ON COMPARING AND REDUCING MAGNETIC OBSERVATIONS. 87 



non-cyclic increment in the midnight to midnight interval. There is 

 apparently considerable variation of value in each element in different 

 months, but, taking the elements separately, a certain order appears. In 

 declination there are nine months in which the values of p, m and /suc- 

 cessively increase ; these are January to April, June and July, and 

 October to December. And the value of p is in ten months of the year 

 negative, whilst that of / is in eleven months positive. Thus it would 

 appear that the selected quiet days on the whole fall at a time of 

 transition from a negative to a positive non-cyclic increment, indicating 

 change from a decreasing to an increasing value of declination, represented 

 in the general mean by values of jt)= — 0'-26, m= — 0'-02,and/= -F0'-30. 



In horizontal force there are three months in which the values of />, 

 m and /successively increase in value; these are April, July and August ; 

 and there are four months in which the values successively decrease ; 

 these are January, March, May and December. In the remaining five 

 months — February, June, September, October and November — the value 

 of m is greater than those either of jo or/, which may indicate a turning- 

 point in the value of the element. But there is altogether a preponder- 

 ance of positive values, showing that the selected quiet days, on the whole, 

 distinctly fall at times of increasing horizontal force. 



In vertical force there are three months in which the values of p, m 

 and / successively increase — these are March, May and October ; and 

 there are two months in which they successively decrease — these are 

 February and June. There are five months — January, April, July, 

 August and September — in which the value of m is greater than those of 

 either j3 or/, and two months — November and December — in which the 

 value of in is less than those of either p or/, which again may indicate 

 turning-points in the value of the element. There is, however, an entire 

 preponderance of negative values, showing that the selected quiet days 

 fall altogether at times of decreasing vertical force. 



All this, it will be understood, refers only to Greenwich. Further, 

 if the cases of progressive increase and progressive decrease of value, of 

 noa-cyclic variation (that is of the values of p, m and /in Table X.) of 

 which there are nine months out of twelve in declination, seven months 

 out of twelve in horizontal force, and five months out of twelve in vertical 

 force, can be taken as representing a real condition of things, and are not 

 due to any haphazard circumstances, it would justify to a certain extent 

 the custom of making quiet day non-cyclic diurnal inequalities cyclic, by 

 assuming uniform increase or decrease of the non-cyclic element through 

 the twenty-four hours. 



The values of p, m and /in Table X. for years support generally the 

 conclusions derived from the values for months. 



In Table XI. comparison of diurnal range in declination and horizontal 

 and vertical force, as variously determined, is made for Greenwich, em- 

 ploying the results for the eight years 1889 to 1896. It is compared as 

 found from the quiet day results (both as observed and corrected for non- 

 cyclic variation) and from the all day results (which require no non-cyclic 

 correction). The difierences between the observed and corrected quiet day 

 ranges, a — b, are generally small in declination and vertical force, but in 

 horizontal force they are larger. As regards declination and horizontal 

 force they have much the same order of magnitude as those for Greenwich 

 appearing in Table VI. (in the comparison for a smaller number of years 

 with Kew). In the comparison of diurnal range found from quiet days 



