A thesis submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Doctor of Philosophy. September 2014, Johannesburg. / Aims: The adsorptive removal of Cu, Mn, Hg and U by silica and maghemite nanoparticles (NPs) under acid mine drainage (AMD) conditions was investigated with the aim of assessing the applicability of NPs for remediation of AMD-contaminated water. The effect of NPs on metal uptake by algae, an increasingly popular remediation alternative, was also investigated. Methods: NP and algal metal removal were quantified by batch experiments using commercially prepared, bare and amine-functionalised silica-carbon hybrid NPs characterised for size, surface area, porosity, crystallinity, elemental composition and hydrodynamic size. Metal uptake by algae was quantified in the presence and absence of NPs. Results: Silica and maghemite NPs can be used for the adsorptive removal of Cu, Mn, Hg and U from AMD-contaminated surface and ground water. NP metal uptake was rapid and equilibria were attained within 5 minutes with silica and maghemite NPs, and within 45 minutes with amine-functionalised hybrid NPs. Adsorption efficiencies for Cu, Mn, Hg and U at pH 3 were 52, 56, 56 and 49%, respectively with silica and 56, 52, 75 and 50%, respectively, with maghemite NPs. Metal removal was enhanced by >10% in solutions containing ferric, manganese or sulphate ions, although Cu removal was inhibited in solutions with a >1 Mn:Cu ratio. Despite the presence of high affinity amine groups in hybrid NPs, Cu removal was only 52% due to the low surface area of the adsorbent. The comparative study with Hg, however, showed that surface area was not the only determinant of adsorption efficiency: maghemite NPs with a specific surface area ~15 times less than silica adsorbed 21% more Hg. Metal removal by Scenedesmus sp. was enhanced by 12-27% in solutions containing NPs due to the greater sorption surface areas. NPs also modified metal partitioning in algal cells: intracellular concentrations were lower and extracellular concentrations higher in solutions containing NPs relative to controls (no NPs). Conclusion: Silica and maghemite NPs can be applied for the adsorptive removal of Cu, Mn, Hg and U from AMD-contaminated water and to improve the efficiency of phycoremediation by Scenedesmus sp. These findings also point to the possibility of retardation of metals by NPs during their transportation from tailings and contaminated sites. Their partitioning to NPs and the strength of the interactions thereof can determine the prevalence of the metals in solution or in the solid phase.

A thesis submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Doctor of Philosophy, / The Ventersdorp Contact Reef(VCR) is a major gold-bearing reef in the Witwatersrand Basin. It occurs between the overlying Klipriviersberg Group lavas and the underlying Central Rand Group sediments, and was strongly altered by hydrothermal fluids circulating in the Witwatersrand Basin. A detailed study of the mineralogy, geochemistry of rocks and minerals, physicochemical conditions, stable isotopes and ages of hydrothermal alteration zones associated with the VCR were carried out at Western Deep Levels South Mine, South Africa. ( Abbreviation abstract) / Andrew Chakane 2019

A dissertation submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Master of Science. Johannesburg, 2016 / The Mesoproterozoic Pilanesberg Complex, South Africa, is the world’s largest alkaline intrusive complex. Mapped geological field relationships suggest the Complex has circular inward dipping layers. However, it is unclear how the dipping layers extend at depth. As a result, the 3D geometry of the Pilanesberg Complex is unknown. Modelling of the Pilanesberg Complex uses 2D forward models as well as 3D forward and inversion, gravity and magnetic data models, to set limits on the 3D geometry of the Pilanesberg Complex. The 2D Bouguer gravity models and geology maps indicate that some of the Bushveld Complex Main Zone shifted to the west of the Pilanesberg Complex during emplacement. This, and a highly faulted country rock, accounts for a portion of how the host rock was able to accommodate the Pilanesberg Complex intrusion. The geometry of the Complex is explored with test gravity models where the model of outward dipping and vertically dipping cylinders are unable to match the Bouguer gravity signal over the Complex, but the inward dipping model matched the data to provide a possible solution for the geometry of the Complex. The Pilanesberg Complex geometry is modelled with 3D magnetic inversion, 3D forward gravity models and 2.5D gravity test profiles that were all constrained by the surface geology. The different models correlate