HALL. — THERMAL AND ELECTRICAL EFFECTS IN SOFT IRON. 621 



On the other hand, it seems to be well established that the numerical 

 value of V increases at a rate not very different from that indicated 

 above, 2.0 X l()~^°per degree, in the region between 32° and 182°. As 

 V is proportional to the tangent of the angle of inclination of the iron 

 line from the horizontal line on the thermo-electric diagram, the state- 

 ment just made means that in the temperature region considered the 

 iron line should be one of increasing steepness with rise of tempera- 

 ture. Taking the value of v 



at 32° as - (757 — 20 X 2) X lO"^" = — 71 7 X lO'^^ 

 and at 182° as —(757 + 130 X 2) X 10"" = — 1017 X 10-'^ 



we find that the tangent of the angle of inclination of the iron line 

 should be about 40 per cent greater at 182° than at 32°. 



In our previous paper ^^ it was pointed out that Battelli's observa- 

 tions on iron gave data which yield the following values for v : 



-283 X 10-1*' at 53° C. -331 X lO"^" at 243° C. 



-319 " " 108° -369 " " 308° 



The mean rate of increase of r here indicated would make the value at 

 182° about 18 per cent greater numerically than the value at 32°. 



Battelli's general method was to surround a certain length of each of 

 two similar bars of one metal by a narrow trough containing mercury, 

 separated fi-om the bar by a layer of varnish or other insulating 

 material, establish a gradient of temperature in the bars by means of 

 ice and boiling liquids, pass a strong electric current up the gradient 

 in one bar and down in the other, then estimate the Thomson effect 

 from the difference in the rate of heating of the two baths of mercury. 

 He does not give sufficient details to assure us that he took adequate 

 account of the tendency of the mercury, which was continually stirred, 

 to reduce the gradient of temperature, and so the Thomson effect, in 

 the part of the bar which it surrounded. Whether the low numerical 

 value which Battelli found for the Thomson effect in iron, as com- 

 pared with that found by us or very recently by Lecher, 1* is due 

 to this peculiarity of his method or to his having used a very different 

 kind of iron from that which we have used, is uncertain, but it is quite 

 possible that his relative values of v at different temperatures were 

 approximately correct ; and the per cent rate of change which his data 

 give for v \vith rise of temperature, though less than that which we 

 find, is large enough to establish a strong presumption in favor of 

 representing iron on the thermo-electric diagram by a line of con- 

 is These Proceedings, 41, 36 (1905). 

 " Annalen der Pliysik, 4, 19, 859 (1906). 



