Crystallography in Canada

continued from Volume 17, Number 4

Lachlan Cranswick was the driving force responsible for generating the Canadian content of these issues. He was a strong believer in the preservation of the history of the development of crystallography and crystallographic computing. He now is a prominent part of the history of Crystallography in Canada.

Early days of protein crystallography in Canada

Protein crystallography in Canada dates back to the 1960s. At the Physics Division of NRC in Ottawa there was already a strong contingent of small-molecule crystallographers led by W. F. Barnes. A member of that group, Maria Przybylska, moved to the Chemistry Division and embarked on a major protein crystallography project, a complex between hemoglobin and haptoglobin. George Birnbaum, who joined her group in 1966, became involved in this project by isolating and identifying the monomer haptoglobin 1-1 obtained from ascites fluids of cancer patients. Many years were devoted to obtaining suitable crystals of the Hb:Hp complex but without success, and the structure was never solved.

The University of Toronto was also a hub of small molecule crystallography, with Stan Nyberg in the Chemistry Dept. and Norman Camerman in the Biochemistry Dept. Theo Hofmann, in the Biochemistry Dept. at the University of Toronto, isolated an aspartic proteinase from the fungus Penicillium janthinellum. He named it penicillopepsin. Theo characterized penicillopepsin biochemically and he recognized the need for structural data in order to understand the catalytic mechanism. Theo grew the first crystals of penicillopepsin, data collection was performed in Stan Nyberg's laboratory on precession photographs and a postdoc, I-Nan Hsu, carried out initial work. The work was subsequently transferred to the University of Alberta in the mid 1970s.

In addition to Cambridge's Laboratory of Molecular Biology, Dorothy Hodgkin's laboratory in the University of Oxford was a Mecca for those wanting to learn the methods of protein crystallography. Two Manitobans were privileged to study with Dorothy in the 1960s, Carol Saunderson and myself. I arrived in 1963 just as Carol was graduating.

In 1965, I traveled to London with two fellow researchers from Dorothy's group, Ken Watson and Tony Cooper, to attend the unveiling of the lysozyme structure at the Royal Inst. David Phillips headed the team of researchers that solved this structure at 2.0 Å resolution. Included in the Phillips' team were Tony North, Colin Blake and Louise Johnson. It was an inspirational experience, to see the structure of egg white lysozyme, the first enzyme to have its structure determined. This research inspired many others to work on the catalytic mechanism of an enzyme that had a three-dimensional structure available. I was determined to work on the structures of enzymes from that day forward.

In 1966, I returned to Canada and joined Dave Hall in the Chemistry Dept. at the University of Alberta. Dave was a small-molecule transition metal crystallographer who was starting one of the first crystallographic laboratories in Alberta. Also present in the Chemistry Dept. was Ray Lemieux, one of the world's leading carbohydrate chemists. I had the good fortune to work with Ray as a postdoc in Dave's laboratory on structures of small monosaccharides with a variety of halo-substituents on the C1 atom of the pyranose ring.

In late 1967, I received a phone call from Larry Smillie and Cyril Kay in the Biochemistry Dept. at the University of Alberta. They wanted to start a protein crystallography group. Having obtained my name from Manpower Canada under the job category of 'crystallographer', Larry and Cyril invited me to join them, provided I received a scholarship from the Medical Research Council of Canada (MRC) to support my research and salary. After being successful with this MRC application, I started my forty-year long career in the Dept. of Biochemistry in July 1968.

Larry Smillie provided me with a highly purified sample of α-lytic protease, an enzyme from the soil bacterium Lysobacter enzymogenes. We managed to crystallize it and my first post-doctoral fellow, Michael Joynson, and I set about determining its structure. This was an interesting and exciting project because Larry had just finished determining the amino-acid sequence of the protein and had shown that it was closely related to the mammalian enzyme α-chymotrypsin. In order to show the evolutionary relationship between these enzymes, it was important to determine that it had a similar tertiary structure. Eventually we showed that it did!

SGPB in 1975

SGPB in 2007

The first protein structure determined in Canada was done in my laboratory in Alberta. SGPB is a proteolytic enzyme from a soil bacterium, Streptomyces griseus, and is closely related in sequence to α-lytic protease. The 2.8 Å resolution structure was published in Nature in 1975. SGPB had the same disposition of aspartate, histidine and serine residues, as did α-chymotrypsin. It also had the two beta-barrels that constituted the major fold of the molecule. SGPB and α-chymotrypsin were clearly related evolutionarily, even though one was from a cow and the other from a bacterium. Other proteolytic enzymes from Streptomyces griseus that our group solved were SGPA and the trypsin-like enzyme, SGT. SGPA constituted the major part of the work that Gary Brayer did for his PhD thesis and Randy Read did the work on SGT for part of his PhD thesis research. Many other people played major roles in the early stages of our protein structural studies: Penny Codding, Wendy Hutcheon, Masao Fujinaga, Louis Delbaere, Bill Thiessen and the biochemists who isolated and purified the original enzyme preps, Lubo Jurasek, Larry Smillie and Peter Johnson.

