The Stillwater Complex was emplaced at 2.7 Ga at a depth of 10-15 km. Laramide deformation and subsequent erosion have exposed the basal hornfels and a thick, but incomplete, section of the complex. The complex is divided into a Basal Series, an Ultramafic Series and a Banded Series. The Basal Series forms an irregular sheet-like mass composed of early-formed cumulate rocks and coeval sills of diabase and sulphide-rich mafic norite. The Ultramafic Series is made up of cycles of harzburgite and bronzitite with conformable layers of chromitite in the lower parts of each cycle. The cyclic units and absence of cryptic variation in the Ultramafic Series indicate repeated influxes of magma and venting of fractionated magma from the chamber. The Lower and Upper Banded Series are composed primarily of norite and gabbronorite while the Middle Banded Series is predominantly anorthosite, troctolite and olivine gabbro. Cryptic variation in the Lower and Upper Banded Series is consistent with crystal fractionation and accumulation. In contrast, plagioclase in the Middle Banded Series has a uniform composition throughout which is consistent with a sorting mechanism in which plagioclase crystals were suspended in a convecting magma for extended periods. In the Lower Banded Series, a PGE-rich sulphide zone (J-M Reef) is associated with the reappearance of olivine, most likely due to a magma influx. Two models have been proposed for the origin of the reef. The magmatic model holds that sulphides became enriched in PGE during batch segregation of an immiscible sulphide melt which came into contact with a large volume of silicate melt during magma mixing. In the hydromagmatic model, it is proposed that PGE enrichments are the result of leaching of underlying cumulates by Cl-rich hydrous fluids exsolved during the late stages of intercumulus crystallization. Crystallization sequences reveal that at least two compositionally distinct magmas were involved in the formation of the complex. The magma that formed the Ultramafic Series was rich in both MgO and SiO2 and appears to be related to the mafic norite sills. The magma that formed the Middle Banded Series was tholeiitic in character. Isotopic and trace element compositions preserve a record of crustal interaction. Addition of older continental crust to the mantle source regions via subduction is the most plausible contamination mechanism.
The Bjerkreim-Sokndal Layered Intrusion is a large (~230 km2), discordant, Late Proterozoic, post-orogenic pluton in the Egersund-Farsund Igneous Province. The intrusion was emplaced shortly after massif-type anorthosite plutons and is cut by jotunite dykes. It contains a >7000 m thick Layered Series consisting of rocks belonging to the anorthosite kindred: andesine anorthosite, leuconorite, troctolite, norite, gabbronorite, mangerite, and quartz mangerite. Cumulates in the Layered Series are organized in 6 megacyclic units (MCU 0 to IV), individually up to 1800 m thick, but varying considerably in thickness and development along strike. The highest-temperature cumulates are troctolites containing plagioclase of ~An54 and olivine of ~Fo77. Phase contacts in the macrocyclic units reflect crystallization of the silicate minerals in the order plagioclase (± olivine), orthopyroxene, Ca-rich pyroxene, pigeonite. Ilmenite crystallized early and apatite appeared as a cumulus mineral at about the same time as Ca-rich pyroxene. Cumulus magnetite followed orthopyroxene and preceded Ca-rich pyroxene in MCU III and IV, but crystallized after Ca-rich pyroxene in MCU IB. MCUs 0, IA and II do not contain cumulates with cumulus magnetite or Ca-rich pyroxene. Olivine (~Fo50) reappears in the uppermost part of the Layered Series where there is a rapid stratigraphic transition to mangerite and quartz mangerite. The basal parts of MCUs III and IV are characterized by thin sequences of plagioclase, plagioclase-orthopyroxene-ilmenite and orthopyroxene-ilmenite cumulates in which there are systematic upward decreases in initial Sr isotope ratios. They are overlain by troctolite (plagioclase-olivine cumulate) and are believed to have crystallized from hybrid magmas. The MCUs, the discordant geometry of phase contacts, the stratigraphic variations in initial 87Sr/86Sr ratio (0.7049-0.7085), and the abundance of xenoliths suggest crystallization of the cumulates at the base of a periodically-replenished, compositionally-zoned magma chamber that was continually assimilating country rocks. The parent, as indicated by medium-grained jotunite along country-rock contacts, appears to have been an evolved, Ti-rich magma similar to ferrobasalt, but poor in diopside components. Systematic stratigraphic variations in initial 87Sr/86Sr ratio at the base of MCU III and MCU IV suggest that influx of magma
