The Hut 6 Story by Gordon Welchman Gordon Welchman M&M BALDWIN Cleobury Mortimer, Shropshire The publishing history of this book is summarised on page 252. The cover has been designed by John Tuck. Foreword Alan Stripp, 1997 Prologue and Parts 1, 2 & 3 The Estate of Gordon Welchman 1982, 1997 Part 4 Frank Cass & Co Ltd Other sections M & M Baldwin, 1997, 1998 ISBN 0 947712 34 8 First impression 1997; second impression with minor corrections 1998 All rights reserved. No part of this publication may be reproduced or transmitted in any form, or by any means, electronic, optical or mechanical, including photocopy, recording or any information storage system, without the prior written permission of the copyright owner. Published by M & M Baldwin 24 High Street, Cleobury Mortimer Kidderminster DY14 8BY Printed by MFP Design & Print Longford Trading Estate, Thomas Street, Stretford Manchester M32 OJT IV To Alastair Denniston, Marian Rejewski, Dilly Knox, and Edward Travis. To the many men and women whose varied talents cind total dedication contributed to the success of Hut 6 in World War // ^nj to those among us who, seriously concerned with our national security will see to it that the story of this success, no longer hidden, ^ fu/h utilized in our anticipation of future dangers. Contents FOREWORD by Alan Stripp xi PROLOGUE I PART ONE Backdrop to Hut Six 1 Britain Went to War 7 2 Germany Prepared for War 18 PART TWO The First Year 3 Cottage and School: September and October 1939 31 4 Ideas and Plans: October and November 1939 59 5 Early Days: December 1939 to May 1940 84 6 Silly Days: May to September 1940 97 VII PART THREE The Rest of the War 7 Hut 6 Faces New Problems: 1940/41 - 1942 1943/44/45 119 8 Bombes and Wrens 138 9 From Interception to Intelligence 149 10 A Comedy of Errors 163 11 My Tasks as A.D. (Mech): 1943 to 1945 170 12 The Bletchley Park Environment 185 PART FOUR An Addendum 13 From Polish Bomba to British Bombe: The Birth of Ultra (first published in 1986) 195 APPENDICES I The Bombe with a Diagonal Board 235 II Biographical Note 250 III Publishers' Note 252 ACKNOWLEDGEMENTS 253 INDEX 257 VIII Illustrations General Guderian's Command Vehicle 27 Bletchley Park 33 Composition of a Typical Enigma Message 36 The Scherbius Enigma, copied from the cover of the original sales pamphlet (The National Archives, Washington) 41 Enigma with Steckerboard Uncovered 44 Letter Substitutions in Six Successive Positions of the Enigma 48 Electrical Connections Between Lampboard, Keyboard, Steckerboard and Scrambler 50 Sheet K (1,4) for Wheel Order 413 66 Portions of Five Sheets 68 Relative Positions of Matrices to be Stacked Over an Illuminated Window Diagrams Dervied From a Crib 79 A Herivel Square 100 Sillies and Chains 105 Inspection and First-Stage Testing of the Output of a Silly Menu 707 A Bombe Menu 743 Diagrams of Scrambler Units 238 Application of Turing Bombe 240 Bombe with Diagonal Board 242-3 Keen's Design of the Bombe 248 IX Foreword It is a great privilege to be invited to commend Gordon Welchman's important book, and a pleasure to congratulate Mark Baldwin on publishing this new edition. It was one of the first to give a clear first-hand account of the remarkable success of Bletchley Park (later to become GCHQ), the wartime centre of British codebreakmg. It lucidly describes both the technical problems which we faced and the ways we found to overcome them. The atmosphere of'BP', with its wartime sense of urgency, its small teams of men and women working all hours to break the ever-changing cipher-settings before the contents of the signals were too stale to be useful; its friendships and tensions--all these are vividly evoked by a man who worked there for the whole six years of the war. So is BP's unorthodox system of recruiting men and women of almost any age and background provided that they had keen, varied and imaginative approaches to complicated problems, and drawing them largely from the universities rather than the traditional source, the armed services. The breaking of codes and ciphers is often described as if it were a single act. Far from it. The process goes on round the clock, as Welchman, one of the leading code breakers vividly relates. By 1941 we were breaking some 4,000 German signals a day at every level from Hitler downwards, mostly within a few hours quite a post bag Like most of us, Welchman had to start XI xii / THE HUT SIX STORY from scratch. He was an apt pupil, soon emerging with bright ideas that worked. He played an important part in improving the techniques by which Hut 6 broke the German army and air force signals and passed the de crypts to Hut 3 for translation and processing, after which they were sent to our Commands at home or overseas Huts 8 and 4 dealt similarly with naval signals. It should be remembered that the Enigma machine was commercially available until the German armed services began adopting it in 1926. By then other countries, including Britain, had bought several, either for their own use or as insurance in case others used it. But two common misunderstandings need correction. The first is the assumption that an Enigma machine endows its owners with magical powers for breaking Enigma signals. On the contrary: all the machine can do is to define the difficulty of breaking it. That task remains to be done--and the total number of possible daily keys is almost 160 million million million. The second is to speak of 'breaking' as if it were a single act for all time, and not one needing to be done afresh once or even twice every day. An earlier book, (Top Secret Ultra) by Peter Calvocoressi of Hut 3, had given a good picture of the Enigma machine, thus correcting Frederick Winter both am unrecognisable caricature (in The Ultra Secret), but it was left to Welchman to give the first complete and accurate description of the significance of its intricacies, and of the extraordinary methods needed to break its signals. Gordon Welchman was an outstanding member of a very talented team which also included Stuart Mimer-Barry, Dennis Babbage and many others. Not only was he a first-class code breaker ('cryptanalyst' in American parlance), but he also saw that BP's excellent work might soon be limited in quality and quantity by their cosy, creaking arrangements, and that the advent of The Real War' in spring 1940 called for big improvements if vital intelligence was not to be held up. That meant more staff, more machinery--especially more FOREWORD xiii 'bombes'--and the rapid but thorough processing of de crypts through all their stages Mimer-Barry's recent tribute sums it up: "If Gordon Welchman had not been there, I doubt if Ultra would have played the part it undoubtedly did in shortening the war." A surprising number of Enigma-breakers had a common background in chess, or Cambridge, or both. It was a bright Cambridge quartet (Hugh Alexander, Stuart Mimer-Barry, Alan Turing and Welchman) who sent the celebrated letter to Churchill, by-passing the Head of BP, Alastair Denmston, who by 1941 was grossly overworked, a letter demanding specific priorities in staff and equipment if work was not to be held up. Churchill's famous "Action this day' note was the sequel. Imagine trying that in the armed services, then or now! Parts of the story of our Enigma-breaking were kept dark surprisingly long after the war, and not only when Britain's security demanded it. For example, the wonderful achievements of the Polish cryptographers were known to few people in Britain until the 1980s, and are still hardly appreciated. Welchman, in the early 1980s, was worse off still. On page 13 he repeats, with apparent approval, William Stevenson's cock and-bull story (from A Man Called Intrepid) about Polish agents ambushing a German truck to steal an Enigma machine, although this is displaced in the following four pages by a substantially accurate account. The Poles had to overcome British complacency and scepticism before they could persuade us to send a small team there, as late as six weeks before Germany invaded them--an act of aggression which, with cruel irony, their signals-breaking had clearly predicted. The British team, and above all the brilliant but quirky "Dilly' Knox, could hardly believe what they found. It is still not too late to salute Tadeusz Lisicki, Marian Rejewski, Jerzy Rozycki and Henryk Zygalski, and acknowledge the debt that BP owed them. Wladyslaw Kozaczuk's book Enigma, giving a clearer picture, first appeared in 1979, but only in Polish; the first English translation followed in 1984, two years after The Hut Six Story. xiv THE HUT SIX STORY Not only was that debt played down, if acknowledged at all. When two of that Polish team eventually arrived here in 1943, instead of being allowed to join BP's Enigma team, whom they had virtually baptised, they were assigned to a tiny unit at Boxmoor, near Hemel Hempstead. There they were relegated to working on "Double-Cassette' traffic--a scandalous example of using race-horses to pull carts. I must also comment on the erratic and ill-informed nature of much British official censorship, not merely during the war but up to our own times. Long before The Hut Six Story appeared in 1982, the Enigma secrets were already known to the professional cryptanalysts of most advanced countries, whether friend or foe, and had become irrelevant because of the immense technical advances in electronics. Nevertheless, Welchman's book had to be published first in the USA, where he had been working since 1948. So this book, with its speculation about future developments in cryptology, was a very tempting target. Even the comparatively low-level subjects mentioned in a book of mine in 1989 were adjudged too risky to be let out--though let out they were. Welchman's distinguished colleague Sir Stuart MilnerBarry pointed out in 1986 that 'the methods of 1939-45 must seem to present-day cryptanalysts rather like fighting with bows and arrows'. The electronic computer, in the early development of which BP's own "Colossus' had played an important part--though not for Enigma-breaking--had already transformed the whole world of secret messages. It is therefore good to have this book, with its clarity, insight, compassion, sense of proportion, humour and expertise, once again available. Alan Stripp Cambridge August 1997 Prologue On a Sunday in the summer of 1974 my wife, Teeny, and I were staying in England with her godmother, whose son-in-law was reading the Sunday Telegraph. Finding an interesting article, he showed it to me. It was the second part of "Deepest Secret of the War" by F. W. Winterbotham, and it was a discussion of secret activities of World War II in which I myself had taken part. Fortunately the previous week's Sunday Telegraph had not been thrown away, and I was able to read the whole article there and then. It was a preview of Winterbotham's forthcoming book The Ultra Secret, which has since become familiar to readers throughout the world. For the first time it became public knowledge that two organizations of the British wartime Government Communications Headquarters (GCHQ) at Bletchley Park, known as Hut 6 and Hut 8, had succeeded in breaking the Enigma machine cipher that was used extensively by the Germans. Furthermore, Winterbotham had revealed that intelligence derived from these two sources, known by the code name Ultra, had been of immense value to the Allies. I was tremendously excited. I had been the man primarily responsible, during the winter of 1939-1940, for building up the Hut 6 organization to a point at which it was ready for action when Hitler struck against Norway and France in April and May 1940. But in the years since World War II I had never felt at Ijberty even to hint at my own part in such activities. During the war it had sometimes been extremely difficult to distinguish between items of information that I had learned from Enigma decodes and those I had picked up from newspapers or radio broadcasts. The possibility that one might give away the secret in an unguarded moment was a continual nagging worry, so I had tried to avoid any discussion of the war except with people who were in the know. After the war I still avoided discussions of wartime events for fear that I might reveal information obtained from Ultra rather than from some published account. Indeed, I was careful in much of my subsequent work to avoid suggesting that many of my ideas were derived from wartime experience. Now, suddenly, what was indeed one of the best-kept secrets of the war had emerged into the light of day. I felt that this turn of events released me from my wartime pledge of secrecy. I could at last talk to my friends, relatives, and colleagues about the activities of one of these two organizations, Hut 6, with which I was closely associated from the outset. I could even write my own account of what actually happened. I began to think of a book. During the twenty-nine years from summer 1945 to summer 1974 I had thought very little about World War II events related to Hut 6, until I joined the MITRE Corporation* in 1962 and found myself involved in studying the information flow needed for coordination of ground and air forces in battle. In researching the optimum design of a communications system to meet the information needs of the future battlefield, I discovered that what I had learned from my work on the German system of World War II was still valid. Technology had changed tremendously, but the principles of handling the coded traffic were much the same. Whereas years earlier I had been involved in breaking down the security of the German communications system, now my task was to protect the security, and also the survivability, of our new battlefield communications systems. The combination of these two experiences has strengthened my desire to discuss what we can learn * The MITRE Corporation of Bedford, Massachusetts, and McLean, Virginia, is a nonprofit, non manufacturing Federal Contract Research Center, providing services in systems analysis, systems engineering, and technical direction for U.S. Government Agencies and other public-interest organizations. from the past that will help us to achieve national security in the future. Thus, in telling the story of Hut 6, I have emphasized those aspects of clandestine activities that are still relevant. The foremost of these is that the way in which Hut 6 was able to break down the security of a major part of the German command communications of World War II contains valuable warnings to the many people--not just cryptographers--on whom the security of our future battlefield communications will depend. Second is that the study of cryptographic security must be widened to include survivability: making sure that communications will be available in working order to handle essential information in times of emergency. Third, that protection against various forms of "special means' such as sabotage, theft, infiltration of agents, jamming, spoofing, and so on is just as important, if not more important, than protection against enemy cryptanalysis. Fourth, because of continually changing world politics we need to be prepared for different kinds of military conflict in different environments. Fifth, that we must not neglect the nonmilitary means by which an enemy may seek to undermine our national will. And finally, that the really frightening pace at which technology is advancing introduces many new problems that are not yet being satisfactorily addressed. In thinking about security measures for a military communications system in the 1980s and 1990s, the near certainty that major technological advances will have occurred in the interval is a real worry. Will the security measures we plan today be adequate in the technological environment of tomorrow? I am convinced that the Hut 6 story can contribute to this planning. The principle of the bombe, a completely new electromechanical device that we built to help us break the Enigma, should have been made known many years ago; it contains a clear warning, much needed today, that our confidence in cryptographic security should not depend solely on the number of possibilities an enemy cryptanalyst must examine to break a code, and on the assumption that he must examine each possibility individually. My study of how the Germans made it possible for us to break a cipher machine that would have been unbreakable if properly used proves that if we are to achieve security, we must pay careful attention to the indoctrination, training, and monitoring of all personnel involved. The Hut 6 story also provides a striking example of the importance of close cooperation between different specialized activities in military action that requires coordinated operations. (We must consider an interdisciplinary approach to the planning of military capabilities and to the conduct of military operations.) We can also see that the German tactical communications system, apart from unknowingly allowing its Enigma to be broken, was an exceptionally effective tool of war. At present, however, our own tactical communications can only be regarded as a glaring weakness in our military defenses. Finally, an analysis of what was effective or ineffective in the exploitation of Hut 6 Ultra intelligence can hardly fail to be of value in planning how to handle intelligence on future battlefields. PART Ml: Backdrop to Hut Six Britain Went to War The Germans invaded Poland on September 1, 1939; Great Britain and France declared war on September 3. Most of us in England were pretty scared. We had attended first-aid classes in anticipation of bombing. The media had told us to expect immediate air attacks if war should come. We had also been told that about the safest place in the house was under the stairs. My first wife, Katharine, and I lived on the outskirts of the largely academic town of Cambridge where I was teaching mathematics at the university. Cambridge could hardly be regarded as a prime military target by anyone who had thought about war. Not having thought about war, however, we followed official advice. On that first night, when the air-raid sirens were sounded in Cambridge--a false alarm, incidentally--we crouched under the staircase with our one-and-ahalfyear-old son, Nick. I must confess that Katharine saw the absurdity of the situation before I did. This incident shows how poorly the British public was prepared for war. The country had been under the appeasement-minded government of Neville Chamberlain, who is remembered for the umbrella that he carried and for the monstrous "Munich Agreement" with Hitler of September 1938, which betrayed Czechoslovakia. Already, in March 1938, Austria had been invaded and annexed, yet most of us in England were far less concerned with the sinister trend of events than we should have been. We even welcomed the Munich Agreement, thinking mistakenly that the danger of war had been averted. After Munich, however, there were signs that the British government was taking the possibility of war seriously. We became increasingly alarmed, but were still abysmally ignorant of what to expect. The general impression was that, from the outset of war, every town and village in Great Britain could expect to be bombed, and that after a feu days there would be no more R.A.F pilots and the country would be defenseless. As I remember, I did once ask a guest, a senior Royal Air Force officer, a question to this effect: "If the Germans were to load all their aircraft with all the bombs they could carry, how large an area could they bomb in one raid?" But I could obtain no answer whatever. Our guest had evidently never considered the question of how much damage the Germans could actually cause. As things turned out, even after the heaviest German bombing attacks, one could still drive around the English countryside and see few signs of damage. The area around Bletchley Park, where so much vital work was done on the breaking of the enemy's codes and ciphers, was never attacked, even though it also contained an important railway junction on one of the main lines to the north. In fact, the threat of country wide German bombing had been grossly exaggerated. For several years before the war, mercifully, many Britons who believed that Hitler could not and should not be appeased had been making preparations. One of these was the naval commander, Ala stair Denniston, head of the Government Code and Cypher Sch(x>l, who had been one of the original cryptographers in Room 40 at the British Admiralty during World War I, first under Sir Alfred Fwing and then under the Director of Naval Intelligence, Captain William Reginald Hall, later Admiral Sir Reginald ("Blinker") Hall--whose nickname referred to a twitch that made one eye flash like a Navy signal lamp. Between the wars, Denniston, seeing the strategic value of what he and his group were doing, had kept the Room 40 activity alive. In 1920 the name of the organization was changed to "Government Code and Cypher School" (GCCS), and it was transferred from the Navy to the Foreign Office. Alastair Denniston's son, Robin, was about twelve years old when World War II broke out. Today Robin is a leading member of the British book publishing establishment. Back in 1939, of course, he had no knowledge at all of what his father was doing, but he still has many significant memories. Before the war, much of Alastair Denniston's social life consisted of sherry parties, golf matches, and other occasions attended by his colleagues of GCCS. In retrospect they seem to Robin to have been a very close, fiercely loyal, defensive, almost cozy group, perhaps under attack from other departments. He went to his father's office very seldom, but knew several of the group quite well, including Josh Cooper, Xigel de Grey, Dilly Knox, and John Tiltman; also Frank Birch, who had returned to Cambridge but kept in touch. All of these will appear in my story. Robin remembers many occasions when his father's colleagues gathered at his home. He feels that the interest they showed in his sister and himself reflected an admiration for their father. Alastair Denniston planned for the coming war. Following the example of his World War I chiefs, he decided that the Universities of Oxford and Cambridge would be his first source of recruits. So he visited them and arranged for polite notes to be sent to many lecturers, including myself, asking whether we would be willing to serve our country in the event of war. He chose Bletchley, 47 miles from London, as a wartime home for GCCS because it was a railway junction on the main line to the north from London's Euston Station and lay about halfway between Oxford and Cambridge with good train service to both. He acquired Bletchley Park, a large Victorian Tudor-Gothic mansion with ample grounds. The house had been renovated by a prosperous merchant, who had introduced what Dilly Knox's niece, Penelope Fitzgerald, in her family memoir The Knox Brothers, was to call "majestic plumbing." In the planning for the wartime Bletchley Park, Denniston's principal helpers, in Robin's recollection, were Josh Cooper, Xigel de Grey, John Tiltman, Admiral Sinclair's sister, and Sir Stuart Menzies. Edward Travis, whose later performance as successor to Alastair Denniston at Bletchley Park was so significant, must have been involved, but he was not one of the "family" team that dated back so many years. Robin believes that, before joining GCCS, Travis was involved in encipherment rather than code breaking As preparatory work was being done, Denniston visited the site frequently, and made plans for construction of the numerous huts that would be needed in the anticipated wartime expansion of GCCS activities. When war actually came, these wooden huts were constructed with amazing speed by a local building contractor, Captain I lubert Faulkner, who was also a keen horseman and would often appear on site in riding clothes. The word "hut" has many meanings, so I had better explain that the Bletchley huts were single-story wooden structures of various shapes and sizes. Hut 6 was about 30 feet wide and 60 feet long. The inside walls and partitions were of plaster board. From a door at one end a central passage, with three small rooms on either side, led to two large rooms at the far end. There were no toilets; staff had to go to another building. The furniture consisted mostly of wooden trestle tables and wooden folding chairs, and the partitions were moved around in response to changing needs. The final move of the GCCS organization to Bletchley was made in August 1939, only a few weeks before war was declared. As security cover the expedition, involving perhaps fifty people, was officially termed "Captain Ridley's Hunting Party," Captain Ridley being the man in charge of general administration. The name of the organization was changed from GCCS to "Government Communications Headquarters" or GCHQ. The perimeter of the Bletchley Park grounds was wired, and guarded by the R.A.F regiment, whose NCOs warned the men that if they didn't look lively they would be sent "inside the Park," suggesting that it was now a kind of lunatic asylum. Denniston was to remain in command until around June 1940, when hospitalization for a stone in his bladder forced him to undertake less exacting duties. After his recovery he returned to Bletchley for a time before moving to London in 1941 to work on diplomatic traffic. Travis, who had been head of the Naval Section of GCCS and second in command to Denniston, took his place and ran Bletchley Park for the rest of the war. In recognition of his achievements he became Sir Edward Travis in 1942. In spite of his hospitalization, Denniston, on his own initiative, flew to America in 1941, made contact with leaders of the crypto logical organizations, and laid the foundations for later cooperation. He established a close personal relationship with the great American cryptologist William Friedman, who visited him in England later. The air flights were dangerous. On Denniston's return journey a plane just ahead of his and one just behind were both shot down. Denniston's daughter was nicknamed "Y," because she was an unknown quantity before her birth. Years later Friedman wrote a letter to her: "Your father was a great man in whose debt all english-speaking people will remain for a very long time, if not forever. That so few should know exactly what he did ... is the sad part." Not very long after the Munich Agreement of September 1938 I had answered "ves" to Denniston's polite note and had been called to London a little later for one or two weeks of preliminary indoctrination. It was a very informal affair. With several other new recruits I was introduced to the established manual methods of cryptography and cryptanalysis, and also to the various types of cipher machines that were in use at that time. I did not see much of Denniston, but I remember that he had a very warm personality and a quiet but incisive manner of speaking. He was neat in appearance, and he must have had a somewhat distinctive gait because, though I could not possibly describe it, I recognized it in his son long afterward, when I met Robin. Denniston told me that, in the event of war, I was to report to him at Bletchley Park* in the town of Bletchley, Buckinghamshire. I remember very little else about the preliminary indoctrination in London, except that I was very impressed by Oliver Strachey, a senior member of the GCCS staff, who during the coming war would head an organization known as Intelligence Services Oliver Strachey (ISOS). He seemed to be giving us an overview of the whole problem of deriving intelligence from enemy communications, and this may well have been a strong guiding influence on my wartime work. Our other instructors dealt with specific examples of manual cryptographic methodology. Dilly Knox, the GCCS expert on the enigma machine, may have been there, but I do not remember meeting him on that occasion. I have the distinct impression that the chances of ever being able to do anything with the German Enigma traffic were at the time considered minimal, at least until access to a machine could be achieved. The enigma was one of a family of cipher machines that lo the thousands of people u ho worked there. Bletchlev Park w ill always be known as R P T ' r lo any of those people u ho may read this book I apologize for not always referring to the place as B.P." hut I feel that to the general reader the abbreviation will hardly have the same nostalgic appeal. produced a highly variable "scramble" of the twenty-six letters of the alphabet by passing electric current through a set of movable rotors, each of which, by its internal electrical connections, contributed to the overall scramble. To discover the internal wiring of the rotors by purely crypt analytical means appeared to be an insoluble problem. In the interim between the preliminary indoctrination in London and the actual outbreak of war, I knew nothing of the GCCS activities. I was back in Cambridge, where I had been a mathematics scholar at Trinity College from 1925 to 1928, and a college lecturer in mathematics at Sidney Sussex College since 1929. I was working on a book, Introduction to Algebraic Geometry, for the Cambridge University Press. I had been writing it for about five years and was within a few months of having the manuscript ready for the publishers, but the war came just too soon. Publication was delayed until after the war. The word "scholar" that I have just used, and will use again, needs explanation, because it has a very different meaning in America. In Cambridge, England, the scholars won their titles (major scholar, minor scholar, or exhibitioner) in a highly competitive examination before they entered the University. Being a scholar carried rights and responsibilities--for example, rooms in college as! opposed to lodgings outside, and reading the lessons at early morning services in chapel. If a scholar needed financial assistance, he was entitled to generous amounts of money toward his tuition and! other university expenses. Able boys, who did not need the money took the scholarship exams simply to win the academic standing that the title implied. | On the morning after Britain declared war on Germany I drove from Cambridge to Bletchley and reported to Denniston, who sent me along to join Dilly Knox's Enigma group of about ten people in a building known as the "Cottage." In that small building in the stableyard I found that an event of immense importance had occur red very recently. I was not told very much about it, but the Poles had given us one or perhaps two replicas of the Enigma machine that had been issued to the German army and air force. We now knew all the electrical and mechanical details, and we all knew the procedures that were then being used for encoding and decoding Furthermore the Poles had given us the full advantage of their brilliant work on Enigma. When I come to describe what happened at Bletchley Park it will become apparent that these gifts were of immense importance in getting us started on the road that led to Hut 6 Ultra. At the time I did not bother about how the Poles had been able to make replicas of the German Enigma, or about how the machine had reached us. It didn't matter to me then, and even now I do not know which of several stories to believe. A point of general agreement is that a meeting was held in July 1939 at a hideout of the Polish Secret Service in the Pyry forest. At this meeting Colonel Gwido Langer, head of Poland's Cipher Bureau, astonished a British delegation, headed by Alastair Denniston and a French delegation headed by Captain Gustave Berrrand, by revealing that his team knew "everything about Enigma," and furthermore that two replicas of the machine were to be handed over, one for Britain and one for France. This much I have learned from Ronald Lewin's Ultra Goes to War, and I have not found any contradiction in other writings, though I must admit that I have not tried very hard. But there are different stories about what led up to the meeting in the Pyry forest and what followed it. On the whole I like the story, told in A Man Called Intrepid, that in early 1939, when the new Enigmas were being delivered to frontier units, a German military truck containing one of them was ambushed. Polish agents removed the Enigma and put another machine in its place. They then made it appear that there had been an accident and that the truck had caught fire. Thus the Germans who examined the wreckage were led to believe that some charred bits of coils, springs, and rotors were the remains of their Enigma. For a reason that will appear immediately, I am inclined to believe that something of this sort must indeed have occurred. There is a divergence of view as to how an Enigma reached England after the Pyry forest meeting. Stevenson, in A Man Called Intrepid, asserts that when Alastair Denniston, then in his fifties, stayed at the Bristol Hotel in Warsaw in August 1939, a large leather bag containing an Enigma machine, or its replica, was left beside a pile of luggage in the hotel's foyer. Denniston, carrying an identical bag containing some dirty shirts and some weighty but valueless books, casually exchanged the bags and returned to England with his prize. Robin Denniston believes this story of his father's exploit to be true, but cannot provide conclusive evidence. Another story, supported by Ronald Lew in in Ultra Goes to War, is that at the Pyry forest meeting two replicas of the Enigma were given to the Frenchman, Bertrand, and that he himself brought one of them to London, where he was met at Victoria Station by Menzies. Robin Denniston believes that this story too is true. And why not? In those dangerous times the Poles may well have decided to send us two Enigma replicas by different routes. Anyway, what does it matter? The vital point is that the Poles managed to give us the details of the machine and the benefit of their expertise in time for us to exploit them early in the war. We would ultimately capture many Enigmas, but, as things turned out, everything was to depend on what we achieved in the first three months and later, in the very first year of the war. In fact, the Hut 6 balloon might never have gotten off the ground at all if I had taken six months instead of three to get things moving. Just as this book was entering the final stages of publication, I received reliable information about the Polish contribution. In February of 1981, Monsieur Jean Stengers, Professor of History at the j University of Brussels, sent me a copy of his article "La Guerre des j Messages Codes (1930-1945)," which had just appeared in the! magazine L'Histoire. The article contains a new account of Polish! work on the Enigma, which fits exactly with what I know of the! early days at Bletchley, and also with what I have guessed. I believe! this account to be far more trustworthy than the many others that! had previously appeared in print. 1 The story starts as early as 1930, when three mathematics graduates of the Polish University of Poznan, Marian Rejewski, Jerzyi Rozycki, and Henryk Zygalski, were recruited by the Polish ciphers bureau in Warsaw. In the autumn of 1932 Rejewski became aware of the Enigma problem. The Polish bureau had acquired a commercialy available Enigma and knew that the German military were using a somewhat similar cipher machine. Rejewski, working aloneB began an investigation that entitles him to be regarded as one of the greatest cryptanalysts of his day In a feu weeks Rejewski achieved a first breakthrough by est am * He died in Warsaw in February 1980. lishing a mathematical approach that, in theory, would solve his problem. But the time required to reach a solution made his method impractical. Then, on December 8, 1932, he was given what appeared to be authentic documents relating to the German military Enigma. Among them was a list of the keys used in the months of October and December 1931. These documents had been offered to the French intelligence bureau by a German named HansThilo Schmidt. Captain Bertrand, later General Bertrand, the bureau's specialist on foreign ciphers, handled the contacts with Schmidt, who became known by the code name Asche. Having obtained the Enigma documents, Bertrand presented them to the cryptanalysts of his bureau, who were not interested. He then turned to the leaders of the Polish cipher bureau, with whom he had already established an excellent rapport. They were enthusiastic, and before long the documents appeared on Rejewski's desk. Rejewski was now able to put his theory into practice After only a month of continuous and highly concentrated effort, he had worked out the electrical connections of the three wheels that were used at that time in the German Enigma. He was able to have a replica of the machine constructed. He began decoding German military messages in early 1933. He was then joined by Rozycki and Zygalski; these three men formed the team that, with the brilliant collaboration of Polish engineers, followed the changes in the German use of the Enigma during the following six years. This team conceived and built two machines: the "Cyclometre," which consisted of two interconnected Enigmas, and the "Bomba," due primarily to Rejewski, which was a combination of six Enigmas. They developed a system of feuilles perforees, or perforated sheets, due primarily to Zygalski, who died in England in 1978. (Rozycki died in an accident in 1942.) The reader of this book will recognize, when he comes to Chapter 4, that Zygalski's system of perforated sheets must have been the forerunner of the system developed in the Bletchley Cottage in the fall and winter of 1939/40. The testing machine used by Dilly Knox's team must have been based on the "Cyclometre." Rejeu ski's It seems that DjIJy Enow in England may uell have arrived at a comparable theory, hut he did not have access to the Asche documents. six-Enigma "Bomba" must have been the origin of the machine on which Turing was working in the fall of 1939. It must have used the idea of double-ended Enigma scramblers that I describe in the Appendix. Rejewski and his team reached their peak of success in the first half of 1938, when they were decoding German Enigma messages almost every day. But this peak of success was followed by a staggering setback. On December 15, 1938, the Germans introduced two new wheels, making five in all, any three of which, in any order, could be employed in the Enigma machine. This resulted in the sixty alternative wheel orders with which the Cottage and Hut 6 were to be faced. Even Jean Stenger's excellent article does not explain how Rejewski came to know the internal wiring of these two new wheels. That he did so is evident from what followed. But as an achievement of pure cryptanalysis it is hard to believe; it seems to me that the Poles must have obtained the new five-wheel Enigma by capture or some other nefarious means. Even then, however, the preparation and use of new perforated sheets would have demanded a far greater effort than the Polish team could provide. Finding them] selves in deep trouble, the Poles sought first to find out whether the! British or the French could help them. An international conference,! organized in January 1939 by Bertrand, showed them that they were well ahead of their friends. They did not reveal their owim success. I In July 1939, sensing the approach of war the Poles decided to make all their achievements known to their counterparts in France and Great Britain. A conference near Warsaw, on July 25, 1939 was attended by a small French group, including Bertrand, and small British group, including Dilly Knox. To the astonishment of their guests, the Poles explained and demonstrated all their remarkable achievements: their Enigma theory, the models of the German Enigma machines, their "Cyclometre," their "Bomba," their per forated sheets, and the decodes of German messages that they had produced when the Enigma machine had only three wheels. In August 1939, just in time, Polish models of the five-wheel German military Enigma were sent to France and England. Jean Stengers appears to have been in close touch both wit Rejewski and with Bertrand. He tells the story of what the Polls and the French were doing during the war. To my surprise, I now learn that an organization at Gretz in France was collaborating with Bletchley for a time. Later, in 1943 to 1945, Rejewski was in England, attached to the Polish army. He was still involved in crypt analysis, but was not permitted to know how Hut 6 had carried on from the start that he had given them. Indeed, he did not know of the end result of his work until he read Winterbotham's book in 1974. Professor Stengers expresses astonishment that, even now, he cannot complete his story with an account of how the British carried on. But he knows the reason for British reluctance to reveal the methods they developed. He has discovered that, after the war, Britain sold a considerable number of Enigma machines to certain countries, no doubt claiming that they were unbreakable. But, says Jean Stengers, British code breakers had no trouble. As I have made clear in this book, it is my firm conviction that what I have revealed should have been made available to our planners many years ago. The over-prolonged secrecy has, in my view, already been prejudicial to our future national security. The reason for the secrecy, made public at last, is hard to accept * I, personally, have heard conflicting statements on this matter, but, on balance, I believe that Jean Stengers is correct. Germany Prepared for War German President von Hindenberg died on August 2, 1934. That same afternoon Hitler proclaimed himself Fuhrer, and ordered the armed forces of Germany to swear an oath of unconditional obedience to him, stating that they would be ready, as brave soldiers, to risk their lives at any time. It was the old Fahneneid, or "flag oath," of the Teuton knights. The entire General Staff recited the oath. Soon afterward a plebiscite approved Hitler's assumption of the presidency and of sole executive power. Hitler, however, preferred to retain the title of "Der Fuhrer." In March 1935 he startled the world by denouncing the clauses of the Versailles Treaty providing for German disarmament. When in the years 1935 to 1940 Hitler was remaking Germany's military capabilities, he and his generals paid considerable attention to the writings of several Englishmen. In 1929 one of these Englishmen, Sir Basil Liddell Hart, had published a remarkable book, The Remaking of Modern Armies. Combining knowledge of the military technology of his day with a study of military history, particularly the history of cavalry tactics, Liddell Hart expounded a new theory of the proper use of tanks in battle that was put into practice by the Germans ten years later with stunning success. But Hitler and his generals improved on Liddell Hart's theory in two important areas: air support for ground forces, and battlefield communications. The original theory was that tank strength should not be dispersed by assigning a relatively small number of tanks to eaeh infantry division. Rather, tanks (Panzers in German) should be concentrated in panzer divisions and panzer armies capable of bursting through the enemy's defense line and wreaking havoc in his rear. The initial penetration of prepared defense lines might require well-coordinated cooperation with massed infantry and artillery, but if and when a breakthrough could be achieved, the panzer formations, accompanied by motorized infantry and self-propelled guns, should exploit the situation by operating on their own, well in advance of foot-mobile infantry and slower-moving artillery weapons. The Germans under Hitler remade their army. They developed a branch of their air force, the dive-bombing Stukas, for the specific purpose of support for their armored ground forces. And they remade their military signaling capabilities to permit effective coordination of all their fast-moving forces. By doing these three things they achieved one of the greatest revolutionary changes in military history. We will be primarily concerned with the developments in communications. The German planners had realized in good time that the successful conduct of their blitzkrieg would call for revolutionary radio communications capabilities. They developed such capabilities, and, to provide the main arteries of secure information flow, they trained a large, highly mobile signals organization equipped with the Enigma cipher machine. Indeed, if there was a single central idea underlying the concept of Hitler's blitzkrieg, it was "speed of attack through speed of communications." That this was so did not occur to me until I read William Stevenson's A Man Called Intrepid, which gives a brief account of Sir William Stephenson's prewar contacts with Hitler's Air Force chiefs At one conference Stephenson heard a description of "blitzkrieg." Dive bombers and tanks would spearhead each offensive, followed by troops in fast carriers. When he asked how their Air Force would get support so far ahead of the armies, General Erhard Milch, State Secretary for Air, replied: Note the confusion between Sir William Stephenson, who was called "Intrepid." and the book's author. William Stevenson. The dive bombers will form a flying artillery, directed to work in harmony with ground forces through good radio communications. You, a radio expert, must appreciate that for the first time in history, this coordination of forces is possible. The Air Force will not require ground support, any more than the armored divisions will need repair units. Tanks and planes will be disposable. The real secret is speed--speed of attack through speed of communications. At the start of World War II, as a result of this general principle, the Germans were overwhelmingly superior to the Allies in their radio communications for the speedy coordination of forces in battle. The Allies could not match the equipment illustrated in Figure 2.1, which shows the great German panzer general, Heinz Guderian, and his signals staff in a command vehicle. The general is standing with field glasses in hand. Next to him is a radio operator. And in the foreground, the man at the complicated-looking typewriter is operating an Enigma cipher machine. It was common for the German regiment, division, and corps commanders to be accompanied everywhere by specially equipped radio command and control vehicles of this sort. So equipped, the German panzer generals were always well informed about the current situation and had means available to control their units, request] air support, and keep in contact with their parent headquarters. An] entire signal battalion was deployed with each panzer division to] provide the necessary communications. Such, then, was the German revolution in battlefield communications. It was not a major revolution in technology. Rather, it was a revolutionary attitude to what existing communications and cryj; to graphic technology could contribute to the combined operatior of fast-moving ground forces and their air support. It was a matte of organization, training, and scale of effort. Above all it was matter of thinking out what problems had to be solved. Because the Germans had done such a good job, the problems with which we were faced were unprecedented. Never before had radio signals and cryptography been employed on such a large scale to provic battlefield communications. In thinking about the German preparations as they affected Hi 6, it will be worthwhile to look into some of the earlier developments both of signaling and of cryptography. One of the pioneers of modern military signaling was an American, Albert James Myer. The U.S. Army had no Signal Corps at all before 1854. In that year Myer, then a young Army doctor, went on frontier duty in Texas and became interested in the problems of military signaling in the field. He invented the "wig-wag" system, which became sufficiently established to be used by both sides in the Civil War. Myer became the originator and first commander of the U.S. Army Signal Corps. The invention of wig-wag stemmed directly from Myer's rediscovery of the work of a Greek historian, Polybius. Around 170 b.c. Polybius described methods of military signaling; in Chapter 46 of his tenth book of history he tells us: The last method that I shall mention was invented either by Cleoxenes or by Democritus, but perfected by myself. This method is