By the beginning of the twentieth century, differentiation had emerged as the leading theory to explain the chemical and mineralogical diversity of igneous rocks. Soret diffusion, liquid immiscibility, compositional stratification of magma by gravity (Gouy-Chaperon effect), volatile transport, crystal settling and other processes had been advocated as mechanisms of differentiation, but no consensus was achieved regarding a dominant mechanism.During a career spent primarily at the Geophysical Laboratory, Washington DC, Norman Levi Bowen (1887–1956) initiated a new approach to petrology. On the basis of experimental studies of rock-forming silicates and on physicochemical principles, Bowen argued against the importance of Soret diffusion, liquid immiscibility, volatile transport and assimilation as major causes of diversity. He formulated a comprehensive theory of differentiation that emphasized the role of separation of crystals from liquid. He reasoned that rocks of the subalkaline igneous rock series, including granite, have been derived from parental basalt by crystal separation, e.g. settling, filter-pressing or armouring of crystals.Apart from its scientific merits, Bowen’s achievement rested on personal, institutional and technical factors that included his determination to dedicate virtually his entire career to solution of the problem of igneous rock diversity; his affiliation with the Geophysical Laboratory; the prior development of the quenching method and the calibration of the temperature scale to very high temperatures; and the influence of Arthur L. Day.