The eight Lower Group chromitite layers of the Ruighoek Pyroxenite in the area west of the Pilanesberg (LGl - LG7), on the farm Zandspruit 168 JP, were sampled in borehole cores drilled through the succession. The stratigraphic succession faIls within Cameron's (980) units B to E of the Critical Zone. The lowermost layer of the Lower Group, the LG 1 chromitite, is located some 440 metres below the MG 2 (Middle Group) chromitite layer, above which the first appearance of cumulus plagioclase in the Complex is seen. By convention, this horizon is designated the top of the Ruighoek Pyroxenite in the Western Bushveld Complex. Homogeneous units of chromite-bearing orthopyroxenite (bronzitite), exhibiting inconspicuous layering defined in terms of variations in orthopyroxene grain-size host all but one of the Lower Group layers; the LG 4 chromitite layer is exposed within an olivine-rich subunit 23 metres in thickness (C₃ subunit). The cumulative thickness of chromitite is 2,92 metres or 0,8 per cent of the studied section, which is 381 metres in thickness. Minimum and maximum thicknesses of the LGl - LG7 layers exposed in drill core are 17 and 81 centimetres, respectively, with minor chromitite layers ranging between 2 and 5 centimetres in thickness. Weighted mean Cr₂0₃ contents of units B to E vary between 1,17 and 3,22 per cent, with the latter estimate representative of the D₂ subunit which hosts the LG 6 chromitite layer. The LG 6 is correlated with the Steelpoort layer of the Eastern Bushveld Complex, and varies between 76 and 81 centimetres in thickness under a large portion of the farm Zandspruit. An undisturbed succession striking N15°E and dipping 12 - 15°E is depicted within the studied area, which is bounded on the eastern side by the north-striking Frank fault. Major folding of the layered succession is evident to the north of the area, where the layering adjacent to the trace of the fault dips 35° to the southwest. Particular attention is paid in the present study to (a) the nature of chromitite layers and their host rocks, (b) the contrast between the mineral chemistry of weakly disseminated chromite and grains within massive ore layers, (c) concentrations of Cr, V, Ni, Co, Sc and Ti in orthopyroxene in relation to stratigraphic height, and levels of Sr, Ba and Zr associated with hypothetically pure, intercumulus plagioclase feldspar, and (d) possible mechanisms which induce crystallization of chromitite layers containing 50 per cent Cr₂0₃ from magma with a Cr content of less than 1 000 ppm. Electron microprobe studies of chromite in relation to mineralogical and textural environment clearly reveal that (a) the proportions of Cr and Al cations are linked to paragenesis: higher Al/Cr ratios characterize olivine-bearing domains, whereas grains intergrown with plagioclase feldspar exhibit low AI/Cr ratios, and (b) Al contents rise with a decline in Mg/(Mg + Fe²⁺) from high values to a value of 0,450, then decrease with a further decline in Mg/(Mg + Fe²⁺). The paragenetically later trend is emphasized in a large population of chromite grains which escaped early encapsulation in orthopyroxene crystals and continued to grow in the environment of intercumulus plagioclase. Within-and between-sample compositional variation of grains in silicate-rich domains is modelled in terms of in situ growth increments, diffusive homogenization of zonal structures, and residence time within interstitial melt. Fractionation trends, as measured by Mg/(Mg + Fe²⁺) ratios in whole-rock and/or microprobe studies of orthopyroxene, are reversed in relation to stratigraphic height towards the top of the B unit and in the overlying C unit. These data are supported, for example, by lower vanadium contents and higher Ni/Sc ratios in hypothetically pure orthopyroxene. Small olivine crystals in chromite-rich domains are enriched in Ni relative to coarse-grained olivine in adjacent dunite: a feature attributed to early isolation of primocrysts from magma in the former case, and in situ equilibration between olivine crystals and Ni-depleted residual melt in the latter case . Similarly, rising Ni contents and Mg/(Mg + Fe²⁺) ratios of orthopyroxene with increasing stratigraphic height in the footwall of the LG 6 chromitite layer, linked to a progressive decline in orthopyroxene grain-size, are effects which may arise out of early separation of interstitial melt from orthopyroxene cumulates. A model is thus proposed which (a) links the thickness of chromitite layers to the vertical separation between successive layers or the thickness of fine-grained orthopyroxenite in the footwall, (b) ascribes copious nucleation of chromite to liquid mixing of this footwall derived, Cr - depleted contaminant with influxes of hot, primitive magma, and (c) tenders the notion that the present modal proportion of mesostasis in the footwall of a chromitite layer serves as a reciprocal measure of the volume of fractionated exudate

