Thursday, February 2, 2012

Intercepted conversations - Bell Labs A-3 Speech scrambler and German codebreakers

One of the most interesting codebreaking successes for the German side, during WWII, was the solution of the American A-3 speech scrambling device. This system was used on the radio-telephone link between Washington-London throughout the war. Up to summer 1943 it was the only speech privacy system employed on this link. Then the new SIGSALY device entered service and both systems were used concurrently.

The efforts of Post Office engineer Kurt Vetterlein have been mentioned in numerous books, so that part of the story is well known. However there was also another team under the Army Ordnance, Development and Testing Group, Signal Branch - Wa Pruef 7 which successfully solved the A-3 system.
As far as I know the work of this second team has not been mentioned in any book or article.

Time to take a closer look at the work of both teams and the intelligence they got from the A-3 system:
The A-3 device

In the 1920’s, when radio-telephone communications began to used by government departments and private citizens, it became clear that there was demand for devices that would protect these communications from eavesdroppers. The first system built by Bell Laboratories utilized speech inversion, meaning that the frequency of the speech was inverted on a fixed point. This system was first used on the radio-telephone circuit between Catalina Island, California and the US mainland in the early 1920’s (1). The inverter device offered protection from casual listeners, as the speech was rendered unintelligible, but the procedure could be reversed by any technically minded individual. In fact a Bell Labs report says ‘This device was thus an effective privacy arrangement against the casual listener but was very easy to crack, even in those days. An inverter can be built by any reasonably competent high school boy and, in fact, there have been instructions on how to build one in QST’ (2).
The next step was to design a device that worked on the principle of band-splitting. The speech segment was divided into separate frequency bands, these were then rearranged and in addition some of them were inverted. This became the A-3 speech privacy system. The A-3 was a 5 band system and since each band could be either in the right side up or inverted there were in theory 3.840 possible combinations (3). Field tests however showed that out of these combinations only a small number ensured speech unintelligibility and out of these only 6 were selected to be used by the A-3 (4). Every 15 or 20 seconds one of these 6 combinations was used and after 36 steps the ‘key’ was repeated (5).

Although the A-3 device was technically complex it was understood even at that time that a determined opponent with the necessary skills and with access to specialized equipment could eventually discover the operating procedure and descramble the conversations. That’s why Bell Labs called the A-3 a speech privacy system and not a secrecy system (6). Still the fact that for most of the war it was the only device available meant that it was used widely by the Anglo-Americans.
The A-3 was used by US civilian and military authorities and on the link Washington-London during WWII. The most important intercepted discussions were those between the leaders of the Anglo-American alliance, Franklin Roosevelt and Winston Churchill. It seems that in at least one case their conversations gave the Germans vital clues on impending military actions.

The Post Office effort:
The Post Office - Deutsche Reichspost was the organization responsible for telephone, telegraph and wireless communications in Germany. The Reichspostminister from 1937 to 1945 was Wilhelm Ohnesorge, a convinced National Socialist with close ties to Hitler. Ohnesorge was interested in the new radio technologies and was willing to fund research in decoding the A-3 device. How he came to know of the existence and use of this machine is not mentioned in any of the books I’ve read.

The unit that handled this operation was the Forschungsstelle der Reichspost (Research Post of the Reichspost). After the initial decision was made, two factors made things relatively easy for the Post Office people. First was the fact that the Post Office already owned an A-3 device. The second factor was their gifted engineer Kurt Vetterlein who headed the effort to decode the A-3. After researching the A-3’s operating procedure, Vetterlein and his team were able to build equipment that decoded the conversations in real-time and carried out this mission from September 1941 till 1945. Each day a large number of calls were intercepted, usually up to 60 and never less than 30 (7).
The equipment and the team were originally based in Noordwijk, Holland where the reception was excellent. However the threat of British commando raids in 1943 forced the German team to move inland to a more secure location near Eindhoven, Holland and in late 1944 because of the advancing Allied armies they returned to Germany.

The transcripts of the intercepted conversations were sent by teletype to the foreign intelligence department of the Sicherheitsdienst (security service of the SS). Then these were forwarded to Himmler, Hitler and other personalities of the Third Reich. Interesting details regarding the intercepted material are given by Dr Hans Wilhelm Thost, a journalist and employee of the SD. Thost had a strange background. In 1935 he was the London correspondent of the Völkischer Beobachter, the newspaper of the National Socialist Party. In October of that year he was ordered to leave the country. What was the reason for his expulsion? It seems that Thost may have taken part in unlawful activities like espionage. Whatever the case he was one of the people who translated the incoming A-3 material and his interrogation TICOM I-190 ’Extracts from report on interrogation of Dr Hans Wilhelm Thost’ is very interesting.

