Embarrassed by Herbert Yardley's revelations in his 1931 book The American Black Chamber, the Japanese Foreign Ministry turned increasingly to machine systems to encipher its messages.
The Foreign Ministry in 1935 adopted a device called the "Type A Machine" to protect its communications.
When the U.S. Army's Signal Intelligence Service (SIS) first encountered intercept from the "Type A" machine in 1935, the personnel nicknamed the system "RED." Color nicknames were used extensively in the U.S. military at that time for plans and programs, so it seemed only natural to the SIS staff to adopt the first color of the spectrum for the first machine cryptosystem they worked on.
SIS, which had been studying machine systems in anticipation of developments such as this, produced its first translation from a RED machine decrypt in February 1937.
Translations of RED messages gave the U.S. government insight into Japanese foreign policy and practice. For example, senior American leaders, as might be expected, were anxiously watching Japanese negotiations with Germany and Italy, which eventually resulted in the Anti-Comintern Pact of 1936. A decrypted Japanese diplomatic message of November 1936 revealed to U.S. policymakers the details of a secret clause in this tripartite treaty which committed each country to support the others in war "regardless of the circumstances."
However, this kind of close access ended abruptly. In late 1938, a cipher expert traveled to Japan's major overseas diplomatic posts to install a new cryptodevice, called the "Type B Machine." Japan's Foreign Ministry introduced this new equipment in February 1939; U.S. Army cryptanalysts nicknamed it PURPLE.
In fact, RED and PURPLE machines were used simultaneously. PURPLE was used for communication with major diplomatic posts -- Washington, London and Moscow, among others --- and RED for less important locations. Some posts, such as Washington, had both machines, enabling lateral communications with other overseas missions as well as directly with Tokyo. Some good intelligence was still obtained from RED, but the material the Americans most wanted was now enciphered on the PURPLE machine.
The machines had a basic weakness that eventually became one of the keys to its exploitation by U.S. cryptanalysts. The machines split the Japanese syllabary into vowels and consonants, and enciphered each separately. Once this 20/6 split was recognized, the basic attack against PURPLE was cribbing.
The Japanese also made two elementary mistakes in procedure: since RED and PURPLE were used at different overseas posts, they frequently sent the same message enciphered in both systems. Furthermore, they enciphered English-language diplomatic texts;
The Americans capitalized on both mistakes to develop cribs for analyzing the system.
The solution of PURPLE was a team effort, under the overall direction of William Friedman, with Frank Rowlett leading the day-to-day efforts. Genevieve Grotjan, Albert Small, and Samuel Snyder, junior cryptanalyts, also made important contributions in solving the system.
Once the system had been solved, William Friedman established what may have been the first compartment for intelligence information in the U.S. Decrypts were marked with the codeword "MAGIC" to show they needed extra protection. According to the story, Friedman chose this codeword because he liked to describe his cryptanalysts to outsiders as "magicians."
When SIS personnel had solved the PURPLE system, they came up with the concept of building their own version of the machine for faster decryption, a device they called the "PURPLE ANALOG." An electrical engineer in SIS named Leo Rosen produced a prototype, according to legend, constructed in his basement. Since the Army lacked the machine shop capabilities to make additional units, SIS asked the Navy to make PURPLE ANALOG copies in its facilities.
Three American-version machines were built initially. The Army and Navy each kept one; the third was to go to Station HYPO in Hawaii, but instead was sent to Britain when the two countries began their cryptologic cooperation.
The Navy, by the way, had referred to this system as the "B-machine," but adopted the Army's nomenclature, PURPLE, once this cooperation was established.
As will be discussed in a later article, the Army and Navy eventually came to a consensus to exploit these communications jointly. In January 1941, by agreement, distribution of MAGIC translations was limited to ten people -- the President; the secretaries of War, Navy, and State; the Chief of Staff; the Chief of Naval Operations; the chiefs of the two services' war plans divisions; and the chiefs of their intelligence organizations. No decrypts were distributed to commanders outside the continental United States.
The diplomatic decrypts were of great benefit to U.S. negotiators in helping them understand Japanese policy prior to the war.
- William F. Friedman, Preliminary Historical Report on the Solution of the 'B' Machine, Army Security Agency, 4 December 1945, SHR-159, CCH Files.
- Philip H. Jacobson, "The Japanese Cipher Machines," NCVA (Paducah, Ky.: Turner Publishing Company, 1996), pp. 54-55.
- Francis A. Raven, "Some Notes on Early Japanese Naval/Diplomatic Cipher Machines," CCH Series File, IV.W.III.23.
- The Role of Radio Intelligence in the American-Japanese Naval War (August, 1941 - June, 1942), 1 September 1942, SRH-012, CCH files.
NOTE: in the middle of World War II, to avoid confusion, the U.S. adopted the British term ULTRA for most decrypt intelligence from high-grade systems; the codeword MAGIC survived for daily summaries of some high-level decrypts.