The Great Dyke of Zimbabwe is unique within the family of large layered intrusions by virtue of its highly elongate form. Apart from the tectonic controls that gave rise to the series of linked magma chambers which together comprise this intrusion, the width to length ratio profoundly affected the layering style, rock-types, mineral compositions and the form of mineralized ore bodies. The intrusion developed as a series of initially isolated chambers which became linked at progressively higher levels during the filling process. The dynamic interplay of crystallization and magma emplacement gave rise to the succession of cyclic units within the ultramafic sequence. The entire length of the Great Dyke (some 550 km) was linked at a level corresponding to the top of the Ultramafic Sequence and at this stage influxes of new magma effectively ceased. The initial magma of the Great Dyke was high magnesian (15.6% MgO), relatively enriched in silica, but with low initial 87Sr/86Sr indicating low crustal contamination. The various primary processes of magma mixing resulting from emplacement of new magma into an expanding chamber gave rise to economically important chromitite layers, while fractionation combined with influx of magma caused the formation of base metal sulphides enriched in platinum group elements. This paper considers the following aspects of the Great Dyke: its tectonic setting, structure, form and development of the magma chambers, initial magma composition, emplacement of magma, crystallization and fractionation, and mineralization.
Convection in magma chambers is driven by small density differences that originate at the margins of a magma chamber or when a new pulse of magma enters a chamber. Buoyancy anomalies at the margins of magma chambers result from cooling or crystallization at the floor, roof or walls of the intrusion. Cooling produces a thermal boundary layer which is typically between 10 cm and 1 m wide with the temperature drop across the layer between 0.05 and 1°C. Compositional boundary layers, produced by crystallization, are much thinner than thermal boundary layers and are no more than a few millimetres wide. The compositional step across them normally lies between 0.6 to 12 wt%. Calculated thermal and compositional flux Rayleigh numbers, assuming convection over the full depth range of the chamber, are typically greater than 1012 and 1019 respectively, well above the critical value of 106 that marks the transition from laminar to turbulent convection. Laminar or cellular convection is only possible in a convecting layer if its depth is less than 10 cm. A new pulse of magma entering a chamber may have a density that is less than or greater than the fractionated magma in the chamber. If it is light it will rise to the top of the chamber as a plume. If it is dense it will form a fountain. In both cases the flow will be turbulent and the input magma will mix extensively with the fractionated magma in the chamber, leading to stratification. If the input magma is hotter than the fractionated magma, the stratified hybrid zone produced at the floor of the chamber by a fountain will consist of hot, compositionally dense magma overlain by cooler, compositionally lighter magma. Because the distribution of heat is unstable the hybrid layer will break up into double-diffusive convecting layers. A plume of hot, light magma will produce a hybrid zone at the top of the chamber that is stably stratified with respect to both temperature and composition. The magma will remain stably stratified until heat loss to the surroundings can overcome the stable density gradient and convection can recommence. Crystallization, dissolving, or melting at the floor or roof of the chamber can also lead to stratification of the magma. A light melt released at the floor, by any of these processes, has an homogenizing influence on the overlying magma whereas light magma released at the roof stratifies the magma at the top of the chamber. The release of a dense magma has the reverse effect; it stratifies the magma if formed near the base of the chamber but has an homogenizing influence on magma if formed near the roof. Where melting or dissolving of the roof produces a light magma it will pond against the roof and the chamber will divide into two layers separated by a double-diffusive interface. Much of the heat required to melt the roof is provided by the latent heat released by crystallization at the floor and it is transmitted across the interface by diffusion. However, little mass is transferred across the interface. That is, assimilation-fractional crystallization is not an important process in basaltic magma chambers.
