Geochemical and geochronological constraints on the geologic evolution of the western Sonobari Complex, northwestern Mexico

Geochemical and Geochronological Constraintson the Nature of the Immediate Basement nextto the Mesoarchaean AuriferousWitwatersrandBasin, South Africa

H. E. FRIMMEL1,2*, A. ZEH1, B. LEHRMANN1, D. HALLBAUER3 ANDW. FRANK4

1GEODYNAMICS & GEOMATERIALS RESEARCH DIVISION, INSTITUTE OF GEOGRAPHY, UNIVERSITY OF WU« RZBURG,

AM HUBLAND, D-97074 WU« RZBURG, GERMANY2DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF CAPE TOWN, RONDEBOSCH 7701, SOUTH AFRICA3ORTSSTRASSE 36, D-07429 DO« SCHNITZ, GERMANY4CEAL LABORATORY, SLOVAK ACADEMY OF SCIENCES, DUBRAVSKA CESTA 9, 84104 BRATISLAVA 4, SLOVAKIA

RECEIVED FEBRUARY 17, 2009; ACCEPTED OCTOBER 6, 2009ADVANCE ACCESS PUBLICATION NOVEMBER 4, 2009

A combined petrological, geochemical, and geochronological (Rb^Sr

and Sm^Nd whole-rock, U^Pb and Lu^Hf zircon, and Ar^Ar

hornblende) study on a section of pre-Witwatersrand basement

drilled at the northwestern margin of theWitwatersrand Basin has

revealed new insights into the nature and tectonic setting of the

likely source area for some of the Mesoarchaean auriferous

Witwatersrand sediments.The protoliths of intersected altered gran-

ite and hornblende metagabbro are of indistinguishable age

(3062� 5 Ma) and have very similar geochemical signatures.

Trace element characteristics typical of calc-alkaline magmatism

and evidence of variable contamination with older crust (subchon-

dritic eNd and eHf in zircon) point to an active continental margin

setting.The Ar^Ar hornblende ages are within error of the magmatic

crystallization age or slightly older. Alteration of presumably primary

magmatic hornblende to magnesio-hornblende immediately after

gabbro emplacement during late magmatic autometasomatism is

suggested.The presence of hydrous melts (44 wt % H2O), compara-

ble with fertile Au-bearing magmatic^hydrothermal mineralizing

systems in Phanerozoic volcanic arcs, is inferred.Thus, a kind of hin-

terland is proposed for theWitwatersrand that compares favourably

with the tectonic domains that are known to host the majority

of post-Archaean gold deposits. Later retrograde hydrothermal

alteration at c. 2720 and 2630 Ma led to variable Pb loss in zircon

and the resetting of the whole-rock Rb^Sr isotope system whereas

the Ar^Ar and Lu^Hf isotope systems in the hornblende and zircon

grains, respectively, were not significantly affected. Comparison

with published data suggests that these alteration events are the

same as those that affected theWitwatersrand Basin fill associated

with major early Ventersdorp flood basalt volcanism and possibly a

pre-Transvaal thrusting event in response to contractional deforma-

tion in the Limpopo Belt.

KEY WORDS: active continental margin; Archaean crust; U^Pb;

Lu^Hf;Witwatersrand hinterland; gold genesis

I NTRODUCTIONThe Mesoarchaean Witwatersrand Basin in South Africaaccounts for about 40% of all known gold (Frimmel,2008) and thus represents by far the single most importantgold depository known in the Earth’s crust. The genesis ofthe ore bodies in theWitwatersrand goldfields has been amatter of intense debate (for a relatively recent summary

*Corresponding author.Telephone:þ49-931-3185420.Fax:þ49-931-3184620. E-mail: [email protected]

� The Author 2009. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 50 NUMBER12 PAGES 2187^2220 2009 doi:10.1093/petrology/egp073

of the arguments see Muntean et al., 2005); however, con-sensus is emerging that the gold entered the conglomeratichost-rocks essentially in the form of detrital particles thatwere subsequently mobilized on a local scale by variouspost-depositional fluids (for a review and summary ofarguments see Frimmel et al., 2005). A palaeoplacer modelinvariably raises the question of the source of all that goldand thus of the nature of the hinterland at the time of det-rital gold transport and sedimentation.The term ‘Witwatersrand Basin’, although widely

accepted in the literature, is misleading because it refersto tectonically different basins whose fills were stacked ontop of each other. Most of the Witwatersrand gold(480%) is concentrated in quartz-pebble conglomerates(reefs) of the52·90 to42·78 Ga Central Rand Group. Incontrast to the underlying 52·98 to 2·91 Ga West RandGroup (contributing55% of Witwatersrand gold produc-tion) for which a passive margin setting is indicated, thepredominantly fluvial sediments of the Central RandGroup were deposited in a foreland basin whose shapewas similar to the current subcrop outline, except for thelater up-doming of the central part as a result of the 2023Ma Vredefort impact event (Fig. 1; Coward et al., 1995;Kositcin & Krapez, 2004; Frimmel et al., 2005).Information on the nature of the source rocks may be

obtained from the study of the clastic sediments.Derivation of the various allogenic components of theWitwatersrand sediments, including gold, from oldergranite^greenstone terranes, such as the BarbertonGreenstone Belt, has been suggested repeatedly by severalworkers (e.g. Viljoen et al., 1970; Robb & Meyer, 1990). Anumber of undoubtedly allogenic minerals indicative ofboth felsic and mafic sources, most importantly zirconand chromite, respectively, are present in the siliciclasticWitwatersrand sediments (for a detailed list of mineralssee Feather & Koen, 1975). From the distribution of rela-tively immobile elements in Witwatersrand shales,Wronkiewicz & Condie (1987) concluded that the propor-tion of granite, basalt and komatiite in the source areasincreased with time at the expense of tonalite.Vennemannet al. (1992, 1995) showed that the oxygen isotopic composi-tion of theWitwatersrand quartz pebbles conforms to thatof mesothermal vein quartz as found in greenstone-hostedorogenic gold deposits. Detrital zircon age spectra fromthe Witwatersrand (Kositcin & Krapez, 2004) agree wellwith available ages from the various granitoid^greenstonebelts to the east, north and west (Poujol et al., 2003; for acomparison see Frimmel et al., 2005).The basal contacts of the numerous auriferous and ura-

niferous conglomerate beds typically represent an ero-sional surface above which older intrabasinal sedimentaryunits have been partially reworked. The sedimentaryunits higher in the succession, for example the 2·71 GaVentersdorp Contact Reef (contributing c. 15% of the

total Witwatersrand gold production) above theWitwatersrand Supergroup, reflect considerablereworking of the underlying sediments. This reworkingof intrabasinal sediments forms a major obstacle in thereconstruction of the primary source rocks and our atten-tion is, therefore, directed towards the potential sourcerocks directly.Interestingly, most zircon grains in the arenitic units

below and above single reefs of the Central Rand Grouphave ages between 2·96 and 2·83 Ga, whereas those fromthe auriferous conglomeratic reefs indicate a dominanceof 3·05^3·09 Ga sources (Ruiz et al., 2006; N. Koglin,unpublished data, 2009). An imprecise Re^Os age of3016�110 Ma obtained by Kirk et al. (2002) for the golditself (3033�21 Ma obtained for rounded pyrite and goldcombined) is clearly older than the age of sediment deposi-tion and thus provides strong support for the palaeoplacermodel. More importantly, these ages conform to the detri-tal zircon age peaks mentioned above for the reef units.These data indicate that 3·05^3·09 Ga pre-Witwatersrandunits are by far more important as a potential source ofthe gold than the various younger, syn-depositional grani-toids in the hinterland and thus they will be at the centreof this study. Numerous studies have been conducted onthe main granite^greenstone terranes of the KaapvaalCraton, especially on the Barberton Belt (for recentreviews and further references see Brandl et al., 2006;Robb et al., 2006; Zeh et al., 2009). Today’s exposed gran-ite^greenstone belts on the craton (Fig. 1) cannot, however,be the source regions of theWitwatersrand sediments andthe gold therein. They are the deeply eroded remnants ofwhat might have been a potential 3·5^2·8 Ga hinterland.Although the finer grained metasedimentary rocks of theWitwatersrand Supergroup and its stratigraphic equiva-lent, the Pongola Supergroup (Fig. 1), may have beensourced in distal hinterlands, the allogenic components ofthe coarse-grained, auriferous and uraniniferous conglom-erate beds are likely to be derived from proximal sources.Within a given reef, both Au and U concentrationsdecrease from the basin margin towards its centre, with asystematic increase in the U/Au ratio down the palaeo-slope. This observation has been explained by hydraulicsorting of allogenic gold and uraninite particles (Minteret al., 1986; Frimmel et al., 2005). Thus the source rocks forthe high-energy detritus, including the gold, have to besought in proximity to the former basin margin.Unfortunately, data on such pre-Witwatersrand units in

the immediate vicinity of the Witwatersrand Basin aresparse because of a lack of outcrop. So far, two types ofbasement have been recognized in the vicinity of theWitwatersrand Basin: bimodal volcanic rocks with minorsiliciclastic sedimentary material of the Dominion Groupas well as various granitoid domes (with minor green-stones). From the spatial distribution of both basement

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2188

types (Fig. 2) it is apparent that both could have suppliedmaterial, particularly into the Central Rand Basin.The precisely dated 3074� 6 Ma (Armstrong et al., 1991)

Dominion Group represents the first supracrustalvolcano-sedimentary succession above the Palaeo- toMesoarchaean granitoid^greenstone basement. Its tectonicsetting has been a matter of controversy. Whereas someworkers have argued for a continental margin settingbecause of the calc-alkaline andesitic character of themafic rocks and depletion in Nb, Ti and Zr (Burke et al.,1986; Crow & Condie, 1987), others have considered thisgeochemical signature not conclusive, inherited from the

mantle source rocks, and proposed deposition in a conti-nental rift setting because of the bimodal nature of the vol-canism and the tholeiitic affinity of the mafic rocks(Bowen et al., 1986; Marsh et al., 1989; Jackson, 1992).Fragments of greenstone belts in the vicinity of the

Witwatersrand Basin have been known so far only fromthe Johannesburg Dome to the north (Anhaeusser, 1973)and from the eastern margin of the Vredefort Dome(Minnitt et al., 1994; Fig. 2). Ultramafic^mafic complexesdominate in the former area. No precise age data exist butthese complexes, which have been interpreted as reflectinga former suture zone (Anhaeusser, 2006), are considered

Fig. 1. Main Archaean stratigraphic units of the Kaapvaal Craton. TheWest Rand and Central Rand Groups constitute theWitwatersrandSupergroup. Bold dashed line outlines the boundary of the Kaapvaal Craton as inferred from aeromagnetic data. The crustal blocks (amalga-mated by 2·8 Ga) are separated by major lineaments; modified from Eglington & Armstrong (2004) and Frimmel et al. (2005).

FRIMMEL et al. WITWATERSRAND HINTERLAND

2189

to be older than the 3200^3340 Ma intrusive tonalitic totrondhjemitic gneisses in the Johannesburg Dome (Poujol& Anhaeusser, 2001). Ultramafic^mafic intrusions in thecore of the Vredefort Dome are metamorphosed at highgrade and a xenocryst age of 3425 Ma has been reportedfrom there (Hart et al., 1990).Some petrographic, geochronological and geochemical

information is available from a variety of granitoid bodiesthat occur in several basement domes to the north andNWof theWitwatersrand Basin as well as in theVredefortDome (Anhaeusser, 1973, 1999; Klemd & Hallbauer, 1987;Robb & Meyer, 1987; Barton et al., 1999; Poujol &Anhaeusser, 2001; Armstrong et al., 2006; for a summaryand further references see Robb et al., 2006). Their agesrange from 3340 to 2777 Ma. Most of them, notably the

majority of the granitoid bodies in the Johannesburg,Vredefort andWesterdam^Coligny Domes, have ages thatcluster around 3·1Ga, whereas the youngest of these intru-sions post-dates the onset of sediment deposition in theWitwatersrand Basin.Whereas the focus of the previous studies has been on

the granitoids along the perimeter of the WitwatersrandBasin, mafic rocks are at the centre of this study. Here wereport the first geochemical and isotope data on maficand associated felsic intrusive rocks that were intersectedin exploration drill holes through a basement horst at thenorthwestern margin of theWitwatersrand Basin. The sig-nificance of this site lies not only in the abundance ofmafic rocks but also in the fact that the pre-Witwatersrand basement is not covered byWitwatersrand

Fig. 2. Surface and subsurface distribution of theWitwatersrand Supergroup, Dominion Group and Mesoarchaean granite^greenstone base-ment domes. Also shown are the palaeocurrent directions during Central Rand Group times (small arrows) as well as the position of the produ-cing goldfields (from Frimmel et al., 2005). Larger open arrow indicates approximate position of the studied drill site (see Fig. 3).

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2190

sediments but is directly overlain by rocks of the lowermostTransvaal Supergroup, as well as in its proximity to themargin of the Central Rand Basin. Consequently, thedrill core can provide insights into the nature and make-up of at least a section of what has been a potential sourceof proximal Central Rand Group sediments. Specifically,we provide petrographic, mineral-chemical and whole-rock geochemical data, U^Pb and Lu^Hf isotope data onzircon grains, Rb^Sr and Sm^Nd whole-rock isotopedata, as well as Ar^Ar hornblende data, all of which areused to constrain the age and to formulate a petrogeneticmodel for this section of proximal source rocks.

DRILL CORE SETT ING ANDPETROGRAPHYThe investigated core comes from a borehole (BH1) thatwas drilled by the Chamber of Mines ResearchOrganization (supervision: D. K. Hallbauer) on the farmRooidraai 85 IQ at 2682104700S, 2780605400E, 30 km west ofCarletonville (Fig. 2). The local geology of the site is domi-nated by a ridge that consists mainly of strongly weatheredgranodiorite with minor pegmatite and quartz veins, over-lain by a 2^5m thick cover of Black Reef Formation quart-zite and conglomerate (the basal lithostratigraphic unit ofthe Transvaal Supergroup), and soil (Fig. 3). The ridge ispart of an approximately north^south-trending horst struc-ture within the northwestern Witwatersrand Basin. Thestructure was shaped during syn-Ventersdorp (2·71 Ga)extension. Block faulting at that time made it possible thatlocally the pre-Witwatersrand basement became elevatedto shallow levels and was eventually exposed at or nearthe surface. The block-bounding faults have, however, aprotracted history of repeated reactivation from syn-Central Rand Group compression to modification by the2024 Ma Vredefort impact (Brink et al., 2000), leavingbehind cataclasites and locally pseudotachylyte.The borehole intersected 35m of pre-Witwatersrand

basement rocks beneath 4m of quartzite of the Black ReefFormation and overlying saprolite and lateritic soil. Thetop 10m of the basement consists of coarse-grained alteredgranite underlain by massive, coarse-grained hornblendemetagabbro (or metadiorite). The contact between thetwo rock types in the core is characterized by a transitionzone with pink alkali feldspar disseminated in the other-wise mafic to intermediate rocks, and by a strong enrich-ment in largely chloritized biotite over a 1m distancefrom the granite into the metagabbro (or metadiorite).Macroscopically visible hydrothermal alteration isrestricted to quartz^calcite^sulphide veinlets that cross-cut particularly the mafic rocks in variable orientations.The altered granite is macroscopically highly heteroge-

neous with equigranular, medium-grained and fine-grained domains. In places, centimetre-size pinkish alkali

feldspar grains give the rock an overall coarse-grained,inequigranular appearance. Under the microscope, themain phases are anhedral microcline (c. 36 vol. %), plagio-clase (c. 29 vol. %), and quartz (29 vol. %) with irregularto serrated grain boundaries. Microcline displays typicaltartan twinning and is, in places, altered to muscovite(2 vol. %). Plagioclase shows polysynthetic albite twinningand is partly saussuritized to epidote and white mica.Quartz is partly recrystallized.Variably chloritized biotiteoccurs in minor amounts (2 vol. %). Apatite, titanite,zircon, and magnetite are present as accessory phases(51 vol. %).The metagabbro (or metadiorite) is, except for the con-

tact with the overlying granite, mineralogically fairly uni-form, but texturally variable. The overall dark green,coarse- to fine-grained rock consists mainly of amphiboles(56 vol. %) and plagioclase (20 vol. %).The generally sub-hedral amphibole grains are mainly hornblende and theyare typically larger (up to several millimetres) than theinterstitial, anhedral plagioclase. Variably chloritized bio-tite makes up as much as 13 vol. %. Irregular patches ofthe hornblende grains are replaced by actinolite. Epidoteand sericite, derived from the saussuritization of plagio-clase, occur in addition to actinolite and chlorite, andquartz (1vol. %) as secondary, metamorphic^hydrothermalphases in minor amounts. Microcline is present in minoramounts (5 vol. %), irregularly distributed throughout themetagabbro (metadiorite) and especially along the con-tacts to veins. Zircon is relatively abundant as is apatite(1 vol. %). Accessory titanite is secondary and formedduring the chloritization of biotite and the oxidation ofpresumably primary ilmenite.The biotite-rich zone in the mafic portion immediately

below the granite is 1·15m in thickness and occurs as adark greenish-grey, fine-grained rock that contains irregu-larly distributed, anhedral, pink microcline grains, severalcentimetres in size. Largely chloritized biotite constituteson average 71 vol. % of the rock, with the remainderbeing predominantly saussuritized plagioclase (15 vol. %).Apatite is present in unusually high amounts (9 vol. %),mainly as euhedral inclusions in plagioclase. Other minorphases include quartz (1 vol. %), zircon, secondary titaniteand relics of a probably primary opaque phase (2 vol. %).Of special interest is a strong enrichment in allanite thatdisplays oscillatory zonation and occurs concentrated aseuhedral, several millimeter long grains at the bottom ofthis biotite-rich zone.All of the above lithotypes are cross-cut by various milli-

metre- to centimetre-thick veinlets that contain variableamounts of quartz, calcite, chlorite, microcline, pyrite andchalcopyrite. Depending on the mineralogy, their colourranges from white, grey, green to pink. Where they cutacross metabasite, the latter is typically affected by hydro-thermal growth of microcline and pyrite.

