GFCI question

Possibly the leakage is happening through some material that isn't exactly an ohmic connection, but rather breaks down under comparatively high voltage—such as a minute air gap. It might be interesting in a scientific investigation sort of way to see if the voltage at the ground lead with respect to either input changes as the element heats up (obviously with the ground disconnected).

I guess it's somewhat of a moot point, really, since the element is plainly defective and needs replacement, regardless of what you can measure with your Fluke. It doesn't take much current leakage to trip a GFCI.