You're seeing two separate phenomena here.

As SteveDG notes, inside a message body, the display of the characters is governed by your browser's character encoding: The particular alphabet represented by the current font settings. It's a genuine Tower of Babel problem--the simple old 128-character ASCII character set's all but disappeared, and now computers use Unicode, which allows switching between tens of thousands of different alphabets, each of which may represent a particular character code with a different symbol ("glyph").

Most of the time, autodetection works OK. When it doesn't, try Unicode UTF-8.

The other problem has to do with subject lines. A long time ago I tried to address the character encoding problem with a patch that substitutes "*" for characters above code 128 (below 128 the character sets are relatively consistent, but above that anything goes). I meant that to be a temporary fix...ha ha ha, you'd think I'd know better by now, huh?

The subject line problem is going to be difficult to fix because I did the patch rather stupidly: Hardwired into some compiled code, rather than a script where I can change it easily. However, the message database stores the raw text you submit, including those fancy characters, which means that I can resurrect them once I remove the patch. Someday.

Sorry for the inconvenience this causes.