Abstract A detailed analysis, based on an Antarctica-Africa finite reconstruction at chron C29 (64 Ma), an assumption of no relative wander between the Marion/Prince Edward and Tristan hotspots, and on recently revised bathymetric maps of the Southern Ocean region, shows that the fixed hotspot reference frame is tenable for “absolute” plate motions. Bouvet hotspot, and probably Trinidade as well, also shows little or no Cenozoic relative motion. Contrary to previous models. Bouvet hotspot is unrelated to the Meteor Rise-Cape Rise seamount chain. Instead, the bathymetric data, when compared with the predicted hotspot tracks, indicate another hotspot exists near the southernmost South Atlantic spreading ridge segment. New geochemical evidence from the latter region supports this hypothesis in showing the effects of “plume enrichment” from a source that is compositionally distinct from Bouvet. The peculiar zig-zag shape of the Cape Rise-Meteor Rise lineament is the result of this hotspot crossing the active transform segment of the Falkland-Agulhas Fracture Zone in Late Mesozoic times, followed by an early Cenozoic ridge-jump to the pre-weakened trace on the then South American plate. From the averaged Cenozoic absolute motions of the African, Antarctic, and South American plates, it is evident that Antarctica has been most nearly stationary in an absolute motion sense.