Unraveling the record of successive high grade events in the Central Zone of the Limpopo Belt using Pb single phase dating of metamorphic minerals (original) (raw)

Unraveling the record of successive high grade events in the Central Zone of the Limpopo Belt using Pb single phase dating of metamorphic minerals

L. Holzer a,∗{ }^{\mathrm{a}, *}, R. Frei a{ }^{\mathrm{a}}, J.M. Barton, Jr b{ }^{\mathrm{b}}, J.D. Kramers a{ }^{\mathrm{a}}
a{ }^{a} Gruppe Isotopengeologie, Min. Pet. Inst., Universität Bern, Erlachstrasse 9a, 3012 Bern, Switzerland
b{ }^{\mathrm{b}} Department of Geology, Rand Afrikaans University, P.O Box 524, Auckland Park 2006, Johannesburg, South Africa

Received 28 November 1996; accepted 3 October 1997

Abstract

Dating of relic metamorphic assemblages can provide important information about the timing and character (metamorphic grade and/or PT-evolution) of early high grade episodes in polymetamorphic provinces. Using Pb stepwise leaching of metamorphic silicates, we have dated multiple granulite facies metamorphic episodes in the Central Zone (CZ) of the Limpopo Belt. Ages of 2.52 Ga were obtained from sillimanite and cogenetic garnet and ages of about 2.01 Ga from titanite, garnet and clinopyroxene. Together with new and published conventional age data from accessory phases and in the context of combined petrological and structural data, these results lead us to a reinterpretation of the tectono-metamorphic history of the CZ. Three distinct high grade events at about 3.2−3.1Ga3.2-3.1 \mathrm{Ga}, 2.65−2.52Ga2.65-2.52 \mathrm{Ga} and 2.0±0.05Ga2.0 \pm 0.05 \mathrm{Ga} are recognized. Each of these is suggested to correspond to a tectonic episode of distinct character: (a) for the 3.2 Ga event magmatic activity can mainly be identified (best represented, for example, by the Sand River Gneisses or the Messina Layered Intrusion). The field relationships concerning the tectonometamorphic history of this Early-Archean event are largely erased by at least two high grade metamorphic overprints. (b) Late-Archean ( ∼2.6−2.52Ga\sim 2.6-2.52 \mathrm{Ga} ) low pressure granulite facies metamorphism was associated with voluminous granitic and charnockitic plutonism. The anticlockwise P-T evolution of these granulites probably reflects deep crustal processes, associated with magmatic underplating (or in-plating), contemporaneous with vertical crustal growth of the Zimbabwe craton around 2.6 Ga . © During the Proterozoic event ( ∼2.05−1.95Ga\sim 2.05-1.95 \mathrm{Ga} ) tectonic thickening was caused by the collision of the Kaapvaal and Zimbabwe cratons. The CZ was squeezed between these two cratons and as a consequence underwent high pressure granulite facies metamorphism with a clockwise P−T\mathrm{P}-\mathrm{T} evolution. The structural, metamorphic and geochronological data can be best explained with a tectonic model that describes this final event as a dextral transpressive orogeny. © 1998 Elsevier Science B.V.

Keywords: Archean; Geochronology; Granulite facies; Limpopo Belt; Palaeoproterozoic; Pb isotopes; Polymetamorphic

[1]

1. • Corresponding author. Tel: 00413163185 33; Fax: 00413163149 88; e-mail: lorenzacmpi.unibe.ch
     ↩︎

1. Introduction

Granulite facies conditions may be reached in many different tectonic environments. Simplifying the large variety of granulite provinces, two endmembers can be distinguished with respect to the tectonic environment and the associated metamorphic character (e.g. Harley, 1989; Percival, 1990): (a) belts that underwent high pressure granulite metamorphism ( ⩾10kbar\geqslant 10 \mathrm{kbar} ) are characterized by clockwise pressure temperature ( P−T\mathrm{P}-\mathrm{T} ) evolution; and (b) belts that underwent low to medium pressure granulite facies metamorphism are usually characterized by an anticlockwise P-T evolution. The first type of granulites is considered to form by tectonic crustal thickening. After peak metamorphic conditions have been reached, a first phase of near isothermal decompression is in many cases initiated by a change to an extensional tectonic regime. Post-collisional exhumation of such buried granulites is the result of erosion following rapid uplift and leads to isostatic equilibration of the crust. Many of the large Archean granulite terranes pertain to the second type with low/medium pressure granulites. Their association with voluminous mafic and (charno-) enderbitic intrusions suggests mantle derived magmas as an important heat source for their high T metamorphism (e.g. Bohlen, 1987). Exhumation of these granulites usually occurs in distinct, later tectonic cycles. These features have characteristics similar to those in recent continental arc environments, but can in general hardly be explained with alpinetype tectonic processes.

In testing possible models of the formation of granulite belts, the reconstruction of the P−T−t\mathrm{P}-\mathrm{T}-\mathrm{t} evolution is of major importance. Therefore combined approaches involving metamorphic petrology and geochronology are necessary. A problem in such studies is that petrological data can often not be easily linked to conventional geochronological data. Whereas thermobarometry relies on thermodynamic equilibrium of major elements between metamorphic minerals, most geochronological data depend on the retrograde closing behaviour of radiogenic trace element systems of accessory minerals. Dating of minerals involved in metamorphic reactions can now be
used to overcome this problem. Age data from metamorphic minerals can be correlated with petrologically deduced temperature estimates in cases where the analysed mineral is a product of an observed metamorphic reaction and the closure temperature ( TcT_{\mathrm{c}} ) is higher than the temperature at which the reaction takes place ( Tgrowth T_{\text {growth }} ). Mezger et al. (1989) for example dated the growth of garnet which formed from the breakdown of biotite during vapour-absent melting. Since P-T conditions for this type of reaction are well established, the age derived from garnet could be correlated with a narrow T range during the prograde metamorphism.

The study of granulite belts is further complicated by their frequent localization at craton boundaries or within intracratonic zones of weakness. Such zones are prone to reactivation during possible subsequent tectono-metamorphic events. This overprint by a second high grade metamorphic event could conceivably cause partial or complete resetting of chronometers and/or petrological thermobarometers. This may lead to wrong interpretations, particularly if chronology is based on different minerals (e.g. zircon) than those used for P T calculation (e.g. garnet, pyroxene, plagioclase), and the need to obtain age data from metamorphic minerals is thereby strongly emphasized. Then, the question of whether or not the radiogenic decay systems (e.g. U-Pb) remain closed during later high grade metamorphic overprint is important.

In this study we address these problems for the case of the Central Zone of the Limpopo Belt in Southern Africa. This province is suggested to have undergone at least two high grade metamorphic episodes, one prior to or synchronous with the Bulai intrusion at ca. 2.6 Ga (Watkeys, 1983) and one at ca. 2.0 Ga (Kamber et al., 1995b; Barton et al., 1994, Holzer et al., 1996). We have, on the basis of paragenesis and texture, selected samples with mineral assemblages judged to belong to the earlier and the later event and carried out Pb/Pb\mathrm{Pb} / \mathrm{Pb} stepwise leaching to constrain the age of metamorphic minerals. This method has been applied successfully to a wide range of silicates in metamorphic and hydrothermal environments: garnet, titanite, hornblende, clinopyroxene, epi-

dote (Frei and Kamber, 1995), staurolite (Frei et al., 1995) and tourmaline (Frei and Pettke, 1996). A multi-experimental study was undertaken by Frei et al. (1997) in order to assess the leaching mechanisms. The aims of our investigations are:
(1) To assess the retentivity of the U/Pb\mathrm{U} / \mathrm{Pb} systems in high grade metamorphic minerals, given a second high grade overprint.
(2) To delineate the P−T−t\mathrm{P}-\mathrm{T}-\mathrm{t} evolution of the youngest high grade event in the Central Zone and to obtain an accurate age for the earlier metamorphic episode.
(3) To compare qualitatively the P−T\mathrm{P}-\mathrm{T} loops characteristic of each of the two events.
(4) To discuss possible tectonic models for the different granulite facies events recorded in the Limpopo Belt in the light of the results.

2. Geological setting

The Limpopo Belt (Figs. 1 and 2) is a high grade metamorphic province which has an elongation of about 650 km (ENE-WSW) and a width of 200 km( N−S)200 \mathrm{~km}(\mathrm{~N}-\mathrm{S}), respectively. To the east it is cut by younger tectonic units. It covers parts of South Africa, Botswana and Zimbabwe and is wedged between two (mostly) Archean blocks: the Kaapvaal craton to the South and the Zimbabwe craton to the North. The margins of the Limpopo Belt are made up by two major thrust zones, along which granulite facies rocks and retrogressed granulites have been thrust onto the adjacent cratons during the late Archean, i.e. the Hout River shear zone (e.g. Smit et al., 1992) in the South and the North Marginal Thrust Zone (Blenkinsop et al., 1995; Mkweli et al., 1995) in the North. The Limpopo Belt itself is divided into three subzones: the Northern and Southern marginal zones (NMZ and SMZ) border the volumetrically dominant Central Zone (CZ). The marginal zones are considered as lower crustal equivalents of the adjacent granite greenstone terranes (e.g. Du Toit et al., 1983) which underwent granulite facies metamorphism during the late Archean. The CZ consists of a wide range of lithologies, including metamorphosed platform sediments (Beit Bridge Complex), quartzo-feldspathic gneisses, tonalitic grey gneisses
(Sand River and Alldays Gneisses) and other intrusive rocks with variable composition and age. The polymetamorphic character of the CZ is documented by the available age data and the complex structural pattern, both of which will be outlined below. The suture zones between the marginal zones and the CZ are defined by large shear zones: to the North the CZ is bounded by a set of ENE-WSW trending dextral strike slip shear zones, including the Triangle, Lepokole and Magogaphate shear zones. The Triangle shear zone represents a belt of transpressive dextral deformation which was active mainly under granulite facies conditions during the Proterozoic (2.0Ga(2.0 \mathrm{Ga}, Kamber et al., 1995a). The timing of movement in the Lepokole and Magogaphate shear zones is not yet determined. The kinematics at the southern boundary of the CZ is more complex. Almost the entire contact with the SMZ and the Kaapvaal craton is overlain by the Proterozoic Soutpansberg and Waterberg sediments. The high grade ‘Tshipise Straightening Zone’ (Bahnemann, 1972) is located to the North of the Soutpansberg trough (see Fig. 2). Foliations trend ENE-WSW, similarly to those in the Triangle shear zone, but they dip more steeply towards SSE. In addition, the southern boundary of the CZ is defined by the 10 km wide Palala lineament (McCourt, 1983; Brandl and Reimold, 1990). The shear sense of this mylonite zone has been a subject of controversy: McCourt and Vearncombe (1992) have interpreted the Palala shear zone as a sinistral strike slip zone. Broekhuizen and McCourt (1995) defined two phases of movement with opposing lateral shear senses. The youngest transcurrent faulting in the Palala shear zone post-dates both the exhumation of the CZ granulites and the emplacement of the Bushveld complex 2060 Ma ago (Barton, 1995).