precise, and capable of signifying everything that happens with the greatest accuracy. A very exact attention, however, is required in using it. I shall have occasion to echo the last sentence. The Polybius method was an alphabetic signaling system. First the letters of the alphabet were arranged in a square matrix or checkerboard, with five numbered row s and five numbered columns This provided positions for twenty-five letters, enough for; the Greek alphabet. For our twenty-six-letter alphabet we could I combine two letters, say I and ], and form a Polybius square thus: j 1 2 S 4 5 1 A B C D E 2 F G H I/J K 3 L M N 0 P 4 Q R S T U 5 V W X Y Z In this matrix any letter can be represented, rather as in a n reference, by a combination of two numerals--the numbers of the horizontal row and the vertical column in which the letter occurs. To transmit a message consisting of a series of letters, it was necessary to send a sequence of signals that would identify ordered pairings of the numerals 1 to ^. For example, the letter R is in row 4 and column 2 and is therefore represented by the combination 4,2; the letter P by 3,f. 1 he pairs of numerals are "ordered" in the sense that the pair 2,4 does not have the same meaning as the pair 4,2, though they consist of the same numerals. In the Polybius system a signaling station, probably on the top of a hill, would have two walls side by side, facing the remote station with which communication is to be established. Each wall was about 10 feet long and a little less than the height of a man. Along the top of each wall were five sockets in which torches could he placed. To signal a letter, say H, the operator referred to his Polvbius square and found that H was in row 2 and column 3. He set torches in two of the holes on the left-hand wall, and in three of the holes in the right-hand wall. The distant observer saw first the two torches, then a little to their left, the three torches. He referred to the second row and third column of his Polybius square, and saw that the letter signaled to him was H. And so on until the laborious process of message transmission was completed. This method triggered Myer's imagination and, with remarkably few additional inventions, led to the alphabetic signaling technology that was to be available to Hitlers planners, Myer retained the Polybius square, but discarded the walls and torches. In their place he introduced a lightweight signal staff with a maximum length of about 16 feet, which could be disassembled into 4-foot lengths for easy portability. During daylight a flag was attached to the top of the signal staff; at night its place could be taken by a torch. To transmit a message a signalman would stand facing the signal station that was to receive it. He would hold his signal staff in a 'neutral position, pointing straight up. Then, to transmit the letter H, represented by 2,3, he would simply "wig" his staff by dipping it twice to his left, and then "wag" it by dipping it three times to his right. I he receiving operator would count the wigs and the wags and so recover the letter H. In principle Myer's system was beautifully simple, and the necessary equipment was very easily portable by an individual signaler. However. as Polybius pointed out, a very exact attention was required in using it. Strict adherence to standard operating procedures was essential. For example, the transmitting signalman would have to make sure that the distant signalman was ready to receive; he would have to indicate, probably by dipping his staff forward, whenever he was about to transmit a letter; he might have to wait for an acknowledgment signal before transmitting another letter; and he would need to indicate when a message transmission had been completed. After the American Civil War, Myer's wig-wag signaling system was adopted by the British army, and by military establishments throughout Europe. However, one very important change was made. The Polybius square was replaced by the more versatile Morse code, which had been developed for the electric telegraph in 1843, and used combinations of "dots" and "dashes" to represent the twenty-six letters of our alphabet, the ten decimal digits 0 to 9, and a few other things. Any combination of dots and dashes could be transmitted by the wig-wag system, using a wig for a dot and a wag for a dash. The electric telegraph depended on the transmission of pulses of electrical current along a wire circuit connecting two terminals. A pulse was transmitted from one terminal by tapping a key connecting the circuit to a source of electric current. At the other terminal the pulse would produce an audible sound, whose duration would depend on the length of time the key had been held down. Thus the operation of the key, or "keying," could transmit any Morse character as a sequence of short pulses (dots) and long pulses (dashes, which were about three times as long as the dots). At the receiving terminal the characters would be recognized aurally. Operating speeds varied from a hundred to two hundred Morse characters a minute, depending on the experience of the operators. On long circuits relay stations were needed to amplify signals before they became too weak. The development of radio, or wireless, introduced a new feature. The signals from an electric telegraph station could only be picked! up at places connected to that station by wire circuit. A radio! transmitter, however, could be omnidirectional; the reception of its! signals depended solely on range and signal strength. This feature of omnidirectional transmission led to the simplex| radio net, which was to be used extensively in Hitler's command! communications system to handle Enigma messages in Morse code. The term "net" implies that several subscribing stations would be communicating with each other. The term "simplex" means that all these stations would use the same radio frequency both for transmitting and for receiving. Thus a transmission from any one station of the net could be picked up by all the other stations. The Polybius dictum on the need for "a very exact attention" remained valid for radio nets. Standard operating procedures had to be developed and rigidly observed. Strict control was needed to ensure, in particular, that no two stations of the net would transmit at the same time. But this control could be exercised as necessary by any station of the net, and could be passed from one station to another. Furthermore, military units using the net could be mobile, and could join or leave the net at any time. All these features would be vitally important for Hitler's far-ranging blitzkrieg. I will have to discuss the operating procedures of the German radio nets in some detail, for they were at the heart of the Hut 6 story. Of what went before I need only say that there were very few revolutionary changes in technology between the "perfected signaling method" of Polybius and the German radio nets that were the concern of Hut 6. Torches on walls to Myer's signal staff to radio transmission. Polybius square to Morse code. That was all! Indeed, Myer's wig-wag system was still in use in World War II, as a means of training our intercept operators in reading Morse code signals. One of the pioneers of modern military cryptography was Au guste Kerckhoffs, who was born at Nuth, Holland, but spent most of his life in France. In 1883 he published La Cryptographic Militaire, which is still considered one of the fundamental books on the subject. In laying down basic principles for selecting usable field ciphers, Kerckhoffs sought to solve the problems that were being introduced by new conditions. David Kahn quotes him as saying "it is necessary to distinguish carefully between a system of encipherment envisioned for a momentary exchange of letters between several isolated people and a method of cryptography intended to govern the correspondence between different army chiefs for an unlimited time." In laying down fundamental principles, Kerckhoffs was the first to make the basic distinction between the general system and the specific "key." He foresaw that if a general system were to be used by too many individuals it would inevitably be compromised, and that secrecy should reside solely in the particular key, which could be changed at will. Thus Kerckhoffs introduced, nearly a century ago, what nowadays is sometimes called the fundamental assumption of military cryptography, that the enemy knows the general system. As to what Kerckhoffs meant by the general system and the key, he was thinking of the manual methods of his day, which I do not intend to discuss. But his meaning will become very clear when we come to the principles of the Enigma machine. The performance of this could be varied in several ways in accordance with instructions that constituted the key. Ability to decode an encoded message depended on knowing this key. Possession of the machine would not be sufficient. Kerckhoffs' teachings are highly relevant to the cryptographic aspects of our story. We and the Germans were dealing with a great volume of radio correspondence taking place over an extended period of time. The Germans did put their trust in keys rather than in the Enigma machine, but they failed to recognize the danger of use by too large a number of individuals. During World War II we were somewhat careless over our terminology, regarding code breaking as part of cryptography. As early as 1920, however, a distinguished American exponent of the art, William Frederick Kriedman, had introduced the new term "cryptanalysis" to mean the methods of breaking codes and ciphers. He used "cryptography" to mean the methods by which a message is rendered unintelligible to outsiders by various transformations of the plain text employing "codes" and ciphers. "Cryptology" was to be the name of the science that embraces both cryptography and cryptanalysis. In my narrative I propose to use Friedman's terminology, since it is widely accepted today. It should be noted, however, that Friedman allowed the term code breaking to include the breaking of ciphers, and that I will talk about "decoding" Enigma messages, even though the Enigma was a cipher machine. There is no clearly defined distinction between codes and ciphers, but for our purposes we can regard a "cipher" as a method of transforming a text in order to conceal its meaning. A "code," on the other hand, is a system in which groups of symbols are used to] represent a variety of things, such as letters, numerals, words, complete phrases, and so on. In the United States there are "zip codes" in which a group of five decimal digits represents a postal delivery zone, and we have already talked of the Morse code for signaling, which uses groups of dots and dashes. In commercial communications many codes have been developed in which thousands of groups of letters or decimal digits are used to represent anything from an individual letter to a commonly used phrase. In sending a cablegram, it is possible to save a good deal of transmission time, and cost, by using a single code group to represent a frequently used form of birthday salutation. Thus, unlike ciphers, codes are not always associated with secrecy. Hut it the "code hook that sets out the meanings of the code groups is kept secret, the content of a message may be concealed. Such were the codes with which the British Code and Cipher School was dealing, as were similar organizations all over the world. Breaking a code implied discovering the meanings of the individual code groups. Moreover, a message that had been encoded by means of a code hook could then he enciphered by a cipher system, a process known as "superencipherment." David kahn, writing in 1967, said that the First World War marked the great turning point in the history of cryptology, the direct cause being the vast increase in radio communications. Commanders soon found that radio was a speedy, convenient means of communication during military operations. The volume of messages soared, interception capabilities grew, and cryptanalysis flourished. Cipher key after cipher key, code after code, were betrayed by needless mistakes, stupidities, or outright rule violations. in his was because so large a volume of messages had to be handled by so many untrained men. In Kahn's view the great practical lesson of world War I cryptology was the necessity of enforcing an iron discipline on the cryptographic personnel. This lesson had been forgotten when World War II came. It could be forgotten again. During the first World War, because the volume of communications traffic had increased so enormously, all the manual enciphering methods that offered a high degree of security had proved unacceptably cumbersome. After the war, therefore, inventors turned their attention to the design of cipher machines. A German, Arthur Scherbius, produced the original version of the Enigma machine and tried unsuccessfully during the 1920s to promote it as a commercial product. (The commercial version was the one originally acquired by the Poles.) In the 1930s the German high command became interested. They decided that the Scherbius machine was well suited to the type of communications system needed for the coordination of their forces in their blitzkrieg, so they designed their own portable, battery-operated version and put it into large scale production. PART ONE The First Year Cottage and School September and October 1939 When I turned up at Bletchley Park in September 1939 I knew nothing about all this historical "backdrop" of communications, cryptology, and German preparations for war. I was absolutely green, and I simply tried to learn what I needed to know as quickly as possible. I was interested in finding out what the problem was, not in how it had arisen. The Victorian mansion of Bletchley Park incorporated samples of a wide variety of architectural styles, as will be evident from the photograph, Figure 3.1. The stableyard was behind the mansion and a little to the right. Further to the rear was a small building known as Elmers School, or simply "the School." When I reported for duty on September 4, 1939, I found that the bow window of Alastair Denniston's office on the ground floor looked out across a wide lawn to a pond, with attractively landscaped banks. There were rose beds near the mansion and near the pond. Before long the landscaping would be impaired, as rose beds and lawn were sacrificed to the much-needed wooden huts. Denniston, who was extremely busy, received me cordially and sent me to join Dilly Knox, the man in charge of work on the German Enigma cipher machine. Knox and his staff occupied a building in the stableyard, the former residence, I suppose, of the coachman. It had already become known as the Cottage. Alan Turing, from Kings College, Cambridge, was already there, and had B_ if "V=j| been working with Dilly for some time. John Jeffreys, whom I knew well, arrived from Downing College, Cambridge. Jeffreys, with Knox and Turing, all now dead, were three key characters in what was to happen. I was to see very little of Dilly. Our paths would diverge within two weeks or so and would not come together again except on one brief occasion. What I know of his achievements I did not learn until 1979, when I read Ronald Lewin's Ultra Goes to War and The Knox Brothers by Dilly's niece, Penelope Fitzgerald. Robin Denniston tells me that his father loved Dilly and admired his work. Before discussing my own exploits, I will give a brief account of him, because it sheds light both on his achievements and on the friction between him and myself, which until recently has puzzled me. Dilly Knox went up to Kings College, Cambridge, as an undergraduate in 1903, three years before I was born. By 1910 he was a fellow, studying Greek literature. When the First World War broke out, since he owned a much-loved motor bicycle, he tried to enlist as a military dispatch rider. But he was called to London instead as part of the expansion of the original crypt analytic effort in I.D. 25, a department of naval intelligence that became better known as Room 40. Dilly did outstanding work in I.D. 25, particularly in a solo performance on the "Flag Code" used by the Commander in Chief of the German Navy. He married in July 1920, at a time when the Room 40 of World War I was in the process of being transferred to the Foreign Office as the Government Code and Cypher School, or GCCS. Then, in 1931, he had the motor bicycle accident that had been long expected by those who knew him well. His leg was badly broken, and thereafter he always walked with a limp. He wore horn-rimmed spectacles, and was helpless without them. His health was not good. He had persistent stomach trouble. By 1938 cancer was suspected, and he had a preliminary operation. Earlier he had been tempted to return to his research in Kings College, but in 1936 GCCS was faced with the Enigma problem, and Dilly chose to stay on to tackle it. At the outbreak of war his old stomach trouble was bothering him, and he never felt well. He slept in the office, returning to his home only once a week. After I lost touch with him, according to Penelope Fitzgerald, he did brilliant and important work on the ciphers of the Italian fleet and the Abwehr (German spy organization), both of which used variations of the Enigma differing from the machine with which Hut 6 was dealing. After hospitalization in 1942 he was unable to return to Bletchley Park, but continued to work at home on a difficulty in the Italian cipher until his death on February 27, 1943. Dilly was neither an organization man nora technical man. He was, essentially, an idea-struck man. He was not interested, as I was, in the administration and automatic routine needed to handle the enormous volume of Enigma traffic generated by the German army and air force. Also, apart from a few lifelong friends, by and large Dilly seems to have disliked most of the men with whom he came in contact. Certainly during my first week or two at Bletchley I got the impression that he didn't like me. I don't remember what I learned in the Cottage, but after a week or so he gave me some sort of test and appeared to be, if anything, annoyed that I passed. Basically, I suppose, my problems were to be far simpler than those that interested Dilly. What I needed to know about Enigma could have been explained to me in less than an hour, and will, I believe, be easily understood by every reader. Anyway, very soon after my arrival I was turned out of the Cottage and sent to Elmers School, where I was to study call signs and discriminants," groups] of letters and figures which were a regular feature in the preambles! of the German Enigma messages. Alex Kendrick, a civilian memberf of Dilly's prewar staff, was sent along to get me started. He was! fair-haired, walked with a stick as a result of a paralyzed leg, and! was noted for the holes burnt in his trousers by a cigarette, or possibly by ash from a pipe. We occupied a fairly large room with bare! walls and no view from the windows; its only furniture was a long table and a few wooden chairs. Nobody else was working in the School, so Kendrick and I felt a bit lonely. In those early days the whole GCHQ staff met at lunch in the dining room of the mansion. Also any trace of loneliness was relieved for me every evening, for I had been assigned a billet, with two other new arrivals, at an excellent pub, the Duncombe Arms! in the hills to the south of Bletchley. Our host made us extremely comfortable, and his billiards and darts room became a social club for Bletchley Park personnel who were billeted nearby. The company was very enjoyable. Dilly's great friend Frank Birch uas the central figure. James Passant, a close friend of mine and the history fellow of Sidney Sussex College, often joined us. So did Dennis Babbage, another close Cambridge friend from Magdalene College, who would join Hut 6 later on. I remember his skill on the billiard table, particularly with strokes played behind his back; before the war he hat! repeatedly defeated me at squash and tennis. When we started work in the School, Kendrick and I were given two collections of Enigma traffic to analyze. And, as things turned out, my banishment from the Cottage to the School proved to be the real start of the Hut 6 organization that was to develop. At that time almost all the British-intercepted Enigma messages that I was to study were being plucked out of the ether by the experienced operators at an Army station on a hill at Chatham, on the Thames estuary. Hidden behind the ramparts of an old fort, this highly efficient and highly secret organization had been producing a little-used output for some time. Hach day's accumulation of messages had been regularly bundled up and sent to Bletchley Park together with a report on the day's traffic. All these bundles and daily reports were made available to Kendrick and me. Colonel 1 iltman, in charge of Army operations at Bletchley Park, instructed his sergeant in charge of records to give me all possible assistance. The sergeant did so; in fact, it was probably he who first told me something about the function of the call signs that I was to study, and about how the Germans operated their radio nets. He also gave me a large collection of intercepts, traffic reports, and other material that had been received from the French but had not yet been studied. 1 he composition of a typical intercepted Enigma message of those days is indicated in Figure 3.2. There was an un enciphered preamble followed by an enciphered text arranged in five-letter groups. The preamble, as transmitted by a German operator, contained six items of information: 1 The call signs of the radio stations involved: first the sending station, then the destinations). 2. the time of origin of the message. 3. The number of letters in the text. 4. An indication whether the message was complete, or was a specified part of a multi-part message (for example, the second part of a four-part message). 5. A three-letter group, the discriminant, which distinguished among different types of Enigma traffic. 6. A second three-letter group, which I will call the "indicator setting." This was related to the procedure for encoding and decoding the text of the message. INTERCEPT OPERATOR'S ADDITIONS a. Frequency 4760 Kilocycles b. Time of Interception 11:10 UN ENCIPHERED PREAMBLE 1. Call Signs: P7J to S-F9 and 5KQ 2. Time of Origin: 10:30 3. Number of letters: 114 4. Single or Multi-part: Part 2 of 4 parts. 5. Discriminant: QXT 6. Indicator Setting: VIN ENCIPHERED TEXT W Q S E U P M P I Z T L J U W Q E H G L /? B I D F W B 0 1 P D A Z H T T B R 0 A H H Y 0 <> J Y G S F H Y K T N T D B P H U L K 0 H U N T I M 0 F A /? L B P A P M X K Z Z X D T S X L 0 W H V L K A G U Z Z T S G G Y I V Figure 3. 2 Composition of a Typical Enigma Message Before this information in the preamble, the intercept operate would include two more items: a. The radio frequency used for transmission, b. The time of interception. Note that the British intercept operator would have warning that a message was about to be transmitted, because the control station of the radio net would have been making the necessary arrangements. Thus he would often be able to enter the frequency and time of interception, and perhaps the call signs too, before the German operator had gotten started on the preamble and the text. The messages from Chatham were handwritten, exactly as taken down by the intercept operator. Sometimes our intercept operator would be uncertain of the correct interpretation of a Morse character, in which case he would give alternatives. Sometimes the signal would be garbled, and he would miss part of it, in which case he would tiivc an estimate of the number of five-letter groups he had missed. Kendrick started to work on the large collection of material from Chatham, and set me a good example by beginning to analyze its characteristics in a methodical manner. His approach was reminiscent of the period some five to ten years earlier when I had been doing research in algebraic geometry and had often been faced with the problem of thinking of something to think about. In those earlier days I had found that the best approach to this problem was to force myself to start writing, and here was Kendrick dealing with the same problem in the same way. I followed suit. We simply started making lists of this and that, and gradually we came across things that seemed of some interest. Soon, however, Kendrick was transferred to another job. I had to carry on alone. It was about then that Josh Cooper, in charge of Air Force operations at Bletchley Park, gave me what was at that time our one and only collection of decoded German Enigma messages. It did not amount to much: decodes of at most two or three days of intercepted traffic. At the time I did not know how the collection had been obtained. When I started to delve into those few decoded German Enigma messages, I had little idea of what I would find. However, the fact that Josh Cooper was able to hand them to me at that time was to have far-reaching consequences, for I believe it was the study of these decodes that gave me my first glimpse of what my job was all about. Previously I suppose I had absorbed the common view that crypt analysis was a matter of dealing with individual messages, of solving intricate puzzles, and of working in a secluded back room, with little contact with the outside world. As I studied that first collection of decodes, however, I began to see, somewhat dimly, that I was involved in something very different. We were dealing with an entire communications system that would serve the needs of the German ground and air forces. The call signs came alive as representing elements of those forces, whose commanders at various echelons would have to send messages to each other. The use of different keys for different purposes, which was known to be the : reason for the discriminants, suggested different command structures for the various aspects of military operations. | Even more important, perhaps, was the impression I got from the messages themselves. Although my knowledge of German was very limited, I could see that the people involved were talking to each other in a highly disciplined manner. They were very polite to each other, in that the originator of a message would be careful to give the full title of the officer or organization to which the message was to be sent. Furthermore, in the signature that came at the end of the message, the originator would be careful to give his own title inl full. These early impressions proved to be of immense importance! later on, and it was fortunate that I had this period of secluded! work. m Before I can explain the early development of traffic analysis in the School, I must say a little about the Enigma machine and how was used. I must beg the reader not to be misled by descriptions of the machine that have appeared in the literature, and I must also ask for patience. For, if I am to explain the weaknesses in the machine and in its usage that were so helpful to us, I must discuss its anad omy with some care. We are concerned with the portable battery-operated machine that appears in General Guderian's command vehicle (Figure 2.1fl One can see that this machine has a keyboard of twenty-six key corresponding to the letters of the alphabet. Behind the keyboard a lamp board When a key, say a letter P, is pressed, some oth^B letter, say Q, lit up on the lamp board will be the encode of letter ^1 But, if the same letter P is pressed repeatedly, the encoded letters ! up on the lamp board will appear to be a random sequence. This^l because the substitution of letters is produced by electrical connections through the wheels of a "scrambler" at the back of the machine, u hose setting is varied as each letter of a message is encoded by pressing a key on the keyboard. How it was arranged that an authori/cd recipient could decode a message u ill appear shortly. The Germans had adopted the principle that the security of their communications must rest not on the machine itself, but rather on a "key" that would determine how the machine was to be set up for a particular purpose. Moreover, they were concerned with both external and internal security. They wanted to prevent their enemies from reading their messages and also to prevent their own units from reading messages that were not intended for them. For example, three of the many different types of Enigma traffic were messages between operational units of the regular army and air force; messages between units of Hitler's private army, the SS or Schutz/.staffel; and messages involved in the training exercises of new signals battalions. All three kinds of messages were enciphered on identical F.nigma machines, but the regular army and air force units were not to be allowed to read the texts of SS messages. Nor were the trainees to be permitted to read the texts of the other two types of traffic. Consequently, different keys were issued for different types of traffic. This, however, did not quite solve the problem, because messages of different types were often transmitted on the same radio net. It was therefore necessary to provide means by which a receiving unit's operator would know what type of clear text was hidden behind the enciphered text, and whether he had the necessary key to read it. The Germans chose to solve this problem by using a three-letter "discriminant" transmitted in the un enciphered message preamble. Let me emphasize: This discriminant was not part of a key. Its purpose was simply to indicate which of many keys was being used. A cipher clerk would examine the discriminant of each incoming message to determine whether he had been issued the key used for its text encipherment. If he did have the key, he could set up his enigma and decode the message. If not, he couldn't. To understand what follows, we must become fully acquainted with the machine that appears in Guderian's command vehicle, figure 2.1. The basic components are the keyboard, lamp board scrambler unit, and steckerboard. The keyboard appears at the end nearest to the cipher clerk who is operating the machine. It has twenty-six letter-keys arranged as follows: QWERTZUI 0 ASDFGHJ K PYXCVBNML The lamp board which appears behind the keyboard (from the operator's viewpoint) is a similar arrangement of letters on small glass windows, any one of which can be illuminated from below by a lamp On looking at the keyboard in Figure 2.1, one might think that it must be like a typewriter keyboard--but one would be wrong. For one thing, the alphabet keys are arranged a little differently. Then again, a typewriter keyboard has about twice as many keys,] including ten for the numerals 0 through 9 and functional keys such as the shift, space, backspace, and carriage return. Because the Inigma keyboard has only twenty-six keys, corresponding to the] capital letters A through Z, numerals in message texts had to be! spelled out in full. Also, pressing a typewriter key causes mechanical motions that may or may not include the printing of a symbol,! while the Enigma had no printing mechanism. Pressing its letter-! keys caused only one type of mechanical motion, that of the scram-! bier, of which more in due course. Figure 3.3 shows the commercially available Inigma machine that the Polish cipher bureau had already acquired when Mariar Rejewski began his studies in 1932. The heart of this Scherbii Inigma was its scrambler unit, built around a set of four movable wheels, or rotors. Protruding flanges of the wheels appear behind the keyboard and lamp board Comparing Figure 3.3 with Figur 2.1, we see that the military version of the Inigma, developed by the Germans for widespread use in World War II, has a scrambler in which there are only three movable wheels. However, the military version, with which we are concerned, had five different wheels numbered 1 through 5, only three of which were placed in the scrambler unit at a time. * Or "glow lamp." Hence the name "Glow lamp Machine" by which the German Inigma was called. Figure 3.3 The Scherbius Enigma (from the original sales pamphlet) To the right of the scrambler's three movable wheels is a fixed commutator, a shallow cylinder containing a circle of twenty-six flat terminals with which twenty-six spring-loaded pin terminals of the right-hand wheel can make contact. The flat terminals of the commutator will be called the "in-out scrambler terminals." They correspond to the letters A to Z in alphabetic order. Each of the three movable wheels in the scrambler has a circle of spring-loaded pin terminals on the right, numbered 1 through 26, which are cross connected to a circle of flat terminals on its left-hand side. (The five wheels, numbered 1 through 5, have different internal cross connections.) To the left of the three movable wheels, in a fixed cylindrical mounting, is a circle of spring-loaded pin terminals, 1 through 26, which are interconnected in pairs. This fixed cylinder at the left-hand end of the scrambler is called the "turnaround wheel," or, in German, ttmkehrwalze. F.ach of the protruding flanges of Figure 2.1 has twenty-six rounded identations into which a spring-loaded bar at the back of the machine is pressed. This bar holds each wheel in one of twenty-six discrete positions to ensure that good electrical contact will be made between the pin terminals and flat terminals of adjacent wheels. But the operator can rotate a wheel to any desired position by finger pressure on the protruding part of the flange. The position of each wheel is indicated by a letter on an alphabet ring, which shows through a window to the left of the flange. These alphabet rings, however, are not fixed to the wheels. On each wheel the ring can be rotated and held in position by a spring clip so that any chosen letter, say K, will be opposite a fixed "zero" position of the wheel. This chosen letter is called the ring setting. Thus the scrambler unit is a package consisting of the in-out terminals A to Z of the commutator, the three movable wheels with their internal cross-wirings, and the fixed interconnections of the turn-around wheel, or umkebrivalze. Electric current entering at one of the in-out commutator terminals on the right will flow from right! to left through the internal connections of the three wheels to the umkehrwalze, which will turn the current around and send it back by a different route through the three wheels to an in-out terminal) different from the one where the current entered. In fact the whole] effect of the scrambler, in any one position of its wheels, is to] interconnect pairs of its in-out terminals. Any such electrical interconnection will work both ways. Thus if current put in at in-out terminal G comes back to in-out terminal Q, current put in at Q u ill come back to G. To understand how we broke the Enigma codes it is important to make a clear distinction between the scrambler and the whole Enigma machine. This distinction was, in fact, at the bottom of our success. An Enigma machine had a keyboard, a lamp board and a steckerboard as well as a scrambler unit. The scrambler consisted only of in-out terminals, three moving wheels, a fixed turnaround wheel, or umkehrwalze, and a drive mechanism, which I will describe later. When I talk about a scrambler I will mean just that. The steckerboard or "cross-plugging board" appears in Eigure 3.4, in which the front flap of the Enigma is folded down, revealing an array of twenty-six pairs of sockets. These correspond to the letters of the alphabet arranged in the same pattern as the keyboard. The purpose of the steckerboard is to vary the interconnections between the in-out terminals of the scrambler unit and the letter keys and lamps. If there is no plug in a particular pair of sockets, say the pair corresponding to letter A, then input-output terminal A of the scrambler is connected to letter-key A and lamp A. In this case we say that letter A is steckered to itself. On the other hand, if one of the two-wire cables shown in Figure 3.4 is used to provide cross-plugging between socket-pair B and some other socket-pair, say G, the result is that in-out scrambler terminal B is connected to letter-key G and lamp G, while in-out scrambler terminal G is connected to letter-key B and lamp B. We say that B is steckered to Cl, or BIG. H>r each type of traffic, keys were issued for a month at a time to authorix.ed units. Each key was valid for twenty-four hours, and was changed at midnight. A complete Enigma key consisted of three items: 1- I he order of the three wheels that were to be placed in the scrambler unit. Example: 413. - I he ring settings of the left-hand, middle, and right-hand wheels in the scrambler. Example: OUB. Figure 3.4 Enigma with Steckerboard Uncovered 3. Cross-pluggings on the steckerboard. Actually only eleven cross-pluggings were specified, leaving four letters steckered to themselves. For example: KY, T/0, Z/L, B/U, W/D, C/X, I/J, G/A'P/H'N/R,andF/M, leaving E, Q, S, and V self-steckered. When two Enigma machines are set to the same key and their three wheels are in the same positions, the electrical connections through their steckerboards and scramblers will produce the same thirteen pairings of the twenty-six letters of the alphabet. Moreover the resulting letter substitution is reversible, since electric current can flow in either direction along the path that results in each pairing. Thus, if pressing letter-key K on one of the machines causes lamp P to be lit, then pressing letter-key Pon the other machine will cause lamp K to be lit. This is the basic principle of encoding and decoding on the Enigma. Once it is fully understood the rest is easy. When the text of a message was to be encoded, the originating cipher clerk was allowed to choose the wheel setting, say RCM, at which he would start the encoding process. I will call this the "text setting" for the message. With his Enigma set up to the appropriate key, he would use the flanges to turn the wheels until the letters R, C, and M appeared in the windows. He would then press the keyboard keys corresponding to the successive letters of clear text, writing down the sequence of letters lit up on the lamp board to obtain the enciphered text. As the keys are pressed, the wheels go through a cyclic series of positions. The right-hand wheel moves on one position every time a key is pressed. When the right-hand wheel reaches one particular position, known as its "turnover position," the middle wheel also moves ahead one position. In due course the middle and right-hand wheels may both be in their turnover positions, at which time pressing a letter-key will cause all three wheels to move Each time a key is pressed, encipherment takes place when the wheels have reached their new position. The three wheels would have to go through a cycle of 26 x 26 x 26 = 17,576 successive positions before they would return to their starting position, RCM. A receiving cipher clerk, with his Enigma set to the same key, would turn his scrambler wheels to the text setting RCM, and tap out the successive letters of the enciphered text he received. The wheels of his scrambler unit would follow the same cycle of positions as those of the encoding Enigma, so that the original clear text would be recovered letter by letter. But for this to happen--and this is vital--the decoding operator would have to be told the wheel setting, RCM, that had been chosen. How this was done introduced yet a further level of obscurity through which a cryptanalyst would have to fight his way. Fortunately, when I arrived at the Cottage on September 4, 1939, we already knew the procedure that the Germans were using. Before encoding the clear text, the originating cipher clerk carried out a preliminary procedure. He chose a wheel setting at random, say VIN, which we will call the "indicator setting." He then set his scrambler wheels to VIN and encoded his chosen text setting twice over, tapping out the letter sequence RCM RCM on the keyboard and noting the sequence of letters, say WQSEUP, appearing on the lamp board This sequence, which is the encode of RCM RCM with starting position VIN, we will call the "indicator." Having obtained his indicator in this manner, the encoding operator turned his scrambler wheels to his chosen text setting, RCM, and proceeded to encode his plain text message letter by letter. When the encoded message was transmitted on a radio net, the indicator setting VIN was sent at the end of the un enciphered preamble, and the six letters of the indicator, WQSEUP, appeared at the beginning of the message text, constituting the first five-letter group and the first letter of the second group (Figure 3.2). The encipherment of the clear text! started with the second letter of the second group. A decoding cipher clerk with his Enigma set up to the same key turned his wheels to the indicator setting, VIN, that he found in the preamble. He then pressed letter-keys W7QSEUP in turn, and saw the letters RCM RCM appearing successively on his lamp board knew from this that the text setting chosen by the originating cipr clerk was RCM; he could proceed to decode the message by setting his wheels to positions R, C, M, tapping out the successive letters of the encoded text, and writing down the letters that appeared on the lamp board Note that the process of encipherment on an Enigma machine was not directly related to the transmission of the enciphered message. In fact the Enigma was what is known as an off-line cipher machine. Getting an Enigma message on the air involved three people: the originator, a cipher clerk, and a radio operator. The originator would prepare the clear text on a standard form, at the top of which he would note such things as the addresses, the time of origin, and, if necessary, the Enigma key to be used. A cipher clerk would then go through the encoding process I have described, preparing a message text and a preamble that would include a discriminant and an indicator setting. The preamble and message text would be handed to a radio operator for transmission in manually keyed Morse code. Long messages would be handled in two or more parts, each encoded separately with its own text setting, indicator setting, and indicator. What I have said about the Enigma machine, the keys, and the procedure for encoding and decoding should be sufficient to prepare the reader to understand the ideas that arose in the School and led to the establishment of Hut 6. However, a few additional comments, combined with recapitulation of important points, may prove helpful. The effect of the scrambler unit in any particular position is to interconnect the twenty-six in-out terminals in thirteen pairs. The combination of scrambler and steckerboard has the effect of interconnecting the letter-keys and lamps in pairs. Current can flow in either direction between each of the latter thirteen pairs, so if any letter-key, say A, is pressed, the lamp representing some other letter, say H, will be lit up. If letter-key H had been pressed, lamp board letter A would have been illuminated. Typical letter substitutions produced by the Enigma in six successive positions of its scrambler wheels are shown in Figure 3.5. For each position, the top row represents the letters of the keyboard. The bottom row indicates what letter of the lamp board would be lit by pressing a letter key. The property of reversibility illustrated by the letter substitutions of Figure 3.5 can be used to explain the principle of encoding and decoding. Suppose that the letters TARGET are enciphered POSITION ABCDEFGHIJ KLMNOPQRSTUVWXY H I U L O S O A BTfDXYEVGWFJ C f> RUN ABCDEFGHIJ KLMNOPQRS T U V W X Y FVLGMADYPNSCEJU ITZKQOBXWH ABCDEFGHIJ KLUNOPQRSTUVWXY NWSEOXHGZKJMLA Y R V P C U T Q B F O ABCDEFGHIJ KLMNOPQRSTUVWXY G U V I H T A E D L XJWYRZSOQFBCMKN A B C D Fff H I K L. MNOPQRSTU V W X Y QZSMPTOLXKJHDVGEAUCFRNYI W ABCDEFGHI JKL.MNOPQRSTUVWXY SEOKBTRVJI DUWZCYXEAFLHMQP Figure 3. 5 Letter Substitutions in Six Successive Positions of the Enigma successively in the six positions. In the first position is substituted for T, in the second F for A, and so on. Thus pressing letter keys I, A, R- G, K, and I in turn will light up lamps J, F, P, A, P, and F. In fact JFPAPF is the encode of TARGF/f. Now suppose that some other cipher clerk, receiving JFPAPF, can set his Inigma to produce the same sequence of substitutions. Then, as Figure 3.5 shows, in the first position pressing letter-key will light up letter T, in the second F gives A, and so on. Thus the second clerk recovers the original clear text word, TAR(1F'T. However, let me repeat that in order to get the same sequence of substitutions, the receiving cipher clerk would have to know two things--the key to which he must set up his Inigma and the setting of the wheels that the originating cipher clerk has chosen to use to encode TARGET. The receiving clerk would have been given the key ahead of time, but the originating clerk would have to give him the wheel setting. When an Inigma machine is being assembled for use, the three chosen wheels are placed in order on a spindle. The left-hand end of this spindle fits into the center of the umkehrwalze, the right-hand end into a central sleeve attached to the commutator containing the in-out terminals. A lever-operated mechanism then pushes the in-out terminal commutator toward the umkehrivalze so that good contact is established between the circles of pin terminals to the right of the umkehrxalze and the three wheels, and the circles of flat terminals on the wheels and the commutator against which they are pressed. When this has been done, electric current entering the scrambler at any in-out terminal, say G, will flow through the three wheels from right to left, be turned around by the umkebrwalze to from back through the wheels from left to right, and will emerge at an input-output terminal. 