Volcano: In the Belly of the Beast: Scientists know what makes a volcano blow but still cannot say when

The blowout that was so evidently seen and heard . . . gave off three muffled cannon rounds like a bombardment, so loud that they could be heard for more than 30 leagues around the base. In the region there were two rivers, the Guali and the La-gunilla . . . both were flooded with melted snow. It didn't really seem like water, but masses of ash and soil, with such a pestilent odor of sulfur that it couldn't be tolerated even from afar.

That vivid eyewitness account could be a description of last week's furious eruption of the Nevado del Ruiz in Colombia. Yet it was written by a Spanish monk, Father Pedro Simon, who observed on March 12, 1595, the 17,716-ft. volcano's previous major convulsion. The similarities between the two events point up both the always present menace of an active volcano and its lethal unpredictability. Although scientists were convinced that Nevado del Ruiz was due for a major burst, they could not pinpoint the time with sufficient accuracy to allow large-scale evacuation of the surrounding towns.

The Colombian eruption, like most volcanic events, is the result of continental wanderlust. According to the widely accepted theory of plate tectonics, the earth's crust forms the top layer of about a dozen major plates and several smaller ones, which range in thickness from 20 miles to 150 miles. These sections float on a gooey layer of partly molten rock known as the asthenosphere. As they move in different directions at an average speed of several inches a year, the plates collide, dive under and buckle against one another, crinkling up into a mountain range here, yanking apart to form a rift valley or oceanic ridge there. Such tectonic clashing was responsible for the violent earthquake that shook Mexico City two months ago, when the Cocos plate of the Pacific, temporarily stuck in its slow but inexorable plunge under the North American plate, suddenly jarred loose and lurched ahead. Last week's burst involved a similar movement of plates, but the result was entirely different. Extending along most of the coast of South America, the dense Nazca plate of the Pacific, moving eastward, subducts, or descends beneath, the lighter mass of the South American plate, which is moving westward. As the oceanic plate dives deeper into a region of high temperature and pressure some 60 miles to 125 miles below the earth's surface, rock in the area begins to soften and form magma, molten rock. Says Robert Christiansen, a volcano specialist with the U.S. Geological Survey (USGS) in Menlo Park, Calif.: "This is one area in which our knowledge is the least advanced."

Christiansen and others suspect that magma is produced in the subduction zone, the border between the diving plate and the lower mantle. In that complicated layer, a variety of phenomena, including high temperatures, changes in pressure and the influx of water, may act to melt the already softened rock. Minerals and water then coalesce with the molten material into viscous, tear-shaped packets known as diapirs. Because they are more buoyant than surrounding rock, the diapirs percolate upward, like bubbles rising through honey, melting more rock as they go. Eventually they accumulate in pockets called magma chambers, located two miles to 15 miles underground. If the magma is very liquid and gases can escape gradually, a volcano may lie fallow for long periods of time.