I Love Typography

So you want to create a font. Part 2

Choices, Choices, Choices

[So You Want to Create a Font — Part 1]

The sheer number of fonts out there (MyFonts.com sells over fifty-five thousand) is a testament to the fact that there are nearly an infinitude of creative choices that can be made when designing a font. Of course there are the basics: serif vs. sans serif (and the numerous subcategories of each of these); handwritten vs. precision print-quality; wide vs. narrow; bold vs. light. But beyond these obvious choices are some specifics you may not have thought of:

  • closed or semi-open or open 4?
  • three-line or two-line Y?
  • descended or base-lined J?
  • two-storey or one-storey g?
  • two-storey or one-storey a?
  • crossed or joined or rounded W?

Examine a bunch of your favorite fonts to get ideas about these specifics and others—see if there’s a method to the decision-making of others. Do three-line Ys seem more traditional to you? Is that what you’re going for with your own font? Does a descended J fit your font, or are you just trying to force it in there?

Here are some other issues you may not have pondered:

  • the height of the horizontal bar of your e
  • the number of points in your *
  • the degree of the slant of your #
  • do your y and q have tails?

Don’t get so bogged down in details that you never get to the actual font-creation. But it does pay to think about some of these things before you dive into creating your font. A little well-spent time outside of your font creation software can save a lot of time inside of it, correcting problems or recreating glyphs.

Vertical Metrics

Another set of decisions to grapple with concern vertical metrics: the measurements that define the heights of your glyphs. Here’s some terminology for you:

  • The Ascender line defines the position of the top of lowercase characters (usually the topmost point of b)
  • The Caps Height defines the height of the uppercase characters (usually the height of H)
  • The x-Height is the height of most lowercase characters, like v
  • The Baseline is where your glyphs sit
  • The Descender line defines the position of the bottom of the lowercase characters (usually the bottom point of p)

Questions regarding vertical metrics you should definitely address before you start creating your font include:

  • Will your tall lowercase letters ascend to a line higher than your capital letters? (Many fonts do this, but not all.)
  • Where will your x-height lie? (You can achieve interesting effects by raising or lowering the “standard” x-height.)
  • How low will your descender go?

There are also some rules of thumb to consider when dealing with vertical metrics of your font.

Glyphs that curve at their bottom generally descend a small amount below the baseline. Likewise, glyphs that curve at their top generally ascend a small amount above the standard x-height or caps-height.

These rules of thumb are in place because generally speaking if a rounded glyph doesn’t ascend or descend more than a straight glyph, it appears to the eye that the rounded glyph isn’t the same size as its straight counterparts. That said, there’s no law that says you have to observe this. If your font works better where all of the glyphs are on exactly the same baseline and heights, than that’s what you should do. But doing it because you didn’t know any better isn’t really a great strategy.

Horizontal Metrics

You will be spending a great deal of time dealing with the horizontal metrics of your font. The major horizontal metric—kerning (to be addressed below)—can take many hours of fastidious work to get right. (Surely you’ve read Johno’s article on kerning, right? No? You really should. Go ahead. I’ll wait here.) But before we get to kerning, we should think about sidebearings.

Good sidebearings can help make for easier kerning, saving you some of those precious hours you’ll be spending at getting your horizontal metrics just right.

U Sidebearings

Sidebearings are the spaces on the left and right of your glyphs. In the image above, the “U” has the same symmetric left sidebearing as right. This is generally the case, but needn’t be. In some cases, one sidebearing can be positive and the other negative—that is, one sidebearing can be inside the font:

j Sidebearings

Notice that the left sidebearing (LSB) for the j is inside the boundary of the actual glyph — it actually cuts off part of the glyph. Why would you want to do this? Well, here’s how this j interacts with neighboring letters:

j Sidebearings

The inside-sidebearing tells your computer to render the j closer to its left neighbors than it would otherwise be. If the sidebearings were more symmetric, the spacing between the j and its left neighbors would be vast and unappealing:

j Sidebearings

This could all be fixed with kerning, but if the j is always (or nearly always) meant to be closer to its left neighbors, then having the negative left sidebearing means that you have less kerning to do, and that for users who don’t use kerning (most word processors, sadly, turn kerning off by default), the letters will be reasonably well-spaced.


Having read Johno’s article on kerning, you’re up to speed on the basics, right? Well, here are some specifics about how kerning will pan out for you as a font designer.

