94 ©. Marignac—Chemical Equivalents and Atomic Weights. 
If we refer to the fundamental hypothesis of the atomic theory, 
which supposes that the a of bodies is not indefinite, 
but that they are formed by the agglomeration of excessively 
small but indivisible particles, or atoms, the theoretical defini- 
tion of atomic weights is of the simplest, as they are the relative 
weights of these ultimate particles. But, however simple the 
definition may be, the determination of the weights is sur- 
rounded with great difficulties, 
thesis of the existence of atoms accounts in such a 
simple manner for that < hs ag equivalent proportions for 
elements which play t me part, that we are natura 
at first sight, to mueiter icas proportions as representing their 
relative _— weights, although this consequence is not rigor- 
ously necessary. It is evident, however, that as neither this 
consideration of chemical equivalence, nor any other considera- 
tio rawn from chemistry alone, has led to a complete and 
arti: cpitanh of chemical equiva alents, we cannot by such con- 
siderations be gui in the choice of all the atomic weights, 
and as these. on account of the hypothesis that is made on their 
nature, cannot be arbitrary, like equivalents, it has become 
necessary to study the physical properties of the elements and 
of compound bodies to find motives for this determination of 
the atomic weights. Among the properties which can be ap- 
pealed to, the most important are the densities of gases and 
vapors, . the specific heats and isomorphism. 
I acknowledge that in some very rare cases these three orders 
of physical properties do not lead to the same result, and [agree 
with Mr. Berthelot that between these three data we must make 
a choice. I am, however. in complete disagreement from him 
in the conclusion that I draw Sane this. If he does not say So 
expressly, his whole argument proves that, in his opinion, no 
account is to be taken of these physical properties, when they 
disturb the usage established for weights that have been adopted 
‘or a long time in chemical notations. On the contrary, I think 
that great account should be taken of these physical properties, 
and that when they all agree we must have eno fear of modifying 
a few formulas which have only long usage in their favor, par- 
re if the necessary modification is unimportant. If, 
oreover, the physical properties do not agree, it is necessary 
i. cite the facts with the greatest care, and see if, in some 
cas isagreement can be explained and then choose the 
vaabe which spite ot i digs the general properties of the 
elements and its 
Is it thipoobes to oe this? The best proof that it is not, and 
that there is even no serious difficulty in determining the atomic 
weight which agrees the best with the physical pro properties, vf - 
be found in this circumstance that there is no disagreem 