FRIMMEL et al. WITWATERSRAND HINTERLAND

2191

Fig. 3. Local geological map of the investigated borehole (BH1) site on the farm Rooidraai 85 IQ.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2192

The spatial distribution of the microcline in the meta-gabbro (metadiorite) is suggestive of post-magmaticK-metasomatism that is probably related to the emplace-ment of the granite or that of pegmatite pods in the vicin-ity of the borehole. Thus, some of the microcline observedin the granite may be secondary, in which case the originalcomposition of the granite would be granodioritic ratherthan granitic, analogous to the granodiorite that occurs inoutcrops near the borehole locality (Fig. 3).

MINERAL CHEMISTRYMineral chemical data were obtained by electron micro-probe analysis (EMPA) on 160 spots of amphiboles, feld-spars, biotite and chlorite, using a CAMECA S50instrument at the University of Wu« rzburg (accelerationvoltage 15 kV, beam current 15 nA). The analytical errorsare less than 1%, except for Na (52%).The amphibole compositions cover a wide range from

magnesio-hornblende to actinolite, irrespective of theassumed Fe2þ/Fe3þ ratio (Table 1), with 7·10^7·72 Si p.f.u.As expected, the Al content (Al2O3 6·56^2·19wt %) isnegatively correlated with Si and positively correlatedwithTi (TiO2 0·74^0·08wt %).Plagioclase in the mafic rocks is oligoclase with XAn of

0·13. Considering the extent of saussuritization, the origi-nal anorthite content must have been considerably higherbut can no longer be constrained. The plagioclase grainsin the altered granite are all essentially stoichiometricalbite. Again, a certain anorthite component has to beassumed for the original plagioclase grains in this rock,taking into account the extent of saussuritization. Overall,bearing in mind that both the felsic and mafic^intermediaterocks experienced the same post-intrusive alteration his-tory, an originally more calcic plagioclase composition forthe latter rocks compared with the granite can be inferredfrom the results obtained, but the initial anorthite contentof the magmatic plagioclase remains elusive. Thus, basedon purely petrographic criteria, it cannot be statedwhether the protolith of the mafic drill core portion was agabbro or a diorite. The alkali feldspar in the altered gran-ite, as well as in the more mafic rocks and the veinlets, isalmost pure K-feldspar throughout.Biotite is compositionally uniform and identical in both

the hornblende-rich mafic portion and the biotite-richzone. Its XFe is consistently 0·37; the TiO2 content is ele-vated and varies between 2·0 and 2·8wt % (Table 2).Secondary chlorite in the hornblende metagabbro

(metadiorite), where it replaces biotite, as well as in thecross-cutting veinlets has an XFe of c. 0·33 (Table 2). It isslightly more enriched in Fe in the biotite-rich zone(XFe¼ 0·45) and even more so where it replaced biotite inthe altered granite (XFe¼ 0·70).

CONSTRA INTS ONMETAMORPH IC^HYDROTHERMAL OVERPR INTThe original magmatic mineral assemblage K-feldspar^plagioclase^quartz^biotite in the granite or granodioriteshows evidence of potassic alteration in the form ofsecondary microcline growth and a low-grade meta-morphic overprint in the form of partial recrys-tallization of quartz and the sericitization andsaussuritization of microcline and plagioclase, respec-tively, as well as the chloritization of biotite. Qualitatively,a retrograde overprint at lower greenschist-facies condi-tions can be inferred.The primary mineral assemblage in the gabbroic or

dioritic rock is not easily reconstructed. The actinoliticdomains are evidently metamorphic but the dominantmagnesio-hornblende could be either a primary magmaticor an earlier metamorphic phase. A magmatic origin forthe hornblende seems indicated by a lack of any pyroxenerelics anywhere in the samples as well as the lack ofany other evidence of medium-grade metamorphismhaving affected the area. However, in the presence ofTi-minerals, such as titanite, in the rock a higherTi contentthan detected would be expected in the hornblende ifit were indeed magmatic. Similarly, the originalmagmatic plagioclase composition remains unknownbecause of retrograde alteration to oligoclase. The biotiteremnants are presumed to be magmatic because oftheir elevated Ti content, indicative of a relatively highformation temperature. A retrograde overprint is reflectedby chloritization of the biotite, formation of actinoliteand the saussuritization of plagioclase, all of which indi-cate lower greenschist-facies conditions, also supportedby the limited extent of quartz recrystallization in theadjacent altered granite. Assuming a relatively low pres-sure of c. 2 kbar (estimated from Na contents in the crystal-lographic B-site of amphibole), application of theplagioclase^amphibole thermometer (Holland & Blundy,1994), specifically the edenite^richterite thermometer,yielded a mean temperature of 439� 488C for 23actinolite^plagioclase pairs, largely dependent on the XAb

(0·86^0·99).The temperature at which the hydrothermal alter-

ation, evident in the various veinlets, took place appearsto have been similar to that of the pervasive retro-grade overprint. Some of the microcline, notably the por-phyroblastic textural variety and those grains thatoccur in the mafic^intermediate rocks adjacent to thehydrothermal veins, must be hydrothermal. The infiltrat-ing fluid, whose pH is constrained by the stability ofK-feldspar, was carbonic and S-bearing, as is evidentfrom the precipitation of calcite, pyrite and chalcopyritein the veinlets.

FRIMMEL et al. WITWATERSRAND HINTERLAND

2193

GEOCHEMISTRYMajor and trace elementsNotwithstanding the low-grade pervasive metamorphicand local hydrothermal overprint, the whole-rock geo-chemical compositions of the three principal rock types,hornblende metagabbro (metadiorite), altered granite andthe biotite-rich zone, have been used in an attempt toassess the genetic relationship between the felsic and moremafic rocks in the drill core and to constrain the likely

tectonic setting of the magmatism. Altogether, 17 hornble-nde metagabbro (metadiorite) samples, five samples of thebiotite-rich zone and seven altered granite samples wereanalysed for their major and trace element concentrations(Table 3). In addition, four samples of intensely veinedmetagabbro (metadiorite) were analysed to assess theeffect of hydrothermal alteration on the overall geochemis-try of the mafic^intermediate rocks. The major elementsand selected trace element contents were determined onfusion disks by conventional X-ray fluorescence

Table 1: Representative analyses of amphibole grains in the metagabbro

Core-metres 14·10–14·35 14·10–14·35 22·70–22·80 24·20–24·40 24·20–24·40 32·00–32·35

SiO2 54·35 49·08 53·13 50·32 52·52 52·60

TiO2 0·08 0·74 0·11 0·38 0·29 0·26

Al2O3 2·19 6·56 3·17 5·37 4·19 3·83

Cr2O3 0·14 0·13 0·19 0·14 0·17 0·12

FeO� 9·24 12·08 8·59 10·35 9·74 9·00

MgO 18·25 15·53 18·58 16·56 17·39 17·98

MnO 0·20 0·30 0·17 0·21 0·22 0·18

CaO 12·75 12·14 12·36 12·39 12·49 12·45

Na2O 0·32 1·04 0·49 0·92 0·58 0·75

K2O 0·08 0·61 0·19 0·35 0·23 0·22

Total 97·60 98·21 96·98 96·99 97·82 97·39

(a) (b) (c) (a) (b) (c) (a) (b) (c) (a) (b) (c) (a) (b) (c) (a) (b) (c)

Si 7·72 7·70 7·66 7·10 7·08 7·01 7·59 7·56 7·50 7·29 7·27 7·22 7·48 7·47 7·41 7·50 7·49 7·44

Al(IV) 0·28 0·30 0·34 0·90 0·92 0·99 0·41 0·44 0·50 0·71 0·73 0·78 0·52 0·53 0·59 0·50 0·51 0·56

Sum T 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00 8·00

Al(VI) 0·09 0·07 0·03 0·22 0·19 0·12 0·13 0·10 0·02 0·21 0·18 0·13 0·19 0·17 0·11 0·14 0·13 0·08

Fe3þ — 0·09 0·32 — 0·15 0·58 — 0·17 0·53 — 0·13 0·42 — 0·08 0·40 — 0·06 0·39

Ti 0·01 0·01 0·01 0·08 0·08 0·08 0·01 0·01 0·01 0·04 0·04 0·04 0·04 0·04 0·04 0·03 0·03 0·03

Cr 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·02 0·01 0·01 0·01

Mg 3·86 3·85 3·84 3·35 3·34 3·31 3·95 3·94 3·91 3·58 3·57 3·54 3·69 3·68 3·66 3·83 3·82 3·79

Fe2þ 1·02 0·96 0·76 1·33 1·22 0·85 0·89 0·76 0·49 1·15 1·06 0·82 1·06 1·01 0·74 0·99 0·95 0·68

Mn — — 0·02 — — 0·04 — — 0·02 — — 0·03 — — 0·03 — — 0·02

Sum C 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00 5·00

Fe2þ 0·08 0·04 — 0·13 0·08 — 0·13 0·10 — 0·10 0·05 — 0·10 0·07 — 0·08 0·08 —

Mn 0·02 0·02 — 0·04 0·04 — 0·02 0·02 — 0·03 0·03 — 0·03 0·03 — 0·02 0·02 —

Ca 1·94 1·94 1·93 1·88 1·88 1·86 1·89 1·88 1·87 1·92 1·92 1·91 1·91 1·90 1·89 1·90 1·90 1·89

Na 0·07 0·14 0·13 0·09 0·11 0·11

Sum B 2·04 2·00 2·00 2·05 2·00 2·00 2·04 2·00 2·00 2·05 2·00 2·00 2·04 2·00 2·00 2·01 2·00 2·00

Na 0·09 0·09 0·02 0·29 0·29 0·15 0·13 0·13 0·00 0·26 0·26 0·16 0·16 0·16 0·05 0·21 0·21 0·10

K 0·01 0·01 0·01 0·11 0·11 0·11 0·03 0·03 0·03 0·06 0·06 0·06 0·04 0·04 0·04 0·04 0·04 0·04

Sum A 0·10 0·10 0·03 0·40 0·40 0·26 0·16 0·16 0·03 0·32 0·32 0·22 0·20 0·20 0·09 0·25 0·25 0·14

Total 15·14 15·10 15·03 15·45 15·40 15·26 15·20 15·16 15·03 15·37 15·32 15·22 15·24 15·20 15·09 15·25 15·25 15·14

�All Fe expressed as FeO.(a) Normalization assuming all Fe to be FeO on the basis of 24 (O,OH); (b) normalization on the basis of 15eNK;(c) normalization on the basis of 13eCNK.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2194

spectroscopy (XRF) using a Philips PW1480 instrument,whereas most of the trace element (including the rareearth elements, REE) concentrations were obtained bylaser ablation-inductively coupled plasma mass spectrome-try (LA-ICPMS), using a Merchantek 266 LUV laser con-nected to an Agilent 7500i mass spectrometer, both at theGeodynamics and Geomaterials Research Division of theUniversity of Wu« rzburg. External calibration was carriedout with the aid of the NIST 612 glass standard (Pearceet al.,1997). Reproducibility was tested by repeated analysesof the NIST 612 and 614 glass standards and the in-housestandard BE-N. The lower limit of detection for most

trace elements is below 0·02 ppm, except for Cr(0·54 ppm), Cu (0·18 ppm), Gd (0·07 ppm) and Pb(0·12 ppm). The results are given inTable 3. An additional13 hornblende metagabbro (metadiorite) and four alteredgranite samples were analysed for both major and selectedtrace element concentrations by XRFonly. The results arevery similar to those above and are used in the data analy-sis but are not included inTable 3.In a total alkalis^silica (TAS) diagram (Fig. 4a) the

strong bimodal distribution of SiO2 contents reflects thepolarity in felsic and mafic rock compositions in the drillcore. Almost all of the felsic samples conform

Table 2: Representative compositions of biotite and chlorite in metagabbro

Biotite Chlorite

Core-metre: 32·00–32·35 12·00–12·10 22·70–22·80 14·10–14·35 14·10–14·35 5·65–5·75

Rock type: Hbl-gabbro Bio-gabbro Hbl-gabbro Vein Bio-gabbro Granite

SiO2 37·37 38·09 27·80 27·29 27·08 24·44

TiO2 1·97 2·76 0·05 0·03 0·01 0·01

Al2O3 15·24 13·88 19·87 19·73 18·51 18·65

Cr2O3 0·40 0·18 0·15 0·00 0·01 0·00

MgO 14·87 14·49 21·71 21·13 17·01 8·83

CaO 0·02 0·67 0·01 0·01 0·04 0·03

MnO 0·11 0·28 0·21 0·33 0·50 0·74

FeO� 15·68 15·49 17·98 18·48 24·63 36·12

BaO 0·01 0·00 0·00 0·00 0·00 0·00

Na2O 0·05 0·09 0·00 0·03 0·03 0·00

K2O 8·65 8·94 0·00 0·01 0·02 0·03

Total 94·37 94·87 87·79 87·03 87·83 88·85

Si 5·63 5·73 Si 5·62 5·59 5·69 5·44

AlIV 2·37 2·28 AlIV 2·38 2·41 2·31 2·57

T-site 8·00 8·00 T-site 8·00 8·00 8·00 8·00

AlVI 0·33 0·18 AlVI 2·35 2·35 2·27 2·32

Cr 0·05 0·02 Ti 0·01 0·01 0·00 0·00

Ti 0·22 0·31 Fe2þ 3·04 3·16 4·32 6·72

Fe2þ 1·97 1·95 Mn 0·04 0·06 0·09 0·14

Mn 0·01 0·04 Mg 6·54 6·45 5·33 2·93

Mg 3·34 3·25 Na 0·00 0·01 0·01 0·00

O-site 5·93 5·75 K 0·00 0·00 0·01 0·01

Ba 0·00 0·00 Sum 11·97 12·03 12·02 12·12

Ca 0·00 0·11

Na 0·02 0·03

K 1·66 1·71

A-site 1·68 1·85 Total 19·97 20·03 20·02 20·12

Total 15·61 15·59 XFe 0·32 0·33 0·45 0·70

XFe 0·37 0·37 T (8C)y 294 298 297 342

Biotite normalization on the basis of 22 O, chlorite normalization on the basis of 36 (O,OH).�All Fe as FeO.yTemperature calculated according to Kranidiotis & McLean (1987) for chlorite with XFe50·50.

FRIMMEL et al. WITWATERSRAND HINTERLAND

2195

Table 3: Geochemical analyses of different rock types in borehole BH1 on the farm Rooidraai

Rock type: Granite Biotite-rich zone

Core- 0·15– 1·65– 2·20– 3·10– 4·25– 5·10– 6·10– 7·00– 7·00– 7·40– 8·00– 9·30–

metre: 0·35 1·75 2·45 3·30 4·40 5·30 6·20 7·30A 7·30B 7·70 8·10 9·55

Depth: 4·25 5·7 6·3 7·2 8·3 9·20 10·15 11·15 11·15 11·55 12·05 13·4

SiO2 (%) 72·12 74·25 70·81 72·97 63·32 70·29 73·26 47·69 44·96 51·55 50·73 53·42

TiO2 0·16 0·07 0·18 0·11 0·50 0·22 0·06 1·67 2·23 1·48 1·12 0·57

Al2O3 13·96 13·48 14·60 13·80 15·81 13·96 13·79 17·65 14·30 12·39 12·08 13·10

Fe2O3� 1·33 0·76 2·19 0·83 4·38 2·42 0·70 12·71 15·06 11·9 11·60 6·56

MnO 0·02 b.d. 0·04 0·02 0·10 0·04 0·01 0·15 0·24 0·23 0·22 0·16

MgO 0·30 0·17 0·90 0·24 1·50 0·59 0·18 6·30 11·14 9·91 11·82 7·95

CaO 0·93 0·59 0·75 1·25 3·08 1·74 0·72 2·91 3·07 3·98 3·05 7·13

Na2O 3·90 3·68 4·86 4·13 5·41 4·07 3·26 4·82 1·17 1·55 1·08 3·46

K2O 5·38 5·44 3·48 4·86 2·25 4·54 6·46 1·43 1·63 1·46 3·20 2·56

P2O5 0·06 0·02 0·09 0·03 0·28 0·06 0·03 0·70 0·66 0·52 0·67 0·37

LOI 0·97 0·76 1·32 1·31 2·86 1·73 0·81 3·51 5·87 5·17 4·26 4·35

Total 99·13 99·22 99·22 99·55 99·49 99·66 99·28 99·54 100·33 100·14 99·83 99·63

Sc (ppm) 3·02 2·02 3·23 2·32 10·22 3·43 1·48 23·84 42·58 36·93 32·30 23·24

V 20·00 b.d. 23·00 b.d. 45·00 11·00 11·00 169·0 145·0 118·0 134·0 103·0

Cr 6·55 7·46 7·35 6·42 8·58 6·76 6·52 104·1 634·8 744·3 872·8 440·6

Co 35·0 43·0 27·0 39·0 40·0 46·0 49·0 35·0 49·0 39·0 59·0 41·0

Ni 3·98 3·45 4·39 2·92 6·38 4·49 2·69 52·00 125·60 121·75 162·80 103·60

Cu 7·45 6·54 8·58 5·62 12·52 8·50 5·37 23·92 27·16 14·33 12·86 40·20

Zn 31·0 17·0 51·0 21·0 94·0 37·0 29·0 208·0 320·0 265·0 284·0 117·0

Ga 21·0 18·0 28·0 17·0 24·0 19·0 19·0 38·0 50·0 26·0 30·0 20·0

Rb 212 213 210 202 140 186 232 110 65 167 471 147

Sr 211 185 124 182 264 236 203 284 105 177 185 320

Y 17·64 9·03 18·40 11·84 50·12 28·76 7·02 98·74 111·00 93·68 151·80 95·24

Zr 183·6 62·7 226·6 99·9 205·2 321·6 53·4 330·0 47·3 85·3 71·8 306·4

Nb 13·00 11·03 13·42 9·57 50·10 14·10 8·09 94·02 165·20 110·00 135·60 41·58

Mo 0·64 0·63 1·35 6·28 0·74 7·43 0·56 0·49 0·42 0·61 0·37 0·53

Sn b.d. b.d. b.d. b.d. b.d. b.d. b.d. 16·00 18·00 21·00 18·00 b.d.

