Predictions Unfulfilled:
Hydroxyacids
In the fall of 1858, Kolbe put his student Carl Ulrich on the problem of hydrolyzing and reducing Wurtz' chlorolactyl chloride. Ulrich found a way to remove the two chlorine atoms in two clean stages, first by generating chloropropionic acid and then by reducing this to propionic acid by means of nascent hydrogen. The fact that the process occurred in two stages neatly underlined Kolbe's point that the two hydroxyl functions of lactic acid were not equivalent, hence lactic acid was not dibasic as Wurtz claimed. Moreover, Ulrich had transformed a lactic acid derivative into propionic acid, proving the close relationship Kolbe had been asserting.[17]
Simultaneously with Ulrich's paper, Kolbe published a summary of his views on the constitution of lactic acid. He rehearsed all of the arguments just published in his textbook; regarding the nonalcoholic character of glycols, he stressed that aldehyde had never been produced from glycol. Debus' glyoxal, interpreted by its creator as a double aldehyde, had been prepared from alcohol, not glycol, and in any case, there was no proof that it was an aldehyde.[18]
In a long article published simultaneously with Kolbe's and Ulrich's,
Wurtz developed his thoughts on reactions of polyatomic alcohols (derivatives of glycerin and glycol) on which he had done so much valuable work.[19] It was the first time Wurtz used Williamson's atomic weights (barred O's and C's indicating doubled conventional equivalents), and he thoroughly discussed the incipient theory of "molecular structure," as he termed it here. Wurtz attempted to build a case for the crucial role his research on polyatomic alcohols had played for the rise of this theory.
I will force myself to be brief; for although I attribute a high value to theory, which must be the foundation and the end of all science, I believe that, above all, facts should be allowed to speak for themselves, and that in chemistry, theory consists only in the direct and judicious interpretation of that which experiment teaches us.
Among other things, "the theory" had suggested to Wurtz that there ought to be an intermediate member between glycerin and normal alcohol; this thought had led to his discovery of glycol. He reported that he was continuing to produce large numbers of glycol derivatives in his lab, as predicted by the theory.[20] The mere existence of glycols was unimportant; what mattered was that it fulfilled predictions, and transformed the "hitherto vague and unsupported hypothesis" of polyatomic radicals into substantiated fact.[21]
Wurtz noted that polyfunctionality means that type or radical formulas in general capture only partial views of molecules, but they are nonetheless of enormous value. "I realize that many people abuse them. But abuse does not condemn use."[22] He felt that one must not reject rational formulas, but that one must also not abuse them by putting too much trust in them. "It is wrong to present these things as the law and the prophets." Formulas cannot provide ultimate depictions of "the intimate constitutions of compounds," but rather should serve as guides to the prediction and interpretation of reactions.[23] Wurtz cited A. S. Couper and Kolbe as two men who had erred by believing their formulas too literally. Kolbe had often averred that multiple types are imaginary, hence useless, because (for instance) the double water type H4 O2 as the basis for formulating sulfuric acid does not exist in nature. Wurtz responded that this objection "is not serious," for the advocates of such types, including himself, had always been careful to specify that the cause of cohesion of multiple types is a polyatomic radical. Williamson had formulated the hydroxyls of sulfuric acid as held together by a diatomic SO2 radical, and Wurtz had formulated the three fatty acids of triglycerides as held together by the triatomic glyceryl radical.[24]