so that best data fit for the Complex is represented by an overall inward dipping structure. Surface geological measurements indicate that the northern edge of the Complex dip out to the north. The 3D forward modelling was able to produce a positive solution that matched the gravity data with a northward dipping northern edge. The dipping northern edge is also observed on the University of British Columbia, UBC, 3D gravity inversion and the Euler deconvolution gravity profile solutions. The depth of the Pilanesberg Complex from 3D forward gravity modelling is estimated to be between 5 and 6 km. The Complex is suggested to have undergone block movement where the northern block and southern block are separated by the 30 km long Vlakfontein fault, which bisects the Complex from the north-east to the south-west. The image processing contact depth, Euler deconvolution solutions and the 3D Voxi inversion model suggest that the fresh bedrock is closer to surface in the north, while the southern block appears to be approximately 1km deeper than the northern block. The northern dip and block movement are explained by complicated structural events that include trap door graben settling which hinged on the northern edge as well as faulting and external block movement during a regional lateral extensional event. The Pilanesberg Complex intruded during a larger system of alkaline intrusions, known as the Pilanesberg Alkaline Province. The intrusions are associated with the Province due to their ages and chemical affinity. This Province includes two dyke swarms that radiate to the north-west and south of the Pilanesberg Complex, as well as smaller circular clinopyroxenite intrusions throughout the Bushveld Complex. The Pilanesberg dyke system and the circular clinopyroxenite intrusions are reversely magnetised with IGRF corrected values ranging between -150 to -320 nT compared to the normally magnetised 166 to 330 nT values of the Pilanesberg Complex. This suggests that a magnetic reversal occurred between the emplacement of the Pilanesberg Complex and the dyke System. The age data of the Complex and dyke Swarm suggest a magnetic reversal could have occurred between the emplacement of the Pilanesberg Complex and the Pilanesberg dyke System. The Complex is dated at 1602 ± 38 Ma and 1583 ± 10 Ma, from two white foyaite samples from the southern edge (using 40Ar/39Ar amphibole spectrum analysis). These ages are vastly different from previously reported ages, which ranged between 1200 Ma and 1450 Ma (Harmer R., 1992; Hansen et al., 2006). The error analysis has improved considerably from the published dates making the proposed dates plausible for the intrusion of the Pilanesberg Complex as the first and main intrusion of the Pilanesberg Alkaline Province. The Pilanesberg dyke System intruded much later between 1219 ± 6 Ma to 1268 ± 10 Ma for the red syenite dyke samples (using 40Ar/39Ar on feldspars spectrum analysis) and 1139 ± 18 Ma obtained for the grey syenite dyke (using 40Ar/39Ar on amphiboles inverse isochronal analysis). The dyke Swarm dates are significantly younger than the previously published ages for the dykes, which were between 1290 Ma and 1330 Ma (Van Niekerk, 1962; Emerman, 1991). / LG2017

The epeiric Transvaal Sea covered the Kaapvaal Craton of southern Africa during the Early Proterozoic and its remnant strata represent one of the oldest known carbonate depositories. A genetic stratigraphic approach has been used in this research on the evolution and syndepositional tectonics of the Transvaal Sea; research also emphasized the development of basement precursors, which influenced the Transvaal Sea. Eight subfacies were initially recognized and their interrelationships through Transvaal Sea time and space were used to identify ten depositional systems. Paleogeographic reconstructions indicate that the depositional systems developed on morphological variations of a distally-steepened carbonate rarp and that the depositional character of each was simply a function of water Backstripping of the depositional systems indicates that the Transvaal Sea was compartmentalized; three compartments are preserved on the Kaapvaal Craton. Backstripping also indicates that the depositional center of the Transvaal Sea lay over the western margin of an underlying rift. Rifting had developed a major, north-south-trending structure, and its geographical interrelationships with the east-west-trending Selati Trough created the compartment architecture of the basement. Interpretation of syndepositional tectonics suggests that six stages of subsidence influenced the Transvaal Sea. Early subsidence consisted of mechanical (rift) subsidence followed by

No description available.

