The aftermath of war

Clive Akass tells how Colossus was reborn after post-war paranoia kept it secret for 30 years and hindered the careers of two of computing’s founding fathers. This is the final part of our series

Tony Sale looks and sounds as if he has stepped out of a John Le Carré spy novel as he stands reminiscing in what was once the hub of Britain’s wartime intelligence.

The fate of nations, and the lives and deaths of countless people, rested on the codebreaking work that was done in these rooms at Bletchley Park, an hour’s train ride out of London.

Today, as it must have been back in 1944, there is the regular thrumming of a paper tape running with surprising speed over multiple rollers feeding data at 5,000 characters a second into a replica of the Colossus Mark 2 proto-computer.

This was used to break the Lorenz encryption used for Nazi high-command messages. By the end of the war 10 of these machines were at work and, if it had not been for Sale, a former MI5 boffin, they might have been largely lost to history. Even today you can find books on computer history that make no mention of them.

This is because the British Government forbade all mention of them, and of the Bletchley codebreaking, after the war. Eight of the machines were broken up and two went to the GCHQ signal intelligence centre at Cheltenham, where they remained in use until 1961. “It took that long for modern computers to catch up,” said Sale.

This secrecy had little effect on the development of computers: their time had come. Those Americans who mattered knew about Colossus, so it probably helped give early US computer projects official credibility and support. Max Newman, who first suggested mechanising Lorenz deciphering at Bletchley, led the team that built the Baby, the world’s first stored-program computer, at Manchester University.

Alan Coombs, who designed Colossus Mark 2, which was completed just in time for D-Day in 1944, went on to build the Post Office’s first computer, Mosaic. “He was an absolutely brilliant electronics engineer. Some of the things he did on the Mark 2 were exceptional,” said Sale. But the secrecy over Colossus hinde red the post-war work of Alan Turing and Tommy Flowers.

Flowers’ inability to cite the precedent of Colossus delayed until 1962 the opening of his great post-war project, an all-electronic phone exchange at Highgate Woods in London. Turing had a similar problem when he went to the National Physical Laboratory to build a machine called the Automatic Computing Engine (ACE).

It used just 850 valves, around a third of the number in Colossus 2 but an enormous number by the standards of the time, and Turing had a hard time trying to persuade his bosses of its feasibility. Finally he gave up fighting the bureaucrats and went to join Max Newman at Manchester. Flowers received an OBE and £1,000 after the war. Turing got not even the grace of a blind eye over his sexuality.

He killed himself in 1954, virtually unknown outside academia, after being convicted of gross indecency. The reasons for his suicide are unfathomable now but professional frustrations after the days of the war surely played a part.

Incidentally Derek Jacobi, the gay actor who played Turing in Hugh Whitemore’s play Breaking the Code, got a knighthood – doubtless deserved for a distinguished career, but a telling contrast nevertheless. And what was the point of all that cussed secrecy?

Sale says the Russians began using Lorenz enciphering machines captured from the Germans; this was at the start of the Cold War and Britain did not want it known that the cipher could be cracked. Actually Sale believes the British were even more devious. A former Bletchley worker, John Cairncross, was exposed in 1951 as a Russian spy who had passed on information during the war.

Sale reckons the British knew all along what he was doing, but allowed him to continue because Russians would not believe intelligence they were given officially by the British – including German plans for the pivotal Battle of Kursk.

Cairncross knew about Enigma, the cipher machine used by German submarines and field units, but like many people at Bletchley he knew nothing about Colossus breaking Lorenz – and neither did the Russians. But after the war, when electronic brains were discussed even in popular newspapers, surely the Russians would realise Lorenz could be cracked?

Sale points out that the Lorenz cipher was very powerful. Some of the methods Bletchley used to crack it were not declassified until five years ago. It was a million times stronger than the Enigma ciphers, and they too were very strong.

“A couple of years back someone found in the archives three unbroken [Enigma] U-boat messages. No context. None of the clues that were used at Bletchley. The only way to break it was by a brute-force attack. A gentleman in Belgium tried it using 2,500 computers linked over the internet. It took three months.”

Still, British reticence about Colossus seems excessive. Compare what happened in the US, where the decimal 18,000-valve Electronic Numerical Integrator and Automatic Computer (Eniac) was completed in 1945, initially to perform artillery calculations. The team who built it held a series of open lectures in Philadelphia where there was a free exchange of information.

Among those attending was Maurice Wilkes, who learned enough to return to Cambridge University and build the first operational computer Edsac. This in turn formed the basis of the groundbreaking Lyons Leo, the first proper business computer.

No-one could talk about Colossus in the UK until the late 1970s – PCW (December, 1978) carried what may have been the first in-depth article about it. Sale knew about it, though. He learned about Bletchley from Peter Wright, the ex-MI5 man who became world famous when the Government tried to ban his memoir Spycatcher.

Sale was 14 at the end of the war, too young for Bletchley but not too young to have built himself a radio and a TV. He was a radar instructor with the RAF at the age of 19 and spent five years as a research assistant at Marconi before joining MI5 as a scientific officer. He describes Wright as “a lovely man”, albeit having a tendency to see communists under every bed.

Wright tells in Spycatcher of how MI5 agents allegedly bugged the Government and plotted to overthrow Prime Minister Harold Wilson. Sale left MI5 in 1963, before all that happened, and set up a succession of IT businesses until 1989, when he became manager of the computer-restoration project at the Science Museum.

While he was there he heard that Bletchley Park, now used as a training centre by BT, might be sold off to be redeveloped as a housing estate. He joined a successful campaign to save it, and turn it into a museum, then launched a project to rebuild Colossus.

He and his wife Margaret put up much of the original funding; they still work as a team, he doing the techie work and she the administration. First Sale had to collect all the information available about Colossus. This amounted to eight wartime photographs, information available from US archives, and fragments of circuit diagrams illicitly copied by engineers – now pinned on the wall behind the rebuilt Colossus.

He also had volunteer help and the memories of surviving members of the original Colossus team. It helped that Sale had learned electronics in much the same era. “Valve electronics was in my blood so I was able to put myself back into the mindset of the designers,” he said.

Tommy Flowers attended the unveiling of the first stage of the rebuild in 1996, just two years before he died. This was based on Colossus Mark 1, but limited to processing just two bits of the two five-bit data streams involved in the decipherment. The fully rebuilt Colossus Mark 2, more versatile and processing all five bits, was unveiled last year.

One of the few parts surviving from the original is an old Post Office transformer used to drive valve heaters, drawing up 1.2amps at 4v. Colossus draws around 5kw, and generates a lot of heat.
It differs in one respect from the wartime version, which was never switched off.

Sale installed a mechanised Variac (variable transformer), as used in post-war valve computers, which slowly increases or decreases the heater voltage when the machine is switched on or off. This minimises the heat stresses that caused valves to blow, earning them the reputation for unreliability that fuelled much of the early scepticism about electronic logic.

Sale points to a valve, one of 40 of wartime manufacture used in the rebuilt Colossus, to show how wrong the doubters were. “Look at that,” he said. “Still going after 64 years.”

Read the first and second parts in our retro news series.