Trowbridge — Spectra of Gases. 421 



obtain by such a discharge. The limit I have reached is the 

 volatilization of silica ; perhaps 1800 degrees. At this tem- 

 perature the spectrum shown by all gases in narrow capillary 

 tubes consists of a continuous spectrum crossed by broad bands 

 due to silica or to an oxide of silica ; the gaseous spectra are 

 completely masked. This masking seems to be due to the 

 greater conductibility of the volatilization products from the 

 walls of the tubes and from the metallic terminals. It seems 

 to me that this variation in conductibility is sufficient to account 

 for the phenomena of masking without recourse to a theory 

 of electrons which provides for suitable damping of electrical 

 oscillations. The electron theory may be an ultimate explana- 

 tion, however, of electrical conduction. 



When terminals of different metals are employed in capil- 

 lary tubes of glass or quartz, and are separated four or five 

 millimeters, complicated phenomena result from powerful con- 

 denser discharges through the rarih'ed gases contained in these 

 tubes. 



All specimens of glass which I have tried, soft G-erman glass, 

 lead glass, Borsilicon glass, or Jena glass, give broad bands due 

 to silica; lead glass gives, in addition, leadlines. Jena glass 

 gives a very strong line of boron at wave length 3451*49. 

 These lines and bands are obscured by a continuous spectrum. 



The narrow capillaries with metallic terminals, which I have 

 used, may be called electric furnaces in which there is no per- 

 manent product or permanent decomposition ; moreover, the 

 spectra which we observe do not reveal all that the capillaries 

 contain. Hydrogen may be present; but it is concealed. 

 Oxygen shows its presence only by probable oxides ; the con- 

 stituents of rariiied air are undoubtedly always there. The 

 conditions which prevail in the case of discharges in such nar- 

 row capillaries seem to be analogous to those in the case of 

 discharges under liquids. In this latter case we also have 

 reversals of metallic lines ; and, moreover, certain character- 

 istic lines of metals are wanting — see "Spectra from the 

 Wehnelt Interrupter," Harry W. Morse. (Proc. American 

 Academy of Sciences, May, 1904.) 



These results make one doubtful in regard to the entire sub- 

 ject of spark spectra which are observed between metallic 

 terminals in ordinary air ; and we are forced to ask, what 

 influence does the environment have upon the character of 

 these spectra — to what must we attribute the presence of oxy- 

 gen ? And even if we take spark spectra between metallic 

 terminals in an atmosphere of hydrogen or nitrogen we are not 

 sure that the results are not modified by the gases which are 

 occluded in the metallic terminals. 



Are we sure that, even in electrodeless tubes, helium is a 



