Overview
The question has never really gone away. Within weeks of 7 December 1941 there were people in Washington asking how a fleet of six carriers crossed the Pacific and arrived unannounced, and the asking has continued through eight official inquiries, a congressional investigation, and decades of declassification.
The strong version of the claim is that Roosevelt knew and stayed silent, because an attack was the only thing that would carry a reluctant country into the war. The weak version is that Washington knew enough to have warned Hawaii and did not. The two are often blurred together, and they are not the same argument. The National Security Agency's own history of the period separates them cleanly, distinguishing writers who argue that American policy was designed to provoke Japan from those who argue that specific warning intelligence was withheld from the Hawaiian commanders and then covered up.
The distinction matters because one of them is substantially documented.
A fifty-year lineage
The argument did not begin with the internet, or with Robert Stinnett. It has a traceable literature.
George Morgenstern, a Chicago Tribune editorial writer, published the first book-length version in 1947, built from a forensic reading of the congressional record. Charles A. Beard — among the most influential American historians of his generation, and president of the American Historical Association in 1933 — followed in 1948 with President Roosevelt and the Coming of the War, 1941, subtitled A Study in Appearances and Realities. Beard's charge was constitutional rather than conspiratorial: that Roosevelt governed on the appearance of neutrality while steering toward belligerency by executive action, without making the case to Congress. The book damaged his standing badly.
Charles C. Tansill's Back Door to War (1952) supplied the phrase the whole tradition is known by. Harry Elmer Barnes collected the strand into an edited volume, Perpetual War for Perpetual Peace, in 1953, and added the claim that an establishment of historians and publishers was suppressing the argument — a "historical blackout." Barnes is also part of the record in a way the tradition rarely mentions: from the mid-1950s he moved into explicit Holocaust denial, translating the French denier Paul Rassinier into English and dismissing the Eichmann trial evidence, and the Institute for Historical Review reprinted Perpetual War for Perpetual Peace in 1982.
Rear Admiral Robert A. Theobald, who had commanded Destroyer Flotilla One from the Dobbin inside Pearl Harbor on the morning itself, published the strong version in 1954: that Roosevelt knowingly withheld the warning to let Japan strike first. John Toland, a Pulitzer winner for The Rising Sun, revived it in 1982 with Infamy. Stinnett's Day of Deceit followed in 1999.
The case that has been made
Stinnett's is the most detailed. A Navy veteran of the Pacific war himself, he argued after some seventeen years of archival requests that the approaching fleet was not silent at all — that its radio traffic was picked up and direction-found, that American cryptanalysts were further into Japanese naval codes than was later admitted, and that the resulting picture was kept from Admiral Husband Kimmel at Pearl Harbor.
Around that sits an older, narrower complaint. Kimmel and General Walter Short were relieved of command in December 1941 and blamed; both spent the rest of their lives arguing they had been denied intelligence held in Washington. That grievance is not fringe, and it does not require believing Roosevelt wanted the attack.
The circumstantial frame is that Roosevelt had spent 1941 moving an unwilling public toward intervention. The best-known artifact of that reading is Henry Stimson's diary entry for 25 November 1941, describing a meeting with Roosevelt, Hull, Knox, Marshall and Stark, at which the President raised the likelihood of an attack and "the question was how we should maneuver them into the position of firing the first shot without allowing too much danger to ourselves."
The caution on that quotation is worth stating with it. Roosevelt is not quoted directly; Stimson dictated the entry from memory afterward, and the anticipated Japanese move under discussion was against Southeast Asia. R. J. C. Butow, examining it in the National Archives' own journal, read "maneuver" as diplomatic positioning rather than an invitation.
The codebreaking question, which is where it turns
Almost everything in the strong version depends on one technical claim, and the claim is checkable.
Two different Japanese systems have to be kept apart. PURPLE was the diplomatic cipher. The Army's Signal Intelligence Service broke it in 1940, it was read currently and well, and its product circulated as MAGIC. JN-25B was the Imperial Japanese Navy's principal operational code, introduced on 1 December 1940 — and it is the system that would have carried anything resembling an attack order.
The reading ability was in PURPLE. The theory needs JN-25B.
By 7 December 1941, American cryptanalysts had recovered roughly 3,800 of some 30,000 JN-25B code groups, along with about 2,500 additives — well under ten percent, and, as Stephen Budiansky put it, nowhere near enough to read current traffic. Of the Japanese naval messages intercepted between September and December 1941, some 26,581 decrypts survive, about ninety percent of them in JN-25B. Roughly 2,413 were translated. They were translated in 1945 and 1946, after the war, not at the time.
David Kahn, the historian of cryptology, drew the conclusion that does the most damage to the theory: even if that code had been fully solved and those messages read before 7 December, they would not have foretold the attack. Nothing in the intercepted traffic named Pearl Harbor as a target.
There is a live specialist argument here, and it is worth representing accurately. Timothy Wilford, writing in a peer-reviewed maritime history journal, has argued that American cryptanalysts could partially read JN-25B from around November 1941, resting on a letter of 16 November from Lieutenant John Lietwiler at Station Cast on Corregidor reporting that "we are reading enough current traffic to keep two translators very busy." Wilford's own conclusion is the honest framing of it: the evidence "buttresses the revisionist positions advanced by Toland and Stinnett, although neither author proved that the USN could read JN-25B in late 1941."
So the open question among specialists is how much fragmentary text was readable. It is not whether anyone read an order disclosing the attack. No one on any side of the argument claims that.
