Web Typography

Set type for screens, not print: legible, readable, and hierarchically clear across viewports, zoom, and font preferences. Work top-down: roles, measure, scale, loading, polish. Fit reading task, language coverage, density, and brand.

The decision order

Make these decisions in sequence; each constrains the next.

1. Font choice — A fast system stack is deliberate for operational UI. For brand, marketing, portfolio, or launch surfaces, start from the design-system interview and choose a display/body stance. Plan web-font loading (step 6) up front.
2. Base size & measure — Start around 1rem and 45–75 characters per line; test the actual face, language, density, and viewport.
3. Type scale — Define a small role scale. Use a modular ratio to start, then set optical values; use fluid sizing only where the role benefits.
4. Vertical rhythm — Start body line-height around 1.4–1.6, tighten display roles, and verify clipping/spacing with the chosen face and scripts.
5. Hierarchy & polish — Use weight/size/space contrast, text-wrap: balance on headings, pretty on body where supported, OpenType features, and optical tracking. Numeric tracking recipes are starting points, not universal values.
6. Loading & performance — Choose font-display, preload only a genuinely critical known resource, prefer WOFF2, and metric-match the fallback where reflow matters.
7. Accessibility pass — Check contrast, zoom to 200%, WCAG 1.4.12 text-spacing overrides, enabled zoom, and rendered justified-reading text.

Derive a type stance from use, not a sample pairing

Quick-start faces and scales show constraints, not defaults. Start from reading task, density, language coverage, brand direction, and loading budget. Public narrative, data-dense operations, and multilingual transactions may need different hierarchy and fallbacks. Record intended reading behavior plus computed-size, wrap/measure, glyph-coverage, and loading evidence; variation must change the type system without weakening those checks.

Non-negotiables (the rules that prevent the common failures)

Apply these rules to novel cases, not just the sample.

• Use user-relative units. Prefer rem for document roles and em for components that scale with context. Pixels are not inherently inaccessible, but a pixel-only system detaches from user preferences and scales poorly.
• Use unitless line-height. line-height: 1.5 inherits a ratio and recomputes per element; line-height: 24px inherits a computed pixel value and breaks on children.
• Bound fluid type. Include a relative-unit contribution and test zoom/text resize: clamp(2rem, 1rem + 3vw, 4rem).
• Constrain measure. Lines over ~75 characters lose the return sweep; under ~45 become choppy. max-width: 66ch is a high-leverage readability fix.
• Default sustained reading to ragged text. Justification needs language-aware hyphenation, adequate measure, and rendered review or it creates rivers.
• Control layout shift. Pair font-display with a metric-matched fallback (size-adjust + ascent/descent/line-gap-override) and measure it. This reduces font reflow but cannot support a zero-CLS claim without observation.

Role map and metric budget

Build the role map before naming faces. Separate scanning roles (labels, identifiers, tables: stable width, clear wraps, numeric behavior) from sustained reading (calmer measure, hierarchy, rhythm). Record job, size/measure, language coverage, numeric feature, fallback, and verification.

Test longest translation, accents, punctuation, dense labels, and comparison values (including tabular figures). A family is a hypothesis until loaded face and fallback render specimens without clipping, substitution, or unstable hierarchy.

For each loaded face, record fallback, metric overrides, critical reflow text, and observed layout shift. A system stack is valid when it fits; a custom face needs coverage and metric proof. Missing measurements are gates, never zero-CLS claims.

Quick-start

Use assets/starter.css for the complete scale and metric-matched font example. Keep contextual roles, readable measure, fluid sizing, language coverage, measured loading/CLS, and no duplicate reduced-motion contract here.

Canonical tokens

Reference files

• references/decision-records.md — ADR rationale/rejected alternatives; read for novel judgments and the "principle, not property" boundary.
• references/foundations.md — web anatomy, measure, hierarchy, pairing, alignment; read for font pairing or unexplained layout problems.
• references/fluid-scale.md — clamp() math, modular ratios, Utopia, and cqi; read for scale or container-driven type.
• references/font-loading.md — loading, WOFF2, subsetting, variable fonts, and metric-matched fallbacks; read for custom-font or layout-shift issues.
• references/accessibility.md — WCAG 2.2 contrast, 200% resize (1.4.4), text-spacing (1.4.12), zoom, readability, and reduced motion; read before accessibility/compliance claims.
• references/modern-css.md — text wrapping, OpenType, text-box, punctuation, lh/cap/ ch/ic, line-clamp, and color-mix; read for polish/enhancement checks.

How to deliver

Explain the type decision, edit existing tokens, and keep unsupported features progressive. Report contrast, 200% zoom, text-spacing, loading, and glyph evidence; pair multilingual work with i18n-ready.

Completion evidence

Close with a five-field evidence ledger. Start Reason with the check ID; record an observed result or explicit verification gate.

Contract (generated)

Canonical detail: contract.json.

• Route: A web interface needs a contextual type stance, scale, readable measure, or font-loading plan (+1 in contract.json); avoid: The request is to write or rewrite product copy (+1 in contract.json)
• Exclude: Do not choose a font from examples instead of the brief (+1 in contract.json)
• Stop / handoff: Pause when a requested font lacks required language coverage and no fallback is agreed (+1 in contract.json); receives [design-system-interview, deslop-ui, improve-existing-website, tasteroll] -> sends [responsive-layout, i18n-ready, a11y-pass, spacing-system]
• Output: A contextual type system with tokens, loading plan, and tested reading constraints
• Evidence: table_with_evidence with status, reason, remediation, evidence, provenance.