The Earth’s crust is divided into mobile plates. Cartographers mapping the outlines of Africa and South America had speculated as early as the 16th century that these continents might once have been joined. In the 20th century geologists began to investigate their relationship more deeply, pre-eminently Alfred Wegener. In 1912 he pointed out that the two continents had similar fossils and stratigraphic patterns, and the occurrence of ancient low-latitude tillites (glacial deposits) in the southern hemisphere would be less puzzling if South America and Africa had later moved both apart and northwards. The biggest problem was how to account for the growth of solid basalt crust between the continents. A breakthrough came in the 1960s. Assimilating remotely-sensed information on the ocean crust, geologists realised that the Atlantic floor was expanding. To compensate, slabs of ocean crust were elsewhere descending into the bowels of the Earth like escalators in London’s underground. It was time for geology to experience its own Copernican revolution. Just as the Earth was not stationary in space, so the arrangement of its continents and oceans was not stationary in time.
Plate tectonics explains many things. Volcanoes occur where an oceanic plate plunges beneath a continental plate (diagram top left). The water entrained in the subducting plate gets squeezed out, heats up and rises into the lithosphere, reducing its melting point and creating pockets of magma. Some of the magma erupts to the surface. Along the margins of subducting plates, mountain ranges such as the Andes and the Zagros-Taurus mountains (including Mount Ararat) consist entirely of volcanoes. Other mountain ranges – from relatively young ones such as the Alps and Himalayas to ancient ones such as the heavily eroded Appalachians – are the result of collisions between continents (diagram top right). Such collision zones are also the places where earthquakes occur, as the strain of two adjacent plates grinding in opposite directions is suddenly released.
At the mid-ocean ridges new crust forms. Magma rises to the surface, spreads out either side of the ridge axis and solidifies, making the rock youngest next to the spreading centre, progressively older further out. With a length of over 65,000 kilometres, this submarine ridge system is the longest and volcanically most active mountain chain on the planet, each year generating over 20 cubic kilometres of new crust. A corresponding volume is subducted under the continents. Although this sounds like a lot, the average spreading rate in each direction is only around 3 cm per year, and volcanism is spasmodic.
Away from the ridges, as the crust gets older, an ever greater thickness of carbonate ooze overlies the basalt crust, much of it consisting of microscopic fossils. Where the ooze is accessible to drilling, the deepest fossils can be used to date the crust, showing that even the oldest parts are relatively young. In contrast to the continents, whose basements go back to the Archaean, the oceans are no older than the Jurassic (blue in the colour-coded image). Only a small part of the eastern Mediterranean is older.
This is why the mutually facing coastlines of the Africa and South America seem to correspond. At the beginning of the Jurassic the supercontinent ‘Pangaea’ rifted apart, dense basaltic magma erupted within the gash and as water poured in, the widening seaway gradually became the Atlantic Ocean. Since then, the entire ocean crust – 70% of the Earth’s surface – has been recycled back into the mantle and replaced by new material from the mid-ocean ridges. Like human skin, the crust is continually being renewed.
When erupting lava cools, the orientation of the Earth’s magnetic field becomes locked in its iron minerals. In ancient rocks this imprint can be compared with the magnetic field at the present location to determine how far north or south a continent has shifted. Similar rock and fossil sequences can also show how distinct continents were once joined together. Techniques such as these demonstrate that during the Permian and Triassic, immediately before the Jurassic, South America, Africa, India, Madagascar, Antarctica and Australia were all parts of a supercontinent.
During the Permo-Triassic, the plate-tectonic system was less active. While vigorous subduction may still have been going on between ocean plates, the land was virtually immobile.
Such deductions are consistent with the kind of marine limestones produced then. In times of low plate tectonic activity, as at present, the calcium and carbonate ions in seawater tend to precipitate (directly or via the secretions of shell-producing organisms) in the form aragonite. When plate tectonic activity is high and submarine volcanism releases higher amounts of ions into the oceans, the ions combine to precipitate another form, calcite. Permo-Triassic limestones are predominantly of the aragonite kind and therefore suggest low hydrothermal flux into the oceans (Steuber & Veizer 2003).
In the periods before the Permian the Earth looked even less familiar (see the series of Palaeozoic snapshots above). Britain, for example, was not a distinct entity at all: Scotland in the early Palaeozoic lay near the equator and was separated from England and Wales, in the southern hemisphere, by an entire ocean. The Scottish highlands were once part of North America – pushed up as the once separate continents collided – whereas the English lowlands lay thousands of miles away to the south. The suture line running NE-SW through Scotland, the Isle of Man and Ireland can still be traced.
The map of the world changed enormously in the course of the Palaeozoic, and since the map underwent even more revolutionary change in the preceding Proterozoic and Archaean, it is clear that the entire Earth’s oceanic crust must have been replaced several times. By contrast, cratons – the cold, old nuclei of continents – have hardly been consumed at all, because they are more buoyant than the ocean floors. Overall, they have been accreting material and thereby growing over time.
Plate tectonics is the driving force behind most geological change. And gravity is the force behind plate tectonics. Being more elevated than the rest of the plate, the ridge exerts a pushing force against the continent, and the plate, being denser, tends to sink. The pull of the sinking slab is by far the stronger of the two forces. As the oceanic crust moves away from the spreading centre and cools, convection currents also play a role 
(though the cells are smaller than implied in the diagram). Hot convecting material rises, softens the crust, thins it, melts and erupts. Other convecting material circulates away from the ridge and draws the plate with it.
