Alan Turing: The Enigma The Centenary Edition

by Andrew Hodges

  Appropriately, the Luftwaffe messages so laboriously and expensively deciphered at Bletchley in March 1940 turned out to consist mostly of nursery rhymes sent as practice transmissions. Even there, where at least they were busy with very exciting work, a sense of unreality and anticlimax was often felt. It was the same at Cambridge. Alan would return there occasionally for leave days, to work on mathematics and to see friends. At King’s they had all dutifully trooped down into air raid shelters (all except Pigou, who refused to compromise with the Luftwaffe), but the promised bombardment had not come. Three quarters of the children evacuated to Cambridge had returned home by mid-1940.

  Yet the war had not been over by Christmas; Alan had exercised his option to suspend his fellowship for the duration of the war on 2 October 1939, and although his course on the Foundations of Mathematics had been advertised in the lecture list, it was not to be given. And there was Finland. Once during this period there was a party in Patrick Wilkinson’s room, where Alan met a third-year undergraduate, Robin Gandy, who was reading mathematics and also rather conscientiously trying to defend the Communist party line. ‘Hands off Finland’ was double-talk such as Alan despised, but he liked Robin Gandy, and instead of marching off in disgust, led him on gently with Socratic questioning to arrive at a contradiction.

  And one thing that was real, even in the phoney war, was the conflict at sea. Just as in the First World War, it was the strength and the weakness of the offshore European island that war with Britain was an attack upon the world trading economy. One third of all merchant shipping was British, and apart from coal and bricks there was scarcely a commodity in which the island was self-sufficient. Despite the blockade Germany could survive by pressing the resources and labour of Europe into its service. But British survival depended upon the ocean lanes. There lay the critical, paramount asymmetry.

  It was the sea war that would become Alan’s particular province. In early 1940 the different Enigma systems were divided among the chief crypt-analysts, who were allocated huts outside the Bletchley mansion. Welchman took over the army and air force Enigma systems, in Hut 6, joined by a number of new recruits. Dillwyn Knox took the Italian Enigma* and that used by the German SD, again with new recruits. These systems, which did not use plugboards, suited his psychological methods. And Alan was allocated Hut 8 in which to head the work on the naval Enigma signals. Other huts housed sections translating and interpreting the output; thus Hut 3 dealt with the army and air force material issuing from Hut 6, while the naval signals, if and when any were produced, would be interpreted by Hut 4, which was headed by Frank Birch.

  Alan probably knew little of the context in which he was working, apart from the general air of urgency that issued from Hut 4. This was probably just as well, for the context was not exactly an encouraging one. He was working for the Admiralty, which only grudgingly had relinquished naval cryptanalysis to GC and CS. Traditionally, the Royal Navy expected autonomy. As possessor of the world’s largest fleet, the Admiralty might be supposed capable of organising warfare for itself. Yet it had signally failed to learn the lesson that navies depended not only upon force but upon information, for guns and torpedoes were impotent unless in the right place at the right time. Like the giant Cyclops, ‘Our Fighting Navy’ was decidedly one-eyed. Naval Intelligence was embodied in an organisation that anyone of the new generation would find absurdly Victorian, if not criminally incompetent.

  Only in the First World War had any Naval Intelligence Division been set up, and this had declined in peacetime into Kafkaesque fantasy. In 1937, the NID was11 ‘… neither interested in nor equipped to collect or disseminate information about the organization, dispositions, and movements of foreign fleets … the situation was very little better than it had been … in 1892…. Large old-fashioned ledgers were used in which to enter in longhand the last known whereabouts of Japanese, Italian and German warships…. These reports were often months old, and only once a quarter [were] the supposed dispositions of foreign navies … issued to the Fleet.’ The Movements Section of the NID (consisting of a single part-time officer) ‘did not even subscribe to Lloyd’s list, which would at least have provided a daily and highly accurate record of all the world’s merchant ships. Reports of the movement of warships from the Secret Service were virtually non-existent… The possibility of locating ships at sea was … even more remote than that of obtaining up-to-date information about them when they were in port.’ The admirals did not really want to know.

  By September 1939 a new man, Norman Denning, had somewhat improved the position. There was a card-index instead of ledgers, a direct telephone link to Lloyds, and a Tracking Room on which a plot of merchant ships positions could be up-dated. Links with GC and CS were not so successful. Indeed, the cryptanalytic organisation, captured by the Foreign Office after the First World War, tended to be treated as the enemy. Denning continued to plot its reconquest by the Admiralty until February 1941.

  But the forward-looking Denning had also been able to establish the principle that a new sub-section of the NID, the Operational Intelligence Centre, which replaced the old Movements Section, should receive and coordinate information from all sources. This had been impossible in the First World War and represented a revolutionary advance. On the eve of war the OIC stood by with a staff of thirty-six. They had many problems to overcome, but the main problem of 1939 was that they had virtually no information to coordinate. Like Tweedledee, the Admiralty could hit out bravely at anything it could see, but it could see very little.

