this chapter (Hoffmann et al. 2018a). These are claimed to be “minimum ages,” but the wording is misleading because the true age can be much younger if the system is open. The incredible age reaching 300,000 years that was found by U/Th in Nerja caveFootnote 2 certainly falls within this predicament (Sanchidrián et al. 2018).

3.4 Thermoluminescence

Thermoluminescence (TL) has been used to date archaeological artefacts for almost as long as radiocarbon. TL is a property of crystals that accumulate energy in traps when they are exposed to ionizing radiations; they can restitute this energy as light when they are heated. The quantity of emitted light is proportional to the time of exposition to radiations, and hence this is a way to estimate the time elapsed since the last heating of the materials (flint), its last solar exposition (grains of quartz), or its time of crystallization (calcite). The difficulty is to evaluate the amount of radiation received over time. This method can be used in the case of aeolian sediments and, more rarely, fluvial sediments, because it is difficult to be certain that the TL chronometer was completely reset before deposition. This factor makes controversial the very ancient dates of between 58,000 and 75,300 years obtained by TL in a decorated rock-shelter in Australia (Fullagar et al. 1996).

The dating of calcite by TL has been experimentally done in a number of Paleolithic caves. The comparison of the dates of concretions sampled below and above paintings allows researchers to bracket the age of the painting between a minimum and maximum date. Experiments held in Cantabrian caves provided interesting results, although the precision of the method is reduced. In Pondra cave (Cantabria), archaeologists could assign an age of between 32,946 ± 3440 (maximum) and 26,972 ± 2747 (minimum) years (González Sainz 2001). In La Garma cave (Cantabria), a TL date of 34,175 ± 3850 years (ante quem) was obtained for an ibex depiction, whereas U/Th for the same concretion gave results between 26,100 and 28,800 years (González Sainz 2003). The relative concordance between both methods, taking into consideration the large standard deviations, acts as a kind of confirmation and allows these paintings to be assigned to the Gravettian period.

3.5 Optically Stimulated Luminescence

Optically stimulated luminescence (OSL) is based on the same phenomenon as TL, except that, in this case, the stimulation is obtained by a visible light. OSL can be applied to quartz and feldspar. An important result was obtained by OSL on grains of quartz covering figurative engravings at Qurta (Egypt). The mean value of ~15,000 cal BP obtained for the rock art was fully compatible with the excavations carried out in the vicinity, demonstrating the reliability of the technology. This Pleistocene age is the oldest found in North Africa, justifying a posteriori the mediatic title of “Lascaux along the Nile” (Huyge et al. 2007, 2011).

OSL can be used to date materials up to 350 ka. At Pech de l’Azé (Dordogne), OSL demonstrated that the site was occupied for more than 130,000 years, as dates ranging from 179 ka (Pech de l’Azé-IV) to 48.9 ka (Pech de l’Azé-I) were obtained by this method (Jacobs et al. 2016).

3.6 Amino-Acids Racemization

Amino-acids racemization (AAR) has also been used for dating archaeological organic materials. Amino acids, the constituents of proteins, are essentially in an L-configuration in living beings. Racemization slowly transforms the L-form into a D-form after the death of the organism until an equilibrium between both configurations is reached. If the rate of racemization is known, the time elapsed since the death of the organism can be calculated. The problem is that this rate is highly dependent on environmental conditions (e.g., temperature, humidity, pH). Although sometimes used in paleontology, racemization is rarely used in archaeology. One study using this method focused on the antiquity of humans in North America (Bada et al. 1974), but the very ancient date of 50 ka obtained from aspartic acid racemization appears doubtful. Attempts have been made to correlate racemization of aspartic acid in high molecular weight proteins with radiocarbon at Pataud rockshelter in Dordogne. The age calculated from racemization is hardly compatible (30.4 ka) with the radiocarbon date for the same Aurignacian layer (ca 35 ka cal BP) (El Mansouri et al. 1996).

