Typography, Color Spaces & Vector Craft

Choosing Modular Type Scales, Line Height, and Measure for Long-Form Reading

A type scale can look orderly in a specimen and still interrupt reading when headings wrap or the column gets too wide. Start with one body size, derive adjacent heading steps between 1.200 and 1.333, and tune the column to 60 to 72 characters on roomy layouts. Then adjust baseline spacing using the actual typeface and copy.

Size ladder Body · 18 px × 1.250 H3 · 22.50 px × 1.250 H2 · 28.13 px Column measure 560 px at about 8.4 px per character Start at about 67 characters, then count wrapped lines Body baseline spacing Check ascenders and descenders. Read across and return to the left. 28 px baseline to baseline
Read the size ladder from left to right, multiplying each size by 1.250. The horizontal ruler marks the 560 px column width, and the vertical bracket spans the 28 px between successive body baselines.

Set the body size on real copy

Build a specimen from a draft EMFLS Studio Journal could actually run. Use three or four paragraphs, a short heading, a heading long enough to wrap, and a passage with italic words, numbers, and inline code. Keep the copy, typeface, and weight identical when comparing sizes. A specimen made entirely of short words can hide awkward breaks and make a cramped setting seem comfortable.

Start with 18 CSS px for body text and 28 px between baselines. At normal zoom, read a full paragraph at your usual distance from the screen. Look at small counters such as those in e and a, and check whether you can distinguish punctuation from neighboring letters without leaning in. Compare 17, 18, and 19 px with the same line-height ratio. If the column changes length enough to influence your judgment, adjust the width for each specimen so the typical line holds roughly the same amount of text.

Use the production font assets and weight in the specimen. If the face has automatic optical sizing, keep the production setting during the comparison. A size change can also change glyph shapes and widths when an optical-size axis is active. Record the chosen settings before working out the heading sizes.

Derive the ladder from the headings you need

For body size B, adjacent multiplier r, and step n, calculate the size as B × rn. Put the body at step zero, H3 at step one, and H2 at step two. With an 18 px body and a 1.250 multiplier, H3 is 22.50 px and H2 is 28.125 px. The H2 is 1.5625 times the body size because it sits two steps above it. Round for display after calculating each step from the body.

Two heading steps from an 18 px body
Adjacent ratioH3, step 1H2, step 2A reason to try it
1.20021.60 px25.92 pxFrequent headings with long technical names
1.25022.50 px28.13 pxSeveral heading levels in a continuous essay
1.33323.99 px31.98 pxShort section headings that mark distinct transitions

You can solve the same equation backward. If the 18 px body is settled and the strongest acceptable H2 is 30 px, the two-step multiplier is the square root of 30 ÷ 18, or about 1.291. That puts H3 near 23.24 px. Set the longest actual H2 at that size and see where the break falls. If its last word keeps dropping onto a line alone, try a smaller step and preserve the hierarchy with weight and space above the heading.

Set column measure by counting wrapped lines

Measure is the usable width of the text column, inside any padding. For an initial width estimate, measure an unwrapped sample of ordinary prose in the loaded body face. Include internal spaces and punctuation. A 100-character sample that spans 840 CSS px gives an average advance of 8.4 px per character. Multiplying by a target of 66 gives about 554 px. Try 560 px in the layout and count the text that actually wraps.

Use an inline element set to white-space: nowrap for that first measurement, and read its rendered width in browser developer tools. Then remove the measurement styling. A width expressed in ch can make a useful font-relative cap. Its reference is the advance of the zero glyph, while the average of a passage depends on the letters, spaces, and punctuation it contains. W3C CSS Values and Units

Count 12 naturally wrapped lines across three paragraphs, skipping the last line of each paragraph. Include spaces between words and count any visible hyphen at the break. For example, if the median is 74 at 560 px, try 560 × 66 ÷ 74, or about 500 px, then count again. Line breaks move a word at a time, so the calculation only proposes a new width. Check the spread as well as the median. One long identifier can leave a short line even when the surrounding paragraph is well set.

Tune baseline spacing by looking at the ink

For a plain multiline paragraph, 18 px × 1.56 gives a baseline interval of about 28.1 px. A unitless CSS line-height lets descendants that inherit it use the same multiplier with their own font sizes. Give headings their own multiplier. At 28.125 px, 1.18 yields about 33.2 px between baselines on a two-line H2. Give that heading separate block margins so its distance from the preceding and following paragraphs can be adjusted independently.

Compare body intervals of 27, 28, and 29 px at the same measure. Look where a descender such as g or y sits above an accented capital or ascender. A face with larger lowercase forms can make an identical interval feel tighter. If the gap looks crowded in several places or your eye picks up the wrong line on return, test the wider interval. If each line seems isolated and the paragraph loses its shape, compare the shorter interval. Keep paragraph margins fixed during this pass.

Check the wrap in the longest heading separately. Two lines that look unrelated may need a smaller interval. A heading with accents or a fallback glyph may need more room. Inspect inline code and footnote markers in body copy too. Their metrics or vertical alignment can make one line box taller than its neighbors. If the gap jumps in one place, inspect the inline run before changing the body setting across the page.

Allow shorter measures under width constraints

At a 360 px viewport with 20 px padding on each side, the column has 320 px available. At the example's 8.4 px average advance, that gives room for about 38 characters before word wrapping. Recheck the body at its intended size and accept the shorter measure. A 60-character minimum would require more width than the screen offers at that size. Test the narrow layout with an actual long heading, a linked phrase, and a long unbroken identifier. Give exceptional tokens a way to wrap without forcing breaks in ordinary words.

Remeasure at the widths around the switch to the full column. A sidebar or generous page gutters can leave less room than a viewport breakpoint suggests, so log the actual content width. Repeat with the fallback font and at 200% browser zoom. When the page gets taller, headings and paragraphs should expand naturally and the prose should stay within the viewport. Exercise the table wrapper separately and check that readers can reach every column.

Verify and record the settings

Use the same draft and widths for each revision. Record the font family and weight, body size, heading sizes, median and range of line counts, and line-height multiplier. Keep a note about the longest heading and the fallback font. These details make a later font update or layout change possible to compare against the original specimen.

  • Recalculate H3 and H2 from the unrounded ladder and check the rendered sizes in browser developer tools.
  • Count 12 interior prose lines at the full measure and confirm that the typical line falls between 60 and 72 characters.
  • Read a full page at normal zoom and check a two-line heading, accented text, and a paragraph with inline code for uneven gaps or clipped glyphs.
  • Repeat the wrap and overflow checks on a narrow viewport, with the fallback face, and at 200% zoom.
  • Apply the WCAG text-spacing settings together. Set line height to 1.5 times font size, spacing after paragraphs to 2 times font size, letter spacing to 0.12 em, and word spacing to 0.16 em. Check for clipped content or unusable controls. W3C guidance on text spacing