3. A compilation of available age data from the CZ

Table 1 is a compilation of the most reliable isotope data, indicating an apparent history for the CZ ranging from 3.8 until 2.0 Ga . These dates cover a nearly continuous spread of ages between 3.2 and 2.0 Ga . Considering only conventional zircon U/Pb\mathrm{U} / \mathrm{Pb} concordia intercept ages from the CZ.

Fig. 1. Geological map of the Limpopo Belt and the adjacent Kaapvaal and Zimbabwe cratons: curved lines in Central Zone mark the trend of foliations. SMZ == Southern Marginal Zone, NMZ == Northern Marginal Zone, Phanerozoic cover is shown in white: Intrusions: 1=1= Entabeni Pluton, 2=2= Schiel Alkaline Complex, 3=3= Matok Pluton; Greenstone Belts (GB): 4=4= Rhenosterkoppic, 5=5= Southerland GB, 6=6= Bubwa GB, 7=7= Gwanda GB, 8=8= Umzingwane GB, 9=9= Belingwe GB; Proterozoic sedimentary basins: 10=10= Blouberg, 11=11= Waterberg, 12=12= Palapye; Localities: M=\mathrm{M}= Mashwingo, R=\mathrm{R}= Rutenga, T=\mathrm{T}= Tshipise.

Fig. 2. Geological profile across the Limpopo Belt. This geological interpretation is based on structural data (discussed in text) and on published geophysical profiles (De Beer and Stettler, 1992, and references therein). The trace of the profile is shown in Fig. 1. Signatures are consistent with those in Fig. 1. M=\mathrm{M}= Matok Pluton, E+S=\mathrm{E}+\mathrm{S}= Entabeni Granite and Schiel Alkaline Complex, B=\mathrm{B}= Bulai Pluton, R=\mathrm{R}= Razi granite.

Fig. 3. Histogram showing a compilation of published zircon concordia intercept ages from the CZ. Three peaks at 3.2, 2.6 and 2.0 mark three distinct geological events, which are discussed in the text. Labels correspond to references given in Table 1.
the data define three peaks at 3.2,2.63.2,2.6 and 2.0 Ga (Fig. 3).

Early-Archean in the CZ (ca. 3.2 Ga ): various isotopic data from the Sand River Gneisses, the Messina Layered Intrusion and the Zanzibar gneisses indicate important magmatic activity between 3.1 and 3.3 Ga (see Table 1). Structural relationships and metamorphic records are largely erased by later high grade tectono-metamorphic overprints. The pre-3.0 Ga geologic history of the CZ is thus still poorly understood.

Late archean in the CZ (2.7-2.55 Ga): the zircon data from Jaeckel et al. (1997) give evidence for two late Archean magmatic pulses. A first period occurred around 2.65 Ga (granodioritic Alldays Gneiss); a second important phase of granitoid plutonism is identified between 2.6 Ga and 2.55 Ga , during which quartzo-feldspathic Singelele orthogneisses were emplaced. Furthermore the Early-Archean Zanzibar gneiss was intruded by granitic orthogneisses at 2.55 Ga (Barton and Key, 1983). The Bulai Pluton intruded at 2.6 Ga (enderbitic phase) and 2.57 Ga respectively (granitic phase; Barton et al., 1994).

Proterozoic in the CZ (2.05-1.95 Ga): the recognition of a 2 Ga granulite metamorphism in the Triangle shear zone (Kamber et al., 1995b) initiated a detailed study of the Messina area during which it was recognized that this area was also affected by important Proterozoic metamorphism (Barton et al., 1994; Holzer and Kamber, 1995; Holzer et al., 1996). U-Pb data from meta-
morphic zircons (Jaeckel et al., 1997; Barton and Sergeev, 1997) indicate that high grade metamorphic conditions were reached between 2.06 and 2.03 Ga . Rb-Sr ages of ca. 1.97 Ga from biotite (Barton and van Reenen, 1992) and 1.92 Ga from retrograde muscovite (Barton et al., 1994) reflect the time of cooling in the CZ.

These three periods with increased magmatic and tectono-metamorphic activity during the Archean and early Proterozoic are largely confirmed by new data from Kröner et al. (submitted).

4. Structural patterns in the CZ and their relations to metamorphic episodes

The Messina-Beitbridge region is a classical study area, situated in the middle of the CZ (Fig. 1). This study area exemplifies a complicated, polyphase deformational history of the CZ, which is summarized below (see also Fig. 4 and Table 6). A more detailed description of the structures is given by Watkeys (1983, 1984).

For an interpretation of the structural evolution of this area the Bulai intrusion is an important time marker. This calc-alkaline pluton was emplaced at 2572±4Ma2572 \pm 4 \mathrm{Ma} (Barton et al., 1994) and reveals the following magmatic contact relationships: it intrudes a crystalline basement, consisting mainly of metamorphosed and multiply folded paragneisses (Beit Bridge Complex). The foliation of metapelitic xenoliths within the Bulai granite is

Table 1
Compilation of age data from the CZ

Table 1(continued)

ap =apatite, bt=\mathrm{bt}= biotite, cpx=\mathrm{cpx}= clinopyroxene, fsp=\mathrm{fsp}= feldspar, grt=\mathrm{grt}= garnet, hbl=\mathrm{hbl}= hornblende, ilm=\mathrm{ilm}= ilmenite, ms=\mathrm{ms}= muscovite, pl=\mathrm{pl}= plagiociase, sill=\mathrm{sill}= sillimanite, wr=\mathrm{wr}= whole rock, zrn=\mathrm{zrn}= zircon.

Fig. 4. Geological map of the study area between Beitbridge and Tshipise in the CZ. The structural features within the Bulai Pluton. the Shanzi and Campbell folds and the Tshipise Straightening Zone are discussed in the text. Numbers indicate sample localities.
defined by granulite facies mineralogy (grt. sill. crd. bt) and is discordant to the foliation in the magmatic host rock. This implies that deformation under granulite facies conditions occurred prior to or during the emplacement of the Bulai Pluton. This intrusion also cuts migmatite structures in the metasediments. Locally, granodioritic Bulai
melt seems to be mingled with granitoid Singelele type melt (Holzer, 1995). The Singelele orthogneisses are interpreted as products of widespread anatexis and mobilization of leucosome which was contemporaneous with the Bulai emplacement. These field observations lead to the following temporal relationships: (a) the Bulai pluton post-

dates high grade metamorphism (M1) which is associated with ductile shearing and folding (D1); (b) the emplacement of the Bulai pluton is contemporaneous with migmatization (M2a) and occurred syntectonically (D2a). The emplacement age of 2.57 Ga of the Bulai pluton is interpreted as a minimum age for the first granulite facies episode in the CZ.

Several phases of high grade deformations postdate the emplacement of the Bulai Pluton. The entire Messina-Beit Bridge area exhibits highly ductile and multiple phase D3-deformations.

The main mechanism of deformation within the Bulai Pluton is ductile shearing. ‘Miniature mobile shear belts’ (Watkeys, 1984) with variable orientation, size and shear sense are the earliest D3 deformations recorded within this pluton. Conjugate sets of shear zones-also with variable orientation and shear sense-developed subsequently. The early D3 deformations in the Bulai Pluton seem to be the product of a NNW-SSE directed shortening ( σ1\sigma 1 ). The latest so-called ‘Limpopo trend shears’ (ENE-WSW) within the Bulai Pluton exhibit mainly dextral shear senses.

A distinct style of D3-deformation evolved in the well layered paragneiss sequences. Locally, ENE-WSW directed compressive stress regimes evolved and a flexural slip mechanism led to the formation of regional scale upright, isoclinal ‘crossfolds’ (F3a). In the Messina area two important F3 folds have been described: the Campbell and the Shanzi structures. The latter is strongly affected by F3b deformations, during which earlier crossfolds have been refolded. This produced the ‘top to the ENE’-shearing observed on Mount

Shanzi (one of our sample localities). F3b foldaxes and lineations in the Messina area are subparallel and dip with about 30∘30^{\circ} towards 260∘260^{\circ}. The contact of the Campbell and Shanzi fold structures with the Bulai body consists of highly ductile shear zones. Undeformed leucosome veins crosscut these D3 shear zones and they occur both in hinges of F3b secondary folds within the Campbell structure, and as post-tectonic mobilizates within the Sand River Gneisses at the Causeway locality (sampled by Jaeckel et al., 1997).

The section from Messina to Tshipise is characterized by a continuous transition into the Tshipise Straightening Zone, in which monotonous ENEWSW trending foliations dip steeply towards SSE. Fold axes and lineations dip moderately towards WSW, subparallel to the F3b-lineations in the Messina area. Fripp (1983) describes two sets of ductile shear zones trending towards 010 and 080∘080^{\circ}. They led to a clockwise rotation of older structures into the dominant ENE-WSW structural trend, suggesting a dextral simple shear component for the area south of Messina.

