Comparing three sample preparation techniques for portable X-ray fluorescence: A case study of Coarse Orange ceramic jars, Veracruz, Mexico

•Ceramics analyzed as sherds, powders, or planchets yield different pXRF measurements.•Analysis of powdered samples result in the highest precision and accuracy.•pXRF measures some elements with a higher precision and accuracy than NAA.•pXRF independently reproduces NAA chemical groups for ceramics in Veracruz, Mexico. Applications of energy dispersive portable X-ray fluorescence (pXRF; Olympus Vanta M Series VMR) for the chemical analysis of ceramic fabrics are gaining importance for archaeology. In addition to the broad accessibility of the instruments and their ability to gain first-look chemical data without destroying archaeological samples, the ability to bring the instrument to the field may be the only option for researchers working internationally. Through a systematic evaluation of quality control measures applied to multiple standard and in-house reference materials, we conclude that the pXRF employed in this study can precisely and accurately quantify many elements, some of which are not reported or possess high detection limits as measured by other instruments, including neutron activation analysis (NAA). We also demonstrate that analyzing ceramics in different states – intact sherds, homogenized powders, or pressed planchets – produce internally consistent results within categories but yield different results across sample preparation techniques. Finally, we re-analyze an archaeological sample of Coarse Orange jars from the Classic period Tuxtla Mountains, Veracruz, Mexico previously studied through petrography and NAA (Stoner, 2013; Stoner et al., 2008). Analyses of samples processed into homogenized powders yield the most accurate and precise results, rivaling the analytical capabilities of NAA for characterizing this particular sample of ceramics. Analysis of intact sherds and pressed planchets yields sufficient results to reproduce the main compositional groups identified through prior NAA, but loses some detail necessary to separate subgroups.