I have a mystery, whereby an overland survey between two fixed points gives a correct pdf plan output with; loop-closure survex (I know it is 'correct' because the aerial photograph image matches the features surveyed) but if I specify; loop-closure therion then one of the fixed points is about 150m misplaced in plan and the surveyed features are clearly in the wrong place. In both scenarios therion.log reports identical average loop errors and specific loop error statistics. The error over a 500m loop is about 2.3%, mainly due to a 12m error in elevation (can not trust those gps readings). The plan error if one of the points has its 'fixity' removed is about 4m, and the plots produced are correct regardless of the loop closure method. It is a rather large dataset with many underground loops, and four fixed surface points. Two of which have overland surveys between them. Only one of the overland surveys has this problem. Using Therion 5.3.5. Can replicate this result on three computers. Any theories as to what might be going on and how to locate the issue? Bruce