Dewar was appointed to the Chair in Chemistry at Queen Mary College of the University of London in 1951. He moved to the University of Chicago in 1959 and then to the first Robert A. Welch research chair at the University of Texas at Austin in 1963. After a long and productive period there, he moved to the University of Florida in 1989. He retired in 1994 as Professor Emeritus at the University of Florida. He died in 1997. Dewar's reputation for providing original solutions to vexing puzzles first developed when he was still a postdoctoral fellow at the University of Oxford. In 1945, he deduced the correct structure for stipitatic acid, a mould product whose structure had baffled the leading chemists of the day. It involved a new kind of aromatic structure with a seven-membered ring for which Dewar coined the term tropolone. The discovery of the tropolone structure launched the field of non-benzenoid aromaticity, which witnessed feverish activity for several decades and greatly expanded the chemists' understanding of cyclic π-electron systems. Also in 1945, Dewar devised the then novel notion of a π complex, which he proposed as an intermediate in the benzidine rearrangement. This offered the first correct rationalisation of the electronic structure of complexes of transition metals with alkenes, later known as the Dewar–Chatt–Duncanson model. In the early 1950s, Dewar wrote a famous series of six articles on a general Molecular orbital Theory of Organic Chemistry, which extended and generalised Erich Hückel's original quantum mechanical treatments by using perturbation theory and resonance theory, and which in many ways originated the modern era of theoretical and computational organic chemistry. Following Woodward and Hoffmann's suggestion of selection rules for pericyclic reactions, Dewar championed an alternative approach to understanding pericyclic reactivity based on aromatic and antiaromatic transition states. He did not however believe in the utility of Möbius aromaticity, introduced by Edgar Heilbronner in 1964, and now a flourishing area of chemistry. He is known most famously for the development in the 1970s and 1980s of the Semi-empirical quantum chemistry methods, MINDO, MNDO, AM1 and PM3 that are in the MOPACcomputer program, and which for the first time enabled the quantitative study of the structure and mechanism of reaction of many real systems. This was illustrated in 1974 by computing the structure of a molecule as large as LSD at a quantum mechanical level. It is worth noting that in 2006, the equivalent calculation takes less than 1 minute on a personal computer. In 2006, the same structure computation can now be completed using high-level ab initio or density functional procedures in less than two days, and semiempirical programs can be used to optimise the structures of molecules with perhaps 10,000 atoms. He was a member of the International Academy of Quantum Molecular Science.