Vapour absent melting, observed in a multitude of lithologies (including the Bulai body), indicates that granulite facies conditions were reached during M3. The youngest granulite event (M3) is characterized by a clockwise P-T path. Spectacular reaction textures in metapelites have been the subject of various metamorphic studies (Chinner and Sweatman, 1968; Droop, 1989; Horrocks, 1983a; Windley et al., 1984). Peak conditions are constrained to 825∘C(±25∘C)825^{\circ} \mathrm{C}( \pm 25^{\circ} \mathrm{C}) and 10−12kbar10-12 \mathrm{kbar}. After a phase of near isothermal decompression (ITD) pressures of 5−6kbar5-6 \mathrm{kbar} were reached. During

Table 2
Modal mineralogical composition

[1]

1. Abbreviations: aln = allanite, ap = apatite, bt = biotite, crd = cordierite, cpx = clinopyroxene, grt = garnet, hbl = hornblende, kfs = K feldspar, mag = magnetite, mon=monazite, pl=\mathrm{pl}= plagioclase, qtz=quartz, rtl=\mathrm{rtl}= rutile, scp=scapolite, sill = sillimanite, spl = spinell, tit=\mathrm{tit}= titanite, zrn=\mathrm{zrn}= zircon. ↩︎

Table 3
Age data. Pb isotopic compositions and experimental data of leach experiments

Table 3 (continued)

Abbreviations: ap = apatite, cpx = clinopyroxene, fsp = feldspar, gtt = garnet, mon = monazite, sill = sillimanite, tit = titanite r1,r2=r 1, r 2= correlation coefficients according to Ludwig (1988).

the subsequent isobaric cooling gedrite was locally formed at the expense of orthopyroxene. This initial rehydration occurred at about 700−800C700-800 \mathrm{C} and 5−6kbar5-6 \mathrm{kbar} (Hisada and Miyano, 1996).

5. Samples and sample localities

From structural and regional criteria, based on the above framework, samples whose metamorphic parageneses were formed specifically in either the D2 M2 or the D3/M3 event were collected, and these were selected for Pb stepwise leaching (PbSLdating) of metamorphic minerals. Modal mineralogical compositions are listed in Table 2 and the sample localities are shown in Fig. 4.

Sample 93 048: 1.5 km NE of ‘Three Sisters’ (Farm Boston). Strongly elongated, up to several km long slices of leucogranitic orthogneiss and metapelite occur within the Bulai Pluton. In one of these metapelitic xenoliths ( 93048 ), rectangular sillimanite pseudomorphs after andalusite (probably chiastolite) are up to 1 cm in diameter. Between the coarse grained, perfectly oriented sillimanite needles ( 84%84 \% ) cogenetic cordierite and garnet occur as interstitial phases ( 15%15 \% ) within the pseudomorphic aggregates. Rutile and hercynite are further accessory minerals. These pseudomorphic sillimanite aggregates are embedded in a fine-grained matrix consisting of cordierite, biotite, garnet, sillimanite and minor rutile, spinel and magnetite. The matrix-foliation anastomoses around the sillimanite pseudomorphs. Sillimanite in the matrix is aligned within the foliation. Garnet forms elongated, irregularly shaped grains and is partly broken. The youngest deformation recorded in this rock thus post-dates growth of both the coarse sillimanite-nests and matrix garnet. We interpret the pseudomorphic aggregates as relics that grew during an early phase of granulite metamorphism. As they are probably after andalusite, they might reflect the prograde phase of an early metamorphic episode.

Sample 93083 is a calesilicate gneiss, which was sampled close to the hinge of the Campbell fold structure. 11.5 km W of Messina (Farm Plaatje). This calesilicate layer was strongly attenuated during the formation of the ‘Campbell cross fold’.

The intense intergrowth of plagioclase, clinopyroxene and titanite in mafic bands parallel to the F3 axial planar surface give evidence that these minerals recrystallized synkinematically during the D3 event.

Sample 93 167: 4 km NW of Messina, on the NE-side of Mount Shanzi (Farm Uitenpas), a sequence from porphyritic Bulai gneiss at the base which ‘merges into a migmatitic zone, contaminated by assimilation of supracrustal xenoliths, overtopped by quartzofeldspathic gneisses and magnetite quartzites’ (Watkeys, 1984, p. 156ff.) is exposed. These paragneisses on top of the sequence mark the easternmost boundary of the D3-Shanzi fold structure. The degree of deformation within the Bulai granite increases towards the contact with the overlying metasediments. Shear bands indicate a transport direction ‘top to the ENE’. The mineral elongation lineations are subparallel to the fold axes and dip moderately towards WSW (ca. 265-35). These structures are ascribed to the D3 F3b deformation. The migmatitic zone at the contact itself is characterized by ductile shear zones with calesilicate enclaves. A slight compositional zoning is observed in these enclaves: the core of one ( 93167 core) consists of plagioclase, garnet, clinopyroxene, titanite and quartz (Table 2). Abundant green hornblende, an increased plagioclase content and lack of garnet and quartz characterize the bright, decimeter wide rim ( 93/167rim93 / 167 \mathrm{rim} ). The calesilicate enclaves have an internal compositional banding (mafic and felsic layers) parallel to the shear planes in the mylonitic host rocks. In the more mafic layers garnet is intergrown with plagioclase, titanite and cpx, indicating a strong synkinematic (D3) recrystallization. In the more felsic layers post-kinematic recrystallization produced coarse grained, polygonally shaped plagioclase.

6. Analytical techniques

Pb stepwise leaching (PbSL) was carried out on 93/048 sillimanite and garnet, 93/083 clinopyroxene, 93/167 (core) garnet and 93/167 (rim) titanite and followed the procedure described in Frei and Kamber (1995) with slight modifications. Table 3

summarizes the experimental parameters as well as the Pb isotope results. Subsequent to hand picking of the mineral separates under a binocular microscope, mineral leaching was performed in 7 ml Savillex ®{ }^{\circledR} screw-top beakers. Individual stepleach solutions were decanted and then dried on a hot plate. Pb was separated using conventional HCl−HBr\mathrm{HCl}-\mathrm{HBr} anion exchange procedures, during which the blank level was less than 130 pg . Pb was loaded on single Re-filaments and measured on an AVCO®90,35 cm\mathrm{AVCO}^{\circledR} 90,35 \mathrm{~cm} radius single collector mass spectrometer. Pb isotope ratios were corrected for fractionation using the values obtained from repetitive NBS SRM 981 runs under similar operating conditions ( 0.001frc/amu±4%0.001 \mathrm{frc} / \mathrm{amu} \pm 4 \%, n=10n=10 ). Isochrons and error correlations are calculated after York (1969), using Isoplot (Ludwig, 1994). Errors assigned to the isochrons are 2σ2 \sigma.

Conventional bulk U−Pb\mathrm{U}-\mathrm{Pb} analysis was performed on monazite and apatite from sample 93/167. Apatite was dissolved in 7 N HCl in a Savillex beaker for 3 h on a hot plate. Monazite was dissolved for 3 h in a mixture ( 3:13: 1 ) of 14 N HNO3\mathrm{HNO}_{3} and HF conc. A mixed 235U−205 Pb{ }^{235} \mathrm{U}-{ }^{205} \mathrm{~Pb} tracer was used for both. Chemical separation and measurement of Pb was the same as applied for PbSL . U was extracted using a conventional HCl−\mathrm{HCl}- HNO3\mathrm{HNO}_{3} anion exchange procedure and measured from a Ta−Re−Ta\mathrm{Ta}-\mathrm{Re}-\mathrm{Ta} triple filament configuration.

No absolute quantities of lead were determined in our leach experiments. The relative quantities released during the individual leach steps can be estimated from the signal intensities. Although the leaching behaviour is not the same for all host minerals, for most experiments highest signal intensities are obtained for the first step(s). The amount of lead released during the final steps is strongly dependent on the presence of microinclusions and on their leaching behaviour.

The lack of absolute amounts of Pb released during the single leach steps is a potential problem for the blank corrections. The laboratory total procedure blank is below the 130 pg level (generally 80 pg ) and its composition is unradiogenic (206 Pb:204 Pb:18.7±0.22σabs,207 Pb/204 Pb\left({ }^{206} \mathrm{~Pb}:{ }^{204} \mathrm{~Pb}: 18.7 \pm 0.22 \sigma \mathrm{abs},{ }^{207} \mathrm{~Pb} /{ }^{204} \mathrm{~Pb}\right. : 15.67±0.3,208 Pb/204 Pb:38.45±0.4)\left.15.67 \pm 0.3,{ }^{208} \mathrm{~Pb} /{ }^{204} \mathrm{~Pb}: 38.45 \pm 0.4\right). From signal intensities, quantities of Pb in fractions were always >20ng>20 \mathrm{ng}. therefore blank contributions were

Fig. 5. Pb isotopic diagrams of PbSL data. For each sample both, the uranogenic ( 207 Pb−204 Pb{ }^{207} \mathrm{~Pb}-{ }^{204} \mathrm{~Pb} versus 206 Pb204 Pb{ }^{206} \mathrm{~Pb}{ }^{204} \mathrm{~Pb} ) and the thorogenic ( 206 Pb204 Pb{ }^{206} \mathrm{~Pb}{ }^{204} \mathrm{~Pb} versus 206 Pb204 Pb{ }^{206} \mathrm{~Pb}{ }^{204} \mathrm{~Pb} ) Pb isotopic compositions are shown. Experimental data, isotopic compositions and mineral abbreviations are given in Table 3.
<1%<1 \%. This, and the close proximity of the blank composition to the regression lines, means that the ages calculated from the leach data are not perceptibly affected by the blank.

