i 



Magnetic Declinatiox. 331 



Annual Period. 



An examination of the monthly means at the foot of Table I. will shew that though the amount of change 

 from year to year he nearly constant, this is not the case from month to month ; on the contrary, the north 

 extremity of the needle at times moves towards the west. It is not very evident, however, from these means, 

 whether the rate or sign of motion has any well-marked relation to the season of the year. In order to render 

 this more apparent, we may separate the variations into two parts ; one, consisting of an easterly motion (the 

 north end of the needle being always considered), at the constant rate of 0'-48 a-month, or 5''70 a-year ; tlie 

 second, of motions which are alternately to the east and to the west of the same mean position, — the latter being 

 evidently the only portion which can have any relation to season. If, then, n be the number of the month 

 from January, and we add the quantity 0''48 n to each monthly mean, we shall obtain the following quantities : 



From which it would appear that the westerly declination was a minimum in March and a maximum in Sep- 

 tember ; secondary minima occurring in June and November, and secondary maxima in January and April. 

 The whole range of these means, however, is only 2'' 12, and as the effect of 10° of torsion in the suspension 

 thread is 0'-84, it is quite possible that some of these variations may be due to this source of error. Though 

 an examination of the amount of torsion, found at different times in the suspension thread, will shew that the 

 error due to torsion is, in general, small,* yet it will be desirable to destroy accidental errors as much as pos- 

 sible by taking the means of groups; making use, for this purpose, of the mean for December 1843 = 18'-72, 

 and for January 1845 = 19''85 (both reduced for secular change to Januai-y 1844), and noting the mean for 

 each three months as the mean for the middle month, we obtain the following quantities for 1844 : — 



Jan. Feb. March. April. May. June. July. Aug. Sept. Oct. Nov. Dee, 



19''58 19'-60 19'-31 19'-25 19'-31 19'-20 19'-70 20'-35 20'-49 20'01 19'-52 19'-57 



These numbers indicate a distinct annual period, consisting of a principal minimum about May, a principal 

 maximum between August and September, and probably of a secondary minimum in November, and of a 

 secondary maximum between January and Februai-y. The variations upon wliich the latter portion of this 

 result depends are too small to be entitled to much confidence alone ; the Observations for 1843 have, accord- 

 ingly, been discussed anew for the purpose of comparison with this result. In the volume for 1843, the 

 monthly means from 9 daily observations were corrected by quantities obtained from the 24 hourly observa- 

 tions of 1844 (1843, p. 221). Correcting the means for 1843, and for December 1842, by the complete series 

 in 1844 and 1845, we have the following quantities : — 



Dec. Jan. Feb. March. April. May. June. July. -\ug. Sept. Oct. Nov. Dec. 



24'-64 25'-50 24'-91 24'-35 23'-79 23'-51 25'-25 23'-59 22'-33 20'-92 21'-75 19'-09 19'-20 



When these numbers, and the mean for January 1844 (20'-60), are reduced for secular change to January 

 1843, and means of each three months taken as in the above instance for 1844, we have, 



Jan. Feb. JIarch. April. May. June. July. Aug. Sept. Oct. Nov. Dec. 



25'-02 25'-40 25'-31 25'-32 26'-10 26'-52 26''60 25'-64 25'-51 24'-9l 24'-81 24'-91 



From these means for 1843, the principal minimum occurs in November, the principal maximum between 

 June and July ; a secondary minimum between March and April, and a secondary maximum in February ; the 

 gi'eatest difference in the epochs for the two years is found in the period of the principal maximum, which 

 occurs two mouths earlier in 1843 than in 1844. This difference may be explained by the gradual destruction 

 of the suspension thread, and the substitution of a new one in June 1843 ; when this and the small range of 

 the variations are kept in remembrance, the resemblance in the results for the two years will appear considerable. 



On taking the mean of the results for the years 1843 and 1844, we have. 



* See the foot-notes to the Hourly Observations of Magnetometers, and the article Declinometer in the Introduction. 



