100 mm 2 ) from the rock art panel using a battery-operated rotary tool equipped with a diamond saw blade. In the laboratory, we micro-excavated each sample in arbitrary spits extending over the surface of each speleothem, creating a series of four (LBLGJ1) and five (LBLGJ2) aliquots per sample measuring less than 1 mm in thickness. The resultant sections consist of two continuous microstratigraphic profiles extending from the outer surface of each speleothem through the pigment layers and into the underlying rock face. The samples overlying both the pig figure and the hand stencil comprise multiple layers of dense and nonporous calcite. In the case of both sampled coralloids, we clearly observed the pigment layer corresponding to each artwork (the red pig figure and the hand stencil) across the entire length of each sample. In total, we obtained nine U-series age determinations for the red pig figure and the hand stencil overlying it (Table 1 and Fig. 6, D and E). Sample LBLGJ1 yielded an older age at the surface of the sample. Because this aliquot (LBLGJ1.1) includes parts of the weathered and friable crust of the speleothem, we interpret this older age as being caused by partial leaching of uranium at the surface. Samples LBLGJ1.2 and LBLGJ1.3 show a slightly reversed age profile (<100>

2000 years). We also interpret this anomaly as per by-product of our microsampling strategy and the convoluted process of coralloid speleothem formation (Materials and Methods). The internal growth structure revealed durante the ciclocross section (Fig. 6E) is obviously more complex, implying verso high probability of ciclocampestre-growth layers mixing durante different proportions among different aliquots during the microsampling process. We therefore consider that the oldest minimum U-series date of 32 ka given by subsample LBLGJ2.4 represents the minimum age of the hand stencil, whereas the maximum age for this particular artwork (73.4 ka) is provided by the U-series date for underlying sample LBLGJ2.5.