LECT. in.] HISTORY OF CHEMISTRY 45
stant proportions ; which, as a matter of fact, he had never
wholly denied. In 1809 he still regards compounds with
variable proportions as possible,40 but in this opinion he stands
isolated. Too much support has now come to the opposing side.
Richter's investigations, carried out from 1791 to 1800, had at
length become known ; Dalton had formulated his atomic
theory, which was irreconcilable with Berthollet's view and
was already beginning to constitute the basis of chemical con-
siderations ; Gay-Lussac's classical work on the proportions by
volume in which gases combine was concluded; and Berzelius
had published his first important papers. Thus the contro-
versy ends, apparently with the complete defeat of Berthollet.
I have treated this subject at length, because I consider it
very important. We have here to do with a general doctrine
which constitutes one of the foundations of our theoretical
considerations, and settles the distinction between mixtures
and compounds. It is for the latter alone that our chemical
laws hold, since mixtures are not subject to them. In any
particular case, therefore, it is very important to know which
class of substances is being dealt with. What, then, are
our means of forming an opinion ?
It is to be found stated in text-books, that compounds
possess a homogeneous character, whereas mixtures can very
frequently be mechanically separated into t heir ingredient s. It
is further stated there that in compounds the properties of the
constituents have disappeared, whilst they are present side by
side in mixtures. Finally the constancy of proportions is then
adduced as characteristic, and I wish to direct attention to
this whole matter. There are cases in which mixtures are,
in their whole behaviour, no longer distinguishable from com-
pounds. We then have recourse to analysis to solve the ques-
tion. We prepare the substance in various ways, and observe
whether it always possesses the same composition. We thus
invert the doctrine discussed by Berthollet and Proust. The
former regarded compounds with variable proportions as pos-
40 M^m. d'Arcueil 2 470.