That's the recipe used by two University of California, Berkeley, geophysicists to produce the first laboratory evidence in support of a theory that explains the formation and persistence of the Earth's hot spots, such as the one underlying the Hawaiian Island chain.

A dozen or more chains of volcanoes like the Hawaiian Islands exist around the world, each a record of one of the Earth's tectonic plates rolling over a stationary hot spot. Geophysicists generally agree that hot spots are the tops of plumes of rock extending 1,800 miles down to the bottom of the Earth's mantle, funneling hot rock to the surface, where it melts and spews out as lava.

The question for theorists has been why plumes persist for tens of millions of years, if not longer, and why they remain in the same spot their entire lives.

"It's a total mystery," said Mark Jellinek, a Miller postdoctoral fellow at UC Berkeley. "Mathematical models do not predict a stationary plume."

Jellinek and Michael Manga, associate professor of earth and planetary science at UC Berkeley, modeled the mantle in a vat of viscous oil and showed by analogy that a thin layer of dense, low-viscosity rock at the base of the mantle could anchor plumes for long periods of time, perhaps for the entire age of the Earth.

The two scientists reported their experimental findings on Sunday, Dec. 8, at the annual meeting of the American Geophysical Union in San Francisco. The meeting ends Dec. 10. The team also detailed their experiment and conclusions in an article in the Aug. 15 issue of Nature.

Over the past few years, evidence from earthquake vibrations bouncing around the Earth's interior have revealed a layer of dense material at the base of the mantle that clumps around the feet of the known plumes. It stands out because seismic waves slow down when they pass through the material, indicating rock that is hotter than the mantle above. Several scientists have suggested that if this dense material is 10 to 100 times less viscous than the overlying mantle, the clumps could anchor the plumes and explain their long lifetimes. The dense layer, which sits on top of the Earth's core, is about 60 to 200 miles (100-300 kilometers) thick.

Jellinek and Manga sought to test this hypothesis with a series of more than 40 experiments in a vat of polybutene - an oil used in two-stroke engines and as a lubricant - that simulated a scaled-down mantle. The vat, about the size of a 10-gallon fish tank, was 17 centimeters (6.7 inches) deep.

To create rising plumes like those in the mantle, they stuck a heater at the bottom of the vat and cooled the top. The heat, simulating that from the Earth's hot outer core, generated convective plumes and buoyant bubbles like those in a lava lamp on in a pot of water on the stove, though they moved around and tended to be short lived.

The two researchers injected a thin layer of less viscous oil at the bottom, simulating the dense, low-viscosity rock at the base of the mantle. This oil, tinted red to distinguish it from the oil above, formed a layer about 1/5-inch thick.

They tried oils of different viscosity, ranging from walnut oil and various vegetable oils to specialized lubricants, until they found one - soybean oil - with the right viscosity to create oil clumps at the bottom of the vat with long-lasting plumes rising above them. Soybean oil is about 110 times less viscous than polybutene.

Apparently, Manga said, as a plume rises, it draws up the less viscous oil to create a peak or ridge, which makes it easier for the more viscous plume to rise above the clump. This coupled flow lets plumes stabilize themselves for long periods. While each plume in the tank lasted about an hour - the duration of the experiment - this time period scales up to about 3.6 billion Earth years.

"We now have a mechanism by which to make hot spots last a long time and fix them," Manga said.

The experiments also predict how closely plumes would be spaced in the mantle.

"Under the Pacific Ocean, we would expect the spacing to be about 1,000 kilometers (630 miles), which is about what we see," Jellinek said. That would mean there could be about 20 hot spots around the globe.

This number accords with some estimates, which identify as many as 15 hot spots under the oceans and continents. Continental hot spots are difficult to confirm, however, because the topography - mountains and crustal deformities - deflect upwelling rock. Some geologists claim to see evidence of nearly 100 hotspots - a count that Jellinek and Manga doubt.

The experiments clearly demonstrate what happens during the lifetime of a plume, which can take 30-50 million years to reach the surface after it starts to rise, Manga said. The mantle rock, which has a viscosity comparable to glass, is heated by the hotter core and begins to rise. As it does so, the head balloons out, trailing a thin conduit, like a soda straw, that connects it to the bottom of the mantle. This conduit, perhaps only six miles (10 km) across, sucks up the dense, low viscosity material through its center, drawing it to the surface along with hot mantle rock.

Above a certain depth, the head of the plume begins to melt and spread out under the crust, until it finds a crack and pours through. The theory of mantle plumes predicts an initial flood of lava lasting tens of millions of years, which means that each hot spot should have an associated flood basalt. The initial burst of lava from the Hawaiian hot spot has been slurped back into the Earth at a subduction zone, but other major flood basalts, such as the Siberian and India's Deccan Traps, are at the end of an arc of volcanoes and thus fit this model.

Lava can flow for several million years and create flood basalts as large as 10 million cubic kilometers - an outflow rate of about 10 cubic kilometers per year. That volume would submerge San Francisco each year beneath 270 feet of lava, with only the top of City Hall's dome peeking out.

After the plume head has dissipated, the conduit remains in place indefinitely, periodically erupting and dotting the surface of the plates with a chain of volcanoes.

Jellinek and Manga hope to find evidence to support their theory.

"We hope to see if we can find evidence in the geochemical record of the entrapped material from the dense layer," Manga said. "This also provides a way to predict what the material is like where the hot spot came from at the core-mantle boundary."

There are two reigning hypotheses about the origin of this dense material. One is that the core has partially melted some of the mantle rock, even at the high pressures near the core. The partial melt would be less viscous than the overlying mantle.

An alternative theory is that bits of the core have become mixed with mantle rock, forming an iron and silicon slurry that would be less dense than the mantle material.

"Because the plume should drag up bits of the dense layer, we should be able to find evidence of its composition in the flood basalts associated with hot spots," Jellinek said.

The two scientists plan to continue their experiments and have built an even larger tank to hold nearly two tons of corn syrup for another simulation of the mantle, including its interaction with the constantly moving crust.