NSA (National Security Agency) "Eavesdropping On Hell", by Robert J. Hanyok

Historical Guide to Western Communications Intelligence and the Holocaust, 1939-1945

Chapter 2: Overview of the Western Communications Intelligence System during World War II

Step 1: Setting the Requirements, Priorities, and the Division of Effort

For the Allies, perhaps the most difficult step in the COMINT process was simply to decide what Axis communications to intercept, decipher, and report. To visualize the potential size of the Axis communications target is to grasp the scope and nature of the problem facing the Allied cryptologic agencies, especially early in the war: thousands of radio terminals on hundreds of radio networks around the world supporting Axis military, naval, diplomatic, security, intelligence, and commercial entities. All of these utilized hundreds of cryptographic systems from simple hand ciphers to complex book codes and intricate machines such as Enigma, Purple, and Tunny. [10] Added to this initially uncharted wilderness were the hundreds of military, diplomatic, and commercial communications networks of important neutral and Axis-friendly countries, notably Vichy France, Turkey, Sweden, Spain, Portugal, Switzerland, the Vatican, and those of Latin America.

These networks, too, had potential as sources of intelligence and could not be ignored; many already were targets of Allied cryptologists. Early in the war, the Allied communications intelligence effort could not adequately monitor the existing Axis communications networks, much less even hope to cover it completely. The British, beginning in September 1939, and later the Americans in late 1941, lacked the facilities, the personnel, and the technology to adequately monitor all of the Axis terminals that mattered. It would take time for the influx of money, training, and the development of administrative and technical support organizations to create the impressive Allied COMINT structure that existed by the end of the war.

Meanwhile, the disparity in capabilities between the Allied cryptologic agencies and the Axis communications networks forced the intelligence staffs in Washington and London to prioritize the Axis communications terminals and cryptographic systems that had to be attacked. This prioritization was determined by a number of general factors. First of all, there had to be either an already demonstrated or a reasonably high expectation that intelligence of value to the war effort could be derived from a terminal’s messages. Another factor, which could override all others at times, was the perceived current strategic or tactical military needs of the moment. For all practical purposes, these factors were military oriented. They dictated target listings throughout the war and dominated all prioritization schemes from strategic priorities to those in individual theaters of operation. The COMINT agencies, too, had their own considerations to add to the calculations for target priority. Usually, these considerations reflected the agency’s own measure of its technical ability to intercept or to exploit cryptanalytically a particular Axis terminal or a general target category such as “Japanese military.”

Above: Arlinton Hall

The feeling within the SIS leadership, for example, was that tasking of radio links was far more complex than to be left to the whims of intelligence officers. In early 1943, the SIS disagreed with the idea that the War Department’s G-2 could just hand it a list of radio terminals to go after. One memorandum stated that there were other factors “in the chain of events leading from G-2’s desires to actual results.” Rather, it continued, the priority of a target radio terminal should be determined first by whether a monitoring station could hear it and then whether any cryptographic systems used in its traffic could be read or were being studied. [11]

As an example of how requirements were levied and how they changed with the war, it is useful to look at the experience of the SIS. For the overall U.S. COMINT effort, the SIS was responsible for intercepting and processing Axis military, weather, and air force communications traffic. It also was responsible for exploiting international diplomatic communications. This was a carryover from its success with the Japanese diplomatic cipher machine known as Purple. In April 1942, G-2 gave the Signal Security Service (SSS), as the Signal Intelligence Service was soon to be called, a set of priorities for collection and processing (which included cryptanalysis and translation). [12] The first priority included all German, Japanese, and Italian military traffic. Second priority was given to all Axis military attaché communications. Axis diplomatic traffic among their respective capital cities was in the third priority. The fourth priority was for all so-called German “administrative” radio nets, which probably referred to German illicit intelligence and security radio networks. The diplomatic messages of other countries, minor Axis and neutrals, were spread across priorities five to eight. [13]

