Gordon Welchman Page 12
All of this did not go down well with Radley and Flowers. Welchman had met them around 22 May 1943 and they had attacked Keen, questioning his competence as an engineer and the actual relay system he had designed. They also threatened to take the matter higher unless their point of view was accepted. Welchman investigated their charges but the people he interviewed were more than satisfied with the work of BTM. Not surprisingly, Keen and his Letchworth colleagues were not happy with this turn of events. Denniston was forced to take the matter up with Radley who pulled back from making a formal charge against Keen or BTM. He did, however, insist that Dollis Hill employed the best relay engineers in the country and that Keen’s relay apparatus would not stand up to continuous use. When Denniston pointed out that Keen’s apparatus was already being used successfully, Radley insisted that Dollis Hill’s valve-sensing equipment (the Cobra attachment) should also undergo the same tests as Keen’s solution. This was a small concession to make to keep the Post Office on board for future developments. Welchman had wanted to launch a formal enquiry but as Radley agreed to withdraw his charges against Keen, Denniston persuaded him to not to proceed with it. Denniston went on to summarize the outcome of his discussions with Radley:
At this point the interview terminated with expressions of goodwill. I am however quite certain that this whole matter will recur at a later date and although I have formed a good impression of Radley I am not convinced that there is not some motive behind some of his statements and actions. He was very ready to deny that he or Flowers had said this or that to Welchman, although Welchman had produced to me written notes of his conversation, written immediately after the interview that I was definitely inclined to doubt Radley’s veracity, in other words, when I made a direct frontal attack he seemed only too ready to pipe down.6
*
In his dealings with those who could provide technical solutions to BP such as Keen’s BTM engineers and Ellingworth’s intercept operators, Welchman had learned that it was a mistake to tell them how to do their jobs. Unfortunately, some cryptographical staff at BP simply approached technologists with demands for solutions without first discussing with them the problem at hand. The large installation of standard IBM and BTM tabulating equipment at BTM under F. V. Freeborn, was a case in point.7 Freeborn was supporting a number of cryptographical activities at BP yet members of these departments tended to try to tell him how to use his machines in support of each problem instead of presenting their problems to him and discussing possible solutions.
BTM had been doing some work in Freeborn’s section for Freddie Jacob of the Military Section at BP and for Wilfred Bodsworth from the Naval Section. In May 1940 an initial installation of Hollerith punch-card equipment was established at BP and when further processing was required, it was decided that this should also be done at BP. The original staff from BTM consisted of Freeborn and two brothers, Ronald and Norman Whelan. They started off in a small hut outside the perimeter fence but then moved into Hut 7 within the grounds of BP and staff numbers quickly increased. They originally had a six-week job but demand was so great that they remained for the rest of the war. Many jobs undertaken were on a once-off basis as their ‘customers’ were not very good at explaining the problem and what was required. To manage the increasing workload, the operation was split into five teams working eight-hour shifts, seven days per week. The operation grew to over 500 staff by the end of the war and was consuming some two million Hollerith punch cards each week.
The daily Hollerith processing carried out in Hut 7 and then in the larger Block C in dealing with Navy Enigma messages was given the highest priority. Messages were split into groups and punched onto Hollerith cards. Naval messages were transmitted in four-letter groups and the Hollerith equipment was used to find a pair of such groups in more than one message, with the separation between them being the same in each message. This could involve searching through around 80,000 cipher text characters in a day. In other words, the Hollerith equipment was being used as an early ‘search engine’ to find cribs which could be used on the bombes.8 Given the difficulty of finding cribs in German naval messages, Welchman had encouraged Alexander to use Freeborn’s facility by carefully explaining the problem to his team. Ronald Whelan recalled one visit that Alexander made to Freeborn years later:
He was a frequent visitor to our building in BP and on one of his calls to discuss processing requirements Freeborn handed him a newspaper cutting which he thought would be of considerable interest. Alexander glanced at the paper for a second or so, apparently with scant attention, before handing it back with the comment that he found it interesting. Freeborn indignantly accused him of not reading the paper, whereupon he solemnly delivered the whole text, word for word, to Freeborn’s astonishment and near collapse.9
In addition to Block C, Freeborn’s team also had equipment located in the village of Drayton Parslow about four miles from BP. The original BTM group had taken up residence in a large house in the village and purpose-built buildings were quickly erected to establish a new machine room. The facility was mainly used for compiling cryptographic material. Whelan recalled that Freeborn drove the BTM group hard and they worked seven days a week from 8.15 a.m. to 6.00 pm. and once the Drayton Parslow facility was in place, they worked most evenings as well. After their first three months at BP, Whelan had one day of leave and after that, his leave amounted to a couple of days every six to eight weeks. They did have some respite on each Christmas Day when they would leave work around 4.30 in the afternoon!
