While I agree that the use of w-hr/hr is dumb, I have to argue that it is not (quite) nonsensical. Presumably input power is electrical, and measured by an integrating power meter with an output selectable between w-hr and kw-hr, with readings taken every hour. In this case, an increment of 0.5 between hourly readings would be expressible as 0.5 w-hr/hr. It's not what I'd call smart, but it's about what I'd expect from a guy with a master's in philosophy who's trying to pass himself off as a scientist/engineer. Of course, he then has to bugger his thermal and optical measurements to convert them to the same units, but this is certainly feasible (although full of possible errors). He quite casually discusses converting lumens to watts here http://www.e-catworld.com/2016/06/15/e- ... mpilation/ Of course, this implies a fairly detailed second-by-second monitoring to the Quark-X's power levels in order to convert a supposed optical measurement to effective power, which would require the continuous use of a wide-band spectrometer, but this sort of inconsistency has never stopped Rossi before.
Of course, Rossi himself clearly indicates that this generous appraisal is wrong. From the same compilation we see
What was the longest period of self sustain in which the output remained steady or increased with zero input power? zero seconds. Always fed half Wh/h
which clearly implies that the input power was fixed at a constant 1/2 watt. Oh well.