520 On the Polarization of the Light of the Sky, 



Tied by actual measures ; and, again, we make the assumption 

 that the polarization of elliptically polarized light is determined 

 by the same laws as the polarization of partially polarized light — 

 an assumption which it would seem desirable to avoid in deter- 

 minations of an absolute character. Now the measure of the 

 ratio in which polarized light and common light are mixed can, 

 theoretically at least, be performed by a direct process which 

 does not involve any intermediate law, whether theoretical or 

 empirical. It is only necessary to provide two apertures which 

 shall admit polarized and common light in known propor- 

 tions, to superpose these lights, and determine by observation 

 the inclination of the glass bundle employed which will neu- 

 tralize the polarization. I have not constructed any such ar- 

 rangement ; I have been unable, in fact, as yet, to devote any 

 resources to the practical part of the question, in consequence of 

 the large demands made on me by the construction of instru- 

 ments in another branch of inquiry ; but I propose to describe 

 the outline of an arrangement for this purpose which might be 

 constructed. 



A simple lens of some feet focal distance is cut in half and 

 mounted in the manner of a heliometer ; the micrometer arrange- 

 ment is not required, only that the two images given may be 

 made to coincide and separate at pleasure. In front of each 

 half of the lens a screen is mounted, with an aperture the area 

 of which can be regulated by a micrometer-screw. A Nicol's 

 prism large enough to supply the whole of one of the apertures 

 with polarized light, is placed in front of it. A simple lens, or 

 Ramsdeir's eyepiece, may be employed to view the images. The 

 sun may be viewed through a dark glass or any ordinary source 

 of unpolarized light directly. The images are separated, their 

 brightness compared, and the movable apertures adjusted until 

 the brightness of the two images is the same. This furnishes 

 the values of the two apertures which admit the same amount of 

 light when the Nicol is in front of one. By the micrometer 

 arrangement either area can be altered in a known proportion, 

 taking care always to keep the aperture behind the Nicola good 

 deal smaller than that of the prism. The two images are then 

 superposed ; and a single image is thus formed, in which polarized 

 and common light are mixed in known proportions. We 

 have only to observe the inclination of the glass bundle of the 

 polarimeter required to compensate this polarization for varied 

 proportions of common and polarized light; and the instrument 

 will be graduated independently of instrumental data, except 

 such as can be verified at any time and independently of any 

 hypothesis. 