7. Results

The PbSL and U−Pb\mathrm{U}-\mathrm{Pb} isotope data are given in Tables 3 and 4. The Pb spectra from leach experiments are plotted in conventional uranogenic (207 Pb−204 Pb\left({ }^{207} \mathrm{~Pb}-{ }^{204} \mathrm{~Pb}\right. versus 206 Pb;204 Pb)\left.{ }^{206} \mathrm{~Pb} ;{ }^{204} \mathrm{~Pb}\right) ) and thorogenic ( 206 Pb−204 Pb{ }^{206} \mathrm{~Pb}-{ }^{204} \mathrm{~Pb} versus 206 Pb;204 Pb{ }^{206} \mathrm{~Pb} ;{ }^{204} \mathrm{~Pb} )) diagrams (Fig. 5(a)-(e)). U-Pb and Pb−Pb\mathrm{Pb}-\mathrm{Pb} data from apatite are plotted in Figs. 6 and 7.

93/048 matrix garnet: increasingly radiogenic Pb was recovered for the first steps (step 1 could not be measured due to a weak signal). After step 3 the radiogeneity again decreased and the least radiogenic Pb was measured for steps 6 and 7 ,
which indicates that leaching was nearly complete. Step 5 marks an irregularity in this pattern, because it is again more radiogenic than step 4. In the thorogenic plot (see Fig. 5(a)) this step lies below the garnet reference line, indicating the influence of a second, low Th/UPb\mathrm{Th} / \mathrm{U} \mathrm{Pb}-source. From our experience with stepleaching we suggest zircon microinclusions as possible contaminants. Zircon is a uranium rich phase which is only attacked efficiently by fluoric acid (step 5). In the uranogenic plot steps 2−4,62-4,6 and 7 define an isochron with an age of 2518±352518 \pm 35 (MSWD 1.42). All steps together, including step 5 , also define an isochron with a slightly higher age ( 2546±412546 \pm 41, MSWD 2.08). However, as discussed, step 5 may be dominated by zircon inclusions and we therefore consider the first isochron to give a more accurate age for garnet growth.

93/048 Sillimanite: a similar leach spectrum as for matrix garnet was obtained for sillimanite,

Fig. 5. (continued)
which forms aggregates pseudomorphic after andalusite. Increasingly radiogenic Pb was recovered as leaching progressed, and the less radiogenic Pb from the residue indicates that step leaching was nearly complete. Regression of all sillimanite steps did not result in an isochron (MSWD 44.1). Also, scattering values of the individual step solutions in the thorogenic plot indicate at least two sources of Pb . In the uranogenic plot steps 3,4 and bulk sillimanite and bulk associated garnet (intergrowth within the sillimanite aggregates) all plot slightly above the tie line between steps 1 and 2 . It is apparent from the thorogenic plot (208 Pb/204 Pb\left({ }^{208} \mathrm{~Pb} /{ }^{204} \mathrm{~Pb}\right. versus 206 Pb/204 Pb{ }^{206} \mathrm{~Pb} /{ }^{204} \mathrm{~Pb} ) that the four datapoints (steps 3,4 , bulk sillimanite, bulk associated garnet) are characterized by a highly uranogenic Pb component whereas leach steps 1 and 2 have more thorogenic Pb components. The data thus define two linear trends and the question arises which of them has age significance. Steps 3 and 4 (together with the associated bulk garnet) plot on a line with an apparent age of 2573±15Ma2573 \pm 15 \mathrm{Ma} (MSWD
1.59). Alternatively, the tie line between steps 1 and 2 has a slope equivalent to an age of 2524±5Ma2524 \pm 5 \mathrm{Ma}. The combination of both leach spectra (matrix garnet and sillimanite) gives evidence for the second approach: leach steps 1 and 2 of sillimanite together with all steps of matrix garnet (excluding step 5) define a perfect isochron with an age of 2521±4Ma(MSWD±1.432521 \pm 4 \mathrm{Ma}(\mathrm{MSWD} \pm 1.43 ), which is considered to reflect the time of both sillimanite (aggregates) and garnet (matrix) growth. The slightly older and more uranogenic Pb signatures of steps 3 and 4 from sillimanite are explained to be contaminated by Pb components from zircon microinclusions. The age of 2573±152573 \pm 15 is considered as a minimum age for these inclusions.

93/167 (core) garnet: very radiogenic Pb components were leached from garnet within the core of a calcsilicate enclave. Except for the residue, PbSL data almost define an isochron with an age of 2010±17Ma2010 \pm 17 \mathrm{Ma} (MSWD 2.24). The scatter in the thorogenic diagram suggests the presence of three different Pb components: the first two steps are

Fig. 5. (continued)
dominated by thorogenic Pb that probably derived from monazite inclusions (data points lie above the garnet reference line in the thorogenic plot). Their colinearity with other fractions in the uranogenic plot shows that these microinclusions are cogenetic with garnet. This is compatible with the age of 2011±20Ma2011 \pm 20 \mathrm{Ma} obtained for a monazite fraction from the same sample (see below). Steps 3 and 4 are interpreted to be dominated by garnet hosted Pb . The residue revealed a uranogenic Pb component (data point lies below the garnet reference line in the 208 Pb;204 Pb{ }^{208} \mathrm{~Pb} ;{ }^{204} \mathrm{~Pb} versus 206 Pb;204 Pb{ }^{206} \mathrm{~Pb} ;{ }^{204} \mathrm{~Pb} diagram), which we interpret to derive from zircon inclusions, attacked during the final dissolution in HF. These inclusions were not in initial isotope equilibrium with garnet as the residual data point lies above the isochron in the uranogenic diagram. Pb from bulk feldspar is in near isotope equilibrium with Pb from garnet, which is reflected by an errorchron ( 2003±18Ma2003 \pm 18 \mathrm{Ma}, MSWD 6.86), defined by garnet steps 1−41-4 and bulk feldspar. The published two point garnet-feldspar isochron of

Fig. 6. Conventional U-Pb isochron diagrams for five apatite fractions of sample 93/16793 / 167. The results are discussed in the text. Experimental data, isotopic compositions and mineral abbreviations are given in Tables 3 and 4.

Fig. 7. Concordia diagram for five apatite fractions of sample 93/16793 / 167. U-Pb isotopic data are given in Table 4.
2011±6Ma2011 \pm 6 \mathrm{Ma} (Holzer et al., 1996) is compatible with the above PbSL result. The position of bulk garnet relative to the garnet reference line in the thorogenic diagram implies that its signature is

more effectively dominated by Pb from cogenetic monazite than from older zircon inclusions. A minimum age of 2337±47Ma2337 \pm 47 \mathrm{Ma} for the zircon inclusions can be estimated from the feldspar-step 5 garnet tie line.

93/167 (rim) titanite: similarly to the above described garnet, PbSL of titanite from the rim of the calcsilicate enclave also released very radiogenic Pb components. The scatter of data in the thorogenic diagram again suggests the presence of three different Pb sources, i.e. host titanite and monazite- and zircon-inclusions: in Step 1 effectively all Pb was leached from cogenetic monazite inclusions, as shown by its very thorogenic Pb signature. In contrast, steps 3 and 4 are dominated by more uranogenic Pb (low 208 Pb/204 Pb{ }^{208} \mathrm{~Pb} /{ }^{204} \mathrm{~Pb} ratios relative to a titanite reference line). The final dissolution in HF (step 4) particularly revealed a predominant influence of Pb from zircon inclusions. Steps 1, 2 and bulk titanite define a perfect isochron of 2007±5Ma2007 \pm 5 \mathrm{Ma} (MSWD 0.05). This isochron is defined by two cogenetic Pb sources: titanite hosted Pb dominates step 2, monazite hosted Pb dominates step 1. In the bulk titanite the zircon source seems to be hardly expressed. In addition, near isotopic equilibrium with titanite is also indicated for monazite and feldspar ( 93/16793 / 167 core), which define an errorchron with an age of 2014±15Ma2014 \pm 15 \mathrm{Ma} (MSWD 10.2). For the zircon inclusions a minimum age of 2422±15Ma2422 \pm 15 \mathrm{Ma} can be calculated from the tie line between feldspar and ‘titanite’ residue.

Initial isotopic near-equilibrium for all measured minerals (titanite, garnet, monazite, apatite and feldspar) from core and rim of the calcsilicate enclave 93/16793 / 167 is demonstrated by a Pb−Pb\mathrm{Pb}-\mathrm{Pb} errorchron ( 2008±10Ma2008 \pm 10 \mathrm{Ma}, MSWD 13.6), which is defined by 14 datapoints. Only steps 3 and 4 of titanite, bulk and step 5 of garnet have been excluded. For these steps the influence of noncogenetic contaminants has been detected in the leach spectra.

93/167 apatite: the lead budget of all five apatite fractions is characterized by relatively high proportions of initially incorporated Pb (ca. 35%). The initial Pb isotopic compositions can be determined from conventional isochron diagrams for each decay system (Fig. 6). However, the five apatite
fractions only define errorchrons and thus the isotopic ratios determined by yy-axis intercepts in conventional isochron diagrams have large errors (206 Pb/204 Pb1:16.78±4.82πabs;207 Pb/204 Pb1\left({ }^{206} \mathrm{~Pb} /{ }^{204} \mathrm{~Pb}_{1}: 16.78 \pm 4.82 \pi \mathrm{abs} ;{ }^{207} \mathrm{~Pb} /{ }^{204} \mathrm{~Pb}_{1}\right. : 16.11±0.916.11 \pm 0.9 ). Within error these values are identical with the isotopic ratios of feldspar ( 93/16793 / 167 core). We therefore used the feldspar composition as a best approximate for the initial Pb component and justify the appropriate correction by the fact that apatite and feldspar are nearly in isotopic equilibrium. The slight scatter may be due to noncogenetic microinclusions. In thin sections for example zircon and monazite inclusions within apatite were observed.

