A historian says a newlyBy Burl Burlingame
declassified codebook proves the
government knew of Japanese
intentions before the attack
The U.S. government knew far more about Japanese intentions before the Pearl Harbor attack than previously claimed, say historians who have recently declassified documents in the archives of the National Security Agency.
The documents are "JN25," the top-secret codebooks used by the Imperial Navy. The official and conventional view given by the U.S. government for more than a half-century is that the codes were too large and complicated to understand prior to the Pearl Harbor attack, but that they were cracked in time to ambush the Imperial Navy at Midway in 1942.
This version of history is a government lie, made "systematically and over a long period of years, trying to minimize speculation that (Japanese codes) could have been read" in advance, said historian Mark Willey of Iowa, who has a special interest in Pearl Harbor history and intelligence efforts.
"The JN25 book is in itself the most important crypto document of all time -- more important than the Enigma and Purple machine or any other codebook because of the significance of Pearl Harbor."
Under the command of Cmdr. Lawrence Safford, the U.S. Navy was in charge of decoding Japanese messages, or "intercepts" in 1941. Following the war, the still-secret materials were transferred to the National Security Agency, the department generally responsible for most of the nation's cryptological activities.
There were several codes used by the Japanese. The code used by diplomats was called MAGIC, and the codes used by the Imperial navy were JN25. "JN25A" was in use in the late 1930s, and "JN25B" in use from early December 1940 to late May 1942, a critical period in U.S.-Japan relations.
The codebook uses five numbers to signify individual words, numbers and phrases, and then enciphered by adding numbers from a table of numbers. The beginning of a message indicated the starting point of the table to the decoding clerk.
The National Security Agency has always maintained that the JN25B code contained at least 55,000 codes or "values." A 1993 article in the "Journal of Intelligence and Counter-Intelligence" by senior agency historian Donald Gish, for example, claimed that there were 57,833 values.
The government has never acknowledged more than a handful of codebreakers assigned to crack the JN25B code. The 1994 NSA monograph "Pearl Harbor Revisited" by Frederick Parker states that no more than two to five codebreakers were assigned to JN25B. According to the National Security Agency's Parker, this "bureaucratic blunder" charges the Navy with dereliction of duty in assigning not enough cryptographers.
So the official position has been too much code, and too few codebreakers, to understand Japanese moves prior to the attack.
The actual JN25B codebook, released to Willey after a drawn-out Freedom of Information Act request, has only 16,409 values, not the 55,000 previously claimed, and nearly 2,000 of these are dummy, "filler" codes. "Thousands of others are insignificant items like pay-rates, Chinese characters and lists of irrelevant minor ship and place names," said Willey.
The earlier JN25A code, previously claimed by the agency to have 30,000 values, only has 5,600 entries.
Also, additional documents like the internal precis "The History of OP-20-GYP-1" show that dozens of codebreakers were at work on JN25B by the summer of 1941, with additional personnel added every week until the attack. The implication, notes Willey, is that JN25B was completely decoded by the fall of 1941.
So a much larger percentage of the JN25B book was decoded than previously thought, and more resources placed on decoding than previously claimed. Why?
"The intent of misstating the size was to bolster the position, which is still the position of the U.S. government, that we didn't read the code prior to Pearl Harbor -- much," said Willey. "But it takes a lot less effort to read a smaller codebook. People reasonably assume that the smaller the book, the more likely that we read it.
"The size issue is directly relevant to the issue of whether we read the messages. The other problem for the NSA is that the messages gave every detail of the attack. So -- to admit we read the code is to admit we knew Pearl Harbor was going to be attacked on Dec. 7."
"It does make you wonder what these American codebreakers knew, and when they knew it," said James Delgado, director of the Vancouver Maritime Museum and author of several Pearl Harbor histories. "This suggests strongly that the government might have had advance warning of the attack -- or at least was capable of knowing."
On the other hand, the initially small numbers of cryptographers assigned to JN25 doesn't surprise USS Arizona Memorial historian Daniel Martinez of the National Park Service. "These guys were, like, geniuses, and geniuses are always hard to come by," said Martinez. "And you can't blame them for concentrating on the diplomatic codes instead of the military codes. We weren't at war, and we were still trying to formulate new foreign policy day-by-day."
Historian and publisher Larry Cott of San Rafael, Calif., an expert in Japanese codes, said he has felt for years that the Navy "deployed far too many men on MAGIC, and not nearly enough on the operational codes such as JN25. To learn that they were busy breaking JN25 as well is surprising. But then, it took 35 years before we even learned there was a JN25A and a JN25B. That was still a secret into the 1980s.
"But I don't know that it changes that much. It's one thing to break a code, it's another to understand it, to learn what things mean by inference."
The messages uncrypted by the JN25 codebooks are still a secret.
"The NSA still will not release any pre-Pearl Harbor raw messages, worksheets or decrypts," notes Willey. "The only specific fact that the NSA will acknowledge is that, 11 months prior to Pearl Harbor, on Jan. 4, 1941, the U.S. knew approximately 2,000 code values of JN25B."
"The question is, why declassify the codes and not the intercepts that the codes helped break?" wondered Delgado. "What is there to hide after more than 50 years?"
Interestingly, notes Willey, while there is one value assigned for each ship in the U.S. Navy in JN25B, the only ship with two values was the USS Arizona.