The war might have ended and my search for witnesses might have been nearing its end, but in 1945 the controversy about the Winds Code was still going strong.
Captian Safford had testified to the Navy that a broadcast that implemented the Winds Code had been intercepted but ignored. He also claimed the records of this interception had been destroyed. So, despite all the evidence to the contrary I had dug up to date from my Army witnesses, there were still some loose ends to be tied up.
One of these was Col. Harold Doud. Before Pearl Harbor, he had been in charge of B Section of the SIS, which was the Code and Cipher Section. His duties were to supervise the “solution” (the breaking and translating) of Japanese radio intercepts for both the diplomatic and the military sides.
We discussed the original Winds Code message of November 19, 1941.1 Doud said that when the intercept was translated, the Army immediately made arrangements to monitor Tokyo radio for a broadcast that would be the so-called execute message. But, said Doud, “I did not see any execute message as thus contemplated and, so far as I know, there was no such execute message received in the War Department.”
I asked Doud to review Safford’s testimony to the Navy that the execute message had been intercepted. Safford had even gone so far as to say that Doud might have had knowledge of the execute intercept. Doud refuted Safford.
“I do not know the basis for this testimony by Captain Safford,” Doud said, “as I do not have any information of an execute message.”
Doud also confirmed for me that it took an average of two days for the Army to process the PA–K2 codes, which led me to conclude that the only way the Army could have processed the two intercepted messages of December 6 any faster would have been to have machines doing the work. This was something I would have to check out.
Another item Doud mentioned confirmed my belief in Dusenbury’s story about having received the fourteenth part of the Japanese diplomatic message on the night of December 6. This meant, of course, that there had been a truly great failure in Washington to distribute that vital fourteenth paragraph. If Washington had received the message at around midnight on December 6, we would have had an additional nine hours of warning. But for reasons yet unknown to me, we had failed to use the time.
The problem was that I couldn’t track this down all by myself. It was more than an Army matter.
The Army’s responsibility for distributing intercepted Japanese messages ended at midnight on December 6. Because of the crazy way the “boogie-woogie” distribution system was set up, the responsibility for translating the Magic decrypts became the Navy’s starting at 12:01 A.M. (or 00:01 hours military time) on December 7, 1941. Digging into what the Navy would consider its own secret internal affairs was not within my authority. But there’s more than one way to skin a cat, I told myself. The trick was finding the right way to do it. It would require time and patience.
I then took a second affidavit from Lt. Col. Frank B. Rowlett, whom I had asked to do some additional checking for me. I had asked him to study the time it took to process the messages in the PA–K2 codes, which was how the Army had designated the codes in which the two messages of December 6 from Honolulu to Tokyo were sent.
The average processing time for nineteen intercepted messages within the period of November 1 to December 6, 1941, was three and a half days. This figure was based on the number of messages that were actually “published,” or distributed throughout Army channels. In other words, there had been little or no chance that the Army—or even the Navy—might have processed a message intercepted late in the day on December 6 so that adequate warning could be gained from it.
Rowlett also confirmed for me that we were unable to read totally any Japanese army or Japanese Military Attaché codes before the outbreak of war. (Our Navy had been equally unsuccessful in breaking the Japanese J-19 naval code up to that time.)
I was interested to know how the Army had processed the all-important messages of December 6, including the fourteen-part diplomatic message. Rowlett recalled the day, saying it was the first time the Army had used a teletype to transmit intercepted material between San Francisco and Washington.
I thought this was rather unusual, what with the speed of processing seemingly so important, but Rowlett was definite that “this was the first time the Army had used teletype facilities to forward traffic to Signals Intelligence.” He had actually helped to operate the equipment in the old Munitions Building that day, and he remembered that the first call to initiate the circuit was placed sometime after six P.M. on December 6. (The situation was even more bizarre because it was a Saturday. His operation usually only worked until midday on Saturday, which meant that people had to be called in from home.)
