A clinopyroxenite intrusion from the Pilanesberg Alkaline Province, South Africa

(This is a sample cover image for this issue. The actual cover is not yet available at this time.)

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Precambrian Research 198– 199 (2012) 25– 36

Contents lists available at SciVerse ScienceDirect

Precambrian Research

journa l h omepa g e: www.elsev ier .com/ locate /precamres

A clinopyroxenite intrusion from the Pilanesberg Alkaline Province, South Africa

R. Grant Cawthorna,∗, Robert M. Ellamb, Lewis D. Ashwala, Susan J. Webba

a School of Geosciences, University of the Witwatersrand, PO Wits 2050, South Africab Scottish Universities Environmental Research Centre, Rankine Ave., East Kilbride G75 0QF, Scotland, United Kingdom

a r t i c l e i n f o

Article history:Received 19 May 2011Received in revised form12 December 2011Accepted 28 December 2011Available online xxx

Keywords:Pilanesberg Alkaline ProvinceSouth AfricaClinopyroxenitesTi-rich lavasMagnetic anomalies

a b s t r a c t

A number of circular negative magnetic anomalies (up to 8 km across) exist within the area encompassedby the western Bushveld Complex (150 km by 100 km) on the Kaapvaal craton in South Africa. They arecovered by up to 700 m of sedimentary rocks of the Karoo Supergroup, which could not produce theseanomalies. Exploration boreholes into one of these magnetic anomalies revealed a hidden volcanic com-plex, called the Elandskraal Volcano. One of these boreholes intersected an olivine–magnetite–apatiteclinopyroxenite body, which we studied. Ages on apatite–clinopyroxene pairs using Sm–Nd dating tech-niques yield a poorly constrained age of 1207 ± 200 Ma, because there is little variation in Sm–Nd betweenall the samples. This age correlates with the Pilanesberg Alkaline Province that spans the time period from1430 to 1200 Ma, but almost all age determinations from this suite give very large errors.

Three other clinopyroxenite bodies closely related to the Pilanesberg Alkaline Province have beenreported, but no geochemical data have been presented. Our mineral and whole-rock geochemical datapermit an interpretation of the genesis of these clinopyroxenite bodies. The basaltic lavas of the Eland-skraal Volcano are extremely unusual in having very high TiO2 (over 7 wt%), high Fe2O3 (16–21 wt%) andhigh incompatible element contents. Modelling the crystallization sequence using MELTS of the moremagnesian lava compositions yields olivine, clinopyroxene, magnetite and apatite as liquidus phaseswithin a temperature interval of less than 30 ◦C, which matches that observed in the clinopyroxenitebody. In our samples the Mg/(Mg + Fe) value for the mafic minerals and incompatible trace elementsabundances in clinopyroxene are consistent with crystallization from these unusual basaltic composi-tions. Contrasts with other clinopyroxenite bodies in the Phalaborwa Complex and the Bushveld Complexare documented in terms of rock associations and mineral compositions (both major and trace elements).

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

The Kaapvaal craton in South Africa (Fig. 1) contains someremarkable igneous suites, including (in stratigraphic order):the type locality for komatiites near Barberton; lavas of theVentersdorp Supergroup (the oldest at 2700 Ma, well-preservedcontinental flood basalt); the Bushveld Complex (the largest knownlayered intrusion), the Phalaborwa (carbonatite) Complex; thePilanesberg Alkaline Province; the Karoo flood basalt province;and numerous diamondiferous kimberlites (including the typelocality at Kimberley). A simplified map showing these igneousoccurrences is shown in Fig. 1. The Pilanesberg Alkaline Provincecontains a wide variety of rock types (as reviewed by Woolley,2001; Verwoerd, 2006) from mafic to felsic lavas, syenites andfoyaites, clinopyroxenites, carbonatites and kimberlites (includ-ing the famous Premier/Cullinan kimberlite suite). They span an

∗ Corresponding author. Tel.: +27 117176557; fax: +27 117176579.E-mail address: [email protected] (R.G. Cawthorn).

age range from about 1430 to 1200 Ma, but all reported ages havelarge uncertainties from tens to over 100 m.y. (Verwoerd, 2006).Within this Pilanesberg Alkaline Province are four examples ofclinopyroxenite (Fig. 1). The Spitskop Complex (1341 ± 37 Ma) isvery isolated, being nearly 300 km from the Pilanesberg Com-plex. It is associated with carbonatite and has no olivine, andhas been well described (Harmer, 1999). The other three, closerto the Pilanesberg Complex, have only received passing mention.Frick and Walraven (1985) reported the results of drilling throughKaroo cover into the Elandskraal Volcano (Fig. 1), considered tobe part of a line of alkaline bodies referred to as the FranspoortLine (Shand, 1923) or the Pienaars River Complex (South AfricanCommittee for Stratigraphy, 1980), and now regarded as part ofthe Pilanesberg Alkaline Province. Frick and Walraven (1985) pre-sented a possible cross-section of this volcanic complex, in whichthey showed a cross-cutting plug of clinopyroxenite, but no furtherinformation on the pyroxenite was presented. Further to the north-west is another buried alkaline complex, the Buffelskraal Complex,identified only from a drilling intersection, which also containsa clinopyroxenite (Frick and Walraven, 1986). Likewise, drilling

0301-9268/$ – see front matter © 2012 Elsevier B.V. All rights reserved.doi:10.1016/j.precamres.2011.12.016

Author's personal copy

26 R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36

Fig. 1. General map of most of the Kaapvaal craton (denoted by long-dashed lines) with major igneous bodies: Ko, the type locality for komatiites in Barberton; V, thenortheastern portion of the Ventersdorp lavas, one of the oldest continental flood basalt provinces; Ph, Phalaborwa Complex; B, Bushveld Complex; Pi, S, Bu, E, F and G,Pilanesberg Complex, Spitskop, Buffelskraal Complex, Elandskraal volcanic complex, Franspoort line of intrusive bodies (Pienaars River Complex), and Goudini Complex(which are all considered part of the Pilanesberg Alkaline Province); C and Pa, Cullinan and Palmietgat kimberlites; Ka, lavas of Karoo Supergroup. The borecore samples forthis study were taken from just west of the Palmietgat kimberlite—Pa. The inset shows a cross-section of the Elandskraal Volcano produced by Frick and Walraven (1985),showing a pyroxenite plug, assumed to be similar to that identified here. Abbreviations for the inset: G, Bushveld Granite; and from the Elandskraal Volcano BT, basalt andtrachyte lavas; A, agglomerate; P, pyroxenite. Width of section 25 km. The short-dashed box indicates the area shown in Fig. 2.

revealed a magnetite-rich hornblende clinopyroxenite, consideredto be part of the Goudini Carbonatite Complex (Verwoerd, 2008).However, no further information exists about these three occur-rences.

The north-central parts of the Kaapvaal craton are partially cov-ered by sedimentary rocks of the Karoo Supergroup (300–180 Ma),which conceal other possible occurrences of such rock types.Clinopyroxenites are also known in the Phalaborwa Complex(2060 Ma, Eriksson, 1989; Reischmann, 1995) and the BushveldComplex (2060 Ma, Walraven et al., 1990).

We obtained access to borecore material from the explorationdrilling programme into the magnetic anomaly described by Frickand Walraven (1985), which proved to be a clinopyroxenite witholivine, magnetite and apatite, and here report our attempts todetermine its age and genesis, and compare it with the above-mentioned clinopyroxenites on the craton.

2. Geological setting

The area of interest is located between the (post-Jurassic) Palmi-etgat kimberlite and the Franspoort Lineament (Fig. 1). Threemagnetic anomalies (Fig. 2), which reach up to 8 km across, areprominent on geophysical maps of the Kaapvaal craton (as shownby Fig. 4 of Frick and Walraven, 1985). These structures were inves-tigated via exploration boreholes (in 1975) that were described byFrick and Walraven (1985). They produced a map of the inferredgeology beneath the Karoo Supergroup cover, as shown in the insetto Fig. 1, and included a clinopyroxenite body. In view of the ref-erence to a pyroxenite plug by Frick and Walraven (1985), one ofthese boreholes was re-examined. The borehole penetrated 780 mof Karoo sedimentary rocks, which overlay 110 m of an ultramafic

Fig. 2. Aeromagnetic map of the area denoted in Fig. 1, showing the pronouncedmagnetic anomaly over the Elandskraal Volcano, and the extreme magnetic anomalyat the position of the pyroxenite plug (28◦12′ E, 25◦12′ S) at its southwest extremity,the present borehole locality. The red plus sign denotes the location of the boreholestudied here.

