1076 



tungsten wire must glow strongly for the formation of Si from Si02, so that the 

 small quantity of Si that might eventually be formed, will directly be vaporized 

 again. It will therefore be better to arrange the experiment with the quartz capil- 

 lary, in such a way that we investigate whether more tungsten in the form of 

 WO3 is precipitated on the wall than in a lamp without quartz. In fact we then 

 obtain a positive result, but the quantity of WO3 is smaller than when glass is 

 used. That reaction between W and SiOj is analogous to that between C and 

 quartz, by which reaction according to W. Hempel ') reduction to Si and formation 

 of CO take place. 



An intensive reaction between SiOa and W does therefore certainly not take 

 place in the lamp sprayed with SiO.-powder. The good influence at the first burning 

 is therefore also much smaller with quartz than with glass, so that already for 

 small quantities of impurities that develop gases, the activity of the SiOj is insuf- 

 ficient. 



Metal-oxides as ZrO^, TliO,, TiO,, Na,() ure iioi tit to replace 

 the action of pliospliorus. Either they give rise to a gh)\v discharge 

 at the first burning of the lamp (by whicli the lamp can be 

 destroyed) or (as in the case of NajO) tliey are vaporized before 

 tiiose elements can be liberated, which possess the same activity as 

 the phosphorus. Besides, traces of impurities, which can accom[>any 

 such substances as Na^O, lower the degree of the vacuum, of the 

 oxides mentioned above SiO, only can be used. But this is a non metallic 

 oxide that intensively emits electrons or ions before or during the 

 decomposition. 



Is there no metal-oxide, less volatile than Na^O, which does not 

 at all or hardly increase tlie glow-discharge at a high temperature? 

 Perhaps Al^O^ might be usefid. This oxide however with the well- 

 known aflinity bet\veen aluminium and oxj'gen is very stable at high 

 temperatures and at all exents did not show the activity that was 

 wanted. A different result was obtained with i\IgO. in fact this shows 

 the phosphorus-activity in a sufHicient degree. 



The application of glass remains however the simplest method. 

 It is somewhat paradoxal, that we only need to rub the bulb of 

 a glow-lamp to powder and to spray the filament with it to obtain 

 a better lamp after sealing it into a new bulb. But this paradox 

 may have found an explanation in the preceding discussion. 



It may still be remarked that we need not confme ourselves to 

 the use of the mentioned substances. We can also apply these in 

 combination with others, the good activity of which is known. 

 It is however remarkable, that the combination glass-pow^der -|- 

 phosphorus gives less satisfactory results. At the first burning of 

 the lamp the phosporus acts perhaps reducing in the vaporizing 

 glass, which causes already a little blackening of the precipitate on 



1) W. Hempel, Zeitschr. f. angew. Chem. 30, 10 (1917). 



