408 Prof. H. BufF on the Thermal Conductivity/ 



A comparison of the two Tables shows that the effects upon 

 the junction were somewhat greater on the removal of the 

 screen, but they did not alter essentially. Since the screen 

 appeared thus superfluous, and might even have disturbed the 

 uniformity of the experiments, it was henceforth emitted. 



The propagation of heat in air, which at first had exceeded 

 that in vacuum, upon cooling by water considerably fell below 

 the latter. Now, since the thermal motion in strongly rarefied 

 air could essentially be caused only by radiation, it follows 

 that air of atmospheric density must exercise a considerable 

 absorptive power on such rays as have proceeded from a source 

 of moderate temperature. In the first experiments, when no 

 cooling-water was used, this property of air must have been 

 concealed by the upward currents conveying heat to the 

 soldered junction. But the same disturbing influence must 

 have manifested itself in the experiments with hydrogen, and 

 favoured the passage of heat through this gas, independently 

 of its supposed conductivity. 



The last two series of experiments likewise show, notwith- 

 standing the use of the cooling-water, an accelerated thermal 

 motion in hydrogen, as Magnus had found, which led him to 

 infer the metal-like conductivity of that gas. It was, however, 

 remarkable (and this was noticed in several experiments the 

 results of which closely agreed) that just when free circulation 

 of the gas was most impeded, viz. when the screen was used, 

 the difference between the passage of the heat on the one hand 

 through dense hydrogen, on the other hand through rarefied 

 hydrogen, was found to be least. But other circumstances led 

 me to suppose that, even with the cooling water, currents were 

 not entirely prevented during the experiments. 



The return of the needle from maximum towards zero by no 

 means corresponded to the decrease in the temperature of the 

 water. Indeed, even upon suddenly reducing the latter to the 

 temperature of the surrounding atmosphere, the needle remained 

 for some time deflected. The diminished influence of the hot 

 water, consequent on its gradual cooling, was therefore par- 

 tially compensated for by the fact that the sides of the cylinder 

 did not with proportionate rapidity yield up to the cooling- 

 water the heat which they had received from the water in the 

 brass vessel. It was, in fact, found after each experiment 

 that the upper layer of the cooling-water had become appre- 

 ciably warmer, though the difference at the height of the sol- 

 dered junction was seldom more than one degree, frequently 

 even less. I noticed, however, that even such small differ- 

 ences of temperature had an observable effect upon the needle ; 

 but these differences of temperature may have been somewhat 

 greater on the inside of the glass cylinder. 



