About Codes & Letters

Codes & Letters takes apart the systems people use to write messages down: ciphers, writing systems, the grammar of the world's languages, public symbols and the shapes of letters. In the name, "codes" means ciphers and the history of breaking them, and "letters" means the letters of alphabets and other writing systems, with their shapes and spacing. The site is not about letter-writing, correspondence or literature.

Who the site is for

The site is for anyone curious about how a message gets encoded: puzzle solvers looking up the Caesar cipher or the Enigma machine, readers of history, students, designers, type and language enthusiasts, and teachers looking for classroom material.

Each article takes one thing apart, such as a single cipher, writing system or part of a letter, and opens with a diagram you can measure rather than a photograph. Articles sit in five sections: Ciphers, Writing Systems, Grammar, Symbols and Letterforms.

Point of view

Reading writing, grammar and symbols as codes is one of the most enjoyable ways to understand them. That is a way of looking, a metaphor rather than a scientific claim. The site does not argue that writing or grammar arose from a need to compress information, and it does not rank any writing system or language by how efficiently it encodes.

Positions the site holds, each open to dispute:

• Irregular English spelling is worth reading as history. It works like a fossil layer; the site calls its cost to readers letter debt and takes no side on spelling reform.
• Codebreaking ran on tables and paperwork, not on a lone flash of inspiration. Biography-style histories tend to leave out the roomful of people who did the counting.
• Good symbols are learned, not intuitive. The icon everyone understands at first sight is mostly a myth.
• Spacing matters more than the letters themselves. A good typeface still fails with uneven gaps, and kerning is the most underrated craft in typography.
• No language is primitive. Grammatical complexity does not rise with how advanced a society is thought to be, and a small language's verb forms can be more intricate than any European language's.

Some subjects stay off the site on purpose: practical advice on modern security or encryption; showing or measuring commercial typefaces, since only fonts under the SIL Open Font License are taken apart; language lessons; puzzles sold on being hard to solve; and disputes over language policy. Requests the site does not take on are listed on Send a Message.

The long-term aim is a shelf of reference pages worth citing, from letter frequency tables for English, German and French to an index of unsolved ciphers updated every year and every writing system on one page. That is the tagline made concrete: every code humans write, taken apart.

How each claim is checked

The method depends on the kind of claim.

Ciphers. Every ciphertext, frequency table and puzzle solution on the site is produced and checked by a program. Getting the plaintext back is not proof on its own: shifting four places forward and four back also restores the text, even if the article says the shift is three. So each worked example is also compared with a published test vector (a known plaintext, ciphertext and key from an independent reference), and each intermediate step the program prints is matched against the rule the article explains.

Writing systems. Details of writing systems outside the Latin alphabet go through three separate checks:

1. Encoding. Code points, character names and Unicode blocks are checked against the Unicode Character Database.
2. Shapes and readings. Letter shapes, transliterations, historical readings and changes over time are checked against named scholarship or images of original objects in museum and library collections.
3. Rendering. This shows only that the characters display. It cannot confirm that the language is right.

A detail that cannot pass the first two checks is left out.

Grammar. Grammatical features and example sentences follow the typological databases WALS Online and Grambank. Features the databases do not record for a language are left out; when the two disagree, both values appear.

Letterforms. Measurements come only from fonts released under the SIL Open Font License, read from the font file by a program. Each measurement names the font, its version, the weight and the coordinate system.

Symbols. Histories of public symbol systems draw first on published explanations from the design bodies or governments behind them. Standards are cited by number, title and edition only.

Sources and numbers. Every article ends with a list of its sources, linked to the original where one is online. Letter counts come from public-domain books with the book and counting rule named, so numbers can be recalculated, and data tables download as CSV files. Every image links to its source and license, and statements that can go out of date carry a year.

When sources disagree. Sources of equal standing that disagree are shown side by side. Character encoding defers to the Unicode Character Database, while readings of ancient writing defer to original objects and named scholarship, not to Unicode. Unsolved and disputed questions are labeled as unproven or disputed.

Limits of a program. A program confirms the counting and the drawing; whether the input is right depends on the sources, which is why every source is listed.

Reusing images and citing pages

Diagrams made for this site are released under CC BY 4.0, each with a credit line ready to copy:

"Diagram title" by Codes & Letters, CC BY 4.0. Source: article URL. License: https://creativecommons.org/licenses/by/4.0/

A link back to the article is welcome. It is a request, not a condition added to the license.

Material from elsewhere, such as scans from Wikimedia Commons, keeps its own license and is credited with its creator, source page and license. A diagram that labels someone else's image follows that image's license, including the ShareAlike terms of a CC BY-SA version.

Drawings that follow a specific protected work, such as a particular symbol set, are not original diagrams of this site and are never released under CC BY; without the rights, the page describes the work in words.

Illustrations. A schematic illustration, where one is used, carries this label: "Illustration: Codes & Letters. Generated schematic image — not a photograph or historical record. Reuse: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/), to the extent any copyright exists in this image." It never stands in for a real object, document or person, and it contains no readable text.

Data. Data from WALS Online and Grambank is used under CC BY 4.0 and credited as each dataset asks, with its version.

Citing a page. Each article ends with a Cite this page block giving the site name, title, first publication date, last substantive revision date and canonical URL; add the date you read it. Articles have no DOI.

Reporting an error

If something here is wrong, write to contact@codesandletters.com, the address at the foot of every page, with the page address, the sentence or number in question and a source if you have one.

Corrections are logged with the date, one sentence on what changed and the reason, at the foot of the page and in the site-wide correction history. Nothing in the log is deleted, and no correction is made silently. A correction to a spec card value updates the downloadable CSV at the same time, and a substantive revision updates the revision date in Cite this page.

What the site measures about visits: Data and Privacy.