164 PROCEEDINGS OF THE AMERICAN ACADEMY 



Before beginning a combustion, the place of the hydrogen globe was 

 supplied by a straight piece of glass tubing, and the air current main- 

 tained through the heated combustion tube until the cupric oxide was 

 perfectly dry. Next the furnace at the extreme left was lighted, and 

 the current continued until the oxygen in the air of the drying tubes 

 had been so far replaced by nitrogen as to remove all risk of sub- 

 sequent explosion. 



Meanwhile the globe and condensing tubes were made ready, and 

 first the globe and next the condensing tubes were placed in position, 

 all the rubber connectors required having been previously dried in the 

 current of dry air, and the joints were so contrived that subsequently 

 the stream of gas came in contact with the smallest possible surface 

 of the rubber connectors. The combustion lasted, as has been already 

 stated, from seven to eight hours, and during this time quite a rapid 

 current of air was drawn through the apparatus. 



Our preliminary experiments plainly showed that the rapidity of 

 the current within practicable limits had no appreciable effect on our 

 results, and this is due to the fact that the current entered the globe 

 and left the condensers under precisely the same conditions. More 

 than nine tenths of the water was condensed during the first half-hour, 

 the drops falling regularly from the mouth of the inlet tube, and after 

 two hours all traces of cloudiness disappeared from this tube or its 

 connections, showing that the air coming over was perfectly dry. The 

 water thus collected was absolutely clear and limpid. 



After the first hour the combustion furnace used for removing oxy- 

 gen from the air, at the extreme left, was taken away, and by the end 

 of the combustion the reduced copper in the combustion tube proper 

 was again completely oxidized, leaving the globe and all the tubes 

 filled with normal air, as at the beginning of the process. It only 

 remained now to remove the condensers, and reweigh them with all 

 necessary precautions. At both weighings the barometer and ther- 

 mometer were observed, and the small, usually insignificant correction 

 for buoyancy caused by a change in the atmosphere during the interval 

 was carefully estimated. 



The question of the time of running the combustion after the pro- 

 duction of water had sensibly ceased was one that was carefully con- 

 sidered. The time mentioned above — eight hours — was far outside 

 the necessary limits, and was reached only after a large number of 

 experiments. That a very long continuance of the current after the 

 combustion was practically ended was unnecessary, was clearly shown 

 by several circumstances. In the first place, the duration of the com- 



