I have always been fascinated by Type. That is – the letters as they are set down on paper or on the screen. The New Yorker Dec 5 issue had a great short article on Mathew Carter, a fairly famous British-born American type designer. I know nothing about type, but I’ve read what I could in my spare time about typeface, its design, and page layout. All of these disappear into the background for someone with my day job, but reading the books you quickly realize there is a whole art form you are missing.

Computer Modern was probably the typeface that first got me interested in type and layout. This is the default typeface used by TeX, the typesetting program that is used almost universally by physicists the world around. The sweet of programs and the CM typeface are written and designed by D. Knuth, and are free and easily downloadable. I think the typeface is really ugly.

It is blocky and dense and makes for difficult reading, especially on the screen. If you make the margins too wide (as the default settings for TeX will do) the page becomes a smeared mess of black with small white borders. Not at all inviting to the eye, regardless of the content.

But the seriously cool thing about TeX, and CM, is how they are designed. At least, seriously cool from my nerd-brain point of view. Knuth wrote a type designing program called METAFONT. You specify a letter in the METAFONT language and then it renders it on the page. The specification isn’t by pixels – you don’t draw the typeface – rather you specify the typeface mathematically. For example:

beginchar("A", 8u#, 9u#, 5u#);
  z1 = ( 0u,  0u);
  z2 = ( 8u,  0u);
  z3 = ( 1u,  8u);
  z4 = ( 7u,  8u);
  pickup pencircle scaled 1u#;
  draw z4 ..  z1 .. z2 ..  z3 .. cycle;
  pickup pencircle scaled 3u#;
  drawdot z1;
  drawdot z2;
  drawdot z3;
  drawdot z4;
endchar;

There are a number of advantages and disadvantages to this. First, frequently you have to re-use parts of letters and you can draw them in common using this language. Second, the scaling program is basically solved. That is – drawing something 8pt and 12pt and 24pt is just a matter of scaling up the mathematical equations – something a computer can do. But a big downside is that fonts change when you scale them or try to render them on a low resolution surface (like a screen). So you have to provide hints. I think of hints as the type-face designers systematic errors: they have to take the longest time to get right! Hints augment the design and turn on or off specific pixels to make the typeface look right at a particular size.

I have never actually designed a type face, or done more than installed METAFONT and tried making a letter based on something that already exists. It takes too much time to do it. And visualizing the math and turning it into shapes was hard – I couldn’t do that in my head. But, I continue to lurk and read articles and books (short books!) when I get a chance.

On a related note. Have you seen the new LCD displays coming out? My Toshiba M200 has one. The resolution is about 50% better than the standard 72 dots-per-inch. Most computer programs render 12 point text as a certain number of pixels rather than an actual size – they don’t take the screen resolution into account. As these new displays penetrate the market that is going to change, however. And with higher resolution displays all typefaces will look better. One of Carter’s more famous and well read typefaces is Veranda, which he designed for Microsoft to be rendered on a 72-dot-per-inch display; with 50% more pixels he will be able to do a much better job. I’m really looking forward to that. Especially as my eyes go in my old age. ;-)