Thursday, May 10, 2012

US diplomatic codes A1, C1, Gray and Brown

The diplomatic communications of the United States during WWII were a major target for the codebreakers of the Axis nations. Unfortunately for the Allied side the security of State department codes was compromised both cryptanalytically (through mistakes by cipher clerks etc) and physically by breaking into the embassies and making copies.

This allowed the Japanese, Italians, Finns and Germans to read US secret communications.
German personnel were able to solve both low and high level systems used by American diplomats. I have gone through their success with the high level M-138 strip

cipher here.
However this was not their only success with State Department codes. The A1 and C1 codebooks were read plus the low level Gray and Brown codes.

According to ‘United States Diplomatic Codes and Ciphers, 1775-1938’, the codes used by the State Department in the late 30’s were the low level Gray and Brown (used unenciphered) along with the enciphered codes A1,B1,C1,D1 and the new M-138 strip cipher. All these systems were letter codes.
Gray was introduced in 1918 and Brown in 1937. The A1 and C1 codebooks, which are the ones that interest us, were used for a long time. A1 was introduced in 1919 and C1 in 1927 (1). Whether it was a good idea to use these codes for so long is something I leave to the reader to ponder. These codebooks were enciphered with substitution tables which changed after several months.

How were these codes used? The low level systems and the strip cipher were used by all embassies. On the other hand it seems the A1, B1, C1, D1 codebooks may have been used only by specific embassies. For example a British report from September ’44 states that the A1 was used mainly by the embassies in Stockholm and Madrid (2).


These codebooks continued to be used during WWII, even though US officials suspected that they were not secure. A security survey of the State Department cryptosystems carried out in late 1943 (3) mentions the Gray, Brown, A, B, C codebooks and the strip cipher but it seems that use of the D code had been discontinued. The findings of the committee were that the enciphered codebooks were probably compromised.





German exploitation of State Department codebooks

The agencies that attacked US diplomatic codes were the Foreign Ministry’s Pers Z , Goering’s Forschungsamt and OKW/Chi. This definitely led to duplication of effort. Still there was cooperation between these agencies. It seems OKW/Chi and the FA concentrated on breaking the enciphering tables, while Pers Z focused on book-breaking.
The German codebreakers had no problem solving and reading the low level Gray and Brown codes throughout the war. The A1 codebook was attacked in 1941 and 80% ‘broken’ before the Japanese gave them a copy. C1 was recovered by Pers Z and the enciphering tables solved by OKW/Chi. According to post war interrogations it took them two months to solve the enciphering tables so in the third month messages could be decoded currently.

Let’s take a look at the relevant reports:

Forschungsamt:
From TICOM I-25 ‘Interrogation of five members of the RLM/Forschungsamt at Schloss Gluecksburg ,near Flensburg on 15th and 21st June 1945’

Q.22 What work was done on American systems?
PAETZEL: We read U.S. plain systems and afterwards a U.S. system consisting of a five letter book recyphered by Mono-alphabetic and bigram substitution though this was not solved currently. The tables changed within the message and this change was shown by an indicator.

Q.23 How long did you take to read messages after a change of elements?
PAETZEL: We only read back material.

Q.24 How old?
PAETZEL: Several months, sometimes over a year. The successor to this system was not broken because there was no indicator in the middle of the message to show where the tables changed.

Paetzel was head of the Forschungsamt’s cipher research department.

Pers Z:
From TICOM I-22 ‘Interrogation of German Cryptographers of Pers Z S Department of the Auswaertiges Amt’

P10 :
81. Herr Zastrow is a bookbreaker and has been engaged on U.S. systems for 27 years. The Americans used mainly 5 letter books with 10 recyphering tables for mononalphabetic and bigram substitution to each book. The tables ran for two to six months. Later other substitution tables began to be used, with 5 indicator groups to each table. The tables were changed in the course of a message, the indicator for the new table being encyphered with the previous table.

 82. The German designation for these systems was B1, B2, B3 etc to B10, (B= Buchstabe) American equivalents  he remembered were:
B6a=A1

B6b=B1
B7=C1

B8=Brown code
The B6a book was known through physical compromise. The ‘Brown code’ had been captured but it had been broken before capture. The reconstruction of the book had taken 21/2 years.

P16:
129. Dr. Schultz is a mathematician and statistician by profession. He worked before the war in the Statistisches Reichsamt and was transferred to Dr. Kunze's part in 1939 at the outbreak of hostilities. He had worked on the Japanese machine until 1940 when it ceased to be soluble. It had been broken before he came to work on it.

130. He then worked on the American B7 (C1) code and solved the recypherment. There were 80 to 100 tables the first of which were difficult to reconstruct, the later ones getting progressively easier. He built up 80 tables himself. When the tables changed, so that the sequence was no longer cvcvc, he continued investigations, but work was discontinued in favor of the American strip ’02’.

OKW/Chi:
From TICOM I-2 ‘Interrogation of Dr. Huettenhain and Dr. Fricke at Flenshurg, 21 May 1945’, p3

Q. WERE OTHER TYPES OF DIPLOMATIC TRAFFIC BROKEN?
A. ENCIPHERED CODE, USING CONVERSION TABLES. BETWEEN TWENTY AND THIRTY  TABLES WERE AVAILABLE FOR EACH MESSAGE, AND THERE WERE TEN DIFFERENT CONVERSION SYSTEMS, WHICH WERE CHANGED QUARTERLY. IT TOOK ABOUT TWO MONTHS TO BUILD UP TABLES, SO THAT WE COULD ONLY READ THE TRAFFIC CURRENTLY DURING THE LAST MONTH OF THE PERIOD.