Having good sidebearings is like having a head-coach who has a good overall gameplan — you’re preparing your font for those who don’t use kerning, and making it presentable in most cases — but to really make things work beautifully in specific cases, you’re going to have to have a good offensive coordinator who gets his hands very dirty with the details — you’re going to have to kern.

One thing I like to do right after setting up the sidebearings is print out a list of my font’s pairs and visually inspect it for trouble spots that will need kerning help. In fact, I wrote a script to generate an Open Office document with these pairs; the document is available here for downloading. Just open in it in Open Office, select all of the text, and change the font to your font. Print and examine.

Kerning Test

Once you’ve identified problematic pairs, it’s time to get your hands dirty and fix things. Here’s an example of one of my fonts, and how the V-e pair looked after generating sidebearings, but before kerning. Notice how big the gap is between the glyphs.

V-e Pre-Kerning

And here’s how it looks after kerning:

V-e Post-Kerning

The idea is to make the letters flow naturally from one to the other. I often like to think of kerning as making my glyphs snug against each other.

A couple of things to keep in mind about kerning:

  • If you find yourself making kerning adjustments to every pair in your font, you might have a problem with your sidebearings. Good sidebearings should generally mean that some of your pairs are set up not to need individual kerning.
  • People will tell you that you only need to kern the most used pairs. For instance, with q you’ll only need to kern qu and maybe qa, not qz, because who the heck is going to be using qz in print? I, for one, kern every pair I can, no matter how obscure. Purists may faint and/or gasp in horror. But who am I to constrain the users of my fonts to only having standard pairs be beautifully kerned? If someone wants to print out qz, they should have qz print out beautifully.
  • Most font-editing programs have an auto-kerning feature. This can be a good place to start, but the received wisdom is that after auto-kerning you should really still go in and tweak things by hand. There’s no algorithm that the human eye can’t best.

Scratching the Surface

We’ve really just scratched the surface, here. And, if left untreated, font-itis might set in. I hope it does. After a couple of years of font designing, I still learn something new every time I fire up FontLab Studio. And I hope I always do.

[Alec Julien is a web developer and amateur typographer living in Vermont, US. He dreams of someday living somewhere warm, and typesetting a novel.]



  1. Another great article Alec.

    There is only one issue I have the order of the series thus far. You have kind of jumped from creating the letters to kerning the letters.

    There should be at least another article (or two) on the process of designing the font. For example: what letters to start with, general rules for width or letters in relation to other letters (w, m, ect. ect.), punctuation, and things like using an ellipse to help with serif and width consistency.

    I hope to see more of the above.

    Overall, great article! Keep ‘em coming.

  2. An excellent follow-up, Alec, and it’s certainly piqued my interest in creating a font.

    You say you want to live somewhere warm. Don’t you get some sunshine in Vermont?

  3. Wow, really good stuff Alec. Keep the information comin’. :)

    David: Vermont, warm? Not to sure about that, but then again I live in California. ;)

  4. Again a very nice writeup.
    For those interested in even more font picking goodness have a look at http://typies.blogspot.com/2006/11/15-tips-to-choose-good-text-type.html The article is a little older but very good.

  5. Cody
    You mention some interesting points. Perhaps we can persuade Alec to write more :) With your background, though, I’m just waiting for you to send me that “killer” article. Love your DIN Slab Serif, BTW. Mind if I post a pic’ here?

    Looking forward to Airey Sans ;)

    Good to see you here again. How’s the verb iLT “fan” club going?

    Great link. Yes, you’re right, it’s relatively old, but it’s a classic. Perhaps we’ll have a revised and updated version here on iLT.

  6. Robert,

    I’m sure California’s much warmer than Vermont, it’s just that living in Scotland, and seeing that Vermont is further south on the globe, I reckon there’s a hotter summer than I experience. Although double-checking it’s location, there’s no doubt a cold winter.


    Airey Sans will be one for every collection. ;)

  7. David
    Having lived in Glasgow for a year, I can vouch for the cold winters. I still shiver now, when I think of them.
    Re Airey Sans: I’ve started writing a review of it already…so far I’ve said that it comes in 6 weights with 9,000 kerning pairs and… ;)

    On a slightly more serious note, I do like the sound of Airey Sans.

  8. @Cody — You’re absolutely right. I did skip over way too much. I wanted to just present some of the crafting issues involved, and avoid the aesthetic issues, but, really, there are plenty of crafting issues with the actual font design, like the issues you’ve mentioned. What I think would be a good way to address these things would be with a case study — a report on the creation of an actual font, complete with details of width selection, etc. I’ll write as much as (or more than) you all can bear, but it’d be really interesting to have someone talk about designing a more “pro” font than the ones I’ve created.

