256 Dr. P. Zeeman on Doublets and Triplets in the 



the grating gives brilliant lines. This quality is possessed by 

 a grating in the possession of the laboratory of the University 

 of Groningen. Jts director, Prof. Haga, kindly invited me 

 to make some measurements with his apparatus, which were 

 in full working order. The grating is mounted in a very 

 stable manner, which, of course, is very favourable for accurate 

 measurements. 



15. For the particulars of the mounting I refer to Haga's 

 paper (Wied. Ann. lvii. p. 389, 1896). The grating (best 

 quality) has a radius of 10 ft. and 10,000 lines to the inch. 

 The source of light used was a piece of asbestos paper soaked 

 with molten salt and introduced into the flame of coal-gas fed 

 with oxygen under high pressure. An image of the sodium 

 flame was formed on the slit by means of a lens. Between 

 the lens and the slit a large nicol was placed ; the distance 

 between the slit and the flame being about 50 cm. Care was 

 taken that there were no absorption-lines in the spectrum of 

 the non-magnetized flame. 



16. The nicol was placed so that its plane of polarization 

 was horizontal. With the putting on of the current the two 

 lines mentioned in § 14 appear (c/. also § 10). The distance 

 between these lines was now measured by means of a micrometer 

 eyepiece. The movable frame carries cross-wires; the cross 

 resembles that of St. Andrew. For spectroscopic measure- 

 ments this cross is recommended (see e. g. Scheiner, Spectral- 

 analyse der Gestirne, p. 74). Illumination of the wires was 

 necessary. Now the position of each of the 2x2 lines formed 

 by the D-lines was read. The difference between two readings 

 gives the distance between the centra for D x and for D 2 in 

 terms of the divisions of the screw-head (one revolution =100 

 divisions). These differences are entered in the following table. 

 The electromagnet soon became very hot by the heat generated 

 by the necessary current (22 amp.) and by the action of the 

 flame. Hence it was impossible to make more than three 

 or four measurements without interrupting the current. 



Thirty-eight measurements give for the distance between D x 

 and D 2 288 divisions. The probable error of one measurement 

 of the magnetic change is 6*5 divisions for D b 4*5 for D 2 . The 

 results have the probable errors 15 and 1*0 divisions. The 

 magnetic change is the same for the two sodium lines, the 

 difference lying within the limits assigned by the probable 

 errors. The intensity of the field (determined by a bismuth 

 spiral) was 22,400 C.G.S. In this field the positive and 

 negative magnetic change of the period amounts to if^. 

 Hence e/m is 1-6. lO" 10 . 



