Narrative cave art in Indonesia by 51,200 years ago

Prehistoric rock art provides important insights into past human cultures, but is typically challenging to date in an accurate and reliable manner. Over the past few decades, solution-based U-series methods have been used to produce early dates for rock art in several regions, including western Europe, Island Southeast Asia and Russia. In Spain, a hand stencil has been dated using solution U-series analysis of overlying calcite to at least 64.8 ka, and is therefore attributed to Neanderthals; however, dating evidence presented for this image has been questioned. Up until now, the earliest evidence for figurative art had comprised a naturalistic painting of a Sulawesi warty pig (Sus celebensis) at Leang Tedongnge in Maros-Pangkep (Fig. 1), which was dated using solution U-series to a minimum of 45.5 ka.

Thus far, only solution-based U-series methods have been used to date calcium carbonate deposits that formed in direct association with rock art. This approach involves the physical microexcavation of arbitrary layers of calcium carbonate and their chemical preparation before analysis using multicollector inductively coupled plasma mass spectrometry (MC-ICP-MS). The technique allows for small analytical error, but has some disadvantages for dating thin calcium carbonate layers associated with rock art, especially those with complex growth history; the latter includes the coralloid speleothems (also known as ‘cave popcorn’), which are most often found in association with rock art in limestone karst caves and shelters in Indonesia. Here we used a laser-ablation U-series (LA-U-series) approach to date these thin calcite accretions in association with rock art. Instead of physically microexcavating arbitrary ‘spits’ using a rotary tool, as was done in the previous approach, the laser is focussed on a polished cross-section and requires no chemical preparation. The small spot size of the laser-ablation system (typically spot diameters of 44 μm) offers several advantages over solution-based methods: (1) it is fast and cost effective; (2) it requires a much smaller sample, and is therefore less destructive; (3) the spatial resolution obtained using this method allows the detailed growth history of the speleothem to be revealed and the age of the oldest deposit closest to the pigment layer to be precisely defined, improving accuracy; (4) it can be easily and rapidly demonstrated that the calcite has not undergone remobilization of uranium (and areas exhibiting these issues can be identified and avoided), which could be problematic for U-series dating (Methods, Supplementary Information and Extended Data Table 1).

Coralloid speleothems found in association with rock art often have complex internal morphologies that reflect their origin as aggregates of a cluster of cylindrical, mound-like calcite structures, leaving overhanging features with gaps between older material that are infilled by carbonate materials of younger age. As the physical microexcavation procedure involves laterally collecting material from an arbitrary depth above the pigment layer—as opposed to sampling individual laminae—the resultant U-series age could, in some instances, be an average of the older mound material and the younger infill. Similarly, the lateral averaging of onion-like undulating layers can mix calcium carbonate material of different ages, sometimes resulting in a series of arbitrary subsamples with ages appearing not to be in chronological order. Such a complicated growth history may account for the small age reversals that are sometimes observed in previously dated samples using mechanical abrasion of arbitrary layers. Here we used our LA-U-series approach to map the U-series isotopes across the surface areas of sample cross-sections. This approach enables us to understand the complex ways in which coralloid speleothems have formed, therefore enabling the analyst to identify, and avoid, small zones affected by diagenesis.

The LA-U-series method generally provides age estimates with larger errors than the solution-based method, but it can result in genuinely older minimum ages for art as calcium carbonate material closer to the pigment layer can be analysed. This error can be improved by integrating a larger area of data; however, it can result in younger minimum ages as calcium carbonate material from later growth stages would need to be integrated. A more efficient way of minimizing this error involves slowing down the speed of the laser stage and increasing the integration time on the MC-ICP-MS, resulting in more datapoints for similar integrated areas. The trade-off is a substantial increase in the time needed to complete the analysis (Methods and Supplementary Information). We found that a spot size of 44 µm with a laser rastering speed of 21 µm s⁻¹ (with a 2.097 s integration time) was optimal for most circumstances.