MAGNETIC D1STUEBANCES AT KEW. 



275 



The amount of disturbance during the 209 disturbed days of 1890 to 1900 was 

 probably somewhat below the average for the 39 days classed as " 2 " at either Kew 

 or Greenwich during the three years 1906 to 1908, but the difference was not 

 great, and if data similar to those for the last three years had been in existence for 

 1890 to 1900, they would undoubtedly have disclosed a similar preponderance of 

 disturbed days. 



If each magnetic storm lasted a day, commencing at, say, Oh. Om. a.m. and 

 terminating exactly 24 hours later, and if the days immediately preceding were 

 invariably free from oscillatory movements, one might, with some degree of assurance, 

 associate with a storm the non-cyclic change observed subsequently. But even then 

 there would be the possibility that during quiet days the earth's magnetism tended 

 to rise above or fall below a stable value, and that the magnetic storm simply played 

 the part of a mechanical shock, whose effects depended partly on the length of the 

 preceding quiet interval. These remarks will, I hope, explain the way in which the 

 non-cyclic results are presented. 



3. Table I. shows the mean non-cyclic increments, the 209 days being distributed 

 amongst the months of the year to which they belong. The first line gives the 

 number of disturbed days in each month. Results for D, already published, are 

 included, but are measured in units of force instead of in angular measure as 



TABLE I. Mean Values of the Non-cyclic Daily Changes. 



previously. Taking for H the mean value at Kew between 1890 and 1900, an 

 increase of 1' in westerly declination means a force of 5'32y where ly = '00001 

 C.G.S. directed to magnetic west. The D, H, and V results are thus all expressed 

 in terms of ly as unit ; they are derived directly from the curve measurements. The 

 data for I are calculated from the corresponding data for H and V ; they are 

 expressed, however, in angular measure. The results for H and I under August 

 include August 20, 1904. If that day is excluded, the August values become for H 

 24'0y and for I + l''87 ; these are the values which had actually to be employed 

 when dealing with the diurnal inequalities. 



2 N 2 



