More notes I do not know if they used a pump, or simply let the water flow from the tap. I have used both methods at various times, and so has Dennis Cravens, although not for such a large flow rate. They said they checked the flow rate "several times" which I assume means it was measured manually, with a bucket and stop watch.
You might think that the flow rate would fluctuate significantly over 18 hours, but in my experience, using either tap water pressure or something like a 700 gallon/hour (gph) pond pump, the flow rate is quite stable over many hours. With a pond pump, you can use a small plastic throttle to set a lower flow rate. It stays constant longer than you might think. I have tested this out of curiosity. With an actual outdoor pond, it will change gradually over a week, as gunk accumulates in the pump. I do not have a 700 gph pond pump. I have a 170 gph circulation pump, and also a 1/3 HP sump pump that I think is rated 25 gpm, about twice the flow of the Rossi test. As far as I can see, the only likely error with this setup would be measuring the temperature too close the energy source within the gadget. Based on the photo, McKubre thought the outlet thermocouple was too close to the likely source of energy. As I mentioned, the NRL 10 kW test bed system has much better arrangement of temperature sensors and flow meters. However, with input power of only ~80 W and a flow rate of 833 ml/s, without excess heat the temperature difference would be 0.02°C. I doubt you could detect that with this arrangement. The difference between 0.02°C and ~5°C is gigantic. - Jed
