70 



so on. By thus proceeding in the case of the Declination at St. He- 

 lena, we have sixty differences thus accruing in the five years of 

 hourly observation, by which we find that the monthly increase of 

 West Declination during these five years amounted on the average 

 to 0'-657, or to an annual increase of 7'' 88. 



" It is not however necessary for this investigation that the system 

 of observation should be hourly : a much less onerous system is suf- 

 ficient, provided that the observations be distributed equably through 

 the year, and that the intervals between the observations of each 

 day be, approximately at least, equidistant. Before the commence- 

 ment of the hourly series there had been fifteen months of two- 

 hourly observations, and after its close the observations were con- 

 tinued for twenty-one months more at five hours of each day, the 

 hours being such as to give by their combination a true mean value 

 for each day. We are thus enabled to take in a more extended period, 

 amounting to ninety-six consecutive months, or eight years, from 

 which to derive the average rate of secular change at St. Helena. 

 Proceeding as before, we find for this period an average rate of O' 1 661 

 for the increase of West Declination in a month, or an annual in- 

 crease of 7'' 93 in a solar year. During these eight years the hori- 

 zontal magnetic direction at St. Helena had consequently changed 

 altogether rather more than one degree. 



" When the number of years are few from which an annual average 

 rate of secular change is derived, it is necessary to be particular in 

 regard to the regular distribution of the observations as to months 

 and hours, because observations made at one time of the year or at 

 one hour of the day, are not strictly comparable with those made at 

 other times of the year or at other hours of the day, unless indeed 

 corrections based on a long series of observations at the same spot 

 or in its vicinity are applied for the annual and diurnal variations. 

 But when the periods of comparison include intervals of consider- 

 able length, the comparative influence of the annual and diurnal 

 variations is greatly diminished, and, if the comparison extend over 

 a great number of years, it may practically be disregarded. Now, St. 

 Helena being a naval station, and frequently visited by navigators 

 of our own and other countries, who have had the requisite knowledge 

 and have been at the pains to take the necessary precautions to 

 make trustworthy observations, we are able to collect from the nar- 



