Thursday, 30 May 2013

Railway Enigma and Other Special Machines

Recently my old friend David Hamer asked me a question about the Railway Enigma, the machines used by the German railways — Deutsche Reichsbahn. His question was: "Has anyone ever located a Railway Enigma  or is aware if one currently exists in a museum somewhere?" I was not able to answer his question because I simply don't know if any of these machines ever survived the war. So I decided to see what would happen if I could create some more interest for this machine and other, similar machines.

In 1998 David Hamer, Geoff Sullivan and myself tried to show that Enigma is not one single type of machine, but rather a family of machines built on the same cryptographic principle, by publishing the Cryptologia article Enigma Variations: An Extended Family of Machines. I think we were successful in establishing this fact but it is still the Army and Naval Enigmas that get most of the attention. This is understandable because these two models were the two main cipher machines used by the German Wehrmacht. The fact they they were broken, first by the Polish and then later by the British and American codebreakers, made them historic. The decodes provided intelligence that in many ways were crucial for the conduct of the war and as a whole signal intelligence is said to have possibly shortened the war by two years. However, the Army and Naval Enigma machines were not alone in supplying this intelligence. The greatest allied undertaking, Operation Overlord, the landing of allied forces on the continent, was heavily dependent on intelligence from two other systems, Tunny and the Abwehr Enigma.

We should be grateful to Brigadier John Tiltman and Dr. William Tutte who made the initial break into Tunny, the Lorenz SZ42 on-line teleprinter machine, and to Dillwyn Knox who made the first break into the Abwehr Enigma, Enigma G. The importance of Tunny in supplying strategic intelligence for Operation Overlord is probably well known. The importance of Dilly Knox's section ISK, Intelligence Service Knox, in supplying the Double-Cross operation Fortitude with information is perhaps less well known. The D-Day deception operations were based on feeding the Germans truthful but false information through the use of turned Abwehr spies. However, these operations would have been nothing more than a lottery without the feedback the ISK and ISOS, Intelligence Service Oliver Strachey, sections provided.

Enigma G is one of the commercial Enigma machines. The other well known commercial machine is Enigma K. They both have their origin in the early commercial model Enigma D, which can be better seen in the Enigma Family Tree. The family tree shows that the Enigma G, which internally in the Enigma firm Chiffriermaschinengesellschaft Heimsoeth und Rinke (H&R) was called Ch. 15b, rather should be called G31. This is to better distinguish it from the earlier Zählwerk Enigma model A28, which was internally called Ch. 15. The Enigma K, internally Ch. 11b, has a similar history of previously being called A27. This machine first received the K designation in 1936. This curious history means that until 1931 the Zählwerk Enigma machines had serial numbers starting with A exactly like the early 3-wheel Heeres Enigma machines equipped with a Steckerbrett, plugboard. The commercial Enigma A27 also carried A serial numbers until 1936 when it became Enigma K and received K serial numbers; the first machine in the new series being K 201.

While the plugboard Enigma machines kept their wheel wirings intact throughout their existence the commercial models Enigma K and G were supplied with a large number of special wheel wirings, Sonderschaltungen. There are a few exceptions to this rule. Some 3-wheel Heeres Enigma machines that were used for special purposes received specially wired wheels. There are two categories of these machines, those equipped with Alpha and Delta wheels and those in the S-series. I have already described these machines in more detail in the Cryptologia article Spanish Enigma: A History of the Enigma in Spain. The machines with Alpha and Delta wheels had normal serial numbers in the A-series and externally they cannot be distinguished from a normal Heeres Enigma machine, while the S-series machines had a letter S attached to their serial number such as A 17245S. 

A very special version of the Heeres Enigma is the B-series machines. Apparently only 24 of these machines were made, B 201 to B 224. The machines have wheel and plugboard markings in red paint and Senior Engineer Willi Korn of H&R mentioned that it was their belief they were used at the Führer Headquarters. Hitler had a mobile communication unit that accompanied him on his special train (Führersonderzug) and this unit might have used some of these machines. So far only one B-series machine is known to have survived the war, B 207 manufactured in 1937, which is at the US National Cryptologic Museum (NCM) at Ft. Meade next to the National Security Agency, NSA. This unique machine was originally standing on a pedestal in one of the museum rooms and it was used as an interactive exhibit for the public to play with. I think his error has now been rectified and it is now on display in a closed and locked display case. Let us hope NCM takes good care of this priceless machine and gives it the a special place among their exhibits. As "Hitler's private Enigma" it might very well get the attention it deserves.

The 1937 manufacturing date is interesting. It shows that these machines were in use very early, long before Hitler built any of his many headquarters. However, its deployment fits with the date for the extension and modification to Hitler's home in the Bavarian Alps, the Berghof, which was finished in 1936. As Hitler then started using the Berghof more frequently it is possible this was the reason they wanted to have another secure communication channel that could not be intercepted and read by the normal Wehrmacht users. Another clue to the users of this machine can be found in a letter from OKW/Ag WNV/Fu II b to H&R dated 11 May 1942. The letter concerns the urgent repair of the Enigma machine B 201 that belongs to the cipher office (Schlüsselstelle) at OKW.

