124 



EEPOET 1882. 



be left out of account. For the full discussion of it we refer to the 

 original paper, and only quote, in conclusion, the summary of results 

 obtained : — 



1. There is a real cause acting in a direction opposed to the law of 

 harmonic ratios, so far as fractions formed by numbers smaller than 70 

 are concerned. 



2. After elimination of the first cause a tendency appears for fractions 

 formed by two lines to cluster round liarmonic ratios. 



3. Most probably some law hitherto undiscovered exists which in 

 special cases resolves itself into the law of harmonic ratios. 



It must be remarked, however, that these conclusions must stand at 

 present on the evidence of the iron spectrum alone, and it is not im- 

 possible that the regularities which have been discovered are due to 

 accident. We can at present only say that the investigation, as far as it 

 has gone, seems to point to the above conclusions. 



There is one fact which points very strongly to another yet un- 

 discovered law which rules over the distribution of lines in spectra. It is 

 often observed that the spectrum of some body contains two or three 

 lines in close proximity, forming a characteristic group. Such doublets 

 or triplets are often repeated several times in the spectrum ; yet, though 

 we might expect, if the harmonic law was true, to find some simple rela- 

 tions connecting the periods of these sets, such is not the case. The lines 

 of sodium, for instance, are all double. In the set of lines given by 

 Thalen the components approach each other as we pass to the more 

 refrangible end of the spectrum more rapidly than they would if the 

 lines were connected by the harmonic law. In the following table the 

 wave-length of the least refrangible of each set of sodium lines is given, 

 together with the distance of the two components. In addition to the 

 pairs observed by Thalen, one pair in the ultra-red, photographed by 

 Abney, and one pair in the ultra-violet, photographed by Cornu, ai-e 

 given. 



Professors Liveing and Dewar' have observed some additional sets of 

 double sodium lines, but the distance between the components of each pair 

 does not seem to follow any law. 



The spectrum of potassium as observed by Liveing and Dewar^ 

 contains, when the metal is heated in the electric arc, five groups of 

 lines, each containing four lines. Though, roughly speaking, the lines 

 of each set are the nearer together the shorter the wave-length, of the 

 set, there seems to be no general and well-defined law. 



The most remarkable perhaps of all the groups of lines observed in 

 the spectra of metals are the magnesium triplets. The well-known set of 

 lines in the green is repeated three times in the ultra-violet ; but Table V. 



' Proo. Boy. Soc. xxiv. p. 398 (1879). 



' Ibidem. 