When Kolbe's and Ulrich's papers appeared in the spring of 1859,
Wurtz responded almost immediately with new compounds supporting new arguments. Chlorolactyl chloride was the starting material for both esterification of the acid group and acylation of the alcohol group, producing "lactobutyric ester" (butyryl ethyl lactate). Wurtz also prepared an ethyl ether/ethyl ester from the same starting material, using two moles of Williamson's sodium ethoxide. These reactions once more emphasized the difunctionality of lactic acid. However, Wurtz conceded the point deduced from Ulrich's reactions, namely, the non-equivalence of those two functions. Lactic acid is not dibasic in the same sense that oxalic acid is, Wurtz admitted. However, it is nonetheless diatomic (that is, it has two replaceable hydrogen atoms).[25] Here he was following Kekulé's ideas and language from a paper on glycolic acid published the preceding year.[26]
Wurtz took this occasion to stress once more the alcoholic character of the glycols and their close relationship to the hydroxyacids. Kolbe's name for ethyl alcohol was "ethyl oxide hydrate" and that for glycol "ethylene oxide hydrate." These two names show even more analogy than Wurtz' "alcohol" and "glycol." Kolbe had said he feared that the definition of "alcohol" would be stretched beyond all recognition or meaning by the inclusion of the glycols in this class. "Let him be reassured on this point," Wurtz wrote; polybasic acids such as oxalic acid have not destroyed the concept of acid! As for the fact that glycol had not yet been converted into an aldehyde, this was but a temporary situation. Debus' glyoxal was indeed the aldehyde analog of glycol, even though it had been prepared from alcohol and not glycol. In any case, no one was denying that methyl alcohol is an alcohol, even though no one had isolated a methyl aldehyde. Finally, several glycols had been converted to acids, analogous to ethyl alcohol being converted to acetic acid, which was Kolbe's own criterion for alcoholic character. As for Kolbe's argument that lactic acid is monobasic, "or rather monatomic," Wurtz thought that his reactions had shown this thesis to be untenable. Besides, where were all of Kolbe's hypothesized isomeric alcohols and aldehydes, the so-called true reduced analogs to the hydroxyacids? Wurtz believed that it was Kolbe, not he, who dwelt in the land of hypothesis.[27]
Kekulé was in complete accord with Wurtz on the matter of formulas and their interpretation, and also on the power of structure theory. As he wrote Lothar Meyer in 1860, "We and science quietly wend our way between the mischief of those who make a game of constitutional formulas, and the indolence of those who deny [rational] formulas, toward the star of a fundamental synthesis beckoning from afar."[28] As it happened, the first fascicle of Kekulé's textbook was being printed while Wurtz' papers just described were being published.
In this fascicle, Kekulé emphasized that rational formulas are de-
rived solely from reactions, and that one must be allowed to write different formulas for the same compound, depending on what functionality was in question for a given reaction. He wrote, "It is clear that even for acetic acid—and all the more so for more complicated compounds—a completely comprehensive rational formula is not appropriate for ordinary use, even if one can be specified in the present state of the science." Rather, one uses whatever formula most clearly makes the point in question. Type notation is handy for many, even most situations, he felt. It clarifies, for example, the different chemical behavior of the "typical" hydrogen in alcohol, or the "typical" hydrogen in acetic acid, from the other hydrogens in the compound. However, type formulas, too, have their limitations.