Author's personal copy

R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36 27

rock, dominated by clinopyroxene, olivine, magnetite and apatite(see below). No layering could be recognised, but there were slightvariations in grain size. This rock is fractured and veined by calcite.We were able to collect some sections of core for this study.

3. Petrography

All thin sections studied are quite similar. All are dominatedby coarse-grained clinopyroxene, with anhedral oxide phases, andminor proportions of anhedral and variably altered olivine, pseu-domorphed by serpentine. Extremely rare apatite grains occur inthe clinopyroxene. No feldspar or feldspathoid is present in anyof the thin sections. The clinopyroxene grains display polygonalto sutured mutual boundaries, and a few grains contain abun-dant small oxide plates in crystallographically controlled directions(Fig. 3a). Some grains show a zonation of these plates with somezones showing extreme enrichment of these lamellae (Fig. 3a) tothe extent that it becomes extremely difficult to accept that theyhave exsolved from an original pyroxene based on mass-balancecalculations (see below). Many of these zones define a shape mim-icking that of a euhedral grain, suggesting that the change fromoxide-bearing to oxide-poor zones represents growth zones, in rarecases showing multiple zones (Fig. 3a and b). The olivine grains arevariably serpentinised (Fig. 3c), but relic atolls are still recognis-able. The grains vary from being subhedral to anhedral relative tothe clinopyroxene. The oxide minerals occur both as small euhe-dral grains within clinopyroxene, and as larger areas interstitial tothe silicate grains (Fig. 3a and c). Both ilmenite and magnetite arepresent as interstitial grains.

Many of the sections contain two sets of fractures, one with ser-pentine veins not confined to the olivine grains, and the other withcalcite.

4. Whole-rock chemistry

Thirteen samples from the 110 m of core available were ana-lysed. After coarse crushing through the jaw crusher (±1 cm size)any pieces with calcite were discarded before fine pulverising.Analysis was performed by Setpoint South Africa, an ISOI 17025accredited laboratory in Johannesburg, by X-ray fluorescence (forthe major oxides) and using a Perkin-Elmer Elan 9000 ICP-MSfor the trace elements. Data are presented in Table 1. To test foranalytical precision, and because of the possible similarity withiron-rich ultramafic pegmatoids (IRUP) of the Bushveld Complex(as discussed below), for comparison we also obtained analyses onfour of these (IRUP) samples that had previously been analysed atthe University of the Witwatersrand by XRF using internationallyrecognised standards (Lennoir, 2004). A comparison of the analysesis given in Appendix.

All the 13 compositions from the current borehole are simi-lar, being rich in CaO, TiO2 and Fe2O3 (total), and poor in SiO2,consistent with them being oxide-bearing clinopyroxenites. Thereis almost no K2O in these rocks (average 0.02 wt%), suggestingthat they are extreme accumulates, as inferred from their texturesand complete absence of feldspars. Na2O and Al2O3 are also lowas expected for feldspar-free rocks. Two of the samples contain2 wt% P2O5, consistent with the observations of very small grains ofapatite within the clinopyroxenite, and which have been separatedfor isotope study (see below).

It is not possible to undertake a rigorous norm calculationbecause of the high ferric iron content in magnetite. However, wehave made an approximation. The compositions of the clinopy-roxene obtained by electron microprobe (see below) yield mg#of 87–83. Hence, we have manipulated the proportion of FeO andFe2O3 in the whole-rock analysis until we obtained mafic minerals

Fig. 3. Photomicrographs of the clinopyroxenites: (a) clinopyroxene showing acentral core with two crystallographically controlled orientations of oxide plates.Anhedral oxide phases can be seen rimming the clinopyroxene; (b) intense con-centrations of several zones of rods of oxide in a single pyroxene; (c) the outergrowth boundary of the oxide plates defines a euhedral crystallographic face, butthe inside of the oxide-rich zone is irregular. Anhedral olivine (now replaced by ser-pentine with chains of magnetite) can be seen on the lower right. Rare, euhedraloxide grains occur within clinopyroxene, but not where there is the concentrationof oxide plates. Long axes of photographs are (a) 4 mm, (b) 3 mm and (c) 2 mm.

in the norm calculation that gave a ratio for mg# of 83 (see Table 1).Eleven of the 13 samples gave a remarkably similar ratio of ilmeniteto magnetite between 0.55 and 0.75, with two samples giving 1.0and 1.23. Some samples contain very minor calcium disilicate inthe norm, which possibly reflects addition of CaO as calcite veins.Normative plagioclase proportions are very low, and all except onesample produced minor (<2%) nepheline in the norm. The analysedclinopyroxenes contain minor Al2O3 and Na2O (see below), and inview of the absence of feldspar in all samples, is considered the rea-son for the minor normative plagioclase and nepheline. One samplehas normative orthopyroxene, which was not seen in the thin

Author's personal copy

28 R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36

Table 1Analyses of clinopyroxenites from this study (samples 780–900) and ultramafic pegmatitic pyroxenites from the Bushveld Complex (GC samples).

Depth (m) LLDa 780 797 802 810 823 832 850 865 867

SiO2 0.05 35.80 40.00 35.80 33.80 33.10 42.00 37.90 36.20 35.70TiO2 0.01 6.52 4.49 6.81 7.48 5.97 3.98 5.84 6.34 6.83Al2O3 0.05 3.19 2.14 3.02 3.15 3.32 3.60 3.09 2.92 3.08Fe2O3 (T) 0.01 26.50 22.90 27.40 30.30 28.70 18.40 24.20 25.80 26.00MnO 0.01 0.30 0.24 0.33 0.36 0.24 0.23 0.27 0.28 0.32MgO 0.05 10.70 14.70 10.10 9.92 10.10 12.70 11.00 10.60 10.50CaO 0.01 14.60 13.50 15.90 14.70 14.70 17.60 16.40 16.00 16.80Na2O 0.05 0.44 0.35 0.39 0.40 0.36 0.45 0.38 0.44 0.40K2O 0.01 0.03 0.02 0.01 <0.01 0.01 0.07 <0.01 0.04 0.02P2O5 0.01 0.07 0.68 0.05 0.04 0.24 0.10 0.03 0.42 0.85V2O5 0.01 0.14 0.07 0.15 0.16 0.12 0.09 0.13 0.13 0.12Cr2O3 0.01 <0.01 0.23 <0.01 <0.01 0.03 0.06 0.02 0.04 0.03LOI 1.61 1.13 0.32 −0.28 2.03 0.93 0.34 −0.02 0.07S 0.01 0.02 0.8 0.37 0.48 2.37 0.33 0.4 0.26 0.4Total 99.92 101.25 100.65 100.51 101.29 100.54 100.00 99.45 101.12

Trace elements in ppmBa 5 75 48 48 44 27 223 44 49 42Ce 0.05 29.7 57.8 32 37.1 54.6 45.8 32.3 57.3 90.6Co 0.1 86.7 106 98.4 135 264 81.5 99.4 94.8 98.4Cu 0.5 577 259 468 848 3280 523 729 612 787Hf 0.02 3.76 2.88 3.44 5.35 5 4.99 4.13 4.38 4.75La 0.1 12.4 24.2 14.9 14.3 24.6 19.4 14.7 25.5 38.9Nb 0.1 4.1 3.8 1.8 14 6.7 11.5 7.8 11.2 7.9Ni 0.5 165 655 183 193 553 208 215 208 221Pb 0.5 4.6 3 9 5.5 5.9 4.4 3.5 4.4 8.9Rb 0.2 2.7 1.7 1 0.7 1.2 4.7 1.3 2.5 1.4Sr 0.5 178 238 140 178 156 157 146 190 222Y 0.1 13.6 14.7 14.2 18.6 19.8 18.5 14.9 19.6 27.4Yb 0.1 1.2 1.1 1.2 1.6 1.7 1.6 1.3 1.6 2Zr 0.5 127 88 104 102 137 122 113 108 123