    @David — Vermont doesn’t count as warm on too many scales. But the summers are gorgeous (short, but gorgeous).

    @Robert — Thanks! (California counts as warm on most scales.)

    @Squawk — Thanks, and cool link!

    @Johno — Glasgow, eh? Brrrr.

  9. I didn’t realize that kerning was such an important part of the font-creation process. Sitting down and going through every single letter combination has got to take a lot of patience.

    This was a pretty fascinating set of articles Alec. It is always neat to learn a little something about this, even though the odds that I’ll ever sit down and try and create an entire font are pretty slim. ;)

    I agree with Cody in that it would be neat to see a little more about the actual font-creation process itself.

  10. Even though this is not directly related to the actual process of creating fonts, this link has four interviews with typographers. http://www.myfonts.com/newsletters/cc/ I only read one, but that one was very interesting and gave somewhat of a behind the scenes look. (Thought I’d share this… yeish, I have to write some stuff for my own weblog sometime…;-)

  11. Alec, terrific stuff! Eventually, I think I’ll be printing all these pieces out—along with side stuff, like the article Squawk provided the link for. I’ll use all that as something of a text if this proceeds the way I sense it’s going. Then I just need to find something FontForge-specific.

  12. Chris
    It certainly increases our appreciation of type, and what actually goes into the finished product.

    Thanks. Another good link. I’ve been enjoying that series—nice to see the people behind the type.

    Yes, Alec’s done a wonderful job. You’ve reminded me that I really need to make a print style sheet for this site, so that you can print just the articles, without the sidebar and header noise.
    “Something FontForge specific”, is also a good idea. I might consider it myself some day, though I’ve only just begun using it. I’ve used FontLab Studio for some time, and it’s taking me a while to get used to a different interface (and time I don’t have much of).

    I think we need a vote for Alec button to persuade him to write more on this topic.

  13. Thank again Alec, for another great follow up! Today’s a good reading day, my copy of Thinking with Type just came in as well.

    I must say that Airey Sans has quite a nice ring to it indeed. One could say that naming a font is like naming a baby … of course, I’m probably not going to name my first-born “Frutiger” or “Helvetica”, but you know what I mean ;)

  14. @Chris — Yeah, there’s a bit of painstaking drudgery to font-making. Though I guarantee I’m one of the few obsessive/compulsive types who sits there and kerns every pair.

    @Steve — Thanks. And good luck finding FontForge help out there. The world of open-source is cool, but not very well-documented, as a rule.

    @Johno — Careful what you wish for. I’m generally a quiet person, but once you get me going it’s hard to shut me up. ;)

    @Hamish — How about naming your first-born “Minion”? It’s the best serif font, and your kid would be an instant bad-ass. Who would dare mess with Minion in the school yard?

  15. Alec
    Whatever you do, don’t shut up. I’ve enjoyed reading these articles as much as everyone else. I think 40,000 page views of Part 1, is evidence that we want more Alec Julien!

    I don’t know… “Frutiger” is not such a bad name for a kid; better than DIN (unless it’s a noisy child, that is). Oh, and Sabon would make a lovely girl’s name.

  16. Johno
    Feel free to post a shot of it up here. “Killer” article?! Do you mean the book review? It’s getting there =P

    Keep writing because I’m sure people still want to read!

  17. Cody
    I was thinking of having a readers’ type section, to showcase, well…readers’ type. I’d like to include some pieces from the your portfolio too (if you don’t have any objections). I’ll put it in there. Do you plan to finish it and publish it?
    I know you’re busy, so no hurry; though I’m looking forward to seeing it.

  18. Yeah! Feel free to take some pieces from my folio. It’s older work, some of which should be tweaked (a widow here and there), but go ahead and post some stuff.

    Yupp, I am planning to finish up the review. Well, planning to get seriously started anyways. I’m still kind of deciding how I feel about the book (I will explain why in the review).

  19. Johno,

    Only 6 weights? In that case I’ll launch it next week. ;)

  20. Nice follow up on part 1!

    With FontLab you can also make kerning classes. That way (for example) ‘ÁV’ or ‘ÅV’ will be kerned just as ‘AV’.

    My latest font Anivers (featured here in September fonts) has 116 class based kerning pairs (which comes down to roughly 1.600 kerning pairs).

