1879.] B u ff' s Experiments on the Diathermancy of Air. 11 



pouring of hot water into the vessel. From this source the heat pro- 

 ceeded through air or any other gas introduced into the cylinder 

 underneath. To measure the heat transmitted, a flattened thermo- 

 electric couple was fixed, in the first experiment, at a distance of 

 23 millims. below the polished surface ; two wires being carried from 

 the couple through the plate of the air-pump to a suitable galvano- 

 meter. 



It immediately appeared that the experiments were vitiated by the 

 heat conducted from the source down the glass cylinder. Below the 

 level of the thermo-electric couple this heat was even sensible to the 

 touch. Currents, Professor Buff rightly inferred, would, under the 

 circumstances, be unavoidable. He would have free convection,. 

 whereas his object was to investigate conduction. He further con- 

 cluded, also rightly, that such currents were present in the experi- 

 ments of Magnus,* although through the disposition of his apparatus,, 

 "he probably failed to appreciate the importance " of this source of 

 error. 



Thus warned, Professor Buff changed his mode of experiment, and 

 the change will be best described and illustrated in his own words : — 

 u I had observed," he says, " that if the hot liquid in the brass vessel 

 was allowed to cool freely, the astatic needle, nevertheless, after some 

 time, assumed a maximum deflection, indicating a state of equilibrium 

 between the influx and loss of heat. It is known that tolerably strong 

 deflections of the needle of a highly sensitive galvanometer corre- 

 spond to but slight alterations of temperature at the soldered 

 junction. If, therefore, we know the temperature of the brass vessel 

 (or rather that of its bottom) at the commencement of the experi- 

 ment, and the temperature of its fluid contents for which the 

 deflection of the needle is greatest, then the thermal equilibrium, f 

 upon which the highest temperature at the soldered junction depends, 

 must be very nearly proportional to the difference of these two tem- 

 peratures of the brass vessel, at least so long as the thermal dif- 

 ferences are small. 



" Accordingly the brass vessel, at the commencement of the experi- 

 ment, was filled with water at such temperature that the needle 

 pointed to 0°, and remained stationary. This having been effected, 

 part of the cold water was taken out and replaced by hot water ; the 

 temperature of the mixture was observed, and simultaneously the 

 time of pouring in was noted. The temperatures were taken by a 



* The existence of these currents was not only inferred, but they were rendered 

 visible to the eye many years ago, by means of the illuminated fumes of chloride of 

 ammonium. See section headed " Proof of Convection," Contributions to Molecular 

 Physics, p. 381. 



f I have found some difficulty in seizing the exact connexion between this 

 expression and the words which follow it. 



