15 



obtained with alloys, which consisted mainly of mercury, appeared 

 very anomalous ; it seemed as if a very small per-centage of even the 

 best conducting metals reduced immensely the conducting power of 

 mercury. But it was suggested to the authors, that the apparently 

 high conducting power of mercury obtained by their method, was 

 probably due to the transference of heat by convection ; that the 

 real conducting power of mercury for heat was low, like its conduct- 

 ing power for electricity ; that the other metal, contained in small 

 quantity in the amalgam, acted by rendering the amalgam viscous, 

 and thereby interfering with the transference of heat by convection, 

 and that the low conducting power of mercury would show itself on 

 merely inclining the vessel used in the experiment, so that the box 

 containing the warm water should be higher than the other. Ex- 

 periment confirmed this view. As the apparent conducting power 

 of mercury was found continually to decrease with an increase in the 

 inclination of the vessel, it was found necessary, in order to obtain 

 correct results, to arrange so that the bar-shaped box containing the 

 mercury or fluid amalgam was actually vertical in the experiment. 

 In this way the authors obtained for mercury the figure 54, on the 

 same scale as before. It is worthy of remark, that mercury comes 

 out the worst conductor of all the metals tried, the figure for 

 bismuth, which had previously been the lowest, being 6 1 . This is 

 in analogy with water, also a fluid, the conducting power of which 

 is known to be excessively low. The conducting power of the more 

 fluid amalgams determined by experiment with the box vertical, 

 proved to be in all cases nearly the same as that of pure mercury, 

 in conformity with the law mentioned by the authors in their former 

 paper, that alloys in which there is an excess of the number of 

 equivalents of the worse conducting metal, over the number of 

 equivalents of the better conductor, do not sensibly differ in con- 

 ducting power from the worse conductor alone. In the case of 

 amalgams generally, the conducting power obtained by experiment 

 was found to agree pretty closely with the number calculated from 

 the per-centages and conducting powers of the component metals. 



In conclusion, the authors give some further experiments on the 

 conduction of heat by compound bars, formed of metals placed in 

 some cases end to end in alternate cubes, in other cases side by side 

 in parallel bars, extending the whole length of the compound bar. 



