I’ve been very frustrated with my type sketches so far. The act of making these precise, controlled contour drawings has felt very alien to me. I’ve never done much calligraphy, so using shaped markers or pens (or simulating their strokes) doesn't quite feel natural to me. Drawing the outlines is difficult because I tend to distort the overall shapes when I focus on the details as I work my way around the character. I’ve had better luck by jotting down some key positions before connecting parts of the contours, so today I tried an experiment that got me over the hump. Instead of trying to adapt my drawing technique to what I thought type drawings should be, I just drew type the way I know how to draw: building up from gestural lines and basic proportions to to overall shapes to greater detail, refining the good lines and slowly eliminating the bad ones. It's how I learned to draw in a way that explores forms as much as it portrays them.
Now, that’s not a great set of letters by any means, but they've got some qualities that I like — not a bad result from a quick sketch. Whereas my drawings from last week were a collage of little details I had in mind, these forms developed as I kept adapting to an overall sense of mass and relationships between parts. Drawing like this uses motion that comes from the shoulder, not just the wrist or the fingers, so the straight lines have a little more tension and the curves have a little more swing.
Certainly these shapes have been informed by the sketches of existing types I’ve been doing all for the last two weeks. I’ve been looking at types that have some specific qualities I’ve been thinking about: open counters, contrast in stroke weight, some angularity to the serifs, curves meeting straight lines. Letting these qualities slowly emerge as I worked my way around the whole shape of the word helped me note some patterns and rhythms that I couldn't quite get before.
My first attempt to trace this drawing and refine the shapes wasn't very successful. I tried to make the second set of letters to clean too quickly, and I paid more attention to the conservative details than the vigorous ones. A second tracing went a little better, especially once I blocked it in in a similar way. I think a few more sets of gesture-based drawings will be a helpful way to explore some directions the type can take that’s a little bit more of me and a little bit less of an assemblage of details of other types.
So on the whole I’m loving things here, but do you notice a pattern in the weather forecast up there? I’m told there were a lot of drought conditions over the summer. Well, that’s sure as hell not a problem anymore. It's crazy wet here all the time. Even when it's not raining, you can feel the moisture all around you.
In a way, though, it's almost like Florida rain. It rarely rains all day; it just rains every day. Nothing ever has time to dry out very well.
I sure must be in type-geek heaven, then, because I’m happy as a clam despite the gloomy weather. (As a matter of fact, I am in type-geek heaven: no supposition about it.)
I also learned yesterday that Allison, the sassy and charming wife of classmate Vernon, is actually the voice of Dee Dee from Dexter's Laboratory! that’s completely awesome, and I’m a teeny bit starstruck. It does explain, however, why I got such crazy pictures after giving her my camera at the party. Dee Dee!
After a slow-ish couple of weeks of introductory sessions and reading, we finally began drawing some type today. It was a slow, clumsy start to be sure, but surprisingly encouraging b the end of the day.
Gerard Unger gave us each a specimen sheet of Morris Fuller Benton's New Century Schoolbook and had us blow it up so that we had letters about 1 7/8 in. (48 mm) high. The goal was to make modified tracings of some sample letters, transforming them from a book weight into a heavy bold weight with modified details. The end result tends to have the same vertical proportions as the Century, but (with any luck) a whole new personality and vocabulary of forms. It's also a good way to overcome the panic of having to draw type for the first time, since you have a basic guide to lean on until you begin going off in your own direction.
In the Unger Method you begin by tracing only the left-side contours of a letter: left side of the stems and serifs, left segments of curves, maybe a dash or two to indicate some tight junctions. Then, you slide your tracing to the left and create a wider stem weight by tracing the right-side contours and deciding how to best join the two halves of the character. You're also encouraged to begin modifying details and stroke widths as you go along, trying to explore new design ideas for the characters rather than just solve the weight problem for the original letter.
