OMFG I Wrote a Book!
Design for Hackers: Reverse-Engineering Beauty (Wiley & Sons, September 2011) will help you see like a designer does.
Sign up for updates now »Amongst designers – especially print designers - Garamond is considered one of the best fonts in existence. It’s timeless, and very readable. But, because of the limitations of current display technologies, it’s not a good font to use in web copy – even with the advent of font embedding methodologies such as TypeKit and Google Font API.
One of the most important principles behind every good piece of design is that the designer has to master his or her medium. With any medium – whether it’s pencil and paper, steel and glass, or pixels – the designer has to work with strengths and limitations. Work with these characteristics, and the design stands a chance to be good – work against them, and there is no chance.
Apple’s lead designer, Jonathan Ive knows this. He recently said
The best design explicitly acknowledges that you cannot disconnect the form from the material – the material informs the form…
Medium and Form in Type History
Typography is the perfect vehicle with which to illustrate this concept throughout history. From the beginning, the forms of our letters have been influenced by the tools we used to create them.
This cuneiform inscribed tablet is an early example of how medium influenced form in written communication. You can see, looking at these pictograms, that they are made up of a series of indentions that are pretty much identical. This is because they were formed using a wedge-shaped stylus.
As this language was replaced in the west by our current roman characters, and the tools which we used changed, so did the form of our letters. Some of the best examples of early typography using roman characters are from – you guessed it – the Roman empire.
This is graffiti from the ancient city of Pompeii. It was created using a brush, and this is apparent in the letterforms. You can see there’s a great deal of variation in the strokes that make up the letters, and they all terminate with a soft point, just like you would expect from a brush.
Here’s a picture I took from Pompeii that I blogged about several years ago – dating back to the same time (remember, this city was frozen in time when it was buried under volcanic ash in 79AD). Only this time, the sign was chiseled in stone – and you can see how this has influenced the letters: all of the strokes of the letters are uniform in width, and to make the ends of those strokes looks nice – serifs were added. You can see little spur serifs from where the chisel was applied perpendicular to the stroke of each of these letters.
Now, moving more quickly through history, we have letters from the column of Trajan (which inspired today’s Trajan font), which were formed first by brush, then by chisel (it would have been awkward to chisel letters like the brush-drawn ones in the earlier Pompeii example). Then we moved on to lead and wood-cut printing, which first imitated work done by scribes with pens.
Once actual drawing tools were a smaller part of the design equation, typographers started to get more theoretical with their designs – creating constraints of their own – fonts like Bodoni are geometrically rationalized, as they were created in a medium (cast metal) with relatively few restrictions.
A Little Too Much Freedom?
In modern web typography, we still have the restriction that the letters of our alphabet take certain forms, but many restrictions have been removed. Rather than only having a couple of fonts available in our typecases, there are thousands. So, this makes it easy for bad habits to develop, such as trapping our information in images, or using fonts that just aren’t good for the web.
So, what makes a font bad for the web? There’s the widely-known issue of availability of fonts on the computers of our audience members – this, of course, is why we’re usually using widely-available fonts like Arial, Verdana, Georgia, Times New Roman etc.. Now there are some pretty feasible ways of using whatever fonts we want – methods like SIFR, Typekit, and Google’s new Font API, but that still doesn’t mean you should use just any font. Even great classics like Garamond can be a disaster on the web, so its better to use a modern font that has been drawn with the screen in mind.
And the reason behind this is that our display technology isn’t up to par with paper. You can see here a comparison of the great classic font, Garamond, blown up (as it might look on paper), next to a detail of what it would be anti-aliased at 12px height on a modern computer screen. You can see that it doesn’t look so good on-screen, because it’s just made up of a bunch of blocks of color.
Working With the Screen
So, the popular web fonts (Arial, Verdana, Georgia, and Times New Roman) are such not only because of their wide availability, but because they are drawn with the screen’s limitations in mind.
