Temperature Coefficient of the Deflection Magnet. xliii 



The observations after July 2^ are reduced to those before that date by the 

 above quantity. 



62. The observations of the balance magnetometer are made in the following 

 manner : — The moveable wire of the right micrometer is made to bisect the spider- 

 cross half the time of vibration in the vertical plane before the minute of observation, 

 and that of the left micrometer as long after the minute ; the mean of the two read- 

 ings gives the position of the needle at the minute. The readings increase posi- 

 tively when the north pole of the needle moves below the horizontal. The tabular 

 observations given, in this column, are obtained thus : n being the observed reading 

 of the needle (generally negative), t that of the thermometer, giving the temperature 

 of the needle, q' the temperature coefficient in micrometer divisions = 7"90, and B, 

 the quantity in the column, " Balance Corrected" 



Jan. I'' 1845— July l^ 1846. R = 700 + g' (t, - 26) + n ; 

 July 3d 1846— Dec. Sl^ 1846. R = 124 + ^ (; - 26) + w ; 



increasing tabular values, therefore, indicate increasing vertical force. 



The Temperature Coefficients of the Deflecting, Balance, and 



BiFiLAR Magnets. 



Deflecting Magnet. 



63. The temperature coefficient of the large deflecting bar (15 inches long), used 

 in the observations for the absolute horizontal intensity, was determined November 

 11, 1843, by hot and cold water experiments, see pages xlii. and xliii.. Introduction 

 1843, for the details : the mean of all the observations gave 



The correction for 1° of Fahr., y = 0000288 



64. The observations from which this result was obtained were very good, con- 

 sidering that the whole angle of deflection was less than 3°, and it may therefore be 

 worth examining the individual results for the highest and lowest temperature. The 

 whole number of results was 15, the mean dift'erence, from the final result given 

 above, =0-000025, and the probable error of a single result was therefore about 

 0-000021. 



The mean of 9 results for the mean temperature 49°-? gave g = 0000289 

 The mean of 6 68°-7 g = 0000286 



so that the temperature coefficient for this bar is constant within the ordinary tem- 

 peratures occurring during the observations in which it was employed. 



