In New Mexico's desert and Russia's Ural Mountains, U.S. and Russian experts are experimenting with simulated "dirty bombs" to see how such radiation weapons and potential terrorist tools might work, officials of the two countries say.

It's a sensitive area in which some information is withheld to keep clues to bomb-building out of terrorists' hands. But American and Russian specialists attending a global conference on dirty bombs disclosed some aspects of recent testing to a reporter because, as a ranking U.S. official said, the public should know everything is being done to deal with the threat.

These so-called "RDDs," for radiological dispersal devices, haven't made an appearance yet, but the Al Qaeda terrorist network, for one, is reported to have shown a serious interest in developing them.

Dirty bombs would combine conventional explosives with strontium, cesium or some other highly radioactive isotope used for such purposes as cancer radiotherapy, searching for oil deposits and sterilizing food.

They wouldn't cause the immediate mass casualties or devastation of nuclear weapons, but they are much simpler to make and the contamination and fear of radiation poisoning could cause general panic and shut down sections of cities for years.

Some 600 scientists, government officials and others at the three-day conference that ended Thursday focused on tightening protection of radioisotopes in use worldwide, stopping illicit trafficking and planning emergency responses to such attacks.

Others, meanwhile, are trying to learn how a dirty bomb would behave if detonated.

For the past six months teams at the U.S. Energy Department's Sandia National Laboratories in New Mexico have been experimenting with basic designs of RDDs, said the U.S. official, who spoke on condition of anonymity. Using materials that simulate the characteristics of the radioisotopes -- except for the radioactivity -- they have been exploding the devices to test the reach of the radiation effect as a result of blast and wind, he said.

Earlier computer modeling allows the testers to assess likely levels of radiation in various areas as a result of the blasts.

Formal results from the U.S. Defense Department tests haven't reached Washington yet. But researchers already know some things, such as that cesium chloride powder, used in large amounts in food irradiators and some older medical devices, is probably the material best suited for dirty bombs.

"It's very radioactive, and the powder disperses well," the official said.

He said the tests will be stepped up to the level of radiothermal generators -- devices packed with large amounts of isotopes, developed by the Soviet Union and the United States during the Cold War to power long-life aviation beacons and other remotely installed military equipment.

Such generators often hold 40,000 curies -- the basic measure of radioactivity -- in strontium or other material. Experts say even 1,000 curies might make an effective dirty bomb.

"A Russian admiral told us there have been many attempted thefts of RTGs reported," the U.S. official said, adding that apparently none was successful.

A Russian scientist, Alexander M. Agapov, told the Vienna conference it's possible 900 such devices were deployed by the old Soviet military, many with radio beacons or small lighthouses along Russia's Arctic fringe. Retrieving and securing that radioactive material will be a major challenge.

In his slide presentation, Agapov, safety chief for Moscow's Atomic Energy Ministry, described Russian computer simulations of dirty-bomb events. In a sign of the sensitivity, however, he blocked out the amounts of TNT and radioisotope used for simulated weapons.

He later told a reporter the Russians determined that radioactive particles from an explosion do not disperse in an oval pattern following wind direction, as usually theorized, but in a much more irregular pattern affected by crosscurrents.

Agapov described a Ural Mountains field test of emergency response, based on a computer simulation of a terrorist bazooka attack on a train carrying cobalt-60 and cesium-137, isotopes common in Russian transport. The test didn't involve an actual attack, but a small amount of isotope was used and controlled fires were set beside a train car, to test detection abilities and response.

He said the conclusion was that people within a mile or two of the attack would have to be evacuated within five to six minutes -- or at least kept sealed indoors. "You don't have time to check wind direction. You just move," Agapov said.