Der IBM-Radiotyp

Und seine Rolle im Zweiten Weltkrieg

Source: http://www.columbia.edu/cu/computinghistory/ibmradiotype.html

IBM Radiotype at US Army Signal Corps Communications in the Pentagon, 1944. Photos in this section: IBM.

IBM Radiotype - another view

Radiotype Army demo 1942

Pentagon comms center 1944

Radiotype transmitting typewriter

Developed in 1931 by Columbia-educated Walter S. Lemmon, president of Radio Indus­tries Corp­oration in New York City, the Radiotype transmits text messages from one electric typewriter to another via shortwave radio or wire, and was widely used by the US Army Signal Corps during World War II. The large photo shows the component parts. Top: Copy holder, transmitting typewriter, paper-tape punch, control panel, receiving typewriter, clock counter. Bottom: Automatic tape reader, tape winder, Morse code keys. WAR (under the Control Panel label) was the call sign of the central Signal Corps radio station.

Screen shots from the Hollywood movie Air Force

Air Force, Howard Hawks, 1943 (the full 2-hour film online for as long as the link lasts). It features not only the IBM Radiotype and other IBM equipment but also early-model B17s (and false propaganda involving Japanese-American fifth-columnists in Hawaii). As the film opens (at 1 minute and 40 seconds) in the "Base Signal Office" some clacking is heard and the sergeant goes over to the Radiotype machine that is receiving a message. We see the machine typing the message using a typewriter-like mechanism. He carries it into the Base Intelligence Office and hands it to the captain, who hands it to a codebreaker, who sits down at an IBM 032 key punch, transcribes the message, and we see two punched cards come out. The captain takes the cards and sticks them into the hopper of what seems to be an ordinary IBM 405, and we see the clear-text message being printed by the 405's type bars, and this message is what sets the plot in motion.

1. Incoming message on Radiotype

2. Closeup of message

3. Transcribing onto cards

4. Closeup of IBM 032 card punch

5. IBM 405 type bars

6. Message coming out of 405

Thanks to Henry "Strontium Black Cat" for most of these screen shots.

Military Messaging in World War II

Throughout World War II US, British, German, Japanese, and other armed forces kept their widely scattered units and command chains in touch by various means including short-wave radio. For the USA, the IBM Radiotype was an important means of short-wave communication as it allowed messages to be sent and received in real time as text, rather than (say) morse Code, thus eliminating the need for specially trained human telegraphers at both ends of a connection (such as my mother and my father, who were both Navy radiomen during the war). But since enemy forces could tune in to the US frequencies (or tap US wires), all messages had to be encrypted. Therefore, great efforts were expended on all sides to create unbreakable ciphers and develop methods of breaking intercepted enemy ciphertext.

Top-secret IBM relay calculator (left) connected to IBM 405 [1]

SIGCUM

The sequence in the film is accurate except in two respects. First, the message would not have arrived on the Radiotype in clear text (image 2), but rather as a nonsensical garble of characters; that is, in code — which is why it had to be transcribed onto cards and fed into another machine to be interpreted. Second, the 405 (which is really not much more than a big card-driven adding machine) would have been merely an I/O device for a decoding machine, such as the still little-known nameless top-secret relay calculator shown at left and built by IBM for the US Army Signal Corps, which is described in some detail in the Pugh book (reference 1 below), which is also the source for the image (thanks again to Henry for scanning it). Another possibility would have been the SIGCUM, a smaller (portable, even) Enigma-like mechanical device, but I don't see any evidence of it being hooked up with any punch-card machines.

Type 040 tape to card punch Photo: IBM

From the descriptions below, it would seem that the same setup could be used for decrypting friendly messages for which the key is known by simply swapping in some different plugboards. Also note that the transcription from paper to punched cards would not be necessary if the incoming message was directed to the Radiotype's paper tape punch, and then the tape was converted to cards automatically; for example, by the IBM Type 040 Tape Controlled Card Punch. Then the cards could be printed without any possibility of errors in the transcription, which would wreck the decryption. IBM notes (on the page just cited) that "Signal Corps and other military organizations used the IBM 40, and its counterpart IBM 57 card-to-tape punch, for defense work during World War II."

Image: US Navy

If you think about it, it takes a great deal of skill and concentration to transcribe encrypted text. Touch typists go fast because their fingers "know" common patterns like T-H-E and I-N-G. But ciphertext is a string of random characters with no patterns. Yet both my parents did this during the war, all day, every day; it was their job at Navy Department headquarters in Washington DC; they transcribed incoming Morse code to paper on big old clunky manual typewriters and then handed the messages off to the decoders, who worked in the same building. They also sent ciphertext in Morse code. Both ways at 120wpm. That's all I know; they never talked about the details and it's too late now to ask them. Reference 6, however, has a whole chapter on their workplace.

Links:

• The film Air Force (1943), Wikipedia, accessed 29 May 2019.
• Wing and a Prayer, a 1944 film. It includes an almost identical sequence: incoming Radiotype message taken to an IBM 032 operator in the "decoding" department, who punches it onto cards which are then fed to the 405, decoded, and printed, but of course the decoding part is not shown.
• Interconnected Punched Card Equipment. Steve Dunwell, who designed the secret relay calculator, was also the one who engineered Wallace Eckert's groundbreaking switch box at Columbia in 1934 which was the first instance of automated scientific computation.

