supecite vs. fill mode?

[Dan Jacobson]

Regarding emacs/22.0.50/lisp/mail/supercite.elc, I note whenever I
reply to somebody (in gnus), if the quoted stuff is more than one
line, and I add more than one line of reply, upon the
snapping of my line into a second line, I get a uninvited "M> " here added:

M> Even google.com isn't red/yellow free!  We're "less broken" than they are
M> (last I checked).  Who exactly makes dnsreport.com any sort of authority,
Firm words indicate you see beyond the traffic light colors. I can now
M> rest assured.

Which I have to clean up by hand. Or am I supposed to add a blank line
before replying? Hurmf. Here some sc line in my .gnus.el
just in case it's all my fault.

(setq sc-auto-fill-region-p nil)
(bbdb-initialize 'gnus 'message 'sc ); 'w3)
(setq sc-preferred-attribution-list
      '("sc-lastchoice" "x-attribution" "sc-consult"
        "initials" "firstname" "lastname"))
(setq sc-attrib-selection-list
      '(("sc-from-address" ((".*" . (bbdb/sc-consult-attr
                                     (sc-mail-field "sc-from-address")))))))
(setq sc-mail-glom-frame
      '((begin                        (setq sc-mail-headers-start (point)))
        ("^x-attribution:[ \t]+.*$"   (sc-mail-fetch-field t) nil t)
        ("^\\S +:.*$"                 (sc-mail-fetch-field) nil t)
        ("^$"                         (progn (bbdb/sc-default)
                                             (list 'abort '(step . 0))))
        ("^[ \t]+"                    (sc-mail-append-field))
        (sc-mail-warn-if-non-rfc822-p (sc-mail-error-in-mail-field))
        (end                          (setq sc-mail-headers-end (point)))))