Dear TypeDrawers community,
I feel so privileged to be a member of this community, even though I do not design type — I only use it. More than once I’ve seen comments from type designers about the lack of direct feedback from the people who use their creations. It occurred to me that I might be in a position to provide some feedback, as a book designer, about the features that I would most like to see in text typefaces. For context, most of the text that I typeset is either literary or scholarly. I tend to work in a classical idiom, though I will sometimes take a more contemporary design approach.
Here, for whatever is may be worth, is my unofficial and unsolicited wish list — consider this to be something like my inner monologue when I am typesetting a book. I will omit the various mental expressions of “Gosh, that typeface is beautiful! Look at that pilcrow!” and the like, and focus only on my gripes. This thread was inspired in part by a recent thread about what characters compositors want to have included in fonts, but not everything below is relevant to that.
JL’s completely unasked-for but possibly informative typographic wishlist
I wish that italic text faces included roman delimiters (parens, brackets, braces) as a stylistic set. I almost always use roman delimiters in italic text, automating this with GREP styles, but then I have to kern them manually. If they were built into the font as a stylistic set, the type designer could include kerning.
I wish that more text faces included Unicode “mathematical angle brackets” — which is to say, proper tall angle brackets that match the other delimiters, not greater-than and less-than signs. I need them now and then — usually for textual scholarship, not math — and I always need to borrow them from another font.
I wish that more text faces included prime marks. These are very often necessary, and seldom included. My usual workaround is to borrow the quotation marks from Optima. But this cannot be automated with GREP, because the quotes are different characters, not just different fonts. And it creates encoding problems for electronic text.
I often wish that parens and brackets were less wimpy. I sometimes borrow delimiters from another font because most of them are too small or light for my taste, and they are spaced/kerned far too close to the text within them. I prefer delimiters to take up the full ascender/descender height, so that lowercase letters are vertically centered within their space. I almost always need to kern delimiters to add space inside them.
I wish that : ; ! ? had larger left sidebearings. Traditionally these punctuation marks are spaced more loosely than the period and comma, but in most modern fonts they are not. (A good example of a modern font that preserves this is the custom Cheltenham used on the NYT website.) Again, I do this with GREP.
I wish that superscript numbers were always proportional. In many faces they are tabular only, which often disqualifies a font, or at least its superscripts, for my use. If a superscript 11 looks more widely spaced than a 22, that presents a problem that cannot be easily solved in InDesign (because an individual superscript note number cannot be adjusted manually).
I wish that em dashes always kerned to create an unbroken rule for bibliographies, etc., even when the dash is designed to be shorter than an em and have sidebearings.
I wish that the copyright symbol was vertically centered on the small cap height, so that the “C” rested on the baseline and the circle descended. Same for phonomark.
I wish that the prebuilt ellipsis character had slightly wider spaces between the dots. Although I usually build my own ellipses, having a more usable single character would be nice for web design and other circumstances where it’s not practical to build my own.
I wish that the italic ampersand were provided as an alternate or stylistic set in the roman font.
I wish that roman superscript numbers were provided as a stylistic set in the italic font, and kerned appropriately. Even in italic text, note numbers are usually roman.
I will very likely think of more items to add later. I hope that this might spark some interesting discussion and maybe provide a bit of information for designers of text faces about how this particular typesetter — who is, in all likelihood, not representative of other typesetters — expects or prefers text faces to work.
With respect and appreciation,
Josh
Comments
Why borrow Optima's quotation mark for prime instead of using the native one of whichever font it is you're using? Is the shape usually inadequate?
As for mathematical angle brackets, are you aware of left/right-pointing angle brackets (U+2329, U+232A)? Some fonts might have the latter but not the former (the names possibly suggest that it's the latter that is aimed at your non-mathematical use).
Ori: I borrow the "curly" quote from Optima, which is actually straight and slanted, and resembles what I would expect a prime park to look like in a typical serifed text face.
Speaking of primes, the degree symbol is another character often missing from text faces. Needed for things like geographic coordinates, in addition to temperatures. Also sometimes used as a relatively unobtrusive footnote marker.
Ori, to return to your first comment, those are indeed the codepoints for angle brackets that I was referring to. Maybe the "mathematical" ones are different. The sample glyphs for these brackets in the Unicode charts give me the impression that they're designed to harmonize with CJK scripts — they are placed within a full em square — but maybe I'm wrong about this. I think they are also supposed to be the codepoints for textual angle brackets in latin scripts. Regardless, I do wish that more text faces had angle brackets that matched the other delimiters in style, weight, and alignment. They are useful for all sorts of things.
About angle brackets, I guess you wish something that isn't encoded. Codepoints 2329 and 232A are deprecated because they duplicate the 3008 and 3009, which are part of CJK punctuation. The 27E8 and 27E9 are mathematical operators so they will most probably not match the text style. The best solution seems to have alternative glyphs for 2039 and 203A.
Could you please give details about how and when do you use these delimiters?
Not only I find it very relevant, as related to actual use of digital type, but it is also particularly meaning for me now that I have decided to (hopefully) make "serious" type design my main job.
For now, let me tell you that, since I love to use delimiters and various brackets, I had been thinking about angle brackets recently and thought to insert them in De Vinne (although obviously they were not historically there in the late 1890s).
