Dilly

by Batey, Mavis;

  The ‘bombs’ were an electro-mechanical machine which the Poles had devised as a means of breaking into Enigma. Dilly’s anger was, in all probability, caused largely by the fact that he had yet to meet anyone on the Polish side who had the faintest idea how to break Enigma. The next day, Thursday 27 July, saw a complete turnaround to the point that Dilly would never mention the previous day’s unfortunate meeting again. The administrators were to meet and make arrangements for the Enigma treasure trove to be transported to Paris and London, which would ensure that GC&CS could take over where the Poles had left off, with the possibility of collaboration when war broke out. Meanwhile, Dilly and Henri Braquenié were taken to the cipher room itself to meet the people who had actually broken the Enigma machine. A worried Denniston managed to break away from the executive meeting to look in on Dilly and was mightily relieved to find that ‘Knox was really his own bright self and won the hearts and admiration of the young men with whom he was in touch. If only that first day of formal disclosure could have been avoided and pompous declarations by senior officers had been omitted, Knox’s mind and personality in touch with men who really knew their job would have made that visit a very real success.’ The men Dilly met were Marian Rejewski and two younger colleagues, Jerzy Różycki and Henryk Zygalski.

  Dilly’s first question to Rejewski, who had broken Enigma theoretically as early as 1932, was, predictably, ‘Quel est le QWERTZU?’, and one can only imagine poor Dilly’s terrible shock when Rejewski told him that it was not random, as he feared, but just ABCDE. This was made worse by the fact that a certain ‘Mrs B.B.’, whose identity has not been traced, had actually suggested this possibility, but Dilly thought the Germans were not stupid enough to make it that easy! ‘Had she worked on the crib we should be teaching them,’ he lamented to Denniston that night in a letter written on the hotel notepaper. Dilly warned Denniston that although the Poles had done ‘very well’, changes to the German use of Enigma that took place in November 1938 had put an end to the Polish success, showing up the limitations of their system of breaking the messages. ‘We should examine their system and statistics (if any) with considerable skepticism,’ Dilly said. He clearly distrusted Ciężki but had the utmost respect for the three young mathematicians, Rejewski, Różycki and Zygalski, who had done the work. ‘I am fairly clear that Ciężki knows very little about the machine and may try to conceal facts from us,’ he told Denniston. ‘The young men seem very capable and honest.’

  The respect was reciprocated. Rejewski soon recognised Dilly Knox as an ace cryptographer. In an interview in 1978, he vividly recalled the meeting in the building in the woods at Pyry. ‘Just how much Braquenié understood I don’t know, but there is no question that Knox grasped everything very quickly, almost quick as lightning. It was evident that the British really had been working on Enigma … They were specialists of a different kind – of a different class.’ Rejewski also recalled the jubilation about their future co-operation and although alcohol was forbidden in the centre, a few bottles of beer were found and they drank to their joint success. Braquenié tells the story of going back to their hotel together with Dilly chanting: ‘Nous avons le QWERTZU; nous marchons ensemble.’

  As soon as Dilly got home – having been delayed for a day in Poznan because his papers were not in order – he sent a letter of appreciation to his new Polish friends with his sincere thanks for their co-operation and patience. Rejewski gave a copy of the letter on Foreign Office notepaper to Władysław Kozaczuk, who published it in 1984 in his book Enigma. As Rejewski had been intrigued by Dilly’s rods and the way they worked, Dilly sent him a set: the French had already made themselves a set of ‘batons’. Rejewski had never thought of this way of breaking Enigma, but in his interview he did not doubt that ‘Knox’s method’ would have succeeded although ‘unfortunately we got there first’. Despite his scepticism following the Ciężki lecture on the first day, having met Rejewski, Różycki and Zygalski and discussed their methods with them, Dilly had fully accepted that they ‘got there first’ and together with the rods he also sent three silk scarves showing a horse winning the Derby. The Poles treasured these souvenirs, one of which remained in the possession of Różycki’s widow and, at Rejewski’s centenary celebrations at the Foreign Office, Marian’s daughter, Janina, told me that she had always known about the gift. It was indeed a nice gesture on Dilly’s part to use the scarves to congratulate the Polish codebreakers on winning the race and must have made Denniston, who – in spite of frequent disagreements – always managed to remain on good terms with Dilly, feel better about the Warsaw visit.

