Pearl Harbor Betrayed

by Michael Gannon

  The burn and destroy signals were decrypted by Army and Navy cryptanalysts, and on 3 December OpNav sent two radio messages to Kimmel paraphrasing their contents. This was one of the first information items based on Magic transmitted to CINCPAC since the previous July, and, in a curious lapse of security, the second of the two messages, initiated by Safford, included not only the phrase “Purple machine” but also the Tokyo message serial number (2444). Kimmel and Layton puzzled over the expression “Purple,” but were later advised by fleet security officer Lt. Comdr. Herbert M. Coleman, a recent arrival from Washington who was in the know, that it was “an electric diplomatic coding machine” used by Japan. In Kimmel’s mind, the messages bore little relation to Pearl Harbor. Cipher and code destruction worldwide would be part of the Japanese war preparation for invading Southeast Asia, as the 27 November “war warning” had predicted Japan would do. And Kimmel noted that the messages did not order diplomatic and consular posts to destroy all their codes and ciphers. Nor was Kimmel surprised when, on 6 December, he learned from Layton that the consulate at Honolulu had been burning papers for the past two days. He himself was sent an instruction from OpNav on the same day authorizing him, “in view of the international situation,” to authorize the destruction of secret and confidential documents on Midway, Wake, and the outlying islands.73

  EIGHT

  IMPERILED

  Even on the morning of December 7, four or five hours before the attack, had the Navy Department for the first time seen fit to send me all this significant information, and the additional fact that 1:00 P.M., Washington time, had been fixed for the delivery of the Japanese ultimatum to the United States, my light forces could have moved out of Pearl Harbor, all ships in the harbor would have been at general quarters, and all the resources of the fleet in instant readiness to repel an attack.

  The Pacific Fleet deserved a fighting chance.

  Rear Admiral Husband E. Kimmel, USN (Ret.), 1954

  What later became the most controversial, though inconsequential, of the Japanese messages intercepted during the weeks immediately prior to 7 December were two so-called Winds Code transmissions from Tokyo to Washington on 19 November, which, because they were sent in the low-priority J-19 cipher, were not decrypted and translated until the twenty-sixth and twenty-eighth of that month. The messages established a code that would be followed in advising the Washington embassy essentially which course or courses of war Japan had decided to take. The embassy was to monitor all Japanese short-wave voice radio news broadcasts around the clock and listen for mention of one or more weather conditions that identified the nation or nations with which Japan was “in danger” of severing diplomatic relations: The phrase higashi no kazeame (east wind rain) would point to the United States; kitanokaze kumori (north wind cloudy) would point to the Soviet Union; and nishi no kaze hare (west wind clear) would point to Great Britain.1 The code was devised as a means of communicating this information when and if commercial international transmission facilities such as those operated by the Mackay System and Radio Corporation of America (RCA) were shut down.

  The War and Navy Departments took the winds code more seriously than they did the more foretokening bomb plot messages, if one may judge from the amount of time and the number of personnel that were devoted to monitoring its “execute.” Several intercept stations of both services were put on twenty-four-hour alert for the key phrases. Teletype traffic from those stations increased from three to four feet per week to 200 feet per day.2 Six three-by-five prompt cards containing the phrases and their meanings were distributed by Admiral Noyes to the Navy’s Magic recipients. The Army’s General Miles and Colonels Bratton and Carlisle C. Dusenbury were equipped with similar cards. The watch officers manning teletypes from the intercept stations were authorized to signal reception of an execute directly to one of the card-carrying senior officers, bypassing ordinary channels.

  In another unusual step, Bratton and Dusenbury, who knew that Short had no access to Magic, either raw, gist, or paraphrase, but who thought that Rochefort’s Station Hypo at Pearl was decrypting the Purple, J-19, and other traffic, drafted a signal, approved by Miles, to Rochefort’s opposite number at Fort Shafter, Col. Kendall J. Fielder (G-2), which read: “Contact Commander Rochefort immediately thru Commandant Fourteen Naval District regarding broadcasts from Tokyo reference weather. Miles.”3 By this means the officers apparently sought to circumvent Marshall’s directive not to give Magic to theater commanders. Rochefort would give it. They were wrong on two counts. Rochefort was not Magic-equipped, though he had been sent the winds code information on the twenty-eighth—“our first piece of Magic diplomatic intelligence in four months,” Layton later wrote4—and had been asked by Noyes to listen for a winds-execute from his post, which in fact he never heard. On the second count, Rochefort, like Layton, was forbidden by Navy Department regulations to pass on to Army personnel any intelligence received from Washington, unless specifically directed to do so.5 For example, Kimmel did not tell Short about the code-destruction messages.