A brief review of the various models proposed to account for the Bushveld Complex shows that there are two main hypotheses. These are the Multiple Intrusion hypothesis and the In Situ Crystallization hypothesis. The latter also allows for multiple additions to the crystallizing magma, and several variants involving the number of these inputs , their composition, volume and timing have been proposed. To facilitate description and investigation of the study section, the stratigraphic nomenclature of this part of the Rustenberg Layered Suite is revised and clarified. It is proposed that the boundary between the Critical Zone and Main Zone be placed at the base of the Merensky cyclic unit, and thus the whole of the Merensky and Bastard cyclic units are included in the Main Zone. Furthermore, the extremely confused terminology for smaller units within the Merensky and Bastard cyclic units is resolved by discarding the term Reef as a formal term and substituting lithological terms such as Merensky pegmatoid, Merensky pyroxenite, Bastard pyroxenite and Merensky mottled anorthosite etc. It is recommended that the term Reef be retained as an informal term to designate the mineralized horizon which may be mined, regardless of lithology. The term "pegmatoid" is restricted to stratiform or lensoid masses of coarse grained feldspathic pyroxenite or harzburgite which are part of the layered sequence. The transgressive vertical pipe-like, coarse-grained ultramafic "iron-rich bodies are termed "ultramafic pegmatites ". The main features of the Merensky and Bastard cyclic units are the regular chemical and mineralogical changes that occur with respect to stratigraphic height in these units. In the Merensky cyclic unit there is a smooth iron enrichment in the orthopyroxenes upward in the succession and a transition from pyroxenite at the base to mottled anorthosite at the top of the unit. The Bastard cyclic unit is broadly similar to the Merensky cyclic unit. A variety of textures and chemical features are in disequilibrium in some samples but not in others, and great complexity is evident wh en individual samples are studied in detail. The initial ⁸⁷Sr/⁸⁶Sr ratios of plagioclase separates and whole rocks from the study section show a distinct step-like increase in the Merensky cyclic unit to .70806 at the base of the, Merensky cyclic unit to .70806 at the base of the Bastard cyclic unit. In contrast , samples from below the Merensky cyclic unit have a constant initial Sr-isotopic ratio, as do the samples from the Bastard cyclic unit. These isotopic and chemical data, and available published geologic relationships suggest that a major new influx of basic magma occurred after the Footwall unit was deposited and that this mixed with the residual magma in the chamber and then precipitated the Merensky and Bastard cyclic units. The crystal settling theory as outlined by Wager and Brown (1968) fails to account for the chemical and stratigraphic variations observed in the study section. The theory of bottom crystallization, initially proposed by Jackson (1961), more adequately explains the features observed. Applying a model outlined by Irvine (1980a & b), it has been established from chemical data, that the Merensky cyclic unit crystallized from a magma layer with a thickness roughly equivalent to the average thickness of the cyclic unit itself (±10m). A similar exercise on the Bastard unit was not possible. The formation of the Footwall unit is still enigmatic. Infiltration metasomatism and sintering can modify the petrographic and chemical characteristics of rocks and minerals after deposition at the liquidus stage. During the solidification of the crystal mush a separate vapour phase may form in the crystal mush, which could move up through the crystal pile. This process may ultimately be responsible for the generation of potholes and pegmatoidal horizons, such as the Merensky pegmatoid. The upward increase in the initial ⁸⁷Sr/⁸⁶Sr ratio within the Merensky cyclic unit is strong evidence that infiltration metasomatism has played an important part in the generation of the Merensky cyclic unit. This process, coupled with fluid enrichment, may also result in the formation of pegmatoid layers. Sintering appears to have been a common process in the mottled anorthosites of the study section and may have severely reduced the amount of trapped interstitial liquid in these rocks.

