Evaluation of thermobarometers for garnet peridotites

The accuracy and precision of a large number of combinations of geothermometers and geo-barometers for garnet lherzolites have been evaluated with a suite of well-equilibrated xenoliths from kimberlites of northern Lesotho. Accuracy was tested by comparison of P-T estimates for a diamond-bearing and a graphite-bearing xenolith with the experimentally determined diamond-graphite univariant curve and by comparison of P-T estimates for phlogopite-bearing xenoliths to the high-temperature stability limit of phlogopite (Eggler and Wendlandt, 1979). Precision was evaluated by measuring the scatter of P-T estimates for each of four xenoliths from a wide range of P and T when many point analyses of the constituent minerals are used for P-T estimation. A thermobarometer composed of the uncorrected diopside-enstatite miscibility gap of Lindsley and Dixon (1976), combined with the uncorrected isopleths for aluminum in enstatite coexisting with pyrope of MacGregor (1974), is most satisfactory. Correction schemes such as those of Wells (1977) and Wood (1974) will ultimately provide a better means of P-T estimation, but at the present stage of development they serve to decrease precision without a demonstrable increase in accuracy. Thermometers based on $\text{Fe}{}^{\mathrm{2+}}\text{Mg}$ exchange reactions are imprecise because of variable and unknown $\text{Fe}{}^{\mathrm{2+}}\text{Fe}{}^{\mathrm{3+}}$ in minerals and xenoliths. The inflection observed in the northern Lesotho paleogeotherm cannot be an artifact of the method of temperature estimation.