C awd.ian tulineralogist Vol. 17, w. 233-256 (1979)

A REVIEWOF SOMERECENT CONCEPTS OF THE BUSHVELDGOMPLEX' WNN PARTICULARREFERENCE TO SULFIDEMINERALIZATION

GERHARD VON GRUENEWALDT Lnstitute lor Geological Research on the Bushveld complex, university of Pretoria, 0083Pretoria, Republic of SouthAfrica

ABsrnAcr de roches mafiques i ultramafiques qui constituenl lo complexeont 6t6 mises en place il y a 2O95(25\ m,a., pr€sque entidrement dans la couverture s6di- lff t of Ni , With resources of close ta 22 x m"ntaire peu d6form6e du craton du Kaapvaal' Bushveld Complex tle layered mafic rocks of the dont les $tructures gouvernent l'em'placementainsi repository of low can be considered the largest que la forme actuelle des diff6rentes infrusions. grade, Ni-sulfide deposits in the world. disseminated Deux -ug^a* ont prodrut la s6quencestratifi6e du are contained in several min- These vast resources complexe; Ie premier, i caracGre ultramaf,ique, a of which the Merenskv Reef. the eralized layers 6t6 suivi par plusieurs 6pisodes de maematisme UG2 chromitits layer and the Platreef are the most thol6itique. Ces magmas parentaux' pauwes en The four mafic to ultramafic compart- irportant. soufre, auraien! par incorporation de S, COz et HzO whicb constitute the layered sequence of the ments lib6r6s lors de rEactionsm6tamorphiques, donn6 les were emplaced. 2095!25 m.y. Bushveld Complex horizons min6ralis6s situ6s i la base du complexe. into a relatively undeformed sedi- ago, essedtially les hypothdsesvisant i. expliquer les con- cover of the Kaapvaal craton. The loci of On revoit mentary centrations de sulfures; la compaction de cumulats as well as the present-day configura' emplacement et la migration ascendantede liquides r6siduelsinter' various intrusions are controlled largely tion of the cumulus (qui constituaientplus de 2O/o d'un empi' structural features of the Kaapvaal craton. by the lement de cristaux cumulus) sont des processus Evidenco is presented in support of a bimodal importants de la formation des adcumulats et des the magmas that gave rise to the layered character of diff6rents types de peqmatitesmafiques. Ces mOmes the complex. The initial magma is con- sequen@ of liquides ont aussi influenc6 (1) la cristallisation des sidered to be ultramafic and to have been followed poches isol6es de magma prds de la base.de.. la by several heaves of magma of tholeiitic character. produisant ainsi le type de min6ralisa' parental of complex were ap- chambre. These magmas the tion du' banc Merensky et (2) la concentration parently poor that incorporation in the in sulfur so des 6l6mentsdu goupe du platine dans les niveaux magrna of S, COs and HzO liberated during meta- de de la zone critique sup6rieure. morphic reactions resulted in mineralized layers (Traduit Par la R6daction) near the floor of the intrusion. The hvpotheses to account for sulfide conoentration are reviewed. and ascending intercumulus liquids, Compaction INTRoDUcTIoN which constitute more than 20Vo of a pile of cu- mulus crystals,. are important processes in tle formation of adcumulates and the different types It is well known that the layered mafic rocks of mafic pegmatites. These liquids must also have of the Bushveld complex house the wor"ld's had an effect, firstly, on the crystalization of largest reserves of the -grot'p tnetals, "isolated" batches of magma near the floor of the chromium and vanadiu,m (Willemse 1969, von give of chamber to rise to the Merensky Reef type Gruenewaldt 1977), estimated to be about 86, and, secondly. in tle concentration mineralization 83 and 64Vo, rcspe*tiv,e1y,by van Rensburg & of platinum-group elements in chromitite layers (L977). These authors consider the of the upper critical zone. Pretorius world's reserves of to be approximately 50 x 100t and South Alricals share in these to be SoMlrerns 5 x 10' t, contained in the Merensky Reef, which is presently being min66 at several local- Vu ses r6serves de ?2 x 10ot de Ni, on peut con- ities. If, however, the estinated tonnages of sid6rer le complexe stratiforme du Bushveld comme nickel metal sontained in the UG2 chromitite gisements le plus vaste entrep6t de de sulfures de layer and tle Platreef (Table 1) are added to du monde. Ces ressources se nickel diss6min6s this figure, total nickel resources in the Bush- trouvent localisdes i plusieurs niveaux min6ralis6s, Complex amount to a substantial' 22 x LU t" dont les bancs Merensky et Platreef et la chromitite veld one of the largest UG2 sont les plus impor.tants. Ires quatre chambres making the Bushveld Complex ?33 234 TIIB CANADIAN MINERALOGIST repositories of low-grade disseminated nickel Tne SsrrrNc AND SHApE oF TrrE Cotupr,sx sulfide in the world. Nowhere in the corn- plex does nickel occur in sufficiently large Regional tectonic setting quantities to be exploited on its own; the con- The assemblageof well-exposed Precambrian tribution of the associated platinum-group rocks on the Kaapvaal craton displays a fairly , Au and Cu make these large dissemi- continuous record of the evolution of the crust nated strlfide deposits economically viable. from the early Archean, some 3600 m.y. ago, This review is confined essentially to tle to late Waterberg times, t1250 m.y. (Hunter latest thoughts and developments concerning the 1J74a, b). Detailed sttrdies on the many rock environmsal and mode of emplacement of the units that comprise the Kaapvaal sraton have complex, the composition of the magmas that shown that the evolution of the crust during gave rise to the layered mafic rocks of the this 2300 m.y. period was dominated by east- complex and the magrnatic sul,fide concentra- northeast and north-northwest structural trends tions contained within them. A basic knowledee and that these structures are also important of the various lithologies that comprise tie controls on the present configuration of the Bushveld Complex is assumed. For this infor- lithologies preserved on the craton. The east- mation, the reader is referred to the many northeast direction is especially well-developed review-type papers (Wager & Brown 1968, in the eastern part of the Kaapvaal craton and Willemse 1969, Hunter 1976, Vermaak & Lee is manifest in the prominent Murchison linea- 1978, Hunter & Hamilton l97S). Although in ment (Fig. 1), a site of repeated tectonic re- moet of these reviews an attempt was made to activation from Archean to late WaErberg cover all aspects and components of the Bush- times, as well as in several other linear east- veld Complex, the approach differs from review northeasterly-trending structures @utton 1973, to review; this is understandable when the large Hunter 1975). The north-northwesterly trend amount of literature on the Bushveld Complex is represented by tle so-called Amsterdam is considered. Well over 800 articles, tleses linea,ment and the Kraaipan trendo as well as and research reports are listed in two recent a broad north-northwesterly-trending arch that bibliographies (Molyneux et aJ. 1976, Knowles is the location of a number of domical base- t978). ment windows. Termed "Vryburg arch" by (1975), TABLEI. PNOMI]IOilSAI'ID RESOURCES OF P@ + AU, NI ANDCU Hunter this consists of a number of III iIINERAI.IZEOLATERS OF THEBTJSHIELO COWLTIT positive, dome-like features, the magnitude of lilEREl{'XY ,,r.r!ffr' which varied at different times during the depo- REET LAYER PLATRETF TOTAT sition of the sedimentary successionsthat over- lie the basement. At other times, however, PGE1 Au Proportlonsxx these positive areas also coincided with the 59 41 42 axes of maximum deposition of sedimentary Pd 25 34 46 basins (Brock & Pretorius t964). Rr t2 4 Rh 3 9 3 According to Hunter (L975), the develop- Ir I t.9 0.8 ment of the Kaapvaal craton, rigid enough to 0s 0.8 0.6 sustain and preserve ensialic basins, took place tu 3.2 0.4 3.4 about 3000 m.y. ago. On this crust a series of Redery grade 5.5 6.9 3.0 9n/t distinct depositional basins developed, the basin Res@rces. t8.15 32.50 12.24 62.89 kg r l0o axes and northern flanks of which migrated across the craton .from southeast to northwest l{l (Anhaeusser 1.973, Hunter 1974b). In these ConEnt , 0.18 varlable variable kr@ery grade 0.7 3.6 basins the sedimentary and volcanic rocks of, Kg/l in ascending order, the Pongola, Dominitrm Res@rces- 4.3 3.8 14,7 tx 10o Reef-Witwatersrand, Ventersdorp, Transvaal and Waterberg Supergroups accumulated. Sedi-

Content z 0.ll l0 varlable mentological studies suggest that, in general, ns@ery grode 0.8 1.8 the southern kg/t margins of these basins reflect t"'fl"i3u 7.30 9.95 gradual subsidense in contras,t to the more aotive northern flanks, and that transport di- xrvalues \fter von Gruen*aldt (1977); bNen fM Co6ins I Vemak fi976). rections werg except for local deviations, mostly from the northern and northeastern sides of the basins (Brock & Pretorius 1964, de Villiers 1967, Visser 1969, Button 1973, 1975). SULFIDES IN TTIB BUSTTVBLD COMPLBX 235

THE STRUCTURALSETTING OF THE BUSHVELDCOMPLEX

o l@ eoo

LEGEND ...-..-.htdddof d€d

2tt@ !ffittcd i) a*rm O o*tty nqi a @tur

.F-- b tbg'dft bts

-.-.- k@*y Rd

BASEI{ENTDOTES ! Chsi66t 8 V@d€fod 2 lrolope 9 Klsl€dop 3 Mdblo Holl lO filolrl&q 4 Domllt@ ll irobpo 5 \k@ 12 Gabqong 6 Joh@6hrg 13Mdo@o 7 Dq6 14 C@@dil3Ris

Fro. 1. The structural settlng of the Bushveld Complex.

An explanation for the development and the inflation caused by subcrustal thermal expanqion. migration of tle sedimentary basins with the During such times the lithosphere would have associated volcanic activity was put forward by been in a state of tension and cond,itions would Pretorius (1965), who considers the basins to have prevailed favoring the formation of dykes' have developed under tensional conditions in In soitrast, the Bushveld Complex was intrtrded an environment where vertical tectonics pre- at the termination of Transvaal sedimentation, vailed. The nature and type of sedimentary and at a time of maximum continental deflation, volcanic rocks are considered to reflect the when compressional conditions resulting from variations in the energy associated with periods zubsidence would have dominated the upper of uplift and succeeding quiescance. crust. Although regional stress conditions at Hunter (1974b, 1975) has attemPted to re- the termination of Transvaal times favored the late the evotrutionary history of the sedimentary formations of sill-like intrusions such as the basins in terms of other events that occurred Bushveld Complex and its assosiated sheets of in the Kaapvaal cnrstal plate subsequent to diabase in the hoor rocks of the complex' this its stabi{ization some 30@ m.y. ago. He argued does not offer a suitable explanation for the (L974b) that periods of magmatic activity, siting of the complex, which is limited to an whether they involved large-scale emplace- area in the northeastern part of a basin that ment of granite or relatively smaller-scale probably extended over the whole Kaapvaal mafic plutonism and volcanism, followed re- craton. gressive sedimentary patterns of basin infiling. Earlier contentions (Daly 1928, H:all L932) Tbis, as well as other features of the Kaapvaal that the Bushveld Cornplex was a lopolith had crustal plate, implies, according to llunter, a to be modified when more detailed mapping connection between events in the mantle, crust revealed that the floor contact was in many and depositional basins. He developed the con- ltrcalities transgressive. Consequently, sevetal cept that both the IJsushwana Complex and the overlapping, essentially funnel-shapp"g Ttto- were emlilaced during periods of sions were proposed to account tor the rorm 236 THE CANADIAN MINBRALOGIST

complex (Wilson 9! _t!e 1956, Wager & Brown tional basin of the same supergroup. According 1957, Willemse 1959, l9@). However, when to Hunter the broad locus of emplacement of interpretations gravity of data (Cousins 1959, the Bushveld Complex is sited within the Trans- Smith et aI. 1962) revealed that the mafic rocks vaal depositional basin close to the interface do not extend beneath the centre of the com- between dominautly domical structures plex, and es- the more modern concept developed, ac- sentially linear structures. cording to which the complex consists of a num- ber.of Stractural rclationship .essentially separate and only partly s gygrJapping intrusions (Willemse 1969, Fiunter Ro!9rts (1970) has shown that sills in large 1975, 1976; Vermaak - 1976b, Vermaal & Lee depositional basins such as tle Transvaal basin 1978, Sharpe & Snyman 1978). ideally would be emplaced close to the strati- Apart from a small body at Nietverdient, graphic level occupying the chord positionn f.e., north of Zeerust in the western Transvaal" the a horizontal to near-horizontal surface. From intrusions have what Hunter (j,g7i) t""-"a u this it follows that any transgressions of a sill "crucifom" outline. They outcrop as four lobes: of large horizontal dimensi;N would r.n""t the western lobe, tle eastern lobe, a south- irregularities in the atti,tude of the bedding of eastem lobe that is continuous with the eastern the intruded sequense. This is diagr.ammatically Iobe br.r,t largely concealed by a younger illustrated in Figure 2 from wniin it is seeu cover, {al9ozolc and a nothern lobe. In this that a downward transgression into lower strati- definition, the four lobes are arranged approxi- graphic levels implies the existence of an uo- mately symmetrically about two pirpendicular warp at the time of emplacement, whereas a axes, each 350 km Iong aligned east_northeast transgression upward in the sequence impfies and north-northwest (Fig. -the i). The east_north- the existence of a downwarp 14 basini as- east axis coincides with the depositional axis suming constant thickness of cover. of the Transvaal basin (Visser 1969, Button With the aid of this model, it becomes pos- 1973) which in turn is oriented paralet and in sible to explain various structural relationslips close proximity to several othei linear, east_ in the Bushveld Complex (von Gruenewatdt northeasterly-trending & structures, u-oogri th"a Sharpe in prep.); these are too numerous to the Murchison lineament. On the othEr hand, elaborate upon here. A ferr of tlese trans- the north-northwest,axis corresponds, accJd_ gressive features have a bearing on the present- ing to Hunter (1975), to a struciurally control- day configuration of the lobeJand on ihe dis- led depressionon the nortleastern flank of the cussion of the tectonic setting as outlined above; Vryburg arch. Although Bushveld ,*t. *. they will be briefly discussedhere. today preserved in this depression there is Both the northern and southeasternlobes are reason to believe that both the areas of the characterized by pronounced transgressions.In northern and tle southeastern lobes were struc- the southeast the layered mafic iocks of the fural highs during or shortly before the time Bushveld transgress,over a distance or emplacement. of 100 km. from a level above the Dullstroom Hunter (L975) volcanic assumed the Vryburg arch rocks at the top of the pretoria Group over to-have originated the as a result of the culmlnation entire group to rest 200 m above the Transvaal of two or more fold axes of different waveterrstn dolomite at Bethal (Buchnan IITS). The trans- and amplitude. Such a broad arch would ie gression in the northern lobe is even more flanked by two major synclines, the north- pronounced in that the Bushveld rocks eastern one trans_ of which would correspond to the gress from above the Magaliesberg basin of sedimentation euartzite of the Transvaal Super- Formation into Archean granite ovei a dis- group in the Transvaal whereas the southwesiern tance of less than 20 km (van der Merwe t976). one would represent the northern Cape deposi_ Both these transgressionsnecessitate the exis-