The structure of penicillopepsin, determined by I-Nan Hsu in collaboration with Theo Hofmann's group at the University of Toronto in 1976, produced a great deal of excitement because it brought with it the structural data that would aid in interpreting the catalytic mechanism of this family of enzymes. The aspartic proteinases are so-named because they have two aspartic acid residues in an amino-acid triplet of Asp-Thr(or Ser)-Gly at the active site. This family is medically important because two of the members are kidney renin, an aspartic proteinase that is critical in the regulation of blood pressure, and HIV proteinase, an enzyme that is crucial for the replication and 'life' cycle of the HIV virus. Because of the medical importance of these two enzymes, they rapidly became key targets for the development of inhibitors, in the regulation of blood pressure and as antiviral drugs. These were certainly very 'heady' days in the field of structure-based drug design and many of those suffering today from AIDS have benefited tremendously from this research.

Other leading protein crystallographers in the Dept. of Biochemistry at that time include Bob Fletterick and Wayne Anderson. Bob worked with Neil Madsen on the structure of rabbit muscle phosphorylase a. Bob collected a data set every day (before the days of flash-cooled crystals) in order to get suitable heavy-atom substituted crystals to solve the MIR phase problem. With two capable postdocs, Jurgen Sygush and Steve Sprang, Bob's team solved the structure of phoshorylase a. The sugar binding sites for the glycogen molecule were visualized and the pyridoxal phosphate bound to lysine was identified. This was a momentous achievement in these early days of protein crystallography.

Wayne Anderson, who graduated from Tom Steitz' laboratory in Yale, came to the MRC Group after a postdoctoral stint in Brian Matthews' lab at the University of Oregon in Eugene. Wayne was part of the team that solved the structure of the DNA binding protein CRO. Wayne worked on Fab fragments of monoclonal antibodies that had been raised against DNA. This was an excellent model for the autoimmune disease, Lupus erythromatosis. With Wayne's help we were successful in purchasing one of the first multiwire detectors from Ron Hamlin in San Diego. This was a huge advance in the data collection speed in Alberta and we used it to tremendous advantage.

The 40 years that I have been a member of the Dept. of Biochemistry have been truly rewarding. I owe a great debt to all of those who have supported me. I want to thank Michelle James for typing this manuscript. I have been fortunate to have received funding from a variety of agencies over the years, the MRC of Canada, CIHR, AHFMR, the Canada Research Chairs program and CFI among others.

Eric Gabe

Eric Gabe (1957)

I graduated from Cardiff University with a BSc in 1957 and went on to postgraduate work in crystal structure analysis under the direction of D. F. Grant, obtaining a PhD in 1960.

During that time computers were becoming available and it was possible to attempt three-dimensional analysis of medium-sized organic compounds for the first time, though it was a very tedious process. There were no integrated systems for structure analysis and it involved the use of several different computers. Thus, the output from a structure-factor calculation on one computer was incompatible with calculating a Fourier map on another computer, and would require manual re-typing of the required paper-tape input. Time on the computers, which were in different cities, always seemed to be allotted to crystallographers during the night. At five in the morning, realizing that the paper-tape going into one side of the reader was not emerging on the other, was not the happiest experience. Typically, the whole process might take a year.

I then moved to the Pure Physics Division of NRC in Ottawa as a postdoc under W. H. Barnes, though I worked mainly with Farid Ahmed using programs which he had written for an IBM 650. After a brief stint in London using a Ferranti Mercury machine - from midnight to 6 am - I moved to A. L. Patterson's lab in Philadelphia where I wrote programs for the in-lab IBM 1620, a rather strange machine which used a decimal structure rather than binary.

Eric Gabe with Picker diffractometer controlled by DEC PDP-8 (circa late 1960s).

In 1966 I moved back to Ottawa and became involved in the automation of a Picker diffractometer in the Mines Branch labs of the Dept. of Energy, Mines and Resources; the first such machine in Canada. This work used a DEC PDP-8 minicomputer and, because of its speed, I was able to use the otherwise idle computer time while the instrument was scanning a reflection to calculate absorption corrections by Gaussian quadrature. This led me to realize that such machines could be used for all aspects of structure analysis. In 1972 I began work in the Chemistry Division at NRC on what was to become the NRCVAX package. The first programs were written in machine code for the PDP-8e computer, that had 4 K 12-bit words and a 1 microsecond cycle time. Working in collaboration with D. F. Grant, A. C. Larson, Yu Wang and P. S. White, we were able to develop a complete Fortran package for data collection and analysis using the PDP-8e. The introduction of floating-point hardware made it possible to solve and refine a 20 to 40 atom structure in less than a working day. The only real restriction was caused by the size of the memory in full-matrix calculations and block-diagonal methods were used.

The diffractometer program, DIFRAC, migrated to control CAD-4 machines using PDP-11 computers and the analysis package moved to the PDP-11 60. Memory size still imposed a limit on full-matrix least-squares and, in fact, was more severe on the PDP 11 than on the earlier machine, because of the less efficient overlay scheme. In the early 70s most structure analysis was done on large mainframes and many people had difficulty in believing that it could be done on in-lab minicomputers. I remember being told at a meeting that such things could not be done on a PDP-8!

