EXPERIMENTS ON THE THERMAL CONDUCTIVITY OF IRON. 175 



Provisional Report of a Committee consisting of Professor Tait, Pro- 

 fessor Tyndall, and Dr. Balfour Stewart^ appointed for the 

 purpose of repeating Principal J. D. Forres's Experiments on the 

 Thermal Conductivity of Iron, and of extending them to other Metals. 

 By Professor Tait. 



In consequence of a misunderstanding, the standard thermometers ordered 

 from the Kew Observatory did not arrive in time to be employed in the ex- 

 periments hitherto made, so that the results now to be stated, besides being 

 only approximate, are, for the most part, confined to a range of temperature 

 of about 100° C. merely. Before the next meeting of the British Association 

 the whole question will have been reexamined with far superior instruments ; 

 but with such thermometers as I had at hand (including some of those used 

 by Principal Forbes, of wlaich, however, I have not succeeded in obtaining 

 the corrections determined by Welsh at Kew), results have been obtained of 

 a character sufficiently definite for publication, though, of course, subject to 

 (slight) future corrections and perhaps limitations. 



The substances experimented on were iron, lead, and copper. Two spe- 

 cimens of the latter metal were employed, one of high, the other of low 

 electric conductivity, the resistances of equal lengths of wires of the same 

 gauge being about 1 to 1-64. The ratio of the thermal conductivities of 

 these bars was at once found to be within 5 per cent., the same as that of 

 their electric conductivities, a result certainly anticipated, but stiU very 

 striking. In specific gravity and specific heat, as well as in chemical com- 

 position, mode of manufacture, and drawing, these bars of copper scarcely 

 differ. As yet they have been treated for thermal conductivity in the hard- 

 drawn state alone ; but annealed wires of the same materials, while showing 

 a slightly improved electric conductivity, maintain towards one another a 

 ratio practically unaltered. 



Two points have been observed which enable us materially to simplify the 

 determination of thermal conductivities by Torbes's method, so long at least 

 as moderate ranges of temperature are concerned ; and we seek no greater 

 accuracy than admits of 1 or 2 per cent, of error. 



1. The Curve of Cooling is practically the same for aU the substances I 

 have tried (even for gas-coke), merely foreshortened or elongated in terms of 

 a parameter, which involves the product of the specific gravity and the spe- 

 cific heat of the substance employed. This was, of course, to be expected, 

 provided the radiating power of the surface be kept the same, and provided 

 conductivity do not interfere with the results. 



2. The Curve of Statical Temperature possesses, practically, the same pro- 

 perty, at least for the four different bars employed. This proves that within 

 the range of the experiments, and subject to the errors of the thermometers, 

 the law of change of thermal conductivity with temperature is the same for 

 lead and copper as for iron. I showed (Proc. R. S. Edin., 1867-68) that 

 Forbes's results for iron agree closely with the statement that the conduc- 

 tivity is inversely as the absolute temperature, a result which is identical 

 with Matthiessen's determinations of electric resistance of pure metals at 

 different temperatures. With a view to foUow up this analogy still further, 

 I have ordered a bar of German silver, a substance whose electric conduc- 

 tivity is but little altered by temperature. The results cannot fail to be 

 interesting. 



Very simple reasoning from the (plotted) curves of Cooling and of Statical 

 f^mperature shows that, to the amount of accuracy before mentioned, the 



