Steve Harrison Oral History

Recorded and edited by
Sondra Schlesinger

March 30, 1999 &
April 1, 1999

Stephen C. Harrison is now the Higgins Professor of Biochemistry at Harvard University and Professor of Pediatrics at Harvard Medical School. The idea that the structure of a virus could be solved by X-ray diffraction originated with J.D. Bernal and his colleagues in the 1930s, but it wasn't until 1978 that the goal was finally realized when Steve solved the structure of tomato bushy stunt virus to a resolution of 2.9 angstroms. Much of this oral history describes the steps that led to this accomplishment. The quest began in 1965 when Steve was a graduate student in Don Caspar's lab at the Harvard Medical School.

Steve doesn't remember a time that he wasn't interested in science and one of the important influences must have been growing up in a family of scientists. His Father, Harold E. Harrison, M.D. was a Professor of Pediatrics at the Johns Hopkins School of Medicine and Chief of the Pediatric Service at Baltimore City Hospital. His Mother, Helen C. Harrison, Ph.D., is a biochemist and collaborated with her husband throughout a long research career. Initially Steve's interests were in chemistry and physics and those were the subjects he chose to major in when he was an undergraduate student at Harvard University. He remembers, however, discovering the excitement of the new biology - molecular biology. One of his first exposures to biology came in 1961 when a friend dragged him to a lecture by Francis Crick. The experiments demonstrating the triplet codons had just been done and Steve recalled that the lecture was "mesmerizing".

In that same year, his first year at Harvard, Steve learned more about DNA from attending lectures by Jim Watson. He recalled:

Steve cited the lectures by Crick and by Watson and reading in Scientific American about Seymour Benzer's work on the genetics of the rII locus in the T4 bacteriophage as the factors that convinced him to explore the field of molecular biology and biochemistry. He spent the summer of 1962 in the laboratory of Charlie Thomas at Johns Hopkins Medical School. This was at a time when the colinearity of DNA and the genome was still an issue. The T-even bacteriophage have a circular genetic map, but a linear piece of DNA. Thomas and his colleagues were able to demonstrate that the order of the pieces in the DNA of these phage was circularly permuted. Steve's job was to make the agar gels for the experiments. Members of the lab at the time included Ken Berns, John Abelson and Fred Blattner.

Steve's introduction to protein structure came in 1962 when Max Perutz was invited to Harvard Medical School to be the Dunham Lecturer. Steve recalled:

Steve's interest in structural virology began when he started graduate school in the Fall of 1963 and that's when the oral history begins. I asked him what he had decided to do after graduation from Harvard and he answered:

SCH I had accepted a slot in the biophysics graduate program at Harvard and by the middle of the Fall, I decided I wanted to get away for a year, partly because I didn't just want to sign up in somebody’s lab and be told what to do. I wanted to figure out what I wanted to do and I had a feeling that that was going to take more than the rest of the year. So I had been assigned to Elkan Blout as a first year advisor. He suggested I try to go to Aaron Klug whom I had never heard of because he wasn't terribly famous yet. So I wrote to Klug and obviously - now that I know how these things work - Blout wrote to Klug on the side and I was quite amazed to discover that Klug took me. So I finished up my first year and went off to England. But in the Spring of that first year, Don Caspar and Carolyn Cohen gave a course on what we would now call structural cell biology, I suppose, which then was just viruses and muscle - that’s about all you could do big scale structure on. That pretty well determined that I wanted to work somehow on structure on a larger scale than an individual protein.

I got very excited, both about muscle and about viruses. I wrote a paper on actin assembly because I read Don Caspar’s review on TMV assembly and tried to apply some of the ideas in his TMV article to thinking about actin assembly. You can imagine that one knew nothing about these things. TMV subunits were just loaves as in the original Don Caspar drawing based on Rosalind Franklin’s stuff, but he had been trying to work out what possible assembly pathways might be for a helix like that - what might nucleate things - and he looked at all the detailed physical chemistry from Lauffer and all sorts of people and was trying to reconcile that with the structure. In retrospect somewhat unsuccessfully because Klug then much later worked out the role of the disc in nucleation and so on. Don still debates some of that also, so in a funny way, it’s never been properly sorted out.

SS So Don talked mostly about TMV and not about the spherical viruses? This was in 1963 which was just after the 1962 Cold Spring Harbor Symposium (where he and Klug published their article on quasi-equivalence and icosahedral viruses).

SCH I’m sure he talked about spherical viruses but I don’t remember. The stuff I remember in the course especially was Carolyn Cohen’s muscle stuff, actually, but there surely must have been spherical virus things. I just don’t remember it precisely because by then I had already read spherical virus things to see what Aaron Klug did. When Blout suggested I go to Aaron Klug then I went to the library and read some of his papers. That would have been my first exposure to icosahedral virus which would have been in the Fall and by the time I took that course I might have read the relevant stuff. So it wouldn't have been new. I remember going by - one thing was funny- I went by to see Don Caspar about something. Don Caspar, Carolyn Cohen, and Susan Lowey all shared a lab on the sixth floor of the Jimmy Fund building. I had gotten to know Carolyn from the course and when I was coming to see Don, she and Susan conspired that after 2 and 1/2 hours they would come rescue me. Indeed, just as they had predicted, Don had talked nonstop and I had said nothing for 2 and 1/2 hours.

I was trying to think if during that year there were any other striking influences. I had done sort of lab rotations. One didn't have lab rotations then but I sort of made up my own rotation as a research course for credit or something like that, that Elkan Blout encouraged me to do. I did one term with Jared Diamond on water transport in the gall bladder and I think my first paper - no my first paper was out of the second rotation which was in Elkan’s lab on refolding of apomyoglobin. A paper came out of the first term as well on streaming potentials in the gall bladder. I never went back to that. There were several things I realized in the Fall of my senior year or in my first year as a graduate student and they all had to do with the fact that I needed to see what something looked like. I really needed to do structure. One was when I was trying to think about membranes somehow thinking about things like a black box even though you were writing formal physics equations - and it was all very quantitative and all very exact - I just didn't know how to think about it. I realized that unless I knew what a membrane looked like, and that was clearly a long way off, then I wasn't going to be able to make headway with that kind of research so I better not try to do it.

