At present, young oceanic lithosphere is positively buoyant, and it does not become negatively buoyant until it is older than about 20 m.y. If, in the past, the mantle was hotter, the oceanic crust would have been thicker and the lithosphere's age of neutral buoyancy greater. On the other hand, the hotter mantle would have had a lower viscosity and convected faster, so the average age of oceanic plates at subduction would have been less than the present 100 m.y. At some time in the past, average oceanic lithosphere would have only just become negatively buoyant as it reached a subduction zone. It is estimated here that this condition occurred when the mantle was only about 50 °C hotter than at present and as recently as 0.9-1.4 b.y. ago. Transformation of basaltic crust to eclogite might have enhanced the vigor of the plates, but possibly only modestly and intermittently. In earlier times, plate tectonics could have operated, but more slowly, so that it could not by itself have accomplished the necessary rate of heat removal from the earth. A different tectonic mode must also have operated, perhaps subcrustal delamination or dripping. Plate tectonics would have gradually taken over from the earlier mode. Plumes are a omplementary mode of mantle convection driven by a lower thermal boundary layer, and so they could have operated in conjunction with both regimes of the upper boundary layer.