VISIT to NATIONAL CASH REGISTER CORPORATION
of DAYTON, OHIO
On December 21st I visited the works at Dayton, Ohio, where the Bombes are being made, with Commander Wenger, Lieutenant-Commander Engstrom, Lieutenant-Commander Meader, Lieutenant (jg) Eachus and Major Stevens. The weather held up our train and we arrived six hours late at 2 p.m. so that we did not have quite so long there as we might have had, but probably sufficient.
The plans for the Bombes are on the whole essentially the same as ours, but there are a number of minor differences which should be noted.
(A) As mentioned in my previous report the machine is intended to stop and reverse whenever there is a ‘stop’, and go back to the position of the stop, and there do further twisting. Engstrom and I are still both rather unhappy about this idea. […]
(B) Wheel Changing. You may remember that the American Bombe programme was to produce 336 Bombes ‘one for each wheel order’. I used to smile inwardly at the conception of Bombe hut routine implied by this programme, but thought that no particular purpose would be served by pointing out that we would not really use them in that way. However it now seems that this programme has actually affected the design of the Bombes, for, assuming that the wheels would not be changed, they have designed the Bombes with different sizes of wheels for the different positions.
The eavesdroppers
The technical visit to Dayton was no doubt all very interesting, but everyone knew the Americans were highly accomplished at engineering. While some input from Prof with his years of experience might be helpful, this type of ‘liaison’ was not exactly vital, although having him available to answer questions showed commitment to the secrecy alliance. From the British perspective, the primary purpose of Prof’s visit was something altogether different. It was not about Enigma or even teleprinter codes. It was about something much more direct and simple as a means of communication, but equally dependent on radio and therefore open to eavesdroppers. It was the problem of the telephone.
Very substantial equipment: the US Navy’s four-wheel Bombe (above), and a reconstruction of the X-61753 speech encipherment machine (below), both at the US National Cryptologic Museum. The X-61753 used phonograph records to provide noise to mask the plain speech.
Bell Labs New York building. Alan had to wait for six weeks while a fight between the top British and American generals broke out as to whether he would be allowed in. During the war, a railway line carrying freight trains ran right through the building.
As with the British naval codes in the Battle of the Atlantic, so with phone calls. The scrambled telephone calls of Winston Churchill had been intercepted and unscrambled by the German Post Office – the old technology in use at the start of the war was insecure and unfit for purpose. As early as August 1940 Churchill had been worried about switchboard operators listening in, and he was still concerned in October 1942, writing to Anthony Eden, ‘I do not feel safe with the present free use of the radio telephone either to USA or to Russia.’ Bell Laboratories, the home of the US Army Bombe, was working on a new device to make phone calls secure, and it might be the solution. There was, however, a problem. Alan Turing still wasn’t allowed into Bell Labs.
AMERICAN RESEARCH AND DEVELOPMENT OF TELEPHONIC SCRAMBLING DEVICE AND RESEARCH OF UNSCRAMBLING TELEPHONIC DEVICES
Although there has been no exchange of information between the British and U.S. on this subject, it was known that the U.S. were carrying out certain experiments and Air-Commodore Lywood, Director of Signals, Air Ministry, raised the question with General Olmstead suggesting that a technical expert might be sent out from England as a consultant with a view to assisting British authorities and making available to U.S. authorities a technical expert of proved ability.
On September 24th. Mr. Albert F. Murray, of the Communications Section of the National Defence Research Committee, wrote to the R.A.F. Delegation stating that permission for Dr. A.M. Turing to visit Bell Laboratories in New York had been approved by the Director of Naval Communications and the Office of the Chief Signal Officer. On receipt of this letter, G.C. & C.S. sent Dr. Turing to the U.S.A. but on application being made to Mr. Murray, it was discovered that the matter was complicated and that permission to visit Bell Laboratories could not be obtained for the present. […] I am informed that the construction of the scrambling device is well advanced and is considered of so secret a nature that Dr. Turing cannot be given access to the investigations in Bell Laboratories – this decision having been taken by General Marshall himself.
