Characterizing obsidian sources with portable XRF: accuracy, reproducibility, and field relationships in a case study from Armenia

Portable X-ray fluorescence (pXRF) has been demonstrated as a powerful tool to assign obsidian artifacts to sources. Newer instruments can even match artifacts from some regions to their sources in a matter of just seconds, not minutes. There remains, however, a reluctance to use pXRF instruments to characterize the sources themselves. Many past studies have used pXRF in a region where the sources have been well characterized using lab-based techniques. That is, earlier analytical work established compositional types for the obsidian sources, and pXRF was later used to sort artifacts into those types. This is due, at least in part, to notions that pXRF instruments are insufficiently accurate or reliable to characterize the sources. The motivations to use pXRF for characterizing sources are similar to those for sourcing artifacts: many more specimens can be analyzed without concern for the financial, practical, and legal considerations associated with instruments in distant facilities. This paper documents tests conducted to investigate the accuracy and reproducibility of pXRF data relative to five laboratory-based techniques (NAA, EDXRF, WDXRF, EMPA, and LA-ICP-MS) with a focus on Armenian obsidian sources. These tests demonstrate that there is no reason to believe pXRF is inherently inaccurate, unreproducible, or otherwise inadequate for source characterization. A case study of the Pokr Arteni source highlights the advantages of pXRF, including the capability to analyze large numbers of specimens, recognize variability, and elucidate field relationships. In these respects, pXRF can facilitate more sophisticated obsidian sourcing studies.