When the system moved to the PDP-11 we encountered portability problems, with compiler inconsistencies and different input/output handling. Programs were modified accordingly, but we realized that the code needed to be made more portable. When the system migrated to the DEC VAX in the mid-70s (and to the increasingly available PCs using Windows) size restrictions were removed and the code itself was revised to use a safe subset of Fortran 77, sometimes referred to as 'Pidgin' Fortran. Areas where compilers in different machines still caused problems were isolated to a small set of routines which could be controlled by system specific parameters. These included graphics, input/output, Boolean and time/date routines. In this way potential problem code could be isolated and easily dealt with if the need arose. The infamous, and grossly overstated, Y2K problem was dealt with by changing one line of code. The NRCVAX system is not, and was never meant to be, a 'black box', though it can be used quite successfully with the built-in defaults. Users with crystallographic knowledge, however, can use the many options available to solve and refine difficult structures.

Osvald Knop

Osvald Knop

One of the first to do crystallography in the Maritime Provinces was Osvald Knop (Ossie). Born in 1922, he arrived in Halifax in 1949 via Mazarek University in Czechoslovakia, Sweden, a spell at Cal Tech with Linus Pauling and Laval (DSc 1957). At Cal Tech, Ossie worked on intermetallic phases attempting to solve the three-dimensional structure of Co₂Na using a Hollerith Punch Card adaptation of Beevers-Lipson strips. A similar machine was installed in the basement of the Mathematics Inst. in Oxford. It was powered by a multi-horsepower diesel motor, which was started by a crank handle, and when it ran (night only) the whole building shook. In Halifax with Ossie there was also Karoline (Lottie) Castelliz, who received her PhD in Vienna in 1932.

In Halifax Ossie first worked on the three-dimensional structures of the lycopodium alkaloids, although once his work showed progress other groups with greater resources and deeper pockets took over. His other major interests were metal oxides, sulfides and intermetallic phases. The early equipment consisted of a Solus Schall generator - with a pre-war model plate - and a band-driven Weissenberg camera. The generator, still faintly working, departed to the dump only 10 years ago - the transformer oil containing no trace of PCBs. The camera still exists; those who remember this model will not be surprised to learn that it is now exhibited as a memorial to the misery of working with a badly designed instrument.

Ossie and his students produced many elegantly accurate phase equilibrium diagrams, some with neutron diffraction data (he was the first to use Chalk River for diffraction experiments). The three main furnaces for this work - constructed by Ken Reid - were called Faith, Hope and Charity and survived in working order for many years. Ossie's time in Halifax included 7 years at Nova Scotia Technical College and many years at Dalhousie. He retired in 1990, became an Emeritus Professor and still comes into the lab most days of the week.

Comments on QTAIM

Three representations of the structure of the guanine-cytosine base pair: (a) classical structure; (b) the network of bond paths linking neighbouring atoms that defines the molecular graph; (c) contour map of the electron density, showing the bond paths and the intersections of the zero-flux interatomic surfaces with the plane of the diagram. The atomic symbols in (b) denote the positions of the nuclei: H (white),C (black), O (red), N (blue). The small yellow dots denote ring critical points, the red dots the positions of the bond critical points: the origin of the unique pair of trajectories of grad ρ, each of which terminates at a neighbouring nucleus, and the terminus for the set of trajectories that define the interatomic surface. There is a wealth of information in the molecular graph. Note that it provides a faithful mapping of the classical structure onto the topology of the density, showing in addition to the three hydrogen bonds a weak interaction between a methyl H and the keto O in cytosine that is not indicated in the classical structure. The presence of the bond path is but a useful way of depicting and summarizing which pairs of atoms share an interatomic surface, as demonstrated in (c). That this shorthand notation mimics the way in which the same information is conveyed by the structures that evolved from experimental chemistry is surely one of the most powerful of all the physical vindications of the zero-flux boundary condition for the definition of a quantum open system.

The QTAIM approach to the examination of electron density has provided crystallographers with a remarkable tool to examine, experimentally, the fine details of the electron density throughout a crystal. Our group here at Dalhousie has been specializing in the collection of high-angle, low-temperature, multi-redundant X-ray diffraction data to examine the weak interactions present in crystals. In a series of papers we are beginning to see that although many of the C-H...X and even H...H interactions are weak, they are, with the insights from QTAIM, clearly detectable. Moreover for many of these weak interactions it begins to appear that they are far more logically organized than was ever suspected. The result is that the species in crystals are often held in place not just by the familiar conventional forces but also by a multitude of Lilliputian tethers.

The resulting theory is called the quantum theory of atoms in molecules, QTAIM,  developed by R. F. W. Bader. It is well documented that the atomic and group properties predicted by QTAIM agree with the additive group contributions measured experimentally, agreement with experiment being the only test of theory.

QTAIM, by providing the quantum basis for an atom in a molecule, necessarily recovers all of the related concepts of experimental chemistry. Thus in addition to the definition of atoms and molecular structure, QTAIM provides the physical basis for the Lewis model and its associated chemical concepts. The pairing of electrons and the associated concepts of electron localization/delocalization are determined by the atomic expectation value of the exchange density and given physical expression in the topology of the Laplacian of the electron density, the quantity ^.2ρ.

The measurable electron density provides the link between the primary concepts of experimental chemistry and quantum mechanics, providing the prediction and understanding of the properties of matter at the atomic level. Every question capable of expression in terms of quantum observables - energy, force, pressure, current etc. - can be asked and answered. QTAIM provides a new way of asking and answering questions concerning structure and its relation to measured properties that previously were matters of intense debate, the very definition of structure - of bonds being present or absent - being a source of controversy. It only remains for the investigator of the density to ask proper and meaningful questions.

