Polarization of Shy Light. 91 



as I could find in these Alpine villages. The sensitiveness of 

 the polariscope was no doubt seriously diminished by the 

 insertion of the mirroi' E, which was a piece of ordinary 

 looking-glass. Of course it ought to have come between the 

 Nicol F and the eye. To add to the difficulties of reading, 

 there were some large dents on the faces of the Iceland spar. 



These things must be my excuse for not reaching the same 

 degree of accuracy in reading as others have. Still I seldom 

 found two readings of the same thing differ by more than 2°, 

 which corresponds roughly to a change of 0"02 in the ratio r. 

 For each observation I took four readings of the angle of 

 incidence, two on one side of the zero and two on the other. 

 So the probable error of the mean is certainly not greater 

 than 1^. The object of separating the glass plates into two 

 piles was to increase the power of the instrument by diminish- 

 ing the amount of transmitted reflected light. Thus, for an 

 angle of incidence of 50° on the five plates, my arrangement 

 gave r = 0*25, while eight plates in one pile would for the 

 same angle of incidence only have given ?'=0"34. These 

 values are calculated, 'not observed. The plate of Iceland 

 spar was about 23 millim. thick. This great thickness was 

 of importance, bringing the four black spots near together. 



It was necessary, not merely to be able to point the polari- 

 meter to the required part of the sky, but also to make the 

 plane of rotation of the glass plates pass through the sun. 

 This was secured by the mounting shown in fig. 2. The 

 polarimeter was fastened with a screw-bolt G to a board, 

 hinged at HH to another board, itself fastened with the 

 screw-bolt K to the head of a tripod belonging to a photo- 

 graphic camera. By rotations about K and H it was easy to 

 point the edge ed to the sun. Then, by turning the polari- 

 meter about the bolt Gr, observations could be taken of any 

 point of the sky distant 90° from the sun. With the same 

 mounting I could have observed the polarization at any point 

 of the vertical plane through the sun. But I have not done 

 so hitherto, chiefly because it would have required accurate 

 pointing of the instrument and another graduated circle. Since 

 the polarization reaches a maximum about 90 degrees from the 

 sun, an error of two or three degrees in the pointing is of no 

 consequence there. Figure 2 shows how the position of the 

 Nicol, and therefore of the mirror E, added much to the con- 

 venience of the observer as well as to the simplicity of con- 

 struction, though it was prejudicial to the sensitiveness of 

 observation. 



