INTIMATE STRUCTURE ON COOLING. 



65 



the geometrical centers of the groups are in the same vertical line. It 

 will be seen from the diagram: 



(1) That the group center is in all four cases almost coincident with 

 the crest of the D band. 



(2) That the distance between D and E is approximately the same 

 in all. 



(3) That the arrangement of bands within the group is essentially 

 the same in all, except for the displacement of band A in the spectrum 

 of the rubidium and of B and C in that of the caesium salt, as mentioned 

 above. 



Further discussion of these discrepancies will be found in a later 

 paragraph of this chapter. 



INTERVALS OF THE INDIVIDUAL SERIES. 



For the consideration of the frequency intervals of the individual 

 series, the values from table 29 have been arranged by series in table 

 30. Distances between the observed positions of neighboring members 

 of each series are given in the column marked "Intervals." To 

 facilitate the detection of systematic departures from uniformity of 

 interval, a column of values calculated by the following method is 

 likewise given: A " center" for each group was found in the manner 

 already employed for determining the group centers. Around this the 

 calculated positions were arranged under the assumption of a con- 

 stant frequency interval equal 

 to the average of the observed 

 intervals for each series sepa- 

 rately. The column marked 

 1 ' Differences ' ' indicates the de- 

 parture of the observed values 

 from those thus calculated. 

 The departures from uniform- 

 ity of interval are unsyste- 

 matic, indicating, for all the 

 salts, that the series may be regarded as having a constant frequency interval. 



This interval has been computed for each series by subtracting the 

 observed frequency of each band from the frequencies of all the other 

 bands of the series and dividing the sum by the total number of inter- 

 vals in question. The results are presented in table 21. 



These data indicate no progressive change of interval with the 

 molecular weight, except that the interval is definitely smaller for the 

 caesium uranyl chloride. The other three salts, so far as this deter- 

 mination goes, must be regarded as having the same average interval. 

 It is likewise difficult to distinguish with certainty differences in the 

 intervals of different bands in a given salt, except that the C band has 

 in general a smaller interval than the other series, or of a given band 

 in the different salts, excepting in the case of the caesium chloride. 



FIG. 54. 