The turning point in the acceptance of the topological theory of structure by X-ray crystallographers came with the demonstration that the topology of the Laplacian of the density provides a remarkable pictorial mapping of the Lewis electron pair concept onto real space, the bonded and lone pairs being defined in terms of concentrations of charge appearing as local maxima in ^.2ρ.

A life in science

Left to right back row: Williamson, Booth, Llewelyn; front row: Boyle, Claringbull, Small. From roof of Birmingham Chemistry Building, 1942.

I am the last surviving member of the group of X-ray crystallographers who worked in the Chemistry Dept. at Birmingham University in the early 1940s. Sir Norman Haworth, who won the Nobel Prize for his determination of the molecular structure of ascorbic acid (vitamin C), was Chairman.

My earliest training in X-ray crystallography involved rehabilitation of a home-made X-ray system by order of E. Cox (later Sir Gordon) in consultation with George Jeffrey and working on the structure of the explosive penterythritolo tetranitrate (PETN) as part of the war effort with a scholarship from the British Rubber Producers' Research Association (BRPRA).

The Bernal team, 1948. Left to right, back: Aaron Klug, Jeffrey, Hirsh, Pit, Helena Scouludi; front: Anita Rimel, Ehrenberg, Desmond Bernal, Helen Megaw, Carlisle. Missing: Booth away in USA.

After postdoctoral work at the BRPRA laboratories, where I designed a large, mechanical, structure factor calculator and a relay Fourier synthesizer, I was offered a lectureship in physics at Birkbeck College London, under the leadership of Desmond Bernal. The 'bio-molecular' laboratory was housed in the Davy-Faraday laboratories at the Royal Institution, where I wrote 'Fourier Technique in X-ray Organic Structure Analysis' (Cambridge University Press, 1948) and taught optics and theoretical physics at the old Birkbeck building.

Following a visit to the USA in 1946 to attend the 1st ASXRED conference at Lake George, I visited most of the computing facilities and on return to England terminated the work on the mechanical Fourier machine at BRPRA and started on an all-electronic version. With a Rockefeller Fellowship I worked with von Neumann at the Inst. for Advanced Study in Princeton, designing two general-purpose computers, the ARC and APEXC.

In 1962 I accepted the post of the Chair of Electrical Engineering with the University of Saskatchewan. The University of Saskatchewan was founded in 1909. The College of Engineering consisted of Depts. of Agricultural, Civil, Mechanical, Chemical and Geological Engineering. The first year enrollment was about 1000, mostly male. None of the departmental chairmen had doctoral degrees and there was practically no research.

Andrew Booth working on a digital multiplier

In my first year I started a number of projects, including the design and construction of a fully transistorized version of the APEXC. With excellent financial assistance from the Canadian National Research Council and the Defence Board this machine was completed within the year. However, universities in the prairie provinces had no interest in X-ray structure analysis and the machine was used for bio-medical and other data acquisition.

Out of the blue came the offer of the Presidency of Lakehead University in Ontario, a new university with splendid buildings. My mandate there was to stabilize the faculty and students, to introduce research and to balance the budget. I managed to do everything but initiate research. I had kept in touch with the area of small computers so I got one of the first TI60 programmable pocket calculators with associated printer unit. I spent a happy Sunday afternoon programming it to calculate and print all of the F(h,k,l) values for PETN from the coordinates in our original paper. It was about 10 times faster than the months of work in the 1940s.

On retirement I also used a PET computer to automate a Fourier refinement of the original PETN coordinates. The program was written in BASIC and the results cleared up several anomalies that we had detected in the original work. I am still very interested in mathematical crystallography and often return to several problems from my early days: uniqueness of the solution in the absence of phase information and the general problem of phase determination.

19th century basics spearhead 21st century progress: some autobiographical details

The team of ICPET (Institute for Chemical Process and Environmental Technology) and IRC (Institute for Research in Construction) crystallographers receives an NRC award for Research Excellence on April 28, 2008. Front: Isobel Davidson, Pierre Coulombe (NRC President), Pamela Whitfield. Back: Yvon Le Page, Patrick Mercier, Lyndon Mitchell. (Credit for photograph: National Research Council of Canada).

In the fall of 1976, I was a postdoc with Gabrielle (Gai) Donnay, a diffractionist with great reputation and a crystallography professor at the Geology Dept. of McGill University. Her husband, Joseph (José or J. D. H.) Désiré Hubert Donnay (1902-1994), was a legend of mineralogy, crystal morphology and twinning. They occupied facing desks in the same office. Together, they were an encyclopedia of mineralogy, crystallography and diffraction. I had been assigned the task of teaching the undergraduate crystallography course to students from the physics, chemistry and geology departments, while the Donnays enjoyed a sabbatical year. Teaching 40 undergraduate and graduate students was a daunting task for a recent graduate in physics. Knowing that the lecture notes would be scrutinized by the Donnays made things really daunting.

I was careful to prove almost everything, including the existence of seven crystal systems and no more, a fact that is usually taken for granted. The students were great. Nobody dropped the course or flunked the final exam, to the amazement of the Donnays who checked that the all exam books were up to their specs for graduation.

