E. W. Morley— Moisture in Gas. 203 



temperature was far below that at which it had been dried. 

 Therefore the hygroscopic state of the interior was constant. 



When dd and efgh had been prepared and weighed, they 

 were connected, as described to the tube i. This tube, in- 

 tended to prevent the passage of moisture from the air pump 

 to the tube efgh, was fused to a tube carrying a syphon barome- 

 ter gauge, and a mercurial stop-cock leading to the air-pump, 

 and connected to a copper vessel holding 54*1 liters. The con- 

 nection of the glass tube to the copper reservoir was immersed 

 in water: this water jacket and the mercurial stop-cock secured 

 the certainty that all air entering the receiver must come 

 through the tubes c to i. The pressure in the receiver was 

 reduced till the flow through h was about three liters an hour, 

 and was measured on the gauge. When the pressure had risen 

 by a suitable amount, it was again measured, and again 

 reduced. When the water in e and / was exhausted, air was 

 slowly admitted to the receiver by breaking off a sealed capil- 

 lary tube ; when the pressure in the receiver was equal to that 

 of the atmosphere, a free communication was made between 

 the air and the tubes i and c, and when the air had a suitable 

 density, the caps were applied and weighings made as already 

 described. 



In an experiment lasting through December, 1886 and Janu- 

 ary, 1887, 726 liters of air passed through c and d. They were 

 expanded to 2428 liters in g, and passed through h at the rate 

 of 70 liters in a day. The excess of the volume passing out 

 was 1702 liters, and the loss of weight in the tube efgh was 

 rather less than the twentieth, of a milligram. In an experi- 

 ment lasting from February to April, 1887, the volume of. air 

 entering was 211 liters, the expanded volume, which passed at 

 the rate of 75 liters in a day, was 284] liters, and the excess 

 was 2630 liters. The loss of weight of the apparatus was 

 rather more than one twentieth of a milligram. The meas- 

 ured volumes may well be in error by one or two per cent. 



The weight of aqueous vapor left in a gas passing at a rate 

 of three liters in an hour through a tube two centimeters in 

 diameter and eight centimeters long compactly filled with phos- 

 phorus pentoxide and glass wool, is therefore so little, that in 

 4300 liters it cannot be detected with much probability by 

 even the most delicate weighing. 



It is obvious that if at some future time physicists should 

 need to know the tension of the vapor of water in presence of 

 phosphorus pentoxide, an experiment by this method must be 

 greatly prolonged. With a current of three liters an hour, the 

 effect measured was not over a decimilligram in two months. 

 To obtain some such quantity as ten milligrams might there- 

 fore require ten years. But for the purposes of the chemist 

 the present two concordant results may for a time be sufficient. 



