Big shocks push volcanoes over the edge.

Earthquakes and volcanoes are alike in at least one way: They cluster along the margins where tectonic plates grate against each other as they move across the planet. Yet scientists have had little evidence that one earth-shaking event can trigger the other. Now, Alan T. Linde and I. Selwyn Sacks of the Carnegie Institution of Washington (D.C.) say that some of the world's most powerful quakes seem to set off volcanic eruptions.

A handful of researchers previously have tried to link individual shocks and eruptions, but "you could never be sure they weren't just coincidences," Linde says. There is one link between a quake and a volcano that everyone agrees was no coincidence, and it caught Linde's eye.

In 1992, a magnitude 7.5 earthquake near Landers, Calif., triggered knots of activity far away. It set off rumbling under Long Valley, a collapsed volcano 400 kilometers to the north. The valley shuddered hundreds of times a day. What's more, sensitive instruments detected swelling of the pool of magma stewing below the surface.

Long Valley did not erupt, but a question burned in Linde's mind. "Suppose the Long Valley system had been poised, just ready to go," he says. "Could this type of thing be the trigger that tips it over the edge?"

To see whether past earthquakes had ever pushed volcanoes over the brink, Linde and Sacks scoured global records dating back as far as the 1500s. They chose earthquakes greater than magnitude 7.0 and volcanoes for which the eruption start date was known.

The geophysicists reported in the Oct. 29 Nature that 8 of the study's 204 earthquakes of magnitude 8.0 or greater seemed to trigger same-day eruptions within 750 km. Altogether, 11 volcanoes blew their tops on the same day as these big quakes, but only 5 or fewer erupted on any of the 1,000 days before or after.

"If that's not just a fluke, it means earthquakes can cause things to happen at a distance where the shaking can't even be felt," says David D. Jackson of the University of California, Los Angeles.

To test the strength of the finding, Linde and Sacks randomly selected dates before or after each of the earthquakes and then recorded the number of eruptions on these dates. In 100,000 trials, only once did even eight eruptions cluster on one of the random dates. The chance of coincidentally having 11 eruptions, as in the historical data, is "terribly, terribly teeny," Linde says.

Jackson points out that the study's parameters limited the possible outcomes. If they had chosen a hypothesis that included different magnitude cutoffs or different time intervals, he says, the results might have been different.

Even Linde says that scientists won't be using earthquakes to predict eruptions any time soon. "You need a coincidence of physics," Linde says. "[The volcano] must be sitting and waiting to erupt for the earthquake to trigger anything."

Although a triggering effect is rare, "this is another tool for looking into how volcanoes prepare for eruption," Linde says. "The more you know about the physics of the volcanic system, the better you can tell what's going on."

Eruptions aren't the whole story, he adds. "There are probably a lot more like the Long Valley case, where the earthquake goes off and there's no eruption, but there is some flurry of activity at the volcano."

Few volcanoes, however, are monitored well enough to detect the kind of small changes seen in Long Valley's magma chamber. Such monitoring is bound to catch on, Linde says. "Just a couple of weeks ago, we put our first strainmeter on the slopes of Vesuvius, and we're hoping to get them on Etna and Mauna Loa as well."