The stratiform 2.8 Ga Windimurra Complex (2300 km 2 ) has a total thickness between 13 km (field stratigraphy) and 5 km (gravity modelling). The complex is surrounded by younger granitoids with sheared contact zones, and the roof is not exposed. Phase layering and cumulus mineral compositions show upwards fractionation, and allow recognition of the following subdivisions: o 1) Ultramafic Series (UMS, 0.5 km thick), serpentinized olivine (Fo 90 ?) - chromite cumulates; 2) Lower Series (LS, 6-11 km thick), mainly anorthositic leucogabbronorites with olivine gabbroids increasingly abundant upwards, with cumulus plagioclase (An 85-64 , 76 vol%), augite ( mg 87-67), orthopyroxene ( mg 85-61), olivine (Fo 80-50 ), and cumulus magnetite 2 km below the top, where mg = 100xMg/(Mg+Fe Total ); 3) Middle Series (MS, 1.5 km thick), mainly magnetite gabbronorites with cumulus plagioclase (An 58 , 57 vol%), augite ( mg 64), inverted pigeonite ( mg 56), magnetite, and ilmenite; 4) Upper Series (US, <1 km thick), mainly magnetite olivine gabbros with cumulus plagioclase (An 57 , 37 vol%), augite ( mg 56), olivine (Fo 35 ), magnetite, and ilmenite. The extensively explored and researched Lower Series dominates the complex, and poses many problems. Stratigraphic correlation of the various outcrop areas is uncertain due to scarcity of markers, lateral changes and discordances, non-outcrop, and faulting. Thickness estimates vary from 3 to 11 km depending on the method. Cumulus olivine increases upwards and orthopyroxene decreases, but both tend to appear and disappear irregularly. Anorthositic leucogabbronorite (80–90 vol% plagioclase) is the main rock type, and troctolite, olivine gabbro, and gabbro are common in the upper parts. Overall, mineral compositions show slight reverse fractionation initially, then remain roughly constant and show slight normal fractionation, but fluctuate irregularly in detail. Modal and phase macro-layering are well developed in the LS. Olivine-free modal macrolayers (gabbronorite - anorthositic leucogabbro-norite) are much more planar, uniform, and laterally extensive than olivine-bearing macrolayers. Anorthositic leucogabbronorite layers are commonly more mesocumulate compared with the adjacent adcumulate mafic gabbronorite layers. Imperfect cyclic layering, commonly reflecting the crystallization order plagioclase + olivine first, then augite, and orthopyroxene last with olivine disappearing, becomes more obvious upwards in the LS. Two 500 m thick intervals of magnetite-rich leucogabbroids are present in the LS - the Canegrass Magnetite Zone (CMZ), which is partly discordant to the upper LS, and the enigmatic Shephards Discordant Zone which may be equivalent to CMZ. The Windimurra Complex is probably one of the most discordantly layered stratiform intrusions, and shows angular discordances between layers, structures resembling unconformities and troughs, non-planar layers, layer terminations, poorly layered intervals with heterogeneous “mixed rocks”, lateral changes in modal proportions (olivine particularly) and mineral compositions, and angular discordance between the magnetite-in horizon and modal layering.
