INTIMATE STRUCTURE ON COOLING. 67 



systematic phenomenon. Ignoring group 7, in which the bands are 

 displaced by absorption in a manner to be discussed later, we find the 

 following shifts to exist. 



TABLE 22. Shift of the groups. 



Group 2 3 4 5 6 



Shift 5.4 5.1 5.3 5.2 5.1 



Average shift from K to Cs, 5.2 



Ths shift is therefore to be regarded as approximately uniform 

 throughout the spectrum. The shift is much greater between NHi and 

 lib than in the other cases, the averages being as shown in table 23. 



TABLE 23. Average shift of groups. 



K-NH 4 1.6 



NH 4 -Rb 2.7 



Rb-Cs 9 



It will be noticed that in this discussion the order of molecular 

 weights used is K, NH 4 , Rb, Cs NH 4 being placed between K and 

 Rb instead of in its proper position. This is in accordance with the 

 results of Tutton, 1 who has shown that in various optical properties of 

 crystals which depend on the molecular weights, NH 4 always lies 

 between K and Rb, as though its effective molecular weight were 

 larger instead of being smaller than K. 



THE EFFECTS OF TEMPERATURE. 



The narrow, line-like bands into which the ordinary uranyl spec- 

 trum is resolved at low temperatures 2 form a rather complex aggre- 

 gation separable into a series of identically arranged groups corre- 

 sponding to the unresolved bands at +20, but related to the over- 

 lapping components of the latter in a manner not easily capable of 

 direct determination. It was deemed of especial interest, therefore, to 

 observe the effect of cooling on the double chlorides, where the relation, 

 owing to the partial resolution at +20, should be more obvious. 



For this purpose a crystal, C, of the salt to be examined was mounted 

 within a long cylindrical Dewar flask, D, with unsilvered walls (fig. 56) . 

 The carbon arc A was focussed on the crystal by the lens L. A water- 

 cell W was inserted between the arc and the condenser. The light- 

 filter F was opaque to all but the violet and ultra-violet rays used for 

 excitation. Observations with the Hilger spectroscope H, a portion 

 of the collimator of which is shown, were made through a second filter 

 E opaque to the exciting light but transmitting the fluorescence. The 

 arc and specimen were well screened by an opaque box BB. When 

 it was desired to photograph the spectrum a camera was substituted 

 for the observing telescope of the spectrometer. 



The control and adjustment of temperature were effected by attach- 

 ing the crystal at the upper end of a vertical copper rod which could be 



1 Tutton, A. E., Crystalline Structure and Chemical Constitution. (London, 1916.) 

 1 See Becquerel and Onnes, 1. c. 



