The garnet–clinopyroxene Fe2+–Mg geothermometer: an updated calibration

Multiple regression analysis on an extended dataset has been performed to refine the relationship between temperature, pressure, composition and the Fe–Mg distribution between garnet and clinopyroxene. In addition to a significant dependence between the distribution coefficient K_D and X ^Grt_Ca and X ^Grt_Mg#, as shown by the experimental data, the effect of X ^Grt_Mn has also been incorporated using data from natural Mn‐rich garnet–clinopyroxene pairs. Multiple regression of data (n=360) covering a large span in pressure, temperature and composition from 27 experimental datasets, combined with 49 natural high‐Mn granulites from Ruby Range, Montana, USA, and Karnataka, India, yields the P–T –compositional relationship (r²=0.98): where K_D=(Fe²⁺/Mg)^Grt/(Fe²⁺/Mg)^Cpx, X ^Grt_Ca=Ca/(Ca+Mn+Fe²⁺+Mg) in garnet, X ^Grt_Mn= Mn/(Ca+Mn+Fe²⁺+Mg) in garnet, and X ^Grt_Mg#=Mg/(Mg+Fe²⁺) in garnet. The Fe²⁺–Mg equilibrium between garnet and clinopyroxene does not seem to be affected by variations in the sodic content of the co‐existing clinopyroxene in the range X ^Cpx_Na=0–0.51. Comparisons between the new and former calibrations of the garnet–clinopyroxene Fe²⁺–Mg geothermometer clearly demonstrate how the various parameters in each case affect the calculated temperatures. Application of the new expression gives reasonable results for natural garnet–clinopyroxene pairs from various rock types and settings, and should be preferred to previous formulations. Using the new calibration to the self‐consistent dataset of Pattison & Newton (Contributions to Mineralogy and Petrology, 1989, 101, 87–103) suggests a systematic deviation with regard to both temperature and composition between their dataset and the datasets used in the present calibration.