Here, we adopt a dynamic model for vein formation in the crustal section underlying Atlantis Bank similar to that proposed by Dick et al. (1991), Robinson et al. (1991), and Dick et al. (1992). The felsic and plagioclase-rich veins must have formed very early as late magmatic melts penetrated a crystalline mush. The sharp planar boundaries of these veins indicate that the rocks were deforming in a brittle fashion and that there was little interaction between the veins and the wall rocks. As the late-stage melts cooled and evolved, they became more fluid rich, perhaps mixing somewhat with seawater. These late-magmatic or hydrothermal fluids, with temperatures generally between 600° and 700°C, followed the original cracks formed by brittle deformation, significantly modifying the original vein mineralogy.
Early in the crystallization history of the crustal section, a detachment fault developed in the newly formed crust as spreading continued. The main detachment fault is marked by the roughly 70 m of amphibole schists and gneisses at the top of the section, which are characterized by well-developed porphyroclastic and mylonitic textures. Considering that these rocks and the monomineralic amphibole veins within them formed essentially at the brittle-ductile transition, this deformation must have taken place largely at the ridge axis. Although some shear zones are present deeper in the hole, the intensity of high-temperature crystal-plastic deformation decreases significantly with depth, as does the abundance of amphibole veins and the extent of groundmass alteration (Shipboard Scientific Party, 1999). The abundance of amphibole in the zones of crystal-plastic deformation indicates that these zones were major conduits for hydrothermal fluids. The remarkably fresh nature of the lower 500 m of core indicates that very few fluids penetrated this part of the section and those that did probably had temperatures <100°C. Thus, there is clearly a close link between high-temperature crystal-plastic deformation and alteration in the Hole 735B gabbros. Brittle fracture of the crust appears to have played a less important role.
It is very difficult to constrain the rate of cooling of this crustal segment, which would have depended primarily on the spreading rate and the amount of seawater that penetrated the crust. The current rift valley on this portion of the Southwest Indian Ridge is ~8 km wide, and the spreading rate to the south of the ridge axis has been 8.5 ± 1.2 mm/yr for the last 25 m.y. (A. Hosford, pers. comm., 2000). If the rift valley was the same width at 11-12 Ma when the Atlantis Bank crust was formed, it would have taken roughly 500,000 yr for the crustal segment to migrate across the valley floor and reach the rift valley walls. The downward decrease in the intensity and temperature of alteration suggests that cooling of the crust was relatively rapid. Detachment faulting associated with the crystal-plastic deformation at the top of the section would have thinned the crust and allowed penetration of cold seawater, leading to rapid cooling.
Circulation of high-temperature hydrothermal fluids would presumably have ceased by the time the crustal segment reached the edge of the rift valley and was transferred to adjacent valley walls. Thus, the brittle-ductile deformation associated with formation of the felsic, plagioclase-rich, and amphibole veins took place relatively early near the ridge axis. The hydrothermal overprinting of the early felsic veins and the formation of the lower-temperature diopside veins probably took place as the section moved across the rift valley floor and brittle deformation became dominant. Uplift of the crust into the rift valley walls would have increased brittle fracturing and reopened existing veins, leading to the development of the late-stage prehnite, smectite, and chlorite veins. The latest oxidative alteration associated with carbonate veins presumably took place after the Atlantis Bank had been unroofed and uplifted into the transverse ridge on the edge of the fracture zone.