Welchman had established a close working relationship with the Army’s intercept station at Chatham and its commander officer, Ellingworth, in 1939. By 1940, Hut 6 had a twenty-four-hour intercept control section working under John Colman which kept in continual telephone contact with Ellingworth’s duty officers. When the main RAF station at Chicksands became operational, its head, Wing Commander Shepherd, was willing to learn from Ellingworth and Colman and before long Colman’s team was also working well with the duty officers at Chicksands. Welchman’s belief that traffic analysis could be used effectively alongside the work of the cryptanalysts in Hut 6 was beginning to bear fruit. Colleagues had not been convinced, as Milner-Barry recalled:
I had been entirely sceptical of the possibility of ever finding or recognizing standard routine messages of the requisite length, and only the persistence and optimism of Welchman, independent as it seemed to me of any evidence, induced me to make the attempt.
However, Colman’s team were not the only people trying to analyse the German Enigma traffic. In Hut 3, Major D. L. B. Lithgow, who had been in charge of a ‘Y’ station in France and then moved to BP after Dunkirk, had established a group called the Special Liaison Party (SLP). Independent of the BP work, MI 8 (the cover designation for the Radio Security Service) had set up a small section in London in May 1940 called No. 6 Intelligence School (later called the Central Party). Hamish Blair-Cunynghame, Neil Webster and Chris Wills had been brought together to break the call-sign system being used by the Luftwaffe. Another group in the same building, including Edward Crankshaw and Bill Tozer, were studying the German armed forces radio network in liht of decrypted German messages. This would ultimately reveal order of battle information.
The German radio communication system was based around ‘stars’ with a control station at the centre through which went all communications for that ‘star network’. Intercept operators kept a log which was a listing of enemy Morse signals as they arrived along with other information at the beginning of each message. Apart from the radio frequency being used by the sender and sometimes preliminary unencrypted chat, there was other information which was of particular interest. The preamble of a message was the opening sequence of characters consisting of time of origin, urgency prefix if any, indicator (three-letter starting position for decryption), number of letters in the message and the first five letter group including the discriminant. Discriminants were three-letter labels which indicated which Enigma key had been
used. Each key could be recognized by the use of any one of four different combinations of three letters allocated to it (e.g. SFY, SZY, ZKQ, BQI). Call signs were labels consisting of three alphanumeric characters which identified separate units of the German Army and Air Force (e.g. P7J, SF9, 5KQ). They were changed daily using a fixed programme laid out in a call-sign book, one of which would eventually be captured. The log readers summarized from the logs the pattern of communications of a star by means of a diagram with arrows giving the directions of the messages, a list of message preambles and notes of significant chat. These summaries made it possible to recognize stars and some of the stations on them from day to day even without knowing the call sign. This was the same data that Welchman had been studying when he first arrived at BP.
The London group moved to Harpenden in the spring of 1941 and log reading became increasingly important. The initial team consisted of Philip Lewis, Hamish Blair-Cunynghame, Edward Crankshaw and Bill Tozer. When the intercept station at Chicksands became operational in addition to Harpenden and Chatham, problems occurred because different logs were being compiled in different formats. A process of ‘reconciliation’ was attempted in order to compare and combine logs from different intercept stations. This inevitably led to a rivalry developing between the team in Hut 6 and the Harpenden group. As reports of their findings came to Welchman’s attention, he was made aware of the re-encryptions which were being detected in increasing number. These were messages which were encrypted more than once on different keys. The log readers would spot them in their summaries of all communications on each star. However, he was having problems with some of the key people involved in the log-reading process. Lithgow had been promoted in Hut 3, much to Welchman’s displeasure, and he wrote a memo to Travis on 9 May 1941 lobbying for ‘a combination of the four mutually interdependent parties, working in a civilian basis under leaders who are themselves experienced in the actual work of their sections and under your leadership’.10
In another memo to Travis Welchman was particularly critical of Majors Lithgow and Tozer:
Both Major Tozer and Major Lithgow are apt to dress up source information as W/T I [wireless intelligence] and to display it in their own shop window. Major Tozer is even less reliable than Major Lithgow. There is a danger that he may destroy the value of accurate intelligence by the addition of wild ideas of their own … in spite of [Tozer’s] enthusiasm he is a nuisance.11
Welchman was keen to move the Harpenden group to BP but instead they were moved to Beaumanor in the summer of 1941. Blair-Cunynghame was put in command with Lewis as second in command. A weekly wireless intelligence summary called ‘The Beaumanor Weekly’ began to be published and it reported on log-reading findings network by network.