3.7 Cation Ratio and Varnish Microlamination

A method called ‘Cation-ratio’ (CR) has been proposed, based on the assumption that the mobility of various cations, such as K+, Ca2+, and Ti4+, in rock varnish on the surface of rock-shelters differs. A calibration carried out in similar conditions allows a minimum age for the varnish to be determined. Another technique to determine the minimum age is the study of the varnish microlamination (VML) that occurs due to variations in the manganese content of varnish deposited in wet and dry conditions. These two methods may be applied to petroglyphs in arid environments (e.g., Mojave Desert, Sahara Desert). Although very difficult to apply, both methods give concordant results as shown by blind experiments led by two teams (Whitley 2013). Minimum ages prior to 13,400 ± 2000 years (CR) have been reported for a petroglyph representing an extinct species of llama in the Mojave Desert, demonstrating the existence of rock art in ‘Pre-Clovis’ times in North America.

4 Discussion

The main problem with the application of these methods to the dating of rock art is that the specialists tend to neglect the archaeological context of the images. Increasingly, archaeometry takes precedence over archaeological reasoning. In this setting, it is not rare, for instance, that a stratigraphic layer that has supplied unclassifiable industry is frequently qualified a posteriori according to its radiocarbon or U-series dates. An example of this is the Abric Agut shelter (Barcelona) which presented a lithic industry initially attributed to the Middle Paleolithic. Subsequently, the industry was attributed to the Late Paleolithic-Early-Holocene on the basis of new chronometric dates (Vaquero et al. 2002).

U/Th dating is the best example of the increasing influence of archaeometry on conventional archaeological reasoning. It is now possible to announce sensational dates, particularly old ones, in a scientific review, disregarding archaeological arguments for accepting or refuting the dates, and even without considering the methodological limitations that could lead to invalid results. For instance, the only argument put forward to prove that calcite behaves as a closed system—a hypothesis that is required for the results to be reliable—is the observation that the ages become older when digging deeper in the calcite crust. It is only argued that “in an open system, preservation of the chronological order of subsamples is highly unlikely” (Hoffmann et al. 2018b, 1). However, “highly unlikely” is a rather vague expression. It is easy to show that a correct stratigraphy may be preserved in an open system if the process of lixiviation occurs regularly over time. If one observes that the age of various subsamples becomes younger as the calcite layer grows, this proves that water brings with it carbonates in a continuous way and the same water flow is able to partly dissolve the previously deposited uranium. In such a case, the age of each subsample will be overestimated but will remain in the expected order. We have created a model using a constant rate of lixiviation to show how the apparent ages of samples taken at different depths are modified in this assumption (please see Fig. 9.7).

Fig. 9.7
A line graph with 2 decreasing trends. The line titled age in absence of lixiviation presents the highest peak followed by the line titled apparent age with lixiviation at the rate U t equals U o exp negative k t with k equals 0.000046 per year.

Ages of various samples of calcite taken at different depths in the absence of lixiviation (closed system) and with the hypothesis of a constant rate of lixiviation over time (open system). A rate of lixiviation of 4.6 10−5 per year was chosen to provide feasible results: a calcite sample crystallized 20 ka ago would have an apparent age of ~60 ka

Full size image

A useful notion is that of ‘outliers’ (Bronk Ramsay 2009). Based on a mathematical definition in the Bayesian treatment of a set of radiocarbon dates, this allows values which significantly deviate from the average of the others to be eliminated. An outlier may be simply described as an aberrant value in a series. For instance, in the 29 dates from Pasiega cave determined by U/Th (Pike et al. 2012; Hoffmann et al. 2018a), one of these dates, and only one, is more than 50,000 years older than the others (please see Fig. 9.6). The rough value is 79,660 ± 14,900 years for the left side of a partitioned quadrangular sign, which corresponds to a “minimum age” of 64,760 years. The very large standard deviation is due to the proportion of detrital thorium that is exceptionally high in this sample. In the same panel, much younger dates are found for calcite crusts of similar size, texture, and color (please see Fig. 9.8). For instance, the right side of the same quadrangular sign provided a date of 3070 years, but these factors have not been taken into consideration. The simplistic reasoning used to accept the age of 64,760 years is based on two principles: (1) The rate of calcite growth may vary to a large extent according to local conditions, and (2) the oldest date is the most likely. These principles would be acceptable if the system was a closed one, but a more reasonable explanation is that the age of 65,000 years is an outlier resulting from uranium leaching, related to the frequent phenomenon of lixiviation in the context of humid caves. The fact that the outer layer of the same sample has a minimum age of 50,470 years provides another argument for lixiviation, because a total absence of calcite growth has never been observed in caves during the last 50 ka (Vogel and Kronfeld 1997; Genty et al. 2005; Genty 2008; Moreno et al. 2010; Baldini et al. 2019).

