On 4 February, 1975, something approaching a miracle took place in Haichen City, Liaoning. According to the Institute of Geophysics at the China Earthquake Administration, a combination of rising groundwater levels, unusual behaviour among the city's animals and a couple of small ground tremors persuaded officials that a major earthquake could be imminent. They promptly ordered a mass evacuation of the population. As a result tens of thousands, maybe hundreds of thousands, of lives were saved when a magnitude 7.3 magnitude quake struck hours later, destroying 90 per cent of the city's buildings. Ever since, mainland seismologists have been convinced it is possible to predict the exact time and location of earthquakes. They have poured investment into trying to measure stresses in rock, observing animal behaviour and even looking for strange lights in the sky and other bizarre portents some believe precede earthquakes. All to no avail; the very next year they failed spectacularly to predict the 7.8 magnitude earthquake which flattened the city of Tangshan in nearby Hebei killing 242,000. That failure should have taught the country's leaders a vital lesson: that accurately predicting earthquakes is impossible. The best anyone can do is enforce tightened engineering standards in seismically active areas, so that when earthquakes do happen, the casualties are minimised. The rapidly rising death toll from Monday's Sichuan earthquake indicates that this lesson has yet to be properly learned. The devastation caused by Monday's earthquake should have been no surprise. According to a recent study by the Benfield Hazard Research Centre - known for its 2003 study of tsunami risk and its 2005 warning of heightened hurricane risk - Sichuan is the most exposed of all China's provinces to earthquake damage (see the chart below). Yet building standards vary widely. While reports from companies in the area indicate that modern industrial plants and office blocks sustained little damage, public buildings and homes fared far worse. Especially shocking is the poor construction standard of school buildings. Whereas the Benfield study shows that factories and office blocks in mainland seismic zones come up to the A or B standards of minimum vulnerability, many school buildings attain only the vulnerable C standard. To put that another way, children in mainland classrooms are at the same level of danger in an earthquake as the inhabitants of caves cut into the friable loess soil of Shaanxi would be. (Friable, by the way, is a geological term meaning 'easily pulverized'.) This neglect of public building standards in earthquake zones is typical of the mainland growth-at-all-costs development model. Smelting plants and offices - structures which directly generate wealth - are built to withstand shocks. Schools are ignored. Visiting the earthquake area on Monday, Premier Wen Jiabao exhorted officials to spare no effort in the aftermath to serve the people. Mr Wen is a geologist by training who should be well aware of earthquake risks. If he means what he says, he should order a thorough structural upgrade of all schools in seismically active areas. Then there would be less need for miracles.