We turned to the tapes of Philby's so-called "confession," which Nicholas Elliott brought back with him from Beirut. For many weeks it was impossible to listen to the tapes, because the sound quality was so poor. In typical MI6 style, they had used a single low-grade microphone in a room with the windows wide open. The traffic noise was deafening. Using the binaural tape enhancer which I had developed, and the services of Evelyn McBarnet and a young transcriber named Anne Orr Ewing, who had the best hearing of all the transcribers, we managed to obtain a transcript which was about 80 percent accurate. Arthur and I listened to the tape one afternoon, following it carefully on the page. There was no doubt in anyone's mind, listening to the tape, that Philby arrived at the safe house well prepared for Elliott's confrontation. Elliott told him there was new evidence, that he was now convinced of his guilt, and Philby, who had denied everything time and again for a decade, swiftly admitted spying since 1934. He never once asked what the new evidence was.
Arthur found it distressing to listen to the tape, he kept screwing up his eyes, and pounded his knees with his fists in frustration as Philby reeled off a string of ludicrous claims. Blunt was in the clear, but Tim Milne, an apparently close friend of Philby's, who had loyally defended him for years, was not. The whole confession, including Philby's signed statement, looked carefully prepared to blend fact and fiction in a way which would mislead us. I thought back to my first meeting with Philby, the boyish charm, the stutter, how I sympathized with him, and the second time I heard that voice, in 1955, as he ducked and weaved around his MI6 interrogators, finessing a victory from a steadily losing hand. And now there was Elliott, trying his manful best to corner a man for whom deception had been a second skin for thirty years. It was no contest. By the end they sounded like two rather tipsy radio announcers, their warm, classical public school accents discussing the greatest treachery of the twentieth century.
"It's all been terribly badly handled," moaned Arthur in despair as the tape flicked through the heads. "We should have sent a team out there, and grilled him while we had the chance."
I agreed with him. Roger and Dick had not taken into account that Philby might defect.
On the face of it, the coincidental Modin journeys, the fact that Philby seemed to be expecting Elliott, and his artful confession all pointed in one direction - the Russians still had access to a source inside British Intelligence who was monitoring the progress of the Philby case. Only a handful of officers had such access, chief among them being Hollis and Mitchell.
I decided to pay a visit to GCHQ to see if there was anything further that could be done with the VENONA program to assist the Mitchell case. The VENONA work was done inside a large wooden hut, number H72, which formed a spur off one of the main avenues in the central GCHQ complex. The work was supervised by a young cryptanalyst named Geoffrey Sudbury, who sat in a small office at the front of the hut. Behind him dozens of linguists sat under harsh lamps, toiling for matches, and hoping to tease out the translations from a thousand anonymous groups of numbers.
Sudbury's office was a joyous menagerie of cryptanalytical bric-a-brac. Huge piles of bound VENONA window indexes piled up in one corner, and tray upon tray of decrypts stood on his desk, ready for his approval before they were circulated up to MI5 and MI6. Sudbury and I had a long talk about how the whole program could be pushed forward. The principal problem was that VENONA, up until then, had been hand matched, and computers were used only for specific pieces of work, such as dragging for a cryptonym. Most of the effort had gone into attacking the KGB and GRU channels directly; the trade traffic channels had been used wherever they formed the back of a match, but otherwise the bulk of it had been left unprocessed. A comprehensive computer-matching program was needed, using the new computers which were becoming available by the early 1960s, in the hope that more matches might be found.
It was a vast undertaking. There were over 150,000 trade traffic messages, and very few were even in "punched" form, suitable for processing through a computer. This alone was a huge task. Each individual group had to be punched up twice by data processors, in order to "verify" that the processed traffic was free from errors. Then the first five groups of each message were computer matched against the whole of the rest of the traffic, involving something like 10 billion calculations for each message.
When I discussed the project with Willis at the Directorate of Science, he was skeptical about the whole thing, so I went to see Sir William Cook at the Atomic Weapons Research Establishment again, with Frank Morgan. I knew that AWRE had the biggest computer facility in the country, bigger at that time than even GCHQ. I explained what I wanted to do. We needed at least three months on his computer to find the matches; once that was done, we could farm them out to NSA and GCHQ for the cryptanalytical work of trying to break the matches out. Cook, as always, was marvelous. I told him of Willis' skepticism, which he brushed aside.
"This is one of the most important contributions AWRE can make," he said, lifting the telephone. He spoke immediately to the AWRE head of Data Processing.
"There's a vital job I want you to start straightaway. I'm sending a chap down with the details. You don't need to know where he works. Please do as he says..."
In two months we had punched up and verified every message, and for the next three months the AWRE computers worked on the VENONA for six hours a night.
At first it looked as if the AWRE computerization program might transform the British VENONA. Early on we got a new match for a message just after the existing week's traffic in mid-September, which we had already broken. The message, when it was partially decrypted, concerned Stanley again. He was to carry no documents which might incriminate him to his next meeting with Krotov. Then, in the midst of a haze of unbroken groups, there was a fleeting reference to a crisis in KGB affairs in Mexico. Krotov was told to refer to Stanley for details, since his "section" dealt with Mexican matters.
At the time of this message Philby was the head of the Iberian section of MI6, which controlled a large swath of Hispanic countries, including Mexico. It was a bitter moment. The categoric proof that Stanley was Philby had come just a matter of months after he defected. Had we broken it out a few years earlier, we could have arrested Philby on one of his regular trips back to London to visit the OBSERVER. This merely intensified fears about the integrity of MI5, since it made the decision in 1954 to close down the VENONA program look deeply suspect. When we checked, we found that the officer who ordered the closedown was the then head of Counterespionage, Graham Mitchell.
Sadly, the Philby fragment was the only real assistance the computerization program gave the British VENONA effort. Matches were made in Mexican KGB traffic and elsewhere in South America which were of enormous interest to the CIA and the RCMP, since Mexico was a principal area where the KGB introduced illegals into North America. But the matches made in British VENONA were almost all trade traffic to trade traffic, rather than trade traffic to the KGB or GRU, which was what we needed. The cryptanalytical effort in Hut H72 went on even more intensely than before, but there was to be no new shortcut.