In their original set-up, Fleischmann and Pons used a Dewar flask (a double-walled vacuum flask) for the electrolysis, so that heat conduction would be minimal on the side and the bottom of the cell (only 5 % of the heat loss in this experiment). The cell flask was then submerged in a bath maintained at constant temperature to eliminate the effect of external heat sources. They used an open cell, thus allowing the gaseous deuterium and oxygen resulting from the electrolysis reaction to leave the cell (with some heat too). It was necessary to replenish the cell with heavy water at regular intervals. The cell did not contain a stirring apparatus, though the authors claimed that the bubbling action of the gas kept the electrolyte well mixed and of a uniform temperature.
The cell was also instrumented with a thermistor to measure the temperature of the electrolyte, and an electrical heater to generate pulses of heat and calibrate the heat loss due to the gas outlet. After calibration, it was possible to compute the heat generated by the reaction.
A constant current was applied to the cell continuously for many weeks, and heavy water was added as necessary. For most of the time, the power input to the cell was equal to the power that went out of the cell within measuring accuracy, and the cell temperature was stable at around 30 °C. But then, at some point (and in some of the experiments), the temperature rose suddenly to about 50 °C without changes in the input power, for durations of 2 days or more. The generated power was calculated to be about 20 times the input power during the power bursts. Eventually the power bursts in any one cell would no longer occur and the cell was turned off.