1888.] Investigations on the Spectrum of Magnesium. 249 



absent. The last traces of the ultra-violet bands of nitrogen are 

 almost as difficult to be rid of with certainty. Frequently unknown 

 lines or bauds make their appearance, and the same tube will at 

 different times exhibit wholly different spectra. This is especially 

 the case with tubes of rarefied gases which oppose much resistance to 

 tbe passage of the electric discharge such as oxygen. 



It is no easy matter to prepare tubes for the observation of ultra- 

 violet rays to which glass is opaque. Our plan is to fit a sort of 

 stopper of quartz to an "end-on" tube (fig. 1). This stopper is a 

 slightly conical piece of rock-crystal with the truncated ends' of the 

 cone ground plane and polished. It is first fitted to the tube by 

 grinding and then cemented in with some vitreous substance more 

 fusible than glass. Formerly we employed sodium metaphosphate 

 which answered fairly, but latterly we have used fused silver nitrate 

 which is easier to manipulate. In any case it is very difficult to prevent 

 the tubes cracking under variation of temperature, but if the tube 

 does not crack it is as effectually closed in this way as if it were all 

 of one piece of glass. It is obvious that nitrogen, oxygen, and silver 

 might be derived from silver nitrate used as cement and might add 

 their spectra to those of the other contents of the tube. But the 

 stopper does not lie in the direct course of the discharge, and we have 

 not found that the silver nitrate is in general decomposed. The pro- 

 ducts of decomposition would at any rate give well-known spectra. 

 The unknown and variable rays we are inclined rather to attribute 

 to substances derived from the glass, either products of decomposition 

 under the action of the electric discharge, or to matters adherent to 

 the surface which become detached under some electric conditions, 

 and adhere again when those conditions are ehanged. We have 

 photographed the spectrum of one tube which had been filled 

 with oxygen several times and exhausted, and which gave a well- 

 marked spectrum containing a number of rays unknown to us. After 

 a time other photographs of the same tube showed an entirely 

 different spectrum, and after a further interval the spectrum was 

 found to be again entirely changed, and finally after a further 

 interval the original spectrum reappeared. Changes in the surface 

 tension between the glass and some adherent film may in this case 

 have facilitated the disengagement of the matter of the film and its 

 after re-adherence. Whatever the cause, such changes of the spectra 

 are none the less confusing and suggestive of caution in drawing our 

 inferences from the phenomena of vacuous tubes. 



The ultra-violet magnesium lines which we have observed in 

 vacuous tubes with magnesium electrodes, when the induction coil, 

 without jar, is employed, are the triplets at \3837, and the lines 

 A2852, 2802, and 2795. These appear whether the residual gas be 

 air, oxygen, hydrogen, or carbonic acid. When a jar is used we have 



