Quite often in the development of an hypothesis there arises an impasse. Techniques used to overcome the difficulty include new assumptions, ad hoc auxiliary hypotheses, procrustean stretching, deux ex machina, etc., or a retreat to a previous stage and reconsideration of the choices that were made. The uncovering of paradox, fallacy, error or failed prediction is often the motivation to retreat or reconsider but the temptation is strong to plunge ahead. The retreat often allows one to develop an even more general and simple view (Occam’s Razor) that not only solves the immediate problem but solves what were thought to be unrelated problems. Newton’s theory of gravitation is the best known example. Newton’s theory explains the motions of planets and apples, glaciers and rivers, tectonic plates and erosion of mountains, and the accretion of planets.

The theory of plate tectonics replaced the ideas of continental fixity, permanence of the ocean basins and Earth expansion and contraction because it provided a simpler explanation of geological and geophysical observations. Although the theory is remarkably successful and has great explanatory and predictive power it seems to fail in regions of distributed continental deformation, large igneous provinces and island chains. Separate hypotheses have been advanced to address these concerns. Part of the problem is the adjective rigid which has been attached to plate tectonics, fixed which has been applied to oceanic volcanic islands and the underlying mantle, and hotspot which has been applied to various volcanoes and volcanic chains. Plate tectonics is a much more powerful concept than generally acknowledged and, as such, should be pushed hard before being amended or before auxiliary ad hoc mechanisms are introduced to explain such phenomena as island chains, swells, continental breakup and midplate phenomena.

Recent discoveries in a variety of fields are converging on a simple model of geodynamics that is inconsistent with current widely held views. These developments include noble gas measurements, mantle tomography, convection simulations, statistics, quantum mechanical equations of state, age dating, paleomagnetism, petrology and techniques to infer temperatures of the mantle. Recognition that density variations as small as 1%, which are unavoidable in the accretion and differentiation of the Earth, can irreversibly stratify the mantle is one such development.

These studies have confirmed that hotspots are not particularly hot, are not fixed, and volcanic chains are not very parallel or always time progressive. Statistical studies have suggested that hotspot magmas reflect sampling processes rather than distinctive large isolated reservoirs. Application of the central limit theorem removes many of the paradoxes associated with mantle geochemical models, if one considers chemical components rather than large geochemical reservoirs that are identified with seismic subdivisions of the mantle. It is now recognized that water content is as important as temperature in controlling rheology and melting, and that chemical variations are required to explain features in the top and bottom boundary layers of the mantle. Convection simulations have not been particularly successful in explaining plate tectonics, the numbers and sizes of plates and the initiation or one-sidedness of subduction. Heat flow, lithospheric and uplift/subsidence studies of hotspot regions do not support a thermal or reheating explanation for large igneous provinces and volcanic chains. Mantle tomography shows that the spectral, spatial, correlation and amplitude characteristics of mantle heterogeneity divide the mantle into three zones; the upper mantle, in the original Bullen sense (Regions B and C), the mesosphere (1000 km to 2000 km), and the abyss (2000 km to CMB). The deepest mantle has a few long wavelength structures, consistent with low Rayleigh number, high Prandl number convection, isolated from the surface and the upper mantle. Effects of pressure, and chemical stratification are required in order to understand these new results. Boussinesq and heated-from-below convection simulations appear to be of limited validity in geodynamics. The effects of pressure on the coefficient of thermal expansion make it easy to chemically stratify the mantle in an irreversible way. The tomography of the upper mantle correlates with present and past plate tectonics and is behaving as expected for a high Rayleigh number fluid cooled from above, and driven by the plates and lithospheric architecture.

Although the mantle is definitly convecting and provides the energy for plate tectonics, the connection between convection and plate tectonics is not completely clear. The mantle does not seem to be acting as a template, rather the reverse seems to be true. Contained fluids heated from below spontaneously organize into convection cells when sufficiently far from conductive equilibrium. Fluids can also be organized by surface tension and other forces at the top. Plate tectonics was once regarded as passive motion of plates on top of mantle convection cells but it now appears that continents and plate tectonics organize the flow in the mantle. The flow is driven by instability of the cold surface layer and near-surface lateral temperature gradients. Plate tectonics may be a self-driven far-from-equilibrium system that organizes itself by dissipation in and between the plates. In this case the mantle is a passive provider of energy and material. The effect of pressure suppresses the role of the lower thermal boundary layer. I suggest that the state of stress in the lithosphere defines the plates, plate boundaries and locations of midplate volcanism, and that fluctuations in stress are responsible for global plate reorganizations and evolution of volcanic chains. Stress controls the orientations and activity of volcanic chains since the lithosphere is an efficient stress guide. The state of stress in the lithosphere is probably more important than the temperature of the mantle in localizing volcanism, although the normal variations of temperature in the mantle influence the topography and stress of the plate. Stress, in addition to water and temperature, also controls the strength of the lithosphere and the locations of incipient and mature plate boundaries. Volcanic chains should be regarded as stress gauges and not as indicators of absolute plate motions. Changes in the orientation and magmatic activity of volcanic chains (e.g. the Hawaiian and Emperor chains) cannot be due to abrupt changes in plate motions but can reflect changes in stress. The concept of absolute fixity for hotspot volcanoes has diverted attention away from the true source of the phenomena, just as concepts of ether, geocentric, phlogiston, impetus, permanence and immutability held back the natural and physical sciences for millennia. The word hotspot itself implies characteristics that are not observed (heat flow, magma temperatures, localization).

A model is developed in which the plates are a self-organized far-from-equilibrium system which provide most of the bouyancy, dissipation, and driving forces of geodynamics. The mantle is the source of energy and material and is organized by the plate-continent-slab system. The deep mantle is isolated from the upper mantle and the plates by its density and viscosity. It communicates via Fourier’s and Newton’s laws and contributes to the geoid, dynamic topography and lithospheric stress but it is convecting as a sluggish low Rayleigh number fluid with little heat input, most of the radioactivity being in the crust and upper mantle. The spectral and spatial characteristics of the lowermost 1000 km of the mantle are distinct from the overlying layers. Flow in the upper mantle (Bullen’s Regions B and C) is driven by plate tectonics and lateral temperature gradients associated with lithospheric architecture, and is closed above 1000 km. Most upwellings are broad and passive, or associated with lateral boundaries or tensile stress (dikes) in the lithosphere. Deep boundary layer instabilities play a minor role in this model, at least for plate tectonics and magmatism.