1878.] of Iron and Steel caused by Magnetism. 117 



The magnetising current was then passed through the coil, and the 

 shunt again adjusted, until the piles neutralised each other. 



It can be readily shown that if x x , x 2 be the units of heat imparted 



x S 



to the water in the two boxes in the same time, -L — when 



<tf i R + S 



the piles neutralise each other. 



If, then, dx 2 be the diminution of flow caused by magnetisation, and 

 cZS the diminution of shunt resistance necessary to again make one 

 pile neutralise the other, 



x 2 —dx 2 _ S— dS 



~~~ ~x~~ "R+S-dS ■ 



dx 2 R x cZS 



~^~ _ S(R+S-dS) 



and thus the percentage of diminution of flow may be determined. 



A few resistance coils were very roughly constructed for use in the 

 shunt, S, and an attempt made to ascertain the actual amount of 

 diminution of conductivity of the hard steel. This, again, appeared to 

 be comparatively small, but as the coils used were not suitable for the 

 purpose, the result is not given. But the writer has little reason to 

 doubt that, with resistance coils suitable for such thermo-electric 

 experiments, he will be able to measure, with fair accuracy, small 

 variations of thermal conductivity in this way. 



It should be added that similar experiments to these were made 

 with soft iron bars, both with and without copper terminals, but the 

 results of the experiments, of which altogether upwards of a hundred 

 have now been made, show that in every case the thermal conductivity 

 of soft iron is diminished by temporary longitudinal magnetisation and 

 increased by transverse magnetisation, whilst in the case of steel, of 



