the Roberts-Austen Recording /'urometer. 77 



The temperatures given in the last column of Table II. 

 (p. &2), were obtained in this way from the Oct. 189G results 

 of couple 11. These temperatures really depend on an extra- 

 polation beyond 445° of the empirical formula given above, a 

 formula which appears to be correct in the case of this 

 thermo-couple as far as it has been possible to verify it. The 

 dotted lines in fig. 6 are the continuations of the curves at 

 temperatures below 0°. They depend on the assumption 

 that the formula holds good at these temperatures, and show 

 that at the absolute zero the E.M.F. would become infinitely 

 great, and that the contact E.M.F. would have a negative 

 value of about 200 micro-volts : it has been stated * that the 

 latter should be zero, as otherwise further cooling could be 



dE 

 effected by the passage of a current. The T^ curve for a 



platinum -lead thermo-couple, calculated from figures given in 

 Dewar and Fleming's f paper on thermo-electricity at low 

 temperatures, is straight between 100° and 0°, but below this 

 temperature it shows a tendency to curve in the direction 

 required to reduce its value to nothing at the absolute zero. 



The theoretical simplicity of the formula given above, which 

 involves a linear relation between the temperature of a junction 

 and the Peltier E.M.F., makes it worth while to consider 

 how far this relation is true in the case of other thermo- 

 couples. In fig. 7, which has been plotted with this object, the 



JTT! 



clotted lines represent the values of -™, and the continuous 



rfE dl 



lines those of T-^p, for the couples 13, 15, and If), and for the 



couple used by Holman, Lawrence, and Barr. 



None of the curves representing the contact E.M.F. are 



so nearly straight as in the case of couple 11 ; but all these 



curves are approximately straight, whilst those representing 



c/E 

 the values of —^ are all strongly curved, 

 dl to J 



This is of great interest in connexion with the parabolic 



formula investigated by Tait. Tait's expression for the 



E.M.I 1 , of a thermo-couple follows from equation (A) if the 



assumption be made that the specific heat of electricity is 



proportional to the absolute temperature ; but the parabolic 



formula is really empirical, and depends on the observations 



that the curves representing the values of -^ are approxi- 

 mately straight in the case of the thermo-couples for which 



* Burton, Phil. Mag% vol. xxxviii. 1894, p. 66. 

 t Phil. Mag. vol. xl. July 1895, p. 95. 



