1 Historical context

1.1 German naval Enigma

The German naval Enigma was a variant of the standard Enigma cipher machine used by the German Navy (Kriegsmarine). It employed a larger set of rotors (eight, from which three were selected daily) and a more complex reflector (the “Umkehrwalze” or “UKW”). Additionally, the naval version used a separate “stecker” plugboard with more connections than the Army or Air Force models. These enhancements made it significantly harder to break than other Enigma variants. Messages were encrypted by selecting a daily key (rotor order, ring settings, and plugboard connections) and a message‑specific indicator.

1.2 Bletchley Park and Hut 8

Bletchley Park was the British codebreaking centre during World War II. Hut 8 was the section dedicated to breaking naval Enigma. Established in early 1940, Hut 8 was led successively by Alan Turing, Hugh Alexander, and others. The team worked in cramped conditions, relying on a combination of mathematical insight, manual labour, and electromechanical devices. Unlike the Air Force and Army Enigma sections (Hut 6), Hut 8 faced the additional challenge of the naval Enigma’s extra security features and the irregular transmission patterns of German naval traffic.

1.3 The role of Alan Turing

Alan Turing, a mathematician and logician, joined Bletchley Park in 1939. He quickly recognised that breaking naval Enigma required not just faster machines but also statistical methods to reduce the search space. Turing developed the theoretical foundation of Banburismus, designing the Bayesian scoring system and the physical procedures used. He also personally took part in Banburismus sessions, training other Hut 8 staff. Although Turing left Hut 8 in 1942, his contributions remained central to the unit’s success.

2 Core principles

2.1 Overlap detection

Banburismus exploited the fact that intercepted messages sometimes contained identical plaintext segments, either because the same message was repeated or because two messages shared a common opening phrase (crib). By aligning two ciphertexts at different offsets, cryptanalysts could detect positions where the underlying plaintext likely overlapped, producing a candidate “align” where both messages were encrypted with the same rotor position at that point.

2.1.1 Identifying cribs and possible repeats

Cribs were known or guessed plaintext fragments, often derived from standard German naval phrases such as “Wettervorhersage” (weather forecast) or “Keine besonderen Vorkommnisse” (nothing special to report). Additionally, when two messages were suspected to convey the same content (e.g., a repeated weather report), their first few letters often matched. Comparing the ciphertext pairs for such repeats gave the cryptanalysts a starting point for Banburismus.

2.1.2 Using the “Banbury sheet” method

The key physical tool was the “Banbury sheet,” a long strip of paper with holes representing the alphabetical positions of letters in the ciphertext. The name derived from the town of Banbury, where the paper was manufactured. By sliding two such sheets (one for each message) past each other, analysts could visually identify positions where the same letter appeared at the same horizontal alignment—indicating a possible plaintext overlap. The alignment was “pinned” with a needle, and the user recorded the offset and the resulting “bigrams” (two‑letter combinations).

2.2 Bayesian scoring

Once overlaps were found, Banburismus used Bayesian probability to evaluate how likely a given alignment was to be correct. This scoring method allowed analysts to focus on the most promising candidates and discard improbable ones.

2.2.1 The Bigram Table

A “Bigram Table” listed the relative frequencies of every possible two‑letter pair (bigram) in German, derived from a sample of plaintext. For example, “EE” is common, while “QZ” is rare. When an alignment produced a pair of ciphertext letters (each from a different message) that corresponded to the same plaintext letter, the actual plaintext bigram formed by the two letters (if the alignment were correct) would follow the German language’s statistical distribution. The Bigram Table supplied the odds for each bigram.

2.2.2 Calculation of odds factors

For each candidate alignment, the cryptanalyst examined the bigrams that would result if the alignment were correct. Using the Bigram Table, they computed a “score” (the product of odds factors for each bigram). A high score indicated that the observed bigrams were more likely under the assumption of correct alignment than under random chance. If the score exceeded a threshold, the alignment was deemed promising; otherwise it was rejected. This Bayesian approach greatly reduced the number of rotor settings that needed to be tried on the bombes.

2.3 Elimination of impossible settings

Banburismus also served as a mechanical filter to rule out impossible or highly improbable rotor orders and ring settings. By combining results from multiple message pairs, cryptanalysts could narrow down the daily key.

