Radiation exposure at the Crossroads.

In 1946, civilian scientists and 42,000 servicemen took part in military exercises associated with two atmospheric nuclear tests at Bikini Island. As part of Operation Crossroads, they entered a lagoon after the detonations to examine radiation levels and blast damage to target ships. Last year, the Defense Nuclear Agency (DNA) computed the participants' test-related radiation exposures, based on dosimeter readings from the 6,300 who had worn film badges, and found no evidence that overexposures had occurred. But at the request of Sen. Alan Cranston (D-Calif.), the General Accounting Office (GAO) reviewed several issues related to DNA's estimation procedures. In its analysis, just reported, the report says DNA's numbers cannot be trusted.

Among factors that GAO believes contribute to this are:

* badge dosimeters that were unreliable for measuring both external gamma and beta radiation.

* no allowance for inaccuracies attributable to the badge's film, even though DNA acknowledges that badge readings could vary by 30 percent (up or down).

* the high number of unmonitored workers.

* no estimate for exposure through open wounds.

* suspicion that decontamination was inadequate; the earliest mention GAO found of a requirement to shower or change clothes after work at contaminated ships was in procedures issued after the last Crossoards detonation.

* no accounting for film processing and reading inaccuracies. A mid-1950s National Bureau of Standards test found that several laboratories erred by plus-or-minus-100 percent in their reading of film badges similar to those used at Crossoards. Based on that, GAO doubts that "readings performed under harsh Crossroards conditions would have been more accurate than those in laboratories."

* DNA's method for estimating internal exposure, which may have underestimated alpha does by a factor of 10.

* DNA's faulty assumption that not eating onboard target ships precluded ingestion of radioactive materials.

GAO recommends that DNA revise its estimates of Crossroads exposures to account for these factors because potential errors in the original calculations are not just academic. The Veterans Administration uses DNA's calculations in adjudicating radiation-related disability claims. The Defense Department challenged many of GAO's assertions and recommendations in a 30-page dissent that has been included in the report; the accounting agency counters each with a point-for-point rebuttal.