CHRISTCHUKCH TERM-DAY OI5SKKYATIONS. 233 



currents or electric-tram currents, magnets \vhieh are properly damped do not get into vibration, and their 

 position at any instant except during specially violent magnetic storms may be regarded as one of 

 equilibrium, answering to the instantaneous value of tho magnetic tit-l<l. 



9. Returning now to the term-day data \vr lind very different types of change. Taking, for instance, 

 the Decl iiintioii changes during the term hours of January 15, February 1, and February 15, 1903 (see 

 p. 230), we have a persistent steady rise in Declination from beginning to end of the hour. On the other 

 hand, during the term hours of July 15 and September 15, 1902, p. 229, the Declination went on oscillating 

 through a narrow range without any marked drift in one direction. Then, on August 1, 1902, we have 

 the magnet not budging by as much as 0'- 1 from the position it occupied at the beginning of the hour. 

 A consideration of these varied phenomena suggested that account should be taken both of the number of 

 changes and of the range for each hour. 



Tables HA to IVA, pp. 239-241, summarise the results from the readings at 1-minute intervals as to the 

 number of changes in each hour of each term day as derived from the corrected values of the elements, 0' 1 

 being the minimum change for 1), and ly the minimum for II and V. In preparing these tables, allowance 

 had to be made for the gaps in the record. This was done on the assumption that changes took place 

 during an interval for which data were lacking at the same average rate as during that part of the same 

 hour for which observations existed. It is these corrected totals that appear in the tables. To show the 

 comparatively trifling character of the uncertainties due to lack of record, the total number of additional 

 changes thus introduced during each day is given in brackets in the last column of each table. With a 

 view to obtaining the diurnal variation in the hourly number of changes, it was also necessary to interpolate 

 values for a few hours e.g. the last two hours of September 15 for which no data existed. These 

 interpolated values enclosed in brackets were arrived at by assuming that the ratio which they bore 

 to the total from the remaining hours of the day was the same as for the average day whose records were 

 complete. 



10. Tables VA and VIA, pp. 242-243, summarise the results for the range of Declination and Horizontal 

 Force throughout each hour of each term day. As a rule, no allowance was made for gaps in the record. 

 This was partly because no one guiding principle could lie applied, but mainly because there was usually 

 reason to believe that the loss of record had had little, if any, influence on the result. An allowance was, 

 however, made when a gap commencing towards the end of an hour extended into the next hour, and the 

 earliest recorded reading of the second hour lay outside the range covered by the readings during the 

 complete part of the first hour. In such an instance the element was assumed to have increased or 

 diminished, as the case might be, at a uniform rate whilst the trace was lacking. Values were also 

 interpolated for a few hours e.g. the last three hours of September 15 for which readings were totally 

 or largely wanting. These interpolated values are enclosed in brackets ; they were arrived at in a 

 similar fashion to the interpolated values in Tables HA to IVA. The two last columns in Tables VA and 

 VIA give respectively the arithmetical sum of the 24 hourly ranges and the absolute range, i.e. the difference 

 between the algebraically greatest and least of the readings obtained at 1-minute intervals throughout the 

 day. These absolute ranges are necessarily never less and are usually larger than the ranges on pp. 177 

 and 178 of the "Physical Observations," which are derived from hourly readings, but the difference is 

 in no case large. On September 15, 1902, and January 15, 1903, there was reason to believe that one of 

 the extreme readings for the day was lost through a break in the readings, so a " + " is put after the rarige 

 as deduced from the readings recorded. 



11. If we examine Tables HA and IIlA in detail, we observe a great variability from day to day in the 

 number of changes. On March 15, 1903, there were 426 changes in the Declination reading, or an average 

 of about 18 an hour, though the total range for the day was only 9'- 2. On December 1, 1902, and 

 February 1, 1903, with the closely similar absolute ranges of 9' '7 and 8' -9, the numbers of changes were 

 respectively only 244 and 181. The least number of daily changes averaging about 5 per hour was 

 recorded on April 15, 1902, although the absolute range for the day exceeded that on the nine immediately 

 subsequent term days. The greatest number of Declination changes recorded in an hour, viz. 41 on 

 March 1, 1903, took place in an early morning hour, for which the normal change due to the regular 

 diurnal inequality was under 0'-2. There are 13 hours in Table IlA during which no change in D was 



2 H 



