The Mesoproterozoic rocks of the Bysteek and Koenap Formations of the Arribees Group are exposed within a NW-SE striking antiformal structure comprised of mafic granulites and metapelitic diatexites, and a number of marble and calc-silicate rock layers. The mafic granulites of the Bysteek Formation show a typological variety of anatectic features, including nebulitic, stromatitic mesosomes, melanosomes, quartz syenitic leucocratic vein networks and syenitic pools. Melanosomes consist of hedenbergitic to diopside-rich clinopyroxene (XMg: 0.40), anorthitic plagioclase (An90), with some quartz, minor apatite and titanite. Anatexis was caused by biotite dehydration melting and formed a melt of probably granitic composition. The leucosome composition ranges from either alkali-feldspar-granitic to plagioclase rich or granitic. This variation is interpreted as a result of variable extraction of melt from the source to granitic pools. The diatexites of the Koenap Formation are most likely of metapelitic or meta-greywacke origin. They are texturally variable but always contain high modal contents of alkali feldspar and quartz which generally form magmatic textures. Almandine-rich garnet (XMg: 0.18-0.25), cordierite (XMg: 0.71) form secondary biotite, sillimanite and magnetite during retrograde breakdown. Thermodynamic modelling of mafic granulite compositions suggests peak P-T conditions of ~865 °C and 8.6 kbar. Occasionally, garnet rich in ferric iron (XAdr: 0.55) forms by plagioclase-clinopyroxene breakdown under oxidising conditions at ~6 kilobar and ~ 800 °C. At the same stage amphibole forms in some melanosomes. P-T estimations for the diatexites based on thermodynamic modelling suggest the equilibration of the assemblage garnet, cordierite, alkali feldspar and melt at ~860 °C and 5.5 kbar. Conditions comparable to the peak pressure in the mafic granulites could not be established. However, since the diatexites and the mafic granulites are closely related in the field and no evidence of juxtaposition after the thermal peak exists, the P-T record of the diatexites might be incomplete