Temperature Coefficient of the Balance Magnet. xlvii 



74. The result from hot and cold water experiments is nearly i more. It ap- 

 pears, therefore, that the determination of the temperature coefficient, by removing 

 the magnet from its position in the instrument and varying its temperature by means 

 of hot and cold water, cannot be depended on. It appears also, that when a suffi- 

 cient number of observations is included, the method of comparison previously de- 

 scribed gives, under very different conditions, consistent, and, therefore, it is pro- 

 bable, accurate results.* 



Balance Magnet. 



75. The temperature coefficient of the balance magnet was determined by means 

 of hot and cold water experiments August 24, September 1 and 2, and November 

 13, 1843, and January 27, 1844. See pages xlii., xliii., and xliv., Introduction, 1841-2, 

 for the details. The mean of the whole observations, properly weighted, gave 



q = 0-000073. 



76. The only good series was that obtained January 27, 1844, which included 

 changes of temperature from 35° to 65° only ; the other series arc too inaccurate to 

 be employed for the determination of the value of q for high and low temperatures ; 

 from series of comparisons of the usual observations of the balance it has been 

 found, however, that the value of q', the temperature correction for 1° Fahr. in mi- 

 crometer divisions, is the same for high and low temperatures, thus — 



Mic. Div. 



From 7 series of comparisons in 1844 and 1845, about the mean temperature 40°, <p'' = 8-33 

 From 8 60°, y' = 8-30 



As the first result is the mean of 7 values of q\ obtained from comparisons of 

 the mean readings of the balance magnetometer for about 170 days, in the months 

 of January, February, November, and December 1844 and 1845 ; and as the second 

 result is the mean of 8 values of q\ obtained from comparisons of the mean readings 

 of the balance magnetometer upon about 190 days in the months of May, June, July, 

 August, and September 1844 and 1845 ; it is extremely probable that the tempera- 

 ture coefficient for the balance magnetometer is constant for the ordinary tempera- 

 tures of observation. 



77. As it was found impossible to determine k the value of one micrometer 

 division in parts of the whole vertical component, by means of the vertical vibra- 

 tions, the value of q obtained from hot and cold water experiments could not be em- 

 ployed, since the observations could not be reduced to parts of vertical force, nor 

 could the value of q be reduced to micrometer divisions. In consequence of this 



* It should be remarked, that these conclusions do not depend wholly upon the results for the 

 Makerstoun instruments, their accuracy has been verified by an examination of the observations made 

 in other places. 



