Will seismologists ever be able to reliably predict the exact location, time and magnitude of earthquakes like the one that just devastated Japan and sent tsunamis racing across the Pacific Ocean? If so, they might be able to save many lives. Consider how many people have been killed by large earthquakes just in the last decade: more than 20,000 people in India in 2001, 30,000 in Iran in 2003, 227,000 in Sumatra in 2004, 86,000 in Pakistan in 2005, 87,000 in China in 2008, and 222,000 in Haiti last year, according to the U.S. Geological Survey. Early reports from Japan suggest that the death toll could be in the tens of thousands.

The pioneers of earthquake studies were the Chinese, who began keeping records of where and when earthquakes occurred as early as 780 B.C. In the second century A.D. the Chinese invented a kind of weathervane for detecting and locating the center of earthquakes. The device consisted of a weight delicately suspended in a large bronze urn, ringed by dragons with hinged jaws. Jostling of the urn tipped the weight toward one side of the urn, causing the jaws of the dragon on that side to swing open and release a ball. The ball would supposedly fall on side of the urn from which the earthquake originated.

Modern seismometers are exquisitely sensitive, capable of calculating the exact location and strength of earthquakes on the other side of the planet. Moreover, the theory of plate tectonics—first proposed by Alfred Wegener in 1915 and finally accepted by other scientists in the 1960s--provides a firm foundation for understanding why earthquakes happen. Temblors tend to occur at the boundaries between the vast, shifting plates that comprise the earth’s crust.

A few decades ago, seismological technology and theory had advanced so far that many researchers became confident they could predict the exact date and location of earthquakes, providing time for evacuation and other life-saving measures. In 1985 scientists working with the U.S.G.S. funded an experiment intended to serve as a test bed for earthquake prediction. The experiment was based in Parkfield, Calif., a small town that sits astride the notorious San Andreas fault. Since the mid-1800s, Parker had been struck by earthquakes of magnitude 6 or greater every 22 years, on average.

Scientists outfitted the Parkfield fault with seismometers, strain gauges and other sensors that ideally would provide warning of an impending quake. The leaders of the experiment claimed there was a 95 percent probability that an earthquake of magnitude 6 or greater would happen by 1993. This claim—and the possibility of a precise prediction--received much attention from the media, including Scientific American. Parkfield was indeed struck by an earthquake—in 2004, 11 years after the initial prediction period had expired.

Other nations, notably China and Japan, have funded earthquake-prediction programs, but they have not been successful, either. One forecasting method focuses on the minor foreshocks that often portend a large quake. Unfortunately, this method is prone to false alarms, because the vast majority of minor tremors are not followed by major ones. Moreover, not all big quakes are preceded by foreshocks. China claims that foreshock-detection allowed it to successfully predict and evacuate people in the vicinity of a 7.3-magnitude quake in 1975. But over the past 20 years the Chinese program has issued more than 30 false alarms, and it failed to predict the 2008 quake that devastated eastern Sichuan.

Many other prediction methods have been proposed and in some cases tested. These involve detection of such alleged quake precursors as surges in ground water; emissions of the radioactive gas radon; fractoluminescence, or flashes of light emitted by compressed rock; unusual tidal activity; low-frequency electromagnetic waves; and unusual animal behavior. One long-running experiment in Japan involves monitoring catfish, which are supposedly sensitive to electromagnetic activity that precedes quakes. None of these approaches has proven reliable.

On the other hand, science and engineering have helped us reduce the damage of quakes. Whatever the final death toll from Japan's quake turns out to be, it would have been orders of magnitude greater if Japan had not designed its buildings, roads, nuclear power plants and other structures to withstand a vigorous shaking. Fewer than 1,000 people were killed by an enormous, 8.8-magnitude quake that struck Chile last year, because Chile has hardened its infrastructure against quakes. Tsunami-warning networks have also saved lives by quickly disseminating alerts to coastal regions.

So even if seismologists never achieve precise, short-term predictions of earthquakes, there is much that science can do, and has done, to protect us from this ancient scourge.

Replica of seismograph invented by Zhang Heng in China in the 2nd century A.D. Photo courtesy Wiki Commons.