Goldfields extend for 300 km around the margin of the Archaean Witwatersrand Basin of South Africa associated with regional greenschist facies metamorphism and deformation. Metamorphic mineral assemblages involving pyrophyllite-chloritoid are associated with gold in all goldfields, reflecting low pressure and 300 to 400 degrees C conditions, and indicating high geothermal gradients. The origin of Witwatersrand gold can be explained by metamorphic devolatilisation to generate auriferous fluids beneath and outside the Witwatersrand Basin, followed by passage of these fluids along large structures and into the Basin. In this model, generation of the fluid is a consequence of the transition to amphibolite facies of mafic rocks such as Archaean greenstone belts, as in Phillips and Powell (2010). In a grain-by-grain scale devolatilisation process, the lower temperature minerals have released their H2O, CO2 and H2S to form a metamorphic fluid which at the time of its formation has already dissolved gold from these unenriched rocks (say at the similar to 2 ppb Au level). A substantial volume of auriferous fluid is inferred given that regional metamorphism is on the scale of thousands of cubic kilometres, all evolving a few per cent of fluid. In the model, this auriferous fluid migrated from the immediate source region along grain boundaries, then in shear zones, and into the Witwatersrand Supergroup via the major thrust faults adjacent to all goldfields. Once in the Supergroup, fluid migration was via bedding planes and especially unconformity surfaces, along cross-cutting faults, shear zones and reef packages. The few thicker shale units and especially the overlying Archaean Ventersdorp basalt pile acted as barriers to fluid flow and facilitated fluid focussing. Deposition of gold from this auriferous fluid occurred where there was abundant carbon and/or iron. The distribution of these two elements was pre-determined by sedimentary distribution, and diagenetic migration. Many bedding features were preserved or mimicked during the hydrothermal replacement process because the volumetric strain in most of the sequence was low. Critical to the success of the hydrothermal process in producing gold deposits were the widespread local concentrations of C and Fe on unconformity surfaces, the reef packages, the major thrust faults adjacent to each goldfield, source rocks of greenstone-like character that were once at (sub)-greenschist facies grade with assemblages that included chlorite-calcite-pyrite, and a regional metamorphic event of high geothermal gradient. Moreover, critical to the exposure and preservation of the goldfields has been the limited erosion of the Basin since the hydrothermal deposition of the gold. The metamorphic devolatilisation-hydrothermal replacement model predicts many features of Witwatersrand ores, and can explain quantitatively the origin of its similar to 100 000 t gold endowment. Calculations suggest that there is adequate gold at background (2 ppb) levels in inferred source rocks to form the necessary auriferous hydrothermal fluids; in contrast, there appears to have been insufficient particulate gold to form a significant placer accumulation. Without a viable source region for detrital gold, and indeed no plausible sorting mechanism to produce such deposits, it would be unwise to accept a placer model.
