Federal energy officials continuously violated their own security procedures over a five-year period, resulting in erroneous declassification of at least eight documents that reveal crucial secrets of the hydrogen bomb, congressional investigators disclosed yesterday.

They testified that, beginning in 1971, the Atomic Energy Commission, now merged into the Department of Energy, reviewed 2.8 million classified documents from "inactive" files and declassified more than half -- 1.5 million -- by the time the program ended in 1976.

At the Los Alamos Scientific Laboratory (LASL) in New Mexico, 25 to 30 reviewers at a time declassified 234,215 of 388,092 documents between Jan. 15 and Feb. 16, 1973. If the reviewers worked seven days a week, that figures out to 237 to 284 declassifications per reviewer per day.

"Some of the reviewers felt the thrust of the informal instruction at this review was 'whenever in doubt, declassify,'" J. Dexter Peach, a General Accounting Office official, told the Senate subcommittee on energy, nuclear proliferation and federal services.

Under AEC and DOE rules, two reviews of a document are required to declassify it. But during the five-year program, Peach told subcommittee chairman John H. Glenn Jr. (D-Ohio), "there was no second review."

The congressional watchdog agency official also said that the agencies used reviewers who "lacked classification expertise" or who dealt with reports beyond their expertise. In some cases, the final decision whether to declassify rested with a reviewer "who was neither a classification specialist nor technically competent in the subject matter," he said.

The AEC "was very interested in getting a large number of documents declassified in a short time," Peach commented.

The central repository of declassified documents was the open stacks of LASL's public library. There, Peach told the senator, it would have been "very easy" for a foreign agent to examine the papers "without raising any suspicion at all."

In this regard, the subcommittee staff revealed DOE data showing that between April 1977 and May 1979, LASL had 630 "official" visitors from 52 countries, including 87 -- the largest single contingent -- from the Soviet Union. How many foreign visitors -- official or unofficial -- went to the library isn't known.

Until yesterday's hearing, the erroneous declassification of only two of the eight highly sensitive documents on thermonuclear weaponry had been divulged. One, "UCRL 4725," was accessible to anyone from July 30, 1975, until last May.

At that time, in connection with the government's effort to prevent publication of a Progressive magazine article on the H-bomb, researcher Dimitri A. Rotow found the document on the open shelves at Los Alamos. His discovery alerted DOE officials to the erroneous declassification.

Later in May, former nuclear weapons designer Theodore B. Taylor told the subcommittee that the exposure of UCRL 4725 had created the nation's "most serious breach of security since World War II." He said the paper describes or directly implies "basic design principles" not only for H-bombs, but also for the "major advances" made in atomic weapons since the war.

The second erroneously declassified document is said by the subcommittee staff to be of comparable sensitivity and the other six also to be highly sensitive.

In a spring 1978 visit to LASL, researcher Rotow came across UCRL 5475, an erroneously declassified document on the TX7 bomb, a Hiroshima, or fission-type, weapon. This disclosure, at a May 1978 Glenn hearing, led to a "crash" DOE review of 19,000 documents that had been declassified from 1971 to 1976 and that had titles indicating nuclear weapon subject matter.

Yesterday, the subcommittee staff disclosed that 2,000 papers had been re-reviewed (including 19 said by DOE to "no longer exist") and that the classification of 104 had been "upgraded," including seven containing "highly sensitive current weapon design information."

Rotow's later discovery of UCRL 4725 led to a new DOE crash review and to Glenn's requests for the staff and GAO investigations. This survey resulted in classification upgrading of an additional 23 sensitive documents, including one rated highly sensitive, for a total of 127.

Glenn termed this number "shocking," said that some of the papers would aid terrorists in building fission bombs. He said he found "no excuse for the . . . hurried, slipshod procedures used by the AEC . . . "

DOE Assistant Secretary Duane Sewell told Glenn of strenuous efforts to prevent recurrence of declassification errors. He agreed with the senator and the staff that the past errors made available to the scientists of many countries information that may be used to damage American security or U.S. efforts to deter nuclear proliferation.

But Sewell acknowledged to Glenn that no concerted interagency effort to assess the damage has begun.

Rotow disclosed last May that he had distributed but later retrieved copies of UCRL 4725, and he made an ambiguous commitment to reveal the recipients' names. Yesterday, Sewell said that despite requesting and receiving immunity from possible prosecution, Rotow has withheld the names.

Last Monday, the Progressive released the text of its article, with the government acquiescing because of the earlier publication by a Wisconsin newspaper of a letter containing similar information. Glenn denounced both publications as "irresponsible." As he did so, Howard Morland, who wrote the Progressive article, distributed copies to reporters in the hearing room.