Gordon Welchman Page 5
Rejewski was the most academically advanced of the three young students who had been recruited into BS 4. He had already been awarded his degree in mathematics and had spent a further year studying the subject in depth at the University of Göttingen in Germany. In October 1932 he was given a separate small room to renew the studies of Enigma abandoned by his predecessors. He was also given the commercial machine which had been purchased earlier as well as several dozen messages daily, encrypted on the military version of Enigma. In early December he received a photograph of a military version of Enigma, user instructions and a schedule of daily keys for September and October 1932. By the end of the month he had achieved one of the greatest feats in the history of cryptography. Marian Rejewski had reconstructed the internal connections within the Enigma machine and identified the indicator system currently being used by the German Army. He had done it entirely using a mathematical technique called permutation theory. Meanwhile, Zygalski had devised a technique which in essence involved cutting square holes in sheets of paper with the positions of the holes determined by some of the data in the indicator at the beginning of an intercepted message. By stacking the sheets produced from a number of messages on a light box, the holes that were aligned would be revealed by the light shining through them. This could be used to work out the ring settings that had been used as part of the daily key.
Early the following year, Rejewski, Zygalski and Różycki were brought together again to work together as a team. They fairly quickly began to read encrypted German Army messages and had considerable success for a number of years. The German Air Force introduced Enigma in August 1935 and the number of different networks to monitor grew rapidly. By 1 February 1936 the Germans had introduced significant changes and complicated safeguards to Enigma. The following year the Polish General Staff transferred BS 4 to a camouflaged high-security new headquarters in the Kabacki Woods near Pyry outside Warsaw. As the Germans changed procedures or tightened their security, the Poles would invent technologies such as the Zygalski sheets and new machines like the cyclometer and the bomba (see Chapter 3) to counter them. When the German Air Force and Army issued their Enigma operators with two additional wheels, on 15 December 1938, the Poles began to struggle even more. With three wheels available to the operator, he could choose from only six possible configurations. Now he could chose three from five which increased the number of possible wheel configurations to sixty. To make matters worse, the Germans had also increased the number of plugboard connections to ten.
The Poles proposed a meeting with the British and the French in the hope that they would have something to contribute as the situation was getting worse. The first tri-lateral meeting between representatives of the cryptographic services of France, Britain and Poland was held in Paris on 9 and 10 January 1939. In attendance on the Polish side were Langer and Ciçżki; on the French side Bertrand and a cryptanalyst called Henri Braquenie; on the British side, Denniston, Knox, Tiltman and Hugh Foss, a cryptanalyst who had joined GC&CS in 1924. The Poles had been instructed not to reveal anything unless they got something in return. As the British and the French had nothing to offer, the meeting, while cordial, was a waste of time.
By May, tensions between Poland and Germany were close to breaking point and, on 30 June, Langer contacted London and Paris with the news that something new had come up since January. He proposed a second trilateral meeting in Warsaw on 24–27 July. As Knox had been included in the invitation, Sinclair instructed Denniston to include him in the British delegation. Also included was Commander Humphrey Sandwith, head of the Admiralty’s interception service.18 Knox and Denniston arrived on the morning of the 24th and stayed at the Hotel Bristol. The French stayed at the Hotel Polonia. The Poles entertained their visitors at lunch at the Hotel Bristol and ironically the fairly banal conversation was conducted in German as it was the only common language of all in attendance.
The key meeting took place the next day at the Pyry Centre. Much to the astonishment of the British and the French, the Poles demonstrated several machines and techniques which they had developed to help break Enigma keys. The news that the Poles were breaking German Enigma keys quite regularly was not received well by Knox, who maintained a stony silence throughout the meeting. Knox, who arguably knew more about the Enigma machine than anyone in Britain, had been unable to break the new military version with the plugboard. His problem had been the connections between the keyboard and the entry drum inside the machine. On the models of Enigma machines that he had successfully broken, the connection pattern followed the order of the keys left to right, row by row and alphabetically around the entry drum. So the Q, W, E, R, T keys were connected respectively to A, B, C, D, E, and so on. On the model of Enigma in mass use by the German Army and Air Force, the connection pattern had been changed and Knox’s team (which included Turing) could not work out the new pattern.
This problem had also stumped Rejewski initially and he described his solution in a paper written in 1980:
What, then, were the connections in the entry drum? It turned out later that they can be found by deduction, but in December 1932, or perhaps in the first days of 1933, I obtained those connections by guessing. I assumed that since the keyboard keys were not connected with the successive contacts in the entry drum in the order of the letters on the keyboard, then maybe they were connected up in alphabetical order; that is, that the permutation caused by the entry drum was an identity and need not be taken into account at all. This time luck smiled upon me. The hypothesis proved correct, and the very first trial yielded a positive result.19
So when they had all gathered at the Pyry Centre, Knox’s first question to Rejewski had been; ‘What are the connections to the entry drum?’ Knox was furious when he heard the answer: ‘A, B, C, —’. In other words, the Germans had wired it up in the simplest possible way, the Q key to Q, the W key to W, the E key to E, the R key to R, and so on.
