Barrett — A Neiv Form of Polarimeter. 201 



In practice a difficulty occurs in determining the precise angle of 

 maximum polarization ; for the extinction of the reflected ray seems to spread 

 over a narrow region rather than occur at a definite point. This error can 

 be lessened by careful attention to the parallelism of the incident rays and 

 the homogeneity and intensity of the liglit. I have employed a small direct 

 vision-prism in the collimator and obtained a sharp spectrum, using, of 

 course, in this case a very brilliant source of white light. In this way I 

 hoped to obtain the angle of extinction for a definite colour, and thus see 

 a dark band pass across the spectrum, as the polarizing angle for each 

 colour was reached. But the result, as might be expected, was not very 

 successful in bodies of low dispersive powers; a faint, broad shadow is 

 observed to move across the spectrum, the best position to read being when 

 the shadow is in the green or greenish-blue : results can then be obtained 

 within 20' to 30', even with this preliminary apparatus. In the case of 

 bodies of very high dispersion, like the nitroso-dimetliyl aniline, as pointed 

 out by Prof. Wood in his excellent work on Optics, the determination of the 

 polarizing angle is easy, as the dark band is then sharp and well-defined. 

 If it were the rofation of the plane of the polarized light, and not the amount 

 of polarization, any of the methods used in saccharimetry could be employed 

 for accurately determining the polarizing angle. Possibly an eifect similar 

 to the half-shadow polarimeter might be obtained by using a divergent 

 incident beam and a split lens, with inclined axes, in the telescope. What 

 is needed is a comparative effect ; and for this purpose a double-image prism 

 in the telescope, instead of a Nicol, would be worth trying. 



SCIENT. PKOC. R.D.S., VOL. XII., NO. XIX. 2l 



