On definite Proportions. 274 
Tt is remarkable that all the intervals from nitrogen te 
water proceed in twelves. According to this computation, 
water consists of 12°413 parts of hydrogen and 87°587 of 
oxygen, which differs but little from one of my experi- 
ments in which the proportions. were 12°23 and 87°77. 
These determinations cannot be precisely accurate, since: 
they depend on experiments. which are not absolutely cor- 
rect. It appears however that the error cannot be very ma- 
terially great, from the coincidence between the composition 
of nitrogen as deduced from calculation, and from. the 
analysis of the nitrous oxide. .1f we might venture'to.as- 
sume, that the weights of the gases in the experiments of 
the celebrated French chemists were perfectly accurate, we 
might easily correct the proportions of the ammonia ac- 
cording to those experiments: but this cannot yet be done 
with any certainty. Mr. Gay-Lussac’s mode of. weighing 
the gases, and determining their composition. from. the 
volumes concerned, is probably the most accurate; and 
when we have obtained some perfectly correct analyses in 
this manner, the rest may be completed by calculation. 
It is therefore to be hoped, that those chemists who have 
begun theseexperiments will shortly repeat them, and carry 
them to the highest degree of perfection of which they are 
susceptible. 
We have still a very comprehensive question left to be 
answered: Why does hydrogen, when combined with oxy- 
gen, always afford water, and nitrogen always nitric acid or 
nitrous oxide? Or, conversely: Why do we obtain, by the 
subtraction of oxygen, always hydrogen from water, and 
nitrogen from nitric acid or nitre, if both these substance$ 
are truly oxides of the same hase? 
In this question there is apparently involved a weighty 
objection against the existence of oxygen in hydrogen. If 
analogy should here mislead me in the outline of my argu- 
ments, and I should endeavour to explain a fact which does 
not exist in nature, I may in this case at least be excused 
for having erred. I shall therefore venture to enter upon 
the subject. 
I have observed that ammonia, considered as an oxidated 
substance, was not likely to contain a compound bases 
which, if it existed, must consist of hydrogen and the base 
of nitrogen. For when ammonia is decomposed by po- 
tassium, it condenses a part of the hydrogen with the ni- 
trogen. If the base of ammonia were a compound, it ought 
to set all the hydrogen at liberty, or to condense the whole 
of it with the base of nitrogen, It cannot therefore be 
believed, 