Denniston later wrote a report on the Pyry conference on 11 May 1948 from memory and using his pocket diary to check dates:
It was only when we got back in the car to drive away that he [Knox] suddenly let himself go and assuming that no one understood any English raged and raged that they were lying to us now as in Paris. The whole thing was a pinch he kept on repeating – they never worked it out – they pinched it years ago and have followed developments as anyone could but they must have bought it or pinched it.20
Knox remained aloof and withdrawn over dinner that night, almost as if he had a grudge against the Poles. The next day, Knox met Langer, Ciçżki, Rejewski, Zygalski and Różycki and was apparently ‘his old self’. However, that still didn’t stop him from writing to Denniston a few days later saying that the Poles had ‘got the machine to Sept 15th 38 out by luck. As I have said only Mrs B. B. had seriously contemplated the equation A = 1, B = 2. Had she worked on the crib we should be teaching them.’ Despite his grumpiness, he ended his letter with kind words for Rejewski, Zygalski and Różycki: ‘The young men seem very capable and honest.’21 Unfortunately, Mrs B. B. has never been identified.
In the end, Denniston attributed Knox’s fit of pique to the formality of the meetings held on the first day of the conference and pompous declarations by senior officers. Knox and Denniston had been friends and colleagues since coming together in Room 40. Yet in a letter to Bertrand, dated 3 August 1939 and written on Hotel Bristol note paper, Denniston is remarkably candid about Knox:
My dear Bertrand,
I have finally had a day off and I take this opportunity to write to you a very personal letter, ‘from the heart’, which seems necessary to me.
I have seen D [Dilly Knox]; in his opinion I may have said something bad about you and that is why I wish to emphasise that we owe everything solely to you and I look forward to the co-operation of our trio and that to reach our goal you must remain in the leading position. In Warsaw it was you who advised me to return and think about it – and you were right.
Maybe you underst
and my problem in the shape of Knox. He is a man of exceptional intelligence, but he does not know the word cooperation. You surely must have noticed that off duty, he is a pleasant chap loved by all. But in the office his behaviour is different.
In Warsaw I had some deplorable experiences with him. He wants to do everything himself. He does not know how to explain anything. He can’t stand it when someone knows more than him. Unfortunately, I cannot do without him, he knows more about the machine than anyone else in the country. He built a machine of the type used by the Spanish, and frequently by the Italians in Spain, which is not to be sneered at, even if not so much has been done as has been done by our friends in Z [the Polish Cypher Bureau].
You must forgive me for being so keen to keep him, but I will tell you in all sincerity, that I will never take him to a conference again if I can only avoid it. From now on, we must establish the rules of our co-operation in order to avoid unnecessary effort.22
On 10 September 1939 the Poles closed the Pyry Centre, destroyed all trace of their machines and evacuated their workers. The Second World War had begun. Ten days later, Bertrand delivered a replica Enigma to Colonel Stewart Menzies in London as a gift from the Poles.
One cannot overstate the Polish contribution to the ultimate success of BP. They had recognized in the early 1930s that the age of machine cryptography had begun and that mathematicians would be needed to deal with it. While it is likely that Knox came away from the Pyry conference with the missing link in his attempt to re-construct the Enigma machine itself, Denniston clearly saw the bigger picture and it is to his credit that he too had already started to recruit mathematicians for GC&CS. More importantly, the Poles had demonstrated to the British that encryption machines like Enigma could be broken, if the right mathematical minds were allowed to concentrate on the problem.
In early drafts of The Hut Six Story, Welchman had carefully avoided saying anything critical about Knox, though after reading Penelope Fitzgerald’s account of Knox and his brothers, he felt he could say something about his treatment at Knox’s hands. He used carefully chosen words such as being ‘turned out of the Cottage’ and more tellingly:
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.
In her excellent and loving biography of Knox, Mavis Batey argues that, quite to the contrary,
The records show that Welchman was not ‘banished’ to the School because, as he thought, Dilly did not like him. Where Dilly was concerned, lack of communication was not a sign of dislike but merely of total absorption in a project to the exclusion of all else.
She goes on to quote from a confidential note from Knox in November 1939 in which he says that Welchman ‘was doing well and is keen. I hope to get him back here to learn about the machines’.
Unknown to Mavis Batey, who was one of Knox’s team in The Cottage and a brilliant cryptanalyst, Welchman had other reasons for thinking that Knox disliked him. Welchman told Winterbotham in January 1975 that Knox had not been happy with his proposed reorganization of BP’s codebreaking activities.23 He had gone to Travis with the proposal and Travis had then got him the full support of Tiltman and Cooper, so he was allowed to go ahead. He went on to say:
Incidentally, this early initiative on my part was never forgiven by Dilly Knox, but please do not mention this to anyone. Dilly was much loved by all of us, and he made a tremendous contribution. On the other hand Denniston soon became enthusiastic about what I was doing.