Frc' 2' Relation of stratigraphic successionto a sill of large horizontal dimensionsemplaced into a gently warped basin at or close to the position of the chord. SULFIDES IN THE BUSTIVELD COMPLEX 237 tence of structurally controlled anticlines in the evidence thus suggestsonly limited and localized area during emplacement of the complex and upwarps along the eastern flank of the Vryburg nof synclines as postulated by Hunter (L975). arch after deposition of the Transvaal sequence Such anticlines within the Transvaal depositional but prior to emplacementof the Bushveld Com- basin were conceivably causedby compressional plex. stress during subsidenceof the basin. Several modeled profiles through the eastern In the western Transvaal, the Bushveld Com- Bushveld Complex based on available and new plex was emplaced semiconcordantly above the gravimetric data as well as airborne magneto- Magaliesberg Quartzite Formation, except in an metric data show that the mafic rocks laterally area west of the Pilanesberg where it trans- peiretrate the dome-like features at Marble gtessesthese quartzites and rests on underlying Hall and Dennilton situated along the western shales (Vermaak 197O, Liebenberg L970). At flank of the eastern lobe. Two sections, based Nietverdient, the Bushveld \Mas emplaced at a on the gravity models of Hattingh (1977) and similar level as in the western lobe and it is of Molyneux & Klinkert (1978), are reprod- therefore evident that no anticline existed at the uced here as Figures 3a and 3b. This relation- loci of emplacementof this intrusion. The pre- ship, together with the pronounced thinning ol sent-day limited extent of this occurrence, with the Transvaal strata over the Marble Hall dome only representativesof the lower half of the (de Waal 1970) in proximity of the depositional critical zone and the lower zone remaining, is axis of the Transvaal basin, points to the pre- the result of subsequent upwarping of the senceof a relatively inert area during subsidense Vryburg arch. Of interest is the extensive meta- of the Transvaal basin that underwent doming morphic aureole of the Nietverdient occurrence prior to emplacement of the complex' Further which affects the Malmani Dolomite up to 60 evidence indicates that the western flank of the km south of the intrusion (Hammerbeck 1970, eastern Bushveld was tectonically active during Klop 1978). This vast metamorphicaureole not emplacementof the complex: (1) the absence only indicates that the mafic rocks extended of the delicate layering typical of the gently for a considerable distance to the south, but dipping portions of the cornplex along its eastern also suggeststhat these lithologies substantially flank (Hunter 1976), and (2) somedeformation transgtessed their floor because the thickness features described by Marlow & van der Merwe of the Pretoria successionbelow the present-day (L977) from the Malope area. floor contact is of the order of 3000 m in this In an alternative interpretation of his gravity area (Klop 1978). It is of interest to note that data of the area southeast of the Dennilton Martini (1976) recognized the presence of a dome, Hattingh (L977) postulatedthe presence paleoridge in the dolomite at the same position of another large, dome-like feature, centred where this transgression is postulated. The approximatelyover Verena (Fig' 1). A surface

E llJonderk6'PitroloPe Phohrm Plsteou ssktlj$uno L€olo Montohs foull

30

Rool roc'ler gronitg gronophyrs,felsits Floor rocks (6stntiolly Ttglsvqol Supsgrow) Mognsilc mofic rcks (Wper zae) Aduoon b@€rnsnt Motlc rocks (undlffer€ntistod Frc. 3. Idealized sectionsthrough the eastern Bushveld based on the geophysicatinterpretations of Hat' tingh (1977) and Molyneux & Klinkert (1978). THE CANADIAN MINERALOGIST

manifestation of this dome could be tle curved Bushveld Complex. I{owever, it is the deep, nature of the Waterberg basin along the southern magma-tapping fractures and their alignment to flank of this anticline. which some importance must be attached when The above evidence suggests the existence, considering the tectonic setting of the complex. beginning in early Transvaal times, of a broad Based on this argument, the layered sequence anticlinal pretoria-Z,ebediela wa4), termed the of the complex should be seen in terms of anticline, along the western flank of the eastern seven shallow, cone-like intrusions, the feeders Bushveld. The warp consists of a number of of which are postulated to coincide with the dome-like features, which, from south to north, observedgravity highs in the Bushveld (Fig. 1). includes the Halfway House dome, the postulated Of these seven intrusions (i) three are aligned Verena dome, Dennilton dome" Marble Hall along a nortl-northeast gravity high in the fragment, the Malope dome and the Chunies- eastern co.mplex, (ii) two are aligned along a poort arch. This north-northeast- to northeasr- north-northwest gravity high in the western trending anticline defines the western flank of complex, (iii) one is sited northwest of pot- the eastern Bushveld Complex and is roughly gietersrus at the intersection of the Amsterdam parallel to the three north-northeast-trending lineament, the Pietersburg lineament and pos- gravity highs (Fig. tr) that have been interpreted sibly also the southern extension of the Grear as a possible dyke-like feeder to the eastern and Dyke direction (van der Merwe 1976), and southeastem sestors of the Bushveld Complex (i9 one is situated at Nietverdient where Bies- (Cousins pretoiia- L959). Reactivation of the heuvel (1970) indicated the presence of a Zebediela anticline post-Bushveld in times is f_eederclose to the position now occupied by indicated by the emplacement some 1650 m.y. the Goudini alkaline complex. ago of a number of diapiric granite intrusions, A model for tle emplacement of the mafic referred to as the Makhutso Granile (de Bruiyn rocks in the Transvaal sequenceis provided by & Rhodes L975, Marlow L976), along the Sharpe & Snyman (L978). They point out thar eastern flank of the anticline (Fig. 1). These extrusion of some 7 km of subaerial basalt and features strongly suggest that the north-north- felsite onto the 7.7 km thick pile of submarine east grain predominated in the eastern part of sedimentsof the Transvaal sequenceresulted in tle Transvaal basin during emplacemenfof the the depression of the floor of the basin into a B3shvgl{ Complex. This rrend parallels the deep regi,me of maximal horizontal compression. This abyssal of Cousins (1959) along whici induced favorable conditions for the intrusion the majority of layered intrusions in s6uthern of a total of. 2.5 km of diabase silts. which Africa are aligned; it is also widely considered f'urther assistedin the subsidence of the basin as having influenced the siting of ihe Bushveld into a regime typical for shear fail.ure. Emplace- Ccrmplex (Hall 1932, Cousins 1959, Vermaak ment of the Bushveld magma into this regime 1.976a,Vermaak & Lee 1978). initiated shallow-angle shear fractures thai ex- It _ _ig the subsequent upnvelling along the tended upward into the sedimentary rock- broad Vryburg arch and the concomitant forma- diabase sandwich, tending to parallel tfie chord tion of a syncline along its northeastern flank at its position below the subaerially extruded in conjunction with reactivation along the linear volcanic rocks. This resulted in a series of east-northeast Murchison trend tlat largely-mafic con- shallow cones, the upper and outer reaches of trolled present-day the configuration of which overlapped in the vicinity of the chord rocks. This is also borne out by the considerably position. younger age of 1,25A-14OOm.y. of the pilanes- berg and related alkaline intrusions (Ferguson Tnr Busrrwr,D MacMA L973);_the majority of these were emplaced along the struotural monocline on the &stern Earlier flank of the Vryburg arch, and contrast with an concepts age of 2095 m.y. for the mafic rocks of the For many years the initial magnna that gave Bushveld Complex (Ilamilton 1977). rise to the layered sequence of the Bushveld Complex was consideredto Emplacement of the mafic rocks be essentiallybasaltic in composition (Hall 1932, Hess 1960, Wager As outlined in the preceding section, the & Brown 1968, Atkins L969). This assumption north-northwes,t and east-northeast aligned was based on the composition of the undifferen- trineaments and arches are considered largely tiated sills of diabase in the floor of the complex r-esponsiblefor the present-day configuration and and the fine-grained marginal rocks of tle the lobate outline of the mafic rosks of the layered sequence. Experimentally deter,mined SULFIDES IN THE BUSHVBLD COMPTEX 239 crystallization sequences from two fine-grained Cameron calculated the weighted average MgO marginal rocks, one corresponding to a high- value for the upper, main, critical and lower alumina basalt and the other to an olivine tho- zones of the eastern Bushveld Complex to be leiite, as well as theoretically deduced crystal- aboat t3/o. He also determined the average lization sequencesfrom their resBective chemical Cr content of the eastern complex to be in composition, have led Tilley et a/. (1968) and excessof 1000 ppm, compared to 170 ppm for Irvine (1970a) to doubt the vatdity of these averagebasalticrocksgivenbyTurekian&Wede- rocls as a measure of the composition of tle pohl (1961); he concluded from these figures parental basalt. that tle initial Bushveld magma was probably composition' Evidencefor an initiarhigh-magnesium magma :jiff*ffT:H",'injliilll"oin In the eastern Bushveld, the Hendriksplaas An evaluation of the bulk composition of the , characterized by nurnerous highly meta- Bushveld magma based solely on a stratigraphic morphosed xenoliths of sedimentary origin, is column of present-day exposures must be viewed usually developedalong the floor-contact of the with some reservations as there seems to be a layered sequencewhere it constitutes part of the considerable difference between the proportions marginal zone. In the vicinity of Burgersfort, of the four layered zones in section and in however, this norite, together with some of the aerial extent. Interpretation of gravity data of underlying quartzite, is separated from the base the easternBushveld (Hattingh 1977, Molyneux .of the intrusion by several hundred metres of & Klinkert 1978) shows a considerable west- an alternating succession of and ward thinning of the mafic rocks from a maxi- . This, together with the semide- mum outcrop position at present (Fig. 3). In tached nature of the Aapiesdoorndraaiperidotite both interpretations the distribution of the ultra- mass and several other separate bodies mafic rosks is restristed to such an extent that in the floor of tle intrusion, led l-iebenberg it is doubtful whether the lower two zones or (lgA) and Willemse (L969) to suggestthe pre- considerable parts thereof are developed in sence of a magma of peridotitic composition in areas overlain by roof rocks today (Fig. 3). early Bushveld times. SubsequentlyGijbels et al. This interpretation is to a large extent substan- (L974) proposed, on the basis of a partitioning tiated by field relations at Malope (Marlow & model for platinum-group elements between van der Merwe 1977) and along the south- magrna and cumulus of the lower and eastern flank of the Dennilton dome. In both critical zones, that the lower zone represents2/s these areas,rocks correlated with the upper part crystallization of an early, magnesium-rich mag- of the critical zone seem to be resting directly ma; this was followed by a second, much larger on the floor of the intrusion. Furthermore, inter- pulse of magma, which mixed with the remain- pretation of the aeromagneticdata indicate that ing Vs of the first batch and from which the about r/s of the mafic rocks concealed by the critical, main and upper zones formed by a Bushveld granite is highly magnetic, i.e., upper- process of fractional crystallization. The second zone type rocks (Molyneux & Klinkert 1978). magma was, according to the authors, enriched It is not only in the eastern lobe where in chromium in order to precipitate chromite in upper- and main-zone type rocks predominate. layers shortly after its emplacement. The valid- A similar situation seems to prevail in the ity of their deductions was criticized by Came- Bethal area (Buchanan 1975) where ultramafic ron (1978), who emphasized that there is no rocks have only a very limited thickness except indication of any break, either structural or perhaps beneath the upper-zone rocks in the compositional, between the lower zone and the vicinity of the very pronounced gravity high to critical zone in the eastern Bushveld and that the northeast of Bethal. In the northern lobe a it is virtually impossible to obtain pure thick succession of ultramafic rocks seems to fractions for analyses at ppb levels because of be restricted to a relatively small area in the the numerous tiny inclusions of sulfides and vicinity of Potgietersrus, whereas rocks of the other minerals in the curnulus minerals. The main and upper zones transgress those of the lack of change in composition of cumulus lower two zones to occupy an area several times bronzite and olivine throughout the great thick- that of the ultramafic rocks (van der Merwe ness of cumulates that constitute the lower zone 1976). Peculiar to this area are the ultramafic and the lower half of the critical zone led Came- satellite bodies of massive bronzitite, some of ron (1978) to propose that both these zones which contain chromitite layers similar in com- crystallized from one batch of magma. position to the LG3 and LG4 layers near- the From available modal and mineralogical data base of the critical zone in the western lobe. 240 TITF CANADIAN MINERALOGIST

