Stratigraphy and Basin Analysis of the West Rand Group, Witwatersrand Supergroup

The Witwatersrand Basin is renowned for its resplendent gold deposits, which sit nestled between quartz pebbles in fluvial conglomerates of the upper 2.9 Ga Central Rand Group (Kositcin and Krapez, 2004; Frimmel et al., 2005; Tucker et al., 2016; Frimmel, 2019). However, the genesis of the basin began about 50 Ma earlier, when well sorted marine sands of the lower West Rand Group were deposited on a basement of recently stabilised Kaapvaal Craton. Although not particularly endowed with gold when compared with the Central Rand Group (historical production of ~35 tons, Tankard et al., 1982), the West Rand Group contains a bonanza of extremely well-preserved sediments (up to 5 km-thick), which host one of Earth’s earliest and largest intracratonic sedimentary successions. This sedimentary archive captures changing dynamics on a Mesoarchaean (~2.95 to 2.90 Ga) shelf and contains several important formations and marker beds that reveal clues about ancient climate, microbial life and palaeo-oxygen concentrations. These include storm deposits, iron-formations and manganese carbonate concretions in the Parktown and Brixton formations (Beukes, 1996; McCarthy, 2006; Cochrane, 2009; Smith, 2009; Smith et al., 2013; Smith et al., 2023), isotopically light organic carbon in the Promise, Coronation and Rietkuil formations (Guy et al., 2012), and widespread, possibly glacial, diamictite deposits in the Promise, Coronation and Afrikander formations (Beukes and Cairncross, 1991; von Brunn and Gold, 1993; Young et al., 1998; Guy, 2012; Hofmann and Bindeman, 2023).

These important formations, members and marker beds have, with very few exceptions, been mapped on all of the first and very good geological maps of Witwatersrand strata by Mellor (1911), Rogers (1922) and Nel (1927, 1934, 1935). Since then, a number of studies have been undertaken on stratigraphic and sedimentological aspects of the West Rand Group. However, due to limited surface exposure and core intersections, these studies normally deal with local areas or very specific aspects of the sequence. These include studies in the Heidelberg area (Camden-Smith, 1980), Klerksdorp area (Watchorn, 1981; Brink, 1986; Tainton and Meyer, 1990; Cairncross and Brink, 1990; Winter and Brink, 1991) and Vredefort area (Minnaar, 1989). The only other regional, but generalised, compilation known is by Hutchinson (1975). Thus, a complete facies succession and succession boundary analysis of the West Rand Group is lacking.

This contribution presents the first comprehensive basin-wide stratigraphic analysis of the West Rand Group and utilises both outcrop and drill core data to construct reference profiles. The reference profiles serve as a basis for lateral stratigraphic correlation and extrapolation when certain areas are characterised by poor outcrop or when little to no data is available. These stratigraphic correlations can be used to define major time-lines in the sequence, i.e., regressive and transgressive time boundaries, which is essential for understanding lateral facies variation within time-related sequences. This contribution focuses particularly on the reconstruction of arenaceous depofacies and their evolution through time with recognition of sediment dispersal patterns, relative proximity to source areas and the nature of disconformities.

The localities of cores and outcrop profiles used in this study are given in Figure 1. Stratigraphic reference profiles were constructed by geological field mapping of the major outcrop areas of West Rand Group strata:

• West Rand: Randfontein, Witpoortjie and Rant-en-dal gorge areas;

• Klerksdorp: Near old Buffelsdoorn Mine, on Buffelsdoorn 389IQ and Welgegund 390IP and on farms Rooipoort and Syferfontein 30 km north of Klerksdorp;

• Heidelberg: East of Heidelberg on farms Blinkpoort, Rietpoort and Bothashoek;

• Parys-Vredefort: Vredefort Dome area on farms Wolwehok, Mooihoek, Eerstegeluk, Mizpah and Eensgevonden, as well as Kopjeskraal (Parys) and Buffelshoek;

• South Rand: Vicinity of the Hexrivier Gold Mine on farms Driefontein 632, Hexrivier 634 and Wolvenfontein 652;

• Ottosdal, Rhenosterspruit, and Ventersdorp.

During geological field mapping, it was found that structural deformation is mainly confined to fault and shear zones, as well as along diabase sill contacts. However, within fault blocks, stratigraphic relationships and sedimentary textures are generally well preserved. For example, in many of the cores, sequences are virtually undeformed with the finest of micro-sedimentary structures preserved in the shales (Martini, 1976). The presence of diabase sills complicated the measurement of stratigraphic thicknesses in some areas, especially of argillaceous rock units, which tend to be preferentially intruded by the sills. In addition, diamond drill cores normally only intersect small intervals of the West Rand succession leading to some correlation uncertainty. However, in most cases, correlation could be done with a high level of confidence because of the stratigraphic reference profiles available for the different areas. Correlation is best done on the basis of stratigraphic relationships between lithofacies units, i.e., lithofacies associations or successions (‘sedimentary cycles’ or ‘increments of sedimentation’) in the sequence.

Palaeocurrent measurements are supplemented with data from Camden-Smith (1980), Watchorn (1981) and Minnaar (1989). In addition, the correlation diagrams and isopach maps are supplemented with data from Coetzee (1960), Button (1970), Jansen (1977), Camden-Smith (1980), Tweedie (1986), Ellis (1991), Winter and Brink (1991) and Fletcher (1996). Grain size variation, palaeocurrent, isopach and sediment dispersal maps can be found in Supplementary Information A. Reference profiles and core logs used in the correlation diagrams and isopach maps can be found in Supplementary Information B. (Supplementary data files are archived in the South African Journal of Geology repository (https://doi.org/10.25131/sajg.127.0015.sup-mat)).

Schematic three-dimensional models of the two major unconformities were prepared using open-source modelling software LoopStructural (Grose et al., 2021). The software was used to interpolate conformable stratigraphic units between drill core observations using the finite-difference implementation of the discrete smooth interpolator (Mallet, 1992). The surfaces were subsequently imported into Blender software to construct block diagrams for visualisation.

The three subgroups of the West Rand Group: Hospital Hill, Government and Jeppestown, are bounded by disconformable contacts situated at the base of the (a) Orange Grove quartzite, (b) Promise diamictite, (c) Buffelsdoorn Reef and (d) Main Reef (Figure 2). In broad lithostratigraphic terms, the three subgroups can be defined as follows:

• The Hospital Hill Subgroup was deposited sometime after 2 954 ± 8 Ma (youngest age of Dominion lavas; Paprika et al., 2021). It is a succession composed of medium to coarse-grained orthoquartzite units interbedded within an extensive sequence of shale that also contains two prominent iron-formations and several magnetic mudstone beds. Argillaceous quartzite and gritty/conglomeratic units are subordinate and become better developed towards the top of the subgroup.

• The Government Subgroup is composed of interbedded conglomerates, gritty to coarse-grained argillaceous quartzites, quartz wackes and argillites. It contains three prominent and laterally continuous diamictite beds. Many of the quartzite units have sharp erosional bases and are poorly sorted. Orthoquartzites are subordinate. Some of the shales are magnetic and one very prominent finely laminated iron-formation is present.

• The Jeppestown Subgroup is composed of slightly argillaceous to argillaceous quartzites interbedded with shale, along with minor orthoquartzite and magnetic mudstone. The quartzites are medium-grained and exhibit a characteristic striped appearance. Very fine-grained quartz wackes are especially well developed. It contains a very good marker bed in the form of the Crown amygdaloidal lava that was dated at 2 914 ± 8 Ma (Armstong et al., 1991). The minimum age of the West Rand Group is 2 902 ± 13 Ma (youngest detrital zircon from the bottom of the Johannesburg Subgroup; Kositcin and Krapez, 2004).

Note that the term ‘shale’ encompasses all argillites ranging from very fine-grained quartz wackestones to siltstone and mudstone. Additionally, the term ‘structural basin’ refers to that part of the original depository preserved after structural deformation and erosion.

The Hospital Hill Subgroup is composed of the Orange Grove, Parktown, Brixton, Bonanza and Eleazar formations. The Orange Grove Formation contains three well sorted orthoquartzite units. The Parktown Formation contains five excellent marker beds: the Water Tower and Contorted Bed iron-formations and the Bulskop, Ripple and Speckled marker orthoquartzite beds. For the most part, however, the Parktown Formation is composed of siltstone and shale. The Brixton Formation comprises three fuchsitic orthoquartzites: the Versterkop, Witkop and Rangeview members. Three shale units are also present, the lower of which is named the Blinkpoort Member and the upper, the Albida Member (Figure 2). The Blinkpoort Member is normally characterised by the presence of three thin quartzite marker beds and a magnetic mudstone unit, which is typically manganiferous. A magnetic mudstone unit is also developed in the Albida Member. The Bonanza Formation is generally composed of coarse-grained, argillaceous quartzite with minor orthoquartzite. Conglomerate beds are developed in the upper part of the formation. The Eleazar Formation is composed of shale with thin interbeds of quartzite.

The Government Subgroup is composed of the Promise, Coronation, Tusschenin, Palmietfontein, Elandslaagte and Afrikander formations. The Promise Formation consists of argillite, two coarse-grained quartzites (Breaunanda and Hamberg members) and a diamictite marker bed (Witfontein Member) at its base (Figure 2). The Coronation Formation contains a basal argillite and three overlying marker beds: an orthoquartzite, a diamictite (Kensington Member) and a finely laminated iron-formation (Silverfields Member). The Tusschenin Formation is a coarse-grained quartzite unit with a sharp basal contact (Coronation Reef). The Palmietfontein Formation is predominantly argillaceous in composition with some magnetic mudstone intervals and a central orthoquartzite marker unit (Townhouse Member). The Elandslaagte Formation is a sharp-based coarse-grained quartzite unit, with the Government Reef at the base and the Elandslaagte Reef near the top of the formation. The Afrikander Formation consists of argillite, quartzite (Noycedale and Steynskraal members) and diamictite (Lagerspoort Member; Antrobus, 1986).

