Why Tracking Changes Between Display and Text Sizes

Set a headline and a caption at the exact same tracking value — zero, say, meaning no adjustment at all — and they won't feel equally spaced. The headline, blown up large, will tend to look loose and disconnected; the caption, shrunk down small, will tend to look tight and cluttered. Nothing about the type changed except its size, but the same numeric tracking value reads completely differently at each end of the scale. The letter-spacing calculator on this site bakes a size-dependent recommendation into its logic specifically because of this, and it's worth seeing the actual numbers it produces.

Why size changes what "even" spacing looks like

Tracking is always a fixed fraction of the em — a proportion of the font size, not an absolute distance. That's exactly the problem: a proportional gap that reads as a comfortable, even amount of air at a small size becomes an increasingly large absolute gap as the font size grows, because the em itself is growing. Blow a 12px paragraph up to 96px for a poster headline without touching the tracking value, and the letters that sat comfortably close together now sit visibly, uncomfortably far apart, purely because the same proportional gap now spans many more actual pixels. The opposite happens shrinking type down: a gap that felt right at reading size can start to make small text look sparse and harder to scan as a coherent word rather than a string of separate letters.

The size bands behind the recommendation

The calculator's guideline moves in five bands by point size: under 10pt, it suggests +40 (in thousandths of an em) of extra tracking; from 10pt up to just under 14pt, +10; from 14pt up to just under 24pt, a neutral 0; from 24pt up to just under 60pt, −10; and 60pt and above, −20. All-caps settings add a flat +50 on top of whatever the size band already suggests, regardless of size, since capital letters lose the varied silhouette lowercase ascenders and descenders provide and read more comfortably with extra room between them at almost any size.

Five sizes, worked out in full

Converting each recommended per-1000-em value into an actual pixel offset (using the standard 96-CSS-pixel-per-inch conversion, where 1pt = 4/3px) makes the trend concrete:

• 9pt (12px), small caption text: recommended +40/1000em → +0.48px of extra letter-spacing.
• 12pt (16px), small body text: recommended +10/1000em → +0.16px.
• 18pt (24px), comfortable reading size: recommended 0/1000em → no adjustment at all — the calculator's neutral middle.
• 40pt (53.33px), a subhead: recommended −10/1000em → −0.53px, a slight tightening.
• 72pt (96px), a poster headline: recommended −20/1000em → −1.92px, a full two pixels pulled in.

The pattern across those five rows is the whole idea in miniature: positive tracking (more air) at the small end, sliding steadily through neutral at a normal reading size, into negative tracking (tighter) at the large end. None of the individual pixel values look dramatic on their own, but applied consistently across a hierarchy they add up to type that reads evenly at every size, instead of type that was only ever tuned for one size and left unadjusted everywhere else.

All-caps changes the story at any size

The +50 all-caps bonus is large enough to flip the sign of the recommendation entirely at some sizes. At 10pt, normal-case text gets a modest +10/1000em; the same size set in all caps jumps to +60/1000em — six times as much air. At 60pt, normal-case text gets a tightening −20/1000em; the same size in all caps gets +30/1000em, a positive value, because the loss of lowercase's varied silhouette outweighs the usual large-size tightening rule. This is worth remembering specifically because it's counter-intuitive: the general "bigger type, tighter tracking" rule of thumb is still broadly true, but flipping a large headline to all caps can genuinely reverse which direction of adjustment is correct, not just soften it.

Applying it to a real headline

Take a 60pt (80px) poster headline. Set in normal case, the calculator suggests −20/1000em, which converts to a CSS letter-spacing of −1.6px — a small, deliberate tightening that keeps the large letters feeling connected rather than floating apart. Flip the identical headline to all capitals, and the recommendation becomes +30/1000em, or +2.4px — now opening the letters up instead, because the all-caps setting has traded away the shape variety lowercase provided and needs the extra room to stay legible as individual words rather than a dense block of uniform rectangles. Two very different CSS values, from the same font size, purely because of the case setting — which is exactly the kind of adjustment that's easy to miss if tracking is set once and never revisited when a design's casing changes.

A guideline, not a law of physics

It's worth repeating what the tool's own documentation is careful to say: this is a general optical-scaling guideline, not a fixed rule every typeface obeys identically. Some typefaces are drawn with generous built-in spacing already and need less manual tracking adjustment than the guideline suggests; others are drawn tight and benefit from more. Treat the size-band numbers as a sensible, evidence-informed starting point — a place to begin adjusting from — and always check the result by eye at the actual size and weight you're using, rather than trusting a formula to replace looking at the type itself.

Seeing tracking alongside the rest of a size's setup

Tracking is only one part of what changes as type moves through a hierarchy — line height and the resulting baseline placement shift with size too, following their own separate logic. The type-setting reference on this site lines up tracking, line-height, and baseline-grid placement for every step of a full type scale in one table, which is a faster way to see how all three move together across a hierarchy than checking each one separately for every size.

This is tracking, not kerning — they solve different problems

Everything above is about a single, uniform adjustment applied across a whole run of text at a given size — that's tracking. It's a separate tool from kerning, which fixes the spacing of one specific letter pair rather than a whole size band, and the two are meant to work together rather than replace each other: set the overall tracking for a heading using the size-band reasoning above first, then zoom in and kern any individual pairs — a capital "A" next to a "V," say — that still look off once the general density is right. A separate guide on this site covers that distinction, and the pairing pattern, in more detail if the difference between the two isn't already familiar.

A three-level hierarchy, tracked consistently

Bring the size-band reasoning together across an actual hierarchy, reusing three of the five worked sizes from earlier: a 9pt (12px) caption, an 18pt (24px) body paragraph, and a 72pt (96px) poster headline, all normal case. The caption gets +40/1000em, or +0.48px — a small nudge that keeps small type from feeling cramped. The body copy, sitting exactly at the calculator's neutral band, gets 0 — no adjustment, left at the type's natural spacing. The headline gets −20/1000em, which at 96px works out to −1.92px, a real, visible tightening that keeps a word this large from feeling like it's coming apart into separate floating letters. None of these three values is large in isolation, but applied consistently, the whole hierarchy reads as evenly spaced at a glance — which is the actual goal, rather than any individual number being "correct" on its own.

From a recommendation to a CSS value

The calculator's px output is the number that goes straight into a stylesheet: for the 72pt headline above, that's letter-spacing: -1.92px;. It's worth setting it as a literal pixel value tied to that specific font size, rather than trying to express it as a generic em value meant to apply at every size — because, as this whole article has argued, the correct tracking genuinely changes with size, so a single em-based letter-spacing value applied uniformly across a hierarchy will be right for at most one size in it and increasingly wrong for the others. If your hierarchy is built with relative units, recompute the tracking value at each size individually rather than letting one value ride along unmodified.

Small caps: a related, easy-to-forget case

Small caps — capital-shaped letterforms set at roughly lowercase height, standing in for a run of lowercase text without shouting the way full-size all-caps does — behave like a milder version of the all-caps case for tracking purposes. They share the same underlying issue: capital letterforms lack the varied ascender-descender silhouette that helps the eye separate lowercase words at a glance, so a run of small caps set at zero tracking often reads as slightly denser and harder to scan than the equivalent lowercase text would at the same size. It doesn't need the full +50/1000em all-caps bonus — small caps are usually set at a smaller effective size than a genuine all-caps headline and already read a little more open relative to their height — but a modest positive nudge, checked by eye rather than assumed, is worth trying before deciding a run of small caps is set correctly.