The same key, a new path
Try AAAAA: at I–II–III, rings AAA, windows AAA, reflector B and no plugs, the result is BDZGO.
The engineer’s notebook / 01
Inside Enigma.
A key becomes a circuit. A circuit becomes a letter. Then the whole arrangement moves. Open the lid and try it.
Inspired by The Most Important Decryption Machine Ever Built by Veritasium.
The machine on your desk
Enigma I · 3 rotorsChoose a lesson, or press the round keys below. All calculations happen in your browser.
Live rotor windows · left to right
Ready at AAA. Rotors move before the current flows.
Lamp board · output
Keyboard · press a letter
Follow one electrical journey
Waiting for a keyRead left to right along each row. Gold is the outward route; blue is the return. Each box shows the letter leaving that component. These are logical stages, not the physical layout.
Press a key to illuminate its route.
The reflector sends the signal back through the same rotors, along their inverse wiring.
What are the window, ring and notch doing?
Window: the visible rotor position. The right rotor advances before every letter. A notch lets the adjacent rotor advance too.
Ring: shifts the alphabet ring relative to the wiring core. Changing it changes the electrical substitution at the same window position. For these rotors, the turnover letter visible in the window remains fixed: I at Q, II at E, III at V, IV at J, V at Z.
Double-step: with I–II–III at ADU, three keypresses give ADV → AEW → BFX. The middle wheel advances on two successive presses. This is not a simple odometer.
Inside a rotor: offset the input by (window − ring), follow the wire, then subtract that offset. On the return trip, follow the inverse mapping.
The machine works both ways
Rewind to the identical starting settings and enter the ciphertext. You get the original letters back. “Decode this output” sets this up for you.
A letter never lights itself
The reflector pairs different contacts. That rules out a letter mapping to itself, even after the plugboard and rotor transformations.
A clue is better than a guess
Crib workbenchThis generated training message contains WETTERBERICHT (“weather report”). Slide that guess—a crib—under the intercepted letters. Any matching letter rules that alignment out.
This exercise is separate from your machine above. Known: I–II–III, rings AAA, reflector B, no plugs. Unknown: the three starting windows. The message is synthetic, not a wartime intercept or the video’s unsolved naval message.
Move the crib until there are no self-matches, then search.
How is this different from the Bombe in the video?
This exercise directly tries all 26³ window settings with the other settings known. A wrong alignment can still pass the no-self-match check; a full crib match is a stronger constraint. A short crib can leave several candidates.
The historical Bombe used a menu of relationships between crib and ciphertext letters. Multiple Enigma-equivalent scramblers tested whether the resulting constraints could hold together. It sought contradictions in possible plugboard assignments rather than enumerating every plugboard.
Turing developed the British Bombe design; Gordon Welchman’s diagonal board added the reciprocal constraint “if A is plugged to G, G is plugged to A.” Harold Keen and his engineering team made the machines. Their work built on the achievements of Polish cryptanalysts Marian Rejewski, Jerzy Różycki and Henryk Zygalski. Operators still had to check candidate stops.
This is an accurate three-rotor encryption model and a deliberately bounded crib-search lesson. It does not recreate the Bombe’s electrical menu, diagonal board, full key search or four-rotor naval Enigma.
The Bombe’s idea: close the loop
A constructed three-link menuSuppose a crib gives three relationships: A ↔ Y at letter 1, Y ↔ W at letter 2, and W ↔ A at letter 3. Connect those three scramblers, and the signal must return to where it began.
This tiny synthetic menu was made with I–II–III, rings AAA, reflector B, windows MCK and plugboard AR BS CT DU. The A–Y–W loop is constructed for this lesson, not taken from a wartime intercept.
Why can we leave out the plugboard?
Write the plugboard as P and each rotor-plus-reflector scrambler as S. One Enigma transformation is P → S → P. Connecting three gives P → S₁ → P → P → S₂ → P → P → S₃ → P. Adjacent plugboards undo one another, because each cable swaps the same pair in both directions.
So a closed A → Y → W → A loop means the core alone must take P(A) back to P(A). At MCK, try R: the actual plugboard pairs A with R. Change the guessed letter or the windows and watch the constraint succeed or fail.
The real Bombe combined a larger menu with electrical propagation and, in the Turing–Welchman design, the diagonal board. This widget tests one loop’s fixed points explicitly. It explains one necessary constraint; it does not simulate the complete electrical Bombe.
The rabbit hole began here
Credit to the people who opened the lid.
I watched Veritasium’s The Most Important Decryption Machine Ever Built and wanted to understand it with my hands. This independent learning experiment was inspired by that video. The video, its presentation and its visuals belong to their respective creators; this page is not affiliated with or endorsed by Veritasium.
Jump into the video: 04:16 · Enigma · 26:30 · Exploiting the flaws · 34:27 · The Bombe
The English captions informed the lesson structure. Wiring and stepping were checked against Crypto Museum’s wiring tables and Virtual Enigma’s technical explanation. The account of the Bombe follows The National Museum of Computing. Visit Bletchley Park and TNMOC to explore the original history.
What other machine only begins to make sense once you can take it apart?
spoken by Sree, shaped into pages with ❤️