Nearly a year later, in March 1943, the priority list had changed somewhat. The priority list for collection and analysis was organized now into groups identified by letters that ranged from “A,” the highest, to “G,” the lowest. For the SSS, the Japanese Army was now the most important target. This was tagged “#1 Special Research Project,” which probably referred to the main Japanese army code that was still unexploited at the time. [14] Weather traffic was also part of Group A, as were certain diplomatic links between Tokyo and Moscow, Berlin, and Rome. German military had fallen to the bottom of Group “B,” along with Japanese diplomatic messages to Europe (other than other Axis capitals) and “Security” (intelligence, espionage and security elements such as police) messages. [15] The latter reference to “Security” traffic is interesting because the SSA still lacked the means to intercept consistently the communications of the German security agencies such as the Abwehr (military intelligence), the SD, and the German Police. The Americans had few monitoring sites that could intercept such communications in the European theater. There were some U.S. Army radio intercept companies in England and North Africa, but they were tasked with collecting tactical German military radio traffic. A note attached to the tasking for these intercept units indicated that the coverage of the German security elements was primarily for the purpose of tracking the volume of traffic and reporting this to Army counterintelligence. [16]

The SIS already was somewhat familiar with German Police communications and cipher systems. In January 1941, then SIS Major Abraham Sinkov, one of William F. Friedman’s original cryptanalytic staff, had headed a small technical exchange mission to GC&CS. It was this mission that provided the British with a working Purple and Red Japanese diplomatic cipher machines. During the ensuing technical discussions, one of Britain’s foremost codebreakers, John Tiltman, provided Sinkov with detailed information on how the so-called German Police cipher worked. He was informed that the descriptor “German Police” applied to the systems used by the Schutzpolizei, the SD, the SS, and the line Order Police battalions and regiments. Sinkov was told that the information in the police messages was useful for mapping associated German military units. [17] However, at the time the U.S. had no way of intercepting these communications; the cryptanalytic information was useless except for training.

As for the remainder of the revamped 1943 SIS overall requirements, these consisted mostly of targeted diplomatic communications from most minor Axis, neutral, and some minor allied nations. As in the previous year’s priorities they remained spread across the lower priority Groups from “C” to “G.” For example, the tasking requirement for intercept of diplomatic messages between Washington and major neutrals was now at the Group “D” level. The requirements to intercept diplomatic messages from the Vatican, Latin America, and European countries to elsewhere other than Washington were covered in Groups “E” and “F.” [18]

In the various combat theaters of operation, cryptologic working arrangements between the British and Americans followed de facto national theater command responsibilities. In Europe the British were preeminent for the first two years of the war. A gradual cooperative effort with the United States Navy’s OP-20-G cryptologists developed during the lengthy U-boat campaign in the Atlantic. In the European Theater of Operations (ETO), the GC&CS, and the supporting British service units, remained the principal Allied cryptologic agency when it came to the Axis military and air force. In the Pacific theater, the American army and naval cryptologic organizations supervised allied intercept and code- breaking operations. In the China-Burma-India Theater, there were British stations in Ceylon and India that had American contingents. Also, Canadian, Australian, and New Zealand personnel served at various Pacific and Asian sites with the Americans and British.

These worldwide operations required that the United States and Great Britain eventually had to establish corresponding divisions of effort and responsibility for setting requirements, intercept, processing, and dissemination. Otherwise, there would be problems with redundant operations, and misallocated resources. Also, the rules, methods, material, and channels for technical and intelligence exchanges needed to be ironed out between the two countries. The two countries had to make a general agreement to govern operations around the world. For the first eighteen months of the war, the GC&CS and the SIS and OP-20-G had a number of separate and limited reciprocal interchanges. By mid-1943, these various working relationships were institutionalized in a series of agreements between the British and the U.S. War Department, which culminated in the so-called BRUSA (Britain-United States of America) Agreement of June 1943. [19] (The U.S. Navy would sign a separate, more limited, agreement. It was known as the Extension Agreement and applied to an earlier 1942 exchange arrangement known as the Holden Agreement.)

