I am recalling Dr. Parkhomov's initial runs in which he devised a simple calorimeter by measuring the amount of water that had boiled off. His tests were limited, in part, by his ability to replenish water at odd hours. Let's hope there isn't a whole lot of difference between that simple measurement and input / output data collected for Mr. Rossi's test. It would be best if power input, water input, steam output, and conductive / convective losses all add up.
Phase change calorimetry is simple and robust, but liable to abuse unless done sensibly. If the steam is not vapour, with the heat of condensation definitely absorbed out of the reaction vessel, it will over-read energy by a very large amount. there are setups that make this easily checkable, and others like many of Rossi's early tests that don't.