  In a memo written immediately after returning from the Warsaw conference, Dilly said:

  At Warsaw, my French colleague and I were admitted into the Bureau itself and a course of lectures was arranged for us. After he had gathered the fact that ‘le QWERTZU, c’est A, B, C’, he took little interest and allowed the lectures to proceed. I tore up the programme and we proceeded by question and answer, writing these down in an odd variety of French. The expert staff of three were charming young men straight from the university. They had a good supply of mechanical gadgets. They were understaffed. They answered all my questions truthfully. What are the facts? When Major S [Ciężki] paraded before us in Paris his secrets de Polichinelle we took him for an ass; in fact he was bluffing. The Poles, getting in at the start had been reading the Enigma till 15th September 1938; at the time of arranging the Paris meeting they may have been in difficulties; at the actual time of that meeting things were going fairly well; they had got out the two new wheels and were reading SD [Sicherheitsdienst – the Nazi Party’s own intelligence service] and military messages. In January, they ceased to be able, or had not enough staff, to read the latter; naval messages had not been read since May 1937. Discovery of le QWERTZU was effected (a) durch Verrat [through treachery], i.e. purchase of details of setting (? and the ‘crib’ message), (b) par hasard (they might have claimed par inspiration) … Doubtless they had also the ‘key-block’ and thus easily equalled the values; in which case it is just possible that they had enough. Anyhow inspiration served. Mr Twinn has just checked up on the genuineness of the crib. Such constatations are important as an answer to the question ‘Is the Enigma safe?’ The answer is that Enigmas of this pattern are definitely unsafe (since a thousand persons may handle the ‘day-setting’, memorise and betray the vital portions) unless the concealed and still more vital QWERTZU (diagonal) is such as cannot be guessed.

  Dilly added that some of the Polish methods were ‘neat’ and tended to rely on the use of the ‘bombs’, which they called bomby. This had spawned in Dilly’s mind the idea of a different type of electro-mechanical machine:

  At the time of my visit I had ideas which seemed to be better, and I have since discussed them exhaustively with Mr Turing and Commander Travis and we believe that we could produce a really good alarm machine – the Geschlechtzylinder. This would give solutions for Army keys till (probably) March 21st: after that perhaps not. Almost all our guesses have been right except one of mine that there might be a naval book. I think we may hand some bouquets to the Poles for their lucky shot, but far more for their surmounting the difficulties after 15th September if only for two months. I could not possibly have finished the tasks I gave myself but for AGD’s [Denniston’s] help even in technical matters. I recommend strongly that if we have the Poles over here we ask AGD to invite one of the young men.

  How far Dilly had been from the winning post himself is shown by the fact that when he received the QWERTZU information, Peter Twinn was able to confirm within two hours that Bertrand’s operators’ manual crib was after all authentic; the wiring of the first two wheels could then be determined, using Dilly’s ‘buttoning-up’ method, but now with an ABCDE diagonal. The way was now open to tackle another one of the ‘Scarlet Pimpernels’ from Bertrand that they had been given, described by Josh Cooper as practice messages ‘made by an agent with access to an E machine’ with no Stecker in pl
ace. Another three days’ work produced these decodes, albeit trivial ones, and Cooper congratulated Twinn for being the first British codebreaker to read a German service Enigma message. Peter himself said modestly:

  I hasten to say that this did me little if any credit, since with the information that Dilly had brought back from Poland, the job was little more than a routine operation. Indeed I had come out of my first few months at GC&CS extremely badly. If I had thought of making the guess, which in retrospect seems such a reasonable possibility, that QWERTZU was simply ABCDEFG, as Rejewski had done in 1933, I could have read a few messages within a fortnight of my arrival at GC&CS and made a debut of unparalleled brilliance. Alas! I failed to grasp my chance. I can only say in defence – and it is not much of a defence – that Dilly Knox, Alan Turing, and Tony Kendrick did no better, and none of that trio was known as an unimaginative dullard. Of course, reading a few scattered messages on a single day in 1938 was a whole universe away from the problems that lay ahead, but it would have been some encouragement; had we thought of it in February 1939, we should have saved ourselves six valuable months.

