Not all Hollywood sci-fi movies are implausible; 1998's Deep Impact even got a good review in Science magazine. More often they're like the recent release The Core, however, described by the New York Times as "monumentally dumb."

	The New York Times called The Core a "monumentally dumb" movie. The true story is a fascinating riddle, wrapped in a mystery.

In the movie, the collapse of Earth's magnetic field is inexplicably signaled by failing cardiac pacemakers, a space shuttle thrown off course (perhaps it was steering by magnetic compass?), and the wandering of the Northern Lights to lower latitudes -- never mind that without a magnetic field there would be no Northern Lights. The cause of the mayhem? It seems Earth's core has stopped spinning!

So much for fiction. In fact, for at least the last 160 million years, as known from evidence preserved in the geological record, Earth's magnetic field has often reversed polarity. The story is told by tiny magnetic domains in layered basalts on land and on the spreading ocean floors, frozen in different orientations.

Core spin isn't implicated, however. The solid inner core turns only once every 120 years or so, relative to the rest of the planet. No one knows the real reason for field reversals.

"We know more about the surface of the sun than the deep earth," says Rich Muller of the Lab's Physics Division, a professor of physics at UC Berkeley. "We can probe it by seismography and by looking at heat signatures, and we have the evidence from changes in the magnetic field. But mostly it's a mystery."

Avalanches down below

Last October Muller published what he considers to be the most plausible hypothesis yet of what's behind geomagnetic field reversals, in an article in Geophysical Research Letters titled "Avalanches at the core-mantle boundary."

Earth's inner core is solid iron, its outer core is liquid iron mixed with other components, and its mantle is dense rock.

What we do know about the core -- that there is a solid inner core of pure iron the size of the Moon, an outer liquid core rich in iron the size of Mars, and an irregular boundary between the liquid core and the bottom of the rocky mantle -- comes mostly from studying how seismic waves travel through the earth.

In addition, geophysicists have recreated the physical states and chemical compositions that could account for the seismic record both in the laboratory, notably in diamond-anvil-cell experiments like those of Raymond Jeanloz at UC Berkeley, and in computer simulations such as those by Gary Glatzmaier of UC Santa Cruz and Los Alamos.

Most scientists agree that Earth's magnetic field arises from convection currents in the liquid outer core, a good conductor of electricity. These currents constitute an amplifying, self-sustaining "geodynamo."

Convection probably starts as iron crystallizes on the surface of the inner core, about 5,000 kilometers beneath Earth's surface; lighter components like oxygen, sulfur, and silicon are excluded and rise toward the core-mantle boundary (CMB) 2,000 kilometers higher, where temperatures are a thousand degrees cooler.

Here the lighter components cool and condense as slushy sediments. Muller theorizes that tens of meters of these buoyant sediments accumulate each million years, "falling" upward onto the uneven topography at the base of the mantle. Even if the slopes of the CMB hills are shallow, like sand dunes, eventually the sediments will slip and slide.

"It's a little like an avalanche on the sea floor, where mud mixes with water, causing turbidity flow," Muller says. The turbid mixture of cool sediments and hot liquid iron causes cooled-off, denser iron to sink back toward the inner core.

The sinking iron would perturb the geodynamo's convection cells, causing frequent "excursions" of the dipolar magnetic field as measured at the surface. This normal process of sediment accumulation and slippage probably goes on all the time.

Upward-sliding "avalanches" at the core-mantle boundary probably occur frequently on many scales, but all would tend to warm the mantle and cool the liquid core. (The inner core is toward the top of this image.) Truly massive CMB avalanches could disrupt the geodynamo and cause Earth's dipole field to collapse; if the mantle got hot enough, a magma plume might form that could reach all the way to Earth's surface.

Rare events could trigger really big avalanches at the CMB, however. When a massive asteroid or comet slammed into Earth's surface at an oblique angle, the lower mantle would jerk sideways, shearing off whole mountains of sediment. As the sediments slide up, a downward-sinking mass of cool iron could completely disrupt large convection cells. Although variously oriented local fields within the core would remain strong, at the surface Earth's dipole magnetic field would collapse.

"My theory is not really a theory of field reversals," says Muller. "Rather it's a theory of how the field is turned off. Then, over thousands of years, as the large convective cells in the core gradually reestablish themselves, the dipole field at the surface would turn itself back on, with a fifty-fifty chance of opposite polarity."

Chaos or cosmic bombardment?

Muller's theory faces competition from one other prominent theory of field reversals, that of Gary Glatzmaier, who with his colleague Peter Roberts used thousands of hours of supercomputer time to simulate hundreds of thousands of years of geomagnetic activity in the core.