Of considerable interest, however, is the huge The extensive partial melting to produce a body of ultramafic rocks at Grasvally, inasmuih MgO-rich magma is considered by Hamitton as these are the most magnesian rocks (Enes, (1977) to be the result of a sudden release of Fom) known from the Bushveld Courplel As- overload pressure on a shallow-level mantle sys- sociated chromitite layers have Cr/Fe ratios tem at or near solidus temperatures. He bases similar to those of the Great Dyke (de Villiers his assumptions on the model of Richardson er al. 1970), This mass of ultramafic rocks is con- (in Hamilton 1977) who expressed the opinion sidered to be upfaulted from lower levels in the that mantle material near solidus temperatures cone-shaped intrusion (Barrett et al. L97g), would rise beneath zones of separation in areas Available evidence therefore suggests that of rifting and that the ascent will be accom- the upper- and main-zone type rocks, which have panied by a rapid increase in temperature and -an ayerage MgO content close to that of typical extensive partial melting. They demonstrated basalt (Cameron 1978) outweigh the ultraliafic that earlier magynas would be MgGrich (20- rocks of the critical and lower zones by a factor 3OVo) f.olloured by magmas with lesser and considerably greater than that which is obtained comparatively uniform MgO values of about from present-day outcrops in the eastern Bush- 7-BVo. veld. The bulk composition of the layered mafic Although the Transvaal basin was in a state rocks should therefore be intermediate between of compresslon during e.mplacement of the that of olivine tholeiite and normal tholeiitic Bushveld Complex, downwarping of the rigid basalt, probably closer to tle latter. Further crust below the Transvaal basin may.have given evidence in_support of an early Mg-rich magma rise to. tensional conditions in the lower grust- ls provlded from recent mapping by M.R. upper mantle regimes and the reactivation of Sharpe (Sharpe & Snyman l97g), who'records deep north-northeast (Great Dyke) and north- th,e presence of fine-grained marginal rocks northwest (Usushwana) tensional fractures adjacent to the lower zone with lith6logies sim- beneath which the initial high-magnesium mag- ilar to that zone. The recognition of such rocks ma could have been generated.The early magma is significant as thesemay yield some indications rose upward along tlese tension fractures but of the chaiacter of the initial Bushveld magma. spread out laterally along the shallow-angle Alsq in the northeastern Bushveld, severafsils shear fractures that developed as a result of of peridotite 10 to 20 m thick occur in sedi- stress conditions in the Transvaal depositional mentary rocks pretoria of the Group several basin beneath the chord positioh & hundred lsharpe metres below the floor contact of the Snyman 1978). This resulted in the initial cone- intrusion (Schwellnus et aI. 1962). sheet intrusion in which the early high-magne- sium Petrogenesis of the Bushveld magmas: magrna crystallized to give rise to the towards a model rocks of the lower zone and lower part of the critical zone. Gradual addition of magma of From .the available evidence it is doubtful more basaltic composition and mixing of these whether fractional crystallization of normal with the partly differentiated earlier'magma tholeiitic b-asalt produced the relatively large resulted, firstln in the very gradual changes quantity in of types encounterid compositional trends of the cumulus minerals in the layered mafic sequence.ihe presenceof with height in the lower parts of the intrusion ultramafic bodies at the base of the complex, (Cameron 1970, L978) and secondly, the simul- such as Aapiesdoorndraai, Grasvally, the-Uit- taneous lateral expansion of the conical magrna loop as well as the ultramafic. fine- chambers. Lateral expansion was governed by grained marginal rocks of the lower zone, clear- the position of the shord at the time of em- Iy point towards an early Mg-rich magrna. The placement, with the result that the individual extensive development of upper-zone id main- cone-sheetscould have become interconnected zone lithologies, on the other hand, indicate that at higher levels at a particular time in the stages subsequentinfluxes of magma were less masne_ of their development. sian, possibly normal tholiiites. It would sErve The last major influx of magma amounting to little purpose at this stage to speculate any at Ieast lOVo of.the volume of the total Bushveld further on_the possible nature and compositlon magma is postulated to have taken place in- of the.Bushveld magmas; suffice it to Jay that mediately prior to crystallization of the pyro- an evaluation of the fine-grained marginal-rocks xenite marker, i.e., after aboat 3/q of the maio of the ultramafic bodies and the lower_zone zone had accumulated (von Gruenewaldt L973) Tgkt should yield important information A gradual reversal in the compositional trend (Jnarpe, Lee, research in progress). of the cumulus minerals is actuallv noted sorne SULFIDES.IN THE BUSHVELD COMPLEX 24L distance below this marker, with a fairly pro- 8'Sr/88Sr ratiq to a level above the Merensky nounced break at the marker itself. The rocks Reef, indicates fairly undisturbed and continuous below the marker correspond closely ,to those fractional crystalliza'tion of a first batch of high- at the base of the upper zone and the volume magnesium magma, the Merensky Reef having of the influx was therefore calculated purely on formed after tle crystallization of. abovt 9OVo the basis of the duplication of lithologies in the of the first magma. Emplacement of a second main zone plus a proportional contribution to batch of magma with a low initial 87Sr/88Sr the upper-zone type lithologies. The addition at ratio after crystallization of the Merensky Reef this level must, however, have been considerably and mixing with the residual magma of the first more than the postulated minimum of lOVo, batch would result in the observed wide fluctua- firstly, because of the gradual reversal which is tions in preserved initial ratios in the main observed some distance below the marker and, zone. After homogenization, continued fras- secondly, because of the dilution effect of the tional crystal'lizationwould produce the gradual- residual magma; a quantitavely much larger in- ly increasing ratios in the upper zone. flux is necessaryto causethe observedreversals The relatively high initial "Sr,/"Sr ratio of than one based on the actual duplications of the 0.70563 of the early magnesian magma of the resultant lithologies. The influx of undifferen- Bushveld is considered by Hamilton (1977) to tiated magma at this level in the intrusion was reflect a relatively high Rb/Sr ratio in the source recently substantiated by pronounced reversals region at the ti.me of melting. Similarly high ratios in Ni and Cr content of rocks from samples for the Great Dyke and other rocks of basaltic collected along severaltraverses across the pyro- composition on the Rhodesian and the Kaapvaal xenite marker in the eastern Bushveld (Marais cratons would seem to indicate fundamental 1977). differences of the initial ratio in the source re- Note here that the pronounced la,teral trans- gions below the cratonic areas on the one hand gressions in the Potgietersrus lobe involve not and for ocean-floor basalt on the other. The YSr/&Sr only rocks of the upper zone but the upper parts high ratio is related to the stabilization of the main zone as well (van der Merwe 1976). of the Kaapvaal craton and the underlying Si.milarly, rocks of the .upper part of the main mantle 2600 to 3000 m.y. ago and the sub'se- zone are also involved in ,the tongue-like trans- quent differences in the evolution of the mantle gressionsof the upper zone north of the Pilanes- beneath cratonic and oceanic environments berg in the western Bushveld (Vermaak pers. (Ilamilton L977, Hunter & Hamilton 1978). comm.). In both these areas and posgibly also the Bethal area, the transgressionsmay be related to the volumetrically large influx of magma Sur,rror MrNBnelrzerroN postulated after a large proportion of the main zone had crystallized. Although sulfides occur in minor quantities A characteristic feature of the layered mafic throughout most of the layered mafic rocks of rocks is their relatively hrgh initial 87Sr/80Sr the complex, a,bove-average concentrations are ratio and the pronounced variation in this ratio restricted to a few layers. Of these, three very in the various zories of the complex (Hamil,ton extensive deposits are of economic importance, 1977). Hamilton interpreted this as indicating i.e., the Merensky ,Reef, the best known although successiveem;:lacements of a number of discrete not necessarily the largest deposit of magmatic magmas derived from a heterogeneousmantle. sulfides in the Bushveld (Table 1), the UG2 Although contamination of the magmas either chromitite layer and the Platreef in tle Pot- on their upward movement through the crust gietersrus area. Platinum mineralization in all or by the felsic roof-rocks of the complex is three of these layers has been known since the rejected by Hamilton; it is especially at the roof twenties, and Wagner (1929) in his classical contact where the existenceof an anatectic melt work on the platinum deposits in South A.frica derived from the Rooiberg felsite (Irvine L970b, not only describesthe Merensky Reef and the '?latinurn von Gruenewaldt 1972) could have causedsome Horizon" of the Potgietersrus area in contamination of the magma. The strontium- considerable detail, bu,t also reports briefly on isotope data were recently reassessedby McCar- the platinum mineralization in the upper chromi- thy & Cawthorn (in press) who consider the tite layers. Since publica,tion of this monumental observed fluctuations to be due to a combination work, the results of numerous investigations on of two major inf'luxes of magrna and the effects the Merensky Reef have appeared in print, but of fractional crystallization-on this ratio. Accord- very little additional information has become ing to their model the gradually increasing available on the other two lavers. 242 THE CANADIAN MINERALOGIST

