HALL. — CONDUCTIVITY OF SOFT IRON. 135 



maining nearly unchanged throughout the two sets.* The elimination, 

 however, would not be perfect ; for the reason that the temperature of 

 the stream which we are considering is not quite the same in the two 

 complementary sets of observations just described, being about 8° warmer 

 in one set than in the other. The question at issue comes, therefore, 

 very nearl}' to this, whether the amount of heat carried away from the 

 stream by the wires, when the stream is 8° warmer than the room, is 

 negligible in comparison with the amount carried away by the disk, the 

 difference of temperature of the two faces of the iron being, as we have 

 seen, usually 1°.4 or more. This question can be answered in the affirm- 

 ative ; for the aggregate cross-section of the four copper wires is little, if 

 any, more than 0.03 sq. cm., and the length of each wire, from the spiral 

 out to the point where it is exposed to the temperature of the room, is 

 not far from 30 cm. The carrying power for heat of a rod of copper 

 30 cm. long and 0.03 sq. cm. in cross-section, the thermal conductivity 

 of copper being taken as eight times that of iron, would be about one 

 five-thousandth part of the carrying power of the iron disk for a given 

 difference of temperature, and not more than one eight-hundredth part, 

 if the difference of temperature were 8° for the rod and 1°.4 for the disk. 

 Nevertheless, after a considerable number of trials had been made with 

 the spirals placed as in Figure 2, the plugs bearing the spirals were put 

 in at the other ends of the supporting sockets, so that any changes of 

 temperature produced in the stream by the copper wires must now occur 

 either before or after the change of temperature noted by the spirals. 

 The mean of the results obtained after this change of arrangement was 

 slightly different from the mean of those obtained before ; but it is un- 

 likely that the difference was due to this change. 



In the observations of July 31 and thereafter a loop of each copper wire 

 was kept in a pocket containing oil on the outside of the water-jacket, so 

 that each wire had at this point a temperature not more than one or two 

 degrees different from that of the spiral with which it was connected. 



Another question was whether the heat carried in or out by the copper 

 wires affected the temperature of the spirals directly by metallic conduc- 

 tion, so as to keep them at a temperature different from that of the water 

 passing them. The reasons for thinking that any such effect was neg- 

 ligible are given in the Appendix to this paper. 



It has been stated that there was a difference of resistance between 

 the two spirals at any given temperature, and that this difference in- 



* Tfie time between the exchange of the water streams and the beginning of the 

 next set of observations was usually rather more than twenty minutes. 



