TEMPERATURE COEFFICIENT OF THE BALANCE MAGNET. xlix 
occur at the same time with the greatest variations of mean temperature ; it is 
believed that it is to this cause chiefly that the differences of the results are to be 
attributed. 
TABLE 8.—Determinations of the Temperature Coefficient of the Balance Magnet. 
Value of Value of 
, 
Period. 
q- q- 
1843. Mic. Div. 1844, Mie. Div. 
Jan. 16—Jan. 21 8-21 Noy. 4—Nov. 3 6-92 
Jan, 23—Jan. 28 : 6-99 Dec. 2—Dec. 7-20 
Jan. 30—Feb. 4 . 7-21 1845. 
Feb. 6—Feb. 11 . 6-69 | Jan. 6—Feb. 7-57 
June 1—June 30 7-82 Feb. 26—Mar. 8-00 
Sept. 6—Sept. 16 Apr. 10—May 10 9-08 
1844. June 2—June 30 . 8-47 
Jan. 1—Jan. 26 July 7—Aug. 6 10-01 
Feb. 5—Mar. 6 Sept. 9—Oct. 13 7-81 
May 9—May 24 Dec. 11—Jan. 10 10-17 
May 29—June 29 : 1846. 
July 4Aug. 3 : Nov. 30—Dee. 26 7-72 
Aug. 4—Sept. 6 
Giving the differences for all the series equal values, and dividing the sums of 
differences of the daily means in micrometer divisions by the sums of differences of 
the daily mean temperatures of the needle, we have 
¢ = 823 mic. div. ; 
but if the results from the bad series for July 7—August 6, 1845, and December 
10, 1845—January 10, 1846, be rejected, the value would be 
‘ = 7-99 mice. div. 
If the whole series were properly weighted, it is believed that the resulting value 
of g’ would be less than 8-00 mic. div. The excellent series, November 30—De- 
cember 26, 1846 (after an adjustment July 1846) gives 
¢ = 7°72 mice. div. 
The adopted value of the temperature coefficient for the balance magnet = 7:90 mic. div. 
It is believed that this value, which has been used in correcting all the observations 
since the commencement of 1843, is within one-tenth of a division of the truth. 
81. The following matters should be attended to in determining the tempera- 
ture coefficient by the previous method. 
1st, The period selected should be free from considerable magnetic irregularities. 
2d, There should be a considerable change of daily mean temperature, the tem- 
perature at the beginning and end of the period being nearly the same. 
3d, The smaller the duration of the period consistently with the 2d the better. 
4th, It will be found best, in general, to correct the daily means at first by an 
approximate coefficient, and 
MAG. AND MET. OBS., 1845 AND 1846. n 