There are several cases where different hydrogen atoms should be equivalent according to the type theory, and are not. For instance, glycolic acid, as well as lactic acid, behave like monobasic acids, although they contain two typical hydrogen atoms. . . . One behaves just like the typical hydrogen of alcohol, the other just like the typical hydrogen of acetic acid. The different behavior of these two hydrogen atoms is apparently caused by the different positions they occupy with respect to the other atoms, particularly oxygen. One hydrogen atom lies in the neighborhood of two oxygen atoms, like that of acetic acid; the other lies in the neighborhood of one oxygen atom, like that of alcohol.[29]
Given the context of the rest of this fascicle, in which Kekulé laid out his founding version of structure theory, there can be little doubt what he had in mind here: lactic acid is hydroxyacetic acid, an alcohol-acid. He indicated here the via media between Wurtz and Kolbe, while also implying the first adequate fully resolved formula for the compound. The implication was made explicit in papers published by Kekulé, W. H. Perkin, and Alexander Crum Brown, all in 1861; Crum Brown made clear that he was only reading between the lines of Kekulé's 1859 quotation, just cited.[30]
But Kolbe never blinked. On the attack once more, he and his student Lautemann published another trio of articles on lactic acid in the February 1860 issue of the Annalen . Given the task by his mentor of converting lactic directly into propionic acid—in other words, foregoing the chloro intermediate—Lautemann found success with his fifth attempted reducing agent, hydrogen iodide. (This marked a significant methodological innovation in organic chemistry, for hydrogen iodide proved to be a very versatile reducing agent.) Concurrently, Kolbe discovered how to convert lactic acid to alanine (Strecker had accomplished the reverse). Both reactions tightened the analogies to the monobasic series, in this case propionic acid.[31]
In short, Kolbe thought that his proof of lactic acid as monobasic
was irrefutable and that Wurtz' own work had only further confirmed this.[32] But just for good measure, he added more arguments to finish Wurtz off. Lactic acid forms no salts with two different metals and has no acid salts, nor does it have a diester. What Wurtz called a diester—his diethyl product using sodium ethoxide—is actually oxyäthylpropionsaures Aethyloxyd , a substituted monoester. As for Wurtz' butyryl ethyl lactate, Kolbe surmised that the compound was actually Oxybutyroxylpropionsäureäther , a hydroxy ketone monoester. Finally, Kolbe excoriated Wurtz for suggesting that both of their different formulas might apply equally to ethyl chloropropionate.
I confess I do not have so broad a chemical conscience, and could never countenance such a doctrine, even if it had to do with more than simply a weak hypothesis. I believe that with these words Wurtz has passed judgment on his own hypothesis. . . . The symbolic expressions for our views on the proximate components of a compound and on their relative positions may of course change. But to assign a compound two different rational formulas at the same time , i.e., to maintain that it possesses sometimes one set of atomic groupings as proximate components, and at other times another set. . . is to maintain an impossible proposition.[33]
This gave Wurtz another opening. As far as their chemical consciences were concerned,
Mine is less delicate concerning formulas. I envision them as expressing the mode of derivation, parental ties, and reactions of compounds, and in no way share the opinion of M. Kolbe, who endeavors to express the exact grouping of the atoms with the aid of his rational formulas. He pretends to know the nature and role of the groups in organic compounds. . . . I express merely parental ties. I express certain reactions, and everyone will agree that it is impossible to express all reactions by means of formulas of this kind.
Wurtz then carefully reiterated his position: lactic acid is indeed monobasic, which explains the absence of dimetal and acid salts. It is, however, diatomic, that is, it has alcohol character, a fact that Kolbe was trying to ignore. Wurtz had no intention of contesting Kolbe's key assertion that lactic acid is related to propionic acid; but it is just as clearly related to propylene glycol, for the latter oxidizes smoothly to lactic acid. Wurtz pointed out that the products of hydrolyzing butyryl ethyl lactate were consistent only with his, and not with Kolbe's, formulation of the compound. Finally, Wurtz presented a table directly comparing his and Kolbe's formulas for the same set of lactic acid derivatives and suggested that chemists choose between them. One might differ over issues of esthetics and informational content, but there is no question that Wurtz' were more compact and simpler.[34]
The following year Wurtz let fly another volley, in conjunction with his student Charles Friedel. They directly compared the two ethyl compounds of lactic acid, namely, the ethyl ester and the ethyl ether; the former was neutral, while the latter was fully as acidic as the parent acid. It would be hard to imagine a clearer demonstration of the replaceability of both "typical" hydrogen atoms and also their chemical nonequivalence.[35]
Kolbe let loose his own shot. "The efforts of some chemists," he wrote in a paper co-authored with Lautemann, "to demonstrate alcohols and aldehydes also for dibasic acids, e.g., to claim ethylene oxide hydrate as the alcohol and glyoxal as the aldehyde of the dibasic oxalic acid, are unscientific frivolities that deserve no notice here."[36] This was a strange outburst, both in the unjust violence of expression, as well as in its logic. Kolbe himself had repeatedly suggested that dibasic acids must have reduced forms—he had simply denied that glycol and glyoxal are the reduced forms of oxalic acid.[37]
Debus was moved to respond. It is possible, he wrote, to recast the Kolbe-Lautemann assertion into "a decent form." However,
Before the judgment of Messrs. Kolbe and Lautemann can make the slightest claim for consideration, the concepts indicated by the words "aldehyde" and "alcohol" must be clarified. Then there must be derived from these concepts, or from a general principle, or from an a priori intuition, and not for example from any set of empirical observations, the impossibility that dibasic acids may correspond to aldehydes or alcohols. Messrs. Kolbe and Lautemann have not to my knowledge offered any demonstration of this sort, and therefore their verdict loses all foundation.
