You can intercept the font substitution process by throwing out the catchall font using some equivalent font-family: userChoice, yourCatchAll , where yourCatchAll is a font that has a common character for all characters.
The problem is finding such a font. The LastResort font distributed by the Unicode Consortium would be ideal, as it also visually indicates the character’s category in broad terms, but its EULA does not seem to allow modification. It is debatable whether this relates to building web font formats (e.g. .eot and .woff).
The Unicode BMP Fallback Font seems to have more liberal usage rules, but it displays the character just like its Unicode number in the field (and only supports the base multilingual plane, although it contains all the characters that most people have ever heard of).
Jukka K. Korpela
source share