A non-quantitative approach to the XRF analysis of archaeological and historical bronzes

Portable XRF spectrometers have long been applied to the study of archaeological and historical bronzes; in most cases the quantitative aspect, i.e. the possibility of measuring absolute compositions, has been favoured, despite the inevitable surface inhomogeneities. The idea that measuring the quantitative composition of a material is good in itself, has often convinced the operators that surface abrasions are an acceptable damage, in contradiction with the main reason for which XRF is usually chosen that is, non-destructivity. This paper describes a non-quantitative - and strictly non-destructive - approach to the XRF investigation of archaeological and historical metals, with special regard for bronzes. The method is based on the possibility of characterizing a given alloy by carring out a significant number of measurements on different points of the (patinated) surface. The outcome is a cluster of data points in the space of the elements' count-rates which is a sort of characteristic fingerprint of the alloy. Compositional (dis)similarities among different alloys can be highlighted by considering the relative position of the corresponding clusters. Three case studies are also presented: a) the byzantine door of the Basilica of St. Paul out of the Walls, in Rome, b) the coins from the Hoard of Misurata, in Libya and c) the statue of the Aphrodite-Nike in Brescia; the byzantine door has thin patinae and flat surfaces and represents the "easy" situation; the coins are surface-enriched in silver, that is why XRF should not be used, but it actually is; finally the statue is covered by a layer of lead carbonate that heavily affects the measurements, so much that a filter had to be defined to remove the most altered data. Each case has a different level of complexity in the interpretation of data, but all of them show that: a) quantitative analyses are not essential for the accurate compositional comparison of alloys and b) it is possible to make up for the uncertainties due to surface inhomogeneities by increasing the number of measurements.