In a recent small note in Volume III of these Annalen I stated that reaction of cyanogen with ammonia solution gives, in addition to several other products, oxalic acid and a crystalline, white substance, which latter was determined not to be ammonium cyanate, although one always obtains it in attempting, for example by a so-called double decomposition, to unite cyanic acid with ammonia. The fact that in combining, these substances seem to change their nature and a new substance is thus formed directed my renewed attention to this matter, and this investigation has yielded an unanticipated result that reaction of cyanic acid with ammonia gives urea, a noteworthy result in as much as it provides an example of the artificial production of an organic, indeed a so-called animal, substance from inorganic substances.

I have previously stated that one obtains the above mentioned crystalline, white substance best by decomposing silver cyanate with ammonium chloride solution or lead cyanate with ammonia solution. By the latter method I have prepared a not insignificant amount of it for use in this investigation. I obtained it in colorless, clear, often more than inch-long crystals, which formed narrow, right-angled, four-sided prisms without appreciable tapering.

With caustic potash or lime this substance evolved no trace of ammonia, with acids it showed absolutely none of the ready decomposition phenomena of cyanate salts, namely evolution of carbonic and cyanic acids and it even failed to give lead and silver salts as true cyanate salts do; it can thus contain neither cyanic acid nor ammonia as such. Since I found that the last mentioned method of preparation gives no by-products and that lead oxide separates pure, I imagined that in the union of cyanic acid and ammonia there might be formed an organic substance, perhaps even a substance analogous to the vegetable salt-forming bases; from this point of view I planned several experiments on the behavior of acids with the crystalline substance. It was however inert to these with the exception of nitric acid, which in a concentrated solution of this substance formed a precipitate of brilliant crystal flakes. After purification by repeated recrystallization, these crystals displayed a decidedly acidic character, and I was already inclined to consider them as a specific acid, when I found that they gave nitrate salts on neutralization with bases from which the crystalline material could be recovered with alcohol with all the properties it had before reaction with nitric acid.

This similarity in behavior to urea led me to conduct analogous experiments with fully pure urea isolated from urine, with the completely unambiguous result that urea and this crystalline substance, or ammonium cyanate if we could call it so, are fully identical substances.

I mention the behavior of this artificial urea no further since it agrees completely with reports in the literature on urine-urea from Proust, Prout and others, and I note only the fact they did not report, that on distillation the urine-urea, like the artificial, besides giving a large quantity of ammonium carbonate, also evolved a striking amount of the pungent odor of cyanic acid analogous to that of acetic acid, just as I have found previously during distillation of mercury cyanate or uric acid and especially uric acid mercuric oxide. During this distillation of urea there forms at the same time another white, apparently specific substance, which I am now occupied with investigating.

If mixing cyanic acid with ammonia really gives only urea, then urea must have the same composition as one finds through calculation for ammonium cyanate according to the constitutional formula I have given for cyanate salts; and this is in fact the case, if one assumes one atom of water, as all ammonium salts contain water, and considers Prout's the best analysis of urea. According to him^* urea consists of:

Ammonium cyanate would consist of 56.92 cyanic acid, 28.14 ammonia, and 14.74 water which means for the isolated elements:

Thus had one not found formation of urea from cyanic acid and ammonia as in the experiment, one could have calculated as above that ammonium cyanate with one atom of water would have the same constitution as urea. Burning cyanic acid over copper oxide yields 2 volumes of carbon dioxide and 1 volume of nitrogen, but burning ammonium cyanate should yield equal volumes of these gases, thus also the same ratio from burning urea, and in fact Prout has found this.

I refrain from all the considerations which so naturally suggest themselves from this fact, especially in respect to the composition ratios of organic substances and in respect to similar elemental and quantitative compositions among compounds with very different properties, as may be supposed, among others, of fulminic acid and cyanic acid and of a liquid hydrocarbon and the olefiant gas, and it must be left to further investigations of many similar cases to decide what general laws can be derived therefrom.

