HYDROGEN INTO ATOMS 153 



pure nitrogen at various pressures, to compare the heat loss in this gas with 

 that in hydrogen. 



The wire used in all lamps was 0.00706 cm. diameter. In every case it 

 was welded to nickel leads with tungsten or molybdenum tips. The length 

 of wire used in the hydrogen lamps ranged from 5 to 9 cm. and in the 

 vacuum and nitrogen lamps, from 6 to 12 cm. It was necessary to use 

 rather short lengths in hydrogen in order to be able to heat the wire to the 

 melting point without using voltages over 200 volts. 



Temperature Measurements. — The temperatures were determined, as 

 has been usual in this laboratory, by the relation 



7.029 — log H ^ ^ 



where H is the intrinsic brilliancy of the filament in international candle 

 power per sq. cm. of projected area. On this scale the melting point of 

 tungsten proves to be 3540° K. (by direct experiment). This value we 

 consider to be more probable than the lower values usually given. 



The temperatures were also determined by two auxiliary methods : 

 by measurement of the resistance and by matching the color of the emitted 

 light against that of a standard provided with a blue screen. In both these 

 methods, however, the primary standard of temperature was the one given 

 above, based on the intrinsic brilliancy of the filament. 



It was found in nearly all experiments that the three methods gave 

 concordant results. The presence of hydrogen or nitrogen did not change 

 either the resistance or the color of the light emitted from a filament set 

 up at a given intrinsic brilliancy. Discordant results were obtained only 

 in the experiments in vacuum or low pressures of gas after the filament 

 had been heated some time above 3200° K. and had evaporated so that the 

 bulb had blackened and the diameter of wire had changed. 



In order to avoid errors in temperature due to the cooling effect of the 

 leads, the candle power was usually determined through a slit, one or two 

 centimeters wide, placed horizontally in front of the lamp. In this way, 

 the intrinsic brilliancy can be determined with accuracy. 



At temperatures below 1800° K., the candle power determinations were 

 too inaccurate to be suitable for temperature measurements. The tem- 

 peratures were, therefore, found in these cases from the resistance, proper 

 correction being made for the cooling effects of the leads. 



Before making measurements on the lamps, they were thoroughly aged 

 by running them for 24 hours with the filaments at a temperature of about 

 2400° K. With the hydrogen and nitrogen of the purity used in these 

 experiments, this treatment produced no perceptible injury of the filament. 

 During the first few hours of aging, the filament, even in vacuum lamps, 



