DECOMPOSITION OF WATER BY HEAT. 309 



by Leslie * respecting its convection of sound, but the parallel 

 does not hold : sound is transmitted imperfectly through 

 rarefied air, and also through hydrogen ; on the contrary, the 

 heat of the ignited wire is most intense in the former, and 

 least so in the latter ; the heat is also very much reduced in 

 intensity in the compounds of hydrogen, ammonia and olefiant 

 gas, or even by a small admixture of hydrogen with another 

 gas. such as nitrogen ; hydrogen, therefore, appears to have a 

 peculiar and specific action in this respect. 



I now pass to the consideration of the effects of the ignited 

 wire on different gases. The ignition was in every case raised 

 to the fullest extent, and the gases after exposure to it were 

 carefully cooled down to their original temperature. 



When the experiments were made over water, the whole 



Fig. 2. 



Fig. 3- 



Library 



eudiometer was immersed in a vessel of distilled water, occa- 

 sionally having an inch depth of oil on the surface (see fig. 2 f ) ; 

 when they were made over mercury, and a long-continued 

 exposure was required, a bent tube was employed, as at fig. 3, 

 the closed end being immersed in water or oil, to prevent the 

 fusion of the glass, which would otherwise have ensued. 



The tubes are much more easily preserved from cracking, 

 and the ignition better kept up with oil on the exterior than 

 with water ; but, as in many of these experiments I might 

 have been considerably misled by a crack in the glass, or a 



* Trans, of the Cambridge Phil. Sec., vol. i. p. 267. 



f In this and in figs. 3 and 5 the lines leading from the platinum loop to the 

 mercury cups represent copper wires. 