This Flash animation that I created illustrates how pixels distort curvilinear form – such as that of typography. It’s the same series of concentric rings, but as it changes sizes, you can see that a moiré effect results from trying to draw these rings out of mere pixels. So, the most web-appropriate fonts are drawn with these limitations in mind.
This illustration shows just what I mean by that. Georgia reads better on screen than Garamond primarily because it has a higher x-height (the height of an “x”), and – as a result – a larger eye. This prevents letters such as “e” from becoming muddled and unreadable, and overall makes the letters actually look larger. The notes on this illustration are in 9px Verdana with no anti-aliasing; and you can see those letters read very crisply, as this font was made for such an application.
Georgia has a huge advantage over Garamond on-screen because it was designed to be displayed as such from the very beginning, when it was designed by Matthew Carter for Microsoft in the mid-90′s. This has manifest itself in sharp serifs on Georgia, rather than more subtly modeled ones on Garamond. Look at little curve on the bottom of Garamond. This gets blurred at smaller sizes, and hurts the legibility of Garamond.
This limitation of screen technology has been embraced, and taken to extremes, though.
Starting in the late 90′s and early 00′s, we saw lots of pixel fonts being used in Flash, such as these from Craig Kroeger’s miniml.com, which are designed to be used at specific sizes, with no anti-aliasing.
When it was more common for computers to have only 256 colors, which caused dithering, designers embraced that constraint to inform their designs. Though ostensibly created to minimize bandwidth (another constraint of medium), designs that were created for the5k embraced dithering and lucidly used every pixel.
The “Web 2.0″ design trends of the last five years or so, are thanks to display quality and bandwidth improving, removing some of this constraint. In 2000, 12% of web users had only 256 colors on their monitors – in 2010, 97% have over 16 million colors (the number of colors available has a big impact on how crisply type, images, or *gradients* are displayed). This has put into the hands of designers a color palette beyond that of CMYK printing, with increased bandwidth to push it through.
Additionally, displays are cramming in more pixels per inch (ppi). The cheap Dell monitor I’m typing this on is displaying at 100ppi, and my MacBook Pro is displaying at about 115ppi. Compare that to the iPhone 4, which displays at an impressive 326ppi. Now, we’re starting to get some display technologies that are approaching the quality of paper when it comes to displaying letterforms readably.
So, maybe some day Garamond can make its comeback.
Pompeii graffiti photo from virtusincertus
Trajan’s Column photo from Silver Tusk


Markupper said,
August 13, 2010 @ 12:58 am
“Now there are some pretty feasible ways of using whatever fonts we want – methods like SIFR, Typekit, and Google’s new Font API, but that still doesn’t mean you should use just any font.”
There are no reason to use flash-based sIFR anymore. Modern alternatives such as Cufón renders faster, better, plays nicer with html and needs no plugins.
adryano said,
August 13, 2010 @ 4:14 am
what a great article!
Thanks for the ancient research and information!
Dan Strohschein said,
August 13, 2010 @ 4:57 am
Please please make more articles like this. I’ve been a programmer for 15 years, and have struggled with design – but this was DEFINITELY eye opening. MORE MORE MORE!
Visitor said,
August 13, 2010 @ 5:54 am
Funny this post discussing fonts, when this very page has a font problem as the upper part of the smallcaps gets cut. Have you checked how this site looks in Safari?
Paul said,
August 13, 2010 @ 5:57 am
More designers need to know this. Working as a developer in this field with an MFA I see too many print designers treat the web like paper. The best designers I’ve worked with in this regard have the same understanding of issues like this as they do of paper textures and ink surfaces that make outstanding brochures. In response to the first comment, from a development standpoint sift is still a reasonable choice for three reasons: efficiency the flash player is 100 times more efficient at rendering text than canvas and vml especially on IE who’s Dom manipulation and js engine leaves a lot to be desired. Aliasing: users without cleartype would see crisper fonts under sifr and cufon but I would still defer to sifr for antialiasing. Size: a sans font I recently used in cufon was 11k and the cufon library is around 20k so the font scheme in this case costs about as much as a banner ad. I’ve had serif fonts in sifr weigh in at around half that. Does it render on iPhone? No. I haven’t used sifr recently enough to find out if the fallback is displayed, but that’s what it is there for.
