Weber's observations on the great variation of the specific heat of charcoal, the diamond and boron: 0° 100° 200° 600° 900° Wood charcoal 0·15 0·23 0·29 0·44 0·46 Diamond 0·10 0·19 0·22 0·44 0·45 Boron 0·22 0·29 0·35 — — These determinations, which have been verified by Dewar, Le Chatelier (Chapter VIII., Note ⁴⁶⁸), Moissan, and Gauthier, the latter finding for boron AQ = 6 at 400°, are of especial importance as confirming the universality of Dulong and Petit's law, because the elements mentioned above form exceptions to the general rule when the mean specific heat is taken for temperatures between 0° and 100°. Thus in the case of the diamond the product of A × Q at 0° = 1·2, and for boron = 2·4. But if we take the specific heat towards which there is evidently a tendency with a rise of temperature, we obtain a product approaching to 6 as with other elements. Thus with the diamond and charcoal, it is evident that the specific heat tends towards 0·47, which multiplied by 12 gives 5·6, the same as for magnesium and aluminium.

I may here direct the reader's attention to the fact that for solid elements having a small atomic weight, the specific heat varies considerably if we take the average figures for temperatures 0° to 100°: Li = 7 Be = 9 B = 11 C = 12 Q = 0·94 0·42 0·24 0·20 AQ = 6·6 3·8 2·6 2·4 It is therefore clear that the specific heat of beryllium determined at a low temperature cannot serve for establishing its atomicity. On the other hand, the low atomic heat of charcoal, graphite, and the diamond, boron, &c., may perhaps depend on the complexity of the molecules of these elements. The necessity for acknowledging a great complexity of the molecules of carbon was explained in Chapter VIII. In the case of sulphur the molecule contains at least S₆ and its atomic heat = 32 × 0·163 = 5·22, which is distinctly below the normal. If a large number of atoms of carbon are contained in the molecule of charcoal, this would to a certain extent account for its comparatively small atomic heat.

With respect to the specific heat of compounds, it will not be out of place to mention here the conclusion arrived at by Kopp, that the molecular heat (that is, the product of MQ) may be looked on as the sum of the atomic heats of its component elements; but as this rule is not a general one, and can only be applied to give an approximate estimate of the specific heats of substances, I do not think it necessary to go into the details of the conclusions described in Liebig's ‘Annalen Supplement-Band,’ 1864, which includes a number of determinations made by Kopp.