My second postdoc in 1977 was at NRC's X-ray diffraction laboratory at the Chemistry Division in Ottawa, working with Larry Calvert, Eric Gabe, Allen Larson and Yu Wang on the nuts and bolts of what would become the NRCCAD Fortran diffractometer program, the NRCVAX structure package and the CRYSTMET crystal structure database. At the IUCr XIth Meeting in Warsaw I presented two contributions, a poster on least-squares weights in accurate refinements and a short proof for the existence of no more than seven crystal systems.

The most pressing problems we had to tackle with diffractometer automation at the NRC were recognition of the metric symmetry of the cell, and derivation of crystal symmetry and its subsequent use in efficient data collection. The popular cell-reduction algorithm at that time was based on distinguishing 44 mutually exclusive cases for the matrix of dot products between edges of the Niggli cell. As Niggli cell reduction requires knowledge of the lattice symmetry, I felt it logically circular to use the Niggli cell to determine lattice symmetry. Mathematically mutually exclusive cases may no longer be mutually exclusive when experimental error steps in, creating problems having to do with pseudo-symmetrical lattices or with the ordering of logical tests. Something more robust was needed.

The scientific library of CISTI for Canada included excellent original editions of Bravais, Mallard, Voigt, Friedel, Wyckoff, Niggli etc. A pristine copy of Seeber (1831) was in a locked room reserved for rare books. I could see that J. D. H. had not exaggerated when he told me that the algebraic proof of the primitivity theorem sprawled there over about fifty pages.

Mastering the derivations in these texts paved the way for the production of many programs I would write during the course of my career, including the CREDUC cell reduction algorithm at the heart of NRCCAD, the MDF checking software, CRYSTMET, Materials Toolkit and MISSYM. MISSYM identifies the symmetry of a structural model within a given tolerance and reduces the possibility of error in space group assignment. It is difficult for crystallographers not familiar with the vigilante campaign that Dick Marsh waged in the 1980s against errors in space group assignment to appreciate the significance of MISSYM. Dick had an incredible talent to 'see' symmetry upon inspection of atomic coordinates. When Dick published a note correcting your space group you were said to have been 'Marshed'. Despite widespread awareness of his efforts to educate contributors, reviewers and editors of articles, Dick went on Marshing about one structure per issue of Acta C for years. Marshing had the unintended adverse effect of raising uncertainty about the accuracy of crystallographic reports in the scientific community at large. MISSYM made it possible to routinely determine space groups correctly. The application of MISSYM to structures in back issues of Acta C flagged all structures previously Marshed and found a few genuine pseudo-symmetrical structures. The NRC gave the IUCr a license for the use of MISSYM and Brian McMahon added it to the IUCr checking package in 1989. Marshable structures became an extinct species in IUCr journals, Dick Marsh refocused his campaign on chemistry journals and MISSYM went on with its own career as a filter in crystal-structure databases.

The CREDUC algorithm was used for interpretation of convergent-beam electron-diffraction patterns for cell volumes or least-squares cell data. Those concepts and developments were applied to rapid phase analysis in electron microscopy.

Twinning is due to metric symmetry or pseudosymmetry of a multiple cell, and is characterized by its twin law with small obliquity and small twin index.

Automated detection of twinning and analysis of its nature is a problem for which no software existed. It was possible to adapt the CREDUC algorithm to detect, identify and analyze twins.

Noting that the performance of PCs was within one or two orders of magnitude of what was needed to perform quantum computations on real systems, Paul Saxe from Materials Design and I set out to automate quantum modeling in MedeA. We automated the least-squares symmetry-general calculation of the elastic tensor from total energy and from stress calculations. Materials Toolkit includes exploratory computation of thermomechanical properties of existing or prospective materials useful in seismology, storage of toxic elements, correction of experimental errors in published elastic tensors and the design of a coating for turbine blades in jet engines that increases their resistance to erosion.

My current focus continues to be upon complementing structural studies with quantitative calculations and modeling, introducing young crystallographers to quantum methods and expanding the range of automated applications of quantum chemistry. The latest such study has been the precise quantum calculation of surface tension on crystal faces for metals.

Quantum modeling of materials, a branch of modern crystallography, is currently undergoing explosive expansion due to greater automation harnessing the exponential increase of available computing power. This expansion will ultimately be a success only if the software that automates it rests on theorems and not on ad hoc considerations suffering from particular cases and exceptions.

Stanley C. Nyburg

Stanley C. Nyburg

I was born in London, UK, in 1924. In 1939, at the outbreak of World War II, my school was evacuated to Weymouth on the south coast. Whilst there in May, we experienced the dreadful evacuation of troops (mainly French) from Dunkirk whom we helped to get fed and accommodated. In 1939 I took my matriculation examinations. We were examined on The Merchant of Venice during a heavy air-raid and (almost certainly illegally) were not sent to an air-raid shelter. We enjoyed the bombardment greatly. Oil tanks in the harbour were struck and the sky was so darkened we had to finish with all the lights on. I passed all the exams.

I enrolled as a part-time university student at Birkbeck College to read for an honours degree in chemistry with physics subsidiary.

After a year at King's College, London, G. M. Bennett found me an opening as a crystallographic trainee at the British Rubber Producers' Research Association (BRPRA), an Institution of London University where I met George Jeffrey ('Jeff') (my supervisor) and Andrew Booth (q.v.). I started X-ray crystallography with primitive high-tension equipment on which I nearly killed myself. E. G. Cox (later Sir) was appointed to a Chair in Chemistry at Leeds University and asked Jeff to join him as lecturer, where I read for a PhD. After a considerable struggle, I managed, by photographic methods, to solve the crystal structure of an organic compound.