The U−Pb\mathrm{U}-\mathrm{Pb} analyses of five apatite grain size fractions yielded two nearly concordant and three slightly reversely discordant data points (Fig. 7). All five fractions together define a discordia line (forced through 0±5Ma0 \pm 5 \mathrm{Ma} ) with an upper intercept at 1983±14Ma1983 \pm 14 \mathrm{Ma} (MSWD 2.0).

93/167 (rim) monazite: a U-Pb analysis of monazite yielded a concordant data point with a 207 Pb/206 Pb{ }^{207} \mathrm{~Pb} /{ }^{206} \mathrm{~Pb} age of 2011±20Ma2011 \pm 20 \mathrm{Ma} (Table 4). This age is consistent with the ages from garnet and titanite of the same sample and partly justifies our interpretation of the Pb leach spectra that monazite microinclusions are cogenetic with the host minerals.

93/083 clinopyroxene: only a small data spread was obtained by PbSL of diopside from calcsilicate 93/083. Due to a weak signal step 4 could not be analysed. PbSL data define an isochron of 1860±320Ma1860 \pm 320 \mathrm{Ma} (MSWD 0.329) (Fig. 5©). The linear array in the thorogenic diagram suggests an inclusion-free sample. Mixing two different components would result in a line only in the case where both sources are cogenetic and when they have an identical U/Th ratio. The age result excludes an Archean age for the cpx, which therefore most likely also formed, or recrystallized during the 2.0 Ga event.

In this study the range of minerals to which PbSL dating ( Pb stepwise leaching) has been applied (garnet, titanite, hornblende, clinopyroxene, epidote, staurolite, tourmaline) is extended to sillimanite. As in previous studies using this method, the leaching technique has shown the great advantage, compared to bulk mineral dating,

that the different components of Pb (host mineral, microinclusions, contaminants) can be detected through their varying uranogenic and thorogenic Pb signatures. In most cases consistent ages could be obtained for minerals with non-cogenetic inclusions by disregarding the leach steps with mixed isotopic signatures. In cases where the total Pb budget is dominated by microinclusions and most steps are a mixture of Pb components from host mineral and inclusions, the age of the host mineral could be determined under the assumption that the microinclusions are cogenetic. A complicated leach spectrum is obtained for sillimanite 93/048 (aggregates pseudomorphic after andalusite), which can be interpreted with confidence because steps 1 and 2 are in isotopic equilibrium with garnet from the matrix of the same sample. However, it can not clearly be distinguished whether the first two steps are dominated by Pb components from the host mineral itself. The high radiogeneity of these steps might indicate the influence of cogenetic microinclusions such as monazite.

8. Discussion

8.1. Implications for geochronology

The results from sillimanite and garnet (93/048) provide evidence for a highly retentive character of the U−Pb\mathrm{U}-\mathrm{Pb} system in these two minerals. In our example the U−Pb\mathrm{U}-\mathrm{Pb} system was not affected during the high grade metamorphic overprint at 2.0 Ga , where peak metamorphic temperatures of 800−850∘C800-850^{\circ} \mathrm{C} were reached. Pb isotopic data from garnet and sillimanite may therefore rather reflect the time of mineral growth or recrystallization than that of a closure during cooling. Garnet and sillimanite are thus important geochronometers in polymetamorphic terranes. Nevertheless, it cannot be uncritically assumed that the P−T\mathrm{P}-\mathrm{T} conditions recorded by the mineral chemistry of, for example, garnet in metamorphic assemblages would be those dated by the PbSL chronometer, as major element systems might have at least partly reequilibrated in the later metamorphic event (Mezger et al., 1989). Information about the P−T\mathrm{P}-\mathrm{T} evolution during
early metamorphic episodes can, however, come from the identification and dating of relic metamorphic mineral assemblages and reaction textures, involving alumosilicates and other metamorphic minerals.

In the following sections the different geological events will be discussed. In Table 5 a simplified summary of the complex relationships in the CZ is presented, in which the structural, magmatic and metamorphic events are correlated with each other.

8.2. The late Archean/early Proterozoic event in the CZ(∼2.6Ga)C Z(\sim 2.6 \mathrm{Ga})

A long period of geological activity in the CZ must be assumed for the Archean-Proterozoic. Magmatism occurred from ⩾2655Ma\geqslant 2655 \mathrm{Ma} (Alldays granodiorite) to ⩽2550Ma\leqslant 2550 \mathrm{Ma} (Singelele and other granitoids). The Bulai intrusion ( 2570 Ma ) postdates high grade structures of the D2a phase and associated anatectic features. Mobilization of large amounts of anatectic melt was contemporaneous with the Bulai intrusion and mingling of the calculcaline magmas with leucogranitic material has been described. However, the Bulai gneisses underwent several phases of high grade deformations, providing evidence for more than one ‘post-Bulai’ overprint.

Our data from metapelite 93/048 give evidence for growth of sillimanite and garnet during a high grade episode at 2521±4Ma2521 \pm 4 \mathrm{Ma}. The textural relationships indicate sillimanite ( + cordierite + garnet) growing at the expense of chiastolitic andalusite ( + inclusions), thereby implying a prograde metamorphic reaction at high temperatures and low pressures. The dating of these relic metamorphic assemblages thus documents a first episode of ‘post-Bulai’ high grade metamorphism. It seems that the CZ underwent several pulses of thermal disturbance associated with magmatism and (tectono-)metamorphism between 2.7 and 2.5 Ga .

The late Archean history of the CZ thus resembles that described for the NMZ by Berger et al. (1995) and Kamber and Biino (1995) in that (a) magmatic activity is constrained between 2.7 and 2.62 Ga (charno-enderbites in the NMZ, granodioritic Alldays Gneiss in the CZ; Jaeckel et al., 1997);

Table 5
Summary of geological events in the Central Zone, Limpopo Belt (modified after Watkeys, 1983)

(b) porphyritic granites and charnockites were emplaced between 2.62 and 2.58 Ga in the NMZ, whereas in the CZ the porphyritic Bulai Pluton (Barton et al., 1994) and the granitoid Singelele orthogneiss intruded between 2.6 and 2.55 Ga (Jaeckel et al., 1997); © low to medium pressure granulite facies conditions prevailed in the NMZ between 2.62 and 2.58 Ga , whereas our data give evidence for (M2-)granulite facies conditions in the CZ prevailing until about 2.52 Ga . Anticlockwise P−T\mathrm{P}-\mathrm{T} evolution is well constrained for the NMZ and is suggested for the CZ from the occurrence of sillimanite pseudomorphs after chiastolitic andalusite; (d) NNE-SSW directed compression was the main foliation producing tectonism in the NMZ, which has its CZ-equivalent in the D2 deformations described by Watkeys (1983).

There is no evidence for important geological activity between 2.5 and 2.0 Ga in the CZ (or SMZ and NMZ), or for retrogression and rehydration of the presently exposed CZ granulites. They probably resided in mid to lower crustal levels during this period. A reason for lack of exhumation after 2.6 Ga might be due to the fact that the late Archean low pressure granulites in the CZ were not produced primarily by tectonic thickening and thus erosion and associated isostatic equilibration of the crust was not an efficient mechanism for post-collisional exhumation.

8.3. Identification of Proterozoic structures in the CZ

In order to date the D3 deformational event, we have chosen samples with mineral assemblages that recrystallized synkinematically (samples 93/083 and 93/167). The data from clinopyroxene 93/083 give an imprecise age of 1860±320Ma1860 \pm 320 \mathrm{Ma}, which nevertheless indicates that the synkinematic recrystallization of cpx, which occured during the formation of the Campbell crossfold (F3a), is not the product of Archean tectonic activity, as previously suggested. The data obtained from calcsilicate enclaves in a shear zone on Mount Shanzi give more precise ages of 2010±17Ma2010 \pm 17 \mathrm{Ma} (garnet), 2007±5Ma2007 \pm 5 \mathrm{Ma} (titanite), 2011±20Ma2011 \pm 20 \mathrm{Ma} (monazite) and 1983±14Ma1983 \pm 14 \mathrm{Ma} (apatite). The deformation on

Mount Shanzi belongs to a phase of refolding (F3b) of the regional scale crossfold structures (F3a). During this phase subparallel fold axes and mineral elongation lineations (ductile feldspars and strings of garnet) were produced, which dip with about 30 towards 260 (Watkeys, 1983; Holzer, 1995). The lineations and fold axes in the Tshipise Straightening Zone, South of Messina, have the same orientation and we therefore suggest that this structure was also produced during the Proterozoic event. It is important to note that apparently there exist domains in the CZ where the Proterozoic D3 deformations are not predominant, such as in the Three Sisters area 20 km NW of Messina.

8.4. The timing of metamorphism during the Proterozoic event (ca. 2.05-1.95 Ga)

The clockwise P−T\mathrm{P}-\mathrm{T} path of the youngest metamorphic event (M3) in the CZ was determined by various authors (Chinner and Sweatman, 1968; Droop, 1989; Horrocks, 1983a,b; Windley et al., 1984; Harris and Holland, 1984; Hisada and Miyano, 1996). In the following section we reinterpret the timing of this clockwise P−T\mathrm{P}-\mathrm{T} evolution in the light of recently published age data and those from this study. The error-envelopes in the P−T\mathrm{P}-\mathrm{T} diagram (Fig. 8) are not based on quantitative error calculations. The temperature-time data and

Fig. 8. Pressure-temperature diagram: combination of new age data with the PT path defined for the youngest high grade event in the Messina area, CZ. The timing of a Proterozoic PT-evolution is discussed in the text. Labels correspond to references given in Table 1.

the temperature errors follow commonly accepted values of the respective (thermo-)chronometers (e.g. TcT_{\mathrm{c}} of Rb−Sr\mathrm{Rb}-\mathrm{Sr} Biotite =350±50∘C=350 \pm 50^{\circ} \mathrm{C}; Dodson, 1979), whereas the pressure errors are speculative.