The main provisions of the BRUSA Agreement were the exchange of technical intelligence (sources and methods) and a division of effort in the daily activities of collection, analysis, and dissemination of intelligence. The two Allies formally agreed to the exchange of finished intelligence. There was no exchange of “raw” (undecrypted) intercepts, except for U-boat messages, and possibly some examples for training purposes. [20] The monitoring sites, though, would send their “raw” intercept to the national center responsible for processing it. The agreement did not include so-called “nonservice” (nonmilitary) intercepts. This uncovered category principally entailed Axis and neutral diplomatic and intelligence-related (illicit espionage and internal security organization) messages. Both countries continued individually to decrypt and exchange diplomatic translations, though a later mechanism for exchange between GC&CS and Arlington Hall was established in August 1943. As a result of this later agreement, the two Allied cryptologic agencies targeted the diplomatic traffic of every major Axis power, minor Axis ally, minor Allied power, and significant neutral, though the total amount collected against each country varied. As for the category of Axis intelligence messages, another division of effort was arranged. The British collected and processed German intelligence and security-related messages in occupied Europe: the United States, principally the United States Coast Guard, collected and processed Axis overseas illicit and overt espionage radio traffic, notably Abwehr messages from Latin America, North Africa, and the Far East.

In essence, the BRUSA Agreement formalized the realities of the division of effort in the combat theaters that had existed since 1942. The Americans took the responsibility for Japanese service and nonservice communications, while the British oversaw that of the German and Italian military and security forces. This agreement did not preclude either country from collection and analysis of each other’s missions. In fact, both sides exchanged personnel who were then integrated fully into each other’s efforts. Three special SSA units were established in Britain to work alongside the British in the areas of collection, analysis, and security. American cryptologists could be found at various British intercept and analytic sites, including Bletchley Park. U.S. radio intercept units stationed in England and North Africa intercepted and analyzed Axis communications in the European and North African theaters, but they passed the take to the British analytic centers. [21]

Great Britain retained authority over the production of communications intelligence in Europe, while the Americans controlled similar activity in the Pacific. Because of this division of effort, German military communications remained a lower priority for the U.S. Army codebreakers at Arlington Hall Station for the rest of the war. [22] Tactical communications intelligence, that is, intercept and analysis of plaintext messages or those encrypted by low-level codes and ciphers that were sent by Axis tactical or operational-level combat units, was produced by Allied cryptologic elements attached to Allied military commands or units in Europe and North Africa. Theater or local commanders controlled the operations of these units and were the main recipient of that type of communications intelligence.

As mentioned above, the priorities for collection and processing shifted during the war according to current strategic needs and capabilities. This shift was especially obvious when it came to diplomatic targets. A good example of the shifting priorities of an individual target country’s communications can be seen in the Allied effort against Switzerland’s diplomatic radio traffic. During previous European conflicts, Switzerland had remained neutral. As a result, that nation had taken on certain roles such as representing the foreign interests of belligerents. These included monitoring the conditions in prisoner-of-war and internee camps, supporting International Red Cross relief activities, and allowing Swiss locales to be used for unofficial contacts between combatants. Considering these historic roles, Switzerland’s diplomatic communications naturally were a target of interest for the Allied cryptologic agencies.

The GC&CS had been analyzing Swiss diplomatic messages since 1939, and the Americans began their separate attack in December 1942. But both agencies produced few translations during this period. There were some technical reasons for this paucity of communications intelligence. One was that for their European message traffic the Swiss relied mostly on the European cable network that was inaccessible to Allied communications monitoring sites. In addition, the Swiss used a large number of codes and ciphers - over ten systems, mostly manual ciphers and codes, and an early version of the Enigma cipher device for their diplomatic traffic. Yet, the most important reason for the small output of translations of Swiss traffic was that, early in the war, the messages to and from Bern appeared to carry little intelligence of value. [23] The intercept of Swiss diplomatic messages continued or the purposes of training and cryptanalytic continuity, but not for reporting intelligence. By early 1943, the Americans downgraded the intercept priority of various Swiss diplomatic terminals to fourth and fifth priority on its list. [24]

