1901.] 



and Melting Points to Atomic Mass. 



11 



Eeese has shown that the 3944 line of aluminium yields, under the 

 Zeeman effect, a quadruplet and the 3961 line a triplet. 



If we confine our attention to the lines of the flame spectra of the 

 metals the diagrams are comparatively simple. The work may be 

 extended, however, by including the more refrangible lines in the arc 

 and spark spectra of several metals. Rydberg's formula, and the 

 formula of Kayser and Runge, will assist greatly in this extension of 

 the work. . 



The study of the effect of a magnetic field on the source of light 

 (the " Zeeman Effect") will be of great importance in confirming the 

 selection of a few of the lines included in the present paper, and in the 

 prosecution of the investigation in the future. 



The more refrangible lines are always feeble and nebulous compared 

 with the less refrangible members of the series, and it is more difficult 

 to obtain accurate measurements of these weaker lines. The principal 

 series of lithium furnishes us with a typical example of this difficulty. 

 The lines were measured in the arc spectrum by Kayser and Runge. 



Wave-lengths. 



Limit of 

 error. 



Intensity. 

 10 = maximum. 



6708 -2 



0-2 



10 



3232 -77 



0'03 



8 



2741 -39 



0-03 



6 



2562 -60 



0-03 



4 



2475 -13 



0'1 



4 



2425 • 55 



o-i 



2 



2394 -54 



0-2 



1 



If we except the line in the extreme red, the accuracy of the 

 determination of the wave-lengths diminishes with the intensity and 

 refrangibility of the line. 



Discussion of the Diagrams {Plates 1 and 2). 



The diagrams were first drawn with the atomic masses as ordinates, 

 and the lines connecting corresponding lines were nearly all curved. 

 Dr. J. H. Vincent, of this college, suggested to me, in February, 1900, 

 that different functions of the atomic masses should be used as 

 ordinates, so that as many of these curved lines as possible should be 

 converted into straight lines. It was found immediately that, by 

 taking the squares of the atomic masses, the line joining the strongest 

 lines of calcium, strontium, and barium — wave-lengths 4227, 4608, 

 and 5536 respectively — was very nearly straight. 



Diagrams have since been drawn to this scale, both in oscillation 

 frequencies and in wave-lengths, of all the lines enumerated above. 



