Bacon Cipher

Selected alphabet mapping

This table changes with the selected version. Each code is the letter’s zero-based position written as five binary symbols, with A for zero and B for one. Unassigned groups are rejected instead of mapped to a guessed letter.

Two presentations, kept separate

This page uses literal A and B letters. It does not hide the groups inside another text or render a carrier with two fonts. In a two-style presentation, each position of carrier text represents A or B through one of two fonts or styles; the receiver needs that style assignment and the grouping. Copying as plain text can lose the style distinction. That carrier method is a different presentation, not an additional alphabet or a confidentiality guarantee.

24 letters versus 26 letters

The older 24-letter alphabet merges I with J and U with V. Here its order is ABCDEFGHIKLMNOPQRSTUWXYZ; J is encoded as I and V as U, and decoding returns I and U. Their original spelling cannot be recovered from those groups alone. The 26-letter variant uses ABCDEFGHIJKLMNOPQRSTUVWXYZ with separate codes for all letters. Selecting one version does not automatically translate groups from the other.

Original encode and decode checks

We created four samples and calculated their A/B groups independently with Python and this page’s JavaScript. Both encoded outputs matched. We decoded each result and checked the expected normalized text, including merged letters in the 24-letter version. All individual letters in both alphabets were also checked. This table was generated from the calculations, not copied from another tool.

To reproduce a row, select its alphabet, paste its original text, convert, then press Reverse result without changing the version. The final sample contains a real newline, displayed as in the table. It becomes a word space. The merged-letter sample restores IIG UOW, not the original J/V spelling.

Input, output and limits

Encode one English-letter text up to 20,000 Unicode characters per run. Lowercase is accepted; decoded text is uppercase. Whitespace is normalized to single word spaces. Digits, punctuation, accented letters and emoji are rejected with a character position, not silently removed.

Decode literal A/B, case insensitive. Spaces and line breaks inside a word are ignored for grouping; every five symbols make one letter. Use / to mark word boundaries. A continuous A/B string is also accepted. Incomplete groups, unknown symbols, empty word segments and groups unassigned in the selected alphabet report errors and produce no partial output.

Encoded input is limited to 140,000 characters and 20,000 five-symbol groups. Output is text or A/B groups with one-click copy. One input per run; no login, registration, uploads, batch mode, saved history or daily counter. Processing stays in your browser.

Can this keep a message secret?

No. Anyone holding the groups and the mapping can restore the letters, subject to the documented merged-letter ambiguity. Use this for puzzles and learning, not protection of secrets. It does not recognize a hidden carrier, identify a cipher or crack a message.

Related classic tools

Use Caesar and ROT13 for shifts, Atbash for alphabet reversal, Vigenere for known-key shifts, Pigpen for geometric symbols, Substitution Cipher for your alphabet mapping, or Cipher Decoder for separate alternatives without identification.

Related representation walkthrough

This existing video demonstrates hexadecimal input and Base64 output, not Bacon carrier decoding.

Loading the player connects to YouTube and shares connection information such as your IP address and playback requests.

Watch the existing encoding walkthrough on YouTube