HALL. — THERMAL AND ELECTRICAL EFFECTS IN SOFT IRON. 617 



the mean Vw will be reduced to about 3.7 per cent ; but even then the 

 smallest Vp will be somewhat larger than the largest Vw 



There is one other important datum, namely, the mean difference 

 of temperature between the main bars, which was measured at only one 

 heating in condition E, namely during February 5-7. This mean value, 

 found from the observed values at the five cross-sections a, b, c, d, e, 

 by plotting a curve, was 0.389°, or 0.39°. An error of about 20 

 per cent in this rather small quantity would account for the whole 

 difference between the Vg and the iv as found. That an error approach- 

 ing this magnitude was made here is shown to be improbable by the 

 fact that the two determinations of the main bar mean differences of 

 temperature made with condition W gave 0.408°, or 0.41°, and 0.391°, 

 or 0.39°, respectively. 



Other possibilities which have been considered in the endeavor to 

 account for the difference in question are the following : 



1. A partial short circuit of the main electric current in the west 

 pot when this contained naphthalin. There is no good reason for sus- 

 pecting such an action. A test made on April 18, with boiling water 

 in the west pot and boiling naphthalin in the east pot, showed that 

 there was no serious leakage of electricity through either liquid. 



2. Difference in the sensitiveness of the thermo-electric spirals on 

 the main bars. 



A difference of two or three per cent, perhaps more, in the sensitive- 

 ness of these spirals is possible. But the observations with these 

 spirals are taken and are combined in such ways that the result ob- 

 tained from a complete set of data for condition E or condition W is 

 dependent on the mean sensitiveness rather than on the differences 

 of sensitiveness of these spirals. We could account for the difference 

 between I'p. and v^ by means of variation in these spirals, if we should 

 assume that the two a spirals, at the east end of the bars, are about 

 seven per cent more sensitive at 204° and seven per cent less sensitive 

 at 115° than the corresponding two e spirals at the west end of the bars. 

 This is very improbable. 



3. Dissymmetry in the shape of the bars or in the placing of the 

 spirals upon them. 



There is none that can account for the difference which we are con- 

 sidering. The bars were ground to 1 cm. diameter by Brown and 

 Sharpe. The spirals are probably not all of exactly the same length, 

 and there may be slight inaccuracies in their emplacement ; but these 

 imperfections, though they might affect slightly the value of v, would 

 have little or no tendency to make a difference between iv. and iv. 



4. Difference in the movements of air within the loose asbestos fibre 



