559 
1909-10.] The Efficiency of Metallic Filament Lamps. 
.The thermopiles employed were of the ordinary Rubens linear type. 
Their resistances were about four ohms each. The telescope of the spectro- 
meter was carefully wrapped round with cotton wadding. The field-glass 
of the eyepiece was blackened except for one small opening by means of 
which the thermopile could be adjusted on the sodium line. The instrument 
was calibrated partly by bringing the thermopile into superposition with 
lines in the visible spectrum and partly by plotting the absorption curve 
of a very thin water film and using the maxima, which have been deter- 
mined by Aschkinass * to be at 1 # 500 /m and T956 /m, and the minimum 
between them at l - 708 /x. 
The galvanometer was a Du Bois-Rubens iron-clad one. Its total 
resistance was 105 ohms, and when the light suspension system was used 
the sensitiveness was 3 10 -10 amps, per half mm. at one metre distance 
for a period of 5 secs. For greater sensitiveness the field was asymmetric. 
This sensitiveness was not all required. When measuring such small 
currents, of course, the keys and binding screws in the circuit must be 
made of copper alone ; otherwise there are always thermo-electric forces 
at the junctions. The instrument was found impervious to magnetic 
disturbance. But it was not slung in a vibrationless suspension, as the 
makers recommend, and whenever there was a strong south-west wind, 
observations had to be discontinued owing to the vibration of the 
laboratory. 
Current was supplied from the laboratory secondary battery. This 
gave voltages up to 260. For the readings at 270 volts the lighting circuit 
was used in series with fifteen accumulators, and the voltage kept constant 
by varying a resistance in the circuit. In order to measure the e.m.f. 
between the terminals of the glow-lamp, the latter was shunted by means 
of a Kelvin centi-ampere balance and a resistance, the total resistance of 
the shunt being 3226-8 ohms. By measuring the current through the 
shunt, the voltage between the terminals of the lamp was obtained. The 
total current through the lamp and shunt was obtained from an ammeter, 
and the current through the lamp obtained by subtracting the current 
through the shunt. The current balance was one of a set of three belonging 
to the laboratory, which had been tested by copper electrolysis some years 
previously, and as they all agree yet within the error of observation, it 
could be assumed to be perfectly correct. The ammeter was checked by 
the current balances. 
* E. Aschkinass, “ fiber das Absorptionsspectrum des fhissigen Wassers und liber 
die Durchlassigkeit der Augenmedien fur rothe und ultrarothe Strahlen,” Wied. Ann., lv. 
1895, p. 401. 
