Last year the Supreme Court gave the CIA absolute authority to keep all of its sources of information secret, even if the sources are not confidential and even if the information is not classified. Writing for a 7-to-2 majority in CIA v. Sims et al., Chief Justice Warren E. Burger declared that Congress had been quite "plain" about the matter when it created the CIA in 1947 and that the "legislative history" clearly showed that the CIA director had been given "very broad authority to protect all sources of information from disclosure."

Now it appears that Burger's opinion was based on a shaky line of reasoning indeed. There is no "legislative history" behind the provision the high court so roundly embraced -- in fact, no indication that Congress ever debated or discussed the matter. And what history there is outside congressional records indicates that the provision in question was an anti-CIA restriction devised 41 years ago to keep the new kid on the intelligence community block from popping off about a very narrow set of secrets.

The scholarly detective work supporting those findings was done by Thomas F. Troy, a highly respected, retired CIA historian who has been, for the past four years, editor of a bimonthly newsletter and book review, the "Foreign Intelligence Literary Scene."

Troy, it should be emphasized, is no bleeding-heart critic of the Central Intelligence Agency. He was as delighted with the high court's decision as the CIA was. It's just that he can't abide bad history, as he makes clear in the latest edition of his newsletter.

The controversy goes back to congressional enactment of the National Security Act of 1947, which established the CIA. The new agency was assigned "to correlate and evaluate intelligence relating to the national security, and provide for the appropriate dissemination of such intelligence within the government" -- with three restrictions.

The first was the so-called anti-Gestapo clause, providing that the agency "shall have no police . . . or internal security functions." The second reassured the other intelligence agencies of the government, the Army's and Navy's especially, that they could continue their work.

The last proviso stated that "the director of central intelligence shall be responsible for protecting intelligence sources and methods from unauthorized disclosure."

Now jump ahead to 1977. The Freedom of Information Act had recently been strengthened; indignation was still high in some quarters about congressionally exposed excesses by the intelligence community.

Two men, Washington attorney John C. Sims and Sidney Wolfe, director of the Public Citizen Health Research Group, sued under FOIA for the names of individuals and institutions that had done research for the CIA's MK/ULTRA Project. Financed from 1953 to 1966 in response to Soviet and Chinese brainwashing tactics, it eventally extended to at least 80 universities, research foundations and similar institutions. It also became notorious for having produced, as Chief Justice Burger noted, some "untoward results," including the deaths of at least two unwitting individuals given dangerous drugs such as LSD.

The CIA refused to supply the requested information, saying it was required under the 1947 law to "protect intelligence sources and methods." The agency's lawyers took the broadest possible position -- that an "intelligence source" was any source of information at all, including even an article from, say, The Washington Post or the London Sunday Times.

The U.S. Court of Appeals here disagreed and formulated its own definition, one that would have required much of the information in the case to be released. The names could still have been kept secret if the CIA had chosen to classify them on grounds that their disclosure would damage the national security. But the CIA refused, looking for a showdown on the "sources and methods" clause.

The case reached the Supreme Court, which handed the agency a complete victory. It rejected the appellate court's definition and came up with its own, even going so far, as Troy points out, to label it "the definition . . . fashioned by Congress." Congress, Burger wrote, "simply and pointedly protected all sources" of information the agency needs.

Indeed, the chief justice emphasized the theme several times, speaking at one point of "the express intention of Congress" in giving the CIA such "sweeping power" and talking at another point of the "plain statutory language."

Despite all that rhetoric, Troy notes, neither Burger "nor anyone else cited one scintilla of evidence that any congressman, any committee or any caucus -- or anybody anywhere -- ever thought or said anything about the words 'sources and methods' . . . Congress never gave those words the time of day."

In fact, the "sources and methods" proviso was concocted in early 1945 following joint military discussions in which Army and Navy officials were opposed to any new intelligence agency. Troy says that when they recognized the inevitability of having to share information with "the new, untried, distrusted CIA," they decided to concentrate on keeping their special secrets safe.

According to Troy's research, Rear Adm. Joseph R. Redman, then-director of naval communications and a man said to be "jealously intent" upon maintaining his turf, devised the formula in a Jan. 8, 1945, memo to a planning group called the Joint Strategic Survey Committee. Redman's office was involved in the wartime business of intercepting, decoding and disseminating enemy radio communications (comint), and, Troy says, that is what the admiral wanted to protect. He took his cue from a June 1944 military study that, in suggesting legislation to counter damaging "comint" leaks, recommended that specific references to "radio intelligence" be avoided and broad "cover" language used in its place. So was born the phrase "sources and methods."

To comint experts, Troy notes, the word "sources" referred to particular kinds of intercepts and thus, the Japanese army, navy, and air force were so many "sources" of intelligence. Similarly, "methods" referred to code books, ciphers and the like.

The Joint Chiefs of Staff incorporated the idea in an overall plan that they approved Sept. 18, 1945. It provided for the sharing of military secrets with the new agency, and in the next breath, made the CIA responsible "for fully protecting intelligence sources and methods which, due to their nature, have a direct and highly important bearing on military operations." Simply put, Troy says, "the phrase was imposed upon the new CIA by a U.S. military fearful that the new agency might compromise comint."

The language was later abbreviated, evidently by military draftsmen, and finally enacted by Congress in 1947, without discussion, without debate.

"Knowing no history and reading only words, the Supreme Court came up with a theoretically plausible but historically untenable view of 'sources and methods,' " Troy concludes.

"What the court would have us believe is that an alert, all-seeing Congress, worried about our intelligence secrets, gave CIA, entrusted CIA, helped CIA and vested in CIA what CIA needed to carry out its mission -- namely, virtually unlimited power to protect all sources, no matter how commonplace and public -- even newspapers, public libraries, road maps and telephone books . . . . What the court failed to learn is that the phrase came from . . . a body of people who simply did not want a CIA . . . . "

The court's ruling, of course, will remain in force, bad history notwithstanding. The law is what the Supreme Court says it is. But perhaps a new corollary can be added: on issues involving intelligence, the law may be what the CIA tells the Supreme Court it is.