This article is from the Japanese FAQ, by TANAKA Tomoyuki tanaka@cs.indiana.edu and Olaf Meeuwissen olaf@IMSL.shinshu-u.ac.jp with numerous contributions by others.
Before anything else, remember that there is e-mail software out
there that is not 8-bit clean. Next, don't forget that you can
never be sure what route your e-mail takes from you to the
addressee, nor that it will always take the same route. That
means that your message may meet e-mail software that is not
8-bit clean. The only fail-safe way around this is to send your
Japanese message in a 7-bit encoding, i.e. JIS, iso-2022-jp or
iso-2022-jp-2. Shift-JIS and EUC-JP are 8-bit based encodings
and may get mutilated on the way.
Unfortunately, some mail software is a bit over-zealous and also
strips the escape character indicating begin and end of encoding
changes. Ken Lunde, wrote a utility called `jconv' that can,
among other things, put stripped escapes back in. The source is
at <ftp://ftp.ora.com/pub/examples/nutshell/ujip/src/jconv.c> and
compiles without any problem. This same program can also detect
the encoding used and convert between encodings if desired.
If you really have to send 8-bit based encoded e-mail and it does
get mangled, you can try sending it uuencoded. The receiving end
will have to uudecode before being able to read anything. With
`jconv' there should not be any real need for this clumsy
approach.
 
Continue to: