Sunday, April 12, 2015

The US Division Field Code

When the United States entered WWII, in December 1941, US military and civilian agencies were using several cryptologic systems in order to protect their sensitive communications. The Army and Navy only had a small number of SIGABA cipher machines so they had to rely on older systems such as the M-94/M-138 strip ciphers and on codebooks such the War Department Telegraph Code, the Military Intelligence Code and the War Department Confidential Code.

Another system prepared for the Army was the Division Field Code. This was a 4-letter codebook of approximately 10.000 groups and in the 1930’s several editions were printed by the Signal Intelligence Service (1). However the introduction of the SIGABA and especially the M-209 cipher machine made this system obsolete. Still it seems that the DFC was used on a limited scale, during 1942-44, by the USAAF and by US troops stationed in Iceland and the UK.
Examples of DFC training edition No 2:

Solution of DFC by German codebreakers
The German Army and AF signal intelligence agencies were able to exploit this outdated system and they read US military messages from Iceland, Central America, the Caribbean and Britain. Most of the work was done by field units, specifically the Army’s fixed intercept stations (Feste Nachrichten Aufklärungsstelle) Feste 9 at Bergen, Norway and Feste 3 at Euskirchen, Germany.

According to Army cryptanalyst Thomas Barthel several editions of the Division Field Code were read, some through physical compromise (2):

The DFCs (Divisional Field Codes).
(a). DFC 15

In use in autumn 42, broken in Jan 43. Traffic was intercepted on a frequency of 4080 Kos from US Army links in ICELAND (stas at REYKJAVIK, AKUREYRI and BUDAREYRI). Stas used fixed call-signs till autumn 43, and thereafter daily call -signs. This field code was current for one month only. It was a 4-letter code, non-alphabetical, with variants and use of "duds" (BLENDERN). It was broken by assuming clear routine messages were the basis of the encoded text, such as Daily Shipping Report, Weather Forecast etc.
(b) DFC 16

This was current for one month, probably in Nov 42. It was similar to  the DFC 15 above.
(c) DFC 17

This was current from Dec 42 to Feb 43. About the latter date one or two copies of the table were captured. Very good material was intercepted from ICELAND, also from 6 (?) USAAF links in Central America, Caribbean Sea etc. Traffic was broken and read nearly up to 100%.
(d) DFC 21

This succeeded the DFC 17. Results were the same.
(e) DFC 25

Current only in CARIBBEAN SEA area, and read in part.
(f) DFC 28 

This succeeded the DFC 21 in summer 43. It was used by the ICELAND links and the 28 (or 29) US Div in the South of ENGLAND. The code was read, Now and again it was reciphered by means of alphabet substitution tables ("eine Art von Buchstabentauschtafel") changing daily. This method was broken because the systematic construction of the field code was known.
(g) DFC 29

A copy of this table was captured in autumn 43. It was never used, PW did not know why.

The War Diary of the German Army’s signal intelligence agency OKH/In 7/VI shows that the DFC was called AC 6 (American Code 6) and several editions were solved in the period 1943-44. Most of the processing was left to field units, with only a few messages solved by Referat 1 (USA section) of Inspectorate 7/VI. The report of March 1943 says that the captured specimen DFC 17 could be used to solve the preceding and following versions (since they were constructed in the same way) and it showed that the code values retrieved by field units and the central department through cryptanalysis were mostly correct (3).
The Luftwaffe’s Chi Stelle was also interested in the DFC and according to Dr. Ferdinand Voegele, the Luftwaffe's chief cryptanalyst in the West, USAAF messages from the Mediterranean area were read (4).

The 29th Infantry Division and the invasion of Normandy
In 1943 the M-209 cipher machine replaced the M-94 strip cipher as the standard crypto system used at division level by the US Army, however older systems like the DFC continued to be used for training purposes. The US military forces in Britain took part in many exercises during the latter part of 1943 and early 1944, since they were preparing for the invasion of Western Europe and some of their training messages were sent on the 28th edition of the Division Field Code.

These messages were intercepted and decoded by the German Army’s KONA 5 (Signals Intelligence Regiment 5), covering Western Europe.  NAAS 5 was the cryptanalytic centre of KONA 5 and its quarterly reports (5) show that training messages from the US V Expeditionary Corps and the 29th Infantry Division were solved.

The solution of these messages allowed the Germans to identify the 29th Infantry Division and considering the unit’s rule during operation Overlord it is possible that they gave the Germans vital clues about the upcoming invasion of France.

(1). Rowlett-1974 and Kullback-1982 NSA oral history interviews

(2). CSDIC/CMF/Y 40 – ‘First Detailed Interrogation on Report on Barthel Thomas
(3).War diary Inspectorate 7/VI - March 1943

(4). TICOM IF-175 Seabourne Report, Vol XIII, p9 and 16.
(5). E-Bericht der NAASt 5 Nr 1/44 and Nr 2/44.

Sources: Frank Rowlett NSA oral history interview - 1974, Solomon Kullback NSA oral history interview - 1982, CSDIC/CMF/Y 40 – ‘First Detailed Interrogation on Report on Barthel Thomas’, War diary Inspectorate 7/VI, War diary NAAS 5, TICOM IF-175 Seabourne Report, Vol XIII ‘Cryptanalysis within the Luftwaffe SIS’, DFC training edition No 2, Division Field Code No 4
Acknowledgments: I have to thank Rene Stein of the National Cryptologic Museum for the Rowlett and Kullback interviews and Mike Andrews for the DFC pics.

No comments:

Post a Comment