The Jeppestown Subgroup is composed of the Koedoeslaagte Formation (orthoquartzite), Rietkuil Formation (argillite and magnetic mudstone), Babrosco Formation (striped quartzite), Crown Formation (lava), Roodepoort Formation (quartzite, wackestone and shale) and Maraisburg Formation (quartzite) (Figure 2). The Buffelsdoorn Reef is developed at the base of the Koedoeslaagte Formation. The Roodepoort Formation comprises two quartzite units (informally referred to as the Lower and Upper Roodepoort quartzites) and two argillite units. The Lower Roodepoort quartzite directly overlies lava of the Crown Formation. Argillite at the top of the Roodepoort Formation grades upwards into coarse-grained quartzite of the Maraisburg Formation, which was previously classified with the Johannesburg Subgroup of the Central Rand Group.

Three major unconformities have resulted in the preservation or erosion of certain lithological units in the basin. This has led to considerable confusion about stratigraphic relationships and correlations in the past. Consequently, it is important to note that:

• The Eleazar and Bonanza formations are absent in the northeastern part of the basin;

• The majority of the Afrikander Formation is absent in the central part of the basin;

• The Maraisburg Formation is absent in the eastern part of the basin.

These three major low-angle unconformities can be distinguished from disconformities, which are virtually parallel to strata on a basin-wide scale. Excellent examples of disconformities include the Coronation, Rivas, Government and Veldschoen reefs.

An isopach map of the thickness of strata from the base of the Orange Grove Formation to the low-angle unconformity at the base of the Promise diamictite (Witfontein Member) illustrates gradual thinning in an east-southeasterly direction from approximately 3 000 m-thick in the Klerksdorp area to 900 m-thick in the South Rand area (Figure 3A). Thickness of strata between the Promise diamictite unconformity and the unconformity at the base of the Koedoeslaagte Formation is remarkably consistent around 1 000 m (Figure 3B). The sequence between the Buffelsdoorn Reef and the Main Reef of the Central Rand Group also thins gradually in an easterly-southeasterly direction from almost 1 300 m-thick near Klerksdorp to around 500 m-thick near Heidelberg (Figure 3C). In the South Rand and Evander areas, much of the sequence has been removed by erosion prior to deposition of the Central Rand Group (<120 m).

Lateral correlations of lithostratigraphic units of the Hospital Hill Subgroup are excellent (Figures 4, 5 and 6). With reference to lateral variations and sediment dispersal patterns within specific lithological units, the following statements can be made:

The Orange Grove Formation forms a remarkably consistent and homogenous sand sheet deposit at the base of the Hospital Hill Subgroup. It rests unconformably on a basement of Mesoarchaean Dominion Group volcanics to the west and Archaean granitoids and greenstones to the east. Towards the east, three distinct orthoquartzites are developed, which are separated by argillite units. However, towards the west, these shales appear to pinch out so that a three-fold subdivision can only be recognised by the development of alternating coarse-grained orthoquartzitic units and fine-grained argillaceous quartzite units (Figures 4 and 6A). The thickness of the Orange Grove Formation ranges from 50 to 150 m and it appears to gradually thin in a southerly direction (Figure S1). It almost pinches out locally at Rhenosterspruit to the west of Klerksdorp. In the Syferfontein area near Klerksdorp, Watchorn (1980) indicated that the Parktown shales are in direct contact with the Dominion Group. Prominent sedimentary structures include small-scale trough and planar cross-bedding, as well as wavy flat-lamination and wave ripple marks. The quartzites are generally composed of well sorted and rounded quartz grains (Blane, 2013). The average grain size of the unit is medium to coarse-grained, but becomes fine to medium-grained in the Klerksdorp, Vredefort and Welkom areas (Watchorn, 1980; Ellis, 1991). Based on the coarsest particles observed, small pebble conglomeratic beds are present to the northeast, i.e., Delmas (Button, 1970) and Heidelberg areas (Camden-Smith, 1980), fining to a maximum grain size of coarse/medium-grained in the Vredefort area. The conglomerate at Rhenosterspruit is abnormal in containing boulder-sized pebbles. Palaeocurrent patterns are polymodal to bimodal/bipolar. Often two sets of bimodal/bipolar palaeocurrent distributions are present at one locality. The main modes are northeast-southwest, north-south and northwest-southeast. The uppermost orthoquartzite bed at Ottosdal displays a strong mode to the southeast.

The Water Tower Bed is the lowermost iron-rich horizon in the West Rand Group and is predominantly composed of magnetic mudstone (i.e., ‘the Water Tower slates’). However, in regions such as Koster, Vredefort and Heidelberg, it can occur as an iron-formation. At Koster, the Water Tower Bed and associated magnetic mudstone are well developed (~50 m) and form the upper part of a broad upward fining sequence originating in the orthoquartzites of the Orange Grove Formation (Roering, 1968). The lower part of the iron-formation is comprised of massive, wavy and podded chert mesobands set in a matrix of magnetite-siderite or magnetite-ankerite felutite (Smith et al., 2013). Contacts between chert mesobands and felutite are gradational and poorly defined. The upper part is composed of laminated magnetite-siderite with well-defined chert bands (Smith et al., 2013). The thickness of the Water Tower Bed generally ranges between 15 and 30 m and thins in a southeasterly direction (Figure S2).

The disconformity at the base of the Bulskop Marker Bed cuts gradually down into the sequence in a northerly direction from the Klerksdorp and West Rand areas towards Koster (Figures 4, 6B and S3). At Koster, the Bulskop Marker Bed is in direct contact with Water Tower iron-formation and incorporates very large pebble-sized intraclasts of this iron-formation near its base. Further to the south, the base of the marker bed is situated a few meters above the Water Tower iron-formation. The Bulskop Marker Bed is apparently absent in the Ottosdal and Rhenosterspruit areas. The marker also thickens and coarsens in a northerly direction from less than 1 m-thick at Krugersdorp to 44 m-thick at Koster. Button (1970) mentions a 12 m-thick quartzite, situated some distance above the Orange Grove Formation in the Delmas area. According to Button’s (1970) profile, this quartzite is closely associated with magnetic mudstone, which could point towards it being an equivalent of the Bulskop Marker Bed. Sedimentary structures include wave ripple marks and hummocky cross-stratification. At Krugersdorp, the orthoquartzite is medium to coarse-grained, extremely well sorted and contains well rounded quartz grains. With reference to coarsest particles present, it fines from containing large rip-up clasts at Koster to medium-grained quartzite near Vredefort.

As far as could be established, the Ripple Marker Bed is sharp and erosively-based at all localities. However, it is apparently absent near Ottosdal. The orthoquartzite thins in a southerly and southwesterly direction from 18 m-thick at Koster and 11 m-thick at Heidelberg to less than 5 m-thick in the Krugersdorp, Vredefort and South Rand areas (Figures 4 and S4). Near Welkom, the unit thickens to around 30 m (Ellis, 1991). In the Krugersdorp area, the quartzite exhibits small-scale trough and planar cross-bedding, as well as wave ripple marks and associated cross-lamination (Figure 6C). Coarsest particles fine from coarse-grained to medium-grained sand in a southerly direction. Palaeocurrent data are sparse with northeast-southwest bimodal/bipolar distributions near the Ventersdorp and South Rand areas.

The Speckled Marker Bed displays two lobes of relatively thick development (70 to 100 m). One lobe is situated in the Koster and Rhenosterspruit areas and the other in the Heidelberg area (Figures 4, 6D and S5). In a time-stratigraphic sense, only the uppermost orthoquartzite bed in the Koster area should be regarded as the equivalent bed as seen on the West Rand. At Koster and other areas of thick development, the quartzite forms part of a coarsening upward sequence and displays gradational contacts with the underlying shale. In other areas like Klerksdorp, Potchefstroom and Krugersdorp, where the unit is relatively thin (<10 m), the basal contact is sharp erosive. In the Krugersdorp area, the orthoquartzite exhibits wavy flat-bedding and may also contain small clasts of angular/subangular chert. Coarsest grain-size development is in the Koster and Heidelberg areas, where small pebble conglomerates are present. However, at Rhenosterspruit, where the unit is ~80 m-thick, coarsest particles observed are in the coarse/very coarse sand size range. Based on the coarsest particles observed, there is a general trend of fining from north and northeast towards the south and southwest, respectively, i.e., towards the Vredefort area. Palaeocurrent patterns are typically bimodal/bipolar with northeasterly-southwesterly and northerly-southerly directed modes dominant.

The Contorted Bed is the most conspicuous marker bed in both the West Rand and Central Rand groups (Camden-Smith, 1980) and represents the only true regionally extensive banded iron-formation (BIF) in the Witwatersrand Supergroup. Its internal stratigraphy is best preserved in the TF1 core near Koster, where it overlies a coarsening upward facies succession of magnetic mudstone, siltstone-shale and greywacke. The bed transitions from an oxide-facies BIF (hematite-magnetite with jasper bands) to a mixed oxide-carbonate BIF (magnetite-siderite-chert) and finally to a carbonate-facies BIF (siderite-ankerite-quartz) (Smith et al., 2013). Chert mesobands are well-defined and microbanded, and can occur as even, wavy or podded structures. Near the upper contact, magnetic mudstone is interbedded with finely laminated iron-formation and thus represents a transitional facies between iron-formation and fine siliciclastics. Although the thickness of the bed is uniform along the margins of the structural basin (10 to 20 m), it is apparently well developed in the Vredefort and Heidelberg areas (~50 m-thick; Figures 6E and S6) but rather poorly developed in the South Rand and Welkom areas, where it occurs as magnetic mudstone and Fe-stained contorted shales with cherty concretions, respectively (Ellis, 1991).

