The tectonic evolution of the Indian Ocean ridge-ridge-ridge triple junction, the junction of the central Indian Ridge (CIR), Southeast Indian Ridge (SEIR) and Southwest Indian Ridge (SWIR), is inferred by mapping structures along its traces on the three plates with a long-range side scan sonar, Geological Long-Range Inclined Asdic (GLORIA). The sonar images to the west of the triple junction show two different styles of tectonic evolution. Some sections of the trace on the African plate are marked by a SWIR-facing scarp containing fine lineaments that are probably normal faults. These correspond with sections of the trace on the Antarctic plate of the same age that have a sharp intersection of SWIR and SEIR abyssal hills that is to be expected from a simple ridge-ridge-ridge triple junction evolution (a ''herringbone'' pattern of isochrons). In contrast there are other sections of the trace on the African plate that have a blocky structure with no definite termination of CIR abyssal hills against those of the SWIR. The corresponding regions of the Antarctic plate show large 10- to 30-km long overlapping fault scarps, tapering to the northeast and ending in asymmetric valleys to the southwest. These major normal faults are each offset by 1--6 km, forming an en echelon series. The two different styles of tectonic structures indicate two different modes of evolution, although why the triple junction has evolved with two different modes is unclear. Periods of asymmetric spreading in the CIR magnetic anomaly sequence collected on the African plate may possibly indicate that the CIR and SEIR have been offset by a fracture zone at the triple junction during one of these two modes.

Echo sounder traverses on the Antarctic and African plates commonly show a steady shallowing of the CIR and SEIR seafloor toward the traces, a steep scarp facing the SWIR and a broad marginal deep at the base of this scarp. This rift flank shape resembles that of the SWIR valley's rift shoulders close to the present triple junction (Mitchell, 1991a), suggesting that the mechanical evolution has been similar to the present day and the relief due to the rift flank uplift has been preserved by the cooling and strengthening of the lithosphere with age away from the ridge axes. If the shape of the rift shoulder uplift along the trace on the Antarctic plate is interpreted with a simple flexural model involving an end-loaded elastic plate, the inferred rigidity (1019--1021 N m or 2--6 km effective elastic thickness) is consistent with the suggestion that the isostatic uplift occurred in young, weak lithosphere, close to the triple junction. Interestingly, the observed ~1-km relief of the trace escarpment and the estimated rigidity suggest that the uplift was produced by a load (2--5¿1011 N/m) that is equivalent to half the present cross-sectional area of the SWIR valley 35 km from the triple junction. ¿ American Geophysical Union 1993