Normative mineralsOr 0.18 0.12 0.06 0.41 0.06 0.24Ab 3.10 2.96 0.40 2.41 1.04 0.76An 6.64 4.21 6.46 6.77 7.41 7.60 6.70 5.87 6.55Ne 0.34 1.79 1.62 1.65 0.76 1.18 1.60 1.83Cpx 52.05 46.44 57.35 52.38 50.11 62.80 59.32 56.44 55.97Opx 15.22Ol 5.80 5.40 2.85 3.51 4.54 5.76 3.98 4.37 3.95Mgt 17.69 14.65 18.31 21.22 20.90 11.72 15.79 16.83 16.96Ilm 12.38 8.53 12.94 14.21 11.34 7.56 11.09 12.04 12.97Ap 0.16 1.58 0.12 0.09 0.56 0.23 0.07 0.97 1.97Lc 0.05 0.05 0.09CDS* 0.05 0.27 0.30mg# 0.81 0.83 0.81 0.83 0.84 0.84 0.81 0.80 0.82

Depth (m) 871 879 888 900 GC 2288 GC 2289 GC 2290 GC 2291

SiO2 31.80 30.30 38.40 37.80 40.60 41.80 38.50 32.60TiO2 9.67 9.18 4.59 5.24 3.52 1.55 3.42 1.96Al2O3 3.38 3.30 2.84 2.82 7.14 2.70 2.65 0.57Fe2O3 (T) 25.30 27.60 22.00 24.70 26.60 29.20 34.20 52.40MnO 0.41 0.35 0.24 0.39 0.33 0.38 0.41 0.63MgO 8.85 8.70 11.10 11.00 9.75 10.80 10.00 12.80CaO 17.60 16.80 18.20 17.30 12.80 13.60 11.80 3.17Na2O 0.41 0.35 0.40 0.42 0.39 0.26 0.19 0.27K2O 0.02 <0.01 0.01 0.03 0.07 0.01 0.02 0.03P2O5 2.68 2.63 0.04 0.02 <0.01 0.61 0.01 0.02V2O5 0.11 0.11 0.12 0.14 0.10 0.14 0.13 0.09Cr2O3 <0.01 <0.01 0.05 0.04 0.04 0.20 0.07 0.07LOI −0.23 0.73 1.07 0.58 −0.71 −0.92 −1.59 −3.59S 0.43 0.82 0.63 0.37 0.36 0.44 0.18 0.04Total 100.43 100.87 99.69 100.85 100.99 100.77 99.99 101.06

Trace elements in ppmBa 28 28 34 32 32 17 19 20Ce 236 194 34 30.7 4.8 17.2 4.75 2.26Co 86.1 136 102 98.4 151 121 146 232Cu 147 1160 816 812 240 251 118 28.4Hf 6.6 1.4 4.33 4.34 0.58 0.59 0.59 0.23La 92.4 81.1 14.7 22.2 5.8 13.2 4 3.8Nb 4.1 4.3 12.1 10.7 3.5 0.8 1.8 0.9Ni 34.2 215 242 254 178 262 162 191Pb 10.1 4.3 8.8 4.8 2.5 6.2 2.5 1.5Rb 2.6 1.2 1.7 2.3 2 0.5 0.7 0.7

Author's personal copy

R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36 29

Table 1 (Continued)

Depth (m) 871 879 888 900 GC 2288 GC 2289 GC 2290 GC 2291

Sr 274 274 169 164 107 45.7 36.6 5.5Y 61.8 50.7 15.8 14.6 6.7 20.6 8.8 3Yb 4.5 3.6 1.3 1.3 0.7 1.9 0.9 0.4Zr 211 145 109 105 25.4 26.7 31.4 40.5

Normative minerals ModeOrAbAn 7.32 7.40 5.92 5.72Ne 1.88 1.60 1.83 1.93Cpx 49.19 46.04 59.52 59.32 56 59 51 14OpxOl 2.55 3.26 4.49 4.29 30 28 30 65Mgt 14.84 17.14 14.35 16.70 14 13 18 21Ilm 18.37 17.44 8.72 9.95Ap 6.21 6.10 0.09 0.05Lc 0.09 0.05 0.05 0.14CDSb 0.22 0.28 2.94 1.73mg# 0.82 0.84 0.80 0.80

bd, below detection level. Samples 780–900 are from this study. GC samples are IRUP from the Bushveld Complex. Numbers refer to depth in m in the borehole. Mode wasdetermined by point counting in transmitted light and magnetite and ilmenite are combined as total oxide.

a LLD is lower limit of detection.b CDS is calcium disilicate.

section, and the reason for the elevated SiO2 in the analysis is notknown. Using this method for distributing ferric and ferrous iron,we find that these rocks contain between 20 and 35% oxides, andbetween 45 and 60% clinopyroxene. Olivine varies between 2 and6%. In terms of the clinopyroxene, oxide and olivine contents thesefigures agree with the observed modal proportions.

For the samples from the Bushveld-related iron-rich clinopy-roxenite we have determined modal proportions by point countingusing transmitted light, and so the magnetite and ilmenite propor-tions are combined as oxide phase (see Table 1). However, becauseof the coarse grain size these modes may not be very precise.

5. Mineral compositions

We analysed the clinopyroxene, olivine, magnetite and ilmeniteby electron microprobe. Wavelength dispersive electron micro-probe analyses were obtained with a Cameca SX-100 instrumenthoused at Spectrau, University of Johannesburg. Operating con-ditions were as follows: accelerating voltage 20 kV, beam current40 nA. A focused beam of 1 �m was used for most analyses, but thediameter was expanded to 20 �m for integrating coarsely inter-grown phases in the magnetite and clinopyroxene. All elementswere analysed on their K� lines and the instrument was controlledby the Peak Site software supplied by Cameca. Matrix correctionswere carried out using the X-PHI software (Merlet, 1994). Calibra-tion standards used included well characterized natural silicate andoxide minerals and synthetic pure metals and oxides.

Compositions are given in Tables 2–5. All clinopyroxene anal-yses have similar compositions (Table 2) with mg# between 87and 83, with most nearer the lower value. Grains also contain over45% of the wollastonite molecule within the pyroxene quadrilateral(Fig. 4), and so have an alkalic rather than tholeiitic affinity. Of theminor oxides, Na2O is close to 0.5 wt%, Al2O3 is 1.1–1.6 wt%, TiO2ranges from 0.7 to 1.6 wt% (with one low value of 0.26 wt%), andCr2O3 is variable with values from 0.06 to 0.45 wt%. We analysedgrains in areas where there was an abundance of oxide lamellae andalso grains where they were absent. The mg# of the pure pyrox-ene is the same inside and outside these oxide-rich zones, and inoxide-free grains. Only a few grains in some of the slides showthis lamellar zonation feature, and so the clinopyroxenes maybe an assemblage from two different crystallizing environments,although there is no difference in mg# when comparing sampleswith this lamellar feature with samples not showing it. We used

a defocused beam to analyse an area where there was abundantoxide phase present. One such analysis is given in Table 2. It hasa composition unlike any clinopyroxene in being too low in SiO2,as shown by its structural formula, and so we conclude that thepresence of these oxide lamellae is not the result of isochemicalexsolution.

Only one sample had small remnant islands of fresh olivinethat we felt confident to analyse. All analyses (Table 3) from thissingle sample gave Fo78. NiO and CaO gave 0.18 and 0.23 wt%,respectively.

The discrete, interstitial magnetite grains from three differentsamples show different minor oxide compositions (Table 4). Aver-age values for the three samples are Cr2O3 4.9, 2.7 and 0.6 wt%;for TiO2 16, 12 and 15 wt%; and for MgO 3.3, 0.5 and 1.1 wt%,respectively. We also analysed the magnetite lamellae in thesample from 871 m depth in the core, and found that theycontain between 9 and 19% Cr2O3, which suggests that theyare unrelated to the interstitial grains. It is difficult to esti-mate the proportion of lamellae in the zones of clinopyroxene,but their very high Cr2O3 contents and constant mg# of theclinopyroxene in both facies of the clinopyroxene grains sug-gest that they could not have formed by exsolution from theclinopyroxene.

Fig. 4. Pyroxene quadrilateral showing the compositions of clinopyroxene from thepresent study compared to minerals in the IRUP of the Bushveld Complex (fromScoon and Mitchell, 1994), the Spitskop Complex (Harmer, 1999) and the Phal-aborwa Complex (Eriksson, 1989). The olivine compositions from the present studyare also shown. The continuous solid line shows the typical range of clinopyrox-ene compositions in tholeiitic intrusions as exemplified by the Bushveld Complex(Atkins, 1969).