Thesis submitted in fulfillment of the requirements for the Degree of Doctor of Philosophy University of the Witwatersrand South Africa 2015 / A multidisciplinary sedimentological, stratigraphic, mineralogical and geochronological analysis of a small, fossil-bearing, Holocene hillslope deposit, flanking a mesa, has enabled a reconstruction of the palaeo-environmental history of the region. The hillslope deposit, located on the farm Heelbo in the eastern Free State Povince, South Africa, overlies Jurassic mudrock and sandstone of the Elliot Formation, Karoo Supergroup. The deposit is located on a steep (~10°) slope and covers an area of ~7 km2 in two separate sections. It extends ~475 m downslope and reaches a maximum thickness of at least 6-8 m towards the base. Mineralogy indicates the deposit is sourced from the mesa but its fine grain size and location on a steep slope position Heelbo outside the scope of traditionally described alluvial fans or colluvial deposits. The hillslope deposit is described as an alluvial slope based on the morphology of the deposit and the grain size distribution against the slope gradient. The deposit comprises fine-sand to silt- grain size, but is found on a steep (~10°) slope. The source of the sediment is shown to be the sediment of the mesa, rather than input from an aeolian source. The grains are described as sub-angular and poorly sorted which also suggests a local provenance for the sediment. Microscopic and XRF analysis confirm the derivation of the sediment from the Elliot Formation, with the lithic fragments derived from subarkoses to arkoses of the underlying bedrock of the Upper Elliot Formation (UEF) in the Karoo Supergroup. Secondary calcite was visible in only two of the thin sections, thus it is likely due to a diagenetic overprint that is constrained by depth from the surface or time and not to specific layers. The deposit is cut by several mature and continuous gully networks with V-shaped profiles in the proximal slope, and combined V- and U-shaped profiles in the medial and distal sections. Gully formation is linked to both the sodium adsorption ratio (SARs), and high soil clay content, which facilitates swelling and shrinking. The Heelbo deposit comprises two palaeosols (BT1 and BT2) and four sediment (B1, B2, RB and TS) horizons. Through luminescence dating, the ages were found to be approximately 6390 ± 740 years BP for the oldest Brown (B1) horizon and 250 ± 170 years BP for the Red Brown (RB) horizon. The radiocarbon ages of the sediment were inconclusive, but the 14C ages of the fossil bones were in agreement with the luminescence ages. The multiple palaeosol horizons identified suggest two cycles of deposition, pedogenesis and erosion of the alluvial succession. The palaeosols and the presence of calcareous nodules and rhizocretions, and smectite and mixed-layer clay minerals, together with the total absence of illite and kaolinite, suggest protracted, dry periods with intermittent short periods of high rainfall. This is a similar climate regime to what the region experiences currently. Main findings: The Heelbo alluvial slope comprises locally derived sediment, rather than an aeolian source. Heelbo suggests that the hillslope deposits classification system needs to be re-evaluated and opens opportunities for wider study of Pleistocene-Holocene hillslope deposits in central / northern South Africa. This study also contributes to climate change debates in the Holocene.

A dissertation submitted to the Faculty of Science, University of the Witwatersrand in fulfilment of the requirements for the degree of Master of Science, School of Geosciences. Johannesburg, 2016. / The measurements of physical rock properties, seismic velocities in particular, associated with ore deposits and their host rocks are crucial in interpreting seismic data collected at the surface for mineral exploration purposes. The understanding of the seismic velocities and densities of rock units can help to improve the understanding of seismic reflections and thus lead to accurate interpretations of the subsurface geology and structures. This study aims to determine the basic acoustic properties and to better understand the nature of the seismic reflectivity of the world’s deepest gold-bearing reef, the Carbon Leader Reef (CLR). This was done by measuring the physical properties (ultrasonic velocities and bulk densities) as well as conducting mineralogical analyses on drill-core samples. Ultrasonic measurements of P- and S-wave velocities were determined at ambient and elevated stresses, up to 65 MPa. The results show that the quartzite samples overlying and underlying the CLR exhibit similar velocities (~ 5028 m/s-5480 m/s and ~ 4777 m/s-5211 m/s, respectively) and bulk densities (~ 2.68 g/cm3 and 2.66 g/cm3). This is due to similar mineralogy and chemical compositions observed within the units. However, the CLR has slightly higher velocity (~ 5070 m/s-5468 m/s) and bulk