Denniston’s support had obviously been important and as Welchman said to Denniston’s son Robin in February 1979:
I had the impression that Dilly Knox disliked me from the start, and may well have complained to your father about my assignment to the Cottage.
Another matter of great personal interest to me is that your father gave strong support to my plans for the creation of Hut 6, and must have followed our progress closely. Before his illness he called me to his office, congratulated me on my achievements, and assured me that I would be rewarded after the war.24
Welchman had also corresponded with another Cottage veteran, Peter Twinn, after The Hut Six Story had been published.25 In answer to Twinn’s question about what Welchman thought of Dilly he said:
Over Dilly, I intend to quote Rejewski’s feeling that he was very quick to grasp what the Poles told him, which suggested that he had got a long way himself. But I would like to say more than that. David Rees, who was one of the Sidney Sussex mathematicians that I recruited in the very early days, left Hut 6 to join Dilly. He felt bad about Dilly’s insistence that he must not tell me about a success. He did tell me about it in confidence, and, though I remember no details, I was very impressed by the brilliance of what Dilly had done. I gathered that Dilly was afraid that I would jump in and take the exploitation of his success away from him. It is sad that he should have felt this way, but he was a very sick man.
So what is one to make of Alfred Dillwyn Knox: an amiable but absentminded professor, or an ill-tempered and secretive loner? His eccentric behaviour has been well documented as has his secretive nature and his apparent dislike of men. Even one of his closest colleagues, Peter Twinn, has said that Knox told him very little. The National Archives hold a number of letters from Knox to Denniston with threats of resignation over one issue or another.26 There was also, of course, his health, as Knox was ill during his time at BP. Set against this was his sheer brilliance as a cryptanalyst. He had broken the basic military Enigma without plugboard on 24 April 1937 and, as Twinn said proudly of the work of Knox’s team: ‘Three different species of Enigmas were solved by us without any help from anyone except the incompetence of the German cypher authorities and their operators.’
It would seem that differences between Welchman and Knox were not personal as both men said complimentary things about the other. Knox was one of the people to whom Welchman dedicated The Hut Six Story. It is clear however, that they had a different vision of the future of cryptography and how best to deal with the threat posed by Nazi Germany. Quite simply, Knox’s approach would not have worked when thousands of intercepted messages came flooding into BP on a daily basis. What remains a mystery is why Knox was unable to make the same guess as Marian Rejewski about the connection pattern of the Enigma keyboard to the entry drum. At the end of 1984 Welchman received a letter from an American academic, author and expert on machine cryptography, Cipher Deavours. In it, Deavours provided an interesting slant on the connection pattern, or as he called it, the entrance permutation:
This business about Knox and the entrance permutation has always puzzled me to no end. Had the British been about their proper intelligence gathering business, this snag need never have happened.
The Enigma company had four models of the machine. The machines were publicly exhibited at both the 1923 Congress of the International Postal Union in Berne, and at the 1924 congress held in Stockholm.
Enigma ‘A’ and ‘B’ were Scherbius Enigmas each employing a common cryptographic method. These machines each had four rotors which were driven by four ‘gaptoothed’ gearwheels of sizes 11, 15, 17, and 19. Each wheel had 6 gaps. The rotor movement was, thus, very irregular with more than one rotor usually moving at once and other rotors pausing. Had this rotor movement afforded more variation, the machine would have been indeed very strong cryptographically. This model of the Enigma was used by the Abwehr during the war.
Enigma ‘C’ and ‘D’ were the glow lamp models with which we are all familiar. Enigma ‘D’ was the widely sold commercial model and it was this version which was used by the Italian Navy (rewired of course) and solved by the British. Enigma ‘C’ seems to have been the model from which the Wehrmacht modification came. This ‘C’ model has the same reflecting rotor construction as the German military version with the slight difference that the reflecting rotor could be s
et in one of two possible positions internally. (Model ‘D’ had a completely settable reflecting rotor.) The point here is that both the keyboard and the glowlamps in Model ‘C’ were in the standard A-Z sequence. Had Knox seen pictures of the machine, he would immediately have seen the possibility that the entrance permutation was not based on the typewriter sequence as in Model ‘D’. Why he did not have this information is beyond comprehension. First, the machine photos were publicly available and second, in 1927, Dr Siegfried Türkel published the work ‘Enciphering with Apparatus and Machines’ (English Translation) which contained numerous photos of all Enigma models from ‘A’ to ‘D’. It would seem impossible for the British to have completely missed the Türkel work.
As you know, the British solved Enigma ‘D’ messages of the Swiss government during and after the war. The method used probable text beginnings in German and French and was equivalent to the ‘Baton’ method. The British obtained the wirings of the Swiss machine through bribery. Knox must have been involved in this and so it seems likely that this was his method. The Baton method sometimes works even when the plugboard is present in the machine, particularly if only a few steckers are in use. I would imagine that Knox tried to refine this approach for better results. Given his background in manual cryptography, this would be quite natural.