In the first field, the Garmin was reading from the cab and the EZG+ was reading from the outside edge of a 25' implement and therefore should've given a larger area than the Garmin but gave an area that was 10% low (half an acre) if my six samples were to be believed (4 were within a 20th of an acre).  After that first trip around I walked the edge of the field with the Garmin for the extra samples and continued to use the outside edge of the implement for the next couple of EZG+ samples.  Implement width was set correctly, as was which edge to use (I triplechecked).

 In the second field where the EZG was WAAAY out, the outside implement edge was used and I walked the field edge with the Garmin.  The EZG had been powered on for a long time, so WAAS corrrections should've been good.  In both instances the EZG didn't lose a lock or flash the "Few Satellites" warning that I get sometimes on treed edges.