Is there a minimum size for small type?

There is a practical floor rather than a rule. Around 11 pixels is where continuous reading starts to suffer, so anything read like a sentence should stay at or above it. Reference matter — a footnote, a legal line, a figure note consulted rather than read — can sit below, which is why the check column separates the two cases.

Why is the caption larger than the footnote?

Because captions are usually read and footnotes are usually consulted. A caption sits under a figure and has to work as a sentence, while a note is looked up after the fact. The defaults reflect that: the caption keeps a slightly larger size and a more generous leading than the footnote at the same body size.

Why does the superscript need a special rule?

Because a raised character set at the normal line height enlarges the line box and opens a gap in the paragraph. Zeroing its line height takes it out of the vertical calculation while vertical alignment keeps the mark raised, so the reference digit sits where it should without disturbing the lines around it.

How is the legal size determined?

It is derived as nine tenths of the footnote size with a leading of 1.35, which keeps the two in a fixed relationship. That means a design that wants a bigger step between notes and fine print edits the exported declaration rather than a control, since the derivation is deliberate rather than adjustable.

Do the ratios work for every typeface?

No. A face with a small x-height needs more nominal size to reach the same apparent size, so the same ratio can look smaller in one family than another. The ratios are a starting structure; the pixel column and a rendered sample are what confirm whether the small roles actually read.

Caption & Footnote Scale

Scale inputs

Body sizepx

Caption× body

Footnote× body

Superscript× body

Sample text

Role	rem	px	line-height	Check	Used for
Body	1rem	16px	1.6	safe	paragraph copy
Caption	0.8125rem	13px	1.45	safe	figure captions, table notes
Footnote	0.75rem	12px	1.4	safe	footnotes, endnotes
Legal / fine print	0.675rem	10.8px	1.35	below continuous-reading floor	terms, disclaimers
Superscript	0.65rem	10.4px	1	below continuous-reading floor	reference marks

Rendered roles

BodySmall type still has to be readable — paragraph copy

CaptionSmall type still has to be readable — figure captions, table notes

FootnoteSmall type still has to be readable — footnotes, endnotes

Legal / fine printSmall type still has to be readable — terms, disclaimers

SuperscriptSmall type still has to be readable — reference marks¹

Small reference matter can go below 11px, but continuous reading text cannot — a caption that has to be read like a sentence should stay at or above the floor.

css output

:root {
  --text-caption: 0.8125rem;   /* 13px */
  --text-footnote: 0.75rem; /* 12px */
  --text-sup: 0.65em;
}

figcaption {
  font-size: var(--text-caption);
  line-height: 1.45;
}

sup, .footnote-mark {
  font-size: var(--text-sup);
  line-height: 0;
  vertical-align: super;
}

You Might Also Need

Type Scale Generator

Generate a modular type scale

Footnote & Endnote Formatter

Format notes consistently

Line Height Optimizer

Optimize leading for readability

CSS Typography Token Generator

Export CSS type tokens

The caption and footnote scale works the small end of a type system: captions, footnotes, legal lines and superscripts, all derived from the body size. The check column flags the sizes that fall under the practical floor for continuous reading.

What is Caption & Footnote Scale?

The caption and footnote scale derives five small roles from the body size and reports whether each one is above the point where small type stops being comfortable to read.

The roles and how each is built:

Body is the anchor at one rem with a leading of 1.6, so every other role is a fraction of it.
Caption defaults to 0.8125 of the body size at a leading of 1.45, for figure captions and table notes.
Footnote defaults to 0.75 at a leading of 1.4, for notes at the foot of a page or the end of a chapter.
Legal and fine print is derived as nine tenths of the footnote size at a leading of 1.35, which is where terms and disclaimers live.
Superscript defaults to 0.65 of the body size, with a leading of one so it cannot push the line box apart.

The check column is the useful part, because small type is where a type system quietly breaks. Sizes below about 11 pixels are flagged as below the floor for continuous reading, and the distinction the tool draws is between reading and consulting. A caption that has to be read like a sentence should stay at or above the floor; a footnote, a reference mark or a legal line is consulted rather than read, and can sit below it without harming anyone. That is why a 16px body produces a comfortable 13px caption and a 12px footnote that is still acceptable in reference matter, while the same ratios applied to a 13px body fall under the floor and need the body size raised rather than the ratios adjusted.

Superscripts get their own treatment for a mechanical reason. A raised digit set at the normal line height pushes its line apart and leaves a visible gap in a paragraph of running text, so the exported rule sets the superscript size as a custom property, zeroes its line height and raises it with vertical alignment. That keeps the reference mark on its line without moving the lines around it, which is the same technique a real footnote mark needs.

The output is a small token block: a caption size, a footnote size and a superscript ratio as custom properties, plus figure caption and superscript rules that use them. Three limits are worth knowing before the block is adopted.

The ratios are not face-specific, and a face with a small x-height needs a larger caption than a face with a large one at the same nominal size, which is the usual reason a legibility floor verdict feels wrong in the browser.
The eleven-pixel threshold is a practical floor rather than a standard, so formal accessibility requirements still come first for anything that has to be read.
The legal size is derived from the footnote rather than adjustable on its own, so a design that wants more separation edits the declaration afterwards.
The superscript ratio sizes the mark but not its raised position, so an unusually high or low mark is a styling decision made where the rule is applied.
Nothing sets the width of a caption, so a long figure note still has to be wrapped or centred deliberately.

Read the pixel column rather than the ratios: it is the number the reader meets, and it is the one that has to clear the floor.

How to use Caption & Footnote Scale

Set the body size, then read the pixel column rather than the ratio column, since the pixel values are what the reader meets.
Adjust the caption and footnote ratios and watch the check column, raising the body size if the roles fall under the floor.
Set the superscript ratio and confirm the line-height rule is copied with it, or the raised marks will open up the paragraph.
Enter a sample line and read the rendered roles against each other, checking that the caption does not look like body copy.

When to use Caption & Footnote Scale vs related tools

Reach for it when a document has captions, notes or legal lines that have drifted out of proportion with the body text, when small type has to be justified against a legibility floor, or when superscripts are opening up the leading in running text. The reading-versus-consulting distinction is what makes the floor usable. For the larger end of the same system, the type scale generator and the heading scale visualizer cover headings, the CSS typography token generator exports the sizes as tokens, and the line height optimizer settles the leading for each role.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.