Debus also pointed out that Kolbe had predicted precisely what he was now claiming to be impossible. And since he had predicted such alcohols and aldehydes, what proof had he that glycol and glyoxal were not those compounds?[38]
Kolbe's search for these missing substances led him to speculate on a possible isomerism phenomenon in glycolic and lactic acids. Might it not be reasonable to think that they could exist in two modifications each? In particular, perhaps the conventional glycolic acid is hydroxymethyl formic acid and is produced from the oxidation of chloroacetic acid. In contrast, Debus' oxidized alcohol may not be identical to this compound but rather it may be the isomer methoxy formic acid, a monoester of carbonic acid. Similarly, perhaps conventional lactic acid is hydroxyethyl formic acid, while the known isomeric compound, lactic acid from meat, is ethoxy formic acid; or the other way around.[39] He privately speculated on the further possibility that oxalic acid may be analogous to glycolic acid, in the sense of being monobasic but di-
atomic. In this case, there should be two chemically distinct ethyl oxalates; a yet unknown isomer of oxalic acid should also exist that is truly dibasic and homologous with malonic and succinic acids.[40] Nothing concrete came of these ideas. Wurtz, too, had suggested the possibility of isomeric glycolic acids, but as early as 1858 Kekulé had asserted the identity of all candidate isomers.
Few of Kolbe's colleagues shared his sense of triumph over Wurtz in the matter of glycolic and lactic acids. Disagreeing with Wurtz over what must have seemed to most observers to be relatively subtle structural or even semantic distinctions, Kolbe generated many predictions, few of which were realized. None of his putative alcohols and aldehydes isomeric with Wurtz' and Debus' compounds, which he considered the true reduced analogs of the acids derived from glycols, were ever found, nor were the predicted isomers of glycolic, lactic, and oxalic acids or ethyl oxalates ever prepared. He himself, in conjunction with Guthrie, refuted his own prediction that one could dehydrate glycols to yield ordinary aldehydes.[41] He also conceded Wurtz' refutation of his interpretation of the constitution of butyryl ethyl lactate.[42] He ultimately adopted Wurtz' and Kekulé's view and language regarding the nature of lactic acid—that it was monobasic and diatomic—but he regarded this as his victory, not Wurtz'.[43]
In the end, his strong uncollegial language in a matter that was even under the most favorable interpretation contestable, and that many considered a losing cause, could only do Kolbe damage. This was one more repetition of the sort of unpleasant polemics that he had waged over the previous ten years against Gerhardt and Williamson. By March 1860, despite his newly productive research program, Kolbe felt isolated and under attack from most sides. The newer type theory continued to attract adherents, including such respected establishment figures as Will, Kopp, and Strecker, a fact that thoroughly mystified Kolbe. Liebig had been acting unfriendly toward Kolbe for years, Berzelius was long dead, and his own mentors Wöhler and Bunsen, although supportive, were neither interested nor active in theoretical matters. In later years, he often reminded the chemical community of this period in which, as he put it, he was considered a "crank."[44]
This unhappy situation was transformed in 1860 due to Kolbe's work with diacids and his predictions of secondary and tertiary alcohols.