( 412 ) 



vary; it is probable that the temperature at both ends does also not 

 remain the same, and only with the aid of three observed tempera- 

 tures we can calculate a. It is evident that in this way errors of 

 observation should acquire a great influence, especially as the deter- 

 mination of the falls of temperature could not possibly be made at 

 the three places at the same time ; the stationary temperatures were 

 about the same in each experiment, but small differences could not 

 be avoided. 



For the calculation, the temperatures, observed at the heated 

 end and in the places ^,J5andC were united in a g-raphical repre- 

 sentation with 31 as abscissa. From this the temperatures in the 

 places «=0, 0,6, 1,2, 1,8, and 2,4 were deduced. These must 

 satisfy the relation. 



tn + 1 



::^ e — 0,6 o _|_ g + 0,6 o_ 



Putting e — c'.en =■ ^i we find k from the equation 



i2_ilL±il±^yt + l=0. 

 ^1 + 1 



Supposing the mean value of k to be correct, I then calculated 

 from the set of 5 temperatures the coefficients p and q, and from 

 these again the temperatures themselves. The result being: 



k = 0,626 



;; = 43,16 



q — 0,311 



Fig. 3 shows the calculated curve. The signs X show the tem- 

 peratures and the places which where used in the calculation, the 

 signs + show the really observed temperature in the places A^ B 

 and C. 



