Colossus, and the operator who handed Bletchley Park a crowbar
On 30 August 1941, a German teleprinter operator on the Berlin–Athens radio link made a mistake the Reich would never know about. He had just sent a long message; the recipient reported the copy was garbled; he retransmitted it — same wheel settings, same text, but this time trimmed of padding and freely abbreviated. He had, without knowing it, handed Bletchley Park a crowbar.
The cipher in question was the Lorenz SZ40, and it was not the Enigma. Enigma was tactical, used by field units; the Lorenz was reserved for Hitler’s High Command — orders flowing between Berlin and army group commanders across occupied Europe (The National Museum of Computing). Twelve interlocking rotors combined plaintext with a pseudo-random keystream, yielding 1.6 million billion possible start positions. German signals officers considered it unbreakable.
Colonel John Tiltman, Bletchley Park’s senior cryptanalyst, spent ten days with the two overlapping intercepts and extracted nearly 4,000 characters of raw key. A Cambridge mathematician named Bill Tutte took it from there. He examined the keystream’s individual bit streams one at a time, tested periodicities, found regularities, and by November 1942 had reconstructed the SZ40’s complete logical structure without ever seeing one (Wikipedia). All twelve wheels, their lengths, their interactions — reverse-engineered from ciphertext alone.
Automating Tutte’s method was the next problem. Max Newman’s team built a machine they called Heath Robinson — a contraption of looped paper tape running at 2,000 characters per second that proved extremely difficult to keep in synchronisation. Tommy Flowers, a Post Office research engineer from Dollis Hill, proposed replacing the mechanical loops with vacuum tubes, generating the keystream electronically inside the machine itself. Bletchley Park declined to fund it. Flowers built it anyway, on his own credit, and his bank account was overdrawn by VE Day (colossus-computer.com).
Colossus Mark I arrived at Bletchley Park on 18 January 1944 and ran its first decryption job on 5 February. The Mark II, with 2,400 valves and five parallel processing streams, went operational on 1 June — four days before D-Day. Among its early outputs were decrypts confirming that German High Command still believed the main Allied landing would come at Pas-de-Calais, not Normandy (colossus-computer.com). The deception held. The landings went ahead.
By May 1945 ten Colossi were running around the clock, decrypting roughly 90 percent of selected Lorenz traffic. Churchill ordered them all destroyed after the German surrender. The General Report on Tunny — the full technical account of what Tutte, Flowers, Newman, and hundreds of others had built — sat classified until June 2000 (Wikipedia).
Tommy Flowers watched ENIAC get declared the world’s first electronic computer in 1946 and could not contradict it. “Publication would have made my name in scientific and engineering circles,” he later wrote, “but I had to endure all the acclaim given to that enterprise without being able to disclose” what he had done first (colossus-computer.com).
Colossus never got the fame it earned in its own time. But it settled a question that would only grow more important: that a programmable electronic machine could outrun any gear or paper loop ever made. The transistor took up where the valve left off.
Sources
- Cryptanalysis of the Lorenz cipher — Wikipedia — Timeline, Tutte’s reconstruction method, Colossus specifications, post-war secrecy.
- Tunny and Heath Robinson — The National Museum of Computing — The 1941 operator error, the Lorenz’s use by German High Command, Heath Robinson’s limitations.
- Colossus — colossus-computer.com — Tommy Flowers’ personal financial cost, first operational dates, Flowers’ post-war statement.