fluid-responsive-typography

name	fluid-responsive-typography
description	Use when browser typography must scale smoothly across viewport or container sizes while preserving zoom, reflow, measure, and hierarchy. Unlike responsive-and-adaptive-layout, this owns type sizing only; face and pairing choices remain upstream.
metadata	{"portable":true,"category":"01-typography-and-fonts","compatible_with":["claude-code","codex"]}

Fluid Responsive Typography

Acknowledgement: Shared by Peter Bamuhigire, techguypeter.com.

Use When

• A type scale must interpolate smoothly between a floor and a ceiling across widths — phone to ultrawide — instead of jumping at hand-tuned breakpoints.
• You are writing the clamp() for headings/display and need the derivation (min, preferred with its vw coefficient, max) rather than copy-pasted magic numbers.
• Line length (measure) drifts too long on wide screens or too cramped on phones and must be capped/controlled across breakpoints.
• A type block lives in a resizable component (card, sidebar, modal) and should size to its container, not the viewport — container-query type.
• An existing page uses font-size: 5vw (or similar) and breaks browser zoom / 200% reflow.

Do Not Use When

• The typeface itself is not yet chosen, or you need the underlying modular scale steps — set those first in 01-typography-and-fonts/font-selection-and-pairing (this skill interpolates between its steps; it does not invent sizes).
• The whole page grid/columns/breakpoint strategy is the question, not just type sizing — use 03-layout-grid-and-composition/responsive-and-adaptive-layout (its sibling pair); set the responsive grid there, size the type here.
• The artifact is a fixed-canvas document or slide (DOCX/PPTX/PDF) with no fluid viewport — fluid clamp()/vw has nothing to interpolate against; use a fixed scale.
• The task is purely choosing weights/pairings or running a slop audit — those are other 01-typography-and-fonts skills.

Required Inputs

Input	Supplied by	Required?	Why
Chosen faces and static scale	Typeface selector	yes	Defines intended hierarchy
Minimum and maximum containers	Layout brief or implementation	yes	Anchors interpolation
Content samples and accessibility targets	Product and QA	yes	Tests measure, zoom, and reflow

• The realistic width range the type must survive — the smallest target width and the largest (e.g. 320 or 360px → 1440px), not an assumed default. The clamp() floor/ceiling anchor here.
• The modular scale steps (px + rem) to make fluid, from font-selection-and-pairing/references/type-scale-recipes.md and doctrine/references/type-scale-and-spacing.md (ratio ≥1.25, real jumps, inverse line-height).
• Which blocks are container-portable (rendered at more than one container width) — these get container-query type, not viewport vw.
• The reading context for measure: long-form body wants ~60–75ch; UI labels are short.

Workflow

1. Anchor the range, then derive — never guess the vw. Pick the floor width (smallest realistic viewport) and ceiling width (largest the design must survive). For each step that should be fluid, set MIN = its size at the floor width and MAX = its size at the ceiling width (both in rem, so zoom works). Then compute the preferred line so it hits MIN at the floor and MAX at the ceiling. With widths in rem (px/16):

   slope (vw coeff) = (MAX − MIN) / (ceilingRem − floorRem) × 100
   intercept (rem)  = MIN − slope/100 × floorRem
   PREFERRED        = {intercept}rem + {slope}vw
   font-size        = clamp(MIN, {intercept}rem + {slope}vw, MAX)
   

   Full derivation, a sign-of-intercept check, and a generated scale in references/fluid-scale-math.md. Generators (utopia.fyi) emit this; you still verify the px endpoints by hand.

2. Always set a rem floor AND a rem ceiling — never a bare vw. A pure-vw size (or a clamp() with a vw bound) defeats browser zoom and 200% reflow — a WCAG 1.4.4 / 1.4.10 failure (doctrine/references/wcag-2.2-criteria.md). The floor keeps phones legible; the ceiling stops ultrawide ballooning. Test by reading at both extremes plus 200% zoom.

3. Do not fluid-scale body copy. Keep --fs-body a fixed rem (typically 1rem/16px). Fluid body text changes the line length as the window moves and harms reading rhythm; let the measure (step 4) adapt instead. Make display/h1/h2/h3 fluid; leave body and captions fixed.

4. Control measure (line length) across breakpoints, independently of font size. Cap the text column so a line never runs past comfortable reading regardless of viewport. Use width: min(100% - 2rem, 66ch) (or inline-size) with margin-inline: auto: ~60–66ch for long-form body, up to ~75ch max, narrower for large display. ch is font-relative, so the cap tracks the chosen face. This is the breakpoint-independent half of "responsive type" that pure clamp() sizing forgets — see doctrine/references/type-scale-and-spacing.md.

5. Set fluid (size-inverse) line-height. Leading must tighten as type grows: small text ≈ ×1.6, large text ×1.3 down to ×1.05 (). Because the size is now fluid, prefer a (it tracks the computed size automatically), or for a block whose size sweeps a wide range, interpolate leading too with so the hero never keeps body leading. Unitless is the default; clamp the leading only when one element spans a big size range.

The Fluid-Type Checklist (run every time)

1. Every fluid step has a rem floor AND a rem ceiling — no bare vw/cqi bound anywhere.
2. The preferred line's vw/cqi coefficient is derived (min/max at named widths), not guessed.
3. Body and captions stay fixed rem — only display/headings are fluid.
4. 200% zoom and 320px reflow both pass — text scales and nothing is clipped (WCAG 1.4.4/1.4.10).
5. Measure is capped (min(…, ~66ch) + margin-inline: auto) independently of font size.
6. Line-height is size-inverse (unitless default; clamp leading only for wide-range elements).
7. Container-portable type uses cqi + container-type, not viewport vw.
8. The modular ratio and weight extremes survive at both extremes — hero÷body still a real jump.
9. The browser root font-size is never overridden (that breaks user zoom and rem math).

If the checklist cannot be satisfied, say so and ask — never ship a vw-only size that breaks zoom.

Anti-Patterns

• Pure-vw sizing: font-size: 5vw (or a clamp() with a vw bound) — illegible on phones, gigantic on ultrawide, and it breaks browser zoom (WCAG 1.4.4 failure).
• Guessed coefficients: a clamp() with a vw number picked by eye instead of derived, so the type misses its intended endpoints.
• Fluid body text: scaling paragraph copy with the viewport — the line length wanders and reading rhythm dies. Body is fixed; the measure adapts.
• No measure cap: full-width body that runs 120ch on a wide monitor — unreadable, and a clear "unconsidered" tell.
• Frozen leading: keeping ×1.6 body line-height on a 64px hero so the hero looks airy and loose.
• Viewport units for component type: vw on a card heading that should respond to its container.
• Overriding html { font-size } in px: kills the user's zoom preference and the rem chain.

Sourcing authority (the asymmetry rule)

Grounds only in human design and web-standards authority — never an AI tool's sizing suggestions, per doctrine/design-doctrine.md §2. Canonical sources:

• The CSS specifications — clamp()/min()/max() math functions, container query units (cqi), and viewport/ch units — admissible as standards, not AI endorsement.
• WCAG 2.2 (W3C) — 1.4.4 Resize Text and 1.4.10 Reflow as the hard floor for any vw/fluid size.
• The Chwezi type-scale doctrine (Gingerich; Paduraru) for ratio, inverse line-height, and measure.

When an AI tool offers font-size: 6vw with no floor/ceiling, treat it as the convergent mistake to avoid — not as a pattern to adopt.

Outputs

Output	Consumer	Evidence / acceptance
Fluid type token table	Design-system implementer	Min, preferred, max, viewport range, and formula
Measure and line-height rules	Component designers	Real content passes target widths
Responsive type evidence	Accessibility and QA	Zoom, 320px reflow, and container tests recorded

• A stated fluid-type decision before any CSS: the floor/ceiling widths, which steps are fluid (and which stay fixed), the derived clamp() for each, the measure cap, and the line-height plan.
• The fluid CSS itself — clamp() scale, capped measure, size-inverse leading, and cqi container-query type where needed — see examples/fluid-type-scale.md.
• A completed fluid-type checklist, within the budgets of doctrine/references/web-performance-budgets-2026.md and the Mission of doctrine/design-doctrine.md §0.

Examples

• Use the worked example in examples/ to verify calculations, zoom, reflow, and measure.

Quality Standards

• Derive interpolation coefficients from declared endpoints; do not guess viewport units.
• Include relative units so browser text zoom remains effective.
• Stop release when text clips, overlaps, loses hierarchy, or fails 200% zoom and 320px reflow.

Decision Rules

Condition	Decision	Wrong-choice failure
Scale follows the viewport	Use viewport-based clamp() with rem bounds	Breakpoint jumps or runaway text
Component width differs from viewport	Use container query units with fallback	Type scales against the wrong space
Dense UI needs stability	Narrow the fluid range	Controls reflow unpredictably
Browser support or rendering is uncertain	Supply bounded static fallbacks	Unsupported clients lose hierarchy

Capability Contract

Read and calculation are required. Editing and browser execution are required only for authorised implementation and verification. Do not change product copy or layout scope merely to make type tests pass.

Degraded Mode

Without browser rendering, produce formulas, static fallback tokens, and a test matrix, marking zoom and reflow unverified. Recover with conservative stepped sizes if container-query support is unavailable.

• examples/fluid-type-scale.md — a real fluid type scale for a 320→1440px range: the stated floor/ceiling, the clamp() derivation shown step by step for each heading (with the vw coefficient computed), a verified px-endpoint table, the capped measure, size-inverse fluid line-height, and a container-query (cqi) heading variant — run against the checklist.

References

• references/fluid-scale-math.md — the clamp() derivation (min/preferred/max + the vw coefficient), the intercept-sign sanity check, container-query (cqi) units, and a generated scale.
• doctrine/references/type-scale-and-spacing.md — the modular ratio, inverse line-height, and measure rules the fluid scale interpolates between (clamp does not invent sizes).
• doctrine/references/wcag-2.2-criteria.md — 1.4.4 resize-text / 1.4.10 reflow: the zoom floor any fluid/vw size must not break.
• 01-typography-and-fonts/font-selection-and-pairing — pick the faces and the modular scale first; see its references/type-scale-recipes.md for the base px/rem steps.
• 03-layout-grid-and-composition/responsive-and-adaptive-layout — the sibling pair: make the whole grid respond there; size the type fluidly here.
• doctrine/references/web-performance-budgets-2026.md — font-display/preload and CLS constraints fluid type must respect.