Washington’s request for teletype service of that day’s intercepts caught San Francisco unaware. They already had forwarded all the current intercepts via air mail, and they complained that they would have to use the station’s file copy to prepare the intercepts for teletype. This required punching tape, and it took some time to get the tapes ready. To the best of Rowlett’s recollection, this intercept traffic was received from San Francisco just after midnight Washington time.
As I have said earlier in my narrative, Rowlett struck me as being one of those totally honest, precise, creative code-breaking experts. He had double-checked the chart as I had requested (see page 104). He tacitly agreed that the second message numbered 25843 was the entire fourteen-part diplomatic message and that it had been received in its entirety around midnight on December 6, give or take fifteen minutes on either side. Now I had documentary evidence that supported Dusenbury’s shocking statement to me that he had all fourteen parts of the message in hand that night, but had failed to deliver them as ordered to General Marshall.
Rowlett gave me some additional leads. I next interviewed Capt. Howard Martin, who had been the Non-Commissioned Officer (NCO) in charge of San Francisco’s station number 2 on the evening of December 6, 1941. He remembered being in his quarters at Fort Scott about eight P.M. San Francisco time when the phone rang. The man on duty at the station said that Washington had called via teletype, asking for a transmission of all that day’s intercepted traffic. It was Saturday night, Martin recalled, and he had only one man on duty. The other personnel were not easy to reach. In other words, they were off base with weekend passes, so Martin went to the station immediately and began punching tape to transmit all of Saturday’s raw intercepted traffic. (The original traffic had been air-mailed as usual to Washington at four that afternoon.) Martin was also positive that the teletype, or TWX (pronounced “twix”) machine had never been used before to send intercepts. As he blandly put it: “Because the following day the Japanese bombed Pearl Harbor, I have always associated [those] things in my memory.”
Well, well, I thought. Here we have a man who did things right on December 6. He was once a sergeant, now he’s a captain. I recalled an old saying: It’s the good sergeants who really run the Army.
In a roundabout way, I was zeroing in on what I wanted to know: the significance of the Army’s use of the TWX machine.
I then deposed Mary J. Dunning, who was currently working for the Signal Security Agency. She, too, remembered the evening of December 6 with great clarity.
Her shop in Washington usually closed down at one P.M. on Saturdays, and that day she left at the usual time to go home. Around two-thirty that afternoon, Colonel Minckler’s office phoned, asking that she return as soon as possible. By three P.M., she was back in Minckler’s office, “ready to work.” She remembered the time precisely, because later on she was asked to come back on Sunday, December 7, at the same time. She asked if she could come in at five P.M. instead, since she was planning to drive out of town.
“I can’t recall being told why we were called back to work [on December 6],” she told me, “but the general assumption was that we wanted to process traffic without delay since the Japanese Ambassador was in conference with the President.” This wasn’t exactly the case, of course, but it was close enough, I thought. It just went to show that if the Army wanted to spend the time and money to do something properly, it could darn well do it. But why hadn’t it put this type of priority on handling the intercepts before this?
Ms. Dunning recalled: “I was asked to work in the ‘cage’ [a room where machine traffic was processed—so called because it had grill-work at its entrance to restrict admittance], where I had not worked for some time. I think that as I entered the room, I was surprised to see a teletype machine. How long it had been installed, I don’t know, but it could not have been for more than a few hours, since I often had occasion to go to the door of the cage and it was clearly visible from the door.”
Around four P.M., the teletype company sent some representatives around to instruct Dunning and the others on how to use the machine. After that, everyone practiced using the TWX.
Dunning said that waiting for something to happen always seems to take longer than it actually does, and she was expecting traffic from San Francisco at any moment. She was getting hungry, but could not leave the cage because no one knew when San Francisco would begin transmitting. She couldn’t even take time to go out and buy a sandwich. She recalled asking Minckler about it, because she was joking with him “about my teaching him to operate the teletype.”
