OP ARTS AND SCIENCES. 213 



thin plates of soft irou, running the length of the glass tube, were 

 placed upon the magnetic poles, thus lengthening the field through 

 which the bar passed. The bar was next heated as before. After 

 several hours, the galvanometer generally showed a permanent deflec- 

 tion of 3o-40 cm., in a direction indicating that the junction at the 

 end of the bar was the cooler. Several observations were made, 

 both when the bar was parallel and perpendicular to the axis of the 

 magnet. The current was passed for about one hour, and in every 

 case there was absolutely no change of the deflection beyond the 

 immediate change due to the direct action of the magnet. 



The result of these experiments seems conclusively to show that 

 longitudinal and transverse majjnetism — at least of the strength used 

 — have no influence upon thermal conductivity of soft iron. It was, 

 however, decided to try a thinner piece of iron than the preceding. A 

 strip of ordinary tinned iron was therefore cut about 1.3 cm. wide, 

 and was bent over and arranged exactly as before. The whole tube 

 was packed with asbestos and cotton to avoid any motion of the strip 

 when the magnet was made. The distance of the flame from the two 

 thermo-electric junctions was 10 cm. A heating of ninety minutes was 

 found to be sufficient in this case for the strip to reach a permanent 

 condition of temperature. The deflection showing the difference of 

 temperature between the junctions was about 13 cm. AVheu the 

 magnet was made, there was no change of the deflection after thirty- 

 five minutes' observation. 



The strength of the magnetic field when the bar was placed perpen- 

 dicular to the lines of force was measured after the preceding experi- 

 ments were made, and was found to be 10,420 times the horizontal 

 intensity of the earth's magnetism at Cambridge. In the C. G. S. 

 system this would be about 1,760. 



Aside from the experiments of Maggi, those of Thomson upon 

 electrical conductivity are the only experiments that seem to be 

 directly opposed to the conclusion that must be drawn from our 

 observations. Magnetism undoubtedly changes several physical prop- 

 erties of iron, but though this renders it probable, a priori, that the 

 thermal conductivity might be changed, yet it does not necessitate 

 8uch a change. The thermo-electric relation of iron is changed by 

 magnetization, but the thermo-electric relation appears to be uncon- 

 nected with thermal conductivity. 



