Nicola Jones

New Scientist, 177 (2385), p 32. 8th March 2003

WELCOME to Hawaii, a beautiful chain of volcanic islands that stretches for thousands of kilometres across the Pacific. According to most geologists, Hawaii is a classic example of a "hot-spot trail", created by a giant plume of magma that rises from the core of the Earth to the surface. As the Pacific plate passes over the plume, the fountain of hot rock leaves behind a trail of burnt spots as it scorches the surface. Such mantle plumes are deeply entrenched in the geological "standard model", going hand-in-hand with plate tectonics.

But a few renegade geologists have been saying for years that plume theory is fundamentally flawed, and have come up with alternative explanations for volcanoes and island chains such as Hawaii. And now others are starting to agree.

Researchers at last year's American Geophysical Union meeting in December presented swathes of such evidence, and more are expected to join ranks with these dissenters at a meeting in Iceland this August. Most aren't convinced that mantle plumes should be dumped entirely. But they're willing to admit that the geological community is standing on the brink of a radical shift in thinking that could completely change our ideas about the inner workings of the Earth.

One of the renegades is Don Anderson. Over the past decade, his work at the California Institute of Technology has made him famous for his belief that mantle plumes simply don't exist. Forget fountains from the centre of the Earth, he says: volcano chains are due to cracks or rents in tectonic plates. After all, plates stretching 10,000 kilometres are unlikely to remain perfectly unbroken. Break the skin and it seems reasonable that molten rock would flood up to fill the gap. Bingo, you've got a volcano - without a mantle plume.

Most volcanoes on the planet form along the edges of tectonic plates, where slabs of ocean floor squeeze under continents and cause the rock to melt and well upwards through faults in the crust. But some volcanoes, such as those in Hawaii, seem to appear from nowhere, right in the middle of a plate. Not only that, but the Hawaiian islands get progressively older towards the north-west, leaving a trail of dead underwater volcanoes. Other strings of ancient volcanoes in the Pacific, such as the Tuamotu Archipelago and the Austral Islands, are lined up almost parallel to the Hawaiian chain and they, along with dozens of other volcanic features around the world, are a long way from the nearest plate boundary. Back in the 1960s, anomalies like these got geophysicist Tuzo Wilson from the University of Toronto thinking. Maybe, Wilson supposed, these volcanoes are caused by sources of intense heat lurking beneath the surface - "hot spots" that remain still while the plates wander about above them.

What could these heat sources be? In the early 1970s Jason Morgan, a physicist at Princeton University in New Jersey, suggested they were fountains of magma rising vertically from the edge of the Earth's core, 2900 kilometres below. It certainly made sense. The plumes would help set up convection currents that would drive tectonic plates around and play an important role in transporting heat from the core of the planet. Their existence could explain why these hot spots seem to stay still relative to each other, and why the magma at these volcanoes appears to have a mix of minerals and gases unlike those from volcanoes at the edges of plates. Mantle plumes could even provide us with a direct glimpse into the centre of the planet. Morgan proposed that plumes existed at around 20 spots, including Hawaii, Yellowstone and Iceland. The idea was seized upon with great enthusiasm and has been entrenched in textbooks ever since. Today, some researchers say there are hundreds of volcanic hot spots caused by plumes.

Yet by the 1990s experts had found plenty of good reasons to question the idea. In 1996, Gillian Foulger, a seismologist at the University of Durham in the north of England, set out for Iceland with a team of researchers to look for plumes. Although Iceland is on the mid-Atlantic ridge, where American and European plates are spreading apart and new ocean floor is being born, the fact that volcanic activity is concentrated in a relatively small area around Iceland suggested to many geologists that a plume must be responsible. So Foulger and her team planned to use a network of seismometers laid across Iceland to measure the seismic waves that bounce through the Earth following an earthquake. Hot, molten rock has a lower density than cool rock, and this reduces the speed at which seismic waves travel. Time the arrival of seismic waves at different points and you build up a map of the rock beneath.

However, the data they collected seemed to show that rather than a long narrow plume of magma coming from deep below Iceland, there was actually a broad reservoir of molten rock less than 400 kilometres down. That left Foulger puzzled - until she realised it would be easy to find an explanation if she dropped the plume theory altogether.

Iceland sits smack at a geological crossroads where the mid-Atlantic ridge crosses an ancient fault line - a "suture" where Europe and North America collided 400 million years ago. This geometry and the presence of volcanic activity on Iceland has to be ignored as a coincidence if you believe a classic mantle plume feeds the island. But Foulger realised that remnants of the crust left behind in the mantle at this fault line have a lower melting point than the surrounding mantle. Where this ancient crustal rock crosses the ridge it should melt more than the rock around it, creating the shallow reservoir of magma that feeds Iceland's volcanoes. She didn't need a mantle plume.

But when Foulger first presented her ideas she met outright hostility and had difficulty getting them published. So when she talked to Anderson in 1999 and realised there were other geologists out there with similar doubts about the existence of mantle plumes, it came as a huge relief.

The problem with mantle plume theory, says Anderson, is that it requires some crucial assumptions. Back when Morgan was developing his theory, the upper mantle was thought to be fairly uniform and not very hot. To explain hot spots in the middle of tectonic plates, geologists needed to find some mechanism to pull magma from deep below.

These days, however, geologists know the upper mantle is highly variable, both in geological make-up and temperature. Besides, the ages of some islands in chains that geologists first thought to be formed by plumes - the Canaries, for example - turn out to vary almost randomly from one island to the next. Even the theory's central tenet - that all the world's hot spots stay still relative to each other while the plates wander about - seems to be wrong. Geologists have discovered that the positions of the top 50 candidates for mantle plumes move relative to each other at the rate of a few centimetres a year, about the same speed as tectonic plates.

