OF ARTS AND SCIENCES. 203 



vitiated to a small extent by the circumstance that when the balloon 

 is exhausted the pressure of the atmosphere determines a slight shrink- 

 age of the external volume, which naturally disturbs the exactness of 

 the compensation between the buoyancy of the air on the balloon and 

 on its counterpoise. Although this shrinkage can be readily meas- 

 ured, as was done by Dr. T. W. Richards under my direction,* and still 

 more recently by Professor Crafts, t who experimented on the balloon 

 used by Reguault, which fortunately has been preserved, it seemed de- 

 sirable to develop a method by which this correction could be avoided. 

 For even if the new method should lead to no more accurate results 

 than before obtained, it might serve to confirm the validity of the cor- 

 rection iu question, and at least would gi\e additional data towards 

 establishing the value of important physical constants. 



The new method we have devised for the purpose consists in first 

 taring the balloon when filled with carbonic acid gas, and then drawing 

 the gas through absorption tubes and determining its weight, as in the 

 well known method of organic analysis. This weight known, the tare 

 of the empty balloon is obviously the difference between the first tare 

 and the weight in question. The practical problem here presented is, 

 however, far more difficult than that of organic analysis. In the last, 

 we expect to determine the weight of only a few decigrams of car- 

 bonic acid within a few tenths of a milligram, while in the problem 

 now before us we must weigh at least nine or ten grams of carbonic 

 acid, not simply to a proportional, hut to an equal, degree of accuracy. 

 We only succeeded in securing such accuracy after many trials and a 

 careful study of all the conditions involved, and our primary object in 

 this paper will be to describe the precautions which are essential to 

 the success of the new experimental method. Incidentally it will appear 

 that our results confirm in a most striking manner the high value of 

 the specific gravity of hydrogen found by Lord Rayleigh,± and the low 

 value of the atomic weight of oxygen found by ourselves. 



The Balance and Weights. 



The balance and weights were the same as those used by us in our 

 previous work.§ The disposition of the apparatus whirl., after many 

 trials, we have found most suitable for accurate work, is shown in 



* The.-e Proceedings, vol. xxiii. p. 177. 1888. 



t Comptes Rend us, vol. cvi. p. 16G'2. 188b. 



j Proceedings of the Royal Society, vol. xlv. p. 426. 1889 



§ These Proceedings, vol. xxiii. p. 159. 1* 



