Determining the Absolute Density of a Gas. 383 



the volumenometer. All must now be let to stand till t ° is 

 attained (half-an-hour will suffice), and the final adjustment 

 of the mercury level to the mark / then effected. By means 

 of a cathetometer the height A, the difference of level between 

 the surfaces at d and/, is read. This will only be some 5 or 

 6 millim. in general. For this a temperature correction is 

 hardly required ; but a capillary correction, ascertained ex- 

 perimentally once for all, must be applied owing to the 

 different areas of the surfaces in the two tubes. This might 

 be eliminated in the construction by conferring an equally 

 small area on d, but it is probably more convenient to make 

 the simple correction required. The barometer is now read, 

 and the pressure of the gas in g computed. Its temperature 

 is 0° C, and its volume that of the vessel, = V. The sphere 

 is now returned to the balance, re- weighed, and the weight of 

 gas, subject to a small correction, found by difference. 



The correction mentioned is due to the shrinkage of the 

 copper sphere on the relief of pressure, and corresponds to 

 that occurring in the case of Kegnault's glass vessel. Ex- 

 periments (two, closely agreeing, have been made on this 

 sphere) reveal a shrinkage which may be taken as 0*1732 cub. 

 centim. on the removal of 4*3525 grammes of air at the 

 temperature 12° C, a fall of pressure of 22*01 atmospheres, 

 closely. At 760 millim. this weight of air occupies about 

 3*367 litres. This would be a needlessly large volume to 

 confer on the volumenometer, except for very special work. 

 Even in the case of hydrogen (1 litre at 760 and 0° C, weigh- 

 ing 0*0895 gram.) two litres will suffice for a very accurate 

 estimation of the weight of gas dealt with. At the pressure 

 corresponding to the compression of two litres into the sphere 

 of 160 cub. centim. capacity, a much lighter weighing-sphere 

 than that previously described will suffice. Thus I use one 

 weighing but 41 grammes, which has been tested up to 500 lb. 

 in the square inch. The thickness of its walls is somewhat 

 less than half a millimetre, its volume a little greater than 

 160 cub. centim. There is no difficulty in weighing such a 

 vessel to the one-twentieth of a milligram., or closer. Using 

 the heavier vessel the correction for shrinkage will be the 

 shrinkage due to two litres, i.e. 0*1029 cub. centim. multi- 

 plied into the normal density of air. This gives a displace- 

 ment effect of 0*000127. This correction is on a weight of 

 2*5878 gram, of air or the correction on w, which is sub- 

 tractive, is for air one part in 25,000 q.p. In the case of 

 hydrogen one part in 1790. Assuming the use of the lighter 

 sphere the experimentally determined shrinkage is 26359 

 cub. centim. for a lowering of pressure of 12*82 atmospheres, 



2 D 2 