1 he whole arrangement of lamp board keyboard, steckerboard, and scrambler unit is shown diagramatically in Figure 3.6, which may help clarify what I have said about the machine. The letter keys Q through L and lamps Q through L are connected electrically to the corresponding upper sockets of the steckerboard. The lower sockets are connected to a set of twenty-six terminals, which I will call the in-out steckerboard terminals. These terminals, in alphabetic sequence A through Z, are connected to the circle of scrambler m-out terminals A through Z by a twenty-six-way connector cable. iIN-OUT SCRAMBLER TERMINALS RIGHT HAND WHEEL----------------MIDDLE WHEEL--------------------LEFT HAND WHEEL--UMKEHRWALTZE --SCRAMBLER UNIT / ? ( If / < | > > ( / If / > ' > < ^ ) > / ) / / } / , ' ' i i - > r / > 9 (, : / 7 * 6' / t f 1 i it > J 1 A ' C IN OUT LAMP BOARD KEYBOARD STECKERBOARD IN-OUT STECKER TERMINALS 26-Way Conntctor CaDI Figure 3.6 Electrical Connections Between Lampboord, Keyboard, Steckerboard and Scrambler When any letter-key is pressed down, the circuit to the corresponding lamp is broken and contact is established between an electric battery and the corresponding upper socket of the steckerboard. Current Hows from this upper socket either to the corresponding lower socket or through a cross-connection to one of the other lower sockets, reaches an in-out steckerboard terminal, and is passed on to the corresponding in-out terminal of the scrambler. Pressing the letter-key will also have caused the scrambler wheels to move to a new position in which the scrambler will produce a return current to another of its in-out terminals. After passing through the steckerboard, this return current will light up a letter on the lamp board With five wheels to choose from, the three wheels to be used in the scrambler can be selected in ten ways, and for each such selection the three wheels can be arranged in six different ways. Thus, if wheels numbered 1,3, and 4 are chosen, the six arrangements are: 134, 143, 314, 341, 413, 431. In all, therefore, there are 10X6 = 60 possible wheel orders to choose from. I have noted that the alphabet ring on each wheel is movable, and can be set to any one of twenty-six positions, by means of a spring clip, which holds a selected letter of the ring opposite a fixed "zero" position on the wheel. This means that the position of a movable wheel in relation to the fixed in-out commutator is not determined by the letter of the alphabet ring appearing in the window, unless the ring setting too is known. Thus the setup of the scrambler that is to be specified by a key involves 26 X 26 X 26 = 17,576 possible combinations of ring settings on top of the 60 possible wheel orders. Altogether there are 60 X 17,576 = 1,054,560 possible ways of setting up the scrambler, any one of which can be specified in a key by a set of three numerals, say 413, determining the wheel order, and a set of three letters, say OUB, determining the ring settings. If the creator of keys had chosen to specify thirteen stecker pairs the number of possible cross-pluggings of the steckerboard would have been 25x23x21 X 19 X 17 x 15 X 13 X 11 X9X 7 x -s * 3, or almost 8,000,000,000,000, a formidable number. Surprisingly, by leaving four letters unsteckered he increased the possibilities by a factor of 26, giving more than 200,000,000,000,000 (two hundred trillion) possibilities, an even more formidable number. As we have seen, the wheel order and ring settings provide more than a million variations, so the total number of possible keys is morel than 200 quintillion (or two hundred million million million Even when the key is known, a message cannot be decoded without an additional item of information--the three-letter text settingl chosen by the originating cipher clerk as a starting point for encoding. This text setting could be chosen in 26 X 26 X 26 = 17,5761 ways. Thus the Enigma, though simple in principle and primitive! in many ways, presented the cryptanalyst with a dazzling numberf of possibilities. When I went from Cottage to School in September 1939, with! instructions from Dilly Knox to study call signs and discriminants, I| had been told about the operation of an Enigma machine and about! the composition of an Enigma key. I also knew that, associated with! each such key, there would be four three-letter groups called "discriminants," such as: KQV, LMY, GJR, and ABO. These discriminants, as I have said, were not part of the key. One! of the four would be included in the un enciphered preamble of anyi message whose text had been enciphered with an Enigma set up to] this particular key. The order of letters in a discriminant could bei varied. For example, QVK or VKQ could be used instead of KQV. 1 I don't know why the Germans chose to use four discriminants at that time, and it doesn't matter. What does matter is the purpose of the discriminant that appeared in the preamble of an Enigma message. This was simply to tell a receiving cipher clerk which, if any,! of the keys issued to him had been used to encipher the text. This! much had been explained to me, and someone, I don't rememberl who, must have told me a little about the operation of the radio nets! of those days and the use of call signs for I had known nothing aboutj such matters when I arrived at Bletchley Park. | The Germans, in their concept of a blitzkrieg, reckoned that many groups of fast-moving fighting, command, support, and staff elements would need effective communication among themselves wherever they might be, and furthermore that the activities of these * After all these years I am not quite certain that the number of self-steckered letters was four. It could have been six, in which case the total number of stecker combinations would have been around 150 trillion rather than a little over 200 trillion. This, however, does not affect the story that I am telling. groups would have to be tied into the higher command system. The elements of each cooperating group were to be served by signals detachments operating a "radio net" on an assigned radio frequency. Under ideal conditions any message transmitted by any radio station operating in the net on the assigned frequency would be heard by all the other stations. One station of the net would act as control, to ensure that no two stations would cause interference by transmitting at the same time. There were to be many such nets, and a station could operate in two or more nets, so that messages originating at any point could be relayed to any other point. The call signs were simply the means of identifying the individual elements that were participating in this overall radio communications system. When messages were passing between elements within a single radio net, the preamble would contain the call signs of the originator and intended recipients) of each message. When a message was to be forwarded to other elements, their identifying call signs would also be included in the message preamble. Thus, by studying call signs we had an opportunity to learn something about the structure of the enemy's forces. As the call signs were changed every day, however, the detective work had to begin anew every twenty-four hours. Our intercept operators listened to the Enigma messages and their preambles, writing them out by hand on standard message forms. The main part of the form was used for the succession of five-letter groups, or "words," which constituted the indicator and text of a message enciphered on an Enigma machine. At the top of the form was a space in which the intercept operator entered the preamble that the German radio operator had transmitted ahead of the message. Indeed the form used by our intercept operators must have been very similar to the form used by the German cipher clerks. The intercept operator, however, also entered the radio frequency on which he was listening and the time of intercept (see Figure 3.2). When I started to study my working material I simply followed Keiulrick's example of painstakingly making lists and charts in the hope that something interesting might turn up. As no one else at Bletchley Park, so far as I could tell, had studied any of the material at all thoroughly I was soon breaking new ground. The enciphered texts of the intercepted messages from Chatham were of course incomprehensible, and anyway of no interest to me at the time, but I took note of the radio frequencies and the times of interception, as well as the discriminants, the call signs and the times of origin that appeared in the message preambles. I charted each day's traffic by radio frequency and time of intercept. When I had succeeded in grouping the discriminants of the day in sets of four that were used with the same key, I marked the entries on the chart by underlining them in different colors according to their keys. At first it was usually possible to identify three different keys, for which I used red, blue, and green pencils. As it turned out I had, out of necessity, designed a format that remained virtually unchanged for half the war. Even the color names stuck--particularly "Red." As part of their daily reports the traffic analysts at Chatham gave the groupings of discriminants, and they sometimes got them wrong. The reason was that they did not fully understand the working of the Enigma machine and the purpose of the discriminants. Since I had come to understand this purpose in the terms I have described, it was obvious to me that the successive parts of a multi-part message, since they would be going to the same addressees), would be encoded with the same key, and hence thar the discriminants in the preambles of all the parts must belong to the same set of four. Thus in analyzing a day's traffic I would start by looking for multi-part messages. Fortunately the Germans nearly always used different discriminants in the successive parts, which often made it very easy for me to establish a group of four. Indeed, one four-part message would ordinarily do the job for me. I knew very little about the "Green" traffic that appeared sporadically in the Chatham intercepts until I studied French intercepts and reports, which told me that this traffic came from the administrative network of the German army. It was not at all easy to intercept at Chatham, as most of it was on frequencies in the medium frequency range and did not carry far. Consequently the French intercept stations did a lot better than Chatham, and their intercepts gave me a good deal of Green traffic over quite a long period. When I analyzed the call signs of earlier Green traffic I found that they were repeating from month to month, so I was soon in a position to give Chatham a forecast of call signs that might be expected on the Green network each day. These two matters of the correct grouping of discriminants and the repetition of call signs may seem trivial, but in the end they proved important. In those early days, their greatest value to me was that they gave me a sufficient excuse for my first venture into the outside world: a visit to the army's radio intercept station at Chatham. I immediately made friends with the officer in charge, Commander Ellingworth, who taught me many things I badly needed to know, and who was to be a tower of strength throughout the war. He gave me a very friendly welcome, introduced me to his staff, took me around the station, and let me watch the operators at work. We had a lot to talk about. Because of security regulations I could not yet tell him about the small collection of decodes that Josh Cooper had given me, but I did talk about my study of discriminants and the French intercepts. For his part, he began to give me a picture of how the German radio nets operated, and informed me about the problems of interception. The reader may well wonder what these things have to do with cryptanalysis, but the fact that I learned about them so early in the game was important for the ultimate success of Hut 6, which was to depend largely on the close coordination of our activities with those of our intercept stations. Ellingworth told me a little about the characteristics of shortwave radio transmissions and the problems involved in intercepting weak signals in Morse code from distant transmitters. I find that some people today who listen only to AM and FM radio find it hard to believe that we in England could pick up German radio signals from places as far distant as Stalingrad and the borders of Egypt. The point, of course, is that for radio transmissions in the AM and FM ranges there is no reflection from the upper atmosphere. Strong shortwave transmissions, on the other hand, can sometimes bounce back and forth between ground and upper atmosphere several times, permitting reception at extremely long distances. Marconi sent the first messages ever transmitted by wireless (from England to Australia on September 22, 1918, using this phenomenon. Ellingworth went on to explain that radio frequency measurements were not exact. The actual transmitting frequency of a German radio station could drift quite a bit from its nominal value, so that intercept operators might well record different frequencies on messages from the same radio net. This drifting had its effect both on the intercept operators and on my analysis. At my end, it was sometimes difficult to distinguish on my frequencytimeof-day charts between different nets using nearby frequencies. At the intercept station an operator might be in difficulty if he were asked to take messages from a net operating on a specified frequency. He would have to hunt around, and could easily pick up a station on another net operating on a nearby frequency. I also learned that a great deal depended on the individual intercept operator. His effectiveness rested partly on experience in listening to the German radio nets and in getting to know their individual characteristics, and also on acquired arts, such as that of picking up a weak signal that no ordinary operator would hear at all. And these were individual talents. Some of the old hands at Chatham were quite phenomenal. The extraordinary human combination of brain and hearing made it possible to pick up signals that would have been missed by any automatic method of reception. By the end of the day Ellingworth and I had agreed on what was to prove a basic procedure throughout the war--that a register of all intercepted Enigma traffic would be sent to Bletchley. The traffic analysts at Chatham had other tasks and could not attempt the sort of analysis that I had initiated, but they could use my results if they could get them quickly enough. For my analysis I did not need the full text of each intercepted message. All I needed was the information that the intercept operator had entered at the top of the message form: the frequency and time of intercept; the preamble, containing discriminant, indicator setting, time of origin, call signs and so forth; and the first two five-letter groups of the message text, which contained the indicator. Ellingworth agreed to send me this information by teleprinter as the messages were intercepted, waiting only for the accumulation of enough items to make a transmission worthwhile. This was our "traffic register." It was agreed that I would telephone each day as soon as I had identified sets of discriminants. I would provide all available predictions of call signs and in general I would make suggestions about what traffic seemed to me to be of particular interest. Thus was born a system of cooperation between Bletchley Park and the inter- cept stations, the critical importance of which will become fully apparent in later chapters. It must have been soon after my visit to Chatham that I acquired my first assistant, Patricia Newman. She took over the job of making entries on the day's frequency time chart as each page of the traffic register arrived by teleprinter from Chatham. Between us we improved the organization of the enormous daily chart, which was about the size of a desktop. When I had identified the Red, Blue, and Green discriminants for the day, and had telephoned them to Chatham, we were able to mark the message entries on the chart accordingly, using colored pencils. With Patric 't doing a lot of the work, I was able to investigate the messages that had discriminants other than those of the Red, Blue, and Green traffic. The Chatham people had tended to follow the radio nets that they knew, so this other traffic appeared only sporadically. However, by frequent telephone calls, Ellingworth and I were able to direct interceptor attention to this "other traffic," and soon could investigate it more thoroughly. As I remember, it was not long before we were following Brown and Orange traffic as well as Red, Blue, and Green. Thus we were distinguishing among several different groupings of elements in the German military force structure, all of them using the same Enigma machine, but with different keys. As our work developed, Patricia Newman was joined by Peggy Taylor, and the empty room in the Bletchley Park Schoolhouse, with its long table and bare walls and no view from its windows, became less lonely than it had been. Nobody else seemed much interested in what we were doing except Josh Cooper and Colonel Tiltman. Cooper and Tiltman were not only heads of expanding sections, they were both distinguished cryptanalysts. I was far too wrapped up in my own work on Enigma to know about the breaking of other codes and ciphers under the direction of these two experts. I remember, however, being told by Josh Cooper that his work was an almost intolerable strain. Success so often depended on flashes of inspiration for which he would be searching day and night, with the clock always running against him. My worries would be of a different nature. In my own work, as I have said, I was quick to realize that my problem was not that of classical cryptanalysis, in which brilliant experts struggled with individual messages. The classical crypt analyst would have paid little attention to the communications system involved, yet for us it was essential first to understand the German radio nets in detail, and to be able to distinguish among types of traffic. It was lucky that, as soon as I went from Cottage to School, I began to analyze intercepted Enigma traffic simply as traffic--worrying about the structure of the communications system rather than the unknown content of the messages that it carried. I was working almost alone at the start, and the progress that I was able to make was small in comparison with the later achievements of others. It was enough, however, to get things moving in the right direction. It must be emphasized here that in building up the organization that was to become Hut 6 to the point at which it could produce decodes of Enigma messages I was dealing with only part of the Enigma material. The Germans employed Enigma cipher machines for many different purposes. There were thus several different categories of Ultra, derived from the decoded Enigma traffic of different users, such as the army, navy and air force, and the spy organization. Eor this reason I have introduced the term "Hut 6 Ultra" to denote the intelligence that was derived from Enigma traffic decoded by Hut 6, which was mostly army and air force traffic. (Enigma traffic was also decoded elsewhere with keys provided by Hut 6.) The immense task of analyzing this and squeezing from it the last drops of intelligence about enemy capabilities, intentions, etc." was to be performed by our sister organization, known throughout the war as Hut 3. From my close association with Hut 3, I am qualified to say a little about what they did, and I intend to do so in the proper place. However, I am definitely not qualified to talk about, for example, Hut 8's decodes of the German navy's Enigma traffic, nor about the production of Hut 8 Ultra intelligence. As Ronald Leu in points out, Bletchley Park activities were highly compartmentalized, largely because of the high level of security involved. Feu of us knew, or wanted to know, anything about what was going on outside our own bailiwick. _i Ideas and Plans October and November 1939 While I was still working in isolation in the School, during the second and third months of the war, I made two somewhat startling breakthroughs. People have asked how these breakthroughs came to me, and it is really very hard to explain. Basically the answer goes back to a memory from childhood; that of being lucky enough, with no purposeful effort on my part, to find myself opposite the vacant chair when the music stopped. In this chapter, however, I will attempt to give a lucid account of something that I find difficult even to understand myself: just how the two important advances came into my mind so soon after my arrival at Bletchley Park. If obscurity creeps in from time to time, it will hardly be surprising. I can only hope that I will be able to satisfy those readers who will want to know how my thinking developed, without infuriating others. If anything is certain, it is that the breakthroughs came to me a lot more rapidly and easily than the explanation I am about to write. But the fact that these advances did actually occur in October and November of 1939 is so important that it must be tackled here. The "fun" part of my story, which will be told in due course, did not start until early 1940. The first of my two ideas had to do with the indicator setting and indicator of an Enigma message. I can identify ten successive steps in my thought process. The reader should not be disturbed if some of them seem trivial. The final results, and some of the intermediate steps, were very exciting. As I have already explained, a three-letter indicator setting, say VIN, appeared in the un enciphered preamble of an Enigma message, and a six-letter indicator, say WQSEUP, appeared at the start of the enciphered message text. I had been told that this indicator setting and indicator were the means by which an originating cipher clerk informed a distant cipher clerk that he had chosen a particular wheel setting, say RCM, as a text setting--the setup of the three wheels of the scrambler at which he had started to encode his message text. Moreover I had also been told that the letter sequence WQSEUP was obtained by enciphering RCM RCM with starting position VIN. From this I knew that the letters W and E in the first and fourth places of the indicator were the encodes of the same letter. Similarly Q and U were the encodes of the same letter. So were S and P in the third and sixth places. This was real solid information. Could we use it? Getting to the point of asking myself this question was Step 1. Having reached this starting point, I proceeded to think, largely by chance, about indicators in which the same letter would appear in places 1 and 4, or 2 and 5, or 3 and 6. For a reason that will soon appear, indicators of these three types came to be known as "females." To distinguish between the types, I will refer to them as 1-4 females, 2-5 females, and 3-6 females. For example the following combinations of indicator setting and indicator might occur: KIE SPESNT (Letter S in places 1 and 4) LTS VBYQGY (Letter Y in places 3 and 6) EGP OHAOCM (Letter 0 in places 1 and 4) RYM XWNPWV (Letter W in places 2 and 5) XXY ZDFJDA (Letter D in places 2 and 5) In the first example, a 1-4 female, letter S in places 1 and 4 must have been the encode of the same unknown letter. Thus the Enigma, by means of its steckerboard connections and scrambler, had been able to pair the same two letters of the alphabet in two machine positions that were three places apart in the machine cycle. Next I must have come to wonder whether this was always possible. This was Step 2. The reader can appreciate my growing excitement at this stage of the thought process by going back to Figure 3.5, in which I showed typical reversible alphabet substitutions that might have been produced by an Enigma in six successive positions of its scrambler wheels. In positions 1 and 4 there is one pair of letters, N and Y, that appears in both letter substitutions. In positions 2 and 5 no letter pairing is repeated. Positions 3 and 6 have no letter pairing in common. Thus it was we? always possible for the Enigma to produce the same letter pairing in two positions three places apart in its cycle. This was the germ, seed, or whatnot from which the success of Hut 6 was to grow. It was Step 3, but I still had a long way to go in my thinking before the ultimate answer dawned. I must have wondered about the probability that two different positions of the Enigma would produce the same letter pairing. Although I had been a professional mathematician before the war, probability was never one of my strong points. Later in the war I was to depend on others, notably Hugh Alexander and Alan Turing, for answers to the many questions of probability that arose. But in October 1939 Hugh Alexander, the British chess champion, had not yet joined the staff, and I hardly knew Alan Turing, who was working with Dilly Knox in the Cottage. I suppose I must have solved the problem myself. Fortunately it was an easy one. In each of the two positions, the Enigma would produce a reversible scramble of the alphabet, one in which if A goes to B then B goes to A. Suppose that the first position produces the scramble ABCDEFGHIJKLMNOPQRSTUVWXYZ RGLYPZBTJWTSCO QMENAKJHX I VDF (I he reversible feature is illustrated by the fact that A goes to R and R to A, B to G and G to B, and so on. The scramble involves thirteen letter pairings: AR, BG, etc.) In a second reversible scramble letter A is equally likely to be paired with each of the other 25 letters, so the chance that it will be paired with R is 1 in 25. Similarly the chance that any particular letter pairing of the first scramble will recur in the second scramble is 1 in 25. Since there are 13 pairings in the first scramble, the chance that one of them will appear in the second scramble is 13 in 25, or approximately 1 in 2. Let us take this a little further and calculate the chance that the same letter will actually be repeated in the first and fourth positions of the six-letter indicator of an Enigma message. And let us use Figure 3.5 as an illustration of successive letter substitutions, or scrambles, that might occur in the six successive positions of the Enigma in which the three letters of the text setting are twice encoded. The mere possibility that the letter repetition could occur depends on whether the two scrambles have a letter pairing in common. We have just seen that the chance of this is 13 in 25. However, assuming that we have hit on a pair of Enigma positions that would permit the same letter to appear in the first and fourth positions of the indicator, we still have to consider that, if this phenomenon is to occur, the first letter of the three-letter text setting must have been one of the two letters that are paired in both positions of the Enigma. The probability of this is 1 in 13.* Thus the chance that the same letter will be repeated in positions 1 and 4 of an indicator will be 13 in 25 multiplied by 1 in 13, or 1 in 25. The chance that the same letter will be repeated in positions 1 and 4, or 2 and 5, or 3 and 6 is 3 times as good, namely 3 in 25, or approximately 1 in 8. In other words, we could confidently expect that around 1 message in 8 would have an indicator with a repeated letter in either the first and fourth, or the second and fifth, or the third and sixth positions. Thus we have established two properties of the females. First, only about half of the 17,576 choices of indicator setting VIN can offer the possibility of a 1-4 female, and the same applies to 2-5 and 3-6 females. Second, we could expect to find females of one of the * In the example of positions 1 and 4 in Figure 3.5, in which there is a common letter pairing, the first letter of the text setting would have had to be either N or Y, in which case the repeated letter in the indicator would have been Y or N. If the first letter of the text setting had been N, the letters appearing in positions 1 and 4 of the indicator would have been Y and Y. three types at the rate of about 1 in 8 messages. These conclusions represent Step 4 in the thought process. Next came the remarkable discovery that I could ignore all those steckerboard cross-pluggings. Up to now I have been talking about the whole Enigma, with its vast number of possible cross-pluggings. But the question of whether or not the same letter pairing can occur in two positions of the Enigma is a property of the scrambler unit that is completely unaffected by different choices of steckerboard interconnections. This is an extremely important point. If, with the added complexity of the fixed steckerboard cross-pluggings, the Enigma can produce the same letter pairing in positions that are three places apart in its cycle, then the same must be true of the scrambler unit by itself, and vice versa. For, if two keyboard letters are electrically interconnected through steckerboard and scrambler, the letters to which they are steckered must be interconnected through the scrambler alone. The reverse is also true, in that if a pair of letters are interconnected through the scrambler, then the keyboard letters to which they are steckered are interconnected through the whole Enigma. Thus, because we are talking about whether or not the same letter pairing can occur in two different positions of the machine, we can completely ignore the more than 200 trillion (200,000,000,000,000) steckerboard cross-pluggings, which are specified in the Enigma key for the day, and concentrate on the effects of the cyclic motion of the scrambler on the repetition of letter pairings. With only the 60 wheel orders and 17,576 ring settings to worry about, we are down to around a million possibilities. In fact we have reduced the odds against us by a factor of around 200 trillion. This was Step 5, and quite a gain! Ect us return to the five typical females: KIE SPESNT (1-4 female) LTS VBYQGY (3-6 female) EGP OHAOCM (1-4 female) RYM XWNPWV (2-5 female) XXY ZDFJDA (2-5 female) Not knowing the ring settings of the key in use, we would not know the actual wheel settings of the scrambler represented by the indicator settings KIE, LTS, etc. Nor would we know the wheel order. But suppose we try all possible combinations of wheel order and ring settings in turn. For each such combination, and there are about a million of them, we could set up a scrambler to KIE and determine the letter substitutions, similar to those of Figure 3.5, that would be produced in the first and fourth positions of the scrambler following position KIE. About half the combinations of wheel order and ring settings will give letter substitutions with a common letter pairing. The others can be ruled out because they could not produce the female KIE SPESNT. Thus this one female rules out about half the combinations. Similarly the next female, LTS VBYQGY, rules out about half the remaining combinations. Together, the five females listed above would divide the number of possible combinations by a factor of 32. If we could find twelve females among the messages encoded on the same Enigma key, we would be able to divide by a factor of 4,096, leaving only about 250 possible combinations, or a little more than 4 possible ring settings for each of the 60 wheel orders. We could expect one female in every eight messages, so, in order to find twelve females enciphered on the same key, we would have to intercept about a hundred messages. At that time, October 1939, we could already expect to obtain this volume of traffic on the two keys that I was calling Red and Blue. Each of the 250 combinations of wheel orders and ring settings that could not be ruled out by the twelve females would have to be examined more closely to determine whether it might be the true combination, but the work involved in doing this 250 times did not seem unreasonable, and represented the outcome of Step 6. What did seem horrendous was the prospect of trying each of the million-odd combinations in turn to see whether it could be ruled out by ofie of the twelve females. The starting point for the last four steps in the thought process must have been the realization that the testing for the possibility of a female need not be repeated for each new key. It need only be done once. For each of the 60 wheel orders and each of the 17,576 starting positions of the three wheels it could be determined whether or not A, a 1-4 female, or a 2-5 female, or a 3-6 female could be produced. The yes-or-no answers could be permanently recorded for easy access. But how? In asking myself this question I was getting hot. Let us first consider the 1-4 females for a particular wheel order, say 413, and let us assume that the ring setting of each of the scrambler wheels is Z. We can represent all possible starting positions of the three wheels by individual squares in twenty-six rectangular matrices on separate thin cardboard sheets, each having twenty-six columns and twenty-six rows. Each sheet will represent one of the twenty-six positions of the left-hand wheel. The columns on each sheet represent positions of the middle wheel, the rows those of the right-hand wheel. A typical sheet is illustrated in Figure 4.1. It is called sheet K (1,4) for wheel order 413 because it records the possibility of 1-4 females occurring when the left-hand wheel is set to letter K with ring setting Z. A circle in a square of the matrix indicates that a 1-4 female could occur with the wheels at the corresponding starting position and ring settings Z'Z. Absence of a circle means that a 1-4 female cannot occur. Thus to test the 1-4 female KIE SPESNT we look at square IE, the intersection of column I and row E. We see that there is no circle, so we have ruled out the possibility that this female could occur with wheel order 413 and ring settings ZZZ. A minor achievement, perhaps, but we are at Step 7, and well on the way. Let us consider what the other squares of sheet K (1-4) will tell us about the female KIE SPESNT. The square JE for instance, which is immediately to the right of IE, tells us whether or not a 1-4 female can occur with starting position KJE and ring settings ZZZ. But this is the same as position KIE with ring settings ZYZ.* Similarly, squares HE, LE, ME, etc." will represent starting position KIE with ring settings ZXZ, ZVV'Z, ZVZ, etc. Going upward from square IE, we see that squares IF, IG, IH, II, etc." represent starting position KIE with ring settings ZZY, ZZX, ZZW, ZZV, etc. In fact the squares of sheet K (1,4) will tell us whether our female * Ib see this, imagine that the middle wheel is held in position J while the ring setting is moved back from Z to Y. The letter showing in the aperture will change from to I. And don't worry if this does not seem obvious. It took me quite a long time to figure out which way the letter in the aperture would go. it /> ABCDEFGHIJKLHNOP-QRSTUVWXYZ Z Y jeT S H f> 0 H M L K J 1 H 6 F E B A Z Y X W V T S f> o f o N H L K 1 H 6 F E D Coo oo oo oo oo oo ooo oo oo oo ooo ooo oo oo oo oo A BCDEFGHIJKLUNOPOKS T U V W X Y Z 6MIDDLE W H f L POSITIONS Figure 4.1 Sheet K(l,4) for Wheel Order 413. Left-hand wheel in position K. Ring Settings ZZZ, Circles indicate possibility of a 1-4 female. could or could not occur for all possible ring settings of the middle and right-hand wheels, with the left-hand wheel at ring setting /. We have now taken Step 8 and we are very near the denouement, the ultimate inspiration, the last piece in the jigsaw puzzle, or whatever. Let us look at Figure 4.2, which shows the area that would be occupied by a sheet like that of Figure 4.1, except that columns ABCDF are repeated on the right, and rows ABCDF are repeated at the top. Figure 4.2 shows the positions in different and separate sheets of five sub-matrices that would be used to investigate our rive females. Fl KIE SPESNT (1,4) P-2 LTS VBYQGY (3,6) P-3 EGP OHAOCM (1.4) PRYM XWNPWV (2,5) F-5 XXY ZDFJDA (2,5) At bottom left is a portion of sheet K (1,4) that would be used to study female F-l. It runs from column I to column M, from row K to row I. The letters in the squares indicate the ring settings of the middle and right-hand wheels that they represent, when used to test the possible occurrences of female Fl. To test the same twenty-five ring settings for female F-2 we would use sheet L (3-6), and the appropriate squares would be in a different part of the sheet, as is shown in Figure 4.2. Similarly the examination of females F-3, F-4, and F-5 would involve sheets K (1,4), R (2,5), and X (2,5). Figure 4.2 shows the positions on the sheets in which the twenty-five ring settings are represented. Note that for females F-4 and F-5 we would not have been able to represent the twenty-five ring settings neatly in a 5 x 5 sub-matrix if we had not repeated columns ABC and rows ABC. Note also that, having done this, we could stack the five sheets on top of each other in staggered positions so that the 5x5 sub-matrices of Figure 4.2 would be on top of one another. Then, if we used a punched hole to record the possibility of a female, light would only be able to penetrate all five sheets in positions representing ring settings that would permit the occurrence of all five females. We would be able to test ABCDEFGHIJKLMNOPQRS TUVWXYZABCOE f 0 c a a / Y X H> V\ F - 2 zr rr >r wr v r f L13-6) izrzxzizvz j| * I f. 3 zrrrxrwYVY zvrvxv*vvv 0 f/,4J zz yz xz wzvz zwrwxwwwvw f f.4 zr_rr_xr_*r_vjf_ u RIZ-5) z_z_rz xzwz_vz_ H L It J I H S F f 8 A ABCDEFSHIJKLMNOPORSTUVWXY Z A B C D figure 4.2 Portions of Five Sheets all twenty-five ring settings at once. If we extended our sheets to represent the whole alphabet twice, we would obtain square sub matrices representing all ring settings of the middle and right-hand wheels from Z to A, and we could stack the appropriate sheets so that relevant 26 X 26 sub-matrices would be on top of each other. We would be able to test all 676 ring settings in one fell swoop! This was it! What we had to do was now clear. Our sheets would have fifty-two columns and fifty-two rows, representing letters A to 7, twice over. On any particular sheet, say K (1-4) for wheel order 413, there will be two P columns and two Q rows, intersecting in four PQ squares. Holes will be punched in all four PQ squares if and only if a test has shown that a 1-4 female could occur with wheel order 413, ring settings ////, and starting position KPQ. The same will be done for all columns and rows. To use the sheets we will need a table with a window the size of a 26 X 26 matrix. This window will be illuminated from below, and sheets representing the twelve females will be stacked in such a manner that the appropriate 26 X 26 sub-matrices are above the window. The relative positions of the five 26 X 26 sub-matrices that we are talking about are shown schematically, and at half scale, in Figure 4.3. To deal with the first of our twelve females, KIK SPFSNT, we take sheet K (1,4) and place it on the table so that the 26 x 26 sub-matrix in the position that I have labeled F-l, K (1,4) will be exactly over the window. Then, to deal with F-2, we take another sheet, L (3,6), and lay it on the table so that its sub-matrix, in the position that I have labeled F-2, L (3,6) in Figure 4.3, will be exactly over the window. Similarly, to deal with the third female, t-3, we take a third sheet, E (1,4), and lay it on the table in a staggered position that will put its 26 X 26 sub-matrix over the window. And so on. The crucial point is that squares that are on top of each other will represent the same assumption of ring settings tor the middle and right-hand wheels. Thus, as I said before, we can test 26 X 26 = 676 ring settings in one fell swoop. As each sheet is placed on the stack, the absence of a hole will block the light in about half the places in which it has not already been blocked. If at any stage we find that no light gets through anywhere, we have ruled out the ring setting of the left-hand wheel that we are investigating, and we move on to another ring setting of that wheel. On AB CD CF GH I KL UN OP OK S T (JV W XYZ AB CO EF GH I KLM N 0 P 0 R ST UV WX Yl R * 1 "r G ** H I T o H I He D C L v * u P I 5 I ' k T J, I e O F N I s I t z F-3 (1,41 F-2 L13.SI F-5 X(2.S F-4 Rli.S) Fl K(lAB CD EF GH IJ KLMNOPQFISTUVWXYZtBCDeFGH IJ KLUNOPOKSTUVWXYZ MIDDLE WHEEL POSITIONS Figure 4.3 Th Relative Positions in their Sheets of Matrices that are to be Stacked Over an Illuminated Window in a table the other hand, if we do find that light shines through all twelve sheets at some particular place, this will be reported to Machine Room experts who will determine whether or not the indicated ring settings and wheel order could possibly be the correct ones. So ended Step 9. The final step in my thought process was the realization that I had hit on a thoroughly practical method of breaking some of the German Enigma traffic. At that time I don't think that I worried over much about the massive task of testing and punching. Nor did I think much about how the experts would do their job. It just seemed to me that the testing and punching could be accomplished in months, and that the sheet stacking and wizardry involved in breaking a key for which we had intercepted twelve females could be accomplished in less than half a day. In which case the decoding of Enigma traffic, which was our entire purpose, was only a matter of months away! Of the twenty-six stackings of selected sheets for any one wheel order, about four would result in light showing through somewhere, indicating that a particular set of ring settings had not been ruled out by the twelve females. This would be reported to the experts. When the correct combination, giving the key for the day, was discovered, the stackings would be discontinued and decoding of the day's traffic could begin. Thus, if we could do the testing and get the cardboard sheets punched, and could find twelve females on the Red or Blue key for a particular day, we could confidently expect to discover that key after an average of 780 stackings. And we could be sure of success provided that nothing had gone wrong. We could be foiled if we had inadvertently used a female that had been encoded on a different key, or if there had been an error in the transmission or interception of the critical nine letters. But we could hope to guard against such gremlins, and the prospect seemed extremely good. So in great excitement I hurried from School to Cottage to tell Dilly about it. Dilly was furious. What I was suggesting was precisely what he was already doing, and the necessary sheets were being punched under the direction of my Cambridge friend and colleague John Jeffreys. Dilly reminded me that I had been told to study discriminants and call signs not methods of breaking the Enigma. He sent me back to the School without telling me how the testing and sheet punching were being done. Indeed I never did know whether Jeffreys designed his sheets in the way that had occurred to me. It is entirely possible that there may have been errors in my thinking at that time, or in my attempt now, forty years later, to reconstruct my sequence of thought processes. But this did not matter then and does not matter now. What did matter was that a sheet-stacking process similar to the one I had suggested was being developed. From my association with Jeffreys before the war, I had a very high opinion of his abilities and felt quite sure that he would do a good job. In fact I was confident that the scheme was going to work. The sheet-stacking process would reduce the possible assumptions of wheel order and ring settings to a manageable number, indicated by positions in which light would shine through a stack. Each of these not-ruled-out assumptions would be called a "drop and would require crypt analytical examination by experts, whom I will call "wizards" with affectionate regard for their many skills. They did the actual breaking of the Enigma keys, but they also performed other functions that can hardly be regarded as crypt analytical ones. I was not told how it was proposed that the wizards would test drops and achieve breaks. One thing I did learn, however, was that the term "female," that I have used in my discussion, had become established in the Cottage. The reason, no doubt, was the analogy between a punched hole, through which light can shine, and a female socket into which a plug can be inserted to make an electrical connection. The reader may find it hard to believe that I had not been made aware of so important a development. Dilly was notorious for not telling anyone anything, though he often thought that he had done so. Moreover Bletchley Park as a whole must have been pretty chaotic in October 1939 and for many months after that. Huts were being built all over the place by the energetic, likable horse-riding Captain Faulkner. Sections were being formed and expanded to handle the many areas of work. The reader should not bother about * The term "drop" was probably derived from information retrieval systems using a stack of edge-punched cards. A search for desired information was conducted by sticking long pins through appropriate holes in the stack and shaking. Cards that might contain what was wanted would drop out. the purposes for which Huts 1, 2, 3, 4, etc." were built, or what went on in them. It would be too confusing. The numbers attached to the actual wooden structures kept being changed so that a section could continue to be known by the same hut number when it moved to new quarters. But with growing pains such as these, it was unlikely that there would be an organized flow of information. As for the strange reaction of Dilly Knox, an explanation of this outburst has recently occurred to me. It is possible that Dilly did not know about the Germans' double encipherment of their text settings until the meeting in the Pyry forest in July 1939 at which the Poles had revealed that they knew "everything about Enigma." It seems probable that, at the same meeting, Dilly had been told about the identical method of breaking the Enigma that would occur to me a few months later in the School. If this was so, Dilly would not have had a chance to think up the idea for himself. It would have been infuriating for him to find that a newcomer, and one whom he had kicked off his Cottage team, had come up with an idea he had not thought of. This first of my two breakthroughs proved to be a lead balloon-in that others, almost certainly the Poles, had already had the same idea. But it proved extremely important that the idea had come to me, too, and so early in the game. The planning for exploitation that I am about to describe was based on the confident expectation that we were going to be able to decode Enigma traffic, and I might not have known this for months if I had not thought of the idea myself. And we didn't have months to spare. The planning for exploitation might have started too late. It may seem odd, but I have no memory of being particularly irritated by Dilly's behavior, or of feeling frustrated by the fact that my idea was not new. Any such feelings must have been submerged in my delight at finding that the Cottage people believed, as I did, that the idea would work and that we were going to break into the Enigma traffic. Back in the School, however, as I mulled over the implications of this exciting development, I had the uneasy feeling that it was regarded in the Cottage as primarily a crypto logical success. No one seemed to be thinking about the activities outside Bletchley Park that would be needed to exploit the break to the full. I started to think about that problem on my own. In the autumn of 1939 we were in the period of the "Phony War," and nothing much was happening. Even in this period of relative quiet, however, we did not have enough operators to intercept all the German Enigma traffic. For only two types of traffic, the Red and the Blue, did we have a hope of sufficient volume to make the Jeffreys apparatus effective when it became available. Furthermore, as a result of my traffic analysis and my contact with Fllingworth at Chatham, I was very conscious of the fact that accuracy of interception was vital and presented a very real problem, because many of the intercepted signals were weak. Added to the problem of poor signal strength was the fact that several pairs of letters, such as L and V, are easily confused when transmitted in Morse. Our intercept operators often had to write down alternate letters to indicate their uncertainty about some of the Morse code characters they received. Kven more evident, perhaps, was the need for maximum coverage of the German radio nets so that, if we were able to break a given day's key, we could provide the intelligence people with decodes of as many messages as possible. And, since the war might suddenly become hot again, we must be ready for a lot more radio traffic, and probably more keys too. To exploit our technique to the full we would need many more intercept operators. It would also be highly desirable to establish intercept stations in widely separated locations, so that a signal too weak for reliable interception at one station might come in strongly at another. Furthermore, our chances of success would be enhanced if we could use only those sets of nine letters (indicator setting and female indicator) which could be confirmed by two or more intercept operators. To satisfy both crypt analytic and intelligence needs to the fullest extent possible with inevitably limited intercept resources, there would be a need for coordination from Bletchley Park. At that time I myself was the only coordinator, and self appointed at that, and I was dealing with only one intercept station, Chatham. The traffic analysis that I had initiated in the Schoolhouse would also need considerable expansion to meet the requirements of hot war and to handle the output of the more extensive interception capabilities I hoped for. The work would have to be highly accurate, because any attempt to break a particular key could be wrecked if a message enciphered on a different key were mistakenly used. Moreover the sheer volume of the registration effort would be far greater. The much larger number of German radio nets we could expect under hot war conditions would, I was sure, require many frequency time charts for each day's traffic, by contrast with the single chart that had sufficed for my own early traffic analysis. In addition to all this, it was clear that we would have to operate around the clock. With my staff of two young women I could not possibly cope with the needs of traffic analysis and coordination of interception. On the crypt analytic side of the picture the situation was equally unsatisfactory. It was obvious that we would need a considerable staff to carry out the sheet-stacking routine within an acceptable time after a sufficient number of females on the same key had been intercepted with adequate confirmation. We would also need a staff