Thanks quantum! This is exactly what happened. How necessary would you say are jumpers when PI testing transformers? I've always been taught to use them but it obviously effects the reading - maybe even in free air?
I had this issue before when the Megger battery was low.
I charged and I retested and was fine.
Another time the weather just changed during the 10 minutes test from cloudy to rain. I retested in same conditions (rain).
If you think about why a PI less than 1 is failing, it makes more sense. That would mean that the insulation is breaking down under test. The reason it isn't as relevant at such high readings, is that the measurement is extremely close to 0 current flow. Any tiny change in conditions could cause a massive change in results. The insulated leads touch, test set gets moved, air moves around the test area; all of those things will cause an extremely small change in current draw, thus decreasing the measured insulation resistance.
Jumpers aren't necessary when performing insulation resistance on a transformer. The windings are interconnected, and thus all energized. See the attached diagrams of the windings for a Delta Wye transformer. If you were to energize 10kVDC H1 with the positive lead without the connections tied together, you're still going to have 10kVDC on H2 and H3.