help with javascript + embedding + japanese characters

Hopefully someone here can help me get my head around an issue I'm
having trying to internationalize our content for Japanese.  The core
of the issue is that we have web content that is a mix of JavaScript
and HTML.  We have an external javascript file that holds a simple
variable definition which is the string representation of the install
path.  We need to be able to represent non-ASCII characters in this
file (we're running into an issue with Japenese).

What is it that determines what text/character encoding should be used
in this file?  I'd rather not rely on any heuristic/interpretive
algorithm that guesses at the character encoding.  Everything seems to
work if the characters are escaped UTF-8, but I'd like to understand
why.  Does the charset of the HTML document that essentially "includes"
the external .js file determine how the bytes are interpreted?  Is
there some standard that I can't seem to find documented?

Please note that I'm a C developer who is handling the embedding of
mozilla, not a web/HTML/Javascript coder nor the one who has written
the installer script (which is in Perl on the mac platform and a
compiled C++ program on windows...ugh), but I have enough knoweldge of
them to discuss these elements.

Any help/insight is greatly appreciated.

J