Technical Summary: Polar plumes have been shown to have primarily unipolar footpoints (DeForest et al, in press for Solar Physics, 1997); however, the weak field around polar plume footpoints shows that most (and perhaps all) plumes have a weak, opposing-polarity region in the vicinity of the main footpoint. This fact, along with the recent magnetic convection work of Hagenaar et al and Title et al, lends credence to the idea that magnetic reconnection may be in part responsible for the plumes' presence.
Consistent with this picture, recent work by Lemy et al using LASCO data (and presented at SoHO V in Oslo) hints that plumes may be episodic in nature: plumes seem to appear and disappear from the LASCO field of view on timescales of less than one day; but subsequent appearances seem to occur in the same location (in the rest frame of the plume's footpoint).
The natural hypothesis is that plumes have persistent magnetic structures (a la Newkirk and Harvey, 1968; and Suess et al, 1997) which are filled with plasma by reconnection-driven heating when (and only when) an opposing footpoint is forced into the proto-plume's footpoint by photospheric convection.