Thursday, June 14, 2012

The Soviet K-37 ‘Crystal’ cipher machine

The Soviet Union used during WWII a large number of 2,3,4 and 5-figure codes of various types. These were all hand systems. When it came to machine ciphers they had in 1941 three different machines in service, the K-37 ‘Crystal’ off-line machine and the M-100 and B-4 cipher teleprinters.

The K-37 was a copy of the Hagelin B-211 with Cyrillic characters on the keyboard. According to before the outbreak of WWII, Boris Hagelin was forced (by the Swedish authorities) to sell two B-211 units to the Russian Embassy. The Russians took the design and copied the machine. At the same time they converted the 5 x 5 matrix into a 5 x 6 one, in order to accommodate more characters. It allowed 30 letters of the Cyrillic alphabet to be used.

According to a very interesting article in production started at Leningrad plant No 209 in 1940 and by the summer of ’41 roughly 150 K-37 machines were in use.

The Germans were able to capture one K-37 machine in 1941 and they evaluated its security. They found that it had low security and could be solved on a 10-letter crib.
The war diary of Inspectorate 7/VI shows that in August ’41 a captured Soviet cipher machine was examined by the cryptanalysts Pietsch, Denffer and Hilburg and a report was prepared.

In September the analysis of the device was complete:

From the TICOM reports it seems that they never had the chance to try their theoretical solution on actual traffic, as the machine was not used by the Soviet forces in the West.
This is confirmed by the war diary of Inspectorate 7/VI, since no further references to the K-37 can be found, apart from a study in October 1942 on intercepted messages that reached the conclusion that they were not enciphered on the K-37.

Information from TICOM reports:

From TICOM I-2 ‘Interrogation of Dr. Huettenhain and Dr. Fricke at Flenshurg,21 May 1945’, p1-2



From TICOM I-64 ‘Answers by Wm. Buggisch of OKH/Chi to Questions sent by TICOM’, p4

This was an electrical machine, almost exactly similar to the French B 211 but without the "Ueberschluesseler"(added E. wheel at one point) of the B211. It was considerably less secure than the B211 and a theoretical solution was worked out which did not need much text. B. had forgotten the details on this. The K37 had been captured, but never really used by the Russians.
From TICOM I-58 ‘Interrogation of Dr. Otto Buggisch of OKW/Chi’, p5

K-37 - A Russian machine, same principle as B211, but more primitive model was captured in 1941, and a theoretical solution worked out by HILBURG and Dr. V. DENFFER. They found it could be solved on a 10 letter crib. The work remained purely theoretical as no traffic in the machine was ever received.
From TICOM I-92 ‘Final Interrogation  of Wachtmeister Otto Buggisch (OKH/In 7/VI and OKW/Chi)’, p4

10. K-37 differed from B211 in lacking the "Surchiffreur", or ‘’Ueberschluesseler’’, a sort of Enigma wheel by which the path of the current was turned to another channel at one point, crossing  over and exchanging positions with another path instead of continuing parallel. Buggisch called this an X effect, and said it greatly complicated analysis, as it was hard to tell when it was being employed in place of the parallels.

Perhaps the K-37 was not used in the Western areas of the Soviet Union because its low security had been discovered by Soviet cryptologists or they learned that one of their machines had been captured and suspected that the Germans had found a solution.
However the German assertion that the Russians never used the K-37 is not correct. It was definitely used in the Soviet Far East in 1945. The Americans intercepted this traffic. It seems reasonable to assume that the K-37 was also used prior to ‘45 in the Soviet Far East.

Postwar history
The captured German K-37 was apparently handed over to the Western Allies at the end of WWII. The Americans built an analog model of the K-37 which they called Sauterne Mark I.

This machine was attacked after the war by the Anglo-American codebreakers. It was used on Red army circuits in the Far East.
In February 1946 US cryptanalysts managed to reconstruct its internal settings, in March the first message was decoded and by April a regular supply of decrypts was being produced.

The US success was short-lived as K-37 traffic dried up by 1947.
Sources:The Secret Sentry’,, various TICOM reports, Intelligence and National security article: ‘Behind Venona: American signals intelligence in the early cold war’, ‘The Russian Target’ by Matthew M. Aid, cryptomuseum

No comments:

Post a Comment