1892.] Relative Densities of Hydrogen and Oxygen. 461 



0*675 0/326 = 0*34:9. Similar operations with tap closed* gave no 

 risible movement. 



The result of the day's experiments was thus 0*3485 for 20 inches, 

 or 0'523 for 30 inches, suction. Similar experiments on January 28, 

 at a different part of the graduation, gave 0*526. On this day the 

 yielding with tap closed was just visible, and was estimated at 0*001. 

 As a mean result, we may adopt 0*524 c.c. The graduation of the 

 pipette was subsequently verified by weighing a thread of mercury 

 that occupied a measured length. 



A part of the above-measured volume is due to the expansion of 

 the water when the pressure is relieved. We may take this at 

 0'000047 of the volume per atmosphere. The volume itself may be 

 derived with sufficient accuracy for the present purpose from the 

 weight of its oxygen contents. It is 2-517/0*00137, or 1837 c.c. 

 The expansion of the water per atmosphere is thus 0*000047 x 1837, or 

 0*087 c.c. This is to be subtracted from 524, and leaves 0*437 c.c. 

 This number applies strictly to the volume enclosed within the glass, 

 but the change in the external volume of the globe will be almost the 

 same.f 



The correction now under consideration is thus the weight of 

 0*437 c.c. of air at the average temperature of the balance room. 

 The density of this air may be estimated at 0*00122 ; so that the 

 weight of 0-437 c.c. is 0*000533 gram. This is the quantity which 

 must be added to the apparent weights of the gases. The former 

 estimate was 0'00056 gram. The finally corrected weights are 

 thus 



H = 0*158531, = 2*51777; 



and for the ratio of densities we have 



15-882. 



This corresponds to a mean atmospheric condition of pressure and 

 temperature. 



If we combine the above ratio of densities with Professor Morley's 

 ratio of volumes, viz., 2'0002 : 1, we get, as the ratio of atomic 

 weights, 15*880. 



If we refer to the table, we see that the agreement of the first and 



* For greater security the tap was turned while the interior was under suction. 



f For a spherical shell of glass of uniform thickness and with elastic constants 

 following Poisson's law, the ratio of the difference of the internal and external 

 expansion to either of them is 4 tfia, where t is the thickness of the shell, and a 

 the mean radius. In the present application the value of /, deduced from the 

 measured circumference and from the weight of glass, is about 110. 



[Perhaps an arrangement in which the external volume is directly measured 

 would have been preferable. No allowance for expansion of water would then be 

 needed. Feb. 17.] 