Even the unusual composition of magma from hot-spot volcanoes does little to back mantle plume theory. Samples of magma from Hawaii are flooded with helium-3, an isotope thought to be left over from the big bang. Researchers had assumed that such high abundances of this primordial helium could only have survived in a reservoir deep inside the Earth. This seemed to confirm that the magma came from near the core. But geochemists have grown worried by this. It turns out that helium-3 doesn't necessarily come from deep underground; it can be trapped and released by any number of rock crystals, such as olivine, a pale green mineral with a high melting point that is common in the upper mantle. And researchers have found that while the magma at some volcanic hot spots releases a flood of helium-3, others show a dearth. Anderson argues that the mantle is simply highly variable in its composition. Sometimes a volcanic hot spot in a plate coincides with high levels of the gas. Other times it doesn't. "At some point you have to say enough is enough," says Anderson. "Maybe there's a better idea." Geochemist Anders Meibom from Stanford University in California agrees. "From a geochemical point of view, the plume hypothesis is way overused. Whenever geologists see an anomalous trace, they yell 'plume!'. It's an easy way out."

Besides, Anderson has a more fundamental objection to the plume theory. He claims that the pressure at our planet's core is simply too high to allow narrow plumes to form in the first place. Like a blob of warm oil in a lava lamp, each bulge of hot magma would have to attain a certain critical mass before breaking away. Anderson's calculations show that at the temperatures and pressures down near the core, these blobs would have to be between 500 and 1,000 kilometres in length and 5,000 kilometres across - a huge chunk of the Earth's interior - before they could begin to rise. If one managed to break through the Earth's surface, it would leave a massive flood plain of lava 10 times as big as anything geologists have seen. And even if a narrow plume of magma just 10 or a 100 kilometres across were created, Anderson argues that the magma itself wouldn't be able to punch through a layer near the lower mantle about 800 kilometres down where there is a chemical change in the rock. He calculates that the pressure and temperature changes at those depths aren't large enough to make the lower mantle rocks less dense than the rock above. They physically shouldn't be able to rise.

So how do you explain the progression in age seen in chains of volcanoes such as those at Hawaii, the trait that led to the birth of mantle plume theory in the first place? The answer might be found in Yellowstone, once an archetype of a plume-fed hot spot. Although there is a chain of progressively older volcanoes marching from Yellowstone towards Idaho, a variety of seismic studies from a number of plume-hunters have shown no signs of hot magma below a depth of 200 kilometres and no disturbance in the mantle below that. "Yellowstone was supposed to be the grandaddy of all continental plumes. It's a huge volcanic centre but its status as a plume has evaporated," says Dean Presnall, a petrologist with the University of Texas at Dallas.

Anderson's idea seems to fit instead, says Foulger, now working with the Volcano Hazards Team at the US Geological Survey in Menlo Park, California. She has worked with colleagues in Yellowstone and says the region's geology could actually be the result of a crack slowly propagating in a north-east direction. Imagine someone ripping a map of the US by pulling the two bottom corners apart. If there were enough hot magma just below the crust, then the freshest volcanoes would follow the tip of that tear up the page. This could explain Yellowstone, says Anderson, and Hawaii too. Stress fractures in the Pacific are already known to form perpendicular to mid-ocean ridges such as the East Pacific Rise that runs roughly parallel to the coast of South America, about 4000 kilometres out in the Pacific. And island chains such as Hawaii and Tuamotu lie pretty much where you would expect stress fractures to form.

One of the more damning pieces of evidence against mantle plume theory is that regions of the crust above suspected mantle plumes don't actually appear to be hot - despite the fact that huge fountains of magma from the hot core should be rising directly beneath. Seth and Carol Stein, a husband-and-wife team at Northwestern University at Evanston and the University of Illinois at Chicago, have been charting heat flow from the ocean floor for decades, in part to see if they can spot these plumes. In Hawaii, however, they found the temperature below the sea floor to be much the same as everywhere else - there is no anomalous heat flow.

Some geologists argue that the heat from a plume takes tens of millions of years to leak up through the crust, and in that time the plates would have moved so there would be no extra heat trace on any single region of rock. The Steins decided to visit Iceland, where the crust is thinner and should let heat leak through about 10 times as fast. Still they saw nothing. In fact, they saw the opposite of what plume theory predicted - a slightly hotter place to the east of Iceland, where the plume supposedly hasn't yet arrived. "You just have to keep making up excuses and modifications to make plume theory work," says Foulger.

Despite the promise of stress cracks, some say they can't explain all volcanic features. Take the island of Réunion in the Indian Ocean, for example. This lies at the end of a volcanic trail stretching back towards India, which passes right over a mid-ocean ridge and spans two separate tectonic plates. It is difficult to see how a stress crack can propagate over a ridge like that and continue from one plate to its neighbour. It's like ripping one piece of paper and expecting a second piece beside it to somehow rip too.

If Anderson is right and plumes don't exist, then plate tectonics alone must explain everything, including rogue volcanoes such as Hawaii. In fact, volcanism and plate motion become one and the same. It's almost a "unified theory" for geologists, jokes Foulger. That simplifies a few things, but it raises a mass of new questions. Morgan envisioned plumes as something that would set up giant convection currents in the mantle, driving plates around like the heat of a stove churning water in a pan. Few people now think things are so simple. Instead, Anderson says that currents of cooler rock do the stirring - but forget simple convection currents: there's a muddle of circulating rock with many layers. And there are no samples of the inner Earth being spat out of the Hawaiian volcanoes as we once thought. Everything is up for grabs.

"We'll have to acknowledge we know far less about the centre of the Earth than we thought we did," says Foulger. But that's not a disaster. It's exciting.