The Reagan administration has begun publicizing once-secret data showing that since the 1960s the U.S. arsenal of nuclear weapons has been plagued by serious mechanical problems, including a substantial number of duds.

To counter Soviet leader Mikhail Gorbachev's call for an end to nuclear testing, the administration has long argued that the United States must continue detonating weapons beneath the Nevada desert to guarantee the quality of the U.S. stockpile.

As evidence to support that argument, the administration has taken the unusual course of publicly questioning the reliability of the U.S. nuclear deterrent, offering details in congressional testimony showing significant problems in a variety of older warheads, including the Polaris and Poseidon submarine-launched ballistic missiles, the Minuteman I intercontinental ballistic missiles (ICBMs), the Sergeant short-range missile and the Army's atomic demolition munitions.

"At times in the past, the warheads for a large part of the U.S. submarine-launched missile force have been found to be deteriorated," according to an unclassified Department of Energy study presented last fall to a House Armed Services subcommittee. "At different times, a large fraction of the warheads either obviously or potentially would not work; they were obvious or potential duds."

More recent examples of problems "that have been revealed by nuclear tests" were contained in a classified report given to the subcommittee by Dr. Roger Batzel, director of Lawrence Livermore National Laboratory, one of the nation's two nuclear weapons-building facilities.

That document along with secret reports to the subcommittee from the departments of Energy and Defense discussed problems in two major U.S. nuclear systems currently in the stockpile, according to a congressional source. They disclosed that random warhead testing uncovered deterioration in one system that was deployed, according to the source.

With the second system, which was about to be deployed, this source said, a normal dismantling of one weapon before testing discovered that "it would not work within the planned operational environment."

The problems have since been corrected, the source said.

During the Feb. 26 House floor debate on a joint resolution calling on the president to negotiate an end to nuclear testing, Armed Services subcommittee chairman Beverly B. Byron (D-Md.) and several panel members referred to testimony on past warhead failures in opposing the legislation.

The House, however, voted, 268 to 148, in favor of the resolution. Supporters of a test ban argued that past failures had been discovered without nuclear test explosions and could be remedied the same way.

Rep. Nicholas Mavroules (D-Mass.), a member of the subcommittee who supported the resolution, said the testimony proved that the administration "is only looking for reasons to continue testing." If the United States is having these kinds of problems, Mavroules said, "you can imagine what is happening to the Soviets."

Some arms control advocates have argued that a ban on nuclear tests would be good precisely because it would result in diminished confidence in the reliability of nuclear weapons over time, leading planners in both Moscow and Washington to be more cautious about considering launching their weapons even in a grave crisis. Opponents of a test ban have argued that Soviet weapons would be more reliable in such circumstances, because they are bigger and simpler than U.S. warheads.

Nuclear weapons are complex mechanisms that contain not only radioactive material, but also chemical high explosives, electronics, metals, plastics and other sealing materials. These materials react with each other over time, and the weapons are expected to last up to 20 years in a variety of storage facilities.

As Batzel put it in his testimony, "Changes do occur during the stockpile life. These metals and salts interact in unpredictable ways." As a result, random tests of non-nuclear components are regularly undertaken. It is at that stage that many of the problems are found.

In 1961, according to the testimony, warheads on the Polaris sub-launched missile, which had first been deployed one year earlier, began to show signs of deterioration in the nuclear materials. Subsequent analysis showed that 20 percent of the warheads had to be modified to eliminate corrosion.

Two years later, it was discovered that a mechanical device put on the Polaris warhead to make it safe was flawed. Chemical reactions with surface components of the weapon prevented the safety from unlocking properly, so the missile would have been a dud if fired.

The Polaris was fixed in 1965, but a similar problem was found two years later. At that time, it was estimated that 50 percent of the missiles could fail, according to the DOE study. As a result, the entire Polaris warhead inventory was rebuilt.

The Poseidon sub-launched missile, successor to the Polaris and still in operation, went into production in 1970. By 1977, scientists running an annual inspection program saw signs of deterioration in the chemical explosive component which, they feared, could prevent the warhead from firing.

By 1979, it was decided that the high explosive in the Poseidon had to be replaced and several thousand were retrofitted.

The first Minuteman warhead developed mechanical problems in its arming device after more than 100 of the weapons had been produced. By the time the redesign was concluded, 160 warheads already deployed or in the stockpile had to be retrofitted.