In the summer of 1950 I spent several weeks in Europe. At Oxford I met the young David Sayre, who told me about his approach to direct methods, which made a deep impression on me. When I returned to Oxford a year later, the calciferol analysis was at a standstill because of the false symmetry of the 3D-Fourier map that is introduced by the phasing from the heavy atom positions alone. I had the idea that it might be possible to break this false symmetry by extending Sayre's method to deal with non-centrosymmetric structures where the Fourier coefficients were associated with phase angles and not merely with signs, plus or minus. For this purpose I developed an equation that was essentially the same as the famous tangent formula developed by Karle and Hauptman. The first step was to make a list of "triples" of strong reflections with indices H, K, H-K that formed, together with the origin, parallelograms in reciprocal space. My plan was to do this graphically, by plotting in ink on transparent sheets the various layers of reflections, moving the origin to each reflection in turn and looking for coincidences. I sat upstairs in a kind of balcony in the large room in University Museum that housed Dorothy Hodgkin’s research group, the same room in which Thomas Henry Huxley had publicly rebuked Bishop Wilberforce in a debate about evolution some 80 years earlier. There, day after day, I overlaid transparent sheet on transparent sheet, looking for parallelograms in three-dimensional reciprocal space. After about two months I took stock; I had completed about 10% of the preliminary, trivial, more or less automatic step in my new method. I estimated I would need to invest about two years of boring, repetitive work before reaching the stage where the method could be tested. I gave up. Ten years later we wrote a computer program that could do the job in a couple of seconds. But by that time the method had been shown to work (23) and the calciferol problem had in any case been solved by other means. (8)
I was in the habit of spending a few hours every week in the Bodleian Science Library reading the latest scientific literature. This was not as voluminous as it is today. One could then read, or at least scan, all the important chemical journals and still have time to do some work on one's own. One afternoon in late spring of 1952 I came across an astonishing proposal from a group of Harvard chemists for the structure of the recently obtained compound, C10H10Fe, an orange colored solid, volatile, insoluble in water but readily soluble in organic solvents. (24) The proposed structure consisted of two parallel cyclopentadienyl rings with the iron atom sandwiched between them. (25) The only physico-chemical evidence offered for this unprecedented structure was the infrared absorption spectrum, which contained, in the 3-4 µ region, a single, sharp band at 3.25 µ, indicating the presence of only one type of C—H bond. It may be difficult today to appreciate just how surprising, unorthodox, even revolutionary, this structure was at the time. At any rate, my first reaction was one of extreme skepticism. On my way out of the Library I met my friend Leslie Orgel, at that time holder of a Research Fellowship at Magdalen College, and asked if he had seen the remarkable structure proposed in the latest JACS number. We retrieved the journal and re-read the article together. He was as skeptical as I was. When we learned that the compound was relatively easy to prepare in crystalline form, we decided to make it and determine the crystal and molecular structure. Or rather, since neither of us had access to facilities in a synthetic laboratory, we decided to try to persuade a friendly organic chemist to carry out the relatively straightforward synthesis. In the nearby Dyson Perrins Laboratory we were fortunate to meet Hugh Cardwell, who agreed that the problem was worthwhile and offered to synthesize a few grams of this unusual and provocative compound. Within a week or two he fulfilled his promise.
According to the record in my laboratory notebook, I made optical measurements on crystals of the new compound on June 9th, 1952 and began to make preliminary X-ray photographs the following day. I soon found that the crystals slowly sublimed in the atmosphere at room temperature and had to be sealed into glass capillary tubes. From the space group alone it was evident that the molecule must sit at a crystallographic centre of symmetry. By the end of the following week, I had made enough intensity measurements to produce two electron-density projections down mutually perpendicular directions. This was possible because, fortunately for me, there was a slack period in the laboratory so that not only one but two X-ray Weissenberg cameras were free for my use. Of course, to get all this done on my own, I had to work long hours, during the evenings and over the weekend too. As the structure began to emerge from the electron-density maps, calculated with Beevers-Lipson strips with the aid of an adding machine, I was becoming so excited that I was working through most of the night as well. By the end of the following week, the structure was solved. Extraordinary as it seemed to me, the Harvard proposal was correct. The rings were parallel, with the iron atom sandwiched between them at a crystallographic centre of symmetry. There was no doubt about it. That was the marvelous thing about crystal structure analysis. When it worked, the result had a satisfying definiteness about it. Even though this aura of definiteness could sometimes be misleading! The crystalline structure of ferrocene occupied me, on and off, for more than thirty years. Later, the apparently staggered orientation of the cyclopentadienyl rings was revealed to be an artifact resulting from crystal disorder. (26) Ferrocene turned out to be trimorphic — at least — and the ring orientation in the low-temperature stable polymorph is eclipsed not staggered. (27)
Back to 1952; there was still the question of how to account for the new kind of bonding in this extraordinary molecular structure. How can the iron atom simultaneously make ten Fe—C bonds? How could the tenfold symmetry be reconciled with the well known tendency of Fe2+ to form 4- or 6-coordinated complexes? Faced with this challenge, within a few days Leslie developed an explanation based on orbital symmetry properties, on the relationships between the symmetry properties of the d-orbitals of the metal atom and the π-molecular orbitals of the cyclopentadienyl rings. This was new terrain. This new type of molecule required a new type of description of its bonding, and Leslie's model, formal and over-simplified as it was, expressed the essence of this. When it was first explained to me I did not understand a word, but by the end of the week I had picked up enough of the group theoretical background of this new language to construct simple statements on my own. In particular, I could see that the model was a generalization of the standard molecular orbital (MO) model of benzene and other aromatic systems. So we wrote a paper, covering both the structure determination and the new theoretical model, and sent it off to Nature on July 2nd, less than a month after we had the crystals, under the provocative title, "Bis-cyclopentadienyl Iron: a Molecular Sandwich." (28) That was the first time, I believe, that this gastronomic epithet had been used in the title of a chemical publication. The name certainly stuck.