Cs 2·88 2·19 9·95 3·24 2·45 2·09 2·19 4·59 6·16 8·95 19·74 2·51

Ba 638 615 357 562 247 622 757 179 270 192 216 452

La 47·6 17·5 7·1 28·4 71·8 95·2 8·5 103·7 88·5 55·8 439·8 49·7

Ce 81·9 30·6 11·5 46·7 132·6 169·8 16·5 205·4 190·2 115·8 895·6 84·5

Pr 8·86 3·56 1·49 5·04 15·26 18·88 1·90 22·26 21·44 13·50 88·74 10·50

Nd 31·54 11·86 6·31 18·22 56·80 67·56 6·51 85·84 78·64 55·13 290·20 43·88

Sm 5·63 2·32 1·87 3·30 11·72 10·98 1·60 18·18 19·16 15·00 48·64 11·84

Eu 0·91 0·68 0·51 0·80 1·25 1·26 0·73 1·25 1·16 1·20 2·09 1·31

Gd 4·52 2·16 2·42 2·71 10·17 8·09 1·38 19·02 20·60 14·90 35·04 12·34

Tb 0·61 0·29 0·45 0·38 1·54 1·06 0·24 2·85 3·26 2·53 5·01 2·20

Dy 3·66 2·02 3·18 2·21 9·31 5·83 1·40 18·00 20·60 16·30 28·24 15·48

Ho 0·71 0·37 0·61 0·47 1·78 1·10 0·27 3·47 3·87 2·99 5·18 3·22

Er 1·70 1·09 1·71 1·35 4·94 2·81 0·82 8·67 10·14 8·46 13·20 9·89

Tm 0·25 0·16 0·25 0·16 0·68 0·43 0·14 1·05 1·26 1·18 1·78 1·58

Yb 1·65 1·12 1·58 1·28 4·41 2·90 0·99 6·25 7·64 7·28 10·33 11·74

Lu 0·24 0·15 0·26 0·19 0·59 0·41 0·13 0·80 0·98 0·96 1·40 1·55

Hf 4·94 2·13 5·97 3·06 5·47 8·11 1·92 8·37 1·23 2·14 1·72 7·45

Ta 1·32 1·04 0·60 0·57 3·22 1·64 1·03 2·67 5·02 3·79 5·43 2·14

Pb 21·74 17·56 10·56 17·60 8·58 15·62 17·30 7·10 7·14 5·06 17·44 4·90

Th 14·06 5·58 7·04 12·76 10·86 17·38 3·04 17·00 20·06 15·10 72·40 2·60

U 5·40 4·71 5·90 7·50 5·10 4·38 2·07 2·47 1·97 4·01 4·04 4·36

(continued)

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2196

Table 3: Continued

Rock type: Hbl-gabbro with veins Hornblende gabbro

dyke

Core- 10·10– 11·00– 12·50– 28·00– 14·70– 15·60– 18·70– 18·80– 19·50– 19·60– 20·40– 21·10–

metre: 10·35 11·25 12·70 28·35 15·00 15·80 18·80 19·00 19·60 20·00 20·60 21·40

Depth: 14·25 15·15 16·6 32·2 18·85 19·7 22·75 22·9 23·55 23·8 24·5 25·25

SiO2 (%) 52·83 51·83 53·00 52·75 51·98 52·72 53·05 53·21 51·61 52·70 51·33 51·23

TiO2 0·53 0·52 0·49 0·43 0·47 0·41 0·33 0·32 0·29 0·34 0·42 0·50

Al2O3 10·02 9·61 8·84 8·45 8·64 9·02 8·78 8·69 7·64 8·49 8·40 8·91

Fe2O3� 9·06 9·32 9·07 8·49 9·05 8·47 8·25 8·42 8·93 8·70 8·94 9·42

MnO 0·17 0·18 0·17 0·16 0·16 0·15 0·15 0·15 0·20 0·17 0·17 0·17

MgO 12·60 13·18 13·32 13·70 13·83 13·82 13·35 13·68 13·73 13·85 14·71 13·57

CaO 8·10 9·00 9·18 9·47 9·23 9·02 8·81 8·99 9·31 8·95 9·04 9·27

Na2O 1·79 1·79 1·62 1·37 1·43 1·57 1·51 1·64 0·75 1·52 1·25 1·50

K2O 2·93 2·42 2·21 2·35 2·22 2·50 2·68 2·37 0·85 2·12 2·05 2·09

P2O5 0·40 0·40 0·39 0·32 0·35 0·32 0·20 0·23 0·18 0·23 0·30 0·38

LOI 2·08 2·67 2·52 2·88 2·20 2·04 2·13 2·04 6·09 2·10 2·49 2·36

Total 100·51 100·92 100·81 100·37 99·56 100·04 99·24 99·74 99·58 99·17 99·10 99·40

Sc (ppm) 29·66 37·72 29·90 32·00 31·16 29·54 28·58 27·90 36·68 29·18 31·84 32·82

V 143·00 144·00 142·00 128·00 122·00 115·00 96·00 98·00 105·00 108·00 116·00 142·00

Cr 891·80 1174·80 914·40 1382·60 1286·20 1204·40 1165·20 1230·00 1509·00 1231·80 1325·00 1059·80

Co 58·00 53·00 64·00 59·00 62·00 59·00 67·00 60·00 65·00 65·00 60·00 69·00

Ni 179·20 236·40 184·60 246·20 221·00 206·60 211·80 220·00 249·40 232·00 246·60 200·60

Cu 49·98 51·94 63·64 69·80 45·04 48·26 50·70 51·56 28·14 56·26 58·18 43·92

Zn 110·00 119·00 112·00 91·00 84·00 77·00 72·00 74·00 137·00 89·00 94·00 88·00

Ga 22·00 17·00 21·00 24·00 12·00 15·00 15·00 8·00 23·00 16·00 9·00 9·00

Rb 169·00 131·60 120·40 135·40 89·00 107·00 97·00 121·40 65·46 116·00 90·00 127·60

Sr 245·00 251·00 244·00 244·00 248·00 311·00 261·00 268·00 128·00 228·00 212·00 277·00

Y 30·22 25·54 18·82 18·52 18·24 16·00 15·90 15·76 48·68 23·04 15·48 17·16

Zr 111·80 120·00 95·64 63·04 99·80 97·84 118·80 112·80 150·80 86·76 70·86 105·00

Nb 28·22 21·78 13·80 14·44 12·58 10·78 10·27 9·51 20·38 15·14 10·51 8·30

Mo 0·47 0·96 9·27 0·68 0·57 0·46 0·43 0·73 0·48 0·55 0·76 0·59

Sn b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.

Cs 2·45 1·53 2·04 2·05 1·43 1·53 1·13 1·37 3·51 2·09 2·10 2·16

Ba 687·00 591·00 513·00 624·00 482·00 531·00 592·00 533·00 171·00 538·00 553·00 557·00

La 26·54 31·20 32·16 23·80 23·74 23·12 18·08 21·04 14·26 26·38 21·56 26·14

Ce 67·42 54·76 57·68 47·42 52·62 49·52 43·92 46·26 39·24 45·72 48·86 52·22

Pr 8·33 6·40 6·53 5·52 6·12 5·72 5·30 5·49 5·64 5·77 5·56 6·30

Nd 31·48 25·50 26·14 22·48 23·36 22·00 20·92 21·68 24·90 23·74 21·26 25·68

Sm 6·40 5·48 5·58 5·06 4·77 4·56 4·38 4·56 6·48 5·49 3·64 5·84

Eu 1·54 1·26 1·37 1·19 1·05 1·05 1·05 1·13 0·99 1·14 1·05 1·38

Gd 5·28 5·05 4·99 4·41 4·28 3·63 3·43 3·94 6·32 4·76 3·45 4·66

Tb 0·76 0·69 0·66 0·53 0·57 0·50 0·48 0·54 1·13 0·72 0·47 0·65

Dy 4·75 3·50 3·90 3·33 3·58 2·95 2·67 3·19 8·33 4·11 2·82 3·74

Ho 0·92 0·64 0·73 0·60 0·66 0·60 0·53 0·55 1·71 0·80 0·52 0·71

Er 2·81 1·57 2·00 1·60 1·81 1·47 1·47 1·57 5·25 2·25 1·49 1·82

Tm 0·49 0·25 0·29 0·20 0·24 0·20 0·21 0·25 0·96 0·37 0·22 0·27

Yb 4·10 1·66 2·10 1·73 1·93 1·57 1·57 1·69 7·54 2·86 1·74 1·72

Lu 0·62 0·23 0·31 0·24 0·28 0·25 0·24 0·23 1·12 0·42 0·23 0·23

Hf 2·67 2·18 2·67 1·71 2·71 2·48 2·83 3·10 3·70 2·55 1·92 2·98

Ta 1·70 1·41 1·48 1·78 1·96 1·48 1·46 1·14 1·30 1·43 1·07 1·30

Pb 5·97 4·51 5·82 5·77 6·33 6·07 6·29 7·11 3·85 6·71 4·86 5·89

Th 1·93 1·52 2·42 2·00 2·04 1·90 1·80 1·89 1·63 1·70 1·54 6·11

U 2·48 1·58 2·49 1·95 2·47 2·23 1·60 2·28 6·55 2·02 1·89 2·29

(continued)

FRIMMEL et al. WITWATERSRAND HINTERLAND

2197

Table 3: Continued

Rock type: Hornblende gabbro

Core- 22·20– 24·70– 25·45– 26·10– 26·55– 29·00– 31·80– 34·65–

metre: 22·40 25·00 25·70 26·30 26·80 29·30 32·00 35·00

Depth: 26·3 28·85 29·6 30·2 30·7 33·15 35·9 38·8

SiO2 (%) 51·38 51·12 50·66 50·52 51·07 51·78 52·72 52·53

TiO2 0·50 0·46 0·46 0·44 0·41 0·50 0·47 0·43

Al2O3 8·71 8·59 8·83 8·84 8·75 8·76 9·25 9·35

Fe2O3� 9·49 9·31 9·27 9·27 9·10 9·07 8·64 8·78

MnO 0·17 0·17 0·16 0·16 0·16 0·16 0·16 0·16

MgO 13·82 13·95 14·10 14·45 14·23 13·56 13·68 14·01

CaO 9·26 9·36 9·07 9·11 8·98 8·86 8·65 8·85

Na2O 1·41 1·50 1·51 1·48 1·36 1·48 1·52 1·61

K2O 2·17 1·80 1·80 1·44 1·86 2·21 2·67 2·51

P2O5 0·40 0·37 0·34 0·35 0·34 0·35 0·31 0·34

LOI 2·23 2·44 2·75 2·80 2·65 2·24 2·06 2·20

Total 99·54 99·07 98·95 98·86 98·91 98·97 100·13 100·77

Sc (ppm) 34·08 33·04 30·34 30·92 29·08 30·44 29·80 30·02

V 145·00 136·00 134·00 127·00 112·00 130·00 127·00 122·00

Cr 1035·50 1122·60 1163·80 1206·20 1218·20 1296·60 1203·60 1229·40

Co 64·00 57·00 70·00 74·00 67·00 67·00 67·00 59·00

Ni 196·00 212·40 230·40 232·80 239·00 228·00 209·40 214·60

Cu 54·98 40·92 46·88 54·32 38·68 54·58 54·24 54·34

Zn 93·00 120·00 81·00 73·00 75·00 91·00 91·00 94·00

Ga 11·00 11·00 16·00 19·00 17·00 10·00 9·00 14·00

Rb 93·00 83·00 110·60 75·00 101·48 101·00 115·00 96·00

Sr 266·00 257·00 277·00 289·00 277·00 255·00 324·00 280·00

Y 16·45 15·76 15·02 16·12 16·24 19·16 18·06 21·50

Zr 89·00 91·04 84·68 57·90 72·80 84·48 78·72 81·86

Nb 7·81 7·38 7·47 7·87 8·01 14·32 11·80 11·70

Mo 0·46 0·49 0·75 3·45 0·85 0·43 2·24 0·62

Sn b.d. 16·00 b.d. b.d. b.d. b.d. b.d. b.d.

Cs 2·28 1·21 2·19 2·46 1·89 2·78 2·09 1·60

Ba 573·00 458·00 520·00 353·00 599·00 579·00 751·00 665·00

La 22·98 20·28 23·48 23·46 24·00 21·94 19·98 22·32

Ce 48·45 44·54 45·24 50·26 50·52 49·22 45·58 51·92

Pr 5·81 5·48 5·43 6·02 6·08 5·81 5·34 6·12

Nd 24·20 22·38 22·68 23·24 24·84 23·68 21·28 24·30

Sm 5·03 5·11 4·62 5·38 5·26 5·53 4·72 5·25

Eu 1·25 1·12 1·19 1·20 1·20 1·18 1·12 1·23

Gd 4·59 4·30 3·99 4·49 4·04 4·57 4·03 4·50

Tb 0·61 0·62 0·56 0·59 0·58 0·64 0·58 0·65

Dy 3·39 3·23 3·37 3·21 3·38 3·72 3·27 3·99

Ho 0·62 0·59 0·57 0·62 0·59 0·70 0·58 0·75

Er 1·65 1·47 1·60 1·39 1·59 1·97 1·80 2·15

Tm 0·22 0·21 0·20 0·20 0·23 0·31 0·26 0·34

Yb 1·54 1·59 1·45 1·56 1·46 2·19 1·97 2·20

Lu 0·23 0·20 0·21 0·18 0·21 0·31 0·28 0·29

Hf 2·36 2·28 2·30 1·82 2·28 2·40 2·27 2·46

Ta 1·16 1·17 0·98 0·76 0·63 1·71 1·13 1·64

Pb 4·69 4·09 5·10 4·51 5·79 5·70 6·04 7·03

Th 2·39 1·94 2·65 2·56 1·92 2·71 1·75 2·47

U 1·82 2·10 2·12 1·65 1·63 3·02 1·88 2·20

�All Fe as Fe2O3.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2198

geochemically to metaluminous to peraluminous graniticcompositions. According to the TAS classification as pro-posed by Middlemost (1994), specifically the SiO2 contentsbetween 50·5 and 53·4wt %, all of the mafic rocks would

be classified as either gabbro or gabbroic diorite. For sim-plicity, they will be referred to in the following discussionas gabbro, with the understanding that a dioritic composi-tion of the protolith cannot be excluded in all cases. In theTAS space all of the mafic samples, except for the biotite-rich zone, are subalkaline. In terms of K contents, boththe mafic and felsic rocks could theoretically correspondto high-K calc-alkaline compositions. The measuredK concentrations are unlikely to be representative ofthe protolith’s composition, however, considering the evi-dence of hydrothermal overprinting and post-magmaticK-metasomatism. At least two stages of hydrothermalalteration can be distinguished: (1) a higher-temperature,more or less pervasive potassic alteration as evident fromthe distribution of microcline in the metagabbro and theformation of the biotite-rich zone near the contact betweenthe metagabbro and the granite; (2) a lower temperature,fracture-controlled alteration along veins.The expected high concentration of K in what is

referred to as the ‘biotite-rich zone’ is notably absent(Fig. 4a). This is due to almost complete chloritization ofthe biotite in that zone. Of interest is an enrichment in Pand Ti in this previously biotite-rich zone (Fig. 5a and b),which reflects the relatively high proportion of apatite andtitanite.The mafic samples with the greatest extent of hydrother-

mal veining do not differ significantly in their major ele-ment concentrations from those samples that are notvisibly veined (Fig. 4). Thus the lower temperature alter-ation did not pervasively alter the whole-rock composition.In contrast, the earlier K-metasomatism must haveaffected most, if not all, of the studied drill core and, con-sequently, any diagram relying on alkali elementdistribution for the characterization of the originalmelt composition cannot be reliable. A more reliable dis-criminant in this regard may be the relative proportionof Fe as shown in a FeO�/MgO vs SiO2 diagram(Miyashiro, 1974). All the hornblende metagabbrosamples plot in a tight cluster in the field of calc-alkalinecompositions (Fig. 4c), even those that contain hydrother-mal veins. The granite samples show a wider spread butall of them, with the exception of two extreme outliers(highly altered granite samples), also plot in the calc-alkaline field.The Ni and Cr concentrations are higher by about two

to three orders of magnitude in the relatively mafic rockscompared with the altered granite, with the biotite-richzone samples plotting in an intermediate position(Fig. 5c). Elevated Cu contents in some samples are relatedto the hydrothermal alteration because the veined meta-gabbro samples contain an order of magnitude more Cuthan the others. The hornblende metagabbro is enrichedin Ni relative to normal mid-ocean ridge basalt(N-MORB) by a factor of 1·5, whereas the biotite-rich

Fig. 4. Harker variation diagrams of Na2O þK2O (a), CaO þMgO(b), and FeO�/MgO (c) vs SiO2 for the various rock types in theinvestigated drill core. Distinction between calc-alkaline and tholeiiticcompositions in (c) after Miyashiro (1974).

