Closely spaced samples spanning 'flow units' at sub-bottom depths of 500 and 590 m in the drill hole at site 332B and a pillow at depth of 460 m, site 335, as well as a few individual samples at various depths in these holes and in the hole at site 332A, were studied with a view to determining compositional changes effected by submarine alteration. The rocks are predominantly plagioclase-olivine phyric tholeiitic basalts, with the exception of the lower unit of 332B, which is a picrite formed by the concentration of olivine megaphenocrysts in olivine tholeiitic basalt, is markedly flow-differentiated, and is assumed to be a sill. Three generations of plagioclase and olivine evident in some of the samples as corroded megaphenocrysts (An86−78, Fo89), euhedral microphenocrysts (An76−72, Fo85−84), and groundmass crystals (An72−61, Fo85−84), record a crystallization history that begins at depth, continues en route to the surface, and ends with quenching on the sea floor. Filling interstices of the groundmass are intergrowths, commonly submicroscopic, of pyroxene–plagioclase and dark, poorly resolved titanomagnetite-charged magmatic residue. The pyroxenes arc augites, ranging to subcalcic and ferroaugite. In places, particularly near pillow and flow unit margins, magma residue partially or completely in-fills vesicle cavities (segregation vesicles). Volatile-bearing phases (notably chlorophaeites, saponite, palagonite, amorphous silica-bearing hydrous iron oxides, and carbonates) tend to be characteristic of the three principal sites in which they are found: palagonite in the glassy margins of pillows and (or) flow units, complex hydrous mineraloids adjoining veins, and chlorophaeites and saponites in the interstices of the crystalline matrix of rocks remote from veins. The latter have the aspect of primary minerals. Palagonitization results in gains in K, Fe, Ti, and Cl, and losses in Ca, Mg, and Na, but net gains and losses in the other sites are less certain. Hydration and oxidation of iron invariably accompany development of volatile-bearing phases, but there is no correlation between these parameters and variation in content of the other analysed elements. Principal component analysis shows that the major part of the compositional variation can be explained by primary factors. In these samples chemical exchange with seawater appears to be limited, possibly because of their rapid isolation by burial from the main body of ocean water.