ATOMIC HYDROGEN 105 



weight in the filament showed that one molecule of nitrogen disappeared 

 for every atom of tungsten that evaporated. A brown compound, WNo, 

 was formed which deposited on the bulb and decomposed when water vapor 

 was introduced, forming ammonia gas. 



From time to time the question kept arising — how good would a lamp 

 be if it had a perfect vacuum? And now, from studies of the character I 

 have described, I began to have an answer. Hydrogen, oxygen, nitrogen, 

 carbon monoxide, and in fact every gas that I introduced, with the ex- 

 ception of water vapor, did not produce blackening of the lamp bulb. The 

 serious blackening that occurred with only small amounts of water vapor 

 depended upon a cyclic reaction in which atomic hydrogen played an 

 essential part. The water-vapor molecules coming in contact with the hot 

 filament produce a volatile oxide of tungsten, and the hydrogen is liberated 

 in atomic form. The volatile oxide deposits on the bulb where it is reduced 

 to the metallic state by the atomic hydrogen, while the water vapor pro- 

 duced returns to the filament and causes the action to be repeated in- 

 definitely. Thus, a minute quantity of water vapor may cause a relatively 

 enormous amount of tungsten to be carried to the bulb. 



The question then arose whether the traces of water vapor, which might 

 still exist in a well-exhausted lamp, were responsible for the blackening 

 which limited the life or the efficiency of many of these lamps. We made 

 some tests in which well-made lamps were kept completely immersed in 

 liquid air during their life, so that there could be no possibility of water 

 vapor coming in contact with the filament. The rate of blackening, however, 

 was exactly the same as if no liquid air had been used. 



Having thus proved that the blackening of a well-made lamp was due 

 solely to evaporation, I could conclude with certainty that the life of the 

 lamp would not be appreciably improved even if we could produce a perfect 

 vacuum. 



Early in 191 1 William Stanley, one of the pioneers in the electrical in- 

 dustry, felt that our company should do more fundamental work in con- 

 nection with heating devices. Since I had become interested in the theorv 

 of heat losses from filaments in gases, I was glad to work along these lines, 

 so I undertook to direct a small laboratory at Pittsfield, Mass., at which 

 I spent about two days a week. Besides studying the heat losses from plane 

 surfaces at various temperatures, I measured the heat losses from wires 

 of various sizes in air at different temperatures, working at first with 

 platinum wires, and was able to develop a theory of the heat losses which 

 enabled me to calculate the loss from a wire of any size at anv temperature 

 in any gas, assuming, however, that the gas did not dissociate at high 

 temperatures. 



Having now a definite theoretical basis on which to calculate the normal 



