The reversed phase equilibrium data of Koziol and Newton (1989) involving grossular (Gr)–pyrope (Py)–almandine (Alm) solid solutions have been analyzed in conjunction with standard state thermodynamic data of Berman (1988) in order to constrain ternary mixing properties for this system. The experimental data are consistent with a single P-T independent ternary interaction parameter, although its magnitude cannot be fixed independently of binary mixing properties. Sensitivity analysis of ternary excess properties to assumed binary interactions offers support for small excess enthalpy and entropy on the Gr-Alm join (Geiger et al., 1987; Koziol, 1990; Berman, 1990), in contrast to calibrations (Ganguly and Saxena, 1984; Anovitz and Essene, 1987) based on the phase equilibrium data of Cressey et al. (1978). The ternary dataare consistent with the strongly nonideal Py-Alm parameters of Ganguly and Saxena's (1984) preferred model only when a more positive ternary excess parameter is used than proposed by Ganguly and Saxena (−15 compared to −66.5 kJlmol on a 12 O atom basis). The experimental data are most compatible, however, with small nonideality on the Py-Alm join. The Py-Alm calibrations of Hackler and Wood (1989), Sack and Ghiorso (1989), and Berman (1990) are consistent with a ternary interaction parameter equal to zero.
The strong influence of the ternary Gr-Py-Alm interaction parameter in thermobarometry calculations is illustrated for several granulites. These calculations indicate that the ideal ternary interaction parameter consistent with available experimental data produces pressures that are lower by as much as 2.4 kbar compared with those computed with the ternary interaction parameter equal to −66.5 kJlmol. Temperatures differ by as much as 120 °C, with the direction of difference being dependent on specific garnet composition.