IN days gone by, the only way to try to anticipate an earthquake was to watch for what were called "empirical precursors" - visually obvious phenomena that were believed to precede a major tremor for reasons not completely understood.
The Chinese, for example, documented in considerable detail the strange behaviour of certain animals prior to a major quake, and it was also observed that water levels in wells might change, as fractures opened up and closed below the ground.
Nowadays, empirical precursors of more subtle kinds are being investigated. One theory which has received considerable attention is called the VAN technique, named after the three Greek seismologists, Varotsos, Alexopoulos and Nomicos, who are its main promoters.
It is based on the fact that weak electric currents flow to and fro in the ground because of minor voltage differences between one place and another on the Earth's crust, and the assumption that these "earth currents" may be disrupted by the kind of crustal stress that precedes an earthquake.
The VAN trio have constructed a network of monitoring instruments in Greece that are, in effect, giant voltmeters, comprising wires perhaps two or three miles in length connected to electrodes inserted in the ground. There is much scepticism in the seismic world about the VAN technique, but the trio claim to have predicted several weeks in advance every major earthquake in the region during the past 10 years.
Another latter-day precursor showing promise is a sudden change in the chemical composition of the local ground water.
This theory came to prominence when it was observed after the event that a week before the major Kobe earthquake in Japan in January 1995, radon levels in the ground water beneath the city had suddenly climbed to 12 times their usual level, before sinking back again to normal.
The theory is that radon may leak from uranium-bearing rocks that are under stress, and so the radioactive content of Japanese groundwater has been monitored very closely ever since.
In the US, scientists are using the Global Position System, the satellite-based GPS, to monitor minute movements of the Earth's crust in the hope that this may reveal a pattern that will prove predictive.
Perhaps the most mysterious empirical precursor of recent times has been the low frequency radio "hum" a Californian scientist noticed a month before the major Loma Prieta tremor in 1989; after monitoring the frequency for five uneventful years, he was rewarded in December 1994, when he heard the same kind of hum again - and another earthquake struck the region shortly afterwards.