Codeword Overlord

by Nigel West

  Italian cryptanalytic successes seem to have been limited. The Army Cryptanalytic Section worked on diplomatic, military attaché, commercial and army systems. The Naval Section concentrated its efforts on British naval and air operational codes. Both sections were small, trained cryptanalysts were at a premium and IBM equipment was difficult to procure. The Army Cryptanalytic Section read the U.S. State Department Brown Code through a compromise, and solved (or purchased) several other American systems including the Military Intelligence Code No. 11-146. According to General Vittorio Gamba, head of the Section, they also read a British diplomatic five-figure code, and an unenciphered four-figure, two-part Foreign Office R Code, as well as French, Turkish and Rumanian systems. The Naval Section read British naval tactical codes, the daily-changing air code enciphering table and an unidentified four-figure ‘Anglo-American Naval code’, a four-figure British naval code was read from 1941 until the North African landings in November 1942.

  The Germans held a low opinion of Italian cryptanalytic capabilities, and considered their procedures to be highly insecure so, consequently, co-operation was limited. When the Italians surrendered to the Allies, the Germans then took over what remained of the Italian assets, and dissolved the remaining organisation in February 1944.7

  Section X of the Hungarian General Staff, with a staff of about fifty, concentrated on Turkish codes and ciphers, as well as some Italian, Polish and Russian systems. A TICOM team retrieved some ninety codebooks from the organisation’s emergency headquarters, withdrawn to Germany in 1945, and found evidence that the agency had been working on codes from sixteen countries. Study of Section X’s records suggested that it had liaised closely with the OKW/Chi and its Finnish counterpart.

  The Finnish military intelligence directorate included a SIGINT unit that included intercept, cryptanalytic and evaluation units, which were highly regarded by their German colleagues, and it concentrated on Soviet military, naval and diplomatic traffic, with some attention given to the secondary targets of Polish, Swedish and American communications. The Finns solved the five-figure Red Army code used at the time of the first Russo–Finnish War and in 1943 broke an unspecified US strip cipher. Liaison with the Finns on Russian traffic was highly valued, and the Luftwaffe seconded personnel to the Finnish sites at Mikkeli and Sortavala. Permanent liaison channels for the exchange of information on common targets, such as an unidentified American strip-code, were established, and some equipment was swapped, but TICOM came to suspect that the Finns had not shared all their knowledge.

  TICOM’s research showed that in 1944 the OKW/Chi gave a training course to selected Bulgarian cryptanalytic personnel, but found nothing to indicate that any significant results had been achieved. However, the OKW/Chi did develop links with the Japanese. In addition, OKM/4 SKL III gave the Japanese military attaché at the Berlin embassy details of the British Naval Cipher No. 3, and in return received some strips and settings for the US Strip Cipher DUPYR. Some data on American systems was given to a cryptanalytic delegation that was to be sent to Japan by U-boat, but no actual intelligence was exchanged.

  According to information supplied to TICOM by OKW/Chi’s Colonel Hugo Kettler, there had been a liaison relationship with the Hungarians since the 1920s, and by the spring of 1944 about one-eighth of OKW/Chi intercepts had been provided by the Hungarians. From April 1944 until January 1945 a Hungarian intercept company was attached to LN Regt 353. The Hungarians also sent Italian, Romanian and Polish traffic to the OKW/Chi in exchange for solution methods.

  The TICOM researchers undertook a trawl through the captured German files for evidence of other liaison partners, but the contact with Spain and Bulgaria seemed slight and insignificant. They were also interested in technical progress made by German engineers in the development of electro-mechanical equipment derived from IBM machines purchased through front companies in Switzerland. In particular, the OKW/Chi had established a facility in Weimar where a device could read the Japanese two-letter transposed code (designated J-19 or FUJI) in less than two hours. According to Dr Erich Hüttenhain, the head of OKW/Chi’s cryptographic section, ‘it did the work of twenty people’ and was also effective against a British meteorological cipher, and with figure traffic employing a stencil, which had required close liaison with the Luftwaffe’s Weather Service.

