In the office I have been banging my head for several weeks against cross-platform, cross-browser font issues. (They’re surprisingly tangible.) Unlike the first few pages of google hits for “browser platform font”, I don’t care about consistency of subjective size; that’s someone else’s problem. My task involves ensuring that a variety of non-alphabetic glyphs will
- display inline,
- tolerate zoom +/- (not indefinitely, but reasonably),
- be available regardless of which fonts one does or does not have installed locally,
- print cleanly,
- not confound assistive reading software.
As a result, I have skimmed several dozen discussions of font issues pitched towards end-users expected to know little or nothing about Web browsers. This isn’t terrible, in that I know less about browser internals than I’d like; it’s depressing in its implication that one might need to produce something like this how-to doc.
So, let’s not. What we’re currently pursuing is constrained by several factors:
- We settled collectively upon these glyphs to represent things extant in the print volumes we’ve published. Some of the choices are abstruse—why should one set of brackets come from the Unicode CJK area?—but they’re not random.
- We love serifs, for better or worse. (The nav / general site text is sans serif, as a point of difference.)
- Users should be able to understand 100% of the rendered result using an assistive reader, if needed.
- Ideally, users shouldn’t have to download a special font or otherwise fiddle with browser settings because some users will be turned off: we want to keep it simple, stupid. As it were.
Two sidenotes:
- Arial Unicode MS contains multitudes. It’s the only font I know containing every glyph we use. (Thumbs up. Everson Mono is in second place for that but looks odd against proportionally spaced text.) Arial Unicode is available only with Microsoft Office since 2002 or from Ascender, however, so thumbs down: not everyone has M$-Office.
- Special font sets created for use by classicists and medievalists—the realm I’m most familiar with—don’t tend to focus on special sorts. Thumbs down to Junicode, Cardo, Gentium… each of which would require a download for 95+% of first-time visitors, anyway.
Unless we generate PDF on the fly as a substitute for print.css, readers will need to turn on their browser’s version of “print background images,” lest they lose several important visual markers. (Oh well—or rather, I’m still chewing on that one.)
Early in March we settled upon Georgia, a font designed for screen reading and bundled with current/recent versions of Mac OS as well as Windows. It has serifs; its characters are relatively “fat” in aspect yet not so fat that it looks terrible in a printout. As a special bonus, unlike Times New Roman and Garamond, it includes the double-prime (″), reducing our special glyph tally by one.
That leaves us with nine glyphs and two spaces (en- and em-) to create as little inline images, in the manner currently employed by OED Online. (I wish I could think of an example that doesn’t require a site license or subscription.) OED Online uses image files for characters that couldn’t be represented by standard fonts as of their initial electronic offering. Though thorn (þ) and its capital form were already available, and though Unicode has caught up since, OED Online continues to prefer images:
<I><!--start_w-->Andreas<!--end_w--></I> 1536 (Gr.) <!--start_qt--><NOBR><IMG SRC="/graphics/parser/gifs/sp/Th.gif" ALT="{Th}" WIDTH=10 HEIGHT=14 ALIGN="absbottom" BORDER=0>ær</NOBR> wæs ælcum <NOBR><IMG SRC="/graphics/parser/gifs/sp/asg.gif" ALT="{asg}" WIDTH=7 HEIGHT=14 ALIGN="absbottom" BORDER=0>eno<IMG SRC="/graphics/parser/gifs/sp/asg.gif" ALT="{asg}" WIDTH=7 HEIGHT=14 ALIGN="absbottom" BORDER=0></NOBR> fram
(hence)
—which renders something like this when translated to Unicode points, if you have a Unicode-compliant font installed:
Andreas 1536 (Gr.) Þær wæs ælcum ȝenoȝ fram
I say “something like” because OED Online uses insular g where I’ve placed a yogh (ȝ). Insular g is in some half-uncial decorative fonts but isn’t part of Unicode 5.0, as far as I know, because it’s “only” a visual variant. (One can still see it on older Irish road signs—final letter on sign here.) You might think that a mere detail, easily discarded, but OED Online wanted to get things right concerning synchronicity: that g indicates something a bit different phonemically in Old English. Similarly, when we mean a double prime (″), we want one to appear, not straight (") or curved-right (”) quotation marks. “Something like” is not enough.
I need to stop adding to this post and release it. 😛 It may have more than one follow-up; adding something this morning lengthened it too much, and I’ll put up that segment in a few hours.
Sidelines:
- Adobe have a page explaining the respective backgrounds of PostScript Type 1, TrueType, and OpenType. (We have been camelCasing capitalization for more than two decades. whee.)
- By search hits, at least, people worried more about cross-browser, font-based display compatibility around 2001-3 than they do now. Has Flash wiped out some of that concern?