PDF reference
Font substitution is what happens when something needs to draw a character and the font it was told to use cannot supply it. The page still renders — that is the point of substitution — but it renders with a different typeface, and usually with different widths, which is why a substituted line so often looks subtly crushed or stretched rather than obviously wrong.
PDF font substitution is a viewer or editor replacing an unavailable font with a different one so the text can still be drawn. It happens when a font is not embedded in the file, or when an edit needs a glyph an embedded subset never contained. Substitutes are chosen to be metric-compatible, but different widths still shift spacing.
It happens at two distinct moments, and confusing them is why the problem is so often misdiagnosed. A viewer substitutes when it opens a file whose fonts were never embedded. An editor substitutes when it is asked to write a character the embedded subset does not contain. Same word, different cause, different fix.
The best outcome is not a lookalike at all. If the original font names a family the tool ships in full — this editor bundles 38 of them, including Roboto, Inter, Lato, Montserrat, Poppins and Open Sans — the real face is embedded and the characters keep the original typeface.
Otherwise a metric-similar face is used: Liberation Sans for Arial, Liberation Serif for Times, Liberation Mono for Courier. These were designed to match the originals’ widths, so the line breaks in the same place even though the letterforms differ slightly.
A condensed face replaced by a normal-width one is the most visible substitution there is — every word gets wider and the line runs long. Width class is read from the original font program, not guessed from its name, which a subset tag may have mangled.
A Poppins SemiBold run takes the real 600 file rather than being flattened to bold. Rounding 600 up to 700 is a small difference that reads as a different font when it sits next to unsubstituted text.
Which face was substituted is recorded and surfaced rather than hidden. A substitution you know about is a decision; one you do not is a defect you find out about from the recipient.
The failure mode this avoids is the classic one: a tool that swaps in its own default font for any unavailable glyph, producing one line in the wrong typeface in the middle of an otherwise perfect document.
If the whole document looks wrong on one machine and right on another, it is render-time substitution — the fonts are not embedded. If one line or one word looks wrong everywhere, it is edit-time substitution introduced when the file was changed.
Run the file through the free PDF Font Embed Checker. Any font reported as referenced rather than embedded is a render-time substitution waiting to happen on every machine that lacks it.
For an edit-time substitution, look at what was typed: accented letters, curly quotes, em dashes and currency symbols are the usual culprits. They are the characters most likely to be outside a subset built from an English-language document.
Zoom in where the substituted text meets untouched text. Different stroke weight, different letter width or a shifted baseline tells you two font programs are drawing one line.
The two substitutions have almost nothing in common except the name.
| Render-time (viewer) | Edit-time (editor) | |
|---|---|---|
| Cause | The font was never embedded | The embedded subset lacks a glyph |
| Scope | Every run using that font | Only the run that was edited |
| Where it shows | Only on machines lacking the font | Everywhere, in every viewer |
| Who chooses the substitute | The reader application | The editor writing the file |
| The fix | Re-export with full embedding | Reuse the embedded font, or embed the real family |
| Detectable before sending | Only by opening it elsewhere | Yes, if the tool tells you |
A metric-similar face matches advance widths so line breaks hold, but the letterforms are not the same. Under close inspection, or in a headline at 40 points, the difference is visible. It is a good compromise, not an invisible one.
If the original is a licensed typeface nobody has, no substitution reproduces it. The only real fix is a source file re-exported with the font embedded.
A font whose glyphs are icons rather than letters has no meaningful metric equivalent. Substituting one produces letters where symbols should be, which is the most obviously broken form of this failure.
It eliminates render-time substitution outright, which is the half of this problem you cannot detect on your own machine.
A phone will do. It is the only reliable test for fonts your own computer happens to have installed.
If the page contains no em dash, do not introduce one. Reusing characters the document already has means no new font program is needed at all.
Typing an apostrophe that gets converted to a curly one introduces a character the file may never have contained — a substitution caused entirely by an autocorrect setting.
It is a viewer or an editor drawing text with a different font from the one specified, because the specified font is unavailable or lacks the glyph required. The text still appears, which is the point — but in a different typeface, and usually with different widths, so spacing changes too.
Almost always because its fonts are not embedded. The file names a typeface and leaves the viewer to find it; on a machine that has it, the document looks right, and on one that does not, the viewer substitutes. Nothing about the file changed — only which fonts were available to draw it.
That the substitute font’s glyphs have the same advance widths as the original’s, so text occupies the same horizontal space and lines break in the same place. Liberation Sans is metric-compatible with Arial, Liberation Serif with Times New Roman. The shapes differ; the layout does not.
Because that word was edited and needed a character the embedded subset did not contain, so a substitute was used for that run alone. The rest of the line still draws from the original embedded font, which is exactly why the mismatch is so easy to spot.
Not from the viewer’s side — it substitutes because it has nothing else to draw with. The fix belongs at export time in whatever produced the PDF: embed the fonts. Once the font program is in the file, no substitution is needed or possible.
No. It is the reason a PDF with missing fonts still opens and remains readable rather than failing. The problem is not substitution itself but silent substitution — being given a document that looks slightly wrong without being told why.
Use the free PDF Font Embed Checker on this site. It lists every font and separates fully embedded, subset-embedded and referenced-only, which is precisely the distinction that predicts whether the document will survive being opened somewhere else.
Yes, and print is where it is least forgiving. A printer’s RIP performs the same substitution, and a difference that is barely visible at screen resolution can be obvious at 600 or 1200 DPI. Commercial print workflows reject non-embedded fonts for this reason.
Every one of these opens the same free online PDF editor — each page covers what makes that particular problem hard, and what to do about it.
A PDF font is an object on the page that carries a font program, an encoding that maps byte codes to glyphs, and a width table for positioning.
A PDF font changes when editing because most PDFs embed a subset — only the characters that document actually uses.
PDF editors break formatting through five distinct mechanisms: converting the file to another format and back, re-saving the whole document rather than appending a change, re-flowing the edited line from scratch, substituting a font the file does not contain, and rasterising pages into images.
PDF layout survives operations that leave existing objects in place — editing a text run, replacing an image, merging, extracting or deleting pages — and breaks under operations that rebuild the page: converting to another format, rasterising, or re-flowing text.
Free, no signup, no watermark, nothing to install — and no round trip through another file format to get your document back.
Open the PDF editor