300 C. G. KNOTT. 



be interesting to establish Ijy direct ex})erinient that the Thomson 

 Effect is exception;! ]ly laru'c in cobalt. 



The downward trend and cDinparatively large inclination (^f the 

 bismnth line are also worthy of note. Uecanse of the ])osition of the 

 line as a whc^le, lying far below tlie lines of all other metals, this 

 large inclination does not greatly influence the electromotive fjrces, so 

 that with bismuth C(juples the electromoti^'e force is very approxi- 

 mately proportional to the temperature. This fact of course prevents 

 us from making a very accurate determination of the coefficient 15, 

 which in the present experiments has a large probable error. Its 

 mean value is a little greater than the ^•alue indicated in liattelli's 

 direct measurement of the Thomson Effect in liismuth^'. 



liio'hi has shown-^ that the electric resistance of lUsmuth is altered 

 in a strcjiig magnetic field. To find if any thermoelectric change 

 accompanied magnetisation in nickel, a bismuth palladium cou])le was 

 set up IjetAveen the poles of a p(3werful electromagnet. Xo effect 

 whatever was obtained, although the arrangement (slightly modified) 

 was sensitive enough to show with great ease tlie thermomagnetic 

 effect discovered by v. Ettingshausen and Xernst^\ 



1) See Wied. Beiblätter, Vol. XI, 1887. 



2) See Wied. Beiblätter, Vol. XIII, 1884. 



3) See Wied. Annaleu, Vol. XXIX, 1886. 



