INTIMATE STRUCTURE ON COOLING. 69 



amounting to about a third of the distance between the original bands. 

 The nature of this apparent shift is as follows : 



Each band at +20 may be regarded as an unresolved doublet, of 

 which in general the member of longer wave-length is relatively so 

 much the stronger that its position determines approximately the 

 location of the crest of the composite band (see fig. 57). The effect of 

 cooling is to resolve this doublet into separately distinguishable bands 

 and at the same time to cause a subsidence of the stronger and an 

 increase of the weaker member. The member of the shorter wave- 

 length usually becomes dominant at low temperatures, and in so far as 

 this occurs the arrangement of the spectrum appears to be undis- 

 turbed but shifted toward the violet by an amount representing the 

 width of the doublet. There are, however, certain exceptions to this 

 rule, so that the relation of the resolved spectrum to that at +20 is 

 not so simple as the above description would imply. The appearance 

 of the group, if this be its real structure (i. e., a set of nearly equi- 

 distant doublets, the distance between the members of all the doublets 

 being nearly the same), would then be as shown in figure 57. 



At +20, B', C', D f , E', and A' are entirely concealed by the over- 

 lapping of the bands. At 185, B, (7, D', E, and A may or may not 

 be visible, according to their intensity or the completeness of the resolu- 

 tion, which in fact varies greatly in different parts of the spectrum. 



It will be noticed that in the lower diagram in figure 57, D and not 

 D' is the dominant component. This is a condition which obtains in 

 the ammonium chloride, with the result that C' and D, which appear 

 to have replaced the strong C and D bands of the spectrum of +20, 

 are near together, D and E' far apart, and the symmetry of the group 

 is impaired. Similar complications occur likewise in the spectra of the 

 other double chlorides. 



To illustrate the application of this assumption, the spectrum of the 

 ammonium uranyl chloride has been mapped in the manner shown 

 in figure 58, in which the fluorescence bands of the 8 groups as they 

 occur at 185 are shown in their relation to a hypothetical grouping 

 given at the head of the diagram. This grouping consists of the set of 

 imagined doublets of which, as in a previous paragraph, the spectrum 

 at +20 is supposed to be made up. The spacing for each doublet is 

 that determined from the observed average shift on cooling and the 

 relative divergence from this arrangement is shown for all the bands 

 of each group. 



A scrutiny of the fluorescence spectrum at 185, group by group, 

 by means of this diagram, affords very satisfactory confirmation of 

 this hypothesis concerning the apparent shift. It is obvious : 



(1) That not all the components B, C, D, E, and A will necessarily be 

 visible in every group of the resolved spectrum. 