With respect to their position in the sequence, by Wagner (L929) and Liebenberg (1970). as well as from petrogenetic considerations,the Where fully developed, tle Platreef consists of above-average concentrations of sulfides in the two units: a harzburgite unit, pegmatitic in complex can be classified into three types: en- places at the base, overlain by a pyroxenite richments at or near the basal contact of the unit, consisting of a lower pegmatitic pyroxenite intrusion, disseminated stratiform enrichments layer and an upper porphyr,itic pyroxenite layer. and enrichments in crosscutting pegrnatitic Cumulus grains of chromite are developed bodies. sporadically, whereas plagioclase and clinopyro- xene together with sulfides constitute the inter- Sullide enrichments near the floor cumulus minerals. This reef, which can be traced of the intrusion along strike for more than 60 km, may attain Sulfide concentrations near the base of the a thickness of 200 m (van der Merwe 1976). intrusion have been describedfrom severallocal- It may rest directly on floor-rocks of the com- ities (Liebenberg 1970). Their composition plex or it may be developedabove a few metres seems to depend largely on the lithologies in of intervening norite and anorthosite (Wagner contact with the floor rocks. They are generally L929, van der Mer:we 1,976). more nickel-rich where the concentrations occur From the desclipions by Wagner (1929) the in ultramafic rosks of the lower zone, e.g., mineralization occurs in three different environ- at Aapiesdoorndraai and in the western Bush- ments. (i) The main mineralization is developed veld, Ni-Cu-rich where the rocks are in the coarse-grainedfeldspathic bronzitite and correlated with those of the critical and main in serpentinized rocks near the base of the Plat- zones as at Potgietersrus (Wagner 1929), and reef. Altlough Wagner referred to these ser- Fe-rich where upper-zone litlologies are in- pentin'ized rocks as representing altered, highly volved, e.g., in the Bethal area (Buchanan 1975). metamorphoseddolomite with which the reef is The lower zone. T\e sulfide concentrations at ized of the lower part of the reef Aapiesdoorndraa,i, investigated by Liebenberg as if these rocks could also represent serpentin- (L970), occur as round droplets near the top in contact over part of the area, it would seem and the floor contact of the ultramafic body. in near-surface environments. The sulfides, Troilite and cubanite are conunon constituents, pentlandite = pyrrhotite ) chalcopyrite with indicating a S-deficient environment. In con- graphite a conspicuous accessory constituent, trast, the sulfides in lower-zone rocks of the occur as irregular disseminatedpatches up to 4 western Bushveld form local disseminations in cm in diameter (Schneiderhdhn 1929, Lieben- layers of feldspathic bronzitite and harzburgite berg 1970). According to Wagner and Lieben- and constitute about two vol. % of the rocks berg the platinurn minerals are predominantly (Liebenberg 1.970,Vermaak L976a). Although and . (ii) Ni-strlfide mineral- Liebenberg considers the mechanisn of sulfide ization is developed up to 30 m away from the precipita.tion in these rocks to be in accordance floor contact of the reef where it is underlain with normal crystallization processes, he be- by highly metamorphosed dolomitic limestones lieved that the S was introduced by a process (silicified dolomite according to Wagner). These of sulfurization of the magma. In both the deposits,referred to as skarn-type mineralization Aapiesdoorndraai area and the western Bushveld, by Liebenberg (1970), contain irregularly shaped the layered mafic rocks transgress the floor patches of sulfide up to 4 mm across. Sulfides rocks to positions well below the Magaliesberg described by Schneiderhiihn (1929) are pent- Quartzite Formation where the intrusion is in landite - pyrrhotite ) chalcopyrite 2 cubanite, direct contact with sulfide-bearing shales and the main platinum-bearing phases being sper- calcareotrs sediments. Sulfur isotope data pre- rylite and stibiopalladinite (Wagner 1929). Mine- sented by Liebenberg (1963) substantiateihis ralized and unmineralized xenoliths of calc- hypothesis in tlat the sulfides are notablv dif- silicates, commonly several tens of metres in ferent in 8sS concentrations compared with diameter, occur in the feldspathic bronzitite. magmatic sulfur in tle sulfides at higher levels (iii) Sulfide deposits are also developed some of the intrusion. distance away from the mafic rocks in the floor '?latreef The Platreef. is a name introduced of the complex, especially in pegrnatites and by the geologists of the Johannesburg Conso- in what Wagner termed crush zones in banded Iidated fnvestment Co. for tle platinum-bearing ironstone. These deposits are fairly massive in rocks in the Potgietersrus area (van der Merwe places, senf4ining up to SVo Cu and 2.5Vo NL 1976) n order to distinguish this layer from Very little is known about the deposits; their the Merensky Reef with which i,t was correlated high Cu content cor'responds to similar deposits S'ULFIDES IN THE BUSFIVBID COMPLEX 243 in tle floor of the Sudbury (Hawley 1962) TA8!E2. Cu/(Cu+ il.t) Aflo pt/(pt + pd) RATIoSroR vARrolSLAIERS Ari0 ptpEs and Duluth intrusions (Bonnichsen 1972). til ru 8l5lfl810 @r.ftrt A fourth type of mineralization seems re- Ptl(Pt + Pd) Cu/(Cu+ lil) Fbf€two stricted to tle area south of Potgiotersrus. In uppar:one, bold @ln @gmtlb lqJror 0.&. 1 this type up to 10 madal Vo sulfides occur lppor zono, beld l@r @gnotlto l!J8r 2 0.45 0.85 2 Betiol !E!, tgp of @lneEllzod lone 0,50 3 evenly dissemina,ted throughout fine-grained bott@ of olrerqllzod a@ 0,40 3 norite. Although the position of this layer in MereGky left asEge 0.70 0.38 4,5 Psoudoreef 0.64 tle sequence is uncertain because of structural Plat@f, arengs 0.44 0.44 4,5 complexities, it is tentatively grouped with the tE2 chMltlto ldj€r lo.zz)r 4,5 marginal zone of the complex by L. Hulbert Vlakfontein olckol plp$, @ntm 0.5 0.07 6 edgo 0.29 0.59 6 (research in progress). on@nocht dmltr plpa 0.99 7 Much more detailed information is required TRfonteln (ln fl@prccts) 0.65 1 xlncl!&s on the various types of mineralization at the lll ln ch@lb. kfeHcos: I Llabonhrg (1970)iZ vd cru|@ldt (t976)i 3 Buch&u base of the mafic sequence Potgietersrus ttezor;5 bn G'*mrdt (re?7)i6 Ya@!k at be- [i3;!]i,trti"ilx;f,,S;i1oi fore they will be understood. The complicated relationships at the floor contact, the develop- Buchanan (1975) indicate that the mafic rocks men't of platiniferous sulfides in the floor rocks of the complex are in places intrusive into and in rafts of floor rocks in the Platreef, all dolomitic limgsfesss of the Transvaal se- point towards a complicated process of inter- quence. These relationships, as well as the action between the Bushveld magma and its presence of intercumulus carbonate minerals floor. Processessuch as sulfurization (Lieben- in the ,mineralized zones, have prompted de berg 1970) or CO, and H:O flushing (de Waal Waal (1977) to suggest that the upper-zone 1977) as coritrolling factors in sulfide precipi- magma transgrcsed the lower-zone rocks but tation at the base of the layered sequence seem in places also dolomite and so became contami- very a,ttractive models on which to base fur,ther nated with COn and IIgO. investigations. Features such as the pegmatitic Notable is the comparatively low Cu/(Cu * nature of the mineralized portion of the reef, Ni) ratio of the sulfides in the Bethal area as the slightly higher Cul(Cu + Ni) ratios as compared with mineralized layers at the base compared for example with the Merensky Reef of the upper zone in the eastern Bushveld (Table 2), the development of a'bundant gra- (Table 2). This feature strongly supports the phite, the skarn-type mineralization and still suggestion made earlier that the upper-zone limited su{fur isotope data (Liebenberg 1968), lithologies in the Bethal area reflect a lateral all point towards volatile activity, probably expansion of the magma chamber subsequent caused by degassing of the dolomite during to the addition of $ubstantial quantities of un- thermal metamorphism. differentiated magma after about 2A crystalhza- The Bethal area. Deep drilling in the Bethal area tion of the main zone. Transgxession of this of the Bushveld Complex has revealed the ex- lnagma across sedimentary rocks, poesibly dolo- istence of sulfide-mineralized upper-zone type mite, during lateral expansion would, result in rocks beneath a cover of Paleozoic sedimentary the incorporation of volatiles (COz, IIO, pos- rocks (Buchanan 1975). The,mineralization is sibly also S) to produce the magnetite and confined to coarse-grained, magnetite-bearing comparatively Ni-rich sulf,ide-bearing pegmatitic pegmatitic in the lower l2O m of the rocks near the base of the succession. This upper zone. Sulfides are irregularly d,isseminated interpretation di'ffers slightly from that of throughout and in places occur in fairly massive Buchanan (L975, 1979) who suggestedthat the patches several cm in diameter. The sulfidee are magma from which the iron-rich upller zone essentially pyrrhotite, usually between 85 and crystallized was injected as a separate igneous 90 modal /o, with only subordinate amounts of event. chalcopyrite and pentlandite. There is, however, Disseminated a grad,ual upward increase in the Cu/(Cu + Ni) statiform enrichments ratio from about O.40 at the base to 0.60 at UG2 chromitite layer. Many of the chromitite the top of fhe 12O m thick rnineralized zone layers, especielly those higher in the sequence, (Buchanan 1972).1Xe qrinslalized zone over- contain small quantities of sulfides and associated lies a 5Q m tlick suscession of harzburgite and minerals of the platinum-group (PGM). The pyroxenite of the lower zone, which seems to highest concentrations occur in the UG2 chromi- imply a major transgression of the upper-zone tite layer (Fig. 4; notation after Cousins & Ferin- magma onto underlying mafic and ultramafic ga l9@) but published information on the plati- rocks. Furthermorg borehole data presented by num-group-element concentrations and the the layer, and noted that Pt/Pd is low near the base but higher in the middle or at the top. The majority of the PGM, usually more than 8O/o, occttt along grain boundaries or rn asso- ciation with interstitial sulfides; only a very small proportion are enclosed in chromite or in the intercumulus silicates. In drill cores from most of the areas investigated, Mclaren found the platinum sulfides , , cooperite and an unnamed mineral with the approximate for,mula Pt Rh(IrEuS to be the most abundant carriers o,f PGE. However, in the Maandags- hoek area of the eastern Bushveld, from where exceptionally high PGE values of about 10 g/t are recorded, the PGE form cornplex alloys with a variety of other metals, notably Pb, Cu, Fe, Mg, Bi, Ten Sb and Ag. Platinum sulfides are rare, Iaurite being the most common. Other sulfides are mostly pentlandite, pyrrhotite and chalcopyrite. As can be seen from Table 1, the Ni and Cu contents of the UG2 layer are considerably lower than those of either the Merensky Reef or the Platreef, which reflects the comparatively sulfideloor nature of this layer. This feature is repeated in the chromitite stringer of the Merensky Reef [except at Marikana (Nevrman 1973, Brynard, et al. L976)l; it contains the highest platinum values, whereas the bulk of the sulfides ocsur in the pegnratitic part of the LEGEND reef. mottlsdcnorthGlts Available information suggests that the re- sourcesof PGE contained in UG2 layer spoitodq$rthoslto ord loncondit€ the are almost twice those of the Merensky Reef. The norilg chromite in this layer is of a rather low qualr'ty; bl@othlc pyrorertlro it has a Cr/Fe ratio of 1.35 and a CrzOgcontent poltnotiticpyorsnlt9 of. 43.5Vo, as compared to values of 1.6 and Ellllds [{nsmllzoibn 46.5, respectively,for the LG6 or main chromi- chmmlt0 tite layer. But, should an economical beneficia- tion process for this low-grade chrome be- 4. Fro. Generalized columnar sections tiro'rgh tho come available, this layer could become the upper critical zone of the eastern and western most important Bushveld Bushveld,ilustrating amongstothers also fle pro- ore deposit in the nounced lateral va"riation in the relative poeition Complex: 5.5 x 1tr t are available to a depth of the componentsof tle Merensky Reef. of 12@ m (von Gruenewaldt 1977). The Merensky Reel. Undoubtedly the most nature of the PGM and other sulfides is limited. peculiar layer in the whole sequence of mafic Although Wagner (1929) records values from rocks is the Merensky Reef (Fig. a). Because many localities, the highest being 19.0 g/t, it of its economic importance, it has been drilled is not always certain whether these represent extensively and has been investigated in rnore the UG2 layer or other layers wiflin the upper detail than other layers of the complex. Ver- goup. Mclaren (1978) has recently published maak & Hendriks (1976) recently redefined some preliminary results of his investig-ations of the reef as comprising a mineralized pegna- borehole material from the eastern and w$tern titic, feldspathic pyroxenite bounded at the top parts of the complex. These results point to con- and bottom by thin ohromitite stringers. Ac- siderable lateral variations of the PGM, even cording to earlier definritions tle terrn referred over short distances. He found the highe.st FCIE to the associated porphyritic pyroxenite. The concentrations near tle bottom and middle of present au,thor considers the definition by X SULFIDF.S IN THB BUSHVELD COMPLEX

