Okay! The culprits are (drum roll, please)....

A changed boundary marker (which does not affect the message signature), a
two-line header moved onto one line, and a deleted header (presumably because
the default transfer encoding is 7bit.)

Now, one might argue that a better design for MIME would have been to sign only
the message and not the headers, but that would open the way to all sorts of
devious attacks involving changing the header, and defeat the entire purpose of
the S/MIME standard. So there we have it.

If I get some time in the near future, I'll look into the source and see if I
can't find a simple way to keep Courier from rewriting the headers.

The details:

------=_NextPart_000_0009_01C05A2D.921B1350 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit

Will it verify? I hope so! ------=_NextPart_000_0009_01C05A2D.921B1350

(as received by QMail)

versus

(as received by Courier)

--=_29106-975525440-1-2 Content-Type: text/plain; charset="iso-8859-1"

Will it verify? I hope so! --=_29106-975525440-1-2