SummaryA gradational increase in concentration of CO2 towards the margins of a very fine-grained basalt dyke has led to the development of a pale marginal facies, enriched in carbonates, particularly siderite. Increases in CO2 from about 3 % in the interior of the dyke to 8 % at the margins are accompanied by decreases in SiO2 and Fe2O3, increases in Al2O3, and less significant changes in the other major components. Cu, Co, and Zn change only slightly, and Cr, Ni, and Li remain constant. Petrographic variation is considerable, even in the superficially homogeneous interior of the dyke, in which it ranges from a type containing a titanaugite (analysed) to one devoid of pyroxene, but containing conspicuous opaque minerals. Microprobe analysis for Fe and Ti shows that these comprise: cotahedral titaniferous magnetite; rodlets, less than 1 µm thick, of rutile partly altered to, or overgrown by, ilmenite; and sub-opaque patches with a very low Ti/Fe ratio. Plagioclase, An55, is the most abundant and constant crystalline phase in the interior of the dyke, but changes to An20 in the marginal facies. Mineral-content of the latter, deduced from optical, chemical, and X-ray data, also includes siderite, serpentine and clay minerals, leucoxene, and apatite. There is no evidence of quartz, sericite, or calcite.Petrographic evidence shows that variations in concentrations of CO2 and H2O affected phase equilibria from the start of magmatic crystallization. Data on the fO2 required for TiO2 and Fe-oxide phases to co-exist at magmatic temperatures indicate that, initially, the concentrations of CO2 and H2O in the interior of the dyke were higher than the values recorded in analyses of the rocks. From this evidence and the field relationships, it is concluded that the intruding magma was rich in volatiles, which diffused towards the dyke margins and, in part, became trapped as the magma congealed, producing a changed marginal assemblage of minerals. The dyke provides a unique glimpse of influences on a basic magma exerted by what S. J. Shand has aptly termed the ‘fugitive constituents’, the transient effects of which are rarely preserved in the rocks.