1081 



The qiiantitj of salt is thus chosen that the chlorine pressure does 

 not become eitiier too high or too low. Not too l)igh for the above 

 mentioned reasons; not too low as then the action is insuffi- 

 cient. We have been able to state that K,T1 C!!, .develops chlorine 

 at 350° C. already. A small quantity of the salt was introduced be- 

 tween the 2 glass walls of a Dewar-glass tilled with liquid air, the 

 external wall of which was heated to 350°. The distillate formed 

 in the high vacuum and precipitated on the cold inner wall contained 

 rathei' much free chlorine and further thalliiim chloride; potassium 

 chloride was not found in it. 



The good results reached with this salt in the lan)p may there- 

 fore doubtlessly be ascribed to the continual presence of a chlorine 

 atmosphere of very low pressure, by which the tungsten precipitate 

 is conveited i?ito WClg, which is only very weakly coloured. A 

 similar action can be stated of copper- and silver-chloride and also 

 of some chlorates and oxides that are easily decomposed as e. g. 

 barium chlorate, barium peroxide, higher lead oxides. 



^ 2. Actio7i of solid substances that are not easily dissociated. 



The action of the above mentioned substances was easily explained 

 by their dissociation. Now H. H. Needham has found, however, that 

 also a substance as cryolite when applied to parts of the lamp that 

 have a temperature between 400° and 800° gives good results, while 

 from other sides the application of small quantities of sodium chloride 

 to the glowing body is recommended. In fact such lamps with cryolite 

 or sodinm chloride prove to blacken less quickly than ordinary vacuum 

 lamps. Needham supposes the cryolite to be decomposed, so that in 

 the lamp a "certain" halogen atmosphere is formed to which the 

 obtained result would be due. As the occuirence of such a halogen 

 atmosphere seemed improbable to us, we subjected this phenomenon to 

 a closer investigation and have made several experiments which 

 might give us a better insight into the action of these substances. 

 First we investigated, whether sodium chloridecould act in the same way 

 as K/nCl,. When introduced in a small tube into the lamp, the 

 NaCl proved however to be perfectly inactive, as might be expected. 

 When sprayed however on the wire and the supports it was 

 very active. When only the supports were sprayed the action 

 was again rather small. The salt on the filament must therefore 

 be the active part. As soon as the lamp glows, however, it is vapo- 

 rized and condensed on the glass wall. This can easily be seen by 



70* 