This situation changed in the summer of 1944 when the Allies discovered that the Swiss diplomatic cables now carried important information on the conditions in the Balkan capitals that were being overrun or threatened by the advancing Red Army. Of particular interest was the situation in Budapest, Hungary. The several changes in governments and the German-supported coup in October 1944 had kept the Hungarians in the conflict. Furthermore, the Germans were in the midst of a roundup, which had begun in the summer, of the sizable Jewish population in Hungary for transport to the death camps and slave labor details in German war industries. A review of the diplomatic translations from this period issued by the SIS shows a marked increase in the number of published translations of Swiss diplomatic messages. Allied interest in the Swiss diplomatic messages continued late into the war as concern grew over the possible fate of prisoners and internees under Japanese control. The Swiss, working in concert with the International Red Cross, reported their findings on camp and prisoner conditions throughout Asia. Another point of interest for the Allies concerned the activities of German banking officials who were negotiating payments and commercial accounts with the Swiss for German purchases of war material and currency exchanges. (See pages 104-110 for more details on this last point.)

One of the problems that bedeviled Allied cryptologists who tried to set requirements was that the intelligence value of the information carried on various communications networks or individual terminals could not always be predicted. The above example of Switzerland shows how reality did not always match expectations. On the other hand, useful intelligence could come from unexpected sources. For example, one of the best sources for insight into Nazi Germany’s strategic plans and Hitler’s own appreciation of possible Allied moves came from the Purple decrypts of Japan’s ambassador in Berlin, Baron Oshima Hiroshi. Oshima had developed a rapport with most of the Nazi hierarchy, especially Hitler and Joachim von Ribbentrop, the German foreign minister. They confided to Oshima some of the Reich’s most important secrets. These personal connections made his long reports to Tokyo a gold mine of information on a number of topics, from German development of new weapons to Hitler’s own appreciation of the Allied troop strength and strategic options for landing in Western Europe in 1944. [25] Oshima reported nothing on the concentration camps or massacres in the western Soviet Union, though a few of messages to Tokyo suggest he was aware of the situation of the Jews. (See pages 100-101)

Left: Japanese ambassador to Germany Baron Oshima meeting with Nazi foreign minister Joachim von Ribbentrop In a similar fashion, British codebreakers were probably surprised by the first bits of information about the massacres and camps that appeared on certain German Police radio networks, such as those servicing the police units in the Soviet Union and SS and SD units in occupied Europe. The Allied intercept and processing of German Police and SS communications originally was performed as a supplementary source to intelligence about the German military, principally to gather administrative data and order of battle information.

The police and SS communications also contained intelligence on German civilian morale and other domestic security concerns, information about the results of Allied air raids, and escapes by Allied POWs.26 The police messages were also targeted since the manual cryptographic system used was similar to systems used by Wehrmacht units. [27]

However, London and Washington would never make gathering intelligence about the fate of Europe’s Jews and other groups targeted for destruction by the Nazis a major requirement for agencies like GC&CS and SIS. Both would collect and process some information about the roundups, massacres, and the camps from Police and diplomatic communications. Yet, this information was a by-product, perhaps even incidental to the coverage security and diplomatic nets. The primary purpose of the COMINT requirements placed on Bletchley Park and Arlington Hall was to gain intelligence dictated by the military exigencies of the war. Whatever advantage was gained from the intercept of Axis communications related to the Nazi atrocities was of little use to the current Allied prosecution of the war, except in a limited way for propaganda against the Axis. [28] Later, the British would gather it with the intent to use intelligence as evidence in proposed postwar war crimes tribunals. (See pages 50-51 for eventual fate of this intelligence.)

Notes

[9] SRH-349, Appendix, “Bulletin production - Average Daily Volume of Translated Messages,” 1 December 1941 - 31 August 1945.

[10] The Enigma and Purple cipher machines are well known from the literature of World War II. Less famous, but important as sources of intelligence, were some other Axis cipher machines. Tunny was the cover name applied to the Lorenz Company’s on-line enciphered teleprinter known to the Germans as Schluesselzusatz 40 or SZ-40. GC&CS first broke it in mid-1941.

[11] Memorandum from Major Brown to Mr. Friedman, 28 January 1943. RG 457, HCC, Box 1432, Folder “SSS Intercept Priorities Memorandum.”