FRIMMEL et al. WITWATERSRAND HINTERLAND

2199

transition zone is depleted (Ni� ¼ 0·75). All analysed maficrocks (without visible veining) are depleted in Cu, withMORB-normalized Cu in the hornblende metagabbrobeing 0·73 and in the biotite-rich zone only 0·32.Considering the altered nature of the studied rocks, even

the FeO^MgO^SiO2 distribution may no longer be repre-sentative of the original composition, and potentiallymore reliable information on the likely setting of the mag-matic protoliths may be obtained from the relationshipsbetween the least mobile elements; that is, the REE andhigh field strength elements (HFSE). The relationshipsbetween Nb, Y, Ta, and Yb in all the granite samples with

a narrow range in SiO2 that is within the range used byPearce et al. (1984) for the construction of their discrimina-tion diagrams conform to those of volcanic arc granites(Fig. 6). The least hydrothermally altered mafic composi-tions were normalized against mid-oceanic ridge basalt(N-MORB). The resulting diagram (Fig. 7) shows an over-all enrichment in the less compatible elements, negativeNb, Zr and Hf, as well as a strong negative Ti anomaliesand a positive Ho anomaly in all mafic samples. The traceelement concentrations in the biotite-rich transition zonefollow a similar pattern but this zone is markedly enrichedin Th and the REE, which can be explained by the highconcentration of allanite in this zone.The same zone is par-ticularly strongly depleted in Zr and Hf, reflecting a lowzircon content.All analysed samples are enriched in the light REE

(LREE) relative to chondrite (Fig. 8) but also relative toN-MORB (not shown). The strongest enrichment is notedin the biotite-rich zone and some of the granite samples.The hornblende metagabbro samples display a very uni-form REE distribution that is very similar to that of thosemetagabbro samples with hydrothermal vein-type alter-ation. Thus, the low-temperature hydrothermal alterationis regarded as not having significantly affected the overallREE patterns. Apart from the overall LREE enrichment,the gabbroic samples show no elemental anomalies.Notably, they lack any significant Eu anomaly. In contrast,the (chloritized) biotite-rich transition zone is character-ized by a marked negative Eu anomaly, whereas mostgranite samples yielded a less pronounced negative Euanomaly (Fig. 8).

Rb^Sr and Sm^Nd isotopesRb^Sr isotope analyses of five hornblende metagabbro andfive metagranite samples (Table 4) were carried out at theInstitute of Mineralogy, University of Mu« nster. Whole-rock powders (c. 100mg) were mixed with a 87Rb^84Srspike in Teflon screw-top vials and dissolved in a HF^HNO3 (5:1) mixture on a hot plate overnight. After evapo-ration and drying, 6N HCl was added to the residue andmixed to homogenization. After a second evaporation todryness, Rb and Sr were separated by standard ion-exchange procedures (AG 50W-X8 resin) on quartz glasscolumns using 2·5N and 6N HCl as eluents. Rb wasloaded with H2O on Ta double filaments and Sr wasloaded withTaF5 onWsingle filaments.The Rb and Sr iso-tope ratios were measured with a VG Sector 54 and aFinnigan Triton multicollector thermal ionization massspectrometer, respectively. Correction for mass fractiona-tion is based on a 86Sr/88Sr ratio of 0·1194. Rb ratios werecorrected for mass fractionation using a factor deducedfrom multiple measurements of the Rb standard NBS 607.Total procedural blanks were less than 15 pg for Rb andless than 30 pg for Sr. Based on repeated measurements

Fig. 5. TiO2 (a) and P2O5 vs SiO2 (b), and Cr vs Cu (c) for the var-ious rock types in the investigated drill core.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2200

the 87Rb/86Sr ratios were assigned an uncertainty of 1%(2s). In the course of this study, repeated runs of NBSstandard 987 gave an average 87Sr/86Sr ratio of0·710223�0·000018 (2s, n¼16).Both the Rb and Sr concentrations and respective iso-

tope ratios are very similar for the hornblende metagabbro

samples, whereas some spread and overall higher Rb con-centrations, and thus more radiogenic Sr is noted for thegranitic samples. Combining all 10 analyses yields anerrorchron, the slope of which corresponds to an impreciseage of 2633�50 Ma (Fig. 9) using a Rb decay constant of1·42�10�11 (Steiger & Ja« ger, 1977). The calculated initial

Fig. 6. Nb vsY (a) andTa vsYb diagrams (b) for the granitic rocks; discrimination of genetic types according to tectonic setting from Pearceet al. (1984). For comparison, the composition of quartz^feldspar porphyries of the Dominion Group is also shown (from Marsh et al., 1989) asa grey shaded field in (a).

FRIMMEL et al. WITWATERSRAND HINTERLAND

2201

87Sr/86Sr ratio is 0·7100� 0·0017. Using only the five granitedata points would result in an errorchron ‘age’ of2721�150 Ma and an initial 87Sr/86Sr ratio of0·7059� 0·0069 (MSWD¼ 226).

The Nd isotope ratios (Table 4) were measured on aVGSector 7-collector mass spectrometer in multi-dynamicmode at the Department of Geological Sciences,University of Cape Town, following the standard chemical

Fig. 7. N-MORB-normalized trace element patterns for mafic rocks in the investigated drill core. Elements are ordered according to decreasingcompatibility. N-MORB composition from Hofmann (1988).

Fig. 8. Chondrite-normalized rare earth element patterns in the principal rock types of the investigated drill core; normalization values fromSun & McDonough (1989).

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2202

separation techniques described by le Roex & Lanyon(1998). A depleted mantle isotopic composition of143Nd/144Nd¼ 0·5131 and 147Sm/144Nd¼ 0·2136 (Henryet al., 2000), and a 147Sm decay constant of 6·54�10�12

(Lugmair & Marti, 1978) were used for the calculation ofthe Nd model ages and the eNd values. The eNd values atthe time of likely formation (3·07 Ga: see Fig. 10 and thesubsequent section on geochronology) obtained formost of the granite and the hornblende metagabbro sam-ples overlap and cluster between �1·8 and þ1·9, but for

one felsic and one mafic sample they are much lower, at�10·4 and �4·8, respectively. The subchondritic eNd

values, corresponding to calculated TDM model ages thatare higher than the age of magmatism (see below,Table 4), might be an indication of contamination by oldercrust. This argument is, however, not conclusive becausethe model age strongly depends on the preferred modelfor the evolution of the mantle composition in theArchaeança controversial topic that is beyond the scopeof this paper.

Table 4: Rb^Sr and Sm^Nd isotope data for hornblende metagabbro and metagranite samples

Core length Rock Rb Sr 87Rb/86Sr 87Sr/86Sr Error (2s) Sm Nd 147Sm/144Nd 143Nd/144Nd Error (2s) TDM eNd

(ppm) (ppm) (ppm) (ppm) (Ma) (3062 Ma)

22·20–22·40 Hbl-metagabbro 97·6 321 0·8842 0·745716 0·000013 5·03 24·20 0·12551 0·511292 0·000009 3106·2 1·85

18·70–18·80 Hbl-gabbro 98·1 293 0·9733 0·746338 0·000013 4·38 20·90 0·12967 0·511192 0·000009 3437·7 �1·78

15·60–15·80 Hbl-gabbro 102 322 0·9211 0·743865 0·000016 4·56 22·00 0·12929 0·511030 0·000008 3708 �4·80

31·80–32·00 Hbl-gabbro 110 342 0·9384 0·746976 0·000016 4·72 21·30 0·13381 0·511336 0·000010 3343·7 �0·59

19·60–20·00 Hbl-gabbro 85·8 257 0·9704 0·746410 0·000015

5·10–5·20 granite 187 231 2·3642 0·798783 0·000013 11·00 67·60 0·098243 0·510640 0·000011 3226·5 �0·13

3·10–3·30 granite 203 177 3·3529 0·837933 0·000018 3·30 18·20 0·109476 0·510859 0·000008 3256·2 �0·29

1·65–1·75 granite 203 179 3·3203 0·836689 0·000027

6·10–6·20 granite 235 203 3·4030 0·839558 0·000015 1·60 6·51 0·148403 0·511133 0·000009 4545 �10·38

0·15–0·35 granite 214 205 3·0502 0·826980 0·000013 5·63 31·50 0·10791 0·510753 0·000008 3358 �1·74

Fig. 9. Rb^Sr isochron plot for the granite and hornblende metagabbro.

FRIMMEL et al. WITWATERSRAND HINTERLAND

2203

U^PB AND HF I SOTOPE DATAON Z IRCONTo constrain the age and to further characterize the sourceof the two principal magmatic rock types in the drill core,single zircon grains were separated from both the variablyaltered granite and the hornblende metagabbro by stan-dard crushing and heavy mineral separation techniques.Polished zircon grain mounts were imaged by scanningelectron microscope cathodoluminescence (CL) using aJEOL JSM-6400 electron microprobe at the Institute ofGeosciences, University of Frankfurt. Selected zircondomains were analysed for their U^Pb isotopic composi-tion by LA-ICPMS at the same institution, using aThermo-Finnigan element II sector field ICPMS systemcoupled to a New Wave UP213 UV laser system. Theresults are listed inTable 5. Analytical details, data process-ing, and error calculations have been given by Gerdes &Zeh (2006, 2008). Concordia and upper intercept ages onconcordia diagrams were calculated using the Isoplot/Ex2.49 software (Ludwig, 2000).Subsequently, the same zircon domains were analysed

for their Lu, Hf, and Yb isotopic composition (Table 6)using the same laser system, with a 40 mm spot size, and aThermo-Finnigan Neptune multicollector (MC)-ICPMSsystem. The procedures for correction of isobaric interfer-ences between Lu and Yb, instrumental mass fractiona-tion, and comparison with standards have been detailedby Gerdes & Zeh (2006, 2008). Multiple analyses by LA-

MC-ICPMS of the GJ1 zircon standard during the periodof this study yielded 176Hf/177Hf of 0·281998�0·000015(n¼10). All uncertainties are reported at the 2s level.The zircon grains in the granite are typically 0·1^

0·2mm in length and a number of them display complexinternal structures. Some grains show a relic oscillatoryzonation, overprinted by several patchy domains thatpoint to recrystallization of the original magmatic zoning(Fig.11a, insert). In some zircon grains, however, only com-positional banding can be distinguished by bright anddull CL domains.Many of the analysed zircon domains are rich in U, with

contents up to 1316 ppm. These domains yielded the lowest206Pb/204Pb ratios and the most discordant results. In con-trast, those domains with the lowest U contents yieldedperfectly concordant age data (Table 5). Two concordantdomains (97 and 100% concordance) gave 207Pb^206Pbages of 3083 and 3045 Ma, respectively. A third concor-dant analysis corresponds to a significantly younger age of2746�62 Ma. The rest of the analyses are variably discor-dant. In spite of this wide scatter of U^Pb age data,almost all (except for two) analysed domains yieldedwithin error identical 176Hf/177Hf ratios (0·28084� 0·00003) calculated for the apparent 207Pb^206Pb age(Fig. 12a, Table 6). This indicates that all these zircondomains crystallized from the same magmatic source butunderwent variable Pb loss as a result of later alteration(Gerdes & Zeh, 2008; Zeh et al., 2009). Two analyses devi-ate from the mean 176Hf/177Hf ratio. One of them has alower 176Hf/177Hf and could reflect inheritance, whereasthe other has a higher ratio, which might point to laterincorporation of radiogenic Hf. Zircon formation duringa later magmatic^metamorphic^hydrothermal event canbe excluded on the basis of the combined U^Pb and Hfdatasets shown in Fig. 12a. If such later zircon crystalliza-tion had taken place, the resulting domains should ploton, or above, the 176Lu/177Hf whole-rock evolution line.From the combined U^Pb and Lu^Hf isotope data it canbe concluded that the oldest concordant ages best reflectthe time of granite emplacement; that is, 3064�20 Ma ascalculated from the two most concordant analyses, A14and A15 inTable 5 (Fig. 11a).The zircon grains in the hornblende metagabbro are on

average larger than (up to 0·8mm in length) but of similarhabit to those in the granite and also show complex inter-nal structures with patchy bright and dull CL domains(Fig. 11b, insert). A large number of zircon domains, typi-cally with U contents of much less than 200 ppm, yieldedconcordant results (Fig. 11b). Eight domains gave concor-dant (99^101%) age data that correspond to a concordiaage of 3064�7 Ma, including the error on the decay con-stant. This age is identical to an upper intercept age of3063�5 Ma obtained on 11 spot analyses (Fig. 11b), whichis regarded as the best constraint on the time of

Fig. 10. eNd evolution diagram for the granite and hornblende meta-gabbro; for comparison the data field for the Dominion Group volca-nic rocks (recalculated data from Marsh et al., 1989) is also shown.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2204

Table 5: U^Pb isotope data of single zircon domains

Sample U1 Pb1 Th/U1 206Pb/204Pb 206Pb/238U2�2s 207Pb/235U3

�2s 207Pb/206Pb2 �2s rho4 Age (Ma)

no. (ppm) (ppm) (%) (%) (%) 206Pb/238U �2s

(%)

207Pb/235U �2s

(%)

207Pb/206Pb �2s

(%)

Con.5

(%)

BH1-metagranite

A1 500 120 0·87 895·52 0·18910 3·2 4·888 3·6 0·18747 1·6 0·89 1116 33 1800 30 2720 27 41

A2 677 77 0·63 663·17 0·07754 10·5 1·721 11·4 0·16095 4·5 0·92 481 49 1016 76 2466 76 20

A3 311 147 0·78 2492·52 0·38819 9·5 11·548 9·8 0·21577 2·5 0·97 2114 173 2569 96 2949 40 72

A4 1088 186 0·46 541·40 0·12430 8·9 2·915 9·6 0·17011 3·4 0·93 755 64 1386 75 2559 57 30

A5 269 186 0·65 867·89 0·51473 8·6 13·515 9·4 0·19043 3·8 0·92 2677 192 2716 93 2746 62 97

A6 312 107 0·57 1294·63 0·23743 10·4 6·166 10·8 0·18835 2·9 0·96 1373 130 2000 99 2728 47 50

A7 303 93 0·84 4960·32 0·27662 5·6 7·148 6·5 0·18742 3·3 0·86 1574 79 2130 59 2720 54 58

A8 408 94 0·83 5374·14 0·20430 5·3 4·790 5·9 0·17006 2·6 0·90 1198 58 1783 51 2558 44 47

A9 426 67 0·85 6272·33 0·13968 5·5 2·776 6·2 0·14415 2·7 0·90 843 44 1349 47 2278 47 37

A10 341 116 0·92 906·62 0·22827 7·2 5·842 7·9 0·18561 3·1 0·92 1325 87 1953 71 2704 51 49

A11 198 114 0·60 1132·41 0·38801 3·3 11·750 3·7 0·21964 1·8 0·88 2114 59 2585 36 2978 29 71

A13 202 115 0·53 6420·44 0·48639 5·9 13·734 6·3 0·20479 2·4 0·93 2555 125 2732 62 2865 39 89

A14 67 47 0·49 21055·69 0·59171 4·0 19·133 4·3 0·23451 1·6 0·93 2996 96 3049 42 3083 25 97

A15 64 45 0·92 5691·13 0·60549 4·5 19·117 4·9 0·22898 1·9 0·92 3052 111 3048 48 3045 31 100

A16 163 92 1·03 30326·75 0·48851 4·2 14·399 5·3 0·21378 3·2 0·80 2564 89 2776 51 2934 51 87

A17 1316 80 0·53 407·67 0·04230 13·3 0·631 13·5 0·10811 2·1 0·99 267 35 496 54 1768 39 15

A18 183 47 1·58 5614·70 0·19251 7·1 5·283 7·3 0·19904 1·9 0·97 1135 74 1866 65 2818 30 40

BH1-hornblende metagabbro

A1 331 66 0·71 5274·18 0·16279 6·4 4·624 6·6 0·20600 1·6 0·97 972 58 1754 56 2874 25 34

A2 335 81 1·82 1301·30 0·17600 3·9 4·430 4·3 0·18255 1·8 0·91 1045 38 1718 37 2676 30 39

A3 510 61 1·41 823·85 0·07949 2·4 1·650 3·4 0·15056 2·4 0·71 493 11 990 21 2352 40 21

A4 163 81 0·68 30538·03 0·44797 2·0 12·878 2·4 0·20849 1·4 0·81 2386 40 2671 23 2894 23 82

A5 358 82 1·11 17357·28 0·19269 2·4 5·836 2·6 0·21965 1·0 0·92 1136 25 1952 23 2978 16 38

A6 99 70 0·49 10869·06 0·60732 1·8 19·290 2·0 0·23036 1·0 0·87 3059 43 3056 20 3055 16 100

A7 92 68 0·64 7895·48 0·60824 1·3 19·335 1·6 0·23055 0·9 0·82 3063 32 3059 16 3056 15 100

A8 106 74 0·34 20377·72 0·60554 2·3 19·376 2·4 0·23207 0·8 0·94 3052 55 3061 23 3066 13 100

A9 128 89 0·36 34568·41 0·60691 2·4 19·397 2·6 0·23180 0·9 0·94 3058 60 3062 25 3065 14 100

A10 104 70 0·21 13214·50 0·60098 2·2 19·169 2·3 0·23133 0·8 0·94 3034 53 3050 23 3061 13 99

A11 116 84 0·35 2943·58 0·60482 2·3 19·320 2·6 0·23167 1·1 0·90 3049 56 3058 25 3064 18 100

A12 216 193 1·52 1579·22 0·61810 1·5 19·752 2·6 0·23177 2·1 0·59 3102 37 3079 25 3064 33 101

A13 302 45 0·66 4501·20 0·11856 3·3 2·994 3·5 0·18316 1·2 0·94 722 22 1406 27 2682 19 27

A14 407 53 0·51 1163·51 0·1071 3·4 2·741 4·1 0·1855 2·2 0·84 656 21 1340 31 2703 37 24

A15 374 93 0·87 1010·09 0·20086 1·9 5·0723 2·1 0·18315 1·0 0·89 1180 20 1831 18 2682 16 44

A16 220 118 0·71 1280·60 0·4381 2·9 13·393 3·2 0·2217 1·3 0·91 2342 57 2708 31 2993 21 78

A17 93 76 0·71 798·30 0·6289 2·3 20·089 2·7 0·2317 1·5 0·84 3145 57 3096 27 3064 24 103

A19 526 351 4·20 66·11 0·0187 19·9 0·53 20·0 0·2053 1·7 1·00 119 24 432 73 2869 28 4

A20 499 313 3·03 85·50 0·4148 3·5 10·94 5·1 0·1913 3·7 0·69 2237 67 2518 49 2754 60 81

A21 542 287 0·77 97·69 0·3263 2·7 7·416 4·3 0·1648 3·3 0·64 1820 44 2163 39 2506 56 73

A22 198 96 0·22 598·54 0·3938 4·9 12·0728 5·1 0·2223 1·6 0·95 2140 89 2610 49 2998 26 71

A23 325 237 0·90 155·36 0·5400 2·5 14·9350 2·9 0·2006 1·4 0·88 2783 57 2811 28 2831 23 98

A24 571 105 0·88 539·58 0·1434 2·6 3·8585 2·8 0·1951 0·9 0·94 864 21 1605 23 2786 15 31

A25 166 98 0·29 7837·33 0·5462 2·7 16·0264 2·8 0·2128 1·0 0·94 2810 61 2878 27 2927 15 96

A26 375 70 1·83 548·62 0·1385 2·1 3·2441 3·4 0·1699 2·6 0·64 836 17 1468 27 2556 44 33

A27 55 41 0·55 3577·39 0·6136 4·0 19·8287 4·2 0·2344 1·1 0·96 3085 99 3083 41 3082 18 100

A28 143 87 0·40 1039·91 0·5287 2·0 16·8013 2·1 0·2305 0·8 0·93 2736 44 2924 21 3055 12 90

A29 82 42 0·65 1804·93 0·3911 3·1 12·49 3·4 0·2315 1·4 0·91 2128 57 2642 33 3063 22 69

Zircon standard

GJ1-4 178 82 1·09 12827·13 0·37576 3·6 11·364 4·2 0·21933 2·1 0·87 2056 64 2553 40 2976 34 69

1Calculated relative to GJ-1 reference zircon.2Corrected for background, within-run Pb/U fractionation and common Pb using Stacey & Kramers (1975) model Pb compositionand subsequently normalized to GJ-1 values.3Calculated using 207Pb/206Pb/(238U/206Pb� 1/137·88).4Rho is the error correlation; that is, error(206Pb/238U)/error(207Pb/235U).5Degree of concordance.