The other area was the sort of golden age molecular biology which friends of mine in college had gone off to do. But again, somehow the classical molecular biology paradigm which was fantastically powerful and very appealing - namely: We had model a, model b and model c and we designed elegant genetic experiments, preferably in lambda and preferably taking no more than a week, to distinguish among those models and the experiment says one of them is right. I was no good at that somehow. It was too non-chemical a reasoning if you wish. So I figured I had to leave that to the Botsteins and the Ptashnes of the world and do something more 3-dimensional and from the Caspar-Cohen course I pretty well decided - I remember telling Carolyn that I wanted to figure out how to do structure on this larger scale that they were struggling to do in their labs. So (in 1964) off I went to Klug who wanted me to do physical-chemical experiments on turnip crinkle virus reassembly and on the assembly pathway of a spherical virus. Initially I was working directly with Reuben Leberman who was a biochemist. I was doing ultracentrifuge experiments with the Model E trying to figure out how TCV (turnip crinkle virus) might assemble and what intermediates in assembly might be.

But I quickly decided I really wanted to do crystallography and so I went to Aaron and said I really wanted to do crystallography. He said well, if you are going to do crystallography then you better read fundamental crystallography. So he gave me the dullest possible book, I think to discourage me. It was an old fashioned book that didn't have an x-ray in it. It was just about interfacial angles and space groups and derived every space group - painfully. It’s a book by Phillips called Introductory Crystallography. I came back a couple of weeks later and said I've read this, what do I do next? And so then I think he realized I was probably serious so I started working under the tutelage of a guy named Bill Longley who had just gotten his thesis with Aaron and was about to go off to Don Caspar as a postdoc.

The problem then was to see whether we could get reasonable X-ray pictures from some rather nice crystals of turnip crinkle mosaic virus that Reuben Leberman had grown. Bill was an interesting guy. He started life as a machinist and in England, if you remember, in those days especially if you started off in a non-university track it was almost impossible to escape from it. Birkbeck College - where Bernal had been - was the one escape. It was a night school actually, sort of like Northeastern or CCNY, but in a different way. The courses were mostly in the evenings or at night so that people who had a full time job could nonetheless earn a University degree. So it was one of the few places where you could escape from how you had been tracked since the age of 11. Bill was a working class guy who had nonetheless gotten a degree and he taught me the importance of machining and things like that. In the X-ray crystallography of those days, you had to be able to fix the X-ray set. You had to be able to build components of a camera and construct a collimator and so forth. So I learned about machine shop stuff from Bill that I didn't know anything about at all. He then went off to Don Caspar sometime in the Spring and I continued to try to get data from these crystals of turnip crinkle virus and work out what the space group was and other things that seem trivial now but were fairly painful at the time. The main thing I found was that the mercury derivatives that we tried to make were actually much better than the native crystals. The native crystals diffracted wonderfully but if you looked at them, they had an inherent packing disorder. I learned to think about packing disorders because I sorted that out. But the mercury derivative was basically the undisordered version of it. Later that came in handy when we actually determined the structure of TCV using bushy stunt as a model and that was the form we used.

SCH He and Don Caspar had collaborated as you recall in 1962 to work out the rules of icosahedral triangulation and so on. And they had interacted well before that. Certainly Don’s paper in 1956, which was the first precession photograph of a virus crystal, showed that bushy stunt was icosahedrally symmetric. That paper was published in Nature (177: 475-476, 1956) and it showed the little spikes of intensity were along the directions of an icosahedral axis. Then in the late '50's John Finch had brought polio virus crystals to England and taken pictures of those - before Rosalind Franklin died. So this must have been between '56 and '58 so that there was the notion that it would be important to try to figure out how to get decent X-ray data from virus crystals. If you think about it myoglobin was getting solved at that time; isomorphous replacement on hemoglobin is 1953, and Kendrew quickly adapted that to myoglobin. By '59 Kendrew had the famous 6 angstrom intestinal-looking model of myoglobin. And at the same time, the descendants of Bernal, that is Rosalind Franklin and then Aaron Klug and John Finch, were trying to figure out how to record X-ray data from these huge unit cell crystals of plant viruses and animal viruses, polio for example.

So when I got serious about trying to work on virus structure when I was at Cambridge, I asked Aaron about it. I thought about staying in Aaron Klug's lab but decided to go back to Harvard and work with Don Caspar instead - I'll come to that in a minute - partly because I had to make the decision early and most Americans nearly had nervous breakdowns from "miserableness" and loneliness over the Christmas holidays in England, because the English in those days were notoriously unfriendly. You could drop dead on the streets in Cambridge on Christmas eve and you wouldn't have been rescued until the day after Boxing Day. But in thinking through all that, I asked Klug what problem might be an appropriate one and he said that bushy stunt and turnip crinkle viruses had been the ones that he and Don Caspar had pretty well settled on as the most likely to succeed and there was kind of an informal agreement that Don would have bushy stunt and Aaron, TCV. Whether that was just Aaron’s view of that I just don’t know but at any rate that’s what he told me. So when I decided to go back to Boston I wrote to Don and said that I would like to work on virus structure and could I determine the structure of bushy stunt for my Ph.D. thesis. He wrote back and said that it was a very important problem and sure I could come do that and he didn't add that "you're crazy" but anyway that’s how I came to settle on bushy stunt virus.

In a way Don and Aaron made the choice partly because they also worked on several other crystals of plant viruses. Turnip yellow mosaic virus was one of the early important ones but those crystals were crystallographically much more challenging than the crystals of bushy stunt and turnip crinkle virus which were simpler packings. TYMV crystallized in a way in which the organization of the particles in the unit cell made it a much more challenging crystallographic problem because you got in effect a much larger unit cell and hence the spots were closer together and there were issues of potential disorder in that packing as well. In retrospect bushy stunt was the right one, but I didn't know that. All I knew was that Aaron and Don thought that TCV and TBSV were probably the best cases. TBSV, Don had already worked on in '56 when he did the first precession photos.

