OP ARTS AND SCIENCES. 161 



generators employed. On turning the stopcock, the side aperture is 

 first closed, and then the direct way slightly opened, so that all the 

 gas evolved now passes into the globe ; and it was found possible to 

 regulate the current with such nicety as not to cause any sudden 

 changes of tension in the generator, which was always provided with 

 an overflow by means of which the tension could be watched, and 

 according to which the stopcock was regulated. 



The filling of the globe was one of the critical points of the deter- 

 mination. It generally occupied from one to two hours, and during 

 all this time it was necessary, with the hand on the stopcock, carefully 

 to watch the tension at the overflow already mentioned, and repre- 

 sented in Fig. 5, and Figs. 7 and 8 of Plate. From the beginning 

 to the end of the operation there was a greater tension in the generator 

 than in the outside air, by about one inch of mercury. When the 

 connection was once established between the generator and the globe, 

 there was absolutely no leakage through the side way, as was tested 

 in several cases by dipping the mouth of the tube at the base of the 

 cock under mercury. 



The whole process of weighing the hydrogen was finally reduced to 

 the following manipulation. The globe was connected by means of a 

 rubber hose with a rotatory air-pump having automatic valves, made 

 by E. S. Ritchie of Boston. It was then exhausted to within 1 mm. 

 of mercury. Next, closing the cock and disconnecting the globe, it 

 was cleaned with distilled water and fine cotton cloth ; at least, this was 

 done five or six times during the determinations. But as the glohe 

 when out of the balance case was always protected by a cylindrical tin 

 box with a cover, from which the exit tubes projected, it was usually 

 only necessary to clean the exit tubes in this careful manner, simply 

 dusting off the globe with a large camel's-hair brush, before hanging 

 it in the balance case. In this part of the operation it was necessary 

 to take care not to communicate to the globe a charge of electricity by 

 rubbing it with a perfectly dry cloth. 



The globe was hung on a wire stirrup, which caught the exit tubes, 

 as the glass joint was sufficiently strong to support the weight of the 

 globe. The globe was so nearly balanced by its equipoise that when 

 exhausted it only required about one decigram to establish equilibrium. 

 The time required to attain perfect equilibrium varied with the con- 

 ditions. If the glass had been previously cleaned as described above, 

 perfect equilibrium might not be reached for forty-eight hours, or even 

 longer, while if the glass had only been dusted, twelve hours were gen- 

 erally sufficient. 



vol. xxm. (n. s xv.) 11 