FRIMMEL et al. WITWATERSRAND HINTERLAND

2205

Table 6: Lu^Hf isotope data for single zircon domains

Sample 176Yb/177Hf1 �2s 176Lu/177Hf1 �2s 178Hf/177Hf 180Hf/177Hf SigHf2 176Hf/177Hf �2s3 176Hf/177Hf(t)4 eHf(t)

5�2s TDM

5 Age5 �2s

no. (V) 3062 Ma (Ga) (Ma)

BH1-granite

A1 0·1010 28 0·00241 4 1·46740 1·88685 7 0·281009 23 0·280883 1·7 0·8 3·31 2720 27

A2 0·0565 40 0·00137 7 1·46731 1·88673 8 0·280916 22 0·280851 0·6 0·8 3·37 2466 76

A3 0·0947 29 0·00223 7 1·46709 1·88610 8 0·280951 24 0·280825 0·1 0·9 3·40 2949 40

A4 0·1068 73 0·00257 15 1·46734 1·88681 7 0·280986 27 0·280861 0·6 0·9 3·37 2559 57

A5 0·1037 31 0·00247 6 1·46719 1·88654 7 0·280951 24 0·280821 �0·5 0·9 3·43 2746 62

A6 0·1078 36 0·00260 11 1·46723 1·88632 7 0·280958 25 0·280822 �0·5 0·9 3·43 2728 47

A7 0·0638 52 0·00160 12 1·46715 1·88631 9 0·280919 26 0·280836 0·2 0·9 3·39 2720 54

A8 0·0846 25 0·00208 8 1·46732 1·88638 9 0·280934 28 0·280833 �0·2 1·0 3·41 2558 44

A9 0·1225 40 0·00302 12 1·46704 1·88593 9 0·280974 26 0·280842 �0·8 0·9 3·45 2278 47

A10 0·09269 74 0·00230 19 1·46724 1·88647 8 0·280969 27 0·280837 0·5 1·0 3·37 2704 51

A11 0·0786 59 0·00200 15 1·46709 1·88607 8 0·280981 23 0·280866 1·6 0·8 3·31 2978 29

A13 0·0835 34 0·00209 9 1·46721 1·88631 9 0·280891 20 0·280776 �1·8 0·7 3·50 2865 39

A14� 0·0409 18 0·00108 4 1·46727 1·88693 7 0·280878 23 0·280814 �0·1 0·8 3·41 3083 25

A15� 0·0337 8 0·00090 2 1·46725 1·88671 7 0·280886 25 0·280833 0·5 0·9 3·37 3045 31

A18 0·0893 42 0·00215 11 1·46735 1·88680 8 0·280959 25 0·280843 0·5 0·9 3·37 2818 30

BH1-gabbro

A1 0·0364 69 0·00114 21 1·46745 1·88674 8 0·280839 21 0·280777 �1·7 0·8 3·49 2874 25

A2 0·0615 49 0·00187 12 1·46711 1·88655 7 0·280865 23 0·280769 �2·3 0·8 3·53 2676 30

A3 0·0255 75 0·00072 18 1·46690 1·88588 8 0·280827 21 0·280795 �1·2 0·7 3·47 2352 40

A4 0·0159 12 0·00059 5 1·46720 1·88657 13 0·280810 19 0·280777 �1·6 0·7 3·49 2894 23

A5 0·0183 4 0·00066 1 1·46720 1·88649 13 0·280823 15 0·280785 �1·2 0·5 3·47 2978 16

A6� 0·0132 10 0·00042 3 1·46721 1·88628 8 0·280833 24 0·280809 �0·4 0·9 3·42 3055 16

A7� 0·0156 12 0·00049 2 1·46717 1·88625 8 0·280803 18 0·280775 �1·6 0·6 3·49 3056 15

A8� 0·0280 37 0·00090 12 1·46720 1·88669 7 0·280859 26 0·280806 �0·5 0·9 3·43 3066 13

A9� 0·0292 35 0·00094 12 1·46711 1·88631 8 0·280846 23 0·280791 �1·0 0·8 3·46 3065 14

A10� 0·0121 15 0·00040 5 1·46713 1·88650 8 0·280815 18 0·280791 �1·0 0·7 3·45 3061 13

A11� 0·0317 107 0·00091 31 1·46717 1·88649 10 0·280855 23 0·280801 �0·6 0·8 3·43 3064 18

A12� 0·0423 111 0·00129 34 1·46733 1·88661 7 0·280868 23 0·280792 �0·9 0·8 3·45 3064 33

A13 0·0415 54 0·00126 14 1·46732 1·88662 8 0·280855 20 0·280790 �1·3 0·7 3·48 2682 19

A14 0·0108 13 0·00035 3 1·46710 1·88654 9 0·280807 25 0·280789 �1·2 0·9 3·46 2703 37

A16 0·0296 23 0·00088 7 1·46704 1·88603 6 0·280826 23 0·280775 �1·6 0·8 3·49 2993 21

A17� 0·0311 108 0·00097 35 1·46738 1·88678 8 0·280826 22 0·280769 �1·8 0·8 3·50 3064 24

A19 0·1386 30 0·00389 6 1·46727 1·88667 8 0·281003 24 0·280789 �1·6 0·8 3·49 2869 28

A21 0·0659 24 0·00200 4 1·46739 1·88695 6 0·280899 31 0·280803 �1·3 1·1 3·48 2506 56

A22 0·0049 3 0·00015 1 1·46723 1·88660 12 0·280786 16 0·280777 �1·5 0·6 3·48 2998 26

A23 0·0655 47 0·00181 14 1·46710 1·88640 7 0·280910 25 0·280812 �0·5 0·9 3·43 2831 23

A24 0·0662 16 0·00181 6 1·46721 1·88664 6 0·280880 21 0·280784 �1·6 0·7 3·49 2786 15

A25 0·0106 4 0·00035 2 1·46714 1·88657 8 0·280812 15 0·280793 �1·0 0·5 3·45 2927 15

A26 0·0762 29 0·00227 5 1·46719 1·88676 7 0·280921 20 0·280810 �1·1 0·7 3·46 2556 44

A27� 0·0216 9 0·00067 4 1·46717 1·88657 7 0·280834 23 0·280794 �0·9 0·8 3·45 3082 18

A28� 0·0371 56 0·00115 19 1·46715 1·88680 7 0·280847 22 0·280780 �1·4 0·8 3·48 3055 12

A29� 0·0926 83 0·00277 24 1·46720 1·88638 7 0·280966 24 0·280803 �0·6 0·9 3·43 3063 22

JMC (50 ppb) (n¼ 7) 1·46718 1·88643 13 0·282162 14

GJ-1 0·0094 13 0·00031 6 1·46722 1·88656 9 0·282004 18 0·282000 �13·9 0·6 609 6

(n¼ 15)

1 176Yb/177Hf ¼ (176Yb/173Yb)true �173Yb/177Hf)meas � [M173(Yb)/M177(Hf))]b(Hf). 176Lu/177Hf calculated in a similar way by using

the 175Lu/177Hf. Quoted uncertainties (absolute) relate to the last quoted figure; Effect of inter-element fractionation on Lu/Hf isestimated to be about 6% or less based on analyses of the GJ-1 and Plesovice zircons.2Mean Hf signal in volts.3Uncertainties are quadratic additions of the within-run precision and the daily reproducibility of the 50 ppb JMC475 solution.Uncertainties for the JMC475 and GJ-1 are 2SD (2 standard deviations).4Initial 176Hf/177Hf are calculated using the apparent Pb–Pb age determined by LA-ICPMS dating (see last two rows).5All eHf and TDM are calculated for the emplacement age of 3062 Ma, TDM is the two-stage model age calculated by using themeasured 176Hf/177Hf of each spot (first stage ¼ emplacement age), a value of 0·0113 for the average continental crust (secondstage), and a depleted mantle 176Lu/177Hf and 176Lu/177Hf of 0·0384 and 0·28325, respectively.�Most concordant analyses.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2206

Fig. 11. Concordia diagrams for zircon grains from the granite (a) and the hornblende metagabbro (b). Data point error ellipses are 2s. Alsoshown are cathodoluminescence images of typical zircon grains from each lithotype; scale bars represent 0·1mm. Lines I represent discordialines from the concordant crystallization age forced through the origin, lines II reflect a later thermal overprint (for further details see text).

FRIMMEL et al. WITWATERSRAND HINTERLAND

2207

emplacement. There are, however, also a number of otherdiscordant zircon analyses that plot to the left of the discor-dia line as shown in Fig. 11b. All analyses, irrespective ofthe level of discordance, yielded within error identical176Hf/177Hf ratios (0·28079� 0·000026) calculated for theapparent 207Pb^206Pb age (Fig. 12b, Table 6). This relation-ship is well reflected even by spot analyses obtained fromdifferent zircon domains that have different internal struc-ture, U contents and 176Yb/177Hf ratios within the samegrain (Fig. 12b, Tables 5 and 6). This indicates that all ofthese zircon grains were formed from an isotopically

homogeneous magma. As with the zircon in the granite,multiple zircon growth can be excluded and the variablediscordance, and 207Pb^206Pb ages, can be explained bymultiple alteration events after zircon growth.Significantly, the U^Pb and Hf isotope data obtained on

zircon domains from both the altered granite and thehornblende metagabbro appear very similar. Their respec-tive emplacement ages and Hf isotope characteristics areidentical within error. The eHf calculated for the time ofemplacement (3063 Ma) ranges from �1·8 to þ1·7 in thecase of the granite and from �2·3 to �0·4 in the

Fig. 12. Calculated 176Hf/177Hf ratio at the time of the apparent 207Pb^206Pb age vs apparent 207Pb^206Pb age for single zircon domains fromthe granite (a) and the hornblende metagabbro (b). The concordant analyses (97^103% concordance level) are shown as filled symbols. Itshould be noted that the corresponding eHf at the time of emplacement are all within a very narrow range close to zero, but slightly lower inthe hornblende metagabbro.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2208

hornblende metagabbro. The mean eHf(t) values areslightly higher for the metagranite (þ0·2) than for thehornblende metagabbro (^1·2). These values correspond toHf model ages of 3·32 Ga for the metagranite and 3·47Ga for the hornblende metagabbro, calculated by usingthe parameters as outlined in the legend of Table 6 and dis-cussed by Zeh et al. (2007).The U^Pb isotope data indicate that the zircon grains in

both rock types experienced multiple Pb loss. A major Pbloss event seemingly occurred at 2720 Ma, as reflected byone concordant data point obtained from the altered gran-ite (spot A1, Table 5) and by several discordant analysesthat yielded 207Pb^206Pb ages around 2700 Ma and ploton a reference line between the concordant datum and theorigin (dashed line II in Fig. 11a). Subsequent Pb loss is fur-ther indicated by a few zircon analyses that yielded youn-ger 207Pb^206Pb ages (Table 5).

AR^AR AGE DATA FORHORNBLENDEThe above petrological and U^Pb zircon age data indicatethat both the granitic and gabbroic protoliths intruded atabout 3·07 Ga, and were altered afterwards, most probablyat 2·7 Ga. At present, it is unclear whether the low-Ti com-position of the magnesio-hornblende in the metagabbro isa result of a Neoarchaean alteration process or a primaryfeature related to the emplacement of the mafic intrusion.To assess the origin of the magnesio-hornblende in themetagabbro, we conducted 40Ar^39Ar analyses on hornble-nde separates from four positions within the mafic portionof the drill core.Between 60 and 120mg of hornblende concentrates were

enclosed in high-purity quartz vials and irradiated at thenuclear research reactor VR-1 in Prague, Czech Republic.Once cooled the samples were filled into annealed Ta cap-sules and subsequently analysed by stepwise heatingexperiments at the CEAL Laboratory of the SlovakAcademy of Sciences in Bratislava. The analytical detailsfor the fully automatic Ar-extraction and purification lineare as described by Frimmel & Frank (1998; at that timethe line was still housed at the former Institute ofGeology, University of Vienna). The paper by Frimmel &Frank also includes details regarding corrections for massdiscrimination and radioactive decay, as well as for thedetermination of the J-value (0·013552�0·4%) and thedefinition of a plateau age.The K/Ca ratio was determinedfrom the 39Ar/37Ar ratio (calculated for the end of irradia-tion) using a conversion factor of 0·247. The 40Ar/36Arratio of the line blank was close to air compositionthroughout the study (299�1·0%).The errors of the calcu-lated ages for single steps are given as 1s. The errors ofthe plateau and total gas ages include an additional errorof �0·4% on the J-value.

The hornblende separates from all four positions in thedrill core yielded similar but not identical results withsome significant subtle differences (Fig. 13, Table 7). Thehornblende from position 15·1^15·5m, nearest the overly-ing granite, gave a total gas age of 3064�19 Ma. No dis-tinct plateau can be recognized (Fig. 13a). Lower ages atthe lowest temperature steps are most probably caused bysecondary overgrowth of actinolite and/or chlorite. Thehighest ages were obtained on relatively low-temperaturesteps and this is regarded as reflecting the incorporationof excess Ar into the hornblende lattice as a consequenceof partial chloritization. Hornblende in drill core 19·3^19·4m yielded an identical total gas age of 3074�21 Ma.Some Ar loss is indicated by lower apparent ages obtainedfor the lowest temperature steps, which are also character-ized by elevated K/Ca (Fig. 13b). Most of the 39Ar released(65%) defines a very good plateau that corresponds to anage of 3078�20 Ma. The analytical data were also usedto construct 40Ar/36Ar vs 39Ar/36Ar as well as 39Ar/40Arvs 36Ar/40Ar isotope correlation diagrams. The agesobtained from these diagrams (not shown) are all indistin-guishable (3072�8·5 and 3071�11Ma, respectively) fromthe plateau age and the total gas age, which is, therefore,regarded as dating the time of hornblende crystallization.The remaining two hornblende separates from the drill

core sections 26·8^26·9m and 30·1^30·2m both yieldedsimilar results with total gas ages of 3108�29 and3101�26 Ma, respectively (Fig. 13c and d, Table 7). Inboth cases, a certain variability in the ages obtained atthe various steps prevents a good plateau from being seenand this is also reflected in large errors in the (compara-ble) ages obtained from 40Ar/36Ar vs 39Ar/36Ar and39Ar/40Ar vs 36Ar/40Ar diagrams (not shown). Very littleevidence of partial Ar loss caused by secondary alterationis indicated for these samples, in accordance with petro-graphic observations.