I have the obligation to try to make amends for having then deprived my co-author of credit for what, at that time, would have turned out to be an important theoretical prediction. Since the same symmetry arguments as we applied to ferrocene could be applied mutatis mutandis to the then still unknown and scarcely imagined molecules dibenzene chromium and dicyclobutadiene nickel, Orgel wanted to include in our paper his prediction that these molecules would turn out to be stable species. I argued that it would be a pity to spoil a good, solid paper by what could be regarded as risky speculation and managed to persuade him to omit the additional paragraphs. As Leslie later ruefully remarked, one characteristic of our collaboration was that we sometimes succeeded in shooting down each other's best ideas.
There is a nice if-then-what? postscript to this story. (29) In September 1951 Pauson and I happened to meet at the IUPAC meeting in New York. He had with him a crystalline sample of dicyclopentadienyl iron , which he intended to hand over to my former research director, Professor J. Monteath Robertson, in the hope that he would be able to establish the correct molecular structure. Robertson, on his way to Cornell University to give the Baker Lectures, was one of the plenary lecturers at the IUPAC meeting. Pauson and I happened to meet at Robertson’s lecture. As we had both been students at Glasgow University, we knew one another and sat together. Pauson, according to a letter he wrote to me in 1990, nearly forty years later, considered whether he might entrust his precious crystals to me rather than to Robertson, as he had first intended. However, with the crystals "burning a hole in his pocket," he decided to stick to his original plan. He entrusted the crystals to Robertson. After his arrival at Cornell, Robertson gave the material to Lynn Professor J. Lynn Hoard, and Hoard then passed them on to a student, who was not successful in solving the structure. By the following summer, when I became involved in the ferrocene work, I had forgotten about our meeting in New York and was in any case completely unaware of Pauson's predicament. Now we come to the "if" possibility. If Pauson had given the crystals to me at the New York meeting, history would have taken a different course. Instead of wasting my time after my return to Oxford with my direct methods approach to the calciferol problem, I could have started work on Pauson's crystals almost immediately. It is quite likely that the sandwich structure would then have established within a few weeks, in time for Pauson to revise the paper he had sent to Nature with the wrong structure and almost certainly well before March 1952, in which case Woodward and Wilkinson would never had had the opportunity to publish their audacious structure proposal. I wonder what I would have thought when the unprecedented sandwich structure began to take semblance in the electron density maps. (30)
Besides our work on ferrοcene, Orgel and I wrote at that time a paper about hydrogen bonds. It was then generally considered that O-H.. .O hydrogen bonds were unsymmetrical, with the hydrogen atom closer to one oxygen atom than to the other. We proposed that the acid maleate anion should have an unusually strong, symmetrical hydrogen bond, and backed this up with spectroscopic observations on crystalline potassium hydrogen maleate. (31) Our proposal was subsequently confirmed by neutron diffraction studies. (32) This may be the first example of what has come to be known much later as a low barrier hydrogen bond (LBHB).
In September 1952, on the invitation of my old Caltech friend Edgar Heilbronner, now returned to to the ETH in Zurich, I visited there for the first time and lectured on my work on cyclobutane and on ferrocene. I had no idea then that I would spend the greater part of my life there.
Around that time, we used to discuss more biological themes with the budding molecular biologist Sydney Brenner, who arrived in Oxford in the autumn of 1952 to work in Hinshelwood's laboratory on bacteriophage. This was an area about which I had then a smattering of knowledge through my earlier contacts with the Delbrück group at Caltech. So Sydney talked to me, and I talked to Leslie Orgel , and it was not long before we were talking together — sometimes all three at once — about possible roles of proteins and nucleic acids, and the growing evidence for the involvement of DNA as the carrier of genetic information. Of course, our education in these areas was sadly incomplete and fragmentary, but what we lacked in erudition we made up for in fantasy. We argued about whatever we happened to have learned, especially about the possible meaning of the latest results from the latest journals, or rather, as I recall, Leslie and Sydney argued while I acted as a kind of umpire. It was tremendous fun.
At the same time, my connection with the crystallographers in Cambridge brought me every few months into exciting though inconclusive discussions over pub lunches with Francis Crick. Hence I was aware that he and Watson, whom I had known from Caltech, were trying to deduce the structure of DNA by model building but without any reliable diffraction data to test their models. As I recall, their mood oscillated wildly between enthusiastic optimism and downcast pessimism. I did not give them much chance. In late 1952 I advised Watson to abandon the project and get down to some more promising project. Naturally, I reported on this work to my Oxford discussion partners. At any rate, in early April 1953, when Francis telephoned to ask me to come to look at their marvelous new DNA model, all three of us traveled together to Cambridge, together with Dorothy Hodgkin and her young assistant Beryl Ougton. We knew enough about the problem to recognize almost immediately that the proposed DNA structure must be correct in its essential features. Did we realise that we were present at the dawn of a new age? Did we feel: "At this place and on this day a new epoch in the history of the world begins, and we shall be able to say that we were present at its beginnings"? (These are Goethe's words, written on September 20, 1792, the occasion being the defeat of the Prussian army by the ragged French militia at the battle of Valmy.)
In August 1953 I married Barbara Steuer and returned almost immediately to Pasadena for the second stay, as described above.