Around June 1942, the Beaumanor group was moved to BP as a result of Welchman’s persistence and the valuable contribution of Blair-Cunynghame. It eventually absorbed Lithgow’s group and became part of Hut 6, using the name of the original London group, the Central Party. One of the Central Party’s main functions was what they informally called ‘fusion’. The ‘Fusion Room’ compared decrypts from Hut 6 with corresponding data extracted by the log readers from the daily radio traffic between enemy stations. This enabled a complete picture of the enemy order of battle to be constructed.
One pleasant by-product of the Central Party’s move to BP was that Welchman could now tell his wife that he was engaged in intelligence work. Welchman had felt that he was the subject of a certain amount of disapproval from the military branch of her family. It seemed to him that his wife’s Aunt Do was the main instigator of hostility towards him. He must have been an easy target as a person who worked in some mysterious office, wore no uniform, and showed no evidence of doing anything useful for the war. Welchman’s experience was matched by many of his BP colleagues. Katharine had volunteered for service with the ATS in 1942 and been posted initially to Aldermaston and then London. She was then sent to the log-reading group at Beaumanor and moved with them to BP.
The log readers of the Central Party had a huge task as enemy wireless traffic increased. They also made use of a growing number of decrypted messages and captured documents. By the beginning of 1943 they were examining some 12,000 pages of logs each day. The material had to be condensed and sorted before use and diagrams continued to be used to show the flow and peculiarities of the traffic. By going through this process day after day, log readers gained an intimate knowledge of groups of wireless stations which were working together as a unit. Even when call signs or frequencies were changed, the log readers could usually recognize the individual stations in a group by monitoring their traffic over a period of time. This huge amount of information was distilled into a weekly report which in 1943 was running to some 100 pages. This drew attention to any unusual wireless activity which in turn might point to interesting enemy movements or the need to increase the coverage of a particular group.
Lewis recalled the work of his team years later in correspondence with Welchman:
If ‘fusion’ meant anything it covered our attempts to identify units and signals procedures on the basis of evidence submitted by the log readers. By then we were fully aware of the patterns of signal procedures, frequencies, participants in nets, etc. that might provide evidence to you of continuity. Though it was not until we moved to BP that, from every source of information, we built up our own large Order of Battle primarily to encourage the log readers to understand the overall picture and the importance of the work they were doing. In essence the Fusion Room was a Common Room where the initiates could meet in seclusion. The term was not used at Harpenden and was of only nominal significance at BP where we had ample space for private discussions. Thus the term ‘Fusion Room Officer’ has no esoteric meaning whatsoever. In my time we never called ourselves thus; such a term would have been risible!12
Another important part of the work of the Central Party was suggesting probable re-encryptions. Re-encryptions of a message from a key already broken were likely to provide an excellent crib. These took various forms including, from one day’s setting into the next day’s setting, from one Enigma key to another or from one Enigma key to a hand cipher.
A steady flow of information was sent to Hut 6 and Hut 3 both as a result of the routine log reading and as the result of investigation suggested by a study of decrypts. For example, a message sent on a certain key might be refused by a station which did not possess that key. The same message might then be sent to the same station on another key. This type of reencryption was spotted immediately by the log readers. They also detected cillies which were often of use to the cryptanalysts in Hut 6. The Central Party provided a similar service to other sections at BP including those working on the Fish13 and German Police systems.
Lewis believed that the work of the Central Party was ancillary and subordinate to that of Hut 6. Its members did not regard themselves as prime producers of intelligence but it was essential that they be fed with as much information as possible as the Enigma keys were broken to enable them to chase up idiosyncrasies of the individual radio networks. They were provided with the whole logs delivered by despatch riders. Lewis remembered that they had an ‘excellent man’ (Sergeant Carteret) who was responsible for collecting all the despatches and making sure the log readers received them as soon as possible. He did not recall receiving any teleprinted material, only the log pads. They were concerned with the radio frequencies, the amount of traffic and the participants in the networks, as well as the preambles with the call signs. Many German operators had their own peculiarities, both in relation to the preamble and the signing-off procedures. Some seemed rather proud of their signing-off flourish and the Central Party was always delighted when they could follow idiosyncratic behaviour.
In addition, Lewis had problems with Lithgow’s team in Hut 3.
Early on we ourselves realised the dangers of what you aptly term ‘tunnel vision’. I mean early on at BP. That was why we had the weekly briefing session I hav
e already mentioned when the whole organization assembled to hear the latest developments in our researches, with call signs becoming real units and their movements linked to the Order of Battle map. So many people were concerned with their own task that they had no idea of the interlinking nature of the whole intelligence exercise. Again various people tried to run their own little empires or elderly regular officers felt they had to have some say, so that overall coordination and control was lacking, which could have been (and maybe was) disastrous or at least time-wasting with regard to priorities. These priorities were of the utmost importance and we needed firm direction from a small control unit in that respect. I assumed that I took my priorities from you. You may have assumed that we were capable of providing more information on our own than in fact we were.