Does the Artemidorus papyrus have multiple lives? Seeking for the answer in the inks through a Raman and PCA analysis

Artemidorus papyrus has been the subject of a longstanding dispute around its authenticity. The 2.5 m long document, presents a multiplicity of writings, on both sides. Some historians support the papyrus’ multiple life hypothesis (PMLH): the different writings appearing on the document, would correspond to three time periods (1st BC–1st AD) called the “three lives of the papyrus”. A possible way to assess whether the document has been written in different eras, is the analysis of the graphic materials employed; a heterogeneous formulation of the inks used throughout the papyrus as a whole, together with a local homogeneity in defined regions, would support PMLH. To unveil this paradigm, we performed an in-depth investigation of the inks on the papyrus, by collecting their Raman spectra on a statistical relevant number of samples belonging to the different regions of the document. We found a wide variety of carbon-based inks that appear to be randomly distributed on the entire surface. Different ordered, disordered, crystalline or amorphous structures might coexist in carbon-based materials, due to the different hybridisations of the carbon. All of the different structures appear as peaks, shoulders or sub peaks that are convoluted in the global spectrum, thus rendering the attribution of Raman peaks as an extremely complex task. In order to support and guide our analysis, we therefore implemented a Principal Component Analysis (PCA) aimed at performing a two-step comparison between the Artemidorus inks Raman spectra and an ample reference spectral library on carbon-based materials that we built specifically for this study. A first PCA screening was used to restrict the analysis for similarity between the experimental spectra and the library of standards to a subset of compounds. It was indeed possible to define a cluster in the Principal Component subspace that contained only a few standards and the whole set of Artemidorus spectra. It is interesting to notice that, among the possibilities in the reference library, our analysis also found two pigments manufactured with modern industrial methods: Carbon black (after 1870) and lampblack (after 1740, modified in 19th c.). A second PCA, performed on the subset of the standard spectra individuated, allowed to highlight the regions of the Raman spectra where the Artemidorus data depict significant differences with respect to their reference standards, thus allowing for a guided deconvolution of the Raman peaks. This study demonstrates that it is not possible to define neat regions on the papyrus that would be identifiable through the usage of different groups of inks, as the same subset of inks is distributed on all of the document. It is therefore impossible to recognise areas of the papyrus where the local usage of specific and region-dependent graphic materials would lead to chronologically distinguishable writings and drawings. This allows us to confute the “three lives hypothesis”.