200 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1924 



tions that commercial lamps are so made that they burn at normal 

 voltages on proper circuits an average of 1,000 hours. If we were 

 satisfied with a shorter life we could have a more efficient lamp, but 

 experience has shown that we would be unwilling to use the short- 

 life lamps in order to secure the added efficiency. To increase the 

 efficiency without shortening the life of lamps much study has been 

 made of distillation of metals in vacua, and of methods for return- 

 ing the distilled metal to the filament. Much has also been done 

 to make the deposit on the glass invisible or white, so as not to 

 interfere with light transmission. Naturally we are always on the 

 lookout for metals, such as the newly discovered hafnium, which 

 might possibly live longer as a filament than tungsten now lives. 

 Thirty or forty years of research work had been spent on high- 

 vacuum incandescent lamps before Doctor Langmuir showed us 

 how to make still better lamps by putting back into the vacuum 

 gases like argon and nitrogen. 



DISTILLATION 



When the material of a filament distills in vacuum it does not meet 

 interference to the motion of its molecules, and the distilling sub- 

 stance proceeds in straight lines from the heated source. This is 

 often observed when an incandescent vacuum lamp arcs or burns 

 out. Metal shadows of interior parts of the lamp are then often 

 cast onto the walls. This is shown more clearly when a metal like 

 gold is evaporated from the surface of a tungsten filament in vacuum. 

 By the interposing of a design, as, for instance, the star in the left- 

 hand bulb of Plate 1, Figure 1, a shadow in gold is cast on the 

 glass as shown on the right-hand bulb of Plate 1, Figure 1. This 

 simple phenomenon is mentioned because it fits in with the kinetic 

 theory of gases and explains many things observed in vacua. The 

 " mean free path " of molecules or atoms is very long in good vacua, 

 and so straight-line distillation occurs. 



IIITTORF BULBS 



In the historic Hittorf experiment two vacuum bulbs, each carry- 

 ing an electrode, were joined together by two glass tubes, one very 

 .short and one exceedingl}^ long, as shown in Plate 1, Figure 2. 

 When electric current was passed from one bulb to the other, it 

 evidently chose the longer instead of the shorter path, because the 

 longer tubes became highly luminous, while the shorter one did 

 not. This is a quality of electrical conduction in vacua where small 

 quantities of gases still remain, but it can not be gone into here. 



