Peter was constantly active in the organization of international scienti"c meetings. Here in Poitiers he will particularly be remembered for the crucial part he played in setting up the present series of Workshops, the "rst of which took place at San Miniato (Pisa) in 1992. Peter's research for his doctorate and directly afterwards concentrated upon electrical machine design. His early postdoctoral work, in 1966, on the numerical solution of eddy-currents problems was immediately singled out as showing considerable insight and invention. This provides the basis for the further work by Jan and Cendes, on the transient analysis of transmission lines, which is presented in this session.
In those early years, Peter, with a small band of other pioneers, began to recognize that the high-speed digital electronic computer held great promise as a tool for use in engineering design, not least in electrical engineering. To go back even further, in 1943 the mathematician Courant published a paper, based upon the Rayleigh}Ritz principle, in which piecewise continuous functions were optimized to approximate the solution of a mechanical vibration problem. This paper could be said to represent the birth of the &"nite element method', but progress in exploiting the new idea was slow because computing facilities at that time could not match the prodigious number-crunching required. However, in the early 1960s, with the advent of the electronic computer, work began in earnest applying the "nite element method to mechanical engineering problems. Peter took note of this work and realized that the method could also very e!ectively be exploited in electrical engineering. He was the author of a paper entitled &"nite-element solution of homogeneous waveguide problems', presented at the 1968 URSI Symposium on electromagnetic waves, Stresa, Italy, which could be said to represent the "rst application of the "nite element method to a problem in electrical engineering. He went on to formulate the universal matrices for the scalar variable simplex "nite elements, of general order in two and three dimensions, which for several decades represented the principal work-horse of the method. Since