Addendum. . Ixix 



ascending and for descending temperatures ; on the whole, however, they indicate 

 that the coefficient deduced from such rapid changes is greater for the same mean 

 temperature (about 65°), when the temperature is increasing than when it is 

 diminishing. The greater part of the observations were made without any inter- 

 mediate temperature, so that there are not sufficient data to determine whether the 

 difference is greater at high mean temperatures than at low mean temperatures. 



Mr S. H. Christie found that the temperature coefficient increases with the 

 temperature, the increase becoming more rapid for temperatures above 80° ; and 

 that beyond 100°, a portion of the magnetism is permanently lost.* The previous 

 results seem to indicate that (in such experiments at least), a portion of the magnet- 

 ism is lost even at the lower temperatures for certain kinds of steel.f 



The observations of deflection with the circle magnetometers have been reduced 

 by the formula (see Introduction, p. xx.) 



m , , . 1 



— = i r* sm u 



X ^ - 



9 "^ 4 



where 



sin Wq 1 + k (bg— by) 



_ dm . 1 -g (ta- t) 



sm u ■= 



Uq being the observed deflection corrected for declination change (see column 9, De- 

 flection, p. 349-351). 



The observations of June 23, 1846, and of January 2, 1847, were made with 

 Professor Forbes' 6-inch circle magnetometer ; the remaining observations were made 

 with Sir Thomas Brisbane's instrument, the dimensions of the magnets employed 

 will be found, p. 351 : the values of one scale division for the suspended magnets 

 are as follow : — 



June 23, 1846. Jan. 2, 1847. May 31 and June 15, 1847. Sept. 11 and 13, 1847. 



0'-995 0'-995 l'-025 2'-017 



The solid bars are about 0*3 inch diameter, and the collimator bars about 0*4 

 inch diameter. 



All the observations, excepting those of September 11 and 13, were reduced by 

 the method of least squares, the equations of condition having the form 



1 + -o P + -IT Q - i r^ sin M — = 



* PhilosopUcal Transactions, 1825, p. 63. 



] The previous results seem to indicate that while it will always be necessary to determine the 

 temperature coefficient for the magnetometers from the observations with the magnet in its box, 

 yet it will be desirable to determine it also by the usual method of hot and cold water experiments. 



MAG. AND MET. OBS. 1845 AND 1846. S 



