A research project submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in partial fulfilment of the requirements for the degree of Master of Science. / Rb-Sr emplacement ages of nineteen kimberlites from the Prieska Province vary from 74 to 174 Ma, Their isotopic, whole-rock geochemical signatures and perovskite REE distributions were also determined. Non-micaceous and micaceous kimberlites from the area show similar petrographic, geochemical and isotopic compositions relative to cratonic kimberlites, indicating similar sub-continental mantle source compositions in the two tectonic environments. Transitional varieties of kimberlite, which can be defined petrographically, geochemically and isotopically are also recognised from the area, but are not prevalent in the cratonic environment. The Prieska Province kimberlites are possibly derived by partial melting processes within a subcontinental reservoir characterised by a spectrum of compositions from time-averaged depleted (HllMU) to enriched (Group II). The occurrence of these components in both on- and off-craton settings indicates that the kimberlite source area is not strictly linked to the suberatonic lithosphere. Mantle' plume sources are not favoured because of the variable source compositions and distribution of emplacement ages within the Province. Transitional kimberlites were derived from a source component with mixed character, not yet noted from within the Kaapvaal Craton. This is the first documented example of isotopically transitional kimberlites from southern Africa. / AC2017

The Cretaceous (133Ma) Kroonstad Group II Kimberlite Cluster is located approximately 200km south west of Johannesburg on the Kaapvaal Craton. The cluster is made up of six kimberlite pipes and numerous other intrusive dike/sill bodies. Three of the pipes are analysed in this study, which includes the: Voorspoed, Lace (Crown) and Besterskraal North pipes. These pipes were emplaced at surface into the Karoo Supergroup, which is comprised of older sedimentary rocks (300-185Ma) overlain by flood basalts (185Ma). At depth the pipes have intruded the Transvaal (2100-2600Ma) and Ventersdorp (2700Ma) Supergroups, which are comprised dominantly of carbonates and various volcanic units respectively. The pipes have typical morphology of South African pipes with circular to sub-circular plan views and steep 82o pipe margins. The Voorspoed pipe is 12ha in size and is characterised by the presence of a large block of Karoo basalt approximately 6ha in size at the current land surface. This large basalt block extends to a maximum of 300m below the current land surface. The main Lace pipe is 2ha is size with a smaller (<0.5ha) satellite pipe approximately 50m to the west. No information is available on the morphology of the Besterskraal North pipe as it is sub-economic and no mining has occurred. Samples from the Besterskraal North pipe were collected from the De Beers archives. The Kroonstad Cluster has been subjected to approximately 1750m of erosion post-emplacement, which has been calculated by the analysis of the crustal xenoliths with the pipe infill. The hypabyssal kimberlite from the three pipes shows a gradational evolution in magma compositions, indicated by the mineralogy and geochemistry. The Lace pipe is the least evolved and has characteristics more similar to Group I kimberlites. The Voorspoed and Besterskraal North kimberlite are intermediately and highly evolved respectively. The gradational evolution is marked by an increase in SiO2 and Na2O contents. Furthermore the occurrence of abundant primary diopside, aegirine, sanidine, K-richterite and leucite indicates evolution of the magma. The root zones of the pipes are characterised by globular segregationary transitional kimberlite, which is interpreted to be hypabyssal and not the result of pyroclastic welding/agglutination. The hypabyssal transitional kimberlite (HKt) is characterised by incipient globular segregationary textures only and the typical tuffisitic transitional kimberlite (TKt) end member (Hetman et al. 2004) is not observed. The HKt contact with the overlying volcaniclastic kimberlite (VK) infill is sharp and not gradational. The presence of HKt in the satellite blind pipe at Lace further indicates that the distinct kimberlite rock type must be forming sub-volcanically. The HKt is distinctly different at the Voorspoed and Lace pipes, which is likely a result of differing compositions of the late stage magmatic liquid. Microlitic clinopyroxene is only observed at the Lace HKt and is interpreted to form as a result of both crustal xenolith contamination and CO2 degassing. Furthermore the HKt is intimately associated with contact breccias in the sidewall. The root zones of the Kroonstad pipes are interpreted to form through the development of a sub-volcanic embryonic pipe. The volcaniclastic kimberlite (VK) infill of the Kroonstad pipes is not typical of South African tuffisitic Class 1 kimberlite pipes. The VK at Voorspoed is characterised by numerous horizontally layered massive volcaniclastic kimberlite (MVK) units, which are interpreted to have formed in a deep open vent through primary pyroclastic deposition. MVK is the dominant rock type infilling the Voorspoed pipe, however numerous other minor units occur. Normally graded units are interpreted to form through gravitational collapse of the tuff ring. MVK units rich in Karoo basalt and/or Karoo sandstone are interpreted to form through gravitational sidewall failure deep within an open vent. Magmaclasts are interpreted to form in the HKt during the development of an embryonic pipe and therefore the term autolith or nucleated autolith may be applied. Debate on the validity of the term nucleated autolith is beyond this study and therefore the term nucleated magmaclast is used to refer to spherical magmaclasts in the VK. The emplacement of the Kroonstad pipes is particularly complex and is not similar to typical Class 1 tuffisitic kimberlites. However the initial stage of pipe emplacement is similar to typical South African kimberlites and is interpreted to be through the development of an embryonic pipe as described by Clement (1982). The vent clearing eruption is interpreted to be from the bottom up through the exsolution of juvenile volatiles and the pipe shape is controlled by the depth of the eruption (+/-2km) (Skinner, 2008). The initial embryonic pipe development and explosive eruption is similar to other South African kimberlites, however the vent is cleared and left open, which is typical of Class 2 Prairies type and Class 3 Lac de Gras type pipes. The latter vent infilling processes are similar to Class 3 kimberlites from Lac de Gras and are dominated at the current level by primary pyroclastic deposition.