  21. @Jos — I have a handful of topics that would make a good “Advanced Topics” piece, and kerning classes would be a great addition. I love Anivers, by the way. Beautiful work!

  22. Oops, I understand. It wasn’t my intention to scratch below the surface :-) I recently discovered this time saving FontLab feature and I found it very easy to use… Thanks for your compliment!

  23. Cody
    Thanks, and thanks for your mail.

    I’m a fan of too. Are you planning to do more with Anivers?
    My favourite though is Fontin—it’s beautiful. It will be mentioned in “November Fonts”.
    Would you consider doing an interview for iLT?

  24. Thanks Johno!
    I’m working on Anivers Bold, Italic and a Cyrillic version at the moment. Bold Italic and most certainly Small Caps (regular) will follow.

    I’m proud to read that Fontin will be in your november list. Really great!

    btw - I’m also working on a serifed Fontin :-)

    About the interview: I would be honored!

  25. Great article! I really learned a lot. I would have loved more info on stuff like ligatures and OpenType, though. But maybe that’s for part 3 :)

    I absolutely agree, people should pay attention to the kerning of every pair imaginable. Especially when the font is used for other languages than English, such pairs may become really important since those languages use different combinations (Czech, for example). Another problem are acronyms (rss, fps), file extensions (qxd), and company or product names (BBC, Studio MX, Windows XP, Xara XS). Those may even contain combinations like “tX” (such as in “DirectX”), which as a font designer, you wouldn’t necessarily think any sane person would ever use in their text.

    Most makers of free fonts also underestimate how many glyphs are really needed to make a font usable for other languages. I lost count of how many times I had to manually attach diacritics to letters (abusing commas, dashes, periods, and colons) in InDesign when working with free fonts from sources like dafont.com and non-English text.

    Consider this: French requires glyphs such as à, á, é, è, û, ç, ê, ë, œ, etc., German requires ä, ö, ü, and ß. Both French and German require special types of quotes, Italian requires è and à. Spanish needs those reversed exclamation and question marks, plus stuff like ñ. All of those need a Euro glyph (€). And everybody, including English-speaking users, would appreciate special glyphs for “…”, all sorts of dashes, hyphens, bullets, #, £, {, [, @, ©, &, and more. If they are not in your font, people trying to typeset text in any of those languages will curse you, no matter how great your font looks.

    @Hamish: Personally, I’d be curious what a girl with a name like Akzidenz Grotesk would look like …

  26. Johno
    Did you enjoy the little message at the bottom of the 3rd page of the PDF? I think there should be an ILT article on those.

    I have been keeping a close eye on your collection for quite a while now. I am really excited for the entire family of anivers to be released. Fontin Serif also looks like it’s coming together very well. Keep up the great work and I look forward to seeing you more on ILT.

  27. @Alec
    Haha! I can just hear it now, “Minion! Get over here!” :)

    Well, I still think Frutiger sounds odd, but DIN seriously had me laughing out loud — Imagine that…

    Sir, allow me to thank you personally for your wonderful specimens of typographic goodness. Keep up the good work!

  28. @ Cody
    Thank you. You will see me here as often as time permits me. I really like this blog. Anivers will take some time, because I have to finish Museo first, but I’ll post some samples of Bold and Italic on my blog soon…

    @ Hamish:
    Sir :-) I’m just someone who’s in love with typography as you are. Thanks again very much for your donation.

  29. I liked your suggestions about examining the details of favorites and thinking about how symbols will look. Wow, I knew there would be a lot to consider in designing a font but… whew! I think I’ll just search through those 55,000 on myfonts! And like Cody mentioned, you haven’t even gone into all the little details of creating the letters! I see you have designed your own fonts, Alec. It would be awesome to see a case study on your process! It seems you could really expand on this topic… perhaps start a companion blog… hmm….

  30. @Lauren — I’m working on a case study, actually (though, as I’ve said, it’d be even cooler if someone who designed a professional non-handwriting font would document that process). I might start blogging again someday on my own site, but it’s way nicer here in the nirvana Johno has set up, and where people actually visit! Love your blog, by the way.

  31. Alec, I’m sure you’d benefit in attracted readership from writing for this blog! But I don’t mind seeing more posts from you over here :) Johno has set up quite an amazing thing here! It’s neat to be a part of it, no matter how small. Thanks for the kind words about my own little bloggy!

