The peculiar gains on 3C286 suggest that perhaps the model was not in fact used. Why don't you run UVPLT on 3C286 with DOCAL true to see if you get a wierd looking point source visbility or the proper heavily-resolved source visibility. Eric Greisen