6S Wright and Downs — Spectrum of the 



4247 to about wave length 4225 came out which appeared to 

 be the same in the case of all the metals employed. When 

 the coatings of a large Leyden jar were connected to the ter- 

 minals of the secondary, the discharge took the form of 

 intensely brilliant sparks following one another in such rapid 

 succession and such energetic detonation as to cause a jagged 

 and almost deafening roar. In general the spectra of the 

 metals employed in the induction flame are less complete and 

 less developed than their flame spectra, owing to the fact that 

 they are obscured by the rather intense continuous spectrum. 



For a more careful study of the spectrum than the one pre- 

 viously described, which was made with the naked eye, it was 

 necessary to have recourse to photography. To obtain the 

 photograph of this discharge a Rowland concave grating hav- 

 ing a radius of 21*5 feet was employed. The upper half of 

 the photographic plate was exposed to the spectrum of the 

 sun's rays for a comparison and the lower half to that of the 

 flame between the terminals of the metal to be studied. 



The plate upon which was the spectrum which it was the 

 primary purpose of this investigation to locate, was stained 

 with erythrosine to make it more sensitive to the rays at the 

 red end of the spectrum. This staining also appears to 

 increase the sensitiveness of the plate for ultra-violet rays. 

 These lines are in the region where the red of the first spec- 

 trum and the ultra-violet of the second overlap, and they begin 

 abruptly at wave length 6127*32 in the red and extend down 

 to 6269-09, or at 3063*66 in the ultra-violet, extending down 

 to 3134*55. Inasmuch as the two spectra overlap, it was un- 

 certain to which spectrum they belonged. 



The first thing that suggested itself was that these might be 

 due to copper from the electrodes. Very few copper lines so 

 far down in the red are given in the very admirable tables of 

 the British Association Report.* This fact renders it impos- 

 sible to make a comparison of any significance with the lines 

 given in this region. The number of copper lines given in 

 the ultra-violet was much larger, but still so much smaller than 

 the number of lines upon the plate that no comparison of any 

 value could be made although there were a few cases of approx- 

 imate coincidence. 



With a view to ascertain to which spectrum the lines belonged 

 various devices were employed. At first plates stained with 

 cyanine were tried, and some preliminary experiments made in 

 obtaining the solar spectrum to get the time of exposure. It 

 was found that an exposure of eighteen minutes gave a good 

 spectrum extending into the red far beyond the situation of 

 the lines in question. Several plates were then exposed to the 



* 1884, p. 384. 