DISCUSS IONAge and geotectonic setting of theanalysed basement rocksThe best constraints on the time of magmatic crystalliza-tion of the investigated granite and hornblende metagab-bro are provided by the concordant U^Pb zircon agedata. They are 3064�20 and 3063�5 Ma, respectively,identical within error (Fig. 11).Younger concordant and dis-cordant U^Pb ages, in particular those obtained from thegranite samples, point to a partial or complete resetting ofthe U^Pb system at 52720 Ma, perhaps by a pervasivealteration process. At this point it is worth noting that thisalteration process left the original Hf isotope compositionunchanged, as is reflected in the within-error identical176Hf/177Hf (c. �1·5 e-units) ratios obtained for nearly allconcordant and discordant zircon domains in the

FRIMMEL et al. WITWATERSRAND HINTERLAND

2209

respective samples (Fig. 12). This clearly indicates that allthe zircon grains or domains formed during the magmaticcrystallization event at c. 3065 Ma, and not during laterprocesses (e.g. during a Neoarchaean or Palaeoproterozoicmetamorphic or metasomatic overprint). Clues about the

age of such an overprint are, apart from the youngerU^Pb zircon ages (Fig. 11), provided by the Rb^Sr isotopewhole-rock data, which yield an errorchron age of c. 2633Ma. However, the geological significance of this Rb^Srage, which is predominantly constrained by the spread of

Fig. 13. 40Ar^39Ar incremental release spectra obtained for hornblende separates from different positions within the intersected hornblendemetagabbro.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2210

Table 7: 40Ar^39Ar analytical data for incremental heating experiments on hornblende

Step T (8C) 39Ar (%) 40Ar1 (mV) Rad (%) 39Ar/37Ar 36Ca (%) 40Ar1/39Ar � (%) Age (Ma) � (Ma)

Sample 15·1–15·5 m

1 800 4·3 478·71 97·9 0·631 1·56 302·58 0·9 2939·3 12·6

2 880 1·4 143·79 97·6 0·147 6·43 272·16 2·8 2787·3 40·1

3 950 9·6 1318·70 99·5 0·055 38·07 376·53 1·0 3263·0 15·2

4 980 34·8 4302·12 99·7 0·074 47·99 338·86 0·6 3105·4 9·4

5 1000 9·8 1116·66 99·7 0·079 51·34 313·06 1·0 2988·9 14·4

6 1015 4·9 556·58 99·8 0·083 52·70 308·58 0·7 2967·9 10·8

7 1045 6·5 730·72 99·9 0·097 64·59 306·97 0·7 2960·3 10·6

8 1095 3·3 374·96 99·8 0·090 55·77 311·53 0·6 2981·8 9·1

9 1160 9·6 1137·33 99·7 0·088 48·24 323·11 0·6 3035·2 8·5

10 1300 15·6 1869·61 99·6 0·079 41·29 329·39 0·6 3063·5 8·7

Total gas age 3063·7 18·91Measured; correcting factors: Daly/HF ¼ 9·17� 2·0%, 39Ca/37Ca ¼ 0·00039, 36Ca/37Ca ¼ 0·00022

Step T (8C) 39Ar (%) 40Ar1 (mV) Rad (%) 39Ar/37Ar 36Ca (%) 40Ar1/39Ar � (%) Age (Ma) � (Ma)

Sample 19·3, 19·4 m

1 720 2·0 236·88 97·5 0·689 1·27 284·63 1·0 2851·2 13·9

2 760 1·3 142·03 97·7 1·195 0·88 264·65 1·7 2747·7 24·1

3 820 5·7 645·60 99·3 1·420 2·52 266·52 0·6 2757·7 8·9

4 850 1·2 142·48 98·0 0·366 3·09 270·54 1·5 2778·8 20·6

5 880 7·7 1081·27 99·2 0·093 21·20 332·69 0·7 3078·2 10·1

6 920 5·6 779·96 99·5 0·073 35·60 329·46 1·0 3063·8 14·4

7 950 4·7 654·01 99·1 0·071 23·12 330·10 1·2 3066·7 17·5

8 1000 35·9 5055·51 99·8 0·069 59·57 333·53 0·7 3081·9 10·9

9 1050 5·6 782·30 99·9 0·061 70·42 332·09 0·5 3075·5 7·6

10 1095 3·3 465·94 99·9 0·063 74·77 333·48 0·6 3081·7 8·4

11 1200 2·2 310·11 99·2 0·066 28·42 332·71 1·5 3078·3 22·6

12 1300 24·8 3742·31 99·7 0·057 55·23 357·17 1·2 3183·7 17·3

Total gas age 3073·7 21·3

65% plateau age 3078·1 19·61Measured; correcting factors: Daly/HF ¼ 9·10� 2·0%, 39Ca/37Ca ¼ 0·00039, 36Ca/37Ca ¼ 0·00022

Step T (8C) 39Ar (%) 40Ar1 (mV) Rad (%) 39Ar/37Ar 36Ca (%) 40Ar1/39Ar � (%) Age (Ma) � (Ma)

Sample 26·8–26·9 m

1 800 5·5 281·94 97·3 0·262 2·98 295·721 1·0 2906·2 15·1

2 880 9·5 414·35 99·1 0·778 3·47 249·922 1·3 2667·5 18·1

3 950 10·9 598·29 99·0 0·073 21·79 316·749 0·9 3006·0 13·5

4 980 18·4 1194·83 99·7 0·057 50·42 376·066 1·3 3261·1 19·1

5 1000 19·6 1164·44 99·8 0·060 59·96 342·593 1·3 3121·6 19·4

6 1015 9·9 508·26 99·7 0·072 53·80 296·435 1·4 2909·7 20·9

7 1045 10·3 656·67 99·8 0·060 60·77 369·105 0·8 3233·0 12·5

8 1095 4·5 343·18 99·1 0·050 25·70 444·576 2·5 3516·8 38·6

9 1300 11·5 764·91 99·9 0·058 75·45 384·311 1·7 3293·9 26·2

Total gas age 3107·5 28·81Measured; correcting factors: Daly/HF ¼ 8·91� 5·0%, 39Ca/37Ca ¼ 0·00039, 36Ca/37Ca ¼ 0·00022

Step T (8C) 39Ar (%) 40Ar1 (mV) Rad (%) 39Ar/37Ar 36Ca (%) 40Ar1/39Ar � (%) Age (Ma) � (Ma)

Sample 30·1–30·2 m

1 800 7·1 831·15 98·63477 0·734 1·819598 319·415 0·7 3018·3 10·7

2 880 1·7 207·82 98·46037 0·144 7·334663 336·103 1·2 3093·3 18·4

3 950 14·6 1968·35 99·4128 0·066 29·60979 370·579 1·3 3239·0 19·6

4 980 16·3 1953·70 99·60721 0·077 37·92846 329·466 1·7 3063·8 24·3

5 1000 9·7 1086·63 99·49839 0·080 33·00942 306·726 1·3 2959·1 19·3

6 1015 4·1 468·07 99·20482 0·081 22·85102 313·544 1·5 2991·2 21·6

7 1045 3·7 412·32 99·13203 0·092 19·73829 304·459 2·6 2948·3 38·4

8 1095 4·6 533·32 98·68499 0·092 13·46074 316·918 1·3 3006·8 19·3

9 1160 13·7 1805·51 99·77105 0·072 50·79623 361·611 0·8 3202·2 12·4

10 1300 24·6 3087·86 99·70786 0·071 46·17452 344·218 0·8 3128·6 11·5

Total gas age 3101·4 26·11Measured; correcting factors: Daly/HF ¼ 8·91� 5·0, 39Ca/37Ca ¼ 0·00039, 36Ca/37Ca ¼ 0·00020

FRIMMEL et al. WITWATERSRAND HINTERLAND

2211

the data from the granite sample, is not entirely clear.Judging from the petrographic observations (formation ofsecondary microcline, even in the metagabbro, and serici-tization), it appears likely that the Rb^Sr errorchon ageof 2633 Ma reflects the time of retrograde hydrothermalinfiltration and alteration. In this context, the somewhatolder ‘concordant’ U^Pb zircon age of c. 2720 Ma could beexplained as a maximum age that results from incompletePb loss from altered (metamict), primary magmaticzircon domains at 2·63 Ga.Surprisingly, and in contrast to the Rb^Sr isotope data,

a pervasive Neoarchaean hydrothermal alteration event isnot indicated by the Ar^Ar data obtained on the hornble-nde from the metagabbro. As shown in Fig. 13, the hornble-nde separates yielded total fusion Ar^Ar ages, which areeither within error of the zircon crystallization ages(3064�19 Ma and 3078�20 Ma) or slightly older(3108�29 Ma and 3101�26 Ma), and provide little evi-dence for a Neoarchaean or younger overprint. This find-ing points clearly to hornblende formation at, or close to,the time of magma crystallization. Judging from the verylow Ti content of the hornblende in the metagabbro, andthe fact that it occurs together with other Ti-bearing(buffer)-phases, it seems most likely that the magnesio-hornblende was not formed during magmatic crystalliza-tion, but rather resulted from post-magmatic, solid-statealteration either of primary magmatic (Ti-rich) hornble-nde or of other mafic magmatic phases (such as pyroxenes)immediately after magmatic crystallization. Possibly, thistransformation was caused by pervasive autometasoma-tism. Such an explanation conforms to the U^Pb and Ar^Ar age data, but also to the very narrow range in Rb/Srand 87Sr/86Sr in the five hornblende metagabbro samplesanalysed (Fig. 9, Table 4). Such an interpretation finds fur-ther support from recent O isotope data obtained onwhole-rocks and mineral separates from the investigatedcore samples (M. Depine¤ et al., unpublished data, 2009).The slightly older 40Ar^39Ar ages obtained for some

hornblende samples, or some steps, might result from theincorporation of excess Ar during secondary alteration ofhornblende to actinolite and/or chlorite in the course of alater (Neoarchaean) low-grade metamorphic and/orhydrothermal overprint. Alternatively, the excess Ar maybe due to infiltration and reaction of late-magmatic fluidsthat caused the autometasomatism suggested above. SuchAr-enriched fluids were perhaps released during crystalli-zation of the associated granites, which formed nearly con-temporaneously with the hornblende gabbro, as suggestedby the within error-identical U^Pb zircon ages and Hf iso-tope data. In fact, this late- to post-magmatic infiltrationprocess, which is related either to granite emplacement orto another unexposed source, could also account for theformation of the biotite-rich alteration zone between thehornblende metagabbro and the granite, and the presence

of microcline throughout the metagabbro. This microcline,although only in minor proportions, is unusual andcannot be explained by fractional crystallization becauseof the low density of K-feldspar and its relatively late crys-tallization. As the same type of microcline is also found inlocally very coarse-grained patches within the granite andconcentrated along veins in the metagabbro, it is mostprobably related to the infiltration of a potassic fluidçpossibly during a stage of pegmatite formation evidentfrom surface outcrops (Fig. 3).Pegmatite formation was followed by a further retro-

grade, hydrothermal overprint as indicated by the chloriti-zation of biotite, sericitization and saussuritization ofmicrocline and plagioclase, respectively, and the formationof actinolite at the expense of magnesio-hornblende andthe various veinlets, especially in the hornblende metagab-bro. It appears likely that this final hydrothermal over-print was responsible for the resetting of the Rb^Srisotope system, in particular of the granite, as reflected bythe Rb^Sr whole-rock errorchron age of c. 2630 Ma.The reconstruction of the tectonic setting of the investi-

gated magmatic rocks is hindered by their complex post-magmatic alteration. Some elements, such as the alkali ele-ments, are certain to have been mobile. For instance, the(Na þ K) concentrations shown in Fig. 4a are likely to behigher than those of the magmatic protoliths because ofthe later K-metasomatism. Consequently, the protolithswere even less alkaline than shown in that diagram andconform to the calc-alkaline series, as is also indicated bythe relatively small proportion of total Fe relative to Mgat variable SiO2 concentrations (Fig. 4c). Despite the pro-blems of some major element mobility, trends of petroge-netic significance can be derived from certain less mobiletrace element distributions.Almost all geochemical indicators point to a calc-

alkaline composition and crustal contamination typical ofcontinental arc magmatism. This includes, for example,the Nb^Yand Ta^Yb relationships in the granite (Fig. 6),the overall enrichment of the metagabbro in the less com-patible elements (e.g. Th), its relative depletion in Nb andits strong depletion in Ti relative to MORB (Fig. 7).Variable crustal contamination is also reflected by LREEenrichment and by chondritic to slightly subchondriticeNd 3·06Ga (mostly between þ1·9 and �1·7) and eHf 3·06Ga

values (þ1·7 to �2·3; Fig. 10, Table 6). Notably, the eHft

values are identical within error to those recently obtainedfor similarly old rocks from other parts of the KaapvaalCraton by Zeh et al. (2009), comprising 3·1 Ga graniticrocks from the Swaziland, Witwatersrand and Pietersburgblocks (Fig. 1). The very low eNd 3·06Ga values of �4·8 and�10·4 obtained from two samples (Fig. 10) result eitherfrom post-intrusive alteration (e.g. Moorbath et al., 1997)or from the assimilation of much older crust, recycledduring the intrusion of the granitic and gabbroic

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2212

protoliths. Notably, similarly strong variations are not rec-orded by the Hf isotope data.Following Pearce (2008), we chose Yb-normalized Th

and Nb as proxies for crustal input. BothTh and Nb havea similar geochemical behaviour in most petrogenetic pro-cesses and both are relatively immobile under conditionsthat range from weathering to medium-grade metamor-phism. With regard to the Th^Nb^Yb relationships(Fig. 14a), the hornblende metagabbro samples plot in aposition similar to enriched (E-)MORB with variabledegrees of crustal contamination. In a similar way,Condie (2005) described the extent of crustal contamina-tion versus mantle composition in terms of Nb/Th vs Zr/Nb and distinguished between plume and non-plumesources based on the Nb/Y vs Zr/Y relationships. Aspointed out by Pearce (2008), the application of the latterdiagram is problematic because of depth-dependentgarnet fractionation.Thus the apparent plume source indi-cated in Fig. 14c may not be real. The low Nb/Th is inagreement with a volcanic arc setting, but the Zr/Nb islower than expected for such a setting (Fig. 14b).The Cs/Nb and U/Nb ratios are supposed to be insensi-

tive to magmatic differentiation by fractional crystalliza-tion and are therefore also used to assess the extent ofcrustal contamination (Fig. 15a). As can be seen from thatfigure, most samples plot away from primitive mantleratios, thus testifying to considerable crustal contamina-tion in both the mafic and felsic portions. One trend pro-jects to particularly high U/Nb ratios at reasonable crustalCs/Nb, whereas one sample plots far away from thattrend, having a particularly high Cs/Nb ratio. The latteris rich in microcline and probably reflects a geochemicalchange during K-metasomatism. The U/Nb ratios arehigh for most samples, even for the mafic rocks, whichshow, on average, higher U/Nb ratios than typical of conti-nental crust (Fig. 15b). Again, considerable crustal contam-ination can be inferred from this observation.The enrichment in U is also evident in Fig. 15b, in which

the U/Nb ratio is plotted against Ce/Pb. The latter ratio isexplicable in terms of extensive feldspar fractionation inthe case of the granite and hornblende metagabbro, andconforms to the observed negative Eu anomalies (Fig. 8).This effect is most pronounced in the biotite-rich zone,because of the elevated, probably hydrothermal, allanitecontent therein.

Formation and alteration of the basementnext to the Witwatersrand BasinThe U^Pb crystallization age of 3062�5 Ma for the plu-tonic rocks adjacent to the ancient Witwatersrand Basininvariably invites a comparison to be made with the ageof Dominion Group volcanism. The volcano-sedimentaryDominion Group underlies the WitwatersrandSupergroup rocks over probably a much larger area thanindicated by the remnants of that group known from

surface and subsurface exposures mainly along the north-western margin of the Witwatersrand Basin (Fig. 2). Theprecise U^Pb single zircon age of 3074� 6 Ma obtainedfor a quartz^feldspar porphyry in the upper part of theDominion Group (Armstrong et al., 1991) is only slightlyolder than the crystallization age obtained in this study.

Fig. 14. (a) Th/Yb vs Nb/Yb plot for the hornblende metagabbrosamples and the Dominion Group mafic volcanic rocks, utilizing theTh^Nb proxy for crustal contamination as suggested by Pearce(2008); (b) Zr/Nb vs Nb/Th, and (c) Nb/Yvs Zr/Y plots for the samesamples. Compositional fields for different tectonic settings and effectsof batch melting (BM) and subduction (SUB), as indicated byarrows, are from Condie (2005); PM, primitive mantle; DM, depletedmantle.

FRIMMEL et al. WITWATERSRAND HINTERLAND

2213

This raises the possibility that the intrusive rocks of thisstudy are the plutonic expression of the same type of mag-matism that formed most of the Dominion Group. A conti-nental rift setting has been generally assumed for theDominion Group (e.g. Clendenin et al., 1988; Marsh et al.,1989; Jackson, 1992) because of the bimodal nature of thevolcanic suite and the tholeiitic affinity of the mafic rocks.The evidence is, however, not unequivocal. Geochemicaldata for the Dominion Group volcanic rocks were

presented by Marsh et al. (1989) and Jackson (1994).The bimodal compositional distribution encompassesrocks with a range of SiO2 concentrations between 51 and61%, with andesitic compositions dominating; however,there is a gap between 61 and 67% SiO2. A bimodal SiO2

distribution has been noted in many Archaean arcs andexplained by shallow subduction caused by a highermantle heat flow resulting in thicker oceanic crust(Abbott & Hofmann, 1984). According to Marsh et al.

