PoLLOK — The Vacuum Tube Spectra of some Metallic Vapours. 203 



liues or even bands that could be readily and exclusively attributed to any 

 deGnite compounds. 



About this time I examined the reproduction of the vacuum tube spectrum 

 of stanuic chloride in Higgeiibach and Kooen's Atlas, and compared it with 

 the vacuum tube spectrum of chlorine and the spark spectrum of tin, and 

 noted that all the lines of the chloride were either the lines of tin or chlorine, 

 and that I could not find any new lines due to the compound. 



Now the lines of the spark spectra of the metals are very well known ; there 

 is no doubt or ambiguity about these spectra, and as a preliminary step to tlie 

 investigation of elementary and compound spectra it seemed desirable to 

 obtain pliotographs of a large number of volatile metals and their chlorides, 

 and see how far their spectra were composed of the Hues of the spark spectra 

 of the metal and the chlorine. 



In my earlier experiments I used glass tubes with quartz windows, and in 

 some eases glass tubes with the centre part of the capillary tube of quartz, 

 and joined the two ends by rubber tube wired on ; but the former cracked 

 and broke, and the latter could not be used with a high temperature, and 

 much trouble was experienced until I invented the simple little quartz vacuum 

 tube shown in the illustration, PL XVI., when all the difficulty vanished, 

 and the photographs of spectra were taken with the utmost ease and rapidity. 

 To measure every line in each spectrum and identify them all would be a 

 very laborious task, and I do not think at all necessary at the present stage 

 of the inquiry. In PI. XV. I have placed the spark spectrum of the metal in 

 each case just over the vacuum tube spectrum, and the lines that obviously 

 correspond are identified by the use of an ivory wave-length scale applied to 

 each spectrum, and the accepted values of tlie lines adopted and tabulated. 



In many cases lines are seen in tlie vacuum tube spectra that are not seen 

 in the spark spectrum of the element. Those lines, when not due to chlorine 

 are generally found to be due to accidental impurities in the metals or 

 chlorides used, or to the presence of nitrogen, water-vapour, or carbon ; but in 

 some few cases they are not so easily accounted for, and will receive further 

 consideration in subsequent investigations. The scope of the present paper 

 is confined to a rapid survey of a large number of metals and their chlorides 

 and the identification of the lines that are common to both spectra, with some 

 indication of the relative intensities of the lines in the spark spectrum, and in 

 the vacuum tube spectra with and without a condenser, together with the 

 approximate position of the bands that are a prominent feature of some of the 

 vacuum tube spectra. The author hopes later to make a detailed examina- 

 tion of a variety of compounds from the same metal, in the case of the more 

 characteristic metallic vacuum tube spectra, to ascertain how far the presence 



2h2 



