SIGABA (ECM Mark II)
The American WWII rotor machine the Axis never broke — fifteen rotors, irregular stepping.
Why This Matters
Enigma stepped its rotors like a car odometer: predictable, regular, exploitable. Rowlett's insight in the late 1930s was that the rotor advance itself should be cryptographically driven. SIGABA added two extra rotor banks whose only job was to decide, on each keystroke, which of the five cipher rotors should step — sometimes one, sometimes four, never the same pattern twice in any reasonable interval. That single design choice, combined with rotors that could be inserted in either direction, made SIGABA the most secure operational cipher machine of WWII.
SIGABA was the joint Army/Navy successor to a tangle of incompatible 1930s machines (M-134, the Navy's CSP-889). Friedman and Rowlett at the SIS provided the analytical work; Safford's OP-20-G added the index permuter. Fewer than 10,000 units were ever built — they were heavy, expensive, and required two operators. The machine handled all US strategic Army and Navy traffic in the European and Pacific theatres. No SIGABA traffic is known to have been read by the Axis. The machine remained classified until 1996, decades after its retirement; modern cryptanalytic analyses (Stamp, Chan; Lee 2000s) suggest SIGABA's effective key space is around 295, well beyond what 1940s electromechanics could exhaustively search.
Three banks of five rotors. The cipher bank performs the actual substitution. The control bank is wired to advance pseudo-randomly, taking input from a four-bit ‘clock' that rotates one rotor per keystroke. The control bank's outputs feed the index bank — five rotors whose contacts encode the question ‘which of the five cipher rotors steps next?’ The result: any given keystroke advances 1–4 of the cipher rotors in a pattern that depends on the entire current rotor state. The simulation in this exhibit models that pseudorandom advance and the resulting per-keystroke substitution.
Enigma fell to two structural weaknesses: regular stepping (so the position of each rotor was predictable as a function of message length) and the reflector (which guaranteed no letter encoded to itself, providing the famous ‘I/I’ crib elimination). SIGABA has irregular stepping by construction, no reflector, and no plugboard self-pairing constraint. The Bombe-style menu-and-stepping attack that worked on Enigma simply has nothing to grip. Modern (post-2000) academic analyses confirm that even with full machine knowledge, the search remains intractable for 1940s computing. SIGABA is the rare WWII machine that actually was as good as its designers claimed.
| Concept from SIGABA (ECM Mark II) | Modern Evolution |
|---|---|
| Make the schedule itself secret | Same lesson reappears in modern stream ciphers (key schedule must be cryptographically strong) |
| No reflector → no constraint | Enigma's elegant reciprocity was also its single biggest exploitable property |
| Compartmented design teams | Friedman and Safford collaborated across services — the rare case of US inter-service cryptographic cooperation actually working |
Ciphertext: LBVIICVKBEYCICOAV
Hint: the rotor seed is the machine's own name.
| Era | WWII · 1940 |
| Security | Never broken in service |
| Origin | William F. Friedman & Frank Rowlett (US Army SIS); Laurance Safford (US Navy OP-20-G) |
| Year | Approved 1940; in service 1941–1959 |
| Key Type | 15 rotors arranged as 3 banks of 5 (cipher × 5, control × 5, index × 5) |
| Broken By | Never broken in operational service; declassified 1996 |
| Modern Lesson | Pseudorandom rotor stepping defeats the regularities that broke Enigma |