connected by single bonds to (say) four other atoms, then in a certain unit of time it will collide with each of these atoms in turn. Now suppose two of the attached atoms are replaced by one atom, then this atom must have two valencies directed to the central atom; and consequently, in the same unit of time, the central atom will collide once with each of the two monovalent atoms and twice with the divalent. Applying this notion to benzene, let us consider the impacts made by the carbon atom (1) which we will assume to be doubly linked to the carbon atom (2) and singly linked to (6), h standing for the hydrogen atom. In the first unit of time, the impacts are 2, 6, h, 2; and in the second 6, 2, h, 6. If we represent graphically the impacts in the second unit of time, we perceive that they point to a configuration in which the double linkage is between the carbon atoms 1 and 6, and the single linkage between 1 and 2. Therefore, according to Kekulé, the double linkages are in a state of continual oscillation, and if his dynamical notion of valency, or a similar hypothesis, be correct, then the difference between the 1.2 and 1.6 di-derivatives rests on the insufficiency of his formula, which represents the configuration during one set of oscillations only. The difference is only apparent, not real. An analogous oscillation prevails in the pyrazol nucleus, for L. Knorr (Ann., 1894, 279, p. 188) has shown that 3- and 5-methylpyrazols are identical.
The explanation thus attempted by Kekulé was adversely criticized, more especially by A. Ladenburg, who devoted much attention to the study of the substitution products of benzene, and to the support of his own formula. His views are presentedLadenburg’s formula. in his Pamphlet: Theorie der aromatischen Verbindungen, 1876. The prism formula also received support from the following data: protocatechuic acid when oxidized by nitrous acid gives carboxytartronic acid, which, on account of its ready decomposition into carbon dioxide and tartronic acid, was considered to be HO·C(COOH)3. This implied that in the benzene complex there was at least one carbon atom linked to three others, thus rendering Kekulé’s formula impossible and Ladenburg’s and Claus’ possible. Kekulé (Ann., 1883, 221, p. 230), however, reinvestigated this acid; he showed that it was dibasic and not tribasic; that it gave tartaric acid on reduction; and, finally, that it was dioxytartaric acid, HOOC·C(OH)2·C(OH)2·COOH. The formation of this substance readily follows from Kekulé’s formula, while considerable difficulties are met with when one attempts an explanation based on Ladenburg’s representation. Kekulé also urged that the formation of trichlorphenomalic acid, shown by him and O. Strecker to be trichloracetoacrylic acid, was more favourably explained by his formula than by Ladenburg’s.
Other objections to Ladenburg’s formula resulted from A. von Baeyer’s researches (commenced in 1886) on the reduced phthalic acids. Baeyer pointed out that although benzene derivatives were obtainable from hexamethylene compounds,Baeyer’s researches. yet it by no means follows that only hexamethylene compounds need result when benzene compounds are reduced. He admitted the possibility of the formulae of Kekulé, Claus, Dewar and Ladenburg, although as to the last di-trimethylene derivatives should be possible reduction products, being formed by severing two of the prism edges; and he attempted to solve the problem by a systematic investigation of the reduced phthalic acids.
Ladenburg’s prism admits of one mono-substitution derivative and three di-derivatives. Furthermore, it is in accordance with certain simple syntheses of benzene derivatives (e.g. from acetylene and acetone); but according to Baeyer (Ber., 1886, 19, p. 1797) it fails to explain the formation of dioxyterephthalic ester from succinosuccinic ester, unless we make the assumption that the transformation of these substances is attended by a migration of the substituent groups. For succinosuccinic ester, formed by the action of sodium on two molecules of succinic ester, has either of the formulae (I) or (II); oxidation of the free acid gives dioxyterephthalic acid in which the para-positions must remain substituted as in (I) and (II). By projecting Ladenburg’s prism on a plane and numbering the atoms so as to correspond with Kekulé’s form, viz. that 1.2 and 1.6 should be ortho-positions, 1.3 and 1.5 meta-, and 1.4 para-, and following out the transformation on the Ladenburg formula, then an ortho-dioxyterephthalic acid (IV) should result, a fact denied by experience, and inexplicable unless we assume a wandering of atoms. Kekulé’s formula (III), on the other hand, is in full agreement (Baeyer).
This explanation has been challenged by Ladenburg (Ber., 1886, 19, p. 971; Ber., 1887, 20, p. 62) and by A. K. Miller (J.C.S. Trans., 1887, p. 208). The transformation is not one of the oxidation of a hexamethylene compound to a benzenoid compound, for only two hydrogen atoms are removed. Succinosuccinic ester behaves both as a ketone and as a phenol, thereby exhibiting desmotropy; assuming the ketone formula as indicating the constitution, then in Baeyer’s equation we have a migration of a hydrogen atom, whereas to bring Ladenburg’s formula into line, an oxygen atom must migrate.