The Versterkop Member generally exhibits a gradational lower contact with shale of the upper Parktown Formation, which is also known as the Observatory Shale Member (Figure 6F). It is sheet-like in form and generally comprises a lower flat-laminated orthoquartzite unit overlain by three upward fining successions of conglomerate/gritstone to medium-grained, small-scale trough cross-bedded orthoquartzite (Figure 6G). Thicknesses decrease in a southeasterly direction from 260 m near Randfontein to 40 m in the Vredefort area (Figures 4 and S7). Thickest known development is 500 m in the Ottosdal area. However, it is possible that this thickness includes quartzites from the overlying Blinkpoort Member due to lateral facies variation. The quartzite is well sorted, medium to coarse-grained and contains well rounded quartz grains (Blane, 2013). Based on the coarsest particles observed, the Versterkop Member fines from containing small pebbles near Krugersdorp to coarse/very coarse-grained sand in the Ottosdal area and medium-grained sand in the Vredefort area. Palaeocurrent distributions are polymodal, with prominent distributions to the southwest and south. In some localities, bimodal/bipolar distributions are also present, i.e., Klerksdorp and South Rand areas.

The Witkop Member forms part of an upward coarsening succession from Blinkpoort shale into trough cross-bedded orthoquartzite and in some locations, topped with gritstone lags. However, in the Klerksdorp profile, the quartzite unit is argillaceous and displays relatively poor grain sorting. The unit thins gradually in a southeasterly direction from about 200 m-thick in the Klerksdorp and Randfontein areas to 70 m-thick in the Vredefort area (Figures 4 and S8). Maximum particle size increases from coarse/very coarse sand in the Vredefort area to granule-bearing in the Heidelberg and South Rand areas, and small pebbles in the Klerksdorp area. Palaeocurrent distributions in the orthoquartzites are polymodal with strong southerly directed modes. In the South Rand profile, a strong east-west directed bimodal/bipolar distribution is present. Palaeocurrents in the Klerksdorp quartzites display a unimodal distribution towards the south-southeast.

The Rangeview Member displays two lobes of lateral thickening and coarsening: one from Vredefort towards Klerksdorp and the other from Vredefort towards Heidelberg (Figures 4 and S9). In the Krugersdorp profile, small-scale trough cross-beds, large-scale wave ripple marks and gritstone beds are developed. In the Klerksdorp profile, virtually the whole unit is coarse to very coarse-grained, poorly sorted and argillaceous. Here, large pebble conglomerates are present with interbedded quartzites displaying unimodal palaeocurrent distributions directed towards the southeast. Similarly, in the Heidelberg area, the central part of the unit contains small pebbles and medium to coarse-grained argillaceous quartzite with poor to medium grain sorting. Palaeocurrents in this unit are unimodally directed towards the south. Palaeocurrent patterns in the orthoquartzites are polymodal with a dominant mode in a southerly direction. In the South Rand area, a bimodal/bipolar mode is developed in a west-northwesterly and east-southeasterly direction.

The Albida Member is composed of interlaminated siltstone and shale, as well as magnetic mudstone (Figure 4). Manganese-bearing carbonate concretions have been documented from this interval near Carletonville (Cochrane, 2009; Smith et al., 2023). The shale is thickest in the Randfontein and Parys areas (270 to 280 m; Figure S10). The contact between the shale and magnetic mudstone of the Albida Member with the lower quartzite bed of the Bonanza Formation is gradational.

The Bonanza Formation is only developed in the western part of the structural basin. This is due to removal by erosion in the eastern part of the basin prior to deposition of the Promise diamictite (Figures 2, 5 and S11). The internal stratigraphy of the formation consists of lower and upper quartzite units, which are commonly bisected by a thin argillite bed (40 to 50 m), and an upper conglomeratic/gritstone unit. In the Carl1 core, the lower quartzite unit consists of stacked scour-based fining upward facies successions ranging from gritstone lags to medium-grained quartzite. The quartzite is commonly striped, with a low to moderate degree of sorting and exhibits ubiquitous small-scale trough cross-bedding and ripple cross-lamination. The upper quartzite overlies the central argillite bed with a gradational contact and coarsens upwards into a zone containing abundant pyritic gritstone bands and small-scale trough cross-beds. The upper quartzite unit is overlain with a sharp scour contact by a sequence informally known as the Upper Conglomeratic and Gritstone unit (Figures 6H and S12). Two conglomeratic beds are developed within this unit. The first is developed in the basal part of the unit and is referred to by Tainton and Meyer (1990) as the ‘Lower Promise Reef’; herein referred to as the Bonanza Reefs. In the Randfontein area, the Bonanza Reefs overly a disconformity resulting in the removal of the central argillite bed and part of the lower quartzite bed. The second conglomeratic unit is developed at the top of the Upper Conglomeratic and Gritstone unit, overlying 30 to 40 m of orthoquartzite, and is referred to as the ‘Upper Promise Reef’ by Tainton and Meyer (1990). However, it is better referred to as the Red Reef to avoid confusion with the overlying Promise Formation (Watchorn, 1980). This reef is very well exposed on the farm Buffelsdoorn 389IQ near the old Buffelsdoorn Gold Mine and consists of a well-packed conglomerate. The Bonanza Formation thickens from 300 to 400 m in the Randfontein and Vredefort areas to 765 m northwest of Carletonville and up to 910 m in Klerksdorp (Winter and Brink, 1991). Coarsest development of the Bonanza Formation is in the Carletonville-Randfontein area. Palaeocurrent distributions are unimodal to weakly bimodal in southerly directions. Polymodal palaeocurrent directions are present in the Vredefort profile (southeasterly and two opposing modes in easterly and westerly directions).

The low-angle unconformity at the base of the Promise diamictite cuts down into the sequence in an easterly direction (Figure 5). The result is that a series of stacked coarsening upward shale-quartzite facies successions are preserved between the Bonanza Formation and the Promise diamictite to the west and north of Klerksdorp. This sequence, known as the Eleazar Formation, outcrops on Eleazar, Welgegund 390IP and Buffelsdoorn 398IP near the old Buffelsdoorn Mine and was intersected in cores KD2 and KD3. A thin pyritic pebble conglomerate is developed immediately below the Promise diamictite. This conglomerate may also be present in the Randfontein and Benoni areas.

A palaeogeological map of the floor of the Promise diamictite illustrates the effects of erosion prior to deposition of the diamictite (Figure 7). To the west and north of Klerksdorp, the diamictite rests on the shale-quartzite sequence of the Eleazar Formation preserved above the Bonanza Formation. The Eleazar Formation is up to 170 m-thick in this area. A thin veneer of the Eleazar Formation is also preserved below the diamictite in the Parys-Vredefort area. However, for most of the central part of the basin, the diamictite rests directly on the Bonanza quartzite. To the east in the Heidelberg area, the Bonanza Formation is absent and the diamictite rests directly on the Albida Member of the Brixton Formation. Further northeast, in the Benoni area, the diamictite rests directly on the upper Brixton quartzite. The amount of erosion is illustrated by the thickness of the sequence preserved from the top of the Rangeview Member to the base of the Promise diamictite. This thickness is in excess of 1 080 m near Klerksdorp, 430 m at Carletonville, 100 m at Heidelberg and 10 m at Benoni.

Lateral and vertical stratigraphic relationships in the Government Subgroup are well illustrated by four correlation charts (Figures 8, 9, 10 and S13) and accompanying outcrop/core photographs (Figure 11).

In most localities, the Witfontein diamictite consists of a single bed. However, in the Randfontein, Krugersdorp and Benoni areas, the diamictite consists of lower and upper beds, each with sharp basal and upper contacts (Figures 8 and 9). The quartzites situated between the two diamictite beds are characterised by sharp erosional bases and are composed of poorly sorted, very coarse-grained to gritty quartzites. The quartzites contain large-scale trough cross-beds (a few meters across and 0.5 m high) and the lowermost quartzite unit contains large dislocated blocks. In the Randfontein area, Roering (1968) documented erosion channels and poorly developed conglomerate bands within the diamictite. In relation to thickness, the diamictite thins from 60 to 70 m in the northern and northeastern parts of the basin (Krugersdorp and Benoni areas) to less than 10 m-thick in the Vredefort and Welkom areas (Figures 8, 9, 11A and S14). Although variation in grain size is difficult to estimate, the largest clasts are typically present in the Randfontein and Klerksdorp areas (decimetres in size) with fining taking place towards the southeast. Clast density in the diamictite also appears to increase towards Klerksdorp and Randfontein. In these two areas, abundant cobble-sized clasts are present, whereas in the Vredefort area, only small pebble-sized clasts were noted. Clast compositions include quartz, quartzite, shale, chert and lava (Camden-Smith, 1980; Watchorn, 1980; Guy, 2012). It is important to note that in the Benoni area, a finely-laminated carbonate-facies iron-formation, about 4 m-thick, overlies the diamictite. This differs to other parts of the basin, where the diamictite is overlain by shale or wackestone.