The mineralogy and petrology of two kimberlites, a peripheral monticellite kimberlite, and its core of phlogopite kimberlite, from the West Auxiliary Pipe at Dutoitspan Mine are described. The mineralogy of the two kimberlites differs mainly in the presence of phlogopite macrocrysts, greater abundance of angular crustal inclusions, more heavy minerals and higher diamond grade in the phlogopite type. Microprobe analyses of olivine, phlogopite, monticellite, oxide minerals and garnet are presented. Silicate compositions are comparable in both kimberlites and zoning of olivine grains is typically towards a rim of Fo₈₉₋₉₀ʻ irrespective of whether cores are more Fe-rich or more Mg-rich. This is caused by re-equilibration after fluidised emplacement in the earth's crust of macrocryst-bearing kimberlite magma. Olivine aggregates were derived from sheared mantle lherzolite and single-crystal macrocrysts were formed at higher mantle levels from a kimberlitic crystal-mush magma. This was emplaced in the crust by rapid gas streaming. The post-fluidisation phenocrysts of olivine and phlogopite which formed then are in general more Fe-rich than macrocrysts. Re-equilibration of ilmenite results in the formation of complex perovskite and titanomagnetite mantles. Phlogopite macrocrysts are preserved in the monticellite contact rock where rapid quenching prevented their resorption and allowed separation of an immiscible carbonate melt, giving the abundant groundmass calcite. Atoll-textured spinels are found in the contact rock. Major and minor trace-element analyses of whole rock samples are presented and discussed, bringing into account the problem of contamination by crustal inclusions. Whole rock chemistry supports derivation of the kimberlites as partial melts of mantle material in the presence of a lithophile-element-enriched fluid. The monticellite contact rock is highly enriched in REE, Nb, and Sr due to rapid freezing of this perovskite-enriched phase. The monticellite type is more enriched in lithophile elements than the phlogopite type, which supports derivation of the monticellite type by a small degree of partial melting, further melting reducing the relative concentrations of lithophile elements to give the phlogopite kimberlite chemistry.

he estimation of post-emplacement kimberlite erosion in South Africa through the study of upper-crustal xenoliths is relatively unexplored; however the presence of these xenoliths has been recognized for well over 100 years. Post-emplacement erosion levels of a small number of South African kimberlite pipes have been inferred through the study of the degree of country-rock diagenesis, the depth of sill formation, the depth of the initiation of the diatreme and fission track studies. Through these studies, several estimates were proposed for the Group I Kimberley kimberlites. Although the 1400 m estimate of erosion remains widely accepted today, this estimate relies on the presence of Karoo-like basalt xenoliths in the Group I Kimberley kimberlites, as their presence proves that basalt existed in the Kimberley area when the kimberlites were emplaced. Basaltic xenoliths were described during the early stages of mining in Kimberley, though only one of these descriptions suggests that the ‘basaltic’ boulders correlate with the Karoo basalts. Because of the discrepancy between these early documentations of upper-crustal xenoliths and because the occurrence of Karoo-like basalt xenoliths in the Group I Kimberley kimberlites is under question, a re-investigation of the erosion levels and the upper crustal xenolith suites in South African, Cretaceous-aged kimberlites, including Melton Wold, Voorspoed, Roberts Victor, West End, Record Stone Quarry, Finsch, Markt, Frank Smith, Pampoenpoort, Uintjiesberg, Koffiefontein / Ebenheuyser, Monastery, Kimberley (Big Hole), Kamfersdam , Jagersfontein, Kaal Vallei, De Beers, Bultfontein, Lushof, Britstown Cluster, Hebron and Lovedale, was conducted. This study presents the analytical results for upper-crustal sandstone and basalt xenoliths collected from dumps, excavation pits and borehole core at the above-mentioned kimberlites, and demonstrates that they correlate with stratigraphic units of the Karoo Supergroup on the basis of mineral and geochemical compositions. These upper-crustal xenoliths are incorporated into kimberlites and down-rafted to levels below their stratigraphic position during kimberlite