  Access to this hardware, which included two teleprinter tapes and photo-electric sensors that could read at seventy-five positions per second, was strictly limited, although it was made available to Pers Z S. When Dr Rohrbach was interviewed by TICOM he recalled that:

  As was to be foreseen at the outset, the total material [US State Department messages] sent in the strip cipher C-25 could not be deciphered by hand on account of its immense size. The number of available qualified workers with sufficient knowledge of English was too small for that. Deciphering … through moving the strips by hand required six to seven minutes on an average, so that the work … would have taken a whole year, provided that four collaborators had worked on it eight hours daily. It was, therefore, of the utmost importance that the automaton should be available for the decipherment of the material at the time when all keys had been worked out.

  According to Dr Rohrbach, it could take up to a month to read an entire message, and most of OKW/Chi’s successes were all relatively low-security systems, and none had required what TICOM termed ‘higher cryptanalysis’. Furthermore, it was concluded that the Germans had ‘developed no important cryptanalytic methods not already used by the Anglo-Americans.’ TICOM was especially comforted that the Germans had failed to achieve any major breakthroughs:

  Although they were successful with the Japanese RED machine they did not solve its successor, the PURPLE machine. They did not solve the U.S. Army Converter M134C (Sigaba), Converter M-228 (SIGCUMK), the Teleprinter Cipher System using double-tapes (SXGIBS) nor, of course, its successor, the One-time Tape System (SIGTOT), nor the U.S. Navy equivalents thereof, nor the joint Army–Navy–British Combined Cipher Machine (CCM). If they were even aware of the existence of the Anglo-American high-security ciphony system (SIGSALY) is very doubtful, as not a single reference to it is to be found in any TICOM document. They did not solve the British TypeX machine. They apparently did not read traffic sent in the Russian B-211, nor the French modified B-211. In their security studies they certainly did not develop and probably were not aware of practical methods of solving their own plugboard Enigma, or their teleprinter cipher attachments.

  The greatest success for the Germans was in the field of transatlantic communications, and specifically the commercial radio telephone circuits between London and Washington D.C. that were monitored, solved, and recorded by the German Post Office Research Laboratories at Eindhoven, and by the press monitoring group, designated Gruppe VI of the OKW/Chi, located at Ludwigsfelde, about 25 miles south of Berlin.

  The recorded conversations included those held between Prime Minister Churchill and Anthony Eden when the latter was in Washington; the Minister of War Transport and his representative; the same minister and the British Shipping Mission; the Foreign Office and the British embassy; the Dutch government representatives in both cities; the Soviet embassies; the US embassy and the US State Department.

  TICOM interrogators interviewed Kurt Vetterlein of the German Post Office, who explained that these radio telephone circuits were enciphered by a ‘frequency scrambling’ principle. The speech frequencies 450 cycles’ wide and the small blocks were rearranged in positions within the speech frequency spectrum, to give the finally enciphered speech:

  Simple frequency scrambling of speech can usually be solved by examining the spectrographic records of the enciphered speech, cutting out the ‘blocks’ of frequencies with scissors, rearranging them by sight into proper order, and pasting them back together. This reveals the pattern or key used. Simpler yet, if the scrambling pattern has a sufficiently long duration, the rearranging can be done electrically, with the ear for a guide.

  On the Washing
ton–London commercial radio telephone circuit, scrambling and recombining of frequencies was achieved by a pattern that remained fixed for 20 seconds, and then changed into another such pattern. There were only 16 such patterns in all, and then the whole, procedure repeated. Thus the grand cycle was twelve minutes. The German Post Office had no apparent difficulty in solving this system. They built a five-bank rotary switch with 36 positions, drove it with a synchronous motor so as to step every 20 seconds, repeating every 12 minutes, and controlled this operation accurately over 24 hour periods with a quartz-crystal-controlled oscillator. Once the German engineer wired this switch correctly to match the patterns, they were able to monitor transmissions 100% and receive the speech instantaneously in the clear, so that they could record the speech traffic magnetically on steel tapes. The pattern cycle was rearranged by the American Telephone and Telegraph Company (that is, the enciphering keys changed) only several times between April 1942 and April 1945; after each change it took German Post Office engineers ‘only a few hours’ to reconstruct the new patterns and their sequence. 153 Oscillographs, spectrographs, magnetophone recorders, and quartz-crystal oscillators for time control were available for this work, but well-trained ears were said to have played the most important role in the solution.

