Editorial method & sources

Inclusion and classification

We parse code points, official names, general categories, combining classes, and decompositions from UnicodeData.txt. We recursively resolve canonical decompositions, retaining uppercase and lowercase Latin letters only when the result is an ASCII A to Z or a to z letter followed entirely by marks. Compatibility decompositions are excluded.

Ligatures, stroked letters, and other Latin forms that fail this rule remain separate from the primary dataset. Selected related forms are also copyable on each letter page, grouped as related letterforms, ligatures or digraphs, and phonetic letters. Their names and codes are read from the same pinned Unicode source. Related forms are an editorial selection, not a claim to catalog every Latin character. The Fahrenheit sign is a symbol, not an accented Latin letter, and is excluded. Names describe Unicode identities; we do not infer language usage from names.

Authoritative sources

• Unicode 17.0.0 UnicodeData.txt
• Unicode Character Database documentation (UAX #44)
• Unicode normalization (UAX #15)
• Microsoft CP1252 mapping hosted by Unicode
• Preserved Unicode license notice

Validation and limitations

The generator records source URLs, retrieval dates, and SHA-256 checksums. Automated checks compare regenerated data, reject duplicates, and test character formatting and site routes. CP1252 shortcuts are mapping-verified, not tested across every Windows setup. Browser fonts and keyboard support vary.

The text tool uses the browser’s normalization implementation. Removing marks is not full transliteration. See the tool’s limitations and the typing guide.

Corrections

Compare a questionable character with the pinned source and include its code point when documenting the discrepancy. Our contact page explains current feedback availability.