  According to Safford’s testimony before the JCC, an execute message came into the Navy’s Cheltenham Station M in Maryland beginning at 0830 Washington time on Thursday 4 December (elsewhere he says the third). The source was an overseas news broadcast from station J-A-P in Tokyo on 11980 kilocycles. From Cheltenham the execute was teletyped to the Navy Department (OP-20-GY) where the “weather conditions” were translated by Kramer. According to Safford, Kramer rendered the code phrases in pencil or crayon directly onto the incoming teletype sheet: “War with England (including NEI, etc.); war with the U.S.; peace with Russia.” Kramer testified that the message arrived on the morning of the fifth, that he did not recall writing on the teletype sheet, and that he definitely would not have used the word “war.” Safford stated the entire message was two hundred or so words in length; Kramer said that it was not more than a line or two. These conflicts of memory were but the beginning of a long and bizarre series of contradictions, inconsistencies, changes in testimony, missing documents, and charges of conspiracy to follow from 1944 through 1946.

  Among the more notable events in that bleak procession were: Kramer insists in 1944 that the teletype sheet contained the phrase meaning a break in relations with the United States, then in 1945 testifies that he is “under the impression” that the phrase he saw was the one that meant instead a break of relations with “England and possibly the Dutch”;6 allegations swirl that the Navy Department bureaucracy, acting in self-protection, brought pressure to bear on Kramer to amend his testimony; the one alleged copy of the execute and all the prompt cards carried by senior officers disappear from the Navy’s sequentially numbered message file sometime before 1945; Japanese communications officers deny after the war that any winds-execute was sent; British and Dutch witnesses come forward to say that they did receive it; five U.S. Navy radiomen from intercept stations separately give sworn affidavits that they knew nothing about the winds code or its execute, each adding in identical (formulaic?) language, “until I read about it in the newspapers”; and one radioman named Ralph T. Briggs, who was on duty at Cheltenham M in December 1941, is not asked by the JCC to sign an affidavit or to testify but comes forward in January 1977 to say that he intercepted the execute and was prevented from saying so by his commanding officer, Captain John H. Harper, who allegedly told him, “Someday you will understand the reason for this.” Furthermore, finding in 1960 that all his signal sheets for the first week in December are missing from the station’s records, Briggs writes another entry specifying his receipt of the “winds message warning code” onto the original duty personnel log sheet, not for 4 but for 2 December.7

  For writers unsatisfied with the knowledge that history is often untidy, like an Agatha Christie mystery without an orderly ending, and who search instead for the logic and clarity that conspiracy provides, here are plot lines aplenty. Using this material it would not be difficult to construct a conspiracy theory according to which the
Navy Department’s top brass tried to prevent a warning message that should have been sent to Kimmel and Short, but was not, from coming to public notice; or, if it did come to light, to extinguish it. Certain popular authors and so-called revisionist historians have taken that route. Others, historians like Gordon W. Prange and Roberta Wohlstetter, have tended toward the position articulated by Vice Admiral H. Kent Hewitt, who in 1945 made an independent investigation of this and other matters relating to the Pearl Harbor attack at the behest of then Secretary of the Navy James V. Forrestal. After exhaustive examination of the winds affair, Hewitt concluded that if an execute had been received on the day (days) specified—and he concluded that none did, the preponderate weight of evidence being that Safford was “honestly mistaken”—“such fact would have added nothing to what was already known concerning the critical character of our relations with the Empire of Japan,” and, furthermore, “would not have conveyed any information of significance which the Chief of Naval Operations and the commander in chief, Pacific Fleet, did not already have.”8 This writer tends in the same direction.

