A high-priority, classified alert issued by the CIA on Aug. 18 quickly caught the eye of senior U.S. policymakers. The bulletin came from the government's Nuclear Test Intelligence Committee, an interagency scientific group, and said that Russia probably had conducted a nuclear test two days earlier on an island near the Arctic Circle.

Officials at the National Security Council swung into action, convening an interagency meeting two days later and ordering a full-court press to collect an explanation from Moscow. The Russian ambassador was summoned to hear a strong complaint at the State Department, and the senior U.S. diplomat in Moscow issued a similar demarche at the Foreign Ministry there.

Although the government kept the report secret, the NSC prepared a statement to be read in case of a leak, which said, "We do have information that a seismic event with explosive characteristics occurred in the vicinity of the Russian nuclear test range" on the island of Novaya Zemlya. When the statement was eventually released on Aug. 27, it raised dark suspicions around the world that Moscow had challenged the nuclear test ban treaty.

There was only one problem: The CIA's report about the location of the "event" was wrong, according to various U.S. intelligence and defense officials, independent scientific experts, and the British, Norwegian and French governments. The event actually occurred roughly 80 miles at sea and, these officials and experts now say, was almost certainly an earthquake.

In the past two months, U.S. intelligence officials say, the CIA has scoured its overt and secret sources of intelligence near the test site and has found nothing to corroborate its initial report -- no sign of unusual radioactivity, no record of telltale underwater blast sounds, no indication of underwater drilling or extraordinary activity of any kind in the Kara Sea off Novaya Zemlya before, during or after the event.

The Russian government has called the charges unfounded and disappointing. But the administration has not yet publicly given Moscow a clean bill of health, a circumstance that some U.S. officials and independent scientists claim is partly due to a lingering distrust of Russia's military operations in the vicinity of the test site and partly to the reluctance of the CIA and senior policymakers to acknowledge that they made a diplomatic and scientific goof.

"I personally think it was an earthquake," said Harold P. Smith, assistant to the secretary of defense for nuclear, chemical and biological defense programs, in a telephone interview on Friday, adding that other scientists at the Pentagon share his belief that the initial CIA report was wrong. "We now know that they would have been well advised to wait" until they had more data and could reach an accurate conclusion, he said.

"Not only was there a mistake made, but there was no effort to retract it," said Paul Richards, a seismologist at Columbia University who consults for the Arms Control and Disarmament Agency and has written a forthcoming article for the scientific journal Nature that criticizes Washington's handling of the event.

What makes the initial CIA report especially surprising is that the event was described from the outset as having occurred at sea by officials at an international monitoring center in Arlington, which was created to collect, analyze and distribute data from a worldwide network of nuclear test sensors -- the same network used by the CIA to provide its analysis in this case.

About an hour after the disturbance occurred, computers at the center -- drawing upon seismic signals from five sensors or arrays in Norway, Sweden, Finland and Russia -- had fixed the probable location at more than 60 miles from Novaya Zemlya and ruled out the possibility it had occurred at the test site. A second report, prepared by the center after more exhaustive analysis two days later, confirmed that conclusion.

An unclassified report on the center's analysis, including a statement that the event was most likely an earthquake, was slated to be presented last month by center director Robert North at a scientific meeting convened by the Defense Special Weapons Agency in Orlando. But at the last minute, the presentation was canceled, after officials elsewhere within the Defense Department objected to making the information public, according to the official who made the decision to withdraw the presentation, who asked that his name not be published.

"The sense I got was that there were some concerns {about the paper} and this thing had not been vetted," the official said, attributing the decision to a problem of timing rather than censorship, as some other officials have privately claimed.

Interviews with White House, Defense Department and intelligence community officials indicate that confusion over the origins of the Aug. 16 event stemmed largely from the fact that Russian technicians were busily engaged in suspicious-looking activities on Novaya Zemlya during the same period. Although no test had been conducted there since 1990 and Moscow had since promised to abide by a treaty banning all nuclear tests, which it had signed but not ratified, U.S. officials were wary.

If Moscow were to carry out such a blast, it not only would undermine the treaty but could have implications for the dormant U.S. nuclear testing program. Congressional legislation barring future U.S. blasts is conditioned in part on continuation of the Russian moratorium, making it possible for Washington to reconsider its position if Moscow is proven to have shifted course.

The CIA was aware before the event that the head of Russia's atomic energy ministry, Viktor Mikhailov, recently had visited the vast test site, and the agency had snapped satellite photographs showing test equipment being lowered into the ground, with telltale diagnostic cables leading away from the holes into nearby buildings.

