U-Th dating of carbonate crusts reveals Neandertal origin of Iberian cave art

See allHide authors and affiliations

Science  23 Feb 2018:
Vol. 359, Issue 6378, pp. 912-915
DOI: 10.1126/science.aap7778

Neandertal cave art

It has been suggested that Neandertals, as well as modern humans, may have painted caves. Hoffmann et al. used uranium-thorium dating of carbonate crusts to show that cave paintings from three different sites in Spain must be older than 64,000 years. These paintings are the oldest dated cave paintings in the world. Importantly, they predate the arrival of modern humans in Europe by at least 20,000 years, which suggests that they must be of Neandertal origin. The cave art comprises mainly red and black paintings and includes representations of various animals, linear signs, geometric shapes, hand stencils, and handprints. Thus, Neandertals possessed a much richer symbolic behavior than previously assumed.

Science, this issue p. 912


The extent and nature of symbolic behavior among Neandertals are obscure. Although evidence for Neandertal body ornamentation has been proposed, all cave painting has been attributed to modern humans. Here we present dating results for three sites in Spain that show that cave art emerged in Iberia substantially earlier than previously thought. Uranium-thorium (U-Th) dates on carbonate crusts overlying paintings provide minimum ages for a red linear motif in La Pasiega (Cantabria), a hand stencil in Maltravieso (Extremadura), and red-painted speleothems in Ardales (Andalucía). Collectively, these results show that cave art in Iberia is older than 64.8 thousand years (ka). This cave art is the earliest dated so far and predates, by at least 20 ka, the arrival of modern humans in Europe, which implies Neandertal authorship.

The origin of human symbolism is a central concern of modern paleoanthropology (1). For the European Middle Paleolithic and the African Middle Stone Age, symbolic behavior has been inferred from the use, presumably for body adornment, of mineral pigments, shell beads, eagle talons, and feathers (27). Cave and rock art constitutes particularly impressive and important evidence for symbolic behavior (8), but little is known about the chronology of its emergence, owing to difficulties in precise and accurate dating (9).

Claims for Neandertal authorship of cave art have been made (10, 11). However, ambiguities of indirect dating and uncertainty in distinguishing between natural and intentional modification (12, 13) leave these claims unresolved. Recent technical developments enable the possibility of obtaining age constraints for cave art by U-Th dating of associated carbonate precipitates (14). This dating approach can provide robust age constraints while keeping the art intact. However, it is a destructive technique, in that a carbonate sample is required (albeit, a very small sample, typically <10 mg) and is taken not from the art itself but from the associated carbonates. The key condition is demonstrating an unambiguous stratigraphic relationship between the sample and the art whose age we wish to constrain. Dating of carbonate crusts formed on top of the art provides a minimum age (15). For art painted on top of carbonates (e.g., on flowstone walls, stalagmites, or stalactites), dating the underlying “canvas” provides a maximum age (15).

With this approach, the earliest results so far are for a hand stencil from Leang Timpuseng, Sulawesi (Indonesia), with a minimum age of 39.9 thousand years (ka) (16), and a red disc on the Panel of Hands in El Castillo, Cantabria (Spain), with a minimum age of 40.8 ka (17). Whereas the art in Sulawesi has been attributed to modern humans, the minimum age for the red disc in El Castillo relates to a point in time when it could be attributed to either Cantabria’s first modern humans or the region’s earlier Neandertal populations (18, 19).

Here we report U-Th dating results of carbonate formations associated with rock art in three Spanish caves: La Pasiega (Cantabria), Maltravieso (Extremadura), and Doña Trinidad (or Ardales; Andalucía) (fig. S1) (20). Our criteria for sample selection and subsequent sampling strategy strictly followed previously described methods (14). The reliability of the U-Th dating results is controlled by quality criteria for the carbonate (14) as well as by the collection and analysis of multiple subsamples of a given crust.