Fig. 15. Cs/Nb vs U/Nb (a) and U/Nb vs Ce/Pb (b) diagrams for the various lithotypes in the investigated drill core. The fields or points foroceanic island basalt (OIB), mid-oceanic ridge basalt (MORB), primitive mantle (PM), lower (LC), middle (MC) and upper continentalcrust (UC) are from Sun & McDonough (1989) and Rudnick & Fountain (1995). Also shown are fractional crystallization trends for amphibole(Amph), clinopyroxene (Cpx), plagioclase (Plag) and alkali feldspar (K-fsp), each arrow representing 50% fractional crystallization.The influ-ence of fractional crystallization on the scale of Fig. 15a is negligible.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2214

(1989) the Dominion Group mafic lavas are enriched in theless compatible trace elements shown in Fig. 7 in relationto MORB but they are depleted in Nb relative to Th, La,and Ce (no data forTa available). Moreover, they are char-acterized by depletion inTi relative to MORB and overallshow a trace element pattern that resembles that of typicalcalc-alkaline basalts. The geochemistry of the DominionGroup volcanic rocks could thus be easily explained by avolcanic arc setting. Interestingly, Burke et al. (1986) pre-viously suggested that the Dominion Group volcanic rockserupted on the landward flank of an Andean-type conti-nental margin.Closer inspection of the dataset of Marsh et al. (1989)

reveals, however, that there are certain differences fromthe geochemical characteristics of the intrusive rocks stud-ied by us. For example, the Nb and Y concentrations oftheir silicic rocks, the quartz^feldspar porphyries, corre-spond to those of volcanic arcs as well as syn-collisionalgranites but their Nb/Y ratios are slightly lower than forthe granites investigated in this study (Fig. 6a). In termsof Th^Nb^Yb relationships, the Dominion Group volcanicrocks follow a trend from N- to E-MORB but lack evidenceof major mantle^crust interaction (Fig. 14a). Minor crustalcontamination might be indicated by their calculated Ndmodel ages (mean age 3233�53 Ma, recalculated fromthe data of Marsh et al., 1989), which are slightly olderthan the age of emplacement. This is, however, subject tothe previously mentioned uncertainty in the projection ofthe Archaean depleted mantle composition in general andthat below the Kaapvaal Craton in particular. TheDominion Group rocks exhibit very different Zr^Nb^Yrelationships that are more akin to typical volcanic arccompositions, except for a higher Nb/Th ratio (Fig. 14band c). It is, therefore, assumed that the intrusive rocks ofthis study are not co-genetic with the Dominion Grouplavas and might be slightly younger. In particular, theabsence of evidence of significant crustal contamination inthe Dominion Group rocks might indicate that theyformed earlier, in a less mature arc above a thinner crust,whereas the 3062�5 Ma intrusive rocks could reflect amore mature stage of arc development.Apart from our Nd and Hf isotope and Nb^Th^Yb^

REE data, crustal contamination at 3·06 Ga seems likelyin view of the presence of older crust in the neighbouringbasement along the northern and northwestern margin oftheWitwatersrand Basin and in theVredefort Dome (for asummary of geochronological data see Poujol et al., 2003;Armstrong et al., 2006). The oldest reported zircon agesare 3480�7 Ma from xenocrysts in a quartz porphyry ofthe upperVentersdorp Supergroup (Armstrong et al., 1991)and c. 3245 Ma from paragneiss in the Vredefort Dome(Hart et al., 1999), respectively; these ages are similar tomany of the Nd and Hf model ages obtained during thisstudy (Table 6). Following trondhjemite^tonalite

emplacement (the most precise age is 3340�3 Ma, Poujol& Anhaeusser, 2001) in the Johannesburg Dome, the base-ment along the northern basin margin was affected bylarge-scale granodiorite emplacement with U^Pb singlezircon ages of 3121�5 Ma (southern JohannesburgDome), 3120�5 Ma (200 km SW of Johannesburg) and3114�2 Ma (southwestern Johannesburg Dome, Poujol &Anhaeusser, 2001). A slightly younger U^Pb zircon age of3031þ11�10 Ma was obtained on a granodiorite near Coligny(Robb et al., 1992). Our new data are comparable with pre-viously obtained Rb^Sr, Pb^Pb and Sm^Nd data obtainedon similar granitoids from the Johannesburg Dome.Granodiorite and granite from that area yielded Rb^Srand Pb^Pb isochron ages of 3081�33 and 3062�26 Ma,respectively, and a Pb^Pb zircon age of 3093�3 Ma(Barton et al., 1999). Significantly older Nd model agesobtained in the same study indicate involvement of oldercrustal components. Based on geochemical data for thesecalc-alkaline granitoids in the Johannesburg Dome,Anhaeusser (1999) concluded that these rocks formed in avolcanic arc setting.The age obtained here for the drilled granite and horn-

blende metagabbro (3062�5 Ma) is identical within errorto that reported by Armstrong et al. (2006) for an aplitedyke in theVredefort Dome (3068�6 Ma).The volumetri-cally minor aplite emplacement in the Vredefort Dome isthe last of three magmatic episodes there, following tona-lite^trondhjemite^granodiorite emplacement probably inan oceanic arc at 3·1 Ga and subsequent high-grade meta-morphism and granite^granodiorite emplacementbetween 3·1 and 3·08 Ga (Armstrong et al., 2006). Thelatter episode has been interpreted by Armstrong et al. toreflect crustal thickening in response to accretion of theVredefort rocks onto an older core of the Kaapvaal craton.Our new results reinforce the idea of Armstrong et al.

(2006) that the central Kaapvaal Craton, which laterbecame covered in places by theWitwatersrand sediments,had not already consolidated into a stable craton by 3·2Ga, as suggested by de Wit et al. (1992) and subsequentlyadopted by many workers, but continued to grow throughthe accretion of magmatic arcs until about 3·06 Ga. Onthe basis of available data so far, all potential source rocksfor the proximal Witwatersrand sediments seem to be of amagmatic arc affinity. The exact geometry of these mag-matic arc systems remains uncertain, but an east-northeasterly trend along the northern margin of theWitwatersrand Block can be distinguished from a north-westerly trend in the Vredefort Dome and in the westernpart of the craton. It may be speculated that the volumi-nous potassic granites of similar age in the Barberton^Swaziland region (Poujol et al., 2003; Zeh et al., 2009),formed in response to extensive crustal heating within thecontinent behind an arc system. This conclusion is in goodagreement with Hf isotope data recently obtained from

FRIMMEL et al. WITWATERSRAND HINTERLAND

2215

granitoid rocks in the eastern part of the Kaapvaal Craton(Zeh et al., 2009). These data indicate that at c. 3·1 Ga newcrust was added to the pre-existing basement of theWitwatersrand block, perhaps in response to roughlysouthward subduction during the accretion of thePietersburg block onto theWitwatersrand block.In contrast to most previous models, we prefer an intra-

arc position for the Dominion Group. This is notwith-standing the bimodal nature of the volcanism (similar tothe bimodal character of the investigated drill core). Inthe light of mounting evidence for andesite magmas beingformed by mixing between evolved, silicic melts and basicplutonic root components (e.g. Reubi & Blundy, 2008), themere presence or absence of a bimodal magmatic suitemight not be a reliable indicator of a specific tectonic set-ting. Instead, intra-arc extension might have provided suit-able pathways for the ascent of felsic and mafic melts.The retrograde, hydrothermal alteration of the various

basement rocks in the studied drill core is similar to thatpreviously described for basement granites along thenorthern and northwestern margins of theWitwatersrandBasin (Klemd & Hallbauer, 1987; Robb & Meyer, 1987).As already pointed out by Klemd (1999), the similar fluidinclusion characteristics of the hydrothermal alteration ofthe basement granites and of theWitwatersrand Basin fillare suggestive of a relationship. The situation is, however,complicated because of the multistage alteration historyrecorded by the Witwatersrand metasedimentary rocks(Frimmel et al., 2005).This ranges from diagenetic dewater-ing to regional low-grade metamorphism, a thermal over-print by the Bushveld event, to brittle deformation andfurther hydrothermal alteration triggered by theVredefortimpact.Our zircon U^Pb and Rb^Sr isotope data are consistent

with previous interpretations that the basement below theWitwatersrand Basin was affected by several alterationevents between 2720 and 2630 Ga (Figs 6 and 11). Forexample, Kositcin et al. (2003) recognized, based on U^Pbsensitive high-resolution ion microprobe (SHRIMP) dataon different hydrothermal xenotime generations, at leastthree stages of hydrothermal fluid infiltration. The oldestof these (2720 Ma) is close to the time of the outpouring ofthe voluminous Klipriviersberg lavas (lower VentersdorpSupergroup) and is identical to the older alteration agesuggested above for the pre-Witwatersrand basement.Considerable heating affected the Kaapvaal crust at thattime, which is also evident from the contemporaneousultrahigh-temperature metamorphism in the lower partsof that crust (Schmitz & Bowring, 2003).The cause of the 2714� 8 Ma Klipriviersberg volcanism

(Armstrong et al., 1991) remains a matter of debate, withboth a mantle plume and/or crustal thinning having beenheld responsible for it. More recently, Silver et al. (2006)suggested flood basalt eruption in a collisional rift. Their

model involves short-term drainage of a molten basalt res-ervoir in the sublithospheric mantle during a change inthe stress field in an overall collisional setting. In thiscase, the collision would be an early stage of amalgamationof the Kaapvaal Craton with crustal fragments now pres-ent in the Central Zone of the Limpopo Belt (see Zehet al., 2009). A northward thrusting event that affected thegreenstone belts along the northern flank of the craton,dated at 2729�19 Ma (Passeraub et al., 1999), is likely tobe an expression of this collision. Furthermore, severalgranites of comparable ages (Robb et al., 1992) in theimmediate vicinity of the Witwatersrand may be relatedto the same tectonic stage.The maximum temperature experienced by the

Witwatersrand Basin fill was attained at different times indifferent parts of the basin (Frimmel et al., 2005). At leastin the upper parts of the basin fill it did not exceed lowergreenschist-facies temperatures (350�508C; Frimmel,1994; Phillips & Law, 1994). Along the northern margin ofthe Witwatersrand Basin, peak metamorphic conditionswere already reached by the end of VentersdorpSupergroup deposition, as indicated by kyanite- and pyro-phyllite-bearing post-Platberg and pre-Transvaal thrustfaults (Coetzee et al., 1995). The inferred younger alterationof the basement rocks investigated in this study could wellbe an expression of this collisional event in the LimpopoBelt because the obtained Rb^Sr errorchron ‘age’ of2633�50 Ma overlaps with the published ages of syntec-tonic granites in the Limpopo Belt (McCourt &Armstrong, 1998; Kro« ner et al., 1999; Zeh et al., 2007, 2009;Millonig et al., 2008). In addition, the high initial 87Sr/86Srshown in Fig. 9 points to the involvement of crustal fluids.Further hydrothermal xenotime growth was noted by

Kositcin et al. (2003) at 2210 Ma (Pretoria Group exten-sion) and again at 2046^2061 Ma, the time of theBushveld event. These younger events do not seem to havefurther disturbed significantly the isotopic composition ofthe investigated basement rocks and zircon grains therein.Similarly, the 2023 Ma Vredefort impact event, whichundoubtedly triggered renewed fluid circulation throughthe Witwatersrand Basin and its surroundings (Frimmelet al., 1999), did not affect the isotope systems of the studiedrocks, analogous to a previous finding by Barton et al.(1999) in the Johannesburg Dome.

Significance for Witwatersrand gold genesisMounting evidence exists for plate-tectonic processeshaving already been operative in Mesoarchaean times(e.g. de Wit, 1998; Moyen et al., 2006; Zeh et al., 2009).Thus comparison with the gold productivity in variouspost-Archaean plate-tectonic settings seems justified. Thevast majority (�87%) of known primary gold deposits(i.e. excluding secondary, placer deposits), appear to havebeen formed along active continental margins, where theyare present as orogenic (including intrusion-related),

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2216

Cu^Au porphyry and epithermal types (Frimmel, 2008).There is no doubt that active continental margins provideby far the best sites for the concentration of Au into orebodies and there is no reason why this should have beenany different in the Archaean.Of significance is not only our conclusion that at least

some of the auriferous Witwatersrand sediments couldhave been sourced in an active continental margin but par-ticularly the finding of ‘autometasomatized’ hornblendegabbro (or diorite) in that source area. The REE patternsof the studied mafic rocks are identical to those that aretypically explained by the fractionation of middle andheavy REE-enriched hornblende. The abundance of horn-blende in these rocks makes this a highly feasible explana-tion. A high magmatic oxidation state is unlikely to havebeen the reason for the noted lack of a pronounced nega-tive Eu anomaly in the hornblende metagabbros becauseof the Mesoarchaean age of the system. This could also bedue to the suppression of feldspar fractionation. Early pla-gioclase crystallization is suppressed when a magma con-tains sufficient H2O to stabilize hornblende as an earlyliquidus phase. By analogy with studies such as those ofRutherford & Devine (1993) and Richards et al. (2001) anH2O content of44wt % is required for the crystallizationof near-liquidus hornblende. Such an elevated H2O contentin the melt is a critical and typical ingredient of a numberof productive magmatic^hydrothermal ore-forming sys-tems worldwide, such as porphyry copper (^gold) or ironoxide^copper^gold (IOCG) deposits. Consequently, theinvestigated basement rocks contain the essential prerequi-sites for particularly high fertility in terms of primarymagmatic^hydrothermal gold mineralization if the proto-lith of the studied mafic rocks was indeed a hornblendegabbro. In the light of a total absence of any pyroxenerelics anywhere in the drilled hornblende metagabbro thisis our preferred interpretation.Derivation of theWitwatersrand placer gold from an Au-

enriched continental volcanic arc also would explain thepreviously noted wide range of generally orders of magni-tude higher Os concentrations in the Witwatersrand goldcompared with any other type of gold investigated sofar, as noted by Kirk et al. (2002) and Frimmel et al.(2005). It has been suggested by these workers that theWitwatersrand gold was originally magmatic andextracted from a 3·1^3·3 Ga mantle source because of aninitial 187Os/188Os value of 0·108 that corresponds to theprojected depleted mantle composition at that time. TheseOs-depleted mantle extraction ages agree well with theNd and Hf model ages obtained during this study. Thewide range in Os concentrations would be explicable by amixture of sources, including disseminated porphyry-stylegold, intrusion-related, IOCG, epithermal or orogenicgold. The extent to which the gold was transported byhydrothermal fluids (as opposed to melts), which is vastly

different between the above sources, would control the Osconcentration of a particular type of gold because of theextremely low solubility of Os in aqueous fluids.Evidence of mesothermal (orogenic), potentially gold-

bearing, quartz veins in the source area exists in the formof O isotope data on quartz pebbles in the host conglomer-ate (Vennemann et al., 1992,1995). Such veins are, however,an unlikely major contributor to the Witwatersrand goldbudget. The required abundance of typical Archaeanmesothermal lode gold deposits in the source area toexplain about 40% of all known gold would have to beunrealistically high. Hallbauer & Barton (1987) previouslysuggested that Archaean greenstone-hosted gold fromquartz veins, as mined in the Barberton greenstone belt,could not have been a major source of theWitwatersrandgold because of compositional differences. Instead, theyproposed that altered granites with as much as 80 ppb Auin the hinterland could be a more likely source.A magmatic^hydrothermal origin of theWitwatersrand

gold solves the mass-balance problem and also explainsthe absence of vein quartz pebbles with visible gold inclu-sions in the Witwatersrand goldfields. It also explains theoverall very small size of theWitwatersrand gold particles(see Minter et al., 1993) and the lack of reasonably sizedgold nuggets (although the latter could simply be due to alack of an oxidizing atmosphere).The comparison with magmatic^hydrothermal minera-

lizing systems, such as porphyry Cu^Au or IOCG systems,makes it tempting to interpret the earlier stages of theobserved hydrothermal alteration as being part of such amineralizing system. Late-stage magmatic alteration isadditionally indicated by our combined datasets for thehornblende metagabbro, for which a late-magmatic stageof ‘hydrothermal’ autometasomatism immediately afteremplacement at c. 3065 Ma is suggested by the very similarAr^Ar ages of very low-Ti hornblende. In contrast, theNeoarchaean hydrothermal alteration at c. 2720 and 2630Ma, as indicated by our Rb^Sr and U^Pb age data,might have nothing to do with enrichment in Au in thecentral Kaapvaal crust, because the bulk, if not all, of thegold in the Witwatersrand Supergroup had already beendeposited within the largely conglomeratic host rocksprior to that time. The only effect this hydrothermal alter-ation had on the gold was that of local, short-range mobili-zation and dispersion within the host conglomerates.

CONCLUSIONSNew lithogeochemical, isotopic and geochronological datafor hornblende metagabbro and granite from a basementhorst at the northwestern margin of the WitwatersrandBasin provide new insights into the nature of some pre-Witwatersrand units that could have been potential sourcerocks for some of the auriferous conglomerates particularlyin the Central Rand Basin. Both the hornblende

FRIMMEL et al. WITWATERSRAND HINTERLAND

2217

metagabbro and the granite yielded indistinguishable U^Pb zircon ages of 3062�5 Ma. It is concluded from wide-spread K-metasomatism and the alteration of presumablyprimary Ti-rich hornblende to Ti-poor magnesio-hornblende at effectively the same time that the graniteintruded into the gabbro shortly after the latter’s crystalli-zation. Both rock types have geochemical signatures thatare typical of calc-alkaline magmatism. This, combinedwith evidence for contamination by older crustal compo-nents based on Nd and Hf model ages for whole-rocksand zircon grains, respectively, leads to the conclusionthat a continental volcanic arc is the most likely magmaticsetting. Our new results support previous suggestions ofsuccessive arc accretion onto the northern and westernmargins of the Witwatersrand block prior to the develop-ment of the various basins that constitute theWitwatersrand ‘successor’ basin.With a previously published age of the Witwatersrand

gold (and associated detrital pyrite) that is within error ofthe age obtained for the pre-Witwatersrand units of thisstudy and the proximity of the studied rocks to the sites ofdeposition of the auriferous conglomerates, the investi-gated rocks could well be a potential source for some ofthe Witwatersrand gold. Inferred primary hornblendegabbro provides evidence for ascending water-rich meltsthat might have been particularly conducive for the trans-fer of Au into the crust. Gold enrichment in the Palaeo- toMesoarchaean hinterland, similar to that found in youngeractive plate margins, might have been a major controllingfactor for the unique extent of gold enrichment in theWitwatersrand Basin fill.

ACKNOWLEDGEMENTSA. Gerdes is thanked for providing the infrastructure forthe single zircon analyses. S. Govender assisted with theNd isotope analyses, and H. Bra« tz and U. Schu« �lerhelped with respectively the LA-ICPMS and XRF ana-lyses. The Rb^Sr isotope analyses were kindly conductedby M. Bro« cker. C. Anhaeusser, J. Barton and an anon-ymous reviewer provided constructive comments thathelped to improve the original manuscript. Financialsupport by the Deutsche Forschungsgemeinschaft(DFG grant FR2183/3-1 and FR2183/3-2) is gratefullyacknowledged. This is a contribution to IGCP 540 (Goldand Fluid Inclusions).

REFERENCESAbbott, D. & Hofmann, S. E. (1984). Archean plate tectonics revisited:

1. Heat flow, spreading rate, and the age of subducting oceaniclithosphere and their effects on the origin and evolution of conti-nents.Tectonics 3, 429^448.

Anhaeusser, C. R. (1973). The geology and geochemistryof the Archaean granites of the Johannesburg^Pretoria Dome.

In: Lister, L. A. (ed.) Symposium on Granites, Gneisses and Related

Rocks. Geological Society of South Africa Special Publication 3, 361^385.Anhaeusser, C. R. (1999). Archaean crustal evolution of the central

Kaapvaal Craton, South Africa: evidence from the JohannesburgDome. South AfricanJournal of Geology 102, 303^322.

Anhaeusser, C. R. (2006). A re-evaluation of Archean intracratonicterrane boundaries on the Kaapvaal Craton, South Africa:Collisional suture zones? In: Reimold, W. U. & Gibson, R. L.(eds) Processes on the Early Earth. Geological Society of America, Special

Papers 405, 193^210.Armstrong, R. A., Compston, W., Retief, E. A., William, L. S. &

Welke, H. J. (1991). Zircon ion microprobe studies bearing on theage and evolution of the Witwatersrand triad. Precambrian Research

53, 243^266.Armstrong, R. A., Lana, C., Reimold,W. U. & Gibson, R. L. (2006).