Metamorphic zircons from metapelites yield an age of 2026±7Ma2026 \pm 7 \mathrm{Ma} (Jaeckel et al., 1997). This date is interpreted to reflect the time of high pressure granulite facies metamorphism. Our PbSL data from garnet and titanite with ages of 2010±17Ma2010 \pm 17 \mathrm{Ma} and 2007±5Ma2007 \pm 5 \mathrm{Ma}, respectively, show that synkinematic recrystallization of garnet and titanite (in calcsilicate enclaves from Mount Shanzi) during M3/D3 occurred almost synchronously and they probably mark the time when near-isothermal decompression was initiated. The PbSL data is confirmed by a concordant datapoint with an age of 2011±20Ma2011 \pm 20 \mathrm{Ma} from monazite of the same sample. Textural evidence for equilibrium between garnet, titanite, clinopyroxene and plagioclase imply that their dynamic recrystallization occurred under granulite facies conditions. Subsequent static recrystallization of plagioclase shows that high temperatures persisted after 2010 Ma . At some localities a late phase of partial melting is observed. The crystallization of these melt patches post-dates the latest ductile deformations (F3b). Undeformed leucosome from within the Sand River Gneiss (Causeway locality) yielded zircon ages of 2005±8Ma2005 \pm 8 \mathrm{Ma} (Jaeckel et al., 1997). These authors interpreted the melt patches as the product of decompression melting around 2005 Ma . The time of the subsequent near isobaric cooling (IBC) is constrained by a U−Pb\mathrm{U}-\mathrm{Pb} concordia intercept age of 1983±14Ma1983 \pm 14 \mathrm{Ma} from five apatite fractions. Furthermore the time of retrograde metamorphism is constrained by Rb−Sr\mathrm{Rb}-\mathrm{Sr} biotite-whole rock cooling ages from throughout the CZ (Barton et al., 1992), which scatter widely but concentrate around 1970 Ma . This scatter may signify regional variance of uplift in the ca. 600 km long CZ. However, it may partly be due to Sr isotopic disequilibrium between biotite and the whole rock, caused e.g. by retrograde recrystallization.

8.5. Towards a Proterozoic tectonic model for the Limpopo Belt-a working hypothesis

This model is based mainly on structural observations from the Messina area (summarized pre-
viously in this paper) and on the data discussed above. Further we try to delineate large scale tectonic relationships of the 2 Ga event, including available structural and P−T−t\mathrm{P}-\mathrm{T}-\mathrm{t} data from the entire CZ. The model is merely an outline and should therefore be considered as no more than a working hypothesis for future research. Four phases of a Proterozoic event are distinguished.

8.5.1. Phase 1. Tectonic thickening of the CZ

( >2.03Ga>2.03 \mathrm{Ga} ) (Fig. 9(a))
The CZ was squeezed between the Kaapvaal and Zimbabwe cratons and as a consequence the CZ was thickened and high pressure granulite facies conditions in the presently exposed crustal levels were reached at or before 2.03 Ga . Early D3-deformations within the Bulai Pluton suggest a NNW-SSE directed shortening. The exact relative positions of the CZ and the adjacent marginal zones (amalgamated to the cratons) during this early stage of the Proterozoic orogeny are unknown. Information about duration and P-T path of the prograde metamorphism are also lacking. The Proterozoic orogeny in the Limpopo Belt appears to have a close relationship with the Kheis and Magondi fold belts at the western margins of the cratons, which were formed at about 2.0 Ga . A better understanding of the link with these orogenic belts could provide important additional information about the geodynamic evolution in the Limpopo Belt.

8.5.2. Phase 2. Dextral transpression ( 2.03-2.01

Ga) (Fig. 9(b))
In the Messina area the formation of large scale crossfolds occurred at or shortly before 2.01 Ga . ENE-WSW directed movements along D3 structures (e.g. thrusting on Mount Shanzi or Limpopotrend shears) indicate a transition from NNW-SSE directed shortening (with strong pure shear component) to dextral transpressive tectonics. However, the shear strain was concentrated in the suture zones bordering the CZ. The moderately SSE dipping Triangle shear belt (Kamber et al., 1995b) to the North and the steeply SSE dipping Tshipise Straightening Zone to the South (Bahnemann, 1972) are both regarded as the product of such right lateral transpressive tectonics. Strike

(a)

Figure 9b:
Dextral Transpression (−2.03−2.01Ga)(-2.03-2.01 \mathrm{Ga})

Fig. 9. Sketch maps illustrating the four-stage tectonic evolution during the Proterozoic ‘strike slip orogeny’. The tectonic model is discussed in the text. For legend and abbreviations see Fig. 1. (a) 1=1= Molopo Farms Complex, 2=2= Nebo Granite, 3=3= Soutpansberg Graben, 4=4= Waterberg Basin, 5=5= Palapye Graben, 6=6= Murchison Thabazimbi Lineament. (d) P=\mathrm{P}= Palala Granite, E=\mathrm{E}= Entabeni Granite, S = Schiel Alkaline Complex.

©

Figure 9d: Final exhumation of the CZ
Transtensional phase (Soutpansberg rifting) associated with postorogenic transcurrent faulting along the Palala

(d)

slip movement in these 30−50 km30-50 \mathrm{~km} wide shear zones occurred under granulite facies conditions. Transpressional tectonics where strike slip and thrusting were contemporaneous may explain the existence of both subhorizontal and down-dip lineations in the Triangle shear zone. Steep lineations are also observed in the Sunny Side shear zone and in the northern domain of the Palala shear zone (McCourt and Vearncombe, 1992). They reflect subvertical movements in the southern and southwestern part of the CZ as a response to the transpressive tectonics. Schaller et al. (submitted) document continuous deformation under changing metamorphic conditions along the Palala shear zone in the Koedoesrand area. Thrusting in the Koedoesrand window occured under granulite facies conditions at ⩾2.03Ga\geqslant 2.03 \mathrm{Ga}, changing to dextral strike slip shearing associated with the exhumation of the CZ shortly after 2.0 Ga . Important migmatization and formation of voluminous granitic bodies at about 2.0 Ga occured in the westernmost part of the CZ, the Mahalapye migmatite complex in Botswana (Chavagnac et al., submitted).

8.5.3. Phase 3. Uplift and cooling (2.01-1.95 Ga) (Fig. 9©)

The right lateral movements of the Zimbabwe and Kaapvaal cratons relative to each other prevailed and thus strike slip tectonics in the Triangle (e.g. Kamber et al., 1995a) and Straightening shear zones lasted until after 2.0 Ga . The various cooling ages between 2.0−1.95Ga2.0-1.95 \mathrm{Ga} from throughout the CZ (e.g. Barton et al., 1992) indicate that the uplift and cooling of the entire CZ was associated with these strike slip movements. Two large shear belts are suggested to have bounded the tectonically thickened CZ at that time (Holzer et al., 1997). Both shear belts are characterized by a set of interconnected shear zones that operated simultaneously: the Triangle-Lepokole-Magogaphate shear system to the North and the Tshipise Straightening Zone-Sunny Side-Palala shear system in the Southern part of the CZ. The detailed kinematics of these shear systems are not yet fully understood. Whereas the northern shear system is predominated by dextral strike slip at high grade conditions, the southern system is much more complex, including strike slip, oblique slip and
subvertical movements under peak to retrograde conditions. The near isothermal uplift from high to low/medium pressure granulite facies conditions in the CZ might be the product of a rapid change from a transpressive to a transtensive tectonic environment. Subsequent to the decompression, thermal relaxation resulted in a phase of near isobaric cooling (pressure =5−6kbar=5-6 \mathrm{kbar} ). By about 1.95 Ga temperatures had decreased to ca. 350 C .

8.5.4. Phase 4. Final exhumation of the CZ and post-orogenic transcurrent faulting ( ∼1.95−1.85Ga\sim 1.95-1.85 \mathrm{Ga} ) (Fig. 9(d))

Strike slip movements persisted after the uplift and cooling of the CZ. However, strain was localized along the southern margin of the CZ, mainly in the Palala lineament. This phase of late- to post-orogenic transcurrent faulting led to the formation of the Soutpansberg and Palapye graben structures. The clastic sediments of the Soutpansberg sequence were transported from the North (CZ). The transtensive tectonics and the contemporaneous sedimentation in the graben structures are thus closely associated with tectonic thinning and final exhumation of the CZ (Barker, 1983). The timing of this transtensional phase is bracketed by the ages of the Entabeni granite (1950 Ma; Barton et al., 1995), which is unconformably overlain by the Soutpansberg basal conglomerates, and the Schiel Alkaline Complex ( 1850 Ma ; Barton et al., 1996), which reflects the final stages of the Soutpansberg rifting tectonics. The formation of low grade mylonites in the Palala shear zone post-dates the uplift and cooling of the CZ and is thus most probably related to the described transcurrent tectonics. A dextral shear sense is described for the late strike slip movements in the Palala shear zone (Brockhuizen and McCourt, 1995).

8.6. Implications for the tectono-metamorphic history of the entire Limpopo Belt

The view presented above of the CZ as a polymetamorphic province is in strong conflict with many of the published tectonic models that include all three subzones of the Limpopo Belt, and with terrane interpretations of the whole of Southern

Table 6
Summary of geological events in the three subzones of the Limpopo Belt

Africa, which argue that the final juxtaposition of the Kaapvaal and Zimbabwe cratons is the product of a suggested late Archean ‘Limpopo Orogeny’ (e.g. Treloar et al., 1992). Our study leads to the following implications concerning the history of the Limpopo Belt (Table 6).

Different types of granulite facies events: In the Limpopo Belt several granulite facies events are recorded. The different types of granulite facies metamorphism signify different tectono-metamorphic genetic relationships: high pressure granulites with clockwise P−T\mathrm{P}-\mathrm{T} paths were produced in orogenic environments including tectonic thickening of the crust (SMZ at 2.7Ga,CZ2.7 \mathrm{Ga}, \mathrm{CZ} at 2.0 Ga ). Low to medium pressure granulites with anticlockwise P-T evolution seem to be confined to the Archean-Proterozoic boundary ( 2.6 Ga in NMZ and 2.52 Ga in CZ). These low pressure granulite facies events, in which mantle derived magmas provided the overlying crust with heat, might reflect vertical crustal growth during the late Archean.