And I am including prime and second marks in my digital De Vinne. De Vinne is not a text face, but I am also developing De Vinne Text (on the original 1898 design), and your suggestions will be precious!
And of course, case-sensitive versions are always welcome. But really it would be pretty unusual, at least in my experience, to be setting a copyright notice in all caps. I wonder if the copyright symbol is almost better thought of as being related to the figure style of the font — with OS figures, it should align with the small caps. With lining figures, it can be larger and sit higher.
This is a fairly typical use of different kinds of brackets in textual scholarship. Of course, as soon as I point out that this book was written by Robert Bringhurst, we know that in its attention to typographic detail it is anything but typical. Surely those angle brackets don't come from Aldus — Bringhurst either borrowed them from some other font or drew them from scratch.
In the Folio Society edition of this book, which is for the most part very carefully designed, the typesetter seems to have given up on this particular nicety and used < > instead.
(The book, for those who are interested, is A Story as Sharp as a Knife.)
2329/232A were added in Unicode 1.1 as part of Miscellaneous Technical block. They are used in math and computing science. But also in Unicode 1.1 the CJK punctuation block was added with the 3008/3009 characters. Probably by mistake, both pairs were set as "canonically identical" —this is, the characters are the same, but with different meaning.
This caused a problem because these brackets in CJK have a large blank space at their left (3008) and right (3009) sides to fulfill the square area of CJK characters. This doesn't work in math and computing notation, when the excessive space in not needed or desired.
This was later solved by setting 2329/232A as deprecated and by adding the new operators 27E8/27E9 in a math-only block. Once Unicode does not remove codepoints, the original 2329/232A remains, but with the deprecated note.
Is there any problem to use them as editorial marks? Probably not. If you add a 2329/232A pair with different design than 3008/3009, you are violating the Unicode definition because the characters aren't identical. But I am not aware of any practical issue caused by this except the one you (Adam) already described when switching fonts.
Andreas, I would have to really take some time to come up with a useful piece of text to disseminate as a testing ground. But here are a couple examples of the kinds of things I routinely do with text faces:
From a literary journal:
Note the upright brackets. If I were doing this now I might even consider adding a little more space within the brackets.
From a preliminary test for another project:
In this case, all the delimiters have been manipulated somewhat. The brackets are resized and repositioned, and the parentheses are from a different font. The semicolon, however, has a pleasing left sidebearing to begin with and needed no adjustment.
Andreas, if you are looking for a test to run, the first thing that comes to mind is to set some text in italic with roman delimiters and see what happens in your typefaces. I think you will see that quite a bit of manual adjustment is required, which could be incorporated into the font if the upright delimiters were included as a stylistic set.
Maybe when I have more time I can cook up some kind of test document that addresses other items on my list.
Here is what I get with the query "angle bracket":
And this full width versions, which only one of my ~200 installed fonts have:
Tried some of the fonts and only Everson Mono and Cardo have U+2329/U+232A:
Here the test string for copy & paste:
([{⟨|⟩}]) (Hhxxx) (pq) (Oqo) [ܰ”Ù] [.,_] Fontname
1. I wish that, when a designer goes with a pelican-neck italic /g/, they give us the choice of switching to a regular, two-storey /g/ as an alternate. Some of us don't like that pelican thingy nearly as much as you think we should. Please don't forget about us.
2. I wish that whoever goes with a finial entry serif on lowercase /a/ gives us a choice of an alternate /a/. (That's right, GT Super.) Some of us don't like that finial, ball-less thingy nearly as much as you think we should. Please don't forget about us.
3. I wish all the kewl kidz grasped that 1997 is long over, and going neo-Brutalist on a text font is deader than disco. Stop chopping off elliptical terminals designed for pleasant, smooth reading. Your neo-Leninist ugly-for-ugly's sake is grating. If you insist on that ugliness, at least please give us the choice of an alternate set made for actual reading. I'm looking at you, Alcala (205TF), Lyon Text, Immortel (205TF), Signifier (KLIM) and their epigones. Knock it off.
4. I with MyFo/MTI stopped trying to digitize and sell everyone's handwriting, and instead they spent a few bucks to expand into further grades some of the old classics. Janson, Garamond 3, Caslon 540, Century Expanded, Palatino -- why don't they come in grades G1-G4, like Hoeffler's fonts, made for different screen, printing, and photocopying conditions? Paper is not dead, you know, and we'll be printing documents long after your cool MacBooks have gone the way of Pets.com.
5. I wish designers chose to sell their fonts in two variants. STD, at a reasonable price, for people who aren't looking to design a handcrafted-cocktail menu for some cool place in Upper West Swaziland; and PRO, at a higher price, for anyone who needs 3,000+ glyphs per style. I'd buy a lot more new fonts if they came in an Extended-Latin set at a decent price. I can't afford to spend 80 bucks per style just to subsidize the work you put into expanding your glyph set for speakers of Friulian, Telugu, and Middle Occitan. Some of us will never need to use Greek or Cyrillic letters, so why are you asking us to shoulder the cost of that?
I can go on, but I'll stop here, for now. Fighting words, indeed.
However, the text string posted didn’t render the angle brackets (U+2329, U+232A), I had to insert them via the glyph palette—why is that?
Konrad: the basic licence for Goodchild (Extended Latin) is ‘only’ $49.