Fig. 9.8
A diagram of samples and their ages labeled. Marking on the top read 2160 and 12600 B P. Markings on the central region read 3070 B P and 64760 B P minute.

La Pasiega C (Cantabria). Ages of samples obtained by U/Th (Hoffmann et al. 2018a). Note that inside the quadrangular sign, parts of animals are enclosed. Modified after a tracing by Breuil et al. (1913)

Full size image

4.1 Preconceptions in Archaeology

When a new theory conflicts with the current state of knowledge, its scientific credibility needs to be evaluated. A possible approach to this is Popper’s principle of ‘falsifiability,’ according to which a new theory can be considered scientific only if it is possible to prove that it is not false (Popper 1959). In most cases, it is difficult to determine whether a theory is objectively valid. In archaeology, we have to confront any hypothesis to the bulk of existing facts. Up to which values can we be confident in a date? Why do we accept the U/Th dates of 14.5 ± 0.13 ka for rock art in Central India (Banerjee and Charkraverty 2015) and we discard the dates of 65,000 years for the Spanish caves, even if the methodological risk of overestimation is the same in both cases? What should we make of the dates close to 300,000 years found in Nerja cave (Sanchidrián et al. 2018)?

When physical measurements conflict with an entire set of well-established archaeological knowledge, it seems imperative to examine in detail the causes of the discrepancy. Indeed, we must be careful not to fall into inherited preconceptions. For instance, we have been taught that ‘art’ started with the arrival of AMH into Europe at the beginning of the Upper Paleolithic. How do we know that? A survey of the literature is required. The examples collected by Lorblanchet (1999) and Majkič (2017) show that, prior to the Aurignacian period, only rudimentary sets of lines and notches were produced (please see Fig. 9.3d and e). These likely constitute a preliminary step of cognition required for the development of drawing. The case of the cross-hatched engravings found in Gorham’s Cave (Gibraltar), tentatively assigned to late Neanderthals (39,000 cal BP), may also be catalogued within this preliminary phase (Rodríguez-Vidal et al. 2014) (please see Fig. 9.3f). In La Roche-Cotard-II cave (Indre-et-Loire), digital tracings on the walls were found at several places (please see Fig. 9.3g), and a strange stone with a rock bridge, into which a piece of bone was inserted, has been interpreted as a “human mask” (Marquet et al. 2016). The presence of a Mousterian industry, along with OSL dates of 75,600 years, confirm that the site was occupied by Neanderthals. In Bruniquel cave (Tarn-et-Garonne), a large circular structure made of broken stalagmites was found and dated to 176.5 ± 2.1 ka by U/Th, but no explanation, either functional or ritual, has been proposed (Verheyden et al. 2017).

Hence, the production of ‘symbolic artefacts’ that could be eventually attributed to Neanderthals is scarce and, in any case, very far from what we call ‘art.’ It probably took a long time between these first attempts and the first figurative depictions in caves. In Spain, it is even uncertain that ‘art’ was produced during the Aurignacian period (Gárate Maidagan et al. 2015).

Many examples of misleading preconceptions can be found in archaeology. These prejudices are particularly strong in the field of rock art because ‘art’ provokes strong feelings among archaeologists and the public. The most famous example is the paintings of Altamira cave (Cantabria), discovered in 1880. Their prehistoric origin was long denied, until the accumulation of other discoveries persuaded skeptics to change their minds (Cartailhac 1902). The story of the recognition of Altamira shows that the delayed authentication of this cave was mainly related to a number of political and religious preconceptions (Freeman 1994). A much more recent case has occurred with Chauvet cave. Despite more than 259 radiocarbon dates obtained, among which twelve concern direct dating of paintings, with an average value of 36,000 cal BP (Quiles et al. 2016), some investigators still deny the age of these paintings (Pettitt et al. 2009; Bahn et al. 2019). Behind pseudo-scientific arguments, the preconception at work is likely the same as in the case of Altamira in the nineteenth century: Chauvet art is too sophisticated to be so old. What is at stake is a widely accepted preconception that the art should evolve from simple to complex. Yet the phylogenesis of art has nothing to do with biological phylogenesis. If we are conscious of this, the timeline of Altamira and Chauvet are more easily understood. Art is a cultural phenomenon, and each culture follows its own path. Another controversy occurred with the discovery of the open-air sites of Foz Côa (Portugal).

