Coarse-grained, discordant, mainly ultramafic, bodies cut the layered succession of the Bushveld Complex. Previous models have subdivided them into magnesian dunite, considered to have been emplaced upward, and more ironrich wehrlite (± oxides), referred to as iron-rich ultramafic pegmatites (IRUP), emplaced downward. We have studied a number of these IRUP bodies in an open pit on Marikana Mining Operations platinum mines property, near Rustenburg, as well as other samples from different levels in the intrusion. A compilation of our and previously published compositions for olivine and clinopyroxene suggests that there are not two discrete suites and processes, but that all bodies originated from multiple magmas with a similar source, resulting in formation of magnesian olivine followed by progressively more iron-rich olivine and clinopyroxene in a fractionating system. In fact, more evolved compositions of discordant pegmatitic gabbro mineralogy exist that have been largely ignored in the past, but are considered here to be cogenetic with the ultramafic bodies. These rocks are all cumulates, based on their simple mineralogy and extreme deficiency in incompatible trace elements in whole-rock compositions.
We question the previous models. The first involved fluid replacement. We question from where the high TiO2 content of these bodies originated and to where the high Al2O3 (from the replaced plagioclase) went, noting that TiO2 and Al2O3 are typically the most refractory of oxides. The second involved highly evolved or immiscible liquids or products of disequilibrium melting, generated within the intrusion, considered to have penetrated downward (kms) displacing a succession of cumulate rocks. We question the mechanism by which a downward penetrating magma could displace pre-existing cumulates, presumably returning them back up the same conduit without any interaction between the two. We suggest that initial upward penetration of magmas, not generated within the layered succession, replaced the existing cumulates initially by infiltration. Subsequently, vertical central channels developed and large volumes of magma flowed through, precipitating cumulus minerals. The removed components and the residual liquids from formation of these discordant cumulate bodies were expelled upward. The composition of the proposed liquid for these bodies is considered to have affinities with sub-alkaline basalt (ankaramite), which occurs in many large igneous provinces, and whose distinctive features are the dominance of olivine and clinopyroxene and absence of plagioclase, and have a high CaO/Al2O3 compared to tholeiitic magmas. Support for this conclusion is based in part on the clinopyroxene compositions in IRUP, which are enriched in the wollastonite component and do not plot near to the tholeiite or Bushveld trend in the pyroxenite quadrilateral. They contain no exsolution which is abundant in all layered rocks in the Bushveld Complex. Examples of such sub-alkaline rocks related to the Bushveld Complex have been previously reported. One (Schoongezicht intrusion), distal from, and at a much deeper stratigraphic level (base of the Ventersdorp Supergroup) than, the complex, is dominated by dunite, wehrlite and magnetitite, comparable to the IRUP.