Two episodes of late Archean metamorphism: two episodes of late Archean activity from 2.72 to 2.62 Ga and from 2.62 down to 2.52 Ga are recorded in the Limpopo Belt. During the first period high to medium pressure granulites were produced in the SMZ, regarded as a product of orogenic tectonics (Van Reenen et al., 1987). Magmatic activity at the same time is also identified in the CZ and NMZ; however, the appertaining tectono-metamorphism has not yet been characterized sufficiently and therefore the tectonic relationships remain unclear. During a second period shortly after 2.62 Ga granulite facies metamorphism with anticlockwise P−T\mathrm{P}-\mathrm{T} evolution occured in the NMZ (Kamber and Biino, 1995). The voluminous charnockitic and granitic magmatism in the NMZ has its analogy in the Chilimanzitype granites, which at the same time intruded the adjacent Zimbabwe craton. The features observed in the NMZ probably illustrate processes at lower crustal levels which were important during the late Archean cratonization. Granulite facies conditions associated with important magmatic activity between 2.6 and 2.52 Ga are also recorded in the CZ. Relic sillimanite aggregates pseudomorphic after andalusite give evidence for a late high grade
metamorphic episode with an anticlockwise P−T\mathrm{P}-\mathrm{T} evolution at 2.52 Ga .

Although the previously discussed similarities between the late Archean histories of the NMZ and the CZ may suggest a common geologic history at about 2.6 Ga , the tectonic relationships are difficult to assess because high grade overprint of the CZ and important strike slip movements along the Triangle shear zone at about 2.0 Ga have erased much of the field evidence and obscured the regional context. The relationships between the CZ and the SMZ are even more diffuse. The timing and style of metamorphism recorded in the late Archean granulites from the SMZ differ markedly from those in the CZ and NMZ: clockwise P−T\mathrm{P}-\mathrm{T} evolution led to the formation of high pressure granulites at about 2.7 Ga (Van Reenen et al., 1992), which is 100 Ma earlier than the anticlockwise episodes in the CZ and NMZ. Uplift of the SMZ was contemporaneous with the syntectonic Matok intrusion around 2.67 Ga (Barton et al., 1992). No evidence for high/medium grade metamorphism younger than 2.62 Ga has yet been identified in the SMZ. A correlation of high temperature events in all three subzones during the late Archean can thus neither be denied nor justified convincingly.

Juxtaposition of the different tectonic units: the marginal zones were thrust onto the adjacent cratons mainly in the late Archean. However, it is quite uncertain whether the five tectonic units (Kaapvaal craton, SMZ, CZ, NMZ, Zimbabwe craton) formed a connected continental mass. It cannot be precluded convincingly that some of these tectonic units were brought together only during the Proterozoic event. The relative positions of the cratons and the three subzones of the Limpopo Belt before the Proterozoic transpressive orogeny (at ca. 2.0 Ga ) are uncertain.

9. Conclusions

The Limpopo Belt is an intercratonic suture zone, which has been repeatedly reactivated during the late Archean and Proterozoic. However, the three subzones are characterized by distinct tec-tono-metamorphic histories, which cannot easily

be correlated, because style and timing of the recorded granulite events are significantly different from each other. The metamorphic history of the Limpopo Belt is thus much more complex than most published tectonic models imply. The Limpopo Belt remains an interesting object for future research where possible genetic models for granulite provinces can be tested, such as continent collision and magmatic arc environments or crust formation by magmatic underplating in the Archean.

The history of the Limpopo Central Zone is depicted by repeated high grade metamorphism associated with deformational and magmatic events which occured at about 3.2−3.1Ga3.2-3.1 \mathrm{Ga}, 2.65−2.52Ga2.65-2.52 \mathrm{Ga} and 2.0±0.05Ga2.0 \pm 0.05 \mathrm{Ga}. The youngest high grade event is described as a dextral transpressive orogeny, during which the Kaapvaal and Zimbabwe cratons collided. The CZ, which was tectonically thickened during this collision, is bounded by two large strike slip shear systems which were active during peak metamorphism and exhumation.

This study has confirmed the importance of age data from metamorphic silicates, because they provide important information that can be used to unravel the geologic histories of polymetamorphic high grade terranes. Lead stepwise leaching (PbSL) is a tool with which important minerals such as garnet, titanite, clinopyroxene and even sillimanite can be dated. These age data can be directly linked with the petrologic interpretation ( P−T\mathrm{P}-\mathrm{T} conditions) of the dated minerals or mineral assemblages and are thus important for the determination of P−T−t\mathrm{P}-\mathrm{T}-\mathrm{t} paths. This study has also shown that PbSL on highly retentive minerals like sillimanite and garnet can provide information about the P−T\mathrm{P}-\mathrm{T} evolution of early metamorphic events that have been overprinted by a later granulite facies metamorphism.

Acknowledgment

Discussions and field trips with Tom Blenkinsop, Dirk van Reenen, Andre Smit and B.K. Paya have contributed much to this work. Dirk and Andre are thanked for their gentlemanly attitude towards
geological controversies. Financial support from the Schweiz. Nationalfonds (Grant 20-40442.94) is gratefully acknowledged. L.H. would also like to thank Alfred Kröner for the many helpful discussions. Klaus Mezger is thanked for his constructive review, which greatly improved an earlier version of the manuscript.

References

Armstrong, R., Compston, W., Dodson, M.H., Kröner, A., Williams, I.S., 1988. Archean crustal history in southern Africa. Research School of Earth Sciences Australian National University, Annual Report 1987, pp. 62-64
Bahnemann, K.P., 1972. A review of the structure, stratigraphy and metamorphism of the Basement rocks in the Messina District, Northern Transvaal, unpublished Ph.D. thesis, Pretoria.
Barker, O.B., 1983. A proposed geotectonic model for the Soutpansberg group within the Limpopo Mobile Belt, South Africa. Spec. Publ. Geol. Soc. S. Afr. 8, 181-190.
Barton, J.M., 1983. Pb-Isotopic evidence for the age of the Messina Layered Intrusion, central zone, Limpopo Mobile Belt. Spec. Publ. Geol. Soc. S. Afr. 8, 39-41.
Barton, J.M., 1995. Constraints on the nature of the Proterozoic juxtaposing of the Central Zone of the Limpopo Belt with the Southern Marginal Zone and Kaapvaal Craton in the Northern Transvaal. Centennial Geocongress of the Geological Society of South Africa, Johannesburg, pp. 170−173170-173.
Barton, J.M., 1996. The Messina Layered Intrusion, Limpopo Belt, South Africa: an example of in-situ contamination of an Archean anorthosite complex by continental crust. Precambrian Res. 78, 139-150.
Barton, J.M., Key, R.M., 1983. Rb-Sr ages and geological setting of certain rock units from the central zone of the Limpopo Mobile Belt, near Zanzibar, eastern Botswana. Spec. Publ. Geol. Soc. S. Afr. 8, 19-25.
Barton, J.M., Sergeev, S., 1997. U-Pb analyses of single grains of zircon from quartzite in the Beit Bridge Group, Central Zone, Limpopo Belt. J. African Earth Sci. (in press).
Barton, J.M., van Reenen, D.D., 1988. The significance of 3000 Ma mafic dykes in the central zone of the Limpopo belt, southern Africa. Chem. Geol. 70, 141-157.
Barton, J.M., van Reenen, D.D., The significance of Rb−Sr\mathrm{Rb}-\mathrm{Sr} ages of biotite and phlogopite for the thermal history of the Central and Southern Marginal Zones of the Limpopo Belt of southern Africa and the adjacent portions of the Kaapvaal Craton. 1992. Precambrian Res. 55, 17-31.
Barton, J.M., Fripp, R.E.P., Horrocks, P., McLean, N., 1979. The geology, age and tectonic setting of the Messina Layered Intrusion, Limpopo Mobile Belt, Southern Africa. Am. J. Sci. 279,1108−1134279,1108-1134.
Barton, J.M., Ryan, B., Fripp, R.E.P., 1983. Rb-Sr and