Vermaak & Hendriks to be too restrictive, be- o1 16s mineralizsd pertisns of the reef is usually sause esonomic sulf.ide mineralization is not less than 3 vol, Vo, Mineralization oscurs as always confined to the pegmatitic part of the irregularly shaped patches occupying, together reef and because the richly mineralized chromi' with plagioclase, clinopyroxene' biotite, horn- tite stringers may be developed in the porphy' blende and magnetite, the interstitial spaces be- ritic pyroxenite (Schwellnuset aJ. 1976, Brynard tween the large cumul'us orthopyroxene crystals; et aI. L976). It is therefore proposed to define (vii) although the sulfides, in decreasing order the Merensky Reef as the basal, pyroxenitic of abundance, are usually pyrrhotite, pentlandite, portion of the Merensky cyclic unit; this in- chalcopyrite and pyrite, this sequenceis reversed cludes the porphyritic pyroxenite, the pegma- at the Atok Platinum mine in the northeastern titic pyroxenite and any chromitite stringers Bushveld where pyrite ) chalcopyrite > pent- that may ,bedeveloped (Fig. 4). landite ) pyrrhotite (Schwellnus et al. 1976). It is not within the scope of this paper to This together with the development of cubanite describe the reef in detail; schematic sections in certain areas (Liebenberg 1970) indicates of the reef are presented (Fig. ) in which lateral fluctuations in the sulfur content variations of the relative positions of the various of the sulfide liquid that separated from components are shown .for different parts of tle silicate magma; (viii) considerable varia- tle complex. For f,urther detail the interested tions have been recorded in the abundance reader is referred to the excellent contributions of the various PGM along the strike of the of Cousins (1969), Liebenberg (L97O), reef (Brynard et aI. 1976). The platinum Schwellnus et al. (1976), Vermaak & Hendriks sulfides cooperite and braggite predominate at (1,976) and Brynard et al. (1,976). For present the Rustenburgmine; the arsenidesand tellurides purposes it is considered advisable to point out sperrylite, kotulskite and moncheite considerably certain aspects of the reef that highlight its outweigh the sulfides at the Western Platinum complex nature, aspectsthat must be borne in mine; the Pt-Fe alloys and laurite are very com- mind when considering its origin. (i) The reef mon at the Union mine. Interestingly' the type forms the basal pyroxenitic part of a cyclic unit and relative abundance of the PGM at Atok referred to as the Merensky unit (Fig. 4), which, and Rustenburg are si'milar even though major accordingto Vermaak (1.976b),forms the most sulfide mineralogy of the two areas show pro- complete of a seriesof broadly similar over- and nounced differences. Furthermore, the major underlying cyclic units; (ii) ttre cyclic units mineralization at Atok is in porphyritic pyroxe- of the Merensky type us,uallydisplay an upward nite and in two associatedchromitite stringers basification of the plagioclase (varr Zyl t970, near the top of the reef, whereas the mineraliza- Vermaak L976b). In the Merensky unit the tion at Rustenburg occurs in pegmatitic pyroxe- An content increasesby a few mol Vo from the nite and chromitite at the base of the reef; base to the top, whereas the associatedcumulus (ix) detailed mineralogical investigations have orthopyroxene displays a normal upward Fe shown that by far the majority of the platinum- enrichment; (iii) the reef has a sharp footwall group minerals are developed along the outer contact. Underlying rocks vary from one local- edges of the sulfide grains and as separate ity to another, but they are usually either mottled minerals between interstitial silicates. Only a or spotted anorthosite, although porphyritic small proportion (- 15Vo) occurs within the (x) pyroxenite may constitute the footwall at places sulfides (Vermaak & Hendriks L976); the the in the easternBushveld. This has widely been in- highest PGE values are usually found within terpreted as indicating a hiatus in the crystalliza- chromitite stringers. The only major exceptions high tion d.uring which the floor was ooeroded"in are in the Western Platinum mine where places prior to deposition of the reef; (iv) the values are commonly found a few cm above relative position of the chromitite layers, the this stringer (Newman L973) and' with areas the reef pegmatitic pyroxenites, the porphyritic pyroxe- of late-stage deuteric alteration of (xi) the reef, nite and the sulfide mineralization in the se- (Crocket et al. L976); within platinum mineralogy quence within the reef shows considerablelateral considerabledifferences in with the sili- variation (Fig. 4); (v) the composition of tle are evident; the PGM associated the layer. chromite within the reef displays, according to cates differ from those in chromitite (Pt higher in the Vermaak & Hendriks (1976), an upward in- Pt/ + Pd) ratios are usually with creasein the Cr/Fe and Mg/Fe ratios, although ohromiti.te layer than in sulfides associated this trend is normal (i.e., decreases) in the the silicates; (xii) the high Pt/(Pr + Pd) Merensky Reef at the Western Platinum mine ratio and tle low Cul(Cu + Ni) ratio (Table (Brynard et al. 1976); (vi) thE sulfide content 2) illustrate the high Pt and Ni content 246 THE CANADIAN MINERALOGIST of the Merensky Reef compared with many nature are pieserved in the so-called "boulder other deposits associated with tholeiitic in- bed" at Rustenburg (Cousins 1969, Figs. 7 and trusions (Naldrett & Cabri 1976). 8) where for some rezlson the process to form Any "hypothesis for the genesis of the Me- a cyclic unit of the Merensky type was not com- rensky Reef, whether by differentiation by pleted. (It is of interest to note that Lee & gravity settling in the mafic portions of the Sharpe (1978) called upon a process of spheri- Bushveld Complex as a whole, or whether from cal agglomeration to account for similar boul- a inore restricted pulse (heave) of magma, ders in gabbroic rocks of the main zone, close create as many problems as they solve" (Lauder to the margin of the complex). Extraction of 1970), The validity of th,is statement was am- FeO for the chromite and silicates lowerd plified by Vermaak's (1976b) thorough treatise the sulfur-carrying capacity of the magma and on the reef and its associated rock types. He resulted in the separation of an immiscible sul- reviewed the various hypothesesand tested them fide melt. This melt settled to fill interstitial against the wealth of information amassedbv the spaces in the pegmatitic basal layer of tle reef, Johannesburg Consolidated Investment Com- in places even percolating through the reef to pany over many years of exploratory drilling become entrapped in interstitial spaces of the and laboratory investigations. On the basis of footwall anorthosite. Through continued crystal- all these data- he postulated a crystall2ation lization of chromite and , the upper model that accounts for the majority of tle chromitite layer and the porphyritic pyroxenite featlrres displayed by the Merensky Reef. This accumulated, tle latter grading upward into hypothesis, also extended to all the cyclic units norite and anorthosite to complete the cyclic in the upper part of the critical ioo", unit. "uo briefly be summarized as follows. The imrniscible Cu-Ni sulfide liquid acted as Fractional crystallization of the Bushveld a collector not only for the PGE but also for magma resulted in an increase of the components small amounts of Ag, Bio Te, Sn, Sb and other of plagioclase until a stage was reaohed when rare metals to for.m the multitude of different this mineral was on the liquidus. Anorthosite PGM encountered in the reef. The presenceof layers that crystallized from this liquid are Cu, Ni and PGE is considered by Vermaak to characterized by an upward basification of the represent a progressive enrichment of the main plagioclase which is ascribed bv Vermaak to body of magma through expulsion of interstitial process post-cumulus floating of the plagioclase, the upward movc- liquids by a of growth or pressing. ment being arrested by a compositional and/or filter He attributes the association of density and/or temperature inversion. Floating base-metalsulfides and PGM with the chromi- of crystallizing plagioclase resulted in a mat of tite layers in the upper part of the critical zone crystals that gradually thickened wing to under- to a decreasein the FeO content during crystal- plating of continued plagioclase crystallization. lization of a chromitite layer, a resultant in- As a result of this process the magrna beneath crease in the sulfur fugacity of the crystallizing magma and the separation of an immiscible the mat was trapped and isolated from the sulfide melt. magma chamber. Simultaneously, the trapped magma was enriched in volatile constituents, Vermaak's model, based on a wealth of data and also in sulfur, PGE, Cu and Ni through the from underground workings and boreholes in utrrward migration of the intercumulus liquid the western lobe of the complex, seems to frm the consolidating pile of cumulates below. account for most of the observed features in Cardinal to the subsequentcrystall2ation of the the reef. His model could be adapted slightly magma below the mat is the degree of imper- to account for lateral variations in the nature meability of the anorthosite ,mat and conse- of the Merensky Reef; the absenceof tle peg- quently its efficiency as a trap for the volatile matitic portion of the reef may indicate that constituents.Removal of the plagioclase caused conditions for the aggregation of boulders were enrishment of the ferromagnesian constituents not met, possibly due to absence of sufficient which initiated crystallization of chromite to volatiles in the crystallizing magma. He. re- form a thin layer at the base of the unit. When cognizes that his model does not account satis- lle ferromagnesihn minerals started to crystal- factorily for certain aspects, e.g., tbe mecb- lize, the volatile-enrished magrna enhanced iapid anisms involved in the precipitation of the second growth and crystal aggregation into spherical chromitite layer, and whether depletion of FeO boulders that sank in the magma, dimiled tle during crystallization of the chromitite was suf- early chromitite layer and merged to form the ficient to precipitate sulfides from the entrapped basal pe8rnatitic pyroxenite. Boulders of this Merensky magma. SIJLFIDES IN THE BUSHVBLD COMPLEX 247

