Nevertheless, the lead registered in the mind of Guy Liddell, then head of Counterespionage. In his diaries he speculated about the possible identity of Elli. Oddly, I learned of this only after Liddell's old secretary brought the diaries to me, asking that I preserve them, as Hollis had ordered that they be destroyed. Once again I paused for thought. Was this chance, or did Hollis have some other reason for suppressing Liddell's diaries?
In 1965 we managed to break a new message out of the VENONA, which transformed the FLUENCY assessment as to whether Gouzenko's Elli was a true bill. The one week of VENONA traffic which we had broken into began on September 15, 1945, with a message to Krotov discussing, with no sense of panic, the precautions he should take to protect valuable ARGENTURA in the light of problems faced by the "neighbors" in Canada. This was clearly a reference to Gouzenko's defection, which had taken place in Canada the previous week. The "neighbors," we already knew, was the KGB jargon for the GRU, for whom Gouzenko worked. The KGB had no reason to fear that any of its agents in Britain had been compromised by Gouzenko. The GRU knew no KGB secrets and, in any case, Philby was there to monitor any unforeseen developments on a daily basis.
However, by the end of the week's traffic, September 22, the tone of the messages is markedly different. The relaxed tone disappears. Krotov is given elaborate and detailed instructions on how to proceed with his agents. "Brush contact only" is to be employed, and meetings are to be reduced to the absolute minimum, if possible only once a month.
The question we needed to answer was: Why had Moscow Center suddenly become so worried about the implications of Gouzenko's testimony? Gouzenko had actually defected on September 5, two weeks previously, and almost immediately the GRU would be making provisional damage assessments and the requisite precautionary arrangements for any assets they feared Gouzenko might have betrayed. By September 12 details of what Gouzenko was saying to his debriefers in Canada was flowing from Peter Dwyer back to Kim Philby in MI6 headquarters in London. Yet it is not until a week later that the KGB became suddenly worried.
The answer lay in the MI6 files for the relevant period. On September 18-19 a telegram reached Philby's desk which first detailed Gouzenko's description of the spy code-named Elli. This was the first time Philby would have been aware of any reference to the spy in "five of MI." The actual copy of the telegram, when we examined it in the 1960s, was folded into four, with grimy edges, as if it had been placed in an inside pocket, and was initialed off HARP (Philby's initials) two days after he received it. Clearly he had removed the telegram during those two days and shown it to his Russian controller in London. No other telegram in the file dealing with Gouzenko had been treated in this manner. This was obviously the telegram which had caused such worry at the tail-end of the week's VENONA.
I asked GCHQ to conduct a search of all KGB traffic flowing from London to Moscow. We could read none of this traffic. The only matches in the VENONA we had were coming the other way, from Moscow into the KGB in London. Sudbury told me that the only noticeable thing GCHQ could detect in the traffic was a message sent on September 19-20, which they could tell was a message of the highest priority because it overrode all others on the same channel. The significance was obvious - it was sent the day after Philby had received the MI6 telegram containing Gouzenko's description of the spy Elli in "five of MI." Indeed, when GCHQ conducted a group-count analysis of the message, they were able to conclude that it corresponded to the same length as a verbatim copy of the MI6 telegram from Canada which Philby removed from the files.
Once we realized London had sent a high-priority message to Moscow, we searched for the reply. There was only one high-priority message in the line going the other way, from Moscow to London. So far we had never been able to read this particular message properly. It was dated at the very end of the week's traffic, but because it was flashed high priority, it was received in London somewhat earlier than other messages which we read. In late 1965 Sudbury and I made a determined attack on this message; using as collateral the guess that it was a reply to a message containing the information in Philby's telegram. Eventually we succeeded in breaking it out. It read: "Consent has been obtained from the Chiefs to consult with the neighbors about Stanley's material about their affairs in Canada. Stanley's data is correct."
I remember sitting in Sudbury's office puzzling over this translation. It made no sense. I wondered at first if we had made a mistake, but Sudbury checked the translation against the other side of the VENONA, and the trade traffic read off perfectly. There was no mistake. Philby, by the time this message was sent, had been a top-class KGB agent and head of Counterespionage in MI6 for the best part of ten years, yet it appeared as if they doubted his intelligence. Why did it need checking? What was it about Stanley's data which had thrown the KGB into such confusion?
Only one explanation could account for all these oddities. The KGB must have been ignorant of the spy in "five of MI" controlled by the GRU. Thus, when Philby relayed to them news of this spy, and the threat to him by Gouzenko, the KGB had to obtain permission from the "Chiefs," the Politburo, to consult with the "neighbors," the GRU, to ask if they did indeed have such an asset in London. Having received assurance from the GRU that they did have such a spy, the KGB realized that the heat was likely to come on in London, so they sent back the message confirming Stanley's data, and followed it up with urgent orders to increase security.
But who was Elli, and where did he work? He was obviously not Blunt or Philby, since we knew that they were never controlled at any time by the GRU. I asked every Russian defector in the West what the phrase "five of MI" signified. All assured me it meant MI5, not Section V of MI6 or anything else. Whoever Elli was, he must have had access to files on Russians, which placed him indisputably in F Branch, where this material was handled. The senior officer in F Branch was Roger Hollis, the very same suspect defined by Volkov's "Acting Head" allegation.
FLUENCY spent years trying to unravel the riddle that lay in the three connected threads of Volkov's "Acting Head"; Gouzenko's "Elli"; and the VENONA with its eight cryptonyms, each of which came together in that one week in September 1945. Was it Mitchell or Hollis? Both or neither? The resemblances between these strands was uncanny. The "Acting Head" and Elli both pointed to the same two men, but the first allegation was KGB and the second GRU. The VENONA had eight spies; Volkov's list talked of seven in London, two in the Foreign Office, and five in British Intelligence. Maclean had been in Washington for a year, so he could not be one of the Foreign Office spies. Burgess probably was one of these. He was working in the Foreign Office Press Department at the relevant time. The best bet for the other seemed to be Krivitsky's "Eton and Oxford" Foreign Office spy, whom Philby used to decoy MI5 away from Maclean as the net closed on him in 1951.