Separation of the Spectral Lines of Thorium g 



error, corrections may be made if needed when physicists are 

 decided upon the accuracy and magnitude of these magnetic 

 fields. 



PART II. GENERAL DESCRIPTION OF THE TABULATED 

 MEASUREMENTS 



Tables I-XI inclusive contain symmetrical lines exclusive of 

 the triplets. 



Table I consists of two 14-component lines. The interior 

 J-components are too weak to be very positive of their presence. 

 The exterior ^-components are diffuse, and not capable of meas- 

 urement with desirable accuracy. The ^-components agree very 

 closely for the two lines. This fact, and the general appearance 

 of the components, leads to the conclusion that they are dupli- 

 cates. The separations are shown as multiples of .17, and these 

 agree as closely as could be expected for the /'-components. The 

 value .17 does not appear to be an aliquot part of the normal a, 

 for a/6 gives .184 and a/y, .158. 



Table II contains two 12-component lines, which are evidently 

 not related. The components of 4086.71 A are related in a simple 

 way. All are probably multiples of .214. This differs little from 

 30/16, which is not an aliquot part of a. The readings may also 

 be represented about as well by multiples of ff/11, which is one of 

 Runge's aliquots of a. The former value gives a distance from 

 component to component for the ^--separations of 4 x .214, and 

 the same value for the /^-components. The latter, a/il, gives 

 uniform spacings for the ^-components ; but for the />-compo- 

 nents we have 8a/n once, and 90/11 twice for the spacings. 

 There can be scarcely a doubt but that the components are equally 

 spaced and that the value is the same for both .?- and p-. This 

 leads to the rejection of the aliquot interval value, a/11. 



Table III gives two 11 -component lines. They are not dupli- 

 cates, but a multiple relationship holds for the components of 

 each line. The repeated values are not rational parts of a. 



Table IV contains two io-component lines. They are not du- 

 plicates and a simple relationship in the components is not 

 apparent. 



97 