This was in a memo from Captain E.G. Hastings, RN, to Field Marshal Sir John Dill, the former Chief of the Imperial General Staff, now Chief of the British Joint Staff Mission in Washington. General George C. Marshall was the US Army Chief of Staff. The Bell Labs Question couldn’t get much higher-level attention than this. Dill wrote to Marshall in early December 1942 and over the course of the following five weeks, while Alan pottered around Washington and Dayton, they exchanged a further seven letters. Alan Turing’s bridge-building liaison mission was turning into an international incident. It boiled down to mutual mistrust on the secret-sharing deal between Travis and Wenger. Every time Marshall thought he had obtained clearance for Alan’s visit he was rebuffed. Finally Dill put his foot down. If it was to be a two-way street, Britain would show the Americans everything in Britain, but the Americans would not, on security grounds, be allowed to exploit in America everything they had seen; conversely, America should show the British everything going on in the US, on condition there was no exploitation of it in Britain. It was a sensible compromise, if the Americans agreed to it. If. There was an ultimatum in there; not set out explicitly, but it was there clearly enough between the lines (‘perhaps rather crudely worded but I did it in a hurry to catch you before you looked into the matter tomorrow,’ wrote Sir John disingenuously). If the Americans persisted in denying Turing access to Bell Labs, the deal for them to get British Enigma intelligence and everything else would be annulled. The Americans blinked. Two days later, in Marshall’s absence, his deputy wrote to Dill’s deputy: ‘I have instructed our Assistant Chief of Staff, G-2, to permit Dr. Turing to visit the Bell Laboratory for the purpose of inspecting the scrambling device experiments which are being conducted there.’ On 12 January Alan Turing’s name was added to the Clearance List for something called the ‘X-61753 Project’, with the caveat that he was not cleared for ‘Cryptographic U.S. Army’. Alan got into Bell Labs on 19 January 1943 and, apart from a couple of weeks away in Washington, he was there until 15 March.
He might be in the building but the question of security clearance was still rumbling on. Scrambling of phone calls used, at root, the same principles as encipherment of text: putting it into codified form and adding a secret ‘key’. The cryptographic key which was added by the X-61753 equipment to the speech was provided by phonographic records, which didn’t have nice American music on them but random screeches and noise. This method was, for a time, thought ‘operationally impractical’, and instead the US Army Signal Security Branch wanted to replace the records with an electromechanical converter called M-228. Alan had not been cleared to see M-228 but ‘it does not appear practicable to exclude certain British representatives, principally Dr. Turing, from gaining knowledge of the principles and construction of Converter M-228’. Twenty thousand other personnel were going to know about M-228’s construction and operation, so ‘withholding these features from the British would appear to be somewhat unimportant’. On 6 February it was decided that M-228 converters would be sent over to Bell Labs for them to study. ‘Bell Laboratories will acquaint Dr. Turing with the cryptographic principles involved in the M-228 but will not permit him to examine or conduct any tests on this equipment in their laboratories. Any and all specific questions which Dr. Turing may have on this converter will be answered by representatives of the Signal Security Branch.’
On 5 February 1943 Alan was, at last, allowed to see the US Army Bombe at Bell Labs. Alan’s report
went over to Travis on 11 February under cover of a note apologising for Alan’s typing (perhaps unfairly, since at least compared with Prof’s book, it is of reasonable, if not secretarial, quality) – ‘unfortunately time does not permit re-typing before catching the bag’. Like the bag, the Bombe was fast, ‘giving a 3-wheel running time of 7 minutes. These advantages do not seem to me to outweigh the disadvantage of the enormous amount of equipment involved.’
