The physical process of mantle convection affects the chemistry of the mantle. The chemical changes occur mostly through melting, directly or indirectly. The resulting chemical differences are then acted upon by physical processes, such as subduction and convective stirring. As a result, mantle chemistry potentially contains a lot of information about the physical processes, and any model of mantle convection must ultimately be consistent with what is known about mantle chemistry. Also, mantle chemistry may react upon mantle convection, most directly through density and buoyancy, as discussed in Chapters 10 and 14.

A great deal of information about the mantle has been obtained from measurements of the chemical and isotopic composition of rocks derived from the mantle, and this is currently a very active field of geochemistry. The mantle, like the crust, contains minor or trace concentrations of virtually every element. Comparisons of concentrations, abetted by knowledge of crystal chemistry, have allowed geochemists to deduce some important conclusions about the mantle, such as that much of the mantle seems to be a residue, after the extraction of the atmospheric gases, the ocean and the continental crust, from a material with an initial composition like that of primitive meteorites. Further, measurements of the proportions of radioactive isotopes and their daughter products yields information on time scales of processes, and sometimes of dates of particular events. Isotopic compositions have also been used to identify distinct sources in the mantle, and such measurements have made it clear that there is a level of heterogeneity in mantle chemistry, and that much of this heterogeneity is quite ancient, of the order of two billion years old.