HALL. — CONDUCTIVITY OF MILD STEEL. 297 



0.1300-0.1325 ^^,^,^, 



y = 0.1325 X (59.2 - 27.2) = -^-OOOb-. 



If we take the values 0.1347 and 0.1316, obtained by including the 

 bracketed values in finding the means, we get y = — 0.0007+. The 

 value — 0.0006 is to be iireferred, although it is evident from the pre- 

 ceding pages that no great confidence can be placed in the digit of 

 this coetficient. A wider range of temperature, a greater number of 

 measurements, and evidently a more careful calibration of the copper- 

 German-silver junctions is needed for a completely satisfactory deter- 

 mination of y. Moreover, no allowance has here been made for the 

 change in the specific heat of water between 27° and 59° C. It is to 

 be observed that the lower temperature is very near the temperature 

 of minimum specific heat of water. If it should be found that the 

 specific heat of water at 59° is two per cent greater than at 27°, 

 which is, to be sure, improbable, the value of y obtained from these 

 experiments would be 0. If the difference is one third of one per 

 cent, which seems not unlikely, it will reduce y to — 0.0005. If we 

 take this as the most probable value, it will, in spite of its uncertainty, 

 be not devoid of interest, for the tables in Winkelmann give no values 

 of y in the case of steel, while for iron they give 



y — — 0.000038 (Lorenz), 

 y = — 0.000517 (Angstrom), 

 y=: -0.0011 (Stewart). 



The one measurement of k for an intermediate temperature, 44°. 2, 

 lies between the mean value for 27°. 2 and the mean for 59°. 2, which 

 is gratifying evidence, so far as it goes, of the precision of the meas- 

 urements. 



It is a curious fact that every measurement of Jc near 27°, except 

 the last, gave a smaller value than the preceding measurement. The 

 reason for this is not evident. 



At the low temperature the mean of h^ is a little greater than the 

 mean of i'j,' At the high temperature the contrary is true. In either 

 case the difference is to be regarded as mainly accidental, and not 

 significant, — a fact already mentioned and explained. 



"We must now consider more fully the question whether A, as de- 

 duced from curves like those of Figure 10, represents accurately the 

 difference between the mean temperatures of the upper and of the 

 under face of the steel disk. It can hardly be doubted that A, derived 