Enigma Machine B 207
Figure 1. Hitler's private Enigma? Rare Enigma Machine B 207. Photo: David Hamer, NCM Foundation

B 207 wheels with red numerals
Figure 2. B 207, close-up showing the wheels with red numerals. Photo: David Hamer, NCM Foundation

The mystery deepens with a Top Secret (Geheime Kommandosache) and very urgent order for two B-series machines in 1943. On 15 May 1943 H&R received a letter from OKW/Ag WNV/Fu II b with an order for two cipher machines for a very special connection (Sonderverbindung). The two machines with the serial numbers B 222 and B 223 were to be wired after the connections in drawing Ch 11 Tz 184 a-c for the wheels I, II and III respectively. Due to the exceptional secrecy surrounding these machines the reflector, (Umkehrwalze — UKW), could only be wired in the presence of an official from Ag WNV/Fu. Furthermore, the acceptance tests of the machines had to be performed with the standard Wehrmacht wheels and UKW because the special wheels could not be left with H&R after they had been wired. The secrecy was indeed exceptional. What happened in May/June 1943 that necessitated such a special, private communication link? The question is still open.

Special B-Series Enigma
Figure 3. Order for special Enigma Machines B 222 and B223

Now that I perhaps have got you interested in Enigma machines with special wirings let us have a closer look at those machine. If you take a look at our article Spanish Enigma you will see that alone in Spain and Italy there were many commercial machines in use with special wiring and this was already before the start of the war. During the war the Enigma G and Enigma K machines were supplied to a large variety of government and official users almost all of them equipped with specially wired wheels. There are only a few Enigma G machines that we know about. We have the famous G 312 that was stolen from BP for so to be returned to Jeremy Paxman at the BBC, several that was used by the Dutch military and now are in closed museum collections there, the Hungarian G 111 that turned up at an auction in Germany and G 316 that was found in Spain some years ago. NSA and NCM are reported to have another two machines, G 194 and possibly G 215. So at present probably not more than 10 or 11 of these machines seems to have survived the war. Personally I find that very strange because there were many Abwehr offices throughout German occupied territories and also in the some of the neutral countries. For instance there must have been several of these machines in Norway but I have yet to hear that any have been found here.

The origin of G 312 has always been a mystery, but now we know it probably is connected with Spain in some way because the wiring of G 312 is the same as G 316. The wiring of G 316 is given in the Spanish Enigma article. This shows that knowing the wiring of a machine is very important when trying to identify the machine and trace its history. The A-series machines that clearly can be identified as Heeres Enigma will most probably have wheels with the well know wiring of these machines, but for all the Enigma G and Enigma K machines we can not be sure what kind of wheels they are equipped with. I would therefore urge all owners of such machines to have the wheels measured to see if their wiring correspond to what is expected for their machines. I am more than willing to help and give advice on how to go about measuring wheel wirings. 

Many of the Enigma K machine in public ownership are what we call Swiss Enigmas, which means they belong to the group of commercial machines that were sold to the Swiss before and during the war and which finally were sold to the public in 1994. Many of these machines are mentioned in the list of known locations of Enigma machines maintained by David Hamer. I might later try to compile a complete list of the serial numbers of the Swiss  Enigmas with information about when they were delivered when that is known. At the moment I am missing some of the information I need to make such a list.

Enigma K machines that are not Swiss Enigmas might be machines with special wirings. Let us now look closer at the Railway Enigma. This machine has some historic importance as it probably was the first German Enigma machine that BP recovered the wiring of during the war. Previously Dilly Knox had recovered the wheel wiring of the machines used by the Italians and Spanish during the Spanish Civil War. The process he developed for recovering  the wheel wiring was then well known and it is also described in detail by Turing in his Treatise on The Enigma. The initial break was yet again made by Brigadier John Tiltman who was an expert on working out messages in depth — messages that are encrypted on the same or very similar key settings. In theory it sounds easy but to succeed you need a very sharp mind and a lot of perseverance. After Tiltman had recovered the plaintext of the messages at the end of July 1940, Peter Twinn quickly recovered the wheel wiring using the boxing method developed by Knox. Peter Twinn started as Knox's assistant in 1939 and worked in the BP Cottage with Knox until early 1940 when he was lent to Hut 8 to work with Turing on Naval Enigma. Later the Railway keys were turned over to Hut 6 and Twinn returned to the Cottage to work with Knox on the Abwehr Enigma. The full story about the battle with the Railway Enigma keys is explained in our Enigma Variations article.

