I have a 1850 KVA single-phase, ungrounded transformer without ground-detection installed. It has a 4000 Amp resistive load attached and uses a saturable core reactor to control output current. It feeds two large electrodes that pass through a grounded metallic plate. The voltage is fairly low 300-800 VAC. There are occasions when I see significant arcing from one electrode to the metallic plate, never both electrodes at once. I am pretending like I know that there are no other direct conductive paths to the "non-arcing" electrode. Is it concievable that there might be some inductive effect that results in this arcing. The air gap is at least an inch, the penetration is round, the amps =3000 and the volts ~ 600VAC.