44 



Art. 9.— T. Terada : 



Table YII. 



Thirdly, taking the records of 1913 only, those cases were 

 chosen in which the phase difference of the two horizontal 

 components was eitlier o or nearly t:, i.e. in which the periodic 

 distnrbancc is "polarized'' in a certain azimuth, and the ratio 

 of the amplitudes was carefully determined which corresponds 

 to the tangent of the azimuthal angle «. The latter angle was 

 counted from N, positive value taken toward W. The results, 

 if plotted in a diagram, witli the hours as abscissa, regardless of 



Fig. 16. 



Co' 



20' 



-10 



-40' 



-60' 