  The “winds” was not the only new code at play in this period. After 27 November the Japanese employed ingo denpo, or the “hidden word” code. A plain language message in that code was translated by the Navy on the morning of 7 December, and it was placed in a Magic pouch that Kramer delivered to Stark that same morning; it may also have reached Marshall’s office in the Army’s first Sunday delivery. As translated it read: “Relations between Japan and England are not in accordance with expectations”9—which was no more revelatory than the putative winds execute. However, a few days after the Pearl Harbor attack, Army cryptanalyst Col. William F. Friedman discovered that the translator had made two errors, and that the message should have read: “Relations between Japan and England and the United States are on the brink of catastrophe”10—which should have triggered the tocsins in War and Navy, particularly since the departments had an even more incendiary alarum on their desks that Sunday morning, as will be shown.

  While fleet and Army intelligence officers in Hawaii were not informed of the Honolulu-Tokyo espionage traffic that was being read in Washington, they did come into possession of one curious piece of transoceanic communication that put them on their guard temporarily. It was a lengthy, two-hundred-dollar radiotelephone call on 3 December from a Tokyo newspaperman to Dr. Motokazu Mori, a Japanese dentist in Honolulu. (Some accounts state that the recipient of the call was Mrs. Mori, but that appears unlikely since in the caller’s last remark he says, “Best regards to your wife.”) The caller was responding, he said, to a telegram sent him by one or both Moris. Tapped by the Federal Bureau of Investigation, the conversation was translated on the afternoon of the sixth and given to Special Agent in Charge Robert L. Shivers, who had long had Dr. Mori on a suspect list. To Shivers’s eyes questions addressed to Mori from the Tokyo caller about daily flights of aircraft, ship movements, the presence of searchlights, and the number of sailors in Oahu appeared to be more than ordinary civilian talk, and Mori’s mention of “flowers in bloom,” the hibiscus, poinsettia, and Japanese chrysanthemum in particular, seemed suspiciously like a code.11 Later that afternoon, Shivers showed the transcript to Army Lt. Col. George W. Bicknell, assistant G-2, who thought it worrisome enough to pass upstairs to his chief, Lt. Col. Kendall J. Fielder, at the latter’s residence. Fielder suggested that they show the transcript to Short, who lived next door. For about three-quarters of an hour the three officers discussed the transcript and decided that they could not “attach any military significance to it.”12 What most struck the three in the end was the very ordinariness of the military information given; and the fact that the hibiscus, poinsettia, and chrysanthemum were in bloom.

  * * *

  In the Station Hypo basement offices, known to its occupants as the “dungeon,” beneath the Administration Building of the Fourteenth Naval District, Rochefort and his cryptanalysis team brooded over the blank pages that once were filled with data on the physical locations of ships and air units that constituted the Japanese Combined Fleet. Prior to 26 November it had been possible to track at least some elements of the fleet by their call signs, which were not enciphered, by radio direction-finding (HF-DF), and by the signature “fists” of individual Morse senders. On that date, however, a general silence fell over the fleet and Rochefort’s data source on the operational forces blacked out, particularly where the carriers were concerned. Perhaps, Rochefort reasoned, some carriers were moored in harbor and were communicating with headquarters by landline. In any event, his Communication Intelligence Summary issued on the twenty-seventh stated: “Carriers are still located in home waters.”13 That was two days after six of the carriers, including the flagship Akagi, sortied from the Kuriles to begin their 3,150-mile voyage through the Vacant Sea.

  On the twenty-eighth Rochefort reported “no indication of movements of any Combined Fleet units.” On the thirtieth he thought he heard a tactical radio circuit in operation between Akagi and several supply ships (which would have been a violation of radio silence). On 1 December he stated, “As to the First [Air] Fleet [the core of the Hawaii Striking Force], nothing indicated that it was operating as a fleet outside of Empire waters.” It did appear, though, that the submarine force had taken up positions eastward of Yokosuka-Chichijima, Japan, in the Marianas, north-northeast of Guam. On the second: “The First Fleet appeared relatively quiet.… Almost a complete blank on the Carriers today.… Not one Carrier call has been recovered.… Carrier traffic is at a low ebb.” On the third there was “no information on Submarines or Carriers.” Nor was there any over the three days following, other than that the Commander Submarine Force “is definitely in the Marshalls.” In the next month’s Roberts Commission hearings, Rochefort told chairman Justice Owen J. Roberts and commissioner Rear Admiral Joseph M. Reeves that he had speculated that the carriers Akagi and Kaga (Carrier Division 1) were in the Marshalls, and that there might be another division in Thai waters.