A full-bird colonel learning how to use a TWX machine, I thought. Now, that’s a refreshing change. Another good man.
When did the transmission from San Francisco end?
That is the key question, as we will see later.
“I believe I went home around midnight or 1 A.M.,” said Dunning. There was nothing more to do.
Dunning also recalled processing traffic from San Francisco, the Philippines and Honolulu that evening. “I cannot say, however, whether it came to us by teletype or not, since the Message Center had been asked to deliver traffic to us as soon as it arrived,” Dunning explained.
I was positive that the fourteenth part was received around midnight on December 6, just before Dunning went home. Once again, it was just as Dusenbury had sworn in his testimony, which has never been contradicted in the past fifty years.
Before I delve further into the matter of what happened to the fourteenth part of the all-important diplomatic message, let me conclude my investigation with my final witness, Louise Prather of the Signal Security Agency.
She, too, clearly remembered the night of December 6, 1941. As she recalled: “I was called at home and told our unit was being placed on a 24-hour basis immediately and that I should report for work at 7 A.M. the following morning, 7 December. When I arrived at the office at this unusual hour, I learned that the teletype was being operated and the reason for the urgent call had been to process this and other traffic as rapidly as possible.”
The material Prather spoke of was yet another intercept from Tokyo instructing the Japanese diplomats in Washington to deliver the fourteen-part message about the break in relations between the two countries to our State Department at one o’clock that afternoon (Washington time), which would be approximately dawn at Pearl Harbor.2 This meant the vital delivery time was in hand at least six hours before the bombs started falling.
I was concerned that the Army had not used the teletype before this to transmit important intercepts to the War Department. The Navy had been using teletypes. Why hadn’t the Army? Why hadn’t it been more on top of the situation? I regret to say that I didn’t learn the answer right away, possibly because the discovery of what had happened to the fourteenth part of the diplomatic message was more important than deciding what was wrong with the system of intelligence processing used by both services.
During the course of my investigation, I had passed on to Sonnett, the Chief Counsel for the Hewitt Investigation, what I had learned from Dusenbury about the fourteenth part of the diplomatic message. In turn, Sonnett had pressed that inquiry into areas that made some people in the Navy highly uncomfortable.
According to these Navy witnesses, the fourteenth part of the diplomatic message, the notice that the Japanese were breaking off relations, was in English, and the Navy officer on watch thought he had delivered it to the Army sometime after midnight, which also supports Dusenbury’s story.3
The message telling the Japanese diplomats in Washington to hand over the fourteen-part reply at one P.M. Sunday came in sometime between 3:05 A.M. and seven A.M. on December 7. It was processed by the Navy and sent to the Army for translation from Japanese into English at seven A.M. on December 7. The Army translated the message and returned it to the Navy at about ten that morning.4
The question that kept niggling at me was this: It appeared that the Navy had broken the agreed-upon rules for translating the Magic material. The Navy was supposed to handle these translations on the odd-numbered days of the month, and December 7 was an odd-numbered day. Why didn’t the Navy have a translator on duty from midnight to seven A.M. at the Navy Department?
The man who handled these translation duties, Capt. Alwin D. Kramer, testified to Sonnett that it would have taken him only “about two minutes” at 3:05 A.M. to translate the final and separate message instructing the diplomats to deliver the fourteen-part message to Secretary Hull at one P.M. Sunday.5 But Kramer had gone home to sleep that night after having delivered the first thirteen parts of the fourteen-part message to the White House, Secretary of the Navy Knox and various admirals. (His wife drove him around town in making these deliveries.) Before going home, Kramer testified, he checked back with the office, found no new intelligence had come in, and ordered the watch officer to telephone him immediately upon the receipt of any further Japanese message. This had been done several times in the past. He lived only ten to fifteen minutes away from the Navy Department. A trip back to the Department was easy.6
No one called him.
