Also invented by Felix Delastelle around 1902, the Trifid cipher takes his Bifid idea into three dimensions: 27 alphabet cells (A–Z plus one padding symbol) are arranged into a 3×3×3 cube. Every plaintext letter becomes a triple of trits (layer, row, column, each 1–3); the trits are read down within a block of period length, concatenated, and re-grouped into new triples to produce the ciphertext. Because each letter is split three ways, Trifid diffuses plaintext structure more aggressively than Bifid.
Each cell has coordinates (layer, row, column) from 1 to 3. Change the keyword above to rearrange the cube.
For the current plaintext, split into blocks of the chosen period. Each column of trits (layer / row / column) is written down, then the rows are concatenated left-to-right and re-grouped three at a time to produce ciphertext letters.
Alphabet. 26 English letters need 27 cells, so a pad character (traditionally + or .) fills the last slot. Change it above — anything not in the cube is dropped from the plaintext.
Period. Small periods leak too much structure; larger periods diffuse better but leave a partial trailing block. Delastelle originally recommended 5–7 for short messages.
Symmetry. Encryption and decryption are inverse operations of the same algorithm — the only difference is how the trits are re-grouped inside each block.