* ) Annals of Philosophy. Vol. XI. 1818, p. 354. [Original: Prout, W. Observations on the Nature of Some of the Proximate Principles of the Urine; with a few Remarks upon the Means of Preventing those Diseases Connected with a Morbid State of that Fluid, Medico-Chirurgical Transactions, VIII, Pt. 1, 1817, pp. 526-549]

** ) This is based on the new atomic weights of Berzelius: thus N = 88.518, C = 76.437, H = 6.2398, O = 100.000, water (H) = 112.479, ammonium cyanate = NH³ + CNO and urea = NH³+CHO+H .
[Note: The original footnote (shown below)uses special symbols. The dot above the H atom denotes an oxygen and barred atoms are to be doubled. This notation derives from uncertainty about multiplicity in atomic weights, e.g. does an amount of nitrogen 88.518% of an amount of oxygen correspond to the same number of atoms, or twice (or half) as many?]

By far the most powerful tool in the early development of organic chemistry was elemental analysis. Answering the following problems will help you understand some of the factors that were involved in this procedure. Nowadays various kinds of spectroscopy have largely replaced elemental analysis, but in general results from combustion analysis are still required for publishing reports of new compounds. This is among the most venerable, if not the most sensible, traditions in organic chemistry.

1. Compare the elemental percent composition of urea using modern atomic weights with Wöhler's values for theory and experiment. What do you notice?

2. How good are Berzelius's values for atomic weights?

3. Liebig and Wöhler found that salts of fulminic acid (HONC) and cyanic acid (HOCN) were isomeric. How accurate would elemental analyses have to be to reveal that "a liquid hydrocarbon" (say C₁₂H₂₆) and "the olefiant gas" (C₂H₄) are not isomeric? What if, by the liquid hydrocarbon Wöhler meant an olefin, such as butylene (C₄H₈)?

4. After consulting Streitwieser and Heathcock, Section 3.4, reconstruct how the elemental analysis of urea could have been done experimentally. (You need also to know that when heated with cupric oxide and copper many substances released their nitrogen as N₂ which could be measured by volume.) In light of this approach examine the theoretical and experimental results in Wöhler's paper carefully for errors in logic or arithmetic or both. What do you conclude?

1. The "experimental" % composition of Prout agrees better with modern atomic weights than the calculated composition based on the "new atomic weights of Berzelius" does. This would seem to underline the primacy of experiment over theory and to teach the following key lesson:

Lesson I: A good experiment carefully observed and recorded will retain its value forever, while results that are tinkered with to make them conform with current expectations are of transient, if any, utility.

This is the answer the question was designed to elicit. One might go further to infer that it would have been better to base atomic weights on Prout's analysis than on Berzelius's experimental data. But deeper detective work teaches further valuable lessons.

Just how good were Prout's experiments? That he gets the right percentage for Nitrogen to 4 significant figures makes it look like he was accurate to better than 0.1%. Accuracy of a part in a thousand would be fabulous. However, in fact he determined Nitrogen by the volume of N₂ gas that was evolved, and he reported a volume of "6.3 cubic inches", presumably accurate to only one part in 63 or +/- 1% (To judge by a drawing of Prout's apparatus, there would seem to be no way he could measure the gas volume to better than +/-1%). Thus it would appear that in being correct to within 0.1% he was in fact only lucky. This would teach an equally important second lesson:

Lesson II: Luck is a useful thing to have on your side in science.

And there is a further twist. It appears the numbers for Prout's "experimental" elemental percentages, as reported by Wöhler, are not truly experimental!

At the time he analyzed urea, Prout was pushing Prout's Hypothesis that all elements are composed of hydrogen, meaning that all atomic weights should be integral multiples of the atomic weight of hydrogen. He used H = 1, C = 6, N = 14, O = 8, Cl = 36. (Note that Prout's carbon and oxygen were half of ours. We'll see Couper correct the carbon, but not the oxygen, in 1858.)

Prout's Hypothesis is not such a terrible idea in the sense that most of the mass of atoms comes from the protons and neutrons in the nuclei, which have nearly the same mass as hydrogen does (there is also a tiny conversion of mass to energy on forming the nucleus). When an element has several abundant isotopes there is a problem, e.g. Cl = 35.45 (where the dominant isotope of natural Cl is 35, but there is also 24% of the 37 isotope).

