ON COMPARISON AND REDUCTION OF MAGNETIC OBSERVATIONS. 219 



For the mean of the years 1890-94 we have 

 H=-18211 C.G.S. units; 

 ■whence 



cY = 530xlO-'x<;D (5) 



where iT) is supposed to be given in minutes of arc. 



Resolution in and perpendicular to the magnetic meridian is some- 

 what inconvenient when the general features of terrestrial magnetism are 

 being considered, owing to the complicated relationships between the 

 magnetic meridians at different stations. Even in dealing with the 

 phenomena extending over a long interval of time at a single .station, the 

 reference to a set of axes whose position is continually altering has its 

 disadvantages. When, however, the components of a force along two 

 rectangular axes are known, the components along any other pair in the 

 same plane are easily deduced when the inclination of the old to the new 

 axes is given. It will thus suffice to state that for the epoch 1890-9-1 the 

 mean magnetic meridian at Kew lay at 17° 36'-2 to the west of the 

 astronomical meridian. 



§ 12. Instead of giving explicitly the components in and perpendicular 

 to the astronomical meridian I propose to consider the magnitude and 

 direction of the resultant horizontal disturbing force itself. Denoting its 

 intensity by p, its inclination to the east of magnetic north by 4', we 

 have to a sufficiently close degree of approximation 



,=v/((^X)2 + (aY)2=2Hsec ;A (6) 



j^=tan-'(-aY/^X)=tan-'(-<D.H/^H) ... (7) 



Remembering that the inclination of p to the astronomical north is 

 \I/ — (17° 36''2), it would be found a simple matter to deduce the com- 

 ponents of the disturbing force along and perpendicular to the astro- 

 nomical meridian from the calculated values of /< and xp. 



The following table, VII., gives the values of n and 4j for the mean of 

 December and January, or midwinter, the mean of June and July, or 

 midsummer, and the mean for the year at each hour. The greatest and 

 least values of p in each case are in heavy type. 



In comparing Tables IV. and VII. the reader will require to notice that 

 the former refers to lO** cH, the latter to 10^^>. Table VII. was got out 

 from arithmetical means proceeding to two places beyond those retained 

 in Table IV., and an extra figure was retained in i>, so that the arith- 

 metical accuracy of any resolution of p into orthogonal coordinates should 

 Vie as trustworthy as its previous resolution into cX and fY, i.e., into cH. 

 and HoD. 



The angle 360° is to be added to the values of d' under the hour 23 for 

 'midsummer,' and for hours subsequent to 18 in 'midwinter ' and ' year,' 

 when it is desired to make out the true increase of (// relative to a previous 

 hour. 



In the case of the yearly mean the vector p has a continuously pro- 

 gressive rotation from east to west through south, like the sun as seen 

 from the earth. This is the general character of the rotation at any 

 season ; but at midwinter there is an unmistakable retrograde motion in 

 the early morning for several hours, and at midsummer there is at least 

 a suspicion of a retrograde motion from 8 to 10 p.m. From 10 a.m. to 



