More significant for Alan, though, than any of this, was that the Battle of the Atlantic had largely been won by the spring of 1943. Alan arrived back safely; he was lucky in his timing. The first three weeks of March 1943 were among the worst ever for sinkings, with 97 merchant ships totalling more than 500,000 gross registered tons going to the bottom. What the Germans described as ‘the greatest convoy battle of all time’ had just concluded. Over a ten-day period, Convoys HX229, HX229A and SC122 had battled with three wolfpacks – 36 U-boats in all – which had sunk 22 ships from these convoys. Nevertheless, this battle was the swansong of the U-boat war. Aided by another capture of secret material from the foundering U-559, the Allies would, from the end of March 1943, always be in the ascendant in the naval code war. Not only had the Enigma blackout ended but, with the closure of the ‘air gap’, convoys were escorted in much greater safety from this point on. Admiral Dönitz abandoned U-boat operations in the Atlantic after May 1943. It was time for Alan Turing to move on.
The Gamekeeper
Hanging beside the stairs at home is a family portrait. It dates from about 1910 or so. It is not by a distinguished painter, it is rather dark and a bit scratched, and it is very large, which is why it is by the stairs. Its subject is Sir James Turing, the ninth baronet. He is holding a shotgun and on the table beside him is a dead pheasant. This is the sort of thing you might expect in a portrait of the only man who could put Alan’s formidable Aunt Jean in her place; and if Aunt Jean was the only woman who could scare Alan’s mother you are beginning to get the picture. We call the picture in question ‘The Gamekeeper’. I am fortunate in being born much too late to experience what Sir James would have thought of such mockery. In their childhood, John and Alan Turing were cleaned up and dressed up and made visits to Sir James in Chichester (and were allowed on outings in Sir James’s red car, complete with its twirly horn and acetylene lamps). John remembered talking to Sir James about some of the family portraits, and Alan would have recognised The Gamekeeper. Family holidays in Scotland and Ireland with Julius and Ethel Turing were fishing holidays; there are fuzzy photos of John and Alan with freshly slaughtered trout. The importance of gamekeeping to keep the stocks secure was well understood by all concerned.
By 1943 Bletchley Park had become an intelligence factory, the workplace for thousands of people. Alan Turing needed a new role.
In 1943 it was time for Alan Turing to become the gamekeeper. Gamekeeping began with the Typex machine. The Germans had not been alone in developing machines to encrypt and decrypt messages. With varying degrees of security, machines similar to Enigma, with rotating coding wheels in an electro-mechanical typewriter-like box, had been developed at around the same time in a number of countries. Britain was one, and Britain had poached. In 1934 the RAF had decided to produce an encryption machine which could print its output. They selected ‘an improved “Enigma” type through the agency of so-called “Type-X” attachments’, and their Typex machines were in operation by 1936. Typex was very similar indeed in its concept to Enigma. To be honest, it was almost certainly a rip-off of German intellectual property. But such niceties as royalty payments to a potential enemy, in relation to a product that was highly secret, were not going to trouble the British.
By 1943 the demand for Typex machines had soared: all the forces in all theatres wanted them, and so did government agencies. Operational communications with the Americans and Canadians in the Atlantic were going to be made more secure by using a modified form of Typex. However, a ‘most secret’ report written in the autumn of 1943 mentioned that out of 6,000 staff then working at Bletchley Park, the resources devoted to the security of Allied communications were ‘the part-time services of only one man plus two or three girls’. Of the armed services, only the Admiralty had scrutinised its own security, and nobody had yet investigated whether, with the improved understanding of cryptanalysis that had been developed in the last ten years, Typex itself was safe. If the British could break open Enigma, Typex itself might equally be vulnerable to attack. Sure, there were technical developments afoot to improve its security: a plugboard (like the one on Enigma), and a pluggable reflector wheel (here anticipating the Germans, who introduced these for Enigma nearer the end of the war). With more services using the machines, the need for confidence in their security was rising too. Back at Bletchley Park, in his new role, Alan was asked to investigate. The Air Ministry wanted to run longer messages through the machine before changing the coding wheels. What was the maximum message length compatible with security? Alan worked out the maximum-length formula, tested it against his experience of codebreaking techniques, and concluded it was all right to double the message length – provided that the pluggable reflector wheel was used.
The problem of encryption which Pat Bayly was working on in America was thus, in all senses, the key. Pat Bayly had also been over to Britain. He was shown round Bletchley Park by Gordon Welchman and then returned the favour when Welchman visited New York. This was the lead-up to a new project for Welchman, Turing and a group of other senior men from Bletchley: ‘the design of new Cypher Machines intended to embody all the lessons learnt during this war’.
D.D. (S) Serial Order No. 117
Machine Co-ordination and Development Section
I am setting up a new Section under Mr. Welchman to deal with all matters arising from the development of new machinery as an aid to Cryptography. […] He will call together a Committee, which should meet at least once a month, and of which I wish the following to be members :-
Mr. Welchman (Chairman)
Professor Vincent
Messrs. Newman, Turing, Freeborn, Wynne Williams [sic] and Major Morgan.
