A major geochemical hiatus occurs in the Bushveld Complex at the level of the platiniferous Merensky reef, close to the Critical-Main zone boundary. The origin of this hiatus and its relationship to mineralization has not been fully resolved. Initial Sr isotope ratios for plagioclase and Cr/MgO values for orthopyroxene from Critical and Main zone rocks are significantly different. Geochemical data through the Merensky and Bastard cyclic units indicate that orthopyroxene in the Merensky and Bastard units originated from magma with a composition like that of the Critical zone magma. It sometimes occurs together with plagioclase that originated from Main zone magma. In a detailed study of the eastern limb, the pyroxenite at the base of the Merensky unit contains plagioclase and orthopyroxene that have compositions typical of the Critical zone, but in the overlying part of the Merensky unit, plagioclase compositions are more typical of the Main zone, whereas orthopyroxene retains a Critical zone signature. Similarly, in the Bastard pyroxenite, plagioclase has a range of Sr isotope values between that of the Main and Critical zones, but orthopyroxene has values typical of the Critical zone. These observations of mineral disequilibrium clearly show that the two major minerals in the Merensky and Bastard units were formed from two different, but coexisting, magmas. A model that accounts for this disequilibrium is proposed here. It invokes the influx of Main zone magma at the base of the Merensky unit during which the Critical zone magma was displaced upward, but the two magmas did not mix. The elevated Critical zone liquid continued to crystallize orthopyroxene, which sank through the influx of magma of the Main zone, due to the density contrast. These grains collected on the crystal pile to form the Merensky pyroxenite. The Main zone magma, into which the orthopyroxene accumulated, crystallized interstitial plagioclase that had an Sr isotope ratio typical of the Main zone.
Whole-rock, major element geochemical data show that a variable proportion of the plagioclase in both the Merensky and Bastard pyroxenites is cumulus. It is inferred to have accumulated with orthopyroxene and has an initial Sr isotope ratio typical of the Critical zone. Thus the two pyroxenites now have a mixed Sr isotope signature of cumulus (Critical zone) and intercumulus (Main zone) plagioclase that varies along strike. Above the pyroxenites, the Sr signature of the norites and anorthosites of both cyclic units is dominated by cumulus plagioclase from the Main zone magma, but some plagioclase with a Critical zone signature may be present. We conclude that the variations in initial Sr isotope ratios do not result from mixing of magmas, but from variable accumulation of orthopyroxene and plagioclase from a higher, isotopically distinct layer of magma into an underlying layer. The Merensky and Bastard cyclic units therefore may have features of either Critical or Main zone magma depending on which chemical parameter (Cr/MgO and Sr isotopes) is considered. They are therefore classified as a transitional unit between the Critical and Main zones. Models that suggest magma mixing was responsible for sulfide saturation and the creation of an immiscible sulfide liquid are not supported by this study and alternative models must be sought to explain the sulfide-saturation event in the Merensky reef.