That year in Cambridge was the year that (David) Phillips, in London, solved lysozyme which was the second protein to be determined at high resolution, I believe. Perutz had built a model of hemoglobin that he actually talked about in his Dunham lecture based on the myoglobin model - the first example of homology modeling because it would be several more years - hemoglobin being a tetramer - it would be several more years before the data collection and computation would reveal an actual experimental model for hemoglobin. But Perutz being impatient had built a model of hemoglobin based on his low resolution 6 angstrom model: the famous balsam wood model with the black and white alpha, beta dimers fitting into each other. It was based on myoglobin, and it was largely right, of course, so that the first homology modeling was actually done painfully over 8 months I think or something like that by Perutz himself in the model room with the Kendrew models. They weren't old then, they were brand new with the Kendrew models in a wire frame. I remember seeing that model and Max still working on it when I was there in Cambridge in '64 to '65. At any rate, I think that the next protein determined at high resolution was lysozyme. The only competitor that I can think of would be carboxypeptidase that was in Lipscomb’s lab - but I think that was later. Yes, it had to be because lysozyme was the first enzyme and carboxypeptidase is also an enzyme. David Phillips had come up and talked about their progress in December or so, I don’t remember exactly. He had come up to Cambridge and brought with him his whole team, so there were four talks including one by a graduate student, Louise Johnson. She was trying to look at a crystal structure of a complex. The talk was called "Two Steps Forward and One Step Back" because they had hoped to solve it but hadn't yet due to some crystallographic nuisance. By April they overcame that and built a model.

David Phillips gave a kind of preliminary unveiling in a morning in April, I don’t remember exactly, at the Royal Institution, where they were. Bragg was head of the Royal Institution, the old Faraday Museum in London. I was about to go - it may even have been the end of March - I was about to go to Italy for my Spring holiday. Everyone from the lab, I remember, packed into the lab van. Max threatened to drive, but somebody fortunately stopped that, so the lab driver drove, but lots of people packed into the lab van. I remember having my bags to go to Italy - sort of crouching in the platform in the rear of the van on top of the bags. Paul Sigler was there, he was working on chymotrypsin with David Blow. I’m sure that other people went by more comfortable means of transportation. By some time in the morning we got to the Royal Institution. There was a lecture in which David Phillips showed the model. I do remember that afterwards you could go and look at the Kendrew model in the frame and stare at all the things.

SCH No, this was lysozyme. The Kendrew model means the model, made of Kendrew metal parts that screwed together, that you remember pictures of. You've seen pictures of them in books. They were called Kendrew models because he and Herman Watson had devised them. They were wire models soldered together so that each atom was represented by its bonds, so that carbon was a little skeleton of 4 pieces of wire soldered together in a tetrahedral array sticking out from the center so that the ends of each bond had a little grove machined into them, so that you could join it to another one with a little sleeve that had tiny screws in it, and you tightened the screws into the groove, and that kept the thing from coming apart. You can imagine building such a thing which is a horror and your fingers got bloody, and it took forever.

That (the lysozyme model) was pretty impressive, and so I figured that proteins had been solved and enzymes had been solved, so one definitely had to do viruses - the protein problem was over! [This statement was followed by appropriate laughter.] By then I was already pretty well committed to determine the structure of something big and the only big things you could crystallize - really big interesting assemblies - were viruses. The only viruses where you could get an adequate quantity and adequate purity at that time to have a serious go at it were the plant viruses, since you needed many hundred of milligrams or probably grams of any protein or virus or whatever at that time to hope to go through all of the trials with the heavy atom derivatives over the period of years that you would need. Now you can solve structures with 10 milligrams or less if you are really lucky. But you definitely needed to get hundreds of milligram to gram quantities trivially, and plant viruses were about the only thing available.

SCH No, actually I had to figure out how to do that. Don had some, so when I got back in the Fall of '65, where we are now, there was some that he had had in the lab for awhile, but it still crystallized. Bushy stunt is very stable. I have some that we grew about 20 years ago that I probably could still crystallize. So I grew crystals of that; they took a long time to grow. Bushy stunt took 3-4 months for the crystals to even begin to appear so it was a little unsettling, but I later learned you could seed and get the crystals rather quicker.

SS Did Harvard have some kind of committee that you had to meet to approve your thesis?

SCH No, that was in the days well before thesis committees. The first thing that Don thought I ought to do was to construct an X-ray camera that would allow us to record decent data. The way you collimated an X-ray beam then, and still do at some synchrotrons after a monochrometer, is with a collimator - a multi pin hole collimator that gave you a well collimated beam. That had two problems - first it threw away a lot of the intensity and second it still wasn't a terribly collimated beam, so there still was an enormous amount of diffuse scattering. While that wasn't a problem for chymotrypsin, it was a hell of a problem for turnip crinkle virus.

Don suggested we try to use focusing optics as people were doing already for looking at muscle diffraction. Hugh Huxley had constructed such a camera - or Ken Holmes, I think, had constructed it, with Hugh, to do muscle diffraction - and they were already using it when I was in Cambridge. In fact, it was right in the same little room with the X-ray set that I was using to do the virus diffraction, so I was pretty familiar with this kind of X-ray optics, and Bill Longley was by then in Don’s lab and he also knew how to do this kind of thing. So that led to my trying to use a double mirror focusing camera in order to record better data. It worked fantastically and gave dramatically better pictures and was, if you wish, the first key step in figuring out how to determine a virus structure, because until synchrotrons came along, that was the sort of optical arrangement we used with the more powerful X-ray sets that then came along. And characteristic of Don, when I went to publish this 3 years later, he didn't put his name on the paper even though I really do think he had suggested the idea. I had done the whole thing and the particular design I had come up with. I pirated the parts from a monochrometer mounting in order to be able to mount the camera.

SCH Yes, I was the only person. Bill Longley was there and was sort of helping me but he, I can’t remember exactly what he was working on, catalase I think. Or no, maybe he was working on muscle things with Carolyn Cohen. Yes he was actually working with Carolyn, but he helped me with the machinery end of stuff. But there wasn't anybody else working on viruses. So step one was to figure out how to get better data than you could get with a collimator, and in the end Don’s advice was critical.

SS So you really had nobody else that you knew was working on viruses except the people back in England?

SCH Right, it didn't seem to matter to me, and it didn't much bother me. And there weren't any other graduate students on the floor either. There were a handful of postdocs - it wasn't a very big lab. Susan Lowey had 3 or 4 people working with her and Alan Weeds came through as either a postdoc on a sort of post-postdoc, I can't remember. Carolyn had a couple of people working with her. The office I had had exactly the same view as my Enders office now has, which is also on the sixth floor, about a block down. The other thing that one had to do then of course was write all one's own computer programs.

SCH No we didn't have a computer- the computer we used was the Harvard computer. And so we got an account at the Harvard computer center which was in Cambridge in the old Aiken lab that just got torn down. At the time they had an IBM, a 7090, and you programmed it with punch cards. I was living in Lowell House as a resident tutor when I was a graduate student. I didn't have to pay rent, and I therefore used that part of my graduate stipend to run a car that I bought in Cambridge, a little MG. So I had a car to go back and forth in, but still I would go in the evening or whatever, and do my computing in the evening, and you could get a couple of turnarounds a day.