Author's personal copy

30 R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36

Table 2Clinopyroxene analyses.

Depth (m) 871 871 871 871 871 871 871 871 871 871Beam Broad Broad Broad Broad Broad Broad Broad Broad Fine Fine

SiO2 53.06 51.4 51.53 52.1 52.18 52.22 52.84 52.44 53.57 51.97Al2O3 0.79 1.66 1.63 1.16 1.16 1.15 0.94 0.95 0.58 1.29TiO2 0.81 1.52 1.5 1.22 1.2 1.22 0.74 0.75 0.26 1.05Cr2O3 0.37 0.06 0.06 0.07 0.06 0.06 0.44 0.45 0.18 0.33FeO 4.39 5.72 5.65 5.38 5.43 5.51 4.78 5.28 4.35 4.8MnO 0.1 0.11 0.13 0.14 0.13 0.12 0.12 0.11 0.12 0.09MgO 16.45 15.49 15.38 15.6 15.57 15.6 16 15.94 16.28 16.06NiO 0.04 0.03 0.04 0.03 0.02 0.03 0.02 0.03 0.03 0.04CaO 23.37 23.39 23.44 23.61 23.58 23.65 23.58 23.5 24.36 23.33Na2O 0.42 0.52 0.53 0.48 0.49 0.49 0.53 0.49 0.33 0.49Total 99.8 99.9 99.89 99.79 99.82 100.05 99.99 99.94 100.06 99.45

Structural formulae per 6 oxygens:Si 1.9539 1.9078 1.9121 1.9316 1.9337 1.9319 1.9483 1.9402 1.9692 1.9281Al 0.0343 0.0721 0.0708 0.0504 0.0504 0.0499 0.0408 0.0412 0.0252 0.0560Ti 0.0224 0.0421 0.0415 0.0338 0.0332 0.0338 0.0205 0.0208 0.0072 0.0291Cr 0.0108 0.0017 0.0017 0.0020 0.0017 0.0017 0.0128 0.0131 0.0052 0.0096Fe2+ 0.1352 0.1762 0.1740 0.1657 0.1672 0.1697 0.1472 0.1626 0.1340 0.1478Mn 0.0031 0.0034 0.0041 0.0044 0.0041 0.0037 0.0037 0.0034 0.0037 0.0028Mg 0.9031 0.8504 0.8443 0.8564 0.8548 0.8564 0.8784 0.8751 0.8937 0.8817Ni 0.0012 0.0009 0.0012 0.0009 0.0006 0.0009 0.0006 0.0009 0.0009 0.0012Ca 0.9221 0.9229 0.9248 0.9315 0.9304 0.9331 0.9304 0.9272 0.9611 0.9205Na 0.0300 0.0371 0.0378 0.0343 0.0350 0.0350 0.0378 0.0350 0.0236 0.0350Total 4.0161 4.0146 4.0124 4.0114 4.0110 4.0162 4.0206 4.0195 4.0239 4.0117Si + Al + Ti 2.0107 2.0220 2.0244 2.0157 2.0173 2.0156 2.0096 2.0022 2.0016 2.0131Mg/Mg + Fe 87.0 82.8 82.9 83.8 83.6 83.5 85.6 84.3 87.0 85.6Wo 47.0 47.3 47.6 47.7 47.7 47.6 47.6 47.2 48.3 47.2En 46.1 43.6 43.5 43.8 43.8 43.7 44.9 44.5 44.9 45.2Fs 6.9 9.0 9.0 8.5 8.6 8.7 7.5 8.3 6.7 7.6

Depth (m) 880 880 880 880 880 859 859 871a

Beam Fine Fine Fine Fine Fine Fine Fine Broad

SiO2 52.13 52.21 51.41 51.26 51.1 51.72 50.95 38.87Al2O3 1.11 1.05 1.32 1.78 1.63 1.29 1.51 1.34TiO2 0.95 1.07 1.52 1.55 1.59 1.1 1.38 4.25Cr2O3 0.49 0.2 0.2 0.26 0.25 0.15 0.08 3.93FeO 4.91 5.16 5.43 5.35 5.54 5.46 5.89 22.83MnO 0.1 0.11 0.11 0.13 0.13 0.12 0.12 0.19MgO 15.53 15.79 15.55 15.33 15.32 15.28 14.92 14.24NiO 0.03 0.04 0.03 0.03 0.03 0.01 0.03 0.13CaO 23.91 23.98 23.67 23.65 23.52 23.6 23.47 17.11Na2O 0.54 0.49 0.53 0.56 0.51 0.48 0.51 0.42Total 99.75 100.11 99.83 99.93 99.65 99.27 98.89 103.39

Structural formulae per 6 oxygens:Si 1.9338 1.9305 1.9112 1.9025 1.9035 1.9303 1.9144 1.550Al 0.0485 0.0458 0.0578 0.0779 0.0716 0.0568 0.0669 0.058Ti 0.0265 0.0298 0.0425 0.0433 0.0446 0.0309 0.0390 0.118Cr 0.0144 0.0058 0.0059 0.0076 0.0074 0.0044 0.0024 0.114Fe2+ 0.1523 0.1596 0.1688 0.1661 0.1726 0.1704 0.1851 0.703Mn 0.0031 0.0034 0.0035 0.0041 0.0041 0.0038 0.0038 0.006Mg 0.8588 0.8704 0.8618 0.8482 0.8508 0.8501 0.8357 0.782Ni 0.0009 0.0012 0.0009 0.0009 0.0009 0.0003 0.0009 0.004Ca 0.9503 0.9500 0.9428 0.9405 0.9387 0.9437 0.9449 0.675Na 0.0388 0.0351 0.0382 0.0403 0.0368 0.0347 0.0372 0.030Total 4.0276 4.0315 4.0338 4.0318 4.0309 4.0259 4.0308 4.040Si + Al + Ti 2.0089 2.0060 2.0115 2.0237 2.0196 2.0179 2.0203 1.726Mg/Mg + Fe 84.9 84.5 83.6 83.6 83.1 83.3 81.9 52.649Wo 48.4 48.0 47.8 48.1 47.8 48.0 48.1En 43.8 44.0 43.7 43.4 43.4 43.3 42.5Fs 7.8 8.1 8.6 8.5 8.8 8.7 9.4

a Analysis 871 used a defocused beam across one of the zones showing abundant oxide lamellae.

Ilmenite only occurs as interstitial grains. It differs in its minoroxide content between samples (Table 5), containing on average 0.1and 0.04% Cr2O3 and 5.5 and 2.8% MgO in the two samples. How-ever, for MgO in the two samples, there is an inverse relationshipbetween the high and low values compared to the magnetite grainsin the same sample. Thus, MgO in ilmenite is higher in the sam-ple from 871 m than the sample from 859 m by a factor of almosttwo, whereas MgO in magnetite is higher in the sample from 859 mthan the sample from 871 m. In view of the considerable variation

in compositions of the oxide phase even within one sample wedo not believe that we can obtain realistic estimates of the finalequilibration temperatures of oxide pairs.

6. Isotopic age determination

Three apatite–clinopyroxene pairs were separated from 20-cm lengths of bore core, but yielded only small quantities of

Author's personal copy

R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36 31

Table 3Olivine analyses.

