For univalent metals, like those of the alkalis, the weight of the equivalent is equal to the weight of the atom. For bivalent metals the atomic weight is equal to the weight of two equivalents, for n-valent metals it is equal to the weight of n equivalents. Thus aluminium, Al = 27, is trivalent, that is, its equivalent = 9; magnesium, Mg = 24, is bivalent, and its equivalent = 12. Therefore, if potassium or sodium, or in general a univalent metal, M, give compounds M2O, MHO, MCl, MNO3, M2SO4, &c., and in general MX, then for bivalent metals like magnesium or calcium the corresponding compounds will be MgO, Mg(HO)2, MgCl2, Mg(NO3)2, MgSO4, &c., or in general MX2.

By what are we to be guided in ascribing to some metals univalency and to others bi-, ter-, quadri-, ... n-valency? What obliges us to make this difference? Why are not all metals given the same valency—for instance, why is not magnesium considered as univalent? If this be done, taking Mg = 12 (and not 24 as now), not only is a simplicity of expression of the composition of all the compounds of magnesium attained, but we also gain the advantage that their composition will be the same as those of the corresponding compounds of sodium and potassium. These combinations were so expressed formerly—why has this since been changed?

These questions could only be answered after the establishment of the idea of multiples of the atomic weights as the minimum quantities of certain elements combining with others to form compounds—in a word, since the time of the establishment of Avogadro-Gerhardt's law (Chapter [VII].). By taking such an element as arsenic, which has many volatile compounds, it is easy to determine the density of these compounds, and therefore to establish their molecular weights, and hence to find the indubitable atomic weight, exactly as for oxygen, nitrogen, chlorine, carbon, &c. It appears that As = 75, and its compounds correspond, like the compounds of nitrogen, with the forms AsX3, and AsX5; for example, AsH3, AsCl3, AsF5, As2O5, &c. It is evident that we are here dealing with a metal (or rather element) of two valencies, which moreover is never univalent, but tri- or quinqui-valent. This example alone is sufficient for the recognition of the existence of polyvalent atoms among the metals. And as antimony and bismuth are closely analogous to arsenic in all their compounds, (just as potassium is analogous to rubidium and cæsium); so, although very few volatile compounds of bismuth are known, it was necessary to ascribe to them formulæ corresponding with those ascribed to arsenic.

As we shall see in describing them, there are also many analogous metals among the bivalent elements, some of which also give volatile compounds. For example, zinc, which is itself volatile, gives several volatile compounds (for instance, zinc ethyl, ZnC4H10, which boils at 118°, vapour density = 61·3), and in the molecules of all these compounds there is never less than 65 parts of zinc, which is equivalent to H2, because 65 parts of zinc displace 2 parts by weight of hydrogen; so that zinc is just such an example of the bivalent metals as oxygen, whose equivalent = 8 (because H2 is replaced by O = 16), is a representative of the bivalent elements, or as arsenic is of the tri- and quinqui-valent elements. And, as we shall afterwards see, magnesium is in many respects closely analogous to zinc, which fact obliges us to regard magnesium as a bivalent metal.

Such metals as mercury and copper, which are able to give not one but two bases, are of particular importance for distinguishing univalent and bivalent metals. Thus copper gives the suboxide Cu2O and the oxide CuO—that is, the compounds CuX corresponding with the suboxide are analogous (in the quantitative relations, by their composition) to NaX or AgX, and the compounds of the oxide CuX2, to MgX2, ZnX2, and in general to the bivalent metals. It is clear that in such examples we must make a distinction between atomic weights and equivalents.

In this manner the valency, that is, the number of equivalents entering into the atom of the metals may in many cases be established by means of comparatively few volatile metallic compounds, with the aid of a search into their analogies (concerning which see Chapter XV.). The law of specific heats discovered by Dulong and Petit has frequently been applied to the same purpose[3] in the history of chemistry, especially since the development given to this law by the researches of Regnault, and since Cannizzaro (1860) showed the agreement between the deductions of this law and the consequences arising from Avogadro-Gerhardt's law.

