VOL. I.VI.] PHILOSOPHICAL TRANSACTIONS. 303 



The specific gravity of each of the above-mentioned sorts of inflammable air, 

 except the first, was trjed in the following manner. A bladder holding about 

 100 ounce measures was filled with inflammable air, in the manner represented 

 in fig. 15, and the-air pressed out again as perfectly as possible. By this means 

 the small quantity of air remaining in the bladder was almost entirely of the in- 

 flammable kind. 80 ounce measures of the inflammable air, produced from zinc 

 by the vitriolic acid, were then forced into the bladder in the same manner: after 

 which, the pewter pipe was taken out of the wooden cdp of the bladder, the ori- 

 fice of the cap stopped up with a bit of lute, and the bladder weighed, A hole 

 was then made in the lute, the air pressed out as perfectly as possible, and the 

 bladder weighed again. It was found to have increased in weight 40-J^ grains. 

 Therefore the air pressed out of the bladder weighs 40|- grains less than an equal 

 quantity of common air; but the quantity of air pressed out of the bladder must 

 be nearly the same as that which was forced into it, i.e. 80 ounce measures: 

 consequently 80 ounce measures of this sort of inflammable air weigh 40|- grains 

 less than an equal bulk of common air. The three other sorts of inflammable 

 air were then tried in the same way, in the same bladder, immediately one after 

 the other. In the trial with the air from zinc by spirit of salt, the bladder in- 

 creased 404- grains on forcing out the air. In the trial with the air from iron, it 

 increased 41-1- grains, and in that with the air from tin, it increased 41 grains. 

 The heat of the air, when this experiment was made, was 50°; the barometer 

 stood at 29-J inches. 



There seems no reason to imagine, from these experiments, that there is any 

 difference in point of specific gravity between these four sorts of inflammable air; 

 as the small difference observed in these trials is in all probability less than what 

 may arise from the unavoidable errors of the experiment. Taking a medium 

 therefore of the different trials, 80 ounce measures of inflammable air weigh 41 

 grains less than an equal bulk of common air. Therefore if the density of com- 

 mon air, at the time when this experiment was tried, was 800 times less than 

 that of water, which, I imagine, must be near the truth,* inflammable air must 



* Mr. Hauksbee, whose determination is usually followed as the most exact, makes air to be more 

 than 850 times lighter than water; vide Hauksbee's experiments, p. 9*, or Cotes's Hydrostatics, 

 p. 159. But bis method of trying the experiment must in all probability make it appear lighter than 

 it really is. For having weighed his bottle under water, both when full of air and when exhausted, 

 he supposes the difference of weight to be equal to the weight of the air exhausted : whereas in 

 reality it is not so much; for the bottle, when exhausted, must necessarily be compressed, and on 

 that account weigh heavier in water tlian it would otherwise do. Suppose, for example, that air is 

 really 800 times lighter than water, and that the bottle is compressed nl (, s part of its bulk ; which 

 seems no improbable supposition: the weight of the bottle in water will thereby be increased by 

 Ti'^TTB °f tli^ weight of a quantity of water of the same bulk, or more than -^ of the weight of 

 the air exhausted : whence the diflference of weight will be not so much as f| of the weight of the 



