Adjacent Romans and Italics Can Collide
Adjacent roman and italic glyphs have an annoying habit of colliding. This is most often seen when italic letters and numbers are adjacent to or surrounded by roman punctuation.
Adobe Jenson Pro’s lowercase italic f is a good example: its descender extends beyond the left side bearing and its ascender extends beyond the right side bearing.

These extensions are called kerns, and in metal typesetting they would be physical overhangs that extend beyond the body of a piece of type. A font with these would be called a kerned font, which is where we get the name for the thing we call kerning today.
There’s nothing wrong with kerns, and when every glyph is angled in a similar way, there aren’t any collisions.
But if we were to surround italics with, say, roman parentheses…

Ironically, we can’t use the modern sense of kerning in order to automatically fix this. In most cases, the roman and italic styles are different fonts, and Inter-Character Features Don’t Work Across Fonts.
- Questions
-
I don’t understand enough about variable fonts to know how complete of a solution they offer.
In fonts where the variations are implemented via glyph substitution, you should be able to add kerning entries between an italic and non-italic glyph. But in most fonts I’ve seen, the roman and italics are delivered as separate fonts.
In fonts where variations are implemented via transforming glyph geometry and metrics, I’m not sure if there’s a mechanism that lets you adjust spacing based on the difference of a design axis between two glyphs.
Dealing With Collisions
One way to resolve this collision is to add a thin space or manually kern the text, which is what I’ve done here:

Of course, this requires the attention of the typesetter throughout the text.
Some typographers and style guides recommend italicizing punctuation following and surrounding italic text.11 I’m being imprecise and assuming we’re talking about italic text embedded in roman text. The rules change if: it’s roman embedded in italic (they invert), the punctuation is logically part of the italicized text (italicize it), the punctuation ends the sentence (exclamation points, question marks are set in roman).

While the 14th edition of The Chicago Manual of Style recommended italicizing trailing punctuation and allowed for italicizing surrounding punctuation (5.4, 5.6), the 17th edition now suggests matching punctuation with the surrounding context (6.2, 6.4).
Hochuli recommends italicizing trailing punctuation (not just to resolve the collision; he argues they look better in this case), but makes no recommendation on surrounding punctuation (Hochuli 44). Bringhurst prefers roman parentheses and brackets almost always (Bringhurst 85).
In some typefaces, using italic punctuation may not fully avoid collisions, although the issue may be relatively inoffensive. For example, there’s still a little bit of overlap in Granjon LT Std’s italics:

I prefer the formatting follow the logical structure of the content, matching punctuation with its surrounding context.
Consider the phrase: CHVRCHES’ second album (Every Open Eye). The parentheses are not part of the album title; should they be italicized?
Or this: CHVRCHES’ second album (titled Every Open Eye). Should the first parentheses be set in roman and the second italicized? That would be absurd, though see Terrible Ebook Typesetting: Contextless, Rule-Based Formatting for someone doing exactly that.
Preferring the roman punctuation in general works more consistently.
METAFONT’s Italic Correction
METAFONT glyphs have an italic correction value, charic
, that adds additional space to the right side bearing. (Knuth, METAFONTbook 315)
For example, here’s the snippet ({\it fluff})
in Computer Modern Text Italic 10. You can see the ff ligature and close parentheses overlap.

The italic correction can be enabled with \/
(Knuth, TeXbook 14), maybe the most ridiculous bit of syntax I’ve seen, which is saying something. So ({\it fluff\/})
becomes:

While it’s not quite how I’d manually kern the text, it doesn’t look as optically unbalanced at its natural size. I think the f’s charic
was chosen for some other contexts, such as math mode, where the italic correction is automatically applied.

Why is there only support for a right side bearing italic correction? I’m speculating, but a left side correction might have simply not been needed. Computer Modern’s italics have relatively small left side bearings, and therefore kerns. Knuth’s types just don’t seem to overlap on the left.

This italic correction value is specific to METAFONT fonts, so it doesn’t apply when using implementations like MathJax or with OpenType fonts.
Details aside, the thing that’s fascinating is italic correction is a feature that adjusts glyph positioning between different fonts, between the italic and roman styles, which is not a thing that static OpenType fonts can do.
Sources
- Robert Bringhurst. The Elements of Typographic Style. 2004.
- The Chicago Manual of Style. 14th ed., 1993.
- The Chicago Manual of Style. 17th ed., 2017.
- Jost Hochuli. Detail in Typography. Translated by Charles Whitehouse, 2008.
- Donald E. Knuth. The METAFONTbook. 1986.
- 1986. . The TeXbook.