A new dating technique using the patina of amorphous silica was experimentally carried out (see above). The results were largely post-Paleolithic (Watchman 1995), against the opinion of the entire community of prehistoric art specialists. Some people, more confident in technology than in archaeology, accepted these dates, likely misled by the incorrect belief that Paleolithic art occurred only in caves and that open-air rock art could not be Paleolithic. To put an end to the controversy, Ronald Dorn (1997) demonstrated that silica glaze forms an open system with continuous exchange of organic matter, so that “contamination from older and younger material is likely.” Due to the likely contamination, the ages provided by this dating technique are meaningless, and they remain compatible with the Paleolithic period. It is noteworthy that a series of fallacious arguments (such as those regarding the depiction of domesticated horses, the superimpositions, the microerosion, or even the geology of the valley) were used to attempt to justify the early dates provided. Once introduced, a preconception can be persistent in its pursuit of supporting arguments.

The same obsessive search for arguments can be observed in the Spanish Levantine art. Here, proponents of a Neolithic age are putting forward very weak arguments, such as the supposed cow udder “proving domestication,” or some arrow points supposedly made of metal. It is not bad faith; the preconception is so strong that their supporters are surely convinced that they are right. Preconceptions and biases are the worst enemy of scientific reasoning.

The fact that numerical dating is becoming so prevalent in archaeology is only one aspect of a more general trend. Broadly speaking, technology is ‘invading’ the whole field of rock art studies. It is now impossible to publish a decorated site in a mainstream journal without using 3D-photogrammetry, orthoplanes, or multispectral or hyperspectral imaging techniques (Ruiz López 2020). There is no doubt that these techniques are useful to researchers in terms of rapidity and accuracy, but they also divert from the true goal of rock art research. The only tool that escapes this critique is DStretch (decorrelation stretch), a digital treatment of images invented by J. Harman (2006), which allows details that are quasi-invisible to the naked eye to be enhanced due to the fading of paintings by natural aging. In fact, DStretch is the most powerful tool for deciphering prehistoric images; this is a revolution comparable to that of the optical microscope in the seventeenth century.

4.2 Cognition, Symbol, and Art

Here it might be useful to return to a semantic viewpoint. The ‘oldest drawing’ found at Blombos, dated to 73,000 years, is considered a “prime indicator of modern cognition and behavior” and as “evidence reflecting cultural modernity and symbol use” (Henshilwood et al. 2018). The tracings of La Roche-Cotard-II cave are approximately the same age and are considered to represent “probable symbolic activities” (Marquet et al. 2016). We agree with these cautious statements. A pattern of crossed lines traced on a stone or a set of parallel lines on a wall may have had a ‘symbolic intent,’ but the hypothesis is difficult to prove (Malafouris 2007, 2008). These types of artefacts only show the ability of humans to “create objects embedded with meaning” (Kissel and Fuentes 2017, 397). These isolated pieces are far from symbols in the sense of Peirce’s semiotics (Peirce 1991), and further still from the modern notion of art. It is worth noting that the first argument for symbolic thought pertaining to a cultural graphic tradition is 60,000 years old and comes from the decorated ostrich eggshells from Diepkloof (South Africa), where 270 engraved pieces were found and attributed to anatomically modern humans (Texier et al. 2010).

If we accept the definition proposed by most anthropologists of art, art is a sensitive form aimed at evoking emotion (Grosos 2017). A work of art starts with a vision in the creator’s mind that is then mediatized in the form of an image to be communicated to a receiver (Belting 2004). In this way, art acts as a mediator in a network of social relations and makes sense only within a given cultural context (Gell 1998). Therefore, art requires an elaborate societal structure which was likely not achieved until the Upper Paleolithic, when Aurignacian people arrived in Europe after a long migration from the Middle-East. There was a large gap between the first traces of drawing activity in South Africa and the first true artistic creations in the Swabian Jura (40,000 years ago) and Indonesia (if we accept the U/Th date of 44,000 years). Since art is social language, it had to undergo a slow progression toward ‘behavioral modernity’ (Vialou and Vilhena-Vialou 2005). Paradoxically, artwork has a symbolic function, but the inverse is not true: a ‘symbolic manifestation’ is not necessarily a work of art.

