( 735 ) 



tion adjusted at the laboratorv accordiu*»- to Coknu and W. Konig ^). 

 On account of tlie considerable astigmatic difference in the images 

 of horizontal and vertical lines formed by a concave grating, the 

 plate was placed near the focus of a third lens in order to enable 

 us to cancel this astigmatism for different parts of the spectrum by 

 comparatively small displacements. The line of demarcation could then 

 be adjusted sufficiently sharply in the S|)ectrum, which Könk; had not 

 succeeded in doing. A nicol followed the mica plate, and then came 

 tlie principal slit. With this arrangement a normal doublet is known 

 to appear in the spectrum as a broken line e. g. thus J ; and on 

 rotation through 90° of the iiicol round the direction of the rays or 

 of the ' I ^ plate round its uertical diameter *. at once appears. 



§ 13. As a rule the samples were mounted in a copper frame- 

 piece and clasped between the polar end-planes ; it is desirable to 

 have an airtight fitting so as to prevent cold currents of air with 

 formation of rime. The level of the liquid air may now rise above 

 the openings so that the sample is quite immersed. The air stag- 

 nating in the bores is effectually dried by the preliminary cooling 

 with solid carbon dioxide. With thin samples we obtain i]i this way 

 a field of 40 kilogauss, which is quite essential for the proper 

 resolution of the quadruplets etc to be described later. With smi- 

 light and a width of 0.05 mm. of the principal slit there was still 

 plenty of light even in the violet ; the Fkalnhofer lines, however, 

 proved so troublesome in many cases that the much weaker arc 

 light had to be used. The spectrum was measured by means of 

 a magnifying glass and a graduated glass scale, the divisions of 

 which amounted to 0.225 mm., exactly corresponding to 0,1 (ifi in 

 the spectrum of the first order. The auto-collimator, which we also 

 used has been described since oui- first communication (see the 

 jireceding papei-). 



x\ll the followijig experiments were made with a longitudinal field, 

 in other words with an axial direction of the i-ays ; many new 

 adjustments would be required after turning round the heavy electro- 

 magnet, so that we hope to extend the observations to an equatorial 

 direction of rays later on. 



§ 14. Third series. Of the large nundjer of coloured com- 



1) A. GoRNU, Gompt. Rend. 125 p. 555, 18U7. — W. Konig, Wied, Ann. 62 

 p. 242, 1897. We found it safer not lo place this arrangement at the end of the 

 beam near the magnifying glass, on account of polarisation by the grating ; cf- 

 P. Zekman, These Proc, Oct. 1907. 



