It looks like ICU doesn't assign a script for "grc" language (Ancient Greek). For example: char maximizedLocale[ULOC_FULLNAME_CAPACITY]; uloc_addLikelySubtags("grc", maximizedLocale, sizeof(maximizedLocale), &status); // now maximizedLocale is "grc" char script[ULOC_SCRIPT_CAPACITY]; uloc_getScript(maximizedLocale, script, sizeof(script), &status); // returns 0, and now script is the empty string. status is success. Since "grc" is not mapped to a script, WebKit will use the font for the Common script, which might not be appropriate, especially for CJK fonts (see bug 88545). We should check whether it's intentional that ICU/CLDR doesn't assign a script to "grc". Curiously, in icu/source/data/misc/supplementalData.txt, there is: grc{ secondary{ scripts{ "Cprt", "Grek", "Linb", } } } So it's possible ICU can return these scripts somehow (though we'd then have to choose between them). Wikipedia uses lang="grc" in articles like <http://en.wikipedia.org/wiki/Pankration> FWIW, Firefox has the same behavior: lang="grc" content isn't rendered using their Greek font setting
Is a modern Greek font optimal for rendering Ancient Greek?