Eric- Re weights: Yes, for data which AIPS FILLM does the baseline flip, the weights from FILLM and CASA agree _before_ accounting for the RL/LR flip. So, the correlation weights are reversed, too, as you surmise. -George