of Enigma experts who would work on the drops thrown up by the routine sheet stacking. Starting every midnight, when keys were changed, this staff of wizards would keep a close eye on the traffic registers coming in by teleprinter from the intercept stations, and on expanded traffic analysis charts similar to those that I had brought into use during the low-traffic period. They would decide when their collection of females was good enough to start a sheet-stacking attempt to break a particular key. At the time I am talking about we had very few people to staff the sheet-stacking operation, and very few qualified Enigma experts. Around-the-clock activity would have been impossible. Finally, on the assumption that we were going to achieve breaks, we would need a decoding room equipped with British cipher machines, which would have to be adapted to operate like German Enigma machines We would need enough decoding personnel on duty at any time of the day or night to start decoding all intercepted messages on a broken key as soon as the break could be achieved, and to get the decodes to intelligence people within an acceptably short time. Remember that the same key would usually have been used for the encoding of more than a hundred messages, all of which would be decodable as soon as we had discovered the key. Fortunately, though I did not know it at the time, the British Type-X cipher machines that were already in production could be adapted without much trouble. Thus there was no need to design, develop, and manufacture duplicates of the German Enigma. With considerations such as these in mind, I formulated an organizational plan and took it to Commander Edward Travis, deputy director, under Alastair Denniston, of all Bletchley Park activities. Travis was a broad-shouldered man of heavy build. His spectacles gave him a curious expression, which I found disconcerting until I knew him better. I think they were of a kind that was unusual in those days, a narrow lens for reading with a flat-topped metal rim over which he could scrutinize distant objects and people to whom he talked. I explained to him that, as soon as Jeffreys and his team had completed the punching of sheets in the Cottage, we would have a golden opportunity that we were very ill-equipped to exploit. We had reason to believe that we could break into a significant part of the German Enigma traffic, even during the quiescent period of the Phony War. If we were to exploit our opportunity under hot war conditions, a lot had to be done in a hurry. Within the Bletchley Park complex we had to build up a completely new twenty-four hour operation of five closely coordinated departments: a Registration Room to perform a continuous traffic analysis of Enigma messages based on traffic registers received by teleprinter from the intercept stations; an Intercept Control Room, which would keep in continuous touch with the intercept stations, helping them concentrate on the most valuable traffic; a Machine Room handling the crypt analytic aspects in close collaboration with the Registration Room and the Intercept Control Room; a Sheet-Stacking Room, which would be called into action by the Machine Room whenever the traffic of a particular day on a particular key merited an attempt at a break; and finally a Decoding Room to handle the messages on any key that might be broken. We had not yet even begun to acquire any such capabilities within our organization. Outside Bletchley Park we were going to need a major expansion of intercept facilities. In particular, in view of the possibility that the Army station at Chatham might be required to concentrate on German army traffic, it seemed very important to have a station operated by the Royal Air Force and dedicated to the interception of enemy air force Enigma traffic. Travis' response was all I could have hoped for. He immediately saw the urgent need for a buildup, approved the plan in full--and wasted no time. He quickly obtained official agreement to the establishment of a new section in the Bletchley Park complex to handle this possibility of Enigma breaks on an interservice basis. Dennis ton, the head of Bletchley Park; Tiltman, head of the Army section; and Cooper, head of the Air Force section, all gave their full support to the plan. It was a time for me of great excitement and great challenge, and in winning approval of this plan at so early a date I probably made my biggest single contribution to the war effort. I had come to Bletchley Park as a green recruit only a short time before and yet, because of the work that Dilly Knox had asked me to do in the School, I had had the luck to stumble on some important aspects of the command communications system that the Germans had developed for their blitzkrieg. I had had a vision of how the Jeffreys apparatus, then being developed in the Cottage under Dilly's direction, could be fully exploited. And I had been able to convince my superiors that a radically new situation called for an entirely new organization. That the necessary organization could be built up speedily, and was ready for action when the Jeffreys apparatus became available, was tremendously important--how important will become apparent as we go on. The second breakthrough that I was able to make before the war was three months old had to do with the design of electromechanical machines called "bombes," which played an important part in the Hut 6 story. The name "bombe" seems to have been attached by the Poles to an earlier concept, but it was also applied to the machines that we actually used. The term "bombe" is simply the French for "bomb"; the connection with our machine is not clear to me, but it may have had to do with the idea of a mechanism that will go on ticking until it reaches a combination that will cause it to produce an output--in our case not an explosive one. Our bombes were said to make a noise like a battery of knitting needles. One of the basic ideas behind our bombe was to move a battery of Enigma scrambler units in synchronism through all possible positions. A group of Polish experts had come to Bletchley and worked with Knox and Turing on the concept of such a machine for a time before going to France. Now Turing was working on the exploitation of their machine; my contribution to its development was to add the principle of the "diagonal board," which greatly increased its power. The fact that the idea came to me so early proved to be extremely fortunate, and again, for those readers who may be interested, I will attempt to reconstruct my mental processes. I was certainly influenced by those precious decodes that Josh Cooper had given me when I started work in the School. As I explained in Chapter 3, I had noticed that stereotyped addresses and signatures appeared in the texts of this small sample of decodes. Now I began to wonder if this gave us the possibility of using these as cribs to break Enigma keys. The notion of a "crib" may need a little explanation. Several I definitions of the word will be found in the dictionary, including "a 1 device used for cheating in an examination Cryptologically speaking, however, one has a "crib" to a cipher text if one can guess the clear text from which some specific portion of the cipher text was obtained. As my analysis of the Enigma traffic began to reveal certain routine characteristics in the preambles of individual messages, I realized that, if we could somehow determine to whom they were addressed, or by whom they had been sent, we might be able to guess a portion of the clear text either at the beginning or at the end of each of the messages, and so have cribs. To illustrate what I had in mind, let us consider the imaginary crib shown at the top of Figure 4.4. Somehow it has been possible to guess that the thirty-one letters CQNZP through UFLNZ at the start of a message have been obtained by enciphering the plain text address TO THE PRESIDENT OF THE UNITED STATES. The numbers 1 through 31 refer to the successive positions of the Enigma's scrambler unit at which the letters of the plain text were encoded. These 31 positions are consecutive ones somewhere in the scrambler cycle of 17,576 positions.t I have explained that the idea of the Jeffreys sheets rested on the fact that it was not always possible for the same pair of letters to be * "And they found in his palms ... what is common in palms, namely dates." t Remember that a scrambler position involves three wheel positions, and that each wheel has twenty-six positions, of which one is a turnover position. In the scrambler cycle the right-hand wheel always advances one notch with the encoding of each letter. The middle wheel advances only when the right-hand wheel is in its turnover position, which will occur once in every twenty-six consecutive scrambler positions. The left-hand wheel advances only when both the right-hand and middle wheels are in their turnover positions, which will occur once in every 676 consecutive scrambler positions. Crib C 0 N Z P V L I i P E U I T oT H P if e S I D N 9 10 /1 /2 13 KTEDCGLOV TOFTHEUH! H 15 If If IS 19 20 21 22 W V G T T D S 23 11 2S 26 U F L N Z T A T E S 27 2S 19 30 31 Diagrams I Positions 3 to 22 Three Loops Seven Letters Nine Links 2 Positions I to 13 One Loop Nine Letters Nine Lints ii rr s m 13 m 3 Positions 14 to 26 No Loop Eleven Letters Ten Links Figure 4,4 Diagrams Derived From a Crib the encodes of each other in scrambler positions three places apart in the cycle of machine positions. The principle of the bombe on which Turing was working may be regarded as a more complex version of the same idea. Turing, however, was now concerned with patterns formed by several letter pairings of a crib, rather than with the occurrence of females, and by the fact that patterns of certain types could not always be produced by a set of scramblers in the relative positions in the machine cycle indicated by the places in the crib in which the letter pairings occurred. Turing intended to use a battery of Enigma scramblers set at relative positions corresponding to a set of letter pairings in a crib that he would select for study. The whole battery of scramblers would be moved in synchronism through all 26 X 26 X 26 positions of each scrambler. In each position of the battery an automatic test would be applied to determine whether his selected set of letter pairings could occur. In this way, as in the case of stacking Jeffreys sheets, he could hope to achieve a major reduction in the possibilities requiring serious crypt analytical study. On examining the crib in Figure 4.4 we see that the encode of E in position 5 is P, that the encode of E in position 8 is I, and that the encode of I in position 10 is P, the letter from which we started. This closure of the loop--P to E to I to P--is an example of the type of pattern that Turing intended to use. Figure 4.4 shows three diagrams that can be derived from the letter pairings of the crib. In the first diagram the loop that I have just discussed is represented on the left. A line numbered 8 links two boxes containing letters E and I. This is a convenient way of indicating that the letters E and I are the encodes of each other through a scrambler in position 8 of the sequence 1 through 31; in fact the line or link between letters E and I represents this scrambler. Similarly the letters E and P are the encodes of each other in position 5, letters I and P in position 10, and so on. The diagram involves three closed loops, seven letters, and nine links indicating scramblers in positions ranging from 3 to 22. The bombe, as it was being developed by Turing from Polish ideas, depended entirely on being able to construct a diagram, similar to the first diagram of Figure 4.4, containing three closed loops. The idea that flashed into my mind in the School was that, by interconnecting the scramblers in a completely new way, one could increase the effectiveness of the automatic test by a very large factor. I saw that we need not depend on obtaining three closed loops. Instead a bombe could be constructed that would make use of configurations such as those shown in the second and third diagrams of Figure 4.4, involving one loop or no loops at all. When this new method of interconnecting the scramblers of a bombe came to me, I couldn't believe it. But I sat down with a few colored pencils, drew a simple wiring diagram, and convinced myself that the idea would indeed work. Armed with this diagram I hurried once again to the Cottage, this time to talk to Turing. On this occasion I had a better reception than I had received from Dilly. Turing was incredulous at first, as I had been, but when he had studied my diagram he agreed that the idea would work, and became as excited about it as I was. He agreed that the improvement over the type of bombe that he had been considering was spectacular. With Turing's support it was not difficult to convince Travis that the development of the improved bombe was urgently important for handling the Enigma traffic of the German army and air force, and the naval Enigma traffic as well. Again Travis wasted no time. Although I did not realize it then, Harold "Doc" Keen of the British Tabulating Machine Company, which was associated with IBM in America, must already have started on the design of the Turing bombe. Travis asked me to work closely with Keen, and this started what was to become both a close collaboration and a close friendship. Keen soon grasped the new idea and set to work on two prototype bombes incorporating the diagonal board. The design proved to be extremely flexible, quite adaptable to changes as they became necessary. This was no doubt due to Keen's long experience with punched-card equipment, which was always being called upon to perform tasks that no one had thought of when specifications were being prepared. As is often the case with revolutionary ideas, mine, after it had occurred to me, proved extremely simple. It involved interconnecting a battery of scrambler units through what came to be known as a diagonal board. But practical realization in the form of a workable bombe required several other ideas Furthermore, the exploitation of the bombes in Hut 6 called for a kind of crypt analytical wizardry that had not yet been developed. Thus the bombe in its final realization was not the brainchild of any one person; and we thought very little, in those hectic days, of who should take credit for what. It will become clear that the purpose of a bombe run was to determine whether, for one of the 60 possible wheel orders, a certain pattern of letter pairings derived from a crib could occur with a set of scramblers in relative positions corresponding to the positions of the letter pairings in the crib. The automatic test, of which I have talked, involved trying to prove that no combination of stecker pairings could allow the pattern to occur. It was accomplished in one fell swoop by electric current flowing to and fro between the battery of scramblers and the diagonal board. Thus the vast number of stecker combinations (around 200 trillion) on which the German cryptographers probably pinned their faith was of no avail against the bombe. In essence our automatic test could examine all possible stecker combinations in less than a thousandth of a second. As in the case of the sheet-stacking idea, I believe it to be very significant that an idea as important as that of the diagonal board came to a complete novice within three months of his exposure to the characteristics and operating procedures of the Enigma. I do not know what types of expertise contributed to the design of the German Enigma system that was in use at the outbreak of World War II, but it must have been a carefully considered design. Yet I, a completely inexperienced recruit from Cambridge University, having been told how the system worked, very soon came up with two complementary methods of breaking the Enigma traffic. What do we learn from this? It seems to imply that professional cryptographers, however good they are, may fail to see how a mind with a different background might find a way of defeating them. And the lesson seems to be that, even against the threat of pure cryptanalysis--by no means the greatest threat to the security of our communications in the 1980s and 1990s--it would be dangerous to depend only on the expertise of professional cryptologists * In the Appendix "The Bombe with a Diagonal Board" I will go into greater detail, explaining the use of double-ended scramblers, how my bombe differed from the Turing bombe, and why the introduction of the diagonal board was so important. who operate under the protective cover of super-high-level secrecy. Whatever systems these professionals may design should be subjected, under stringent conditions of secrecy, to close scrutiny by non professionals who may possibly come up with new ideas. For example, as I see it, the security of our cryptographic systems seems too often to be reckoned by the number of possibilities that an enemy computer would have to examine one by one. In the bombe, houever, particularly with my addition of the diagonal board, we were able to examine an enormous number of possibilities in one fell swoop. Thus the mere number of possibilities proved to be invalid as a measure of system security. It could happen again. cryptological threats to the security of our future communications systems might also be detected by someone other than the experts who devise the cryptographic systems. This possibility svill become more apparent later on, as we see how the security of the German Enigma was undermined by the errors of its users rather than by those of its designers. Should we not subject our cryptographic systems to close scrutiny by many different minds before they are accepted, and to thorough monitoring when they are in use? Furthermore, as I will be better able to argue toward the end of this book, we should make sure that the noncryptanalytic threats to the security and survivability of our communications are very thoroughly examined from many points of view. In failure to do so lies the greatest danger. Thus ends my discussion of the first three months. We can now move on to the remaining nine months of the first year, up to the time when Hitler abandoned his plans to invade England and decided to deploy his forces to other theaters. Early Days December 1939 to May 1940 Recruiting for Hut 6 went fast. Travis produced a scientist, John Colman, to take charge of the Intercept Control Room, which was to maintain close contact with the intercept stations. Colman was soon joined by another scientist, George Crawford, a former schoolmate of mine at Marl borough College. Travis also persuaded London banks to send us some of their brightest young men to handle the continuous interchange of information with intercept stations. Thus, very soon, we had an intercept control team large enough to operate round the clock. They quickly established close and very friendly relations with the duty officers at Chatham. For my part, I quite shamelessly recruited friends and former students. Stuart Mimer-Barry had been in my year at Trinity College, Cambridge, studying classics while I studied mathematics. He was not enjoying being a stockbroker, and was persuaded to join me at Bletchley Park. He arrived around January 1940, when the Hut 6 organization was about thirty strong, bringing with him the largest pipes I have ever seen smoked. Stuart in turn recruited his friend, Hugh Alexander, who had been a mathematician at Kings College, Cambridge, and was then Director of Research in the John Lew is Partnership, a large group of department stores. They brought us unusual distinction in chess: Alexander was the British Chess Champion, while Mimer-Barry had often played for England and was chess correspondent for the London Times. Two other friends from Cambridge undergraduate days joined me later: Harold Kletcher, top mathematician in the college entrance scholarship exams in 1925, and Houston Wallace, who had become a Chartered Accountant. Another mathematical friend from Cambridge, Dennis Babbage, had come to Bletchley in the early days as an Army officer and, after working with Dilly Knox in the Cottage, became one of the original wizards of the Hut 6 team. He and I had been members of a group of geometers known as Professor Baker's "Tea Party," who met once a week to discuss the areas of research in which we were all interested. Before the war, when I was College Lecturer in Mathematics at Sidney Sussex College, Cambridge, John Jeffreys had been a Research Fellow in Mathematics at Downing College. For two years or so he had been helping me supervise the studies of my math students. We had become close friends, and were both considered pretty good math supervisors. Consequently we found it not too difficult to persuade some of our best students to join us, including three very good young mathematicians, David Rees, John Herivel, and John Chamberlain. Other nonmathematical Sidney men, Howard Smith and Asa Briggs, joined the team and distinguished themselves. Nor did I stop at the university level in my forays for bright young men. Before going to Trinity College, Cambridge, in 1925, I had studied at Marlborough College under an outstanding math teacher, A. Robson. I had kept in touch with him, and we had spent summer vacations together in the English Lake District collaborating on mathematics textbooks. So I now appealed to him to send me his best young mathematicians. This produced another influx of very good people, including John Manisty. At the same time other men were being recruited, mostly, I believe, on the basis of personal contacts such as those I was exploiting. Probably this was inevitable. Recruits could not be told what kind of job they were going to do, so they needed assurance from someone they could trust that what they were being asked to do was really worthwhile. Aitken, the chess champion of Scotland, was one man acquired by this means; another was David Gaunt, a very able young classics scholar from Cheltenham College. This kind of piracy was to be curtailed in 1941. The government decided that the use of the best young brains in the country should be regulated. C. P. Snow, of Christs College, Cambridge, whom I had known before the war (and now a celebrated author), was put in charge of allocation of all scientists and mathematicians, and from then on I had to recruit my male staff through him. I could not tell him what we were doing in Hut 6, and I do not know to this day what, if anything, he had been told about our operation. Nevertheless he was extremely cooperative, and sent me some very good men Perhaps it was another fortunate accident that Snow happened to know me fairly well and believed what I said about the importance of our work. One problem that arose later should perhaps be mentioned now. Some of the young men who were sent to Hut 6 because of their brains found themselves trapped there by the demands of security. They longed for active service in the air force, the navy, or the army, but they knew too much about our success with the Enigma for their capture by the enemy to be risked. They were doing an exhausting job, and it was obviously helping the war effort, but many of them longed to play an active part in the fighting. There was, too, the inevitable feeling that not being at the front was somehow dishonorable; one young man received a scathing letter from his old headmaster accusing him of being a disgrace to his school. Recruitment of young women went on even more rapidly than that of men. We needed more of them to staff the Registration Room, the Sheet-Stacking Room, and the Decoding Room. As with the men, I believe that the early recruiting was largely on a personal-acquaintance basis, but with the whole of Bletchley Park looking for qualified women, we got a great many recruits of high caliber. I was very little involved in the recruitment of the important female component of the Hut 6 staff, but I may perhaps venture one anecdote. I had been married in 1937, and my wife, Katharine, and son, Nick, were still in Cambridge, so I returned there as often as I could. While there I hunted around for staff for Hut 6. I recruited June Canney, the daughter of a doctor, and in due course drove her from Cambridge to Bletchley. On the way she remarked that she had been wondering why I had asked at our initial interview if she were colorblind. When she arrived in Hut 6 she could see for herself: We depended on color discrimination. My original key colors, red, blue, green, brown, and orange, had been added to. Indeed, before long the British supply of colored pencils ran out, and we had to appeal to America for the supplies we needed for our Registration Room. It may seem absurd, but those colors really helped, and we could never think of an adequate substitute. Men from the Machine Room and the Intercept Control Room would keep wandering into the Registration Room to examine the traffic charts, and the identification of keys by color was a great help to them. June Canney's color vision was fine, but she didn't stay long in the Registration Room. She became my secretary, and did much to help me in running Hut 6 and making it a happy organization as well as an efficient one. In the very early days, of course, we had nothing to decode, but we managed to acquire a staff for the Decoding Room who had learned to operate our modified Type-X machines as if they were German Enigmas. Thus while we were waiting for the completion of the "Jeffreys apparatus," we assembled a strong supporting staff, and developed an organization that would be ready to go into action as soon as code breaking became possible. In fact the punching of the Jeffreys sheets was near completion, so we did not have much time, and the buildup of our intercept control capabilities was particularly urgent. When Travis brought in Colman he had hoped that, even before we could start breaking the Cierman Enigma traffic, a scientific analysis of message volume might 'yield valuable intelligence. Colman's most important function at that time was to establish the closest possible cooperation with intercept stations, but he gave Travis' idea a good try. He concentrated on the Red and Blue traffic, because at that time no other Enigma keys were offering a volume of messages that would give the Jeffreys apparatus a chance. Colman and his staff found out how to use naval intercept facilities for direction finding (I)/E). From close contact with Chatham they were able to get the naval D/F stations onto specific Enigma messages, with the intention of determining the locations of the transmitting stations. By now Colman was also getting help from the R.A.F intercept station at Cheadle. He found that, whereas transmitters of Red traffic were pretty widespread, those of the Blue traffic were concentrated in the northwest of Germany, an area from which military operations might be expected to originate. Consequently Colman concentrated his attention on Blue traffic, analyzing its fluctuations, which were considerable. On several occasions an unusual peak in Blue caused us to believe that an outbreak of hot war was imminent. We got quite excited about it. However, when the Jeffreys apparatus became available and we finally broke the Blue traffic, we found that it was nothing more than training exercises. The Germans, with their phenomenal planning, foresight, and thoroughness, were busily training the Enigma-equipped signals detachments that were to accompany German ground and air forces on their fast-moving rampages throughout Europe and northern Africa. Of course, when we discovered that the Blue traffic was only an exercise for the German signals organization, we concentrated on the Red, which at that time was genuine Luftwaffe traffic. As we were to discover, the Red traffic became the operational traffic of the German ground and air forces as soon as the war became hot. During the Phony War, the combat units of the German army had no need for radio communications. They could, and no doubt did, use established telephone-type communications. But as soon as they moved out of their homeland, the German army needed radio communications among their own mobile units and with those of their air force. They used the Red key, which was intended for this purpose. The fact that the Red key was being used for army--air coordination must have been apparent during Hitler's invasion of Norway. It certainly became obvious during the invasion of France. Yet the myth that Red was an air force key has persisted; even Ronald Lewin in his excellent Ultra Goes to War still accepts it. But I am guilty too. The myth was not dispelled from my own mind until I argued with Ronald about the tactical army-air messages that we decoded during the Battle of France--messages that gave a picture of what was going on at the combat level, rather than at the higher command level. In the picture of Guderian's command vehicle shown in Lewin's book and reproduced here (Figure 2.1), there is an Enigma that must have been used for messages on the Red key. But neither Guderian himself, nor his cipher clerk and Enigma operator, were air force personnel. Furthermore, many of the crucial messages at the higher command level that are discussed by Winter both am and Lewin were quite definitely army messages (for example, orders from the German Commander in Chief, von Brauchitsch, to van Bock's Army Group B and to von Kluge's 4th Army). Surely an Enigma key that was used to encode such messages cannot be regarded as an air force key. But let us get back to our "Early Days." In anticipation of the advent of the Jeffreys apparatus, it was necessary to introduce a three-shift schedule in other sections as well as in Colman's Intercept Control Room to cover the twenty four hours of each day. At the time, this was a novelty in Bletchley Park, and in addition to organizational problems, it created an unexpected problem concerning "proper" behavior. In the early days of three-shift operation, my night staff would consist of one man in the Intercept Control Room with one girl assistant, and perhaps two or three girls in the Registration Room. This situation was considered highly improper by the Eoreign Office administration of Bletchley Park! Fortunately, the problem did not come to a head in rime to hamper our operation; we were able to keep going until the advent of the Jeffreys apparatus called for more people on night shift and therefore, perhaps, safety in numbers. We did not have to wait long. While I was busy building up the Hut 6 organization and working with Doc Keen on the functional design of the bombes, John Jeffreys and his team were patiently performing a monumental task in the Cottage. I was not a participant myself, but I have already given a broad picture of what must have been involved. I do know that the Cottage people had a very simple manually operated testing machine that would determine in a matter of seconds whether a particular assumption of wheel order and starting position could or could not produce a 1-4 female, a 2-5 female, or a 3-6 female. Though each test was simple, there were 3 x 17,576 tests to be made for each of the 60 possible wheel orders, and the result of each test had to be recorded by punching or not punching a hole in one of the Jeffreys sheets. It was a great achievement by the Cottage staff. Once the sheets were punched, however, we had a permanent record that could be applied in the breaking of any daily Enigma key. As I have already explained, the actual breaking of a key would depend on skilled work by the experts in the Machine Room. The sheet-stacking did no more than reduce the number of possibilities that had to be investigated. In the School days I did not attack this problem seriously. I was not told hum Jeffreys proposed to solve it. When Hut 6 became operational, Jeffreys was in charge of Sheet Stacking and Machine Room activities, while I worried about Registration, Intercept Control, Decoding, and relations with the intelligence people in Hut 3. I can see how the breaks might have been achieved, but will not discuss the matter. The important fact is that the breaks were actually achieved by John Jeffreys and his team. Thanks to our early planning and recruiting, and the establishment of twenty-four-hour shift operations, we achieved a workable routine as soon as the Jeffreys sheets were ready for use. From midnight on, three copies of the teleprinted sheets of the traffic register from Chatham were studied in the Registration Room, in the Intercept Control Room, and in the Machine Room. The staff of the Registration Room would start the day's traffic sheets, determining the sets of discriminants that identified the various keys. The Intercept Control Room Staff, using the charts in the Registration Room, would be trying to concentrate our interception resources on the traffic that we wanted most. The Machine Room personnel would decide when enough suitably confirmed females had been found on one key to justify an attempt to break that key for the day. The Red key, rather than the Blue, would have priority. Once the Machine Room had found a satisfactory set of Red females, the people in the Sheet-Stacking Room would begin their routine. From time to time they would obtain a "drop," which would be tested by the Machine Room staff. Most of these drops would prove false; indeed, the Machine Room decision to attempt a break was based on a calculation of the number of false drops that could be expected from the data it had given to the Sheet-Stacking Room. With luck the sheet stacking would at last produce a true drop that would enable the Machine Room to determine the Red key for the day. When this happened, a shout of triumph would be heard, and the Decoding Room staff would be activated. As an indication of how long all this might take, the shout of triumph for the key of a day's Red traffic would usually be heard in the early hours of the next morning. Soon after the Jeffreys sheets came into use, I invited Commander Ellingworth of Chatham to spend a night with me. In the evening he watched what was going on in the Sheet-Stacking Room, and then we settled down in my office to discuss the operation in general. At about one in the morning we heard the shout, and hurried along the corridor to the Decoding Room. Soon Killing worth was seeing the first decodes of the previous day's messages as they came out of the Type-X machines That night together in Hut 6 was a great help to us both. Ellingworth got a clear picture of our activities, and we were able to talk realistically about the problems of handling Inigma traffic. For example, there was the question of message texts. The teleprinted tnifhc register, which had been initiated after my first visit to ( Chatham, contained all the information that we could use in breaking an Inigma key--preamble, with discriminant and indicator setting, and the first two text groups containing the twice enciphered text setting, or indicator. Once the key was broken, however, we needed the message texts for decoding. These texts were recorded by intercept operators on standard forms and bundled together for dispatch to us. Until we started breaking, nobody had had any use tor these bundles, but now there was suddenly an urgent need to get them to Bletchley Park in a hurry. During that night Kllingworth and I discussed how the bundles of messages could be transported to Hut 6 in time for instant decoding whenever a key was broken. Our teleprinter facilities were utterly inadequate to handle so much volume, so we had to continue to depend on dispatch riders. The debt that we owe to these riders, who faced all kinds of weather on their motorcycles, has never, to my knowledge, been properly recognized. There was also the question of giving certain messages priority, and for that we did use the teleprinter. The problem became a great deal more complicated later on, but I believe that the idea of the welchman Special" was born that night. Even in those early days it had become apparent that we in Hut 6, by studying the teleprinted traffic register, could often identify messages that would be of particular interest from the point of view of the intelligence experts In such cases, Kllingworth and I agreed, dolman's people would telephone the intercept station to ask that the anticipated For example, messages on a particular radio net. appearing each Jay v ith the same time of origin. might have been found, by previous decodes, to be daily reports or daily commands of considerable significance. messages be given special treatment, whereupon the Chatham people would immediately transmit them by teleprinter. The term "Welchman Special" was introduced at Chatham shortly after Ellingworth's visit to Hut 6. The idea of "Welchman Specials" was to be helpful to us in many ways in the later phases of the Hut 6 activities. However, let us first think of the importance of the idea for intelligence. Wonder has been expressed at the speed with which information from Hut 6 decodes could sometimes reach Allied commanders. It has even been said that on some occasions these commanders would actually receive information before the German addressees did. Readers of Winterbotham's The Ultra Secret will know that one of his tasks in World War II was to pass to Winston Churchill those decoded messages that Churchill would regard as highly significant. It seems probable that these messages would usually have been to or from high-level German commanders, in which case they would have been messages that we would have identified as potentially important from our analysis of the teleprinted traffic registers. Thus a high proportion of the decodes mentioned in Winterbotham's book as having been passed to Churchill had reached him earlier than would otherwise have been possible because they had been handled as "Welchman Specials." The same applies to many of the sensational decodes whose sanitized content is now known to have been passed to commanders in the field After September 1940 the "Welchman Special" technique was used to speed up the breaking of keys as well as the decoding of important messages; but much needs to be explained before this becomes fully intelligible. The name "Welchman Special" was coined at Chatham to denote messages that were given special handling as a result of descriptions provided in advance by Hut 6. The use of my name resulted from the happy personal relationship that I had established with Chatham in October 1939. But the exploitation of this procedure was always a routine matter, handled by the Hut 6 staff. A more appropriate name would have been "Bletchley * The word "sanitized" may need explanation. If the exact translation of a decode had been sent, and had somehow been read by the enemy, it might have revealed the fact that the information could only have been obtained by decoding, thus revealing our highly secret capability. Consequently the information derived from a decode had to be presented accurately, but in a manner that would not obviously relate it to an Enigma message. Special," or "Hut 6 Special." What is of significance is that this important procedure was originated when Commander Ellingworth of Chatham visited Hut 6 early in 1940. I was continually bothered by the possibility that this "welchman Special" procedure might be a dangerous breach of security. If we in Bletchley Park could tell an intercept station that a particular message was of special importance, the clear implication was that we could decipher that message. My concern was not so much that the telephone wires might be tapped--one worried less about that in those days--but that the special attention to certain messages might reveal to the intercept operators that the traffic was being broken. I discovered, however, that the intercept operators believed that all their intercepts were decoded. The protection of our secret depended on the fact that the intercept operators had sworn not to tell anyone anything about what they were doing, not even the fact that they were intercepting enemy radio signals. In those early days of Hut 6 some important related activities were beginning to develop elsewhere. All will be discussed more fully in later chapters, but they bear some mention now as part of the context of our early operations. One such activity was the now famous one known as Hut 3. Even before the Jeffreys apparatus became available it was confidently anticipated that Hut 6 would start breaking Enigma traffic. Therefore it was necessary to have intelligence analysts, with an intimate understanding of the German language, ready to begin working on our decodes Arrangements were made for a small Foreign Office intelligence group to occupy a tiny hut between Hut 6 and the main building, then designated as Hut 3. This group, headed by Commander Malcolm Saunders and with an early membership that included two Cambridge dons, E. L. Lucas and Harold knight, went into action as soon as Hut 6 started to produce. In the early days Winterbotham introduced an air intelligence section under Wing Commander Humphreys. My early contacts with intelligence, however, were with Saunders and his group of civilians. I understand that there was an organizational change early in 1942, after which Commander Saunders was definitely in charge for a * I do not believe that anyone in Hut 6 had more than a smattering of German time. At some point the intelligence effort moved to a larger hut, adjacent to Hut 6, which was again named Hut 3. Very confusing! As Winterbotham has noted, the content of the early decodes was not very exciting, because it was still a "phony" war. The German ground forces were not on the move, and much of their traffic was probably going by landlines. However, the mere fact that we had begun to "crack" Enigma keys was realized to be of great significance. As Winterbotham recalls, his boss, Colonel Menzies, soon to become Chief of the British Secret Service, was enthusiastic. Menzies and Winterbotham put a great deal of thought into plans for the dissemination of Ultra intelligence. This was not my concern. What was of immense importance to the future of Hut 6 was the fact that, thanks very largely to Winterbotham, official enthusiasm was sufficient to convince the British Army and Royal Air Force to establish a joint intelligence unit at Bletchley Park. Thus began an extremely close relationship. Throughout the war the Hut 6/Hut 3 complex operated as an interservice organization under Foreign Office administration. For the moment I am concerned with what happened in the first year, but let me mention that much later in the war Saunders was asked by Travis to undertake another important pioneering job: making all necessary preparations for the operation of the many bombes that were to come. In this job Saunders was assisted by Harold Fletcher of Hut 6. Another important activity was starting up in London in the Military Intelligence organization, MIS. A small group of analysts was beginning to study the German radio nets that were being intercepted at Chatham. Their objectives were very different from mine: While I was concerned with our hopes for breaking the Enigma traffic, they had started with the assumption that the Enigma traffic was unbreakable. Their objective instead was to derive intelligence from a detailed study of the operation of the German radio nets. I will return to this matter in Chapter 9; for now it is enough to say that, to avoid simultaneous transmissions and to counter the problem of drifting frequencies, German radio nets depended for their proper functioning on continuous coded conversation between the control station and the subscriber stations. This chitchat permitted frequency settings to be constantly readjusted. Our intercept operators logged this German chitchat, and the thew MI8 group in London proposed to study all the logs of all the Chatham intercept operators. By doing so they had a good chance of discovering all the call signs used on each net on a particular day. As one of its objectives, this log-reading organization set out bravely on the task of breaking the thew annual call sign books of the German army and air force. These books contained sequences of call signs one of which would be assigned to each unit that uould originate or receive radio transmissions. Each sequence provided a different call sign for every day in the year; thus a full year of log reading would be needed before call sign recurrences would help to identify enemy units by their radio transmissions. In the early days I saw very little of this log-reading group. By June 1941 my principal contact was the author Fdward Crankshaw. Subsequent expansion was headed by I lamish BlairCunynghame, and the senior liaison man at Bletchlcy Park was a Colonel Harry Sayer. My close relationship with these three men helped resolve problems that might have impeded the development of the Hut 6/Hut 3 partnership. Thus early in 1940 the plan that had originated in the School was being implemented, and important relationships with associated efforts outside Hut 6 were being developed. The sheets that had been designed and produced in the Cottage were working well, and when the war became hot again, with Hitler's invasion of Denmark and Norway on April 9, 1940, we were able to intercept and decode a good deal of the German operational command traffic. We were in close touch with the people in Hut 3 who were translating our decodes, and the messages we read were disheartening. They gave a picture of the efficiency of the small German forces involved, and of the inept coordination of the British efforts. I remember being particularly impressed by the speed with which the Germans established their battlefield communications. Apparently the well trained Enigma-equipped signals teams accompanied the first assault troops wherever they went. During the Battle of France in May 1940 the decoded messages were equally disheartening. We were reading a great deal of the operational traffic being exchanged among the German field commanders, as well as their reports to the higher command and the directives they received. For example, when any one of their ar mo red units was held up by an Allied defensive position, we would probably decode an Enigma message from the unit's commander requesting air support. A little later we would hear that an attack by dive-bombing Stukas had been effective, and that the armored unit had resumed its advance. At intervals all the major Panzer commanders would report their progress and their assessments of Allied capabilities. We in Hut 6 would pass the decoded messages to the intelligence officers of Hut 3, providing them with a clear overall picture of events on the battlefield, as well as a detailed picture of each sector. The intelligence derived from this flood of Hut 6 decodes could have had little if any effect on the course of the Battle of France--for one thing, as part of the overall Allied unpreparedness, the use of such a prolific source of intelligence had not been worked out in advance--yet it seems to me that in this campaign, Hut 6 Ultra achieved its first major success, albeit a negative one. Our decodes must have given early warning that the military situation was utterly hopeless. This mass of combat intelligence can hardly have failed to speed the organization of the extraordinary fleet of miscellaneous boats that brought so many men back from the beaches of Dunkirk. Indeed, it seems probable to me that assembly of this fleet was initiated in England by senior commanders with access to Hut 6 Ultra before the commander of the British Expeditionary Force, Lord Gort, announced his decision to evacuate. The true story has yet to be told. Be that as it may, the days of the Jeffreys apparatus were numbered. By a simple change in procedure, the Germans had already dealt us a blow that could well have been fatal. The exact date of this blow eluded me until Jean Stengers sent me his article in the February 1981 issue off'Histoire, to which I have already referred. From my own memory it seemed that the blow must have fallen in May 1940. It now appears that the Germans made their procedural change on May 10, 1940, the day of their invasion of France. This, indeed, would have been good cryptographic strategy. Let us see what was involved and how, miraculously, we were able to break the Enigma during the Battle of France. Silly Days May to September 