A multivariate statistical approach of X-ray fluorescence characterization of a large collection of reverse glass paintings

Abstract

We present an X-ray fluorescence spectroscopy (XRF) study combined with a multivariate approach that allow to detect compositional differences and similarities among the glass supports of a large set of reverse glass paintings belonging to the collection of the Mistretta museum. Reverse painting on glass is an old decorative technique used since the Roman time consisting in applying a cold paint layer on the reverse side of a glass support. The collection shows a large spreading of provenience and dating of the items. In consideration of the current classification solely based on stylistic criteria, we applied a multivariate analysis on the XRF measurements data set to find a more objective classification criterion based on the elemental composition of the glass support. Results showed that glass supports can be divided into different typologies on the basis of their different ele- mental composition. Assignment of these typologies to the current assignment to a given geographical area or a given period is not straightforward, due to reuse or original erroneous attribution.