References:

1. Pugh, Emerson W., Building IBM: Shaping an Industry and its Technology, The MIT Press (1995). Radiotype: p.95. Deciphering enemy codes: pp.98-106 (figure caption, p.102: "A relay calculator specially built for cryptographic work by IBM is shown attached to an IBM Type 405 Alphabetic Accounting Machine."
   [Steve "Red"] Dunwell's solution [to the inefficiency of using standard IBM tabulators that generally operated at 150 cards per minute, to test thousands of keys against a ciphertext] was to obtain a specially designed processor built with electromechanical relays ... A unique feature of the equipment specified by Dunwell and built by IBM was the function of its many plugboards [seen in photo as rectangular enclosures on the face]. These permitted the machine to be wired to recognize the sets of code-book units that corresponded to the most frequently used words and to supply their probabilities. Cards were fed into the machine [i.e. into the attached IBM 405] so that each card that contained part of an encrypted message was followed by one or more cards with a hypothesized key. For each key the machine performed the decryption, assigned probabilities to the resultant words, and calculated the probability that the message segment had been correctly decrypted. The relays were fast enough that the entire analysis could be done between the reading of one card and the next — at the rate of 150 cards per minute. The first unit worked exceptionally well, and more units were put in service.
2. Stephen W. Dunwell, IBM Builders Reference Room, IBM Archives (Page 2), the text of "World War II Code-Breaking Exploits Unveiled By Retired IBM Engineer After 50 Years Of Secrecy," an IBM Europe/Middle East/Africa Corporation press release distributed on February 12, 1992:
   Stephen Dunwell, 78, waited 50 years to go public with selected information on the black art of how a small team of engineers "hotrodded" commercial IBM punch card machines with special relay calculators. Using data processing techniques, they were able to obtain from machines running at a speed of 150 cards a minute the equivalent of more than one million comparisons each second -- an astonishing feat unheard of, in public, until now ... Within days of being tapped to be technical director of the machine branch of the new U.S. cryptographic center, Dunwell found himself out of IBM and in the Army with a gun strapped to his waist, and unlimited funding and resources. Between round-the-clock efforts to adapt punch card-fed relay computers to new uses, Dunwell found time to woo and marry a former school teacher who became the secretary of the cryptographic center's Commandant. "We fitted our honeymoon into a working trip," he says. "By 1943, the machine was designed, built, and working."
3. Stephen Dunwell, IEEE Computer Society Computer Pioneers by J.A.N. Lee (1992) [PDF]:
   "During World War II he received a direct commission to the Army Security Agency, whose mission was cryptography and code-breaking, using IBM machines with attached relay calculators. He received the Legion of Merit for this work, and returned to IBM at the end of World War II with the rank of Lt. Colonel.
4. An Interview with Stephen Dunwell, OH 153, Conducted by William Asprey, Charles Babbage Institute, Center for the History of Information Processing, University of Minnesota, Minneapolis, 13 February 1989:
   [In] 1938 I was asked to come down to New York to what was then called [IBM] World Headquarters ... At that time they broke the organization up and I came back to the [IBM] Laboratory and worked on the thing that was called "radio-type." I don't think many people realize that before World War II IBM had a radio-operated typewriter that they actually sold to the services. I stayed there until the advent of World War II ... the Signal Corps came to IBM and said they'd like to interview several people for possibly going into the cryptographic operation, which was then down in Washington, and later went to Arlington Hall station ... There were two general ways of approaching the cryptographic machinery problems. One of these was [the bombe, like at Bletchley, which was basically an Enigma machine in reverse, with spinning things] and the other used punched-card equipment. We came out about on a par with the [bombe]. If anything, our punched-card equipment came out better ... In the course of all of that, I had gone to IBM and arranged for some array processing equipment to be provided for us ... which would allow us to read a punched card and punch back [or print] the decrypted result in the same card ... I had [at Arlington Hall] 14 punched-card machines [meaning an IBM 405 coupled to the special relay calculator] built by IBM that would read a card and punch the result back in the card with the computation being made as an array computation in a very short space of time with relays."
5. They also Served, an album of IBM Special Products, Volume 1; pages 3-5:
   With the U.S. entry into World War II, the Signal Corps ordered quantities of the Radiotype machines to equip its stations in San Francisco, Honolulu, Panama, Puerto Rico and elsewhere, forming the Army Command and Administrative Network (ACAN, see map). Long-distance transmissions, e.g. from a ship in the western Pacific to Washington DC would be done in relays. Radiotype installations in New York, Dayton, Omaha, Seattle were linked to the US Signal Corps communications headquarters and its central radio station WAR in the Pentagon. These various stations handled a wartime peak of 50 million words a day.
6. Mundy, Liza, Code Girls, Hachette Books (2017). An in-depth look at World War II military codebreaking.
7. SIGCUM, Wikipedia, accessed 31 May 2019.
8. Thompson, George Raynor, et al, The Signal Corps: the test (December 1941 to July 1943), Center of Military History, United States Army (2003), Chapter VIII: "Signal Equipment: Wire and Radio". The IBM Radiotype used a unique 6-bit code, not the common 5-bit Baudot (Teletype) code. It could transmit and receive up to 100 words per minute — slightly slower than the fastest touch-typer. It was a stopgap solution used until a more general all-purpose Radio-Teletype (RTTY) network was developed that would have built-in end-to-end encryption and decryption, but this took years and years so the Radiotype was used throughout the war (as were, indeed, Morse code transmission... and carrier pigeons!).
9. Radiotype Wireless Data Transmission, IBM 100 (IBM 100th anniversary material).