1940 On May 10, 1940, with no warning whatever, our traffic registers showed a change in the preambles of all messages. It was a small change that might have seemed unimportant to anyone not intimately familiar with the encoding procedure of the German Enigma operators. It was just an additional three-letter group in each preamble, but to our Machine Room staff this change was shattering. We guessed at once that the Germans had dropped the double encipherment of the message setting. This was soon verified, and it meant that the Jeffreys apparatus had suddenly become useless. To explain what had happened, let me recapitulate the procedure, described in Chapter 3, that had been followed before the change. The operator originating an Enigma message on a particular key would choose an indicator setting, say VIX, and a text setting, say RCM. With the wheels set at YIN he would encode RCM RCM and obtain six letters, say WQSEUP. The letters VIN would be transmitted in the preamble, and WQSEUP would be the first six letters of the message text. The receiving operator, with his Enigma set up to the same key, would set his wheels to VIX, press the WQSEUP keys, and recover the RCM RCM sequence. Xow, however, the originating operator would set his wheels to VIX as before, but he would encipher RCM once only, obtaining WQS. He would then put the two three-letter groups VIX and WQS in the message preamble. Our "females" were lost to us. Nor could our preparations for the arrival of the bombes offer us any immediate help. When the Jeffreys apparatus began to work I had asked Mimer-Barry to make a careful study of decodes. I wanted him to develop an intimate knowledge of the people who were communicating with each other on the German radio nets as this might enable us to find cribs, and cribs, with the bombes to help us, could lead to breaks. Mimer-Barry had indeed acquired an intimate knowledge of the traffic, and he had found that the addresses and signatures were often both lengthy and stereotyped. Thus, by the time the change came, he was well able to produce cribs. But the two prototype bombes would not come into action for many months, and the cribs were no use without the bombes. Fortunately, with the help of German procedural errors that had already been discovered, we were able to improvise new methods of breaking right away. Some of our people had been studying the habits of the German Enigma operators from a very different point of view, and had found some remarkable quirks. The two astonishingly bad habits that now enabled us to go on breaking Enigma even without females were dubbed the "Herivel Tip" and the "Sillies." The first often enabled us to guess the ring setting of each wheel to within two or three letters; the second often allowed us to guess text settings. Both had to do with the lazy habits of the German operators. Herivel's attention was drawn to a quirk in machine setup practice. When an Enigma operator was changing the setup of his machine to a new key, he had-to choose the correct set of three wheels out of the five available, set the alphabet ring on each wheel, insert the wheels in the machine in the correct order, and close the cover. To set an alphabet ring on a wheel, he would probably hold the wheel in one hand so that the clip position was facing him, and then rotate the ring until the correct letter was opposite the clip position. There the clip would engage. Herivel's contribution was to realize that, when the operator inserted the wheel into the machine, the letter determining the ring setting would probably still be facing him, and when he closed the cover it was quite likely that the three letters appearing in the apertures would be pretty close to the ring settings of the new key. Indeed, if the operator was lazy he might leave the wheels in their initial position when he encoded the text setting for his first message of the day. If so, the letters of the indicator setting in the preamble of this message would be pretty close to the ring settings in the new key. To exploit this habit, the midnight-to-eight-A.M. Machine Room watch, working on the traffic registers, would first identify the discriminants of the Red traffic for the new day. They would then look tor the first message on the new key originating from any radio transmitter. The indicator settings of these messages would be entered on a "Herivel Square," Figure 6.1. Indicators HDR, IK'/, )F\Y, etc." would be entered by writing the letter R in the square in column H and row D, Z in square TK, \Y in square OF, and so on. Before too long a cluster would appear somewhere in the square. In Figure 6.1, the cluster consists of the entries (iRI, HSk, (iTK, anil FRJ. The Machine Room would then ha/ and a guess that the ring settings of the three wheels must be: Left-hand wheel F, G, or H Middle wheel R, S, or T Right-hand wheel I, -' \r = ---- , h~ " ~ --- , , -H -A -F A/ "I -- I -\ 1ABCDEFGHIJKLMNOPQRSTUVWXYZ HDR TKZ JFW A OR GRI BOD OKT TFO RLI HTO SAME ABX HSK ZRH KYG OEW AUY SAH JQO GKX GTK GLO XWA FRJ XEI CAV DTR RWI SDM UZR Figure .1 A Herivel Square, with Entries Representing 3O /rid'cutor Settings QWERTZUI 0 ASDFGHJ K PYXCVBNML He had obviously chosen as his indicator settings the alternate keyboard diagonals QAY, KDC, and T(iB. We would conclude that his text settings would be the alternating diagonals: \YSX, RFY, and ZHN. If the text setting of the first part of the message was WSX, we knew that, with wheels set to QAY, MPR was the encode of WSX. Similarly at setting F.IXi, \.w was the encode of RFY, and at setting TGB, \"KA w as the encode of /I IN. Thus this set of three "Sillies" gave us nine letter pairings. We knew that, at the wheel setting following QAY in the scrambler cycle, M and W were paired, that at the next u heel setting P and S were paired, and so on. It was, in fact, a form of crib. Suppose that, in addition to this three-part message, we could find on the traffic register two single-part messages with indicator settings QWF and QAR Looking at the keyboard with an eye to pattern and the known habits of individuals, we might well guess that the corresponding message settings were ASI) and OKI,. This would give us six more letter pairings which, combined with the Herivel tip, would probably enable us to break the Red key for the day. Unbelievable! Yet it actually happened, and it went on happening until the bombes came, many months later. Indeed, though I cannot remember when the two prototype bombes arrived, it seems to me that we must have been entirely dependent on Herivel tips and Sillies from the invasion of France to the end of the Battle of Britain, right up to the final crunch on "F.agle Day," September 15, 1940. By then the R.A.F was nearly exhausted, and it appears that Hitler was not prepared to invade until his Luftwaffe had completely knocked them out. Therefore Goering planned a conclusive triumph. Hut 6 Ultra revealed his plans for that critical day, and helped the R.A.F to make the best use of its remaining capabilities. Goering's attempt to knock out the R.A.F failed, thanks in part to Hut 6 Ultra. Two days later a Hitler directive, received via Hut 6, made it clear to the intelligence staff at Bletchley that invasion plans had been abandoned. What I have called "Eagle Day" should perhaps be called "The Last Eagle Day." Reg Jones, the author of The Wizard War, tells me that the Germans had mounted an earlier Eagle Day (Adler Tag) on August 13, 1940, when they flew 1,485 sorties. Later, on August 15, 1940, they are said to have mounted 1,786 sorties. The objective on each occasion was elimination of the R.A.F as a force that could oppose an invasion. The last "Eagle Day" attack, on September 15, involved fewer than a thousand German aircraft sorties, and the British-claimed 185 enemy aircraft shot down. But this was the day when Churchill was told that the R.A.F had no reserves. There were other types of Sillies that sometimes allowed us to guess a message setting. Occasionally, for example, a lazy operator would get in the habit of using the same three letters both for the indicator setting and for the text setting. This would mean that, having encoded his text setting, he would only have to move the right-hand wheel back three places to encode his message. This type of Silly was known as JAB JAB from the letters used in its first appearance, discovered by Babbage. But in JAB JAB too our help came primarily from the use of keyboard patterns. If, for example, we found a three-part message in which the indicator settings were QAY, WSX, and EDC, it would be a good bet that the text settings were the same. It was extremely fortunate that both the Herivel tip and the keyboard Sillies had been discovered before the Germans changed their indicating procedure and defeated our method of breaking Enigma keys with the Jeffreys apparatus. Had we been foiled then, we would not have been breaking at the time of the last Eagle Day, and we would have lost the continuity that was to prove so essential later on in our use of cribs. Those first few days after the change on May 10, 1940, were quite fantastic. Many people contributed bright ideas. Even at the time it was hard, and after all these years it is quite impossible, to determine who thought of what. We were like a pack of hounds trying to pick up the scent. Till now the observed habits of individual Enigma operators had been regarded as interesting oddities, rather than as a means of breaking Enigma keys. Suddenly they were all we had, and we had to find ways to take advantage of them. As it turned out, I believe, either there was no gap in our breaking, or any gap there may have been lasted merely a few days and we subsequently caught up with the skipped traffic. We were--though by a bare margin--still in the game. During the early Hut 6 days Jeffreys was running the Machine Room and Sheet-Stacking Room; I was in charge of the Registration Room, Intercept Control Rn to the exciting early days of Hut 6, I chose to put the detailed description of the bombes in the Appendix, which might nou be read as an introduction to the actual use of the bombes after they began to be available in the second year of the war. Soon after his arrival in August 1941, Harold Fletcher became involved in the administrative aspects of the bombe program, and his memory has been extremely helpful to me in reconstructing the sequence of events. In summer 1941, when Fletcher joined us, there were four to six bombes at Bletchley Park, housed in what was then known as Hut 11, and a similar number in converted stables at Adstock. the village where Travis lived throughout the war. Another "bombe hut" in the village of Wavendon was nearing completion, and, between them, Bletchley, Adstock, and Wavendon must eventually have accommodated some twenty-four to thirty bombes. Already authori/.ed production at BTM was going steadily on, and Kletcher u as asked to help the Admiralty requisition Gayhurst Manor and build a bombe hut there with a capability of utilizing some sixteen bombes. These early expansions of our bombe capabilities were not too difficult, but involved a problem of timing. The buildings had to be completely finished and ready for use before any bombes could be installed. No outsider could be allowed inside after bombes had been delivered. The machines themselves were operated by members of the Women's Royal Naval Service, who were called "Wrens," though the official spelling was WRNS. When lay hurst became operational, bringing the total to some forty to forty-six bombes, it was still felt that the limit would be about seventy bombes, which would require suitable accommodation for some seven hundred Wrens. In 1942, however, for reasons that I have explained, the picture changed completely. We would need far more than seventy bombes. Commander Saunders, then in charge of Hut 3 operations, was asked by Travis to devote his entire effort to the expansion of our bombe capabilities. In this task he was to be assisted by Harold Fletcher of Hut 6. Because the bombes were being built at the BTM factory in l.etchworth, not far from Bletchley, we were able to keep in close touch with the needs as they arose. Saunders worked with the Managing Director, a Mr. Bailey, while I continued to work closely with Doe Keen. The existing BTM factory was still able to provide a secure area for the assembly line and development laboratory, but it had become necessary to establish a new and quite large factory for the manufacture of parts, particularly those requiring machine tools. Keen directed the entire engineering effort from start to finish, did all the design and worked with his assistants on the final assembly and on troubleshooting and modifications after delivery. He visited Hut 6, as Ellingworth of Chatham had done, to get the flavor of our problem. He, together with a few of his principal assistants in the development laboratory and the man in charge of the new factory, were told the whole story of Hut 6. The workers on the assembly line and in the factory, however, never knew what the machines were intended to do. I hope this book will tell these people, and their children and grandchildren, that what they did was enormously worthwhile. There were other workers involved in bombe manufacture besides those at BTM, for in support of the war effort part-time work had been organized in villages all over the country. Within a wide radius of Letchworth some of these village teams were making such parts for the bombes as did not have to be made by machine tools and other factory equipment. Some of the people who worked in those village halls may remember that they were making twenty six-way cables, or other electrical subassemblies involving the number twenty-six, in which case they also may be glad to know that they were almost certainly making an essential contribution to Ultra. There were problems of financing, because the first contracts with BTM were covered by Secret Service funds, whereas the expanded program was funded by the Admiralty, which caused a slight hiatus in the summer of 1942. Bailey explained to Fletcher at that time that one of the BTM workshops, making the basic frames for the bombes, had almost completed work on its current contract. If a further order was not received in time, he would have to put this workshop onto other war work, after which it would not be possible to transfer the workshop back to bombe work for a considerable time. Fortunately Bailey accepted Fletcher's personal assurance that a new order for at least as many bombes as they had already contracted for would shortly be forthcoming. It took another month or two for the new order to materialize, but no delay in the delivery of the bombes resulted. Late in 1942 or early in 1943 BTM began to have difficulties in getting raw materials, which they could not resolve themselves. Malcolm Saunders was brought in, attired in his navy uniform, to apply Admiralty pressure on suppliers and Government departments. Though rather unapproachable, he did this sort of thing very well. He had an impressive manner and the much-needed supplies quickened up under his pressure. He also gave helpful pep talks to factory workers to encourage their efforts, as, for example, when he went with Fletcher to the Spirella corset factory at Bal dock, which, instead of corsets, was making parachutes and miscellaneous bombe parts. The biggest problem facing Malcolm Saunders and Harold Fletcher was where to put the projected quantities of bombes and the Wrens who would operate them. Fletcher recommended that a private school, Stowe, which was within easy reach of Bletchley, be requisitioned, but the suggestion was turned down at a very high level. I le and Saunders had to settle for some brick office buildings at Stanmore, on the northern outskirts of London, which were nearing completion. This actually proved to be a very good solution to the problem. It was possible to provide nearby accommodation for the large staff of Wren operators. Furthermore it was possible to provide exclusive teletype and telephone lines to Bletchley Park. For efficient operation of the anticipated number of bombes we needed large establishments, and, as a precaution against enemy air attack, we needed dispersed sites. So, having found the Stanmore scheme so satisfactory, Saunders and Fletcher readily settled for another site in Eastcote, in the western suburbs of London, where two buildings were nearing completion and there was space for a third, with effective encouragement from Saunders, in his navy uniform of course, the third block was erected in around six months. It housed some eight hundred to nine hundred Wrens. Bombes from wavendon and Adstock were moved to Stanmore and Eastcote, but Gayhurst remained in operation throughout the war. A few bombes remained in Bletchley Park, but they were used for demonstration and training purposes only. To understand the task of the Wren operators we must first consider the initial setup of a bombe in accordance with a menu prepared in Hut 6. In the Appendix, when I explain the idea of the diagonal board, I use the second diagram of Figure 4.4 and show, in Figure A.3, how the in-out and out-in terminals of nine double ended scramblers would be connected by twenty-six-way cables to rows of the diagonal board in accordance with the diagram, which uses some of the letter pairings that occur in the first thirteen positions of the crib. The same diagram is repeated as the main chain in the bombe menu of Figure 8.1, which is based on the same thirteen positions of the crib and on the assumption that there is no turnover in that stretch. But because our bombes had twelve scramblers we could use three more letter pairings: those of the two-link subsidiary chain R to L to S and also one of the single-link chains, say Q to O. This will reduce the number of drops by providing feedback from rows R, L, S, Q, and O. The Z to H link will not be set up on the bombe, but might be employed in testing drops. The letters under the crib positions at the top of Figure 8.1 indicate initial positions of the top, middle, and bottom drums of the bombe's scramblers for the start of a run. Thus a Wren would set the first scrambler to AAA and connect its in-out and out-in terminals to rows C and T of the diagonal board. The second scrambler would be set to BAA and connected to rows Q and O, the third to CAA and connected to rows N and T. Because the letter pairing Z to H in position 4 is not to be used, the fourth scrambler will be set to EAA and connected to rows P and E, the fifth to FAA and connected to rows V and P, and so on, until the twelfth scrambler is set to MAA and connected to rows I and N. When this initial setup has been completed for a selected wheel order and the test register of Figure A. 3 has been connected to row E of the diagonal board, the bombe is ready for a run, during which its twelve scramblers will be driven in synchronism through all 26 X 26 X 26 possible positions of the drums of each scrambler. Let us consider how Keen handled this mechanical motion and the associated electrical sensing. Remember that in each position of the battery a test had to be applied to determine whether some letter of the test register was not connected to any other letter. Thus, if current input was at letter A of the test register, Keen's sensing circuitry had to signal a drop when current failed to reach all the terminals of the test register. In fact he had to sense the occurrence of one of two possibilities. The first was a position of the bombe THE CRIB. C Q N Z P V L I L f> U I KTEOCeLOVWVG T U F L H Z rOTHEPRES IDENTOFTHEUNI T D S T A T S I234S6789IOIIIZ 13 14 IS 16 IT 18 19 ZO 21 iZ S3 24 25 26 27 2S 29 30 ABCDEFGH I K L M AAAAAAAAAAAAA AAAAAAAAAAAAA the: main chain Positions I to 13 SUBSIDIARY CHAINS jl--m--9~--T7 ---0 B T4 (not unit i I Tut fftfistff Coiu> ct to row E of tfi Diagonal Board Figure 6.1 A Bombt Menu scramblers in which current reached no terminal other than A, which would mean that, for this position, the bombe had failed to rule out the possibility that E is steckered to A. The second was a position in which current failed to reach some other terminal, say J, which would mean that the possibility of It's being steckered to J had not been ruled out. For his sensing circuitry, Keen designed an electromagnetic relay that could operate in one millisecond. (It was similar to a Siemens relay that was about the fastest available at that time.) The top drums of all the scramblers were driven as fast as the sensing device would permit. The sensing for each position was timed to occur during periods when the drums' brushes would be making good contact with the commutator terminals. As the top drums made a complete revolution, sensing would occur in all twenty-six positions. If no drop occurred in a revolution of the top drums, sensing was discontinued and a signal went to the drive mechanism to cause all the middle drums, and sometimes the bottom drums as well, to move to their next positions. Another sensing cycle was then initiated. When a drop was sensed, the top drums were not stopped but the motion of the middle and bottom drums was inhibited. Indicators would tell a Wren operator which letter of the test register was involved, and the position of the still moving top drum of the first scrambler in which the drop had been sensed. She would also note the positions of the middle and bottom drums of this scrambler and the wheel order being run. Having noted all this information, which would be reported to Hut 6, the Wren operator would reactivate the drive mechanism for the middle and bottom drums. The remote possibility of two drops in one revolution of the top drums was taken care of. The time occupied in dealing with drops would depend on the number of drops and on how quickly a Wren operator could get to the bombe. It is Harold Fletcher's recollection that an average bombe run took about twelve to fifteen minutes. When a run on a particular wheel order had been completed, another wheel order would be tried. It was possible to choose a sequence of wheel orders that would not call for more than one set of drums (top, middle, or bottom) to be changed between successive runs. Thus, allowing for the initial setup time, changing drums, and dealing with drops, it took one bombe rather more than twelve to fifteen hours to run one menu on all 60 possible wheel orders. When we had plenty of bombes we could put several bombes to work on the same menu, greatly reducing the time required to achieve a break. The diagram of a bombe in Figure A.4 (page 308) shows an open door at the back. All the in-out and out-in terminals of the twelve scramblers were connected by an enormous cable to twelve pairs of twenty-six-way female jacks on the inside of the door, which also contained twenty-six jacks for the rows of the diagonal board, one jack for the test register, and the commoning boards. Thus the Wren operator would do all the cross-plugging on the door. The maintenance staff would have good access to the drive mechanism behind the front panel of the bombe. When, very likely at the instigation of Travis, it was agreed that the operators of the bombes would be provided by the Womens Royal Naval Service, no one could have anticipated that the number (it bombe Wrens would rise to around two thousand. In spite of this heavy demand, however, the high level of qualifications laid down at the outset was maintained to the end, and the Wrens were exceptionally good. So good that one felt that the pick of the recruits were being assigned to bombe duty in the WRNS organization known as HMS Pembroke V. Although the bombe Wrens could not be given a taste of the sea, the WRNS authorities insisted that proper Naval nomenclature be used. This is why the organization to which the Wrens were assigned was given the name of a ship. Moreover, in their "ship" there were wardrooms, galleys, cabins, and berths. At one or two sites there was even a quarterdeck, which all Wrens had to salute. Harold found it refreshing to hear the humble minibus or shooting brake, which took Wrens to the station when going on leave, referred to as "The Liberty Boat." F.ven so, he always felt a great deal of sympathy for them. Most of them had joined the Wrens, as opposed to any of the other services, in the expectation of seeing something of the sea, of warships, and of sailors. Instead they were stuck a long way from the sea, doing monotonous shift work that they couldn't talk about at home or anywhere else, with very little promotion, and not a male naval uniform in sight. It seemed to Harold that there was considerable danger of loss of morale, resulting in mistakes in operation and possible threats to security. So he devised a talk explaining in simple terms the problem the bombes were helping to solve, and its place in the chain of processes that, with luck, resulted in decodes. This talk was an immediate success, and Harold was invited to deliver it, as a matter of urgency, some ten to twelve more times, so that all the bombe Wrens who had already arrived could hear it. Thereafter he repeated the talk for new Wren arrivals. The R.A.F mechanics also attended, and, later in the war, so did the U.S. Army personnel who operated a few of the bombes at Eastcote. Harold also gave or arranged for lectures on specific problems affecting bombe operations, such as the appearance of umkehricalze D late in the war. He always tried to give, at the end of each talk, some comparatively innocuous items of intelligence in an attempt to show the work had a real connection with the war. At the time Fletcher kept fairly quiet about the contents of these talks, in case he should be stopped on security grounds. He himself was quite certain that he was actually improving security, an opinion that would be confirmed by modern experience. Indeed the talks cannot have failed to make a real contribution to the welfare and efficiency of the Wrens, and they were certainly remembered by some. For several years after the war it was not unusual for him to meet socially a lady who would tell him that she had heard him lecture. The last occasion when this happened illustrates the deep and inviolate sense of security that Lew in mentions in Ultra Goes to War. During a tour of duty in Singapore in 1956 or 1957 a Lieutenant-Colonel arrived to join the intelligence staff of Far East Command and became aware of Harold's trade. His wife, on being introduced to Harold at a cocktail party, said, "I was with the Wrens during the war and went to hear your lectures." At which the Lieutenant-Colonel, with a triumphant grin on his face, turned to his wife and said, "So that's what you did in the war." Inevitably the day came for an inspection of Hut 6 by the Director of the WRNS, Mrs. Laughton-Matthew s, who held the appointment throughout the war. The inspecting party consisted of the Director; Mrs. Mackennie, who commanded our Wrens with the rank of Chief Officer; Captain Bradshaw , R\, the senior naval officer at Bletchley Park; and Fletcher. After a short preliminary talk in Fletcher's office, the party proceeded to a Registration Room, uherc. of a total staff of about thirty, around ten were Wrens. The Director at once asked Fletcher, in rather supercilious tones, "Why are my Wrens working with civilians?" Fletcher, caught oft balance by this unexpected question, came out with the unadorned truth, which was that we were using Wrens because we couldn't get any more civilians. This shook Mrs. Laughton-Matthews to the core, and Captain Bradshaw retired behind a pillar for a giggle. Mrs. Laughton-Matthews bore Fletcher no ill will, however. Indeed he met her later on a number of occasions, and she couldn't have been nicer or more helpful. Fletcher was the man primarily responsible for establishing standard procedures for using the bombes and for maintaining harmonious relationships among the various parties involved, particularly the Wrens who operated the bombes, Sergeant Jones and his men who maintained them, Doc Keen and his men who built them, and the watch in Hut 6 who gave them "menus" and used their output. Sergeant Jones was one of the original maintenance engineers. Travis had decided to give him overall responsibility for maintenance of all the bombes and, working out of Bletchley Park, Jones established effective maintenance procedures. This experience of his techniques was extremely valuable, particularly in the care of around 15 million delicate wire brushes on the drums that had to make reliable contact with the terminals on the templates. (There were 1014 brushes per drum, 720 drums per bombe, and ultimately around 20(1 bombes.) Later in the war, when other people tried to operate them, we realized how lucky we had been to have had Sergeant Jones. Me was a tireless worker, very ingenious, and willing to take on any technical problems that Hut 6 might throw at him By the end of the war he had been promoted to Squadron leader. All the bombes had names. The first prototype was called Agnes, 1'itt after the first few bombes had arrived, the responsibility for naming them was given to the Wrens, who tended to use warships"ui ports. Thus one might hear a Wren duty officer call out "Put Warspite, Victorious, and Tiger on Chaffinch job 81." This was better for morale than saying "Put machines 110, 111, and 112 on job 9-81." Later in the war the German navy modified its Enigma by splitting the umkehrwalze into a fixed left-hand part and a rotatable right-hand part, creating a machine with four rotating wheels instead of three. When Keen was brought into the picture, he immediately produced a sound approach to the problem, and was soon manufacturing four-wheelers instead of three-wheelers. And his solution was simplicity itself. When he had started building his bombes he did not know how fast he could drive his drums and still obtain reliable contacts between their brushes and the commutator terminals, so he based his drum speeds on the testing speeds that he expected to achieve with electromagnetic relays. Now, however, he knew from experience that he could safely drive his drums much faster, and it was an easy matter to use already well established electronic techniques to perform the very simple testing function at speeds much higher than :' those achievable by his relays. So Keen simply added a set of twelve super-slow drums below the top, middle, and bottom drums of his existing bombes. This involved making higher frames and providing a simple addition to the bottom of his drive mechanisms. The drums were already in large-scale production and no change in the electrical circuitry was called for, other than the simple matter of introducing electronic sensing of electric current reaching the terminals of the test register. Hey presto! With very little additional effort Doc Keen was building four-wheel bombes that could do twenty-six times as many tests as the three-wheel bombes in only twice the time. From Interception to Intelligence The big expansion of our intercept facilities began in 1941, when the R.A.F opened a big station at Chicksands, a little to the east of Bletchlcy, under Wing Commander Shepherd. In the meantime we depended almost entirely on the highly experienced intercept operators at the army station at Chatham. This station was a good site tor interception, except that, being on top of a hill, it probably did not have enough level space for the large rhombic and Beveridge aerials that would be needed to intercept traffic from considerable distances. But another consideration limited its suitability. Chatham was too exposed to enemy air attack. Accordingly, the Chatham intercept organization had to be moved. Our crack intercept operators' new home was Beaumanor. a country estate in Leicestershire some 50 miles to the north of Bletchley. On a flat part of Beaumanor's grounds the necessary large aerials were erected, and several huts were built. One hut housed Commander FTlingworth and his controllers, and had teleprinter circuits to Bletchley Park. The other huts were for the intercept operators. The interception huts all had pneumatic tubes leading to the control hut; these were used for sending across hand- u'itten forms containing logs and intercepted messages. I he pneumatic tubes were laid underground, and in the winters, which w ere quite rugged, they would collect condensed moisture; thus the arrival in the control room of a container from one of the other huts would often be announced by a spurt of water. Astor the motorcycle dispatch riders, they had a somewhat shorter ride to Bletchley Park than they had had from Chatham, but it was still a long one in bad weather and in the blackout I visited Kllingworth several times at Beaumanor, and found that the dedication of the operators and of the staff was just as impressive as it had been at Chatham. I don't know how many new operators had been recruited, but I believe the station had grown considerably. Nor was interception at Chatham entirely defunct, for Fl ling worth kept a small group of intercept operators there so as to have diversified reception. This could make the difference in the case of the weaker signals, some of which might be more audible at Chatham than at Beaumanor, or vice versa. The R.A.F intercept station at Chicksands, a large station from the outset, opened in 1941 with male operators. In August 1942 it was ; augmented by operators from the wAAF (Women's Auxiliary Air| Force). Late in 1942, an auxiliary station was established at Shaftesj bury, initially staffed by wAAF operators, to provide diversified < reception. Thus by early 1943 we had tour widely separated principal intercept sites: Beaumanor in Leicestershire, Chatham in Kent, Chicksands in Bedfordshire, and Shaftesbury in Dorset. Wing Commander Shepherd, at Chicksands, was a tremendous help to Hut 6. He was eager to cooperate with our Intercept Control Room, and he reali/.ed that his operators and staff had much to learn about the interception of German radio nets carrying F.nigma traffic, particularly those that were almost inaudible. The R.A.F was now running a training school for intercept operators, which must have helped a good deal. The Chicksands operators were well indoctrinated and well led from the outset; my early visits there gave me the same impression of dedication that I had observed at Chatham and Beaumanor. In a remarkably short time Chicksands was making a valuable contribution. The quality of its interception went on improving, and the station became more and more expert at * wc gradually acquired more tele tv pc equipment, and when we moved to our n u brick building in 1^43 we had enough to carry virtually all the intercepted Knigm.i mess gcs. By that time we were also receiving intercepts from overseas, sent by radio after iK'ing neoded on I ype-\. The double encoding produced a good many garbles, so these intercepts ere not too popular in the Hut 6 Decoding Room, but they produced a lot of good stuff to Hut 3. picking up weak signals from Africa, the Balkans, and the Russian front. During the war the dedicated operators and staff of these intercept stations were doing a job that was absolutely vital to the production of Hut 6 Ultra intelligence. Recently, thirty years after the end of the war, winterbotham's The Ultra Secret made it known that what they did resulted in intelligence of immense value, but, not having been in touch with the intercept stations, he does not pay adequate recognition to their contribution to I hit 6 Ultra. The tribute that I am able to pay in this book is long overdue. When, in 1971, I retired as a regular employee of Mil RK, another of the many happy chances of my life led me to choose the historic and beautiful city of Newburyport, about an hour's drive northeast from Boston, as my retirement home. It was a fortunate move for my story, as well as for myself, because in Newburyport my wife already knew Diana Lucy, who, as Diana Stuart, had been a WAAF operator at Chicksands. She has given me an account of how she came to be there and what it was like to be an intercept operator. Quite early in the war young unmarried women who were not in "reserved" occupations such as teaching or nursing were subject to a system of "directed labor," which might mean having little or no choice as to one's assignment. One could end up working in a munitions factory or on a farm, occupations that might be congenial to some, but not to others. However, one could volunteer for the assignment of one's choice before one was actually called up. Diana was training as a speech therapist, which was not considered a "reserved" occupation, so at age nineteen, in October 1941, she volunteered for the Women's Auxiliary Air Force (WAAF), known today as the Women's Royal Air Force. She became Leading Aircraft Woman 2082928 and was sent for six weeks to a camp outside Gloucester for basic training and aptitude evaluation. It was decided that she was to be a radio operator, and she was sent to Manchester for an intensive six-month course in the dreary Post Office building, where her instructors were GPO telegraphists. After further training in Blackpool, Diana and many other " (.c-ncr. il Post Office. VVAAF radio operators were posted to RAP tighter stations on the! south coast of England. Then, in the summer of 1942, after an intensive course on the maintenance of radio equipment, she found herself at Chicksands Priory, having been told only that she was to do "specialized radio work." At the Chicksands intercept station conditions were primitive. The WAAFs slept in Nissen huts and shared bathing facilities with] the men. They worked eight-hour shifts, and reported for duty in aj cobblestoned courtyard of the old Priory, where a roll call was! taken. Then the thirty-odd members of a shift would proceed to aj "watch room" high up in an old wing of the Priory, with cold stone] floors, Gothic windows, and a vaulted ceiling. The only access used by the WAAFs to this eerie watch room was a narrow circular stone staircase, with a stone column in the middle and steps that showed: centuries of wear. The radio sets were lined up on shelves above working benches running along the walls. Bats had been accustomed for some rime to : make use of the vaulted ceiling of this watch room. On one occasion the bats were disturbed and swooped screeching all over the place. Pandemonium ensued. Interception ceased, because the WAAFs were under the benches, until the bats were somehow caught or driven out. As soon as possible much better working conditions were provided in new concrete block structures with fluorescent lighting. Late in 1942 Diana and a party of around twenty WAAF interceptors were sent to the Chicksands satellite station at Shaftesbury in Dorset to operate an evening watch, roughly from six p.m. to two a.m." when reception of distant radio transmitters was particularly good. Diana much enjoyed this assignment in a particularly lovely corner of Fngland. The WAAFs were billeted with families in the town, who had no idea what was going on. The intercept facility was referred to simply as the "radio station," and there were plenty of those around the countryside. Diana herself was somewhat reluctant to return from the coy atmosphere of a billet to the regimentation of Chicksands in the summer of 1943. The Shaftesbury activity continued to expand after that date. Diana has the distinct impression that the pressure on intercept operators increased considerably toward the end of the war, starting with the winter of 1943/44. There seemed to be more stations on each German radio net, and more traffic too. It was just about all she could do to record on log and message pad all that was going on. When one remembers that the German retreats in Russia started early in 1943, and that their troubles in Italy started only a few months later, this seems understandable. They would have to deal with deep infiltrations and clandestine operations. More and more they would need to keep track of what was going on everywhere. To indicate how specialized was the task of our intercept operators, I must say a little more about how the Germans controlled the radio nets that carried the Enigma traffic. I understand this far better now than 1 did during the war. Only recently have I learned that the frequency stability of radio sets in those days was very poor, particularly under field conditions. Diana tells me that the drifting of German radio frequencies often made it extremely hard for our intercept operators to keep in touch with the stations of a net. There was no problem when the signals were coming in strongly, but she was often listening to distant nets on the Russian front. Usually the control station of a net was relatively easy to pick up; the problem was to identify the very weak signals from the outstations, which would fade out with the slightest drift in transmitting frequency. Our intercept operators would be continuously tuning, trying to pick up outstation transmissions, and it was only too easy to pick up a signal From a different net on a nearby frequency. By following a net for a long enough period, an intercept operator could learn to recognize the individual German outstation operators by their "fists"--the way they keyed their Morse code transmissions--and this proved the most reliable method of keeping in touch with all the stations of a net and avoiding confusion with other nets. When contact had been established with a particular net, prediction of call signs could be a great help to the intercept operator, but often he or she would be looking for new nets, created to meet the requirements of a new tactical situation. It the task of our intercept operators was not an easy one, neither was that of the controller of a German radio net. The Germans had great difficulty in keeping their nets operating efficiently. The procedures they developed to cope with this problem were very thorough, further demonstrating their regard for the importance of reliable communications in fast-moving military operations. But the problem continued to plague them throughout the war. The same frequency drift and fading that made it difficult for our intercept operators to pick up signals on a specified German radio net hampered the German operators in hearing each other. Let us think about what this implies. We are talking about a very simple and flexible type of radio net. Its radio frequencies are in the high frequency (shortwave) range to permit communication over considerable distances. A number of radio stations use the same frequency both for transmission and for reception. All the stations monitor this frequency all the time, one of them acting as net control. Stations may join or leave the net as necessary, and the control function can be transferred from one station to another. If there were no problems of drifting and fading, the function of the control station would simply be to control traffic. Any message transmitted by any one of the stations would be heard by all the stations. The controller would simply have to ensure that no two stations transmitted at the same time. But the Germans' problem was not so simple. Diana, and many hundreds of skilled British interceptors, would follow the chitchat by which the German net controller struggled to keep his net in working order. This involved a lot of very short messages using the call signs of the individual stations and a so-called Q-Code. This needs a little explanation. The creation of Q-Codes, still widely used today, dates back to the sinking of the Titanic on April 15, 1912. Radio distress signals from the ship were picked up in New York, but although there were ships of several nationalities near enough to have assisted, they would not have been able to understand the distress signals even if they had picked them up, because of language difficulties. This incident showed that to handle emergencies at sea it was necessary to introduce an international code that, in combination with figures (e.g." for latitude and longitude) and names (e.g." of ships), would permit effective communication anywhere in the world, regardless of language differences. The three-letter international code that was soon introduced was called the "Q-Code" because each three-letter group began with a Q. Typical meanings of code groups are: QBiA What is the name of your station? QRG What is my exact frequency QRM Is my transmission being interfered with? QUO Shall I increase transmitter power? QRQ Shall I send faster? QTH What is your location? QTX Will you keep your station open for further communication with me? QUA Have you news of---? In the international code of today there arc some forty such three letter groups, together with a few special signals, such as the well known SOS. To control their military radio nets in World War II, the Germans needed a specialized form of Q-Code. After all these years Diana does not remember all the code groups, but she believes that the list was similar to that used today by the Amateur Radio Relay League. Without going into further detail, we can now understand the general nature both of the German task of net control, and of our task of following the "whispers" of the German nets. For our intercept operators would record all the chitchat on logs, and the analysis of these logs, known as "log reading," was the basic tool of the "Central Party," whose radio intelligence activities I am about to discuss The individual stations of a German radio net identified themselves by call signs usually consisting of two letters and one figure; these changed at midnight, as did all the Enigma keys and their identifying discriminants. When the new net controller came on at midnight, his first task was a roll call. Using their call signs and the appropriate Q-Codes, he asked each outstation to transmit an acknowledgment if it heard him. If he failed to get an acknowledgment from one of the outstations, he would ask another one to try to make contact with the one that had not answered. This initial procedure gave our interceptors a chance to record on their log for the day all the call signs of the nets they were covering. We in Hut 6, however, would be aware only of the call signs that were involved in message traffic, because only these would appear on our traffic register. I hi c.irl\ log-reading effort in MIS was mentioned at the end of Chapter 5. Having established contact with all his stations, the German controller proceeded to arrange for the handling of traffic. If he had an Enigma message for a particular station, he would make sure that' the station was ready to receive before transmitting the mes sag preamble and text. If he had an Enigma message for all the stations! of the net, he would announce it with an all-station (CQ) call. Hei would also ask each station in turn if it had a message to transmit,] and give permission for the transmission. For a message addressed! by one station to two or more others, the controller would check! that the addressees were receiving from the sender before permit-i ting the transmission. From midnight to midnight, successive watches of German net! controllers would struggle to keep the communications lines operative so that when a message needed to be sent it would reach its | destinations). Sometimes the recipient of a message would be un! able to receive the whole text, and would ask for a repeat; or one of) them might be unable to decode the message at all, and would ask for a repetition of the indicator setting and indicator. Sometimes we found ourselves doing better than the German recipient. Provided that one or more of our intercept operators was steadily following a particular net, the log would show all the chitchat involved both in keeping the net operative and in arranging for the transmission of messages. All call signs would appear, and the existence of the net would be revealed even if no messages were passed. The existence of the net! Looking back I am sorry that I did not recognize the significance of the fact that log reading could reveal the existence of a German radio net even without message traffic. I have been told that, in the official accounts of Bletchley Park activities, I have been regarded as the originator of traffic analysis, commonly shortened toTA To put the record straight, however, I had no hand in developing the possibilities of chitchat analysis, as distinct from the analysis of Enigma message traffic. Rather naturally, because I