La Pasiega is part of the Monte Castillo cave art complex, a World Heritage Site that also includes the caves of El Castillo, Las Chimeneas, and Las Monedas. Together, these caves show continued human occupation throughout the past 100 ka. At La Pasiega, the rock art comprises mainly red and black paintings, including groups of animals, linear signs, claviform signs, dots, and possible anthropomorphs (21). Maltravieso was episodically used by hominin groups during the past 180 ka (22); it contains an important set of red hand stencils (~60), which form part of a larger body of art that includes both geometric designs (e.g., dots and triangles) and painted and engraved figures (23). Ongoing excavations have shown that Ardales was occupied in the Middle and Upper Paleolithic. Its walls feature an impressive number (>1000) of paintings and engravings in a vast array of forms, including hand stencils and prints; numerous dots, discs, lines, and other geometric shapes; and figurative representations of animals, including horses, deer, and birds (24).

We obtained U-Th ages for 53 samples removed from 25 carbonate formations stratigraphically related to paintings in these caves. The full details of our methods and data are described in the supplementary materials (20). Here we present and discuss the results that are most meaningful for the antiquity of the art.

In La Pasiega gallery C (fig. S2), a cauliflower-type carbonate formation on top of a red scalariform sign [panel 78 of hall XI (Fig. 1) (20)] yielded U-Th dates for three subsamples (outer, middle, and inner) that increase in age with depth—that is, toward the pigment layer. They provide a minimum age of 64.8 ka (sample PAS 34) (Table 1) (20) for the sign.

Fig. 1 Red scalariform sign, panel 78 in hall XI of La Pasiega gallery C.

This panel features the La Trampa pictorial group (21). (Inset) Crust sampled and analyzed for a minimum age (64.8 ka), which constrains the age of the red line. See (20) for details.

Table 1 U-Th results of samples discussed in the text.

More details and additional results can be found in table S4 (20). All ratios are activity ratios. Analytical errors are at the 95% confidence level. Spl ID, sample identifier.

View this table:

In Maltravieso (fig. S7), we dated samples from five locations on various carbonate formations overlying the same red hand stencil (motif GS3b) (Fig. 2) (20). Carbonate deposits almost completely obscure this hand stencil, making it difficult to see with the naked eye and challenging to record by conventional photography. Figure 2 therefore also shows a version of the photographic documentation after we used the DStretch software (25) to enhance the image. For subsamples in all locations, the expected depth-age consistency was confirmed. The oldest date provides a minimum age of 66.7 ka (MAL 13) (Table 1) (20) for the hand stencil.

Fig. 2 Hand stencil GS3b in Maltravieso cave (minimum age 66.7 ka).

(Left) Original photo. The inset shows where the overlying carbonate was sampled for MAL 13. (Right) Same picture after application of the DStretch software (25) (correlation LRE 15%, auto contrast) to enhance color contrast. See (20) for details.

In Ardales (fig. S9), we dated layers of five carbonate curtains from three areas of the cave (II-A, II-C, and III-C) that had been painted red. In three cases we were able to obtain both maximum and minimum ages by dating samples from immediately underneath the pigment and from carbonate that subsequently formed on top. These age pairs constrain one or more episodes of painting to between 48.7 ka and 45.3 ka ago (ARD 14 and 15), 45.5 ka and 38.6 ka ago (ARD 26 and 28), and 63.7 ka and 32.1 ka ago (ARD 6 and 8) (Table 1) (20). A further two samples yielded minimum ages of 65.5 ka (ARD 13) (Fig. 3), indicating an earlier episode of painting, and 45.9 ka (ARD 16), consistent with the other episodes (fig. S42) (20).

Fig. 3 Speleothem curtain 8 in section II-A-3 in Ardales cave with red pigment, painted before at least 65.5 ka ago.

(Left) Series of curtains with red paint on top, partially covered with later speleothem growth. The white rectangle outlines the area shown at right. (Right) Detail of curtain 8. The black square indicates where carbonate, overlying the red paint, was sampled for ARD 13. See (20) for details.

