186 PROCEEDINGS OP THE AMERICAN ACADEMY 



sured us that this could not be the case. During the first stage of the 

 combustion, when pure hydrogen was passing into the combustion 

 tube, and while water was dropping into the condenser (Figure 4, 

 page 163), there would often be several minutes — during which the 

 larger part of the water was condensed — when no residual gas what- 

 ever would be seen to escape, and the bubbling of the gas through 

 the sulphuric acid at the bend of the U tube made the least overflow 

 perfectly evident. Again, the oxide of copper in the combustion tube 

 was always reduced to a perfectly definite limit, leaving at least seven 

 eighths of the tube in which the black oxide was apparently wholly 

 unchanged. Further, it is not probable that an error arising from the 

 imperfect combustion of the hydrogen would have a constant value. 

 The unconsumed residue must vary greatly with the conditions of the 

 experiment ; and such an agreement as that exhibited by the results on 

 page 173 could never have been obtained under such circumstances. 



It seems unnecessary to add, that every precaution was taken in our 

 work which our experience could suggest, and that a great amount of 

 labor was spent on such details which does not appear in the published 

 results. Both the balances and weights employed were most carefully 

 verified. The water formed by the combustion was tested, and the 

 dissolved air taken into account. We mention these points because 

 they have been noticed by correspondents ; but many similar details 

 which were worked out and set one side we have not thought it ne- 

 cessary to describe in our paper. In writing such a paper elementary 

 principles must be assumed. 



In adopting Regnault's method for weighing the hydrogen used in 

 our determinations, we assumed with him that the glass balloon used 

 in the work remained practically constant, whether exhausted or filled 

 with gas. We never questioned this assumption, not only because we 

 had the greatest confidence in all Regnault's work, but also because 

 we knew that he had himself carefully investigated the behavior of 

 glass bulbs under pressure ; and indeed he treats the subject fully in 

 the paper immediately preceeding his classical paper on gas density.* 

 Moreover, we made with our apparatus a preliminary determination 

 of the density of air, and obtained Regnault's number within the 

 limits of the uncertainty in regard to the value of the force of gravity 

 at this place. Regnault's values for the weight of one litre, not only of 

 air, but also of nitrogen, oxygen, hydrogen, and carbonic dioxide, have 



* Memo-ires de 1'Acad. Hoy. des Sciences de l'lnst. de France, vol. xxi. 

 pp. 106 and 121. 



