JEtlier " Drift " and Rotary Polarization. 393 



the axial adjustment of the system and the beam of sunlight, 

 as, if the latter were not perfectly uniform in intensity over 

 its section, a rotation of the trough would show a spurious 

 effect in the half-shade, destroying the match. This difficulty 

 was many times greater than would have been experienced 

 with an artificial source which could have moved with the 

 system ; but the increased light allowed a much smaller 

 azimuth of the strip and a consequent increase in the sensi- 

 bility. It was found that the light, when it reached the 

 analyser, was slightly depolarized; and I sought to correct 

 this with a thin mica compensator, in the way I have already 

 described *, placed over the end of the cell S ; but this did 

 not eliminate it, showing that the light was irregularly de- 

 polarized. The compensator was therefore dispensed with. 

 A portion of this residual light came from diffusion, by the 

 liquid, of the intense initial beam of sunlight within the same. 

 The rest was . due to imperfect compensation of all the rays 

 arising from differences in density. The disturbing effect on 

 the field of this residual light was, however, small, and did 

 not interfere seriously with the settings for a match. At the 

 beginning of a set of observations, the two halves of the field 

 might show complementary colours on rotating the polarizer, 

 indicating inequality between the combined paths. The arm c 

 was then shifted until the complementary colours disappeared 

 on rotating P in both directions, the entire field having a 

 bright orange hue. 



To avoid magnetic effects due to the earth's field, the trough 

 was placed normal to the mean magnetic meridian. A com- 

 pass, mounted in a slide upon the edge of the trough, could 

 be set at equal intervals along its length and the position of 

 the needle noted. This was found to vary by several degrees. 

 The mean for the various positions was taken and found to be 

 21° N.E., so that the effect of the drift could be reduced only 

 a few per cent. The trough was then set at right angles and 

 always reversed in this direction. 



Much difficulty was experienced on account of the ingress 

 of the oil to the nicols and other optical pieces. On this 

 account, and also because the oil, if left in contact with the 

 air for some time, turns a yellowish colour, the trough was 

 always filled and emptied at each observation. The residue 

 on the mirrors and optical pieces also produced a scattering 

 of the light, reducing the final sensibility perhaps by a half ; 

 so that all pieces had to be dismounted, cleaned, and re- 

 cemented, before remounting and refilling the trough for a 



* Phil. Ma^. /. c. 



