Winston Churchill called the cracking of the German Enigma codes “the secret weapon that won the war.”1 Fred Winterbotham, the RAF officer responsible for distributing the secrets gleaned in Enigma decrypts, wrote “[the war’s outcome] was, in fact, a very narrow shave, and the reader may like to ponder [...] whether or not we might have won had we not had Ultra."2
The intelligence that resulted from reading the encrypted German messages, codenamed Ultra by the British, was instrumental in winning the war. The breaking of the codes started in Poland in 1933. By 1944, when WW II engaged much of the world, the British were able to read most of the enciphered traffic within two days of transmission, usually sooner. By one estimate, roughly half a million messages were deciphered by the end of the war3, contributing greatly to the strategies of the Western allies.
How was it possible that something so important as the military plans of the entire German Wermacht were not better protected? How was the decipherment of Enigma accomplished?
This paper will assume that the reader has a basic understanding of the Enigma machine and how it enciphers messages. A very thorough description of how it works can be found at http://en.wikipedia.org/wiki/Enigma_machine.
During the 1930’s and 40’s the Enigma machine was considered uncrackable using the resources available to a code-breaking team. The number of permutations created by a three-wheel Enigma was astronomical and it was estimated by the designers that it would take 900 million years for an eavesdropper to work through them all4. There were, however, two weaknesses that were exploited by the Allies.
The first weakness, the machine design itself, was a minor one. When Enigma enciphered a letter, it was guaranteed that the resulting enciphered letter would not be the same letter. For example, ‘A’ could be translated into any letter other than ‘A’, ‘B’ would never be enciphered as ‘B’, etc. By itself would not have allowed the Allies to read Enigma ciphertext but it did assist cryptographers in finding the human errors.
The more important weakness was the human one. The mistakes made by the Enigma operators and the poor procedures put into place by the German military were legion; without them the Poles and the British would have had no hope of cracking them. Gordon Welchman, one of the officers in the British program, said “the machine as it was would have been impregnable if it had been used properly” but pointed out twelve serious errors in procedure that, if corrected, would “have stopped us cold.”5 Some of these errors will be discussed further.
The Beginning – The Poles
The cracking of Enigma starts in France when the executive of the German Defense Ministry Cipher Office approached the French embassy in Berlin inquiring how to sell secret documents to the French Secret Service. On November, 1931, Hans Thilo Schmidt met two French agents in a hotel Belgium and handed over manuals explaining how the German Army operated the Enigma machine. Schmidt would continue to supply the French government with Enigma-related documents until his capture and execution in 1943.6 The French cryptology department was small and could not figure out how to discover the keys used by the Germans so they shared the knowledge with Poland, with whom they had recently signed an alliance.7
The Poles were in an unenviable position, a new country carved out between the losers of the WW I, both of which were eyeing their old territories. Looking for any edge, Poland created the Polish Cipher Bureau (PCB) and started offering a cryptography course at the Mathematics Institute at the University of Poznan. They had had some success during the 1920’s with simpler ciphers but had been stopped when the Germans switched over to the Enigma machine. The documents supplied by Schmidt, however, gave them enough clues that they could begin to think about cracking the new codes. The largest advancements were made by three graduates of the Poznan Math Institute who took advantage of a major procedural blunder.
Before an Enigma operator enciphered his message, he would set the wheels on his machine to the correct key for the day. He would then pick a random three-letter code, tap it out on the keyboard, and transmit the enciphered results. He would then switch the wheels to match the three-letter code he had chosen and type out the message. The advantage of this was that only the first three letters of the message would be enciphered using the daily key. If hundreds of messages were enciphered this would significantly lower the amount of ciphertext that an enemy cryptographer could analyze when trying to determine the key. This was all spelled out in Schmidt’s documents.
