On Feb. 20, 1933, with the help of a mustachioed federal prosecutor in New York named Thomas E. Dewey, the U.S. government impounded the manuscript of a book on national security grounds for the first -- and still the only -- time in American literary history.

Its unexciting title was "Japanese Diplomatic Secrets: 1921-22," and it had been stitched together by a free-lance journalist. But the prospect of its publication had put the Justice, State and War departments in a panic for months because the true author -- the man who supplied the information -- was the father of American cryptography, Herbert O. Yardley, founder of the nation's first "Black Chamber," the man who had broken the Japanese code in 1919.

The seizure was of dubious legality at best. The espionage statute cited as the excuse for such action required proof of intent to injure the United States. Yardley, by all accounts, had no such intention. But Dewey threatened prosecution unless Yardley stopped trying to get the manuscript published. Finally, Yardley relented.

The sequence marked the origins of the law now being invoked by CIA Director William J. Casey in his warnings of possible prosecution of The Washington Post and other news organizations for publishing information about communications intercepts reflecting U.S. code-breaking capabilities.

It was not until 1950 that Congress enacted the law, but it really began with Yardley. A brilliant Indiana boy, he came to Washington in 1912 as a State Department telegrapher and taught himself cryptography. He got his chance in World War I when he was commissioned a lieutenant in military intelligence and assigned to take charge of MI8, the Cipher Bureau.

"MI8," historian David Kahn wrote recently, "soon discovered spy letters written in invisible ink, solved German diplomatic codes and ciphers, and found specialists to read obscure German shorthand systems."

By war's end, Yardley was pressing for a permanent peacetime organization "for code and cipher investigation and attack." His wish was granted, and by the summer of 1919 he had set up shop in a rented New York brownstone -- "a suitable place," he later wrote, "where the famous American Black Chamber could bury itself from the prying eyes of foreign governments."

One of Yardley's first assignments was to crack the codes of Japan and within a few months Yardley had done it, out of a sound sleep. The solution came in plenty of time for the Washington Armament Conference of 1921-22, a series of postwar arms control talks at which Japan was hammered down to a fleet whose size, thanks to Yardley's Black Chamber, the United States knew it would accept.

Peacetime work dwindled after that and so did Yardley's budget, by then almost totally funded by the State Department. In 1929, President Herbert E. Hoover's new secretary of state, Henry L. Stimson, was provided with a few selected translations so he could become acquainted with the Black Chamber's skills.

Stimson was shocked. "Gentlemen," he later pronounced, "do not read each other's mail." Having solved the codes of some 20 nations, the Black Chamber was forced to shut down.

With the Great Depression staring at him, a devastated Yardley went back home to Indiana, and decided to write a book. He thought that Stimson had made a terrible mistake and that public reaction might cause the government to think again about what it had done. Besides, he reasoned, "now that the Black Chamber has been destroyed, there is no valid reason for withholding its secrets."

A gifted writer, Yardley produced a bestseller, "The American Black Chamber." Advance serialization in The Saturday Evening Post fueled the sales. The book, with chapters devoted to "Japanese Secret Codes" and "The Washington Armament Conference," was an especially big hit in Japan where the Foreign Ministry became the scapegoat.

The Saturday Evening Post announced the series March 28, 1931. "Emergency meetings were called in Washington to decide on a course of action," wrote James Bamford, author of "The Puzzle Palace," a 1982 history of the code-breaking National Security Agency and its predecessors. "Prosecution was considered, but rejected, because it was felt that such a trial would be compromising as well as embarrassing to the government. Suppression was also considered, but, there being no precedent or legal basis, this too was rejected."

The more things change, the more they stay the same. Central Intelligence Agency Director Casey, sources say, asked the Justice Department recently to seek a court order restraining The Washington Post from publishing an article on U.S. intelligence capabilities, but was turned down. The Post has not decided whether to publish the article and has declined to comment on it. Casey has warned The Post that he will recommend criminal prosecution if the article is published.

In the 1930s, the secret that the Japanese code had been broken was already out, but the government was determined not to allow Yardley an encore. He went on the lecture circuit at first, touting U.S. successes in cryptology and warning of a bleak future without them. But as Bamford has written, Yardley wanted to be accepted as a serious historian too, and he hit on the Japanese role at 1921-22 conference as the subject for his next book. He had a cache of thousands of intercepted messages to be culled for the text.

Yardley hired a free-lance journalist known for her speed, and by the summer of 1932, rumors of a new Yardley book were circulating. This time the publishing business played ball with the government. Yardley's first publisher, Bobbs-Merrill, rejected the manuscript and alerted the Justice Department. His agent then turned to the Macmillan Co. Its president, a former military intelligence officer, delivered the manuscript to a federal grand jury at Dewey's behest. It remained in federal custody for another 47 years, with portions still marked secret.

Congress, meanwhile, moved to erect a legal barrier against a recurrence. The House passed a sweeping proposal making it a crime to disclose any government document, whether classified or not, so long as its release could be shown "prejudicial to the safety or interest of the United States." But news organizations woke up at that point and demanded revisions.

