On the night of Sept. 22, when nuclear scientist believe an atomic explosion went off in the skies south of the Cape of Good Hope, a task force of South African warships was conducting a secret exercise at sea.

The precise location of the sea-based exercise is unknown, but the Central Intelligence Agency has told a few select committees of the House and Senate that the exercise was held at roughly the same latitude and longitude where the explosion appeared to take place in the atmosphere.

"That's one of the main pieces of evidence we've heard so far for this mystery explosion," confided one congressional aide briefed by the CIA. "That suggests that the source of the explosion could have been a rocket launched from one of those South African ships."

The CIA passed on another piece of information to Congress about the suspected Sept. 22 explosion. That same night, scientists peering into space with the world's largest radio telescope witnessed a ripple moving through the ionosphere in the skies over Puerto Rico a few hours after a Vela satellite saw a double flash of light in the southern hemisphere some 4,000 miles away.

"There were some questions raised about this ripple, but it appeared to be coming from the right direction and at the right velocity to have been caused by a nuclear explosion near South Africa," another congressional aide said. "If not, the ripple in the ionosphere is an awfully strange coincidence."

Ever since the State Department first announced the suspected nuclear explosion four months ago, the Carter administration has tried to discount it with explanations that seem to some scientists to depend heavily on coincidence.

A panel of outside scientists convened by the White House Office of Science and Technology first said that what the Vela satellite saw could have been a superbolt of lightning or a freak strike of lightning at the same time a meteor began to burn up in the atmosphere.

The panel has since discounted those explanations, suggesing instead that the satellite could have malfunctioned or seen a double glint of sunlight off another satellite or a descending meteor. The panel's final report is due to be made public next week and is understood to reach no conclusions.

Meanwhile, evidence is piling up on the side of a man-made nuclear explosion. Besides disclosing the presence of a South African naval task force and a ripple in the ionosphere the night of Sept. 22, the CIA is telling Congress that two sensors instead of one aboard the Vela satellite detected the double light flash charateristic of a nuclear event.

For weeks, Congress and the public have been told that an optical sensor called a "bhangmeter" on the Vela satellite picked up the double light-pulse of a nuclear explosion in the skies near South Africa at 3 a.m. Sept. 22

The double flash was described at Los Alamos Scientific Labratory as the "unmistakable" signature of an atomic explosion: a fireball blacked out for an instant by the shock wave surrounding the explosion, then reappearing 99 times more instense as the shock wave dissipates.

It turns out that there are two "bhangmeters" on the Vela satellite, one very sensitive instrument to spot the fireballs of small nuclear explosions and one not-so-sensitive to make sure large atomic blasts don't drown out the first instrument with their light.

"Both bhangmeters saw the double-flash the night of Sept. 22," a congressional aide said."This kind of rules out any mistake by one of the instruments."

The best corroborating evidence for an explosion is the sighting the same night by the radio observatory at Arecibo in Puerto Rico of a ripple in the ionosphere, suggesting that a shock wave had moved through the skies and displaced the electrons in at least one layer of the ionosphere as it moved along.

In calculating the velocity of such a shock wave, scientists figured that a shock wave starting about 4,000 miles away in the southern skies near South Africa at 3 a.m. Sept. 22 would have arrived in the skies over Puerto Rico about the same time they saw the ripple in the ionosphere.

Scientists on the White House panel have challenged those calculations. Scientists making the calculations insist they are correct. The White House is having them checked by another independent group of scientists.

Meanwhile, no radioactive debris has turned up in southern hemisphere rainwater to confirm that an explosion took place. This doesn't surprise scientists at Los Alamos, who remember not being able to find fallout from planned U.S. explosions and who point out that it took the United States as long as three weeks to sample the rainwater in the southern hemisphere after the suspected explosion last year.

Although the CIA cannot confirm an explosion, it has told Congress that if there was an explosion it suspects South Africa of having set it off.The CIA also mentions Israel as a suspect and rules out Pakistan and India. The mystery of Sept. 22 is still a mystery.