I found it: RTF is damn archaic!
It works with Unicode but internally convert it in legacy Windows format based on code pages (as of Windows 95 and earlier), 8 bits per characters, but the fragment of text is prefixed with the Windows code. I saw this encoding when I types Unicode characters (code points above FF) in WordPad. If also supports Unicode escapes like
\u1234?
. The question marks denoted fallback for the programs not supporting Unicode ("show ?"). Also RTF does not support code points beyond BOM in any way (by all new versions of Windows do support it via surrogate pairs; as far as I remember — since some service pack of Windows 2000).
You can simply try type what you need in WordPad, save the file and open it with the plain-text editor, to make sure your understand how it should be coded in all details. I just did it.
See
http://en.wikipedia.org/wiki/Rich_Text_Format[
^].
Why using such archaic legacy stuff? What's wrong with HTML? PDF, to the worst end?
—SA