may be used like hydrogen, for the purpose of determining the valency of other elements. This is all the more satisfactory, as their compounds are much more numerous than those of hydrogen. A comparison of the fluorides, chlorides, bromides and iodides of the first four families shows that their compounds correspond to the four types which have just been mentioned, e.g.:

The chemical valency is constant for each family, and with an increasing atomic weight, rises from one family to the next by one unit.

The following families behave in a similar way, but at first sight the relationship appears somewhat more complicated, the composition of the typical hydrides indicating that the valency decreases, thus:

Certain chlorine compounds of these elements have an analogous composition:

But these and similar compounds do not contain the maximum amount of chlorine or other monovalent atoms, with which the elements are capable of uniting. Compounds of the following composition are known:

Some of these compounds are very unstable: ammonium chloride, phosphorus pentachloride, and iodine trichloride decompose on volatilisation, e.g.: NHCl=NH2+HC1; PCI, =PCI,+Cl2; ICl2=ICl+Cl2; others can be converted into vapour without decomposition, e.g.: NbCl, TaCl, TeCl, MOCI,, WC. This difference in the behaviour of the compounds is explained by the fact that the affinity of the negative elements for chlorine and its analogues is feeble, and consequently the chlorine atoms are easily separated from the compound. The behaviour of sulphur is very remarkable: the tetrachloride SCI, can only exist at -20° C.; even at 0° C. it begins to decompose into SCI,, and on distillation half the residual chlorine is lost. Sulphur does not form any definite compounds with bromine and iodine. Phosphorus combines with five atoms of fluorine to form a stable compound. PC1, and PBr, easily part with two atoms of chlorine or bromine respectively but phosphorus can only unite with three atoms of iodine; as a rule it only combines with two, forming PI,.

We may assume that the difference in the behaviour of the elements belonging to families V. and VI. towards the nonmetals is not due to a difference in their valency, but is probably caused by a difference in the force with which they attract the monovalent elements.

This assumption is confirmed by an examination of the oxygen compounds. As an atom of oxygen is equivalent to two atoms of hydrogen and is divalent, any other atom which has the power of uniting with one atom of oxygen is also divalent. One trivalent atom could combine with oxygen in the proportion of one to one and a half, or, more correctly, two trivalent atoms can unite with three atoms of oxygen. The radical

hydroxyl-OH is formed by the union of one atom of hydrogen with one atom of oxygen; as the oxygen still has the power of uniting with a second atom, the radical is monovalent. The valency of an element is represented by the number of hydroxyls with which an atom combines, or by twice the number of oxygen atoms which in its oxide would be united with one atom. In order to permit of a uniform comparison, the formulæ of the oxides in the following table are given as though they contained two atoms of the other element, even where the molecule may only contain one atom.

This table shows the remarkable relation which exists between

valency and atomic weight. The number of each family indicates its valency. In families I. to VII. only oxygen and fluorine are missing; no oxide of the latter element has as yet been prepared. In VIII. iron is omitted; the highest oxide, the anhydride of ferric acid, is unstable and has not been satisfactorily investigated. The same is true of the highest oxides of Co, Ni, Rh, Pd, Ir, Pt.

A few examples of the hydroxides will suffice:

Here again the valency is indicated by the number of the family, showing the close relation between valency and atomic weight.

It is to be noted that families V., VI., VII. do not exhibit the same valency towards hydrogen and the metals that they do towards oxygen. Beginning with family IV., the valency increases one unit for oxygen and other negative elements, but decreases one unit for hydrogen and the positive elements, as is shown in the preceding tables.

§ 41. Possible Errors in the Determination of Chemical Valency. There are many other compounds, besides those mentioned in the preceding section, which can be used for determining the chemical valency of an element. Most of these yield a smaller value, but some give a larger value for the valency.

The valency is always too low when the element in question is not combined with the maximum number of atoms; that is to say, when some of its affinities remain unsaturated. The tetravalent carbon atom can unite with four monad atoms, such as hydrogen or chlorine, or two divalent oxygen atoms :

When a limited supply of oxygen is passed over an excess of red-hot charcoal, then the carbon atoms cannot take up the maximum amount of oxygen, but form carbon monoxide. The molecule of this compound is represented by the formula

*-C 0.

It would be incorrect to conclude from the composition of this compound that carbon is divalent; for two unsaturated affinities are also present, as indicated by the asterisks. This is shown by the fact that the compound unites with two atoms of chlorine, forming phosgene gas, or carbonyl dichloride.

The valency of nitrogen cannot be deduced from the composition of nitric oxide (NO), nitrogen peroxide (NO„), nitrosylchloride (NOCl), and similar compounds: they are known as unsaturated compounds, or as compounds containing unsaturated affinities. Those metals which only contain one atom in the molecule (§ 27), e.g. mercury, cadmium, and zinc, belong to this class.

There is another class of compounds which cannot be regarded as unsaturated, although the number of monovalent atoms does not correspond to the valency of the polyvalent atoms, viz. all those compounds containing more than one polyvalent atom in the molecule. In such compounds a certain number of valencies are used up in the union of the polyvalent atoms with one another, how many are in this way employed it would be difficult to decide.

It would be very difficult to ascertain the correct valency of carbon if we were only acquainted with the following compounds: ethane CH, ethylene C,H,, and acetylene C2H2.

A correct determination of the valency may be attempted by taking into account the affinities required for the purpose of linking together the polyvalent atoms: but this process may lead to erroneous conclusions. If we apply it to ethane

we obtain the correct result: 6 affinities for H and 2 for C; total 8. But in the case of ethylene we only obtain 2 for C and 4 for H, total 6; and for acetylene 2 for H and 2 for C, total 4. According to these last results carbon might erroneously be considered as only di- or tri-valent.

But the composition of marsh gas shows that carbon is tetravalent, and the fact that the carbon is combined with less hydrogen in ethylene and acetylene than it is in ethane, is accounted for by the hypothesis that the carbon atoms are attached to each other by more than one affinity, thus:

It follows from these considerations that the chemical valency can only be deduced with certainty from the composition of those compounds which only contain one atom of