2.3.1 Scritch and the “Banburismus menu”

“Scritch” was the term for the manual process of listing all possible rotor and ring‑setting combinations that were consistent with the surviving alignments. The resulting list—the “Banburismus menu”—was then passed to the Bombe section. Scritching was laborious but necessary; it might eliminate over 90% of candidate settings before a single Bombe run.

3 Operational procedure

3.1 Preparation of intercepted messages

Before a Banburismus session could begin, incoming radio intercepts had to be organised.

3.1.1 Decoding and grouping by date

Messages were first transcribed from Morse code and decrypted (if already known) or simply logged as ciphertext. They were then grouped by the day’s date, because the daily Enigma key changed at midnight. Only messages from the same day could be compared, since they used the same rotor order and ring settings.

3.1.2 Sorting by message indicator

Each message had a three‑letter indicator that denoted the message‑specific rotor starting position. Messages with similar indicators were more likely to share the same plaintext opening. Sorting by indicator allowed analysts to quickly pick pairs that might be repeats, a crucial first step.

3.2 Running a Banburismus session

A session typically involved two or three people at a table, armed with ladders of Banbury sheets for the selected messages.

3.2.1 Pinning the Banbury sheets

The Banbury sheets for two messages were aligned side‑by‑side on a flat surface. The analyst slid one sheet relative to the other while a partner watched for matching letters at the same horizontal position. When a match was found, the sheets were pinned to hold that offset.

3.2.2 Scoring alignments

At each pinned offset, the analyst recorded the bigrams formed by the two ciphertext letters at each position. Using the Bigram Table, they calculated the overall score for that alignment. A score above a set threshold (e.g., 100:1) was considered good.

3.2.3 Rejecting low‑probability candidates

Alignments that failed to reach the score threshold were marked for discard. This step could be performed quickly because the scoring table was precomputed. Only high‑scoring alignments were retained for further analysis.

3.3 Integration with Bombe attacks

Banburismus did not break a key completely; it reduced the number of rotor orders and ring settings to a manageable number for the electromechanical Bombes.

3.3.1 When Banburismus alone was insufficient

If only one or two messages were available, or if no plausible overlaps were found, Banburismus could not proceed. In such cases, more messages had to be intercepted, or the Bombes had to try a larger set of possibilities based on other intelligence.

3.3.2 Handover to the Bombes

After a Banburismus session, the list of surviving candidate settings (the menu) was sent to the Bombe hut. The Bombes would then test each candidate against a known‑plaintext crib. If a setting led to a decrypt, the daily key was broken. Banburismus typically reduced the number of required Bombe runs by a factor of 10 to 100, saving precious machine time.

4 Impact and legacy

4.1 Contribution to the Battle of the Atlantic

By speeding the breaking of naval Enigma, Banburismus played a key role in the Allied victory in the Battle of the Atlantic. Convoys could be rerouted away from U‑boat patrol lines, and wolf‑pack tactics were anticipated. During some months, Banburismus allowed Hut 8 to break the naval key before the Bombes could even be allocated, giving an early warning advantage that saved countless lives and tons of shipping.

4.2 Influence on later cryptanalytic methods

4.2.1 Bayesian inference in modern cryptography

Banburismus was one of the earliest practical applications of Bayesian inference in cryptography. Its use of prior probabilities and likelihood ratios presaged modern techniques in cryptanalysis, error‑correcting codes, and machine learning. The same statistical principles are still employed in state‑of‑the‑art attacks on symmetric ciphers.

4.2.2 Comparison with other manual techniques (e.g., Herivel tip)

The “Herivel tip” was another manual method used at Bletchley Park, which exploited operator laziness to deduce Enigma ring settings. While the Herivel tip was simpler and required less data, Banburismus was more systematic and could handle the naval Enigma’s extra complexity. Banburismus also led directly to the development of the “Turing‑Welchman Bombe,” combining statistical and logical methods.

5 See also

6 References

  • Copeland, B. J. (Ed.). (2004). *The Essential Turing*. Oxford University Press.
  • Kahn, D. (1996). *The Codebreakers*. Scribner.
  • Mahon, A. P. (1945). *The History of Hut 8* (Public Record Office, HW 25/1).
  • Turing, A. M. (1940). *Treatise on the Enigma* (National Archives, HW 25/3).
  • Welchman, G. (1982). *The Hut Six Story*. McGraw‑Hill.