The SIGABA cipher machine

SIGABA was the American rotor cipher machine of the Second World War, and its record is unusual: as far as anyone knows publicly, nobody broke it while it was in service. The Enigma stepped its rotors in a fixed order. SIGABA used a second bank of rotors to decide which of the others should move, so its stepping looked random. It was big, heavy and expensive, so it was reserved for the highest levels of command.

How it works

SIGABA had fifteen rotors in three banks of five.

• The cipher rotors scrambled the letters, as in the Enigma, but without a reflector.
• The control rotors had a current passed through them, and it came out on some of their contacts.
• The index rotors turned those outputs into signals that told the cipher rotors which of them should step and which should stay put.

The result was pseudorandom stepping instead of a fixed pattern. An Enigma's right rotor moves on every key press and the others move at predictable moments. In SIGABA, after a letter was typed, anywhere from one to four of the cipher rotors could move, depending on what the control and index rotors did. The control rotors themselves stepped in a simple odometer pattern, so the whole machine was still deterministic. The pattern it handed to the cipher rotors was far from simple, though.

The design is credited chiefly to William Friedman and Frank Rowlett, who developed the core idea of using one set of rotors to advance another. Joseph Wenger, Laurance Safford and Commander Seiler later developed it for the Navy. The Army learned of the system in early 1940, and the Army and Navy adopted it as a joint system in 1941. It went by several names: ECM Mark II (Electric Code Machine), Converter M-134 for the Army, and CSP-888 and CSP-889 for the Navy, with a modified Navy version called CSP-2900. Just over 10,000 machines were made.

Why it held

The irregular stepping took away the weakness the Poles and the British had exploited against the Enigma, which depended on the rotors stepping regularly. With no reflector, a letter could be encrypted to itself, so the "no self-encryption" shortcut for fitting cribs didn't work. Its record is mostly a record of what didn't happen. No successful cryptanalysis of the machine during its service life is publicly known, and neither Germany nor Japan made any real progress against it. SIGABA stayed in use into the 1950s, longer than nearly any wartime machine.

It wasn't perfect, though. A machine without a reflector can't decrypt just by being run again, so it needed a way to run backwards, and SIGABA had one. Operators still had to keep the machine in good condition and follow the procedure. A cipher machine is only as good as its handling, and the Germans' failures with Enigma were mostly failures of procedure.

Why it wasn't everywhere

SIGABA was large, heavy, mechanically complex, fragile and expensive, which made it impractical at the front. Tactical units used other systems, such as the M-209, a smaller, lighter and much cheaper machine, and, in the Pacific, Navajo code talkers. SIGABA served headquarters and ships. The Combined Cipher Machine, in which a British Typex was adapted to work with SIGABA, let the Allies exchange messages.

Timeline and names

The Army called it Converter M-134 and the Navy CSP-888 and CSP-889 (with CSP-2900 for a modified version), and together they called it ECM Mark II. The name SIGABA is the one that survived. Compared with the British Typex, which gave the Enigma's idea a better set of notches, SIGABA changed the way the rotors were driven.

What it shows

• Irregular stepping is the main improvement. An attacker who can predict how the rotors move can model the machine. One who can't has no starting point.
• Cost and security pull against each other. The strongest machine was the one the army could least afford to carry.
• A long record is the best evidence we have. Years of service with no known break counts for a lot, though it proves nothing, since attacks on a cipher in service are usually secret.

Common questions