Earth scientists have been arguing for 30 years about what the inside of the Earth looks like. At long last, an answer seems to be emerging from the depths, thanks to painstaking analysis of seismic shock-wave data from around the world, by Dr Rob van der Hilst of the Massachusetts Institute of Technology and colleagues, published in the 10 April issue of the science magazine Nature.

The answer promises to usher in a new era of understanding about the structure and history of our planet. The researchers show how, in the normal course of plate tectonics, the surface layers of the Earth can be sucked down almost to the Earth's core, in contrast to stalling halfway down, as many researchers would prefer.

The Earth has a kind of layer-cake structure. The crust and the upper parts of the underlying mantle form the icing, a layer called the lithosphere. Beneath the lithosphere lies the more viscous parts of the rocky mantle, which extends all the way down to a sharp boundary with the liquid-iron outer core, 2,900 km beneath our feet.

This new evidence may well turn the standard 'layered' model of the Earth's interior into a more dynamic entity, dominated by wholesale mantle convection, but with a significant overlay of more local, shallower mantle structures that reflect discontinuities within the mantle itself.

At the Earth's centre is the solid, metallic inner core. The lithosphere is about 100-km thick beneath the oceans, but up to 400-km thick beneath the continents. Although continental crust tends to be stable on timescales of 2,000 million years or so, oceanic crust is never much older than 100 million years. This is because it changes constantly: the crust is divided into a series of 'plates' that move independently, jostling one another at the edges and causing earthquakes and volcanism. New oceanic crust wells up from the mantle in mid-ocean ridges, and old oceanic crust is 'subducted' back into the mantle in the oceanic trenches that fringe continental margins.

In addition, new ocean crust emerges within plates themselves by the formation of 'hotspots', caused by 'plumes' of mantle material. The Hawaiian island chain is still being formed by the slow drift of the Pacific plate across a stationary mantle plume. When a plume coincides with a mid-ocean ridge, the result can be dramatic -- Iceland is the product of such a union.

Over the course of thousands of millions of years, material subducted at ocean trenches could, once again, re-emerge at mid-ocean ridges, completing a cycle of convection in the mantle. The question is how deep the convection goes. Does subducted crust make it all the way down to the core-mantle boundary, or does it halt halfway down, isolating the lower mantle from the outer layers of the Earth? And at what depth to mantle plumes originate?

Many researchers attach significance to a distinct boundary in the mantle at a depth of 660 kilometres. Seismic evidence shows a marked `phase change' at this depth, perhaps caused by a shift in the structure of the minerals that make up the mantle. There is considerable geochemical support for a division into an 'upper' and a 'lower' mantle, separated by this boundary. Other researchers, though, think that slabs of subducted crust can penetrate the boundary, sinking far more deeply into the Earth -- and that, in general, mantle convection involves the whole mantle rather than being restricted to superficial layers above 660 km.

There is also accumulating evidence that some of the more dramatic mantle plumes start their journey much deeper than 660 km.

Until now, though, seismic evidence has not been strong enough to convince the sceptics. Dr van der Hilst and colleagues present carefully reevaluated seismic data that show how some subducted slabs seem to remain coherent as structures down to depths of at least 1,700 km, and may extend all the way to the core-mantle boundary.

©Macmillan Magazines Ltd 1997 - NATURE NEWS SERVICE