128 CLASSICS OF MODERN SCIENCE 

 mable substance possessed the property of depriving the air of part of 

 its elasticity, and that in consequence of this it becomes more closely 

 compressed by the external air. In order now to help myself out of 

 these uncertainties, I formed the opinion that any such air must be 

 specifically heavier than ordinary air, both on account of its contain- 

 ing phlogiston and also of its greater condensation. But how per- 

 plexed was I when I saw that a very thin flask which was filled with 

 this air, and most accurately weighed, not only did not counterpoise 

 an equal quantity of ordinary air, but was even somewhat lighter. I 

 then thought that the latter view might be admissible ; but in that case 

 it would necessarily follow also that the lost air could be separated 

 again from the materials employed. None of the experiments cited 

 seemed to me capable of showing this more clearly than that accord- 

 ing to the tenth paragraph, because this residuum, as already men- 

 tioned, consists of vitriolated tartar and alkali. In order therefore to 

 see whether the lost air had been converted into fixed air, I tried 

 whether the latter shewed itself when some of the caustic ley was 

 poured into lime-water ; but in vain — no precipitation took place. In- 

 deed, I tried in several ways to obtain the lost air from this alkaline 

 mixture, but as the results were similar to the foregoing, in order to 

 avoid prolixity I shall not cite these experiments. Thus much I see 

 from the experiments mentioned, that the air consists of two fluids, 

 differing from each other, the one of which does not manifest in the 

 least the property of attracting phlogiston, while the other, which 

 composes between the third and the fourth part of the whole mass 

 of the air, is peculiarly disposed to such attraction. But where this 

 latter kind of air has gone to after it has united with the inflammable 

 substance, is a question which must be decided by further experiments, 

 and not by conjectures. 