was developing my kind of TA as an aid to breaking Enigma keys, I was concerned with the messages that we hoped to decode. Very early on it seemed certain that we could achieve our goal with the information contained on our traffic register-information related to the transmission of an Enigma message. I was hardly aware of the Q-Code signals used by the Germans to keep their nets operating. I never read a log myself. It never occurred to me that the Germans might inadvertently reveal their intentions simply by exercising a radio net, without passing any message traffic on it. I lou ever, the leaders of the expanding log-reading effort were well aware of this possibility. Their objective was pure radio intelligence, derived from detailed analysis of enemy radio transmissions. & ' They studied the chitchat in the logs, they used direction-finding facilities to locate the transmitters, and they collected call signs used on successive days by the same station so that, when the yearly repetitions appeared, log reading could begin to identify military units served by radio stations No doubt they used the intercept operators' identifications of the German radio operators by "fist" when they could. They studied the preambles of the encoded messages, and particularly the re transmissions of the same message on different nets, in order to piece together the whole structure of the German radio communications system, which had to reflect the command structure and the order of battle (what units were where under whose command). By getting to know the regular pattern of message traffic in relatively quiescent periods, they could hope to detect something unusual and to guess what it might mean. Occasionally the logs would contain short messages in the clear, giving names of individuals or units. All this effort had started in a small way in London in the early days of the war, while I was beginning to study message traffic in the School. When the Chatham intercept operators moved to their huts in the grounds of the Beaumanor mansion, the log readers moved into the mansion itself, and grew to a large organization. During this period they must have gained a good deal from close proximity to the intercept operators, but as the R.A.F station at Chicksands expanded, and carried an increasing portion of the load, Beaumanor became a less advantageous location. I believe it was in 1^42 that the log readers moved to Bletchley Park, where they became known as the Central Party. By now the Central Party was a large organization of British army men and ATS (Auxiliary Territorial Service) women. They worked in a separate hut, but they became an integral part of the I his identification became much easier when the German call sign books were captured, but by that time the log-reading party had gone a long way toward the reconstruction of the tmnks Hut 6 organization. At last, now that they were with us, I was able] to tell them that we were breaking a great deal of Enigma traffic,} how we were doing it, and how their log reading could help. Their I leader, Philip Lewis, joined my management group, and the Cenf tral Party was included in the circulation of weekly reports. They! were reinforced by a group of U.S. Army men. The arrival of the Central Party at Bletchley Park was a boon to I me for a personal reason. My wife, Katharine, having young children, was not subject to conscription, but, like many young married women, she had volunteered for service in the ATS, leaving] the children in the care of her mother. She went first to the ATS Depot at Aldermaston and was then assigned, as Private No. W/75856, to 1st AAMT Company ATS at 8 Rutland Gate, London, which meant that she could easily get home on leave. However, after a period as a driver, she was posted to the log-reading group at Beaumanor and moved with them to Bletchley Park. Now she could live at home, and she was permitted to know what I was doing. One of the tasks in which she became involved was the continual updating of a huge wall chart showing all the German radio nets on the Russian front. The Central Party's intimate knowledge of the German radio nets was of immediate value to the Hut 6 Intercept Control Room, and they were soon able to cooperate a good deal with the Hut 6 crib hunters, with the Registration Room, and even with the Decoding Room. I think we were all amazed by how far they had been able to get without ever seeing a single decode. They maintained a detailed picture of the structure of the German communications system, and this was updated as the logs from the intercept stations were analyzed. They were already familiar with the routine messages that interested our crib people, they had followed successive re transmissions on different radio nets, and they could help Hut 6 to detect cases in which a message originally encoded on one key had been reencoded on another key before being retransmitted. They would notice situations in which the recipient of a message who was having difficulty with the decoding would ask for a repeat of the indicator setting and indicator and might receive a correction that would help our Decoding Room. The Central Party was sometimes able to assist Hut 6 in breaking Enigma keys and in exploiting the breaks. Sometimes its contribution was a matter of helping the Hut 3 people develop the full implications of Enigma decodes; sometimes the Party people were able to produce independent intelligence. Always they were trying to keep up their expertise, so that they could act as a backup to the key breaking. If Hut 6 had had a temporary setback, as could easily have happened, the Central Party could have continued to produce intelligence, and could have been a great help in attempts to start breaking again. Moreover, if Hut 6 had become unable to break Enigma keys altogether, the Central Party would have become the principal source of radio intelligence derived from nets carrying Enigma traffic. Fortunately that never happened, but if it had, the Central Party's close association with Hut 6, and still more with Hut 3, would have greatly enhanced their ability to produce valuable radio intelligence without decodes. I will have more to say in Part Four about the potential value and actual achievements of radio intelligence. At this point I am concentrating on activities directly related to the breaking of Enigma traffic by Hut 6. I have spoken of the expansion in bombes, interception, and log reading. We now come to intelligence derived from the decodes. The expansion of the Hut 3 intelligence activity was vital to exploitation of Hut 6 Ultra in an increasingly complex war. Peter Calvocoressi, who was first deputy head and later head of the Air Section of Hut 3, has given some indication of the magnitude of the task in an article of November 24, 1974, in the London Sunday Times Weekly. As Hut 6 succeeded in breaking more and more Enigma keys--tor example, at some point we began to break the SS traffic as well as the operational traffic from the many fronts--it was obviously necessary to expand the organization that was to squee/.e the last drop of reliable intelligence from the decodes. Calvocoressi points out that this was no simple matter. The expanded Hut 3 revoked around its twenty-four-hour watch, which occupied a room about 30 or 40 feet square containing a large horseshoe table. The Head of the Watch sat in the middle of the horseshoe, with some ten members of the watch facing him around the outside--all working on the stream of decodes from Hut 6. After the move to our thew brick building the decodes were transported from the Hut 6 "Voiding Room to the Hut 3 watch by a conveyor belt that never stopped. Another omvevor tx-'it linmght the raw intercepted messages from the teleprinters to the Hut 6 Registration Hi nun Some of these people had been schoolteachers; others came from universities, museums, and business. One thing they all had idj common was that they knew German. Most of the members of th watch belonged to one of two main Hut 3 sections--Army and Airl During Calvocoressi's time as deputy head and head, the Air Se tion grew to a strength of around fifty. At any time of day or night two army and two R.A.F "advisersl were working in the corner of the watch room. Their job was to take the translated decodes, annotate and comment on them, an--not to retell old war stories, but to discover how up-to-date it seems. What I learned then keeps providing fresh insights into today's problems of secure battlefield communications, on which I have been working at the MITRE Corporation. In many other ways also, the Hut 6 experience of 1939 to 1945 seems to have its meanings for the present generation. The Hut 6 story, in fact, deserves to be reviewed and related to what has happened since, and to what i may happen in the future. I will continue my personal story in Part! Four, in which I will make a big jump from the World War II; experience to the military's communications problems of today as I see them. The Bletchley Park Environment In my discussions of successive phases of Hut 6 activity I have, I hope, given many glimpses of what it was like to work in England and in particular at Bletchley Park during the war years. Vet this book would he incomplete if it did not contain a few additional comments on our overall environment and on the marked contrast between the peaceful life at Bletchley and what was going on elsewhere. Like everyone else, we had to contend with routine problems such as food rationing, gasoline rationing, the blackout, and acute shortages of important supplies. Because of the nature of our work one of the most irritating shortages was that of ordinary writing paper. We used the reverse sides of no longer needed documents. Knowing this, an American, whose office I visited in Washington, teased me by deliberately tearing up an unused sheet of scribble paper. It hurt. I he national food rationing in Britain was well organized. What food was available was evenly distributed, and price subsidies made it possible for everyone to buy what was needed. Indeed it was said that those with the lowest incomes were better fed during the war than ever before. Kor some reason the traditional British fish and chips, when it U'as available, could be bought without coupons. I found this a great boon. Because my work schedule as head of Hut 6 was apt to be extended beyond the time for which official transport was proviided for shift workers, I was allowed enough gasoline coupons fa journeys to and from home--and this allowed me to drop in at if village fish and chip store quite often. Kven so, because of the paj shortage, I still had to provide my own piece of newspaper fo wrapping before I could take my meal home and eat it. The scarcity of liquor was a problem, too. One could usually bu) a drink of some kind in a pub, but it was hard to obtain a privates supply of hard liquor. I was fortunate in that I knew the sale representative of a firm that had sold wine and liquor to Sidney! Sussex College for many years before the war. I le was able to send me a case of twelve bottles from time to time. Once, however, aj wooden case arrived, was ceremoniously opened by Katharine and! myself, and was found to be full of coal! Reporting this, I found the liquor was insured against theft, and that another case would be sent. Later we decided that we might as well use the coal in our; stove, so we took it out of the case. At the bottom we discovered a! bottle of rum. F.evidently the thieves were only interested in the gin! and whisky that had constituted the remainder of the consignment." I reported this also, but was told that the insurance claim had: already gone through, and that I had gotten a free bottle of rum. Billeting was a problem from the start, but this got worse in summer 1940 when the Phony War came to an end and section after section had to operate around the clock. The problem became even more acute with the major expansion of activities that started in 1942. The distances at which people had to be billeted increased steadily, and in the end a good many were living in Bedford, some 20 miles away. An extensive bus service had to be instituted. The three eight-hour shifts were hard on billet or and billetee alike. Each would have to be quiet while the other needed to sleep. It was impossible to settle down to a regular routine, because the billetces would change from one shift to another. One billet or might have two billetees on different shifts. Moreover, there being a big railway junction close by, many billetors were railway workers, who also had to work on the night shift. Nevertheless the people who came to Bletchley Park found themselves at least in pleasant rural surroundings. This, combined with the pleasant atmosphere of the whole place, contributed to our overall well-being. Moreover our minds were on our work, and we were happy to be doing something that was really helping the war effort. For The great majority of "B.P.-ites," that converted estate with its jumble of huts was a very happy place in which to live and work. As I have said, many of us look back to Bletchley Park Days as the best period of our lives, and it has been a real trial that until recently we have been unable to discuss our activities there. As the numbers of personnel expanded, so did our social activities. We had a fine collection of people drawn from a great variety of peacetime activities, including the stage, the arts, the sciences, and the professions. In spite of the problems of shift operations and transportation to and from scattered billets, people managed to get together for many social activities, large and small. Because my wife, Katharine, was a very good amateur singer and pianist, I was more in touch with the musical activities than with others, and we certainly had plenty of very good musicians, professional as well as amateur. Concerts were held in the cafeteria, and they were excellent. I think the musical community at Bletchley Park put on Purcell's opera Dido and Aeneas, and I am sure that the theatrical community put on several plays. The two groups joined forces in at least two highly successful revues. I also remember a madrigal-singing party on the bank of a canal on a perfect summer evening. When Katharine and I moved to an old house on the Wading Street in Stony Stratford, we would often fill our paneled drawing room with musicians who would perform as the spirit moved them. On these informal occasions I would carry in enormous jugs of draught beer from a pub across the road. Incidentally, our Stony Stratford house, with its Queen Anne front and a much older rear portion, was halfway between two famous pubs, the "Cock" and the "Bull." It is said that the term "cock and bull story" originated here, as someone in one pub would recount what he had heard from someone in the other pub, who had no doubt got his information from someone in the first pub, who had ... In The Bletchley Park grounds we had some extensively used hard tennis courts. There were a few very good players--one had been at Junior Wimbledon--and many lunch hours were enlivened by exhibition games. Shortly after WE day, matches were arranged between British and American teams at cricket and baseball. Appropriately enough the United States won the cricket and the United Kingdom won the baseball. Most of the sections held dances at regular intervals. The Hut 6 Dance Committee was particularly efficient and the dances they arranged, about three or four times a year, were very successful. A particularly memorable dance, arranged for early June 1944, caused a nice problem in security. Leave had been canceled and everyone knew that D-Day would be at the end of May or the beginning of June, but in Hut 6 only Mimer-Barry and Fletcher knew that the planned date was June 5. The dance had been planned for the evening before the D-Day operations were due to start, and Fletcher asked Mimer-Barry if it was all right to proceed; no one wanted to do anything to prevent the utmost effort being available for the period of the invasion of France. After consideration Mimer-Barry said yes, and how right he turned out to be. He was afraid, of course, that if he said no he would be giving a strong indication that D-Day was planned for June 5. In the event, the invasion had to be postponed twenty-four hours anyway, because of the weather, and so took place on June 6, giving ample time for the dissipation of any hangovers acquired at the dance. From this account of some of our social activities, it will be evident that we had play as well as work. Even during working hours, the fact that most of us were involved in friendly teamwork meant that we were in an atmosphere conducive to good comradeship and the development of lasting friendships. There were many Bletchley marriages. Yet many people at Bletchley were, to say the least, disturbed by the contrast between our happy existence and what was going on elsewhere. In impotent rage we watched the flames of Coventry. We saw huge bursts of flame over London during the big air raids. Those of us who knew through the Enigma decodes what the Germans were saying about their military successes had some faint idea of what hell our troops went through, particularly in the early days when they were so completely out matched by German preparatory planning. No wonder many of the brightest young men in Hut 6 pleaded to be allowed to get into the fighting, and were disappointed to be turned down because, knowing our secrets, they could not be permitted to come within the enemy's reach! It was a time of strange contrasts. When our third child, Rosamond, was on the way Katharine left the A IS and went to live in Rippington Manor, a medieval priory in the village of Great Gransden that we had bought just before the war started. We had to share our territory with a good many Italian prisoners, who were quartered in Nissen huts in the grounds and probably worked on farms. Nevertheless the old manor, with its walled garden sloping down to what must once have been the monks' carp pond, was and still is one of the most peaceful places I have ever known. Fortunately it was near the railway line from Bletchley to Cambridge. The Bletchley Park staff were encouraged to take a day off once a week, to keep fresh. Even in the war there were four or five trains each way on this line on weekdays and two on Sundays.* I could take an evening train from Bletchley and return the next evening or the following morning. I left a bicycle at the nearest station, Gamblingay, and as I came down the hill into Great Gransden I would often hear the old carillon in the church tower playing one of its tunes. The whole atmosphere seemed so peaceful and so remote from what we were thinking about at Bletchley. For others, however, the contrast between the peace of the village and their war work was far greater. Just a few miles away was an airfield from which bomber crews would fly at night into the hell of antiaircraft hre over Germany. I never had to endure one of those nights in London during the period of heavy bombing, when the underground stations were used as sleeping quarters. But toward the end of the war, after the A-bombs had been introduced, I had to go to London several times. I was full of admiration for the way people just carried on. Once, having spent the morning in the communications room at the Admiralty, I went into a pub in Whitehall for a sandwich lunch. The drone of a B-1 flying bomb was heard through the open door to the street, but nobody seemed to take any notice, and I followed suit. On another occasion I was at a meeting in May fair when a V-l was heard; the only reaction was that "perhaps we had better move a little further away from the window." I he striking contrast between our life in Bletchley Park and that t Londoners was brought home to me after the war when I joined 1 here are none today the John Lewis Partnership and talked with Michael Watkins, t man who had run the business throughout the war. By keeping t Partnership operating in spite of bomb damage, Watkins perform a valuable wartime service. Moreover he was, I believe, the principal originator and organizer of the "utility goods" scheme undi which the manufacturers of a variety of merchandise, including clothes, furniture, and china, agreed to produce only a limited! number of designs. The selected designs were good, and the schemes enabled people to meet their essential needs at the lowest possible prices while releasing a considerable part of the manufacturing labor force for other work. Discussing our respective experiences, Michael Watkins and I found many differences. I lived in quiet surroundings that werej never subjected to anything more than trivial accidental bombing. (I do not remember hearing of anyone's being hurt.) When I had to travel any distance on business, I was provided with a car and a female chauffeur. Because of the importance of my work I was excused from Home Guard duty. Watkins, though a man of some prominence doing an important, exacting job, was shown no such consideration. He lived at Cookham on the Thames, in normal times a railway journey of about an hour from London. At the end of each day's work at his office in Cavendish Square near Oxford Circus, he had to struggle across London to Paddington Station in the hope that a train would take him home. The journey was often nightmarish, delayed either by bomb damage or by actual bombing attacks. Yet when he got home to Cookham he would do his full | share of Home Guard duty. I Certainly our work at Bletchley Park was important. Still, our ; lives were very easy in comparison with those of civilians like Watkins who were doing important jobs under fire. Furthermore, no matter how hard some of us may have worked, we made no sacrifices that could remotely be compared to those made by the men who were doing the actual fighting on the land, on the sea, and in the air. In my own case exemption from military service involved a curious sequence of events. At the beginning of the war, when I became a temporary civil servant in a branch of the British Foreign Office, I was thirty-three years old. In due course my age group was called up, and I received a notice telling me to report to a unit of the Royal Artillery somewhere in the north of England. I took the notice to the Foreign Office administrative people. They assured me that they would handle the matter, and that I was to do nothing. A little later I received a polite letter from the Colonel of my artillery unit, saying that there was no doubt a good reason for my nonappearance, but would I please report at once. I took this letter also to the administrative office. Again I was told that the Foreign Office would handle the matter. The next development was a phone call to my mother-in-law, from her brother, Ned, who as chance would have it was Chief Constable of Buckinghamshire, the county in which I was living. He had a warrant for my arrest. This raised an intriguing point. Bletchley Park was enclosed by a high fence and was under military guard. Its cafeteria was open night and day, and sleeping accommodation was available. Suppose I had kept on living and working there and never emerged? I suspect the police might have had some difficulty in arresting me. As it turned out, however, I did not need to resort to any such dramatic delaying action. The Foreign Office and Army Administrators tin all}' resolved the matter. One problem remained. Army regulations included no means of simply letting go of a man who had been called up but had not enlisted. The regular discharge procedure applied only to those who had gone through the enlistment process. It developed that, in order to sever my relationship with the Gunners, I would first have to enlist. I had to report at a Royal Artillery establishment, and it was arranged that I should go to the nearest one, which was a few miles south of Bletchley Park. I was given a gasoline allowance and drove my own car. The "establishment" turned out to be a small office presided over by a sergeant. I he sergeant had received detailed instructions, and after filling in a few forms, he shook me by the hand, congratulated me on being a Gunner, and said that he would arrange for me to be discharged some other day. When I explained that my office did not want me to take time off for a second trip, he said that he could not discharge me at once because a medical examination was needed and the doctor would be at lunch. I had to get back to Bletchley as soon as possible, so I discovered where the doctor lived, dashed round and just caught him before I went to lunch. A few minutes later I had whatever medical cert if cafes were necessary for my discharge, only to find that the sergean| had gone to lunch. I found him in the nearest pub and persuade him to come back to his office. After filling out a few more forms, he told me that I was now a civilian again. My length of militar service was almost exactly twenty minutes. Then, having arrange my discharge, the sergeant gave me a few appropriate papers, one which I treasured for many years. It urged me to join the Homd Guard, where my experience in the Army would be extremeh valuable. So much for the lighter side of my wartime experience. In the last part of my book I will be talking very seriously about the dangers that we face today. In my view we are in a real mess. Something; needs to be done urgently if what we value most is to survive. PART i=cim Addendum rADDENDUM / 195 Reprinted by permission from vt' i<." and Lalionul Secuniv, vol. 1, "Y . I published by Frank C ass & Company, 900 Eastern Avenue, Ilf'ortl, Essex, England. Copyright Frank Cass & Co Ltd. From Polish Bomba to British Bombe: The Birth of Ultra ihl; poles, the british and the french Until just before the Second World War a small Polish team of three muthematician-cryptologists, headed by the brilliant Marian Rejewski, had been happily breaking the German military cipher machine, the Enigma, for many years. A small British team under the First World War cryptanalyst, Dilly Knox, was near to success, but was foiled by failure to make a guess which, in retrospect, seems an obvious one. The French cryptanalysts do not appear to have tried, but Captain Gustave Bertrand, involved in French espionage, achieved a coup without which the Polish breaks and the subsequent British successes might never have been achieved. The Poles kept their secret to themselves until July 1939 when, with the German invasion of their country imminent, they gave all their knowledge, as well as working replicas of the Enigma machine, to the French and British. In England, at Bletchley Park, we were quick to exploit the golden opportunity that the Poles had handed us. In France Bertrand established an organisation near Paris, code-named Bruno, at which Marian Rejewski and his associates Henryk Zygalski and Jerzy Rozycki, having escaped from Poland, continued to work on Enigma. There was collaboration between Bletchley and Bruno until the German advance on Pans forced an evacuation. Ultra intelligence, based on decodes of the Enigma traffic of the German army and air force, was born early in 1940. Its heyday started after the battles of El Alamein and Stalingrad in the autumn of 1942 and the Allied landing in North Africa in November 1942. By that time the Allies had become strong enough to take advantage of this extra ordinarily prolific source of intelligence, and a satisfactory means of sending Ultra information to commanders in the field had been developed. As early as 1941 and 1942 Rejewski wrote reports in Polish on the pre ^ruf)pen) were used in the preambles of messages to indicate which crypto net was involved. I used the term 'machine setting' to mean the positions of the three wheels in an Enigma Scrambler that is already set up in accordance with the daily key of a crypto net.27 The encoding of an individual message involved two machine settings: 1. A 'text setting' at which encipherment would start 2. An 'indicator setting' (Grundstellung) used to encipher the text setting. Rejewski calls the text setting a 'key', or 'message key'. He calls the indicator setting the 'basic position'. I will stick to my terminology here. When I came into the picture in September 1939 I was told how the Germans were using their Enigma at that time. The daily keys for a month were issued to all units of a crypto net. Individual messages were enciphered at text settings chosen by the originating operator. These text settings were enciphered twice at an indicator setting also chosen by the operator. The indicator setting was transmitted in the preamble of the message, and the double enciphered indicator setting, which I called the 'indicator', was transmitted as the first six letters of the enciphered text. The receiving operator, with his machine set up to the same daily key would set his wheels to the indicator setting, decipher the indicator, and so obtain the text setting repeated twice. He would then set his wheels to this text setting and decipher the message. Let me reiterate that Dilly Knox never told me about Rejewski's work, which I will describe in part V. I and the people who joined me in Hut 6 knew that the Poles had given us a replica of the Enigma, and we soon had modified Type-x machines that operated like German military Enigma machines. We knew Zygalski's principle of perforated sheets. We knew the purpose of the discriminants and the way in which the units of a German crypto net would set up their Enigmas to a daily key. And we knew the indicating procedure used at that time for the encipherment and decipherment of individual messages. That was all! But it was enough. Knowledge of the other methods that had been used by the Poles would not have helped us. What really mattered was the machine itself and the stimulus that came from knowing that the Enigma traffic could be broken. REJEWSKI'S BRILLIANT CRYPTANALYSIS Six Successive Periods At the end of his recent article Stengers remarks: What the Poles themselves did was the result, primarily, of: Rejewski's creative spirit. But Rejewski himself could have^ remained impotent without the Asche documents, and the Aschei documents were just a bit of luck. A bit of luck and history is changed. In my chapter, "A Comedy of Errors', I discuss the principal German errors that led to our success and the simple ways in which we could have been defeated. I concluded with the remark that "We were lucky Rejewski and his two colleagues had different problems but similar luck. At one point29 he remarks that a German slip-up (mixing plain text with code) made it possible to break into the Enigma traffic of the Sicherheitsdienst or SD, an achievement that, combined with another German error, led to the recovery of the wiring of wheels IV and V. He also remarks that the Germans would have been better off if they had not enciphered their text settings. To get rid of misconceptions we must distinguish between the very different situations that existed in six successive periods, as follows: 1. From the turn of 1927/28 to September 1932, when Rejewski was assigned to work on Enigma. 2. The brief period from September 1932 to the turn of 1932/ 33 during which Rejewski, working in isolation, broke the German military Enigma. 3. The relatively quiet period from early 1933 to the end of 1935, during which Rejewski and his team achieved regular breaks of daily keys. 4. From 1 January 1936 to 15 September 1938, a period during which the work load increased, due in part to the introduction of new crypto nets for which daily keys had to be recovered. 5. The three months from 15 September 1938 to 15 December 1938 during which the Polish team were dealing with a major change in the German indicating procedure. 6. From the German introduction of two new wheels, IV and V, on 15 December 1938 to the Pyry disclosures to the French and British of July 1939. I. Turn of 1927/28 to September 1932 The interest of the Polish Cipher Bureau in Enigma was aroused at the turn of 1927/28 when, on a Saturday afternoon, a package from the German Reich arrived at the Warsaw Customs Office. According to the accompanying declaration it contained radio equipment. The German firm's representative demanded very strenuously that the package be returned to the Reich even before going through Customs, since it had been shipped by mistake. The customs officials were suspicious and notified the Cipher Bureau, which was interested in new developments of radio equipment. The package could not be returned until after the weekend, so personnel from the Bureau had plenty of time to investigate. They carefully opened the box and found that it contained a cipher machine. They examined the machine minutely and carefully closed the box again."" Rejewski insists that this Enigma was a commercial model with no steckerboard. The military model had not yet been put into use. Indeed the first machine-enciphered messages did not appear on German military radio nets until 18 July 1928. The importance of the incident lay in the fact that it revealed German interest in the Enigma. The Polish Cipher Bureau bought one of the commercial machines. At the turn of 1928/29 I they organised a cryptology course in Poznan for mathematics students who were fluent in German. This proved to be good thinking. 2. September 1932 to the turn of 1932/33 The second period started on 1 September 1932 when three students who had attended this course, Rejewski, Zygalski and Rozyki, were hired to work permanently at the Cipher Bureau in Warsaw. Rejewski was soon separated from his two colleagues, given a separate room, and instructed to study Enigma. (Earlier studies had been abandoned.) The commercial machine was placed at his disposal, but was of no assistance. Each day he > was given several dozen messages enciphered on the military machine. At that time the German method of using their Enigma was very; different from that with which Hut 6 was to be faced seven years later. Of the items in the daily key the wheel order was only changed once a quarter | and only six pairs of letters were steckered, leaving 14 unsteckered. Thel indicator setting was specified as another item in the daily key. From 1 January 1936 the wheel order was changed once a month; from 1 October 1936 it was changed daily, and the number of sleeker pairs, instead of| being fixed at six, began to vary from five to eight. The indicator settingj continued to be part of the daily key until 15 September 1938, after which! date the indicator setting for each message was chosen by the operator. It! was fortunate that Rejewski was set to work before these improvements! in procedure were made. The fact that the Germans continued for so long to use the sam el indicator setting for all messages on the same daily key, an extraordinary! error, was brilliantly exploited by Rejewski. It became apparent to him! that the first six letters of each message text, that I call the 'indicator',31! were obtained by enciphering the three-letter text setting twice at the| same indicator setting. In other words, for all messages on the same daily key the six-letter indicators were the result of encipherment at the same sequence of six positions of the Enigma. Rejewski denotes the letter permutations produced by the Enigma in these six positions by A, B'C, D, E, F. He started by writing the six-lette indicators of all messages on a daily key underneath each other. All the indicators that had the same first letter also had the same fourth letter, obtained from the first by the product AD of permutations A and D. The same applied to the second and fifth, and to the third and sixth, involving the products BE and CF. His next move was to have far-reaching results. Starting with any one of the indicators he wrote down the first letter, say d, and next to it the fourth, say v. He then sought out another indicator that had v as its first letter and wrote its fourth letter, say p, next to d and v. Thus, if three indicators were: d m q v b n v o n p u y p u c f m g he would obtain the sequence: d v p f continuing this process he would obtain a cycle of letters such as: dvpfkxg z yo which was closed by the fact that, if the first letter of an indicator was o, the fourth would be the starting letter d. The remaining indicators would give further closed cycles of the permutation AD. The same procedure would be applied to BE and CF. giving three sets of cycles such as: AD = (dvpfkxgzyo)(eijmunqlht)(be)(vw)(a)(s) BE = (blfqveoum)(hipswizrn)(axt)(cyg)(d)(k) CF = (abvikrigfcqny)(duzrehlxwpsmo) This simple method of representing the permutations AD, BE and CF. which Rejewski hit on right away, turned out to be of immense importance. He found that the composition of the cycles was different each day. Later on he would call the three sets of cycles the characteristics' of the permutations AD, BE and CF. Rejewski quickly developed a mathematical theory of these characteristics and, by guessing that some German operators might select three identical letters, such as aaa, bbb, for their text settings, he was able to recover the six permutations A, B'C, D, E, F and so determine the text settings of all messages on the same daily key. In fact, without knowing either the internal connections of the wheels or the set up of a daily key. ie had broken the indicating system. To form the three complete characteristics of a daily key all he needed was about 60 messages on that he >. It sounds incredible, but it happened Still working in isolation, Rejewski's next step was to develop a mathematical representation of the working of the Enigma machine. He was hoping that the knowledge of permutations A to F would enable him to work out the wiring of the wheels. He had reduced his problem to a set of six equations involving three unknown permutations, and he was wondering whether they could be solved, when, on 9 December 1932, at just the right moment, he was given four documents. He did not know it at the time, but these documents had been obtained by Bertrand from the German traitor Asche. The four documents were a Gebrauchsanweisung (operating instructions), a Schlusselanleitung (keying instructions), a table of daily keys for the month of September 1932, and a table of daily keys for the month of October 1932. It was extremely fortunate that the two months, September and October, occurred in different quarters, during which different wheel orders were in use. The known daily keys for each month, combined with the equations he had developed, enabled him to work out the internal wiring of the two different wheels that appeared on the right. Finding the connections of the third wheel and the umkehrwalze presented him with no great difficulties. For each wheel he was able to determine the correct torsion of the sides with respect to each other and the turn-over positions of the alphabet rings. The establishment of all these details depended on attempts to read several messages. Fortunately each of the two monthly tables of daily keys included a sample plain text and its encipherment at a stated daily key and text setting. At one point in this recovery process Rejewski had seemed to be near < defeat. In formulating his equations he had assumed that the wiring of the entry wheel (equivalent to the connections of the 26-way cable in my Figure 3.612) was known to him. At first he assumed that the connections were the same as those of the commercial Enigma, in which the terminals of the scrambler's input-output ring were connected to the terminals of the keyboard and lamps in the order in which letters appeared on the keyboard, namely: qwertzuioasdfghjkpyxcvbnml : This assumption, being wrong, was getting him nowhere, when, at thej end of December 1932, or perhaps in the first days of January 1933, he| wondered whether the Germans might have used the alphabetic sequence a b c ... z instead of the keyboard sequence q w e ... 1. This inspired guess proved correct. The very first trial yielded a positive result.j Then, from his pencil, as if by magic, began to issue numbers designating the connections of the right-hand wheel." Now that the wiring was known, the Poles could modify their commercial Enigma to operate like the German military Kmgma. At (his point Rejewski was allowed to initiate Zygalski and Rozycki so that, using the daily keys supplied by Bertrand, they could decipher messages that had been intercepted during September and October 1932. Rejcwski continued (o work in isolation on the third part of his task. He had broken the system of enciphering text settings. He had worked out the electrical and mechanical details of the military Enigma. He still had to find a means of recovering daily keys, and this, as he modestly says, was 'hardly easy'. He developed what he called the 'grill' method. It was very laborious, hut. combined with the fact, shown up by decodes of September and October 1932. that many plain texts began with the three letters A N X. it was effective. Thus in January 1933 the Poles were in business. They were able to decode current Enigma traffic. What an extraordinary achievement for a period of a little over four months! And what a brilliant one man triumph for Rejewski! j. Far/y /9jj (o f/zf end of /9J3 When Rejewski reported his success, the Cipher Bureau ordered a series of replicas of the German military Enigma to be built. (Rejewski calls them 'doubles'.) Then five or six young persons were hired to decipher intercepted messages, the keys to which were soon forthcoming. Zygalski and Rozycki were assigned to work permanently with Rejewski. Day by day the daily keys were recovered by this three-man team. For three years, until the end of 1935, the Germans introduced no essential changes, so time could be devoted to improving methodology. In his mathematical representation Rejewski had used letter Q to denote the permutation produced by the middle wheel, left-hand wheel and Mm^f/zrtva//zf. A catalogue was made of all the possible permutations Q so that, when the early stages of the grill method had determined the setting of the right hand wheel and the permutation O. reference to the catalogue would at once reveal the settings of the other two wheels. Ro/ycki worked out a 'clock method' which, in many cases, could determine which wheel was on the right. This became important when changes of wheel order occurred monthly, and then daily, instead of quarterly. At some point German operators were forbidden to use text settings consisting of three identical letters, but they developed other habits which still allowed the Polish team to determine the permutations A to F. "? / /dmmn /V.?6 m /J &?prfm6 v /9.?g The fourth period saw two very important developments: the invention of the Cyclometer and the breaking of the SD Enigma traffic. Already, from August Wx the German air force had created its own crypto net with its own daily keys. Gradually other military and paramilitary organisations joined in, forming additional crypto nets with their own daily keys. This implied a great increase in the work load of the three-man team. Furthermore, from 1 October 1936, five to eight pairs of stecker were used, which made it difficult to apply Rejewski's grill method. It became necessary to look for other methods. Thoughts went back to the characteristics that Rejewski had investigated in September 1932. It dawned on the team that, if they could make a complete catalogue of all possible characteristics, the recovery of the three characteristics of a new daily key, achieved as usual by an analysis of indicators, would lead very quickly to the complete recovery of that daily key. The time-consuming demands of the grill method would be avoided. Rejewski then invented a machine, the Cyclometer, that would permit the construction of this catalogue. It consisted of two Enigma scramblers with the right-hand wheel of the second scrambler displaced three positions with respect to that of the first scrambler. The overall layout is shown in his Figure E-4, the interconnections of the two scramblers in Figure E-5.u The cycles of a characteristic always occurred in complementary pairs of equal length. When current was turned on at any letter, all letters in the same cycle and in the complementary cycle would be shown up by lamps. This is shown in Figure E-5, in which the 'reversing drum' represents the permutation Q, which is assumed to remain unchanged. When a switch puts current into the first (left-hand) scrambler at position 1, it returns at position r, enters the second scrambler at this position and returns at position w and, after going through the second scrambler unit again, returns to the starting position 1. Thus, for the particular setting of the wheels that is being tested, we have two complementary cycles (lz) and (r,w). Using other switches reveals the other pairs of complementary cycles. The cyclometer had a rheostat because the number of lamps to be lit would vary. In September 1932 Rejewski obtained three characteristics for the permutations AD, BE and CF involved in the double encipherment of J text settings. Now, for each of the six wheel orders and each of the 26 x 26 x 26 possible positions of the first scrambler, the cyclometer would enable him to obtain a characteristic. The cycle lengths of each characteristic would be entered on a card, probably in order of decreasing magnitude, and the 6 x 26 x 26 x 26 cards so obtained would be arranged i as a card index. Once this major task was accomplished, the recovery of a new daily key would be greatly simplified. The first step, as before, would be to accumulate enough intercepted messages on a new daily key to permit the construction of the characteristics of the three permutations AD, BE and CF. These three characteristics would be looked up in the card index and. as a rule, the break would he completed in l(>-20 minutes." Rcjewski does not say how the turn-over problem was handled, but. except in the worst possible case, at least one of the permutations AD. BE. CF would be unaffected by turn-over and this would greatly reduce the labour involved in completing the break. The Polish team had a severe setback when on 2 November 1937. soon after the card catalogue of characteristics had been completed, the Germans introduced a new umkehrwaltze. At about the same time, however, luck smiled on them once again. In September 1937 a new crypto net had appeared. It was used by the Sicherheitsdienst. or SD. The usual methods provided the wheel order, indicator setting and stacker but attempts to complete a break by looking for messages that began with the letters A N X failed. The ring settings could not be determined. This problem arose from the use of a four-letter code to represent the clear text before it was enciphered on the Enigma. But the Germans enciphered the word 'ein~. for which it seems that there was no four-letter code. With perhaps one and a half ounces of luck this was spotted. The Polish team proceeded to break the four-letter code, and the breaking of the SD crypto net became routine. 