Quality in Typefaces & Fonts

What makes for quality type? What’s the difference between typeface quality and font quality? Who makes quality typefaces/fonts? Today’s post is partly an education for the beginner, but also a plea to my colleagues at other companies for more testing.

One of the things that attracted me to work in the type group at Adobe, when I was dreaming of such things a decade or more ago, was my belief that Adobe made the best fonts. Now, of course, I didn’t know all the world’s type foundries then – and with the ever-growing number of font vendors out there, I still don’t. However, Adobe certainly makes very good fonts, and my concern with quality has continued to this day….

First, let’s go ahead and make that distinction between typeface and fonts. A typeface for my purposes is the actual aesthetic design, and may encompass a range of styles and be implemented in multiple ways. A font, in this digital era, is a specific file (or in the older formats sometimes a pair of files) that instantiate that design on your computer.

For example, Times is a typeface. Linotype’s Times type family in digital form comprises four fonts, a regular, italic, bold and bold italic. Monotype’s Times New Roman may or may not be the same typeface (a debatable point, as there are some tiny aesthetic differences), but is certainly a different set of digital fonts.

So typeface quality is about how the design does in terms of aesthetics, and also the optical principles of perception.

Some explanation of these optical principles may be in order. They are illustrated in the graphic below, an edited screen shot of Myriad Pro Regular from the FontLab Studio 5 font editing program.
screen shot of vector font glyphs

Even Myriad, a relatively “monoline” sans serif, in which the lines are drawn to be all the same thickness, needs to have vertical lines slightly heavier (think 5-10%) than horizontal lines for them to look the same thickness. Similarly, round shapes need to go a tiny bit higher and lower (1-3%) than corresponding squarish shapes to look the same size – meaning that the top and bottom of the “o” needs to go just a tad higher and lower than the “X.” Curves also need to be just a smidge thicker than equivalent straight strokes to look the same (88 vs. 90 units in the example). Places where strokes join at a tight angle often need some tapering to avoid creating a dark spot on the letter, especially in bolder weights – so for example where the curve joins the straight on the left side of an “n” is subject to this effect.

Now, that’s just the easy part of the optical principles of type design – there’s a lot more to this. Achieving balance on curves and avoiding “lumps” can be tricky. Knowing which stroke thicknesses, heights, and other features should be consistent across letters is difficult. Spacing is especially troublesome – round shapes like “O” need less white space on either side than straight shapes like “H,” of course, but there’s also the question of how much white space the whole font needs. Sometimes this is done partly so as to balance the rhythm of vertical strokes, so a really bold face will actually have tighter spacing between the letters than a really light space, mirroring the smaller spaces between vertical strokes in a letter such as “n.”

Another aspect of aesthetic quality is originality…. Of course, there are few things that have never been done before in type design. But to have a few novel elements or combine things in a novel way is important, lest the design be completely derivative.

Sometimes people ask me why the world needs more typefaces. “Aren’t there enough already? How can you do something new when the differences are so small? In response, I often compare type design to furniture design: it’s not like we “need” another kind of chair, but people make them anyway. As with text fonts, the functional requirements of furniture put a lot of limitations on what one can do – yet it is still possible to do interesting and innovative things with furniture design.

One could get into legibility, but that’s got to be an entire article unto itself.

For type quality, Adobe does very well, I think. We tend to be on the conservative side in terms of typeface design choices, mostly aiming to produce classic typefaces that might be equally popular in another decade or even a century from now.

A number of our colleagues/competitors also do fine work in this line. some are companies with a long history, such as Monotype and Linotype (I’m particularly impressed with Linotype’s recent premium-priced updates of typefaces such as Optima and Sabon, and soon Palatino). Others are more recent companies, but still have strong classical design values underlying their work, such as The Font Bureau, Carter & Cone, Hoefler/Frere-Jones, Orange Italic, and Porchez Typofonderie, among others.

But it is certainly possible to do more “trendy” or even specialized/historical typeface designs while still having all the other attributes of quality design. In this department, designs from P-22, House Industries and Underware spring to mind.

(In both of the above lists I apologize in advance to the many friends and colleagues I haven’t mentioned. This is not an exhaustive list of everybody who does quality typeface designs!)

