When the filament of a thermionic valve emits and a thermionic current flows from the filament to the anode and through an external circuit back to the filament, the filament current is necessarily different at all points along the filament.An ammeter in each filament lead and a voltmeter across them will all read differently when the valve emits, the actual deviations depending on the electronic emission and the circuit used. The only exceptions are the case of a filament connected directly to a source of P.D. with no intermediate rheostat, when the voltage will be maintained constant by the supply, but the currents will vary in any case. The other exception is that of the valve with the separately heated cathode, the potential of which is uniform.The variations of the reading of the filament voltmeter and ammeters have been investigated experimentally, and a circuit has been devised in which there is no change of reading of the voltmeter or either ammeter when the anode current is switched on and off.The current read is the filament battery current, and with this circuit tests may be repeated on valves with the certainty of the filament conditions being constant for different tests.An investigation was also made of the validity of the practice of readjusting the filament voltage after switching on the anode potential, and it was found that this practice is not correct as regards maintaining the filament under constant conditions, except for those valves intended to be run without a rheostat.All tests were made with valves used as diodes, as in this case the emission is usually greater than the anode current when the valve is used as a triode.For standard tests on valves used as triodes the precautions necessary to maintain constant filament conditions are of less importance.For standard valve tests the circuit devised should be adhered to, and the practical means of ensuring constant conditions are indicated.The method of reducing the asymmetry of the temperature gradient of large valves used as rectifiers is of some interest here, and the conditions have been calculated.The experimental work necessary shows how fine precision measurements may be made with commercial instruments, as numerous checks are possible, most of which show up the accuracy of the majority of the readings, which in most cases involved a change of less than one scale division.Finally, a valve was tested with its filament supplied with A.C., and various circuits tried. It was found immaterial to which point of the filament circuit the anode return was connected, on account of the symmetry of the temperature gradient, at any rate at 50 similar and above.