There is a technique that Celani can apply to his test system that will reduce the effects of wire resistance changes to a remarkable degree. Some question has arisen as to how much of the measured excess power output variation is due to changes in heating power from the source as the resistance of the wire varies with loading, etc. These variations can be reduced by a factor of almost 40 to 1 for a 10% change in wire series resistance by adding a series resistor to his input drive voltage supply equal to the nominal value of the driven wire.
The RF design world has used impedance matching for many years for a similar purpose and I propose that Celani do the same. Of course the penalty is that he would be required to deliver two times the power from his supply as before and also find a resistor network that can handle the 48 watts of power dissipation with a continuous rating. In my opinion this would be a small price to pay if we then would know that the drive power variable would essentially vanish from consideration. I will be happy to assist in selecting a resistive network and more powerful supply if requested. An efficient switch mode power supply operating at this power level would not be difficult to obtain. Dave