Proportions of the Constituents of Water, &c. 208 
when the water was poured out it effervesced like mineral wa- 
ter. 
Thus it would appear that the mode adopted by Mr. Perkins 
in these experiments, was rather inadequate to the purpose he 
had in view. I would therefore, in conclusion, suggest, that in 
any future trials he may be disposed to make, he should previ- 
ously fill the bottles to the bottom of the corks with water, that 
the whole of the air may be removed: and, to render the result 
still more conclusive, the bottles ought to be inclosed in a cage 
of iron or copper. After these precautions, should he find that 
the bottles were broken when brought up, he might then justly 
conclude, that the re-expansion of the compressed water had 
been the cause, since the counter-resistance of the water within, 
must have presented the effect of pressure from without so con- 
siderably as to prevent the breaking of the bottles from that di- 
rection: the only way, therefore, in which they could be broken, 
would be by the external column of water compressing the con- 
fined portion, and foreing an additional quantity into the inte- 
_vior: and by this again resuming its former volume when the 
acting pressure is diminished. I remain, dear sir, 
Yours respectfully, 
Joun DEUCHAR. 
DS ee 
XLVIII.— New Determination of the Proportions of the Con- 
stituents of Water; and the Density of certain Elastic 
Fluids, By MM. Berzexius and Dutone*. 
As modern chemists in their analyses claim a degree ef accuracy 
embracing the thousandth part of the elements employed, it is 
evident that the fundamental data from which they make their 
deductions should be free from that degree of error which they 
profess themselves able to avoid in their experiments. Of these 
data, that of the constitution of water is one of the most impor- 
tant, and the most frequently employed. ‘The proportions ge- 
nerally adopted of Jate years have appeared to be beyond suspi- 
cion of error, both by the means employed to obtain them, and 
the ability of the observers who have conducted them. But we 
have had some reason to believe that this number was liable to 
be affected by some slight error ; and as the subject was of con- 
siderable importance, we resolved to conduct, in concert, the ex- 
periments necessary to ascertain this point. M. Berthollet, 
whose liberality has so often been useful to science, gave us 
* From the Annales de Chimic et de Physique. 
Cc every 
