What really happened at No Gun Ri?

It's clear that the Americans felt it necessary at that point in the conflict to put a stop to the refugees' movements. In some cases during the Korean War, the policy against refugee movement across American lines was explicit. One of the 7th Cavalry's "sister" units in the No Gun Ri area, for example, received a telephoned instruction relayed on June 24 by a regimental liaison at the division's headquarters: "No refugees to cross front line. Fire everyone trying to cross line. Use discretion in case of women and children."

This order, Bateman notes, is not recorded in any of the other regiments' existing message logs, or in the division log itself. But it is impossible to say if the same sort of message reached the 7th Cavalry because -- significantly, the AP reporters say -- the unit log of the 7th Cavalry for this crucial period is missing. Even if no one can find any documentary evidence that any such order was given to the 7th Cavalry, numerous documents on the Web site for the AP reporters' book show that the deliberate targeting of refugees was almost standard operating procedure in the early days of the conflict.

Then there's the grisly question of the disputed body count. The most significant finding of the analysis of the Aug. 6 aerial reconnaissance photos is that the bodies that were supposed to be lying around in the open until approximately Aug. 10 -- according to one of the Korean villagers who returned to the site -- are nowhere to be seen. Nor was there any evidence in the photos of soil excavation, funeral pyres or mass graves. Even most of the abandoned American foxholes in the immediate area, which might have served as expedient graves, were still open to the sky.

This finding was so troubling to Korean government investigators when they were shown the photographs during the joint Korean-American investigation prompted by the AP story that they essentially accused the U.S. government of having forwarded bogus images. Hanley, too, in an e-mail, noted that the South Korean technicians "suggested to [Sang-Hun Choe, the South Korean reporter on the AP story] the photos were tampered with." Understandably, the U.S. government categorically repudiated that suggestion.

When I asked Hanley about the fact that the National Imagery and Mapping Agency's examination of the photos did not find any bodies lying around in the open, he told me that the bodies weren't out in the open by that point, because the No Gun Ri villagers had found the burial job facing them too overwhelming. Instead, he says, they opted to temporarily stack the dozens of bodies in the tunnels formed by the concrete trestles of the bridge and cover them there with a layer of soil. Most of the individual burials did not take place until much later, and some of the bodies remained under the bridge "through the winter."

When I told Bateman about this theory regarding the lack of bodies out in the open, he scoffed. If 100 or 200 people had been killed under the bridge, he said, there would be literally tons of human remains to cover. Even covering them shallowly would take a lot of soil, and that soil would have to have been dug from somewhere near the bridges, excavations which would have appeared in the photos. Covering 200 corpses within the confines of the tunnel space would be like covering half a dozen Volkswagens. "Don't you think they would stick out just a little?" he asked me. "That photo, taken at a mild slant angle, shows 10 to 20 feet under the bridge but there's nada there. And as for excavations outside the bridge, hell, do you see any large excavation at all?"

I didn't. But something else in the government report on the photographs did catch my eye. The river that flowed under the railway bridge ran toward No Gun Ri, not away from it. It is extremely unlikely that the people of No Gun Ri would have left dozens of bloody, putrefying human bodies in the streambed of a river that led directly past their village. Even a brief flurry of rain in the nearby mountains could have washed the carrion onto their doorsteps.

Hanley dismissed my observations regarding the watercourse as "amateur hydrological analysis," and returned to his theme: the witnesses' testimony -- and the AP reporters' ability to discern the truth from those accounts and their own examinations of the site -- trumped everything, even expert imagery analysis. "You haven't been to the NGR area. We have. Many times. You haven't seen the lay of the land. You haven't talked to the villagers ... I simply don't buy the idea of some unnamed analyst 50 years later, viewing photos based on 50-year-old technology, never having been to the area ... making some judgment about what he saw there."

But the age of the technology is moot, says Bateman, when the photos' resolution is good enough to discern individual railroad ties. And, he says, "if the NIMA, which discovered mass graves in Bosnia from overhead reconnaissance, is not qualified to evaluate [the film], who is? Hanley?"

Never mind the pictures, Hanley said. The photographs and the flow of the river are virtually irrelevant to the question of how many were killed at No Gun Ri: "There are many possible answers, but dismissing the word of honest local villagers who remember bodies stacked under the trestle ... GI after GI remembers bodies stacked under the bridge when they were pulling out! ... and basing that on divining water volume and flow a half-century later, is not an answer."

For his part, Bateman believes that many of the stories the villagers told the AP reporters about coming under deliberate American fire are essentially true, and certainly some civilians did die under that railway bridge on July 26 and during the following days. But he also believes that at least some of the South Koreans' memories refer to other incidents and were conflated with the No Gun Ri incident when a South Korean author gathered them together into a 1994 "novel" that seeded the AP's interest in the event.

"I believe that the accounts of the South Koreans are a collage of several different events that occurred at several different places over the course of a few days," he writes in "No Gun Ri." And in fact, American troops had mistakenly fired on a large group of refugees accompanied by friendly Republic of Korea forces only a day or two earlier, in another nearby area, thinking they were the vanguard of a North Korean advance. The incident at the bridge was a tragedy, Bateman says, but it wasn't the "massacre of war crimes proportions" that most people thought it was after reading the AP team's story.

The argument really shouldn't come down to how many civilians were killed at No Gun Ri, but inevitably it does. Cynically speaking, after 50 years, the AP story would not have been Really Big News, much less warranted the Pulitzer Prize, unless the death toll of the No Gun Ri incident was high enough to qualify it as a "massacre" that was being covered up by the U.S. government.

"The number of dead at No Gun Ri has been -- as was clear in our September '99 journalism -- a secondary matter and a question we knew would never be resolved with precision," Hanley wrote to me while on assignment in the former Soviet Republic of Georgia. "The primary matter is the policy within the U.S. high command to officially, in writing, target refugees and other civilians." Bateman's book, Hanley says, "seems aimed at knocking down the casualty estimate to below 100, 'up to 70.' All to what end? Is the killing of 70 somehow more acceptable than the killing of 400? As [Lt. Gen. Bernard E. Trainor] said after the investigation, the numbers are just a matter of degree."

• View blog reactions