Depth (m) 880 880 880 880

SiO2 38.78 38.88 38.86 38.83TiO2 0.01 0.02 0.01 0FeO 19.9 19.78 20.15 20.17MnO 0.36 0.34 0.37 0.36MgO 40.59 40.54 40.44 40.45NiO 0.23 0.23 0.23 0.23CaO 0.18 0.18 0.17 0.19Total 100.08 99.99 100.24 100.24

Structural formulae per 4 oxygens:Si 0.9980 1.0005 0.9992 0.9986Ti 0.0002 0.0004 0.0002 0.0000Fe2+ 0.4283 0.4257 0.4333 0.4338Mn 0.0078 0.0074 0.0081 0.0078Mg 1.5572 1.5551 1.5502 1.5508Ni 0.0048 0.0048 0.0048 0.0048Ca 0.0050 0.0050 0.0047 0.0052Total 3.0024 2.9996 3.0009 3.0016Fo 78.4 78.5 78.2 78.1

apatite (1.5–6.7 mg). Samples were digested in HF–HNO3–HCl andaliquoted for (i) 143Nd/144Nd, 87Sr/86Sr, Rb and Sr concentrationand (ii) Sm and Nd concentrations, respectively. Samples werespiked with (i) 84Sr and (ii) 149Sm and 145Nd. Rb, Sr, Nd and Smwere separated using cation (Rb, Sr, REE) and anion (Sm, Nd)exchange chromatography. Sr, Nd and Sm samples were analysedon a VG Sector 54-30 mass spectrometer; Sr and Nd isotope ratiomeasurements were made in multi-dynamic mode and Sm andNd isotope dilution measurements were made by static multi-collection. During the course of this study NIST SRM987 gave87Sr/86Sr of 0.710252 ± 18 (2SD) and the in-house JM reference Ndsolution gave 0.511520 ± 10 (2SD) for 143Nd/144Nd, which are bothwithin error of the long-term mean. Isochron fitting was performedusing Isoplot/Ex (Ludwig, K.R., 2001, Berkeley Geochronology Cen-ter Special Publication No. 1a). All three pairs were analysed for87Sr/86Sr and Sr concentration by isotope dilution mass spectrom-etry and Rb concentrations by solution ICP-MS. Rb concentrationswere correctly anticipated to be lower than is able to be accu-rately quantified with our routine 87Rb spike (i.e. mixture 85Rb/87Rbwould have approached zero with the 85 peak being below whatis able to be measured with a Faraday detector), so were analysedby ICP-MS using an Agilent 7500ce instrument. Measured Rb con-centrations would be affected by non-quantitative yields but yieldsare unlikely to be so low as to significantly increase Rb/Sr.

For the small apatite samples, there are likely to be significantweighing errors which will greatly amplify the uncertainty in theelemental concentrations. However, the 147Sm/144Nd determina-tion should be reliable because both Sm and Nd were determinedon the same sample using normal spike weights so that any uncer-tainty in spike to sample ratio will cancel out in the 147Sm/144Ndcalculation. Both apatite and clinopyroxene have Rb/Sr ratios thatare so low that the measured 87Sr/86Sr will approximate the initialSr isotope composition with the largest age-corrected initial ratiosbeing less than 0.00005 lower than the measured value.

The two apatite–clinopyroxene pairs span a small range in147Sm/144Nd, but are adequate to define a broadly isochronousrelationship. Taking the minimum uncertainties in 147Sm/144Ndto be the propagated internal errors from the mass spectromet-ric measurements (probably an underestimate) yields an age of1207 ± 200 Ma with a MSWD of 9.3. While this is a “Model 3” fit thatshould be treated with caution, it is likely to be the best indicationof age that can be extracted from a rock of this mineralogy. The ini-tial 143Nd/144Nd is 0.51097 ± 0.00015 (Fig. 5), which correspondsto a value for C-- Nd of −2.1.

Fig. 5. Plot of 147Sm/144Nd versus 143Nd/144Nd (present) for apatite and clinopyrox-ene mineral separates (Table 6) and calculated age and initial ratio.

As mentioned above, nearly all attempts to determine the agesof suites of rocks in the Pilanesberg Alkaline Province have resultedin large uncertainties (reviewed by Verwoerd, 2006; Hanson et al.,2006), whatever method was used. Petrographic descriptions ofall of these rock suites refer to alteration. In this present exam-ple, olivine is almost entirely replaced by serpentine. Such generalalteration may have resulted in the large uncertainties in all agedeterminations, and could have affected the one sample of clinopy-roxene that showed an anomalous initial Sr isotope ratio. The threeapatite determinations, with their very much higher absolute Srabundances gave an initial ratio identical to the other two clinopy-roxene samples. However, since the Sr–Rb isotope data were notused for age determination such possible alteration does not affectthe possible age calculated above from the Nd/Sm data. We con-clude that this clinopyroxenite body is most probably of Pilanesbergage (1430–1200 Ma according to Verwoerd, 2006). Further discus-sion is presented below.

7. Possible magma compositions

These clinopyroxene–olivine–oxide rocks are obviously cumu-lates (based on their very low abundances of incompatibleelements, such as K2O) and the petrographic textures. The nature ofthe magma that might have crystallized such a liquidus assemblagecan be investigated. In the same area (inset to Fig. 1) are bore-holes that intersected lavas of the Elandskraal Volcano (Frick andWalraven, 1985). They include basalts and trachytes. All the basaltsanalysed are unusual in containing extremely high TiO2 and Fe2O3(total) concentrations of up to 7 wt% and 22 wt%, respectively, andhigh incompatible trace element abundances (Table 7). The highestTiO2 contents that we know of in an alkali basaltic magma suite areabout 5–6 wt% (Thompson et al., 1998), excluding individual sam-ples that may be highly porphyritic or cumulus-enriched. However,we note that similarly high TiO2 and relatively high Fe2O3 val-ues have been reported in experimentally produced glasses frommeimechite from the Siberian flood basalts (Elkins-Tanton et al.,2007).

Frick and Walraven (1985) described many features in thin sec-tion resulting from alteration, referring to the presence of calcite,epidote, sericite and chlorite replacement of primary grains. Wenote that there are very large variations in the ratio of K2O versusNa2O in this suite of lavas (Fig. 6). Such an unusual relationship

Author's personal copy

32 R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36

Table 4Magnetite analyses.

Depth (m) 871 871 871 871 871 871 871 859 859 880 880Beam width 20 �m 20 �m 20 �m 10 �m 1 �m 1 �m 1 �m 1 �m 1 �m 1 �m 1 �mType Grain Grain Grain Grain Grain Lamellae Lamellae Grain Grain Grain Grain

SiO2 0.03 0.04 0.03 0.01 0.02 0.41 0.39 0.04 0.03 0.02 0.03TiO2 12.65 11.46 12.20 11.43 10.64 12.67 14.08 15.12 15.61 16.10 16.57Al2O3 2.11 2.28 2.27 2.21 2.28 2.53 0.08 2.27 2.26 2.24 2.20Cr2O3 2.73 2.74 2.78 2.97 2.91 9.18 19.77 0.61 0.60 4.91 4.90V2O3 0.33 0.37 0.39 0.36 0.39 0.21 0.23 0.25 0.30 0.22 0.26Fe2O3

* 37.72 40.25 38.58 40.13 41.96 28.91 18.47 35.25 34.50 31.38 30.59FeO* 39.99 39.10 39.77 38.98 38.32 38.92 41.57 42.08 42.76 39.37 39.90MnO 1.15 1.15 1.37 1.16 1.23 2.30 0.91 0.51 0.50 1.36 1.54MgO 0.57 0.61 0.45 0.58 0.60 0.27 0.16 1.23 1.12 3.38 3.25CaO 0.02 0.01 0.01 0.03 0.02 0.82 0.86 0.07 0.11 0.00 0.00NiO 0.24 0.26 0.29 0.27 0.32 0.22 0.20 0.24 0.22 0.30 0.30ZnO 0.34 0.32 0.25 0.32 0.30 0.06 0.08 0.08 0.10Total 97.88 98.59 98.39 98.45 98.99 96.45 96.72 97.73 98.10 99.35 99.64

*Formulae per 3 cations calculated for charge balanceSi 0.001 0.002 0.001 0.000 0.001 0.016 0.015 0.002 0.001 0.001 0.001Ti 0.363 0.327 0.349 0.327 0.302 0.366 0.409 0.431 0.444 0.444 0.456Al 0.095 0.102 0.102 0.099 0.102 0.115 0.004 0.102 0.101 0.097 0.095Cr 0.082 0.082 0.084 0.089 0.087 0.279 0.604 0.018 0.018 0.142 0.142V3+ 0.010 0.011 0.012 0.011 0.012 0.006 0.007 0.008 0.009 0.006 0.008Fe3+ 1.084 1.148 1.103 1.147 1.193 0.836 0.537 1.007 0.982 0.865 0.842Fe2+ 1.277 1.240 1.264 1.238 1.211 1.251 1.343 1.335 1.353 1.207 1.220Mn 0.037 0.037 0.044 0.037 0.039 0.075 0.030 0.016 0.016 0.042 0.048Mg 0.032 0.034 0.025 0.033 0.034 0.015 0.009 0.070 0.063 0.185 0.177Ca 0.001 0.000 0.000 0.001 0.001 0.034 0.036 0.003 0.004 0.000 0.000Ni 0.007 0.008 0.009 0.008 0.010 0.007 0.006 0.007 0.007 0.009 0.009Zn 0.010 0.009 0.007 0.009 0.008 0.000 0.000 0.002 0.002 0.002 0.003Total 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000

* ferrous and ferric iron content calculated to yield charge balance.

suggests significant metasomatic alteration. However, TiO2 is con-sidered an immobile element, and so these high, and near-constantvalues, are considered primary. We selected the most magnesiancomposition to study its crystallization sequence (sample num-ber 30, Table 7). We note that this composition contains 4.1 wt%CO2 and these rocks are reported to show calcite veining. Fur-ther, it has a K2O/Na2O value of 4, whereas no other sample

Table 5Ilmenite analyses.

