able. But I could not stop to ascertain this matter strictly by
analysis.

696. I believe, however, that there is such an oxide; that it is often
present in variable proportions in what is commonly called protoxide,
throwing uncertainty upon the results of its analysis, and causing the
electrolytic decomposition above described[A].

[A] In relation to this and the three preceding paragraphs, and also
801, see Berzelius's correction of the nature of the supposed now
sulphuret and oxide, Phil. Mag. 1836, vol. viii. 476: and for the
probable explanation of the effects obtained with the protoxide, refer
to 1340, 1341.--_Dec. 1838._

697. Upon the whole, it appears probable that all those binary compounds of
elementary bodies which are capable of being electrolyzed when fluid, but
not whilst solid, according to the law of liquido-conduction (394.),
consist of single proportionals of their elementary principles; and it may
be because of their departure from this simplicity of composition, that
boracic acid, ammonia, perchlorides, periodides, and many other direct
compounds of elements, are indecomposable.

698. With regard to salts and combinations of compound bodies, the same
simple relation does not appear to hold good. I could not decide this by
bisulphates of the alkalies, for as long as the second proportion of acid
remained, water was retained with it. The fused salts conducted, and were
decomposed; but hydrogen always appeared at the negative electrode.

699. A biphosphate of soda was prepared by heating, and ultimately fusing,
the ammonia-phosphate of soda. In this case the fused bisalt conducted, and
was decomposed; but a little gas appeared at the negative electrode; and
though I believe the salt itself was electrolyzed, I am not quite satisfied
that water was entirely absent.

700. Then a biborate of soda was prepared; and this, I think, is an
unobjectionable case. The salt, when fused, conducted, and was decomposed,
and gas appeared at both electrodes: even when the boracic acid was
increased to three proportionals, the same effect took place.

701. Hence this class of compound combinations does not seem to be subject
to the same simple law as the former class of binary combinations. Whether
we may find reason to consider them as mere solutions of the compound of
single proportionals in the excess of acid, is a matter which, with some
apparent exceptions occurring amongst the sulphurets, must be le