AugtLst 1 8, 1887] 



NATURE 



;69 



I 



equations, and the equations formed a complete expression 

 of the reactions, so far as the weights of matter taking part 

 in them were concerned. Now it was experimentally 

 proved that every definite compound possessed a constant 

 quaUtative and quantitative composition, and it seemed 

 to chemists as something of the nature of an axiom that 

 to a given composition one and only one compound could 

 correspond. So convinced were they of the truth of this 

 unproved and, as the event showed, totally erroneous 

 proposition that, when in 1823 and 1824 the first cases of 

 isomerism, or identity of composition together with 

 diversity of properties, were discovered by Wohler and 

 Liebig, the results obtained by these eminent chemists 

 were generally set down to faulty analysis. But the cases 

 of isomerism multiplied rapidly, and chemists had to 

 make their account with this altered state of things. But 

 here the inadequacy of the empirical formulae became 

 evident Wherever a case of isomerism occurred, the 

 empirical formula was ambiguous, and the equations in 

 which it was employed were neither a complete nor a pre- 

 cise expression of the reactions. 



To some, the discovery that constancy of composition 

 no longer involved constancy of properties may have 

 seemed to sap the very foundations of chemistry as then 

 understood. But this was not the case. The discovery 

 necessitated an extension of the atomic theory, not its 

 abolition. In fact, isomerism afforded a remarkable 

 proof of the correctness of the view that matter con- 

 sisted of atoms. On the alternative hypothesis that 

 mitter fills space continuously and homogeneously, 

 isomerism is incapable of explanation ; as it is incon- 

 ceivable that the same given quantities of the same given 

 kinds of matter, continuously and homogeneously filling 

 space, should produce more than one compound. A dif- 

 ference of properties in such a case bespeaks a difference 

 of arrangement of the component parts ; and further, as 

 each such compound displays, even in the state of the 

 finest mechanical subdivision, perfe:t uniformity, the 

 component parts, by the arrangement of which the dif- 

 ference of properties is produced, mast be exceedingly 

 small. We are thus led back to the atomic theory, whilst 

 at the same time the extension is indicated which it was 

 necessary to mike in this theory in order that isomerism 

 might find its proper place and explanation under it. It 

 was necessary to determine, so far as possible, the mode 

 of arrangement of the atoms in the various compounds. 

 The results of this attempt are embodied in the constitu- 

 tional formulae which have been employed by chemists at 

 various times. 



The method resorted to in solving this problem was 

 very similar to that which had been employed in deter- 

 mining the ultimate composition of compounds. Just as 

 when, after isolating from a compound, or introducing [ 

 into a compound, some particular kind of elementary 

 matter, chemists concluded that the compound actually 

 contained that particular kind of matter, so, when in a 

 reaction a particular group of atoms was eliminated 

 bodily from a compound, or introduced bodily into a 

 compound, they concluded that this group existed as such 

 in the compound. Unfortunately, the conclusion is not 

 always quite so warrantable in the case of atomic groups 

 as in the case of elements. The reaction, for example, 

 by which anatomic group is eliminated from a compound 

 involves the destruction of the parent compound, and in 

 this process, which is generally more or less violent, it is 

 only too easy for the atomic groups to undergo re-arrange- 

 ment. In this way, alcohol (C2H,tO), from the fact that it 

 may be split up into ethylene (C2H4) and water (H-^O), 

 was at one time regarded as containing these two atomic 

 groups — a view which at all events is not that at present 

 held. We thus see that from two different reactions, two 

 totally distinct and mutually incompatible constitutional 

 formulae may be deduced for the same compound. 



It would carry us too far to trace all the steps by which 



constitutional formulae gradually became more precise 

 and less self-contradictory, but a few important disco- 

 veries may be mentioned which have mainly tended to 

 bring about this result. In the first place, the develop- 

 ment of the idea of the molecule as distinct from that of 

 the atom, and the discovery of a means of determining 

 the molecular weight of bodies, led to the division of 

 isomerides into two classes : those in which the propor- 

 tions of the various atoms were the same, but the total 

 number of atoms in the molecule was different — this 

 mode of isomerism being distinguished as polymerism ; 

 and isomerism proper, in which both the proportions of 

 the various atoms, and the total number in the mole- 

 cule, are the same in the different compounds. But 

 the knowledge of the molecular weight aided chiefly in 

 the construction of constitutional formulae by determin- 

 ing the exact number of atoms in the molecule, and thus 

 facilitating the task of arranging these atoms by stating 

 precisely how many atoms had to be arranged. The law 

 of valency also exercised a most important influence, 

 simplifying matters by greatly limiting the number of 

 legitimate arrangements. In fact, in the case of some of 

 the simpler compounds, such as methane (CHJ, ethane 

 (CH3 . CH3), propane (CH3 . CHg . CHj), methyl alcohol 

 (CH3.OH), and others, only one mode of arrangement 

 is, according to the laws of valency, possible for each 

 compound. 



A modern constitutional formula, therefore, takes the 

 various atoms of a compound in the proportions indicated 

 by the empirical formula, and in the absolute number 

 prescribed by the molecular weight, and arranges them 

 in that way which, within the limits of the laws of valency, 

 will best account for the reactions of the compound. 



Let U5 consider what elements of uncertainty are in- 

 volved in each of the various operations here enumerated. 



The correctness of the empirical formula of a compound, 

 as calculated from its percentage composition, depends 

 upon the correctness of the atomic weights assigned to 

 its component elements. In the case of organic com- 

 pounds, to the consideration of which we shall confine 

 ourselves here, the atomic weights of the component 

 elements may be regarded as determined with a degree 

 of probability approaching to absolute certainty. (This 

 does not, of course, refer to the question whether the 

 atomic weight of an element has been determined within 

 1/10,000 more or less of its true value, but whether, for 

 example, in the case of carbon the atomic weight is 12, 

 or some multiple or sub-multiple of 12.) The empirical 

 formulae of correctly analyzed organic compounds may 

 therefore be regarded as standing on as sure a foundation 

 as almost anything in the range of science which is a 

 matter of deduction and not of direct observation or 

 experiment. As regards the second point, the molecular 

 weight, an almost equal certainty may be said to prevail 

 in most cases in which the compound can exist in the 

 state of vapour. Avogadro's law, that " when two gases 

 or vapours are at the same temperature and under the 

 same pressure, the number of molecules in unit of volume 

 is the same in both gases or vapours"— this law, originally 

 advanced as an hypothesis, has been shown to follow as a 

 mathematical deduction from the kinetic theory of gases, 

 a theory almost as well established at the present day as 

 the atomic theory itself. Avogadro's law places in our 

 hands a means of determining the molecular weight of 

 substances which are capable of existing in the form of 

 vapour, the only uncertainty attaching to its determina- 

 tions being that occasioned by the tendency which many 

 compounds exhibit, either to undergo decomposition, or 

 to be incompletely vaporized, in passing into the gaseous 

 state. But in the case of all compounds capable of 

 existing undecomposed in the gaseous state throughout 

 any considerable range of temperature, the molecular 

 weight may be determined with a very high degree of 

 probability. In cases where the compound is not volatile 



