Using the Bacon tool
Type on the left and each letter comes back as a group of five As and Bs. Swap the boxes to decode. Anything that isn't an A or a B is ignored on the way back, so you can paste in groups separated however you like.
Alphabet picks between the 24 letters Bacon actually wrote, where I covers J and U covers V, and a modern 26-letter table that gives every letter its own code. The classic setting is the one you want for historical text.
Counting to 32 with two letters
Five slots, two choices each, gives 32 combinations. That's more than enough for 24 letters. A is AAAAA, B is AAAAB, and the pattern counts upward in binary with A standing for 0 and B for 1.
Bacon's own alphabet ran to 24 because Elizabethan English treated I and J as one letter, and U and V as another. The 26-letter version is a later tidy-up. They produce different ciphertext for anything past H, so a message encoded with one won't decode cleanly with the other.
Bacon's 24-letter alphabet
The cipher was never meant to be seen
Francis Bacon published it in 1605 in The Advancement of Learning and set it out properly in De Augmentis Scientiarum in 1623. He called it a biliteral cipher, and the As and Bs were not supposed to appear anywhere.
The point was to hide them. Write a harmless cover message in two slightly different typefaces, one standing for A and one for B, and the five-letter groups disappear into the shapes of the letters. Bacon's phrase for the goal was writing "omnia per omnia", anything by means of anything. It's steganography rather than encryption, and the secret is that a secret exists at all.
That idea attracted cranks. Ignatius Donnelly in 1888 and Elizabeth Wells Gallup in 1899 both claimed to have found Bacon's cipher hidden in the typography of the Shakespeare First Folio, proving Bacon wrote the plays. William and Elizebeth Friedman, who between them founded American cryptanalysis, spent years on it and published the demolition in 1957. Sixteenth-century printers used whatever type was in the case, and a determined reader can find any message in noise.
Related conversions
Binary does the same counting with 0 and 1 and eight slots instead of five, and the resemblance isn't a coincidence. Morse also builds letters from two symbols, but its codes vary in length so common letters stay short.
For a scheme that numbers the alphabet rather than encoding it, try A1Z26.
Questions
What is Bacon's cipher?
A code from 1605 that represents each letter as a group of five characters drawn from a two-symbol alphabet, usually written as A and B. It's the first published binary encoding of the alphabet.
Why do I and J share a code?
Bacon was writing for an alphabet of 24 letters. Elizabethan English treated I and J as the same letter, and U and V as the same letter, so his table has 24 entries.
Is Bacon's cipher the same as binary?
The mechanism is the same. Five binary digits per letter, counting upward. Bacon wrote A and B where we'd write 0 and 1, and he published it more than 300 years before computing.
Did Francis Bacon write Shakespeare?
No serious evidence supports it. The claims rested on finding his cipher in the First Folio's typefaces, and William and Elizebeth Friedman showed in 1957 that the readings were not reproducible.