Several authors (Hamilton 1977, McCarthy As noted in many of the chromitite layers in & Cawthorn in press) have recently proposed the Bushveld Complex, the chro'mite commonly that the Merensky Reef represents a relatively shows signs of late-stage reaction and equilibra- late-stage product of fractional crystallization tion (Sampson 1932, van Zyl 1970, Schwellnus of an early magnesian magma. This, as well as et al. L976). Therefore, it is suggestedthat the Vermaak's (L976b) contention that the sulfides PGM of tle chromitite layen for'med at a late in the Merensky Reef originated essentially by stage when deuteric liquids percolated upward a process of filter pressing of intercumulus through the pile of cumulates. Reaction of liquids from underlying curnulates raises the these liquids with the chromite and the intersti- question as to whether such an origin for the tial silicates of chromitite layers resulted in con- reef can be reconciled with the comparatively ditions favorable for the precipitation of discrete low Cu/(Cu + Ni) ratio of 0.38 for the sul- PGM. Whether the PGE were transported and fides in the reef. A concept of intermittent introduced by late hydrous solutions (Stumpfl & rqilenishment of the crystallizing rnagma seems Tarkian L976) or were present in the spinel more attractive to account for the relatively low either as minute mineral inclusions (Cousins Cu/ (Cu + Ni) ratio. Crystallization of several & Vermaak 1976) or in solid solution (Naldrett thousand metres of bronzitites and interlayered & Cabri 1976) is debatable.The exact processes harzburgites, dunites and without involved are by no means clear at this stage, periodic replenishment of the magrna would but the extremely small grain size of the PGM have extracted considerable quantities of Ni and associated Ni-Cu sul'fides and their bulk from the mag(na, so that a higher Cu/ (Cu * composition testifies against their precipitation Ni) ratio would be expected than that observed from the overlying magma during crystalliza- for the reef. A further problematical aspect is tion of the chromite. tle enrichment of PGE in certain of the chromi- The upward increase in PtlPd ratio described tite layers relative to the sulfide concentrations by Mclaren (1978) from various localities of in the silisates. This chromitite, described by the UG2 layer might possibly be ascribed to Wagner (L929) as behaving in effect "like a the preferential collection of Pd in sulfides at sponge" with the PGE, not only characterizes the base of the layer during the upward mi- the Merensky Reef and the UG2 chromitite gration of the deuteric fluids. The very low layer, but also xenoliths of shromitite in the Cu/(Cu + Ni) ratio of the UG2 chromitite Onverwacht pipe (see below) and, to a lesser layer cannot be evaluated at this stage because extent, other shromitite layers in the upper half the analytical values available represent only of the critisal zone. This feature is not satisfac- bulk samples of the reef. Mclaren (1978) torily accounted for merely by precipitation of records Ni average values of 0.tVo or greater. a sulfide liquid in response to decreasing S The bulk of this Ni would clearly seem to be solubility of the magma during chromitite crystal- bound in the chromite structure when the values lization. In such case, the composition of the are compared with analyses given by de Waal sulfide precipitated should be similar to tle (1975) for purified chromite from the UG2 bulk of the s'ulfides in the silicates, e.9., in the layer. From the description of the UG2 layer, pegmatitic pyroxenite of the Merensky Reef. pgntlandite seems to be the most abundant sul- The grain size of ,the PGM in the chromitite fide present, so that the Cu/ (Cu + Ni) ratio is usually half or less than that associated with of the sulfides in the reef is proba:bly also sulfides in the pegmatitic portion of the Meren- relatively low, a fea.ture that may be related to sky Reef. Although this latter feature may be subsolidus equitribration between sulfides and ascribed to the smaller grain size and closer chromite. packing (Vermaak of the chromitite I976b), Sullide enrichments in the upper zone. Con- the presence of fairly large amounts of inter- centradons of disseminated sulfides are known (mostly plagioclase) cumulus silicates would from various layers in the upper zone of the militate against such an explanation, especially complex (von Gruenewaldt 1976).In the lcvwer if it is assumed that the inmiscible sulfide part of this zone, sulfides are developed at crystallization of liquid formed in response to the top of anorthosite layers, directly beneath the chromite and . Grain sizes of the thicker of the magnetite layert, where PGM and associatedsulfides in the UG2 chromi- they usually occupy the intercumulus spaces tite layer vary considerably, but average values of the plagioclase cumulate. In contrast, presented by Mclaren (1978) indicate dia- the sulfides in the upper half of the zone meters of the order of 5.0 and 0 p,m, reE ec- occur evenly distributed as tiny, spherical, tively. droplike bodies included in silicates; they THE CANN)IAN MINERALOGIST usually make up about 0.5 vol. Vo of the rock. lizing magma was sufficiently reduced, as for Larger concentrations, both as tiny droplets and instance during formation of the thicker magne- as larger, lenticular bodies several cm in dia- tite layers, did small quantities of immiscible meter, are developed within the thick upper- sulfide liquid separate and settle to the floor most magnetite layer. where they became concentrated in the inter- The composition of the sulfide assemblagein cumulus spaces of tle plagioclase cumulates the rocls of the upper zone shorrs a gradual and directly belorr the rmagnetite layers. During systematic variation upward in the succession crystallization of the upper half of the upper In the mineralized anortlosites near the base ot zone, the magrna must have been very close the upper zone, the sulfide assemblage contains to saturation in sulfur, because crystallization the highest chalcopyrite content of the layered of only small quantities of silicates resulted in mafic sequencq in the order of 55 vol. Vo. Both the separation of numerous drop.like bodies of chalcopyrite and pentlandite gradually decrease sulfide liquid evenly dispersed throughout the in abundance upward in the sequence with a cumulates. concomitant increase in pyrrhotiite the conterit. Mafic and ultramalic pegmatites Fyrrhotite attains its maximum concentration, between 9O to 95Vo of the sulfide assemblage, Mafic pegmatite bodies displaying a large in the droplets of the rocks in the upper half variety of mineralogical com.positions are widely of the zone. developed in the layered sequence of the Bush- (Willemse Characteristis of the sulfides of the upper veld Complex 1969, von Gruenewaldt 1,973). The bodies zone is their very high Cu/ (Cu }' Ni) ratio, are usually concentric and pipeJike although pegmatite unsually about 0.85 or,more. A Ptl(pt + pd) diallagite com- ratio of 0.45 is recorded by von Gruenewaldt monly forms large, irregularly shaped masses, (1976) for the mineralized anorthosite below especially in the critical zone (Cameron & lower .magnetite layer 2, the only value so far Desborough 1964). Only the hortonolite dunite pipes available from the upper zone. Little is known and the nickeliferous bronzitite pipes a'bout the lateral distribution of the sulfides in carry significant quantities of PGE and nickel, the upper zone, mainly becauseof poor exposure respectively, and these will therefore be de- in most of the areas underlain bv the rocks of scribed in more detail below. However, small quantities this zone. The above information on the sulfide of sulfides are developed in many of the concentrations in the upper zone pertains to an otler mafic pegmatites, e.g., in the diallage pegmatite area of exceptionally good exposure in the surrounding the magnetite plug at eastern Bushveld. It would be unwise at this Kennedy's Vale (van Rensburg 1965), the vermi- stage to extrapolate observed trends to the other cutte.bearing breccia pipes in the upper zone parts of the complex, as the available evidence (von Gruenewaldt 1973), and the anorthositio pegmatites suggests that the rock sequences in the upper in the upper zone. The dissemina.ted zone of the Potgietersrus, the western and even sulfides associated with these pegmatites have yet the.southeastern (Bethal) area of the complex not been investigated. deviate considerably from those in the easlern Nickelilerous bronzitite pipes. More than 200 Bushveld. From this it must be deduced that nickeliferous bronzitite pipes have been located crystallization histories in the upper parts of the in an area of about 2O0 km' situated to the west complex differed from one localitv to the next of the Pilanesberg in the western Bushveld and that the nature and composition of sulfide (Verrnaak 1976a). The area is underlain by rocks concentrations may differ accord,ingly. Dissomi- of tle lorver zone and of the lower part of the nated Cu-rich sulfide concentrations are known critical zone, i.e., mostly by bronzitite and to be associatedwi.th certain of the lon'er magne- harzburgite. The bod,ies are generally pipe- tite layers in the Potgietersrus area and in-the like but they may show extensive mushrooming western Bushveld. in the pyroxenite, usually directly beloq' harz- The presence of sulfides in the rocks of the burgite layers. Most pipes are overlain ,by opaline upper zone is ascribed by von Gruenewaldt gossans, which are considerably larger in sur- (1976) to a gradual enrichnent of sulfur in the face diameter and commonly also in volume crfstalliz-ing magtna due to processes of frac- than the underlying sulfide orebody. Although tional crystall,ization. The pronounced iron en- the largest gossaD measures 25 m n diameter, richment in the Bushveld magma prevented tle most of the gossan occurrences are considerably formation of immiscible sulfide droplets during smaller and have an average diameter of 7 m crystallization of tle lower half of tle upper (Vermaak 1976a). zone. Only where the FeO content of the crystal- Idealln the nickel pipes consist of massive SI'LFIDES IN THE BUSEVELD COIV1PLBX 249 sulfide surrounded by what Wagner (L929) and that these liquids migrated into zones of termed "poikilitic oreo', which in turn has fairly weakness to form the orebodies. The sulfur is sharp contacts with the country rock. Many of considered to be essentially of sedimentary the country rocks do, however, contain dissemi- oriein, and Liebenberg therefore related the in- nated sulfide, and Liebenberg (1970) noticed tedtitid zulfides and c6nsequently also the nickel an increase in these d'isseminations towards the pipes to the downward transgression of the Bush- pipes. The mineralogical composition of the ore veld magma into sulfur-bearing argillaceous varies; pyrrhotite always predominates, consti- rocks in the area to the west of the Pilanesberg. fitng TAVo or more of the sulfides, followed Detailed mapping in the western Bushveld led by pentlandite, chalcopyrite and mackinawite Vermaak (1970, I976a) to recognize areas of (Liebenberg L970). Although Liebenberg does intense basification (serpentinized harzburgite) not describe any pyrite from the nickel pipes, of bronzitites along fault and fracture zones this sulfide seems to predominate in some of with which the nickel pipes are associated.He tle pipes in which nickel values are particularly postulated that the basification of the bronzitite low (Vermaak L976a).. 'poikilitic layers developedas a result of testonic oYerpres- The main ore type is the so-called sure. The metals required for the Ni pipes are oreo, which commonly constitutes ap to 80Vo considered to have been liberated from the sili- of the pipe. ft consistsmostly of euhedral ortho- cates during basification, whereas sulfur was in- pyroxene crystals set in a matrix of sulfide. In troduced into the fracture zones from the under- exceptional cases the poikilitic ore contains lying sediments by circulating waters that reac- large, euhedral olivine crystals partly replaced 'intruded' ted with the metals to form sulfides. These in by pyrrhotite and is by a later ortho- turn were remobilized and concentratedin favor- pyroxene poikilitic ore. Detailed descriptions able structures along the fault zones during re' of the mineragraphy of the ore are provided newed overpressure.Most of the fracturing and by Liebenberg (1970) and will not be repeated faulting is related by Vermaak (1976a) to em- here. Of interest, however, is the zonation in placement of the Pilanesberg alkaline complex' the sulfide pipes: the Cu sulfides increase with the result that formation of the nickel towards the margins of the pipes. Cu/(Cu * pipes is considered to postdate the complex. Ni) ratios consequently increase considerably from about 0.07 in the central, massive ore to Temperatures during emplacement of the 0.59 along the edge of the pipes. Similarly the nickel pipes must have been high, probably in Ptl(Pt + Pd) ratio changesfrom about O.5 in excessof those at0ainedin the area of maximum the centre to 0.3 or less irear the edge (Vermaak pipe developmentduring intrusion of the Pilanes- 1976a). berg. This deduction'is made from var'ious Other minerals observed by Liebenberg features described by Liebenberg (1970), (1970) and Vermaak (L976a) are chromite amongst others the replacement of niccolite by within the peripheral ores, and graphite, biotite, maucherite and the development of orthopyro- chlorite and talc around the outer circumference xenb and olivine in the poikilitic ore' all of of the pipes. Contacts of the pipes with their which indicate temperatures well in excess of country rocks are usually gradational over less 800'c. than 3 cm with a fine-grained variety of the De Waal (1977) recently suggesteda process poikilitic ore developed along the contact. whereby release along fractures of carbon Both olivine and orthopyroxene in the poiki- dioxide and water from metamorphic reactions litic ore are enriched by 10 to lSVo of.the Fe-rich in the floor of the complex could be held res- end-member compared to the country rock. This ponsible for the nickel pipes at Vlakfontein. feature is in keeping with the majority of the Where the COz-HzO fluids entered the magma' transgressivepegmatite bodies; all are consider- precipitation of sulfides and spinel took place ably enriched in iron. if the magma was driven to saturation wit} Three diverging hypotheses have in recent respect to these two minerals. The emergence years been advanced for the origin of the nickel of these fluids into the magma will, according pipes. All three, however, have two aspects in to de Waal (t977') be focused at points along commoq i.e., that the siting of the pipes is the fault and fracture zones. In the immediate structurally controlled and that the sulfur is vicinity of these COI-HO "fountains", f(Os) partly derived from an external source. Lieben- in the magoa was raised and precipitation of berg (1970) considered the pipes to have origi- sulfide liquid resulted, a process that was aug- nated by segrega.tion of sulfide-enriched inter- mehted by sulfur in the fluid. New magma' cumulus liquids under conditions of tectonism from which the metals of the sulfides were 250 THE CANN)IAN MINERALG}IST extracted to form the pipe'like bodies as the after their discovery in the early twenties. pile of cu,mulates grew, was supplid by slow Consequently a considerable amount of infor- convection of the magrna in the chamber. The mation on the PGM of the Driehoek pipe is zones of basification are considered by de Waal avarlable, as rfl.as recently summarized by to represent areas where near-consolidated Tarkian & Stumpfl (L975). These authors found bronzite cumulates reacted with the carbon tlat the PGM.in the Driekop pipe are iron- dioxide-water fluids to form harzburgite ac- bearing platinum (50Vo), sperrylite and gever- cording to the reaction: pyroxenite * vapor -+ site (30vo), minglals, of the hollingworthite- harzburgite * melt. iranite group (LSVo) and a variety of sulfides Hortonolite dunite pipes. Numerous bodies of and ant:monides (SVo). They point out that dunitic and pyroxenitic pegrnatites occur in the mineral association differs distinctly from rocks of the critisal zone in the eastern and the sulfide- and telluride-bearing assemblagesof western Bushveld (Wagner L929, Cameron & the Merensky Reef. Desborough 1964, Willemse 1969). They are From detailed field and microscopic observa- either lentisular flat parallel bodies to the tions, Ca,meron& Desborough (l9M) presented stratif,ication, irregular in outline, or pipe-like. convincing evidence that the olivine dunite It is these pipe-like pegmatitic dunite bodies surrounding the hortonolite dunite at the Onver- that have become well-known in the geological wacht pipe owes its origin to replacement of literature for their remarkably platinum high bronzite in the original chromite-bearing pyro- content (up to 20@ Only e/t). the Onver- xenite by olivine. Although the investigations wacht, Mooihoek and Driekop pipes contained of these authors were not conclusive regarding sufficiently high platinum values to warranr the origin of the central hortonolite dunite, they exploitation. drew analogies with similar rocks in other ir- Both the Mooihoek and Onverwacht pipes regularly shaped dunite pegmatites and con- consist of a circular, carrot-shaped core of cluded that the dunite also formed during the coarse-grained hortonolite dunite (Fooo), the replacement process. They were of the opinion main platinum-bearer, surrounded by medium- that water-rich, high-lsmperature fluids were grained dunite (Foao). At Mooihoek an outer the agents for pegrnatite formation and, al- shell of coarse, pegmatitic diallagite and feld- though they could not specify the time of spathic pyroxenite is developed (Wagner l9Z9), formation and the source of such fluids, tley The country rocks of the Onverwacht pipe are pointed out that all the constituents required the pyroxenites below and above the main in the replacement reaction are available in the chromitite layer. Prior to rnining, huge rafts rocks of the critical zone through which tbe of the chromitite were preserved in a relatively fluids must have passed. On the basis of their undisturbed position in the pipe, and these rafts 'hcted investigations of the PGM of the Driekop Fip€, like spoDges so far as platinum was Stumpfl (1974) and Tarkian & Stumpfl (L975) concerned" (Wagner 1929, p.68). In contrast, supported the conclusions reached by Cameron the country rock of the Mooihoek pipe is anor- & Desborough (1964) and considered that tle thositic norite, and the pipe is therefore situated PGE were introduced by and deposited from at a slightly higher stratigraphic level. fluid phases during the late stages of formation The Driekop pipe differs from the previous of the dunite pipes. two in that the pipe is not zoned and the central, platinumjbearing portion consists of hyalo-- siderite-chrysolite (Fozr-ea) composition. Al- DrscussloN though Wagner (1929) described tle platinurn- bearing dunite to be enriched in iron compared Various authors (e.8l.,Willemse 1969, Lieben- to the barren dunite and to consist of a large berg L97O, Vermaak L976a) have commented number of irregular and composite bodies in on the S-poor natule of the Bushveld magma. the more magnesian dunite, Heokroodt (1959) A shallow u,pper,mantle origin in a S-deficient could establish no relationship between the zone'as postulated by Naldrett & Cabri (1976) platinu,m content of the ore and tle composition seems therefore indicated. From the available of the olivine. The pipe is losated in norite of evidence at present, an early immiscible su[- the cnitical zone at a stratigraphic position fide melt that could have separated from tle considerably higher tlan the Onverwacht and Bushveld magma to form a massive deposit Mooihoek pipes. Exploitation of the pipe con- near the floor is therefore not to be expected. tinued until 1961, in contrast to the other two Wherever large concentrations of sulfides are pipes at which mining operations ceased soon developed near the base of the complex, as in SIJLFIDBS IN THB BUSH\IBLD COMPLBX 25r the Potgietersrus area, the western Bushveld and tle Bethal area, evidence is at hand that the Bushveld magma assimilated sedimentary sulfur and other gases such as would prcmote separation of an early sulfide fiquid. Concentrations of sulfides at highel levels in tle intrusion must be related to a combination of processes, such as periodic replenishrnent of undifferentiated magma to maintain a relatively low Cu,/(Cu + Ni) ratio in the magnra at least up to the level of the Merensky Reef, enrich- ment of the magma in sulfur by normal pro- cessesof fractional crystallization and expulsion of the intercumulus liquids from the accumulat- ing pile of crystals as a result of compaction. These highly fractionated liquids, which as- cended through the pile of cumulates, possibly augrnented by COr and HzO-rich fluids from degassing metamorphosed sediments, may be of considerable importance in consideration of the origin of the large variety of rnafic peg- matites developed in the complex. The initial porosity of the settled crystals in the accumulating mush is estimated as being between 20 and SOVoby various authors (Hess 1960, Jackson 1961, Wager & Brown 1968), with the result that the mechanism of enlarge- ment of cumulus crystals to form adcumulates or monomineralic rocks without pore material is difficult to explain. Quantitative infonnation on the amount of intercumulus liquid present directly after settling of pyroxene and plagio- clase in cumulates of the main zone can be obtained from point counts of the so-called "black norite" developed to the west of Belfast (Fig. 1). Numerous tiny needles of rnagnetite and ilmenite in the plagioclase impari the dark color to the rock. In several of the specimens investigated the needles are restricted to the lat6:r; central, evidently cu,mulus parts of the plagio- clase grains, whereas the adcumulus overgrowths are free of needles (Fig. 5a). Point counts of Fro. 5. Featuresof plagioclasecrystals in tle matn these sections showed that the overgrowths zone. (a) Fine magnetite-ilmenitedusting of the (b) on the plagioclase make up between 1O and cumulus portions of plagioclasecrystals; In- plagioclase; (c) ISVo of. the rock. If this figure is extrapolated teqrenetrating crystals of ry' in interpenetrating p)noxene, velopment of myrmekite to other cumul.us minerals, especially plagioclase crystals. it can be shown that adcu,mulus overgrowths constitute more than 20 vol. Vo of. gabbraic cumulates in the main zone. supernatant magma; thirdly, a conbin&tion of Preferred explanations for the mechanism both processes could have occurred. The first of enlargoment of cumulus crystals in adcumu- of these processes can be rutd out as lates were pointed out by Cameron (1969): G.B. Hess (L972) consluded from tlermal firstly, settled crystals continued to grow after considerations that crystallization of the pore shallow burial by diffusion of the required con- material occurs only long after deposition and stituents into the crystal mush down a slight that ionic diffusion botween crystallizing pore temperature gradient; secondly, enlargement of material and the bulk of the nagma is not crystals occurred while still in contadt with the poesible. 2s2 THE CANADIAN MINERAI,OGIST