SCH I had learned Fortran from a course that the Chemistry Dept gave when I was a junior, just a few lectures in the afternoon, attendant on buying their first computer, which was a 1620. That was so small a computer that it couldn't hold the FORTRAN compiler and your program at the same time. So you fed the FORTRAN compiler in as a program and read your program in as the data for it. This was reading a stack of cards. The FORTRAN compiler was about an 8 inch high stack of cards, I remember, and then it compiled your program, and if you had not made any errors you got an object deck out, but if you had made errors, then it told you that you made an error and you had to correct the error and go through the whole process again, recompiling the program. At any rate, it was open to anybody in the Department who wanted it, and even though I was just a junior, I could sign up for a little bit of time, and I taught myself how to program that way. I wrote a program to do the computations in some lab exercise that we did in physical chemistry. Then in England, in order to process the data I was trying to collect, I wrote a program under Ken Holmes’ instruction adapting a program he had written from TMV. So my first real serious programming was a program I wrote in England to process precession data from virus crystals using something called Fox-Holmes scaling which became the standard way of scaling crystallographic data, basically to this day.

There you only got one turnaround a day, because in England, at Cambridge, they didn't have a computer, and the computations were done in London at the IBM Center or at Imperial College, it switched some time in the middle of the year. One of the so-called "computer girls" would take the 3:36 (PM) train to London, I remember, so that by 3:10 you had to have your cards all punched and set up in the box that she would grab and hop in a cab to get to the railroad station which was only 10 minutes away and take off for London, and then she would come back on some evening train with the output.

SCH If there was a mistake you did it again. You got literally one chance a day and that was still considered pretty nifty because the computer in London was quite powerful. So the computing when I was a student in Don’s lab went on in Cambridge, Mass and I was in Boston Mass but that was much more proximate and easier to use than in England, and I could get a bunch of turnarounds if I just didn't come to the lab (in Boston) but sat in the Computer Center for awhile. So that worked out pretty well. So my thesis basically was about how to get decent data. I think it’s fair to say that I showed that with focusing optics you could get data good enough to go on to do something sensible. And it was an effort to determine a very low resolution structure using solvent density differences that I can explain in a minute if you want. In retrospect that was probably - I've never gone back to see how right that was - but in retrospect it may not have been very right. But it was a perfectly valiant effort and in any case I saw my thesis as a step on the route to something else.

SCH Well, we knew what it looked like in the EM. Since the solvent differences were basically an attempt to see an outline of it in a way, I was using possible solutions that looked like the EM as a guide. So in any objective sense I probably had not gotten very far with the phase problem. What I had done was figure out how to use solvent density differences to get data that would represent an envelope of the structure. You can think of the differences between a crystal in a low density solvent and a crystal in a high density solvent as being an image of a negatively stained envelope of the particle. In that sense it was very useful because what happened just a little later was that we were able to use the 3D reconstruction from negatively-stained EM images of bushy stunt virus to phase the solvent density differences since they corresponded to negatively stained images, and that's one of the two things that really got us going - that's jumping ahead a little bit. This happened about 3 years later. So, at any rate, I think that I learned how to collect decent data that could solve the phase problem, but I hadn't properly figured out how to solve the phase problem in my thesis.

SS During this period of time - were you following molecular biology or protein chemistry? How did you feel your work was fitting in with what was going on in the rest of biology or were you more focused on your problems of X-ray crystallography?

SCH No I was following molecular biology a fair amount since I was at Harvard and a lot of it was happening here. But, I did have, as I remember, some very long range and patient view of things so it didn't seem to bother me, looking back, that it seemed a long way before what I was doing in the lab would connect to the exciting results that were coming out of seminars I followed. The year I was in England was the year that the genetic code was cracked, and Sydney (Brenner) and Francis (Crick) shared an office, and into that office by phone or regular mail - there were no other modes of communication as I recall - would come Khorana's latest codon, and it would go up on the blackboard.

SS That can't be right because it was 1961 that Marshall Nirenberg cracked the code.

SCH Right, but then getting all the codons came later. I remember codons coming out even in '62 or '63 so it probably was the finishing up of it and the realization of the redundancies and certainly termination had just been done in that T4 phage experiment when I arrived and I remember hearing about it at a lab meeting. So it was a late stage but the code was getting completed. By the time I came back I noted that the genetic code was done. So I was certainly following all that but somehow I don't remember it bothering me.

I was certainly following also the advances in protein X-ray crystallography which were pretty slow, that is chymotrypsin and carboxypeptidase. You could read every structure up to about '71 which was when things really began to take off. And that was marked by that '71 Cold Spring Harbor Symposium that you mentioned. So that’s in a funny way a kind of key watershed, in fact you can use that as a kind of key demarcation date. I had finished my thesis by the end of '67. Officially I got my degree in '68 and decided either I would go back to Klug’s lab or stay in Caspar’s lab if I was going to make progress on this. There was nowhere else I could really have the facilities. I decided to stay on in Don’s lab, and I got a Helen Hay Whitney Fellowship for the first year and then I got one of these Harvard Junior Fellowships.

Let’s look a little bit at the phase problem, because I said I don't think I had gotten very far. Let's go back. How do you solve the phase problem? Or how did one at the time solve the phase problem with protein structures? Perutz in 1953 had shown in principle that you could do it with a heavy atom derivative, and by ‘59, and then at high resolution at about '61 or so, Kendrew had shown in practice that it would work. But with very large structures it was not obvious that it would work for several reasons: first, the method for finding the heavy atom derivative position to begin with involved something called a Patterson function, which would be hopelessly complicated for a virus and was often too complicated even for a simple protein if you had more than 3 or 4 derivative sites per protein molecule. So your first derivative had to be a simple one that had only 1 or 2, so that the difference Patterson could be studied and interpreted by inspection. The second problem was simply getting accurate enough data that the differences would be reasonable. And third, since even if the differences were reasonable they would be less accurate than with a smaller protein, how many derivatives would you need to get in order to actually determine the phase problem that way?

