1885.] Fluctuations of Declination at Kcw and Stonyhurst. 3b'9 



We do not know of a single instance in which the fluctuation is not 

 in the same direction at both observatories. 



We have given the G.M.T. of the commencement and end of each 

 fluctuation at each observatory. Practically speaking, the times at 

 both places are so nearly simultaneous that we do not feel justified in 

 asserting that they are not quite so. Occasionally, however, there 

 are indications that certain short period fluctuations are not precisely 

 of the same duration at both places. In what follows we have 

 rejected such cases ; also we have adopted the durations as recorded 

 at the Kew Observatory, rejecting however all cases when these are 

 less than five minutes, inasmuch as an accurate measure of duration 

 is essential to our method. 



Let us now, simply as a conjecture which may be of service in 

 indicating the best method of treating the observations of Table I, 

 suppose that in these disturbances two causes are in operation, and 

 that the result is due partly to true magnetic changes, and in part to 

 secondary currents caused by these changes. 



Let K denote the whole observed value of the disturbance at Kew, 

 and of this let h denote the portion due to strictly magnetic change, 

 also let txk<j)(f) be the portion of the whole disturbance caused by 

 secondary action, a. being a constant which may conceivably be either 

 positive or negative, and t denoting the duration. Hence K = 

 k(l + a(fr(t)). In like manner let S denote the whole Stonyhurst 

 change. 



We are, perhaps, justified in putting S = k(B + 70(0), /3 and 7 

 being constants. 



Hence we shall have Jl— /( ^^\ that is to say, ~ will be a 

 K 1 + *0(O J K 



function of the duration. 



S 



It thus appears that the value of — will, according to this or indeed 



according to any probable hypothesis of this nature, be independent 

 of the values of S and K, and be a simple function of the duration. 

 These reasons have induced us to construct the following table (II), 

 S 



in which the ratio — is ascertained for disturbances of varying dura- 

 tions. 



