no compound is formed with evolution of oxygen and containing oxygen, hydrogen, and chlorine (or bromine); this formula-HCl-is recognised on other grounds as the molecular formula of hydrochloric acid. Hence, it is argued, the hydrogen in the molecule of water is divisible in chemical changes into two parts, but the oxygen is not divisible, and hence, the simplest molecular formula for water is H2O; but if this is so, the atomic weight of oxygen cannot be less than 16. Assuming the atomic weights of iron and oxygen to be (in round numbers) 56 and 16 respectively, the formula Fe,O, is deduced, from analyses, for ferric oxide as representing the smallest quantity of this compound which neutralises acids, forms double salts, can be acted on by chlorine to form Fe,Cl, &c.; hence this formula represents the minimum molecular weight of ferric oxide. But similar reasoning leads to As,O, as the minimum molecular formula of arsenious oxide; now we know that the gaseous oxide has a molecular weight expressed by the formula AsO. Hence the method of analogies does not always lead to the adoption of the true molecular weight of a compound. Sometimes the method of analogies becomes very indirect. Thus, the molecular formula of ferric chloride is Fe,Cl, that of ferrous chloride is either FeCl, or Fe,Cl. Ferric chloride is produced by the action of chlorine on ferrous chloride; now the general action of chlorine is either to add itself on -to other molecules, or to decompose molecules and then substitute itself for some one or more of the atoms formerly constituting these molecules. If ferrous chloride is FeCl,, the action of chlorine on this molecule is represented by the equation 2 FeCl + Cl, Fe,Cl, but this reaction is abnormal. If ferrous chloride is Fe,Cl, the action of chlorine is represented by the equation FeCl + Cl2 = FeCl ̧, and this reaction is analogous to other actions of this element; hence the molecular formula of ferrous chloride is probably not smaller than Fe,Cl ̧1. = 1 V. Meyer has recently obtained results regarding the vapour density of ferrous chloride which seem to him to point to the conclusion that, like stannous chloride, = The chemical method of determining minimum molecular weights, as applied to acids and bases, generally resolves itself into determining the basicity of the acid, or the acidity of the base. Thus, the results of analyses of sulphuric acid are satisfied by the formula H2SO; the fact that this acid is dibasic leads with a fair degree of certainty to the conclusion that I, and that the molecular formula of the compound is therefore H,SO,. The simplest formula which can be given to citric acid consistently with analytical results, and with the atomic weights C = 12, O= 16, H = 1, is CHO,; that the molecular formula is probably not greater than this is shewn by the tribasic character of the acid. Reasons have been already given for adopting NH, as the molecular formula of ammonia: analysis shews that the alkaloid quinine cannot have a smaller molecular weight than that represented by the formula C,H,NO (C = 12, H = 1, N = 14, O= 16), but the quantity of this alkaloid which neutralises that amount of hydrochloric acid which is neutralised by NH,, is 2C,H,,NO, therefore the molecular formula of quinine is probably not less than C,,H,N ̧O 20 12 2 This method may be also applied to determine the formulæ of salts. Thus if sulphuric acid has the molecular formula H,SO,, the molecule of sodium sulphate is probably represented by the formula Na,SO,, because the atom of sodium being very probably monovalent', the amount of sodium 'equivalent' to H, is represented by Na,. So, although boric acid is non-volatile, its ethyl salt has been vaporised and found to have the formula (CH),BO,, hence, knowing that boric acid is tribasic, we deduce for it the probable molecular formula H,BO,. The so-called 'law of even numbers' enunciated by Gerhardt led to the revision of many molecular formulæ: Gerhardt stated that the sum of certain elementary atoms (hydrogen, chlorine and its analogues, nitrogen and its anathis compound possesses two molecular weights expressed respectively by the formulæ FeCl and Fe,Cl; Ber. 14. 1455. 1 That is, capable of combining directly with not more than one atom of hydrogen, chlorine, bromine, iodine, or fluorine to form a compound molecule. See chap. II., pars. 56, 57. logues) contained in any molecule is always an even number1. Thus analysis leads to the formula C,H,O, for tartaric acid, and as the acid is dibasic this formula is apparently molecular; but the hydrogen atoms must be expressed by an even number according to Gerhardt's law, therefore the formula was doubled. Similar reasoning applied to the formulæ of nitric oxide and indium chloride would require that these should be written N,O, and In,Cl, respectively, but we know that the molecular formula of these compounds are NO and InCl,, hence Gerhardt's 'law' must be applied with care2. 37. The chemical methods for determining molecular and atomic weights differ in two main particulars from the physical methods which have been already discussed. The chemical methods as a class do not attempt to distinguish between solids, liquids and gases—so far as the application of these methods is concerned the molecular weight of a solid, liquid or gaseous substance is the smallest quantity which takes part in a chemical reaction-the physical method for finding molecular weights is only applicable in any strictness to gases. The chemical methods also generally begin by determining, if possible, the atomic weights of the elements composing a given compound, and then argue as to the molecular weight of the compound; the physical method, on the other hand, begins by defining molecule, and then, applying this definition to chemical reactions, arrives at a definition of atom, both definitions being so stated as to indicate the data which are required before the relative weights of either atoms or molecules can be determined. 38. In the following table I have sought to summarise a considerable amount of facts concerning the atomic weights of the elements: it is well that the student should have placed before him a synopsis of the evidence on which these allimportant numbers are based. 1 See Laurent, Chemical Method, p. 46 et seq. * For further examples of the application of chemical methods to determinations of molecular and atomic weights see Watts's Dict. vol. 1. pp. 457-8 and 460-1; also Williamson 'On the Atomic Theory,' C. S. Journal, 22. 328. directly PHOSPHORUS PH, PC, PI, PFs, SULPHUR CHLORINE POTASSIUM POCI, PSCI, Pala PH, PN,CI, &c. Isomorphism: compounds compared [See note A, p. 84.] Li compounds with analogous CN compounds with those of F, NH, compounds with those of alkali metals metallic fluorides with analogous compounds of Cl, Br and I Na compounds with those of other alkali metals Mg compounds generally with those of Zn, Mn, and Fe (in ferrous salts) with Cr, Mn and Fe in R2O, and derivatives directly: sp. ht. varies much with C, Zr, Sn and Ti in comwith temperature SH, SO2, SO3, SOCI, directly CIH, CI(CH), CITI, none none SCANDIUM none TITANIUM TiCl, VANADIUM VCI, VOCI, indirectly: doubtful directly directly sp. heats of some compounds de- sp. heats of a few compounds pounds of type RO, phosphates with vanadates and arsenates, organic compounds of P with those of N, As and Sb with Se compounds, with Te compounds of type R"Te. Salts of H2SO, with those of H.SeO. and H,TeO Chlorides, with analogous compounds of Br and I K compounds with those of other Ca compounds with those of Sr, sp. heats of one or two com- Vanadates with phosphates and arsenates |