The result, signed into law by President Roosevelt on June 10, 1933, was what is now Section 952 of Title 18 of the U.S. Code. It makes it a crime, punishable by up to 10 years in prison and a $10,000 fine, to disclose any official diplomatic code or any material that had once been prepared in such code. The law remains on the statute books, but it appears to be a dead-letter. No one has ever been prosecuted under it, perhaps because it is so sweeping in its coverage of material prepared in diplomatic code and at the same time so restrictive, covering only "diplomatic" codes.

Congress, in any case, rested until the end of World War II when it took up a bill to safeguard "military information." With so many people leaving for civilian life, Rep. Francis E. Walter (D-Pa.), one of the chief advocates, said in floor debate on Oct. 25, 1945, "it certainly seems to me that their lips should be sealed when it comes to the discussion of some of the things that made possible the winning of the war."

Critics, however, pointed out that the measure went far beyond its professed intent of protecting U.S. code-breaking and cryptographic successes and played hob with the First Amendment by covering everything that had been transmitted in code, any code, U.S. or foreign. Several critics said it would have kept Gen. Johnathan Wainwright, the hero of Bataan and Corregidor, from publishing his memoirs. Others objected that it would condemn the Washington news corps to a life of official handouts.

"This bill is so broad that it strikes at the very roots of freedom of speech and freedom of the press in this country," said Rep. Clarence J. Brown (R-Ohio). "No one objects, of course, to protecting the secrets of our military codes or equipment, but why stifle all information as to that which has gone before?"

Rep. Charles A. Halleck (R-Ind.) agreed. "We fought a very successful war over the past four years," he said. "We have just won that war. We did not find this sort of legislation necessary then."

The measure failed to pass, but the military establishment, unhappy with the laws at its disposal, kept trying.

"At present, there are two acts affording limited protection to cryptographic information," acting Secretary of the Navy W. John Kenney said in a 1948 letter to Congress. "These are the Espionage Act of 1917 and the. . . act of June 10, 1933. Under the Espionage Act, unauthorized revelation of information can be penalized only if it can be proved that the person making the revelation did so with the intent to injure the United States. Under the 1933 act, only diplomatic codes and messages transmitted in diplomatic codes are protected."

Kenney submitted a bill that he said would meet earlier objections since it would not penalize disclosure of "information transmitted in United States codes and ciphers."

The Senate accepted an amendment suggested by the American Society of Newspaper Editors and passed what is now Section 798 of Title 18, the so-called "Comint (communications intelligence) law" a few months later. The House followed suit on May 1, 1950, following a Judiciary Committee report that summed up the measure in a a single sentence:

"The present bill is designed to protect against knowing and willful publication or any other revelation of all-important information affecting United States communication intelligence operations and all direct information about all United States codes and ciphers."

The House report then proceeded to give a misleading bit of history about Yardley, asserting that his first book had caused "irreparable harm" and suggesting that the Japanese had made their codes so complex that the United States was helpless on the eve of Pearl Harbor.

"It can be said," the House report stated, "that United States inability to decode the important Japanese military communications in the days immediately leading up to Pearl Harbor was directly ascribable to the state of code-security consciousness which the revelations of a decade earlier had forced on Japanese officialdom."

In fact, William F. Friedman, who some have called "the world's greatest cryptologist," had broken the new Japanese Foreign Office Code for the Army Signal Corps in 1936 and then the more-complex Purple Code on Sept. 25, 1940. Washington knew what was coming that December day in 1941, even though it did not get word to Hawaii in time.

The 1950 Comint law has never been used against a news organization. It was first invoked in 1954, according to Bamford, when a former NSA employe, Joseph S. Petersen Jr., was prosecuted for passing secrets to a World War II friend from the Netherlands who was stationed in Washington. Petersen was sentenced to seven years in prison.

Its most recent use is in connection with the case against Ronald W. Pelton, a former NSA communications specialist awaiting trial as an alleged Soviet spy.

The most celebrated case involving Section 798 came in 1977 when Christopher Boyce, a 24-year-old code-room clerk at a West Coast firm with with a direct link to the CIA, was accused with his boyhood friend, Andrew Dalton Lee, of selling thousands of secrets to the Soviet Union over a two-year period. Government prosecutors tried to limit the public disclosures by producing testimony about only one project that might have been compromised, a CIA communications satellite plan code-named "Pyramider" that never got past the drawing board.

Sentenced to 40 years in prison, Boyce appealed, contending, among other points, that the information in the Pyramider documents was old stuff, available in standard engineering texts, and thus improperly classified. The 2nd U.S. Circuit Court of Appeals in San Francisco rejected the claim with a ruling that underscored the bite of Section 798.

The law makes it a crime for anyone to transmit -- knowingly and willfully -- to an unauthorized person, or publish "any classified information" concerning U.S. or foreign communications intelligence activities or "any classified information. . . obtained by the process of communications intelligence from the communications of foreign governments."

"Under Section 798," Circuit Judge Shirley Hufstedler wrote in the Boyce case, "the propriety of classification is irrelevant. The fact of classification of a document or documents is enough to satisfy the classification element of the offense."

"It's a tough statute," Bamford told a reporter. "It's the closest thing to an Official Secrets Act that the United States has."