From the 1954 published structure of rubber.

In 1949 I obtained an assistant lectureship at the new University College of N. Staffordshire at Keele. At the BRPRA I stretched a sheet of natural rubber on a frame and put it in a freezer. A small fragment of frozen rubber was mounted in a primitive cold stream in a collimated beam of X-rays. Excellent diagrams were obtained at 30 deg rotational intervals and all intensities estimated by eye. The Bragg intensities could be explained by a disordered crystal structure. The structure has been widely accepted to be true.

The correct molecular connectivity of aspidospermine, as determined by crystallographic methods in the late 1950s.

At Keele, we built our own X-ray generator, determined Bragg intensities photographically and determined a number of crystal structures including a natural product that had resisted structural characterization despite over 40 years of intense chemical study.

In 1958 G. A. Jeffrey, who was at the University of Pittsburgh, offered me a Fulbright Fellowship. Here I started my investigations into how crystalline Cl₂ came to have its extraordinary herring bone structure.

Overbutting of two nitrogen molecules in a ruthenium complex.

In 1963 I was offered the post of full professor at the University of Toronto. In the new Dept. of Chemistry I was given substantial funds with which to get started. Automatic diffractometers were becoming available and I decided on the four-circle Picker machine. This proved a good choice, the machine being operative for more than 40 years. During its 23 or so years in Toronto the Picker employed a variety of computers and devices to attain full automation. We used the NRCVAX system of structure determination written by Eric Gabe and collaborators at the National Research Council Laboratories in Ottawa. We provided a structure-solving service for the department and were engaged in other research activities. Our most exciting determination was the discovery by A. D. Allen in 1965 of a ruthenium complex containing molecular nitrogen as the ligand. Traditionally, it had always been considered that molecular nitrogen was too inert to form chemical compounds.

From a study of the Cambridge Organic Crystal Database we made scatter-plots of non-bonded distances between identical atomic elements, showing that for the halogens especially the effective van der Waals shapes are not truly spherical but flattened. In 1974 we wrote a program BMFIT to give the closest match between molecules having some structural features in common. We also collaborated on crystallographic matters with the Department of Pathology at Mt Sinai Hospital in Toronto.

y = 0 electron-density map of β-C24,C26, showing decreasing C-atom site occupancies towards the end of the C-atom chains.

During the 70s we conducted crystal studies of normal alkanes for Esso. We discovered that the published crystal structure of n-octadecane was wrong and corrected it. Mineral oils contain alkanes of many molecular lengths and when cooled this oil deposits waxes which greatly handicap the extraction of the oil. Despite enormous effort by the oil companies the crystalline nature of these waxes was not known. We decided to examine a binary mixture of alkanes as a model of a wax. The C20:C22 phase diagram examined by our Australian colleagues showed there were new phases in the 50%:50% region. We obtained excellent crystals from such a 50%:50% mixture. Weissenberg photographs were taken with the crystal cooled to 10°C. From these we inferred that the space group was orthorhombic while the two pure components were triclinic. We managed to find the crystal structure. Either of the two alkane molecules can occupy the same position, their combined symmetry being a mirror normal to the z axis, a symmetry element which neither alkane possesses. This fascinating structure forms the basis of the alkane wax structures. Later a study was made at King's College of the binary pair n-docosane:n-tetracosane. This truly remarkable crystal structure containing molecules with twenty-four and twenty-six carbon atoms shows twenty-seven Fourier peaks that confirmed our earlier results.

In 1987 I returned to King's College, London, as an honorary senior fellow, bringing my faithful Picker, which Kings purchased for a few thousand dollars. Its delicate goniometric parts were covered in a slow-setting sheath of polystyrene by someone who had done similar work for Ford automobile engines. It was shipped (all 0.75 tons) over the Atlantic in a Russian freighter to London, UK.

The Picker was newly fitted to a PC, an IBM 286, and Gabe's Fortran program for data collection was rewritten in Gbasic. The Picker behaved excellently and Adrian Parkins and I completed a large number of crystal structures for staff members.

Unfortunately, the Dept. of Chemistry at Kings College, London, was closed permanently by the College in 2006 and I am now an honorary senior fellow at University College London continuing normal alkane studies.

I. David Brown

I. David Brown

I studied for my bachelor's degrees in general science and honours physics at King's College, London, UK, at the time when Rosalind Franklin and Maurice Wilkins were determining the structure of DNA. For graduate studies I moved to the Royal Institution (RI) where the group working with the director, Sir Lawrence Bragg, was unraveling the first crystal structure of a protein. For my PhD under the supervision of Jack Dunitz, I solved the crystal structures of two organic complexes of Cu^I, doing most of the calculations by hand, but experimenting with writing machine-language code for a computer with around a hundred and twenty memory locations which had to store both program and input. When Jack accepted an appointment at the ETH, I followed and spent my final PhD year in Zurich. In 1959, as was customary at the time, I crossed the Atlantic to take up a postdoc at McMaster University with Howard Petch. My job was to finish building a single-crystal neutron diffractometer at the newly opened McMaster swimming pool reactor. During this time I developed an interest in the structure of inorganic compounds, a term which I use to include any non-metallic crystal that contains no C-H bonds. This was an area much neglected except by mineralogists who were, however, focused on natural minerals. Solid-state physicists were not prepared to work with anything more complex than a cubic crystal, preferably containing only one kind of atom, while chemists thought of a crystal as a kind of freezer in which a dormant compound could be indefinitely stored until it was needed to carry out some real chemistry in solution. Those who called themselves 'inorganic chemists' were more interested in preparing organic molecules that could bind to a transition metal atom. The neglect of inorganic solid-state chemistry arose in part from the lack of an effective model for the systematic description of infinite structures. Most descriptions were based on an eclectic mixture of words such as 'packing of spheres', 'ionic interactions', 'covalent bonds', 'space group symmetry',' polyhedral packing' and 'structure type'. Unlike the chemical bond descriptions used in organic chemistry, descriptions of inorganic structure tended to be vague and unconvincing.

