156 PHENOMENA, ATOMS, AND MOLECULES 



in the effects in the two gases. In nitrogen, the values decrease steadily 

 as the pressure is reduced, and in hydrogen, at the lower temperatures, the 

 same decrease is observed. But at higher temperatures, in hydrogen, the 

 power consumption is considerably greater at lower pressures than at 

 atmospheric pressure. 



The conclusions drawn from these measurements will be considered 

 later, together with the results obtained at still lower pressures. 



Lower Pressures. — In these experiments, measurements of the power 

 consumption were made at a series of temperatures, with pressures ranging 

 from o.oi mm. up to 13 mm. The measurements were made while the 

 lamps containing the filaments were connected to a vacuum system consist- 

 ing of two McLeod gages (one for lower and one for higher pressures) 

 and a Gaede mercury pump. The lamps were first exhausted at 360° for 

 an hour and a trap immersed in liquid air placed directly below the lamps 

 condensed the moisture given off from the glass. 



The candle power was measured during the experiments by a portable 

 Weber photometer. The characteristics in vacuum at temperatures ranging 

 from 1500 to 2500° K. were first determined and then a few centimeters 

 (pressure) of pure hydrogen was admitted to the system and pumped out 

 to a pressure of 13 mm. A series of measurements of current, voltage and 

 candle power was again made. In this way, in the first experiment, the 

 characteristics were measured at pressures of 13. i, 6.5, 2.5, 1.2, 0.53, 

 0.218, 0.102 and 0.045, 0.020 and o.oio mm. and at o.oooi mm. In a second 

 experiment, measurements at a few pressures were made as a check on the 

 preceding. 



In the third experiment, made a few weeks later, a much more complete 

 series of measurements was made, extending the range of observations 

 from 1100° to 2900° K. Higher temperatures were avoided, in order not 

 to cause any evaporation of the filament. The pressures employed in 

 this set of tests were 4.4, i.io, 0.207, 0.039, O-OiS- ^^'^ 0.00007 "i"!- 

 At each pressure about 30 readings at different temperatures were 

 taken. 



The results of the third experiment were in excellent agreement with 

 those of the first two. Since the former covered so much wider a range, 

 only these results are given in this paper. 



The tungsten wire used for these experiments was taken from the same 

 spool as that used in the experiments at higher pressures. 



The filament was mounted horizontally in a straight piece 9.85 cm. long. 

 Because of the cooling effect of the leads, it was assumed that the effective 

 length of the wire for calculation of the watts per centimeter was 9.4 cm. 

 The bulb was a cylindrical bulb with rounded ends, about 10 cm. diameter 

 and 18 cm. long, and the filament was- supported in its axis. 



