Eavesdropping On The Rising Sun


Our training was radically different from that at the university. Japanese characters could not be transmitted in the dit-dahs of Morse code. The words had to be vocalized and written out in the Roman alphabet (known as romaji in Japanese). No longer did we have to memorize characters, the hardest part of tackling languages written in ideographs. Besides that, terse military messages in any language require only a fairly limited, if specialized, vocabulary.

I quickly learned to my surprise that I had joined one of the most secret and vital programs of the war. Our specialized studies began under Professor Edwin 0. Reischauer of Harvard, author of the textbook we had used at Columbia and the government's guiding light in accelerated Japanese instruction. He would become the postwar ambassador to Japan, navigating a tricky path toward a new relationship with that country. Reischauer and a pair of associates put us through our daily paces, drilling us in listening to Japanese and translating sample messages.

While the ABCs of cryptology (the making and solving of "secret writing") are simple enough, the practice is wickedly complex. A code is any "language" known only to the sender and recipient of a message, and codes have been used by diplomats, generals, merchants, and clandestine lovers since ancient times. All the two parties need to know is what certain terms signify. For instance, "need cookies" could mean "meet me tonight in front of the bakery," or "send more cavalry." But code- books are necessarily limited in scope. A cipher, on the other hand, substitutes for each letter of "plain text" a different letter, symbol, or number drawn from some "key" table shared with the recipient. That opens an entire dictionary to the creator of a cipher. Without the key the enciphered text resembles gibberish, especially when written without the usual breaks between words but in arbitrary groups.

Any cipher is vulnerable to solution because all languages—or at least all those used by World War II combatants—have certain words, letters, and combinations of letters that are used more frequently than others. (Think of "the" in English.) By isolating these common words and cracking their meaning often by painstaking trial and error, the cipher breaker can use these clues to recover more letters. (Edgar Allan Poe provided an excellent cipher primer in "The Gold-Bug.")

The code maker's job is to disguise these repetitions as much as possible by many layers of enciphering—for example, substituting numbers for letters and then adding yet other numbers to those. But ultimately those buried patterns will resurface somehow, and any cipher can be broken by someone with enough accumulated traffic to work on. IBM machines in the operations building spit out countless punch cards, which sorted thousands of intercepted messages into every conceivable combination of their words and letters. Our job as translators was twofold: to help the cryptanalysts by scrutinizing strings of letters that they brought us in search of Japanese words, and to translate messages already solved. My first assignment was dazzling. One morning I was a student, and the next I was reading a less-than-48-hour-old report from the Japanese ambassador in Berlin to his Tokyo superiors.

My work hinged on one of grandest achievements of the American war effort, the cracking of Japan's major diplomatic code, a machine-generated cipher known as Purple. Like Germany's Enigma machine or our own Sigaba, Purple relied on electromechanical impulses transmitted through changeable arrays of rotors and switches to activate the keyboards of printers that encoded and decoded messages. While such a system vastly speeded up the output of code clerks in the message centers, it also multiplied the possibilities for disguising plain text and tremendously complicated the task of the code breaker. The scientific genius behind the breakthrough, recognized as one of the American fathers of 20th-century cryptanalysis, was Russian-born William Friedman, who worked with mathematicians Frank Rowlett, the principal inventor of the Sigaba, and Solomon Kull- back. By the end of 1940, the trio had cracked the Japanese code and oversaw the building of a replica of the Purple machine in the tiny, secluded offices of the pre-Pearl Harbor Signal Intelligence Service.

The three of them, present at the creation of today's globe-blanketing electronic surveillance system, had become high-ranking officers at Arlington Hall during my stay there, though I don't believe I ever saw them in person. It was a little like working in a telephone and telegraph exchange while Alexander Graham Bell and Samuel F. B. Morse were still hammering out improvements in office suites upstairs.

In January 1943 I saw my first long, partly translated Purple message, which contained Ambassador Hiroshi Oshima's summary of Hitler's intentions in the spring of 1943 from information elicited from Nazi officials. I could not believe that a skinny 20-year-old kid from Queens was translating material that would become part of a top-secret daily summary of signal intelligence findings labeled "Magic," to be read the next morning by President Roosevelt and General Marshall!

It was a thrill that of course could not be shared. The demands for absolute secrecy were obsessive, justified, and sobering. Any breach of security that tipped off the Japanese that their code was compromised would have caused an immediate change, nullified years of effort, and cost lives. Like my companions still wet behind the ears, I was suddenly loaded with a host of frightening adult responsibilities.