Separation of the Spectral Lines of Thorium 5 



from the standpoint of Voigt's equation, is an experimental 

 method of five times the accuracy which one may possibly obtain 

 with some of the new forms of interferometer. Such a method 

 was used by Gmelin (Phys. Zeit., vol. 9, p. 212), who found the 

 mercury line 5790 A to have a dissymmetry proportional to the 

 square of the field strength. This results from the hypothesis of 

 linkages among the electrons (Voigt, Phys. Zeit., vol. 9, p. 353). 

 The field strength of the two sets of plates here studied differed 

 only by 20 per cent, and were not adapted to< an investigation 

 of that point, like Gmelin's observations, whose field strength 

 varied two and one-half fold. However, the following consider- 

 ations will show that my measurements should have given me 

 some intimation of such a change with the change in field 

 strength. If one designates the distances of the two ^-compo- 

 nents from the null position by "a" and "b" the distance from 

 the middle is (a-\-b) /2, and the deviation of the apparent middle 

 is (a — b)/2. For 20 per cent greater field, on the assump- 

 tion of Gmelin, the former is increased 1.2 times, and the latter 

 (1.2) 2 times. The weaker of my dissymmetrical lines were meas- 

 urable only upon one set of plates, ' but those lines which could 

 be measured upon both sets of plates gave practically the same 

 results when the separations of the weaker field were increased 

 1.2 times. It would not seem that all of the difference for all of 

 the lines was due to experimental error. If the result is granted, 

 it would mean that the dissymmetry is proportional to the field 

 strength. However, greater variation in field strength would be 

 more competent to settle the question. From the Voigt equation 

 the above half sum is also the usual separation and the half dif- 

 ference the dissymmetry. The latter value is a constant, and is 

 independent of the field strength. Now there are cases in thorium 

 where a equals 3b, and therefore the magnitude of the dissym- 

 metry is b. Maintain this value of b constant and diminish the 

 field one-half. The distance from the apparent middle is then 

 (a-|-&)/4 which equals b. One component then falls in the null 

 position, and only the other component is displaced. This is 

 meaningless. And 20 per cent difference in field should have 

 given me something different from what was observed if there 



Q3 