The Breaunanda orthoquartzite is sheet-like in form and normally overlies shale with a coarsening upward transitional contact (Figures 8, 9 and S15). However, on farm Buffelsdoorn 389IP near Klerksdorp, the contact is sharp and to the south of Klerksdorp in the R1 core, it is composed of three upward-fining successions, each with a sharp scour base. The orthoquartzite thins slightly in a westerly direction from 110 m-thick near Heidelberg to about 50 m-thick in the Klerksdorp and Vredefort areas. The orthoquartzite is moderate to well sorted, medium to coarse-grained and contains small to large-scale trough cross-beds (Figure 11B). Coarsest development is in the Klerksdorp area, where abundant conglomerate and gritstone bands are present (Figure 11C). The unit also coarsens in a northeasterly direction from Vredefort towards Heidelberg. Palaeocurrent directions are weakly polymodal with a southerly directed mode dominant. However, in the South Rand area, distributions are distinctly polymodal with a dominant mode directed towards the southeast.

The interval between the Breaunanda Member and the overlying Hamberg Member is normally composed of shale interbedded with quartzite. However, in the Klerksdorp area (BAB1 and R1 cores), the interval coarsens dramatically and contains a series of stacked upward coarsening facies successions (Figure 8). These arenaceous successions are less prevalent in the eastern part of the basin (Figure 9), and appear to be absent in the Welkom area (Figure 10; Ellis, 1991). In the Carletonville and Welkom areas, observations of an ‘intermediate diamictite’ (~10 m-thick, Fletcher, 1996) and zones of deformation and mud-cracks (Ellis, 1991) have been noted within this interval.

The Hamberg Member displays a coarsening upward transitional contact from the underlying shale. In the Krugersdorp area, it consists of a basal unit composed wavy flat-bedded quartzite truncated by a series of stacked, erosively-based gritty argillaceous quartzite facies assemblages. This is overlain by finer-grained quartzite displaying flat-bedding, as well as small to medium-scale planar and trough cross-bed sets. The Hamberg quartzite is thickest in the Carletonville and Randfontein areas (140 to 150 m). It thins towards Vredefort (60 m) and appears to pinch out near Welkom (Figures 8, 9 and S16). Coarsest development is in the Randfontein area, fining towards the south and southwest. It is important to note that at the South Rand profile, a well-packed large-pebble conglomerate is developed. Palaeocurrents are generally directed in a southerly direction. However, in the upper slightly argillaceous part of the unit in the Heidelberg profile, they are directed towards the west.

The Coronation Formation has a very persistent internal stratigraphy consisting of a basal shale unit (Figure S17) followed by a thin orthoquartzite (Figure S18), diamictite (Kensington Member; Figure S19) and a finely laminated iron-formation (Silverfields Member; Figure S20).

It is interesting to note that unlike the Promise diamictite, which rests on a marked low-angle disconformity, the Coronation diamictite rests persistently with a sharp contact on a sheet-like thin orthoquartzite bed. The diamictite thins in a southeasterly direction from 70 m-thick to the west of Klerksdorp to less than 10 m-thick in the Vredefort and Edenville areas. However, it also appears to thicken in a northeasterly direction towards Evander (Tweedie, 1986). In core, the diamictite is massive with poorly defined bedding and slump-fold structures (Figure 11D). Grain size varies sympathetically with thickness, with cobble-sized clasts present in the Klerksdorp area, large pebbles common in the Heidelberg and Benoni profiles, and small pebbles near Vredefort and Edenville. In the latter area, lonestones have been documented within the overlying Silverfields iron-formation (Guy et al., 2012). Variable clast compositions, including quartz, chert, fuchsitic quartzite, siltstone, shale, jasper, granite and lava, have been documented from the diamictite (Rogers, 1922, Camden-Smith, 1980; Guy, 2012).

The iron-formation overlying the diamictite is a finely laminated magnetite-silicate ferhythmite and typically contains no chert mesobands (Smith et al., 2013). It is well developed in the northern parts of the structural basin but grades laterally into magnetic mudstone, wackestone and carbonaceous shale (Figures 11E and S20). This observation is consistent with the detection of only minor magnetic anomalies in the Klerksdorp region relative to the West Rand and Vredefort areas (Krahmann, 1930; Paver 1939; Table A6 of Camden-Smith, 1980). The Coronation Formation is overlain by the Coronation Reef of the Tusschenin Formation with a sharp erosive contact.

The disconformity at the base of the Coronation Reef is subparallel to bedding over very large areas. It is only with detailed long-distance correlations that slight erosional discontinuities become obvious, particularly in the eastern areas of the basin, where the Coronation Reef rests almost directly on the diamictite (Figure 9).

The Tusschenin Formation is a sheet-like sand deposit and exhibits significant lateral facies variation. In the Klerksdorp area, the unit is composed of stacked scour-based successions of coarse-grained argillaceous quartzite with poor grain sorting. Further to the east in the Carletonville, Benoni and Heidelberg areas, interbeds of orthoquartzite are commonly observed (Figure 11F), whereas in the south near Vredefort and Edenville, the quartzites are fine-grained and contain interbeds of shale/ wackestone (Figures 10 and S13). The formation is thinnest in the Krugersdorp and South Rand areas (50 to 80 m) and thickens to around 200 m in the Klerksdorp and Edenville areas (Figure S21). Average grain size varies very little with coarse to very coarse sand prominent along the northern perimeter of the structural basin and medium to coarse sand in the southern and central areas. Palaeocurrent distributions are polymodal with dominant modes in southerly directions. Bimodal/bipolar east-west orientated distributions are present in the Heidelberg and South Rand areas.

Previously known as the ‘Government Shale’, the Palmietfontein Formation is a predominantly argillaceous sequence that ranges from 150 to 280 m in thickness. It normally contains an ortho-quartzite unit, known as the Townhouse Member, in the central part of the formation (Figures 8 and 9). However, in the Heidelberg and Edenville areas, a sharp-based orthoquartzite is developed between the Townhouse Member and the Tusschenin Formation (Figure 10). This quartzite, with its disconformable base, is informally referred to as Quartzite 1 of the Palmietfontein Formation (Figure S22). It may be a correlative of the sharp-based orthoquartzite unit forming the top of the Tusschenin Formation along the northwestern perimeter of the structural basin (Figure 8). This correlation implies that the shale unit present between Quartzite 1 and the Tusschenin Formation has been removed by erosion. Thus, in the Klerksdorp, Krugersdorp and Vredefort areas, Quartzite 1 is in direct contact with the Tusschenin quartzite.

The Townhouse Member is a thin, 10 to 50 m-thick sand sheet deposit correlatable over a very wide area (Figure S23). Thickest known development is in the northern part of the basin (~50 m in Carletonville and Benoni). It thins and fines in a southerly direction with a possible depositional limit near Welkom (Ellis, 1991). The unit is characterised by planar and trough cross-bedding. It is slightly coarser-grained in the Benoni area relative to the Klerksdorp and Vredefort areas. Palaeocurrent distributions are polymodal with southerly and southeasterly directions dominant.

The upper part of the Palmietfontein Formation is characterised by a very persistent symmetrical cycle of mudstone-wackestone-mudstone. Slump folds, load casts and silt/sand dykelets are abundant. The upper and/or lower mudstone units of the cycle are commonly magnetic. It is interesting to note that this symmetrical cycle has been intersected in core ZA1 near Clarens and in core LA1/68 near Ladybrand.

The Government Reef rests with a sharp disconformable base on argillite of the Palmietfontein Formation (Figure 11G). The disconformity is virtually parallel to bedding and detailed correlations indicate only slight variations in its stratigraphic position (Figures 8, 9 and 10). The only area where the disconformity may cut down prominently through strata of the Palmietfontein Formation is in the South Rand area (Figure 9). Although this could be explained by normal faulting, a similar condensed stratigraphic section has also been documented from the Evander area (Tweedie, 1986).

Like the Tusschenin Formation, the Elandslaagte Formation exhibits significant lateral facies variation. In the western part of the basin, i.e., in the Klerksdorp, Bothaville and Carletonville areas, the quartzites are argillaceous, coarse to gritty, erosively-based and poorly sorted. However, towards the east and southeast in the Benoni, Heidelberg, South Rand and Edenville areas, the quartzites become orthoquartzitic and separate into two distinct units (Figures 9, 10 and S24). The formation appears to thin along a central arcuate axis, connecting Welkom, Vredefort, Parys and the South Rand. Maximum known thickness is just over 120 m near Klerksdorp and around 100 m near Edenville and Heidelberg (Figure S24). Palaeocurrent directions are unimodal to the south-southeast near Klerksdorp and polymodal in the Vredefort and South Rand areas.

The Afrikander Formation is best preserved in the eastern part of the basin (Figure 9), where the maximum known vertical sequence of the formation is preserved below the Buffelsdoorn Reef unconformity (Figure 2). The formation is virtually absent in the central part of the basin due to erosion and only part of the basal shale unit is preserved (Figure 8). This shale, which is normally on the order of 20 to 40 m-thick, is thus a very good marker unit. Towards the northwest, in the Klerksdorp area, more sequence of the Afrikander Formation is preserved below the Koedoeslaagte unconformity.

The Afrikander Formation is composed of a basal argillite, which grades upwards into a prominent quartzite unit containing the lower Noycedale and upper Steynskraal members. The quartzite unit is succeeded by wackestone/shale, which is then sharply overlain by a thick diamictite unit known as the Lagerspoort Member (Antrobus, 1986) or ‘Blue Grit’. This diamictite is in turn overlain by shale forming the uppermost unit of the Afrikander Formation and Government Subgroup.

