In several studies of chromite-rich cumulates in layered intrusions, the Cr/Al ratio of chromite varies significantly within centimetre-to-decimetre scale layering. In the G-H chromitite section of the Stillwater Complex (United States), Cr/Al ratios are consistently higher, by up to relative 20%, in the most chromite-rich rocks compared with immediately adjacent layers. Systematic relationships between Cr/Al ratios and chromite modes are also observed in cumulate rocks of the Bushveld Complex (South Africa). This is a critical observation for models of chromitite genesis involving mechanical sorting of chromite from silicate minerals in gravity flows; there is no plausible mechanism for sorting small chromite crystals on the basis of subtle variations in composition. A possible explanation could be in the combined effects of subsolidus re-equilibration and trapped liquid reaction. To test this possibility, a series of model calculations was carried out, solving for conservation of mass, relationships between distribution coefficients, composition and temperature, and stoichiometry. Although Mg/Fe ratios are indeed predicted to be highly mode-dependent, Cr# values (atomic Cr/Cr + Al) are insensitive to post-cumulus effects where chromite proportions are greater than 20 wt%. Variation comparable to natural data is only possible for chromite modes less than about 10 wt%. The difference in Cr# between rocks with 20–80 wt% and > 80 wt% chromite, and the short-range cyclicity of Cr# within individual massive chromitite layers, cannot be explained by post-cumulus processes. This variation, therefore, must reflect primary liquidus compositions. This argues in favour of in-situ boundary layer crystallisation models over ones invoking mechanical deposition and hence supports models of in-situ growth of chromitites from large convecting magma bodies.
Barnes et al. (Sun,) studied this question.