The relative merits of the formulae of Kekulé, Claus and Dewar were next investigated by means of the reduction products of benzene, it being Baeyer’s intention to detect whether double linkages were or were not present in the benzene complex.
To follow Baeyer’s results we must explain his nomenclature of the reduced benzene derivatives. He numbers the carbon atoms placed at the corners of a hexagon from 1 to 6, and each side in the same order, so that the carbon atoms 1 and 2 are connected by the side 1, atoms 2 and 3 by the side 2, and so on. A doubly linked pair of atoms is denoted by the sign Δ with the index corresponding to the side; if there are two pairs of double links, then indices corresponding to both sides are employed. Thus Δ1 denotes a tetrahydro derivative in which the double link occupies the side 1; Δ1.3, a dihydro derivative, the double links being along the sides 1 and 3. Another form of isomerism is occasioned by spatial arrangements, many of the reduced terephthalic acids existing in two stereo-isomeric forms. Baeyer explains this by analogy with fumaric and maleic acids: he assumes the reduced benzene ring to lie in a plane; when both carboxyl groups are on the same side of this plane, the acids, in general, resemble maleic acids, these forms he denotes by Γcis-cis, or shortly cis-; when the carboxyl groups are on opposite sides, the acids correspond to fumaric acid, these forms are denoted by Γcis-trans, or shortly trans-.
By reducing terephthalic acid with sodium amalgam, care being taken to neutralize the caustic soda simultaneously formed by passing in carbon dioxide, Δ2.5 dihydroterephthalic acid is obtained; this results from the splitting of a para-linkage. By boiling with water the Δ2.5 acid is converted into the Δ1.5 dihydroterephthalic acid. This acid is converted into the Δ1.4 acid by soda, and into the Δ2 tetrahydro acid by reduction. From this acid the Δ1.3dihydro and the Δ1 tetrahydro acids may be obtained, from both of which the hexahydro acid may be prepared. From these results Baeyer concluded that Claus’ formula with three para-linkings cannot possibly be correct, for the Δ2.5 dihydroterephthalic acid undoubtedly has two ethylene linkages, since it readily takes up two or four atoms of bromine, and is oxidized in warm aqueous solution by alkaline potassium permanganate. But the formation of the Δ2.5 acid as the first reduction product is not fully consistent with Kekulé’s symbol, for we should then expect the Δ1.3 or the Δ1.5 acid to be first formed (see also Polymethylenes).
The stronger argument against the ethylenoid linkages demanded by Kekulé’s formula is provided by the remarkable stability towards oxidizing and reducing agents which characterizes all benzenoid compounds. From the fact that reduction products containing either one or two double linkages behave exactly as unsaturated aliphatic compounds, being readily reduced or oxidized, and combining with the halogen elements and haloid acids, it seems probable that in benzenoid compounds the fourth valencies are symmetrically distributed in such a manner as to induce a peculiar stability in the molecule. Such a configuration was proposed in 1887 by H. E. Armstrong (J.C.S. Trans., 1887, p. 258), and shortly afterwards by Baeyer (Ann., 1888, 245, p. 103). In this formula, the so-called “centric formula,” the assumption made is that the fourth valencies are simply directed towards the centre of the ring; nothing further is said about the fourth valencies except that they exert a pressure towards the centre. Claus maintained that Baeyer’s view was identical with his own, for as in Baeyer’s formula, the fourth valencies have a different function from the peripheral valencies, being united at the centre in a form of potential union.
It is difficult to determine which configuration most accurately explains the observed facts; Kekulé’s formula undoubtedly explains the synthetical production of benzenoid compounds most satisfactorily, and W. Marckwald (Ann., 1893, 274, p. 331; 1894, 279, p. 14) has supported this formula from considerations based on the syntheses of the quinoline ring. Further researches by Baeyer, and upon various nitrogenous ring systems by E. Bamberger (a strong supporter of the centric formula), have shown that the nature of the substituent groups influences the distribution of the fourth valencies; therefore it may be concluded that in compounds the benzene nucleus appears to be capable of existence in two tautomeric forms, in the sense that each particular derivative possesses a definite constitution. The benzene nucleus presents a remarkable case, which must be considered in the formulation of any complete theory of valency. From a study of the reduction of compounds containing two ethylenic bonds united by a single bond, termed a “conjugated system,” E. Thiele suggested a doctrine of “partial valencies,”