I don't have any opinion on whether this thing works or not (other than the usual "this is an extraordinary claim, the evidence is not extraordinary"), but the test apparatus is pretty well-described in the paper. So, y'know, if you think it works, build one. I am certainly no expert in this but I can imagine that calorimetry with short 300kW power transients may be prone to measurement error. Having your prototype tested and effect confirmed by a skeptical third party with expertise in this kind of thing (say, a university physics or engineering department) would go a long way towards showing you aren't fooling yourself somehow. -n