According to him the Post Office minister Ohnesorge distrusted the military and did not want to give them the transcripts of the intercepted communications. That was the reason for the Reichs Post-SD connection. Thost says that the address for Washington was Republic 2020.

In his interrogation he lists the memorable calls as follows:

a). Between War Office ,London and British Army staff ,Washington. Most of the time the caller was Brigadier Leslie Dawes and in London Brigadier Owen Young. The discussions concerned British orders of American military equipment. Cover words were used for the items (‘grapefruits ‘, ‘pineapples‘)
b). Between the Ministry of War Transport, London and British Shipping Mission, Washington. Talks concerned the allocation of shipping space. Theatres of war were referred to by cover name. (‘Arthur’s place’, ‘John’s place’)

c). Ministry of War Transport, London and representative of same organization in Washington. Talks concerned the allocation of tanker shipping space. Thost says that there was a serious shortage of tanker ships.
d).Concerning political and diplomatic matters:

Cases include British Embassy, Washington to Foreign Office, London , Dutch Government, London to its representative in Washington, in one or two cases Soviet ambassador Maisky to Soviet ambassador in Washington. Also conversations between Eden in Washington and Churchill in London.
e). Concerning economic matters. (Foreign Economic Administration , United Nations Relief and Rehabilitation Administration – UNRRA)

The most interesting calls were of course those between Roosevelt and Churchill. Their conversation of 29 July 1943 alerted the Germans to the impending Italian surrender and allowed them to take swift measures against the Italian army (8).

Other top level officials were also recorded: General Mark Clark, Lord Halifax, Averell Harriman and Harry Hopkins.
Walter Schellenberg, head of SD foreign intelligence, was the recipient of the transcripts and he mentioned the Roosevelt-Churchill talks in his memoirs and in his postwar interrogations. In ‘Report on interrogation of Walter Schellenberg 27 June- 12 July 1945’, p31 he said:

Amt Vi telephone monitoring of the Trans-Atlantic telephone service between London and Washington was very successful. This monitoring was effected from Holland, and a highly complicated machinery was used for that purpose. Before the Teheran conference, SCHELLENBERG received a report of a conversation between CHURCHILL and ROOSEVELT. Most trans-Atlantic calls referred to questions of supply. Decoding of these talks was difficult as the essential words were coded twice. Oberpostrap VETTERLEIN supervised the monitoring service in Holland. This service offered great difficulties from the technical side.
In his autobiography ’The memoirs of Hitler’s spymaster’, p418 he said:
Early in 1944 we hit a bull's eye by tapping a telephone conversation between Roosevelt and Churchill which was overheard and deciphered by the giant German listening post in Holland. Though the conversation was scrambled, we unscrambled it by means of a highly complicated apparatus. It lasted almost five minutes, and disclosed a crescendo of military activity in Britain, thereby corroborating the many reports of impending invasion. Had the two statesmen known that the enemy was listening to their conversation, Roosevelt would hardly have been likely to say good-bye to Churchill with the words, 'Well, we will do our best—now I will go fishing.'

The Post Office operation was undoubtedly a great success but it was not the only effort against the A-3 device.
The Army directed its own separate operation through the Army Ordnance, Development and Testing Group, Signal Branch Group IV - Waffenpruefung Abteilung 7/IV.

Alfred Muche and the 5B machine
Now I said earlier that all the books on codebreaking mention the Vetterlein-Post office story. However that is only 50% of the puzzle.

In TICOM report I-213 ‘Report on interrogation of Alfred Muche’, recently declassified by the NSA, a German engineer named Muche working for the Army Ordnance, Development and Testing Group, Signal Branch Group IV Section E (Wa Pruef 7 /IVe), describes his own successful effort versus the A-3 scrambler.

The WaPruef 7/IV agency was involved in special tasks during the war. They analyzed and decoded Soviet speech scramblers and built equipment that intercepted and printed Allied multichannel radioteletype traffic. Section E, headed by Dr Loetze, did research in speech privacy systems. Muche was an engineer with Section E. His life story was as follows: from 1927-37 he worked for Heliowatt Werke and in 1937 moved to WaPruef 7/IVe. For the period up to 1940 he studied domestic and foreign speech privacy systems. At the end of 1940 or the beginning of 1941 his department started the study of the encrypted transatlantic telephone link. Under Muche’s direction single sideband receivers were built and the traffic recorded at ‘Nordwyk,Holland’ [note that this was the same area that the Post Office used for their interception facility]. In order to build the receivers he got assistance from Professor Koomans of the Dutch Post Office (PTT-Staatsbedrijf der Posterijen, Telegrafie en Telefonie)

