Non-destructive quantitative analyses of obsidian are performed at Geochemical Research Laboratory (GRL) on a QuanX EC (Thermo Electron Scientific Instruments Corporation) energy dispersive x-ray fluorescence (edxrf) spectrometer. X-ray spectra are acquired and elemental intensities extracted for each peak region of interest, after which matrix correction algorithms are applied to specific regions of the x-ray energy spectrum to compensate for inter-element absorption and enhancement effects. Following these corrections, intensities are converted to concentration estimates by employing least-squares calibration lines established for each element from analysis of up to 30 international rock standards certified by the U.S. Geological Survey, the U.S. National Institute of Standards and Technology, the Geological Survey of Japan, the Centre de Recherches Petrographiques et Geochimiques (France), and the South African Bureau of Standards.

Trace element measurements are expressed in quantitative units (i.e. parts per million [ppm] and/or weight percent composition [%]), and matches between unknowns (i.e. archaeological artifacts) and known geologic obsidian chemical groups are made on the basis of correspondences (at the 2-sigma level) in diagnostic trace element concentration values (typically ppm values for Rb, Sr, Y, Zr, Nb and, when necessary, K, Ca, Ba, Ti, Mn and Fe₂O₃^T) or Fe/Mn ratios. Artifact-to-obsidian source (geochemical type) correspondences are considered reliable if diagnostic mean measurements for artifacts fall within 2 standard deviations of mean values for geologic source standards. Composition measurements are reported to the nearest ppm (or, for Fe₂O₃^T, to nearest hundredth wt. %) to reflect calibration-imposed resolution capabilities of non-destructive edxrf spectrometry. Current resolution limits at GRL for the determination of Rb are about 4 ppm; Sr about 3 ppm; Y about 3 ppm; Zr about 4 ppm; and Nb about 3 ppm; Ba about 13 ppm; Ti about 15 ppm; Mn about 10 ppm; and Fe₂O₃^T about .02%. When counting and fitting error uncertainty estimates (± values) for an obsidian sample exceed the element-specific resolution limits given above, the larger number incorporates both composition variation and measurement error arising from differences in sample size, surface and x-ray reflection geometry.