Rossmann and Blow had already noticed that if you had high non-crystallographic symmetry there were substantial constraints on the phase problem. They pointed out that in principle you could even determine phases from scratch if you knew exactly where non-crystallographic symmetry operations were and you knew the envelopes of the objects in the unit cell that were governed by that non-crystallographic symmetry. Michael Rossmann spent a lot of time working on the consequences of those observations, and David Blow did also. What influenced me most were a couple of papers by Tony Crowther - they were basically Tony’s Ph.D. thesis - which recast that problem. This happened when I was a graduate student. I had met Tony who was a starting Ph.D. student the year I was in Cambridge. He succeeded in recasting the problem of the determination of phases using non-crystallographic symmetry as a problem in linear algebra and went a long way toward coming up with a formulation that could be thought of in a sensible way in terms of the computational approaches. Michael was meanwhile working away in what was essentially a parallel effort, but a little bit harder for me to understand at the time, but in fact came up with essentially all of the same conclusions. So one of the reasons I was encouraged to push ahead with trying to figure out how to get good experimental data was that I had confidence that others more mathematical than I were pushing away at figuring out how to use this powerful information to make advances in the phase problem. And I played around with that some and programmed some of the relevant computations while I was a graduate student, and they were important in these first efforts in my Ph.D. thesis to go from the solvent density differences to an effort to think about a map.

Meanwhile electron microscopy had come a long way and the year I was in England a guy named Jack Berger was working as a postdoc for Aaron Klug commuting to Manchester, I think it was, where there was an optical diffractometer that he was using to analyze electron micrographs of phage tails, as I recall. At any rate analyzing helical repeating elements in order to try to turn the analysis of electron micrographs from something that was utterly qualitative and subjective into something quantitative and rather akin to the way one looked at an X-ray crystallographic pattern. This was ultimately the direction that helped win Aaron (Klug) his Nobel prize. By the late '60’s Tony Crowther had joined Klug as a postdoc to do all of this computationally and that culminated by 1971 - at the time of the Cold Spring Harbor Meeting - in the first 3-dimensional image reconstruction from electron micrographs of negatively-stained particles. In this case it was indeed bushy stunt virus. Tony actually stopped in Boston en route to Cold Spring Harbor, and we compared notes and started planning how we would actually put the information from the two together. harrisonCSH71.jpg (8869 bytes)

It took about 2 more years for that actually to come to pass, but it was already clear to me that that was the best quantitative way to put things together. Qualitatively, I had already put them together in the following way. What I did between about '68 and '71 was to try to figure out how to solve a heavy atom derivative, and since it took several days to get even one X-ray photograph, one had to devise some kind of heuristic method for screening a derivative, since normally you would have had to take several pictures and so on in order to see whether even a given derivative effort had ever have successfully produced changes. So my criterion ultimately was that under similar conditions of soaking crystals you could get reproducible differences, and you just looked. Evaluating the data was painful - I’ll go into that in a minute - so that the criterion had to be that you could see reproducible difference from week to week, as it took a week to do an experiment.

SCH On the X-ray pattern itself. With crystals of very large unit cells the spots on the X-ray pattern are extremely close together so in these focusing optics that I had developed which did give a wonderfully improved signal to noise-the spots were very tiny. The scanning densitometers of the time when I was a graduate student were relatively mechanical one-dimensional scanners - the Joyce-Lobel scanners that were also used for other purposes and involved a servomoter linkage between the transparency being scanned and a pen tracing out peaks on a piece of graph paper. And that was simply not sensitive enough and not mechanically fine enough for the virus patterns. So I had to evaluate all the data by eye using a comparison set of spots since the eye is a very good matching device. That had been used classically for small molecule crystallography as a way of estimating spot intensities, and probably as a graduate student I measured 10,000 intensities simply by comparing spot by spot the spots with a reference set of spots that had been exposed actually on an ultracentrifuge plate for time x, 2x, 3x, 4x, 6x, 8x and so on, so that one had a calibrated scale. That was intensely painful as you can imagine, therefore you had to have all sorts of heuristic pattern recognition methods for figuring out where you were getting with the search for a heavy atom derivative. By 1971 I was pretty sure that a platinum derivative that I had identified was going to work and had a tentative solution for it. I think I did, I will have to go back and look at the Cold Spring Harbor paper. We should do that, because that sort of summarizes basically where one had gotten with crystallography, in parallel with Tony’s paper, which was just in front of mine and which showed where one had come with electron microscopy. It's fair to say that up to that point they were pretty well tied, in the sense that crystallography had not gotten beyond about 25 to 30 angstroms, which is exactly where electron microscopy had. But of course what then happened was that crystallography could rapidly move ahead, because we knew we could get data out to 3 angstroms whereas it has taken until about now for electron microscopy to progress to beyond about 15 to 20 angstroms. But it was clear that one wanted to use the electron microscopy to get a jump-start on the phase problem and figure out some way of transferring the information about the image and hence about the phases to the X-ray data in order to be able to find the heavy atoms. As I said the difference Patterson methods were unlikely to work easily - I'll get to that in a minute - and therefore it was clear that if you could use the electron microscopy results to get X-ray phases and then use those to find the heavy atoms even at very low resolution, then you could perhaps refine the heavy atom positions in a boot-strapping way to higher and higher resolution and so move forward. I also played around with another method, as a graduate student, for finding heavy atom derivatives, though that wasn't what I was using it for at the time. And it is indeed the standard method for finding heavy atom derivatives until recently, when we've gone back to the old EM approach. So in a funny way what I'm saying is that the phase problem in bushy stunt got solved by a kind of complex tinkering, but the easiest summary is that we got phases from EM, found heavy atoms with them, and moved forward

SCH Still no one. We'll get to some virology in a minute as well. So another way of trying to find the heavy atom derivatives would be to use a more automated method to solve a difference Patterson function rather than the inspectional methods that were being standardly used. Computing was getting powerful enough that you could imagine programming a systematic inspection of the 3-dimensional difference Patterson function using the criterion that any solution had to have icosahedral symmetry. Using the icosahedral symmetry not just to help you solve the phase problem, as I had said Rossmann and Blow had seen you could do, had encouraged me to push ahead figuring out how to get data that you could use on it because the theory and computation were slowly emerging, but also to find the heavy atom coordinates to begin with. That's called Patterson search methods in the lingo and I had learned about them actually from an old paper by Berger, I think, and by their use in a paper on a disaccharide structure, I think it was. Anyhow a mono or disaccharide structure - from Lipscomb’s lab, and then Michael Rossmann had thought about the very same thing in some hemoglobin context, and there was a paper that I remember of his in which he diagrammed searching a Patterson from horse hemoglobin by looking for the colored hooves of the horse to be the heavy atom sites. It was in a book that I can’t recover. He thought I might be the only person who remembered the horse's hooves. Ultimately there is a paper by Argos and Rossmann that presents computations using that method, which is how they solved the heavy atom coordinates for southern bean mosaic virus. And I've got it in my thesis, but I never published that approach, because I never really used it successfully for something.