  He seems to have forgotten the mysterious Mrs B.B.’s feminine intuition! Although perhaps he never knew.

  The Polish treasure trove, which included the replica Enigma machine, arrived at Victoria station on the Golden Arrow on 16 August under diplomatic seal, the arrangements having been made by ‘Biffy’ Dunderdale. It was handed over by Bertrand to Colonel Stewart Menzies, the deputy head of SIS, who was wearing the rosette of the Legion d’Honneur in his buttonhole. ‘Accueil triomphal’ was how Bertrand remembered it. As soon as he returned from Warsaw, having thanked the Poles, Dilly’s first thought was to get in touch with Alan Turing to tell him how Rejewski had originally solved the Enigma problem and for an inquest to be held into their lack of success. Closeted in Dilly’s study at Courns Wood, they went over the whole story step by step. It is clear from the report he had written on 4 August after Turing’s weekend visit that what really bothered Dilly was that Rejewski had, like Dilly, assumed that the diagonal was as in the commercial machine QWERTZU, but when this would not work on any of the messages he had, he guessed ABCDE as the entry to the wheels. However, Dilly thought that the solution, which had so held him up, could probably have been arrived at mathematically. Turing recognised that Dilly had a good grasp of mathematics, albeit a disregard for evaluating probabilities, and reassured him that this was very unlikely with the limited amount of information GC&CS had. Rejewski himself had never denied that the ‘intelligence material furnished to us should be regarded as having been decisive to the solution of the machine’.

  It was Rejewski, however, who was the first to apply mathematics to cryptology through permutation theory and it was Ciężki who recruited him from Poznan University’s Institute of Mathematics in 1932 to work part time to seek a solution for Enigma. He seems to have produced his theoretical result independently, a few months before the arrival of Bertrand’s intelligence material. Rejewski’s permutation theory was not published until 1979 but Gordon Welchman, a mathematician from Sidney Sussex College, Cambridge, who joined Dilly Knox’s section in September 1939, was able to see the early history of Enigma breaking in perspective. When he came to write The Hut Six Story in 1982, having been its first head, he professed admiration for Rejewski’s initial solution by permutation theory; however, he went on to say that the time required to turn this into codebreaking made his mathematical solution impractical and it was only when Bertrand’s ‘pinches’ appeared on his desk that he was able ‘to put this into practice’, adding that ‘it seems that Dilly Knox in England may well have arrived at a comparable theory, but he did not have access to the Asché documents’. Welchman dedicated his book to Rejewski and Knox, who had contributed so much to the early history of Hut 6. It appears then that Dilly’s boast to Menzies that he believed himself to be ‘the only man who can claim to have reconstituted with workable accuracy an Enigma machine without any purchase by Secret Service Funds’ may well have been correct at that point. When Welchman came to recruit recently graduated Cambridge mathematicians for Hut 6 in 1940, he is on record as saying that the work they would be doing ‘did not really need mathematics but mathematicians tended to be good at it’.

  Before the release of certain classified material in 2001, a misconception had arisen among historians’ interpretations of the information then available that although Dilly successfully broke un-Steckered machines he was ‘defeated’ by the German Steckered Enigma owing to his lack of mathematics and that success was only achieved with the advent of the mathematicians Turing and Welchman. Even Stuart Milner-Barry, a key Enigma player, who joined Hut 6 early in 1940, described the codebreakers at Bletchley Park as consisting of

  a few old-time professionals who had worked in Room 40 at the Admiralty. Such for example as Dillwyn Knox, a Fellow of King’s who died during the war … and new recruits such as Welchman and Alan Turing. Knox had so, I understand, been defeated by the Enigma, and the main credit for solving the Enigma and subsequently exploiting its success should (subject to the Poles) probably go to the other two.

  This, as we now know, is simply not true. Defeat was not a word in Dilly’s vocabulary and it was only those who actually knew Dilly’s work on Enigma before the war and the setting up of the operational huts, notably Denniston, Tiltman, Foss, Bodsworth, Kendrick, Twinn, Cooper, ‘Nobby’ Clarke and Birch, who knew how much he had achieved. Dilly’s early experimental methods for breaking Scherbius’s original commercial machine had been a vital first step, allowing later complications to be dealt with as they arose.

