Cantarell Critique

(This is a repost of my post in the Google font forum)

Some months ago I stepped up as the maintainer of Cantarell, a typeface originally designed by Dave Crossland and later adopted as the default font of the GNOME project. The vast majority of design decisions and actual glyph drawings were done by others, I just tried to make a coherent, shippable whole out of it.

Cantarell serves duty as an on-screen font in OpenType/CFF form, no .ttf so far, because I love the hell out of Adobe's contributed CFF engine in FreeType and must force it down everybody's throat. My work was mainly focused on clean-up and expansion of language coverage. Pooja Saxena worked on the typeface before me and later forked the design into Cambay, to add Devanagari. I back-ported a lot of her redrawn Latin glyphs. The font now covers most of Adobe Latin 4 and Adobe Cyrillic 2.

I'm no type designer, so I'd love to get some feedback!

Main repository:
Get the fonts from:
Specimen: see below.

Known issues:
  • No kerning so far
  • Oblique variants are generated (and without correct alternative shapes for Macedonian, Serbian and Bulgarian)
  • The 5, 6, 8 and 9 need a hard look
  • Cyrillic spacing is probably somewhat tight


  • Ray LarabieRay Larabie Posts: 1,237
    I don't  think your oblique requires alternate Cyrillic italic forms. discussion
  • Spontaneous impressions, all very subjective and possibly fleeting:

    /B curves away from the horizontal too soon on top and bottom (especially in Bold), weird weight distribution in Regular lower bowl.

    Regular /G might profit from lowering the chin a bit. Why is its arc more closed than /C?

    Bold /J: Bumpy curves.

    Kissing /K out of character among the heavy joins in other diagonal letters? Bold a bit too light?

    /M: Heavy tops and symmetrical thinning toward central join feels weird.

    Regular /O seems a bit pointy.

    /Q: Arm rides a bit high for my taste.

    /S feels significantly wider in Bold than Regular.

    /U is very flat at the bottom, especially in Bold.

    Stroke weight seems to diminish in the Bold /T–/Y range and suddenly revert at /Z.

    There's something about the /g that sticks out to me, not quite sure what. Maybe the aperture of the tail is larger than anything comparable elsewhere? I'm mainly getting this impression from the screenshot above; it's much less noticeable in the PDF.

    Regular /m might be a bit compressed?

    The /r might profit from a lower join.

    The /s is a bit angular, especially in the Regular.

    The /t's foot is too short in the Regular.

    /z is top-heavy, especially bold.

    Figures: Yeah, /8 and /9 are really weird.

    The /a in /@ feels really large.

    The /fi ligature is really weird.

    Spacing might need some more tuning, not just kerning. For instance, the combination «Gr» in Regular «Gregor» is too tight. I'm often surprised at how much space the /G's front needs... The /i also strikes me as a bit narrowly spaced.
  • Oh, and Cyrillic /К and /Ж are way too narrow. And according to Leksandra, there's no reason to use curved arms in a humanist typeface; you could re-use the Latin /K's design:

  • John HudsonJohn Hudson Posts: 2,485
    Lots of proportion and some shape problems in the Cyrillic. Sorry I don't have time to go into details, but look at the optical proportionality of the various letters with two vertical stems.
  • Thanks for the feedback so far! I knew there was a lot of work left for a 1.0 :) And thanks for the links @Christian Thalmann and @Ray Larabie :)

    Regarding spacing, I actually respaced the bold face according to Phinney's spacing video but not the regular face (yet). On my todo list.

    @Frode Bo Helland: Regarding Å, the ring doesn't actually merge? Or do you mean at text sizes? And what exactly do you mean by undefined terminals? They don't seem uniform?
  • Nikolaus WaxweilerNikolaus Waxweiler Posts: 31
    edited March 2016
    Thanks for the explanation :) I also found many terminals lacking, I'll venture out to find some inspiration. What do you mean by the capital eng is useless?

    By the way, the Cantarell linked in the first post is the canonical one. The one in the Google fonts directory is quite old and will be replaced with mine as soon as I get a properly spaced oblique going. Which depends on a stable regular, which needs more work and which is why I'm here ;)
  • I'm confused: what is the canonical repository for Cantarell? I see the one you linked has 0.0.23 as the latest version but has 0.0.24.
  • Nikolaus WaxweilerNikolaus Waxweiler Posts: 31
    edited April 2016
    @Frode Bo Helland Ah! Ok, will look into it.

    @Stefano Costa Ah yes, there are basically 3 repos, all under my control. The GNOME repo on GitHub is the mirror of and my GitHub repo is the one where pull requests can be sent. The newer version tag is just cosmetic currently. I was just trying to say that the version in the Google Fonts directory is horribly outdated :)

  • Ray LarabieRay Larabie Posts: 1,237
    @"Frode Bo Helland"

    Are you sure the southern Eng is such a problem in Sámi? I've been looking for clues to how unreadable a southern Eng would be to Sámi readers and I can't find anything. Both forms have shared the same Unicode location for 25 years so I imagine they've seen plenty of southern Engs. Since the design is a large form of the lowercase eng, the form is already familiar. It's hard to track down because there are so few Sámi speakers. I couldn't find on site made for Sámi that was actually written in Sámi.

  • @Ray Larabie
    Besides a few exceptions like this book, most current Sami documents use the N-shape Eng. Some experts say the n-shape is wrong but I wouldn’t go that far although the N-shape is clearly preferred (i.e. the most frequent). The situation might not have been that clear before (there were fewer documents and both shapes can be found).

    There are quite a few Sami sites, some with PDF files as well.
    For Sami language sites: major news media like,,, newspaper (they use N-shape Eng in print in the paper I have), government sites like or others like or Notice the Diŋgo in the menu of There are many more.

    In Sami, capital Engi only occurs in all-caps words as it it never word initial.

    For other languages, in Australia, N-shape Eng seems to be the current preferred shape. For African languages the majority of languages users prefer n-shape Eng, but I’ve met publishers from Niger who prefer the N-shape Eng (while publishers of the same language in Mali prefer the n-shape Eng). In many of those languages Eng can be word initial.

    @Nikolaus Waxveiler
    Both variants should be in the fonts, the 'locl' feature can show the preferred shape based on the language used.
Sign In or Register to comment.