In 1960 I was joined at McMaster by Chris Calvo, who in his short life published a series of high-quality structures of phosphates that still compare favourably with structures published today. We set up a small X-ray laboratory with a generator and a precession camera, where we were later joined by Colin Lock in chemistry and by Doug Grundy in geology. Subsequently this laboratory was expanded and provided structure determinations for other faculty members, while students had the opportunity to learn how to determine crystal structures. Today this laboratory is managed by Jim Britten, and is equipped with the latest X-ray diffractometers and software.

After my postdoctoral years I joined the McMaster Physics Department, where I spent my first ten years wandering in my own scientific wilderness. Although I gained experience solving structures for my chemistry colleagues I could not see where my research was going. As many can corroborate, it is difficult in the wilderness to know what one is looking for, or even if there is an answer to the question that has yet to be discovered. When Gai Donnay came to McMaster to show how Pauling's electrostatic valence model could be improved by taking account of the correlation between bond valence and bond length I failed to pick up on the clues. My epiphany came a couple of years later in 1971 when Bob Shannon arrived at McMaster on a year's leave from du Pont. Within an hour of our first meeting he had convinced me of the importance of Donnay and Allmann's work. The rest of that year we spent extending and simplifying their approach. In our paper we showed how bond valences could be calculated from a wide range of experimentally observed bond lengths, how they are normalized by arranging that they sum to the valence (oxidation state) of each atom, how they can be used to check the validity of newly determined structures, and how they provide a quantitative measure of the strength of a bond that, unlike the bond length, is independent of the nature of the atoms involved.

At the end of his stay, Bob Shannon left me six boxes of computer cards on which he had laboriously punched the unit cells, symmetry operators and atomic coordinates of over four hundred inorganic crystals, but this database was useless because there was no way to retrieve the information it contained. The experience made me realize the importance of creating a well-structured, machine-readable database if I wanted to carry out systematic studies. A large part of my subsequent career has been devoted to simplifying crystallographic data retrieval, first by creating an annual bibliography of inorganic crystal structures (BIDICS 1969-1981), and later proposing a common file structure to allow the interchange of crystal structure databases between programs and laboratories (this has now evolved into CIF). At the same time, Guenter Bergerhoff and I started the Inorganic Crystal Structure Database (ICSD). By 1985 the ICSD had reached the point where Daniel Altermatt and I were able to use it to prepare a systematic set of bond-valence parameters.

Subsequently, I expanded the bond-valence model to show how it can be used in structure modeling of both solids and liquids, how it predicts chemical stability and reactivity including aqueous solubility, and how the bonding geometry can be analyzed into separate contributions from chemical bonding, steric strain and electronic anisotropies. An invitation from Henk Schenk to present a series of lectures in Amsterdam in 1994 led to the publication in 2002 of my book 'The Chemical Bond in Inorganic Chemistry' (OUP) which brings all these threads together.

In 1986 I acted as local chair for the ACA meeting held at McMaster, the last ACA meeting held on a university campus. Subsequently, I persuaded the ACA to establish a Canadian Division and I served as its first representative on the ACA Executive, an experience which convinced me of the importance of the ACA in bringing together North American crystallographers.

My students and academic colleagues will probably remember most affectionately the daily morning coffee breaks when my group would meet together informally. We discussed many topics on these occasions, including the problems of structural chemistry, sometimes focusing on the particular problem of an individual, sometimes raising more general questions. On these occasions we would let our imaginations run wild, but always challenged each other to test and justify each new idea. These gatherings were the forge in which the bond-valence model was hammered out.

A.C. Larson in Canada

Allen Larson in a style of clothing more typical of Canada (circa 1981).

I had joined the Los Alamos Scientific Laboratory (LASL) in 1956 after graduating with a PhD from Washington University. Working with Don Cromer and Brad Roof studying the structure of alloys of plutonium, one of my principal activities was developing software for determining and refining the crystal structures of these materials. I quickly established myself as a software developer, using the LASL-developed Madcap language on the MANIAC (a LASL-developed computer) and very soon after that Fortran on the IBM computers at LASL. In 1966, I was elected to the ACA and IUCr (chair 1975-1978) Computing Committees.

We noted that many of the alloy structures we were finding were similar to already known structures. In an effort to enable us to solve the structure of the new phases that our metallurgists were creating, we decided that developing a computer-based data set of the known structures was necessary. In the summer of 1960 we had Joe Finny search the literature and enter the details of known metal and alloy structures into our data file.

