Hello Typophile
I search on Typophile didn’t give me anything on this problem:

I’m currently designing a sans serif and during conversation with a fellow typedesigner he posed the question if I should group the characters by likeness and then harmonize the width of those characters by group, in order to achieve a better or more harmonious rhythm? Something rang a bell, and in Walter Tracy’s ‘Letters of Credit’ p.72 I did find a system that does exactly that, but only for adjusting the sidebearings, not the total character width, nor the width of the character sans sidebearings.

Then I took a look within Univers and Frutiger to see if there was a system, and indeed there was. There was a clear set of different groups of character widths (incl. sidebearings). Univers’ a e v x y all had a width of 556 units, while b d h n o p q u had a width of 611, and so on. Same deal with Frutiger. Other faces has the same pattern visible to a lesser or greater extent.

Another thing is of course how such a width adjustment would change the shape of the individual character.

My question is whether fewer groups with more characters grouped together and adjusted to the same total character width are better than more groups with fewer characters? My intuition tells me that there probably is a sweet spot somewhere in between the extremes, but where, and are there some pointer on where to go other than trail and error?

Claus Eggers Sørensen

PS: I hope I’m making sense here, as it is getting rather late.