I tried to at least double the stem width for my letters, and I played with some forms to try out four different style ideas for the new glyphs: one with fairly short wedge serifs, one with very bracketed serifs with a trace of spring to the contours, one with softer serifs and details, and a very heavy egyptian. Witness my ham-handed first round of letters, traced from the Century specimen and then altered:
After looking at the set, I decided to develop the second style: the bracketed slab serif with some subtle angles to the horizontals and a fair amount of contrast to the stroke widths. Unlike the egyptian direction, it was easier to figure out the counter shapes when there was a little contrast in the mix. There was also some more leeway to move off the horizontal plane for some details, which gave the characters some spring and some variety along the baseline. The ball terminals definitely had to go, but I had some trouble figuring out the best way to replace them. This shows my final result: tracings of tracings, shown against the original specimen:
The "s" is the end result of a couple of attempts to get the curves and the terminals in better shape: a tracing of a tracing of a tracing. The "n" is also a 3rd-generation form, picking up and correcting details from the "d" and the "h" as needed.
Still clumsy overall, but at least some shape vocabulary is starting to emerge.
In Counterpunch, Fred Smeijers talks about the importance of word shapes in how we read:
At the root of configuration or overall visual arrangement is the design of the word. . . . Though it might seem that the type designer's aim is to make new characters, the real goal is to create a new word-image. [ch. 4, p. 29]
Good type leads to good word shapes which leads to good comprehension. Makes sense, right? It's a concept I’ve read about elsewhere, but as I try to develop a brief for this whole math type issue, it leads to some additional questions. Namely, how do we read word shapes in math? As I’ve mentioned before, equations don't really work as words quite the same way, and comprehension is a little harder since you can't easily rely on context to clarify individual characters. Equations have basic rules for structure and notation, though, so there are conventions that aid comprehension in that context. So what kinds of patterns emerge form those rules? What are the functional "word shapes" that make math notation comprehensible? And — most importantly — how can type be optimized to make those mathematical word shapes easy to read?
Spacing in math is, in general, much looser than in text since it's often a collection of symbols and operators. In a way, each symbol is a noun and each operator is a verb. Rows of symbols are usually shorthand for values multiplied by one another, and looser spacing helps clarify their separation from one another. You may find actual words like "sin" (sine) or "cos" (cosine) scattered within, set in roman rather than italic. These words ought to be spaced normally. You also have changes in vertical positioning all the time &mdash: superiors, inferiors, division, limits, matrices, etc. Readability, then, has to factor in the different meanings of space, different meanings of style, and different reading directions. This is pretty different from anticipating word shapes that move in one direction along one line.
Smeijers, Fred, Counterpunch: making type in the sixteenth century; designing typefaces now. London, Hyphen Press, 1996
The flatmates and I threw the first party of the year last night, since we haven't really gotten a chance yet to socialize with the rest of the department outside of classes yet. We're a big group this year — about twice the size of last year's MA group — and things have been a little too hectic in the classroom to talk much so far. It turns out that all we needed was a generous supply of beer, wine, and snacks to loosen up and get acquainted. My camera was seized by other people early on in the evening, and these are some of the things I found on it this morning:
Vernon gets started
Michal and Fernando chillax
Michael listens a little
Rob listens a lot
Jasso struggles with the world's most complicated corkscrew
Whatever Ian's talking about, it's big
Alice is much prettier than this drunken party photo suggests
Valentin throws Swiss gang signs
Shoko and Tim are jaded, and amused by the new group of MAs
Cormac isn't even a type designer, he's just Irish!
And those aren't even the worst of the lot. Tipsy party photography is never kind to anyone — the tea-totallers. You don't even want to know how goofy I look in all of them. (Well, you may want to know, but I'll never show the evidence, since it's my site.) Our little get-together seemed to do the trick and break the ice, though, even with a small handful of no-shows. Just in the nick of time, too, since we may be too busy to have fun for the next eleven months. By then, we'll probably share a bond more like war buddies who've survived a difficult assault, just like the groups that have preceded us.
(Note to Gerry [our landlord/program director]: notice that no one is breaking anything or smoking inside your flat. Everyone was very well-behaved. No need to fret!)
I will never have hair that is not completely tragic. I have since shaved it all off again.
Because of some freakish quality of my jawline, I often look surprisingly chubby in photographs. I am actually skinnier than I have been in ages. (Three cheers for poverty!) God forbid the camera ever notices that, though.
Rob and I are such chivalrous gentlemen that we would gladly use plates to shield sensitive lady designers from the harsh Iberian sun.