In the summer of 1973 I was caught by a funding crisis at LASL that resulted in a reduction of staff. Since I did not want to abandon crystallography as they insisted, I opted to leave the laboratory and moved with my family to Austin. I then contacted several people around the US and in Canada, including W. B. (Bill) Pearson who was at the NRC in Ottawa. One of my concerns was the metals data file that Don Cromer and I had started to construct at Los Alamos. We felt that the project would be lost after I left Los Alamos. I asked Bill if there was anything I could do to find it a new home. He invited me to bring it to the NRC where Larry Calvert was working on the IUCr Structure Reports metals data.

I flew to Ottawa with the metals data file and with the software that we had developed to access the file on a magnetic tape. Larry Calvert and Eric Gabe had a DEC PDP-8 which we used to further develop the data file and its software.

After completing the transfer of the file, Calvert and Gabe asked me to install MULTAN on the PDP-8 as I had worked on doing that on a small computer for Syntex in Cupertino late in 1974. After we finished, it was agreed that I should install other portions of the Los Alamos Crystal Structure package on the PDP-8, including the space group symbol interpreting software that I had developed. However, licensing issues involving the Syntex installation led us to develop a completely new program. The accuracy of the new program was checked by comparing its output with that of the original program, that had already been checked manually against the International Tables for X-ray Crystallography, 1952. The new code uses matrices and vectors to develop the space groups and is the one that is used in the GSAS package today.

In the summer of 1978, Eric Gabe and I presented the NRCC software package at the IUCr Computing School in Enschede, The Netherlands.

I also spent time at the University of Calgary working with K. Ann Kerr, again installing and developing crystallographic software (June 1975 and March 1976). My last major stay in Ottawa was from November 1980 to February 1981. On this trip, Eric Gabe and I started to incorporate a powder data refinement capability into the software package and I was back in June 1981 for the IUCr, after which I returned to LASL.

My duties on returning to LASL included support for the diffraction instruments at the Los Alamos Neutron Scattering Center (LANCE) that were used for both single-crystal and powder studies. Bob Von Dreele (Arizona State) was one of the people involved with developing the powder diffraction capability and he assisted me in expanding the capability of the Los Alamos software to be able to refine structures against powder data.

One of the criteria that I always used in developing software was to enable the integration of new sources of data. With this in mind, Bob and I continued the process of inserting powder data refinement into the Los Alamos Crystal Structure package. I had been thinking of a name for our software system and came up with 'Generalized Structure Analysis System' that evolved into the 'General Structure Analysis System', GSAS, as it is still known and used today.

An interesting interlude with pyrolytic graphite

Larry Calvert (left) and Eric Gabe (right).

In 1972, after a brief visit to New Zealand, Larry Calvert and his colleague Eric Gabe called in at the (then) Chemical Physics Division of CSIRO in Melbourne. Discussion with the Chief of the Division, A. L. G. Rees, led to Larry coming to spend an extended spell with us in 1973-1974.

At that time, I was interested in certain aspects of extinction of an experimental nature, rather than the more usual theoretical concerns. Originally, the subject had dealt with the shape of single-crystal diffraction peaks. Since most single-crystal X-ray peaks are too narrow to handle experimentally, my attention had been directed to pyrolytic graphite, of which the 000l peaks were of a more feasible magnitude for experimental observations.

Joining us in these investigations was Reg Killean (School of Physical Sciences, University of St. Andrews, Scotland). One aim was to investigate whether the level of polarization changes across a diffraction peak. A source of polarized X-rays of adequate intensity was needed. It turned out that graphite 0006 could be quite successfully used with Cu Kα. The basic arrangement for this experiment involved a Picker diffractometer with a pyrolytic graphite crystal set to provide the polarized beam. This beam of polarized monochromatic radiation passed through a hollow shaft, constituting the collimator of a second small diffractometer with the second pg crystal (the analyzer). Mounted on the shaft was the detector arm, which could rotate about the shaft. One could therefore take readings of intensity at 0, 90, 180 and 270 degrees and deduce the polarization ratio at each angular setting of the specimen crystal at specific 2θ settings. The four readings allowed for internal checks. Thus, due to physical symmetry, readings at 90 and 270 degrees should be equal.

This allowed a complete study of the change in polarization of the 0002 reflection of pyrolytic graphite (a ZYA specimen). This procedure allows for accurate average values of polarization ratios for any crystal monochromator system. A further experimental study was possible using the polarized monochromatic beam and the small diffractometer. This was to investigate the interaction with a turbostatic structure. For this purpose, the polarized beam is passed along a hollow shaft which constitutes the inlet collimator to the small diffractometer. Also attached to the shaft is the detector unit with the second (specimen) crystal adjustment. The intensity for each setting of the specimen crystal is determined and can be converted to an estimate of absorption.

From the material studied in this last work, it was possible to obtain an estimate of the X-ray attenuation coefficient of carbon for Cu Kα₁ radiation. Larry was an enthusiastic and stimulating colleague and the three of us worked well together. The collaboration was both enjoyable and fruitful. The work carried out then has developed over the years and has led to a modified viewpoint on extinction.

As we went to press, we heard the sad news that the body of Lachlan Cranswick, who had gone missing in January (see ACA RefleXions, Spring 2010, p15), has been found. A full obituary will appear in the next issue.