  Occasionally, Coastal Command aircraft would catch sight of U-boats, and the RAF had been persuaded to inform the Admiralty when this happened. Aerial reconnaissance was limited to the hiring of a commercial pilot to take shots of the German coastline. Information from agents in Europe was ‘scanty. The best … came from a black market dealer in silk stockings with a contact in the German Naval Post Office, who from time to time was able to give the address of mail for certain ships, thus providing some fragmentary clues to their whereabouts.’ When the Rawalpindi was sunk in November 1939, the Admiralty were unable to discover even the class of ship responsible. And as for signals, not only were the Enigma-enciphered messages indecipherable, but the German Navy12

  went over to wartime wireless procedure shortly before the attack on Poland, putting an end to the possibility of following its movements by correlating call-signs with the results of direction-finding, and it was to be months before work on the German naval signals system at GC and CS and in the [OIC] … made it possible to produce even tentative deductions on the basis of Traffic Analysis. The first step was to distinguish U-boat from other German naval communications, and it is some indication of the extent of the black-out that this elementary advance was not made until the end of 1939.

  Until the outbreak of war, ‘the naval sub-section of the German Section’ of GC and CS ‘which was started with one officer and a clerk as late as May 1938, still had no cryptanalysts.’ It was just one aspect of the failure even to try to meet the German challenge. The prospects were better now, with the help of the Poles and with Bombes on the way, but the overall picture was dire:13

  Since the outbreak of war GC and CS had continued to give work on the G[erman]A[ir]F[orce] variant of the Enigma priority over its attack on the naval traffic. It had done so for two good reasons. The GAF traffic was more voluminous. Over and above that, those who worked on the naval Enigma had been held up first by the fact that the German Navy used the machine more carefully than the GAF, so that by the beginning of 1940 GC and CS had been able to break the settings for only 5 days of 1938, and then by the discovery that, sometime about the outbreak of war, the naval machine had undergone more radical modification than had the GAF’s. During 1940 small amounts of captured naval cypher material had confirmed that, while both still used only three wheels at a time, the naval Enigma’s wheels were selected from … 8 instead of from 5.

  To make any headway, Alan would need something more to go on. �
��From December 1939 GC and CS had left the Admiralty in no doubt about the urgency of this … requirement, but the Admiralty had had little opportunity to meet it.’ But there was (at least at sea) a war on, which meant that the German authorities had to work on the assumption that the Enigma machine itself would be captured. Indeed this was so; the Polish revelations had only given GC and CS seven months start in this respect, for ‘Three Enigma wheels had been recovered from the crew of U-33 in February 1940.’ But this ‘had not provided a sufficient basis for a further advance.’ Possession of the naval machine, while necessary, would have been far from sufficient. If the German navy used the machine ‘more carefully’, then its key systems were perhaps much less transparent than the foolish repeated triplets exploited by the Poles. And a few days of sparse peacetime traffic would provide a slender basis upon which to mount an attack.

  Then the sea war spread to the land, with the German attack on Norway forestalling British designs. The Anglo-French response was not helped by the fact that the German cryptanalytic department, the Beobachter Dienst, was able to read a number of their messages, as indeed they had been doing all the time since 1938, and that these were used with great effect. At the end of the campaign, the Commander-in-Chief of the Home Fleet complained that ‘it is most galling that the enemy should know just where our ships … always are, whereas we generally learn where his major forces are when they sink one or more of our ships.’ In the final withdrawal from Narvik, the aircraft carrier Glorious was sunk by the Scharnhorst and the Gneisenau on 8 June. The OIC did not know the position of the Glorious, let alone of the German warships, and learnt of the sinking from an open victory broadcast.

  Norway brought Bletchley Park into the war, inasmuch as the main Luftwaffe key, and an inter-service key, were read ‘by hand methods’ throughout the campaign, and revealed a good deal about German movements. Even on the naval side, Hut 4 was able to do work on traffic analysis which could have helped with the Glorious. But there were no arrangements for putting the information to use – not that the conditions in Norway itself were such that it could have been used to much advantage. One negative achievement was that the OIC was now obliged to take some notice of Bletchley. The desperate need for better naval intelligence was now clear. ‘At the outset of the campaign the Admiralty’s own ignorance was complete. When it intervened to give the orders which resulted in the first battle of Narvik on 9 April, it did so in the belief, based on Press reports, that one German ship had arrived there, whereas the German expedition to Narvik had reached the port in ten destroyers.’

  It was in this context that an almost miraculous chance of helping Alan’s work on naval Enigma was thrown away. For14

  On 26 April the Navy captured the German patrol boat VP2623, while she was on passage from Germany to Narvik, and took from her a few papers… More might have been achieved if VP2623 had not been looted by her captors before she could be carefully searched; and the Admiralty at once issued instructions designed to prevent such disastrous carelessness in the future. As it was, except that they provided some information on the extent of the damage sustained by the German main units during the Norwegian campaign, the decrypts were of no operational use.

  The capture of cipher hardware was to be expected and allowed for; the taking of flimsy, water-soluble pages* of current instructions for the use of the machine was a very different matter.

