The idea that parts of the earth have moved slowly relative to each other over distances comparable to the size of the globe belongs mostly to the twentieth century. There were some earlier suggestions of catastrophic global displacements, but it was in the twentieth century that large slow displacements of the continents were proposed and systematically advocated, and eventually their existence was decisively established.

Historically, the idea of mantle convection is closely entwined with the ideas of continental drift and plate tectonics. The idea that the earth's interior is mobile can be traced back at least to the middle of the nineteenth century, but it became the focus of sharp debate early in the twentieth century with the acquisition of seismological evidence that below the crust is a solid, rocky mantle extending about halfway to the centre of the earth. To many this seemed to make continental drift impossible.

After Holmes proposed mantle convection as a possible mechanism of continental drift around 1930, most thinking about continental drift and the emerging plate tectonics was strongly conditioned by expectations of how such convection would work. One can argue that this interaction of ideas actually held back the recognition of the pattern of movements on the earth's surface (Section 3.8).

Others have told the story of the theory of continental drift and its ultimate evolution into the theory of plate tectonics [1–6]. There are several reasons for recounting it here, rather than simply proceeding to a description of the plates and how they work. Mainly, I want to include the complementary development of ideas about the mobility of the mantle, which I think played a larger role in the story than has been appreciated.