When the US bombing of Afghanistan started on October 7 2001, an official "counting of the dead" was deemed unnecessary. The public was assured that American and British military planners would go to great lengths to avoid civilian casualties. The combination of newer, precision-guided munitions and the fact that bombing would take place in remote areas would mean that, in this war, only the "bad guys" would get killed. Subsequent events have proved these claims wrong. But how wrong? Everyone now accepts that civilians have died in American bombing raids in Afghanistan, but exactly how many is hotly disputed.
Given the lack of official interest, the counting of the dead fell upon interested individuals and non- governmental organisations. To date there have been nine studies, of which eight have been made public.The first study was my own, published in December last year. Relying on wire services, NGO and worldwide newspaper reports, I attempted to survey the bombing incidents to date and concluded that more than 3,500 Afghan civilians had been killed. A weakness of the initial study was some double-counting due to confused site names - the figure for the October to December period should have been between 2,650 and 2,970 civilian deaths.
Soon afterwards, a couple of cursory estimates were made by Le Monde and Reuters of about 1,000 dead civilians. At first sight, these seem considerably lower than my own, but this is because only a sample of bombings was examined. Reuters looked at just 14 incidents, which reportedly killed 982 Afghans. If one extrapolates out from the sample, the count broadly tallies with my own.
In February, the Wall Street Journal announced that Human Rights Watch was sending three researchers to Afghanistan - headed by William Arkin, a supporter of the war - to produce the "correct" tally of Afghan dead. HRW officials, it was widely reported, had "said privately" that they estimated the civilian death toll at between 100 and 350 in December, figures consistent with the group's record of severe undercounting in the 1999 Nato campaign in Yugoslavia. The HRW study has never appeared, though it has - absurdly - had some influence: the number 350 is still bandied about as if it had some scientific basis.
Around the same period, a major study was released by a prominent US thinktank, the Project on Defense Alternatives, arguing that US bombing in Afghanistan had killed civilians at a rate four times higher than the Nato bombing of Yugoslavia. By January 1 2002, the report calculated, between 1,000 and 1,300 civilians had been killed. The bombing campaign "failed to set a new standard for accuracy" because of the mix of weapons used, the unreliable nature of intelligence and the decision to bomb al-Qaida and Taliban leaders in their houses, where little margin of error existed. The PDA study was authoritative. Its total was lower than mine only because it relied exclusively on western sources. This made it more palatable to the media, but meant it involved a restricted number of incidents.
On February 11, the Associated Press released its counter-study, boldly reassuring an increasingly alarmed public: "Hundreds lost, not thousands". Its astonishingly low figure of 500-600 was reached "by examining hospital records, visiting bomb sites and interviewing eyewitnesses and officials." The report was beset by methodological problems. Most Afghan deaths are not recorded in hospital records because people are buried immediately; no details were given of interview methods or which bombing incidents were included; many bomb attacks were not reported; and Afghan officials have been shown often to seriously underestimate civilian casualties.
A far better survey - of 14 sites bombed by US warplanes, which resulted in 830 civilian deaths - was published the same month by John Donnelly and Anthony Shadid of the Boston Globe. The authors noted: "Because the 14 sites represent only a small fraction of the total sites targeted... since October, the total is estimated at 1,000 or more." The prime culprits for civilian deaths were: faulty intelligence; imprecision of aerial warfare; and "the selection of targets in civilian areas". Another compilation, by the Los Angeles Times, came up with a death toll of between 1067 and 1201 between October and February. But neither raw data nor sources were disclosed.
Last month, the NGO Global Exchange released a preliminary report of civilian casualties caused by US bombing since the beginning of the war. The study of 11 sites purported to document 812 deaths. This report is seriously flawed. We are not told which bombed places were visited (though we do know that only four of Afghanistan's 30 provinces were included). No raw data is produced and the number of bombing incidents is not divulged. Without this context, the low count of 812 dead is meaningless.
Finally, Dexter Filkins of the New York Times last month published his study of 11 bombing incidents in which 396 Afghan civilians reportedly perished. My own database reveals that in the same 11 incidents between 408 and 509 civilians died. Filkins points to the use of overwhelming force as causing many of the casualties. His study drew an immediate rebuke from Donald Rumsfeld, the US defence secretary.
In the eight months since I published my original study, I have updated and corrected my database, and incorporated the civilian deaths resulting from British and US special forces attacks. My most recent figures show that between 3,125 and 3,620 Afghan civilians were killed between October 7 and July 31. This is compatible with the sample counts done by Donnelly- Shadid, Filkins and (probably) the Reuters study. Comparison with the PDA and Los Angeles Times reports is difficult to make as they do not reveal raw data and exactly which sources were employed. The AP count is flawed both in coverage and methodology and the Global Exchange report is incomplete.
In war, counting is not value-free. To overlook or underestimate the civilian dead gives rein to the enthusiasts of precision-guided weaponry. It is an invitation to proliferation of war. The thousands of Afghan civilians who perished did so because US military and political elites chose to carry out a bombing campaign using extremely powerful weaponry in civilian-rich areas (the isolated training camps were largely destroyed during the first week).
For political reasons, it has been necessary to hide the human carnage
on Afghan soil as much as possible from the western public. Given that many
of the bombing attacks - such as those on civilian infrastructure (cars, clinics,
radio stations, bridges) and those during November and December on anything
rolling on the roads of southern Afghanistan - violated the rules of war,
there are war crimes that need to be investigated. An inadequate count will
make it impossible for the families of those wrongfully killed to get the
compensation to which they are entitled. It will also impede international
The author is an associate professor at the University of New Hampshire, US. His writing on the human costs of the Afghan bombing campaign can be found at www.cursor.org; his database of Afghan civilian casualties is at http://pubpages.unh.edu/~mwherold