Kerning pairs you can’t do without

Claudio Piccinini

While I am fine-tuning the kerning of a typeface I am finalizing, I was wondering — bearing in mind the expected needs of the average-to-advanced user of typefaces aimed at extended text setting — which could be the kerking pairs (or groups of pairs) which one might not immediately think about that is effectively handy to have incorporated in a professional font.

Or, inversely, these you can easily do without (ordmasculine and ordfeminine are used with letters?).
Any thought or consideration is welcome! :-)

Yves Michel

Claudio Piccinini said:

...which could be the kerking pairs (or groups of pairs) which one might not immediately think about that is effectively handy to have incorporated in a professional font.

Hello Claudio,
One of the texts I use in Fontlab 8 to test the kerning is at the following address:
Kerning Test · Typefacts
I hope it can help you.

Claudio Piccinini

Yves Michel said:

Claudio Piccinini said:

...which could be the kerking pairs (or groups of pairs) which one might not immediately think about that is effectively handy to have incorporated in a professional font.

Hello Claudio,
One of the texts I use in Fontlab 8 to test the kerning is at the following address:
Kerning Test · Typefacts
I hope it can help you.

Thanks much Yves. I have been compiling a text from many sources for some time.
I was thinking more if anyone has certain pairs (not necessarily "linguistic") which have proved to be handy/important, pairs which maybe one does not always think about.

Clearly at some point it becomes a choice how much to incorporate, but I was thinking of possible suggestions from typesetters rather than from us typedesigners (i.e. "I’d love not to have to kern that pair by hand each time!").

Adam Jagosz

One peculiar instance I had to kern manually was r/A (as in Adobe Pr/Ae). Something you would never give a second thought until it appeared in a navbar and not in running text. Ideally these would be also kerned with a space between, a narrow space between, etc... one can have dreams.

Claudio Piccinini

Adam Jagosz said:

One peculiar instance I had to kern manually was r/A (as in Adobe Pr/Ae). Something you would never give a second thought until it appeared in a navbar and not in running text. Ideally these would be also kerned with a space between, a narrow space between, etc... one can have dreams.

Yes, that’s the kind of things I was thinking about. I do have strings for kerning the slash in my kerning text documents, like this. I think it’s not so uncommon to have instances like, say Pear/Apple.

/a\ /b\ /c\ /d\ /ð\ /e\ /f\ /g\

/h\ /i\ /j\ /k\ /l\ /m\ /n\ /ŋ\

/o\ /p\ /q\ /r\ /s\ /t\ /u\

/v\ /w\ /x\ /y\ /z\ /þ\ /æ\

\a/ \b/ \c/ \d/ \ð/ \e/ \f/ \g/

\h/ \i/ \j/ \k/ \l/ \m/ \n/ \ŋ/

\o/ \p/ \q/ \r/ \s/ \t/ \u/

\v/ \w/ \x/ \y/ \z/ \þ/ \æ/

I was also thinking of more conventional situations, belonging to historical typography, which one might not consider a priority. I know there are professional designers which do not even consider numerals a priority, so I was wondering about the various experiences.
Thanks Adam!

FettleFoundry

Anything digital related really. Passwords looking like they have a space could be problematic? Underscore to underscore is one example I saw a tweet about recently, due to the use of them side by side in usernames. r/ is another example for anyone that wants to refer to threads on Reddit. Also CamelCase due to brand names such as UserZoom.

Edit: another one often overlooked is phonetic glyphs that are often poorly kerned or clash. 

K Pease

Consecutive underscores are not so much a kerning pair issue as a matter of changing uses. Traditionally they are designed to run together for the purpose of setting long blanks. In today's world, when it would be helpful to be able to count the number of underscore characters in some strings, we must decide what the default spacing should be based on which use case seems more likely. Perhaps unfortunately, as type designers, we've all learned the "proper" way and are slow to reconsider it.

James Hultquist-Todd

I always make sure to check // as in http://

James Puckett

James Hultquist-Todd said:

I always make sure to check // as in http://

Kern HH//HH and then you can kern the lowercase to the slashes with nn//xnn and nnx//nn, etc.

Thomas Phinney

I always check the “nT” pair as in “OpenType”—and usually look at “nTn” and “oTo” combinations just to see that I am keeping things balanced.

Nick Shinn

f? f* f’ f) f® 
These should also be designed to fit nicely.

Claudio Piccinini

Thomas Phinney said:

I always check the “nT” pair as in “OpenType”—and usually look at “nTn” and “oTo” combinations just to see that I am keeping things balanced.

Yes! While I do not want to "over-kern", I make sure to do a general check of the lowercase before Uppercase letters.
I use a string like this, adapted to all lowercase letters:

aAarm Basic aCard Date Far
aJackal Karma Lad aMage aOath
aPasqua Rail aTake aValue
aXandar aYale

For economy reasons, I tended to exclude pairs which for the usual forms of letters are very unlikely to happen, but /M is included as, while in many alphabets can have vertical sides, when the inclination of vertical stems is marked, it could benefit from some kerning.

Craig Eliason

Why "aPasqua"?

Chris Lozos

What about:  McCarty?

Claudio Piccinini

Craig Eliason said:

Why "aPasqua"?

I guessed that in a "aPa" instance, the eventual kerning pair "Pa" could benefit from being checked side-by-side with the "aP".

Claudio Piccinini

Chris Lozos said:

What about:  McCarty?

My /c "basic" string so far is:

cAct cCcel Eco Fclub cJcain cKcal
cLc cMcnally cOcra Pci Rcine
cTcine cVclose cXcity cYcle

So, yes, I thoughr about possible cC<lowercase> benefitting from kerning, but in general the cC pair fares fine by itself, and so the Ca (which is nonetheless included in my /a basic string. It’s a work in progress.