In developing the concept that chemical plumes have a deep-mantle origin, I propose that plumes result from original chemical inhomogeneities in the Earth. According to the hypothesis of in-homogeneous planetary accretion, the terrestrial planets formed with refractory cores and volatile-rich outer shells, that is, they are layered according to the sequence in which compounds condense from a cooling nebula. The primitive deep mantle is enriched in CaO-Al2O3-TiO2 and the refractory trace elements and depleted in MgO-FeO-SiO2 and the volatile elements relative to normal olivine-pyroxene mantle. The refractory trace elements include W, Ir, Y, Zr, Nb, Ba, Sr, rare-earth elements, and, most important for the present discussion, U and Th. In the deep mantle, this refractory material is less dense than ferromagnesian silicates and will rise until phase changes and loss of the low–melting-temperature fraction permit density equilibration. The low–melting-temperature fraction of this primitive refractory assemblage is anorthite, and a widespread anorthosite event probably occurred early in the history of the Earth. I propose that chemical inhomogeneities still exist and because of their high U and Th abundances provide the heat source for driving upper mantle asthenospheric convection. A conservative estimate indicates that the heat flow above frozen-in plumes is at least twice the normal heat flow at the base of the low-velocity layer. This heat is transmitted efficiently by convection to the base of the lithosphere, providing the “melting spot” required to explain linear island chains in the oceans and igneous traces on continents. When a continent comes to rest over one of these radioactive hot spots, the temperature at the base of the continental lithosphere is perturbed, causing doming or swelling and eventual fracture and magmatism. It is also possible that kimberlite and carbonatite intrusions are a result of thermal perturbations that result when continental lithosphere overrides a chemical hot spot. The locations of chemical plumes may control the patterns of continental breakup and dispersal even if they are dynamically passive.