I have always been interested in electronics engineering ingenuity used to solve “impossible” problems. One such problem was the task of deciphering the German Army's most secret transmissions during WWII via a German teletype machine named Tunny (that's what the British called tuna). The SZ40 was one such cipher machine used as a teleprinter attachment.
The German SZ40 cipher machine (Image courtesy of Stanford University, Reference 2)
Due to a key mistake made by a German operator on an SZ40 in 1941, the details of which the British have never revealed, the cryptographers were able to determine that the German machine had wheels with variable pin settings, plus that there were 12 of these wheels, the sizes of the wheels (which were in essence the number of variable pin settings on each) were 23, 26, 29, 31, and a continued progression. They also were able to determine that the rotors moved in relation to one another as a message was enciphered. This was the first of two things that helped change the course of the war.
The second one was that in order to decipher the SZ40, they needed a very fast rate of operation in a British cryptanalysis machine which was not presently available in the mechanical devices used in order to determine the patterns around the circumferences of all 12 wheels in the SZ40 at a particular point in time since the Germans were regularly changing the pin-setting patterns. After that, they would need to figure out the settings of the wheels at the beginning of the message. Maxwell Newman, a 45 year-old mathematician from the University of Cambridge, was chosen to head up the effort to break the German teletype traffic.
That same year, 1943, work was in progress on Colossus, a high-speed programmable machine. Thomas H. Flowers was the chief engineer on the Colossus project, who had been looking into how he could use electronics in the electro-mechanical telephone switching systems which were being used before WWII. This effort proved critical to his work on Colossus.
Flowers suggested that one of Heath Robinson’s tapes, which had the unchanging data on the internal mechanism actions of the SZ40 machines, be replaced by Colossus (The reliability of mechanical machines was poor in reading at high speeds). There was only one tape that needed to be read at high speed and Colossus would perform that task with great reliability.
The problem here was that tube failures were common in electronics and Colossus had 1,500 vacuum tubes. Flowers said that the Colossus system would work reliably if it was not turned off and on, which was the common cause of vacuum tube failures at that time. They ran 24 hours per day. Later, in 1944, the Mark II was introduced with 2,400 vacuum tubes assisted by 800 electromagnetic relays.
The Mark II machines were able to work in parallel on five successive characters simultaneously and could read 5,000 characters per second to process characters at an astounding rate for that time of 25,000 per second.
In 1943, the British government Code and Cypher School, operated by the Women’s Royal Naval Service, began using the Colossus I, a high speed programmable vacuum tube-based computer, to help solve an “impossible” to break cypher code from the German teletype cipher machine called Tunny. (Image courtesy of Reference 1)
In a speech in London in 1986, Hinsley told an IEEE audience, how deciphered Tunny messages gave British intelligence in the campaigns in North Africa, Italy, southern France, and eastern Europe. But the biggest contribution of Tunny decryptions was right before the Normandy invasion.
1 Electrotechnology in World War II, Breaking the enemy's code, Glenn Zorpette, Associate Editor, IEEE Spectrum, 1987.
2 The Lorenz Schluesselzusatz SZ40/42, Stanford University.