380 Thermoelectrics of Platinum-indium Sfc. 



6. Change of Curvature, incidental. — The departure from 

 the linear relation of thermoelectric powers above 1200° may 

 of course be real. But there are agencies here at work into 

 which it is necessary to inquire. In the first place, at these 

 excessively high temperatures the fire-clay insulators*, as 

 well as all parts of the furnace, conduct electrically. It is a 

 curious fact, too, that the acid silicates, although the more 

 viscous, are better conductors than the more nearly liquid 

 basic silicates, co?t. par. Thus a certain part of the thermo- 

 electric power is possibly tapped off through the insulators. 

 Again, and more seriously, under the given circumstances 

 volta contacts are introduced, consisting, in one case, of 

 rhodioplatinum/silicate/platinum, and in the other of iridio- 

 platinum/silicate/platinum ; and it is altogether improbable 

 that these effects will be equal. It is obvious, also, that the 

 discrepancies in question would increase with temperature, 

 compatibly with the chart. Finally, the hot gases themselves, 

 particularly the flame-gases, cannot at once be considered free 

 from disturbing influences. 



Some experiments made at relatively low temperatures 

 with the burner in action seem to bear out these points of 

 view. Thus between 1050° and 1150° a series of 7 measure- 

 ments led to the factor e'2o/e 20 = 0*7689 + 0*0009 ; 6 measure- 

 ments between 950° and 1200° to 0*07682 ±0'0295, &c. 

 Hence, when a flame envelops the crucible, and the insulators 

 are hotter, the factor is apparently larger than in the case of 

 a furnace with the burner , removed, and cooling by radiation. 



Remembering, therefore, that the deviation is not large, 

 I think it maybe reasonably asserted that the ratio of thermo- 

 electric powers of the two couples is probably constant 

 throughout the whole interval of over 1700°. 



7. Thermoelectric Law. — Here, therefore, are two thermo- 

 couples, differing in chemical composition, and differing also 

 in percentage composition, the thermoelectric powers of 

 which, in their variation with temperature, show a constant 

 ratio up to 1200° at least, and probably at all laboratory 

 temperatures (1700°). I think, therefore, that I can make a 

 strong point by claiming that the change of electromotive 

 force with temperature, in both these couples, is thoroughly 

 normal ; for if there were anomalies present, they would have 

 to be similarly and symmetrically situated in two different 

 couples — a much more improbable assumption. In short, the 

 thermoelectric behaviour of iridioplatinum-platinum and rho- 

 dioplatinum-platinum couples may safely be looked upon as 



* See a forthcoming paper by Joseph P. Iddings and myself on the 

 electric conduction of rock magmas of different composition (Am. Journal, 

 Sept. 1892). 