Primary thicknesses of the Afrikander quartzites appear to increase towards the Klerksdorp and Edenville areas (Figure S25). These values represent minima because some of the strata has been eroded prior to deposition of the Buffelsdoorn Reef. The lower Noycedale Member consists of two orthoquartzite units, which in some locations, are separated by a thin (<6 m) argillaceous bed (e.g., Edenville and Benoni areas). Gravel/granule-stone beds are ubiquitous in the Welkom, Klerksdorp and Benoni areas relative to the Edenville area, where it consists of banded grey orthoquartzites (Figures 9 and 10). In the Rietkuil syncline, the quartzites are scour-based, gritty, conglomeratic and argillaceous in character (Figure 8). The Noycedale Member is around 70 to 100 m-thick along the southwestern and northwestern parts of the basin, but thins dramatically to the east around Heidelberg and the South Rand (30 to 40 m). The quartzites are coarse to very coarse-grained in the Welkom, Klerksdorp, Benoni and Heidelberg areas and fine to medium-grained in the Edenville area. Palaeocurrent distributions are unimodal towards the southeast in the Heidelberg and South Rand areas.

The upper Steynskraal Member exhibits a much smaller pre-Koedoeslaagte erosional preservation than the Noycedale Member. Maximum thickness is around 90 to 100 m in the Edenville area but thins rapidly towards Heidelberg and the South Rand (<20 m) (Figures 9 and 10). In the Benoni area, it is composed of slightly argillaceous grey-white orthoquartzite with interbeds of khaki-green argillaceous quartzite, whereas in the Edenville area, it is composed of banded grey, green and white orthoquartzite.

Diamictite of the Lagerspoort Member is only preserved in the eastern part of the structural basin (Figures 9, 10, 11H and S26). It appears to thicken in an easterly direction from 200 m-thick in the Heidelberg area to over 500 m-thick near Evander (Tweedie, 1986). The thickness of 70 m preserved in the AM1 core represents a minimum because part of the unit has been removed by erosion prior to deposition of the Koedoeslaagte Formation. In the South Rand profile, the diamictite appears to be massive throughout. Overall, it fines upwards from clast-bearing wackestone at the base to granule-bearing wackestone in the upper parts. Chert, quartz, quartzite, siltstone, shale and blue-green (volcanic?) clasts are common. In addition to sedimentary clasts, clasts composed of granite and porphyritic lava have also been documented (Camden-Smith, 1980). It is interesting to note that the diamictite appears to fine in a west-northwesterly direction with the largest clasts observed in the South Rand area, up to 15 cm in diameter, whereas at Heidelberg they are only ~8 cm in diameter and at Edenville between 4 and 8 cm in diameter.

The nature of the structure that led to erosion of part of the Afrikander Formation is well illustrated by a model of its preserved thickness below the Koedoeslaagte Formation of the Jeppestown Subgroup (Figure 12). The Afrikander Formation thickens in a southeasterly and northwesterly direction away from a central axis of thinly preserved strata. Along this axis, which trends southwest-northeast, the Buffelsdoorn Reef of the Koedoeslaagte Formation rests on the basal shale unit of the Afrikander Formation. In the far northwestern reaches of the basin, it is possible that the Lagerspoort diamictite may be preserved below the Buffelsdoorn Reef unconformity.

Correlation of various units of the Jeppestown Subgroup is excellent over the entire structural basin and transgressive surfaces subdivide the sequence into a layer-cake fashion (Figures 13 and 14).

The unconformity at the base of the Buffelsdoorn Reef appears to be a very flat surface in general. However, it is known that in the Rietkuil syncline, to the west of Klerksdorp, the surface is channelised up to 45 m deep (Figure 14A; Watchorn and O’Brien, 1991). This channel-fill sequence includes the so-called Outer Basin Reefs 1-4 and can only be observed in good underground exposure. Therefore, they do not represent a practical base, in lithostratigraphic terms, for the Koedoeslaagte Formation. Consequently, the Koedoeslaagte Formation should be consistently defined as an orthoquartzite unit with a basal sharp- or scour-based conglomerate or gritstone bed, i.e., the Buffelsdoorn Reef or No. 5 Outer Basin Reef (Figure 14B). The Buffelsdoorn Reef is apparently only developed along the western part of the structural basin, where it ranges from very large pebbles in the Rietkuil syncline to large pebbles at Bothaville and small pebbles near Welkom. Further to the east, the Buffelsdoorn Reef is typically gritty in character.

The Koedoeslaagte orthoquartzite unit, which overlies the Buffelsdoorn Reef, represents a sand sheet deposit between 20 and 40 m-thick. It thins in an easterly direction towards Benoni, Heidelberg and the South Rand, where it is less than 10 m-thick (Figures 13 and S27). The orthoquartzite commonly displays trough cross-bedding and flat lamination. Grain size varies from coarse to very coarse-grained sand along the western periphery of the structural basin to medium to coarse-grained sand in a southeasterly direction towards Edenville. It is also slightly coarser-grained (coarse/medium) in the Heidelberg area. Palaeocurrents are directed in southerly to southeasterly directions, sometimes unimodal and in other localities polymodal. At the Afrikander Mine, in the Rietkuil syncline, palaeocurrent directions in the basal channel-fill are unimodal to the south-southeast (Watchorn and O’Brien, 1991). In contrast, the overlying orthoquartzite, which is known locally as the ‘White bar’, displays a bimodal-bipolar palaeocurrent distribution in a south-southeasterly and north-northwesterly direction (Watchorn and O’Brien, 1991).

The argillaceous Rietkuil Formation consists of a lower interval composed of interbedded siltstone and carbonaceous shale and an upper interval composed of magnetic mudstone. The contact with the underlying Koedoeslaagte Formation is gradational. At the South Rand profile, the Bird Reef of the Central Rand Group rests unconformably on Rietkuil argillite so that all other units of the Jeppestown Subgroup are absent. At Evander, the Jeppestown Subgroup is absent and the Bird Reef rests directly on the Lagerspoort diamictite of the Government Subgroup. The thickest known development of the Rietkuil Formation is ~180 m in the JY8 core to the south of Klerksdorp (Figure 13). This also represents original post-compactional thickness of the unit because it has a gradational upper contact with quartz-wackestone. Such gradational contacts are rare because in most cases, the argillite is overlain with a sharp disconformable contact by Babrosco quartzite or conglomerate of the Veldschoen Reef.

The Veldschoen Reef forms the base of the Babrosco Formation and is in direct disconformable contact with Rietkuil argillite. However, in cores JY8 and DHK, the reef is developed a short distance above the base of the formation on quartz-wackestone of the Rietkuil Formation or quartzite of the Babrosco Formation (Figure 13).

The Babrosco Formation thins in a southeasterly direction from 250 m-thick near Bothaville to about 60 m-thick in the Edenville area (Figure S28). The formation exhibits a characteristic striped appearance, which is in more detail, is composed of ~1 m-thick stacked scour-based fining upwards associations (Figure 14C). Each association is composed of small-scale trough cross-bedded cream-coloured quartzite overlain by grey, small-scale trough cross-bedded or flat-laminated fine-grained argillaceous quartzite. Greenish orthoquartzite interbeds are also present. Grain size decreases sympathetically from an average of coarse/very coarse-grained sand near Klerksdorp to medium-grained sand near Edenville. This fining trend also applies to the Veldschoen Reef, which is known to be a small-pebble conglomerate near Bothaville and a coarse-grained quartzite at Edenville. Palaeocurrents are mostly directed in a southeasterly direction, but near Krugersdorp, a southerly directed mode is dominant. Watchorn (1981) indicated that palaeocurrents are bimodal/bipolar with modes directed to the southeast and northwest in the Klerksdorp area.

The Crown lava represents one of the most persistent and reliable markers in the West Rand Group (Watchorn, 1980). Previously known as the ‘Jeppestown Amygdaloid’, the Crown lava is green to grey in colour, aphanitic and commonly amygdaloidal. Using the Zr/TiO₂-Nb/Y discrimination diagram (Pearce, 1996, after Winchester and Floyd, 1977), Humbert et al. (2021) classified lavas from the Edenville area as andesite or evolved basaltic andesite. In this area, but also in the South Rand and Free State areas, several lava flows have been detected and are typically marked by changes in colour (weathered brown, grey and green) and changes in the distribution of amygdales/ phenocrysts (Nel, 1933; Camden-Smith, 1980; Humbert et al., 2021; Supplementary Information B). In relation to thickness, the formation thins from 40 to 60 m-thick along the northern and northwestern periphery of the structural basin to 10 to 20 m-thick at Heidelberg, Bothaville and Edenville (Figures 13, 14D and S29). In the Edenville area, the occurrence of a sandstone unit between two lava flows is most likely related to fault duplication (Supplementary Information B). A reliable estimate for the Vredefort area is lacking. Minnaar (1989) indicated the lava to be in excess of 100 m-thick near the Vredefort Dome but this value could not be verified as the lava was not detected in the Parys outcrop profile. On the De Bron Horst, the Crown lava is about 40 m-thick but from there it thickens rapidly in a southerly direction to 110 to 180 m-thick south of Welkom. Note that agglomerate and tuff have been documented in core logs from both the eastern and western edges of the basin (Supplementary Information B). Hyaloclastites have also been documented in the Free State and Edenville areas (Humbert et al., 2021; Supplementary Information B).

The Roodepoort Formation consists of two quartzite units and two argillaceous units. The Lower Roodepoort quartzite bed overlies the Crown lava with a sharp contact. It thins in an easterly and northeasterly direction from about 50 to 70 m-thick in the Bothaville and Klerksdorp areas to less than 10 m-thick in the Parys and Heidelberg areas (Figures 13 and S30). It increases to about 25 m-thick at Delmas according to a profile by Button (1970). However, the contact with the overlying argillite is gradational so it is difficult to determine an exact upper contact. The Lower Roodepoort quartzite is similar in character to the Babrosco quartzite situated immediately below the Crown lava. Grain size varies from coarse/medium-grained sand along the western perimeter of the structural basin to medium/fine-grained sand in a northeasterly direction.