5. The three months 15 September 1938 to I? December 1V38 The change of umkehrwaltze on 2 November 1937 was nothing compared with the blow that hit the Polish team in September 1938. Every single technique that they had used until then had depended on the inexplicable German error of using the same indicator setting for the double encipherment of the text settings of all messages using the same daily key. Suddenly, on 15 September 1938, new regulations called for the arbitrary selection by each operator of the indicator setting for each message. The doubly enciphered text setting the indicator still appeared as the first six letters of the enciphered text. The chosen indicator setting was included in the preamble. The response of the Polish team to this change was amazingly rapid. Within a few weeks Rejewski estimates one or two they developed ways to realise two new ideas. Furthermore and this was a real stroke of luck they could still read traffic on the SD crypto net which had not yet applied the new procedure. There were two inventions the 'bomba' by Rejewski and the 'perforated sheets' by Zygalski. Of these the bomba was a development from the cyclometer, which grew out of Rejewski's very early discovery of the characteristics. Its importance has been greatly exaggerated. Thinking about the characteristics now became focused on whether or not a characteristic contained a pair of single-letter cycles. In the example that I have quoted from Rejewski, the characteristics of AD and BE do, whereas that of CF does not. When the characteristic of AD does contain single-letter cycles, it is possible for the same letter to turn up in the first and fourth positions of an indicator. Otherwise it is impossible for this to happen. Similarly for BE and CF. Until I read the Rejewski papers I thought that the term 'female' for an indicator in which the first and fourth, or the second and fifth, or the third and sixth letters were the same, had been introduced at Bletchley Park. It now seems that an equivalent term was introduced by the Poles. I also thought that Turing's ideas for a British bombe must have been based on what he had learned of the Polish bomba. Again I was wrong, because I had no knowledge of what the bomba was designed to do. It is strange that Kasparek and Woytak, with all the evidence available to them, should have endorsed this mistake of mine in Appendix F of Kozaczuk's book. When the Germans made their operators responsible for selecting the indicator setting for each message, the Poles were no longer able to; recover the characteristics, but, for each message with its chosen indicator setting, the occurrence of a female in the indicator would in dica that the corresponding permutation, AD, BE or CF, had a characteristic! that contained single-letter cycles. This was the basis of both inventions. Rejewski says:16 Given sufficiently ample cipher material it may happen that on a given day there will be three messages with keys (indicator settings and indicators in my terminology) such as: R T J H P L D Q X W A H RAW D W J W I K K T W MWR The bomba depended on the assumption that the letter W wa unsteckered. It required little engineering development because it wa essentially three cyclometers, set to the relative positions at which females had occurred, and then turned automatically (as opposed manually in the generation of the catalogue of characteristics) through! possible positions." In each position there would be a simple automati test of whether (assuming letter W to be unsteckered) the three femald| could have occurred. Six bomby were quickly built, one for each of the s possible wheel orders. Rejewski says that they worked reasonably welli long as the number of sleeker pairs varied from 5 to 8. From 1 January, 1939 this number was increased to 7 to 10, and the bomby be can ineffective. By the time the British became involved the number stecker pairs had settled at ten. Thus the idea that the British would have wanted to copy the bomba is sheer nonsense. Zygalski's invention of a method based on perforated sheets was a different matter.3* It involved no assumptions that certain letters would be unsteckered, and it was immediately copied by the British after Pyry. Because the indicator setting was now chosen by the operator for each message, the characteristics AD, BE and CF would no longer apply to all messages on the same daily key. It would no longer be possible to recover them. On the other hand the occurrence of a female such as: K I E S P E S N T (a 1-4 female) would mean that the AD characteristic associated with the indicator setting K I E must contain a pair of single-letter cycles. The same would apply to the BE and CF characteristics of females such as: R Y M X W N P W V (a 2-5 female) LTS VBY QGY (a 3-6 female) What was needed, therefore, for each wheel order, was a catalogue of indicator settings, whose characteristics contain single-letter cycles, and a means of comparing females appearing on the same daily key with this catalogue. It was found that about 40 per cent of the characteristics, shown up by the cyclometer and recorded in the card index, contained single-letter cycles, and Zygalski found a way of making the comparison. For each of the six possible wheel orders a paper sheet was used to represent each of the 26 possible positions of the left-hand wheel. On each sheet a large rectangle was divided into 51 x 51 small rectangles. The two sides, the top, and the bottom of the large rectangle were lettered 'a' to 'z', and again 'a' to 'y'. This provided a co-ordinate system in which the little rectangles corresponded to positions of the middle and right-hand wheels. In each small rectangle a hole would be perforated if the characteristic of the corresponding setting of the three wheels contained a single-letter cycle. Each such occurrence would call for as many as four perforations. Given enough females on the same daily key, it was possible to stack sheets on top of each other in accordance with the indicator settings that had given rise to the females. The number of visible apertures would steadily decrease and any left open would represent ring settings that would permit the females to occur. Note that, because the turnover notches were on the alphabet rings, it would be known whether a turnover of the middle wheel would occur between the encipherments of the two identical letters. Consequently any female that would involve a turnover could be discarded. It is perhaps of interest to note that, when I thought of the same 1Qlea, independently, I knew nothing of Rejewski's characteristics. In my attempted reconstruction of my thinking process at the beginning of my Chapter 4, I realised, as Step 3, that it was not always possible for the Enigma to produce the same letter pairing in two positions three places apart in its cycle. This is equivalent to saying that a characteristic will not always contain a pair of single-letter cycles. From then on my thinking was probably very similar to Zygalski's. Rejewski and his team had to do the enormous job of cutting about a thousand apertures in each sheet, and they cut them with razor blades! They needed six series of 26 sheets each for the six possible wheel orders and the 26 possible positions of the left-hand wheel. By 15 December 1938 they had made only two series. On that day the Germans started using two new wheels IV and V, which had been issued to all formations, including the SD.4" This meant 60 possible wheel orders, calling for 60 series of perforated sheets. 6. 75 December 1938 to July 1939 The quick recovery of the wiring of the two new wheels by cryptanalysis the achievement that I found hard to believe was made possible by yet another major German error. The SD crypto net introduced the two new wheels on the appointed date, but it was still using the old system of enciphering text settings, which had been abandoned by all other crypto nets on 15 September 1938. Knowing this, the team found a day on which the right hand wheel was on of the original three and applied Rejewski's original grill method. Then, assuming that the left-hand and middle wheels were a known one and an unknown one, they found the connections of the latter wheel in the same way that Rejewski had found the wiring of the wheel that had not appeared on the right in September or October 1932. Knowing the wiring of all five wheels, the Poles, still a three-man team, continued to read messages of the SD crypto net.41 This reading was intermittent because although Rozycki's clock method sometimes revealed which wheel was on the right, the grill method the only method of breaking still applicable sometimes failed because from 1 January 1939 the number of sleeker pairs had risen from seven to 10. This increase in the number of sleeker pairs would have reduced the effectiveness of The bomba, and in any case limited Polish resources prohibited The conslruclion of 54 new bomby lo deal with the additional wheel orders. The preparation of 58 more series of perforated sheets wa another virtually insoluble problem. Il was only possible to read military messages when the three original wheels happened to be in use in the t combinations covered by the existing perforated sheets. When, on 1 July, 1939, the SD crypto net shifted to the new indicating procedure, the of grill method ceased to be effective there too. This was the situation in July 1939, when the Poles told Alastai Denniston and Dilly Knox all they knew. What I knew of what happened after my arrival at Bletchley Park at the outbreak of war is outlined in The Hut Six Story. I still feel very strongly, as a result of some 20 years of research on today's military problems, that this story contains many valuable lessons that are sadly neglected. But my account has two serious gaps. I still do not know what Dilly Knox achieved before Pyry, or what he was up to in August 1939, assisted by Twinn, Kendrick. Turing, and a few other assistants. Nor have I found it easy to obtain reliable information about the wartime collaboration between Dilly Knox and Rejewski's team when they were operating first at Bruno, near Paris, and later, under the code-name Cadiz, near the coast of Vichy, France. From what I have now learned from Polish writings, it seems to me that the achievements of the brilliant Dilly Knox have been belittled in his own country. Even if his crypt analytical methods of around 45 years ago still need to be concealed, which I doubt, it seems strange that a broad outline of what he actually achieved, and might well have achieved with a little luck, is still subject to veto by today's Government Communications Headquarters. VI DILLY KNOX AND BRUNO Marian Rejewski formed a very high opinion of Dilly Knox at the meeting in Pyry in July 1939. In a conversation with Woytak he said: Just how much Braquenie understood, I don't know; but there is no question that Knox grasped everything very quickly, almost as quick as lightning. It was evident that the British really had been working on Enigma. So they didn't require many explanations. They were specialists of a different kind.J: Rejewski also said: I have the fullest grounds to believe that the British cryptologists were unable to overcome the difficulties caused by the connections in the entry drum. When the meeting of Polish, French and British cipher bureau representatives took place in July 1939, the first question that the cryptologist Dilwyn Knox asked was: What are the connections in the entry drum?41 Like all the statements that Rejewski made from his own personal experience, these two ring true. And they make one wonder whether Dilly had actually thought of all the major Polish ideas and had been held up only by failing to make Rejewski's guess that the permutation of the entry drum might be identity. It seems to me entirely possible that Dilly went through much the same thought processes as Rejewski and his team. He could certainly have discovered the characteristics, just as Rejewski did, and he could have gone on to break the indicator system. It is said that Bertrand brought some of the Asche documents to England, so Dilly may have received the four that were so helpful to Rejewski. In that case he could have been well on the way to recovering the wheel wirings, but prevented from doing so only by failure to guess the wiring of the entry drum. After the Germans stopped using the same indicator setting for all messages on the same daily key, it is possible that Dilly would have thought of Zygalski's idea of using perforated sheets to provide a catalogue of wheel settings that could produce females. After all, I had the same idea myself, and I had no previous experience of cryptanalysis. Having female indicators on his mind, Dilly could well have thought of associating two Enigma scramblers set three positions apart, as was done in Rejewski's cyclometer. With Turing around it is quite possible that, before the Pyry conference the idea of mechanising the movement of the two scramblers would have emerged. The Polish bomba was essentially a combination of three mechanised cyclometers, and the idea of driving a set of scramblers automatically through all possible positions is the only feature of the Polish bomba that was used in the British bombe. It is even conceivable that, before Pyry, Turing and Dilly may have begun to think of using a larger battery of scramblers, not just to handle three females with the assumption that the letter involved would be self-steckered but to handle textual cribs with no such assumption. But, of course, the ideas of the perforated sheets and the battery of automatically driven scramblers could not be exploited without a knowledge of the wiring of the Enigma wheels, which was provided by the Poles at Pyry. The development of the British bombe involved four new ideas, descriptions of which will be found in The Hut Six Story: 1. Loops derived from a crib (p. 79) 2. The double-ended Enigma scrambler (p. 297) 3. The diagonal board (p. 304) 4. Taking advantage of the 'filling up' of the test-register (p. 301) In addition the design engineer. Doc Keen, used what I call 'drums' instead of wheels to get greater speed.4' Lisicki has confirmed that none of these ideas were known to Rejewski. In a letter to me Lisicki said: None of the ideas which you listed were Rejewski's. He was happy with the sheets and discarded the idea of improving his bomba. The loops, the double-ended Enigma, the diagonal board, and the filling up of the test register were all British ideas, and Rejewski in his letters to me several times mentioned that he had no idea how to mechanize the search for the keys and thought that the British mechanized the sheets, but that would be useless after May 1940. In Brunohe had absolutely no time for creative work. The running of a number of Enigma scramblers was the only idea which the Poles first used and perhaps was born from the Cyklometer. My suggestions of what Knox may have done before Pyry are pure conjecture, based on Rejewski's feeling that he must have done a lot. Kozaczuk and Kasparek had no right to assert, as they appear to do, that none of the Polish ideas had been thought of by the British. Furthermore, as Deavours says in the review of the Kozaczuk book that I have quoted, their thesis that 'virtually all major crypto logic techniques that the British used to break Enigma in World War II had been thought of and used by the Poles earlier' is simply not true. On the other hand the Poles did give us the wirings of the Enigma wheels, and I still maintain that, had they not done so, British breaking of the Enigma might well have failed to get off the ground. Indeed, it is deplorable that the official history of British Intelligence in the Second World War has tried to establish that the Polish contribution had little effect. More about this in the next section. One would like to know how far Dilly had got with his Enigma studies before Pyry, what he did between Pyry and the outbreak of war, and also what he achieved in the Cottage up to his death in February 1943. It is public knowledge that his organisation, ISK (Intelligence Services Knox). broke an Italian naval cipher system and an Abwehr system both of which used Enigmas different from the German military version. He was probably in touch with Turing's work on naval Enigma. But at the moment I am concerned with matters relating to Huts 6 and 3, and I have found it very hard to determine what Dilly was doing while Hut 6 was getting established. This is no doubt partly due to Bletchley's wartime policy that individuals should know only what they needed to know, but also to Dilly himself. Babbage, who joined Knox around Christmas 1939, has said in a recent letter to me: I gradually got to understand the Enigma machine and the problems it posed, but this was mainly through people like Twinn and Kendrick. Dilly was a most entertaining person, but definitely not very informative, as you found. Indeed, it is probable that even Twinn and Kendrick were not fully aware of what Dilly was up to. It is certain that, when I was banished from the Cottage to work in the School, Dilly had told me nothing at all about Pyry, about the perforated sheets that were already being punched, or about Enigma breaks that had been attempted in the Cottage. Nor was I told about the collaboration with Rejewski's team at Bertrand's Bruno, which seems to have been established in November 1939, when Travis and I were setting in motion the build-up of Hut 6.1 am inclined to think, but have no supporting evidence, that this collaboration was handled by Dilly himself, or by ISKif it existed at that time, and that he was, as usual, 1 'not very informative'. In trying to sort out the sequence of events, I have come to believe that > Turing took a complete set of the perforated sheets that Jeffreys produced to Bruno on 17 January 1940 and spent a few days with RejewskM and his team. The anecdotes that Rejewski relates about the farewell! supper given before Turing returned to England are convincing.45 Forj example, Zygalski wondered why each little square in the British version! of his perforated sheets had so peculiar a measurement about eight and! a half millimetres on a side. "That's perfectly obvious', said Turing,} laughing. "It's simply one third of an inch." From my memory I would have guessed a later date for the availabilit of the Jeffreys sheets in Hut 6, but John Herivel's arrival fits in with the ll January date. He arrived at Bletchley Park on 29 January 1940 and we straight to Hut 6, where Jeffreys and his sheets were already installed. Wi| cannot have been having much success with the sheets at that tir because, once he had got the hang of the machine, Herivel found hir continually wondering how to find a way to break into it. Then, evening in early February, at his digs, he had the idea of the Herivel tip or Herivelismus as Kendrick used to call it. He remembers that, whenl related his idea to colleagues the following day, it was immediat recognised as a possible way into the Enigma. He thinks that the idea looking for clusters on a "Herivel Square' came from me,46 but I have always thought it was his own. He believes that nothing significant' observed until the German blitzkrieg of May 1940, when suddenly^ number of Enigma operators were careless and the 'neighbourhood' i the ring settings for the day stood out clearly on a Herivel Square, was providential, in view of the fact that the perforated sheets suddenly become useless. I myself had nothing to do with the actual breaking of daily keys uu the crisis of May 1940, so it is not surprising that, when I was writing I book from memory, I thought that both the Herivel tip and wh mistakenly called 'sillies' were new ideas that occurred to us at that tin Actually the idea of Cillis had been worked on in the Cottage. It is i conceivable that Dilly was aware of them before Pyry. One would like know. There were a lot of activities in the Cottage of which I was I nothing, as was brought to my attention by Polish accounts of BruB A record was kept at Bruno of all keys broken and exchanged bet B.P. and Bruno from 17 January 1940 to 21 June 1940. What may be deduced from this record is discussed in section VIII. The first Polish break of wartime Enigma came immediately after Turing delivered the Jeffreys sheets, and it seems that Hut 6 started breaking at about the same time. But, in view of Rejewski's impression at Pyry, it is by no means obvious that Dilly and his team did not achieve breaks at a much earlier date. Again one would like to know. It seems a great shame that the accomplishments of a man who did so much for his country, and indeed for the world, have not been made known. He did outstanding work in Room 40 in the First World War, and stayed on with Alastair Denniston. Around 1936 he was tempted to return to academic life at King's College, Cambridge, but chose to continue his work on Enigma. Not very much is known of his successes with the Enigma machines that were used in the Spanish Civil War. And hardly anything is known of his achievements after that. Of the men who were closely associated with Dilly, both before and after Pyry. the only one alive is Peter Twinn. One would have thought that he would have been encouraged to write about Dilly, but he has been refused permission to do so. One would have thought, also, that the major achievements of other old-timers of GC & CS, such as Oliver Strachey, Josh Cooper, John Tiltman and Hugh Foss would have been made public by now. Indeed the attitude of the British authorities to the people to whom so much was owed. British as well as Polish, is hard to understand, let alone justify. Two stories involving Dilly Knox are worth mentioning. Just after the Warsaw-Pyry conferences of 24 and 25 July 1939, Dilly wrote a thank you note to Rejewski, Zygalski and Rozycki saying, in Polish, "My sincere thanks for your co-operation and patience'. He enclosed for each of them a set of little paper 'batons', inscribed with the letters of the alphabet,41 Rejewski's comment was "I don't know how Knox's method was supposed to work. Most likely he had hoped to vanquish Enigma'. Deavours, who published an article on the method of batons in Crypto logia of October 1980, believes that Dilly had actually used batons to break the commercial Enigma during the Spanish Civil War, but has been unable to confirm this. The other story, as yet unexplained, concerns what was known to the Poles at Bruno and Cadiz as The Knox Method'. Lisicki tells me that this involved the MeteoCode, theHerivel tip, and a method of using operator carelessness to determine wheel order. The Meteo Code was a three ettercode used by the Germans for communicating weather information ^tween an airfield and aircraft in flight. It first appeared at the turn of j"/40, but apparently was not considered worth bothering about until around the time of the invasion of Norway, in April 1940. It was found at a reciprocal permutation was being applied to the letters of each code group in a message. Then the astonishing discovery was made that this permutation was the sleeker permutation of the military Enigma key for the day. Thus the first step of the "Knox Method' was to break the Meteo Code for each day, which was not difficult. The next step was to use the Herivel tip, which began to perform well in May 1940, according to Herivel's memory. The third step was to determine the wheel order by an analysis of indicator settings and indicators that was different from the Cilli approach. The rest was easy. This story of a method that Dilly seems to have made known to the Poles but not to Hut 6 is, I believe, only one of many instances in which Dilly generated ideas and did things without telling people who could have used the information. It would be interesting, for example, to know what Dilly did with the daily keys recovered by the Poles and sent to B.P. from Bruno. These keys could have been valuable in the catalogue of broken keys that was kept in Hut 6 by Reg Parker.48 Knowledge that the Poles were breaking Green traffic would certainly have been of value to me, but it was not communicated. It seems, in fact, that even if Peter Twinn is allowed to write an account of Dilly's activities before and during the war, there will be parts of the story that can never be told. VII THE BOMBE WAS NOT ALL THAT M/ TTERED On page 184 of his official history of British Intelligence in the Second World War Hinsley states that the first bombe arrived in August 1940. This I can believe, though from memory I would have guessed September. On page 494 of his Appendix Hinsley changes the time of arrival to the end of May 1940 and goes on to say that 'it is possible to arrive at an actual measure of the Polish contribution to the successes against the wartime Enigma'. His argument leads to the conclusion that, in the absence of I Polish assistance, the first bombe would have been delivered in January 1941 instead of in May 1940. This, in my opinion is utter nonsense.f Furthermore, as I will attempt to show, the bombe was not all thatj mattered. The January 1982 edition of Cryptologia intained Re jew ski remarks,| dated 2 December 1979, on Hinsley's Appendix 1, a copy of which ha been sent to him by Woytak. In comments on 34 of Hinsley's statements! Rejewski shows that the Appendix gives a very inaccurate account of I Polish work. Hinsley is also misleading in his discussion of the Britis effort. His greatest error, in my view, is his complete failure to grasp I importance of the people who were involved. It was extremely important that we were able to recruit enough hig quality people in time to take advantage of the opportunities that came our way. Hinsley was not at Bletchley in the early days and may not have been told of the sheer piracy that we were able to employ in our recruiting until the spring of 1941, when C.P. Snow was put in charge of the allocation of all scientists and mathematicians. Thanks to the Poles we got started quickly and recruited enough key people to see us through the crisis of May 1940. The success of this first round of recruits made it possible to go on recruiting for the expansion of our problems that lay ahead. Without assistance from the Poles, our recruitment of high quality people would have been too little and too late. To be more specific, if the Poles had not given us the details of the Enigma at Pyry, the British GC & CS would probably have continued to think that the Enigma problem was hopeless without a capture. Even if Turing had thought of his bombe, there would have been little or no justification for its engineering development. I would probably not have been assigned to work on Enigma, and who else would have thought of the diagonal board? The need for a production-oriented organisation would not have been apparent. If Herivel had not been recruited in January 1940, who would have thought of the Herivel tip, without which we would have been defeated in May 1940-unable to maintain continuity until the bombes began to arrive many months later? Let there be no misconceptions about this last point. Loss of continuity would, at all stages, have been very serious, if not disastrous. I feel confident in making this statement, even though I myself knew very little about how the Hut 6 team of 'wizards' dealt with their crypt analytical problems. I can claim to have made their recruitment possible, early enough and in sufficient numbers. They did the job. Hinsley was not the only one to concentrate far too much on the bombe. Another, as I have learned only recently, was Oscar Oeser, who was appointed in 1942 to be the spokesman for Hut 3 in matters of priority. In March of 1983 Jean Alington (now Mrs. Jean Howard), who had been Oeser's deputy, was asked by the BBC to prepare a statement on the preparations that were made at Bletchley for D-Day. A year later, after talking to a lot of people who had been involved, she submitted a statement49 and sent me a copy thinking it would be of interest. It certainly was. What Jean had found most extraordinary about her research was how little everyone knew about the whole picture. For instance no one appeared to remember the large log-reading group that had been built up ty M18, or the means by which interception had been coordinated with Hut 6 activities from the very beginning. She remarked: "Each individual, working flat out, thought that they knew everything. In fact each indiv'dual had tunnel vision." What happened, I think, is that people who arrived in Hut 3 after the first two years were put into slots in a wi established organisation Their assigned tasks kept them working flat oufj and, as Jean says, they did develop tunnel vision. Jean herself arrived early enough to take part in the formative periodic Hut 3. The Air Index, which was to prove of enormous importance, hal been set up by Squadron Leader Reggie Cullingham in early 1941. Jean joined him in May 1941, at which time the whole index was contained ig one shoebox. Oeser, then a Flight Lieutenant, was already one of th R.A.F representatives in Hut 3, which was then headed by Commanded Saunders. The index, compiled from Hut 6 decodes, soon became &! means of indoctrinating new arrivals in Hut 3. Jean remembers indoctrinating Peter Calvocoressi when he arrived early in 1942. Of the authors who have written about Ultra, Calvocoressi, in his Top Secret Ultra* seems to me to have the best appreciation of the importance both of? interception and of the overall co-ordination achieved by Hut 6 in the early days and maintained throughout the war. During 1942, just as I was feeling the need for a much closer interrelationship between Hut 6 and Hut 3, the organisation of Hut 3 was changing. Travis asked Saunders to focus his efforts on the obvious need for a greatly expanded bombe programme. Group Captain Jones came in to direct the equally obviously needed expansion of Hut 3. At my request H he appointed Oeser, now a Wing Commander, as the spokesman for Hut| 3 on matters of priority between our two organisations. When, in 1943, Oeser set up an organisation known as 3L, Jean joined him, leaving the| Air Index activity which had grown to six girls on each shift. I had hoped that Oscar would grasp the whole picture of what realty mattered, making it known to key people in both Huts. But it now appears, from what Jean tells me, that this did not happen. He concentrated on bombe time and decoding, for which he developed 'coefficients of importance and urgency'. He himself was often away, leaving the donkey work of the bombe time exercise to Jean and his other assistants. J| He took little or no interest in the overall picture, and he did little td jf inform key people in Hut 3 of what was going on elsewhere. This isJJ evidenced by the fact that, at an Anglo/ Yugoslav Symposium at thei| Imperial War College, Jean heard Ralph Bennett, who had been a Hut31| Duty Officer, state blandly: I suppose we just covered frequencies by luck." She blew her top! f Fortunately the management of interception was in the hands of well qualified people, Commander Ellingworth and Wing Commander* Shepherd. When I started to analyse Enigma messages in 1939,1 established a close working relationship with Ellingworth at Chatham. At thC-J turn of 1939/40 Hut 6 already had a 24-hour team under Colman, whit kept in continual telephone contact with Ellingworth's duty officers hen. in 1941. the big R.A.F intercept station at Chicksands was opened, ing Commander Shepherd co-operated closely with Ellingworth and il man Very soon Colman's team was working just as smoothly with the ity Officers at Chicksands as they were with those of Ellingworth, who d moved his main station to Beaumanor. The organisation that was ablished in the first two years worked remarkably smoothly for the rest the war. Phe interception and analysis of radio nets carrying messages entered on the German military Enigma was not as simple a matter as it 5 been made out to be. The term 'frequency' as used by Bennett and lers is misleading, for the radio sets of those days would drift badly, eh controller of a German radio net would have to struggle to maintain ntact with and among his outstations, using chit-chat for this purpose/1 eh of our intercept operators, struggling to keep in touch with the stations of a radio net, would be continually tuning to pick up the chit it. It would have been useless to tell our operators to keep their sets led to specified radio frequencies. On the other hand our experienced erators could identify individual German operators by their habits, >n when their transmitters had drifted quite badly from assigned fre sncies. The analysis of the radio transmissions started with the intercept erat or who recorded all chit-chat and identifications on sheets of a g'. When the German net controller had paved the way for the trans ssi on of an Enigma message, this message would be recorded on a >ga rate sheet. The log sheets would go to the M18 log-reading group, which started in London and moved first to Harpenden, then to Beau nor, and finally to Bletchley. Except for the time when the group was Beaumanor, all log sheets reached them by despatch rider. The ssages were sent to Hut 6, also by despatch rider, for a good part of the r. Unfortunately the term "Traffic Analysis' orTA when applied to real time radio nets carrying Enigma traffic, means different things to fe rent people. To some it means obtaining information from a study of logs. To others it involves enciphered Enigma messages as well as the t-chat in the logs. When I used the term in my book I meant the Jlysis of enciphered Enigma messages and their preambles. This, from : outset, was done in Hut 6. The log-reading was done by the Ml8 up. who ultimately joined forces with Hut 6. Mthout going into too much detail I want to show that our method of "idling message analysis, well established in the first year and virtually ^hanged throughout the war, mattered a great deal. It was based on the affic Register', which should perhaps have been called the "Message gister'. It contained the preamble of each message and the first six letters of enciphered text. This was all that we needed in Hut 6 until decided how we were going to attempt to break a daily key. The re gist was sent, page by page, by teleprinter, so there was very little del between the interception of a message and the time at which its pre am reached Hut 6. Until May 1940 the register revealed female indicators. Then it en abl us to work immediately on Herivel tips and Cillis.. Later on it enabled us! to call for important messages to be sent to Hut 6 by teleprinter rather; than by despatch rider. In Hut 6 three copies of the register went to three destinations, to the Registration room, where the messages were charted, to the crypto logical wizards, who used them to plan their attempts at breaks, and to Colman's team of intercept co-ordinators, who used them in their constant telephone contacts with duty officers at the intercept stations, who had their own copies of the register pages that they had transmitted. It proved to be a very speedy and efficient system of information exchange between the speciali sed teams whose contributions were essential to our success. From Jean's research it now seems that leading people in Hut 3 did not realise how this Traffic Register system reduced the delays involved both in breaking and in giving them the most important decodes. It also seems that Oeser and other leaders in Hut 3 contributed little. But there was probably not much that they could do. When the heyday of Ultra had arrived, Mimer-Barry's team of wizards, Colman's team of co-ordinators, the large log-reading effort, and the experienced people at the intercept stations could do a good job on their own. It was extremely important that this group of teams were allowed to collaborate without uninformed outside interference. Indeed the British success in developing and using Hut 6 Ultra was largely due to the early establishment of excellent communication, collaboration and co-operation between many speciali sed activities. In Chapter 3 of The Ultra Secret, Winterbotham recalls how the intelligence part of the overall plan was born. Menzies showed him the 'first results from Bletchley four decodes of German air force Enigma messages and asked him to take them to the Director of Air Intelligence at the Air Ministry. On the following morning Winterbotham presented Menzies with a plan for handling the output of Hut 6. Anticipating problems that would arise later, he proposed that an intelligence organisation be set up at Bletchley to work closely with Hut 6, and also that Special Liaison Units (SLUs) be established to protect the security of Ultra in the field. Menzies said, "All right, you can go ahead if you can get the approval of the Directors of Intelligence'. The Director of Air Intelligence gave his approval immediately. Then, says Winterbotham, "As luck would haveiti the next signals to be caught and unbuttoned were from the German Army'. So he approached the Director of Military Intelligence at the War Office and won his immediate approval. Thus, within a few days, it became possible to establish an inter service intelligence activity at Bletchley and to start building up an organisation of SLUs. That this was achieved so early proved to be of immense importance. However, writing from memory, after more than 30 years, Winterbotham got his dates wrong. He made it seem that all this happened in early April 1940. This error was repeated by Lewin,52 who also gave the erroneous impression that Hut 6 was not put on a 24-hour footing until just before Hitler's invasion of Denmark and Norway. Hinsley repeats both these errors in his official history. Fortunately a glimpse of the true story is provided by a list of the broken Enigma keys that were exchanged between Bruno and Bletchley during the period of co-operation. It now seems that Hut 6 Ultra was born in mid-January 1940, not in early April. VIII LANGER'S LIST OF 126 BROKEN ENIGMA KEYS Colonel Langer wrote a 48-page report probably in Algeria in autumn 1940-on the activities of his Cipher Bureau. This report contains a listing of broken Enigma keys, which is shown in a modified form in Tables 1 and 2. Table 1 starts with the last daily key broken in Poland before the war a key for 6 July 1939, broken on 26 August, 1939. The other entries in Table 1 are daily keys of 1939 broken at Bruno or Bletchley in 1940. Table 2 is concerned with daily keys of 1940. In this table it has been convenient to show the delay in days between the date of a daily key and its entry in Langer's list of broken keys. Polish breaks would have been recorded at once, but Bletchley breaks could not be entered until they had been communicated to Bruno, which evidently took many days in the early months of collaboration. Unfortunately Langer's list does not indicate which of the German crypto nets (Red, Green or Blue) was involved in each break. Nor does it say which of the breaks were achieved at Bruno. But we have clues, and it is intriguing to speculate on what may have happened. In his rebuttal of much of the content of Hinsley's Appendix 1 to Volume 1," Rejewski agrees that 83 per cent of the keys in Langer's list were broken at Bletchley. This means a score of 105 for Bletchley and 21 for Bruno. He also points out that at Bruno everything had to be done by Zygalski. Rozycki and himself, whereas Bletchley already had far more People at work. Indeed the Polish achievement of 21 breaks is quite remarkable, remembering that this same team of three would be de-I ciphering messages on daily keys broken at Bletchley. | It seems clear that the first break at Bruno was the daily key for 28 1 October 1939, broken on 17 January 1940, immediately after the arrival I TABLE 2 DELAYS IN DAYS BETWEEN THE DATE OF A DAILY KEY OF 1940 AND ITS ENTRY ON LANGER'S LIST OF BROKEN KEYS of perforated sheets from Bletchley. This was the Green key used by the administrative centres of the German army. It also seems clear, though I surprising to me, that the Green key for 25 October 1939 was broken at i Date of Delays Delays Delays Delays Delays Delays Daily Key Jan. Feb. March April Mav June ABABABABA B AB Bletchley at about the same time."4 Apart from this one key I suggest that 1 the remaining nine daily keys of Table 1 were Polish breaks. TABLE 1 DAILY KEYS OF 1939 (shown on Langer's List of Breaks) 1 14 i o 2 6 24 0 3 1 1 4 7 14 20 5 7 ,) (W 5 10 7 420 8 15 4 1 (, 9 236 10 4 Date of Daily Key Date of Break (in 1939) 11 1 0 12 15 454 () July 6 August 26 1939 August 2 January ? 1940 September 3 January 28 September 10 March 17 September 13 March 17 September 19 February 28 September 29 February 23 September 30 February 13 October 25 January 27 October 28 January 17 In the period covered by Table 2, there are 16 cases in which twoi different daily keys for the same day were broken. I suggest that the PolrtJ broke the first of the keys shown in the table for the following seven daysg 6, 16, 18, 26 January, 24, 27 February and 20 March; also that they broki| the second keys on five days: 18 January, 21 February and 2, 21, 2K March. This accounts for the Bruno score of 21 breaks. Those shown Oft , &" Table 2 are italicised. Much of this is pure conjecture, but it may well D|| pretty accurate. * It appears that, after the sheets were received on 17 January 1940, Polish three-man team worked backwards, leaving work on current kc]K to Bletchley. I suspect that they concentrated on Green traffic, while concentrated on Red. As I discovered in my studies of September aflft October 1939, the French intercept stations were far better placed f rtflj| interception of Green traffic than was our station at Chatham und | Commander Ellingworth." Furthermore, I found that the call signs were repeated monthly, so the French traffic analysts, as a result of observe tion over a long period, would have been able to provide a compl^B forecast of call signs for each day. This would have been a great help IB 13 14 4 8 o 14 9 2li 15 1 6 16 43 10 2 1 s 17 11 i 18 34 41 2 19 , 5 20 6/711 21 72.? 2 19 11 () 22 12 4 8 11 1 1 l 23 83121 24 15 23 l [ 25 8110 26 28 i , 27 31 2 13 0 0 ^ 2 1 29 10 4 10 30 14 X 9 1 i 31 5 X X 0 broken"8 A 3nd B a" W for C3SeS '" Wh'Ch tW different dail>' keys for the same daY were intercept operators because, as Ellingworth explained to me, the German radio net carrying Green traffic used an unusual method of operation wnich made it difficult to intercept.56 Anyway, using available Green ntercepts, the Poles attacked the Green key for 6 January 1940, which ey broke on 19 January , only two days after the arrival of the perforated eets. Next, on 28 January, having no doubt spent a lot of time decoding "Usages on the two broken keys, they broke the Green kev for 3 eptember 1939. In February and March of 1940 they went back to Green ira -c of 30, 29, 19, 13, and 10 September 1939. This is shown in Table 1 They also managed to attack current traffic of several days in January, February and March of 1940, as shown in Table 2: a truly remarkable achievement by Rejewski, Zygalski and Rozcyki. Because I have no recollection whatever of an early break of Green in Hut 6, I think that the Bletchley break of the Green key for 25 October 1939 must have been made in the Cottage around 17 January when a complete set of perforated sheets had become available. The Hut 6 activities were based on intercepts received from Chatham, and I am pretty certain that these would not have permitted a break of Green at that time. On the other hand French intercepts may well have been made available to Dilly's team in the Cottage.57 Let us suppose that, after completion of the Jeffreys sheets, the first two Bletchley breaks, both made in the Cottage, were the Red key of 17 January 1940,5S and the Green key of 25 October 1939. This would tie in with Winterbotham's account of the Enigma decodes that Menzies showed him. As yet, I have not been able to determine just when Jeffreys, with his team and their sheets, moved from the Cottage to Hut 6. It was certainly before 29 January 1940, when John Herivel arrived at Bletchley and found Hut 6 operational on a 24-hour footing. He remembers that early in February, when he had got the hang of the machine, he was concerned by our lack of success (the table shows no breaks between 31 December and 8 February) and found himself continually thinking about what could be done about it. Then, one evening in his digs, when he was imagining what it would be like to be a German operator preparing to send off his first message of the day, the idea of his 'tip' flashed into his mind. His recollection, however, is that the tip did not become significant until the German Blitzkrieg of 10 May 1940, when suddenly a number of German operators simultaneously showed the appropriate form of laziness. Looking at Table 2, it seems possible that the move to Hut 6 occurred between 17 and 22 January, and that the broken keys of 22,24,29,30 and 31 January were Red. But I am still puzzled by my memory that the first Hut 6 break was into the Blue crypto net used for training.59 Perhaps the key for 22 January was of Blue and the double breaks of March were of Red and Blue.601 feel sure that the 24-hour operation of the supporting activities in Hut 6 had started before Jeffreys arrived, in fact in early January 1940, or in late December 1939." This was one of the things that really mattered. We now come to the spectacular changes in Langer's listing that occurred in April, May and June of 1940. Going by the delays in Langer's recordings of Bletchley breaks, it seems that direct teletype communications between Bletchley and Bruno were not established until late March or early April. More important is the fact that regular breaking of Red * traffic began on 8 April, when the Phoney War ended with preparations or the invasion of Denmark and Norway on the following day." The gap in breaking between 14 and 20 May was. of course, caused by he German change of procedure, which made our perforated sheets jseless." Note that the date on which the change became effective was 15 vlay. which fits in with my memory, not the commonly accepted 10 May. he day on which Hitler's invasion was launched. Indeed, as was shown in section VI above, the Germans seem to have developed a habit of making major changes on the fifteenth day of a month. The resumption of Hut 6 breaking on 20 May, after a lapse of only live jays, was not due, as Hinsley claims on page 494 of his Volume I. to the irrival of the first bombes. It was due to German operators w ho. working w doubt under unprecedented pressure, suddenly made the Hcrivcl tip effective and provided enough Cillis to exploit it. Note (he increasing lumber of zeros in Langer's list. This was due to the fact that the Herivel ip depended on the first messages sent by lazy operators after a midnight tey change. With the old method we had to wait until enough female ndicators had appeared before we could start stacking perforated sheets. Now. if German operators provided enough Cillis as well as a good kernel tip, the break of a daily Red key could sometimes he achieved in 4ut (Sby the midnight to 8 a.m. watch. Again, as with the old method, the Fraffie Register from Chatham provided our Machine Room with all the leccssary information. It may seem barely credible, but comparable success, using the -lerivel-Cillis method, was maintained until the actual arrival of the )omhes. On page 184 of his Volume 1, Hinsley contradicts his othertatcment, saying that the first bombe was delivered in August 1940. hut 1uspect that it may have been even later. My guess would he that the )ombes did not begin to be effective until September 1940. We were ndccd lucky. REFERENCES References arc given in chronological order of publication. Not all arc mentioned in thertiele. I Da\iJ Kahn. TVif CoaVbrcuAcrj. TVic Swry <)/ofn the same day. m. 60. Hinslcy, vol. 2. p. 65 . gives 29 January as the date on which Blue was first broken, but he gives 6 January fir Red which seem wrong. 61. The Hill .S'l'.v Slorv. } S9. 62. The second breaks into keys for 9, 1(1, 19, 21. 22. 23 and 25 may have been into the Yellow key. which, according to Hinslcy (Vol. 2, p. 662) was first broken on 10 April. I have no recollection of the introduction by the German army of a second crypto net for operations in Norway. The key that we broke regularly from 8 April on was certainly the Reil key. used in battle lor army-air co-ordination. To me the Yellow crypto net is still a puy/le. 63 Tin' hiii Si\ Story, p.97. Appendix I The Bombe with a Diagonal Board THE DOUBLE-ENDED SCRAMBLER The principle of the bombe that was to be used by Hut 6 occurred to me in the School at Bletchley Park in November 1939, before the war u as three months old. That the idea came so early was indeed fortunate, because many months of engineering design and development were needed to go from a blue-sky concept to a working machine that came into operation late in 1940. 