For natural C, N, O the relative atomic masses do truly involve integers to within less that 0.1% and H is off by only 0.8%. So for compounds containing only these atoms Prout's theoretical atomic weights work better than Berzelius's experimental atomic weights (see Question 2).

In fact the "experimental" weight percentages that Wöhler took from Prout's analysis of urea are theoretical. Although he doesn't say so, Prout calculated them by using his integral atomic weights and an atom proportion (CH2NO, using his atomic weights) that he inferred from his approximate analytical results. Because relative atomic weights from Prout's Hypothesis are pretty good, his "results" agree with modern expectations to within 0.1 to 0.7%, even though his actual experiments were correct to within only 1.5 to 4%. The same is true for two of three other compounds whose analyses Prout reported in the same paper. With the fourth compound it seems he just messed up.

Lesson III: Juggling experimental results to make them agree with expectations may succeed in some cases, but only if you're really lucky.

My advice - Don't fudge data, we now recognize that it is dishonest and a recipe for disaster. Worst of all, adjusting the data pretty much assures that you'll never learn anything really new, because you make everything fit the Procrustean bed of preconceived notions.

2. If oxygen were 100.000, then N should be 100 * 14.01 / 16.00 = 87.56 and Berzelius's value of 88.518 is 1% too heavy. Similarly his C is 2% too heavy and his H is 1% too light.

3. C₁₂H₂₆ is 12 * 12.011 / (12 * 12.011 + 26 * 1.008) = 84.6% by weight Carbon. C₂H₄ is 85.7% by weight Carbon. So the amount of CO₂ from combustion would have to be correct within 1.1/85.7 = 1.3%, which is less than the error in Berzelius's atomic weights. However, the hydrogen percentages are 15.4 and 14.3, respectively. So the amount of H₂O would have to be correct only to within 1.1/15.4 = 7%. It would be possible to tell this difference with the kind of precision Berzelius obtained. Since C₂H₄ and C₄H₈ have the same C:H ratio they could not be distinguished by elemental analysis, only by determination of molecular weight, a process which was not agreed upon by organic chemists until after 1860.

4. This whole thing is a comedy of errors. At first one might think that in reporting Prout's analytical results Wöhler, and Prout before him, were so honest as to report the observed amounts of elements, even though they did not sum to 100%. However, Carbon was measured as CO₂, Hydrogen as H₂O, and Nitrogen as N₂, with Oxygen determined by difference, that is by subtracting the amount of C,H,N from the original total weight (assuming that no other elements were present). Thus both sets of percentages (experimental and theoretical) must sum to 100%!

The reported results show that in these early days of analysis Wöhler, a very intelligent person but until recently a student, was so naive as not to notice that components should sum to 100%. This glimmer of frailty helps us establish a fellow feeling with Wöhler, Prout and the other giants who developed organic chemistry and makes it clear that you don't have to be perfect to make an important contribution, just honest, diligent, perceptive, and lucky.

There is an addition error in the "experimental" case. The elements should sum to 99.945, not 99.875. This addition error was corrected in a subsequent French version of the paper. Most of the remaining difference from 100 is due to two cycles of round-off error in Prout's two-stage calculation of weight of the elements in 4 grains of urea. Prout's round-off errors were partially offset by Wöhler's mistake in reporting the % of hydrogen as 6.670 rather than 6.650. (Remember that these results were not truly experimental, as explained in the answer to Question 1 above.)

In the calculated case for ammonium cyanate the numbers were added correctly, but 20.198 was rounded down to 20.19, rather than up, and more importantly two digits were transposed in 26.24. These errors were not corrected in the French, so even when Wöhler (or an editor) went over the numbers, he didn't notice the problem of failure to total 100%.

We can't blame Wöhler too much for this error, after all this was early days, and quantitative precision didn't have a lot to do with the point of the paper. Still, especially if analysis is all you have, it is important to do it right. It is remarkable that no one else (as far as I know) from these later, more sophisticated days has commented on the arithmetic errors in this widely quoted (if not carefully read) paper of great historical significance. [Analyses in papers of Berzelius, Wöhler's teacher, did total 100%.]