The mantle is hot because with increasing pressure temperature increases. The decay of radioactive elements in the lower mantle also produces heat. Present rates of radioactivity are insufficient to account for the heat given off at the surface. Only around 20% of the heat flux seems to originate from radioactive elements – the rest appears to represent long-term cooling. Extrapolated back over billions of years towards higher temperatures, the cooling ends in ‘thermal catastrophe’, a ‘first-order paradox in reconstructing Earth’s cooling history, since mantle temperature is believed to have been lower than ~1800° C even in the Archean’ (Korenaga 2006). Present rates of radioactivity also appear insufficient to drive convection currents. The mantle is (now) solid, and convection calculations do not exhibit plate tectonic behaviour unless it is imposed by the modeller (Tackley 2000).
Since seafloor spreading rates affect the rate of all large-scale geological processes, an important question is whether rates of spreading have always been as slow as today’s 1–15 cm per year. There have been various studies to investigate this. Generally, they conclude that rates have been much the same as now, at least during the Phanerozoic. However, the reasoning is somewhat circular, for spreading rates depend on mantle viscosity, which depends on heat generation, which depends on the rate of radioactive decay, and the rate of radioactive decay – driving the clock by which geological time is measured – is assumed to have been constant: had it been faster, spreading rates would have been faster. Fundamental investigation of past spreading rates is precluded.
Spreading rates today vary from ultra-slow (less than 1 cm per year) to ‘ultra-fast’ (12–15 cm per year). Generally, rates are faster where the opposite margin is ‘active’ and the sinking of the slab imparts momentum. Although all such speeds are tiny, the different rates reflect significant differences in magma supply. With ultra-slow spreading, supply is intermittent and shut off most of the time; with ultra-fast spreading it is more continuous, though still episodic. Magma supply has a marked effect on ridge architecture, rock chemistry, and on how faults are segmented, so we can determine whether rates have always been in the current range by investigating whether older ocean floor has features characteristic of slow spreading rates. The evidence suggests that, in the past, rates were consistently ‘ultra-fast’, and thus at least 12–15 cm per year. Perhaps the most obvious evidence is the increasing smoothness of ocean floor away from the ridges. Slow-spreading centres are characterised by deep, fault-bounded valleys, fast-spreading centres by elevated and much more regular topography.
The evidence that spreading rates were faster suggests that mantle convection in the past was more vigorous, and rates of radioactive decay higher. This would explain why the Earth is radiating more heat than can be explained on the basis of present decay rates. The excess heat from higher rates in the past is still feeding through. Because the Earth is much younger than the age based on constant rates of decay, ‘thermal catastrophe’ is not implied.
Another of geology’s great mysteries is that rocks of all kinds are completely missing from the first ‘600 million years’ of Earth history. The only traces of rock to have survived from that time are a few detrital crystals of zirconium silicate. Otherwise that former ‘Hadean’ world (4.6–4.0 Ga) has gone.
The world that disappeared was the antediluvian world. Rates of radioactive decay were then much faster, causing the temperature and pressure of the Earth’s interior to rise. For a time, a subterranean 
ocean isolated the continent, but eventually the pillars sustaining the continent collapsed. In a brief onslaught known as the Late Heavy Bombardment, asteroid impacts completed the demolition, the most powerful of them creating lava-filled cratons that became the nuclei of multiple new continents. Meanwhile the sinking continent of the old world surrounding these cratons flooded, thereby becoming the floor of a world ocean, and cracks in the floor became the spreading centres for new ocean crust. Thus was initiated the new regime of plate tectonics. Sliding under the buoyant margins of the cratons, the Earth’s primeval crust was cycled back into the mantle, and the antediluvian world became, quite literally, the Hadean underworld.
Diapirism effected an enormous transfer of radioactive isotopes, and of contained heat, from the lower to the upper crust within a short period of time in each region, and greatly increased the petrologic stratification of the crust. This transfer allowed the lower crust and subjacent mantle to cool markedly below prior temperatures and thus stabilized cratons and stiffened lithosphere.
W. B. Hamilton, Precambrian Research 91:160-61 (1998).
Cratons formed by vertical tectonics (diapirism), riding on top of hot, mobile mantle, constantly on the move. Towards the end of the Archaean, cratons began to collide and amalgamate. Only then did horizontal (plate) tectonics play a major role in continental growth, as volcanic arcs accreted onto craton margins and the crust above subduction zones thickened.
There is a distinction to be made between the generation of continental crust and the generation of oceanic crust over time. While the formation of continental crust began to tail off after the Proterozoic, ocean crust continued to form at high rates. Large rises and falls in Palaeozoic sealevel can be correlated to high/low plate tectonic activity and imply much faster rates of ocean crust generation than in the present. Today the ocean floor is cooler and less buoyant, and sealevel at a historical low.
The planet’s terrestrial crust has been replaced only once, and this, not gradually over the whole of history but near the beginning, when the antediluvian world perished suddenly and violently. The still continuing process of ocean-crust replacement is a revelation of how it is possible for entire slabs of crust to disappear into the mantle and be completely destroyed.
