Closing the book on Pearl Harbor

Cryptologia, Apr 2000 by Budiansky, Stephen

But-perhaps the most crucial fact of which Stinnett is ignorant-on December 1, 1940, an entirely new code book was introduced, superseding the previous code book. Then on February 1 and August 1, 1941, new key books were introduced as well. The crucial point is that "solving" or "breaking" a code is an ongoing process. Sometimes it is indeed a daily process, as was the case with the German Enigma traffic; every day the key changed and a huge cryptanalytic effort was required to recover the day's keys before messages could be read. Although the theory of how to break the Enigma had been determined by the first months of the war, there were still many days throughout the war when the traffic was never read. Likewise, the fact that the Japanese naval Operations Code had been "solved" in the fall of 1940-at least for year-old back traffic-does not constitute proof in the least that operational messages were being read subsequently. The entire history of cryptanalysis in World War II is one of repeated success and setbacks as enemy codes were changed or modified.

The month-by-month progress reports of OP-20-G provide even more decisive proof. These reports are clearly contemporaneous; each is date-stamped, and each is typed up in a continuous, scrolling fashion from one page to the next, so that there is no possibility of pages having been removed. They report on the complete work of OP-20-GY, the cryptanalytic section of the Navy's codebreaking bureau, and they provide precise tallies of the number of code groups and additive groups recovered each month in each system under study. They include reports on work far more sensitive than the Navy's attempts to read the Japanese naval codes, also listing progress against French diplomatic codes and Italian naval and naval attache codes. (Until only the last few years, the National Security Agency adamantly refused to declassify any documents that contained references to the fact that the United States had ever attempted to read codes of neutral or friendly nations, even during the 1920s, '30s, and '40s. The agency continues to oppose declassification of postwar documents that discuss American scrutiny of neutral nations' coded traffic.) The list of systems being worked on in these reports squares precisely with a complete list of OP-20-GY personnel assignments dated November 1941.5 The point is that we can be extremely confident that nothing has been held back in these reports. Nothing has been removed, redacted, expurgated, or censored. They were prepared as progress reports at the time the work was done; unless we grant the officers of OP-20-G the prescience in January 1940 that they would need to start concealing their work on Japanese naval codes right then and there so that FDR would be able to hide the truth two years later, we have to accept what they are plainly telling us. The progress they report on the Operations Code squares completely with the account of the "History of GYP-1"; both establish unambiguously that as a result of the code book and key book changes in the Operations Code beginning in December 1940, no current decryption at all was taking place at the time of Pearl Harbor. Indeed, they confirm that no Operations Code traffic had ever been read currently by December 7, 1941, and that no Operations Code traffic sent at any time in 1941 was read at any time before Pearl Harbor. On December 1, 1941, only 3,800 of the 30,000-plus code groups and 2,500 of the 50,000 additive had been recovered from the current system, i. e., the code book that had gone into effect December 1, 1940, and the key book that had gone into effect on August 1, 1941. (Other accounts note that most of those 3,800 code groups were the code groups that stood for numerals-which would have severely limited the ability to read any traffic, even traffic that happened to have been enciphered using the 5 percent of the key book that had been recovered.) This was simply not enough to render any messages comprehensible. A contemporaneous report which tabulates the number of messages broken out of the Operations Code in 1941-43 lists the number from 1941 as "none,"6 and this is corroborated completely by the testimony of the navy officers involved in the work and by other first-hand accounts.