MAGNETIC DISTURBANCES AT KEW. 30 & 



Tliis would make the standard of disturbance lower than in the present paper, and 

 lower than the average interpretation of the international scale " 2." If, however, 

 only 1 or 2 days were taken from each month, a very large number of years' data 

 would have to accumulate before anything like a reasonably smooth diurnal inequality 

 could be deduced for individual month's of the year, and this even when years of 

 many and years of few sunspots were grouped indiscriminately. Taking even as 

 many as 5 days a month, it would probably be necessary to combine all the months 

 of one season to derive moderately smooth inequalities from the data of one year. 

 But the accumulated data of an 11 -year period would probably suffice to give 

 tolerably smooth results for individual months, treating years of many and of few 

 sunspots separately. 



In some months, no doubt, to get as many as 5 days, one would have to include 

 days which did not attain to any very high standard of disturbance, but it must be 

 remembered that many of the days which are classed " 1 " on the international 

 scale are so nearly of standard " 2 " that considerable divergence of view as to 

 their classification may exist even at stations so near together as Greenwich 

 and Kew. 



33. In connection with highly disturbed days, there is the difficulty that 

 disturbance occasionally leads to loss of trace, and at some stations this is rather a 

 frequent occurrence. To show clearly the smaller movements on quiet and ordinary 

 days a high degree of sensitiveness is desirable, and this is hardly compatible 

 with including in the width of the photographic sheet the range of the largest 

 disturbances. To obviate this, the magnets in some magnetographs carry more than 

 one mirror, so that when the light from one gets beyond the edge of the photographic- 

 sheet, light from another may remain on. This, however, intixxluces some compli- 

 cation, and at times of large rapidly oscillatory disturbance confusion may be 

 introduced. A better plan, if financial conditions allow, is to have two magneto- 

 graphs, one sufficiently insensitive to preclude loss of trace under any contingency 

 that can be reasonably anticipated. 



34. In the case of highly disturbed curves, I do not think that ordinary smoothing 

 by hand is at all likely to prove satisfactory. An hourly mean derived from the area 

 between the curve and base line and the limiting ordinates may be more satisfactory, 

 but during rapidly oscillating disturbances this would be a difficult method to apply. 

 In the case of the " Discovery " Antarctic curves, ordiuates were measured at the hour 

 and at 20 minutes before and after, and the mean of the three accepted. With the 

 aid of a suitable scale this proved a comparatively short process, and it unquestionably 

 produced a remarkable, smoothing of the results. By adopting some similar method 

 it would be possible to obtain fairly smooth inequalities from a comparatively small 

 number of disturbed days. I hope the opportunity may be found at Kew to ascertain 

 the effect, if any, of the smoothing process adopted there in the case of ordinary day 

 curves. 



VOL. ccx. A. 2 R 



