ALBERT

Notes: Abstract Anjouan is one of four volcanic islands comprising the Comores Archipelago. Three (arbitrarily defined) categories of basic magma are recognised on Anjouan: ‘hypersthenenormative’, ‘alkalic’ and ‘basanitic’, which appear in that order with an eruptive sequence involving 1) shield construction, 2) peripheral fissure-controlled activity, and 3) rejuvenescent (posterosional) eruptions. Differentiated magmas have evolved within the three chemical groupings, and trend mainly towards undersaturated trachyte and phonolite. These trends are considered to have developed by initial removal of olivine and clinopyroxene, followed by Fe-Ti oxides, apatite and amphibole from arrested liquid pools within and beneath the volcano. The appearance of feldspar on the liquidus was clearly inhibited by the high contents of normative diopside in most Anjouan magmas, although late stage plagioclase fractionation is probably responsible for development of peralkaline phonolites at shallow depths, assisted eventually by alkali feldspar. Lherzolite-xenolith-bearing lavas are likely to be directly mantle-derived liquids. Three analysed representatives with 100 · Mg/Mg + Fe2+ (atomic) ratios approaching 70 are characteristically rich in normative diopside and may confirm the suggestion of primitive ankaramitic melts in mantle regions. High pressure fractional crystallisation may involve fractionation of ‘eclogite’, orthopyroxene or clinopyroxene, with or without olivine. It is believed however that such processes do not adequately explain the compositional change from hypersthene-normative basalt towards basanite, as they imply unacceptable degrees of Fe-enrichment. Problems of large ion element enrichment and fractionation would also remain, while the high contents of Cr and Ni place limits on the extent of crystal fractionation. A partial melting model of magma genesis is able to explain the main features of basalt geochemistry, especially if the upper mantle low velocity zone is regarded as having been severely fractionated due to upward migration of large ion elements. An ‘open’ system of magma production in the context of regional plate movement, and the implied decoupling between lithosphere and asthenosphere, accomodates realistic degrees of partial melting and allows a greater potential volume of mantle available for melting than possible ‘closed’ systems. The model also accounts for the migratory pattern of Comores volcanism.