The most telling line on the point is Stinnett's own. In his endnotes he conceded that he had found no reliable evidence establishing how much of the code text "could be deciphered, translated, and read by naval cryptographers." The thesis's leading proponent does not claim the traffic was readable.
The winds execute
One strand of the theory has its own long history. Japan had pre-arranged an open code: weather phrases inserted into news broadcasts to signal a diplomatic break. That much is documented, and Allied warning of the scheme is on the record — a message from Batavia on 3 December 1941 and a telegram received at the State Department on 4 December both describe it. Both point at Britain and Southeast Asia. Neither mentions the United States or Hawaii.
Captain Laurance Safford, who founded and ran the Navy's cryptologic section, maintained afterward that an execute message covering the United States had been intercepted before the attack and had then disappeared. The NSA's documentary history of the controversy found that Safford gave materially different accounts to different inquiries, that the Hewitt Inquiry found no evidence from any source other than Safford himself, and that the documentary evidence he said existed "simply did not, nor did it ever, exist."
The Defense Department's published study of the MAGIC material reached the same place: there is no document in American communications intelligence files indicating receipt of any winds-code message announcing a break with the United States.
Kahn added a technical objection to the 1980s revival of the story. A genuine winds execute would have been broadcast by voice in a news bulletin. What the radioman who revived the claim described copying was Morse on a Navy weather circuit, which is not the same thing.
Where a researcher went wrong, specifically
The general charge that revisionist readings do not survive a return to the documents can be made concrete.
Toland placed two Japanese carriers some four hundred miles north of Honolulu on the strength of the diary of Captain Johan Ranneft, the Dutch naval attaché in Washington. Kahn went and read the diary. It says "beW." — an abbreviation for westerly. The northward reading came from interviews conducted later, not from the document Toland cited for it.
On the tracking claim more broadly, Kahn's objection is that a single bearing does not fix a position: the line of bearing from the Philippines runs through the Kurile Islands, from which the strike force sailed, but also through Japanese home waters, and direction-finding error at that range ran easily to ten or twenty degrees.
The part that is documented
Here is the finding that complicates any tidy dismissal, and it is the reason the theory has kept its grip.
The Army Pearl Harbor Board, reporting in October 1944, found that the War Department had held the relevant information and had not passed it on: "The War Department had the information. All they had to do was either to give it to Short or give him directions based upon it." It found Short was not kept advised of the growing tension, and was not given, on the evening of 6 December and the morning of 7 December, the critical information indicating an almost immediate break with Japan, though there was ample time.
The Navy Court of Inquiry, the same month, found Kimmel's dispositions proper given what he had, and criticised higher command for withholding important information during the critical period of 26 November to 7 December. Both reports were kept secret, and both officers were refused the courts-martial they asked for.
The congressional joint committee in 1946 found that the errors made by the Hawaiian commands "were errors of judgment and not derelictions of duty," and faulted the Intelligence and War Plans Divisions in Washington for failing to appreciate the intercepted Tokyo-to-Honolulu messages of 24 September, 15 November and 20 November — the ones concerning the berthing of ships in the harbour.
The Dorn Report in 1995 concluded that responsibility "should not fall solely on the shoulders of Admiral Kimmel and Lieutenant General Short, it should be broadly shared," while declining to absolve them.
And in 2000, Congress enacted the finding into law. Section 546 of the National Defense Authorization Act for Fiscal Year 2001, signed on 30 October 2000, records the sense of Congress that Kimmel performed his duties "competently and professionally" and that the losses at Pearl Harbor "were not a result of dereliction in the performance of those duties." The Senate had voted 52 to 47 for a non-binding version on 25 May 1999; the language that became law came the following year. Its request that the two officers be advanced on the retired lists has not been acted on by any president.
Four official findings, across fifty-six years, agree that Washington held intelligence Hawaii never received. That is not the conspiracy claim. It is also not nothing, and the two have been easy to confuse ever since.
The case against the strong version
Historians who work on the attack have been, in the main, unpersuaded, and their objections are specific rather than dismissive.
Gordon Prange, whose research remains the standard reference, put the practical objection bluntly: if the point was to be attacked, why permit the attack to land on the battle line? Wrecking the Pacific Fleet's capital ships in the opening hour is a strange way to enter a war you intend to fight.
The second objection is structural. Foreknowledge on the scale described would have required the sustained silence of a large number of officers, several of whom were later free to speak, and none of whom did.
The third is the explanation most historians actually favour, and it is not flattering to anyone. Roberta Wohlstetter's study of the intelligence failure, which won the Bancroft Prize in 1963, located the problem in noise rather than suppression: "we failed to anticipate Pearl Harbor not for want of the relevant materials, but because of a plethora of irrelevant ones." She also noted that no single person or agency ever held all the signals at any one moment, and that the handful of officials cleared for MAGIC saw the traffic briefly, had no time to review it critically, and each assumed the others had drawn the same conclusions.
Why it persists
Part of the answer is the record itself. Material stayed classified for decades, which meant each release arrived as a revelation and invited the question of what remained closed.
Part is that the official verdict genuinely shifted. Kimmel and Short were blamed in 1941 and the blame was formally qualified by an Act of Congress fifty-nine years later, which is not a reassuring sequence for anyone told the matter was settled.
And part is that a true finding sits directly beside a false one. Washington did withhold intelligence from Hawaii; four official inquiries say so. That the withholding was inadvertence rather than design is the harder thing to hold in the mind — a warned, expectant, code-breaking superpower surprised anyway is a claim about institutional failure, and institutional failure has never been as satisfying as a decision made by one man.
The document most often placed at the centre of the design argument is treated separately in The McCollum Memo.