U-Th-Pb studies of the Sand River Gneisses, central zone, Limpopo Mobile Belt. Spec. Publ. Geol. Soc. S. Afr. 8, 9-18.
Barton, J.M., van Reenen, D.D., Roering, C.A., 1990. The significance of 3000 Ma granulite facies mafic dykes in the central zone of the Limpopo belt, southern Africa. Precambrian Res. 48, 299-308.
Barton, J.M., Doig, R., Smith, C.B., Bohlender, F., van Reenen, D.D., 1992. Isotopic and REE characteristics of the intrusive charnoenderbite and enderbite geographically associated with the Matok Pluton, Limpopo Belt, southern Africa. Precambrian Res. 55, 451-467.
Barton, J.M., Holzer, L., Kamber, B., Doig, R., Kramers, J.D., Nyfeler, D., 1994. Discrete metamorphic events in the Limpopo belt, southern Africa: implications for the application of P T paths in complex metamorphic terranes. Geology 22, 1035-1038.
Barton, J.M., Doig, R., Smith, C.C., 1995. Age, origin, and tectonic significance of the Entabeni granite, northern Transvaal. South Africa. S. Afr. J. Geol. 98, 326-330.
Barton, J.M., Barton, E.S., Smith, C.B., 1996. Petrography, age and origin of the Schiel alkaline complex, northern Transvaal. South Africa. J. Afr. Earth Sci. 22, 133-145.
Berger, M., Kramers, J.D., Nägler, T.F., 1995. Geochemistry and geochronology of charnoenderbites in the Northern Marginal Zone of the Limpopo Belt, Southern Africa, and generic models. Schweiz. Mineral. Petrogr. Mitt. 75, 17-42.
Blenkinsop, T.G., Rollinson, H., Fedo, C., Paya, B.K., Kamber, B., 1995. The North Limpopo Thrust Zone (NLTZ): the northern boundary of the Limpopo Belt in Zimbabwe and Botswana. Centennial Geocongress of the Geological Society of South Africa. Johannesburg, pp. 174-177.
Bohlen, S.R., 1987. Pressure-temperature time paths and a tectonic model for the evolution of granulites. J. Geol. 95, 617-632.
Brandl, G., Reimold, W.U., 1990. The structural setting and deformation associated with pseudotachylite occurrences in the Palala Shear Belt and Sand River Gneiss. N. Transvaal. Tectonophys. 171, 201-220.
Broekhuizen, A., McCourt, S., 1995. Structural evolution of the Koedoesrand hills and surrounding area, North-West Transvaal, South Africa. Centennial Geocongress of the Geological Society of South Africa, Johannesburg, p. 280.
Chinner, G.A., Sweatman, T.R., 1968. A former association of enstatite and kyanite. Miner. Mag. 36, 1052-1060.
De Beer, J.H., Stettler, E.H., 1992. The deep structure of the Limpopo Belt. Precambrian Res. 55, 163-186.
Dodson, M.H., 1979. Theory of cooling ages. In: Jäger, E., Hunziker, J.C. (Eds.), Lectures in Isotope Geology. Springer, Berlin, pp. 194-202.
Droop, G.T.R., 1989. Reaction history of garnet-sapphirine granulites and conditions of Archaean high-pressure granu-lite-facies metamorphism in the Central Limpopo Mobile Belt, Zimbabwe. J. Metam. Geol. 7, 383-403.
Du Toit, M.C., van Reenen, D.D., Roering, C., 1983. Some aspects of the geology, structure and metamorphism of the

Southern Marginal Zone of the Limpopo Metamorphic Complex. Spec. Publ. Geol. Soc. S. Afr. 8, 89-102.
Frei, R., Kamber, B.S., 1995. Single mineral Pb - Pb dating. Earth Planet. Sci. Lett. 129, 261-268.
Frei, R., Pettke, T., 1996. Mono-sample Pb−Pb\mathrm{Pb}-\mathrm{Pb} dating of pyrrhotite and tourmaline: Archean vs Proterozoic gold mineralization in Zimbabwe. Geology 24, 823-826.
Frei, R., Biino, G.G., Prospert, C., 1995. Dating a Variscan pressure-temperature loop with staurolite. Geology 23, 1095−10981095-1098.
Frei, R., Villa, I.M., Nägler, Th.F., Kramers, J.D., Przybylowicz, W.J., Prozesky, V.M., Hofmann, B.A., Kamber, B.S., 1997. Single mineral dating by the Pb−Pb\mathrm{Pb}-\mathrm{Pb} stepleaching method, assessing the mechanism. Geochim. Cosmochim. Acta 61, 393-414.
Fripp, R.E.P., 1983. The precambrian geology of the area around the Sand River near Messina, central zone, Limpopo Mobile Belt. Spec. Publ. Geol. Soc. S. Afr. 8, 89-102.
Harley, S.L., 1989. The orogins of granulites: a metamorphic perspective. Geol. Mag. 126, 215-331.
Harris, N.B.W., Holland, T.J.B., 1984. The significance of cor-dierite-hypersthene assemblages from the Beitbridge region of the Central Limpopo Belt; evidence for rapid decompression in the Archaean. Am. Miner. 69, 1036-1049.
Hisada, K., Miyano, T., 1996. Petrology and microthermometry of aluminous rocks in the Botswanan Limpopo Central Zone: evidence for isothermal decompression and isobaric cooling. J. Metam. Geol. 14, 183-197.

Holzer, L., 1995. The magmatic petrology of the Bulai Platon and the tectono-metamorphic overprint at 2.0 Ga in the central zone of the Limpopo Belt, Southern Africa. Unpublished M.Sc. thesis, University of Berne.
Holzer, L., Kamber, B.S., 1995. Reconstruction of a clockwise PTt-path: the 2.0 Ga event in the Limpopo Belt, Southern Africa. EUG 8 Strasbourg, Terra Abstracts, Vol. 7, p. 44
Holzer, L., Kamber, B., Kramers, J.D., Frei, R., 1996. The Tectonometamorphic event at 2 Ga in the Limpopo Belt and the resetting behaviour of chronometers at high temperature. Spec. Publ. Geol. Surv. Namibia 1, 127-138.
Holzer, L., Kramers, J.D., Blenkinsop, T.G., 1997. Granulite facies metamorphism in the Limpopo Central Zone (Southern Africa): evidence for a Proterozoic continental collision between the Kaapvaal and Zimbabwe Cratons. EUG 9 Strasbourg, Terra Abstracts, Vol. 9, p. 39-5B23.
Horrocks, P.C., 1983a. A corundum and sapphirine paragenesis from the Limpopo Mobile Belt, southern Africa. J. Metam. Geol. 1, 13-23.
Horrocks, P.C., 1983b. The precambrian geology of an area between Messina and Tshipise, Limpopo Mobile Belt. Spec. Publ. Geol. Soc. S. Afr. 8, 81-88.
Jaeckel, P., Kröner, A., Kamo, S.L., Brandl, G., Wendt, J.I., 1997. Late Archean to early Proterozoic granitioid magmatism and high-grade metamorphism in the Limpopo Belt, South Africa. J. Geol. Soc. Lond. 154, 25-44.
Kamber, B.S., Biino, G.G., 1995. The evolution of high T low P granulites in the Northern Marginal Zone sensu stricto,

Limpopo Belt. Zimbabwe-the case for petrography. Schweiz. Miner. Petrogr. Mitt. 75, 427-454.
Kamber, B.S., Blenkinsop, T.G., Villa, I.M., Dahl, P.S., 1995a. Proterozoic transpressive deformation in the Northern Marginal Zone, Limpopo Belt, Zimbabwe. J. Geol. 103, 493-508.
Kamber, B.S., Kramers, J.D., Napier, R., Cliff, R.A., Rollinson, H.R., 1995b. The Triangle Shearzone, Zimbabwe, revisited; new data document an important event at 2.0 Ga in the Limpopo Belt, Precambrian Res. 70, 191-213.
Ludwig, K. R., 1988. ISOPLOT for MS-DOS, a plotting and regression program for radiogenic isotope data for IBM-PC compatible computers, version 1.00. In USGS Open-File Report, OF 82-820.
Ludwig, K. R., 1994. ISOPLOT, a plotting and regression program for radiogenic isotope data. In USGS, Open-File Report, OF 91-445.
McCourt, S., 1983. Archean Lithologies of the Koedoesrand Area, NW Transvaal, South Africa. Spec. Publ. Geol. Soc. S. Afr. 8, 113-119.

McCourt, S., Vearncombe, J.R., 1992. Shear zones of the Limpopo Belt and adjacent granitoid-greenstone terranes: implications for late Archaean collision tectonics in southern Africa. Precambrian Res. 55, 553-570.
Mezger, K., Hanson, G.N., Bohlen, S.R., 1989. U-Pb systematics of garnet: dating the growth of garnet in the late Archean Pikwitonei granulite domain at Cauchon and Natawahunan Lakes, Manitoba, Canada. Contrib. Miner. Petrol. 101, 136−148136-148.
Mkweli, S., Kamber, B.S., Berger, M., 1995. Westward continuation of the craton-Limpopo Belt tectonic break in Zimbabwe and new age constraints on the timing of the thrusting. J. Geol. Soc. Lond. 152, 77-83.
Percival, J. A., 1990. Archean high-grade metamorphism. In

Condie, K.C., Archean Crustal Evolution. Elsevier Science, Amsterdam, pp. 357-396.
Retief, E.A., Compston, W., Armstrong, R.A., Williams, I.S., 1990. Characteristics and preliminary U-Pb ages of zircons from Limpopo Belt lithologies. Abstract volume, Limpopo Workshop, Rand Afrikaans University, Johannesburg, South Africa, pp. 95-99.
Smit, C.A., Roering, C., Van Reenen, D.D., 1992. The structural framework of the southern margin of the Limpopo Belt, South Africa. Precambrian Res. 55, 51-67.
Treloar, P.J., Coward, M.P., Harris, N.B.W., 1992. Himalayan-Tibetan analogies from the evolution of the Zimbabwe Craton and Limpopo Belt, South Africa. Precambrian Res. 55, 571-587.
Tsunogae, T., Yurimoto, H., 1995. Single zircon U-Pb geochronology of the Limpopo Belt by secondary ion mass spectrometry. Geochem. J. 29, 197-205.
Van Reenen, D.D., Barton, J.M., Roering, C., Smit, C.A., Van Schalkwyk, J.F., 1987. Deep crustal response to continental collision: the Limpopo Belt of southern Africa. Geology 15, 11−1411-14.
Van Reenen, D.D., Roering, C., Ashwal, L.D., De Wit, M.J., 1992. Regional geological setting of the Limpopo Belt. Precambrian Res. 55, 1
Watkeys, M.K., 1983. A retrospective view of Central Zone of Limpopo Belt, Zimbabwe. Spec. Publ. Geol. Soc. S. Afr. 8, 65−8065-80.
Watkeys, M. K., 1984. The Precambrian geology of the Limpopo Belt north and west of Messina. Unpublished Ph.D. thesis, University of Witwatersrand, Johannesburg.
Windley, B.F., Ackermand, D., Herd, R.K., 1984. Sapphirine/kornerupine-bearing rocks and crustal uplift history of the Limpopo belt, South Africa. Contrib. Miner. Pet. 86, 342-358.
York, D., 1969. Least squares fitting of a straight line with correlated errors. Earth Planet. Sci. Lett. 5, 320-324.