previous chapter
7— Kekulé, Wurtz, and the Rise of Structure Theory
next sub-section

The French Connection

The deep background of valence was a long tradition of speculations regarding submolecular and even subatomic structures. Examples from the early nineteenth century include the provocative ideas of Avogadro, Prout, Ampère, and Gaudin, which were familiar to most chemists of the middle decades of the century and have been studied by modern scholars. It is well known that Dumas was fascinated by some of these writers, and his explicitly philosophical writings of the 1820s and 1830s, as well as some of his scientific papers, suggest that he took the idea of submolecularity very seriously as a way to explain some of the more puzzling common phenomena of chemistry. Nor was Dumas speaking for himself alone when in 1836 he affirmed "that the chemists'. . . atoms are nothing but molecular groups."[2] What he meant, in modern vocabulary, is that elementary molecules normally consist of more than one atom. But Dumas always let this suggestion remain vague and undeveloped, for the details seemed inaccessible to experiment.

The proximate background to the rise of valence involved the work of Gerhardt and Laurent in the years after 1842. In 1846 Laurent provided consistent definitions of equivalents, atoms, and molecules, defended two-volume formulas, and applied this axiomatic base to chemical theory. His calculational control for the two-volume criterion was the even-number rule.

Laurent divided all elements into two classes: monasides or monads , such as carbon, oxygen, sulfur, and selenium, and dyodides or dyads , such as hydrogen, nitrogen, chlorine and other halogens, phosphorus, arsenic, sodium, and silver. The even-number rule states that the number of atoms of all dyads in a molecule must be even. If it was not, the formula was incorrect and had to be doubled or otherwise modified. Following his rule, Laurent could tell at a glance the falsity of formulas such as those for water = HO, cyanogen = CN, and ethyl = C2 H5 . He explained the rule by supposing that dyadic elements exist in the form of "binary" molecules such that each atom must always be joined to a complement (a moitiè complémentaire ). In contrast, monads can exist either singly or in pairs. Thus, dyads always appear in twos, while monads may appear in any integral number.[3]

But Laurent encountered difficulty in applying his principles to certain metals. "It was then that I asked myself," he wrote soon before his untimely death in the spring of 1853, "whether the atoms of chemists might not be divisible." He then echoed the very words of Dumas in supposing "that the chemical atoms are but molecular groups, composed of a certain number of minuter atoms."[4] This hypothesis re-


moved the anomalies. For instance, by hypothesizing two sorts of iron atoms of different sizes—composed of different numbers of identical iron sub atoms—both the ferric and ferrous series of compounds would conform to the even-number rule. The hypothesis would also account for systematic chemical differences between the ferric and ferrous series and could easily be applied to other metals such as copper, mercury, and platinum.[5]

There is no indication in this passage as to when Laurent arrived at his idea of subatomic particles, but as early as 1843 he was already applying it to certain elements. For instance, if the atoms of manganese metal are composed of groups of twenty-four "atomes plus petits," then the three series of salts of manganese could be represented as compounds of three different smaller conglomerations of manganese subatoms. Gerhardt adopted Laurent's notational convention, if not his physical hypothesis, for his Traité de chimie organique .[6]

Laurent's proposals of physical hypotheses to explain the empirical phenomena covered by the even-number rule and the varying valence of certain metals broached the idea that subatomic structure might explain the varying capacities of different atoms to unite chemically with atoms of other elements. The same concept was hidden within his insistence (here following his friend Gerhardt) that atomic weight formulas HCl, H2 O, NH3 , CH4 , and so on are actually more empirical and more consistent than the equivalent weight formulas H2 Cl2 , H2 O2 , N2 H6 , C2 H4 , and so on. Indeed, by the time of his death he had come to the conclusion that so-called "equivalents" as defined and used by chemists since Wollaston and Davy were nothing less than a rival set of atomic weights, no more empirical and no less hypothetical than Berzelius' atomic weights. This was a philosophically correct conclusion that was inadequately heeded both by nineteenth-century chemists and by twentieth-century chemical historians.[7]

Even before Laurent's death, Alexander Williamson explored the implications of Laurent's work by arguing for the "bibasic" character of various radicals and, at least by implication, the oxygen atom. Shortly thereafter, William Odling, an associate of Williamson, wrote of the "replaceable, or representative, or substitution value" of the atoms of a number of elements. From a different starting point, but also consciously following substitutionists such as Dumas, Laurent, Hofmann, and Wurtz, Edward Frankland provided the first explicit and reasonably general statement of valence regularities, even though he formulated the argument in equivalents.[8] But it was Adolphe Wurtz who proposed a clear—indeed, virtually the only—physical hypothesis that could account for why atoms have "polybasic" character, or "sub-


stitution values," or "saturation capacities," to use the language of Williamson, Odling, and Frankland.

