Deleted several posts, as observations were weird symptom(s), not cause(s).

It looks like I was selecting different points on a curve, on different trials.
Maybe when selecting a closed curve, selected point/direction does not matter?

- Brian