Character set problem with data being sent through SQL(?)

To begin with, this is only my problem because it's a customer's problem, and it's affecting our application.

They're sending data into a PF through what APPEARS to be some sort of SQL connection, maybe ODBC or JDBC (or it COULD be something else). The original on their end is in UTF-8. If they had the connection configured right, it would of course be translated to EBCDIC, all under the covers.

But it's acting as if their connection THINKS they're sending 8859 Latin 1. With the result that "lowercase e with acute accent," which is hex "C3 A9" in UTF-8, ends up in the file as hex "23 62 B4," which is <highlight reverse>, "Capital A with tilde," "copyright symbol."

Any idea what could be configured wrong here?

--
JHHL