Summary

During the 1950s and 1960s the Tanzanian volcano Oldoinyo Lengai erupted a highly alkalic lava composed almost exclusively of the Na-K-Ca carbonate minerals nyerereite and gregoryite and containing about 32% Na₂O and 7% K₂O. For many years the lava was considered a petrological curiosity but gradually the idea developed that alkalic carbonatite magmas might develop generally during the evolution of commoner carbonatite magmas. In 1981 Le Bas, followed in 1982 by Woolley, ascribed to the Oldoinyo Lengai magma a parental status and erected a scheme whereby other commoner carbonatite rock types are derived from it; the alkalic carbonatite magma is one of two derived by immiscible separation from a nephelinitic magma, the other being ijolitic. The carbonatite liquid has 500°C of superheat and loses alkalis progressively as the magma cools to its liquidus temperature of 400°C–600°C, by which stage it has become a calcitic-dolomitic liquid. This liquid then differentiates to produce the commoner types of carbonatite rocks.

In rejecting such a scheme we argue as follows: such a scheme would require the magma to be water saturated at the moment of immiscibility, and progressive loss of water and alkalis would induce crystallization thus preventing the formation of a calcite-dolomite magma; the 500°C of superheat is a consequence of the design of the Freestone & Hamilton experiments and does not exist; carbonatite magmas generally begin to crystallize at temperatures greater than 900°C; rare-earth-element compositions do not permit derivation of alkalic carbonatite magma from nephelinite magmas by liquid immiscibility; the mafic silicate mineralogy of carbonatites demonstrates alkali enrichment rather than depletion.

We propose that the commonest parental magma of carbonatites is a mildly alkalic olivine sövite composition and that most carbonatite rock types are derived from it by fractional crystallization, allowing the development of cumulates that are well lubricated by intercumulus liquid and capable of much subsequent movement, deformation and re-intrusion as a crystal mush. As this parental magma crystallizes, continued fractionation of alkalideficient and anhydrous minerals increases both its alkali and water contents until water saturation is reached. This in turn causes the development of an aqueous fluid which limits the alkali content of the magma. The composition of the magma at which water saturation develops is controlled by the rate of rise of the magma, and therefore aqueous fluids with various Na:K ratios can develop and control the type of fenitization that occurs. Under plutonic conditions alkalis in excess of the amount that can be fixed as silicate minerals are carried away as fenitizing fluids. Alkali loss does therefore occur, and through the medium of an aqueous fluid, but calcite and/or dolomite crystallize throughout the magmatic cooling history rather than simply at the low-temperature end of this history.

The alkalic carbonatite magma of Oldoinyo Lengai type also develops through fractional crystallization of the same alkali-poor olivine sövite magma, but under essentially dry conditions where the magma is kept liquid by alkalis and halogens, and alkali loss is prevented by the absence of an aqueous phase. Alkalic carbonatite magmas are therefore late differentiates of a more normal mildly-alkalic olivine sövite magma developed under low water fugacity. They are of very small volume and require long quiescence to develop. They are not to be considered parental to other carbonatite rock types.