1872.] 



Periodicity in Terrestrial Magnetism. 



309 



perature of the room has been maintained in a generally equable state, and 

 in later years it has been remarkably uniform. 



It is easy to see that an error of a single day, or of a large fraction of a 

 day, in the beginning of each period, is of no importance, provided that 

 the errors are not permitted to accumulate. It was allowable, therefore, to 

 take successive periods of 26, 26, 27, 26, 26, 27, &c. days ; and in 

 instances when a single day was omitted, or even two days, no sensible 

 error would be introduced by interpolating between the numbers for the 

 days immediately preceding and following the omitted days. 



The years selected for this examination were 1850, 1851, 1852, 1868, 

 1869, 18/0; and the beginning of the first period in each year after the 

 first was thus found : — Fourteen periods of 26-J- days each amount to 368f 

 days. For convenience after completion of the annual winter adjustments, 

 the first period in 1850 was made to commence on January 17 ; therefore 

 the first period in 1851 was commenced on January 21, and that in 1852 

 on January 25. Similarly, the first periods in 1868, 1869, 1870 com- 

 menced on January 1, 4, and 8 respectively; and the beginnings in the 

 three later years are not unconnected with those in the three earlier years : 

 for, from 1852, January 25, to 1868, January 1, are 5820 days, and 221 

 periods of 26-i- days each are 5819f days; but as the years are widely 

 separated, and a small error of period would produce a large discordance, 

 it has appeared best to exhibit the results of the two three-years' groups 

 separately. 



Some periods, in which there were unusually large interruptions, or which 

 were partly occupied with experiments, were omitted entirely. The fol- 

 lowing is a complete list of periods omitted : — In 1850, that beginning with 

 December 24 for horizontal force ; in 1851, that beginning with March 14 

 or western declination, and those beginning with March 14, June 28, July 

 24, for horizontal force; in 1852, those beginning with February 20, May 9, 

 December 6, for both elements ; in 1868, those beginning with February 

 23 for declination, and January 1, January 27, February 23, and De- 

 cember 8 for horizontal force; in 1869, those beginning with October 21 

 and December 12 for both elements; and in 1870, those beginning with 

 June 15 and December 16 for declination, and that beginning with 

 December 16 for horizontal force. Interpolations of three days occur only 

 in the following instances :— 1850, (dec.) Feb. 4-6, (h. f.) Feb. 9-11, 

 July 23-25; 1851, (dec.) Feb. 18-20, Oct. 20-22, (h. f.) June 9-1 1 ; 

 1852, (dec.) Feb. 7-9; 1868, (dec.) Feb. 15-17, (h. f.) none; 1869, 

 (dec.) June 6-8, (h. f.) June 6-8; 1870, (dec.) Sept. 24-26, (h. f.) 

 Sept. 24-26. 



The mean values of each element for each progressive day in every 

 period of the several years, uncorrected for the proportional part of secular 

 change through the 26 days, and omitting the imperfect 27th day, are as 

 follows : — 



VOL. XX. 



z 