After studying the recordings with specialized equipment the Section E party found out that ‘the speech spectrum was being split into 5 bands, inverted and translated ’ …. ‘it was found that the cipher controlling the switching repeated cyclically after 36 sections.’ Since the operating procedure of the A-3 had been solved a descrambling machine known as 5B was built. The equipment was moved from Nordwyk to Ludwigsfelde (near Berlin) in late 1942. Ludwigsfelde housed a large army intercept station.
According to Muche the 5B machine became operational in the summer of ’43 and one of the conversations between Roosevelt and Churchill foreshadowed the Sicily landings and allowed the Germans to withdraw their forces with minimal losses. Unfortunately for the Germans the 5B machine was destroyed by aerial bombardment in late 1943. Muche then spent 8 months building an improved version and completed that task in the summer of 1944. The machine continued to intercept and decode the traffic till ’45 when the unit was forced to move. He did not know what happened to the machine at the end of the war. For his efforts he was given the Kriegsverdienstkreuz and a Speer reward of 10.000 marks.

Regarding the equipment he used the following companies are mentioned :
1. AEG-parts for the SSB receivers

2. Siemens - ring modulators
3. AEG - ‘‘star‘‘ modulators

4. Filters for the 5B machine - Dr Vierling ( of the Feuerstein laboratory)
5. Speech analysis equipment – Breusing Tonsystem, Berlin

More information about the disposal of the 5B machine is given in TICOM I-203Interrogation of Herbert MARINIOK and Herbert Korn, Former Members of the Reichspost and OKW/CHI, p4

KORN confirmed MARINIOK’s statement that X Geraet was invented by Dr. LOTZE assisted by Mr MUCHE to enable the Germans to intercept transatlantic telephone conversations. Asked about the history of the apparatus, however, he stated that he had been concerned with it only since its installation at Ludwigsfelde in August 1944. He had worked with it until April 1945 when it had been taken to the Schliersee, and he had himself been one of the group which accompanied it by truck. KORN had also been on the raft which had dropped the apparatus into the Schliersee on 1 - 2 May 1945. Contrary to MARINIOK's statement, KORN claimed that the apparatus was sunk in several sections and not in one piece. The frame, which had consisted of three large parts, was dropped in sections, and the compartments of the apparatus itself were dropped separately. KORN was certain he could pick out the exact spot where the parts were sank, and stated that the nearest village was MIESBACH. Although he was certain that the construction of the apparatus could be seen should the parts be retrieved , KORN thought that no part would be usable because of the corrosive influence of the water.
Muche was just as successful as Vetterlein but until now his story was not known. Although the army’s effort wasted resources, by duplicating the Post Office operation, it was nevertheless successful and provided valuable information during the war.

By eavesdropping on the Allied conversations the Germans got military, diplomatic and economic intelligence. In at least one case (Italian surrender) the information they received allowed them to take swift military action and preempt the Allied plans. That event alone justified the resources spent on the A-3 both by the Post Office and the Army.

The intercepted communications between Roosevelt and Churchill are an embarrassing episode in the signals intelligence war. However the Allies knew the A-3 system was vulnerable and the SIGSALY machine which replaced it was a quantum leap in terms of security. In theory thanks to SIGSALY the Allies had absolute security from mid ’43 onwards. However it seems that the device installed in London did not work properly till October ‘43 and only became fully operational in April ‘44. Even then officials continued to use the A-3 for most of the traffic since the only Sigsaly link could be accessed at the Cabinet War Rooms and only a few people had authorization to use it (9).
Overall the story of the A-3 scrambler and the German efforts against it is an interesting chapter in the history of communications security.

(1). Bell Labs report ‘History of speech privacy systems-1970’, p2

(2). Bell Labs report ‘History of speech privacy systems-1970’, p3
(3). According to Bell Labs report ‘History of speech privacy systems-1970’, p4 the bands used on the A-3 were A: 250-800Hz, B: 800-1.350Hz, C: 1.350-1.900Hz, D: 1.900-2.450Hz, E: 2.450-3.000Hz. The possible combinations are: (5x2)x(4x2)x(3x2)x(2x2)x(1x2)=3.840.

(4). Bell Labs report ‘History of speech privacy systems-1970’, p4 says: ‘In other words, a certain amount of intelligibility could be obtained by just listening, particularly if the listener practiced a bit. It finally turned out that there were only about eight truly private combinations which were reasonably proof against an expert listener. Even these could be understood by an expert if one combination was used for quite a time. Later tests made in the laboratory on a 5 band system resulted in the choice of just six out of 3.640 possible combinations. These are used today in the A3 system. This taught us a considerable amount about the "toughness" of speech’.
(5). Historian David Kahn says after 20 sec in ‘The Codebreakers’ and ‘Hitler’s Spies’ while the Bell Labs report ‘History of speech privacy systems-1970’, p4 says almost 15 sec.

