1901.] 



and Melting Points to Atomic Mass. 



5 



the metals given above : — Tin, lead, bismuth, chromium, manganese, 

 iron, cobalt, nickel, ruthenium, rhodium, palladium, and iridium. 



The above nineteen metals may be* classified by their spectra, as 

 shown, into groups in agreement with Mencleleeff's Periodic Law. The 

 spectra of the metals in each group are similar to one another, and they 

 are quite different from the spectra of the other groups. 



The Diagrams. 



It is apparent, when comparing the spectra of each group, that the 

 positions of the strongest lines, and of the others in order, change 

 regularly with the increase of atomic mass of the metals. The change 

 in position is apparently so simple that it suggested a graphical method 

 of representing the spectral lines as functions of the atomic mass. 

 The lines were plotted as abscissae, and the atomic masses as ordinates. 

 Two diagrams were drawn at first, one from the oscillation frequencies 

 of the lines, and the other from the wave-lengths. Connecting lines 

 were then drawn through the corresponding lines in homologous 

 spectra. These were mostly curved, and it was thought that the 

 equations to these curves might be discovered by further study, and 

 also that some relation might possibly be discovered between the 

 equations. 



Several curious results were obtained by observations on the points 

 in which the converging lines, drawn through the corresponding 

 members of doublets and triplets, intersected. These points were on 

 the same horizontal line in each group, but it was difficult to determine 

 their position accurately, and the results have since been regarded 

 more as coincidences. The method, however, gave promise of yielding 

 valuable results, and the research has been continued, partly by a 

 further investigation of flame spectra and partly by an improvement 

 in the method of drawing the diagrams. Use has been made also of 

 some of the work of Rydberg, Preston, and others, to whom reference 

 will be made in the paper, to make the research as accurate and 

 complete as possible. 



Discussion of the Spectra and the Lines. 



It may be stated at once that nearly all the lines observed in the 

 flame spectra of the metals considered have been included in the 

 diagrams. A few lines, selected from arc and spark spectra, have 

 been used, but only in cases where our knowledge of flame spectra is 

 incomplete. 



There is no difficulty in selecting the corresponding lines from some 

 spectra. 



Calcium, Strontium, Barium. — The oxyhydrogen flame spectrum of each 

 * Hartley and Hamage, ' Trans. Roy. Dublin Soc.,' N.S., vol. 7, pt. 12. 