Artefacts dating back to the Lower Paleolithic have been presented as objects of “possibly iconic or symbolic meaning” (Bednarik 2008). This is the case of the figurine in volcanic stone found at Berekhat-Ram (Israel), in a layer putatively dated to 280,000–250,000 years, which supposedly reflects female anatomy. Even if this natural form is demonstrated to have been purposely modified by hominins, it is difficult to conclude that it is “possibly the earliest example of representational art” (d’Errico et al. 2000). It would be, at most, “the first witness of the recognition of forms” (Leroi-Gourhan 1965, 213). The search for unusual forms, or ‘curios’ as Leroi-Gourhan calls them, is a constant in the human mind. d’Errico et al. (2000) make a distinction between ‘gradualists,’ those who admit that the cognitive capacity for symbolism was already present in ancient forms of hominins, and ‘discontinuists,’ those who believe that a qualitative jump occurred at the dawn of the Upper Paleolithic. However, the truth probably lies somewhere in between: the ability to recognize forms in natural objects (and give them some symbolic meaning) was likely acquired progressively as the syntax of language became more complex (Botha 2008), but the ability to create images only originated when the making of those images became a social endeavor. This was only possible once the social organization and relationships between individuals reached a certain level of complexity.

5 Conclusion

Prehistoric archaeology has progressively shifted to archaeometry, relinquishing the sequencing of the events in the early history of humans to physicists. “Absolute” dating techniques are now archaeologists’ favorite tool. However, an examination of the different techniques indicates that most of them are not sufficiently accurate and reliable, due to several sources of error that are often underestimated. Some methods (such as TL and OSL) suffer from intrinsic problems that make them difficult to apply to rock art research. Other techniques, like radiocarbon dating of organics imbedded in crusts (oxalate, amorphous silica) overlapping paintings or engravings give only terminus ante quem dates that are subject to overestimation in many cases, due to open systems. U-series has proven to be a reliable method in other domains, but its application to rock art dating is still controversial because it is impossible to be certain that calcite behaves as a closed system. Crossdating with other methods (such as TL or 14C) is still rare. To date, the most reliable method is radiocarbon, thanks to the progress made in recent years concerning purification sampling techniques. The Bayesian statistical treatment of large data sets now allows the prehistorian to be relatively confident in so-called “absolute” values. However, it remains that the scope is limited to 45,000–50,000 years, with an accuracy rate that remains too low.

Absolute dating methods are not a miraculous tool able to provide us with a global chronology of rock art. In this setting, the current competition to discover the oldest art in the world is flawed by the sources of error that I have examined in this paper. We should therefore be very cautious before accepting a radical transformation of our understanding of Paleolithic societies. Moreover, we need to develop more convincing arguments than those provided by U/Th to establish the chronology of rock images.

In Europe, the archaeological data seems to indicate that the portable art in the Swabian Jura and the parietal art from Chauvet emerged during the Aurignacian period, but there is not undisputable evidence indicating that ‘art’ (in the modern sense of the term) was produced earlier (White et al. 2019). In this context, it is important to keep in mind that the production of ‘art’ does not only depend on cognitive development, but is mainly related to a number of social factors. While this is the subject of passionate controversies, it seems that ‘art’ appeared only when a group of people has reached a certain level of social complexity. In particular, the development of a visual language providing a long-lasting memory essential for the survival of the group was a necessary requisite. Neanderthals were excellent hunters, skilled tool makers, had remarkable cognitive abilities, and were well adapted to their environment, but we have still no archaeological evidence that Neanderthal societies reached the level of complexity required to produce cave paintings. For a number of reasons that are difficult to establish, they did not develop what we call ‘art.’

In this context, the ‘revelation’ in Science that cave paintings were made by Neanderthals undermines academic science. Social networks celebrated the denial of the scientific view according to which Anatomically Modern Humans were the first to create art. The loss of confidence in scientific archaeology is one of important collateral damages of such a mediatic announcement.

If the ambition of prehistoric archaeology is to build a reliable history of past societies (i.e., a genuine socio-cultural paleoanthropology), then we should reflect on these issues and avoid being exclusively temporally driven and determined by absolute dating methods. Many other avenues for archaeological thought remain to follow.