In the past three decades, the cumulus terminology developed by Wager and co-workers has provided the framework for understanding texture development in crystal mushes. Much of the debate has concerned the conditions necessary for development of adcumulate rocks and has involved discussion of mechanisms of heat and mass transfer within mushes. In this article the historical development of ideas is reviewed and aspects of the nomenclature are discussed. The development of primary and secondary textures in mushes are then discussed, principally with respect to the relative roles of crystal overgrowth, compaction, and cementation. Most crystal accumulation in moderate- to large-sized layered intrusions occurs on the floor, where crystal mushes develop by either in situ crystallization or crystal sedimentation. Except where a preferred crystal shape orientation occurs as a result of directional growth from a substrate, there are no definitive textural criteria for distinguishing in situ crystallization from crystal sedimentation in the accumulation of mushes. Mushes inherit primary textural characteristics that influence the subsequent texture development within the crystal pile. Primary porosity and permeability are influenced by initial packing and clustering characteristics of crystals which are a function of the way in which crystals accumulate and any subsequent mechanical reorganization. Crystal growth, solution/replacement, cementation, compaction, and recrystallization are competing processes involved in the secondary texture development of the crystal pile. The densification of a crystal mush involves the reduction of primary porosity of the cumulus grains. This may be by overgrowth on the grains or compaction. Either process will be restricted by the nucleation and growth of poikilitic grains which cement the granular crystal framework. These processes are analogous to syntaxial overgrowth, compaction and cementation involved in sediment diagenesis. Whether crystals grow under near-isothermal conditions or during cooling depends upon whether the mush is open or closed to melt percolation but is independent of the mechanism of heat and mass transfer within the mush. Compaction, necessarily an open-system process, involves deformation (dislocation creep) or solution/reprecipitation of grains (diffusion creep) and usually results in an increase in the degree of local textural equilibration. However, recrystallization (e.g. by thermal annealing) also results in textural equilibration. Growth, compaction, and recrystallization are all competing processes and it is commonly not possible to isolate their contribution to any given texture; all produce rocks with the textural characteristics of adcumulates. The extent of densification of a mush of cumulus grains depends critically on the timing of nucleation and growth of poikilitic cementing phases. In any given magma composition this is a function of the local phase relationships. A cyclicity will develop in the texture in a crystal mush that is a function of the balance of densification and poikilitic cementation. Repeated replenishment of a magma chamber may result in suppression of the cementation cycle and allow mushes to become highly densified. On the scale of an intrusion the texture which develops depends upon the interaction of fronts of densification and cementation and hence is dependent on intrusion geometry.
The Duluth Complex and associated subvolcanic intrusions comprise a large (5,000 km2) intrusive complex in northeastern Minnesota that was emplaced into comagmatic volcanics during the development of the 1.1 Ga Midcontinent rift in North America. In addition to anorthositic and felsic intrusions, the Duluth Complex is composed of many individual mafic layered intrusions of tholeiitic affinity. The cumulate stratigraphies and cryptic variations of six of the better exposed and better studied intrusions are described here to demonstrate the variability in their cumulus mineral paragenesis. Although the general paragenetic sequence is: O1(±Pl) → Pl+Ol → Pl+Cpx+Ox(±Ol±Opx) → Pl+Cpx+Ox+Ap(±Ol) considerable differences exist among the six intrusions in the relative order and timing of cumulus arrivals, most notably with regard to augite (Cpx) and Fe-Ti oxide (Ox). The various cumulate stratigraphies and cryptic variations represented by the six intrusions described here largely reflect differences in the degree of open-system behaviour to recharge, eruption, and country-rock assimilation, but also may have been influenced by differences in parent magma composition, in the efficiency of fractional crystallization, and in the conditions of crystallization (e.g. Ptotal, fO2 PH2O). The Sonju Lake intrusion, wherein cumulus augite arrived before ilmenite, formed by essentially closed-system fractional crystallization. However, its compositional evolution may have been affected to an uncertain degree by assimilation of a granitic hanging-wall. The monotonous, thick troctolitic cumulate sequences of the lower parts of the South Kawishiwi and Partridge River Intrusions appear to represent the effects of frequent magma recharge coupled with in situ (or boundary layer) crystallization though other interpretations have been proposed. Extensive Cu-Ni sulphide mineralization at the base of these intrusions is attributed to country-rock contamination of the earliest intruded magmas. The cyclical progression of cumulates in the Layered Series at Duluth, wherein cumulus augite and oxide arrived nearly simultaneously, formed in a moderately open system characterized by periodic eruption and recharge. The Wilder Lake Intrusion is different from other intrusions in that cumulus ilmenite appears before augite and that olivine and augite composition define an inverted cryptic variation. The latter phenomenon may be related to a strong upward gradation toward lesser amounts of trapped liquid in the cumulates. Finally, the incomplete cumulate stratigraphy of the Bald Eagle Intrusion, which is composed of approximately equal thicknesses of troctolite (PO) and gabbro (PAO) adcumulates, is unique in that Fe-Ti oxide did not arrive as a cumulus phase despite prolonged crystallization of cumulus augite. These different cumulus parageneses probably produced a variety of derivative magmas, which may have contributed in part to the compositional diversity of the Midcontinent rift volcanics.
