G98 TD/CIS Opposite sign of the components of length- and velocity dipole moments?

Dear CCL,
 I have some questions regarding the output os CIS and TD jobs in G98
 (Rev. A7). I can't find anything on these issues in Gaussian
 documentation or past CCL posts, although the questions are rather
 simple. So I hope it makes sense...
 The output of TD and CIS calculations give the x, y, and z components of
 the "Ground to excited state Transition electric dipole moments",
 "Ground to excited state transition velocity dipole Moments", and the
 "Ground to excited state transition magnetic dipole Moments".
 1) Why are the components of the electric and velocity derived
 transition dipole moments oppositely signed? I thought that they should
 be identical in a complete basis. I would have thought that the sign
 ought to be the same. Which values should I trust?
 2) Is it possible to say which of the computed oscillator strengths
 (dipole-length or dipole-velocity derived) provide the best estimate of
 experimental absorption intensities?
 3) I suppose that the computed electric and magnetic dipole moments are
 origin dependent. But are the magnetic transition dipole moments derived
 in a manner which make them related in to the transition velocity dipole
 moments, rather than the transition electric dipole moments? (e.g. in
 terms of origin dependence etc.)
 Any help or comments will be greatly appreciated!
 Peter W. Thulstrup
 Chemistry Department / Royal Veterinary and Agricultural University
 Thorvaldsensvej 40 / DK-1871 Frederiksberg C / Denmark
 Phone: +45 35282464 / Fax: +45 35282398 / E-mail: pwt ( ( at ) ) kvl.dk