Design Nomenclature / Type

Subsetting

Stripping a font file to the glyphs and tables a document actually needs. A Latin subset of a pan-European face can shed most of its bytes. Unicode-range splits a family into subsets the browser fetches only when the characters appear, which is how CJK webfonts become deliverable at all.

Three rows, each labelled with a glyph count and a file size. The first names the blocks a full font carries — Latin, Greek, Cyrillic, Vietnamese, symbols — set as words in Georgia rather than shown as glyphs; 3,400 glyphs, 480 KB. The second names Latin alone; 230 glyphs, 34 KB. The third sets the string 'Zoë Ø ₺ —', where everything past basic Latin — the ë, the slashed O, the lira sign and the em dash — is drawn in Arial while the rest stays Georgia, so the fallback is visible mid-string.

In practice

Over-subsetting is the failure mode: strip too far and a name, a currency symbol, or a user-generated comment falls back mid-sentence. Subset to what the site can contain, not to what today's copy contains.

Not to be confused with

How this term connects

CJKFontHintingOpenTypeVariable fontSubsetting
Related Confused with