By a stroke of luck we know exactly when the German Reichsbahn started to use the Enigma K. The first traffic that BP detected in July and decoded in August 1940 ceased on 27 August 1940 and new traffic was not detected before at the end of January 1941. Later the traffic increased and especially from regions in Eastern Europe, Russia and the Balkans.  It would therefore seem that the use of the machine started some time in early 1940. In the documents recovered from H&R at the end of the war there are three documents that refer to the Railway machines. The first document is a letter from Oberkommando des Heeres (OKH)/Chef HNV IV to H&R dated 22 January 1940. It concerns an order for 18 Enigma K machines for the Generaldirektion (administration) of the Eastern railway (Ostbahn) in Cracow. The document is shown in Figure 4. The 18 machines have the serial numbers K 448 to K 465 and the wiring diagrams have the drawing numbers Ch 11 Tz 271 for the reflector (Umkehrwalze — UKW) and Ch 11 Tz 272 a-c for the three wheels I, II and III respectively. Another document that gives an overview of a large number of wiring diagrams together with their creation dates give 17 January 1940 as the date for these drawings. This fits well with the handwritten note added by H&R that refers to them placing an order for these machines with their production firm Konski & Krüger (K&K) on 17.01.1940. Another interesting fact is that even if the order is for machines for the Reichsbahn the order is processed through the Army High Command (OKH). However, another handwritten note seems to indicate direct contact between H&R and the German Ministry of Transport (Verkehrsministerium). This probably explains how H&R was able to place their order with K&K before the arrival of the letter from OKH/HNV (Heeres Nachrichten Verbindungsabteilung).

Figure 4. The first order for Enigma K machines for the German Railways

On 18 April 1940 H&R receives another letter from OKH but this time from the Chief of Army Equipment and Commander of the Replacement Army (Chef Heeres Rüstung und Befehlshaber der Ersatzheeres) referring to the order 222-25-09/39 from 1939 concerning Enigma K machines. From the previously delivered K-machines OKH now wants H&R to rewire the wheels for the ten machines K 432 to K 441 with the Reichsbahn wiring. OKH stresses that the order is extremely urgent and they ask for speedy execution of the rewiring. The ten machines shall be delivered to the Reichsbahn administration in Berlin. In a handwritten note at the bottom of the letter the destination of the machines are indicated. Two machines were delivered to Berlin, while each of the following cities got one each: Essen, Frankfurt am Main, Hamburg, Karlsruhe, Cologne, Mainz, Munich and Stuttgart. The letter also refers to the machines that were delivered in January but know they list the serial numbers as K 448 to K 467, 20 instead of 18 machines. If this is an error or if the order was subsequently increased to 20 machines is not known. It might be the order was increased because H&R was usually very quick to spot such errors and they would then normally make a note to this effect. The information in these two documents suggests that the Reichsbahn started to encipher their communications with the Enigma some time early in 1940. It appears that Enigma communications were already being conducted in April 1940 and that a realistic date for the start of these communications is probably late February or early March.

The last order for the Reichsbahn that I know of arrived at H&R on 17 June 1943. The order was from OKW/Ag WNV/Fu II b and it concerned 20 Enigma K machines with the serial numbers K 893 to K 912. The wheels were to be wired exactly as the other Reichsbahn machines. The delivery of the machines were programmed to start on 1 October 1943 with a delivery of 10 machines per month. Unless there were other orders that I don't know about we know that the German Reichsbahn had at least 50 machines and that their serial numbers are: K 432 - K 441, K 448 - K 467 and K 893 - K 912.

So let the hunt start. Who will be the first to find a Railway Enigma machine? What is there to gain? Perhaps not very much apart from some fame and being the proud owner of a relatively rare Enigma specimen. The other benefit will be that we will know the German wiring of this machine. In our Enigma Variations article we gave the recovered wiring of the wheels based on the information in Turing's Treatise on The Enigma. Unfortunately, I did not read Turing carefully enough and therefore the wiring we gave are not for the wheels in their neutral position, Ringstellung A, but rather for ring position Z. Furthermore, BP happened to wrongly identify the wheels such that wheel I and wheel III are swapped, but even then the wiring will not be exactly as the German original wiring. When recovering wheel wirings cryptanalytically you are free to select which pin you label as 1, hence there are 26 possible ways to number a given pin. This is called an offset or clip difference, but there is yet another problem and that is called twists. After you have found the connections between the pins on one side of the wheel and the pads on the other side you still don't know how the pads are physically related to the pins. If you imagine a wheel as two disks with pins on one and pads one the other with flexible wires in between it is clear that you can rotate one disk freely with respect to the other. This is the wheel twist. Here you also have a freedom of 26 twist positions, such that totally you have 26 x 26 = 676 possible equivalent wirings of which only one is corresponding to the German wiring. Unless you have access to a lot of other information such as many messages with exact indicators etc. you are not able to identify the German wiring from the 676 possible choices. Today the only way to identify the German wiring is probably to find one of these machines.

You can download the Railway Enigma Documents here. Good luck with your hunt and please keep me posted about what you find.

Sources: Personal research notes and documents from the TICOM files T1715, T1716, T1717 and T1718 in the TICOM collection at the German Foreign Office, Bestand Rückgabe TICOM, Politisches Archiv des Auswärtigen Amts, Berlin.

No comments: