MAGNETIC DECLINATION NEAR THE MAGNETIC EQUATOR. 675 



ence of an inflexion or slight minimum near noon in the months about December, 

 seem to show that Trevandrum belongs more properly to the northern hemisphere 

 (to which it belongs geographically), than to the southern hemisphere, to which 

 it belongs magnetically. 



In comparing the curve of February with that of April, and that of Septem- 

 ber with the curve of November, it would seem probable, that at some other epoch 

 than the middle of the months of March and October, the amount of the range 

 would be still less than that here shown. A consideration of monthly mean curves, 

 corresponding to each week of the year, shows that this is the case ; but as the 

 date of minimum movement is not the same in each year, a combination of the 

 means at the same date in different years will not show the smallest amount 

 of the movement. I have, however, projected (Plate XL.) the mean curve cor- 

 responding to the middle of March 1864, which shows the mean movement 

 limited to about one-third of a minute of arc (the range of the hourly observa- 

 tions is 0'-35). 



We have thus a very near approximation to an extinction of movement in the 

 mean curve for the month of March 1864 ; and if the mean curve represented the 

 usual daily movement, we might say that we have here a case in which the 

 needle remains stationary night and day. This is, however, not the fact ; the 

 result is to a considerable extent arithmetical, and is due, in part at least, to a 

 combination of different movements. 



In the third column (B) of Table II. I have given the means of all the 

 daily ranges for each month. It will be perceived from these, that though the 

 diurnal oscillation is still a minimum in March and October, yet it is a much 

 more marked oscillation than appears in the mean curves. In the year 1856, for 

 which the mean of the daily ranges is least, the mean of those for March is 1-51 

 and for October 1*76; these, though very small mean ranges, are still more 

 marked than we might expect from the mean curves. 



The fourth column of Table II. contains the ratio of the mean of the daily 

 ranges (B) to the range of the mean variation (A); this ratio is greatest in March and 

 October. In the five months of May to September it is nearly constant, or I'll ; 

 while, in the four months of November to February, it increases from 1*26 to 

 l - 72. From these ratios we may conclude that the diurnal variation obeys nearly 

 a constant law in the months of May to September ; while, in the other months, 

 either considerable differences occur in the times of maxima and minima, or the 

 law changes sign, as in March and October. 



Column five contains the hypothetical ranges of mean curves, when we suppose 

 the agreement of the range of the mean curve with the mean of the daily ranges 



T> 



to be the same in all the months, as in May to September, or that j = 1-11. 



In this case the ranges for December and January approach in magnitude those 

 for June and July. 



