A series of basaltic compositions and compositions within the simple system CaO-MgO-FeO-Al2O3-SiO2 have been crystallized to garnetclinopyroxene bearing mineral assemblages in the range 24-30 kb pressure, 750°-1,300° C temperature. Microprobe analyses of coexisting garnet and clinopyroxene show that KD(Fe2+/MgG+/Fe2+/MgCpx) for the Fe-Mg exchange reaction between coexisting garnet and clinopyroxene is obviously dependent upon the Ca-content and apparently independent of the Mg/(Mg+Fe) content of the clinopyroxene and garnet. The Ca-effect is believed to be due to a combination of non-ideal Ca-Mg substitutions in the garnet and clinopyroxene. Our data and interpretation reconciles previous inconsistencies in the temperature dependence of KD- values determined in experimental studies of simple systems, complex basalt, grospydite and garnet peridotite compositions. Previous differences between the effect of pressure upon KDas predicted from simple system theory (Banno, 1970), and that observed in experiments on multicomponent natural rock compositions (Råheim and Green, 1974a) can now be resolved. We have determined KDas a function of P, T, and XGtCa(grossular) and derived the empirical relation {Mathematical expression}. This empirical relationship has been applied to garnet-clinopyroxene bearing rocks from a wide range of geological environments. The geothermometer yields similar estimates for garnet-clinopyroxene equilibration for neighbouring rocks of different composition and different KDvalues. In addition, temperature estimates using the above relationship are more consistent with independent temperature estimates based on other geothermometers than previous estimates which did not correct for the Ca-effect. An alternative approach to the above empirical geothermometer was attempted using regular solution models to derive Margules parameters for various solid solutions in garnets and clinopyroxenes. The derived Margules parameters are broadly consistent with those determined from binary solution studies, but caution must be exercised in interpreting them in terms of actual thermodynamic properties of the relevant crystalline solid solutions because of the assumptions which necessarily have to be made in this approach. © 1979 Springer-Verlag.