Depth (m) 871 871 859 859

SiO2 0.03 0.02 0.01 0.03TiO2 52.63 53.04 53.5 52.75Al2O3 0.03 0.02 0.01 0.02Cr2O3 0.12 0.08 0.02 0.04V2O3 0.14 0.08 0.07 0.01Fe2O3

* 4.41 2.91 0.94 0.94FeO* 35.87 37.52 41.01 42.23MnO 0.63 0.99 1.01 1.29MgO 6.03 5.1 3.38 2.15CaO 0.02 0.04 0.01 0.04NiO 0.07 0.06 0.04 0.04ZnO 0.02 0 0.02 0.02Total 100.00 99.86 100.01 99.55

*Formulae per 2 cations calculated for charge balanceSi 0.0007 0.0005 0.0002 0.0007Ti 0.9562 0.9710 0.9900 0.9897Al 0.0009 0.0006 0.0003 0.0006Cr 0.0023 0.0015 0.0004 0.0008V3+ 0.0027 0.0016 0.0014 0.0002Fe3+ 0.0802 0.0533 0.0174 0.0176Fe2+ 0.7247 0.7638 0.8439 0.8810Mn 0.0129 0.0204 0.0211 0.0273Mg 0.2171 0.1850 0.1239 0.0799Ca 0.0005 0.0010 0.0003 0.0011Ni 0.0014 0.0012 0.0008 0.0008Zn 0.0004 0.0000 0.0004 0.0004Total 2.0000 2.0000 2.0000 2.0000

* ferrous and ferric iron content calculated to yield charge balance.

has a value greater than 1.5, and most are slightly less thanunity. When run through the MELTS programme at 1 kb pres-sure (Ghiorso et al., 2002), this composition crystallizes leuciteas the liquidus phase, and is obviously inappropriate. We thenselected the composition with the second highest MgO content(sample number 47, Table 7) which only contained 2 wt% CO2(the lowest of all samples) and had a K2O/Na2O value of 1.4.For this composition titaniferous magnetite and clinopyroxeneare the liquidus phases, but no olivine formed at any tempera-ture. We then added 2 and 4 wt% olivine of the composition givenin Table 3 to the analysis of sample number 47. The composi-tion with 2 wt% olivine added did not produce any olivine in the

Fig. 6. Plot of Na2O versus K2O for volcanic samples from the Elandskraal VolcanicComplex (data are from Frick and Walraven, 1985) illustrating the wide scatter dueto alteration.

Author's personal copy

R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36 33

MELTS programme, but that with 4 wt% olivine added producedolivine, clinopyroxene, magnetite and apatite within 30 ◦C of theliquidus temperature. The olivine produced in this calculation hasa composition with mg# of 79. The clinopyroxene has an mg# of82, and contains 1.9 wt% TiO2 and 3.6 wt% Al2O3. The magnetitecontains 25 wt% TiO2 and 6 wt% MgO. Ilmenite did appear in thecalculated crystallization products. Using buffer conditions rang-ing from nickel–nickel oxide to hematite–magnetite made littledifference to the calculated assemblages, or oxide compositions.Because of the high TiO2 content in the magnetite its crystalliza-tion inhibited the formation of ilmenite in the MELTS programme.Differences in minor oxide content between observed compo-sition of clinopyroxene and magnetite with those predicted byMELTS may reflect subsolidus re-equilibration or uncertainties inthe modelling parameters for such unusual compositions for whichonly limited experimental data may be available. In summary,we suggest that the type of Ti-, Fe-basaltic magma compositionsidentified by Frick and Walraven (1985) could have crystal-lized the mineral assemblage observed in this clinopyroxenitebody.

In view of the possibility of exsolution and compositional re-equilibration of phases during slow cooling, it is relevant to returnto the abundant oxide lamellae seen in growth zones in the clinopy-roxene (Fig. 3). Their abundance makes it impossible for them tobe the product of exsolution, and so they probably reflect epitax-ial adherence or heterogeneous nucleation of elongate oxide grainsonto the crystal faces of clinopyroxene grains. Variations in the den-sity of these lamellae may relate to fluctuations in oxygen fugacityor nucleation rates.

The trace-element abundances in clinopyroxene, the clinopy-roxenites and these lavas can also be compared with referenceto known partition coefficients. The abundance of Sr is plottedagainst CaO in Fig. 7. In the clinopyroxenites, clinopyroxene andapatite are the only phases that contain CaO. Olivine and oxidescontain no CaO. Hence, the whole-rock clinopyroxenite composi-tions fall within a triangle defined by the contents in clinopyroxeneand apatite (Tables 2 and 6) and the origin (point for oxideand olivine). The Sr concentrations of 180–321 ppm (Table 6)are extremely high for clinopyroxene. However, the lavas fromwhich we suggest they were derived also contain extremely highSr contents of 1500–1900 ppm (Table 7), even for alkaline mag-mas. The partition coefficient for Sr between clinopyroxene andliquid is in the order of 0.13 (Bédard, 1994). Thus liquids with1500–1900 ppm Sr should crystallize clinopyroxene with 195–250,exactly as observed.

The good agreement between the observed mineral assem-blage (and mineral compositions) in these clinopyroxenites andthe calculated liquidus phases for these adjacent lava composi-tions suggest that such a magma could have been the parent tothese clinopyroxenite bodies studied here. The close juxtapositionof lavas and their cumulates as inferred for this Elandskraal Volcanois quite unusual, but supports the concept that the two rock typesare consanguineous.

The only other magma compositions (on Earth) with compa-rable high TiO2 and Fe2O3 that we know of are the immiscibleliquids from alkaline lavas described by Philpotts (1982), but theseare low in MgO (having formed near the end of fractionation)and could not produce the high mg# olivine and clinopyroxenereported in Tables 2 and 3. The close agreement between observedlava and its possible cumulate products is consistent with theserocks being part of the Pilanesberg Alkaline Province, but the gen-esis of such high Ti and Fe alkali basalt magma remains to beinvestigated. The lava samples investigated by Frick and Walraven(1985) were from borecore (at present, untraceable), and so fur-ther study on these distinctive chemical compositions may not bepossible.

Fig. 7. Plot of CaO versus (a) Sr, (b) Zr and (c) Ce for clinopyroxenites from thepresent study (solid dots) contrasted with data from the IRUP of the Bushveld Com-plex (shown as open squares from Table 1, and solid squares from Scoon and Mitchell(1994)). The open circles denoted Cpx(E) in (a) refer to the specific clinopyroxeneanalyses reported in Table 6; the solid circles denoted Cpx (E) and Cpx (IRUP) in(b) and (c) refer to probable compositions of clinopyroxene from this study (Eland-skraal) and Bushveld Complex (IRUP), respectively, required to produce the observedwhole-rock concentrations. Abbreviations: Ol, Mgt, Ilm and Ap—olivine, magnetite,ilmenite and apatite. The magnetite and ilmenite in this study probably containsome Zr, so their exact compositions are uncertain and not shown in (b). In (a) theapatite has 2000–4000 ppm Sr (see Table 6) and over 50% CaO, and so is not plotted.Likewise, in (c) apatite is assumed to have very high Ce (by comparison with of thevery high Sm and Nd contents in Table 6), and there are two whole-rock composi-tions (with high P2O5 contents) that have very high Ce (197 and 236 ppm Ce), andare not plotted.

Author's personal copy

34 R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36

Table 6Rb–Sr and Sm–Nd isotope results.

