Plate tectonics: the moving ground

The ground is the thing we use as our definition of "still". Yet it moves at about the speed your fingernails grow, and over a hundred million years that is enough to open an ocean. In The universe in one year: a cosmic calendar the supercontinent Pangaea was assembled on 23 December and began to split on the 26th. This post is about the machinery behind that.

On 6 January 1912 a German meteorologist and polar explorer, Alfred Wegener, told a geological society in Frankfurt that the continents had once been joined and had drifted apart. He expanded this into a book, first published in 1915. His evidence was good:

The geologists were not convinced, for good reasons as well as bad ones. Wegener had no workable mechanism. He suggested that centrifugal forces and tides pushed continents through the ocean floor, and the geophysicist Harold Jeffreys showed that those forces were far too weak. His numbers also failed: he thought Greenland was moving metres per year, while the true figure is about two centimetres. The standard alternative was sunken "land bridges", and at a famous 1926 symposium in New York most speakers attacked drift. Part of it was snobbery, too: a meteorologist was telling geologists about the Earth.

Wegener died in November 1930 on the Greenland ice sheet, aged 50, on his last expedition. He was right about the fact and wrong about the cause, and for the next thirty years science mostly waited for the cause.

The answer came from the bottom of the ocean, which military sonar and magnetometers had begun to map during and after the Second World War. Marie Tharp, working with Bruce Heezen, drew the Atlantic floor and saw a deep rift valley running down the middle of the Mid-Atlantic Ridge. Heezen at first dismissed this as "girl talk"; later data convinced him. The ridge is one piece of a chain about 65,000 km long.

In 1960-62 Harry Hess proposed that the ocean floor is made at ridges and swallowed at trenches, like a conveyor belt. Robert Dietz named it sea-floor spreading. Then came the clinching evidence, the magnetic stripes.

The mechanism, step by step:

Fred Vine and Drummond Matthews published this idea in 1963; Lawrence Morley had the same idea independently, and his paper was reportedly rejected as cocktail-party talk. The reversal dates were calibrated on lava flows on land, so the stripe widths could be turned into speeds. In 1968 the drilling ship Glomar Challenger confirmed it: sediment sitting on the basalt gets older with distance from the ridge, and almost nowhere is ocean floor older than about 180 million years. Continents hold rocks of about 4 billion years. The oceans are young because they keep being recycled.

Between 1965 and 1968 Tuzo Wilson, Dan McKenzie, Jason Morgan, Xavier Le Pichon and others had combined the pieces into plate tectonics: the outer shell, the lithosphere, is about 100 km thick, broken into seven big plates and many small ones, riding on hotter, softer rock below.

Divergent boundaries create crust. Iceland is the Mid-Atlantic Ridge above water, and at Thingvellir you can walk in the rift. In September 2005 a fissure about 60 km long opened in Afar, Ethiopia, within roughly ten days, a glimpse of how an ocean basin begins.

Convergent boundaries destroy crust, or pile it up. Where an ocean plate meets a continent, the denser ocean plate dives. Where two continents meet, neither sinks, so they crumple. India has been pushing into Asia for about 50 million years, and the top of Mount Everest is marine limestone with fossils, former sea floor lifted almost 9 km.

Transform boundaries slide sideways. The San Andreas Fault in California is the famous one; the Alpine Fault in New Zealand is another.

The sinking slab is cold and heavy, so it keeps pulling the rest of the plate behind it. Most geophysicists regard this slab pull as the biggest force moving plates, with the push from the ridge and the drag of mantle flow as smaller helpers. The whole system is powered by Earth's internal heat.

Two things then follow from the picture. Earthquakes: the plate contact (the megathrust) is huge, and the slab keeps breaking as it sinks, to depths of about 700 km. Volcanoes: at about 100 km depth the slab squeezes out water, the water lowers the melting point of the hot rock above, and the melt rises to build a chain of volcanoes. That is why a "Ring of Fire" of trenches around the Pacific holds about 90% of the world's earthquakes and about three quarters of its active volcanoes. Volcanoes also form at ridges, where rising mantle melts as the pressure drops, and at "hotspots" such as Hawaii, far from any plate edge.

Plates do not slide smoothly. The fault is locked by friction, so the rock on both sides bends and stores elastic energy until it suddenly slips. Harry Fielding Reid worked this out from the 1906 San Francisco earthquake, which offset the ground by up to about 6 metres.

Check the arithmetic: 3.5 cm a year for 150 years is 5.25 m, the same order as 1906. Magnitude is logarithmic, so each whole step is about 32 times more energy and two steps are about 1,000 times. The biggest quakes happen at subduction zones: Chile 1960 (magnitude 9.5, the largest recorded), Sumatra 2004 and Tohoku 2011 (both about 9.1). We can map where quakes will happen very well and when they will happen hardly at all.

A fingernail grows about 3 mm a month, so roughly 3.6 cm a year, and that is the same ballpark as the Atlantic opening or the San Andreas slipping. Multiply it out: 2.3 cm a year for 130 million years is about 3,000 km, which is about the narrowest gap between Brazil and West Africa. Hawaii gives another check. Kauai is about 5 million years old and roughly 500 km from the Big Island, which is 10 cm a year, close to the Pacific plate's estimated speed of 7 to 10.

Today GPS stations measure plate motion directly, and the results agree with the geological averages over millions of years.

Run the speeds backwards and the continents reassemble, as Wegener guessed. He named the supercontinent Pangaea, Greek for "all land". Reconstructions have it fully assembled about 335 million years ago and starting to split about 200 million years ago, surrounded by one world ocean, Panthalassa. The central Atlantic opened first, the south Atlantic about 130 million years ago, and India raced north at 15 to 20 cm a year until it hit Asia. Pangaea was not the first supercontinent: Rodinia, about a billion years ago, was an earlier one, and the opening and closing of oceans is often called the Wilson cycle after Tuzo Wilson.

Established: the continents move; ocean floor is made at ridges and destroyed at trenches; plate boundaries explain where earthquakes and most volcanoes occur.

Still discussed: the exact balance of forces driving plates, how mantle plumes like Hawaii's work, and when plate tectonics began, with estimates ranging from around 4 billion to under 1 billion years ago. Venus and Mars apparently have no active plate tectonics, and why Earth does is open, perhaps linked to its water.

Hypothesis: the cycle will repeat. Models place a new supercontinent 200 to 250 million years ahead, with several candidates (Pangaea Proxima, Novopangea, Amasia). Nobody knows which, if any, is right.

Plate tectonics also touches the story of life, since some theories about how life started rely on vents along ridges: see Life's origin: what we know and what we don't. And Wegener's story is a good companion to Five ideas that built modern science: an idea is not accepted when it is first right, but when someone finds the mechanism.