The Southwest Indian Ridge is well described as fracture-zone country. Ridge segments are usually <100 km long, and in many cases they are <50 km apart (Sclater et al., 1981; Fisher and Sclater, 1983; Patriat et al., 1997). In some cases, including that of the Atlantis II Fracture Zone, the transform offsets are actually longer than the lengths of immediately adjacent ridge segments. Generally in the Indian Ocean, long-offset transforms typically have high transverse ridges, usually on both sides (Fisher and Goodwillie, 1997). The relief of transform ridges is substantial, in many cases as much as 5 km, and portions of several of them reach nearly to sea level. They dominate both the topography and the geoid along the Southwest Indian Ridge (Rommeveaux-Jestin et al., 1997).
Between Legs 118 and 176, two conceptions of the process of crustal accretion near transform faults at spreading ridges gained currency. The first is the general idea of ridge segmentation, of which transform faults are only one aspect but probably the most important one at slowly spreading ridges. Geophysical evidence now indicates that between the closely spaced transform faults at slowly spreading ridges magma supply is not uniform. Both geoidal bulls-eye patterns (Kuo and Forsyth, 1988; Lin et al., 1990; Detrick et al., 1995) and seismic studies (Tolstoy et al., 1992; Magde et al., 1997, 2000) centered on rift valleys in the North and South Atlantic show that the crust thins away from segment midpoints and toward the transform faults. Ostensibly, there is less basalt and gabbro in the crust near transform faults. This does not altogether explain the plutonic rock assemblages so commonly found on transverse ridges, but since these structures arise nearer the tapered tips of the accreted crust than the centers of segments, geophysics says that there are certainly fewer superficial rocks to offset by faulting in order to expose peridotite.
The second conception is that of asymmetric faulting on low-angle detachment faults leading to exposure of core complexes, also termed megamullions, on the transverse ridges of slowly spreading ridges (Cann et al., 1997; Tucholke et al., 1998) and of gabbroic and ultramafic rocks (Cannat, 1993; Lagabrielle et al., 1998). Although Dick et al. (1991a) considered that an assymetric detachment mechanism was responsible for separating basalts, mainly, from gabbros drilled in Hole 735B, the general structure of transverse ridges was not yet at all understood from the perspective of marine geophysics. The grooved surfaces of core complexes seen in the high-resolution bathymetry of a number of core complexes in the Atlantic Ocean, with the grooves evidently representing the striations produced by one block of rock sliding over another, were as yet unknown. Now, however, transverse ridges are known to consist at least in part of core complexes, which thus far, however, are only documented adjacent to rift valleys. Some of the grooved summits of these are quite flat although they steepen toward the rift valleys, suggesting that exhumation of the blocks is accomplished by rotation along curving detachment surfaces, rooted in the rift valleys, which roll over and flatten as the blocks gain elevation.
The postcruise surveys of Atlantis Bank where Hole 735B was drilled show that it, too, is a core complex (Dick et al., 1999; Hosford et al., 2000). This one happened to reach sea level. There, the grooved fault surface was partly skimmed off by erosion and covered with a thin patina of platform carbonates after the summit began to subside (MacLeod et al., 1998, 2000). The core-complex hypothesis explains the occurrence of intense crystal-plastic deformation at many places in the gabbros. It explains particularly why this deformation is so intense near the top of the hole. It explains why this deformation occurred at such high temperature with magmas present beneath the floor of the rift valley and prior to removal of some 2 km of overlying rock. It explains why the orientation of all of the foliation in the gabbros, when considered in relation to magnetic inclination, dips to the north toward the rift valley. Finally, it explains why magnetic inclinations up and down the core are steeper by an average of 19° than they should be for the magnetic paleolatitude at the site. Whether drilled by accident or not, Hole 735B penetrates the heart of one of these structures.
The surveys also show that Hole 735B projects to a point about midway between the center of the ridge segment lying to the north of the site and the trace of the transform fault in the transform valley. In the ideal geometrical sense, the original thickness of the crust beneath the site, before it was rent asunder by asymmetric faulting, was probably somewhat less than at the segment midpoint. This may explain why no truly primitive gabbros have been cored at the site (Dick et al., 2000) or dredged from the transform wall and why, in general, they are absent in gabbroic assemblages dredged from transverse ridges (e.g., Francheteau et al., 1976; Fisher et al., 1986; Bloomer et al., 1989).
To all of this, we add a more general consideration about the Southwest Indian Ridge. There and on the Central Indian Ridge transform offsets are numerous and usually closely spaced. Most of them do not displace ridge segments greatly; others, like the Atlantis II Fracture Zone, separate ridge segments by >150 km. The Southwest Indian Ridge itself is propagating relatively eastward into the gap created by the divergent directions of spreading of the Central and Southeast Indian Ridges (Patriat and Parson, 1989; Patriat et al., 1997). At the inception of the Atlantis II Fracture Zone when it was nearer the propagating tip, the ridge segments offsetting it were so close to the Indian Ocean triple junction that they cannot have been very far apart. The transform offset thus lengthened over time. This has been a continuing process. Magnetic profiles obtained during the original site survey (Dick et al., 1991b) revealed a slower spreading rate toward the north from the ridge segments adjacent to the fracture zone than to the south. Spreading thus has been steadily asymmetric for at least the past 20 Ma, although whether offset lengthening also resulted from propagation of nontransform offsets into the transform is uncertain. Ultimately, the ridge segments came to be >150 km apart. As the transform fault lengthened, the relief of the transverse ridges also increased, although by what mechanism is also uncertain. Nevertheless, we must construe the section cored in Hole 735B in terms of three aspects of structural asymmetry—the first being an original unequal distribution of magma along axis, the second being that associated with detachment faulting and formation of a core complex orthogonal to the ridge axis, and the third recognizing that the state of stress along the transform fault has continually but perhaps not evenly changed as it lengthened. We do not yet know how pulses in magmatism and formation of core complexes are tied to the long-term development of the Atlantis II Fracture Zone. Hole 735B represents but an instant in time in this history. We must be extremely cautious in extrapolating results from there to the broader problem of the origin of ocean crust.