Xenolith-derived mantle geotherms: whither the inflection?

Three possibilities exist for the geometry of the upper mantle geotherm determined from study of garnet lherzolite xenoliths from Cretaceous kimberlites of northern Lesotho, southern Africa: (1) the geotherm is inflected to a lower thermal gradient at greater depth (low-T inflection); (2) the geotherm is uninflected; and (3) the geotherm is inflected to a higher thermal gradient at greater depth (high-T inflection). In the past two years all three possibilities have been advocated. Finnerty and Boyd (1984, 1987) found that many independent thermometers yield similar P-T arrays, so that features of xenolith geotherms cannot be artifacts of the method of temperature estimation. Hence the current controversy centers on the barometers used for pressure estimation. Bertrand et al. (1986) calibrated a new aluminous-enstatite barometer using 50–100 kbar data of Yamada and Takahashi (1983), and presented a southern Africa geotherm displaying a low-T inflection. The high-P alumina solubility data are incompatible with lower-P data, however, with the result that the new barometer underestimates pressure: a diamond-bearing xenolith falls at least 5.7 kbar into the stability field of graphite. Thus, the Bertrand et al. (1986) barometer does not adequately test the reality of inflected geotherms. Carswell and Gibb (1987 a, b) modified the aluminous enstatite barometer of Nickel and Green (1985) to account for Jadeite molecule in orthopyroxenes containing relatively high concentrations of Na. When applied to xenoliths of northern Lesotho the apparent inflection is minimized but still evident. In this suite Na content of orthopyroxenes increases systematically with greater T or greater depth. Sodium correlates poorly with T (and depth) in a suite of xenoliths from Farm Louwrencia, Namibia, and application of the Nickel and Green (1985) barometer (with or without modification) destroys the correlation of T with P expected for a geotherm. The decorrelation of P from T in the Louwrencia suite is caused by errors in the Na correction. The minimization of the inflection in the northern Lesotho suite is caused by the correlation of Na with T (and depth) in that suite and does not result from an improved correction scheme for the aluminous enstatite barometer. Hence, the Carswell and Gibb (1987a, b) formulation of the barometer does not support the absence of an inflection in the northern Lesotho geotherm. Adams and Bishop (1986) recalibrated the olivine barometer (Finnerty and Boyd 1978) and presented a southern African P-T array that appears uninflected. Their barometer, however, underestimates pressure by 10–20 kbar: all xenoliths from the southern African diamond-bearing kimberlites plot well within the graphite stability field. This barometer is also too imprecise to judge whether an inflection is present or absent. Finnerty (1989), employing an empirical fit to data for Ca solubility in olivine and several different independent thermometers, presented northern Lesotho P-T arrays that satisfy the diamond-graphite constraint with minimal scatter and display high-T inflections. Because the inflection is evident with independent thermometers and independent barometers, it cannot be an artifact of the method of P-T estimation. The arguments contesting such an interpretation are flawed and so it is concluded that a high-T inflection is a fundamental property of the Cretaceous upper mantle geotherm beneath southern Africa.