xviir.] AN UNEXPECTED RESULT. 251 



is the longest. Although the spectrum of the substances as they 

 exist in A would contain as many lines as would the spectrum 

 of the substances as they exist in D ; each line would be 

 thick at one temperature and thin at another. It would be 

 therefore longest at one temperature and shortest at another. 

 Hence, since the longest lines at one temperature will not le the 

 longest at another, the ivhole fabric of " impurity elimination" 

 based upon the assumed single molecular grouping falls to pieces. 



To take a concrete case. Let us suppose that in Fig.- 87 the 

 four furnaces represent the spectra of iron broken up into 

 different finenesses by successive stages of heat. It is first of 

 all abundantly clear that the relative thicknesses of the iron 

 lines observed will vary as the temperature resembles that of 

 A, B, c, or D. The positions in the spectra we may assume to be 

 the same, but the intensities will vary ; this is the point. The 

 strongest, and, therefore, as a rule the longest, lines will vary as 

 we pass from one temperature to another. 



But how does the whole fabric of impurity elimination fall to 

 pieces ? In this way. 



Let us suppose that manganese is a compound body, and that 

 one of its constituents is a form of iron represented in furnace B. 

 Suppose the photograph of the spectrum of iron I compared 

 with that of manganese is taken at the temperature represented 

 by furnace D, and that the photograph of manganese is also 

 taken at the same temperature ; now, to eliminate the impurity of 

 iron, in the manganese, we look for the longest (strongest) lines 

 in the iron photograph among the manganese lines. If we 

 do not find them we say there is no impurity of iron ; but 

 although the longest lines are absent we get the shorter 

 familiar ones. 



It will be seen at once, then, that the consideration of the 

 question gave as a natural consequence precisely the state of 

 things detailed in page 232 et seq., which on the old view 

 seemed so inexplicable. 



