Prior studies have proposed and examined the hypothesis that India and Australia are separate rigid plates separated by a wide, near-equatorial, E-W striking, plate boundary. Attempts to place narrow limits on the location of the Africa-Australia-India triple junction have been hindered, however, by the lack of useful magnetic profiles crossing the eastern Carlsberg Ridge and northern Central Indian Ridge. Herein we present near-ridge portions of new profiles from an aeromagnetic survey of the Carlsberg Ridge east of 66 degrees E and of the Central Indian Ridge north of 19 degrees S. These new data are used to estimate 35 new spreading rates averaged from the middle of chron 2A (3.03 Ma) to the present. All other plate motion data along the Central Indian and Carlsberg ridges are also analysed to investigate the present kinematics of the Indian Ocean, especially the motion and boundary between the Indian and Australian plates. Unlike prior efforts, we objectively estimate uncertainties in the strikes of transform faults along the Carlsberg and Central Indian ridges. Indian Ocean plate motion data are unambiguously inconsistent with a model in which India and Australia lie on the same rigid plate, but remain consistent with the existence of distinct and rigid Indian and Australian plates. The plate motion data also remain consistent with closure of the Africa-Arabia-India plate motion circuit. The data are consistent with closure of the Africa-Antarctic-Australia plate motion circuit and place an upper bound of 7 mm yr-l on the summed deformation around the Rodriguez triple junction if we have accurately estimated the errors in the data. From data along the Carlsberg and Central Indian Ridges, 95 per cent confidence limits on the location of the Africa-Australia-India triple junction are 6.2 degrees S to similar to 9 degrees S if the boundary between India and Australia is discrete (i.e. very narrow) where it intersects the Central Indian Ridge. When closure is enforced about the Owen and Rodriguez triple junctions, these limits decrease to 8 degrees S-9 degrees S, which is centred on the Vema fracture zone and is more than 10 times narrower than we found without the new data. The resulting specific prediction of the location of, and velocity across, a hypothetical narrow boundary between the Indian and Australian plates permits us to test the hypothesis of a narrow boundary. Available data suggest, but do not prove, that no such narrow boundary accommodating the predicted direction of motion exists. We think it likely that the motion is taken up on many fracture zones between similar to-8 degrees S and similar to-4 degrees S, which is an interval containing most of the near-, but off-ridge, seismicity. The confidence limits on the new angular velocities, especially those describing the motion between India and either Africa or Australia, are much smaller than we found before. In particular the NW-oriented axis of the confidence ellipse of the India-Australia pole of rotation is seven times smaller than we found before. The shrinking of these uncertainties is accompanied by a shrinking of the uncertainties in the predicted velocities between India and Australia. For example, a point (8.5 degrees S, 68.3 degrees E) assumed to be on the Australian plate near the Central Indian Ridge moves 5.1 +/- 2.2 mm yr(-1) toward S38 +/- 12 degrees E (95 per cent confidence limits) relative to the Indian plate, indicating net divergence of Australia from India along a N-S profile near the Central Indian Ridge. A point (11 degrees S, 90 degrees E) assumed to be on the Australian plate near the Ninetyeast Ridge moves 9.8 +/- 2.4 mm yr(-1) toward N27 +/- 12 degrees E relative to the Indian plate, indicating net convergence of Australia toward India along a N-S profile through the Ninetyeast Ridge. These now very specific predictions remain consistent with independent data that indicate the orientation and (to a lesser extent) the rate of deformation in the zone between India and Australia. The data reinforce our earlier conclusion that the most important way in which shortening is taken up in the non-subducting convergent portion of the plate boundary, i.e. the eastern portion, is by strike-slip faulting and north-eastward delivery of oceanic lithosphere to the Java-Sumatra trench.
