As you already observed, Unicode support was added in more recent versions, which allows to use pretty much any character. Memory is hazy on this, but I think we added Unicode support for v15.
Older versions of ME10 relied on the HP Roman8 and Shift-JIS encodings (Western European languages and Japanese, respectively). See
http://demo.icu-project.org/icu-bin/convexp?conv=hp-roman8 for an encoding table for HP Roman8.