remarkable fact that neither Dalton nor Gay-Lussac accepted the views of the Swedish chemist. The author of the atomic theory obstinately maintained his first idea of deducing atomic weights solely from the ponderable relations of elements in combinations. GayLussac, again, confined himself to the immediate consequences of his discovery, not without forcing them to some extent, in certain cases, by hypotheses upon the forms of condensation of the combining gaseous elements. He and Berzelius expressed the composition of bodies in volumes, the latter admitting that the relative weights of these volumes represented atoms, Gay-Lussac refusing to consider these weights as anything more than ponderable relations,' and inclining rather to the views of Davy. The latter, deviating to an equal extent from the profound conceptions of Dalton, and with the idea of completing them by the discoveries of the French chemist, confined himself strictly to established facts and to the consideration of 'proportional numbers.' After the ingenious but ignored attempts of Avogadro and Ampère, and the unfruitful effort of Berzelius, Dalton's conception would have been sentenced to sterility and oblivion, had it not happened that, at the period of which we are speaking, fresh discoveries and new ideas drew attention to it. We allude to Prout's hypothesis, to the discovery of the law of specific heats, and to the discovery of isomorphism.

CHAPTER III.

PROUT'S HYPOTHESIS-LAW OF SPECIFIC HEATS

ISOMORPHISM.

DULONG AND PETIT-MITSCHERLICH.

I.

WE must first return to Prout's hypothesis, not that it is of such great importance from our present point of view, but because it preceded the important discoveries which we shall presently mention.

1

The anonymous author of a memoir which appeared in 1815 upon the relations between the densities of bodies in a gaseous state and the weights of atoms, tried to prove that the densities of oxygen, nitrogen, and chlorine are integral multiples of that of hydrogen, and that the atomic weights of certain elements are similarly integral multiples of that of hydrogen. Amongst these elements we meet with some metals, the atomic weight of which had been determined by the author or by other chemists by the following excellent process: the quantities of metal were determined which, combined with oxygen, formed quantities of oxides capable of neutra

Annals of Philosophy, vol. vii. p. 111.

lising the same quantity of an acid. The results appear in the following table :—

We will make no remark upon the author's considerations concerning the relations which may be traced between these numbers and those which express the atomic weights of other elements determined with less accuracy. These considerations were obscure and erroneous.

The important point was raised by Prout in 1816, in a work to which he appended his name. It is very advisable,' he remarked, 'to adopt the same unit for specific weights and atomic weights, and to take as this unit the weight of one volume of hydrogen. The same numbers will thus give the densities of gases and the atomic weights, or a multiple of these weights. If,' proceeds the author, these numbers are whole numbers, the fact under consideration may be interpreted by admitting that hydrogen is the primordial matter which forms the other elements by successive condensations. The figures expressing these condensations-that is to say, the densities—would at the same time give the number of volumes of primordial matter condensed into a single volume of a given element, and the weight of this volume, expressed by a whole number, would represent the atomic weight of the element.'

The determinations of atomic weights, and even those of the densities of gases, were too inaccurate, at the time of which we are speaking, for Prout's hypotheIsis to be taken into serious consideration. It was a conjecture. It has, as we know, been lately again taken up with great energy by Dumas.1

But the accurate determinations of a number of atomic weights by Stas, notably those of chlorine, potassium, sodium, and silver, by confirming or slightly rectifying the results formerly obtained by Marignac, have entirely annihilated the celebrated hypothesis in question. Unsuccessful attempts have been made to revive it, by taking as the unit, not the atomic weight of hydrogen, but the half or the quarter of this weight. There are well-known atomic weights, particularly that of potassium, which are not a multiple of the fraction, nor even of 4. If, however, we retained the idea, which is, moreover, striking and profound, of a primordial matter the sub-atoms of which were grouped in different numbers to form the chemical atoms of hydrogen and the various simple bodies, and attributed to these sub-atoms

1 Dumas made a communication to the Académie des Sciences (January 14, 1878) relative to the atomic weight of silver, discussing the error which had arisen in the determination of this atomic weight, from the property which metallic silver possesses of retaining about 10 of cxygen, if the latter has not been carefully expelled by heating it in vacuo to 600° C. Dumas maintains the number 108, which he had previously adopted. He remarks that other atomic weights, described as forming exceptions to Prout's hypothesis, might probably be included in the rule, if, in the process of weighing, account were taken of errors similar to those which he had pointed out in the case of silver.

a weight inferior to a quarter of that of hydrogen, assigning to it, for example, 1 of that weight, such an hypothesis would, I say, escape all experimental verification, as the differences which it would then be our object to establish between the atomic weights of the various simple bodies would fall within the limit of errors of observation. Such an hypothesis, though reasonable, ceases to be legitimate, and positive chemistry for the present must abandon this theory of Prout's, this dream of the ancients upon the unity of matter and the compound nature of chemical elements. Does this mean that Prout wrote in vain? By no means. His idea gave rise to valuable work and important discussions, and the example which he set of referring atomic weights to that of hydrogen has been followed, for all chemists have adopted this unit. But at that time, as in our own, Prout's conception produced no argument in favour of the atomic theory, and has exercised no influence upon the development of that theory.

II.

The discovery which we are about to mention is, on the contrary, one of the foundations of the atomic theory. It draws our attention to the relation which exists between the atomic weights and the specific heats of solid elements. In a memoir published in the Annales de Chimie' in 1819, Dulong and Petit gave the specific heats of a great number of solid bodies, particularly metals, and remarked that these specific heats were generally inversely proportional to the atomic weights.