46 Mr. J. S. Ames on Relations between 



Groups j3 and h are excluded because, containing doublets and 

 triplets, it is impossible to determine just what wave-number 

 should be used. 



It is difficult to say whether or not the agreement in the 

 above figures would be better if the wave-numbers were 

 known more accurately; but I am inclined to think that 

 natural constants for the two metals are indicated by them. 



It will be noticed that nearly all the homologous groups are 

 in the ultra-violet. This seems to be true of all spectra. It 

 is only, apparently, when the wave-numbers, or frequencies, 

 are sufficiently large, that the similar modes of vibrations of 

 the molecules become evident. 



The resemblance of the spectrum of magnesium to those of 

 zinc and cadmium is much less marked than the one just 

 noted. In fact, I think nearly all the points of similarity 

 given by Prof. Hartley are wrong. Magnesium has, it is 

 true, ten groups of triplets ; but their physical properties bear 

 no apparent relation to those of the zinc and cadmium triplets. 

 The beautiful series of five lines at wave-length 2779*9 in the 

 magnesium spectrum surely has no companions in the zinc 

 and cadmium spectra. There is, however, one striking feature 

 common to all three. This is the presence of one broad re- 

 versed line. In magnesium it is at wave-length 2852-2 ; in 

 zinc, at wave-length 2138*3 ; in cadmium, at wave-length 

 2288*1. Liveing and Dewar have observed the line at wave- 

 length 2852*2 in the arc-, spark-, and flame-spectra of mag- 

 nesium ; so it is evidently in some sense a fundamental line. 

 The differences of the wave-numbers in each magnesium 

 triplet are also nearly constant, being close to 4 and 2. 



The spectra of calcium, strontium, barium, and mercury 

 also contain triplets ; but their number is small in comparison 

 with that of the lines not so included. 



It is hardly necessary to refer to the series of pairs in the 

 spectra of sodium, copper, and silver ; but it may be well to 

 note that each of these spectra has in the ultra-violet one 

 prominent pair, always occurring as impurity lines, very u per- 

 sistent," and coinciding with the solar spectrum. The sodium 

 pair has the wave-length 3303*0 and 3302*4 ; that of copper, 

 3274*0 r. and 3247*7 r.; that of silver, 3282*9 r. and 3280*7 r. 

 The presence of silver in the sun is at last proved. From a 

 study of their physical properties 1 think the series of sodium 

 pairs is in reality made up of two overlapping series, thus 

 bearing a resemblance to the two series of potassium pairs. 



The spectrum of lithium is similar to that of hydrogen. It 

 has a series similar to the hydrogen one and obeying almost 

 identically the same law, beginning at wave-length 3232*8. 



