174 



DEFINITIONS OF DISTURBANCE. 



For illustration, take the case where numerical increases of 1' in D and of lOy in H and in V occur at 

 each of the two stations Kew and Christchurch. We have for 1902-3 : 



As a matter of fact, in the calculations I took the Meridian of Kew, not that of Greenwich, as the 

 yz plane, measuring I from it, but to the degree of accuracy aimed at that is immaterial. 



If one desired to obtain the components along the fundamental x, y, z axes of the components of the 

 disturbing force AN and AE towards geographical north and east respectively, this would be readily 

 effected by supposing D = in the formulae, and writing AN for AH and AE for H AD. 



90. A question calling for some consideration is : what is to be regarded as a disturbance, and how is 

 its magnitude and direction to be determined ? SABINE, whose work on Terrestrial Magnetism still 

 commands respect, regarded the value of an element at any particular instant as disturbed when it 

 departed from the mean value of the element at that time of day contributions from disturbed days 

 having been removed by more than a specified limiting value. This regards the departure from an 

 undisturbed mean value as the measure of a disturbance. The difficulty, of course, is to arrive at the 

 undisturbed mean value. SABINE'S method, theoretically considered, was to do this by a process of 

 sifting, rejecting first such individual values as departed notably from the mean derived from all, then 

 forming a new mean from the individual values retained and repeating the process, and so on. 



In practice the method would be very laborious if strictly followed. A serious difficulty is that the 

 amplitude, and sometimes the type, of the regular diurnal inequality varies largely throughout the year, 

 and that there is a large sun-spot influence on the amplitude at least, if not on the type. 



From the point of view of SABINE'S definition, the natural thing would be to take AD, AH, and AV as 

 given at any instant by the departures of D, H, and V respectively from mean undisturbed values 

 appropriate to the hour. But to make even the pretence of doing this satisfactorily it would be necessary 

 to have a knowledge of what might reasonably be regarded as normal or undisturbed values. To obtain 

 such normal values for the Antarctic seemed wholly impracticable, and to have obtained them even for 

 the co-operating stations would have required regular diurnal inequalities to have been formed appropriate 

 for the several months of the years 1902 and 1903. One could, of course, have derived diurnal inequalities 

 from the hourly observations on the international term days, but two days a month afford a very slender 

 basis for diurnal inequalities. 



In the case of the larger Antarctic disturbances comparatively little could be done, because one at least 

 of the traces was pretty sure to be off the sheet, and even if all three traces happened to be on, the 

 oscillatory movements were so large and rapid that it was difficult to assign the times with sufficient 

 accuracy. Most of the measurements that proved practicable referred to disturbances of the special type 

 dealt with in Chapter X. 



On considering the situation, I decided that the most hopeful course to pursue was to focus attention on 

 changes in the values of the magnetic elements occurring during comparatively short intervals, especially 

 on the commencing movements introducing magnetic storms. In dealing with changes of short duration 

 we are comparatively independent of anything but disturbance pure and simple, especially when the 

 changes occur at hours when the regular diurnal changes are slow, or at seasons when the regular diurnal 

 changes are small. This seems to be practically the same conclusion as that reached many years ago by 

 Dr. BALFOUK STEWART, when he gave his attention to what he called " peaks " and " hollows " in the 

 magnetic curves, i.e. turning-points separated by no long interval of time. 



