16 



PKOCEEDINGS OF THE AMERICAN ACADEMY 



tang^ 25° = .21. As each of the following observations is the mean 

 of four, the probable error is reduced one-half. 



The first series of observations were made on the light of the sky. 

 The instrument was screwed into a post and levelled, the altitude and 

 azimuth of the sun taken, and the instrument then directed towards the 

 points to be observed. Most of these were in the same vertical plane 

 with the sun, so that it was only necessary to determine their altitudes. 

 The line of junction was then brought parallel to the plane of polar- 

 ization ; that is, turned until it was vertical, since it then lay in the 

 plane passing tlirough the sun. The four positions of the Nicol's prism, 

 in which the two images were equally bright, were then observed, read- 

 ing the angles to tenths of a degree, and taking the mean. The per- 

 centage of polarization was finally obtained from Table IX. In 

 Table X., series 1 to 9 were taken at Waterville, N. H., in a valley at 

 a height of about 1500 feet, surrounded by mountains about 4000 feet 

 high. The air there may be regarded as very pure. Series 10 and 11 

 were taken upon the top of the building of the Institute of Technology, 

 Boston. The first column in each case gives the altitude of the point 

 observed, the second its distance from the sun, and the third the polai'- 

 ization corresponding to the mean of the four observations. It soon 

 became evident that the polarization depended on the solar distance of 

 the point under observation, and not on the altitude. This is more 

 evident from Fig. 5, in which abscissas give the solar distances and 

 ordinates the polarization of the above points. 



TABLE X. 

 Sky Polarization. 