emplacement, consequently recording the broad stratigraphy into which each kimberlite is emplaced. Therefore, the Cretaceous lateral extent of the Karoo Supergroup is inferred and post-emplacement erosion estimated by reconstructing the stratigraphy based on upper-crustal xenolith suites for each kimberlite and calculating the total thickness of the now-eroded units. The distribution of sandstone xenoliths indicates that during the Cretaceous the lateral extent of the Dwyka, Ecca and Beaufort Groups encompassed all of the examined kimberlites, while the ‘Stormberg’ Group was constrained to an area outlined by the Voorspoed and Monastery kimberlites. Similarly, basalt xenoliths occur in all of the Group II and transitional (143 – 100 Ma) kimberlites but only in the Group I (90 – 74 Ma) kimberlites that lie within close proximity to the western outcrop margin of the outcrop area of the Drakensberg Group basalts (Lesotho Remnant), namely Monastery, Jagersfontein and Kaal Vallei. This trend implies an eastward-retreat of the inland erosion front of the Karoo basalts between 140 and 90 Ma and subsequent erosion of the underlying sedimentary units. It also suggests that a thicker succession of Karoo strata was present at the time of Group II and transitional kimberlite emplacement and that there has been more post-emplacement erosion in these kimberlites than the younger Group I kimberlites, except for Monastery, Jagersfontein and Kaal Vallei. Estimates are unique to each kimberlite as they are dependent on both stratigraphic location, elevation and present country rock, and range from approximately 1000 – 2500 m for the older kimberlites and less than 700 m to 1400 m for the younger kimberlites. Furthermore, the upper-crustal xenoliths found at the Group I Kimberley kimberlites and the coinciding trend of basalt erosion demonstrate that Karoo basalts were eroded from the Kimberley area by the time the Group I Kimberley kimberlites erupted (~85 Ma). Therefore, basalts are omitted from the Group I Kimberley kimberlites post-emplacement erosion estimate, and the upper Beaufort Group is considered the upper limit of the stratigraphy that was present at the time of the eruption of the Group I Kimberley pipes. Therefore, the erosion estimates decrease from a previous estimate of 1400 m down to 400 to 1100 m, where 850 m is considered a dependable intermediate estimate.

M.Sc. (Geology) / This treatise presents the first comprehensive investigation in petrography and geochemistry of a mantle-derived eclogite xenolith suite from the Balmoral kimberlite. Eclogites form a minor component of the Earth’s mantle however they play a vital role in our understanding of geodynamic processes, i.e. the subduction of oceanic crust (Jacob, 2004) and the crystallization of diamond within the sub-cratonic lithosphere. A large portion of eclogites from the Balmoral kimberlite pipe is comprised of bimineralic (garnet and clinopyroxene) rocks with the rest being corundum-bearing. Mica with average modal abundances ≤10 vol% is observed as an accessory phase in bimineralic xenoliths. Modal abundances of corundum in corundum-bearing samples range between 1 and 6 vol%. Textures are ambiguous in Balmoral eclogites and thus chemical criteria of McCandless and Gurney (1989) places all Balmoral eclogites into Group II. As typically observed in garnets from eclogites (Hills and Haggerty, 1989; Jacob, 2004), garnets from Balmoral eclogites are chromium- and manganese-poor. They have a general trend from pyrope-rich towards grossular-rich compositions, with some almandine. Garnets from the bimineralic eclogites have disparate suites of low- and high-MgO samples. High-MgO bimineralic garnets are pyropic in composition with averages at Pyr63Gros22Alm15, whereas garnets from the low-MgO suite are widespread from relatively less pyropic towards grossular-rich compositions with average compositions of Pyr49Gros40Alm11. Garnets from the corundum-bearing eclogites are homogeneous and characterised by the highest grossular component (averaging at Gros47Pyr28Alm25). The clinopyroxenes for Balmoral eclogites are omphacitic in composition. Jadeite content is highly variable (ranging between 8 and 58 wt%) in these clinopyroxenes. The clinopyroxenes in bimineralic eclogites are characterised by a wide variation from diopside-rich towards jadeite-rich compositions. Clinopyroxenes in corundum-bearing eclogites have the highest jadeite levels.