This study comprises a petrographical and mineralogical investigation of rocks from an area 850 sq. km in size, situated about 80km northeast of Middelburg. Roughly half of the area is occupied by rocks of the epicrustal phase of the Bushveld Complex, and consists largely of Rooiberg Felsite and granophyre as well as leptite, microgranite and granodiorite. Numerous veins of finegrained granite traverse the leptite which is considered to be highly metamorphosed felsite. These veins of fine-grained granite probably owe their origin to the melting of the leptite. The coalescence of these products of melting gave rise to the thick sheet of.granophyre between the leptite and the felsite. Rocks of the Layered Sequence occupy the eastern half of the area and consist of the Main and Upper Zones which were subdivided into various subzones on the basis of characteristic rock types and marker horizons. Mineralogical investigations are restricted to the minerals from rocks of the Layered Sequence, namely orthopyroxene, plagioclase, apatite and the sulphides of the Upper Zone. In Subzone A of the Main Zone, the orthopyroxene is present as cumulus crystals, but it changes in texture to ophitic in the lower half of Subzone B where small discrete grains of inverted pigeonite are also developed. Inverted pigeonite is present in the upper half of Subzone B and in rocks of the Upper Zone, whereas the orthopyroxene-pigeonite relationships in Subzone C of the Main Zone are a repetition of those observed in the underlying rocks. The phase-change from orthopyroxene to pigeonite takes place over a transition zone in which both phases crystallized from the magma. It is envisaged that the first pigeonite to have crystallized from the magma at high temperatures had a lower Fe/Mg ratio than the hypersthene precipitating at slightly lower temperatures, with the result that the early formed pigeonite was unstable and reacted with the magma to form hypersthene. This caused the formation of groups of grains of hypersthene which are optically continuous over large areas and which may contain a few blebs of augite exsolved from the original pigeonite. A few pigeonite grains were effectively trapped in other minerals, mostly augite, and consequently escaped reaction with the liquid. These inverted to hypersthene at the appropriate temperature and contain numerous exsolution-lamellae of augite. As fractional crystallization of the magma continued, it moved further into the stability field of pigeonite and out of the stability field of hypersthene with the result that the formation of hypersthene by the reaction of pigeonite with magma was replaced by inversion of pigeonite to hypersthene. This inverted pigeonite is also present as groups of grains optically continuous and contains pre-inversion exsolution-lamellae of augite orientated at random, and post-inversion exsolution-lamellae which are orientated parallel to the (100) plane of the orthopyroxene throughout a unit. The inverted pigeonite is orientated in such a way that its crystallographic c-axis lies close to or in the plane of layering. This is explained as being due to the load pressure of the superincumbent crystal mass during the inversion. Textural features of the plagioclase revealed interesting information on the postcumulus changes in the rock. Reversed zoning, interpenetration and bending of plagioclase crystals as well as the presence of myrmekite are described. These are considered to be due to increased load pressure prior to and during crystallization of the intercumulus liquid. It is considered that the various types of pegmatoids may have originated by an increase in pressure on the intercumulus liquid which was concentrated to form pipe-like bodies by lateral secretion or filter pressing. Cumulus apatite is developed in the olivine diorites of Subzone D of the Upper Zone. From unit cell dimensions it seems as if it changes in composition from a fluor-rich hydroxyapatite at the base of this subzone to a relatively pure hydroxyapatite 70m below the roof. There seems to be a substantial increase in the fluor content of the apatite in the topmost 70m of the intrusion. Rocks of the Upper Zone contain considerably more sulphides than those of the Main Zone. This is ascribed to an increase in the sulphur content of the magma owing to fractional crystallization. The magma reached the saturation point of sulphur when rocks of Subzone D of the Upper Zone started to crystallize with the result that these rocks contain numerous small droplets of sulphide which constitute on an average about 0, 5 per cent by volume of the rocks. A concentration of the sulphides in these rocks would not yield a deposit of economic interest because of the unfavourable composition of the sulphide phase, which consits of more than 90 per cent pyrrhotite. Sulphides in the rocks below this subzone are intercumulus and a concentration could be of economic importance because the sulphide phase contains appreciable amounts of chalcopyrite and pentlandite. Although no economic concentration of sulphides are known from the Upper Zone, this study has revealed the presence of a mineralized anorthosite below Lower Magnetitite Seam 2 which contains in places up to 1 per cent Cu, 0, 18 per cent Ni and 1, 6g/ton platinum metals. Continuous, slow convection and bottom crystallization probably gave rise to the homogeneous rocks of the Main Zone. Injection of a considerable amount of fresh magma took place at the level of the Pyroxenite Marker which resulted in a compositional break and gave rise to a repetition in Subzone C of the rocks of the Main Zone below this marker. The oxygen pressure during crystallization of the magma was probably low, causing a gradual enrichment in iron in the magma and gave rise to the appearance of magnetite at the base of the Upper Zone. Intermittent increase in the oxygen fugacity is considered to be important in the formation of magnetitite seams. As a result of fractional crystallization the volatile content of the remaining magma gradually increased. This is seen firstly, by the appearance of biotite secondly by the appearance of cumulus apatite and droplets of sulphide and lastly by hornblende in the rocks of the Upper Zone. Some water-rich residual liquids apparently also intruded the overlying leptite, causing additional melting of the latter and the formation of irregularly shaped veins and pockets of granodiorite. A lateral change in facies of the rocks of the Layered Sequence in a southerly direction is described. This is considered to be due to crystallization of the magma at slightly lower temperatures because of the more effective heat loss where the magma chamber was thinner. Two parameters of differentiation for layered intrusions are proposed, viz. a modified version of the differentiation index and a modified version of the crystallization index. The former seems more applicable for intrusions such as the Bushveld Complex, whereas the latter seems to be more applicable for intrusions in which there is a considerable development of ultramafic rocks. These two parameters can also be used to indicate the differentiation trend if they are plotted against height in the intrusion.