Structure Theory and the Philosophy of Chemistry

During the eighteenth and much of the nineteenth centuries, the science of chemistry was more thoroughly imbued with the methodology and general culture of natural history than with that of the physical or mathematical sciences. Chemists were usually depicted—and depicted themselves—as naturalists exploring the earth's great garden of chemical species, in just the same way that the true naturalists were widening our horizons on biological species. This was the reason why systems of chemical classification were regarded as so central to the science, why Linnaeus' search for a natural botanical system was emulated by chemists, why genetic and parental relationships among chemical compounds were emphasized, and why biological metaphors were so often employed in chemistry.^[15] It also partially explains the antitheoretical attitudes of many nineteenth-century chemists.

But this naturalist's scientific self-image was also mixed with some of the naturalist's scientific self-doubt. Many chemists felt the need for a general theory to guide their investigations, on the model of the exact sciences, but this goal proved to be elusive. Lavoisier, as much a physicist as a chemist, sought to create a general theory centered on oxygen and embracing physics and physiology as well as chemistry. Dalton's efforts to extend chemical atomism to physics were largely unsuccess-

ful. His atomic theory became the closest thing to a generalized theory for the science, but most chemists innocently maintained the fallacy that this theory was coterminous with the law of definite proportions. Berzelius attempted to develop an electrochemical theory that would span the science, but he ran into increasing difficulties. Advocates of the radical and type theories of the 1830s and 1840s likewise tried to create general theories that could serve as guides to chemical investigations. Both theories came to be seen as insufficiently flexible and general until the advent of structure theory, which combined some of the best features of each.

As they began to search for a general theory analogous to that of the physicists, chemists also increasingly adopted the hypothetico-deductive methodology that had already begun to penetrate the exact sciences.^[16] For a time, the inductivist commitments of most chemists kept hypothetico-deductivism suppressed, at least on the rhetorical level, but by mid-century it was breaking through into the science. As early as 1836 Dumas opined

Theories have always been regarded as things quite different from truth; for this reason theories have long been accorded an importance proportional to the service that they render. . . . In chemistry, our theories are crutches; to show that they are good, they must be used to walk. . . . A theory established by twenty facts must explain thirty, and lead to the discovery of ten more; but nearly always it is modified or succumbs to ten new facts added to all that went before.^[17]

Dumas may have acquired his taste for deductivism from Ampère, who was a particularly adept user of the method,^[18] or from the physical chemist C. L. Berthollet. It appears that he communicated the same predilection to his protégé Wurtz.

In England, John Herschel's popular Preliminary Discourse on the Study of Natural Philosophy offered fervent praise of deductivism (despite the Baconian overlay); the book is known to have strongly influenced Michael Faraday. Both Herschel and Faraday were, of course, chemists as well as physicists. Herschel's book may also have been influential for Alexander Williamson; not only Williamson's hypothetico-deductivism, but also some of his language (the search for verae causae and for experimenta crucis ), may well have derived from Herschel.^[19]

As far as Germany is concerned, throughout the 1820s and 1830s Liebig was inclined toward hypothetico-deductivism, as was Berzelius, who was so influential for the direction of German chemical theory. Given the prevailing inductivist atmosphere in chemistry, however, both men were rhetorically cautious about their methodology and were

also no doubt to a degree self-deluded. In 1862, after the method of hypothesis had become more generally accepted, Liebig composed a devastating critique of Baconian induction.^[20] German physics was beginning to become perceptibly hypothetico-deductivist as early as the 1830s, with the work of such men as Gauss, Weber, and Neumann.^[21] Kolbe, a student of the inductivists Wöhler and Bunsen, was an instinctive hypothetico-deductivist from the time of his earliest scientific work. His first extended paper contains repeated and explicit references to the method of hypothesis, at a time (1845) when this was uncommon in chemistry.^[22] Liebig or Berzelius are the most probable sources for this predilection.

It is clear from all of this that the beginnings of a trend toward development of a strongly generalized and hypothetico-deductivist pattern of theorization modeled on the exact sciences could be discerned in chemistry at the time of Kolbe's entry into the field around 1840. These trends, however, had not yet entered the lifeblood of chemistry, certainly much less so than for physics and astronomy. The model of natural history was still dominant for chemists. In this methodological sense, it was the theory of structure that made the difference and created a new era for the field. Structure theory was a general theory that could be applied to inorganic as well as organic chemistry. In a stronger sense, its success also demonstrated the success of the underlying and more general atomic theory; after the 1860s, few chemists ever again seriously questioned the theory or equated it with the law of definite proportions.^[23]

Moreover, structure theory was well suited for application of hypothetico-deductive method. Once its principles were understood and accepted, the theory generated hypotheses almost effortlessly, each of which could suggest one or more experimental tests. One of the qualities that made the theory so powerful for the method of hypothesis was the gradually acquired ability of organic chemists to synthesize new artificial compounds. As their synthetic repertoire rapidly enlarged during the middle decades of the century, chemists were increasingly able to ask and answer their scientific questions experimentally.

Perhaps the earliest legitimate example of this theoretical technique is Williamson's ether synthesis of 1850. Ether had first been prepared from alcohol in the sixteenth century, which illustrates the point that synthesis per se was nothing new to the nineteenth century. However, Williamson's innovation created a way to produce novel "tailored" ethers at will. In doing this, his intent was not the naturalist's ambition to add as many new chemical species as possible to the world's store of knowledge, but rather to provide a specific test, a crucial experiment,

for two theories that could thereby be made to dictate two different experimental outcomes. Whether or not retrospective logical analysis can sustain Williamson's claim to have provided an irrefutable test, empirical history supersedes logical necessity, for as we have seen, his argument was soon universally regarded as compelling.

The polemic between Kolbe and Wurtz during the years 1857-1861 on the structure of lactic acid provides innumerable additional examples of this sort of approach. Kolbe and Wurtz not only explored new oxidations, reductions, and halogenations, they also created larger molecules by esterifications and ether syntheses, and each of these reactions represented an attempt to provide evidence for a particular point of view. Kolbe's predictions of the synthesis of new diacids and new secondary and tertiary alcohols, which soon led to their preparation, are particularly striking examples. Finally, it needs to be stressed that synthesis now could be placed beside analysis—indeed, by the 1860s it had already usurped the pride of place—in structural investigations. We have seen several examples of this trend in the previous section.