(6). General Marshall’s testimony on A-3 insecurity, Bell Labs report ‘History of speech privacy systems-1970’, ‘A History of engineering and science in the Bell System: National Service in War and Peace’.
(7). ‘Hitler's Spies: German Military Intelligence in World War II’, p173

(8). ‘Kriegstagebuch des OKW - 1943 Teilband II’ by Percy Schramm, p853-4
(9). ‘The woman who censored Churchill’, p112-3

Sources: European Axis Signals Intelligence, NSA website, ‘The Codebreakers – The Story of Secret Writing’,  ‘Hitler's Spies: German Military Intelligence in World War II’, TICOM reports I-190 , I-203 , I-213, ‘Nazis in pre-war London’ , ‘The memoirs of Hitler’s spymaster’ , Wikipedia , ‘The woman who censored Churchill’, KV 2/95 ‘Walter Friedrich SCHELLENBERG: rose to be No. 2 in the S.D. and was close to Himmler’, National Cryptologic Museum library - David Kahn collection, Bell Telephone Laboratories report ‘History of speech privacy systems-1970’, National Defense Research Committee reports: ‘Final report on project C-43 Continuation of Decoding Speech Codes’, ‘Speech Privacy Decoding - Final Report, January 31, 1942’, ‘Kriegstagebuch des OKW-1943 Teilband II’, ‘A History of engineering and science in the Bell System: National Service in War and Peace’, Cryptologia article: ‘Review of Forschungsstelle Langeveld: Duits Afluisterstation in bezet Nederland’, General Marshall’s testimony on A-3 insecurity.
Acknowledgments: I have to thank Rene Stein of the National Cryptologic Museum for the reports from the David Kahn collection, William Caughlin and George Kupczak of the AT&T Archives and History Center for the report ‘History of speech privacy systems-1970’ and Randy Rezabek for sending me information from ‘A History of engineering and science in the Bell System: National Service in War and Peace’.


  1. NSA used the example "transcripts of Churchill and Roosevelt's conversations were on Hitler's desk" in training materials in the 60's, along with photos said to be of the voice scrambler system. IIRC the example was still highly classified in those days, and I felt sorry for my history major friends.

  2. Hi Christos,

    You stated:
    "Its security was limited against a determined opponent that could devote scientific personnel and equipment for the task of decoding it."

    I thought David Kahns description of the A-3's limited security was important.

    "As a not-at-all extreme example, some Bell Telephone Laboratories engineers recovered an average of 47 per cent of the words scrambled by the A-3 simply by listening to it several times. This means that almost half the intelligence leaked through. In one test, intelligibility rose to 76 per cent, or three quarters of what was said. This is enough to give an eavesdropper the gist of a conversation"
    The Codebreakers Page 558 David Kahn 1996


    1. Speech scramblers with a single and fixed inversion point are indeed breakable by a trained ear.

      Multiband frequency domain scramblers offer somewhat better protection but still leak residual legibility, however not on the orders of magnitude listed by Kahn. I believe he meant a single inversion system OR the A-3 was rather a poor implementation of the concept. Kahn [in an article on origins of spread spectrum] states the interval between code changes as 20 seconds which kind of contradicts the keystream repetition period of 36 seconds given in this article.

      BUT any non-time-domain scrambler (and some time-frequency-domain ones) gives the eavesdropper insight on things like pace, emotion and tone of the conversation. That happens because the amplitude component remains unscrambled, the silence gets encrypted as silence and vowels as vowel (no necessarily 'human' vowels).

  3. Wasnt the overall operation called "tubular bells"? Or was that an American/British operation? I have trouble finding anything on it except modern applications of the term.

    1. Do you mean operation Ivy Bells?

  4. No, this was definately a WW2 operation I am talking about. I have only seen it referenced once in my life. Of course I cannot remember the book. It was either the Black orchestra's operation on the US/Britian, or another code name for the A3.

  5. what do you know, if anything, about a 26 November 1941 German intercept and descramble of a Churchill-Roosevelt a potential Japanese attack on Pearl Harbor?

    1. I haven't read anything about that.

    2. Do you recall where you learned this?
      I have read a similar account of a call from Churchill about 12:00 AM notifying Roosevelt that a large convoy of Japanese troops had been observed leaving Shanghai bound south, towards SE Asia or perhaps the Philippines. This proved to be the invasion force for Malaya. Since there were still Brits in Shanghai in the International Settlement, this wouldn't have been a security violation compromising Allied code-breaking efforts - like using the A3 scrambler to transmit information obtained from reading Japanese coded messages. I'm trying to find where I read this, so if you can recall the source I'd much appreciate a reply here.

  6. and...I should have said: what a great site!