Kerning
Why good typefaces need kerning
The best solution for an overall system is not always best for an individual—thus kerning.
Type designers are making the best of a tough situation: trying to whip 52+ little shits into coherent words while following some kind of masochistic design system. Not every use case is (or can be) considered, and designers bear the brunt of that through kerning.
How type designers failed you
Typefaces are designed and kerned with a specific use in mind (often text), and literally anything outside of that will likely require kerning.
Historically for digital fonts, the best case scenario was for a type designer to provide multiple fonts: a display version for headlines, and a text version for... text. The display would be generally tighter, maybe have a lower x-height, perhaps some more contrast, and the text would be the opposite—optimizing for maximum legibility at small sizes.
And that was the best case scenario! Typically you’d just have a sans spaced and kerned for text. Then every single time you used it in a headline you’d have to kern it. And by “you,” I mean the intern.
How design software failed you
Design software can’t make up the difference, at least not reliably.
Blood feuds with design software aside, kerning on the design software side is a tough job. Imagine having to write software that can reliably kern any typeface at a range of different tracking settings. Possible, perhaps, but definitely not an easy task.
Here’s some display vs text
spacing & kerning:
Workhorse Nouveau
spaced & kerned for headlines
Workhorse Nouveau
spaced & kerned for text
(Default +2% spacing)
Crimson Text
spaced & kerned for headlines
Crimson Text
spaced & kerned for text
(Default +1% spacing)
The great divide
There are two schools of thought on spacing/kerning: those who believe letters should never touch, and the rest of us who live in the real world.
Just kidding. But seriously, either way you kern, you’ll have someone breathing down your neck. You can’t win here.
Rules of thumb
Here’s when you should kern
When you track things
The way design software adds/removes space between letters can make awkward gaps and inconsistent spacing in a variety of kerning pairs.
In particular when you negatively track, take care that letter’s don’t combine to look like other letters. For example, sometimes v’s can look like w’s when they’re too close together, particularly at small sizes.
At a size it wasn’t spaced for
If a typeface was designed to perform well at a text size, the spacing can be awkward at headline sizes—particularly with uppercase-to-lowercase interactions.
Unique applications, like logos
Type designers have to space and kern within a larger system. Sometimes, when you narrow the system down to a handful of letters, the spacing and kerning can be improved.
From a practical standpoint this makes the most sense for really important things, like logos.
Parting thoughts
There are some typefaces out there that probably never need kerning.
Anything with flat sides is probably safe at any size or tracking. Neue Haas Grotesk also comes to mind. Given how tight the display version is, and how fucking stupid it would look tracked out at display sizes, you’ll probably never encounter a scenario where you need to kern it.
For better or worse, anything on the web can’t be kerned,
Every tool I’ve seen for kerning on the web messes with SEO and accessibility, which make’s it a bad idea for just about every website out there. It’s possible we might see some kerning features in the future, but I wouldn’t count on it. More likely is designers producing fonts with better, variable kerning features.