300 E. Sissingh on Kerr's 



clined to each other at a small angle, viz. ahout 1 minute *. 

 The tube in which the prism is fixed is attached by means 

 of three pressure- and three tension-screws to the rim of 

 a second tube, which is placed at the centre of the graduated 

 circle. 



Following Gauss' method, and observing the images of the 

 telescope spider-lines reflected from the faces of the prism, 

 the axis of the latter can be set parallel to the collimator axis. 

 The error of this setting was less than 1' of angle. The posi- 

 tion of the polarizer is such that its axis coincides with that 

 of the incident ray of light. The axis of the incident ray is 

 found by means of a cap with a fine hole through its centre 

 which fits over the collimator lens. 



The analyser is a Nicol prism with its faces inclined to each 

 other. The dark band was most clearly seen with such a 

 prism. The condition for the proper position of the analyser 

 was that the ray of light reflected from the face of the Nicol 

 should describe a circle round the axis of the incident ray 

 when the prism was rotated. The axis of rotation of the 

 analyser was also made to coincide with that of the ray 

 falling upon it. 



This care in the adjustment of the Nicol prisms is abso- 

 lutely necessary, in order to observe the dark band at every 

 angle of incidence, and especially so if the angle of incidence 

 is small. The new method of observation puts us in a posi- 

 tion to determine small rotations with a degree of accuracy 

 hitherto unattainable f. 



9. As the zero- and minimum-rotations only amount to a few 

 minutes of angle in the case of magnetic equatorial reflexion, 

 and the small circles of the Nicol prisms are divided in 

 degrees and have no vernier, these rotations were measured 

 by a mirror and scale method. At a distance of 2'5 metres 

 from the polarizer were placed a telescope and scale, the 

 former receiving the image of the scale from mirrors attached 

 to the polarizer. These mirrors were placed on the faces 

 of a four-sided hollow prism surrounding the polarizer- 



* Lippich, Wien. Ber. xci. p. 1079, 1885. 



f This method, which was used by me previously in the direct de- 

 termination ot the principal positions of Nicol prisms in which the planes 

 of polarization are parallel and perpendicular to the plane of incidence, is 

 now applied to the measurement of small rotations of the plane of polari- 

 zation by magnetic reflexion. In this manner the principal positions 

 are determined with a far greater accuracy than in the indirect method 

 hitherto almost universally adopted. Cf. Sissingh, Dissertation, p. 28; 

 Arch. Neer. xx. p. 8 ; Kighi, Ann. de Cliim. et de Phys. (6 ) iv. pp. 446, 

 456 (1885) ; ix. pp. 127-131 (1886) ; x. p. 208 (1887). Cf also § 13, 