Sample Minerala 87Sr/86Sr ±2SEa Rb (ppm) Sr (ppm) ±2SE 87Sr/86Sr(i)b 143Nd/144Nd ±2SE Sm (ppm) ±2SE Nd (ppm) ±2SE 147Sm/144Nd

BL1780 ap 0.704958 (17) 1.8 1999.8 (0.4) 0.70491BL1780 cpx 0.708144 (20) 0.1 205.8 (0.1) 0.70812BL1867 ap 0.704831 (20) 0.5 4400 (9) 0.70483 0.511732 (7) 328.4 (0.04) 2034.8 (0.7) 0.0976BL1867 cpx 0.704820 (20) 0.04 321.1 (0.5) 0.70481 0.511837 (10) 18.5 (0.008) 102.4 (0.15) 0.1093BL1879 ap 0.704921 (17) 1.5 3840 (2) 0.70490 0.511762 (7) 411.9 (0.082) 2501.4 (0.2) 0.0996BL1879 cpx 0.705151 (20) 0.09 180.0 (0.4) 0.70513 0.511988 (11) 15.0 (0.002) 70.3 (0.006) 0.1291

a ap = apatite, cpx = clinopyroxene and SE = standard error from within-run precision.b Approximate initial ratio (1207 Ma) calculated using ICP-MS Rb concentrations (see text).

8. Comparison with other clinopyroxenite bodies of thePilanesberg Alkaline Province

There have been several attempts to determine the age of thePilanesberg Alkaline Province, all producing ages with very largeuncertainty. K–Ar dating of biotite has produced ages of 1250 ± 50and 1250 ± 60 Ma (Snelling, 1963), and of whole rock 1193 ± 98 Ma(Harmer, 1992). A SHRIMP U–Pb age on titanite of 1397 ± 47 Ma wasreported by Hanson et al. (2006). Our attempt to date eudialyte bythe U–Pb method failed because of the very high common lead con-tent (of up to 750 ppm). Our X-ray diffraction study of this mineralproved very inconclusive, suggesting that it may be metamict. Anumber of “unpublished” dates for members of the Pienaars RiverComplex were included in the compilation by Hanson et al. (2006)using U–Pb on zircon yielding ages from 1395 to 1334 Ma, but allwith uncertainties in the order of tens of m.y. The age obtained hereof 1207 ± 200 Ma suggests that this clinopyroxenite belongs to thePilanesberg Alkaline Province.

Clinopyroxenites have been reported from the Pilanesberg Alka-line Province, but the Pilanesberg Complex itself is composed oftrachytic to phonolitic lavas that were intruded by syenites andfoyaites, concentrically arranged in a near-perfect circle, 25 kmacross (c.f. Verwoerd, 2006, and references therein). It contains norocks with a significant mafic component, except for some veryminor aegirine-rich foyaites. There are a large number of dykesthat partially radiate from the central body, but are mainly ori-entated north-northwest and that can be traced for up to 100 kmbeyond the central complex. They are assumed to be part of thePilanesberg Province based on their ages by Rb–Sr dating of biotiteof 1290 ± 180 and 1330 ± 80 Ma (Schreiner and van Niekerk, 1958;van Niekerk, 1962) found in the Witwatersrand basin some 100 km

Table 7Composition of lavas used to model formation of clinopyroxenite.

Sample No. 30 47 47 + 4% Ol

SiO2 34.33 37.49 39.04TiO2 7.34 7.79 7.78Al2O3 8.19 6.33 6.32Fe2O3 (T) 18.10 21.80 22.04MnO 0.33 0.51 0.51MgO 6.92 6.09 7.71CaO 12.55 11.93 11.92Na2O 1.14 1.71 1.70K2O 4.69 2.47 2.46P2O5 1.21 1.29 1.28H2O 1.81 1.22 0CO2 4.16 2.10 0Sr 1951 1496Rb 170 83Ba 713 535Zr 598 555Y 57 51Ni 74 126

Sample numbers refer to those of Frick and Walraven (1985). The sample 47 + 4%Ol is sample 47 plus 4% olivine from Table 3, and the analysis, without volatiles,recalculated to 100%.

to the southeast. Some of these dykes are more mafic, some beingalkali gabbros, with olivine, purple titanian clinopyroxene, oxides,biotite and both plagioclase and alkali feldspars. Chemical analysesfrom this dyke swarm were presented by Ferguson (1973).

A further suite of alkaline complexes lies to the east of thePilanesberg along the Franspoort Line or Pienaars River Sub-province Lineament (F in Fig. 1). These suites intrude the mafic andfelsic parts of the Bushveld Complex. They again consist of felsicalkaline volcanic and intrusive members. Rb–Sr dating of eight ofthese complexes by Harmer (1985) produced a range of ages from1430 to 1300 Ma. The initial 87Sr/86Sr value obtained by him onthe entire suite of all samples was 0.7047 ± 2, although individualcomplexes gave a larger uncertainty. This value is very close to theaverage value of 0.7049 obtained for apatite in this study (Table 6),and suggests minimal crustal involvement. However, it should benoted that the Sr content of many lavas analysed by Harmer (1985)and discussed above (Table 7) exceed 1000 ppm, and such high con-centrations would minimise the effect on the initial isotopic ratioof any crustal assimilation.

The Spitskop Complex (Fig. 1) is geographically very remotefrom the rest of the Pilanesberg Alkaline Province, but because ithas been dated at 1341 ± 37 Ma, and consists of clinopyroxenites,ijolites, nepheline syenites and carbonatites (Harmer, 1999), it isconsidered part of the Pilanesberg Alkaline Province (Verwoerd,2006). Most of these rocks have been variably fenitised. Theclinopyroxenites contain variable proportions of nepheline andphlogopite, and so have higher abundances of K2O and Ba, muchgreater than in the present suite.

Some 50 km to the northwest of the Pienaars River Lineamentoccur four complexes, three of which are predominantly carbon-atites (Verwoerd, 2006). A fourth, the Buffelskraal Complex (Fig. 1),lies along the extension of the Pienaars River Lineament and isagain only known from four shallow boreholes (Frick and Walraven,1986). It contains clinopyroxenites that are variably converted tobiotite and fenitised by a network of intrusive ijolites, and basalticlava with clinopyroxene phenocrysts. Some mineralogical simi-larities exist with the material from the present study. Likewise,drilling revealed a magnetite-rich hornblende clinopyroxenite con-sidered to be part of the Goudini Carbonatite Complex (Verwoerd,2008), but no further information on these occurrences is available.

9. Contrasts with other clinopyroxenite intrusions on theKaapvaal craton

9.1. Discordant bodies in the Bushveld Complex

Geographically in close proximity to the south and west of thisstudied body, and cross-cutting the layered rocks of the BushveldComplex, is a suite of rocks referred to as iron-rich ultramafic peg-matoids, or IRUP, reviewed by Scoon and Mitchell (1994). Theyconsist of irregular, but essentially sub-vertical, elongate bodies,with coarse-grained olivine, clinopyroxene and oxide phases. Theyhave not been independently dated, but are assumed to be inti-mately related to the crystallization of the Bushveld Complex at

Author's personal copy

R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36 35

2060 Ma (Walraven et al., 1990). We have analysed 4 samples ofIRUP together with the present suite to make comparison, andthe results are given in Table 1. Major oxides are very similar,both with diagnostic low Al2O3 and K2O, the former emphasizingthe absence of feldspar, and the latter characterizing the cumu-late nature of both suites. Since both suites of rocks are cumulates,they provide little direct information on the nature of the parentalmagma. However, the trace elements Zr, Sr, Ba and the LREE areall enriched by factors of 2–5 in the present suite compared toIRUP. These differences are demonstrated in Fig. 7. No trace ele-ment data are available for clinopyroxene mineral compositionsfrom the IRUP suite, but it can be seen that the mineral must havevery different Sr, Zr and Ce contents in the two suites of clinopy-roxenites. The clinopyroxene compositions found in the IRUP havea range of mg# values from 0.64 to 0.34 (Scoon and Mitchell,1994), and so are all significantly more evolved than those in thepresent suite. Hence, the lower concentrations of Sr, Zr and Ce in theIRUP cannot be related to different degrees of fractionation. Fun-damentally different trace-element abundances must have beenpresent in the parental magmas to the two suites of clinopyroxen-ites.

9.2. Phalaborwa Complex

The 2060 ± 0.5 Ma Phalaborwa Complex (Eriksson, 1989;Reischmann, 1995) consists of five distinct rock suites, namelydunite, clinopyroxenite, carbonatite, foskorite and syenite (Fig. 1).The relationships between these different rock types are not wellunderstood, and are generally considered to be discrete intrusiveevents (Eriksson, 1989). The clinopyroxenite contains variable pro-portions of phlogopite and apatite, but no olivine or oxide. They aremore magnesian and have a higher initial 87Sr/86Sr value (greaterthan 0.711) than the samples in the present study. Further, the TiO2and Cr contents of the Phalaborwa clinopyroxenes are less than 0.3%and close to detection limit, respectively, and so are very differentfrom those reported here.