Wurtz' valence hypothesis emerged in the context of his work in following up Williamson's research on polybasic radicals. Upon the appearance of an article wherein Williamson described the formation of trinitroglycerin and triethoxymethane and used formulas suggesting that both were combinations of tribasic radicals, Wurtz prepared a French abstract of the paper for the Annales de chimie , and published a related article of his own in the same issue (April 1855), immediately following Williamson's.[9] Developing Williamson's argument more fully than Williamson had done, Wurtz depicted glycerin as a tribasic radical schematically derived from the propyl radical by abstraction of two more hydrogen atoms. The tribasic glyceryl radical, derived from a triple water type, can form a bond (lien ) tying the three hydroxyl groups together. He also reinterpreted Berthelot's description of mono-, di-, and triglycerides as analogous to mono-, di-, and tribasic acids. Wurtz argued that the correct analogy was to mono-, di-, and trisalts of a tribasic acid.

As he later described these events, this work led him to wonder if it might be possible to test the truth of all of these ideas by synthesizing the intermediate link between tribasic glyc erin and monobasic alcohol , namely, a dibasic (or "diatomic," as he now began to call it) substance he chose to name "glyc-ol." On 24 March 1856, Wurtz acetylated and then hydrolyzed ethylene iodide and isolated the expected product, the first dialcohol.[10] He had opened up a huge new field, and he pursued it aggressively, preparing in the following years an impressive array of polyfunctional organic alcohols and acids.

Wurtz' valence hypothesis was founded on an atomic analogy to the polyfunctional radicals he had begun to introduce and study. It was proposed in an offhand fashion at the end of his July 1855 paper on mixed radicals, in the context of a general defense of the reform movement of Laurent, Gerhardt, and Williamson. He thought that Williamson's and Odling's single, multiple, and mixed types could be explained by considering all of them as formed from multiply condensed hydrogen. Water, for instance, was nothing but a double hydrogen type, with the dibasic oxygen atom replacing two of the four hydrogen atoms. Ammonia was three hydrogen molecules with one atom from each molecule replaced by a single tribasic nitrogen atom. A double water type, as in Williamson's sulfuric acid formula, was quadruply condensed hydrogen, and so on.[11]

Citing the reformers as well as earlier work by Dumas, Ampère, and himself on the compound nature of hydrogen and other elemental


molecules, he adopted the symbols previously used by Odling to indicate "basicity": oxygen is dibasic, O", and nitrogen is tribasic, Az'". A footnote reads: "This property of nitrogen can be explained by supposing that it is itself formed from 3 juxtaposed and inseparable atoms." Thus, Az'" could be written as az3, ammonia as H3 az3 , phosphorus (P'") as p3 , and phosphorus trichloride as Cl3 p3 . "But as this notation rests on considerations which are not susceptible of a rigorous demonstration, I renounce them for the moment."[12]

Wurtz was suggesting that polyvalent atoms are polyvalent because they are accretions of monovalent equivalents; nitrogen has a valence of three and an atomic weight of fourteen because it consists of three equivalents adhering together in some fashion, each equivalent weighing 14/3 = 4.67 and each capable of forming a single bond to another atom, such as hydrogen. He never developed this intriguing speculation to any degree of detail, though he subsequently referred to it many times, calling the particles p or az "little atoms" or "subatoms."[13] He said he relinquished the idea after the discovery of the variability of valence; it was hard to imagine how the phosphorus atom, for instance, could be alternately p3 and p5 .

Although in this paper Wurtz adopted Laurent's notational convention and elements of his theory of subatoms, he was here applying the notion differently than Laurent had. This was an original hypothesis, the first of its kind, designed to explain the law of atomic valence—neither of which, theory or law, Laurent had developed. Wurtz' idea provided a conventient visualization of both valence and equivalence, since the valence is equal to the number of subatoms in an atom and the equivalent is the weight of a subatom. In the 1860s the shorthand definition of equivalent became "atomic weight divided by valence," a simple but imprecise and misleading formula still often used by writers of elementary chemistry textbooks.[14]

Wurtz' subatomic speculation was founded on an analogy between atoms and molecules. Just as Williamson, Odling, and he himself had already shown that there were "polybasic"—or as Wurtz began to call them in 1856, "polyatomic"—radicals, Wurtz thought (following Odling and implicitly Williamson as well) that the atoms themselves must also be polybasic or polyatomic. The word atomicity had already been introduced by Gaudin, with a denotation similar to the twentieth-century one, namely, referring to the number of atoms in an elementary molecule. Wurtz borrowed the term to denote what we now refer to as valence, and he may well have wanted it to carry a similar implication to Gaudin's usage: the "triatomic" (trivalent) nitrogen atom is itself a kind of "molecule" consisting of three monovalent subatoms.


Despite the offhand character of Wurtz' proposal, colleagues took notice. In the 1860s the subatomic speculation was adopted and defended by such chemists as Alexander Crum Brown, Emil Erlenmeyer, Charles Delavaud, and Christian Blomstrand, and in the 1850s it may have been influential for Archibald Scott Couper and August Kekulé as well (as we will see in the next section). A notational convention introduced in 1859 by Kekulé, which was probably based on Wurtz' idea, was adopted in several variations during the 1860s by Wurtz, Alfred Naquet, Joseph Wilbrand, Pierre Havrez, George Carey Foster, and Blomstrand.[15] But after 1869 neither the notation nor the theory was seriously defended.

previous chapter
7— Kekulé, Wurtz, and the Rise of Structure Theory
next sub-section