A multivariate analysis for enhancing the interpretation of infrared spectra of plant residues on lithic artefacts

The reliability in using chemical techniques for the characterization of organic and inorganic ancient residues has been proved by several scientific works in the archeological literature. Among the different proposed approaches, Fourier Transform InfraRed Spectroscopy (FTIR) demonstrated to be a valuable method to detect and interpret the use-related residues entrapped in lithic chipped stone tools after their use. However, FTIR is affected by some restrictions that could be partially overcome thorough a statistical treatment of its results. With this contribution, we propose a methodological study based on Principal Component Analysis (PCA) as a statistical method to analyze infrared spectra collected on experimental residues on stone tools used to process vegetal materials. Here we show the ability of the PCA multivariate analysis to cluster samples with similar characteristics and to mine subtle spectral differences. According to our results, we proved the technique to be very suitable to discriminate vegetal materials, especially if belonging to herbaceous plants and underground storage organs (USOs) species where the likelihood to be recognized account for about 86% of possibilities. Our results demonstrate how the application of statistical methods such as PCA may amplify the interpretative capacity of the InfraRed Spectroscopy, especially when applied to archeological materials. Moreover, we contribute to build a representative spectral database of vegetal residues useful for scholars dealing with infrared measurements focused on the interpretation of the function of the archaeological artefacts.