"Glatzmaier has the best model for flow in the liquid core, which is very hard to do in 3-D," says Muller. "His simulation incidentally produced one rapid, spontaneous field reversal -- not surprising, but presumably reassuring to him, since his field-reversal theory is chaos theory."

	Gary Glatzmaier of Los Alamos and UC Santa Cruz used a supercomputer to model flow patterns in Earth's liquid core, source of the geodynamo. Complex fields in the core contribute to the dipole field at the surface. (Field lines flowing inward at Earth's North Pole are blue; outward flowing field lines at the South Pole are yellow.)

However, says Muller, "when I look at the geological record, I see things that don't look like chaos to me." Instead of random magnetic field excursions and reversals, Muller sees a constellation of suggestive associations with a common theme.

"There is a long literature on the association of asteroid/comet impacts and magnetic field reversals," Muller says. "The association of mass extinctions with impacts is well known, of course. And there is also an association of mass extinctions with flood basalts."

But how could impacts lead to flood basalts -- outpourings of thin fluid lava covering vast areas? Muller surmised that a sufficiently powerful oblique impact would unleash a CMB avalanche big enough to strip a patch of the lower mantle bare of insulating sediments. Hot iron would heat the exposed mantle rapidly; within a few million years a plume of magma would rise to the crust and burst out in titanic eruptions.

Thus Muller's CMB avalanche theory could link impacts and flood basalts, leaving as evidence a specific sequence of events:

• the oblique impact of one or more large asteroids or comets;
• which may be followed immediately by mass extinctions;
• which may be followed in a few thousand years by the collapse of the magnetic field;
• which may be followed a million years or more later by flood basalts;
• which will sooner or later be accompanied by the rebuilding of the magnetic field, possibly with reversed polarity.

A series of impacts, as from a comet shower, could spread these events over more than a few million years, but several appear to fit the model — including mass extinctions during the Permian and at the end of the Triassic, Jurassic, and Cretaceous periods.

Although the Cretaceous-Tertiary extinctions are associated with the huge Deccan Traps basalt flows in India, however, no magnetic field reversal occurred then. Muller thinks the reason is straightforward, and involves the "smoking gun" of the Cretaceous-Tertiary extinctions, the large Chicxulub crater on the edge of the Yucatan Peninsula.

	The asteroid that caused mass extinctions at the end of the Cretaceous Period probably struck the Earth vertically and did not disrupt the geodynamo.

"The evidence suggests that the Chicxulub crater was made by a vertical impact," Muller says. Unlike an oblique impact, a vertical impact would cause no shear forces at the CMB, thus no large CMB avalanches and no disruption of the geodynamo.

On the other hand, a powerful oblique impact could strip most of the CMB of its insulating sediments. This would be followed not only by enormous flood basalts but also by a very long period during which there were no magnetic field reversals at all, because not enough sediment had accumulated for avalanches to occur and trigger them.

The realization led Muller to make an inadvertent "prediction" of a phenomenon already well known to geologists — although unknown to him.

Predicting the past

"There is a so-called 'long normal' period in the history of magnetic field reversals," says Muller, "a quiet period that started 120 million years ago and lasted 35 million years, during which there were no field reversals at all." At the end of this period reversals began again, very infrequently, and have gradually but steadily became more frequent up to the present.

"I wrote up a draft of my paper in which I pointed out that my theory predicted a huge basalt flow coincident with the beginning of the long normal period," Muller says. "To my annoyance, there was no record of such a basalt flow. I could only hope one would be discovered some day."

He mentioned his disappointment to geologist Walter Alvarez, one of the scientists who proposed the impact theory of the Cretaceous-Tertiary extinctions. Alvarez pointed out that Muller had been using a table that listed only continental flood basalts; the largest flood basalt in the world, the Ontong-Java plateau, lies under the Western Pacific Ocean. Its formation 120 million years ago precisely coincided with the onset of the long normal period.

The largest flood basalts in the world formed immediately before a 35-million-period during which there were no reversals of Earth's magnetic field.

"So I had the pleasure of knowing I'd 'predicted' it," Muller grins, "although it had already been discovered long before."

Circumstances support Muller's CMB avalanche model of geomagnetic reversals in many ways, but critics have come up with lots of other ways to explain the clues away. He concedes that hard evidence won't be easy to come by -- to see first-hand what's going on in the core would require a journey to the center of the earth. Unfortunately, that adventure is likely to remain in the realm of unlikely science fiction for the foreseeable future.

Additional information

• "Avalanches at the core-mantle boundary," by Richard A. Muller, Geophysical Research Letters 29 (2002): p. 41



• More about Rich Muller and his group