my first reaction to the Q would be how many variables exist, proper grounding of the heater, non-FPE breakers, number of conductable paths, as well as conductables.
would it be riske' to assume that normally there would be a spike of power, until an OCPD reacts, that would include more umph* within the unit, and decreasing umph as the distance from it increases?
*technically speaking.....