[12] The Signal Intelligence Service underwent many title changes during the war. From 1929 to 1942, it was the Signal Intelligence Service. In June 1942 it was briefly renamed the Signal Intelligence Service Division. One month later it was again renamed the Signal Security Division (SSD). In 1943 the SSD became the Signal Security Service. In late summer 1943, the SSS became the Signal Security Agency (SSA). It remained the SSA until 15 September 1945 when the SSA was renamed the Army Security Agency (ASA).

[13] Memorandum, “Expansion of Signal Intelligence Service.” 18 April 1942, SRH-145, 25.

[14] The SIS would achieve its first cryptanalytic breakthrough against the main Japanese army code in April 1943. In January 1944, Australian troops captured the entire cryptographic library of a Japanese Army division that clinched the mastery of the Imperial Army’s codes and ciphers. See Edward Drea, MacArthur’s Ultra, 61-92, for details of the year-long struggle.

[15] That the German military traffic was placed in Group B was a further indication that the two countries were arriving at a division of effort dictated by geography and expertise. Throughout 1942, SIS had been working on an ability to exploit Enigma, but lacked sufficient intercept, expertise, and technical information. The British refused to allow the SIS to join in the exploitation. However, by mid-1943 a compromise for exchange was worked out that formed the basis for a more formal exchange of all Axis intelligence. See Benson, 99-108.

[16] One interesting multiforce, multinational intercept operation against clandestine Axis radio stations in North Africa occurred in North Africa in late 1944. A special effort included personnel from the SIS, the OSS and Free French forces, who operated intercept and direction-finding stations in Algiers, Tunis, and Morocco. RG 457, HCC, Box 1417, Folder 4619, “Special Report on Clandestine Radio Activity in North Africa.” The OSS, though excluded from performing cryptanalysis by FDR’s directive of July 1942, and prohibited from receiving COMINT by a May 1942 agreement among the Army, Navy, and FBI, still established its own intercept and D/F stations outside of the U.S.

[17] Major Sinkov’s Report of Cryptographic Mission, 1941.” NARA, RG 457, HCC, Box 1296; also see Benson, 19-20.

[18] Memorandum, “Priorities for Operations of the Signal Security Service.” 8 March 1943. SRH-145, Collection of Memoranda on Operations of SIS Intercept Activities and Dissemination, 1942-1945. (Fort George G. Meade, Maryland, 1983), 92.

[19] The story of the BRUSA Agreement, which proceeded from earlier negotiations, is, itself, a tale of bureaucratic complexity. The American Army and Navy each made agreements with the British. They then had to make one between themselves. See Benson, 97-133, and Bradley Smith, 105–170. See also “Army-Navy Agreement Regarding Ultra.” RG 457, HHC, Box 1413.

[20] This major exception, the processing of undecrypted U-boat radio traffic by both the United States and Great Britain, began in earnest in late 1942 following the Holden Agreement of October 1942. After the installation of the American-built “bombes” at the OP- 20-G Headquarters on Nebraska Avenue, N.W., in Washington, DC, the U-boat cipher text was transmitted there for decryption. The resulting translations were then transmitted to the appropriate Allied naval commands. Eventually, the bombes were turned against German Luftwaffe and Wehrmacht Enigma traffic. For more on the Holden Agreement, see Ralph Erskine, “The Holden Agreement on Naval Sigint: The First BRUSA?” Intelligence and National Security, (Volume 14, No. 2, Summer 1999), 187-197.

[21] SRH-349, 32.

[22] Memorandum: Assistant Chief of Staff for Intelligence, Subject: Priorities for Operations of the Signal Security Service.” March 8, 1943, SRH – 145.

[23] David H. Hamer, Geoff Sullivan and Frode Weirud, “Enigma Variations: An Extended Family of machines.” Cryptologia (Vol. XXII, No. 3, July 1988); SSA “Effort against the Swiss Cipher Machine (SZD).” NARA RG 457, HCC, Box 1284; History of the Signal Security Agency in World War II, Vol. 2, 237.

[24] Memorandum, March 8, 1943. SRH-145.

[25] Boyd, 178-9.

[26] Phillips, 5-10.

[27] Hinsley, Vol. 2, 670.

[28] On how the Allies handled this information, among others see Walter Laqueur, The Terrible Secret (New York: Owl Books, 1998), 202-4.

Continue to Part 6