1083 



The action could however be iiiHiienced ') neither bv the applica- 

 tion of an accelerating or a retarding field nor bj means of a solenoid. 

 Moreover it is not probable that the number of electrons with velo- 

 cities above 1.23 >; 10^ cm sec. is sufficient. When no accelerating 

 or retarding electric fields existed in the lamp, there would be at 

 the normal glowing temperature of 2350° abs. oidy one such an 

 electron to every 250 vaporized molecules. In reality the electron 

 current is much smaller than the satuiation current, so that doubt- 

 lessly the number of high velocity electrons will be still smaller. 

 A decomposition of the salt by the impact of electrons is therefore 

 very improbable. Bnt moreover we could prove directly, by a 

 chemical method, that in the lamp no decomposition of the NaCl 

 takes place. 



A lamp with a side-tube was constructed. During the burning 

 the side-tube was immersed in liquid air. After the experiment no 

 trace of a chlorine condensate was found to be present. A drop of 

 mercury at the bottom of the tube did not leave a trace, when 

 shaken, as is always the case when traces of chlorine are present. 

 Further the rest of the salt should react as a basis. But even with 

 the method of F. Mylius and F. Förster with 0,001 normal 

 solutions and iodine-eosine as an indicator we were not able to 

 find an alteration in the alkalinity of the salt. Evidently no decora- 

 position of the salt takes place. 



Further we stated that the salt was just as active when the burn- 

 ing lamp was immersed in liquid air. This made it exceedingly 

 improbable that we had to do with a decomposition of the salt and 

 the decolorating action by a halogen atmosphere. Moreover it was 

 proved that also other halogen compositions as Na Br, Kl are just 

 as active at the temperature of liquid air as at room temperature, 

 while for Br and especially for / an exceedingly small activity might 

 be expected at liquid air temperature. 



We also considered the possibility that the velocity of vaporization 

 of the tungsten might be lowered by the formation of a covering 

 layer of salt. M. Knudsen ') had stated a similar decrease of the 

 velocity of vaporization in the case of impure mercury and 

 1. Langmuir ') found in his investigations that even at 3300° K. during 

 some time tungsten can remain partially covered with a layer of 

 oxygen (of atomic thickness). It is therefore not excluded that a 



1) In these investigations we could state that the vaporized W particles are 

 not charged. 



2) M. Knudsen, Ann. d. Phys. (47), 697, 1915. . 



^) I. Langmuir, Journ. Amer. Ghem. Soc. (38), 2221, 1916. 