This study is an account of the stratigraphic sequence, the petrography, mineralogy (microprobe investigations of orthopyroxene, clinopyroxene, olivine and plagioclase feldspar), and whole-rock major- and traceelement geochemistry of the silicate cumulates of the Upper Critical Zone in the western Bushveld Complex. Two parts of the study - an investigation of a 350m column incorporating the MG3 and UGI Footwall Units, and a comparison of two additional Upper Critical Zone profiles with a previously compiled profile between the UGI and Bastard Units - are focused on RPM Union Section in the northwestern sector of the Complex. The third part is a detailed vertical and lateral investigation of the Bastard Unit at the top of the Critical Zone, which draws on sampling and data compilation from seventeen profiles in the western limb of the Complex. The MG3 Unit (45m) is made up of a lower chromitite layer overlain by a norite-pyroxenite-anorthosite sequence while the UGlFW Unit (295m) is composed of a related series of lower chromitite layers (MG4) overlain by a pyroxenite-norite-anorthosite sequence capped by the UGI chromitite layer. These mafic cumulates display a distinctive pattern of oscillating cryptic variation in whole-rock Mg/(Mg+Fe), FeO/Ti0₂, Cr/Co and Ni/V ratios through the sequence. Sympathetic oscillations are recorded for compositions of orthopyroxene and plagioclase feldspar and eight subcycles are recognised through the UGlFW Unit. The entire sequence is characterised by the presence of small, spheroidal, embayed and irregularly shaped plagioclase grains which are poikilitically enclosed in cumulus orthopyroxene grains of both pyroxenites and norites. This texture is indicative of partial resorption of pre-existing feldspar primocrysts within the melt prior to their being incorporated into the host orthopyroxene grains. Textural, geochemical and isotopic data suggest that this sequence was built up by periodic additions of fresh, relatively primitive liquid into fractionated resident liquid, and subsequent mixing within the magma chamber. The Bastard Unit sequence, described in Chapter 4, is the last and most complete cyclic unit (c. 60m) of the Critical Zone, and its upper contact defines the boundary between the Critical and Main Zones of the Complex. This Unit can conveniently be sub-divided into a lower part, where orthopyroxene occurs as a cumulus phase, and the upper part which is composed entirely of anorthosite (Giant Mottled Anorthosite). The basal part of the Unit (≤ 18m) comprises a thin chromitite layer < O.5cm) overlain by a pyroxenite-melanorite-norite-leuconorite sequence. The basal pyroxenite is orthocumulate in character and rapidly gives way to norites and leuconorites. A distinct threefold subdivision emerges within the Giant Mottled Anorthosite which is predominantly an adcumulate which becomes orthocumulate in character at its top. Apart from minor deviations in thicknesses these lithologies are recorded over the entire strike-length covered in this study. Profiles of cryptic variation are compiled for orthopyroxene, plagioclase and whole-rock data and show that the Bastard Unit displays a characteristic pattern which is maintained throughout the western Bushveld Complex. A minor yet distinctive reversal in cryptic variation is revealed at a level which is stratigraphically variable within the lower Giant Mottled Anorthosite, and results in a double cuspate pattern. A remarkable feature of the basal Bastard pyroxenites is that although the modal proportion of mafic to felsic constituents varies systematically away from the northwestern sector, the Mg/(Mg+Fe) ratio of orthopyroxenes remains constant at 0.804 over a lateral strike distance of 171km. Within the upper part of the Unit the orthopyroxene is markedly Fe-rich and it is here that inverted primary pigeonite appears for the first time as a cumulus phase. In addition, K-feldspar, oscillatory zoned plagioclase grains and high levels of incompatible trace elements are noted at this level. On the basis of the data presented it is concluded that the Bastard Unit represents the crystallisation of a final, relatively large influx of hotter primitive liquid, with upper Critical Zone affinities, and subsequent mixing with a column of cooler (less dense) supernatant liquid which had in part hybridized with the overlying Main Zone magma. It is hypothesised that this new liquid was emplaced as a basal flow beneath supernatant liquid and that it initiated the deposition of mafic cumulates at its base. The supernatant liquid is interpreted as representing the fractionated residuum produced by crystallisation of earlier cyclic units, with plagioclase on the liquidus, and that it contained an abundance of small plagioclase primocrysts in suspension. Development of the Unit can be viewed as a two-stage process. In the lower half of the unit, chemical and physical parameters typical of the new magma dominated the crystallisation process, and resulted in cumulates very similar to other relatively complete Upper Critical Zone units. In the upper, leucocratic sequence, above a minor reversal, crystallisation was from a liquid which was the product of mixing of a minor pulse of primitive liquid with the reservoir of hybridized supernatant liquid. Although the Bastard Unit is not continuous around the entire Western limb of the Complex, it is concluded that it developed in a single, or connected, magma chamber and that its irruptive feeder zone was located in the proximal northwestern facies of the Complex.