The blunder was that the procedure specified that the random three-letter code was to be typed out twice to limit the possibility of errors. Marian Rejewski realized that the patterns created by the repeated characters could be used to determine the order of the rotors and their relative positions to each other. The PCB called these letters ‘indicators’ and began to create a catalog of indicators and the possible rotor configurations. Any time a known indicator was found in a message the number of rotor configurations that could be used to create that message was reduced by 40%. Every additional indicator reduced the set of configurations another 40%. If the size of the set was reduced enough then the key could be found using trial-and-error.8
Using this technique, along with others, the Poles were able to decipher their first message in 1933. 9 Over the years the catalogue of indicators became larger and more Enigma messages could be deciphered. In 1938 the catalogue became large enough that sorting through them by hand became too laborious so cryptographer Henryk Zygalski came up with an idea of translating the logic of the indicators into large perforated sheets.
Each sheet represented a different indicator and holes were cut into the sheet to show all the possible rotor positions that could create that indicator. As messages came in with different indicators, sheets where stacked on top of a glass table with a light shining underneath, each sheet reducing the number of holes that could be seen. If enough sheets were stacked only a few holes could be seen, allowing the cryptographers to quickly find the few possible rotor positions.10
The indicators did not give the final answer, however, just the relative rotor positions. The three letters of the key and the plug-board cables still needed to be determined. Rejewski wired together six Enigma machines to simultaneously run through the different permutations not ruled out by the indicators. These machines, called “bombe” (plural “bomby”) by the British and French, would click when the correct combination was found.
By 1938, using the sheets and bombe machines, the Poles were able to determine a key in about two hours time11 and they decrypted about 75% of Enigma traffic.12 This changed in December of that year when the Germans made a change to the Enigma machine that defeated the Poles. Up until that time there were only three different rotors that could be inserted into the three slots in the Enigma machine. Since they could be inserted in any order there where six possible configurations of the rotors. In December the Germans added two more rotors, increasing the number of configurations to sixty, outstripping the resources available to the PCB.13
In 1939 Poland realized that war was soon approaching and in July invited her British and French allies to meet with the PCB. There they revealed everything they new about the German Enigma and cracking their codes.
Before their meeting with the PCB, neither the French or British had deciphered a single Enigma message. Many intelligence officers considered it impossible. In 1938 the French Secret Service, the Deuxieme Bureau, floated a plan where French agents would leak information to the Germans pretending that the Enigma had been solved. The hope was that the Germans would switch to a different scheme, preferably one that was easier to solve.14 Fortunately the plan was scratched.
The British and the French were completely taken aback when they learned of the success of the PCB at the Warsaw meeting. The perforated sheets were explained, along with the bombe machines. The PCB also supplied both allies with Enigma replicas, complete with all the wheels.15 After the war started many of the PCB cryptographers evacuated to France and later—after France surrendered to the Germans—to Britain.
The Warsaw meeting was important to the British efforts to crack Enigma in two ways. First it provided them with the techniques to determine the daily keys. Second it proved that cracking Enigma was possible, something that many British officers had deemed impossible. Because of this doubt the British cryptography department, the Government Code and Cipher School (GC&CS), was a small operation with little funding. When the success of the Poles became known the British war department stepped up its commitment to the CG&CS.16
The GC&CS was housed in a Tudor building bought by the Foreign Office in Bletchley Park, called BP by all the workers. It was here that the gold mine supplied by the Poles was taken and digested. With much greater resources on hand the British were able to create Zygalski sheets to encompass all sixty of the new rotor settings. In addition, mathematician Alan Turing met with the Polish cryptographers in France and was able to design a more powerful bombe to help sort through the possible key settings for each day.
With Polish help the cryptographers were able to decipher their first Enigma message, which had been sent the previous October, on January 17, 1940. By April the BP codebreakers were able to read messages within 24 hours.17 However, the British military did not yet have any process in place to pass such vital information on to their generals. It took time for a process to be created and for the military to trust the intelligence. When the military leaders realized the quality of the intelligence and its timeliness they acted quickly to set up a system to disseminate it the commanders in the field.
