line chlorides, bromides, or iodides are exposed in the hottest part of the same Bunsen burner, we shall find that under these conditions, when the amount of heat furnished in a given time is sensibly the same, these several salts take a very different length of time to volatilise, which is almost inversely proportional to their molecular weights, so that by multiplying the times of volatilisation by the molecular weights we obtain products which, though not identical, sensibly approximate to a mean value (4977). The following table gives, according to Bunsen, the times of volatilisation in seconds for one centigramme of various haloid salts. We have added the corresponding molecular weights, and the products of these weights by the times of volatilisation:

We see that the products vary between the numbers 4659 and 5337-that is to say, in the proportion of 1 to 14-while the times of volatilisation vary from 29.8 to 114, in the proportion of 1 to 3.8. This result, though only approximate, is significant, and we must remember that the difficulties attending the experiment scarcely

admit of greater accuracy. It shows that in an equally hot flame, and in a given time, the same number of molecules of the haloid salts are volatilised, or, in other words, that the molecular volatilities are the same.

VII.

The New System of Atomic Weights is in Harmony with the Law of Isomorphism.

The demonstration of this point will be neither long nor difficult. We have already observed (p. 60) the assistance which Berzelius derived from isomorphism in the determination of certain atomic weights, such as those of aluminium and iron. The principle which guided him in these considerations was correct. atomic weights of simple bodies must be so determined that analogous and isomorphous compounds shall receive similar formulæ. This principle is respected by the new system of atomic weights.

The

We need not here enumerate all the cases of isomorphism presented by the combinations of analogous elements, and will therefore only mention the following--the oxides,

the sulpharsenite and sulphantimonite of silver,

ASS,Ag
SbS.Ag,;

and the isomorphorous phosphates, arsenates, and vana dates of the apatite group

(PO4), Ca,Cl apatite,

(PO),PbCl pyromorphite,
(ASO),PbCl mimetesite,

(VO),PbCl vanadinite.

We shall return to the latter compounds. Here we need only remark that since the researches of Roscoe the atomic weight of vanadium has been altered, so that vanadinite, which is isomorphous with apatite, is represented by a similar formula; secondly, that calcium and lead, bivalent metals in the new system of atomic weights, cannot be replaced in the isomorphous compounds under discussion by univalent metals, such as potassium or sodium.

The latter point is important and requires explanation. The law of isomorphism teaches us that the alkaline metals, amongst which we have included silver, because it also is univalent, form a separate group, distinctly separated from the several groups of bivalent metals, such as magnesium, calcium, barium, strontium, lead, &c. Thus the sulphate and selenate of silver,

SO,Ag2
SeO Ag2

are isomorphous with the anhydrous sulphate and selenate of sodium,

SO,Na
SeO.Na..

On the other hand, the alkaline sulphates, selenates, permanganates, and perchlorates are isomorphous with each other, but not with the corresponding salts of the magnesium series. Thus we have

SO,KH isomorphous with SeO,KH (Mitscherlich),
SO,Na2 + 10H2O isomorphous with

CrO,Na2 + 10H2O,
SeO,Na,+ 10,0.

The same remark applies to potassium perchlorate and permanganate, which, by the new system of atomic weights, are represented by similar formula

CIO,K potassium perchlorate,

MnO4K potassium permanganate.

In the equivalent notation they were written-
C10,.KO perchlorate of potash,

Mn,O,.KO permanganate of potash.

The isomorphism of the chromates and manganates with the sulphates and selenates also deserves notice; it led Berzelius to halve the atomic weights which he had formerly attributed to chromium and manga

nese.

The isomorphism of the alkaline chlorides, bromides, iodides, and chloroplatinates is so well known that there is no occasion to lay stress upon it here. It is admitted that silver sulphide and cuprous sulphide' are isomor

The isomorphism of cuprous sulphide and silver sulphide suggested the following reflections to V. Regnault, which are given at p. 346 of vol. ii. of his Cours élémentaire de Chimic, 2nd edition :'Native sulphide of silver is isomorphous with the native sub-sul. phide of copper, Cu,S; these two sulphides seem to have the power

phous, at least in their compounds. These sulphides are represented by the following analogous atomic formulæ :

Ag2S silver sulphide,

Cu,S cuprous sulphide;

whilst in the equivalent notation they received the dissimilar formula

of replacing each other in all proportions, as, for example, in the varieties of fahlerz. We have maintained that this isomorphism exists only between bodies possessing the same chemical formulæ, and we have frequently referred to this law in fixing the equivalents of elementary bodies. But sulphide of silver would form an exception to the law if we give it the formula AgS-that is to say, if we adopt the number 1350 as the equivalent of silver. This consideration has led several chemists to give to sulphide of silver the formula Ag.S, and Ag2O to protoxide of silver, and to take the number 675 as the equivalent of silver. This view has been confirmed by several other circumstances which demand our attention for a few moments. Physicists have shown by a great number of experiments that there exists a very simple relation between the specific heats of bodies and their chemical equivalents. This law states that the specific heats of elementary bodies, within narrow limits, vary inversely as their equivalents. Now, silver will only satisfy this law when the number 675 is received as its equivalent. Moreover, a law similar to that which we have just indicated for elements has been recognised for the specific heats of compounds. This law may be thus stated: The specific heats of compounds possessing the same formula, within narrow limits, vary inversely as the numbers which represent the chemical equivalents of these compounds. Now, the sulphides of silver and copper satisfy this law if we represent sulphide of silver by the formula Ag2S.

But if we write the formula of sulphide of silver AgS, and consequently that of protoxide of silver Ag2O, we must write the formula of soda Na,O, and not NaO, as we have hitherto done, for we have seen that the sulphate of silver is isomorphous with the anhydious sulphate of soda. The salts of potash and of lithia being isomorphous with the corresponding salts of soda, when they contain the same quantities of water of crystallisation, we must give to potash the formula K2O and to lithia Li2O,' &c.