density (~ 2.78 g/cm3) than the surrounding quartzite units probably due to higher pyrite content in the reef, which increases the velocity. The hangingwall Green Bar shale exhibits higher velocity (5124 m/s-5914 m/s) and density values (~ 2.89 g/cm3-3.15 g/cm3) compared to all the quartzite units (including the CLR), as a result of its finer grain size and higher iron and magnesium content. In the data set it is found that seismic velocities are influence by silica, iron and pyrite content as well as the grain size of the samples, i.e., seismic velocities increase with (1) decreasing silica content, (2) increasing iron and pyrite content and (3) decreasing grain size. Reflection coefficients calculated using the seismic velocities and densities at the boundaries between the CLR and its hangingwall and footwall units range between ~0.02 and 0.05, which is below the suggested minimum of 0.06 required to produce a strong reflection between two lithological units. This suggests that reflection seismic methods might not be able to directly image the CLR as a prominent reflector, as observed from the seismic data. The influence of micro-cracks is observed in the unconfined uniaxial compressive stress tests where two regimes can be identified: (1) From 0 - 25 MPa the P-wave velocities increase with progressive loading, but at different rates in shale and quartzite rocks owing to the presence of micro-cracks and (2) above stresses of ~20 - 25 MPa, the velocity stress relationship becomes constant, possibly indicating total closure of micro-cracks. The second part of the study integrates 3D reflection seismic data, seismic attributes and information from borehole logs and underground mapping to better image and model important fault systems that might have a direct effect on mining in the West Wits Line goldfields. 3D seismic data have delineated first-, second- and third-order scale faults that crosscut key gold-bearing horizons by tens to hundreds of metres. Applying the modified seismic attribute has improved the imaging of the CLR by sharpening the seismic traces. Conventional interpretation of the seismic data shows that faults with throws greater than 25 m can be clearly seen. Faults with throws less than 25 m were identified through volumetric (edge enhancement and ant-tracking seismic attributes) and horizon-based (dip, dip-azimuth and edge detection seismic attributes) seismic attribute analysis. These attributes provided more accurate mapping of the depths, dip and strikes of the key seismic horizon (Roodepoort shale), yielding a better understanding of the relationship between fault activity, methane migration and relative chronology of tectonic events in the goldfield. The strato-structural model derived for the West Wits Line gold mines can be used to guide future mine planning and designs to (1) reduce the risks posed by mining activities and (2) improve the resource evaluation of the goldbearing reefs in the West Wits Line goldfields. / LG2017

Geochemical, mineralogical and tenor variation studies were carried out on the cores from UMT040, UMT064 and UMT063, located in the Flatreef on Turfspruit, in the southern sector of the Platreef. The investigation comprised three objectives (1) to identify controls on mineralisation in the upper section of the Platreef, (2) to construct a 3D PGE tenor model, and (3) to test whether the new geological interpretation on Turfspruit correlates with the eastern and western limbs of the Bushveld Complex. The above-mentioned holes were used for geochemical analyses of major and trace elements; and UMT064 was also used for orthopyroxene microprobe analyses. For construction of the tenor model, a total of 276 drill holes were used. The distance between drill holes for geochemical studies is 145 m and 175 m respectively; and for the tenor model is approximately 100 m. The outcomes of the study have demonstrated that the upper section of the Platreef comprise at least four lithological units i.e. the topmost portion of T1, lowermost portion of T1, T2 Upper and T2 Lower that can be interpreted to have been formed by four separate pulses of magma. Considering the possibility of the T2 Lower being metamorphosed calc-silicates, this reduces the number of magma pulses to three. This is supported by the mineralogy, geochemical content and tenor variations. The T1 is made up of an orthopyroxenite (feldspathic in places), the T2 Upper comprised a pegmatoidal orthopyroxenite (also feldspathic in places) and the T2 Lower is made up of a harzburgite. High grade PGE, Ni and Cu mineralisation was found to occur within two zones in the upper section of the Platreef i.e. the T1m and T2 (Upper and Lower). Mineralisation occurs in the form of base metal sulphides such as pyrrhotite, pentlandite and chalcopyrite; and PGEs are dominated by amphoterics (PGE-arsenides, tellurides, antimonides and bismuthinides). These amphoterics are associated with base metal sulphides. The sulphides