Late last week, in a steel container inside a sealed chamber carved out of solid rock 1,000 feet below the Nevada desert, a chemical explosive was detonated around a coin-like sample of less than half a pound of plutonium.

The experiment, code-named Oboe, was part of a $4.5 billion effort to guarantee the safety and reliability of the U.S. nuclear arsenal without violating the Comprehensive Test Ban Treaty, now being debated on Capitol Hill. Even though the treaty has not been ratified, the United States has adhered to it since 1992, using experiments such as Oboe, sophisticated computer modeling and other techniques to be sure that nuclear weapons are in working order.

Whether these efforts are sufficient, whether other countries can duplicate them and whether foreign scientists can cheat by setting off undetectable nuclear tests are key questions in the debate over the treaty. But they also are cloaked in high technology and secrecy.

Nuclear explosions occur when a critical mass of plutonium or enriched uranium, weighing a minimum of about five pounds, is pressed together by a chemical explosion. A neutron splits an atom, releasing energy and more neutrons, which split more atoms in an instantaneous, uncontrolled chain reaction. In the most powerful weapons, this fission reaction is used to set off a second reaction in which the nuclei of hydrogen atoms are fused together, releasing still more energy.

Because Oboe did not involve a critical mass--the amount of radioactive material needed to sustain a chain reaction--it is considered a "subcritical" test, a category allowed by the test ban treaty.

Last month, U.S. intelligence agencies detected activity at the Russian nuclear test site at Novaya Zemlya. According to a senior Clinton administration official, "What we see is similar to what the Russians would have seen" if their satellites had been above Nevada during Oboe. He believes, though he cannot prove, that the Russians, too, are undertaking subcritical tests.

While the United States was developing nuclear weapons from 1945 to 1992, almost all of its 1,030 atmospheric and underground nuclear tests were aimed at perfecting new designs, according to Sidney Drell, a nuclear weapons expert and a member of the President's Foreign Intelligence Advisory Board.

"Few of the tests were for stockpile confidence"--attempting to be sure that the existing arsenal was safe and reliable, he said.

Over the years, however, the United States began taking 11 warheads out of service annually for reliability review. Ten were examined visually, thoroughly inspected for aging, and then reassembled. One warhead each year was destructively torn apart and its nuclear materials examined for decay.

According to scientists, there are nearly 4,000 parts in some nuclear weapons. Radiation and simple aging can take a heavy toll on these materials, which range from epoxy glues to plastics, electronics, wiring and chemical explosives.

By the late 1980s, the United States had scaled back nuclear tests to fewer than 10 per year. And in 1990, a congressional study recommended a series of just 15 more tests. Two years later, however, President Bush approved a moratorium on testing as part of a broader arms control program. The moratorium was continued by President Clinton, who signed the test ban treaty in September 1996.

Energy Department scientists had already begun expanding the annual inspections. In recent years, they have added the subcritical tests and an effort to create the world's most powerful computer at Los Alamos National Laboratory in New Mexico to perform three-dimensional simulations of nuclear explosions. Lawrence Livermore National Laboratory in California also is building the $1.5 billion National Ignition Facility, a laser that will be used to study nuclear fusion.

It will be several years before the computer and laser projects are fully operational. But Victor Reis, former head of the Energy Department's weapons program, said yesterday he is confident the certification program for existing weapons will be sufficient in the interim.

Others are less certain. The 1960s-era Polaris warhead, the first nuclear weapon designed for submarines, "suffered from radioactive deterioration, corrosion, and had to be refitted," former defense secretary James Schlesinger reminded the Senate Armed Services Committee this week. "The warhead for the Minuteman had a problem with its high explosives, and so on. If you look at the history of weapons, problems will crop up from time to time."

Schlesinger added that five years ago, the U.S. nuclear laboratories opposed a test ban and recommended a treaty "permitting low-yield testing up to one, two or three kilotons, so that the weapon could be tested to the point of nuclear ignition."

Yesterday, however, the directors of those same labs told the Armed Services Committee that the current "stockpile stewardship" program could ensure the reliability of the U.S. arsenal without testing. Energy Secretary Bill Richardson also testified that if the need arose, the United States could abrogate the treaty and resume testing within a year.

Meanwhile, disagreements continue over how well U.S. intelligence agencies can monitor cheating.

Sen. Richard C. Shelby (R-Ala.), chairman of the Senate Intelligence Committee, said he doubts that the United States could detect a very low-level, underground nuclear test overseas. "Simply put," he told another congressional panel yesterday, "I'm not confident that we can now, or can in the foreseeable future, detect any and all nuclear explosions prohibited under the treaty."

In a statement to reporters Tuesday, Sen. Jon Kyl (R-Ariz.) said there was "reason to believe that both China and Russia" have conducted recent nuclear tests. But Kyl apparently was referring to the activity at Novaya Zemlya, which experts say may have been subcritical.

Drell, for example, told the Senate Armed Services Committee yesterday that he is familiar with the U.S. intelligence and is "not persuaded by the evidence that nuclear yield-producing testing has occurred in violation of" the test ban. He noted that in August 1997, similar intelligence surfaced about possible Russian tests, which turned out to be earthquake activity.

Moreover, Drell said, even if a nuclear power gets away with performing a small test explosion, "the value of a half- or one-kiloton test does not present a threat to our security." That opinion also was voiced by Defense Secretary William S. Cohen. Russia or China could "cheat at the margins or conduct some tests that could go undetected," Cohen said this week on "The NewsHour With Jim Lehrer." "But would that be militarily significant in terms of undercutting our strategic capability? Our judgment is no."

Two Methods of Nuclear Testing

Critical (Last U.S. use in 1992)

In a deep shaft, researchers detonate a nuclear device surrounded by a canister packed with measuring instruments. A cavity forms, and a layer of molten material falls into the cavity and hardens. As the cavity cools, pressure decreases and fractured rock falls in.

A thousand feet down or deeper, a fission reaction shoots shock waves through the Earth that can be detected at great distances.

Subcritical

(Allowed under test ban treaty)

Inside an 8-inch metal cube, researchers detonate a small explosive device, splattering apart a coin-size piece of plutonium without setting off a nuclear chain reaction. The cube, a steel drum surrounding it and a metal protective wall have glass panels for laser holography.

Plutonium test amounts

Critical 64 oz.*

Subcritical 2 oz.

*Minimum

SOURCE: Lawrence Livermore National Laboratory, Nevada Test Site