Mr. Welchman will at the same time take over all questions regarding the supply etc. of cyphering and deciphering machines. He will delegate the work of Type X machine supply to Major Carr, who is now responsible for this, only himself co-ordinating this matter with the whole subject of cypher machine supply and selection. […]
E.W. Travis
10th September, 1943
Deceiver of men
In his new job as guardian of secrets, and designer of machines, Alan gravitated towards another secret government establishment, located less than five miles from Bletchley, and hiding under another innocuous cover name. This was the ‘Radio Security Service’ at Hanslope Park. Here Alan was going to try to design a secure encrypted telephone which weighed less than 50 tons.
Alan started work on his speech system in May 1943. Speech encipherment was not simply for the likes of Churchill and Roosevelt, for whose conferences 50-ton monsters might be acceptable. Alan’s new machine was to be used on the go, for ‘tank-to-tank and plane-to-plane work’. As well as portability it would have three units: an audio sampler, a key unit to produce the cipher (which, by flipping a switch, could also serve to decipher), and a combiner to add the key to the signal. Alan, being Alan, rolled up his sleeves and got stuck straight in with a soldering iron.
Hanslope Park at around the time of World War Two.
Donald Bayley, newly commissioned into the Royal Electrical and Mechanical Engineers, was ordered to Hanslope Park. He was rather surprised at his first encounter with Prof:
He was a bit slapdash; I was very well-organised. I came into the hut, they said just see what’s what to start with. This chap had his shirt hanging out. There were resistors and capacitors, as fast as he’d soldered one on another would fall off. It was a spider’s nest of stuff – a complete mess. We made up a ‘breadboard’ sheet of plywood, you soldered between strips of metal, to make up the board. He hadn’t worked on it like that at all, soldered anyhow, and hoped they’d hold together. He was annoyed I mentioned his shirt hanging out. He took it for granted. He said I shouldn’t have mentioned it.
The importance of being able to make it yourself had been imbued from Hazelhurst days. Alan’s brother John kept a carpentry shop in a garage or a garden shed until the end of his life, constantly making cupboards an
d bookshelves for leisure; from time to time he even indulged in bricklaying, and built a playhouse for his daughters in Guildford. As for Alan:
He was intrigued by devices of every kind whether they were abstract devices or concrete, and his friends felt it would be better if he kept to the abstract devices but that didn’t deter him. He wasn’t at all gifted at the concrete embodiment of theoretical ideas but he loved to do it.
Hanslope Park, unlike Bletchley, was run as a military establishment, so his shabby clothes must have stood out markedly. To develop his project Alan really needed help. On 16 June 1943 Alan reported on his resourcing needs for the development of ‘Delilah’. He needed more space and a second oscilloscope; he asked that ‘provision will be made to enable Lt. Bayley to be assigned to me officially’. Alan’s other assistant, also commissioned, was a young mathematician whose degree at King’s College, Cambridge, had been interrupted by the war. This was Robin Gandy, who was to become Alan’s closest friend.
Robin Gandy (centre), in his King’s College matriculation photograph from 1938.
When I arrived he had already started building this speech encoder and he liked doing things himself – very much – he regarded it as very important that one ought to be able to do things oneself. He was fairly clumsy, and he had to me a very curious habit. It’s called a breadboard – actually his was called the bird’s nest, it was a great mass of wires, condensers and resistors and so on – and he was always swapping them round to try and improve something or other. And so he would take the soldering iron. He always soldered with a high-tension left arm, I can’t think why. My first memory of coming into the shop is there was this rather scruffy looking civilian in an old tweed jacket bent over the bird’s nest and every now and again he’d go ‘OW! OW!’
Sometimes Prof could take people by surprise in other ways:
It was at Bletchley [wrote Alan’s brother, John] that for some reason Alan was attached to an army unit, where he was treated with that brand of tolerant amusement which the armed forces reserve for boffins. This did not suit Alan at all so when he heard that there was to be a cross-country race he asked modestly if he could join in. The request was granted; all in the mess looked forward eagerly to the Prof trailing in well behind the rest. Of course Alan came in three minutes before anyone else.
Delilah. Combining, power, and key units. The key unit is also shown with its top open to reveal the scrambler wheels at the rear and valves (one removed) at the front.
It is said that Robin Gandy named the speech device Delilah, because Delilah was a deceiver of men. With his Hazelhurst grounding in Scripture (which meant the Old Testament, so as to avoid troublesome theology), Alan probably knew the story of Samson and Delilah better than Robin, but he didn’t complain that Delilah was a spy working for the other side. In any case, like Delilah, the machine ‘pressed him daily with her words, and urged him, so that his soul was vexed unto death’1, and that probably summed it up for all of them. So Delilah she became.
Delilah’s biggest problem was her ability to deceive, or to put it technically, her key unit.
The system requires a random voltage (k(t)) to be produced simultaneously at each end of the transmission path. This problem presents formidable difficulties. Of the possible solutions, the two that received most serious consideration are (a) recording random noise on discs or tape and using those recordings simultaneously at the ends of the transmission path [this is what the Americans had done with X-61753] and (b) generating identical voltages at each end. The first has the advantage that the keys are truly random and identical but has the disadvantage that the mechanical difficulties of starting and maintaining the keys in synchronism are large and, furthermore, the number of discs or reels of tape required becomes prohibitive. The second scheme is the one that has been tried in practice.
