Right for Fifty Years and Not Believed

Alfred Wegener published the case for continental drift in 1912, and it was good.

The coastlines of South America and Africa fit together, and so do the continental shelves, which is a better fit than the visible coasts. Identical fossil species appear on both sides of the Atlantic and nowhere else: Mesosaurus, a freshwater reptile that could not have crossed an ocean, in Brazil and in southern Africa. Rock formations and mountain chains match across the gap. Glacial deposits appear in India and Africa, in places now tropical, with scratch marks pointing in directions that only make sense if the continents were arranged differently.

The evidence was substantial and it was dismissed for half a century, and the reason was not stupidity.

Wegener proposed that continents plough through the ocean floor, driven by tidal forces and centrifugal effects from the Earth's rotation. Geophysicists calculated this and showed it was impossible: the forces are far too weak, and continental rock cannot push through oceanic rock like a ship through water.

They were right about the mechanism, and they rejected the theory along with it.

That is a general pattern worth having. Strong evidence for a pattern is frequently dismissed until somebody supplies a process, because without a mechanism, a correlation is a puzzle rather than an explanation, and the objection "but how?" is a legitimate one rather than mere conservatism. Continental drift was correct and unbelievable at the same time, and it took a completely unexpected line of evidence to change that.

That evidence came from the sea floor, mapped seriously for the first time during and after the Second World War.

The ocean floor turned out not to be an ancient featureless plain. It has a continuous mountain range running down the middle of the Atlantic and through the other oceans, with a rift along its crest. And magnetic surveys found something very odd: bands of alternating magnetic polarity running parallel to the ridge, and (this is the crucial part) mirrored symmetrically on both sides of it.

The Earth's magnetic field reverses at irregular intervals, which was known from lava flows on land. Vine, Matthews and Morley put the pieces together in 1963. Molten rock rises at the ridge, cools, and locks in the direction of the field at that moment. As more rises, the older crust is carried outward in both directions. The sea floor is a tape recorder, and the stripes are the recording.

Sea floor spreading supplied the mechanism. The continents do not plough through the ocean floor; they ride on plates, and the ocean floor is being manufactured at the ridges.

The obvious objection, that the Earth must be expanding if new crust is constantly created, is answered by subduction. Where an oceanic plate meets another, the older, colder, denser one sinks beneath and is consumed. Creation at ridges and destruction at trenches roughly balance.

This also explains something nobody had thought to ask. No oceanic crust anywhere is older than about 200 million years, while continental rocks reach nearly four billion. The ocean floor is continuously recycled; the continents float on top and persist.

The theory was accepted within about a decade, which is fast, and one observation did more than any other to make it compelling.

Plot every recorded earthquake on a world map. The dots do not scatter. They trace narrow lines that outline the plates: down the mid-Atlantic, around the Pacific rim, through the Mediterranean and the Himalaya. Volcanoes do the same.

Tectonics was proposed to explain the positions of continents and the shapes of coastlines. It turned out to predict where earthquakes happen, which it had not been designed to do.

That is the strongest evidence a theory can offer: not that it accounts for what it was built to account for, but that it gets right something nobody was aiming at.

What drives the plates is still being worked on, and the current answer is mostly slab pull. Old oceanic crust is cold and dense, and once it starts sinking at a subduction zone it drags the rest of the plate behind it. Ridge push from elevated new crust and drag from the convecting mantle contribute. The evidence pointing at slab pull is that plates with long subducting margins move fastest, which is the kind of quiet quantitative check that settles questions long after the dramatic phase is over.