  ROBERTS: But if there were two in each of those places it would have left them a maximum of six [unaccounted for]?

  ROCHEFORT: Yes, sir, including the converted carriers it left them that.

  REEVES: There were either four or six whose whereabouts you didn’t know?

  ROCHEFORT: Yes, sir, but these we did—

  REEVES: Well, when did you hear of these carriers again?

  ROCHEFORT: The 7th of December, sir.14

  Equally deaf to the presence of the six-carrier Striking Force in the mid–North Pacific was Station Negat, or Code and Signal Section, the naval code-breaking unit of OP-20-G, the Operational Intelligence Center in the sixth wing of Main Navy. There a memorandum on Japanese fleet locations, produced on 1 December, placed all ten carriers in home waters at Kure and South Kyushu.15 So Washington was no help. Like Rochefort, Layton was concerned that some of the carriers had left their nests, but with no radio traffic from them excepting the unexplained Akagi–supply train blip, and with an unexpected change of call signs made by Japan on 1 December, there was nothing definite that Layton could report in his daily intelligence presentation to Admiral Kimmel. During the morning briefing of 2 December, as Layton remembered it, Kimmel responded:

  “You mean to say that you are the intelligence officer of the Pacific Fleet and you don’t know where the carriers are?”

  “No, sir, I don’t.”

  He then said, “For all you know, they could be coming around Diamond Head [on Oahu], and you wouldn’t know it?”

  I answered, “Yes, sir, but I hope they’d have been sighted by now.”16

  Had Rochefort and his talented team of cryptanalysts at Hypo not been limited by Washington to breaking, if they could, the Japanese so-called Flag Officers Cipher—which proved month after month to be a fruitless assignment—and had they instead been tasked to penetrate the Japanese Navy’s operational cipher, known as the Fleet General Purpose System, a “five-numeral” cipher that became known eventually as J
N-25-B, Hawaii’s knowledge of Japanese operational intentions in October, November, and December likely would have advanced into a galaxy of irresistible clues. But the authority to attack JN-25 was held tightly in Washington.

  Navy cryptanalysts in Negat had acquired a commendable success rate in breaking into the Japanese diplomatic ciphers, starting with high-grade Purple and descending through J-19, J-17K6, J-18K8, J-22, then PA-K2 and LA, and they had made a start at JN-25B recoveries. “On 4 January 1941,” as one of Rochefort’s traffic analysts, (later) Captain Jack S. Holtwick, Jr., remembered, “it was reported that about 2,000 values had been recovered out of 33,000 possible” in JN-25B—which was slightly more than 6 percent.17 Captain Safford in OP-20-G stated from memory in August 1970, “By December 1st 1941, we had the code solved to a readable extent.”18 The Philippines’ Station Cast (which had moved in November to the island fortress of Corregidor), British cryptanalysts at Singapore, and the Dutch crypto unit in Java were given equal credit for the accomplishment. When, after 7 December, Hypo was at last relieved of the Flag Officers Code (which never was cracked sufficiently to provide intelligence) and given shared responsibility for JN-25B, the list of code group recoveries sent it from Washington numbered about 10 to 15 percent of the total—meaning that about 10 to 15 percent of each message broken surrendered some information, though not necessarily text.19 The percentage did not mean that 15 percent of the intercepts could be read, or even that 15 percent of the words in a given message could be read.

  It is important to recognize that no naval operational message text in JN-25B was read by the United States prior to 7 December, several revisionist writings to the contrary.20 Ten to 15 percent of code groups was not sufficient to read message texts and very little usable intelligence derived from that level of penetration. Our living authority for that asseveration is Rear Admiral Donald M. “Mac” Showers, USN (Ret.), who joined the Hypo team as an ensign in February 1942 and stayed until January 1945. He crowned his career as chief of staff of the Defense Intelligence Agency. At a colloquium on Admiral Kimmel held in Washington on 7 December 1999, Showers told the audience, “If you would write a letter only using ten percent of the words in the dictionary, I challenge you to complete your task. We were not reading ten percent or fifteen percent of the text of JN-twenty-five messages in those days [prior to Pearl Harbor].”21