As one intelligence analyst said, activities at the site on both Aug. 14 and Aug. 16 were "a dead ringer for {those in} test shots" by Moscow over the past 10 years. The Air Force Technical Applications Center (AFTAC), a little-known organization based in Florida that conducts classified studies of Russian nuclear blasts, was alarmed enough to order a plane equipped with radiation detectors to fly downwind from the site on the first of these two dates, but it found no trace of a nuclear explosion.

The first sign of a seismic "event" on Aug. 16 was recorded at a station operated by Russia's defense ministry at Norilsk. The station has already been designated as one of 320 nuclear monitoring sites that will form a global system meant to sort through roughly 20,000 seismic disturbances annually to determine if any were caused by a secret nuclear blast.

The signal from Norilsk was transmitted automatically by a direct link to the International Data Center, which the Pentagon created as a prototype for a more elaborate center to be completed in Vienna by 1999. Additional data were transmitted there within minutes from two monitoring sites in Norway, one in Finland and one in Sweden, and they fixed the time of the event at close to 5 a.m. at the site, a circumstance that matched the punctuality of past Russian nuclear tests.

When the Air Force group -- which learned about the event from the center -- passed the information to the CIA on Aug. 18, the agency's analysts were alarmed by the coincidences, including indications that some of the signals recorded by seismometers looked like those from past nuclear blasts there.

The CIA organized an emergency meeting of the Nuclear Test Intelligence Committee, consisting of roughly 10 specialists from different agencies, and took a new look at the data. They tossed out the signals from the sites in Russia and Sweden on the grounds that their sensors were not properly "calibrated," and determined that the location of the blast, while probably at sea, could have been at the test site.

In the initial alert message to policymakers, however, this analysis was garbled through an error that intelligence officials said Friday they could not explain. The "event" was described without any qualifiers as having definitely occurred at the test site, and the wrong location coordinates -- those from a past Russian nuclear test on the island -- were inadvertently listed in place of the most probable location at sea, according to intelligence officials.

"We were trying to be very, very careful," said one official, adding that it was nonetheless a "fast answer" based on partial data. One of the policymakers who received the report said it conveyed "very high confidence that it was explosive . . . and right at Novaya Zemlya."

Smith said it was his sense that the test site activities had made the intelligence community "leaning forward and spring-loaded" to reach this conclusion. Eugene Herrin, a professor at Southern Methodist University who for the past 15 years has chaired the military's principal seismological advisory panel, said he agreed that "somebody jumped the gun. Based on what I know {from both classified and open sources}, it was not an ambiguous event. . . . It's an earthquake."

The committee's report also was challenged by Norwegian scientists, who told the Pentagon on Aug. 21 that "the event appears to be quite confidently located offshore," according to a copy obtained by The Washington Post. British government scientists reported to the Pentagon on Sept. 11 that the event "has a similar location and mechanism" as an earthquake that occurred in the Kara Sea 11 years ago.

But the nuclear intelligence committee, which the CIA chairs, did not formally begin backpedaling until two weeks after the event, causing one official to describe it as "the last to join the crowd." A new, classified report by the committee in early September, incorporating additional data, declared that the test site activities and the seismic event were not linked and affirmed that the event occurred at sea.

According to several officials, the CIA has no evidence it was "an explosive event," but remains reluctant to call it an earthquake because of a paucity of data on previous earthquakes in that area. "We like our judgment to be based on positive evidence," one intelligence official explained.

Some officials are also interested in exploring the possibility that the shock waves were caused by a sudden compression of the hull of one of the outmoded nuclear submarines that Russia has dumped into the Kara Sea; an alternative concern that one of the submarine reactors might have become "critical" and exploded has now largely been dismissed. The intelligence official said the CIA's scientists will try to obtain better data more rapidly when the next such "event" occurs.

But the CIA's conclusion that there is no evidence of an explosion evidently has not yet reached the NSC. The director for defense programs there, Robert Bell, said Friday that based on what he has been told so far, the event should still be considered ambiguous, with neither an explosion nor an earthquake ruled out.

"We are still trying to talk with the Russians, {but} our assessment from the technical side . . . {is} it was more likely than not explosive in nature," Bell said.

"We think this makes the case for the Comprehensive Test Ban," which provides for consultations and on-site inspections to resolve suspicion, said a White House official.

Moscow has informed Washington that its activities at the test site were scientific experiments related to its nuclear weapons program that did not involve nuclear fission. The Energy Department conducted similar experiments in Nevada last summer, after the administration spurned a suggestion from arms control groups that it allow international observers to monitor the work and allay any foreign suspicions about it. A senior Pentagon official called the idea "not necessary, . . . not cost-effective, and not prudent" in a letter to one of the groups.