Magnetic Field on Thermal Conductivity. 533 



After the flow of heat had attained a steady state the 

 reading corresponding to the temperature-difference between 

 the middle points o£ the bars was taken, and also the de- 

 flexions representing the actual temperatures o£ each middle 

 point with reference to a cold junction. The deflexion cor- 

 responding to the temperature-difference was initially small, 

 but when the magnet current was switched on, it increased 

 with comparative rapidity for 20 minutes, and thereafter 

 more and more gradually for 10 minutes longer, when it 

 became constant. When the field was removed the difference 

 diminished again, usually requiring from 40 to 45 minutes 

 to regain its steady value. This process was repeated several 

 times and the mean of these rises and falls for each field is 

 represented by each of the numbers given in the table ; the 

 first fall and the subsequent rises and falls were less than the 

 first rise, doubtless on account of residual magnetism. By 

 means of the observations of the actual temperatures of the 

 middle points of the bars, it was shown that the changes in 

 temperature-difference were due to a rise in the temperature 

 at the middle point of one of the bars, and an approximately 

 equal lowering of the temperature at the middle point of the 

 other ; on the removal of the field the temperatures fell and 

 rose respectively, until the original small difference before 

 the field was applied had been approximately regained. The 

 changes were proved to be not in any degree due to heat 

 transferred from the electromagnet coils. The bar which 

 was not between the poles was in a stray field of about 1900. 



By using this bridge method, the alteration in conductivity 

 of bismuth was found to be 0*3 per cent, for a field of 8500 ; 

 but this result is probably unreliable, since the length of the 

 bismuth bars (8 in.) was afterwards found to be too great to 

 justify neglecting the loss of heat by escape through the felt 

 lagging. 



The effect of a longitudinal field was measured in two ways 

 — by the " bridge method," and by the method of observing 

 directly the quantity of heat which flows along a magnetized 

 and an unmagnetized bar in a given time. Preliminary 

 trials showed that a decrease of conductivity would be de- 

 tectable by the latter method, although they did not provide 

 data for evaluating the change numerically. Two similar 

 bars were set up with their ends in the copper block of the 

 steam-box and insulated from it. One bar passed through a 

 coil which was wound anti-inductivelv, and arranged so that 

 the two halves could be joined in opposition so as to produce 

 zero field and yet have the same heating effect as when they 

 were put in conjunction to generate a field, which in^tliis 

 case amounted to 41 o.G.s. units. To the top ends oi' the 



