Copper andiron for Heat at different Temperatures. 165 



In like manner the probable error in the value 15*19, or the 

 percentage alteration in conducting-power for 100° C, equals 



±0-3392. 



k 

 If, for the calculation of -k, the formula 



CO 



^=a{l-at+ftt*) 



be taken, then Xe 2 would certainly be diminished ; but since the 

 observations fall within too narrow limits to admit of the obtain- 

 ing of ft with sufficient accuracy, I regard it as unnecessary to 

 give this calculation. 



The alterations of temperature which are expressed by he 

 should be made to enter into a, the value of the coefficient of 

 temperature which was obtained above. But inasmuch as these 

 differences, so far as is known, are very slight, at any rate within 

 the range of temperatures of the foregoing observations, and inas- 

 much as for a rise of temperature 8 diminishes and c increases, 

 it is evident that they must have very little influence upon the 

 value of the coefficient of temperature. 



Let 



cS= 0-84476, 

 then 



£=58-94 (1 -0-001519 /). 



k 

 Calculating the value of -~ from the two last determinations for 



iron given in the Table, which are fully comparable with one 

 another, there is obtained 



^ = 13-458 (1-0-002874/). 



Putting 



c8= 0-8862, 

 it follows that 



it = 11-927 (1-0-002874/). 



The first of the observations for iron given in the Table, and 

 which was not used in the obtaining of the formula, may serve 

 for a further control of the correctness of observations ; and the 

 more so as it was made with other thermometers, and the holes 

 for the reception of the thermometers were 2 millims. in dia- 

 meter, while in the last experiments they were 3 millims. in 

 diameter. 



For comparison I have reduced the observations given in the 

 former paper, and taken upon the bars of copper used in former 