Palaeocene igneous activity in the Rum Central Complex culminated in the formation of the ultrabasic and gabbroic rocks of the Layered Suite. Its three components, the Eastern Layered Series, the Western Layered Series and the Central Series, represent a continuum in time during which replenishments of picritic (MgO 15-20 wt.%) and basaltic magmas ponded in thin sill-like bodies at the Lewisian gneiss - Torridonian sandstone unconformity, each contributing incrementally to a layered cumulate sequence. The magmas were probably guided during ascent by the long-lived Long Loch Fault. Peridotite (olivine - chrome-spinel) cumulates formed from picritic magma. The residual (basaltic) magma mixed with resident residual magma from earlier batches, and with small amounts of siliceous rheomorphic melts from country rocks, forming (isotopically contaminated) allivalitic (= troctolitic), plagioclase - olivine cumulates or, less commonly, gabbroic (plagioclase - olivine - clinopyroxene) cumulates. Residual basaltic magma was probably also intruded as gabbroic sheets and plugs, and extruded as lavas. Widespread slump and shear structures indicate mechanical instability of unconsolidated cumulate mushes, especially in the allivalites. Ultrabasic breccias are common in the Central Series, and are attributed to (i) disruption of earlier cumulates as new batches of magma rose along an elongate, north-south feeder zone and (ii) collapse of cumulates into this zone during episodic magma withdrawal. Equilibrated textures, lack of compositional zoning in olivine and pyroxene, offsets between compositional and modal variation at unit and other lithological boundaries, the occurrence of finger structures and other replacement features, and the compositional modification of ultrabasic rocks adjoining late-stage gabbroic veins, all attest to the pervasive influence of migrating intercumulus liquids during crystallization and consolidation of the cumulates.
The 160 km 2 Fongen-Hyllingen Complex, situated 60 km southeast of Trondheim, Norway, is a synorogenic, layered mafic intrusion of Caledonian age. Lateral correlation along 40 km of strike length through ca. 6 km of layered cumulates has allowed reconstruction of the form and evolution of the magma chamber. The chamber has the shape of a double-bottomed bowl with an asymmetrical, wide lip which extends towards the south. The northern (Fongen) part of the intrusion occupies the deepest part of the chamber. The southern (Hyllingen) part occupies the lip-like protrusion. Initial 87 Sr/ 86 Sr ratios (Sr 0 ) in cumulates range from 0.70308 to 0.70535 and reflect mixing between uncontaminated replenishing magma (with ca. Sr 0 0.70308) and resident magma contaminated by partial melts of metapelitic country rocks. Mineral chemistry and Sr 0 show a good correlation through most of the layered series, with more evolved mineral compositions having more contaminated isotopic signatures. Assimilation and fractional crystallization therefore were linked during much of the magma chamber evolution. The intrusion, which has a dioritic bulk composition, is divided into four evolutionary stages based on mineral chemistry. Stage I comprises a basal reversal. Stage II has fairly constant compositions in any profile normal to modal layering and is characterized by the presence of numerous, large, raft-like, dominantly metabasaltic, inclusions. Stage III consists of a major compositional regression, ending with the most primitive assemblages in the intrusion (with olivine Fo 75 , plagioclase An 63 ). Stage IV comprises a sequence showing strong normal fractionation, ending with quartz-bearing syenites with low-temperature end-member mineral compositions at the roof. A major feature in the Hyllingen Series is the presence of systematic lateral compositional variations in mineral chemistry, with more evolved compositions along the strike of modal layering approaching the southern margin. These discordant relations between modal and cryptic layering developed as a result of the crystallization of compositionally zoned magma along an inclined floor. The upper part of the magma chamber became zoned with respect to Sr-isotopes and chemical composition. The lower part remained isotopically homogeneous but chemically zoned until progressive mixing took place with new, uncontaminated magma during the formation of Stage III. Extensive compositional zoning developed both from the roof down (by the generation of roof melts) and from the floor up (by repeated, quiescent emplacement of dense magma along the floor and by mixing as a result of new, dense magma fountaining into buoyant, resident magma).