  Radio telephone conversations between Moscow, Leningrad, Irkutsk, Alma Ata, and Tscheljabinsk, involving Red Army and People’s Kommissariats, up until 1943 were enciphered by two simple methods that were said to be easily solvable by German engineers at the Army Ordnance, Development and Testing Group, Signal Branch (Wa Pruef 7). According to Corporal Karrenberg of the OKH/GdNA, these two methods of Russian enciphering were inversion, employing superimposed modulation of several audio frequencies; and distortion, by artificial raising of amplitudes of speech harmonics.

  German scientists were able to solve these two simple enciphering methods by recording the enciphered speech, making spectrograms from the recordings, and analysing them. Evidently the voice engineers could see the results of the inversion and distortion, on careful inspection, and could readily identify the frequencies and methods used for encipherment. They tried it only a few times according to Karrenberg, but were successful at will. However, access was terminated at the beginning of 1944 when the Soviets improved their enciphering methods.

  Exactly how much Anglo-American teleprinter traffic was read by the Germans is unclear but TICOM doubted they never solved any teleprinter enciphering machines themselves. When questioned on this narrow topic, Dr Hüttenhain denied that the OKW/Chi had ever worked on American teleprinter traffic, but acknowledged having received some teleprinter intercepts from Wa Pruef 7. In addition, Dr Ferdinand Voegele stated that from April to October 1944 his Chi Stelle OBd intercepted plaintext American teleprinter messages that concerned aircraft movements between the United States and North Africa, but he mentioned no other non-Morse.

  In his interviews Corporal Karrenberg said that the OKH/GdHA had a section, Gruppe VI, Referat 2A, that ‘undertook preliminary evaluation of British and American wireless teleprinter and automatic Morse traffic’, and that another section, Gruppe VI Referat 2 B, ‘picked up the traffic evaluated in Referat 2A’ and that another section, Referat I BJ, charged with cryptanalysis of Russian secret teleprinters also ‘worked on British and United States (non-Morse) systems’ but he made no references to any actual reconstruction of American or British teleprinter cryptographic apparatus.

  According to interviews with Dr Otto Buggisch, some Soviet-encrypted teleprinter traffic may have been solved by the FA in 1943, but he knew little more and asserted that the traffic had dried up, presumably because the equipment was considered redundant.

  TICOM also learned that the Swiss diplomatic (K-Type) Enigma was read regularly, probably by the OKW/Chi, although Dr Hüttenhain, who made this disclosure, did not identify which agency had made the breakthrough. The Swiss usually changed their Enigma rotors every three months, but not on the Berne–Washington D.C. link at the same time as they were made on the Berne–London link, so duplicate messages sent by the Swiss to Washington and to London during the changeover period provided the ‘cribs’ required to reconstruct the new settings. The Croat Enigma, used for both diplomatic and military traffic, was read regularly by Inspektorate 7/VI, mainly because the rotor wirings were acquired from the Berlin manufacturer, Konski & Krueger. The task was all the easier because the Croats failed to incorporate some of the more sophisticated circuitry in their version of the machine.

  TICOM recovered an impressive paper on the K-Type Enigma in the Pers Z S files that showed how the problem had been tackled, but the methods adopted were very familiar to Anglo-American cryptanalysts at Bletchley Park. Another monograph, written by Dr Buggisch on the subject of the British TypeX, was also retrieved and proved that the TypeX, originally constructed on an Admiralty contract, was based on the Enigma, and after an initial study of 10,000 letters of cipher text, all further work was abandoned.

