MAGNETIC DISTURBANCES AT KEW. 297 



where the phase angle varies largely with the season of the year, the arithmetic mean 

 value gives the better idea of the mean amplitude of the forces to which the Fourier 

 " wave " in question is due. A very close accordance between the two sets of values 

 indicates that the phase angle is nearly constant throughout the year, with the 

 possible exception of months in which the " wave " in question is relatively small. 

 Another pretty safe inference in such a case is that the monthly values have not 

 suffered seriously from defects in the observational data. 



With the exception of Cj in H and I, the arithmetic mean and the yearly inequality 

 values for c t and c s in Table XIII. differ but little. There seems to be a considerable 

 seasonal variation in a : in H and I, but c^ is so small in some winter months that 

 more than usual uncertainty attaches to the values for a^ at this season. The 

 difference between the arithmetic mean and yearly inequality values for c s and c 4 is 

 more conspicuous, and probably reflects uncertainty in the monthly data rather than 

 true seasonal variability in the phase angles. There is, in fact, indication of but a 

 small seasonal variation in the values obtained for a 3 and 4 in the case of H. 



The values obtained for c^ and c. 2 in V from the smaller number of more highly 

 disturbed days are much larger than those derived from all the disturbed days. 

 There is, however, no such enhancement in the winter value of c^ in H, and the 

 results for c 3 and c 4 are in this respect somewhat conflicting in all the elements. The 

 8- and 6-hour waves may in reality be but little influenced by disturbance. Another 

 possible explanation, however, is that the greater irregularities existing in the data 

 from the smaller number of days tended to neutralise the increased amplitude of 

 disturbance. It is the latter explanation presumably that applies to the winter value 

 of <?! in H. 



21. Table XIV. gives the ratios borne by the amplitudes of the 12-, 8-, and 

 6-hour waves to that of the 24-hour wave. Older results for D are included. In D 

 and I the disturbed day results are based on the total number of disturbed days. The 

 relative importance of the shorter period waves tends to diminish on disturbed days, 

 especially in V. The 12-hour term in H presents, however, rather a marked exception 

 to this rule, and the same is true to a lesser extent in I. 



The great reduction in the relative importance of the shorter period waves in V 

 arises from the enormous influence of disturbance in increasing the amplitude of the 

 24-hour wave in that element. If we calculate the ratio which the amplitude of the 

 24-hour wave from all disturbed days in the mean diurnal inequality for the year 

 bears to the corresponding amplitude from quiet days, we find it to be 2 '18 in D, 

 1-59 in I, and 1-52 in H, but 5 "25 in V. 



Large as is the value of c t in V derived from all the disturbed days, it is exceeded 

 by 40 per cent, by the corresponding value derived from the sunspot maximum 

 years. 



It would be interesting to know the corresponding sunspot effect on quiet days, 

 but this unfortunately was not determined in (A). 



VOL, COX, A, 2 Q 