Thanks to its magnificent exposures and extraordinarily complete sequence of strongly differentiated rocks, the Skaergaard Intrusion has long served as a prime example of shallow magmatic differentiation and as a testing ground for a wide range of petrologic concepts. During the magmatic episode accompanying the opening of the North Atlantic about 55 Ma ago, a moderately evolved tholeiitic magma was intruded, apparently in a single, prolonged pulse, into Archean gneisses and Tertiary basalts close to the eastern edge of Greenland. Over a period of about 10,000 years, the Layered Series crystallized on the floor, while similar sequences crystallized on the walls and under the roof to form the Marginal and Upper Border Series. In all three places, the minerals follow parallel trends with plagioclase progressing steadily from basic labradorite to sodic oligoclase and olivine and Ca-rich pyroxene evolving to pure fayalite and hedenbergite. The original magma, which was unusually rich in phosphorus and titanium, followed a trend of differentiation characterized by exceptionally strong iron enrichment and relatively little increase of silica until the very latest stages of differentiation when the magma split into two liquids, one very rich in iron and the other in silica. The original compositions and textures of the rocks have been altered by late-stage metasomatism which, in extreme cases, resulted in closely associated anorthosites and pyroxenites. The principal mechanism of crystal-liquid fractionation during formation of the Layered Series was compaction, but convective fractionation seems to have become important in the late stages of evolution.
The layered mafic rocks of the Bushveld Complex are derived from three magmatic lineages - a lower part comprising the Lower and Critical Zones which crystallized from high-Mg and-Si parent liquids, an isotopically distinct Lower Main Zone derived from more evolved aluminous tholeiitic liquids, and a succession above the Pyroxenite Marker, which includes the entire Upper Zone, stemming from the mixing of residua of these earlier liquids with a final major injection of tholeiitic liquid. The varied mineralogy and thickness of the Marginal Zone indicate that it is not a quenched magma, but represents variable cumulus enrichment into several chemically different magmas, only some of which may be representative of magma producing the layered rocks. The complex was emplaced by repeated injections of magma ranging in volume from small to large. Evidence supporting the concept of multiple injections lies in the presence of distinct breaks in initial Sr-isotopic ratio, textures indicating partial resorption of earlier crystallizing phases, abrupt reversals of fractionation trends defining saw-tooth profiles through successive units, and protracted reversals in mineral compositions through hundreds of metres of section that culminate in primitive olivine-rich cumulates in the Lower and Critical Zones. The small influxes yielded localized partial cyclic units < 1 m thick. The size of the larger influxes may be gauged from the thickness and areal extent of chromitite layers which require hundreds to thousands of times their own thickness of liquid to satisfy the Cr budget for their formation. Studies of the Critical Zone in the Western limb, for which more comprehensive data are available than elsewhere, reveal a gradation along ca . 200 km of strike from a more primitive proximal facies in the northwest to a more evolved distal facies in the southeast. This concept helps to explain major problems in the Cr budget of the entire limb, the greater proportion of chromite- and olivine-rich cumulates in the proximal facies, and the more feldspathic character of the distal sequence. In the Eastern limb changes in relative spacing between, and thicknesses of, chromitite layers from north to south suggest a similar process, but the transition is abrupt rather than gradational. No single process can account for all forms of layering. Despite certain limitations, largescale magma mixing is the most plausible mechanism to bring the liquid composition into the primary phase volume of chromite to produce chromitite layers. By