400 Lord Rayleigh. Chi the [Feb. 18, 



The agreement is satisfactory, especially when it is remembered 

 that both gases were prepared by different methods in the two sets of 

 experiments. Feb. 17.] 



But there is still another and a more important correction to be 

 introduced. In my former paper it was shown that when the weigh- 

 ings are conducted in air the true weight of the gas contained in the 

 globe is not given by merely subtracting the weight of the globe 

 when empty from the weight when full. When the globe is empty, 

 its external volume is less than when full, and thue, in order to obtain 

 the true weight, the apparent weight of the gas must be increased by 

 the weight of air whose volume is equal to the change of volume of 

 the globe. 



In order to determine the amount of this change of volume, the 

 globe is filled to the neck with recently boiled distilled water, and 

 the effect is observed npon the level in the stem due to a suction of, 

 say, 20 inches of mercury. It is not advisable to carry the exhaus- 

 tion much further for fear of approaching too nearly the point at 

 which bubbles of vapour may be formed internally. In the earlier 

 experiments, described in the preliminary note, the upper surface of 

 the liquid was in the stem of the globe itself (below the tap), and 

 the only difficulty lay in the accurate estimation of a change of 

 volume occurring in a wide and somewhat irregular tube. The 

 method employed was to produce, by introduction of a weighed 

 quantity of mercury, a rise of level equal to that caused by the 

 suction. 



The advantage of this procedure lay in the avoidance of joints 

 and of the tap itself, but, for the reasons given, the readings were 

 not quite so accurate as might be desired. I wished, therefore, to 

 supplement, if possible, the former determination by one in which 

 the change of volume occurred in a tube narrower and of better 

 shape. With this object in view, the stem of the globe was prolonged 

 by a graduated tubular pipette attached with the aid of india- 

 rubber. The tubes themselves were treated with gutta-percha cement, 

 and brought almost into contact. It had hardly been expected that 

 the joint would prove unyielding under the applied suction, but it 

 was considered that the amount of the yielding could be estimated 

 and allowed for by operations conducted with tap closed. The event, 

 however, proved that the yielding at the joint was scarcely, if at all, 

 perceptible. 



The pipette, of bore such that 16 cm. corresponded to 1 c.c., was 

 graduated to O'Ol, and was read by estimation to 001 c.c. In order 

 the better to eliminate the changes due to temperature, readings 

 under atmospheric pressure, and under a suction of 20 inches of mer- 

 cury, were alternated. On January 28, 1892, a first set gave 0'648 

 0-300 = 0-348, a second gave 0'6645-0'316 = G'3485 and a third gave 