The Green Hornet
Vast equipment was something Alan was getting used to. For the problem of secure voice encryption, Bell Labs had built something huge. Their device filled an entire room, used 40 racks of equipment, and weighed 50 tons. It needed 13 people to operate it, and it took 15 minutes to set up a phone call. Called Project X-61753 when Alan viewed it, it was later renamed Sigsaly, and then nicknamed the Green Hornet because its enciphered product had a nasal sound a bit like Rimsky-Korsakov’s Flight of the Bumblebee. It worked by taking samples of the spoken messages and measuring their strength. The amplitude of each sample could be expressed as a number, and then enciphered like text by the addition of a key. At the receiving end, you just reversed the process. Simple enough, except that human speech is not a monotone, speech is not writing, and (as Alan had been learning for several years) encipherment was not wholly resistant to attack.
At Bell Labs, Alan met the mathematician and engineer who explained how all this was solved by the new equipment. The man was called Claude Shannon. Shannon had done his master’s thesis on Boolean algebra, establishing that electrical circuits could be used to execute logic problems – the basis for modern computer logic. He had then done a stint with John von Neumann at Princeton. Claude Shannon and Alan Turing are reported to have met frequently in the Bell Labs cafeteria. There was much to talk about. Shannon’s areas of interest overlapped significantly with Alan’s, if Shannon’s output of technical papers during this period is anything to go by:
• Four papers on pulse counting, pulse modulation and pulse shape, all connected with the capture, digitisation and transmission of speech.
• ‘A Mathematical Theory of Cryptography’, of 114 pages in length, finished in 1945.
• Two papers on calculating machinery, one theoretical and the other practical.
Later, in 1949, Claude Shannon wrote a paper on how to program a computer to play chess. Designing a chess-playing algorithm was at the back of Alan Turing’s mind: could a mechanical process be used to carry out the combination of logical and intuitive steps which are needed to play the game? Talking to Shannon about the logical design would have been tremendously good fun, and, from a security perspective, totally safe. But to imagine that this is what they discussed is speculation, and there were no computers at Bell Labs in 1943, nor indeed anywhere else. Work was going on elsewhere in the building to build a Relay Calculator to help with the computations needed to aim anti-aircraft guns successfully, but Alan Turing was not at Bell Labs to look at calculators. He was there to see if the Americans’ solution to secure telephony would actually work.
Claude Shannon had cracked the monotone problem of speech encryption, by seeing that it was a problem of sampling. X-61753 listened to the amplitudes of the spoken signal in ten wavebands across the frequency spectrum, and sampled each of these 50 times a second. Each of these samples would then constitute part of the digitised signal to be encrypted. Then the key would be added. And here the second problem came in. Telephone conversations are not like telegrams, which you can decipher at your leisure. To have a real-time conversation, the key needs to be stripped off the enciphered speech at the receiver’s end at precisely the same moment as it is put on at the speaker’s end. The encoding and decoding machines have to be operating exactly in synchrony. No wonder Bell Labs needed those 40 racks and 50 tons. Thirdly, there was the problem of adding a secure key which wasn’t going to be cracked open like Enigma, and whether this was going to be done using phonograph records or the converter M-228.
On 15 February 1943, while Alan was still somewhere among the 40 racks, General Sir Hastings Ismay – the Chief of the Imperial General Staff who had already had to deal with the Wicked Uncles’ troublesome request for more resources at Bletchley – sent a note to the Prime Minister:
A United States Officer has just arrived in London with instructions to install an apparatus of an entirely new kind for ensuring speech secrecy over the radio-telephone. The apparatus is an American invention. We know little or nothing about it, except that it requires three rooms to house it and six men to operate it. […] The Chiefs of Staff have been considering the position. They think it essential to establish beyond doubt the effectiveness of the equipment. If it is not one hundred per cent secure, it would be extremely dangerous. The only Englishman who has so far been allowed to see it is Dr. Turing of the Government Code and Cypher School. The Chiefs of Staff are not sure that he is sufficiently qualified on all aspects to be able to give a final opinion. […] The fact that the Americans desire to retain complete control of this apparatus, and to prevent our experts from becoming familiar with it, is perhaps strange. Nevertheless, the Chiefs of Staff do not recommend that any objection should be raised by us at this stage. They feel it will be time enough to ask to be let into the secret when the apparatus has been installed, and has proved its value.