So if you wish, there were kind of two parallel things going on in the general problem of how you determined a complex structure like this. One is how you used non-crystallographic symmetry effectively to solve the phase problem: do you have to find heavy atom derivatives or, once you had starting phases, to improve those phases or whatever, and on the other hand how do you get data that were good enough to determine the structure in the first place. And I think it's fair to say that I was worrying especially about the data problem and Michael was spending a lot of time worrying about the phase problem. And then - just to jump ahead for a minute - what finally happened is that an approach to the phase-problem computations quite different from what Tony Crowther and Michael (Rossmann) had been thinking about came from another graduate student of David Blow named Gerard Bricogne who realized that by doing the computation in a totally different way - the theory was the same if you wish but there was an algorithm that actually was computationally feasible as opposed to what looked like the straightforward algorithms, which were computationally and still are computationally unfeasible. And that broke open the ability to finish the problem. We’ll get back to that much later.

I got to know David Baltimore somewhere around this time. He had come to MIT. I first saw him, although I didn't really meet him, at one of the Vietnam war protest rallies. MIT had a sort of week of antiwar activities in probably '69 - the March-something movement and there was a day of little workshops of breakout groups or something that were publicized enough, and somehow the one that David was running was on some issue that seemed interesting. I don't recall in detail- but I went to that and that was my first exposure to him. Then at some point I approached him about trying to repeat the poliovirus crystallization and getting ahead with poliovirus even though we were hardly ahead with bushy stunt virus and I did spend a couple of afternoons in his lab actually setting up crystals of polio virus. I remember vividly that they were handling polio virus sort of the way I handled bushy stunt virus. There didn't seem to be a biohazard worry at the time and we set up some crystallizations that didn't succeed - in David Rekosh's fridge I remember. As a result, much later when Jim Hogle wanted to work on it, I fixed him up with David, who in turn fixed him up with Marie Chow, and that was actually the fruition. I had gone to David and said: "Look why don't we also start going forward on polio since now we know how to take X-ray patterns from such things" and he agreed. But it took some 15 years or so for that really to get anywhere. So there were things like that that kept me aware of things that were happening in virology. I gave a couple of lectures on virus structure in a course on virology that David gave at MIT.

SS You gave a lecture in our course (at Washington University) in Virology - probably in the very early '70s.

SCH I remember I came to St. Louis. That was right after I got back from Germany. That would be the fall of '72. And I sat through the (virology) course whether it was the year before I actually gave a lecture or the same year. I think it was the same year and I went to many of the lectures. That’s where I first learned about the emerging knowledge of strategies of viral replication.

SCH No in the late 60s, before Heidelberg - may have been 1970 - after the March 5 movement or whatever that was called- May Fifth or whatever the date was. This was David’s (Baltimore) course at MIT. It probably was something like either '69 or '70 or '71 because I remember lecturing in it twice, I think, and sitting through it at least once. And during that time reverse transcriptase was discovered in his lab. I remember the first year I went to the course there was the puzzle of why retroviruses should be sensitive to actinomycin D, and I certainly was aware of the excitement when they had found the solution to the negative strand RNA virus problem that there was an enzyme in VSV, but I remember not being shocked at the reverse transcriptase result - that is somehow I was prepared for it by the lectures - by having taken the course. It was obviously on David’s mind that that might be a solution, not that it was in the virion but that you would have to have such an enzyme. So when they found it, I remember thinking it was quite wonderful but not thinking of it as some total upsetting of my world view, but I think that was because I had been a bit shaped by the way David was approaching the problem.

SS I'm not sure virologists were so shocked by it because by that time both Pox viruses and VSV had enzymes in them. It explained everything so clearly so it was very easy to accept it - except for a few people.

SCH It's from that period that I began to really start thinking that virology as such was interesting. When I went to work on bushy stunt virus, it was because I wanted to work on the problem of macromolecular assembly and there wasn't anything else you could work on. It wasn't because it was a virus, but because it was an example of large scale macromolecular architecture and had I been able to work on actin - had somebody said I could also work on actin and that it was accessible - it was not - I imagine I would have picked actin.

SS And Don (Caspar) must also have been interested, but not because they were viruses.

SCH That's right, and he continued to be interested in things that way, examples of the physics and physical chemistry of assembly rather than as examples of viral infectivity. But the thing that really got me interested in virology was actually - I'm jumping ahead again - was the first year I taught at Harvard which was '72 to '73, Don Wiley and I gave a course called Introduction to Structural Biology and just by arrangement - initially a convenience of scheduling - it was back to back with a course that Alice Huang, who had just joined the Harvard Faculty, gave in Cambridge on general virology, and so I sat - and I think Don did too - through her lectures. It was just a Harvard version of the course that David had given at MIT two years before, organized by positive strand RNA viruses, negative strand RNA viruses - the now traditional categorization in terms of replication strategy, which was of course a quite excitingly novel way of looking at viruses rather than by disease. That's I think where I really learned that virology posed both exciting problems and how interesting it was. She gave a memorable - to me - lecture on what were then very new experiments from Dan Nathan's lab on restriction fragment mapping - probably only 6 months old but it was certainly my first awareness of it and that's where I first learned about subgenomic RNA in Sindbis virus and several other things so that obviously made a big impression on me and made me realize that it might be fun to work on viruses not just as examples of something else but for their own sake. Let's see, that' jumping a bit ahead

SS We're back to tomato bushy stunt and I guess the evolution of getting a structure that was much more detailed than the electron microscope.