SHRIMP zircon age constraints on Mesoarchean crustal develop-ment in the Vredefort dome, central Kaapvaal Craton, SouthAfrica. In: Reimold, W. U. & Gibson, R. L. (eds) Processes on the

Early Earth. Geological Society of America, Special Papers 405, 233^253.Barton, J. M. J., Barton, E. S. & Kro« ner, A. (1999). Age and isotopic

evidence for the origin of the Archaean granitoid intrusives of theJohannesburg Dome, South Africa. Journal of African Earth Sciences28, 693^702.

Bowen, T. B., Marsh, J. S., Bowen, M. P. & Eales, H. V. (1986).Volcanic rocks of the Witwatersrand triad, South Africa, I:Description, classification and geochemical stratigraphy.Precambrian Research 31, 297^324.

Brandl, G., Cloete, M. & Anhaeusser, C. R. (2006). Archaean green-stone belts. In: Johnson, M. R., Anhaeusser, C. R. & Thomas, R.J. (eds) The Geology of South Africa. Johannesburg: GeologicalSociety of South Africa and Council for Geoscience, pp. 9^56.

Brink, M. C., Waanders, F. B., Bisschoff, A. A. & Gay, N. C. (2000).The FochThrust^Potchefstroom Fault structural system,Vredefort,South Africa: a model for impact-related tectonic movement overa pre-existing barrier. Journal of African Earth Sciences 30, 99^117.

Burke, K., Kidd, W. S. F. & Kusky, T. M. (1986). Archean forelandbasin tectonics in the Witwatersrand, South Africa. Tectonics 5,439^456.

Clendenin, C. W., Charlesworth, E. G. & Maske, S. (1988). An earlyProterozoic three-stage rift system, Kaapvaal Craton, SouthAfrica.Tectonophysics 15, 73^86.

Coetzee, D. S., van Reenen, D. D. & Roering, C. (1995). Quartz veinformation, metamorphism, and fluid inclusions associated withthrusting and bedding-parallel shear in Witwatersrand quartzites,South Africa. South AfricanJournal of Geology 98, 371^381.

Condie, K. C. (2005). High field strength element ratios in Archeanbasalts: a window to evolving sources of mantle plumes? Lithos 79,491^504.

Coward, M. P., Spencer, R. M. & Spencer, C. E. (1995). Developmentof the Witwatersrand Basin, South Africa. In: Coward, M. P. &Ries, A. C. (eds) Early Precambrian Processes. Geological Society,

London, Special Publications 95, 243^269.Crow, C. & Condie, K. C. (1987). Geochemistry and origin of late

Archaean volcanic rocks from the Rhenosterhoek Formation,Dominion Group, South Africa. Precambrian Research 37, 217^229.

deWit, M. J. (1998). On Archaean granites, greenstones, cratons andtectonics: does the evidence demand a verdict? Precambrian Research91, 181^226.

de Wit, M. J., Roering, C., Hart, R. J., Armstrong, R. A., deRonde, C. E. J., Green, R. W. E., Tredoux, M., Peberdy, E. &Hart, R. A. (1992). Formation of an Archean continent. Nature 357,553^562.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2218

Eglington, B. & Armstrong, R. (2004). The Kaapvaal Craton andadjacent orogens, southern Africa: a geochronological databaseand overview of the geological development of the craton. SouthAfricanJournal of Geology 107, 13^32.

Feather, C. & Koen, G. M. (1975). The mineralogy of theWitwatersrand reefs. Minerals Science and Engineering 7, 189^224.

Frimmel, H. E. (1994). Metamorphism of Witwatersrand gold.Exploration and Mining Geology 3, 357^370.

Frimmel, H. E. (2008). Earth’s continental crustal gold endowment.Earth and Planetary Science Letters 267, 45^55.

Frimmel, H. E. & Frank, W. (1998). Neoproterozoic tectono-thermalevolution of the Gariep Belt and its basement, Namibia/SouthAfrica. Precambrian Research 90, 1^28.

Frimmel, H. E., Hallbauer, D. K. & Gartz,V. H. (1999). Gold mobiliz-ing fluids in the Witwatersrand Basin: composition and possiblesources. Mineralogy and Petrology 66, 55^81.

Frimmel, H. E., Groves, D. I., Kirk, J., Ruiz, J., Chesley, J. &Minter, W. E. L. (2005). The formation and preservation of theWitwatersrand goldfields, the largest gold province in the world.In: Hedenquist, J. W., Thompson, J. F. H., Goldfarb, R. J. &Richards, J. P. (eds) Economic Geology One Hundreth Anniversary

Volume. Littleton, Colorado: Society of Economic Geologists,pp. 769^797.

Gerdes, A. & Zeh, A. (2006). Combined U^Pb and Hf isotope LA-(MC)ICP-MS analyses of detrital zircons: comparison withSHRIMP and new constraints for the provenance and age of anArmorican metasediment in Central Germany. Earth and PlanetaryScience Letters 249, 47^61.

Gerdes, A. & Zeh, A. (2008). Zircon formation versus zircon altera-tionçnew insights from combined U^Pb and Lu^Hf in-situ LA-ICP-MS analyses, and consequences for the interpretation ofArchean zircon from the Limpopo Belt. Chemical Geology 261,230^243.

Hallbauer, D. K. & Barton, J. M. (1987). The fossil gold placers of theWitwatersrand. Gold Bulletin 20, 68^79.

Hart, R. J., Andreoli, M. A. G., Tredoux, M. & deWit, M. J. (1990).Geochemistry across an exposed section of Archaean crust atVredefort, South Africa: with implications for mid-crustal disconti-nuities. Chemical Geology 82, 21^50.

Hart, R. J., Moser, D. E. & Andreoli, M. (1999). Archean age forgranulite facies metamorphism near the center of the Vredefortstructure, South Africa. Geology 27, 1091^1094.

Henry, P., Stevenson, R. K., Larbi,Y. & Gariepy, C. (2000). Nd isoto-pic evidence for Early to Late Archean (3·4^2·7 Ga) crustalgrowth in the Western Superior Province (Ontario, Canada).Tectonophysics 322, 135^151.

Hofmann, A. W. (1988). Chemical differentiation of the Earth: therelationship between mantle, continental crust, and oceanic crust.Earth and Planetary Science Letters 90, 297^314.

Holland, T. & Blundy, J. (1994). Non-ideal interactions in calcicamphiboles and their bearing on amphibole^plagioclase thermo-metry. Contributions to Mineralogy and Petrology 116, 433^447.

Jackson, M. C. (1992). A review of the Late Archaean volcano-sedi-mentary Dominion Group and implications for the tectonic settingof the Witwatersrand Supergroup, South Africa. Journal of AfricanEarth Sciences 15, 169^186.

Jackson, M. C. (1994). Geochemistry and metamorphic petrology ofthe Dominion Group metavolcanics in the Vredefort area. SouthAfricanJournal of Geology 97, 62^77.

Kirk, J., Ruiz, J., Chesley, J.,Walshe, J. & England, G. (2002). A majorArchean gold and crust-forming event in the Kaapvaal Craton,South Africa. Science 297, 1856^1858.

Klemd, R. (1999). A comparison of fluids causing post-depositionalhydrothermal alteration in Archaean basement granitoids and theWitwatersrand Basin. Mineralogy and Petrology 66, 111^122.

Klemd, R. & Hallbauer, D. K. (1987). Hydrothermally altered peralu-minous granites as a provenance model for Witwatersrand sedi-ments. Mineralium Deposita 22, 227^235.

Kositcin, N. & Krapez, B. (2004). SHRIMP U-Pb detrital zircon geo-chronology of the Late Archaean Witwatersrand Basin of SouthAfrica: relation between zircon provenance age spectra and basinevolution. Precambrian Research 129, 141^168.

Kositcin, N., McNaughton, N. J., Griffin, B. J., Fletcher, I. R.,Groves, D. I. & Rasmussen, B. (2003). Textural and geochemicaldiscrimination between xenotime of different origin in theArchaean Witwatersrand Basin, South Africa. Geochimica et

Cosmochimica Acta 67, 709^731.Kranidiotis, P. & MacLean, W. H. (1987). Systematics of chlorite

alteration at the Phelps Dodge massive sulfide deposit, Matagami,Quebec. Economic Geology 82, 1898^1911.

Kro« ner, A., Jaeckel, P., Brandl, G., Nemchin, A. A. & Pidgeon, R. T.(1999). Single zircon ages for granitoid gneisses in the CentralZone of the Limpopo Belt, Southern Africa, and geodynamic sig-nificance. Precambrian Research 93, 299^337.

le Roex, A. P. & Lanyon, R. (1998). Isotope and trace element geo-chemistry of Cretaceous Damaraland lamprophyres and carbona-tites, northwestern Namibia: evidence for plume^lithosphereinteractions. Journal of Petrology 39, 1117^1146.

Ludwig, K. R. (2000). Isoplot/Ex, Vers. 2.2. Berkeley Geochronology Center,Special Publications 1a.

Lugmair, G.W. & Marti, K. (1978). Lunar initial 143Nd/144Nd; differ-ential evolution of the lunar crust and mantle. Earth and Planetary

Science Letters 39, 349^357.Marsh, J. S., Bowen, M. P., Rogers, N. W. & Bowen, T. B. (1989).

Volcanic rocks of the Witwatersrand Triad, South Africa. II:Petrogenesis of mafic and felsic rocks of the Dominion Group.Precambrian Research 44, 39^65.

McCourt, S. & Armstrong, R. A. (1998). SHRIMP U^Pb zircon geo-chronology of granites from the Central Zone, Limpopo Belt,southern Africa: Implications for the age of the LimpopoOrogeny. South AfricanJournal of Geology 101, 329^338.

Middlemost, E. A. K. (1994). Naming materials in the magma/igneousrock system. Earth-Science Reviews 37, 215^223.

Millonig, L., Zeh, A., Gerdes, A. & Klemd, R. (2008). Neoarchaeanhigh-grade metamorphism in the Central Zone of the LimpopoBelt (South Africa): Combined petrological and geochronologicalevidence from the Bulai pluton. Lithos 103, 333^351.

Minnitt, R. C. A., Reimold,W. U. & Colliston,W. P. (1994).The geology

of the Greenlands Greenstone Complex and selected granitoid terranes in the

southeastern quadrant of the Vredefort Dome. Information Circular of the

Economic Geology Research Unit, University of the Witwatersrand,

Johannesburg 281, 46 pp.Minter, W. E. L., Hill, W. C. N., Kidger, R. J., Kingsley, R. J. &

Snowden, P. A. (1986). The Welkom goldfield. In:Anhaeusser, C. R. & Maske, S. (eds) Mineral Deposits of

Southern Africa. Johannesburg: Geological Society of SouthAfrica, pp. 497^539.

Minter, W. E. L., Goedhart, M. L., Knight, J. & Frimmel, H. E.(1993). Morphology of Witwatersrand gold grains from the BasalReef: Evidence for their detrital origin. Economic Geology 88,237^248.

Miyashiro, A. (1974). Volcanic rock series in island arcsand active continental margins. American Journal of Science 274,321^355.

FRIMMEL et al. WITWATERSRAND HINTERLAND

2219

Moorbath, S.,Whitehouse, H. J. & Kamber, B. S. (1997). Extreme Nd-isotope heterogeneity in the early Archeançfact or fiction? Casehistories from northern Canada and West Greenland. Chemical

Geology 135, 213^231.Moyen, J.-F., Stevens, G. & Kisters, A. F. M. (2006). Record of mid-

Archaean subduction from metamorphism in the Barberton ter-rane, South Africa. Nature 442, 559^562.

Muntean, J. L., Frimmel, H. E., Phillips, N., Law, J. & Myers, R.(2005). Controversies on the origin of world-class gold deposits,part II:Witwatersrand gold deposits. SEGNewsletter 60, 7^19.

Passeraub, M., Wu« st, T., Kreissig, K., Smit, C. A. & Kramers, J. D.(1999). Structure, metamorphism, and geochronology of theRhenosterkoppies Greenstone Belt, South Africa. South African

Journal of Geology 102, 323^334.Pearce, J. A. (2008). Geochemical fingerprinting of oceanic basalts

with applications to ophiolite classification and the search forArchean oceanic crust. Lithos 100, 14^48.

Pearce, J. A., Harris, N. B. W. & Tindle, A. G. (1984). Trace elementdiscrimination diagrams for the tectonic interpretation of graniticrocks. Journal of Petrology 25, 956^983.

Pearce, N. J. G., Perkins, W. T., Westgate, J. A., Gorton, M. P.,Jackson, S. E., Neal, C. R. & Chenery, S. P. (1997). A compilationof new and published major and trace element data for NISTSRM 610 and NIST SRM 612 glass reference materials.Geostandards Newsletter 21, 115^144.

Phillips, G. N. & Law, J. D. M. (1994). Metamorphism of theWitwatersrand gold fields: a review. Ore Geology Reviews 9, 1^31.

Poujol, M. & Anhaeusser, C. R. (2001). The Johannesburg Dome,South Africa: new single zircon U^Pb isotopic evidence for earlyArchaean granite^greenstone development within the centralKaapvaal Craton. Precambrian Research 108, 139^158.

Poujol, M., Robb, L. J., Anhaeusser, C. R. & Gericke, B. (2003). Areview of geochronological constraints on the evolution of theKaapvaal Craton, South Africa. Precambrian Research 127, 181^213.

Reubi, O. & Blundy, J. (2008). Assimilation of plutonic roots, forma-tion of high-K ‘exotic’ melt inclusions and genesis of andesiticmagmas at Volca¤ n de Colima, Mexico. Journal of Petrology 49,2221^2243.

Richards, J. P., Boyce, A. J. & Pringle, M. S. (2001). Geologic evolu-tion of the Escondida area, northern Chile: a model for spatialand temporal localization of porphyry Cu mineralization.Economic Geology 96, 271^305.

Robb, L. J. & Meyer, M. (1987).The nature of the Archaean basementin the hinterland of the Witwatersrand Basin: I. the RandAnticline between Randfontein and Rysmierbult. South African

Journal of Geology 90, 44^63.Robb, L. J. & Meyer, F. M. (1990). The nature of theWitwatersrand

hinterland: Conjectures on the source-area problem. Economic

Geology 85, 511^536.Robb, L. J., Davis, D., Kamo, S. L. & Meyer, F. M. (1992).

Ages of altered granites adjoining the Witwatersrand Basin withimplications for the origin of gold and uranium. Nature 357,677^680.

Robb, L. J., Brandl, G., Anhaeusser, C. R. & Poujol, M. (2006).Archaean granitoid intrusions. In: Johnson, M. R., Anhaeusser, C.R. & Thomas, R. J. (eds) The Geology of South Africa.

Johannesburg: Geological Society of South Africa and Council forGeoscience, pp. 57^94.

Rudnick, R. L. & Fountain, D. M. (1995). Nature and composition ofthe continental crust: a lower crustal perspective. Reviews in

Geophysics 33, 267^309.Ruiz, J., Valencia, V.A., Chesley, J.T., Kirk, J., Gehrels, G. &

Frimmel, H.E. (2006). The source of gold for the Witwatersrandfrom Re-Os and U-Pb detrital zircon geochronology. GoldschmidtConference, August 2006, Melbourne. Geochimica et Cosmochimica

Acta 17, Suppl. 1, A543.Rutherford, M. J. & Devine, J. D. (1988).The May18,1980, eruption of

Mount St. Helens. 3. Stability and chemistry of amphibole in themagma chamber. Journal of Geophysical Research 93, No. B10.11949^11959.

Schmitz, M. D. & Bowring, S. A. (2003). Ultrahigh-temperaturemetamorphism in the lower crust during NeoarcheanVentersdorprifting and magmatism, Kaapvaal Craton, southern Africa.Geological Society of America Bulletin 115, 533^548.

Silver, P. G., Behn, M. D., Kelley, K., Schmitz, M. & Savage, B.(2006). Understanding cratonic flood basalts. Earth and Planetary

Science Letters 245, 190^201.Stacey, J. S. & Kramers, J. D. (1975). Approximation of terrestrial lead

isotope evolution by a two-stage model. Earth and Planetary Science

Letters 26, 207^221.Steiger, R. H. & Ja« ger, E. (1977). Subcommission on geochronology:

Convention on the use of decay constants in geo- and cosmochro-nology. Earth and Planetary Science Letters 36, 359^362.

Sun, S. & McDonough,W. F. (1989). Chemical and isotopic systema-tics of ocean basalts: implications for mantle composition and pro-cesses. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the

Ocean Basins. Geological Society, London, Special Publications 42,313^345.

Vennemann, T.W., Kesler, S. E. & O’Neil, J. R. (1992). Stable isotopecompositions of quartz pebbles and their fluid inclusions as tracersof sediment provenance: Implications for gold- and uranium-bearing quartz pebble conglomerates. Geology 20, 837^840.

Vennemann,T.W., Kesler, S. E., Frederickson, G. C., Minter,W. E. L.& Heine, R. R. (1995). Oxygen isotope sedimentology of gold- anduranium-bearing Witwatersrand and Huronian Supergroupquartz-pebble conglomerates. Economic Geology 91, 322^342.

Viljoen, R. P., Saager, R. & Viljoen, M. J. (1970). Some thoughts onthe origin and processes responsible for the concentration of goldin the Early Precambrian of southern Africa. Mineralium Deposita

5, 164^180.Wronkiewicz, D. J. & Condie, K. C. (1987). Geochemistry of Archean

shales from theWitwatersrand Supergroup, South Africa: Source-area weathering and provenance. Geochimica et Cosmochimica Acta

51, 2401^2416.Zeh, A., Gerdes, A., Klemd, R. & Barton, J. M. (2007). Archaean to

Proterozoic crustal evolution in the Central Zone of the LimpopoBelt (South Africa^Botswana): Constraints from combined U^Pband Lu^Hf isotope analyses of zircon. Journal of Petrology 48,1605^1639.

Zeh, A., Gerdes, A. & Barton, J. M., Jr (2009). Archean accretion andcrustal evolution of the Kalahari Cratonçthe zircon age and Hfisotope record of granitic rocks from Barberton/Swaziland to theFrancistown arc. Journal of Petrology 50, 933^966.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 12 DECEMBER 2009

2220