10. Conclusions

A clinopyroxenite body has been identified in borecore, overlainby sedimentary Karoo-aged rocks in the middle of the Kaapvaal cra-ton, South Africa. It consists of clinopyroxene, magnetite, ilmenite,apatite and minor altered olivine, and contains no felsic phaseat all. Chemical compositions suggest that it is a cumulate rock.Attempts have been made to determine its age using the Nd–Smmethod on mineral separates of clinopyroxene and apatite. Anage of 1207 ± 200 Ma has been obtained, with an C-- Nd value of −2.The minerals also gave an initial Sr isotopic ratio of 0.7049. Theage corresponds to the period during which the Pilanesberg Alka-line Province was emplaced. Other clinopyroxenite bodies havebeen reported from same general area of the Pilanesberg Alka-line Province, but little information exists about them. The muchmore distant Spitskop Complex has been documented by Harmer(1999). These other bodies are generally associated with a car-bonatite suite. In the present instance, alkali basaltic lavas thatshow carbonate alteration have been reported in adjacent bore-holes, but no carbonatite has been suggested. These lavas haveextremely unusual compositions in having high TiO2 (7%), Fe2O3(up to 22%), and high Sr, Ba and Zr. Of major relevance is that theMELTS programme predicts that such compositions would producethe liquidus assemblage of olivine, clinopyroxene, apatite and mag-netite, exactly matching the observed cumulate mineralogy in theclinopyroxenite body. The high Sr content in clinopyroxene also fitswith crystallization from this distinctive magma type.

Acknowledgements

We thank Joe Aphane for the mineral separations, and AngloPlatinum (especially Gordon Chunnett) for access to borecore andpermission to publish these findings. Dr Christian Reinke (Univer-sity of Johannesburg) helped with microprobe analyses. The Councilfor Geoscience provided the aeromagnetic data. RGC acknowledgesthe sponsorship from Anglo Platinum, Impala Platinum and Lon-plats, and financial support from the National Research Foundation,South Africa. Lyn Whitfield and Di du Toit drew some of the dia-grams. We thank two referees for their significant contributions.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.precamres.2011.12.016.

References

Atkins, F.B., 1969. Pyroxenes of the Bushveld Intrusion, South Africa. J. Petrol. 10,222–249.

Bédard, J.H., 1994. A procedure for calculating the equilibrium distribution of traceelements among minerals of cumulate rocks, and the concentration of traceelements in the coexisting liquids. Chem. Geol. 118, 143–153.

Buchanan, P.C., Koeberl, C., Reimold, W.U., 1999. Petrogenesis of the DullstroomFormation, Bushveld magmatic province, South Africa. Contrib. Mineral. Petrol.137, 133–146.

Elkins-Tanton, L.T., Draper, D.S., Agee, C.B., Jewell, J., Thorpe, A., Hess, P.C., 2007.The last lavas erupted during the main phase of Siberian flood volcanicprovince: results from experimental petrology. Contrib. Mineral. Petrol. 153,191–209.

Eriksson, S.C., 1989. Phalaborwa: a saga of magmatism, metasomatism and miscibil-ity. In: Bell, K. (Ed.), Carbonatites: Genesis and Evolution. Unwin Hyman, London,pp. 21–254.

Ferguson, J., 1973. The Pilanesberg Alkaline Province, Southern Africa. Trans. Geol.Soc. S. Afr. 76, 249–269.

Frick, C., Walraven, F., 1985. The petrology and geochemistry of the pre-Karoo Eland-skraal Volcano, South Africa. Trans. Geol. Soc. S. Afr. 88, 225–243.

Frick, C., Walraven, F., 1986. The mineralogy, petrology and geochemistry of theBuffelskraal alkaline complex south of Thabazimbi. Abstr. Geocongress’86. Geol.Soc. S. Afr., 803–807.

Ghiorso, M.S., Hirschmann, M.M., Reiners, P.W., Kress, V.C., 2002. A revision of MELTSaimed at improving calculation of phase relations and major element partition-ing involved in partial melting of the mantle at pressures up to 3 Gpa. Geochem.Geophys. Geosyst. 3, doi:10.1029/2001GC000217.

Hanson, R.E., Harmer, R.E., Blenkinsop, T.G., Bullen, D.S., Dalziel, I.W.D., Gose, W.A.,Hall, R.P., Kampunzu, A.B., Key, R.M., Mukwakwami, J., Munyanyiwa, H., Pancake,J.A., Seidel, E.K., Ward, S.E., 2006. Mesoproterozoic intraplate magmatism in theKalahari Craton: a review. J. Afr. Earth Sci. 46, 141–167.

Harmer, R.E., 1985. Rb–Sr isotopic study of the Pienaars River Alkaline Complex,north of Pretoria, South Africa. Trans. Geol. Soc. S. Afr. 88, 215–224.

Harmer, R.E., 1992. The geochemistry of Spitskop and related alkaline complexes.Ph.D. Thesis, University of Cape Town, 287 pp.

Harmer, R.E., 1999. The petrogenetic association of carbonatite and alkaline mag-matism: constraints from the Spitskop Complex, South Africa. J. Petrol. 40,525–548.

Lennoir, M., 2004. A petrological and geochemical study of an iron-rich ultramaficpegmatite, Western Bushveld Complex. Unpubl. Hons thesis, University of theWitwatersrand, 40 pp.

Merlet, C., 1994. An accurate computer correction program for quantitative electronmicroprobe analysis. Mikrochim. Acta 114–115, 363–376.

Philpotts, A.R., 1982. Compositions of immiscible liquids in volcanic rocks. Contrib.Mineral. Petrol. 80, 201–218.

Reischmann, T., 1995. Precise U/Th age determination with baddeleyite (ZrO2), acase study from the Phalaborwa Igneous Complex, South Africa. S. Afr. J. Geol.98, 1–4.

Schreiner, G.D., van Niekerk, C.B., 1958. The age of a Pilanesberg dyke from the centralWitwatersrand. Trans. Geol. Soc. S. Afr. 62, 198–199.

Scoon, R.N., Mitchell, A.A., 1994. Discordant iron-rich ultramafic pegmatites in theBushveld Complex and their relationship to iron-rich intercumulus and residualliquids. J. Petrol. 35, 881–917.

Snelling, N.J., 1963. Age determination unit. Annual Report of the Overseas Geolog-ical Survey, 30 pp.

South African Committee for Stratigraphy, 1980. Stratigraphy of South Africa. Part1 (comp. L.E. Kent). Handbook Geological Survey of South Africa. Pretoria 8, 690pp.

Shand, S.J., 1923. The alkaline rocks of the Franspoort Line, Pretoria District. Trans.Geol. Soc. S. Afr. 24, 81–100.

Thompson, R.N., Gibson, S.A., Mitchell, J.G., Dickin, A.P., Leonardos, O.H., Brod, J.A.,Greenwood, J.C., 1998. Migrating Cretaceous-Miocene magmatism in the Serra

Author's personal copy

36 R.G. Cawthorn et al. / Precambrian Research 198– 199 (2012) 25– 36

do mar alkaline province, SE Brazil: melts from the deflected Trinidade mantleplume? J. Petrol. 39, 1493–1526.

van Niekerk, C.B., 1962. The age of the Gemspost dyke from the Venterspost goldmine. Trans. Geol. Soc. S. Afr. 65, 105–111.

Verwoerd, W.J., 2006. The Pilanesberg Alkaline Province. In: Johnson, M.R.,Anhaeusser, C.R., Thomas, R.J. (Eds.), The Geology of South Africa. GeologicalSociety of South Africa, Johannesburg, pp. 81–394.

Verwoerd, W.J., 2008. The Goudini Carbonatite Complex, South Africa: a reappraisal.Can. Mineral. 46, 825–830.

Walraven, F., Armstrong, R.A., Kruger, F.J., 1990. A chronosratigraphic frameworkfor the north-central Kaapvaal Craton, the Bushveld Complex and Vredefortstructure. Tectonophysics 171, 23–48.

Woolley, A.R., 2001. Alkaline Rocks and carbonatites of the World. Part 3: Africa.Geol. Soc. Lond., 372 pp.