I wish my handwriting were nearly as nice as that of Laura Roslin, former president of the Twelve Colonies. I guess handwriting isn't as much of a lost art in a society where computer networks are considered a Very Bad Thing. I just wonder of that’s really her writing, or a Penmanship Stunt Double.
And a few other ideas to file away for the practical work:
The STIX folks have been keeping up with developments to Unicode, and are maintaining tables of recommended Type 1 character names. Look into that.
Structured documents can have a tough time dealing with the level of granularity that might be required for complex font changes, unlike TeX documents that seem to thrive on it. What can be done to anticipate fitting and usage problems in when it's not easy to change back and forth between fonts? For instance, can OpenType contextual alternates be used to insert terms like "sine" and "cosine" that wouldn't be set italic like other characters in math?
Also, spacing would have to be very different for characters that get used for equations, so a different font that spaces the glyphs differently may be needed: probably not monospaced, but certainly set wider, and with no italics that kern or overset their bounding boxes.
Should the type be optimized for screen display, publication, or some happy medium?
Italics should be fairly upright, but definitely italic in style rather than staying too close in form to the roman. They will need to be very distinct from one another.
Not too much contrast, or any details that are too delicate. However, not too blocky and informal, either.
The equation might may also benefit from having short ascenders and descenders to minimize trouble with spacing of overbars, stacked combining symbols, and divisions.
In addition to standard text and numerical glyphs, a good family for dealing with math would need a pretty robust set of agate glyphs for superiors, inferiors, dense tables, etc.
Mitja's "Reflection on Practice" essay talks about some good qualities that would be relevant to what I’d like to do, especially the notion of case-sensitive punctuation and operators.
An article by Donald Knuth and Hermann Zapf about the development of their Euler fonts for typesetting math gives me a lot to chew on. More than I can lucidly process right now, so instead let me jot down a few notes to file away for further thought or inquiry:
Knuth mentions a lot of qualities that mathematicians expect to see that are based on blackboard-writing conventions. Are those still relevant at this point, or is more teaching and research with math being done with electronic tools. If so, how do those tools present the math?
Optimization for screen display could be a big factor with the practical work.
Track down the digital Euler fonts themselves. The AMS only offers a few of the fonts as part of their TeX resources, and Linotype seems to have the full set, but only as part of a fairly pricey collection of Zapf's work on CD.
What other math development projects have there been? Something must be happening with the STIX fonts, right? What kind of research went into Microsoft's Cambria Math? What about Lucida Pro's math? What other major efforts were there before the Euler project?
Maybe a general idea for the dissertation could be an investigation of the various efforts that have been made to address the type-for-math problem. Every time I read about one, it seems to have been formed out of nothingness, without much inquiry into what's come before. that’s probably not the case, but it could help to dig up antecedents and follow them through to more contemporary efforts.
Knuth and Zapf talk a lot about the scripts and frakturs and such, not just the romans, italic, and Greek. Look for more examples of all of those.
Knuth, Donald E., and Zapf, Hermann, "AMS Euler — A New Typeface for Mathematics" Scholarly Publishing, April 1989, pp 131157
It's interesting to read "Future Tendencies in Type Design" in 2006, 20 years after Hermann Zapf first wrote this article about whether or not there is any point in updating classic typefaces for yet another new type technology. (Short version: He says "no.")
I tend to agree with him, for many of the reasons he cites. Typefaces are very much products of their own era and its technologies, and attempts to carry them over into other contexts lose a fair amount of the spirit inherent in the source. At the same time, it's not as if the basic designs of type should be laid to rest just because new technologies call for adaptation. Instead, it would be wiser to openly acknowledge the source and inspiration, but solve the problems of the new context from scratch without holding too slavishly to the model.
Of course, any revival of an old typeface is forced to do this to some degree or another. The problem, though — one which Zapf (and plenty of other people I’ve heard/read) feels has mostly been handled badly — is one of typefaces getting badly updated without enough regard to the past to accurately match them, or enough thought about the future to adequately evolve them in to something else.
It strikes me as a very Modernist stance to take: form follows function, so if the function (or manufacture or reproduction methods) changes, then the forms should adapt accordingly in order to give the best result. Zapf has seen his own work designed for metal go through some poor adaptions from film to digital, and wishes that market forces would have made it easier to create new versions altogether rather than corrupt the original ideas and slap the same names onto them.
