A discussion of “Mineralogical controls on garnet composition in the cratonic mantle” by Hill et al., Contributions to Mineralogy and Petrology (2015) 169:13

Abstract

opx, cpx, spinel and melt) that might occur in mantle peridotites. Unfortunately, in considering their results, the authors ignore data in the literature on actual variations in garnet compositions within documented peridotite samples. In these previously published cases, garnets show changes of compositions as a result of re-equilibration because of either changing P–T conditions, or because of metasomatism by melts, or both (e.g. Ivanic et al. 2012; Burgess and Harte 1999). We believe that actual examples of varying compositions in zoned garnets from peridotitic xenoliths point to a quite different explanation of the Hill et al. ‘harzburgite trend’ and simultaneously clarify reaction relationships for their ‘lherzolite trend’. Hill et al. (2015) claim to show the occurrence of the ‘harburgite trend’ for the first time, but they fail to consider other publications on mantle xenocrysts/xenoliths in which basically similar limiting trends of garnet compositions may be seen. Thus, Grutter et al. (2006, Figs. 7, 8) show similar limiting trends to garnet compositions running sub-parallel to the GDC (graphite–diamond constraint) in Cr2O3–CaO plots. Likewise in Fig. 1b (from Ivanic et al. 2012), a similar zone of garnet compositions running subparallel to the GDC line may be picked out for the garnet compositions in blue (from the Newlands kimberlite pipe, RSA) and in grey for the Kimberley pipes (RSA). In examining their ‘harzburgite trend’, Hill et al. (2015) take garnet compositions (marked in green in Fig. 1c) from both ends of the trend and calculate possible reactions for converting the garnet at one end of the trend to the garnet at the other end. [Note that there appears to be a change of terminology within the Hill et al. paper—the end-member garnet compositions referred to as “Lo-Cr Gar” and “Hi-Cr Gar” in their Fig. 4 (shown here as Fig. 1c), are referred to as “Lo-Ca Gar” and Hi-Ca gar” in their Table 2.] Several of the equations calculated by Hill et al. (2015, Table 2) for Hill et al. (2015) present evidence on the controls of garnet compositions in mantle rocks by considering the array of garnet compositions found in heavy mineral concentrates from the Canastra 8 kimberlites, Brazil (Fig. 1a). They identify two major trends of garnet compositions in their Cr2O3 versus CaO plot (Fig. 1a, Hill et al. 2015; see also Fig. 1a herewith). The first trend, referred to as the ‘lherzolite trend’ shows strongly increasing Cr2O3 with slightly increasing CaO and runs parallel to the G10/G9 separating common garnet compositions in harzburgite and lherzolite mantle xenoliths. The second trend, referred to as the ‘harzburgite trend’, shows markedly increasing CaO with moderately increasing Cr2O3. Hill et al. (2015) attempt to explain these two trends by taking garnet compositions from the ends of each trend (marked by the green compositions in Fig. 1c) and writing balanced equations for converting one ‘end-member’ composition into the other, using the potential compositions of associated phases (olivine,