The federal government has published and made publicly available at least a half-dozen documents outlining ways to break into nuclear power plants and other atomic facilities.

"Barrier Penetration Database," an illustrated Nuclear Regulatory Commission report issued in June 1978, describes 32 physical security barriers commonly found at nuclear plants. The report suggests specific bolt cutters, crowbars and other tools as well as the exact amount of explosives needed to breach each barrier. It even tells how long penetration would take, down to a fraction of a minute.

The report is based on the more extensive "Barrier Technology Handbook," prepared in 1977 for the Department of Energy by Sandia Laboratories of New Mexico, which describes methods to break through more than 130 physical barriers considered for use at DOE facilities.

The Nuclear Regulatory Commission has distributed both publications to many of its more than 130 public document rooms, which are at libraries near every operating or proposed nuclear plant in the U.S. They also can be bought by mail from the government's National Technical Information Service in Springfield, Va.

Other reports obtainable from the service include:

A U.S. Army assessment of the weaknesses of detection systems at nuclear fuel sites.

A Mitre Corp, guide for evaluating security barriers, alarms and surveillance systems;

A Sandia Corp. outline of a sophisticated computer code for determining the best route for a plant saboteur to follow.

The purpose of the publications, federal officials say is to help agencies and operators of nuclear plants and other installations evaluate security at their sites. Because the reports describe security measures in general rather than at specific sites, officials believe they would not be of great help to potential saboteurs.

Even so, many involved in compiling and publishing the reports say they are uneasy about their ready availability.

"I don't think it's really increased the risk of a break-in at a particular plant, but if I had my druthers I would not have distributed it publicly," says Robert Clark, chief of the NRC's safeguards branch.

"The bureaucratic answer is that we've followed all the necessary guidelines and the information released is okay," says William Myre, Sandia's nuclear security systems director. "My personal answer is yes, I'm concerned, and yes, we do worry about it."

"These kinds of things make us very uneasy," says Frank Graham of the Atomic Industrial Forum, a group that represents the nuclear industry. "It could put ideas into a lot of people's heads."

"It's the old story," says Leonard Weiss, staff director of a Senate nuclear proliferation subcommittee that has held a series of hearings on government distribution of potentially sensitive security information. "Any one piece of information by itself is nothing, but when you see it all together (in a manual), you begin to get a picture that could be very helpful to a terrorist."

The Department of Energy and the Nuclear Regulatory Commission have review groups that cleared the reports for public release.

"It's a double-edge sword," said one DOE safeguards official, who asked not to be named. "We're trying to get important information to the DOE and NRC communities without tipping off the wrong people."

NRC officials say "Barrier Penetration Database" was reviewed and cleared last year for public release. But some concede they would be more comfortable if the 48-page handbook were not on public shelves.

Written by two nuclear safeguards experts at Brookhaven National Laboratory on Long Island, the handbook analyzes methods of breaking through 32 security fences, walls, roofs, ceilings, floors, doors and window at plants. For the serious student, there's a bibliography of other, more detailed manuals plus an appendix that tells how long it takes to walk, run or climb a ladder while carrying 30 pounds of burglar tools.

Coauthor Anthony Fainberg says the handbook's material is drawn largely from Sandia Labs' 300-page "Barrier Technology Handbook." He contends the information is "more or less common knowledge -- you can find these barriers everywhere, not just at nuclear plants."

Fainberg and NRC officials say the data would be of little help to trained terrorists, who know how to break through physical barriers. But Graham of the Atomic Industrial Forum says he's more concerned about amateurs the handbook might inspire.

"There are a lot of people out there who want to make statement," Graham said. "They may go to the White House steps to protest nuclear power, and then again they may come and try to break into a plant".

Officials at the National Technical Informationn Service (NTIS), which handles 70,000 new government publications each year, share Graham's concern.

"It seems to me that someone might use this thing for the wrong purpose," says NTIS Director Melvin S. Day, who says his agency plans to raise the question with the NRC.

NTIS officials recall only two other instances in recent years when releasing reports was challenged. The most noteworthy came last year, when the DOE quietly withdrew 16 documents concerning atomic energy and the design of the first atomic bomb from NTIS shelves after the nuclear proliferation subcommittee, chaired by Sen. John H. Glenn (D-Ohio), urged a review.

NTIS Deputy Director Peter Urbach said the 16 reports -- some of which were publicly available for 10 years or longer -- are still under review, and some way eventually be returned to the shevles.

A reporter's recent visit to the Springfield headquarters of NTIS turned up serveral other reports similar to the Brookhaven and Sandia barrier handbooks:

"Capability for Intrusion Detection at Nuclear Fuel Sites," prepared for the NRC by the Army's mobility Equipment Research and Development Command at Fort Belvoir, is a study of security vulnerabilities at three NRC-licensed nuclear fuel facilities.

In order to keep the report unclassified, its authors don't name the sites and instead combine them into a composite facility. Still, the report provides a detailed analysis of microwave detectors electronic sensors and other highly sophisticated surveillance and alarm equipment.

"Guide for the Evaluation of Physical Protection Equipment," one of a series of security studies done for the NRC by the Mitre Corp. of Bedford, Mass., also assess an array of security equipment, including fences, doors locks and detection and alarm systems, and discusses methods for breaching them.

"MINDPT: A code for Minimizing Detection Probability Up to a Given Time Away from a Sabotage Target," was written by a Sandia analyst for the Department of Energy The code describes a method for safeguards experts to develop computer codes. The codes' purpose: to find paths through nuclear fuel sites that would "minimize an adversary's chances of being detects until it is too late for guards to respond and interrupt the sabotage activity."

"Barrier Technology: Perimeter Barrier Penetration Tests" and "Perimeter Intrusion Detection and Assessments Systems," also prepared by Sandia for DOE, assess physical security barriers and detection systems at the perimeters of nuclear sites and include photos, graphs and charts on how the security systems are set up.

Federal officials contend these publications would be useless to intruders unless they also obtained security plans for specific sites. Those plans, officials insist, are under lock and key.

NRC officials also insist nuclear plant security has been considerably improved since 1977, when a General Accounting Office report called security "at best, inadequate." The GAO said plants' security precaution varied widely and said guards were generally badly trained and ineffective.

Brookhaven safeguards expert Fainberg agrees with the NRC that security has improved. But he cautions that although "on paper, things look very good . . . there are examples of security breaches that don't make our feel terribly optimistic."

Fainberg's conclusion: "Security is much better than it used to be, but probably not as good as it should be." CAPTION: Picture, "Barrier Penetration Database" contains material some say shouldn't be public.