[daip] Polarization mystery

[egreisen]

A VLBA user has done a data reduction and then imaged his
polarization calibrator (BL Lac).  The results are point
sources, with the Ipol at the phase center and Q and U
offset about 20 mas south.  Can anyone think of a way to
phase shift the crosshands without phase shifting the
parallel?  I attach his image.

Note he used LPCAL with 2 regions and states that something
like this has happened to him before on other sources.  The full
correspondence is in the AIPS Helpdesk desk.

Thanks

Eric