We propose a tight fit at 200 Ma for India, Madagascar, Sri Lanka and Antarctica against the east coast of Africa that avoids the difficulty of accounting for the gaps between continents required by looser fits. We have followed three lines of argument: (1) A credible animation for the modes of dispersal of these fragments is necessary that is consistent with new ocean-floor topographic data (http://kartoweb.itc.nl/gondwana); (2) Pre-drift rifting has also to be 'undone' and we argue for a separation of only 50-100 km between extant (but now widely separated) rift shoulders that mark the boundaries of the rigid plates of ancient crust that has not been distended by rifting; (3) The margins of plates concealed by younger cover formations are defined through interpretation of aeromagnetic anomaly data. Our new fit is substantiated by the integrity of the (largely pre-200 Ma) Karoo rift basins of Tanzania, Kenya and Madagascar and by the co-linearity of a zone of shears, interpreted from aeromagnetics in northern Mozambique, with the line of rupture between India and Madagascar. Even tighter fits for earlier times would follow, logically, the undoing of Karoo rifting ($300-200 Ma) but this requires further work. Three published reassemblies that compared Precambrian geology across present plate boundaries are presented as figures in their original form and with the fragments reassembled to our new template. They present no real difficulties for our model. Euler rotation parameters for our reassembly are listed to allow others to test our model on their own data.