Also, I showed that some dark channels see the signal as well.  An alternate route we could take is to make dark detector maps and use that to put an upper limit on how much B signal could come from it.  Calibrating the maps correctly would be a little tricky, but at least they wouldn't have any CMB confusing things.