i on account of the heated stove generating currents of air in the room and magnetometer boxes. 
TEMPERATURE COEFFICIENT OF THE BALANCE MAGNET. xlix 
Mondays; the remaining five-sixths are of the readings on days with those on the 
days immediately following :— 
1844. 123 Comparisons, g’= 8:33. Approximate value of ¢, by temperature experiments and vibrations, = 3:8 
eel AG K COM PATIEODE NG OO). | denaveasc-<nastsuceessssncasonccganabissvance you acmanesaueacecwesueddusutenicascseedes o> = 30 
69. The most severe method of testing the accuracy of the coefficient thus 
obtained, is that of artificially heating the Observatory, and of comparing the in- 
strumental readings on days when so heated with those on days when the tempera- 
_ ture of the magnet depends on natural causes ;* the results of such a series of com- 
parisons are given below. In order to render the series perfectly trustworthy, all 
selection has been avoided. From January 1 till March 16, 1843, the stove in 
the Observatory was lighted every day, with the exception of January 23; it was 
not lighted on any other days of the year, excepting March 22 and 23, October 19, 
November 25, and December 21; the means of the instrumental readings, corrected 
for temperature by the adopted coefficient (7-9), on each of these days were compared 
with the means on the four or five days immediately preceding or succeeding, and 
the means on March 13, 14, and 15 (the last three days on which the stove was 
continuously lighted), were compared with the five succeeding days on which the 
stove was not lighted. The sums of the differences 3 A (¢) and 34 (y) are given in 
the following Table, together with the sums of the differences of the external tem- 
_ peratur es 2 a (T) for the same days :— 
TABLE 18.—Results of Comparisons when the differences of the Temperature 
depend on Artificial Causes. 
Day of No. of 
com- Days compared with “| compa- 
_ parison. risons. 
Error of | Resulting 
coefficient. | coefficient, 
1843. ° ic. Diy. | Mic. Diy. | Mie. Div. 
18—21, 2428 +1-41 6.49 
16—21, +242 | 5.48 
16—21, +1-31 6-59 
16—21, +0-90 
16—21, 24—28 +0-11 
16—21, 2428 —0-73 
14—18, 20—25 — 2.58 
. 21—24, 27Dec. 1 + 0-67 
16—20, 2227 +0:07 
womomoooruanded 
NAY . 
The results of all the comparisons are— 
2A (T) = — 55°11, 2 A (®) = 697°-5, = A (y) = + 1040 mic. div.; error of adopted coefficient = + 0°15 mic. div. 
* This method was that first tried for the determination of the coefficient, but ultimately abandoned 
MAG. AND MET. oss. 1843. n 
