78 



heat, It was an important improvement to contrive the ap- 

 paratus in such a manner that the hydro-carbon should be con- 

 stantly maintained at an adequate temperature by the proxi- 

 mity of its own combustion. This was at last effected by very 

 much shortening the burner and gas-ways, and connecting the 

 brass tube which held the wire-gauze with the burner, so that 

 the heat of the flame was transmitted directly downwards, by 

 conduction, to the wire-gauze and hydro-carbon. 



*' Hydrogen, as obtained by the processes of the laboratory, 

 is exceedingly expensive ; it therefore became necessary to 

 ascertain whether 1 might substitute for it that very light car- 

 bureted hydrogen which is procured by passing the steam of 

 water over charcoal or coke, maintained at a bright red heat 

 in an iron retort. This gas contains so little carbon, that, for 

 practical purposes, it may be considered as hydrogen ; and it 

 was essential to know whether, when used in the above-men- 

 tioned manner, it would create a sufficiency of heat to volati- 

 lize the hydro-carbon. 



" In order to ascertain this and other important particulars, 

 I caused an iron gas retort of the ordinary kind, with a well 

 ground mouth-piece, to be built into a furnace erected for that 

 purpose. The retort was capable of containing two pounds 

 of Newcastle coke ; it was furnished with a tube and stop- 

 cock at one end for the admission of steam : and a tube at the 

 other end for carrying off the gas generated by the decom- 

 position of the water into a gasometer containing milk of 

 lime. 



" When the retort was maintained at a bright red heat, and 

 the steam let on, gas was generated in such torrents that at 

 onetime I feared some untoward result. By calculation from 

 this experiment I found that to furnish 1000 cubic feet of gas 

 168 avoirdupois pounds of coke should disappear. The inte- 

 rior of the iron retort had been previously protoxidized, to 

 prevent, as much as possible, its contributing to the produce 

 tion of hydrogen. But 1 found that, although the scale of 



