The onset of plumes where broad-scale convection already exists is investigated along with the thermal history of the earth. Geological evidence relating to the onset time of mantle plumes is not definitive: no well-defined hotspot tracks older than Mesozoic have been identified, but it is difficult to tell whether this absence is a reliable indication of a late onset of mantle plumes, since only continental crust of this age exists. Using the hotspot track distribution for the Pacific, the absence of Pre-Mesozoic hotspot tracks on the Canadian Shield cannot easily be attributed to chance. However, young continental hotspot tracks are controversial, and it is not clear whether tracks such as the Monteregian Hills/White Mountain Magma Series would be recognized as such without considering the contributing evidence of the adjacent oceanic crust and its well-defined seamount chains. Our simple models yield potentially testable predictions: the instability of plume formation in a system where broad-scale convection already exists depends on horizontal conduit flow in the basal boundary layer toward the plume source. Geographically, plumes are most likely to form where the boundary layer is thickest and where broad-scale ascending flow already exists. Parameterized convection indicates that the heat transport by conduit flow in the boundary layer toward plumes depends linearly on the ratio of effective viscosity for broad-scale convection to effective boundary layer viscosity. In terms of material properties, this ratio is the ratio of viscosity across the boundary layer to about the 0.45 power for the case where the broad-scale convection is heated from below. The viscosity contrast across the boundary layer depends on the temperature contrast across the layer. A late onset of mantle plumes is implied if the temperature contrast at the base of the mantle has increased with time, that is, if the core has cooled more slowly than the interior of the mantle. In one scenario, before about 3 b.y. B.P., the mantle temperature varied slowly because surface heat flow was nearly balanced by internal radioactivity.

During this time the mantle was hotter than present, and its internal viscosity lower. The temperature contrast at the base of the mantle was thus small. After 3 b.y., internal radioactivity no longer balanced surface heat flow, and the mantle cooled about a few hundred degrees Celsius, as indicated by petrological observations. The core cooled slower than the mantle, and the temperature contrast across the bottom boundary layer increased with time, leading to a late onset of plumes. Phenomenological inferences are obtained, as Venus may have cooled more rapidly than the Earth. Its mantle may be too cool for rapid plate tectonics, and the gravity field and the convective pattern are dominated by hot upwelling regions. ¿ American Geophysical Union 1988