SCH. As I said, '71 and the Cold Spring Harbor Symposium are a good summary of that point. There’s an interesting aside about that. Michael Rossmann had spent a sabbatical in Uppsala where there was a group under the direction of Bror Strandberg trying to solve satellite tobacco necrosis virus. For some complicated reasons a rotation function calculation or set of them that they had carried out suggested to them that STNV might be octahedral in symmetry rather than icosahedral. Caspar and Klug had pointed out as had Crick and Watson before them that any of the three symmetries - tetrahedral, octahedral or icosahedral - that characterize the closed shell could in principle characterize such a structure, but that the icosahedral was likely to be the most efficient because you got the "biggest bang for your buck", so to speak, that is the largest number of equivalent subunits and hence the greatest genetic economy, and since the analysis of electron micrographs that Aaron (Klug) had then been carrying out in order to try to confirm the outlines of that theory had been so successful and all the structures he had analyzed had been icosahedral, it became increasingly likely in most people’s minds that they would all be icosahedral in symmetry and that the efficiency argument was indeed a powerful one and evolution had respected it essentially universally. Nonetheless it was perfectly possible and so the Uppsala claim had the semblance of a sort of minor shocker and also they came along with it that way. I remember I came a day late, and Bror Strandberg, I think it was, came running up to me and said "how do you know that bushy stunt virus is icosahedral?". I said: "well, there was the analysis of the intensity spikes in 1956 that was clear, and the 3-dimensional reconstruction from electron micrographs", which I knew all about - and indeed Tony had just visited - "demonstrated it", and so on. So it was clear they had some little bombshell they wanted to burst. They were trying to be rather coy about it, but sometime before the session there was finally a discussion about it at the beach at Cold Spring Harbor, and Aaron was convinced it was wrong, but didn't quite see the solution to it. So there was a famous confrontation at the session itself, where Strandberg showed the slide of the rotation function calculation, and Aaron sitting in the front row called out - as Strandberg was about to go on to the next slide - "Wait a minute, you are about to come up with a clearly erroneous interpretation and the reason for it is in that slide, but I need one more minute to figure it out". And of course that antagonized the entire audience, who sided with Strandberg. Poor Strandberg didn't know what to do so he just left the slide there and was standing speechless at the podium - quite an incredible scene. And while I didn't find the manner particularly forgivable, I knew damn well Aaron was right. I was the next talk as well and found it quite an awkward situation. At any rate, it took Aaron a few more days to figure it out, but he did, and there’s a discussion about it in the symposium book.

I wasn't feeling threatened by the microscopy; on the contrary, I was excited about it and realized how much it was going to help. Because I knew it was going to get stuck at the negative stained level and I had recorded diffraction patterns that went on out to 3 angstroms or beyond so it was a matter of now using it to help me to push forward as fast as I could. By that time, Caspar and Cohen had thought about moving to the Rockefeller, where Gerry Edelman was spearheading an effort to recruit a major structural biology group to Rockefeller. It was about to happen, and I was going to go along as the Assistant Professor, but Gladys Caspar put her foot down and said she wouldn't move to New York, and so that got scotched. I was offered a job at Harvard, and I was too lazy to think otherwise, so I just took it. Stanford tried to get me out to give a seminar, but I didn't want to travel to California. So I had taken a job at Harvard, but I wanted a little break from living in Boston, and so Ken Holmes had invited me to come to the Institute or Division that he had started at the Max Planck Institute at Heidelberg about 3 years before that. So I decided to do that and Harvard gave me leave for a year. One other important event that happened - uncorrelated as far as I knew - was that Harvard hired Don Wiley a few months after they hired me.

SCH Harvard was the Department of Biochemistry and Molecular Biology which was a brand new Department that had been formed as a secession of chemists - Konrad Bloch, Paul Doty and one or two others and biologists, John Edsall, Matt Meselson, Jim Watson and again one or two others. One of their very first hires was Mark Ptashne - who was hired straight away as a full Professor from his junior fellowship.

Don Wiley and I collaborated as graduate students. He came to Harvard to work with Don Caspar originally. He was an undergraduate at Tufts and came to the Harvard Biophysics Program attracted by Don Caspar. As an undergraduate, he worked with the guy at Tufts who won the Nobel prize for computer tomography as I remember and (Don Wiley) was doing electron diffraction on phage heads as a senior honors project or whatever it was. At any rate, he spent about a year in Don’s lab but then got wooed away by Lipscomb to work on aspartate transcarbamylase. He actually showed up at the Cold Spring Harbor meeting also, with a low resolution structure of ATC (aspartate transcarbamylase) which was quite important. At any rate, by good fortune this new Department at Harvard hired me and Don.

SCH Jim Watson and Bill Lipscomb, probably. Lipscomb was in chemistry and so not a member of the Department. Watson was obviously the person driving it.

I was dilly dallying accepting the job because I just didn't want to face "growing up". One day Watson came by. I had invited him to an undergraduate meeting at Lowell House and he turned to me and said: "You’re going to take this job aren't you?" And I said to myself: "I guess I am, after that". And so I did. I was hired explicitly to be the head tutor in biochemical sciences as well. I had been a Harvard undergraduate and had this experience as a resident tutor.

SCH I wasn't an undergraduate at Harvard in biochemistry. I was a chemistry and physics major and we didn't have tutors. A man named Karl Strauch was my advisor, but he just signed my study card once a term. I came to know him a trace better later, because he was head of the Cambridge electron accelerator and in the synchrotron context but didn't get to know him all that well. He may not remember that he used to sign my study card.

Don (Wiley) and I had collaborated, when he was a graduate student and I was probably a postdoc by then, writing programs. It was a very informal indirect collaboration, writing programs for the then brand new rotary drum film scanner, which I realized immediately would replace my eyeball in getting accurate data. The first one was actually built at MIT by Cy Levinthal and he almost certainly pioneered that, at least in crystallography. Don and I knew each other when he had been through Don (Caspar's) lab. So it was great that he got hired also, and I remember he came by and we sat in my room in Lowell House and decided we ought to have the labs smoothly shared. Our role model for that had been Susan Lowey, Carolyn Cohen and Don Caspar at Children's where their labs were completely shared as far as technical facilities and all the equipment. One saw it as one smoothly continuous group. So, that was sort of the history of how we got set up.

