MAGNETIC DISTURBANCES AT KEW. 279 



Dr. VAN BEMMELEN,* who has made a special examination of the phenomena at a 

 number of stations, found in most cases at least a trace of an oscillatory movement at 

 the start, but the position at the end of 5 or 10 minutes as pointed out long ago by 

 W. ELList is usually a marked rise in H, as at Kew. At the Winter Quarters of 

 the " Discovery, "J in the Antarctic, the commencing movement appeared usually to 

 be highly oscillatory, both movements being large, but the first of the two was 

 sometimes a rise, sometimes a fall. These commencing movements are experienced, 

 apparently simultaneously, all over the world, and for their elucidation we must look 

 to the inter-comparison of data from a variety of stations. They are referred to here 

 because a physical connection between them and the non-cyclic phenomena is not 

 improbable. There is also occasionally a very direct but " accidental " connection, 

 which may best be illustrated by reference to two individual cases. On November 4, 

 1892, a sudden commencing movement took place between 2 and 3 a.m., the immediate 

 conseqxience of which was a rise of over 30y in H in the course of a few minutes. 

 The non-cyclic change on this day had first to wipe out this rise, but notwithstanding 

 it left the value of H at the second midnight lower by 63y than at the first midnight. 

 On the other hand, on March 12, 1892, when the non-cyclic change during the 

 24 hours was a fall of no less than 209y, the value at the first midnight was enhanced 

 by at least 95y above the normal owing to the sudden commencing movement, which 

 took place between 10 and 11 p.m. on the llth. March 12, 1892, further illustrates 

 in a remarkable way the important part that may be played by the mere accident of 

 time. The enhancement of H, following the sudden commencement of this storm, 

 continued only for a few minutes after a.m., and was then followed by a rapid fall. 

 If the non-cyclic change had been taken for the 24 hours beginning at 0'15 a.m., it 

 would not have amounted to a quarter of that actually recorded. 



Diurnal Variation. Absolute Itanges. 



^ 



7. The usual aim when deriving diurnal inequalities is to obtain results repre- 

 sentative of average normal conditions. In the case of Terrestrial Magnetism there is 

 the complication that the amplitude of the regular diurnal inequality alters markedly 

 with sunspot frequency. Highly disturbed days present, however, more outstanding 

 sources of uncertainty. Let us suppose for a moment what is not really true that 

 the value of, say, H at a particular hour is on disturbed days purely "accidental" so 

 far as local time is considered. The diurnal inequality resulting from an infinite 

 number of disturbed days would then vanish. But for a finite number of days a 

 pseudo-inequality would naturally present itself, and its range would tend to increase 

 as the number of disturbed days was reduced. If we came down to a single highly 



* 'K. Akademie van Wetenschappen te Amsterdam,' 1908, p. 773. 



t ' E. A.S. Notices, 1 vol. 65, p. 520. 



| National Antarctic Expedition, 1901-1904, "Magnetic Observations," p. 179. 