Q. WHAT TYPE OF CONVERSION TABLE WAS USED?
THERE WERE RECTANGLES OF APPROXIMATELY TWENTY COLUMNS OF RANDOM MIXED ALPHABETS WITH THE PLAIN TEXT ALPHABET RUNNING VERTICKILY ALONG THE SIDE. AT FIRST THE CODE GROUPS WERE MADE UP OF FIVE LETTERS IN THE SEQUENCE — CON. VOWEL CONS. AT FIRST VOWELS COULD BE CONVERTED ONLY TO VOWELS, BUT LATER THEY COULD ALSO BE CONVERTED TO CONSONANTS. THE ORIGINAL VOWEL TO VOWEL CONVERSION WAS A WEAK P0INT IN THAT IT REDUCED THE POSSIBILITIES. ABOUT THIRTY GROUPS WERE ENCIPHERED PER TABLE, THEN AN 'ENCIPHERED INDICATOR (USING OLD TABLE) REFERRED TO THE NEXT TABLE. THIS INDICATOR WAS ALSO USED IN BREAKING AS IT USUALLY OCCURRED WITH THE SAME INDERVAL.

Q.WHAT OTHER DIPLOMATIC TYPES WERE BROKEN?
A. UNENCIPHERED FIVE LETTER CODE, WHICH HAS BEEN IN USE SINCE 1920. THIS WAS NOT IMPORTANT TRAFFIC. WE ALSO BROKE A DOUBLE TRANSPOSITION SYSTEM BY MEANS OF DEPTH.

Huettenhain was the chief cryptanalyst of OKW/Chi.

From TICOM I-199 ‘Report on interrogation of Oberregierungsrat Nikolai Rohen head of Anglo-American section of OKW/Chi’, p2-3
He remembered the following captured material:

American: Brown Code. Gray code. A3 Code, all from the Japanese. Military Intelligence Code. Attache code; both from the Italians.
………………………………………………………………………………………………………………………………………………………

American Cyphers: They also read a system with ten bigram substitution tables, which used a new table after every 20-25 letters.

From TICOM I-201 ‘Interrogation of Franz Weisser , Dr Phil Studienassessor of Anglo-American section of OKW/Chi’, p2-3
c) C-1: a code of about 52,000 words enciphered by tables (two 2 letter and one 1-letter) 

The name of the code was, I believe C-1. The code was solved by the Auswaertiges Amt, the tables subsequently solved by OKW/Chi, the last time in the winter of 1943-1944 in the traffic of Spain and Switzerland (the news dealing principally with the Black List). The code seems to have been abandoned.
d.)  A-1: a code of about 160,000 words enciphered  by one letter tables

The code was compromised by Japan and the tables solved by OKW/Chi.
e) Gray Code.

The code had been worked on from the beginning and was completely deciphered by 1936,

f) Brown Code.
Work was started immediately after the introduction in collaboration with the Auswaertiges Amt with the result that the code was nearly solved when in May 1941, we got a copy from the Japanese. This was the one code which justified the existence of the Anglo-American department in the last phase of the war.

In addition to the aforementioned reports a message in June ’44 by the Japanese military attaché in Budapest (ULTRA/SAC/JMA/309) has clues concerning German success with the A1 and C1 codes. That message says that the German authorities would probably share the A1 substitution tables used up to 31 December ’43.The new tables valid from January to April ‘will probably have been readable in May’. Regarding the C1 the Germans would offer the parts of the code which had been recovered and the enciphering tables valid up to 31 December ’43 (still being used at that time in the Berne traffic).


Conclusion


The solution of the mid/high level A1 and C1 together with the M-138 was an important defeat for the Allied side. Even the supposedly low level Brown code carried important information from around the world, as can be seen in decoded US messages found in NARA RG 457.
Although by late 1944 the Americans went over to machine ciphers the harm was done. During the period 1941-44 the German leadership could eavesdrop on the secret American diplomatic initiatives. This gave them valuable information.


It should also be mentioned that the personnel of the OSS station in Berne used diplomatic codes when their own systems were overloaded. This was an additional benefit for the Germans.
It seems that one of the advantages of their success was the ability to keep an eye on attempts by minor Axis nations to exit the war. One such case has been described by David Kahn and concerns a failed Finnish attempt to negotiate with the Americans. This initiative failed because it became known to the German foreign minister Ribbentrop through codebreaking (4).

Another similar case concerning talks between Hungarian and American officials seems to have been betrayed by weak codes (5).
How many events like these took place but are not known to us? There is no doubt that the information from US diplomatic sources had a significant impact on German policy. It is the job of researchers to uncover more information.

 
Notes:

(1) ‘United States Diplomatic Codes and Ciphers, 1775-1938’, p249-50, Report titled: ‘Communications systems in use by the Department of State
(2) HW 40/132 ‘Decrypts relating to enemy exploitation of US State Department cyphers, with related correspondence’.

(3). US National Archives and Records Administration - collection RG 457- Entry 9032- box 1384 - 'JCS Ad hoc committee report on cryptographic security of government communications', State Department security survey 1943
(4). ’Finland's Codebreaking in World War II’ in ‘In the Name of Intelligence: Essays in Honor of Walter Pforzheimer’

(5). ‘Sparrow Mission: A US Intelligence Failure during World War II’, Intelligence and National Security 27:2

No comments:

Post a Comment