The greater impact of these early messages was that BP was able to create a picture, or a catalog, of the types of messages that different branches of the German military used. This soon became critical when the Germans finally fixed their huge security flaw.
On May 1, 1940, just before the invasion of Norway, the Germans changed their encryption procedure. The Enigma operators no longer determined the three-letter key from the distributed codebooks but instead chose one at random. This random key was transmitted in the clear. The operator would then choose another three-letter key, encipher it using the original key, and transmit the resulting ciphertext, this time only once. Lastly the rotors were changed to match the second random key and the body of the message was encrypted.
The British had been expecting the Germans to stop repeating the initial key sequence as soon as they found out about it. BP codebreaker Dilly Knox, when he learned of the blunder, commented that the German procedure “may at any moment be cancelled”. 18 Turing kept this in mind as he was developing his bombe and designed it to look for matches of ciphertext against possible plaintext, known as “cribs” (described below). If a match was found then the key could be inferred. The early decrypts supplied them and without them the bombe would have been much less effective.
When the Germans changed the key procedure BP was momentarily blacked out. The techniques created by the Poles could no longer be used and the perforated sheets were now useless. The search was on for other ways to guess the Enigma configurations. The main attack was made using cribs.
Cribs were another example of poor security procedures used by the German military. Many encrypted messages were created using a template where most of the words were the same. One example of such messages were the daily weather reports sent in by the U-boats. Other cribs were created when the same message was sent over a long stretch of days. One unit sent the message “nothing to report” every day to headquarters over a span of months because it was required to send a daily status report.19
If a message was expected to contain a crib it would be entered into one of Turing’s bombe machines to let it try to work out the key. Even the bomby could not try every key permutation and so other techniques were required to narrow down the possibilities.
There were many different tricks to determining the configuration of the Enigma rotors. One of the first to be useful was a realization by a young cryptographer John Herivel. Herivel imagined how a lazy cipher clerk might save himself trouble when setting up the Enigma machine for the new key each day. The clerk would first choose the correct three rotors, then set the alphabet ring on each rotor, and place it in the machine. If a clerk was lazy, Herivel thought, he might close the lid and, instead of changing the rotors to some random value, just use the ones that were already displayed. Herivel’s contribution was in realizing that, when the operator inserted the rotor into the machine, the letter determining the ring setting would probably still be facing him. It was likely that the three letters appearing in the window would be close to the ring settings of the new key. This technique was known as the “Herivel tip”. 20 When the BP codebreakers looked at the daily intercepts, Herivel was right. Given enough messages a few operators were bound to be lazy and a grouping of the first three letters could vastly narrow the search for the ring settings. Welchman remarked that “the 17,576 possible ring settings had been reduced to six probables.” 21
There were other examples of sloppy practices used by the operators. The cryptographers called them “cillis”, not only because they were silly mistakes made by the operators but because one of the first settings worked out by BP was ‘CIL’.22 One operator was asked to send a test message and he encrypted a message consisting only of ‘T’s. An alert cryptographer noticed that the ciphertext had no ‘T’s and deduced what had happened, which led to the cracking of that day’s settings.23 When the Germans introduced a new rotor one operator sent a message forgetting to use the new rotor. Realizing his mistake he re-encoded the exact same message with the new rotor and sent that one as well. BP was able to crack the first message and then used it to figure out the internal wiring of the new rotor.24
Putting together all these techniques allowed BP to read the messages of most branches of the German military. BP had the most success reading the Luftwaffe messages, codenamed “Red” by the British cryptographers. It was broken on May 22, 1940, and was read uninterrupted throughout the rest of the war. This was largely because the volume of intercepts and different cribs was very high.25
The German army never realized that the Allies had solved their Enigma encryption. This oversight has astounded historians, especially since the Germans themselves had some success in cracking some of the codes of their enemies. Peter Calvocoressi, a BP officer in charge of relaying Ultra intelligence to the RAF, decided this was because the Germans, in starting the war, had an aggressive personality and could not fathom that a passive tool like code-breaking could be so important. The British, ever protective of Ultra, had teams who’s sole responsibility was to monitor the practices of it’s own radio operators to make sure they weren’t making “cillies”.