The lower shale unit is composed of graded wackestones and ripple cross-laminated siltstone in the western parts of the basin (Welkom, Bothaville, Klerksdorp and Carletonville areas). However, towards the east and southeast, near Benoni and Edenville, magnetic mudstone is present.

The Upper Roodepoort quartzite normally rests with a transitional contact on the underlying argillite. However, in the Heidelberg area, it has a sharp disconformable contact and contains large-scale low-angle planar cross-bed sets (1.2 m high with foresets up to 8 m long). This disconformity is thought to be equivalent to the base of a sequence of stacked, scour-based, fining upwards facies associations of gritty and poorly sorted quartzites developed in upper part of the unit in the Carletonville area. The quartzite thins and fines in an easterly direction; decreasing from ~240 m near Klerksdorp and Welkom to 10 to 25 m in the Parys, Heidelberg and South Rand areas (Figure S31). Palaeocurrent directions are dominantly south and southeasterly. At Parys, a bimodal/bipolar east-west direction is also present.

The upper argillite is about 180 m-thick in the Klerksdorp-Carletonville area and 150 m-thick near Edenville, but thins to 50 to 80 m-thick along a southwest-northeast trending axis connecting Bothaville, Parys and Benoni. The unit is typically composed of graded to flat-bedded, fine to very fine-grained wackestone and siltstone with minor mudstone (Figure 14E). In the Heidelberg area, hummocky cross-stratification and shallow wave ripple marks were observed. Magnetic mudstone is present in the central and southeastern parts of the basin near Parys and Edenville, respectively (Figure S13).

The upper Roodepoort argillite coarsens upwards into the basal orthoquartzite unit of the Maraisburg Formation. The orthoquartzite unit is overlain by argillaceous quartzite composed of stacked scour-based fining upward facies assemblages. The sequence becomes gritty and conglomeratic towards the top. The upper part of the Maraisburg quartzite has been removed by a low-angle unconformity prior to deposition of the Main Reef. The unconformity cuts deeper down into the sequence in a southeasterly direction from Carletonville towards Heidelberg (Figure 13), and southerly direction from Klerksdorp towards Bothaville, Welkom and Edenville (Figure 10). The result is a general decrease in thickness of the Maraisburg Formation from over 240 m-thick at Carletonville to 20 m-thick at Edenville (Figure S32). At Heidelberg, all of the Maraisburg Formation has been removed by erosion prior to deposition of the Main Reef (Figure 14F). The Maraisburg Formation is gritty to pebbly in the Carletonville-Klerksdorp area and may fine to the southeast. However, this apparent fining may be related to the removal of the upper coarse-grained part of the unit by erosion prior to deposition of the Main Reef.

Detailed descriptions have been presented on each of the formations, members and marker beds in the West Rand Group, as well as the unconformity surfaces that bind them. In the sections that follow, this information will be used to track both the temporal evolution of depofacies and the lateral migration of depocenters in each of the three sequences (I, II and III). These elements, along with basin architecture and nature of disconformities, will then be collectively assessed to determine the geodynamic configuration of the depository.

Three broad depofacies are recognised in the West Rand Group: marine shelf, fluvial braidplain (FB) and debris flow (D) (Table 1). Marine shelf deposits can be further subdivided into inner shelf sands (MSS), transitional zone (SM-), outer shelf muds (SM) and starved shelf subfacies (SS). Vertical stratigraphic relationships between the depofacies are shown in a composite reference profile for the West Rand Group (Figure 15). The succession is subdivided into three major sequences:

In Sequence I, extending from the unconformity at the base of the Orange Grove Formation to the base of the Promise diamictite, proximal marine shelf orthoquartzite deposits are well developed in the lower and middle parts of the sequence (MSS). Fluvial braidplain quartzites become more prominent in the Bonanza Formation (FB) near the top of the sequence. Wedges of fluvial braidplain deposits first appear in the second Brixton quartzite (Witkop Member) at Klerksdorp and in the third Brixton quartzite (Rangeview Member) at Klerksdorp and Heidelberg. The majority of proximal shelf orthoquartzites in Sequence I, particularly the three main Brixton quartzites, have transitional lower contacts and are thus of the progradational-aggradational variety (P-MSS). Orthoquartzites characterised by erosional bases that overly shale with sharp contacts represent erosively regressive marine shelf sand deposits (ER-MSS), typically associated with rapid rates of sea level fall (Plint, 1988). The Orange Grove quartzites and those of the Bulskop and Ripple markers are good examples of these disconformities, which are also known as type 2 unconformities. The Speckled Marker quartzite displays progradational-aggradational transitional bottom contacts with shale in areas where it is well developed, like at Koster and Heidelberg. However, rapid rate of sea level fall apparently led to the development of a sheet of erosively regressive marine shelf sand at the end of deposition of the unit.

Two main varieties of inner marine shelf facies are recognised in Sequence 1: current-dominated (MSS-C) and storm-dominated (MSS-S) marine shelf sands. The current-dominated inner shelf facies (MSS-C) is represented by medium to coarse-grained and well sorted orthoquartzites with small-scale trough cross-bedding and polymodal palaeocurrent directions. Minor structures include small-scale planar cross-bedding, as well as wavy flat-bedding and ripple marks. Orthoquartzites of the Orange Grove and Brixton formations are characterised by these structures. The abundance of small-scale trough cross-bedding suggests that small dunes were the dominant bedform in the sand shelf environment, which were occasionally reworked into small sand bars to produce planar cross-bed sets and ripple marks. These features, together with the absence of clay partings, suggests continuous winnowing by currents and a setting above normal wave base on the shelf. Flow and dispersal patterns were complex as suggested by the polymodal palaeocurrent directions. Tide, meteorological and oceanic currents may have played a role but it is difficult to ascertain which current was dominant. Hence the use of the general term ‘current-dominated’. However, it is known that tidal (Stride, 1963) and oceanic currents (Flemming, 1978) can produce very large-scale bedforms and associated cross-bed sets; features not characteristic of West Rand Group orthoquartzites. Further, the ubiquity of polymodal palaeocurrent directions (relative to bimodal/bipolar distributions) argues against a tide-dominated system. Evidence for a storm-dominated shelf system (MSS-S and SM-S), such as bidirectional and multidirectional sole marks, hummocky cross-stratification, symmetrical wave-ripple marks and flat lamination, are present in the orthoquartzites of the lower Parktown and Brixton formations (Beukes, 1996). Thus, it is most likely that meteorological currents, such as wind-driven, wave-drift and storm-surge currents were dominant during deposition of the Orange Grove, Parktown and Brixton quartzites.

The argillites of Sequence I were deposited in transitional to distal shelf environments. Wavy and lenticular laminated shale/siltstone/quartzite and wave/current-ripple cross-laminated siltstone/quartzite beds are indicative of traction transport by bottom currents, wave action and storm surges (SM-C, SM-W and SM-S). Iron-rich rock units, such as the Water Tower and Contorted Beds, represent starved shelf deposits, or condensed sections, that were formed during periods of maximum rate of sea level rise in the depository (SS). Maximum overall rate of sea level rise, i.e., accommodation space in the basin, was most probably attained during deposition of the Contorted Bed BIF.

Sequence II, extending from the low-angle unconformity at the base of the Promise diamictite to the low-angle unconformity at the base of the Koedoeslaagte Formation, contains three diamictite deposits (D) of which the Lagerspoort diamictite (Blue Grit) near the top of the sequence is thickest (Figures 2 and 15). The wide lateral extent of the diamictites, their sheet-like form and association with shelf deposits are difficult to explain by slope-induced gravity flow processes. Perhaps the diamictites are related to gravity flows associated with glacial events, or alternatively, basin tectonism and sudden basin subsidence associated with thrust loading or normal faulting. Rapid eustatic sea level fall may create increased slopes but the shelf would become highly incised by fluvial stream erosion and gravity flow deposits would be channelised.

Fluvial braidplain depofacies are very prominent in Sequence II and are characterised by the presence of sharp erosively-based deposits consisting of argillaceous quartzites, gritstones and small-pebble conglomerates with trough and planar cross-bedding, moderate to poor sorting, unimodal palaeocurrent directions and stacked scour-based fining upwards facies successions (FB). Disconformities associated with fluvial braidplain depofacies must be related to rapid rates of sea level fall (small to negative accommodation space) in the depository. Regressive (P-MSS) and transgressive (T-MSS) orthoquartzitic marine shelf sands are associated with fluvial braidplain quartzites and are commonly developed at their bases and at their tops (e.g., Hamberg Member). Some of the fluvial braidplain quartzites grade laterally into orthoquartzitic marine shelf sands, e.g., the Tusschenin and Elandslaagte formations. Maximum marine flooding was most probably obtained during deposition of the Coronation microbanded iron-formation (SS), immediately following deposition of the Coronation diamictite. Such close association of lowstand diamictite deposits and highstand iron-formation deposits could be reconciled with a glacial event followed by rapid break-up of an ice shelf and the development of a major transgressive starved shelf iron-formation.

Sequence III commences from the low-angle unconformity at the base of the Koedoeslaagte Formation to the base of the Main Reef (Figures 2 and 15). Although it starts with a transgressive sequence (T-MSS), i.e., the Koedoeslaagte orthoquartzite and Rietkuil argillites, Sequence III is essentially an upward coarsening sequence composed of stacked shale-quartzite facies successions (P-MSS). The striped quartzites of the Babrosco and Roodepoort formations appear mineralogically immature, yet they display marine-like bimodal/bipolar palaeocurrent distributions (Watchorn, 1981). Although their true depositional setting remains uncertain, they could represent immature marine shelf sands. Rapid aggradation due to basin subsidence and high rate of sediment supply may have been responsible for the relative immaturity of the sands in a shallow marine shelf setting. Increased rates of sediment supply associated with basin subsidence are also suggested by the abundance of graded turbiditic, fine to very fine-grained quartz wacke deposits in the argillite units (MSS-Tb and SM-Tb). Abundant loading and dewatering sedimentary structures, such as ball and pillow structures, contorted bedding and sand dykelets in the wackestone, are suggestive of rapid rates of deposition. These turbidite deposits may be related to storm wave action or active delta progradation on the shallow shelf. Volcanism, reflected by outpouring of the Crown lava, does not appear to have had an effect on the sedimentary pattern in the sequence. The Babrosco quartzite, immediately below the lava, and the Lower Roodepoort quartzite, immediately above the lava, are virtually identical in character. Thus, if outflow of the lava was related to some tectonic event or feature, that feature must have been situated outside the perimeter of the presently preserved structural basin. Fluvial braidplain deposits are well developed in the upper parts of the Roodepoort (T-FB) and Maraisburg formations (R-FB).

The distribution of depofacies and sediment dispersal patterns, and their progression through time, can be depicted by a series of diagrams for each of the three major sequences. However, it is important to note that wherever reference is made to the direction of shorelines or source area, that does not imply the exact geographic position of the shoreline or source area. These positions remain largely unknown because only part of the original depository is preserved in the structural basin. Marine palaeocurrent types indicated on the diagrams (longshore, on-offshore, etc.) are inferred from a combination of lithofacies, thickness and palaeocurrent data. Such currents are normally independent of palaeoslope. Only fluvial palaeocurrent directions are thought to be normal to depositional slope, which in turn may either conform with, or be oblique to perpendicular to, the erosional slope.

With reference to Sequence I, a rather consistent sedimentary pattern appears throughout (Figures S1 to S12). During deposition of the marine quartz sands of the Orange Grove, Parktown and Brixton formations, the shoreline appears to have been arcuate in form along the northwest, north and northeast periphery of the structural basin. Two areas of coarser sediment input appear to have been present rather consistently, specifically to the northwest of Klerksdorp and to the northeast of Heidelberg. Existence of a northeastern source area is well illustrated by the occurrence of conglomeratic units in the Orange Grove Formation near Delmas and Heidelberg (Camden-Smith, 1980). A similar situation is apparent in the lower Brixton quartzite (Versterkop Member). This northeastern source area, or depocenter for coarse marine siliciclastics, was also present during deposition of the Ripple and Speckled marker beds and perhaps less so during deposition of the Bulskop Marker Bed. It was coupled with the formation of thick coarse sand shoal deposits in the northwestern part of the depository in the vicinity of Klerksdorp, Ventersdorp and Koster. A local palaeohigh was apparently present in the floor of the depository to the west of Klerksdorp during deposition of the Orange Grove Formation. However, on west side of this palaeohigh, the Orange Grove Formation thickens again indicating an area of subsidence. This area of subsidence appears to have been very active during Versterkop times to allow for thick accumulation of marine sand shoal deposits.

The lowermost first real indication of fluvial braidplain deposition in the West Rand Group emerges in the Klerksdorp area in the second Brixton quartzite (Witkop Member). Depositional slope was to the south-southeast. This fluvial input becomes much more pronounced during deposition of the upper Brixton (Rangeview) quartzite with fluvial lobes present near Klerksdorp in the northwest and near Heidelberg in the northeast (Figures 2 and S9). The next phase of deposition is represented by major fluvial input from the northwest during the Bonanza Formation. Likewise, depositional slope was to the south-southeast. Distal marine reworking took place by apparent east-west longshore currents and southeasterly directed offshore currents (Figure S11). During and after deposition of the Bonanza Formation, major upwarping and erosion took place in the east and downwarping in the west. Thus, during the early phases of deposition of Sequence I, a northeastern source area was dominant up to deposition of the Versterkop Member. Thereafter, a northwestern source area becomes more prominent in upper Brixton and Bonanza times. The apparent arcuate pattern of the shoreline during the early phases of deposition of Sequence I resulted in radially-centred offshore marine currents and arcuately orientated longshore marine current patterns.

During deposition of Sequence II, input of coarser sediment from a northern arcuate source area persists (Figures S14 to S26). This is especially the case during deposition of the Promise diamictite and the quartzites of the Breaunanda, Hamberg and Townhouse members. However, in the cases of the dominantly fluvial Tusschenin and Elandslaagte quartzites, transport of coarse sediment was mainly from the northwest and depositional slope directed towards the southeast and south-southeast. These two units also appear to be fluvial-dominated in the northwest and marine-dominated in the southeast (Figures S21 and S24). Polymodal palaeocurrent distributions in the latter setting can be interpreted as a confluence of onshore-offshore and longshore marine currents. At the end of Sequence II, preceding the Buffelsdoorn Reef unconformity, an eastern source area was present during deposition of the Lagerspoort diamictite (Blue Grit).

During deposition of Sequence III, the northwestern source area appears to have been dominant (Figures S27 to S32). The unconformity at the base of the Koedoeslaagte Formation is apparently smooth for the most part. However, in the Klerksdorp area, it is channelised, filled with fluvial deposits, with a palaeoslope directed towards the south-southeast. During deposition of Sequence III, there is also an indication of sediment input from western and southwestern source areas, i.e., thickening and/or slight coarsening of sands of the Babrosco and Roodepoort quartzites in that area. This coarsening is also apparent in the argillite units of the Rietkuil and Roodepoort Formations. Quartzites of Sequence III consistently tend to become more orthoquartzitic in an easterly direction. Palaeocurrents are mainly offshore or downslope in a southeasterly to south-southeasterly direction.

Stable conditions prevailed over widespread areas of the platform during deposition of the West Rand Group. Lateral facies changes are for the most part very gradual. In general, correlations are excellent, and transgressions and disconformity surfaces can be traced on a basin-wide scale, allowing for chronostratigraphic subdivision in a layer-cake fashion into three major tectono-sedimentary phases:

Sequence I was deposited during a period of stable tectonic conditions and relative highstand of sea level in the basin. This implies active creation of accommodation space with little sedimentary bypassing and accumulation of abundant shales. Sands are supermature with polymodal palaeocurrent distributions implying that sediment input was in equilibrium with subsidence so that effective marine reworking could take place during progradation and aggradation (Figure 16). Consequently, contacts between shales and orthoquartzites are commonly gradational. Accommodation space decreased at the end of Sequence I with deposition of fluvial deposits, particularly those of the Bonanza Formation at the top of the Hospital Hill Subgroup. Erosional cut-out of the Bonanza Formation between Randfontein and Krugersdorp is very rapid, suggesting that a pre-diamictite monoclinal fold structure, possibly related to the development of a forebulge, was responsible.

In contrast to Sequence I, Sequence II formed during a period of lowstand of sea level in the basin. This led to ineffective creation of accommodation space and very effective sediment bypassing of the basin (Figures 3B and 16). As a result, quartzites are dominantly fluvial and have sharp disconformable contacts with underlying shallow outer shelf argillites. Unimodal palaeocurrents become more pronounced, particularly along the northwestern perimeter of the basin (Watchorn 1980). It was also a period of deposition of diamictites in the sequence. Although speculative, it is possible that these diamictites were related to glacial activity; a factor which would conform to overall lowstand of sea level.

Sequence III was deposited during a period of highstand of sea level, i.e., high rate of creation of accommodation space coupled with an oversupply of sediment (Figure 16). This would have resulted in rapid aggradation of sediment, and formation of turbidite, i.e., quartz wacke deposits and rather poor sorting of shallow marine shelf sands. It would also have resulted in progradational sands, which in proximal areas, exhibit gradational contacts with underlying shales. In more distal areas, the sands may develop sharp disconformable contacts with shales. The latter type of contact would have occurred as soon as sediment supply exceeded accommodation space and sediments started overflowing or bypassing the depository.

Basin fill of the West Rand Group is distinctly asymmetrical, a fact especially well illustrated in west-east sections across the structural basin, which show thicker accumulations of strata in the west and thinner strata in the east (Figures 2 and S33). However, on the scale of these cross-sections, the amount of asymmetry in basin fill is very gentle. Maximum known values for the Hospital Hill Subgroup are 3.0 km thickness in the west relative to 1.2 km thickness of strata in the east, extended over a horizontal distance of 200 km. It is also important to note that the coarsest sedimentary material is typically associated with thicker basin fill in the west. This implies that the area of maximum subsidence was situated relatively close to the sediment source.

The nature of disconformities in the structural basin is also very interesting. Almost all tend to cut deeper down through the sequence towards the east (Figure 2), so that more of the sequence is preserved below disconformities in the west/northwest. The latter is also the area in which gradational coarsening upward lithofacies successions between shale and quartzite are preferentially developed with disconformities higher up near the top of successions. However, in easterly directions, these disconformities cut out transitional lower parts of successions so that quartzites rest with sharp disconformable contacts on outer shelf argillites. These features, combined with asymmetry of basin fill, strongly suggest crustal flexure as the main mechanism of basin formation. In this flexural basin, downwarping preferentially took place towards the northwest, i.e., Klerksdorp-Randfontein areas, and bulging towards the southeast, i.e., Parys-Heidelberg areas (Figure 2). During the final stages of deposition of Sequence II, immediately prior to deposition of the Koedoeslaagte Formation, the sequence became arched with the axis of the arch situated in the central part of the present-day structural basin (Figure 12). Arching resulted in erosion and development of the low-angle unconformity at the base of the Koedoeslaagte Formation (Buffelsdoorn Reef).

