100 Mr. J. C. McConnel on the 



Standardization. 



As has appeared frequently in what precedes, a great diffi- 

 culty in these observations is to translate the readings of the 

 pile into absolute measure. It is impossible at present to do 

 this satisfactorily by calculation, for we do not know the laws 

 of the reflexion of polarized light at the surface of glass with 

 sufficient accuracy. Each instrument must therefore be 

 standardized by direct experiment. Three methods have been 

 proposed for doing this : — The first, by the aid of photography, 

 was suggested to me by Capt. Abney, who has used platino- 

 type paper for measuring the brightness of the sun and sky. 

 It consists in exposing sensitive paper to the light from a small 

 portion of the sky for measured intervals of time, interposing 

 a Nicol prism in the path of the light. By placing the Nicol 

 in the two rectangular positions, we could determine the ratio 

 of the principal intensities, while simultaneous observations 

 were being made with the optical polarimeter. Difficulties 

 might occur in the light reflected from the sides of the Nicol 

 and in the great length of exposure, and at any rate the 

 method would require a good deal of preliminary testing. 

 Bosanquet has proposed another method, viz. to mix common 

 and completely polarized light in known proportions with the 

 aid of a divided object-glass (Phil. Mag. Dec. 1875), and then 

 take the reading with the polarimeter. 



The simplest and most convenient method, and with proper 

 precautions the most accurate, seems to be that invented by 

 Arago and already described as being used by Rubenson, 

 and it is the one I have adopted. Bosanquet throws some 

 suspicion on the method as resting on an uncertain theoretical 

 basis, but this suspicion is by no means deserved. The main 

 assumption involved is that a beam of plane polarized light 

 may be treated as composed of two beams, polarized in rect- 

 angular planes, of intensities proportional to sin^ (f) and cos^ 0, 

 where (f) is the angle between one of these planes and the 

 original plane of polarization. This is a direct consequence 

 of the principle of the superposition of small motions, which 

 is universally admitted to be applicable to light. This assump- 

 tion granted, we have only to add that in the thick plate of 

 quartz one component is retarded with respect to the other by 

 a large number of wave-lengths, and that, therefore, after 

 leaving the quartz there is in homogeneous light no permanent 

 phase-relationship between the components. The emergent 

 light, then, is partially polarized in the ^^rincipal plane of the 

 quartz, and the ratio of the principal intensities is tan^ (fj. A 

 certain amount of light is, of course, lost by reflexion at each 



