Accuracy versus precision in locating reaction boundaries; implications for the garnet-plagioclase-aluminum silicate-quartz geobarometer Available to Purchase

Determination of the P-T location of a reaction boundary is a fundamental prerequisite to geologic thermobarometry and experimental petrology. Too often, experimentalists make no distinction between how well a given data set fits a linear model (precision) and how well the same data set estimates the actual P-T location of a reaction boundary (accuracy). The two are distinct features of a calibration, often of very different magnitudes, with accuracy limits generally much larger than precision limits. Precision limits are minimized by highly precise bracketing experiments; accuracy limits are minimized by large numbers of bracketing experiments. The differences between and magnitudes of these uncertainties are demonstrated by consideration of the anorthite-breakdown reaction. Using data from five precise experimental studies, we derive the position of the reaction boundary for anorthite breakdown as P (in MPa) = (2.20 ± 0.15)T (in K) - [6.2 (± 3.0) × 10²], where accuracy limits are at the 95% confidence level.