112 Barrerr, Brown, AND HaprirLD—Researches on the 
means of a pair of compasses from time to time, until it was found that a perfectly 
steady fall of temperature along each rod was attained. 
In order to have a strictly comparative series of experiments, a standard rod 
of nearly pure iron, marked B by the maker, was included in each set of determina- 
tions, so that seven alloys, together with the standard B, were simultaneously 
tested. The rod B was of the same length and cross-section as the other rods, and 
was the specimen used as the standard in the previous determinations of the 
magnetic properties of these alloys. It contained only 0:03 per cent. of carbon, 
the total impurities amounting to less than 0°3 per cent. This rod, and all the 
iron alloys tested, had been carefully annealed. It would be desirable to repeat 
the experiments with hardened rods for the reason subsequently stated, and this 
we hope to do later on. 
In the tables that follow, the squares of the lengths of the melted wax, that is, 
the relative thermal conductivities, are compared with the electrical conductivities 
of the same specimens, the iron standard used in both cases being taken as 100. 
It will be seen at once that both the thermal and electrical conductivities follow 
the same order. 
This relationship between thermal and electrical conductivity, first remarked 
by Forbes, and more fully investigated by Wiedemann and Franz, has been the 
subject of much discussion and experiment. Wiedemann and Franz concluded 
that not only was the same order followed, but that the ratio of thermal con- 
ductivity & to electrical conductivity *%’ was the same for all metals and all 
temperatures. This, however, has been disputed by Tait, Weber, and Berget, 
though all agree that the order of conductivity in both cases holds good. Lorentz,* 
whose experimental results have been regarded as among the best yet obtained, 
found that this ratio //k’ was nearly constant for all good conductors at any one 
temperature, but increased in value nearly in the same proportion as the rise of 
absolute temperature. For 0° C., Lorentz found this ratio, when the con- 
ductivities were reduced to C. G. S. units, to be 1500 for the best conductors, and 
2000 for the worst conductors. But, as Professor Callendar has pointed out, this 
relation cannot hold generally true, for in a specimen of cast iron—the thermal 
and electrical conductivities of which he himself carefully measured—this ratio 
amounted to 12,500. 
All, therefore, that we can assert at present is that the ionic theory of conduc- 
tivity receives considerable support from the fact that some close relationship does 
appear to exist between thermal and electric conductivity, in that the causes which 
affect the one appear also to affect the other; albeit the diffusivity of heat through 
a metal is enormously slower than the diffusivity of electricity. 
* Wied, Ann. xiii.,“p. 422. 
