Yes, like a bar magnet, in the sense that an ocean is like a bowl of-water.

Magnetic materials found in ancient rocks show that every so often, about once every half a million years, but with no sign of regular­ity, the Earth's magnetic field flips polarity, reversing magnetic north and south. We're not sure exactly why, but mathematical models suggest that the magnetic field can exist in these two orien­tations, with neither of them being totally stable. So whichever one it's in, it eventually loses stability and flips to the other one. The flips are rapid, taking about 5,000 years; the periods between flips are about a hundred times as long.

Most of the other planets have magnetic fields, and these can be even more complicated and difficult to explain than that of the Earth. We've still got a lot to learn about planetary magnetism.

One of the most dramatic features of our planet was discovered in 1912 but wasn't accepted by science until the 1960s, and some of the most compelling evidence was left by those flips in the Earth's magnetism. This is the notion that the continents are not fixed in place, but wander slowly over the surface of the planet. According to Alfred Wegener, the German who first publicized the idea, all of today's separate continents were originally part of a single super-continent, which he named Pangea ('All-Earth'). Pangea existed about 300 million years ago.

Wegener surely wasn't the first person to speculate along such lines, because he got the idea, in part, at least, from the curious similarity between the shapes of the coasts of Africa and South America. On a map the resemblance is striking. That wasn't Wegener's only source of inspiration, however. He wasn't a geolo­gist; he was a meteorologist, specializing in ancient climates. Why, he wondered, do we nowadays find rocks in regions with cold cli­mates that were clearly laid down in regions with warm climates? And why, for that matter, do we nowadays find rocks in regions with warm climates that were clearly laid down in regions with cold cli­mates? For example, remains of ancient glaciers 420 million years old can still be seen in the Sahara Desert, and fossil ferns are found in Antarctica. Pretty much everyone else thought that the climate must have changed: Wegener became convinced that the climate had stayed much the same, give or take the odd ice age, and the con­tinents had shifted. Perhaps they'd been driven apart by convection in the mantle, he wasn't sure.

This was considered a crazy idea: it wasn't suggested by a geol­ogist, and it ignored all sorts of inconvenient evidence, and the alleged fit between South American and Africa wasn't all that good anyway, and-to top it all, there was no conceivable mechanism for carting continents around. Certainly not convection, which was too weak. Great A'Tuin may lug a planet around on its back, but that's fantasy: in the real world, there seemed to be no conceivable way for it to happen.

We use the word 'conceivable' because a number of very bright and very reputable scientists were busily making one of the subject's worst, and commonest, errors. They were confusing 'I can't see a way for this to happen' with There is no way for this to hap­pen.' One of them, it pains one of us to admit, was a mathematician, and a brilliant one, but when his calculations told him that the Earth's mantle couldn't support forces strong enough to move con­tinents, it didn't occur to him that the theories on which those calculations were based might be wrong. His name, was Sir Harold Jeffreys, and he really should have been more imaginative, because it wasn't just the shapes of the land on either side of the Atlantic that fitted. The geology fitted too, and so did the fossil record. There is, for example, a fossil beast called Mesosaurus. It lived 270 million years ago, and is found only in South America and Africa. It could­n't have swum the Atlantic, but it could have evolved on Pangea and spread to both continents before they drifted apart.

In the 1960s, however, Wegener's ideas became orthodox and the theory of 'continental drift' became established, though the ancient supercontinent was renamed Gondwanaland because it dif­fered in some ways from Wegener's conception of Pangea. At a meeting of leading geologists, a Ponder Stibbons-like young man named Edward Bullard and two colleagues enlisted the aid of a new piece of kit called a computer They instructed the machine to find the best fit between Africa and South America, and North America, and Europe, allowing for a bit of breakage but not too much. Instead of using today's coastline, which was never a very sensible idea but made it possible to claim that the fit wasn't actually that good, they used the contour corresponding to a depth of 3200 feet (1000 m) underwater, whose shape is less likely to have been changed by ero­sion. The fit was good, and the geology across the join matched amazingly well. And even though the people at the conference came out just as divided in their opinions as they'd been when they went in, somehow continental drift had become the consensus.

Today we have much more evidence, and a fair idea of the mech­anism. Down the middle of the Atlantic Ocean, and elsewhere in other oceans, there runs a ridge, roughly north-south and about midway between South America and Africa. Volcanic material is welling up along that ridge, and spreading sideways. It's been spreading for 200 million years, and it's still doing it today: we can even send deep-sea submarines down there to watch. It's not spreading at speeds humans can see, America moves about three-quarters of an inch (2 cm) further away from Africa every year, about the same rate that your fingernails grow, but today's instru­ments can easily measure such a change.

The most striking evidence for continental drift is magnetic: the rocks on either side bear a curious pattern of magnetic stripes, reversing polarity from north to south and back again, and that pat­tern is symmetric on either side of the ridge, making it clear that the stripes were frozen in place as the rocks cooled in the Earth's magnetic field. Whenever the Earth's dynamo flipped polarity, as it does from time to time, the rock immediately adjacent to the ridge-line, on either side, got the same new polarity. As the rocks then spread apart, they took the same patterns of stripes with them.

The surface of the Earth is not a solid sphere. Instead, the con­tinents and the ocean-beds float on top of large, essentially solid plates, and those plates can be driven apart by upwelling magma. (Oh, but mostly by convection in the mantle. Jeffreys didn't know what we now know about how the mantle moves.) There are about a dozen plates, ranging from 600 miles (1000 km) across to 6000 miles (10,000 km), and they twist and turn. Where plate boundaries rub against each other, sticking and slipping and sticking and slip­ping, you get a lot of earthquakes and volcanoes. Especially along the 'Pacific rim', the edge of the Pacific Ocean up along the west coast of Chile, central America, the USA, along down past Japan, and back round New Zealand, which is all the edge of a single gigantic plate. Where plate boundaries collide you get mountain ranges: one plate burrows under the other, lifting it up and crush­ing and folding its edges. India was once not part of the main Asian continent at all, but came crashing into it, creating the world's high­est mountain range, the Himalayas. India hasn't fully stopped even now, and the Himalayas are still being pushed up by the force of the impact.