contrast, the trace-element chemistry of magnetitite layers suggests that they were derived from relatively thin liquid layers, and that magma addition and mixing did not occur. The origin of the incomplete cycles which may range from pyroxene- and/or olivine-rich to feldspar-rich layers of the Lower, Critical, and Main Zones appears to be intimately bound up with the emplacement of fresh magma batches. These mixed with partially crystallized residual liquids to produce cumulates bearing incompletely resorbed plagioclase inclusions, highly variable Sr-isotopic ratios, and non-cotectic proportions of phases. Chemical fingerprinting of pyroxenes in Critical Zone rocks shows that gravitational sorting must have played some part in the generation of pyroxenites and in yielding non-cotectic norites. Prominent layering near the top of the Main Zone in the Eastern limb cannot be explained by injection of fresh magmas, oscillatory nucleation or crystal settling, leaving the action of density currents as the most probable mechanism. Whereas layering is well developed in parts of the Upper Zone, the cyclicity of the Critical Zone is absent. The existence of an extremely thick liquid column in the chamber during the later stages is indicated by the 2500 m-thick, isotopically homogeneous sequence above the Pyroxenite Marker. Layering and reversals in mineral composition within this interval may have resulted from the breakdown of stratified liquid layers or convective overturn, rather than addition of magma.
Layered mafic intrusions are significant sources of the platinum-group elements, base metal sulphides, chromite, magnetite, and ilmenite. The distribution of these ores is reviewed, with special attention to the economic deposits and subeconomic occurrences. The geological setting, composition, mineralogy, and textures of the ores are described for the Bushveld and Stillwater Complexes, the Great Dyke, the Munni Munni Intrusion, complexes in Finland, and some smaller intrusions. Both the platinum-group element (PGE) mineralization and the often associated base metal sulphides have characteristic geochemical and mineralogical styles; these are variable in even a single layered intrusion, and are even more so when different intrusions are compared. The distinction between constant and variable metal contents in relation to thickness variations of the PGE sequences is emphasized. Oxide ore deposits are less variable but the compositions, especially for chromite, are specific to the layered intrusion in question. Subsolidus re-equilibration and ore-mineral alteration are usually present as variable processes in all the mineralized sequences. Mineralization models are briefly addressed in the light of these variations. The primary geochemical character of PGE ores, and the occurrence and character of the oxide ores, probably reflect the influence of the magma source region at depth rather than processes in the magma chamber at the site of emplacement.
Variations in whole rock and mineral chemistries allow the recognition of four distinct groups of peridotite nodules within these kimberlite occurrences. Equilibration temperature estimates together with general geo-chemical considerations strongly suggest that the relatively ferriferous Group 1 (Cr-spinel wehrlites and lherzolites) and Group 2 (Al-spinel lherzolites) nodules represent basaltic cumulates of likely lower crustal origin. By contrast the more magnesian peridotites are considered to have been derived from the upper mantle with the Group 3 nodules (Cr-spinel lherzolites and harzburgites) originating from shallower levels than those of Group 4 (garnet harzburgites). These data imply the existence of a Cr-spinel peridotite zone up to 30 kms thick at the top of the mantle beneath this region, at the time of kimberlite emplacement. Nodule samples of such rocks commonly show subsolidus deformation and recrystallisation effects leading to the development of mosaic and symplectite textures. Element partition considerations indicate that such textures have developed during cooling from an earlier temperature maximum; this cooling may have accompanied slow diapiric upwelling in the uppermost mantle.