The PM, whose notes were shorter than his phone calls, replied, ‘Good. WSC 16.ii.43.’ The Joint Staff Mission sent a Most Secret Cypher Telegram to London a few days later, explaining that there were no British officers in Washington technically qualified to examine the thing, so Alan’s verdict, as follows, would have to do:
Bell System depends on electronic translation of speech into numerical code and any standard reciphering process can be applied. It was originally intended to apply a process equivalent to onetime table which would have provided absolute security. In order to simplify construction U.S. propose to adopt modifications and the proposed process is a machine method which should provide adequate security though definitely inferior to one-time table. If the equipment is to be operated solely by U.S. personnel it will be impossible to prevent them listening in if they so desire.
The full report, addressed to Travis, was sent by air mail later. Travis was put on the spot. Was X-61753 all right, or not?
I am in some difficulty in expressing an opinion on the security of the U.S. project X-61753. Dr. Turing’s report is incomplete as the authorities at the Bell Laboratories would not, for security reasons, permit him to include certain details and drawings. Dr. Turing left New York on 18th March and I should prefer to await his arrival so that I can consult him before giving a definite opinion.
Subject to the caveats on eavesdropping and machine encipherment, and the secrets of M-228, it was a favourable verdict, and 50 tons of equipment were duly delivered to London so that Churchill could badger Roosevelt with flights of eloquence in private. There was not enough space for X-61753 in Whitehall, so the basement of Selfridges department store was taken over for the purpose. They stuck to phonograph records for the encryption method; M-228 was renamed Sigcum and kept for encryption of teleprinter traffic.
The return of Prof
By the time of Alan’s return to Britain in the spring of 1943, Bletchley Park had evolved into a radically different place from four years before. Dilly Knox was dead, and with his passing the sense that codebreaking could be done if you were given enough brainpower, pencils, and squared paper had passed too. Machinery was giving the Enigma settings which enabled Bletchley to produce both decrypts and intelligence. The organisation was now being run by Commander Travis, and it was an intelligence factory, with the various groups (still nostalgically called Hut 6, Hut 8 and so on) now housed in functional, bomb-proof brick buildings and communicating with intercept stations, London and beyond over high-speed teleprinter links.
Towards the middle of 1942, Alan’s role at Bletchley had already begun to change. With the mechanisation of decryption, it was no longer right to have the intellectual
s running the process. Alan’s friend Donald Michie, who joined Bletchley Park in the autumn of 1942, later described the Bletchley legend that grew up around this. It’s a legend, and so it’s probably a myth, but tales about classical heroes are usually worth retelling.
Although [Alan’s] intellectual leadership was absolutely unchallenged, he had no powers of administrative leadership, and in the organisation that worked on the Enigma he was the official head; although Hugh Alexander was a member of the same section, and was superlatively gifted in organisational skills. And this situation just became more and more absurd – and here I am now speaking from hearsay because I was not a member of that section myself, but I had good friends who had been in that and they told me about it – what began to happen was the inevitable, that everybody went to Alexander to know what to do and eventually any order forms would get signed by Alexander although Turing was the titular head. And this was quite characteristic of him, that he wouldn’t know how to deal with such a situation. It wasn’t that he necessarily objected to being displaced from a position in which he wasn’t appropriately placed, but he wouldn’t know how, himself, to bring about the transition which other people might take in their stride – they might go to Alexander and say, ‘Look Hugh, you’re really running this show and I don’t like running things so shall we swop around’ – but it wouldn’t come naturally to Turing. That resolved in the most extraordinary way; he turned up at the entrance to the Park late one morning and anybody who did had to sign their name in a book, and one of the columns in the book was for the name of the head of your section, and he wrote down that day his own name and the time and then in the head of section column he wrote ‘C.H.O’D. Alexander’ and nothing more was said, but finally throughout the whole organisation all the administrative records were changed. And without a word being spoken Alexander from that day on was the titular head of the section.
Prof Page 14