in both the T1m and T2 are interstitial. PGE minerals such as michenerite, sperrylite, hollingworthite and cooperite are present in the study area. In the T2, an increase in sulphide content and PGE grades are generally associated with the presence of the “main” chromitite stringer/ seam at the contact between a feldspathic pyroxenite (T1) and the coarse-grained to pegmatoidal pyroxenite (T2 Upper). Several other chromitites might be present above and below this contact, and it has been noted that not all of those chromitites are associated with an increase in sulphide content and higher PGE grades. PGE grades were found to be higher in T2 Upper than in T2 Lower, whereas, the base metals (particularly Ni) are generally higher in T2 Lower. In the T1m, mineralisation occurs close to the contact between the feldspathic pyroxenite and orthopyroxenite, but only the orthopyroxenite hosts the sulphides. Chromitite stringers rarely occur in the T1m. Where present, they generally occur at the top of the mineralised zone. The sulphides in both the T1m and T2 are thought to have resulted from separate and different processes. Those in the T1m are thought to have resulted from a fractional segregation process, whereas the sulphides in the T2 were thought to have been emplaced in bulk or as a mush. The 3D PGE tenor model has demonstrated that the tenor in both the T1 and T2 are higher than previously thought for the northern limb of the Bushveld Complex. Tenors in the T1 reach up to 700 ppm, whereas tenors of about 1000 ppm were recorded in the T2. These tenors are comparable to the tenors measured from the sulphide melt inclusions elsewhere in the northern limb. Tenors are the highest in the northwestern part of Turfspruit and they gradually decrease towards the southeast. Both the T1 and T2 are dominated by a tenor of about 100 ppm. In the T1, a tenor of 250 ppm is dominant in the northwestern and southeastern parts of the study area, whereas in the T2 it is only dominant in the northwestern part. Vertical variation in tenors has shown that the tenors are the highest in the topmost portion of the T1 and at the top of T2 Upper (below the T1-T2 contact). The T1 has been found to be comparable with the Merensky Reef due to the following: (a) similarities with respect to major and trace element compositions of whole rock analyses; (b) Cr/MgO ratios which are greater than 80; and (c) similarities in the pyroxene content excluding the MgO/((MgO)+FeO) ratio and CaO contents; and in the PGE tenor of the sulphides. In addition, the T1 has many of the characteristics of the model proposed by Naldrett et al (2009) for the formation of the Merensky Reef in the western Bushveld. However, the differences between the T1 and the Merensky Reef were also noted i.e. the MgO/((MgO)+FeO) ratios in this study are lower (averaging 0.71) than the Mg# reported for the Merensky Reef (Mg#78-83), the upper Critical Zone (Mg#78-84) and elsewhere in the Platreef north of Turfspruit. The CaO levels (representing the wollastonite component of the orthopyroxene) at Turfspruit are higher (>2 wt %) compared to the Merensky Cyclic Unit and other parts of the northern limb where the CaO levels are below 2 wt %. This is an indication that the pyroxene composition at Turfspruit is more evolved compared to the Merensky Reef.

The Ecca Group of Karoo Supergroup is a sedimentary rock sequence that deposited between the Late Carboniferous (Dwyka Group) and the Late Permian-Middle Triassic (Beaufort Group). The Ecca Group investigated in this study is situated in the Eastern Cape Province of South Africa and it comprises mainly of shales, mudstones, siltstones and sandstones. The Ecca Group sequence contains considerable carbon content and suitable thickness to make it an ideal target for shale gas exploration. Previous studies put more emphasis on the geology and stratigraphy of the Ecca Group, this study revised the stratigraphy, and put new insight on the petrography, depositional processes, sedimentary facies, provenance, paleoweathering, tectonic setting, subsidence rates and history, electrical resistivity, source rock characteristics and diagenesis of the potentially feasible sandstone and mudrock reservoir rocks of the Ecca Group. Based on the lithological features, sedimentary structures and facies characteristics, the stratigraphy of the Prince Albert, Whitehill, Collingham and Fort Brown Formations of the Ecca Group is now subdivided into two informal members each, i.e. Lower Member and Upper Member. Furthermore, the Ripon Formation is now subdivided into three informal members. Each member has been asigned a lithological name. The grain size parameters show that most of the Ecca Group sandstones are very fine to fine grained, poorly to moderately well sorted, mostly near-symmetrical and mesokurtic in grain-size distribution. The linear discriminant function analysis is dominantly indicative of turbidity current deposits under deep