Senior managers at BP said overnight they had made crucial progress quelling the leak from their blown-out oil well in the Gulf of Mexico.

But a US government agency said the leak rate may have been five times greater than feared, making the disaster already significantly bigger than Exxon Valdez.

BP managing director Bob Dudley said that the initial "top kill" procedure of pushing mud down into the well and stopping its leakage into the sea had worked, but cautioned that the oil and gas beneath remained a potent force and was trying to push its way back. It would take until later today or even tomorrow, before it can be declared a success, he predicted.

"We are wrestling a beast that is about the same strength as us," he said in an interview at the BP command centre in Houston. "And we are trying to overcome it". The nearer the battle comes to a head, the harder it becomes, he added. BP won't declare victory "until the well is killed, until we have driven a stake through it... [and] that won't be before some time, perhaps 24 to 48 hours". Assuming, that is, nothing goes wrong.

The US Geological Survey said yesterday that as much as 12,000 to 25,000 barrels of oil had been spewing from the well every day n far greater than the 5,000 barrels a day that had been widely cited. That means that as much as 39 million gallons may have leached into the Gulf of Mexico since the blast on 20 April, compared with the 11 million gallons spilled in the Exxon Valdez disaster in 1989, the previous worst US oil spill.

"Now we know the true scale of the monster we are fighting in the Gulf," said Jeremy Symons, vice president of the National Wildlife Federation. "BP has unleashed an unstoppable force of appalling proportions."

Even as the top kill effort continued into last night, BP said it was preparing an ancillary operation that will involve injecting a viscous gunge into the crippled blow-out preventer on top of the well, essentially to try to clot it. If that works, the plume that we see now that consists mostly of mud coming from the pumps above should also vanish. All the mud would instead then be flowing into the well.

That the whole operation ends successfully has become paramount for BP as the public sees new pictures every day of fouled marshlands in Louisiana and as politicians in Washington multiply their attacks on the British company. BP faced additional difficulties as media reports suggested that it had skimped when choosing well-casing materials at the well to save money. It was also reported that BP managers on the rig had failed to carry out at least one normally required quality control test on the well hours before the blast.

News that the mud was successfully being driven into the well was cause for at least tentative celebration. It had never been tried at such depths before. "Even this is a quite remarkable achievement at the frontiers of human experience," Mr Dudley claimed.

However, in a discouraging development, independent scientists last night reported finding a new giant underwater plume of what they believe to be oil. The researchers said they had found a plume at least 22 miles long stretching from the leak towards Mobile Bay, Alabama.

The company, meanwhile, attempted to play down the new leak-rate findings. Mr Dudley said that other observers had put the rate of loss even higher at between 70,000 and 100,000 barrels a day and those he still considered "alarmist". He said that the best-known 5,000-day figures had never been BP's but had emanated from the government. "We never estimated a rate of flow," he insisted. Indeed, according to BP it remains impossible, even now, to be sure how much oil has entered the ecosystem. "I think you will see a good deal of study and refining of that in the future," Mr Dudley said.

Of the mud-pumping still going on in the Gulf, the BP executive compared it to pushing two extremely strong springs against each other - the mud against the rising oil. "The more you push them together, the harder and harder it gets," he said.

Company and government engineers have always known that the top kill carried risks. Push the oil back too swiftly and there is the risk of making things much worse. For example, further damage to the blow-out preventer could make the gush considerably worse than before.