GOLDSCHMIDTCONFERENCE TOULOUSE 1998 The formation of the PGE bearing Bushveld chromitites and the Merensky Reef by magma mixing: a combined Re-Os and Rb-Sr study R. Schoenberg F. J. Kruger J. D. Kramers The Critical Zone of the Bushveld Complex is characterized by repeated cyclic crystallization series (Cyclic Units), each comprising a chromitite layer, enriched in PGE-bearing sulphides and platinum metals, at its base. It is now widely accepted, among workers in the Bushveld Complex, that injections of fresh melt followed by mixing with residual liquid forced the composition of the crystal- lizing hybrid magmas into the stability field of chromite (Scoon et al., 1994 and references therein) and thus lead to the accumulation of prominent chromitite layers. However, details of this process, like the composition of the magmas involved and the origin of the crustal component displayed by initial isotope ratios (particularly Sr isotopes, Kruger and Marsh, 1982) are still under investigation. From inverse geochemical fractionation trends (e.g. Mg number of whole rocks and pyroxenes, Ni/Yi of pyroxenes) at stratigraphic levels, where new melt was injected, Eales et al. (1990) and others concluded that the new influxing magma must be of primitive composition. Based on immediate increases in initial Sr ratios at the same stratigraphic horizons, Kruger & Marsh (1982) and others concluded that the new magmas were derived from a partial crustal source or were crustally contaminated prior to the influx into the magma chamber. In contrast to chromite accumulation, the PGE enrichment within chromitites is still ascribed variously to (1) the orthomagmatic model: PGE are collected by an immiscible sulphide melt, being formed after magma mixing, (2) the fluid flow model: PGE are transported by upward-migrating hydro- thermal fluids into the chromitites and platinum reefs (Ballhans et al., 1986) and (3) the chromitite control model: PGE are concentrated in sulphide-poor chromitites, probably due to local S saturation (Scoon et al., 1994). Gruppe Isotopengeologie, Mineralogisch-Petrographisches Institut, Universitiit Bern, Erlachstrasse 9a, 3012 Bern, Switzerland Hugh Allsopp Lab., University of the Witwatersrand, Wits 2050, South Africa Gruppe Isotopengeologie, Mineralogisch-Petrographisches Institut, Universit~it Bern, Erlachstrasse 9a, 3012 Bern, Switzerland In order to test these hypothesis and to investigate the correlation between magma mixing and crustal contamination, we used combined Re-Os and Rb-Sr data on chromite and gangue (interstitial phases in chromitites) samples from the Critical Zone and the Merensky Reef, as well as pyroxenites from the overlying Merensky Cyclic Unit (Fig. 1). Os concentrations in gangues exceed those in associated chromites by up to 30 fold, demonstrating the presence of the Pt-alloys as interstitial phases. Chromite concentrations of different dry splits from the same sample may vary as much as 300 %, which is most probably caused by micro-nuggets and/or micro- inclusions of laurites. Initial 187Os/lSSOs ratios of chromites as well as gangues of the lower group chromitites lie only slightly above the contemporaneous mantle value (i.e. 0.113 at 2061 Ma). A general increase of initial Os isotopic composition with stratigraphic height culminates in the value of interstitial feldspathic pyroxenite of the Merensky Reef (sample AE-26), which lies within the range of laurites determined by Hart and Kinloch (1989). However, they also measured two erlichmanite grains (OsS2), which perfectly reproduce the mantle value at that time. Four feldspathic pyroxenites, situated at different levels within the Merensky Cyclic Unit, define a Re-Os isochron (Fig. 2), yielding an age of 2043 _+ 11 Ma (MSWD = 0.707), which agrees within error with the generally accepted Rb-Sr age of 2061 _+ 27 Ma. Offsets in initial Os isotopic composition between chromite and gangue samples, as observed throughout the upper Critical Zone may be the result of magma mixing or in situ crustal contamina- tion of a single magma, while crystallization is already in progress. Decreases in initial Os isotopic compositons of the UG1 layer and the Merensky Cyclic Unit samples relative to underlying horizons, contradict bulk in situ crustal contamination of a 1349