marine environment for Prince Albert, Whitehill and Collingham Formations, shallow marine environment for Ripon Formation, while the Fort Brown Formation is lacustrine-deltaic deposits. Modal composition analysis and petrography studies revealed that the detrital components of the sandstones are dominated by monocrystalline quartz, feldspar and lithic fragments. The sandstones are compositionally and texturally immature and can be classified as feldspathic wacke and lithic wacke. The provenance analysis revealed plutonic and metamorphic terrains as the main source rocks with minor debris derived from recycled sedimentary rocks. The detrital modal compositions of these sandstones are related to back arc to island and continental margin of tectonic setting. Based on the detailed sedimentological analyses of outcrop and borehole data, fourteen lithofacies were identified and seven facies associations (FAs) were recognised. The facies associations are: FA 1: Shale and mudstones intercalated with siltstones, FA 2: Carbonaceous shale, mudstone with subordinate chert and sandstone, FA 3: Mudstones rhythmite with thin bedded mudstone and lenticular siltstone, FA 4: Greyish medium bedded sandstone intercalated with laminated mudstone, FA 5: Dark-grey medium to thick bedded mudstone and siltstone, FA 6: Thin to medium bedded sandstone alternated with thin bedded carbonaceous mudstone, and FA 7: Varved mudstone rhythmite intercalated with siltstone and minor sandstone. Sedimentological characteristics of the identified facies associations indicate four deposition environments, namely, deep marine basin, turbidite, shallow marine and lacustrine environments, which constitute a gradually regression sequence as a result of sea-level dropping and shallowing of the basin during the developmental processes. Geochemical analysis of the Ecca mudrocks and sandstones revealed that the rocks are of quartzose sedimentary provenance, suggesting that they were derived from a cratonic interior or recycled orogen. The petrography and geochemistry of the sandstones indicated that the source areas are composed of plutonic and metamorphic rocks with a minor component from sedimentary rocks. The geochemical diagrams and indices of weathering suggested that the granitic source rocks underwent moderate to high degree of chemical weathering. The tectonic setting discrimination diagrams support passive continental margin setting of the provenance.

D.Phil. (Geology) / A specific geological event has been characterized with the aid of an integrated metamorphic and fluid inclusion study of data obtained from syn-tectonic vein-quartz associated with thrusting and bedding-parallel shear along the northern margin of the Witwatersrand Basin. The vein-quartz associated with this event occurs as boudin-shaped bodies with their long and intermediate axes orientated within the foliation-, bedding- or fault-planes. The length of the quartz lenses which are spatially confined to shear zones often exceeds the thickness of the shear zones. These phenomena and the fact that quartz-fibres are orientated parallel to and not at right angles to the foliation confirms the syn-tectonic nature of the quartz veins. Heterogeneous P-T condition is indicative of imbrication, i.e. crustal thickening which is also substantiated by the random growth of pyrophylite and kyanite in shear zone assemblages, indicating that metamorphism outlasted deformation. Metamorphic studies of aluminous schists and vein-quartz with pyrophylliteand pyrophyllite - kyanite selvages established the development of two critical mineral assemblages: 1 Kaolinite + 2 Quartz = 1 Pyrophyllite + 1 H20 ... (1) and at higher P-T conditions 1 Pyrophyllite = 1 Kyanite + 3 Quartz + 1 H20 ... (2). The schists and quartz vein assemblages are quartz-oversaturated in contrast to the study material of Wallmach and Meyer (1990) which is quartz-undersaturated. Peak metamorphic conditions, therefore, are closely constrained by the position of the reaction curve (2) in P-T space, as is also substantiated by the presence of coexisting kyanite and pyrophyllite which are closely associated with syn-tectonic vein-quartz at the Florida Lake, Monarch Shaft and Krugersdorp localities. The nature of and circumstances under which the equilibrium aSsemblage pyrophyllite + kyanite + quartz has formed support an univariant situation, i.e. this assemblage can only coexist along the pyrophyllite kyanite isograd. The mineral assemblages that equilibrated during peak metamorphism are still present in the rocks of the shear zones, and show only incipient rehydration. The quartz-oversaturated nature of the rocks in the shear zones and the fact that kyanite formation is ascribed to reaction (2), cannot explain the abundance of quartz veins. Accordingly it is concluded that there must have been an external source from which Si02 was imported into shear zone to give rise to the formation of the large quantities of vein-quartz.