When, last October, the Lancet published a study which concluded that the Iraq war had caused 655,000 more deaths than would have been expected if the conflict had not happened, the Prime Minister's official spokesman dismissed its findings as nowhere near accurate. He said the survey had used an extrapolation technique, using a small sample from an area of Iraq that was not representative of the country.
Now, documents released under the Freedom of Information Act reveal that British government experts backed the methodology used by the scientists responsible for the study. If the Lancet estimate is correct, it means 2.5 per cent of the Iraqi population - an average of more than 500 people a day - have been killed since the invasion. Of these, 601,000 died in violent acts - the majority involving gunfire.
Asked by officials to comment on the survey, the chief scientific adviser to the Ministry of Defence, Sir Roy Anderson, concluded: "The study design is robust and employs methods that are regarded as close to 'best practice'." He recommended "caution in publicly criticising the study".
The documents, released to the BBC World Service, show civil servants suggesting that ministers should not "rubbish" the Lancet report. In one email, an official, apparently from No 10 but whose name has been blanked out, asks: "Are we really sure the report is likely to be right? That is certainly what the brief implies."
Another nameless official replies: "We do not accept the figures quoted in the Lancet survey as accurate", but goes on to say: "The survey methodology used here cannot be rubbished, it is a tried and tested way of measuring mortality in conflict zones."
The documents advise ministers to use figures from the Iraqi health ministry, which estimates the number of deaths at less than 10 per cent of the Lancet's figure.
The ministry in Baghdad relies on hospitals to report the number of victims of terrorism or military action. But, critics say, the ministry did not start counting until well after the invasion and required busy hospital staff to report daily.
A statistician at the Department for International Development was also asked for an opinion of the Lancet study. The techniques used were "tried and tested", he said. If anything, the method "should lead to an underestimation of the deaths in the war and early post- invasion period".
The survey published by the Lancet was conducted by the Johns Hopkins School of Public Health and compared mortality rates before and after the invasion by surveying 47 randomly chosen areas across 16 provinces. The researchers spoke to nearly 1,850 families. In nearly 92 per cent of cases, family members produced death certificates.
Asked how it was possible to accept the methodology yet reject the findings, the government said: "The methodology has been used in other conflict situations, notably the Democratic Republic of Congo. However, the Lancet figures are much higher than statistics from other sources, which only goes to show how estimates can vary enormously according to the method of collection."
Owen Bennett Jones presents Newshour on BBC World Service