Earth's mantle, the vast layer of viscous rock between its molten iron core and the outer shell of tectonic plates, may look much like one of those lava lamps that were all the rage in the '70s. Three studies in this week's Science suggest that the mantle consists of two layers that don't mix, although the lowermost layer bulges upward in some places and squeezes close to the mantle floor in others.
Geophysicists have debated the mantle for decades. Does it resemble a giant layer cake, neatly divided at a depth of 660 kilometers into two layers that never mix, or a boiling pot of water, churning from top to bottom over the eons? Seismic images of sinking ocean plates piercing the 660-kilometer "barrier" upset the layer cake model last year. But geochemical data pose problems for the one-pot model, by suggesting that some of Earth's ingredients are sequestered in an isolated part of the mantle.
Analyses of the speed at which earthquake waves traverse the Earth provide seismologists with information about temperature and composition of the mantle rock. Using these data, Rob van der Hilst and Hrafnkell Kárason of the Massachusetts Institute of Technology confirmed that cold slabs of ocean plate clearly sink through the 660-kilometer barrier. But below 1700 kilometers or so, seismic velocities vary widely from point to point in no recognizable pattern, suggesting that the lowermost mantle is subject to a different regime (Science, 19 March, p. 1885). Another hint of a deep boundary comes from a study by seismologists Satoshi Kaneshima of the Tokyo Institute of Technology and George Helffrich of the University of Bristol in the United Kingdom: They found evidence of a thin, chemically distinct slice of rock between 1400 and 1600 kilometers down; it could be a very old crustal slab come to rest on the top of the lowermost mantle layer (Science, 19 March, p. 1888).
To see if the deep lower mantle could stay isolated and chemically distinct over the eons, Van der Hilst and two colleagues constructed a computer model of a mantle with a bottom layer 4% denser than the overlying rock (Science, 19 March, p. 1881). They turned on the radiogenic heat of the lowermost mantle, sent slabs descending from the top, and found that the bottom layer swelled upward in places and thinned beneath slabs--just like a lava lamp. Yet the layer survived as a distinct entity for billions of model years.
"This might be an answer to our dilemma," says geochemist Albrecht Hofmann of the Max Planck Institute for Chemistry in Mainz, Germany. But other researchers are far from convinced. Seismologist Thorne Lay of the University of California, Santa Cruz, notes that the seismic images from below 1700 kilometers may be muddy simply because seismic data are poor at those depths. And there are seismic signs, he says, that the deepest mantle could be more dynamic than allowed by the layers in the lava lamp model.