Prof. Magnus on the Propagation of Heat in Gases. 89 



' Thermochrose,' page 59. As far as 14 0, 5 the strengths of the 

 current were proportional to the deviations. 



In order to fill the apparatus with any given gas, it was first 

 exhausted, the gas admitted by the stopcock H, then again 

 exhausted and filled a second time with the gas, and so on for 

 four times, upon which the atmospheric air could be considered 

 to be completely removed. With gases like cyanogen and 

 ammonia which attack the pump, the filling was performed by 

 displacement, the gas being admitted by the stopcock C, and 

 escaping by the stopcock K, under the plate T T. For this pur- 

 pose the whole apparatus, with the plate T T, was removed from 

 the air-pump and placed on a tripod. On the lower part of the 

 stopcock K a glass tube was fitted, through which the gas passed 

 into an absorbent liquid. If the gas was to be used in a rarefied 

 state, the rarefaction was effected, after the filling was complete, 

 by means of the air-pump. The rarefaction thus produced was 

 either directly observed by the barometer, or a manometer was 

 introduced, which was read off by means of a cathetometer. Thus 

 the most different gases could be examined as to their capacity 

 of transmitting heat, with the exception of those which attacked 

 the metal of the pile. This excluded, to my great regret, all 

 coloured gases. 



The gases were prepared exactly as in the experiments on 

 conduction. 



As the intensity of the galvanometer might have changed in 

 the course of time, in almost every case before the apparatus was 

 filled with a new gas the radiation through atmospheric air was 

 determined. In this way the relation of the radiation in the 

 particular gas to that of atmospheric air was obtained. This 

 method of comparison I have always retained, for it ensures 

 great certainty. In experiments with boiling water this compa- 

 rison was ultimately found superfluous, for the galvanometer 

 remained so unchanged that the values obtained at different 

 times for atmospheric air agreed very closely. Nevertheless in 

 the following Table the observed deflections are so arranged that 

 the control experiments with atmospheric air are found in one 

 column, and the gases examined either directly before or after 

 are placed opposite, being separated from the rest by a horizontal 

 line. 



For the gases which exhibit the greatest deviation, the radia- 

 tion has been determined at different times. Since the num- 

 bers obtained agreed as closely as could be expected with such 

 experiments and with such angles, I have only adduced one series. 