  TICOM found that virtually all the German SIGINT agencies had devoted considerable resources to the Converter M-209, which had been employed by the US Navy and US Army Air Force, and several theoretical solutions were found. Dr Huettenhain claimed that M-209 depths were found by using a Fasensuchgeraet, a rapid analytic apparatus that may have been identical to the IBM-based device deployed against the teleprinter traffic.

  Dr Buggisch provided a statistical formula used in conjunction with IBM machines to help identify M-209 pin patterns. This consisted of writing the cipher text out ‘on the width’ of one of the wheels. However, Buggisch did not claim that he was ever able to make the test work, and TICOM believed that it would not have. Leutnant Muentz, of the OKM/4 SKL/III, developed a statistical theory for solving M-209, based on the frequent use of ‘Z’ as a word separator. However, this never worked on actual traffic. A Kriegsmarine administrator named Schultze also developed a theory on how to guess plain text from a statistical study of the cipher text, but he was never able to make it work.

  In conclusion, TICOM reckoned the Germans made little progress with the French-modified B-211 Hagelin and the Russian B-211, and Dr Hüttenhain said a French B-211 was captured and believed that an eight- to ten-letter crib could solve the wheel settings, pin settings, and pluggings, if the cipher wheel wirings were known. A Russian B-211 was also captured and a theoretical solution devised, but since no traffic was received this solution was never tested in practice:

  No great new cryptanalytic methods were developed by German cryptanalysts to assist in solving additive super-enciphered codes. They were solved as such codes usually are, by superimposing identically keyed texts (by virtue of identical indicators and by means of repetitions), removing the additive from the depths, and reconstructing, from the resultant relative code values, the basic code – unless the code book is already known.

  An example of the foregoing type of solution by the Germans is noted in the case of their cryptanalysis of the British War Office Cypher, a 4-figure super-enciphered code used between Army, Corps, and Division … which was read during the campaign on North Africa in 1940.

  Another example is the Turkish 4-figure diplomatic code enciphered by repeating additives, which was solved without regard to indicators simply by superimposing sections of messages at the period of the additive (in this case a period of 20) thereby obtaining enough depth to eliminate the additives and reconstruct the code.

  TICOM’s view that the War Office Cypher (WOC), a high-grade hand system, had been compromised confirmed reports from other sources that one copy had been captured in Norway, and another during the evacuation from Dunkirk. These coups had allowed the Germans to read the WOC traffic from August 1941, and because the WOC was used extensively by the Eighth Army in North Africa down to division level, it was of immense assistance to Rommel until January 1942, when enciphering improvements were introduced.

  The Kriegsmarine’s solutions of British naval codes were the most completely described in the final TICOM reports, as the
se had extended from before 1939 to the end of the war. When the Royal Navy introduced the S.S. Frame in December 1943, German cryptanalysts were able to analyse correctly the new cryptographic system, and read messages superenciphered in this new system for one month.

  On 1 January 1944 the British changed their basic codes and soon after began to use doubly enciphered indicators for the S.S. Frame. Consequently, the Germans could no longer read the traffic, as Wilhelm Tranow of the OKM/4 SKL/III explained:

  We came to the conclusion that we could not recover a system of this kind within six months, without having the basic book. However it was clear to us that if we were able to capture the book, we should then be able to break this system in a very short time. We provided our own proof for this … We constructed synthetic messages of our own on the pattern of the British originals. We began the first trial with 200 messages a day and broke all of them within three weeks … We then carried out a second trial with 100 messages. The staff was much more practiced and succeeded with a smaller number of messages in a shorter time.

  The Polish government in London used an additive super-enciphered code for military attaché messages, which was read regularly by the OKW/Chi until about 1943, when the Poles changed their methods of obtaining the additives. The Poles had introduced their version of the British S.S. Frame at the suggestion of the British and their stencils had from twenty-eight to forty randomly placed apertures, rather than 100 as in the British version. The German cryptanalysts, having the Polish code book from their previous solutions, were able, with it and with depths obtained by IBM searches for repeats, to reconstruct additives, discover the irregular positions of the stencil apertures, reconstruct the stencils and read the messages.