So, once again, while Washington slept, from three to nine precious hours were lost, this time by the Navy, in terms of our having one last chance to alert our forces in the Pacific.
Kramer insisted that he understood the significance of the final part of the fourteen-part message, and also the meaning of the one P.M. delivery time. It meant war.
Instead of delivering these available messages in the predawn hours of Washington time, it was not until around nine A.M. on December 7 that Kramer returned to the office to deliver both messages simultaneously to Captain McCollum in Admiral Stark’s office. The two men conferred. There was a map on the wall of Kramer’s office showing the different time zones around the world, and he pointed out to McCollum that one P.M. Washington time would mean seven-thirty in the morning Honolulu time, or approximately sunrise. It would also be two A.M. in the Philippines. The exact significance of these Pacific times was not understood, but if an attack were coming, it looked as if it were timed for operations in the Far East, or possibly on Hawaii.7
McCollum took the messages from Kramer and gave them to Stark, telling Stark about Kramer’s comments on the time zones and the potential threat of attack. McCollum testified: “We had no way of knowing, but because of the fact that the exact time for delivery of this note had been stressed to the ambassadors, we felt there were important things which would move at that time, and that was pointed out … to Admiral Stark.…”8
In making his other deliveries that Sunday morning, Kramer pointed out the importance of the time zones to Secretary of State Hull’s private secretary and to Colonel Bratton (who also happened to be in Hull’s outer office at the moment). All in all, Kramer said he told about eight to ten people of the significance of the timing for the Japanese reply.9
What all this meant was that when Kramer began his deliveries after nine A.M., there were only three hours left to warn Pearl Harbor. It might have been enough time if all had gone well in Marshall’s office and with his attempt to send a last-second warning to Short. But two other truly fateful errors had been committed earlier: one was Dusenbury’s failure to deliver the decrypts to Marshall, or the other Army recipients, and the other was the Navy’s failure to have a translator on hand at the crucial moment after midnight on Sunday morning.
The question of why the Navy was processing the diplomatic messages on December 6 still hadn’t been answered to my satisfaction, however. The Army was supposed to process the intercepted material on the even-numbered days. And Kramer went so far as to mislead Sonnett by swearing that “the Navy was responsible [for the sixth of December], [the messages] being in the key that the Navy was handling that day.” This was false testimony.
It wasn’t until much later, after I had read all the testimony before the Hewitt Board and the Congressional Committee, that I understood. The key that Kramer spoke of had nothing to do with the matter at hand.
It all came back to the fact that before December 6, the Navy was receiving intercepts from its listening posts by teletype, while the Army was still using air mail.
To explain: The Navy received the first intercepts of the fourteen-part diplomatic message via teletype on December 6, and began processing them instead of sending them immediately to the Army.10 Apparently, this was done on the orders of Rear Adm. Leigh Noyes, the chief of the Navy’s Communications Division.
According to Safford’s testimony before Congress in 1946,11 the first “five or six parts of the long 14-part message were received in the Navy Department, I believe, about ten minutes to twelve, just before noon [on Saturday, December 6].” The Naval Duty Officer telephoned his Army counterpart and learned that the Army was securing (closing down) its office at one P.M. “because they were observing the normal working hours prescribed by the Civil Service Commission.”
According to the working agreement between the Army and the Navy dated November 12, 1941, the Army was supposed to process the messages on the even days of the month and the Navy on the odd-numbered days. This meant that only one service at a time was supposed to do the decoding, the “exclusion” of the code, the translation and the final preparation for delivery. (The Navy made all deliveries to the White House, and the Army to the Secretary of State.)
This explains the callback of the Army communications people to their offices on Saturday, December 6. The Navy was running on all eight cylinders; the Army was playing post office. It also explains the Army’s belated installation of a TWX machine at headquarters to communicate with its intercept stations. But while the Navy had properly evaluated the situation on December 6, it could not handle by itself the mass of material that flooded in. It had to call on the Army for help that afternoon in processing portions of the first thirteen parts of the fourteen-part message.