Geoff Forster said,
August 13, 2010 @ 7:29 am
Hi, Originally the serifs were added as endpoints by stonemasons to prevent cracking and splitting of stones or tablets. They were soon seen as embellishments and used in calligraphy.
Mathieu Federspiel said,
August 13, 2010 @ 8:11 am
Serifs are supposed to help the eye move across the page, and this seems true while reading books. So why is it that sans-serif fonts seem better on computer screens?
My guess is that the “back-lighting” on screens makes sans-serif stand out more, but it may have to do with the pixel rendering of serifs as you described above. Only a few web sites use serif fonts to good effect. I have never seen this question answered; perhaps it is worthy of another article.
Thank you for this excellent article!
Tim Farland said,
August 13, 2010 @ 8:45 am
Top post! I like how you espoused a general principle (truth-to-materials) through a specific example. This seems a great way to teach.
garrick van buren said,
August 15, 2010 @ 7:37 am
Given the increased resolutions of computer displays, I encourage web designers to worry less about readability and use fonts _appropriate_ to from a visual design and licensing perspective. Just as print designers have done for decades.
Jason said,
August 16, 2010 @ 7:49 am
Great article, now can you do another on why you don’t use Myriad Pro on the web? lol.
moioci said,
September 28, 2010 @ 11:36 am
I think the singular of grafitti is grafitto.
michael said,
September 29, 2010 @ 7:52 am
I propose you “design” an RSS logo on the top of your post page so people can subscribe. Not everyone enters this site from the home page.
jaygeek said,
October 2, 2010 @ 8:13 pm
Great article… Just a note on something that was perhaps a little apples:oranges. For the Web 2.0 section statistics, your numbers were a little off. Thanks for the reference link that shows “In 2000, only 34% of web users had 16 million colors and in 2010, 97% have that… (In 2000, as much as 12% used only 256 colors on their monitors… Today, it’s 0%.)”
David Kadavy said,
October 6, 2010 @ 7:01 pm
Hey Michael – I’m confused, since there is already an RSS logo at the top. Perhaps I need to make it more prominent.
Gene Wirchenko said,
October 27, 2010 @ 12:16 pm
I love clear explanations like this. Thank you very much.
Sincerely,
Gene Wirchenko
Oliver said,
August 11, 2011 @ 3:50 pm
Brilliant write-up, have noticed this “rule” myself, despite using Garamond for practically everything I print. I’ve been using Cambria in my applications as it was specifically made for easy reading at small sizes on screen which is the situation I’m usually in.
@Jason there’s a Myriad Web Pro as it’s name suggests it’s Myriad Pro but optimized for the web so you’d want to use that if your set on myriad.
Matthew Cowan said,
September 30, 2011 @ 8:44 pm
I would disagree with the notion of not using Garamond online, just based on this site alone: http://www.mcsweeneys.net/tendency
Pat Aldridge said,
December 2, 2011 @ 3:30 am
Very interesting, thank you. I got here via your even more interesting article on hue and shade while I was checking online what Impressionists did about shadows without using black. I am trying to write fiction and will definitely look you up again if I ever get as far as trying to set up a website.
Laurel L. Russwurm said,
January 6, 2012 @ 5:47 pm
Thanks for an interesting article. I agree that the serif fonts were more problematic for digital display when resources were more limited, but that has been changing rapidly. Everyone ought to be able to design and present their website as they choose; but I suspect the biggest real impediment to using “off the beaten path” fonts is more due to font licensing issues than actual technical difficulties.