xl INTRODUCTION. 
It was evident that no good result could be obtained without an extra de¢lino- 
meter, owing to the necessity of reversing the needle, in order to eliminate the 
changes of declination, when the slightest alteration of position would have far more 
effect than the greatest change of temperature. 
47. Having had a small wooden house erected at some distance from the Ob- 
servatory, and having formed an extra or unifilar magnetometer, observations were 
made to determine the temperature corrections, according to the method recom- 
mended in the “ Revised Instructions,” by the Committee of the Royal Society. 
The magnets were placed in a copper trough, as in the previous observations, 
and the temperature varied by means of ice and water of different temperatures. 
The declination magnet was observed at the same seconds as the deflected magnet, 
and the bifilar immediately before and after each temperature. 
Owing to the small deflection, the observations on the first day were valueless, 
and are not given here. The centre of the balance magnet was placed afterwards 
at a distance of about 3 feet from the suspended magnet. 
48. The following tables contain the observations. 
The results in the last column are obtained from the formula 
=3 _u—u,—r d—d, , & (6—b) 
aa’, D (¢—4) t—t, 
where u and w, are the unifilar readings, d and d) the declinometer readings, 6 and by 
the bifilar magnetometer readings at the temperatures ¢ and 4 of the deflecting 
magnet ; D the mean deflection ; r the coefficient for reducing the declinometer scale 
divisions to those of the unifilar ; and & the value of the bifilar magnetometer scale 
divisions, in parts of the whole horizontal force. 
During the observations for the temperature correction of the bifilar magnet, 
another magnet was substituted for the bifilar magnetometer ; for it 
k = 0:0002263 ; 
during the observations for the balance magnet, the following were the values of K 
Aug. 234 1843, 
Sept. 1 1843, +z=0-0001204 Nov: 38 ibid, } #=0-0001800. 
Sept. 2 1843, aes 2 
Owing probably to currents of air, the difference of the declinometer and 
unifilar scale readings at the beginning and end of the series Aug. 23, 24, and 31, 
Table 20, had varied considerably. The change was supposed uniform ; and the 
declinometer readings adopted are those interpolated for the change. 
