Kimberlite breccias from several South African pipes or pipe-like enlargements on dykes contain abundant fragments of country rocks including huge, often brecciated, xenolithic masses which are located at positions well below their original stratigraphic levels. The xenoliths show no pyrometamorphic effects indicative of relatively high temperatures during their incorporation in the kimberlite. A striking petrographic feature of these rocks is an emulsion-like texture resulting from the presence of numerous phlogopite-rich, globular segregations, which are set in a matrix which sometimes consists almost solely of primary calcite but in other instances is comprised of variable proportions of calcite and serpentine. Evidence is presented which indicates that in the latter cases serpentine has replaced calcite. The clear separation of the mica-rich segregations and the essentially calcitic matrix is attributed to the development of low temperature, immiscible K-rich silicate and carbonatitic liquids during emplacement of these rocks. The carbonatitic liquids are considered to have acted as the primary transporting fluids during liquid-solid fluidized intrusion and emplacement of the rocks is considered to have taken place as follows. After the crystallization of early phenocrysts the residual, ascending kimberlite magma differentiates at relatively high crustal levels into immiscible silicate and carbonatitic liquids accompanied by a coexisting gaseous phase. Continued intrusion is accompanied by the separation of the gaseous phase and its accumulation at the head of the magma column. Upon further uprise a stage is reached where the internal gas pressure exceeds the lithostatic load and the diatreme is formed by explosive breaching of the cap rocks. The resulting rapid pressure drop is accompanied by the upsurge of partly degassed magma fractions which incorporate explosively disrupted cap rock fragments and material which slumps from the walls of the diatreme.