Another mechanisn whereby the amount of (1976), Vermaak & Hendriks (1976) and intercumulus liquid aan be reduced is com- Mclaren (1978) sugge$t$that all the PGE in paction of the crystal mush prior to cementation chromitite of the Bushveld can ,be accounted "if this were attended by re-solution of crystals for in discretePGM. Data presentedby Mclaren at points of contact and redeposition in in- (1978) for the UG2 chromitite layer show terstices" (Cameron L969). However, Cameron furthermore tha,t the proportion of FGM con- states that owing to lack of evidence, this tained as inclusions in chromite grains is, on process has found litde favor with students of the average, only about 4Vo of. the total PGM. magmatic sediments.Interestingly in this regard, Although most authors agree tlat PGE have Cameron (1975) concluded that minimal post- an affinity for chrome spinel at high temper- cumulus overgrowth has taken place in chromi- ature, be it in solid solution or as minute in- tites and bronzitites of the Bushveld Complex clusions, the abnormally high concentrations and consequentln minimal amounts of post- in the iron-rich UG2,and Merensky Reef chromi- cu,mulus equilibration. This was interpreted by tite as compared with chromitite layers lower Flynn et al. (1978) as being indicative that in the sequence remains problematical. Of in- intercumulus liquids were in close equilibrium terest in this regard is that where Cr/Fe in the with the cumulus crystals, situations which Steelpoort Chromitite layer was reduced by would prevail under conditions of crystalliza- equilibration with late liquids during formation tion close to the magmatic solidus. of the Onverwacht pipe, the chromitite rafts In plagioclase-bearing cumula.tes, however, acted as collectors for the PGM. It is uncertain compaction of the crystal mush is considered where the PGE in the hortonolite dunite pipes to be one of tlte most important processeswhere- were der,ived from. Their concentration in the by large quantities of intercumulus liquid were central portion of the pipes as a result of the pressedout of the pore spacesinto the overlying reaction replacement of the surrounding pyro- magma. This is evident from, several textural xenites in which they are present on a ppb level features displayed by the plagioclase crystals (Gijbels et al. 1974) seemsa distinct possibility. (von Gruenewaldt 1971), such as interpene- The enrichment of the chromitite rafts in the tration, myrmekite textures, bent crystals and Onverwacht pipe suggeststhat a related process reverse zoning (Figs. 5b, c). An especially involving late liquids could have resulted in the important mechanism whereby adcumulus enrichment of the iron-rich upper chromitite growth can take place seems to be the process layers, including the thin chromitite layers in of re-solution, causing interpenetration and the the Merensky Reef. redeposition of this material in the interstices. The ascending intercumulus liquids could have become concentrate{ in places to ,form trans- Coxcr,usloNs gressive pipe-like or sill-like mafic pegmatites. It is envisaged that where their movement to (1) The broad loci of emplacement of the higher levels was ha'mpered by layers where Bushveld Complex is in the Transvaal basin at adcu,mulus enlargement of crystals effectively the intersection of several prominent structural sealed the intercumulus spaces, these liquids directions on the Kaapvaal craton. spread out laterally, reconstituting the cumulates (2) The present-day configuration of the com- to form the irregularly shaped pegmatites com- plex is controlled by a series of dome- and monly displaying gradational as well as sh,artrr basin-like features, the presence of which has contacts with the country rocks, as described characterized &e Kaapvaal craton since earliest by Cameron & Desborough (1964). Such layers f:mes; these developedin.response to interfering in which substantial secondary enlargement of north-northwest - east-northeast oriented anti- crystals took place at an early stage could have clinal and synclinal warps. formed at the top of the mush during periods (3) Emplacement of an early, magnesian mag- of intense convection near the mush-liquid ma was into low-angle conical shear fracture.s interface. that developed in the Transvaal sequence in Conflicting views. prevail concerning the as- response to depression of the basin below the sociation of FGE with chromitite. Cousins & chord position after subaerial extrusion of the Vermaak Qnq have expressed doubt whether Rooi:berg felsile and emplacement of numerous any of the PGE can be accornmodated in spinels diabase sills. Subsequent replenishments of mag- at high temperatures as has been suggested ma resulted in the lateral extension and inter- amongst others by Naldrett & Cabri (1976). connection of the initial 'rnagma chambers, to Evidence presented by Cousins & Verrnaak form the large sheeted nature of the upper two SULFIDES IN THE BUSFIVELD COMPLEX 253

gravity Geol. Soc. S. Atr' zones of the complex. Postulated centres of revealed by a snrvey. Trans.8L, (in press). intrusions are aligned along deep seated north- northwest and north-northeast tension fractures. Brrsnntrwr-, K. (1970): An interpretation of a Pilanes- (4) Petrographic and field evidence suggests gravimetric survey in the area west of the Western Transvaal. fz Symposium the existence of an early, high-mngnresium mag- berg in the on the and other followed by emplacement of normal tho- ma, Layered Intrusions (D.J.L. Visser & G. von leiite. eds.), Geol. Joc' S. Alr. Spec' of Gruenewaldt, (5) Concentrations of sulfides near the floor Publ. 1,266-282. the intrusion have 'been in'terpreted by various BoNNrcHsEN,B. (1972): Sulfide minerals in the result of degassing of the authors as being the Duluth conplex. In Tbe Geology of Minnesota floor rocks during mefamorphic reaction and (P.K. Sims & C.B. Morey, eds.), Geol. Surv, the addition of S, COg and IIgO to the crystal- Minn, Cent. Vol., 388-393, lizing rnagrna. Bnoc& B.B. & PREroRrus,D,A. (1964): Rand (6) Several cyclic units characterize the upper basin sedimentationand tectonics. /n The Geol- part of the critical zone. The Merensky unit ogy of some Ore Deposits in Southern Africa I ls the most complete 'and laterally tlte most (SJI. ffaugnton, ed.), Geol, Soc. S. Afr., Johan' constant; it contains the largest sulfide con- nesburg. 'features centrations. The of the reef suggest BRYNARD,H.J., or Vn-r'rnns, J.P.R' & VuoeN, a hiatus in the crystallization im,rnediately prior E.A. (1976): A mineralogical investigation.of to its formation, followed by crystallization from tho Merensky Reef at the Western Platinum m'ine, an isolated batch of magrna near the floor of near Marikana, South Africa. Econ. Geol, 71, the chamber and enrichment of sulfur prior 1299-1307. or during crystallization. BucnANAN,D.L. (1972): The Petography ot (7) Concentrations of FGE in the iron-rich Rocksol the Bushveld Complex Type Intersected chromitite layers in the upper part of the by Boreholes in the Bethal lrea. M.Sc. thesis' critical zone and in the Onverwacht pipe seem Unin Pretoria" South Africa. to be related to late solutions that percolated (1975): The petrographyof the Bushveld upward through the pile of semiconsolidated Complex intersected by boreholes in the Bethal cu,rnulates. area.Geol. ^9oc.J. Afr. Trans. 78' 335-348. (8) Concentrations of sul'fides in the upper (L979): A combined transmissionelectron zone are ascribed .firstly, to sulfur enrichment microscope and electron microprobe study of in the magma as a'result of normal fractional Bushveld pyroxenes from tle Bethal area. /. crystal,lization, and secondly, to the crystalliza- Pettolog! 20, (in Press). tion of magnetite and ferromagnesian silicates. Brrrrory A. (1973): The depositionalhistory of (9) Mobiliza.tion of intercumulus liquids by the Wolkberg proto-basin, Transvaal' Geol. Soc. com$action and their ascent to hrgher levels is S, Afr. Trans. 76, 75'26. considered i,mportant in the formation of the (1975): A palaeocurrent study of the mafic pegmatites in the comPlex. Dwaal Heuvel Formation, Transvaal Supergroup. Geol. Soc. S, Afr. Trans.78, t73-t84. in ACTNOWT-NNGEMENTS CeurnoN, E.N. (1969): Postcumuluschanges the eastern Bushveld Conrplex. Amer. Mineral. 54, 754-779. The author thanks Dn. T.N. Irvine, N.J (1970): of certain coexisting Page and M,R. Sharpe for their constructive Composition phases in the eastern part of the Bushveld criticism of the manuscript. bomplex. /z Symposiumon the Bushveld Ipeous Complex and other Layered Intrusions (DJ.L. Vissei & G. von Gruenewaldt, eds,), GeoI. Soc. RSTBRENcES S. Atr. Spec.Publ. l' 46-58. (L975): and subsolidusequili' Arnenussm, C.R. (1973):The evolution of the Postcumulus and coexisting silicates in early Precambrian crust of Southern Africa. bration of chromite Complex. Geochfun,Cosmo- Roy. .Soc.Lond. Phil. Trans. M73,359-388. the easternBushveld chim, Acta 39' 1021-1034. ArnNg F.B. (1969): Pyroxenesof the Bushveld the eastern Atrica. I. Petrology t0' 222' (1978): The lower r-ote of intrusion, South trough. 249. Bushveid Complex in the Olifants River J. Petrology 19, 437462. BenRett, D.M., hcoosnr.l, f.B.E., McCsrrc, T.S. & Crwrrrow, R.C. (1978): The structure of & Dssronoucu, G.A. (1954): Origin of the Bushveld Complex south of Potgietersrus,as certain magnetite'bearingpesatit€s in the east- 254 THE CANADIAN MINERALOCIST

ern part of the Bushveld Complex, South Africa. FLlr.[N, R.T., Ur,rvrnn,G.C. & Surrmn, C.F. (1923): Econ. Geol. 59, 197-225, Petrogenesis of the eastern Bushveld Complex: CousrNs, CA. (1959): The structure of the maJic crystallization of tle middle critical zone. l. portion of the Bushveld igneous Complex. Trans. Petrology tg,. l3GlS2. Geol. Soc. S. Afr.62, 179-ZOl, GunEls, R.H., Mnteno, H.T., Dnsnonoucn, G.A. & ------(1969): The Merensky Reef of tle Bush- BARTEL,AJ. (1974): Osmium, ruthenium, iridiu- v.eld igneous Complex. In Magmatic Ore &posits and uranium in silicates and chromite from the {II,.D.B. Wilson, ed.), Econ. Geol. Mon. 4,23g_ eastern Bushveld Complex, South Africa. Geo 251. chim. Cosmochim. Acta 98, 319-337. (1932): & Frnnvca, G. (1964): The cbromitite de- Htl,- A.I-. The Bushveld igneous Complex posits of the western belt of the Bushveld Com_ of the Oentral Transvaal. Geol. Saiv. S. Afr. Mem. plex. In The Geology of some Ore Deposits 28. iu Southern Africa tr (SJI. Haughton, ed.), HAam,ToN, J. (1977): Sr isotope and trace elemenr Geol. Soc. S, Afr., lohannesburg. studies of the Great Dyke and Bushveld mafic phase and proterozoic & VSRMAAK, C.F. (1926): The contribu- their relation to early magma genesis petrology tion of Southern dfrican ore deposits to the in Southern Africa. l. geochemistry of the metals. Ecoz. t8, u-52. Geol. 7t, 287-305. HaI{MERBEc& E.C.L (1920): On the genesis of y. lead-zioc and fluorspar deposits in the south_ Qgcrnr,- J.H., TEnur,a, & Ganru, I. (1976): vi/esterD Marico District, Transrraal. Ann. refltivg importance of sulfides, spinels, and Geol. ln9. ^lary. ,S. Alr. 8, 103-110. platinoid minerals as carrien of pi, iO, lr and {.u in the Merensky Reef at Western itatinu- HATnNcu, P.J. (1977): Die struktuur van die Bos- Limited, near Marikana, South Afica. Econ, veldkompleks in die Grobendal-Lydenburgi rn Geol. 7I, 1308-1323. Belfastgebiede soos afgelei unit regionale eravi- tasieopnama Inst. Geol. Res. Bushveld. Coiplex, DALv, R.A. (1928):,Bushveld igneous Complex of Univ. ,Pretoria, Re.r. Rep. 8. tho I ransvaal. Geol. Soc. Arner. Ball, gg,703-76g. HAwLEI l.E. (1962): The Sudbury ores: their DB BRUrrN, H. & Rnoors, R.C. (1975): A new mineralogy and origin. Can. Mineral. T, l-207. variety of Bushveld granite in the Dennilton Hrcnooor, R.O. (1959): geology area, Transvaal. Geol. ^goc. J. Afr, Trans. 7g, The around the 89-95. {*it" prp. on Driekop (Eastern Transvaal). Geol, Soc. S. Alr. Traw, 62, 59-7t. DE VILLTERgJ.S. (1970): The structure and petrol- Hnss, G.B. (1972): Heat and mass ogy of the mafic rocks of the Bushveld Cimplex transport during crystallization of the Stillwatel igneons put-h of Potgietersrus. .lz Symposium on the Complex. /z Studies in Earth and Space (Shagam Bushveld Igneous Complex and- other Layerett Sciences et al, eds.), Geol. Soc. j03- Inlrusions (D.J.L. Visser & G. von Amer. Mem. 1SZ, Gruenewaldt, 522. eds.), Geol. .9oc. J. Afr, Spec. publ t, 23-35, Hrss, H.H. (1960): Stillwater igneous DE VnLrERs, S.B. (1967): Aanvoerrigtings Complex, van A quantitative mineralogical sedimente van die Sisteme Loskop {oqtaqa. study. en-Wa:terberg Geol. Soc, Amer. Mern, 80. in Noord-Transvaal soos weerspieel deur kruis- gelaagdheid. Ann. Geol. Surv. S. Afr. 6, 63-67. HuNrER, D.R. (1974a): Crustal development in tle Kaapvaal precambrian DE_W_AAL, (l9jA): craton. I. Tbe Archaean. .S.A. Interference_folding of Res, t, 259-294. Bushveld-i€neous-complex age in the ii*ivu"r system north of Marble Hall, Transvaal. fz (1974b): Crustal development in tle Kaap- precambrian Symposium on the Bushveld Ieneous compiei and vaal craton II. The Proterozoic. Res. other Layered Intrusions (D.J-.L. Visser & C. v"n L, 295-326. pn11::::::::::::::::1ey-{dt, eds.), Geol, Soc. S. Afr, Spec. pubt. -_ (1975): The Regional t,283-298. Geological Setting of the Bushveld Complex (An Adjunct to the (1975): The mineralogy, chemistry and Provisional Tectonic Map of the Bushveld Com- plex). certain aslrects of reactivity of chromitite- from Econ. Geol. Res. Unit, Univ. Witwaters- thc.Bushveld igneous Complex. Nat. Irut. U"iall. rand, Johannesburg, South Africa. \Johannesburg) Rep. 1709. (1976): Some enigmas of the Bushveld (1977): Carbon dioxide and water from Complex. Econ. Geol. 7L, 229-248. metamorphic reactions as agents for sulohide & HaunroN, PJ. (1978): The Bushveld s$nel precipitation 1nd in mafic magmas. baol. Complex. /z Evolution of the Earth,s Crust (D.H. Joa J. Afr. Trans.80, 193-t97. Tarling, ed.), Academic Press, Iondon. FERcusoN, J. (1973): pilanesberg The alkaline Irralw, T.N. (1970a): Cqrstallizarion sequencesiD province, southern Africa Geol. soc. s, 76, Zfr, fians. the Muskox intrugion and other layered' se- ug-nu quenc€6. I. Olivine-pyroxene-plagioclase relations. ST'LFIDES IN THE BUSTTVELD COMPLEK 255