Now, that’s typeface quality. There’s also font quality. By this I mean the aspects of quality that are specific to the digital format. For example, does the font technically conform to the specifications for the font format it is made in, and any additional limitations imposed by the operating system, or most applications? Achieving outstanding font quality requires a combination of considerable expert human attention and examination with extensive automated tests.

Font quality is an area that Adobe tends to excel in. As far as I know, only the operating system vendors, Apple and Microsoft, have the same history of testing their fonts so extensively. Perhaps some of the printer vendors or the companies supplying them, too. But historically the type industry did not pay close attention to testing, and has not had a lot of people whose sole job was to test fonts. As far as I know, Adobe was also the first font vendor to institute a program of outside beta testing specifically for fonts. None of which should be taken to suggest that our fonts are completely bug free, of course. But on average they are quite a bit less buggy than most.

It’s worth keeping in mind that all the OpenType goodies that I am routinely promoting also create many more possibilities for bugs. Although these are mostly manageable, the one area that has proved particularly tricky is contextual font features, where glyphs change depending on which other glyphs are nearby. I’m going to do a big juicy post about this some time soon, but for now, I’ll just say some very cool things have been done with script fonts using this technology. Unfortunately, many of the third-party fonts I’ve looked at that have used this capability to something like the same degree as Adobe’s Caflisch Script Pro or Bickham Script Pro seem to me to be very buggy in their implementation of contextual substitutions. Our own heavily contextual fonts aren’t perfect, but they exhibit relatively few bugs, because of the months of testing that went into them.

What’s the point of this little story, anyway? It’s mostly aimed at type foundries, to say that:

(1) testing your fonts is important, and
(2) the nature and required depth of that testing depends on what kind of fonts you’re making.

I bring up these issues not to slam the competition, but to encourage those of our colleagues/competitors who have not yet dealt with this to do so. Adobe actually has a vested interest in all fonts being as reliable and bug-free as possible, because as often as not font bugs are reported to us as if they were application bugs, and we bear the burden as much as the font developer. Adobe has already made our key internally developed testing tools publicly available as part of our Font Development Kit (FDK) for OpenType, and we continue to improve these tests and release new versions.

I’m reminded of a (probably apocryphal) moment in the investigation of the Challenger disaster when a politician said “this isn’t rocket science!” Famed physicist and commission member Richard Feynman supposedly leaned over to a microphone and said, “actually, Senator, this is rocket science.” Fonts aren’t rocket science, but I’ll be darned if they don’t keep on getting closer to it.

(edited 14 Dec 2005 to change reference to “all” fonts to “many”)

5 Responses

  1. John Nolan says:

    Good points, Thomas.As you know, FontFont recently offered Meta Pro for download with the tied st ligature mistakenly included in the default ligatures. Clearly OpenType features need a bit of testing!

  2. Don McCahill says:

    I’m surprised to see no comments yet (perhaps T hasn’t had time to process them).Overall, this is a very useful article, and I hope Thomas will create a PDF of it and store it somewhere accessible for future reference.

  3. Robert Scoble says:

    The Cleartype team at Microsoft was talking on video about this same issue: http://channel9.msdn.com/Showpost.aspx?postid=146749

  4. Pam Bishop says:

    Richard Feynman is the man. It’s unfortunate for the cause of clear thinking that he isn’t around today to expose drivel and puncture pomposity. (Though even Feynman was defeated by school-board obscurantism.) Where can we find his like?

  5. Chris Lozos says:

    Thomas,Thanks for posting this. It will be useful to point people to when they have questions about type and its value. I often get questions about Opentype from non-technical people and this fills the void between the more nitty-gritty stuff and the not-enough-said stuff.Chris

Comments are closed.


Thomas Phinney

Adobe type alumnus (1997–2008), now VP at FontLab, also helped create WebINK at Extensis. Lives in Portland (OR), enjoys board games, movies, and loves spicy food.

Type 1 to OpenType compatibility

Thomas Phinney · December 2, 2005 · Making Type

Lying fonts - good or bad?

Thomas Phinney · December 23, 2005 · Making Type