AT KEW OBSERVATORY, 1898 TO 1912. 



155 



phenomenon. In all the seasons, especially summer and equinox, the phase angle is 

 decidedly larger in the second period, but the seasonal variation in the phase angle 

 is markedly less in that period. . The data from 1862 to 1864 discussed by EVERETT 

 made the 24-hour wave much larger relative to the 12-hour wave than the 1898 to 

 1904 data did. Thus the 1905 to 1912 data show no progressive change, but rather 

 a reversion. 



TABLE VIII. Comparison of Periods 1898 to 1904 and 1905 to 1912. 



In E 1( while recognising a considerable " accidental " element iu individual monthly 

 values of amplitude and phase angle in the 8-hour and 6-hour waves, I concluded that 

 the general features were genuine. The results from the two periods in Table VIII. 

 show a closer resemblance than I had ventured to hope. The accordance in the phase 

 angles is truly remarkable, considering that the time equivalent of 1 is only 

 lj- minutes for a 3 and 1 minute for a,. The 8-hour like the 24-hour term shows a 

 well-marked seasonal variation, both in amplitude and phase angle ; but unlike the 

 24-hour term it has the phase angle largest in summer. The 6-hour term agrees with 

 the 12-hour term in having the phase angle least in winter, and in showing 

 comparatively small seasonal variation of amplitude. 



13. To ascertain exactly how the difference between the results for the 24-hour 

 wave from the two periods came in, I calculated the 24- and 12-hour Fourier coefficients 

 for each individual year from 1898 to 1912. The results appear in Table IX. Fewer 

 figures are retained than in the previous tables because the diurnal inequalities for 

 individual years had been calculated only to O'l volt. The 24-hour term data fluctuate 

 much more from year to year than do the 12-hour term data. The 1911 value of Cj is 

 fully five times that of 1901, while the extreme values of a, differ by 126, 

 representing 8 g hours of time. It is the earliest four years that are mainly responsible 

 for the smallness of i for the epoch 1898 to 1904. There is, however, nothing 



x 2 