Criteria for reliable minimum (or maximum) ages (14) were met by all samples. The oldest minimum ages from the three caves are consistent and, at 64.8 ka or older for each site, substantially predate the arrival of modern humans in Europe, which has been variously estimated at between 45 ka and 40 ka ago (26, 27). Our dating results show that cave art was being made at La Pasiega, Maltravieso, and Ardales at least 20 ka before that. In this age range, Iberia was exclusively populated by Neandertals, as indicated by numerous diagnostic osteological remains, including articulated skeletons (28, 29). The implication is, therefore, that the artists were Neandertals.

All examples of early cave art dated thus far were created in red pigment, and comprise dots, lines, disks, and hand stencils (30). This is a restricted and nonfigurative set of subjects and could represent the extension of Neandertal body art to the external world. Regardless of whether concentrations of color, dots, disks, and linear motifs can be conceived as symbolic, hand stencils (which, unlike positive hand prints, cannot be created by accident) require a light source and previous selection and preparation of the coloring material—evidence of premeditated creation. Because a number of hand stencils seem to have been deliberately placed in relation to natural features in caves rather than randomly created on accessible surfaces (31), it is difficult to see them as anything but meaningful symbols placed in meaningful places.

This cave painting activity constitutes a symbolic behavior by definition, and one that is deeply rooted. At Ardales, distinct episodes over a period of more than 25 ka corroborate that we are not dealing with a one-off burst but with a long tradition that may well stretch back to the time of the annular construction found in Bruniquel cave, France (32), dated to 176.5 ± 2.1 ka ago. Dating results for the excavation site at Cueva de los Aviones, Spain (2), which place symbolic use of marine shells and mineral pigments by Neandertals at >115 ka ago (33), further support the antiquity of Neandertal symbolism.

Cave art such as that dated here exists in other caves of Western Europe and could potentially be of Neandertal origin as well. Red-painted draperies are found at Les Merveilles (France; panel VII) (34) and El Castillo (Spain), whereas hand stencils and linear symbols are ubiquitous and, when part of complex superimpositions, always form the base of pictorial stratigraphies. We therefore expect that cave art of Neandertal origin will eventually be revealed in other areas with Neandertal presence elsewhere in Europe. We also see no reason to exclude that the behavior will be equally ancient among coeval non-Neandertal populations of Africa and Asia.

The authorship of the so-called “transitional” techno-complexes of Europe, which, like the Châtelperronian, feature abundant pigments and objects of personal ornamentation, has long been the subject of debate (35, 36). Direct or indirect (via acculturation) assignment to modern humans has been based on an “impossible coincidence” argument—that is, the implausibility that Neandertals would independently evolve the behavior just at the time when modern humans were already in or at the gates of Europe. By showing that the Châtelperronian is but a late manifestation of a long-term indigenous tradition of Neandertal symbolic activity, our results bring closure to this debate.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S42

Tables S1 to S4

References (3751)

References and Notes

  1. Minimum ages are calculated by subtracting the 95% uncertainty from the mean; maximum ages are calculated by adding the 95% uncertainty to the mean.
  2. See supplementary materials.
Acknowledgments: This research was financially supported by the Natural Environment Research Council (UK) (grant NE/K015184/1 to A.W.G.P.), the National Geographic Society (USA) (grant EC0603-12 to D.L.H.), the Max Planck Society (Germany), and a Royal Society Wolfson Research Merit Award (to A.W.G.P.). The work of M.G.-D. was supported by the Research Group IT622-13 of the Basque government. We thank the governments of Andalucía, Cantabria, and Extremadura for sampling permissions. We are grateful for fieldwork support by J. C. Aguilar, M. Batut, J. R. Bello, D. Garrido, R. Gutiérrez, and C. Hoffmann. The data described are presented in the supplementary materials. We dedicate this paper to the memory of José Antonio Lasheras.

Stay Connected to Science

Navigate This Article