  32. Alec - great article.

    A few years ago, I found a book from my great grandmother’s collection - a Speedball lettering book from 1907.

    Part catalogue and part technique guide, there were two alphabet samples in there. A gorgeous, light humanist face, and an interesting Asian-influenced brush script. I scanned them, and they’ve been languishing on my computer just waiting to be digitized.

    After reading this, I think I’ll tackle those faces again.Thanks for the inspiration.

  33. @Dave — that’s awesome! Make sure to keep us updated on your progress…

  34. Minion the best serif font, Alec? I think not. I mean, I like it. But there’s just something … I don’t know, missing, to call it the best. Like when I first got a load of Palatino—seriously, back in 1989, on my first Macintosh—I was instantly blown away; interestingly, Avant Garde did the same (and I wasn’t particularly impressed with Helvetica back then—but I digress from serifs). And then Adobe Garamond, I found that quite a special typeface family. Something visceral just reached out.

    Later Sabon. And for two days now, having just discovered it thanks to Type The Secret History of Letters, which I blogged about, moments ago, here, Goudy Old Style. This last is something special, I think. I look at it and I nod my head up and down, smiling. Yes, Goudy Old Style.

  35. Yeah, I think there are more beautiful fonts than Minion (e.g., Palatino), but personally I think there’s no font more readable over long stretches than Minion. And Minion was the first font I ever saw with genuine small caps and oldstyle figures, so I have a soft spot for it.

    I remember my first (and only) course in aesthetics when I was in grad school for philosophy—by the end of the semester the one thing I was convinced of is that systematizing beauty is an absurd project. And thus that saying that anything is the most beautiful or best boils down to, for better or worse, an unjustifiable feeling. My final paper was a (surely unoriginal) treatise on the uselessness of aesthetics. It got a B.

  36. I’m still trying to get around to putting my lefthanded Speedball penpoints into a pen and do something. Dave, do those two alphabet samples come with arrows pointing the way to which strokes and in what order? Is there any chance, you’d be willing to emal me the scan(s) of the light humanist face?

  37. Ah! Jos I love your site! It threw me there for a second… hehe

  38. Johno:
    Ha! I thought you would see that sooner or later. Sorry I didn’t ask first I just wanted to make one before someone else did. I just felt it was necessary since virb if focused on the design community anyways. I’m waiting for them to come out with their new version of virb they have been talking about. Maybe then I will be able to feed the RSS via a spring widget or something. Right now it’s very limited, but all in good time.

    Those of you who don’t know what virb is check it out! It’s the future. ;)

  39. Robert
    Certainly no need to ask. I was flattered. I need to spend more time around virb.

  40. @ LaurenMarie - Thanks! I hope my fonts were polite :-)

  41. Very good post, it is helpful. Thanks.

  42. Thanks for this great and helpful article, I am in the process of creating a font and this will definitely help.

  43. Cool, Mirko! I hope you’ll show us your work when you’re done…

  44. Kay

    This was a really great article you wrote!
    I will use it when I’m gonna go to create a font. Thank you for that.

  45. My pleasure!

  46. Hello, excellent site, very rich in content
    and correctly carefully thought out,
    personally I found here much interesting
    and useful

  1. teaching type—Nov 4, 2007

previous post: Sunday type: Spiekermann

next post: Apostrophes don’t swing both ways

May Fonts April Fonts March Fonts February Fonts January Fonts December Fonts November Fonts October Fonts September Fonts August Fonts July Fonts June Fonts May Fonts April Fonts March Fonts February Fonts January Fonts January Fonts November Fonts October Fonts September Fonts August Fonts July Fonts June Fonts May Fonts April Fonts March Fonts February Fonts January Fonts december Fonts October Fonts September Fonts August Fonts July Fonts June Fonts May Fonts April Fonts March Fonts February Fonts January Fonts December Fonts November Fonts October Fonts September Fonts August Fonts July Fonts May Fonts April Fonts March 2011 Fonts February 2011 Fonts January 2011 Fonts December 2010 Fonts November 2010 Fonts October 2010 Fonts September 2010 Fonts August Fonts July Fonts June Fonts May Fonts April Fonts March Fonts February 2010 featured fonts December Fonts November Fonts October Fonts September Fonts August Fonts July Fonts June Fonts May Fonts April Fonts March Fonts February Fonts January Fonts December Fonts November Fonts October Fonts September Fonts August Fonts July Fonts June Fonts May Fonts April Fonts March Fonts February Fonts January Fonts December Fonts