Dulong and Petit, having determined the specific heat of a number of solid elementary substances, observed that as the atomic weights of the elements increase, their specific heats decrease, and that the product of the specific heat Q into the atomic weight A is an almost constant quantity. This means that to bring different elements into a known thermal state an equal amount of work is required if atomic quantities of the elements are taken; that is, the amounts of heat expended in heating equal quantities by weight of the elements are far from equal, but are in inverse proportion to the atomic weights. For thermal changes the atom is a unit; all atoms, notwithstanding the difference of weight and nature, are equal. This is the simplest expression of the fact discovered by Dulong and Petit. The specific heat measures that quantity of heat which is required to raise the temperature of one unit of weight of a substance by one degree. If the magnitude of the specific heat of elements be multiplied by the atomic weight, then we obtain the atomic heat—that is, the amount of heat required to raise the temperature of the atomic weight of an element by one degree. It is these products which for the majority of the elements prove to be approximately, if not quite, identical. A complete identity cannot be expected, because the specific heat of one and the same substance varies with the temperature, with its passage from one state into another, and frequently with even a simple mechanical change of density (for instance by hammering), not to speak of allotropic changes, &c. We will cite several figures[4] proving the truth of the conclusions arrived at by Dulong and Petit with respect to solid elementary bodies.

Li Na Mg P
A = 7 23 24 31
Q = 0·9408 0·2934 0·245 0·202
AQ = 6·59 6·75 5·88 6·26
Fe Cu Zn Br
A = 56 63 65 80
Q = 0·112 0·093 0·093 0·0843
AQ = 6·27 5·86 6·04 6·74
Pd Ag Sn I
A =106 108 118 127
Q = 0·0592 0·056 0·055 0·0541
AQ = 6·28 6·05 6·49 6·87
Pt Au Hg Pb
A =196 198 200 206
Q = 0·0325 0·0324 0·0333 0·0315
AQ = 6·37 6·41 6·66 6·49

It is seen from this that the product of the specific heat of the element into the atomic weight is an almost constant quantity, which is nearly 6. Hence it is possible to determine the valency by the specific heats of the metals. Thus, for instance, the specific heats of lithium, sodium, and potassium convince us of the fact that their atomic weights are indeed those which we chose, because by multiplying the specific heats found by experiment by the corresponding atomic weights we obtain the following figures: Li, 6·59, Na, 6·75 and K, 6·47. Of the alkaline earth metals the specific heats have been determined: of magnesium = 0·245 (Regnault and Kopp), of calcium = 0·170 (Bunsen), and of barium = 0·05 (Mendeléeff). If the same composition be ascribed to the compounds of magnesium as to the corresponding compounds of potassium, then the equivalent of magnesium will be equal to 12. On multiplying this atomic weight by the specific heat of magnesium, we obtain a figure 2·94, which is half that which is given by the other solid elements and therefore the atomic weight of magnesium must be taken as equal to 24 and not to 12. Then the atomic heat of magnesium = 24 × 0·245 = 5·9; for calcium, giving its compounds a composition CaX2—for example CaCl2, CaSO4, CaO (Ca = 40)—we obtain an atomic heat = 40 × 0·17 = 6·8, and for barium it is equal to 137 × 0·05 = 6·8; that is, they must be counted as bivalent, or that their atom replaces H2, Na2, or K2. This conclusion may be confirmed by a method of analogy, as we shall afterwards see. The application of the principle of specific heats to the determination of the magnitudes of the atomic weights of those metals, the magnitude of whose atomic weights could not be determined by Avogadro-Gerhardt's law, was made about 1860 by the Italian professor Cannizzaro.

Exactly the same conclusions respecting the bivalence of magnesium and its analogues are obtained by comparing the specific heats of their compounds, especially of the halogen compounds as the most simple, with the specific heats of the corresponding alkali compounds. Thus, for instance, the specific heats of magnesium and calcium chlorides, MgCl2 and CaCl2, are 0·194 and 0·164, and of sodium and potassium chlorides, NaCl and KCl, 0·214 and 0·172, and therefore their molecular heats (or the products QM, where M is the weight of the molecule) are 18·4 and 18·2, 12·5 and 12·8, and hence the atomic heats (or the quotient of QM by the number of atoms) are all nearly 6, as with the elements. Whilst if, instead of the actual atomic weights Mg = 24 and Ca = 40, their equivalents 12 and 20 be taken, then the atomic heats of the chlorides of magnesium and calcium would be about 4·6, whilst those of potassium and sodium chlorides are about 6·3.[5] We must remark, however, that as the specific heat or the amount of heat required to raise the temperature of a unit of weight one degree[6] is a complex quantity—including not only the increase of the energy of a substance with its rise in temperature, but also the external work of expansion[7] and the internal work accomplished in the molecules causing them to decompose according to the rise of temperature[8]—therefore it is impossible to expect in the magnitude of the specific heat the great simplicity of relation to composition which we see, for instance, in the density of gaseous substances. Hence, although the specific heat is one of the important means for determining the atomicity of the elements, still the mainstay for a true judgment of atomicity is only given by Avogadro-Gerhardt's law, i.e. this other method can only be accessory or preliminary, and when possible recourse should be had to the determination of the vapour density.