Guido Bertolaso (right) said earthquakes could not be predicted
A scientist's claim to have forecast an earthquake that killed dozens of people in central Italy has been challenged by authorities and experts.
Italian researcher Giampaolo Giuliani said he had forecast a major quake in the area by measuring emissions of a radioactive gas, radon.
But the head of Italy's civil protection agency dismissed the claim.
Other scientists have also cast doubt on the possibility of predicting earthquakes with any accuracy.
Shortly before Monday's earthquake in and around the city of L'Aquila, Mr Giuliani was reported to the authorities for spreading alarm when locals apparently took fright at his warnings.
That was after a group had toured the area late in March, telling people through megaphones to leave their houses because of an imminent earthquake.
Last week Guido Bertolaso, the head of Italy's civil protection agency, had reportedly denounced "imbeciles who spread false information" after Mr Giuliani's prediction of a major quake on 29 March failed to materialise.
After Monday's disaster, Mr Giuliani, a researcher attached to Italy's National Institute of Nuclear Physics, was quoted as saying he was owed an apology.
"We have been able to predict these kind of events for 10 years," he told the Corriere della Sera newspaper.
"Over the last three days we were seeing a sharp increase in quantities of radon, over and above the level that is safe."
But speaking at a news conference on Monday, Mr Bertolaso insisted there was "no possibility of making any predictions on earthquakes".
"This is a fact in the world's scientific community," he said.
Italy's National Institute of Geophysics and Vulcanology said in a statement that "at the current state of knowledge" it was not possible to forecast the location, timing and force of an earthquake.
This was the case even when there were a number of tremors ahead of an earthquake - as there had been in L'Aquila, starting in January - the institute said.
Most current practice focuses either on very long-term forecasting of earthquake probability 10 or more years in advance, or on "early-warning" systems that can detect a major quake just seconds before it happens, according to Ross Stein, a geophysicist at the US Geological Survey in California.
Predicting major earthquakes days or weeks in advance is not thought to be possible.
Precursory events including foreshocks, deformation or straining of the ground, or the release of fluids or gases do not provide detectable patterns that can be used for forecasts, Mr Stein told the BBC.
"You can always find individual cases where one of those phenomena has preceded an earthquake but they're not reliable," he said.
He acknowledged that there was some evidence that trapped gases had "played a role" in a major earthquake in the Italian region of Umbria in 1997, and that the thinking around radon emissions as an indicator could change with new discoveries.
But he added: "Internationally, this method has no better track record than any of the other loved but then abandoned earthquake precursory phenomena."