I don't know much about Windows, and nothing about how its font mechanism works

Windows (NT/2K/XP, anyway) uses UCS2 (or was it UTF16?) for output. For legacy applications, it has code pages, mapping 8-bit character references into UTF16 (or was it UCS2?).

Not all fonts have all of (or even useful bits of) the code points defined by Unicode in them.

Problems with Java and fonts are probably down to the JVM using the 8-bit font stuff, rather than the 16-bit font stuff.

Peter knows more about Unicode under Windows than I do -- he did most of the Unicode support in Rio Music Manager.