1 will discuss the concept as it stood at the end of 1939 and the main features of the design, but let me first remind the reader that an Enigma machine had a keyboard, a lamp board and a stecker board, as well as a scrambler unit. The scrambler consisted only of in-out terminals, three moving wheels, and a fixed turnaround wheel, m umkehrwalze. When I talk about a scrambler, I mean just that. Let us first consider the very simple purpose of what I have called a double-ended scrambler. In the German Enigma the scrambler had only one set of twenty-six in-out terminals, which were connected by a twenty-six-way cable to the lower sockets of the steckerboard. Thus electric current entering the scrambler at one of its in-out terminals would return to another terminal of the same set, and would go back to the steckerboard along a wire of the same twent\-six-way cable. For reasons that will become evident in a moment, in order to perform the tests we had in mind Turing and IJ wanted the return current to come out at a second set of terminals. This second set would be connected, by a twenty-six-way cable, to) in-out terminals of another scrambler for the purpose of making i logical deductions At the time I don't think I worried too much about how this would be done, once I had satisfied myself that it could be done. It may have occurred to me that the fixed commutator at the right of a scrambler, the moving wheels, and the fixed umkehrwalze on the left could all have two concentric circles of twenty-six terminals, instead of one circle. In each of the three moving wheels the internal connections between the right-hand and left-hand terminals of the outer circle would be as in a standard single-ended scrambler; the two inner circles of terminals would be cross-connected in exactly the same manner. In the umkehrwalze each terminal of the outer ring would be connected to a terminal of the inner ring. Thus a single interconnection between terminals X and Y of a single umkehrwalze ring would be replaced by two connections: X of the outer ring to Y of the inner ring and Y of the outer ring to X of the inner ring. This arrangement would have the effect Turing and I desired. There would be two sets of twenty-six in-out terminals, each of which could be connected by twenty-six-way cable to another scrambler. Current entering at any letter of either set would go through the wheels to the umkehrwalze and return to a letter of the other set. The letter substitution so obtained would be the same as that of the single-ended scrambler. To proceed further I need a simple way of representing what I am talking about. Unfortunately I cannot remember how I drew the original diagram I showed to Turing and later on to Keen. Several years after the war I asked to see it at GCHQ, and did in fact see it. but I understand that since then it has been lost or destroyed. However, Figures AI, A.2, and A.3 will serve the purpose rather better. Figure AI introduces a diagrammatic representation of a double-ended scrambler. At the top, in (a), I have replaced the permanent connections between the single-ended scrambler and steckerboard of an Enigma machine by plug gable connections. One * Not for the purpose of lighting a lamp under the Enigma lamp board APPENDIX I / 237 Of two terminal jacks has twenty-six pin terminals that can be plugged into the sockets of the scrambler's in-out terminals. This jack, shown in Figure A.l(a), is connected by a twenty-six-way connector cable to another jack, not shown, whose twenty-six pin terminals can be plugged into the sockets of the in-out steckerboard terminals shown in Figure 3.6, which are connected to keyboard and lamp board The scrambler itself is represented by a rectangular box, indicating its in-out terminals and internal structure of moving wheels and fixed umkehrwalze. Figure AI (b) represents the double-ended scrambler I am talking about, with its two sets of in-out terminals, each of which can be connected by a twenty-six-way terminal jack and connector cable to any selected set of female sockets. Note that the two sets of in-out terminals are interchangeable in that, if current entering at the upper P terminal comes out at the lower Q terminal, then current entering at the lower P terminal would come out at the upper Q terminal. In Figure AI (c), I show the condensed representation of a double-ended scrambler (b) that I will use in Figures A. 2 and A. 3 to illustrate the principles of the bombes, both Turing's and mine. The twenty-six-terminal jacks are represented by single lines. The box between them represents the double-ended scrambler with its two sets of in-out terminals. TEST LOGIC OF A TURING BOMBE Back in Figure 4.4 (q.v.), I postulated a "crib," which produced three diagrams of letter pairings involved in the encoding. The basic idea of the bombes, both Turing's and mine, was to interconnect a battery of scramblers in accordance with some such diagram. Initially the scramblers are set at the relative positions indicated in the diagrams. Then the whole battery moves in synchronism through all 26 x 26 x 26 = 17,576 positions of each scrambler. In each position a test is applied that determines whether that position is a "drop" requiring further analysis. Let us first see how a Turing bombe would have been applied to the first diagram of Figure 4.4, which is reproduced as part of Figure A.2. The top scrambler unit in Figure A.2 is to be set initially at position 10 in the crib sequence, in which letter P is the encode of letter I. As I have just explained, single lines represent 26 Terminal Jack 26 Way Canntctor Cat/* magnitude. But this is absurd, because we know that the three angles of a tn angle add up 0 exactly r\u> right angles. Therefore the assumption is false. '10 Positions Tesf Register TTIITTT.TTTTI1I1III 1 A B C D e f G H I KLMNOPQR S T U V W X Y Z oooooooooooooooooo oooooooo Current Input (Assumption IA ) Figure A. 2 Application of Turing Bombs APPENDIX I / 241 26 possible positions and, in each position, test whether current gets back to any terminal of the test register other than A. I laving three closed loops, for each of which the chance of current's returning to A is 1 in 26, we would expect only one drop per wheel order. Fortunately we do not have to repeat the test run with current inputs at other terminals of the test register, because, whenever current gets back to the test register at a terminal other than A, it will go through the loops again. Indeed with three loops feeding back, the reader will not find it hard to believe that, in most positions of the born be, the test register will "fill up"--in other words, current will reach all its twenty-six terminals. Note, however, that if the true stecker of I is some letter other than A--say, P--then, when the scramblers of the bombe are in their correct positions, and we are trying the correct wheel order, current put in at P, representing the assumption IP, would not come back to the test register at any other terminal. Consequently current from terminal A could not reach terminal P. Thus, in order to detect a "drop," the bombe needs to look for one of only two situations: a case in which current does not get back to any test register terminal other than A, or one in which there is some terminal other than A to which current docs not get back. In the vast majority of positions the test register will till up. When a particular wheel order is being run, each "drop" will tell us that, for some set of positions of the scramblers, the bombe has been unable to reject the possibility that we have hit on the correct wheel order and wheel positions. More work will be needed to determine whether we have a false drop. PRINCIPLE OF THE DIAGONAL BOARD Let us now consider how my version of the bombe would be applied to the second diagram of Figure 4.4, which uses positions 1, 3, 5, 6, 8, 10, 11, 12, and 13 of the crib. The right-hand portion of Figure A. 3 shows a diagonal board, which is simply a 26 X 26 matrix of terminals. It acquired its name because, for example, the terminal in row E and column A is connected diagonally to the terminal in row A and column E. The cross-connection is no more than an electrical equivalent of the fact that, if E is steckered to A, then A must be steckered to E. I believe that the name "diagonal board" was introduced by Doc Keen. Scramblers Connectors Jacks Jacks Positions teA B C D F G H IJKLMNOPQRS T U V W X Y Z I' i 1-- l r i i TAT T I --1-- 1 I I --1--1--1--LJ--L Test Register Current Input (Assumption E/A ) Diagonal Board A B C D E F G H i KLMNOPQftS T U V W X Y Z r __ < ---F1 => A =i B =1 C !//// 4 B >^' ' -^ ~5b =j D =1 f C D ^r: : : : : : : : i -13 f cA =1 6 ( __ L => H S: n / i ---=3 J l __ ^ K i __ =1 L i ---3 H & i --Zl 0 N O cA Z! f 1 ---n 0 ' D f f> t ---3 Sci 3 T S= 3 f & D ^ i ---- i ---3 W -, X * Y& 4/---- ------ , E /^/ \^ 7y // U P 6 V C 4Right Hand Side Diagonal Board with Jacks Crib Diagram Figure A.3 Bombe with Diagonal Board Left Hand Side Connections between Scramblers, Jacks, and Test Register APPENDIX I / 243 The terminals of each row of the diagonal board are also connected to a twenty-six-terminal female jack shown to the left of the row. These same jacks appear also in the left-hand portion of Figure A.3, which shows how twenty-six-way cables connect them with a set of double-ended scramblers and a test register. As in Figure A.2, the test register is associated with a letter, in this case E, and current input at terminal A is equivalent to the assumption E/A. My new idea, which made so much difference, was simply to connect the test register and the scramblers to the diagonal board, in the manner shown in Figure A.3. The point was that, if the input current could reach the terminal of the diagonal board in row X and column Y, then that X is steckered to Y must be a logical deduction from the assumption that E is steckered to A. Furthermore the current would be connected diagonally to the terminal in row Y and column X (meaning Y is steckered to X), and would have a chance of getting back into the scramblers via the twenty-six terminal jack associated with row Y. As I said, Turing, though initially incredulous, was quick to appreciate the importance of this new twist in Enigma theory, which greatly reduced the number of bombe runs that would be needed to ensure success in breaking an Enigma key by means of a crib. I had no difficulty in explaining the idea to Doc Keen at BTM, and once he got it, he was really with it. I do not remember any reaction from Dilly Knox, but I shall never forget the facial expression of a representative of the U.S. Navy when the validity and importance of the idea dawned on him. Looking at Figure A.3 a little more closely, it will be seen that five different twenty-six-way cables, from the test register and from the four scramblers, in positions 11, 8, 5, and 12, had to be connected with the twenty-six-way female jack of row E of the diagonal board. The problem of making these connections was easily fixed by providing sets of commo ned female jacks that would interconnect twenty-six-way cables plugged into them. For example, if a set of six commo ned female jacks was available, the five male terminals of the twenty-six-way cables from the four scramblers and the test register could be plugged in, and a single cable could connect them all to row E of the diagonal board * Similarly three cables had to go to row I, three to row F and two each to rows N and T. This could be handled by using other sets of commo ned female jacks. APPENDIX I / 245 The input current from terminal A of the test register goes to this "commoning board" and proceeds to terminal E/A of the diagonal board. This is connected to terminal A/E, which does us no good because row A of the diagonal board is not connected to a scrambler. At the same time, however, the input current is getting to scramblers in positions 11, 8, 5, and 12, giving deduced steckers for the letters D, I, P, and U. The deduced steckers for I and P feed into scramblers in positions 13, 10, 10 (from the other end), and 6, giving deduced steckers for N, P, I, and V. We thus have eight immediately deduced steckers: two each for P and I; one each for D, U, N, and V. These eight deductions result in electric current's energi/.ing terminals in rows P, I, D, U, N, V of the diagonal board. The diagonal connections will energize terminals in other rows, which may feed back into the scramblers and produce further deductions. Since nine of the twenty-six rows of the diagonal board are so connected, there is a good chance that further deductions will be made. Furthermore, since each scrambler unit is connected to one or more of the other eight, any single deduction flowing back from the diagonal board is likely to have an effect comparable to that of the input assumption E/A. The first round of deductions, which is roughly equivalent to what the Turing born he could produce, could lead to a return flow from eight rows of the diagonal board, each of which has a 9-out-of-26 chance of reentering the scrambler interconnections. Furthermore, each such reentry can produce another round of reentries. VALUE OF THE DIAGONAL BOARD Why was this idea so important, and what can it teach us today? Well, let me say at once that my two ideas were by no means the complete answer to the often-asked question of "How did we do it?" As I stated at the beginning, my ideas did no more than reduce--by enormous factors--the number of possibilities that had to be examined by the wizards. The methodology was worked out by others, and I was never personally involved. I think I can see how it could have been done, but I will not go into details. I might so easily be wrong. Perhaps some of the surviving wizards may provide a true account of what they did, pointing out any errors in detail that I may have made in this book. However, although my memory of my unrecorded thinking of forty years ago may not be completely accurate, my two ideas, the first of which had already been thought of by others, did in fact reduce the task of the wizards to manageable proportions. I have stated that in the use of the diagrams of Figures A.2 and A. 3 the scramblers of the bombe would be set to the relative positions indicated in the crib diagrams of Figure 4.4. But we must worry about the fact that a turnover of the middle wheel will occur somewhere in each sequence of twenty-six encipherments of the crib letters, and we have no means of knowing where. Thus, to apply Turing's bombe in the manner indicated in Figure A.2 we would like to assume that no turnover will occur in the nineteen consecutive transitions between the positions in the scrambler cycle that we have arbitrarily numbered 3 through 22. Then, in the initial setup of the scramblers for the bombe run, we could put all left hand and middle wheels to the same settings. Thus, if we use ring settings ZZZ as a standard, we could set all left-hand and middle wheels to AA. But, to assume no turnover in nineteen consecutive positions out of 26 would have given only a 7 in 26 chance of success. The positions involved were 3, 5, 6, 8, 10, 13, 15, 21, and 22, for which the appropriate letter settings of the right-hand wheel are C, E, F, H, J, M, O, U, and V. The biggest gap is between positions 15 and 21. If we assume that the turnover occurs in this gap, we can set the starting positions of the left-hand and middle wheels for positions 21 and 22 to AB instead of AA. But a second run on this assumption would add only another 6 in 26 chance, to give a total of 1 3 in 26, or 1 in 2. Not too good; and even then we would be risking the possibility that a left-hand wheel turnover might accompany the middle-wheel turnover. To allow for all turnover possibilities would require a lot of bombe runs. In contrast to this, the diagonal board idea makes it possible to obtain a usable diagram of scrambler interconnections from a much shorter section of the crib. Thus in Figure 4.4, showing diagrams derived from a crib, the second and third diagrams, both of which are made feasible by the diagonal board, depend on positions 1 to 13 and 14 to 26. We can confidently assume that there will be no turnover in one of these two stretches. For each of these two assumptions we can set all left-hand and middle wheels to the same APPENDIX I / 247 positions, AA, for the starr of a born he run. Between them these two born be runs will cover all turnover possibilities. A big gain. ENGINEERING DESIGN OF THE BOMBE Doc Keen of the British Tabulating Machine Company (BTM), the designer of our bombes, adopted a configuration for a double ended scrambler that used rotating "drums" rather than wheels. The general idea is illustrated in Figure A.4. A flat template, (a), contains three commutators, each of which has four concentric rings of twenty-six terminals with a hole at the center for a shaft. Three drums can be mounted on and driven by the three shafts, as is indicated in Figure A.4(b). Each drum has four concentric rings of terminals making contact with the terminals of a commutator. Wiring inside a drum interconnects the two outer rings. Separate wiring interconnects the two inner rings. I these interconnections are the equivalent of those between the right- and left-hand rings of terminals on an Inigma wheel. Keen obtained the desired effect of a double-ended scrambler Infixed wiring at the back of the template. A set of twenty-six in-out terminals was connected with the outer ring of the top commutator on the template, which would be connected through the top drum with the second ring of the top commutator. This ring was connected, behind the template, to the outer ring of the middle commutator, which would be connected through the middle drum to the second ring of the middle commutator. Similarly the second ring of the middle commutator was connected, behind the template, to the outer ring of the bottom commutator, and the bottom drum would provide connections with the second ring of the bottom commutator. This ring was connected by the equivalent of the umkehrwalze cross-connections to the third ring of the lower commutator and thence through the two inner rings of terminals on the drums and commutators to a set of twenty-six out-in terminals associated with the innermost ring of the top commutator. Finally a battery of twelve double-ended scramblers was mounted on a frame, as indicated in Figure A.4 (c). I'o obtain reliable electric contact and fast mechanical motion, Keen mounted wire brushes, rather than spring-loaded pin terminals, on his shaft-driven drums. These brushes were standard on Z48 / THE HUT SIX STORY ' :.. *: ': o ::: i. vA .; \--:.:...:.: (a) (b ) Figure A. 4 Keen's Design of the Born be (a) Template (b) Drums on a Template (c) A Born be APPENDIX I / 249 BTM punched-card equipment, so they were already being manufactured in large quantities. The use of brushes meant that the drums could rotate in one direction only. One big advantage of Keen's configuration, as opposed to a wheel-type scrambler, was that the shafts on which the drums were mounted could be driven by a mechanism on the other side of the templates. In each of Keen's scramblers the top, middle, and bottom drums correspond to the right-hand, middle, and left-hand wheels of an F.nigma scrambler unit. All the top drums were driven together in synchronism. Similarly the middle drums were all driven together, as were the bottom drums. As far as I remember, each drum was about 5 inches across; the bombe, mounted on casters, was somewhat over 4 feet long. A door at the back served as the frame for a diagonal board, testing device, and other electrical circuitry. It was connected by an enormous cable to the in-out and out-in terminals of the twelve scramblers. With the door open the drive mechanism could be reached. Keen's first step was to design and build two prototype bombes, with ten scramblers each. Me made use of standard BTM brush sensing technology and well-known electromagnetic relay switching, but even the prototypes were essentially experimental machines. Indeed the bombe was totally different from any machine that Doc Keen or anyone else had ever designed. le and his principal assistants paid many visits to the prototypes after they were installed, to correct faults, to try out improvements, and to learn from the technical experience of the two engineers in charge of maintenance, both of whom were experienced in the operation and maintenance of standard BTM equipment. The two prototype machines were soon followed by twelve-scrambler production models. the organization for operating the bombes, and the way they were used by the Hut 6 watch, are discussed in Chapter 8. Appendix II GORDON WELCH MAN a biographical note by Mark Baldwin, MA MSc PhD William Gordon Welchman was born in Bristol on 15 June 1906, son of the Reverend William and Elizabeth Welchman. He went to Marlborough College on a scholarship from 1920 to 1925, when he won a mathematics scholarship to Trinity College, Cambridge. In 1929 he became a Research Fellow at Sidney Sussex, and a Fellow in 1932, and later Dean. In September 1939 he joined a small group of mathematicians and others at Bletchley Park, who were to form an outstandingly successful cryptographic team under Alastair Denniston He remained at Bletchley Park for the entire war, largely working on the cryptographic problems raised by the German Enigma ciphering machine. Work at Bletchley Park was divided between huts, and specialist teams became known by the hut number where each had first been established Welchman's particular responsibilities lay with deciphering Enigma radio signals broadcast by the German air force and army, this was carried out in Hut Six. As a leading member of the Bletchley Park staff, he had wider responsibilities for planning and implementing broad strategies not only for deciphering radio signals, but also for digesting, assimilating and transmitting the intelligence (known as "Ultra') derived from the deciphering process. On one famous occasion in October 1941, a personal letter was sent by Welchman and three colleagues (Alan Turing, Hugh Alexander, and Stuart Mimer-Barry) to Winston Churchill, ; demonstrating most forcibly the need for more support staff if'J the flow of vital intelligence was not to be impeded. Churchill's! APPENDIX 11/251 response was the legendary "Action This Day', aA The intelligence kept flowing. In 1943 he became Assistant Director in charge of mechanisation for Bletchley Park, and his wider responsibilities included cryptographic liaison with the United States. His wartime service was recognized by the award of the OBE, at a time when the work of Bletchley Park was still secret, and awards to those involved in intelligence were deliberately modest. Perhaps Sir Stuart Mimer-Barry's words, spoken much later, summa rise his work better: 'if Gordon Welchman had not been there I doubt if Ultra would have played the part it undoubtedly did in shortening the war'. After the war, he became Director of Research for the John Lewis Partnership, but moved to the USA in 1948, attracted by developments in computing and information technology He first worked at MIT, teaching that college's first computing course This was followed by spells with Rand Remington and Ferranti and in time he became an American citizen. From 1962 until retirement in 1971 he worked with the MITRE Corporation on the development of communications systems for the US forces and was then retained as a consultant by MITRE until 1932 In 1982, his book The Hut Six Story was published by McGraw-Hill in America, and by Alien Lane in Britain. He had hitherto kept loyally silent about his war-time work, and might have been forgiven for assuming that the publication of Winterbotham's The Ultra Secret in 1974, soon followed by other, better, books on the subject, demonstrated that the need for silence had passed. Nevertheless, official disapproval of The Hut Six Story was such that he lost his security clearance in the USA, and received very unsympathetic treatment at the hands of the US National Security Agency (see Nigel West's The Sigjnt Secrets, published in New York by William Morrow in 1988) He died on 8 October 1985, aged 79, at Newburyport Massachusetts, survived by his widow, Elizabeth, and their three children, Nicholas, Susanna and Rosamond, and two stepsons Michael and Thomas Wimer. APPENDIX III: Publishing History f The Hut Six Story was first published by the McGraw-Hill Book I Co. Inc. in 1982, with a British edition appearing in the same I year under the Alien Lane imprint. A paperback edition was $ published by Penguin Books in 1984. f In the present venture, we are grateful for the co-operation | of McGraw-Hill Inc. and of the family of Gordon Welchman. it We also much appreciate Alan Stripp's support for this work, || and in particular his Foreword. i 7*. In this new edition, the text of Parts One to Three has been f reproduced in its entirety, with the following amendments: | page 41: the caption has been corrected (and on page ix) 1 page 44: the original illustration has been replaced by one I matching the caption (and the author's intention) t page 45: in line 29, 'middle' has been substituted for 'left-hand1 f page 138: in line 2, '1940' has been substituted for '1941' | The original Part Four, entitled Today' and relating largely to the USA and the international position in the early 1980s, has been omitted entirely, as being inappropriate for the new edition. However, the book has been materially improved by the inclusion of a highly relevant paper by Gordon Welchman, originally published in 1986, and representing his latest thoughts on his work during the war. (Indeed, he died only a few months before its publication.) It extends, updates, and occasionally corrects, the account he had compiled in 1982. We are most grateful to Frank Cass & Co Ltd for their permission to reprint this paper. The inclusion of a bibliography in this paper precludes the need to reprint the original bibliography, but a new index has been prepared to reflect the changes made in this new edition. Finally, we gladly acknowledge that Tony Sale, Museum Director of the Bletchley Park Trust, first suggested that the time had come for a new edition of this most valuable book. ACKNOWLEDGMENTS In this book I pay belated tribute to many individuals and organizations who played essential roles in the story of Hut 6. I describe several postwar experiences that, combined with my memories of Hut 6, have proved valuable in my work on problems of today. In the Bibliography I have included several writings not mentioned in the text that I have found particularly helpful. Yet many more acknowledgments are needed. I owe so much to so many. First I want to acknowledge that the success of the activity that I initiated at Bletchley Park owed a great deal more than has yet been recognized to F. W. Winterbotham, the author of The Ultra Secret, I never met him, but I can now see that the full implementation of my plan for Hut 6, the fruitful partnership that developed between Hut 6 and the intelligence organization known as Hut 3, and the exploitation of Ultra intelligence on the battlefields of World War II were all due in large measure to two of his personal contributions. First, as a result of his prewar experience in the British Secret Service, Winterbotham was instrumental in arranging that the exploitation of our breakthrough would be handled as an interservice venture under Foreign Office management. Second, his expansion of specialized Secret Service communications and his organization of Special Liaison Units in the field were of critical importance. Next, I would like to acknowledge the great debt that I owe to the Governing Body--the Master and Fellows--of Sidney Sussex College, Cambridge. They allowed me to leave my college duties at the outbreak of war. They augmented my foreign office salary by continuing my fellowship stipend throughout the war, even though I was doing absolutely nothing for the college. And they wished me well when, at war's end, I did not feel that I could return to my academic life. The war had given me the opportunity to start doing what I have been trying to do ever since; to absorb new ideas from experts in many fields and to find ways of combining them with other ideas, old and new, to solve some worthwhile problem. I want to acknowledge the help that I have received from all these experts, and the enjoyment I have derived from my contacts with them. I can mention but a few. When I emigrated to America in 1948, I was extremely fortunate to find a position on the staff of Project Whirlwind at MIT. This was, in my view, the most enterprising of the many pioneering computer projects in that, from its inception towards the end of the war, it was aimed at problems of real-time physical control in which computation was only a means to an end. On joining the project in the summer of 1948 I found a wealth of fascinating ideas that were new to me, and it was my job, as head of Applications Research, to think about what could be done with them. I owe a great debt to the project leader, Jay Forrester, and two of his principal assistants, Bob Everett and Norm Taylor, for drawing me into their highly innovative thinking. In 1951 I left MIT for eleven years in industry. I first worked on computer applications for Engineering Research Associates and Remington Rand in the U.S. and for Ferranti in the U.K. I then had contacts with Ferranti research establishments in England, Scotland, Canada, and America. From 1959 to 1962 I worked for the Itek Corporation. I will mention only two of the many debts that I incurred during this period. While I was personal assistant to the manager of the Ferranti computer department, we were exploring ways of combining the resources of Ferranti and Powers Samas Accounting Machines. This enabled me to work closely with the head of the Commercial Research Branch of Powers Samas, a Mr. Whitwell, to whom I am particularly grateful for all he taught me about principles and practice in the punched-card world. Secondly, at the end of my period in industry, I had the great pleasure and stimulation of leading an engineering group at Itek. The members were Brian O'Brien, Heinz Zeutschel, Don Oliver, and Dick Malin. I was on the staff of the MITRE Corporation from 1962 to my retirement in 1971. Since then I have been a consultant, working on a variety of projects. A major part of my work has been supported by contracts between the U.S. Air Force Electronics Systems Division and MITRE. During these nineteen years I have incurred many more debts to colleagues. It has been extremely exhilarating to have had contact with many fresh young minds, and I would like to make special mention of Lou Williams and Larry Hill. But, in connection with the development and ACKNOWLEDGEMENTS / 255 realization of this book, my greatest MITRE debts are to B. J. Workman and Bob Coltman. In summer 1971 I was assigned to work on a project headed by Workman, concerned with the need for speed of communications on a battlefield. Our collaboration was to continue for seven years. It was at Workman's suggestion that I made contact with IISS and RUSI during a vacation in England, a contact that was to prove very valuable. Together we searched military literature for clues to problems of the future. We were much concerned with the development of scenarios depicting what may be expected to happen on a future battlefield. A tireless student of military affairs and a prolific producer of ideas, Workman was extremely good at impromptu presentations, but was so involved in making further progress that he could hardly spare the time to refine and consolidate what he had already achieved. Refinement and consolidation were my roles, then, in what proved to be an extremely happy and productive working relationship. The President of MITRE, Bob Everett, is an old friend from the days of Project Whirlwind. He has provided encouragement and strong practical support for the writing of this book. In summer 1976 he arranged that Bob Coltman, with whom I had worked on MITRE projects, could help me with my book. We had no idea that this help would continue until the fall of 1980, when a "definitive draft" was handed to Bruce Lee, Senior Editor of the General Books Division of the McGraw-Hill Book Company, another person to whom I owe a major debt of gratitude. Bob Coltman is a MITRE writer-editor. A few readers may perhaps have heard his published recordings of folk songs. We worked hard together, in perfect harmony, and this was very important to me because the nature of the book was changing. By 1976 my work for MITRE had already shown me that the story of Hut 6 and other clandestine activities of World War II contains much that is of value today. But in summer 1978, when my collaboration with Workman came to an end, the focus of my work switched to the security of the communications system on which we were working. Later, at the end of 1980, the focus switched again to the utilization of electronic intelligence gathering systems on future battlefields These two new areas of interest led, of course, to contacts with still more colleagues, to whom I am most grateful. They also led to a fuller appreciation of the value today of the Hut 6 story of World War II. Bob Coltman's steady collaboration in revision after revision was a tremendous help in creating what is now Part Four, and in improving the earlier parts. In April 1978 McGraw-Hill had undertaken to publish my book, and I had the first of many enjoyable and stimulating meetings with Bruce Lee. He was extremely sympathetic to my interest in lessons for the future and asked if I could extend this aspect of the book by writing a chapter on the present situation in communications. He wanted me to emphasize the applied history and advanced planning features. The new draft that resulted from this meeting was read towards the end of 1978 by a young writer, Neill Rosenfeld, who had been asked by Bruce to produce a lot of questions. The resulting comments, written informally as he read the draft, were very helpful. But, when I had finished answering Neill's questions, I realized that a great deal of improvement was needed, and set to work on another draft. By that time, however, the focus of my MITRE work had already changed and I had thought more about lessons from World War II. With help from Bob Coltman and Bruce Lee, the principal themes were clarified and a better organized draft was submitted in November 1980. After another writer had suggested further improvements, Bruce Let took personal charge of the final stages of editing. Under his guidance the structure of the book was greatly improved. I have a happy memory of the last few days of work in his office, when we were tying up the loose ends. During the final stages, Kristina Lindbergh of McGraw-Hill made a major contribution. She managed to keep track of all the material that I sent to Bruce, and kept on getting me out of difficulties by digging out something of which I did not have a copy. I am also very grateful to Philip White of MITRE, who drew all the illustrations. In conclusion, I want to say a very special word of thanks to Bobbie Statkus, head of MITRE's Word Processing Department. For more than four years she did a magnificent job, working closely with Bob Coltman and battling imperturbably with revision after revision. She did most of the work herself, and had to cope with two installations of new equipment, which made it difficult to use previously recorded text. The quality of the drafts she produced was spectacular. jt INDEX To pages 1-252, but excluding the References and Notes on pages 231-234 Ahwehr. 219 AJlcr Tug. 102 Adstock. bombes at. 139. 141 Agnes. 147 An Index. 224 Aitken. Mr.. 85 Alamein (see El Alamein) Ahin Turing The Enigma (Hodges). 197 Alexander. Hugh, 61. 84. 104. 1 13. 1 15. 126. 176. 183 Alington. Jean. 223-224. 226 American Magic (Lewin). 202 Andrew. Christopher. 196 Annals of the History of Computing. The. 196 appeasement policy. 7, 8 A rim. U.S.: cryptologists in. 173. 174-175 at Kasteote. 146 at Hut 6. 135. 158 Signal Corps of. 22 Asche. 209. 218 ATS (Auxiliary Territorial Service). 157-158 Habbage. Charles 173-174. 182-183 Habbage. Dennis. 35. 85. 102. 104. 1 13-1 14. 126. 219 Bailey Mr.. 139. 140 batons, method of, 221 Baudot code. 173. 176. 177 Ba\lv. B. de F. "Pat." 170-173, 174. 176. 199 Beaumanor. intercept station at. 149-150. 157. 201. 225 Bennett. Ralph. 224. 225 Berlrand. mst ave 13. 14. 15. 16. 195-196. 209. 21 1. 218. 220 Ban-ling. Arne. 177 Bible, military intelligence and. 161-162 Bnch. Frank. 9. 35 Blair-Cunynghame. Hamish. 95 Bletchley Park. 139. 141. 175 described. 30. 186-187 expansion of. 76-77. 121 huts at. 9-10, 31. 72-73 location of. 8." prewar development of. 9-10 social activities at. 187-188 (see also Hut 6 Ultra) Blitzkrieg, 8. 19-20. 189-190 radio communications essential for. 19-20. 25. 28. 52-53 Block D. Blue traffic. 87-88. 227. 230 "Bomba'. 15-16. 77. 196-197. 214 invented by Rejewski. 213 bombes. 138-145. '199. 201 British bombe not derived from Polish Bomba. 204. 205. 214. 218-219 cribs necessary for. 80. 164 defined, 77 diagonal boards for. 81. 83. 110. 199 effect of, 222 Germans' failure to anticipate. 164 maintenance of. 147 manufacture of. 139-141 mechanical and electrical functioning of. 142-145, 148 names of. 147-148 operation of. 141-143 purpose of. 77. 82. 120 B.P. (see Bletchley Park) Bradshaw, Captain. 128. 146-147 Braquenie, 217 Briggs. Asa, 85 British Intelligence in the Second World War. 196. 199. 201. 204. 219. 222, 227. 231 British Security Coordination (BSC). 175 British Tabulating Machine Company (BTM). 81. 123. 139-141." 147. 197. 244. 247 Bruno. 19 199-200. 217. 219. 220. 221. 222, 227 keys broken at. 227-231 BSC (sec British Security Coordination) Cadiz. 217. 221 call signs. 197. 198 encoded by Germans. 125 in log reading. 95. 157 purpose of 35. 38, 53. 155 Calvocoressi, Peter. 156-161. 204. 224 Canney. June. 86-87 Central Party. 124. 1.35. 155. 157-159' chains. 105-106. 109 Chamberlain. John, 85 Chamberlain, Neville. 7 Chatham, intercept station at, 37. 53. 55. 76. 140-150. 198. 200. 201. 224. 228. 230. 231 Cheadle. intercept station at, 87 Chicksands. intercept station at. 149. 150. 152. 157. 201. 225 chitchat. 225 defined. 94-95. 154 log reading of. 155. 156. 157 Churchill. Sir Winston, 92. 102. 128. 160. 175, 250 Ciezki. Maksymilian, 20.3 Cillis (also Sillies). 98, 99-102. 104-107. 110. 119. 135. 165. 167. 168 . 200, 220. 222. 226. 2.31 Welchman's mistake over spelling. 220 cipher machines: commercial, 27-28 off-line. 47. 176 on-line. 173. 176-177 Polish. 15-16 rotor-type, 176 scrambling method of. 11-12 Type-X. 75 . 87, 106-108. 110. 176 (see also Enigmas) ciphers, codes vs." 26-27 Codebreakers, The (Kahn). 174 codes, ciphers vs.. 26-27 Colman. John. 84. 87. 126. 200. 224. 225. 226 Colossus. 174. 176-179 communications: developments in. 19-25. 27. 172-173 dispatch riders in 91. 150 improvised. 129. 149-150 communications (cont.) teletype as. 150w. 171. 172-173 wig-wag signaling as. 22. 23-24. 25 (see also German radio communications) confirmations, 108, 109. 110, 114 Cooper, Josh. 9. 37. 57. 77. 78, 179. 221 Cottage. Enigma group in, 12, 31, 98-90, 113. 114, 197. 198. 200, 219, 220. 230 Crankshaw, Edward. 95 Crawford. George. 84 Crete. German attack on. 123 Crib Room, 120, 158 cribs. 78-80. 82. 98, 101. 110, 120, 131-133. 164. 166. 237, 243-244 crypto nets. 205-206. 213. 216, 227 Cryptographic Militare, La ' (Kerckhoffs). 25 Cryptologia. 196. 203. 221. 222 cryptology. 10. 182 in business. 37 cryptanalysis vs.. 26 cryptography vs." 26 defined, 26 development of, 25-28 security and, 3, 17. 27..39. 82-83, 93 terminology of. 26, 78 in World War I. 8. 33 Cullingham. Sqn Ldr Reggie, 224 cyclometer ("Cyclometre'), 15, 16, 197.211.212.214,215.218,219 Deavours. Prof Cipher. 203. 219, 221 Decoding Room (DR). 75. 76. 87. 90, 103. 104. 110. 126. 129. 158. De Grey. Nigel. 9 Denniston. Alastair, 8-12, 13. 31, 76, 77, 197, 202, 216-217. 221, 250 Denniston. Robin. 8-9. 11. 14. 33 diagonal board: function of. 110. 137 Welchman's development of, 77. 81, 82, 83. 199. 218. 223 235-249 Diffie. Whitfield, 287 Dilks. David. 196 discriminants. 36 38 39. 52.. 54. 125, 164. 197. 205 dispatch riders, as communication system. 91. 150 drops'. 72. 90. 237. 241 Drucker. Peter. 171 INDEX /259 Dunkirk. 96 Eagle Day. 101-102. 120 F.astcote. bombes at. 141. 146 I'll Alamein. battles of. 124. 195. 201 Ellingworth. Commander. 55-56. 90-93. 149. 150. 181. 200. 224. 225. 228. 229 Engstrom. Howard T.. 174 Enigma (Kozaczuk). 196. 202. 219 Hmgma messages: color designations of. 56. 87. 123-124 composition of. 35-37 Enigmas. Enigma machines. 19-21. commercial. 27-28. 41. 208. 211 components of. 40-43. 204-206 encoding and decoding procedures for. 45-49. 97 German method of using. 38-39. 43-46, 125 German modifications of. 125. 136-137. 148 German naval. 203. 219 Great Britain secures replicas of. 12-16 invention of, 27-28 keyboard of, 38.39-40. 99-101. ' 102. 209 keys to. 43-45. 123-124. 129-130. 205-210 lamp boards of. 38. 39-40 mechanical and electrical functioning of. 39-43. 49-52. 63. 168-169 ring settings of. 42. 43. 67-69. 98 scramblers of, 39. 40-4.3. 47. 63. 78n. 295-299 scrambling method of. 11-12 in Spanish Civil War. 221 steckerboards of. 39. 43. 44. 45. 47. 63. 168-169. 204-205. 208 terminology, differences in. 204-206 text settings of. 45. 98. 102. 164 HA-<."/;wa/r<."of,43. \ 13-115. 133 137. 148. 205. 209. 211. 213 235-2.37. 247 Hmgma War, The (Garlinski). 196. 203 cp (an passant). 136 Evvmg. Sir Alfred. 8 Faulkner. Hubert. 10. 72 female indicators. 60. 62-69. 72. 97. 164. 214-215. 226. 231 'fists', in interception. 153. 157. 225 Fitzgerald. Penelope. 9. 33. 34 "Flag Code'. 33 Fleteher. Harold. 85. 94. 126 135. 188 as administrator of bombe project. 139. 141, 145-147 recruited for Hut 6. 121-123. 201 as source. 122. 128-129. 138. 144 Flowers. TH.. 178-179 Foreign Office. British. 89 (see also Hut 3) Foss. Hugh. 179. 221 France. Battle of. 88. 95-96 Freeborn. Mr.. 176. 181-182 Friedman. William F.. 10-11. 26. 174. 202 Functions of tin; Executive. The (Bamardi). 283;; Garlinski. Jozef. 196. 203 (iaunt. David. 85 Gayhurst Manor, bombes at. 139. 141 Gazala. 201 GCCS (see Government Code fe Cypher School) GCHQ (see Government Communications Headquarters) Geheimschreiber. 177 General Motors. 171 George VI. King of England. 128 German radio communications. 19-21. 88 Enigma used in, 19. 20. 21. 38-39 errors in, 163-169. 207. 208. 213. 220. 2.30. 231 lazy operators in. 98-99. 102. 1 1. "113-114. 119. 130. 132. nets. 205. 208. 225. 229 simplex nets in. 24-25. 52-53 94-95. 120-121. 124. 152-154. 155-156 Germany: air force of. 19. 88. war fronts extended by. 120. 123-124 war preparations of. 17-20. 24-25 "Glowlamp Machine' (see Enigmas) Goenng, Hermann. 101 Gort. John. Lord. 96 Government Code & Cypher School (GCCS). 8-10. 33. 197. 221. 223 Government Communications Headquarters (GCHQ). 10. 34. 179-181. 183 Great Britain: billeting problems in. 122. 186 blackouts in. 129 Great Britain (cont.) Enigmas secured by. 12-16. food rationing in. 185-186 war preparations of. 7-10 Green traffic. 222. 227. 228. 229. 230 Guderian. Heinz. 20. 21. 38. 39. 88 Hall. Sir Reginald "Blinker'. 8 Hart. Sir Basil Liddell. 18 Hempsted. Pat. 103."3 Herivel. John. 85. 98. 165. 200. 220. 223. 230 "Herivel square'. 99. 100. 112-113 Herivel tip. 98. 101. 102. 104. 112. 165. 167. 168w. 200. 220. 221-222. 223. 226. 231 Hmsley. Harry. 201. 204. 222. 223. 227. 231 Hitler. Adolf. 7. 8. 18. 101. 120. 241-242 Hodges. Andrew, 197 Howard. Jean. 223-224. 226 Humphreys. Wing Commander. 93 Hut 3. 90. 103. 111-112. 125. 129. 199. 200. 201. 202. 219. 223. 224 Battle of France reported by, 95-96 expansion. 224 function of. 58. 159-161 Hut 6 priorities set by. 133-134. indexing system of. 160-161. 224 staff of. 93-94. 134 Hut 6 and Hut 6 Ultra. 90. 188. 199. 201. 206. 219. 221, 222. 226. 227. 230. 235 beginnings of. 35. 58. 198. 220 chess players in. 84. 85. 183 defined. 58 described. 10 effectiveness of. 101. 120 management team of. 126. 135. 158 in 1940. 84-120. 164-165. 175. 231 in 1941. 121-124. 138-139. 150. 166. 201 in 1942. 124. 129-134. 139-141. 150. 152. 157-159. 224 in 1943. 124-129. 135-137 150. 152 recruitment for. 84-86. 121. 200.201 teamwork in. 126-127 Hut 8. 58. 125. 199. 201 Hut 11. 139 I.D. 25 (Room 40). 8. 33 indicator settings, 206. 209 purpose of. 36. 46, 59-60 in Sillies. 102 indicators, female. 60. 62-69. 72, 97, 164,218.225 Intelligence Service Knox (ISK), 219, Intercept Control Room. 76. 84, 89 90, 103,"3. 126 interception, difficulties of. 37. 53. 56, 74. 111. 120, 153 interservice friction, 174-175 "Intrepid' (Sir William Stephenson), 19, 175-176 Introduction to Algebraic Geometry (Welchman), 12 ISK (see Intelligence Service Knox) Italian naval cipher. 219 JAB JAB 102 Jeffreys. John, 33. 71, 72. 85, 89-90, 103, 119, 125. 197, 200, 230 Jeffreys sheets (see perforated sheets) John Lewis Partnership. 183. 190 Johnson, Brian. 204 Jones. Gp Capt, 224 Jones. Reg V.. 102. 204 Jones. Sergeant. 147 Journal of the U.S. army War College, 203 Kahn, David, 25, 27. 174, 177 Kasparek. Christopher. 196. 203, 214. 219 Keen, Harold "Doc'. 81. 138. 139-140, 142-144. 148. 205. 218. 236. 241, 244. 247. 249 Kendnck. Alex. 34-35, 37, 53 Kendrick. Tony. 197. 217. 219, 220 Kerckhoffs. Auguste. 25-26 keys: in cryptology, 25-26 (see also Enigma, keys) (see also indicator settings) kisses. 136 Knight. Harold. 9.3 Knox. Dillwyn "Dilly', 9. 11. 15, 16, 31-34. 7.3. 195. 196. 197.204, 206. 217. 218. 219.220. 222,230 contacts with Welchman, 33, 34, 52, 71-73, 77. 244 Knox Brothers, The (Fitzgerald), 9, 33 Knox Method. The. 221-222 Korda. Alexander, 175-176 Kozaczuk. Wladyslaw. 196. 202-203, 214.219 INDEX /261 "La Guerre des Messages Codes (1930-1945)' (Stenaers). 14. 96 Langer. Guido. 13. 203. 227-229 Laughton-Matthews. Mrs.. 146-147 Letchworth. bombes manufactured in. 123. 139. 140 letter substitutions. 47. 48 Lewin. Ronald. 13. 14. 33. 58. 88. 146. 202. 204. 227 Lewis. Philip. 158 Lewis. Spedan. 183 L'Histoire. 14. 96. 196 Lisicki. Tadeusz. Col.. 196. 203. 218. 221 log reading. 95. 155. 157. 223. 225 Lucas. FL.. 93 Lucy. Diana. 15 1-153 Luftwaffe. 19. 88 MacArthur. Gen.. 202 McCormack. Alfred. 202 Machine Room. 76. 89-90. 97. 99. 103-104. 108. 126. 200. 231 Magic (Japanese cipher machine), 202 Man Called Intrepid. A (Stevenson). 13. 19. 175. 176 Manisty. John. 85 Marconi. 55 Mayer. Col Stefan. 203 menus. 104. 106-108. 109. 141. 143 Menzies. Sir Stuart. 9. 14. 94. 226. 230 Meteo Code. 221-222 MIS (Military Intelligence). 94-95. 124. 223". 225 Milch. Erhard. 19-20 Mimer-Barry. Stuart. 84-85. 98. 110. 120. 121. 126. 188. 226. 250. 251 Missing Dimension. The. (Andrew & Dilks). 196 MITRli Corporation. 2. 183 Monroe. John. 121. 122-123. 129. 136. 138 Morgan. G.W. "Gerry'. 176. 177 Morse code. 24. 25."27. 74. Ill Munich Agreement, 7-8 Myer. Albert James. 22. 23. 25 Navy. U.S.: bombes used by. 135. 148 crvptolocistsm. 173. 174-175. 179 Newman. Max. 176-178 Newman, Patricia. 57 Oeser. Oscar. 134-135. 223-224. 226 Pacific, war in the. 201 Panzer divisions: radio communications essential for. 19. 20. 22.3. 226 (see also World War II) Parker. Reg. 113. 131. 222 Parkerismus. 131. 167. I68w Passant. James. 35 Pearl Harbour. 202 perforated (Jeffreys) sheets. 199. 219. 220. 221. 2.30 Jeffreys' development of. 71-72. 89-90. 95. 164. 169. 197. 199 Polish development of. 15. 16. 197. 206. 213. 218. 229 rendered useless by Germans 96 119.231 Welchman's theories of. 65-72. 197 (see also Sheet-Stacking Room) Phony War. 74. 76. 88. 94. 186 Polish Cipher Bureau. 207-208. 21 1 buys commercial Enigma. 208 Polish contributions compared to British. 202-204. 218-219 Polish cryptologists. 12-16- 28. 77. 195-200. 202-223 Polybius. signaling system of. 22-24. 25 preambles. 35-36. 52. 53. 97 Pyry Forest meeting. 13-14. 16. 73. ' 197. 199. 207. 215. 217. 218. 219. 220. 221. 223 Q-Codes. 154-155 radio, 24. 27 shortwave transmission of. 55. 154 (see also Communications: German radio communications) Radlev. Mr. 178-179 R.A.F (see Royal Air Force) Randell. Brian. 174. 177-179 Red traffic. 87. 88. 90. 1 19-120. 227 228. 230. 231 Rees. David. 85. 114 Registration Room. 76. 89. 90. 103. Ill, 126. 1.36. 226 Rejewski. Marian. 14-17. 40. 195. 196. 197. 199. 200. 202-222. 227. 230 develops theories to analyse Enigma. 209-210 invents cyclometer. 212 Remaking of Modern Armies, The (Hart). 18. Report on Colossus (Randell). 174. 177 reversibility. 47-48 Ridley. Captain. 10 Robinson. 178 Robinson. Heath. 178 Robson. A.. 85 Rockex. 173 Rommel. Erwin. 123. I'll. 201 Room 40 (1.0.25). 8. 33. 221 Rowlett. Frank B.. 174. 202 Royal Air Force (RAK). 101-102. 150 (sue a/jo Cheadle: Chicksands) Rozycki. Jerzy. 14. 15. 195. 196. 197. 208.211.216. 221.227. 230 'sanitized', defined. 92w Saunders. Malcolm. 93-94. 134. 139. 141. 224 Saver. Hurry. 95 Schcrbius. Arthur. 27-2X Schmidt. Hans-Thilo. 15 School. 31. 34-35. 52 (.tvu /.; Hut 6 and Hut 6 Ultra) scramble, electromechanical production of. 11-12 scrambler. 235-239S'ecTtV H' r. TVn' (Johnson). 204 Shaf(esbur\. intercept station at. 150. 152 Sheel-stucking Room. 75. 76. 90 Shepherd. Wing Commander. 149. 150. 224. 225 S'ic/n'njif.st/icM*; (SO). 207. 216 Siemens machine. 144. 177 signaling systems. 22-25 Sillies (se Cillis) SLUs (.*