90 Mr. J. C. McConnel on the 



This gives the polarization of the subsidiary light below the 

 sun at a zenith distance of 46°. It is rather loose reasoning 

 to apply this result to points opposite to the sun at zenith 

 distances of 26°, 55°, and 83^°; but the consequent errors will 

 probably be small and should have opposite signs in the first 

 and third cases. Taking the mean of the readings at 90°, 

 we have 



3/ = -293(i< + «). 



From these two equations we have .2j='373w, 3/ = *403m at 90° 

 from the sun. At 60° and 120° we have 



r=- 



" and v = wcos^60°. 



u-\-x 



Inserting the values of x and ?/ we find 7' = '477, which agrees 

 with the observed values of r quite as well as could be ex- 

 pected, considering the uncertainty of my interpretation of 

 his ; readings. A similar process applied to the afternoon 

 observations gives at 70° and 110° r = '366. 



Description of Apjyaratus. 



The arrangement of my polarimeter will be understood 

 from PI. VII. fig. 1. The light entered through the aperture A, 

 and traversed the movable pile B of five glass plates, the fixed 

 pile C of three glass plates, and the thick plate D of Iceland 

 spar, cut at right angles to the axis. It was then reflected 

 by the mirror E, and reached the eye through the Nicol F. 

 Thus B and C depolarized the light, while the depolarization 

 was tested with the polariscope DEF. On turning the five 

 plates the black cross became gradually fainter and finally 

 disappeared, being soon replaced b}^ four black spots. I took 

 the reading halfway between the disappearance of the cross 

 and the appearance of the spots. These spots were of course 

 the spaces between the white cross and the first white ring. 

 The optic axis of T> gave a fixed direction for the observed 

 light, so the angle of incidence on the plates C remained 

 constant so long as C and D were undisturbed. It was about 

 50°. The plates B were fastened to two uprights a b, pro- 

 jecting from a round disk of tin plate, dotted in the figure, 

 resting on the side of the box. From the centre of the disk 

 a spindle passed through the side of the box carrying a 

 pointer c, shown in fig. 2, by whose means the plates could 

 be rotated about the spindle and their angular position read 

 oft". The instrument was roughly made and had various 

 defects. The Nicol and Iceland spar I got from England, 

 but the rest was made either by myself or by such workmen 



