428 Prof. Tyndall on the Absorption arid 



the knowledge of this fact has rendered me careful to remove my 

 polished plates every evening from the apparatus, and to keep 

 them in perfectly dry air. Still, when it is remembered that 

 the air on entering the tube is raised in temperature and thus 

 enabled to maintain a greater amount of vapour, and that the 

 tube and plates of rock-salt form the channel for a flux of heat 

 from the radiating source, the likelihood of precipitation occur- 

 ring will seem but small. On examining the plates after the 

 undried air of the laboratory was experimented with, no trace of 

 precipitated moisture was observed upon their surfaces. 



But, to place the matter beyond all doubt, I abolished the 

 plates of rock-salt altogether, and operated thus : — An india- 

 rubber bag (B) was filled with air, and to its nozzle a T-piece, 

 with the cocks Q Q', was attached. The cock Q' was connected 

 with two tubes, U 7 U', each of which was filled with fragments 

 of glass moistened with distilled water. The cock Q was con- 

 nected with the tubes U U, each of which was filled with frag- 

 ments of glass moistened by sulphuric acid. The other ends of 

 these two series of tubes were connected with the cocks 0'; and 

 from the T-piece between these cocks a tube led to the end E' of 

 the open experimental tube T. The cock A at the other end 

 of the experimental tube was placed in connexion with an air- 

 pump. The pile P, the screen S, and the compensating cube C 

 were used as in the other experiments. E is the end of the front 

 chamber, and C the source of heat. In some experiments I had 

 the end E closed by a plate of rock-salt, in others it was allowed 

 to remain open, a distance of about 12 inches intervening 

 between the radiating surface and the open end E' of the expe- 

 rimental tube. 



Closing the cocks Q and 0, and opening Q' and ; , gentle 

 pressure being applied to the bag B, a current of moist air was 

 slowly discharged at the end E' of the experimental tube. The 

 pump in connexion with A was then worked; and thus by degrees 

 the air was sucked into the tube T. The deflection of the galva- 

 nometer was 30°, when the moist air filled the tube as completely 

 as the arrangement permitted*, — this deflection being due to 

 the* predominance of the compensating cube over the radiating 

 source C. 



The cocks Q' and 0' were now closed, and Q and opened ; 

 proceeding as before, a current of dry air was discharged at E', 

 and this air was drawn into the tube T in the manner just 

 described. The moist air was thus displaced by dry ; and, while 

 the displacement was going on, the galvanometer was observed 

 through the distant telescope. The needle soon commenced to 

 sink, and slowly went down to zero, proving that a greater quan- 



* Still, of course, only partially. 



