48 



DISCUSSION OF T II K II O K 1 X O X T A 1. C' O M PONE N T 



variation are reached. Owing to the prominent annual variation near 2 P. M., the 

 range of the diurnal variation between the morning minimum at 11 A. M. and the 

 afternoon maximum at 3| P. M. is of more interest in the discussion of the diurnal 

 fluctuation of the horizontal force than the 6 A. M. and 1 1 A. M. range, which 

 latter range, as we have seen, is slightly greater than the first one. 



To find the turning epochs of the annual variation, the monthly values for the 

 hours 9 A. M. and 2 P. M., when it is best developed, were taken from Table IV., 

 and each value was again compared with its annual mean. 



Casting the eye over the columns headed "differences," we see by the change of 

 sign and the magnitude of the values that the transition from a positive to a nega- 

 tive value occurs some time after the equinoxes, and that the maximum variation 

 is reached about the time of the solstices a result in close correspondence with the 

 conclusions reached in the discussion of the annual inequality in the diurnal varia- 

 tion of the declination (Part II. of the discussion). For convenience in the analy- 

 tical treatment, a column headed "mean difference" has been added to Table V., 

 obtained by changing the signs of the 2 P. M. differences (the annual variation 

 being then opposite to the morning values), and taking the mean of the 9 A. M. 

 and 2 P. M. differences. The values in this column are tolerably well represented 

 by the following formula: 



A a = +0.00129 tin (6 + 79) + 0.00018 *m (20 + 191), 



the angle 6 counting from January 1, at the rate of 30 a month. Accordingly, we 

 find the transition to take place shortly before the middle of April and October, or 

 about twenty-two days after the equinoxes. This is about twelve days later than 

 the epoch found in Part II. for the declination. 



Analysis of the Solar-Diurnal Variation of the Horizontal Force. For convenience 

 of investigation and proper comparison with similar results at other localities, the 

 values given in Table I. have been put in an analytical form, and are represented 

 by the following expressions. It will be seen that the difference between any 

 monthly normal mean and the corresponding mean in Table V. of Part IV., which 

 latter mean is affected with the disturbances, does not exceed 2^ scale divisions. 

 Tins small difference includes also a small effect due to the necessity of different 