The key unit randomised the signal by taking six lumps of noise through networks which generated 26 voltage outputs. Twenty-six is a rather handy number if you want to add further disguise by using coding wheels from a cipher machine – and any cipher machine will do. So the key unit could plug into ciphering wheels, then there were further recombinations, a plugboard, and a modulator to normalise the output signal to sound more like white noise. It was very complicated, but it was very compact and it would be possible to use Delilah with any text-encipherment machine.
Gordon Welchman was assigned to work together with Alan specifically on the development of cipher machinery to handle teletype traffic; versions of Pat Bayly’s teleprinter cipher machine were going to be produced at Hanslope Park. So, after the autumn of 1943 Welchman too was spending time at Hanslope Park. Balancing the demands of Bletchley and Hanslope might have implied that Alan should remain in his lodgings at the Crown Inn, Shenley Brook End, but the congeniality of working at Hanslope won out. On D-Day, Alan Turing submitted a progress report on Delilah, in which he concluded that the key settings should be changed after every eight minutes of chat; shortly afterwards, Alan moved into Hanslope Park itself, first living in the old house and then later in a cottage which he shared with Robin Gandy. Discretion was needed, possibly even concealment; both Robin and Alan knew personally why it was necessary to be a deceiver of men.
As an epilogue to the story of Delilah, there was some idea that the Post Office had itself invented a speech scrambler. In January 1945 Prof was asked to investigate, although the Post Office were told they could not ask him about the Americans’ speech secrecy machine. So Prof went to Dollis Hill again. His report was not encouraging: ‘with the ciphering problem they have made practically no progress’. The problem was, none of the British experts who could help, could actually help, as they were all off-side, what with the promises made to the Americans that knowledge of the X-61753’s secrets would not be exploited in Britain. Alan took the opportunity of letting the right hand of British bureaucracy (the Cypher Policy Board) know what the left hand was already doing: building a secure speech device at Hanslope Park. Despite this, the Cypher Policy Board was still arguing about developing speech devices several years later. Delilah was destined to remain buried between the pages of the Old Testament.
Once Delilah had turned into an engineering project, it wasn’t going to occupy Alan Turing full-time. At Bletchley Park he was still an important figure. The subject since mid-1942 had been the design of more machines, because machines were being used to process Fish.
The processing of Fish
Early in 1940 some British policemen listening for German spies had intercepted a rather strange type of transmission emanating from across the Channel. Like Engima messages, they were enciphered, and so they found their way to Bletchley. Like Enigma, a machine was converting plain-text into cipher-text. That is about where the similarities with Enigma stopped. The messages were sent not in Morse but in the Baudot-Murray teleprinter code – the very code which Pat Bayly’s machine was to keep secure. Bletchley Park called this type of traffic ‘Fish’. From small beginnings, Fish was going to have a profound impact on the shape of codebreaking at Bletchley Park, the strategic intelligence available to Allied Supreme Command, and the post-war career of Alan Turing.
The British had never seen the machine which was involved in creating Fish and had no idea how it worked. Everything to uncover the secrets of Fish would have to be done from first principles, and by hand. Fortunately for Britain, the Germans made mistakes, and the classic one of retransmitting a garbled message in the same key in mid-1941 gave Bletchley Park their first break. By mind-boggling feats of analysis, John Tiltman and another Bletchley code-master called Bill Tutte reverse-engineered the structure of the German teleprinter cipher machine. The Germans called it the Schlüsselzusatz, or cipher attachment, made by the Lorenz company in Berlin. The Lorenz machine worked – like Pat Bayly’s machine – by overlaying a ‘key’ of random letters which had to be reversed out to reveal the original. Unlike Bayly’s machine, the Lorenz key was supplied by ten wheels, each of which had a different number of pins wh
ich could be set to ‘on’ or ‘off’ to make electrical pulses whenever contact was made with a pin in the ‘on’ position. There were two groups of wheels, which were named ‘chi’ and ‘psi’ because the Bletchley mathematicians couldn’t help using Greek. Chi wheels always rotated between letters, but psi wheels were more erratic, pausing for one or more letters before moving on.
Worse than Enigma. The Lorenz machine was connected to a teleprinter and used to encrypt the highest-level German signals.
As with Enigma, knowing the machine is not the same as knowing the key. Alan was transferred from his role in Hut 8 in mid-1942 to help out with research on this new and all-but-impenetrable system. His contribution was a technique for revealing pin-patterns, which, as tradition demanded, was named ‘Turingismus’. Alan Turing’s method began with the idea that you could track the changes introduced by the chi wheels by adding the five-bit code for each letter in the message to the five-bit code for its neighbour. Using the mathematics of binary arithmetic, this would eliminate the effect of the psi wheels (assuming they hadn’t moved), because the contribution of the psi wheels to the cipher key was being added a second time – just as it is in decrypting a message. With this insight, you could start looking for the effect of the chi wheels.
Prof Page 15