The interesting questions come from the time this article was written, when digitization of type was really in its infancy. From the vantage point of a couple of decades later when we have more sophisticated type technology and more processing power and storage capacity for handling digital type, we're probably in much better shape to produce more faithful historical revivals. However, whether or not to do so is a big decision. Some foundries, thankfully, are coming out with newer, more sophisticated versions of their initial adapations of older fonts (Adobe Garamond Premier Pro, Linotype Sabon Next, Monotype Bembo Book), but they still involve compromise. At the same time, there seem to be more and more families like Hoefler & Frere-Jones' Mercury, which may start from some historical models but really blossom once they are adapted for contemporary usage problems and take full advantage of contemporary production technology.
Zapf's essay is followed in the same volume of Visible Language by an essay from Matthew Carter, in which he describes his historical references and the extent to which he followed or departed form them in his design for ITC Galliard. Carter based Galliard very directly on the work of Robert Granjon in the 16th Century, but describes in some depth how his own design for a contemporary typeface required many adjustments for technical reasons, market demands, and & perhaps most importantly & to preserve the actual spirit of the source material. To get something to work in film and eventually digital setting, a slavish recreation was less useful than an informed, sensitive tribute.
Zapf, Hermann, "Future tendencies in type design: the scientific approach to letterforms." Visible Language, vol XIX, no 1, 1985, pp 2333
Carter, Matthew, "Galliard: a modern revival of the types of Robert Granjon." Visible Language, vol XIX, no 1, 1985, pp 7797
The more I think about it and discuss it with people, the more I like this idea of concentrating on a typeface that’s optimized for setting math and other technical material. Aside from the amount of time I’ve spent grappling with the typeset math in my own work, conversations with folk like James Montalbano, Nick Shinn, and various folks at ATypI have made it clear that math is an interesting, complex topic with a market niche waiting to be filled, and one that other people seem to keep hoping someone else will figure out for them.
I didn't get a chance to do more than introduce myself to him, but at ATypI I met a German named Johannes Küster who has already done a great job of sorting out a lot of the issues of math characters that are often overlooked (he had some great PDFs from presentations he's given here), so that’s at least one person I hopefully can share my ideas with as go along. His work has confirmed (and gone much further with) a lot of the "rules" of handling technical characters that I’ve been filing away for years, but there's still a lot to think about in terms of how the text characters of the typeface can be optimized for publishing that kind of material.
Certainly I have a number of thoughts based on how I would have liked to redesign the ASME books, but I’m trying to keep my eyes and ears open for other points that haven't occurred to me already.
While reading Alison Black's Typefaces for Desktop Publishing (specifically, chapter 3, "Typeface performance in context," where she discusses the pros and cons of aligning and non-aligning numerals) I began thinking about some of the problems of character heights in the weird, mixed environment of math and technical material. With so many occurrences of all caps, abbreviations, and numbers mixed in with words, the overall texture of the type can get really ugly, and it's hard to imagine most users of the typeface taking the care to switch to small caps or oldstyle figures when those would be better choices. (Also, that kind of attention to detail also falls pretty far outside of the realm of structured documents, where the notation required to format text like that would be considered stylistic aberration rather than good semantic mark-up.)
A slight exaggeration to the height difference between caps/numerals and ascenders could help a lot, though. Minimizing the height of basic caps and numerals would keep them from sticking out so dramatically without toning them down too much. If handled well, you could get the emphasis of caps, but not more of it than you really need. (Obviously I need to keep an eye out for good examples of this, which must be out there already.)
Adobe's Tom Phinney was kind enough to gather up PDFs for many of the presentations from this year's ATypI in Lisbon and post them on his blog. He promises he'll post any others that speakers send him, so it might be worth checking back occasionally. Not all the presentations are useful as reference materials, of course, but they'll at least be useful to jog your memory if you were there, or point you toward some key concepts if you missed them.
The first day of school is always stressful. New faces, new challenges, a new environment, and many questions about what will happen from this point forward. Naturally, I’m all nervous about whether or not I’ve pick a cute outfit to wear. I may have jeopardized my whole future by looking cuter yesterday, when I only spent the day hiking in Surrey with my pal Tim.