  Although Rejewski had been given a commercial machine soon after he arrived in the Cipher Bureau in September 1932, he went straight on to an analysis of the indicators on the German army messages enciphered on their Steckered machine, from which he would produce his mathematical theory; this would only hold good for the period of the fixed Grundstellung, when the six columns of repeated indicator letters for that particular setting, known collectively as a key-block, were in fact substitution ciphers. His detailed analysis of the indicators left him with six complex equations, each with three unknowns: the Stecker, the wiring to the entry plate and the wiring of the right-hand wheel (for the time being he could disregard the wiring of the other two, which only moved occasionally). Bertrand’s gift of key settings for September and October 1932 revealed the Stecker for those months. Like Dilly, he too attacked the Wehrmacht machine on the basis that the diagonal was the same QWERTZU pattern used on the commercial machines. He was foiled for several weeks, but eventually tried the idea that it might be alphabetical. ‘This time, luck smiled on me,’ he said. ‘The hypothesis proved correct.’

  His correct guess that the diagonal was in alphabetical order meant Rejewski had solved the wiring of the right-hand wheel and because Bertrand had supplied the keys for two different months, using different wheels, he was quickly able to work out the wiring of the remaining wheels and the reflector. In an outstanding feat of cryptanalysis, Rejewski had worked out the wiring of the Wehrmacht Enigma by the end of 1932. Having obtained the details of the machine, and with only six possible wheel orders, the Poles were then able to read messages more or less currently using different methods, including ‘grilles’, a catalogue of patterns for the enciphered indicators (known at Bletchley as ‘box-shapes’); ‘perforated sheets’, invented by Henryk Zygalski; and bomby.

  Dilly also worked on ‘boxing’ the message key indicators (known as key-blocks). Unlike the Poles, Dilly did not use mathematics but a simple mixture of logic and guesswork with the key-block repeats and reciprocal characteristic of the machine. This of course meant a lot of hard slogging to get a sufficient number of ‘alphabets’, as Dilly called the clear and cipher pairs in the columns for the next stage of buttoning-up. Dilly was then concentrating on improving rodding techniques for the un-Steckered machines being used in the Spanish Civil War, which involved solving texts and not indicators and there
is no working record of the ‘comparable theory’ that Welchman said Dilly had produced for developing his work on boxing. It was only when the new German indicators not enciphered on a Grundstellung appeared on Luftwaffe messages in November 1938 that Dilly’s interest in indicators was rekindled and he produced what he called ‘statistics’ for a new line of attack even before the Warsaw meeting.

  The Germans had perceived the folly of having a fixed Grundstellung for the indicators, which once discovered would render the system insecure, and had instructed operators henceforth to choose their own. Although this may seem less secure, as they were then forced to put the Grundstellung in clear in the preamble of the message, they had to set up their chosen indicator on the key setting given in their schedules and encipher it at the beginning of the message, thus removing the cryptographer’s gift of a set of indicators on the same basic setting. However, the operators were still required to repeat the enciphered indicators, for the sake of Morse clarity, and Dilly now looked for remaining repeat characteristics. Like Rejewski, Dilly noticed that sometimes the repeat enciphered letter from positions 1 and 4, 2 and 5 or 3 and 6 of the indicator was the same, for instance AZN BZE, and these would only appear in a limited number of places. So if Z was not affected by Stecker, Dilly said (only six were in use in the early days), it would be ‘possible to find the places on the cycle of keys’.

  Unfortunately, at the same time the Germans had added two new wheels to their machines, but the Poles had a stroke of luck in September 1938 when the Sicherheitsdienst, originally set up by Heinrich Himmler as a secret police force for the Nazi Party, was made into a state intelligence organisation and provided with the armed forces Enigma machine. There was a delay before they changed to the new indicator procedure. When the new wheels were introduced two months later, the Poles were able to break them on the old method, thanks to the Sicherheitsdienst. GC&CS was not able to intercept Sicherheitsdienst messages at that time and so Dilly lost the chance of continuing to work on boxing indicators when the change came and he was getting nowhere with Bertrand’s ‘crib message’.