I went off to Heidelberg for a year having written my grant with Don Caspar's help, and I remember he told me that the Cancer Institute had a lot of money so I should figure out how I should direct my grant there. You underlined 10 key words in the abstract in those days and some guy sat in some division and assigned the grants to one of the Institutes based on this and so - carefully clued in - I wrote "growth control" and "cancer" and other things as the reasons one wanted to understand macromolecular assembly, and my grant became CA 13202, which was a pretty early number still and was nicely funded, so it bought me an X-ray set that was on the loading dock when I got home from Heidelberg a year later. Needless to say, in those days Universities gave no setup money, none at all, nothing from Harvard. They gave us the first floor of Gibbs, an old building where Lipscomb was and Don had worked in, and there was space that Kistiakowsky had had, but he had just retired and that had been used during WWII to make Napalm, so it had a pretty grim and unpleasant history. We got the first floor and put a cold room in and repainted the place. Don had the difficult task of actually supervising these renovations and trying to get something started while I was half way off to Heidelberg, and when I got back I found him slightly demoralized as a result of all of that. Also he hadn't really figured out what he was going to work on. He was straight out of graduate school, so he didn't have time to think very much about what to do next. Most of his first year was actually spent finishing his own contribution to ATCase (aspartate transcarbamylase) and Lipscomb didn't really have any talented successors to Don so he was kind of quasi-directing the ATCase project for which he was getting negative academic credit from BMB (Biochemistry, Molecular Biology), as you can imagine. By the time I got back - he originally thought he might try to work on gene regulation - to work on transcription factors which were then - had just been isolated by Gilbert and Ptashne across the street. And he knew that group very well as a graduate student. I think he actually wrote a grant on CAP, which wasn't funded but meanwhile he had decided to try to work on - or by the time we had sat through Alice's (Alice Huang) course - I don’t quite remember whether it happened before or after - I think before - he decided to try to work on viral glycoproteins. The first thought was to work on VSV, and indeed he was taking small angle pictures of VSV, which we were trying to interpret in terms of cylindrical symmetry and so on. So he was sort of struggling ahead in that direction, when John Skehel's publication of little crystals of hemagglutinin, which in retrospect was certainly neuraminidase, appeared and Don immediately arranged to go and spend a term in John’s lab. The rest is history as they say. One of my tasks the first year was to kind of try to rev up the morale in the lab by focusing on things that were quickly going to get ahead and were doable.

SCH In Ken’s lab I actually worked on bushy stunt probably 50% of the time. What had happened by '71 was that I was pretty sure the platinum derivative would work. I desperately wanted to find a mercury derivative, which are usually better because of the sulfhydryl chemistry, and I thought I had a solution to the platinum coordinates from a slightly complex route that didn't involve Pattersons but didn't yet involve the EM phases directly. It involved something called icosahedral harmonics - which I don't think I’ve got time to explain today. It basically has to do with expanding in spherical coordinates, which came out of the work on small angle scattering, the Fourier transforms instead of being sines and cosines (which is a Cartesian expansion) and from the small angle pattern and from the computation of these harmonics, one could straight away get the phases for the crystallographic terms that involved the spherically symmetric part of the structure, but one also could make a guess at phases for some nonspherically symmetric terms based on what the electron micrographs told you in a qualitative way. That is, that it had to have towers on the twofold axis that were in fact the projecting domains, meant that two particular harmonics dominated the pattern and they had to be phased in a certain way and therefore you could make a crude guess about the phases of another 20 or so low resolution reflections, which you could use to try to calculate a difference Fourier for the platinum derivative, and that actually in retrospect worked. What I did in Heidelberg was get somewhat better data, because Ken Holmes had the new rotating anode generators and had set up mirrors, since he had indeed originated that kind of optics for muscle in Cambridge, and I simply played around computationally with efforts to prove that my guess about where the platinum complexes were, was right and to improve my knowledge of where they were. And an important contribution was that Ken had pioneered in his laboratory something that within a year or two I would be doing back in Boston or Cambridge, but hadn't yet, that was using interactive computers; that is, small computers you had in the lab, where instead of submitting a job you sat there and punched in stuff just the way you do everything at a work station these days or in a bigger computers interactively. So that I was able to get a lot of tiny jobs run in an evening just by sitting and doing things and the computing was free instead of something you paid for the cpu-minute or cpu-hour, and therefore I could literally play around with a much more trial-and-error approach, or semi-trial-and-error approach, to improving the platinum computation, rather than figure out how I would do it mathematically and submit some job and hope that it told me a better answer.

SCH And that was an important revolution in computing. I assume that in computing, it is regarded as that, namely the introduction of the much more interactive basically PDP computers. Ken had a PDP11, or whatever it was, and I sat at that and played and by the end of that year I was pretty sure I did know where the platinum was, and I had conventional R factor statistics to support that, which really wasn't true before. And I had also found conditions to get a mercury derivative and probably gotten better pictures. Yes, I had gotten pictures to 8 angstroms by that time. I knew the data went to 3 angstroms, but hadn't done precession photographs beyond about 16 angstroms in Caspar's lab. But in Ken's lab because we had much more powerful rotating anodes, I could then get data to 8 angstroms.

SS At this point did you really have any picture in your head of what you were going to see of tomato bushy stunt virus?

SCH Not beyond the electron micrographs. Definitely not beyond the EM. And since even at 5 angstroms we got it partly right and to some extent a bit wrong, it wasn't until one had a high resolution map that it was clear what was going on. So no, I didn't have the vaguest idea.

SCH No I didn't trace the chain until we got the bushy stunt map in '77. In fact, when we got the 5 angstrom map, or 5 and one-half angstrom map in Jan '76 and I interpreted certain things as likely to be beta sheet domains - it was conventional wisdom that you couldn't identify beta sheets at that resolution and people thought I was going way overboard - whereas in fact the map was so clear that I was completely right about that, but people were so unused to seeing so beautiful a map at 5.5 angstrom resolution because they had never seen a non-crystallographically averaged map - a map where the phases were really good. The phases you get from multiple isomorphous displacement are really crummy but they are good enough that with visual or now more automatic interpretation you can trace a chain because we know so much about the chemistry of proteins.

In those days you didn't, so you had to have a very good map with many, many derivatives. The maps were not always wonderful, whereas the map we had at 5.5 angstroms resolution was what people thought they should have at 3.5 angstroms resolution or something, in terms of what you could interpret. And the high resolution map at 2.8 angstroms resolution map was the prettiest map at that resolution anybody had ever seen straight, because the phases were so much better. But anyhow we're jumping ahead - so about half my time in Ken’s lab was spent in making major advances and solving the phase problem for a structure like this and in learning about new computational - not so much methods but new computational styles. And then the rest of my time was doing a few different things. A membrane bilayer project on sarcoplasm reticulum with another postdoc in Ken’s lab, somebody from Hasselbach’s department next door, and assisting them with the follow up of the first synchrotron radiation experiment. Before I got there, Ken, with some collaborators, had done the first synchrotron diffraction experiment. I think it might be the first diffraction experiment, whatever you were diffracting, at a synchrotron. It was certainly the first biological diffraction experiment at a synchrotron and I think it might have been the first diffraction experiment at a synchrotron. And so Gerd Rosenbaum, who was a student of Ken’s, and others were then trying to build a much more satisfactory apparatus for doing muscle diffraction at the Deutches Electronen Synchrotron DESY) and I sort of went along and helped. That was my introduction to synchrotrons.