202 E. W. Morley — Moisture in Gas. 



required brass weights on the other side were also so nearly 

 equal, that no variations of a decimilligram were noted in 

 the eight, ten or fifteen days used in the weighings. Subtract- 

 ing the difference of weight of caps and fat from the difference 

 of total weight, we have the difference of weight of tubes efgh, 

 and eld with their contents, solid or gaseous. Now, since care 

 was taken to close the tubes at the end of the experiments at 

 an hour when the barometer and thermometer showed the 

 weight of a given volume of air to be the same as when the 

 tubes were closed at the beginning of the experiment, a com- 

 parison of the two differences thus determined at the beginning 

 and end of an experiment would show the loss of weight of 

 the contents of efgh. It was assumed that the difference of 

 volume between H 2 0+P„0 6 before; and 2HP0 3 after, an ex- 

 periment was negligible for the quantity of water used. 



Using many precautions too minute to be detailed, there was 

 no difficulty in determining the difference of weight of the two 

 apparatus as accurately as the difference of two brass weights. 

 The whole weight on each pan was 35 grams ; a Becker 

 balance intended to carry 100 grams in each pan was used. 

 Weighings could therefore be made to the fortieth of a milli- 

 gram. The tubes were left untouched for many days, in a 

 room without artificial heat, surrounded by rooms kept at a 

 pretty constant temperature, and their weights compared from 

 twenty to fifty times. The results obtained were as constant 

 as were obtained on the same balance in ascertaining the rela- 

 tions of the set of weights used with it. 



The details just stated show that it is not impossible to 

 secure equality of hygroscopic state in two equal surfaces 

 freshly made of the same glass. The outer surfaces of dd and 

 of efgh therefore do not interpose any serious difficulty in the 

 way of accurate weighing. But the inside of dd has to be in 

 the same hygroscopic state at the beginning and the end of an 

 experiment lasting many weeks. 



To secure this, dry air from c was passed through d for a 

 long time while it was heated to 200° or 250° C. When d had 

 thus been made as dry as possible at this temperature, the end 

 of d remote from c was closed with its weighed cap and charge 

 of fat. efgh was connected with i at one end, and closed by its 

 cap at the other, dd and efgh were left side by side for a day 

 to attain equality of temperature and pressure, disconnected 

 from c and i, and closed with their second caps at the same 

 time. After weighing, dd and efgh were connected as shown 

 in the figure. Now, since during the experiment, only air 

 dried in c passes through d, and since the experiment proves 

 that this air contains no appreciable amount of moisture, the 

 interior of d absorbed no moisture ; and it lost none, for its 