  While the parliamentary upheaval meant that Winston Churchill now ceased to be responsible for this and other muddles, and instead took on the far greater muddle which was called the war effort, the ‘instructions designed to prevent such disastrous carelessness in the future’ were symbolic of an equally significant change. This time it would not do to have the military men behave as in some glorified Footer match, with the old masters giving earnest pep talks from touchline, and back room boys running dutiful errands. The lesson of the public schools was obsolete, for patriotism was not enough. They had to apply intelligence, at all levels, or they were lost. This was the conflict that would dominate the British war.

  Meanwhile the work on the Luftwaffe Enigma, the Bletchley success of early 1940, was taking the first steps towards military usefulness. The steps were faltering, for on 1 May 1940, the ‘German authorities introduced new indicators on all Enigma keys except the Yellow’,† The perforated sheets had come only just in time to start off the treasure hunt; now they were almost useless. But there were ‘German mistakes in the few days after the change of 1 May’, very likely the classic one of sending out messages in both old and new systems. So by 22 May, Hut 6 was able to find out the new (‘Red’) system for the main Luftwaffe signals, and from that date to break it virtually every day thereafter. By that time, however, the German forces were at the Somme and closing on Dunkerque. The Bletchley success did not come soon enough to reveal German intentions during the first phase of the western attack. Indeed, ‘For a fortnight ignorance of what the enemy was up to was so great that, in the records of the Cabinet and the Chiefs of Staff, discussions of the fighting continued to be headed “The Netherlands and Belgium”.’ By the time they found out, it was too late to make any difference.

  But it was now that the first Bombes came into operation – probably a Turing prototype in May 1940 and then more with the diagonal board after August. Naturally, the machines ‘greatly increased the speed and regularity with which GC and CS broke the daily-changing Enigma keys.’ The Bombes were installed not at Bletchley, but at various out-stations such as Gayhurst Manor, in a remote corner of Buckinghamshire. They were tended by ladies of the Women’s Royal Naval Service, the Wrens, who without knowing what they were doing and without asking the reason why, loaded the rotors and telephoned the analysts to say when a machine had come to a stop. They were impressive and rather beautiful machines, making a noise like that of a thousand knitting needles as the relay switches clicked their way through the proliferating implications.

  Military officers attached to Bletchley were vividly impressed by the Bombes in operation. The secret service officer, F.W. Winterbotham, would refer to the Bombe as15 ‘like some Eastern Goddess who was destined to become the oracle of Bletchley’, and at the OIC, too, they spoke of ‘the oracle’. It was a usage that would have amused Alan, for he too had conceived of an oracle that would produce answers to unsolvable problems. What they began to discover, however, was that interpretation of the utterances was itself a major enterprise. If cipher machines had brought military communications into the Edwardian age, then the effect of the Bombes was to jolt military intelligence into the mass production era.

  In the First World War, Room 40 had worked hidden away in the Admiralty, its productions never coordinated with the results of sightings and interrogation. Only in the autumn of 1917, when the U-boat offensive was at its height, had the officer responsible for tracking them been allowed access to its information. The left hand did not know what the right hand was doing. And although the Navy’s cryptanalytic work had been16 ‘incomparably better than that of any other power, or of the British War Office’, Room 40 had operated in such a way that ‘there were no records, no cross-indexing, and what was not of immediate operational interest went into the waste paper basket.’

  It was not until the fall of France, when the war ceased to be a re-run of 1915, that the Room 40 flavour began to give way. The Poles, Welchman, and Alan Turing had put a Bombe under the British establishment, and nothing could ever be the same again.17 ‘Based on a machine and broken on a machine, the Enigma’s cyphered mesages were mechanically converted direct into plain language; so that it yielded up its end-product in cornucopian abundance once the daily setting had been solved.’ Here was the opportunity to capture not just messages, but the whole enemy communication system. Indeed, it was essential to do so, for the ‘cornucopian abundance’ required a second level of code-breaking in order to interpret it:18

  Apart from their sheer bulk, the texts teemed with obscurities – abbreviations for units and equipment, map a
nd grid references, geographical and personal code names, pro-formas, Service jargon and other arcane references. One example is furnished by the fact that the Germans made frequent use of map references based on the CSGS 1:50,000 map of France. This series had been withdrawn from use in the British Army. Unable to obtain a copy of it, GC and CS was obliged to reconstruct it from the German references to it.

  The Hut 3 filing systems, therefore, had to mirror the German system as a whole, in order to give meaning to the cipher traffic as a whole. Only when this had been done could the Enigma decrypts yield their real value – not so much in juicy secret messages, but in giving a general knowledge of the enemy mind. Without them, Europe was an almost complete blank, out of which anything could emerge. With them, they had some insight into what was possible.

  No precedent existed for a ‘cornucopian abundance’, and no means existed for making use of it. In 1940 the immediate problem was that of convincing anyone of the information thereby gained, without explaining its origin. At first it was passed off as the effects of spying. The result was that no military commander could take it seriously, since the offerings of the secret service were regarded as ‘80% inaccurate’. They had only just begun to consider more satisfactory arrangements for using the Luftwaffe decrypts in France, when events made the oracle irrelevant.