All of the different services in the German military used different key settings and in some cases different procedures. The German Navy created a lot of problems for BP because they used the most sound practices of all the branches. They stopped using the double-key encipherment in 1937 and they always had more rotors available, with a total of eight by the end of the war. 26 In 1942 the navy added a fourth stationary wheel that confounded BP for eight months. Because of their increased security, the techniques used to crack the other Enigma channels wasn’t enough to solve the Naval Enigma. The British, and later the American, Navy was instructed to try and apprehend the codebooks from captured or sinking ships and U-boats. The first success with procurement came was the sinking of U-33 in January 1940 within a mile of the coast of Scotland. One of the sailors forgot to toss some of the rotors in his pocket overboard and they were discovered by the British. The following April a German vessel, attempting to pass itself off as the Dutch fishing boat Polares, was captured along with most of its Enigma documents. Without these successes and other raids it is likely that the Naval Enigma would have remained too secure to be cracked with any regularity.
A discussion of Enigma is not complete without a short list of examples where Ultra was used by the allies. BP contained a group whose sole job was to forward important information to the front-line commanders, each of who had a two- or three-man team to help him make sense of the intelligence. By 1944 this group was sending out over 100 messages a day, all deemed important enough to notify the commanders immediately.27
The starkest example of the importance of Ultra was in the war in the Atlantic against the U-boats. Intelligence from Ultra allowed convoys to be re-routed around known positions of the U-Boat wolf packs. At the same time anti-submarine ships and airplanes were arranged to attack the U-Boats. A graph of tonnage sank during the war by the U-boats could be matched almost exactly with the progress at BP. An increase in tonnage sunk coincided with a decrease in BP’s ability to read the naval Enigma due to changes in procedure. In 1942, when BP was mostly in the dark, the battle was almost lost. When BP finally mastered the new procedures the tonnage went down immediately. Churchill was confident that the British could survive the war if they could win the battle in the Atlantic.
Another example occurred in North Africa against the famed German Afrika Corps, led by General Rommel. In 1940, after the Battle of Britain, the Germans called off Operation Sealion, their planned invasion of the British isle; Ultra intercepts confirmed this. Churchill then ordered tanks that had been held for the defense of the homeland to instead be sent to North Africa to battle against Rommel’s armies. Ultra also told the British of the timetables for German supply ships crossing the Mediterranean from Italy to bring new troops and material to the German army. This enabled the Navy and RAF to sink a large percentage of Rommel’s replacements. Together these led to the Afrika Corps being strangled in the desert. 28
Near the end of the war Ultra depicted the disarray of the German High Command. Hermann Goering sent a message to Himmler saying that Hitler was no longer fit to rule the country and that he was taking over. A day later this was rescinded by another message sent by a contrite Goering.29
The breaking of the Enigma code was a monumental achievement by the Polish and BP cryptographers. Churchill declared that the war would have lasted two more years without Ultra. The German failure to protect their codes had disastrous results and illustrated the importance of using secure cryptography. Perhaps the biggest lesson for the modern cryptologist is that the most secure cipher is only as strong as the laziest clerk. Most of the mistakes made by the Enigma operators were expressly forbidden by the training they received and the manuals that came with the machines30. The procedures for enciphering plaintext need to take into account the stress of war and limit the number of “cillies”.
Calvocoressi, Peter, Top Secret Ultra, (New York, Pantheon Books, 1980)
Freedman, Maurice, Unraveling Enigma, (London, Leo Cooper, 2000)
Sebag-Montefiore, Hugh, Enigma: The Battle for the Code, (U.S., Wiley & Sons, 2000)
Wikipedia.org, various topics and authors. When possible, the reference given by the Wikipedia.org article will be listed.