It is thus quite possible that the West Rand Group was deposited in some type of foreland basin, with an axial zone (thickest sediment accumulation) towards the west/northwest and a peripheral bulge (outer arch with disconformities) towards the east/southeast (Burke et al., 1986; Beukes and Nelson, 1995; Catuneanu, 2001; Nhleko, 2003). Flexures have very low amplitudes and wide wavelengths suggesting the presence of strong elastic plate strength, i.e., thick stable cratonic conditions. Low amplitudes of flexures resulted in the deposition of relatively thin axial zone sedimentary successions and abundant overflowing (sedimentary bypassing) of the basin; a common feature of West Rand Group strata. However, it must have been a special type of foreland basin, allowing for various tectonic components, such as the axial zone and peripheral bulge, not to migrate very actively through time. Rather, these elements appear to remain virtually stationary throughout the depositional history of the West Rand basin. However, the sequence coarsens and becomes slightly less mature with time, which may suggest a slow secular migration of source area closer to the axial zone of the basin.

The nature of this rather static shallow foreland basin, with abundant sediment bypassing, corresponds extremely well with the model presented by Hoffman and Grotzinger (1993) for foreland basins associated with windward orogenic fronts. The model, based on geological observations and geophysical modelling, predicts that, apart from possible variable tectonic processes, prevailing windward and leeward mountain fronts evolve different tectonic styles because of contrasting rates of precipitation and erosional denudation. Windward orogenic fronts have very effective and deep rates of weathering and erosion so that uplift barely exceeds erosional denudation preventing effective loading by mountain building, and propagation of fold-thrust belts onto foreland basins. This results in shallow foreland basins with abundant indications of overfilling, strong progradation of sedimentary successions, compositionally mature dominantly molassoid sediments, and narrow fold-thrust belts (Figure 17). Most of these features are present in the West Rand Group. Although, the position of the predicted fold-thrust belt zone is largely unknown, evidence for metamorphism in the source area has been inferred by the observation of detrital grains of strained quartz, tourmaline and epidote, as well as an abundance of lithoclasts and detrital flakes of chlorite and muscovite in Government and Jeppestown quartzites (Blane, 2013).

In three dimensions, the geodynamic configuration of the West Rand foreland basin may have been much more complex than depicted in Figure 17 since it is known that sediments also came into the basin from northern and northeastern directions during deposition of the Hospital Hill and Government subgroups (Figures 16 and S34; Watchorn, 1980). Thus, the possibility of an evolving arcuate fold-thrust belt must be considered, i.e., interference of belts originating from both northerly and westerly directions (Fripp and Gay, 1972; Allen and Homewood, 1986; Catuneanu, 2001; Schmitz et al., 2004; Zeh et al., 2013; Laurent et al., 2019). Such complex three-dimensional geometries for foreland basins are well known; for example, the Adriatic Sea is a foreland basin flanked on three sides by fold-thrust belts (Allen and Homewood, 1986).

Geological field mapping and detailed logging of drill core from localities throughout the basin were carried out to understand the genetic and stratigraphic development of the West Rand Group. Reference profiles were constructed at type localities to calibrate vertical stratigraphic boundaries and interpret genetic relationships. On this basis, lateral correlation of facies successions and sequences over long distances allowed for a rigorous three-dimensional analysis of the basin, with emphasis on basin architecture and evolution of depofacies.

Depofacies in the Hospital Hill Subgroup shifted from dominantly marine shelf depofacies in the lower parts of the succession to fluvial braidplain depofacies in the upper parts. In the overlying Government Subgroup, fluvial braidplain quartzites are interbedded with, or transition into, shallow marine shelf depofacies. While the latter depofacies prevails for most of the Jeppestown Subgroup, fluvial braidplain deposits are developed in the upper part of the sequence. Source areas, or areas of coarse sediment input, were initially located to the northeast during deposition of the lower Hospital Hill Subgroup. Thereafter, input of coarse material from a northern arcuate source area, particularly from the north and northwest, continued up until the low-angle unconformity at the base of the Koedoeslaagte Formation. In the Jeppestown Subgroup, input of coarse clastics from the northwest was accompanied by thickening and/or coarsening of sands to the west and southwest, signalling a westward migration of source area (Figure S34).

Lateral correlation analysis shows that a number of formations have been preserved or eroded beneath low-angle basin-wide unconformities (e.g., the Bonanza and Eleazar formations below the Promise diamictite and the Afrikander Formation below the Buffelsdoorn Reef). These erosion surfaces were apparently developed in conjunction with structural flexuring and are typically linked to the propagation of forebulges through foreland basins (DeCelles, 2011). It is thus possible that strata of the lower Hospital Hill Subgroup represent the initial phase of an underfilled foreland basin (Catuneanu, 2001), allowing for the thick accumulation of shales in the Klerksdorp and Randfontein areas, which corresponds to the most actively subsiding parts of the basin (Figure 3A). The overlying sediments record a more complex narrative reflecting the interplay between underfilled and overfilled conditions. Overfilled conditions are best exemplified by fluvial strata of the Government Subgroup, which in proximal locations, are characterised by the presence of numerous disconformities, immature sediment characteristics and indications of sedimentary bypassing (Figure 3B). Periods of fluvial expansion in the West Rand Group were repeatedly curtailed by the development of low-angle unconformities, suggesting renewed episodes of thrusting and orogenic loading in the hinterland (Catuneanu, 2019). The reestablishment of a compressional regime during deposition of the Jeppestown Subgroup led to the resumption of marine sedimentation (i.e., foredeep subsidence) and also gave rise to stratigraphic condensation in the central and eastern parts of the basin (i.e., forebulge uplift; Figure 3C). The conclusion that the West Rand Group was deposited in an evolving foreland basin system is consistent with collisional orogenesis on the northern and western margins of the Kaapvaal Craton (De Wit et al., 1992; Tinker et al., 2002; Poujol et al., 2003; Eglington and Armstrong, 2004; Schmitz et al., 2004; McCarthy, 2006; Zeh et al., 2013; Smart et al., 2016; Laurent et al., 2019; Gibson et al., 2022).

Nicolas Beukes, together with colleagues at the Rand Afrikaans University (University of Johannesburg), produced a series of unpublished progress reports on the stratigraphy of the West Rand Group for the Anglo American Corporation (AAPS) between 1991 and 1992 (see Supplementary Information A). Abhen Pather and Mike Buxton assisted during field mapping. Wilma Clark carried out petrography and electron microprobe analyses. In the acknowledgement sections of the progress reports, Helge Dirr was routinely thanked for preparing line drawings on a Macintosh SE30 and Elsa Maritz was thanked for typing. On the basis of this work, Nicolas Beukes established a stratigraphic subdivision and correlation of the West Rand Group over the entire basin that was largely based on unconformities. In 1995, he distilled his work into a final report entitled ‘Stratigraphy and Basin Analyses of the West Rand Group with Special Reference to Prospective Areas for Placer Gold Deposits’.

The importance of this contribution was underscored a few years later, when the South African Committee for Stratigraphy (SACS) created a Task Group in the late 1990s, whose mandate was to standardise the stratigraphy and nomenclature of the Witwatersrand Supergroup. Whilst the lithostratigraphy of the Central Rand Group was well known, knowledge of the West Rand Group was comparatively sparse. Fortunately, Nicolas Beukes was granted permission by the Anglo American Corporation to make his findings available to the Task Group and these findings were incorporated into the final document. The Task Group achieved consensus on the correlation and nomenclature of the Central Rand Group and accepted Nicolas Beukes’ correlation and nomenclature for the West Rand Group. The revised nomenclature was completed in 2000 and published in 2006 in Issue 42 of the Council for Geoscience SACS Lithostratigraphic Series (SACS, 2006). In this regard, Anglo American Corporation (AngloGold Ashanti Ltd.) are sincerely thanked for granting permission to publish and for providing access to drill core. In addition, the Council for Geoscience and Gold Fields Ltd. (Sibanye-Stillwater Ltd.) are also thanked for arranging access to drill core. The National Research Foundation and Rand Afrikaans University (University of Johannesburg) are acknowledged for funding and organisational support.

In addition to his own research, Nicolas Beukes supervised a number of post-graduate students working on the Witwatersrand and Pongola supergroups (Jan P. Nelson, Noah Nhleko, Albertus J.B. Smith, Craig H. Blane, Justin M. Cochrane and Bradley M. Guy). On behalf of the group, as well as many other students in the PPM Research Group, we are indebted to him for his guidance and wonderfully detailed (yet almost indecipherable) hand-written thesis/manuscript corrections. We remember his humility, humour, friendship and passion for South African geology.

In this contribution, Herman Dorland, Matthew Hales, Alan. J. Kaufman, Robyn Ormond, Albertus J. B. Smith, Marylou Vines and Lindsay van der Wiele are kindly thanked for assisting with the collection/scanning of outcrop photographs and figures. The reviewers, Terence McCarthy and George Henry, as well as the editor, Jens Gutzmer, are thanked and appreciated for their insightful recommendations and attention to detail.

Nicolas J. Beukes: Conceptualisation, field mapping, core logging, investigation, writing (1995 final report). Bradley M. Guy: Investigation, writing (redrafting 1995 final report), writing – review and editing. Sam T. Thiele: 3D Models, writing – review and editing.

Editorial handling: J. Gutzmer.