The culmination of all this was that while Noyes was right to order the Navy to work on the intercepts on December 6, even though it wasn’t the Navy’s day to do so, the Navy did not have the requisite number of translators to handle the load on December 7. Kramer was a one-man band. He had to go home and sleep, relying on the watch officer to wake him. But this the watch officer failed to do.
So, when Congress pinned down this failure, the testimony went like this:
Mr. Murphy [Counsel to the Committee]: Is it not a fact that shortly after 5 o’clock (A.M.) [on December 7], or 7 hours before Pearl Harbor, in your department, while you were not in your department, they knew that the warning was given that 1 o’clock (P.M.) was the deadline? Isn’t that right?
Captain Safford: Brotherhood [the watch officer] did, that is correct.…
Mr. Murphy: Where were you at 5 o’clock in the morning [of December 7]?
Captain Safford: I was at home.
Mr. Murphy: At home. The fact is further that the 7th was the Navy day for translating, was it not?
Captain Safford: Yes.
Mr. Murphy: There was no interpreter who knew Japanese in your department, was there?
Captain Safford: There was not.
Mr. Murphy: And you are over at home at a time when you think war is coming, because you have told this committee that war was coming on Saturday or Sunday, you knew that there is going to be a time fixed which will fix the deadline and you leave on Saturday afternoon at 4:30, and you do not inquire as to anyone under you until after the war has started; that is right?
Captain Safford: That is right.…
Mr. Murphy: The fact is that you had no interpreter there on the day you expected war to start, did you? Kramer was a subordinate of yours. You had no interpreter there, did you?
Captain Safford: We had no interpreter there at the time.
Mr. Murphy: The 7th was the Navy’s day?
Captain Safford: Yes, sir.
Mr. Murphy: And it was the day that you expected war to start, wasn’t it?
Captain Safford: Yes, sir.
Mr. Murphy: And you are still in your pajamas having breakfast at 2 o’clock [Sunday afternoon]?
Mr. Murphy: Do you have any sense of responsibility for the failure of this 1 o’clock message to get to the proper people in time? Do you feel responsible?
Captain Safford: Not in the least.…12
Now, that is a pretty rough bit of cross-examination. And, in fairness to Safford, he claimed he went home on Saturday because he was totally worn out. Physically, he didn’t believe he could have done any more. The greatest problem in his professional assignment was that of mental breakdowns. There had been several in his department. Safford was afraid he might be next. His not feeling any responsibility for the failure to handle the intercepts more expeditiously, however, was to my mind another symptom of the Navy’s arrogance in its handling of intelligence.
On the other side of the coin, one must say that if Washington had relied solely upon Army intelligence, and its penny-pinching air-mail delivery of crucial intercepts, the fourteen-part diplomatic message and the all-important final message saying when the news should be given to our State Department would not have been decoded, translated and delivered until after Pearl Harbor.
The point I wish to make is that the intelligence-gathering process of the time was not up to the job. The system for handling communications intelligence was neither operationally efficient nor accurate in its analytical predictions. And while the two services might have paid lip service to the concept of mutual cooperation, when it came to the crunch, their good intentions were nothing more than that. Their rivalry had to be stopped. A single, centralized intelligence service had to be created. God forbid that because of a similar intelligence failure, we might endure a second Pearl Harbor. One with nuclear weapons.
1 See appendix, SIS 25432, p. 322.
2 See appendix, SIS 25850, p. 351.
3 PHA Hearings, pt. 36, p. 532.
4 Ibid.
5 Ibid., p. 349.
6 Ibid., p. 348.
7 Ibid., p. 531.
8 Ibid.
9 Ibid., p. 532.
10 PHA Hearings, pt. 8, p. 3744.
11 Ibid., p. 3558.
12 Ibid., pp. 3716–3717.