| Abstract: | Recently published activity–composition (a–x) relations for minerals in upper amphibolite‐ and granulite facies intermediate and basic rocks have expanded our ability to interpret the petrological evolution of these important components of the lower continental crust. If such petrological modelling is to be reliable, the abundances and compositions of phases calculated at the interpreted conditions of metamorphic equilibration should resemble those in the sample under study. Here, petrological modelling was applied to six granulite facies rocks that formed in different tectonic environments and reached different peak metamorphic pressure–temperature (P–T) conditions. While phase assemblages matching those observed in each sample can generally be calculated at P–T conditions that approximate those of peak metamorphism, a consistent discrepancy was found between the calculated and observed compositions of amphibole and clinopyroxene. In amphibole, Si, Ca and A‐site K are underestimated by the model, while Al and A‐site Na are overestimated; comparatively, in clinopyroxene, Mg and Si are generally underestimated, while Fe2+ and Al are typically overestimated, compared to observed values. One consequence is a reversal in the Fe–Mg distribution coefficient (KD) between amphibole and clinopyroxene compared to observations. Some of these mismatches are attributed to the incorrect partitioning of elements between the predicted amphibole and clinopyroxene compositions; however, other discrepancies are the result of the incorrect prediction of major substitution vectors in amphibole and clinopyroxene. These compositional irregularities affect mineral modal abundance estimates and in turn the position and size (in P–T space) of mineral assemblage fields, the effect becoming progressively more marked as the modal abundance of hornblende increases; hence, this study carries implications for estimating P–T conditions of high‐temperature metabasites using these new a–x relations. |