In Symposiumon the Bushveld Igneous Complex and Aberdeenshireolivine Complexes.J. and'other Layered Intnrsions (DJ.L. Visser & Petrolog! 20, in press. G. von Gruenewaldt,eds.), Geol. Soc. S. Atr. Mcl-ennx, C.H. (1928): A minsnlsglsal investiga- Spec.Publ. t,441476, tion of the platinum-group metals in the UG2 (1970b): Heat transfer during solidification layer of the Bushveld Complex with special re' from of tayireO iot*.io*. I. Sheetsand sills. Can, J. ference to tle recovely 9f the ^ mine?ls narti Sa. ?, 1031-1061. the ores.Metals Minerals ProcessingL(4), t9'27. JecrsoN,E.D. (1961): Primarytextures and mineral MoLyNEUx,T.G., voN GRUENWALDT,G. & Kwow' associationsin tle Ultramafic' Zono of the Still- LEs,RA. (1976)t A bibliographyof the geology water Complex. Montana. US. Geol. Sum. ProI, of ihe Bushveld Complex. Inst. Geol. Res.Bush- Pap.358. veld. Complex, ()niv, Pretoria, Res. Rept. 2. I0-or, A.A.C. (1978); The MetamorphosedSedi' & KLnKERT, P.S. (1978): A structural ments of the Pretoria Group and the Associnted interpretation of part of the eastern mafic lobe Rocks Northwest of Zeerust, Western Transvaal, of tfe BushveldComplex and its surrounds.Geol' M.Sc. thesis, Univ. Pretoria, South Africa. Soc.S. Afr. Trans.81, (in Press). KNowr-ns, R.A. (1978): Supplementto researcb NALDRETT,A.J. & CeaRr,LJ. (1976): Ultramafic report 2: a bibliography of the geology of the and related mafic rocks: their classification and Bushveld Complex. Inst. GeoI. Res. Bushveld. genesiswitl specialreference to the concentration Complex, Univ, Prebrta, Res. ReP. L2. of nickel sulfides and platinum-group elements' LluoBn, W.R. (1970): Oridn of the MerenskyReef. Econ,Geol. 71, 1131'1158. Nature 227, 365-366, NrwrvreN, S.C. (1973): Platinum. Inst. Mining Lrs, C.A. & SmnrE, M.R. (1978): The formation Metall, Trans. 82, 452-468. spherical aggregat€sin the Bushveld of mafic DA. (1965): The evolution of Pre' to potential silicate liquid PREtoRrus, Complex in response basins in South Africa. Econ' Geol' immisqililify. Inst. Geol. Res.Bwhveld Complex, cambrian (Jnit. Univ. Wtwatersrand,7th Ann' Rep' Univ. Pretoria, Res.ReP. t4. Res. 15-17. (L964): Die Rand van die Boneld' LrrnnxsERc,L. intrusion of magmainto suid van Burgersfort, Iwlui' RoreRrs,J.L. (1970): The stollingskompleks, of Igneous Intrusions Sekere Peridotiet-liggame en die Gemeta' brittle rocks. In Mechanism tende Rast, eds.), Liverpool Gallery morfoserde Serie Pritoria. M.Sc' thesis, Utriv. (G. Newall & R. Spec,Issue 2,287'338' Pretoria, South Africa. Press,Geol. I. (1968): The Saltides in the I'ayered' Se' SAMPsoN,B, (t932): Magmatic chromite delnsits quenceof the Bushveld lgneous Complex. Ph.D. in soutlern Africa Econ, Geol. 27, ll3't44' thesis, Univ. Pretoria, South Afriq- ScrniBmsRHdHN,H. (1929): The mineragraphy, (1970): The sulfides in the layered se- spectrography and genesis of platinum-bearing quence of tle Bushveld Igneous Complex. /z o'i"t"t-py.ittdtit" ores of the Bushveld Complex' Symposium on ttre Bushveld Igneous Complex irl fne'ptatinum Mines and Deposits of South and otler Layered Intrusions (D.I.L. Visser & Africa (P.A. Wagner, ed.), Oliver and Boyd, G. von Gruenewaldt,eds.), Geol. Soc. S. Afr. Edinbureb. Spec. Publ. t, L08-2n7. Scnwrlr,Nus, J.S.I., Eucrr-oREclrr,L.N'J', CoERlzE, Maurs, CL. (1977): An Investigationof the Pyro' F.J., RussEL,H.D.' MAI-HERBE,SJ', vANRooJEN, tenite Marker and the Associated Rocks in the b_'f. t Coorr, R. (1962): The geologyof the Main Zone of the Eastem Bushveld Cornplex. Olifants River area" Transvaal' Geol' Surv' S' MSc. thesis, Univ. Pretoria, South Africa- Alr. Expt, Sheets 24298 (Chuniespoort) and Mmrow, A.G. (1976); The Geologyof the Bush' 24304 (Wolkberg). veld Cornplexon the SekhukhunePlateau, Eastern -,IIreMsrRA, S.A. & GesBennnn'E. (1976): Trunsvaal, M.Sc. thesis, Univ. Pretorira, South The Merensky Reef at the Atok Platinum mine Africa. and its environs. Econ. GeoI. 7t, 249-260' prn MEnwE,M.l. (1977): The gsol- & vaN M.R. & SttlMAN, J.A. (197S): A model potential economic significance of SEAnre, ogy and the the emplacement of the eastern compart- north-eastern Bushveld Com' for the Malope area, of the- Bushveld Complex' Inst' GeoI' Res' pbx. Geol.,Soc.J. Atr. Trans. 8A' lt7'L24. ment ti^i"ita Complex, univ. Plaoria, Res' Rep' ll'' (1976): The fluorite depositsin Mmru.n, J-EJ. & Gouan, D'I' (1962): Series of the Marico District SMrrs,- P.J., HALrs, A.L. the Dolomite of the Re4rublic of South Africa. Econ. Geol. 7l' 625' fne gaiity survey Transvaal, South S, Afr. Hb' 3' 635. Africa- Geol. Sum. platinun lVtcCelrnx, T.S. & ClwuroRN, R.G. (1979): Frac- Srtncrr-,- E.F. (1974): The genesis of Eng' s' tional crystallization effects on initial Sf?/S?0 deposits: furthir tloughts' Minerals sci' ratios - theory and application to the Bushveld L20-14L. 256 TIIE CANAI'IAN MINEMLOGIST

& TnnseN, M. (1976): Platinum genesis: Vtssrn, J.NJ. (1969) z tn SedimenologieseStadle now minenlogical evidence, Econ. Geol. 7t, van die Sede Prebrta in Transvaal,ph. D. thesis, r4st-1460. Univ. Orange Free State, Blemfontein. TannnN, M. & S'ruMpFL, E.F. (L975): platinum voN GRUENEWALDT,G. (1971): A petrographical minsplo'gy of tle Driekop mine, South Africa. and Mineralogical Investigation of the Rocks Mineral. Deposita 10, 7l-85. of the Bushveld Igneous Complet, in the Towes- frylt C.E., Yotrn, H.S., Jn & Seanrn, J-F. hoogte-RoossenekalArec of the Eastern Trans- (1968): Melting relationsof igneousrock series. vaal. DSc. tlesis, Univ. Pretoria, South Africa. yearbook Carnegie Inst. Wash, 6$, 450457. (L972): The origin of the roof-rocks of the TunrruN, K.K. & Ivrogonr,, KJ{. (1961): Dis- Bushveld Complex between Tauteshoogte and tribution of the elements in some maior units Paardekop in the Eastern Transvaal. Geol, Soc. of tle Earth's crust Geol. Soc.Amer. Buil. 72, S. Afr, Trans,76, t27-134 (see also ?6, g3-94). t75-t92. (1973); The main and upper zonesof the vAN DERMERwn, M.J. (1976): The layered se- Bushveld Complex in tle Roossen€kalarea, East- guenceof the Potgietersruslimb of the Bushveld ern Transvaal. Geol. Joc. ,S. Afr. Trans. 76, Complex. Econ. Geol. 7L, 1337-l3SL 207-227. (t976): Sulfrdes vAN- \ENsluRg, W.C.J. (1965): The mineralogy in the Upper Zone of the of the titaniferous magnetite and associated$I_ easternBushveld Complex. Econ. Geol.7t, t3Z4- fides. on_Kennedy's Vale 36lKT, Lydenburg 1336. District, Transvaal.Ann. Geol. Jzrv. J. Afr. 4, (1977): The mineral resourcesof tho rt3-r27. Bushveld Complex. Minerals Sci. Eng. g, g3-95. -l- & Pnerouus, D,A. (1977): Souih Alricds Wacrn, L.R. & BRowN, G.M. (1957): Funnel- Strategic Minerals - pieces on a Continental shapedlayered intrusions. Geol. Soc. Amer. Bull. Chessboard. Valiant publishers, Johannesburg. 68, l07t-1074. vAI ZyL, J.P. (1970): The petrologyof the Meren_ & - (1968): Layeredlgneous Rocks. sky Reef and associatedrocks on Swartklip 9gg, Oliver and Boyd, Edinburgh. Rustenburg. Iz Symposium on th€ nusned WacNnn, P.A. (L929): The platinum Deposits and Iq:":" Complex and other Layered Intrusions (D.J.L. Mincs of South Africa. Oliver and Boyd, Edin- Visser& G. von Gruenewildt,eds.), Geol. burgh. Joc. J. Afr. Spec.publ. l, g0-107. I[rrrarvrsr, J. (1959): The'floof' of the Bushveld VnnMeeK,C.F. (1970): The geologyof the lower portion igneous,Complex and its relationshilx, with spe- of the Bushveld Complex and its rela- cial reference tionship to the Eastern Transvaal. G€ol. to the floor rocks in tG area west of the Joc. ,S.Afr. Proc. 62,2l-90. P-ilanesbergWestern Tra^nsvaal. /z Symposiumon (1964): thg Byhvel{ IgneousComplex anO ritnJr A brief outline of the geology of Layered rhe igneous Intrusions (D.J.L. Visser & G. von Bushveld Complex. In The Ceotogy Gruenewaldg of some eds.), Geol. Joc. ,S.Afr. Spec,publ. 1,242-265. Ore Deposits in Soutlern Africa ii (S.H. Hauehton, ed.) GeoI. Soc. S. Afr,, Johan- (1976a): The nickel pipes of ylaKontein nesburg. and vicinity, Western Transvaal Econ. Geol, ------7t, 26t-286. (1969): The geologyof the BushveldCom- p1ex, the largest repository (L976b): of magmatic ore The Merensky Reef _ thoughrs. depositsin tle world" In on its Magmatic Ore Deposi8 environnent and genesis.Econ. Geol.-7!, (H.D.B. Wilson, eA.),Econ. 1270-1298. GeoI. Mon.4,,-l-22. Wu,mN, H.D.B. (1956): Structure of lopoliths. & HEronxs, L.p. (t976): A review of the Geol. Soc.Amer. Bull. 67,289-300. mineralogy of the Merensky Reef, with special referenoe to new data on-the piecious hetal Received.September L976, mineralogy. revised manascript ac- Econ. Geol. 71, 124+1269. cepted lanaoy 1979.