How do new disciplines develop in certain universities more than in others? The history of computing in France suggests a model to describe the development process that shapes the geography of science. In the 1950s a few professors of numerical analysis, often associated with a school of electrical engineering, created threefold structures comprising courses in applied mathematics, a computing facility, and a research laboratory. Such local configurations initiated a cumulative development process, attracting more resources and opening the field to novel investigations in computer science. In other universities, these configurations were not completed, and computing remained confined to technical training. In the 1960s pioneers became leaders in the growing field and controlled the definition of computer science curricula. As the institutes they had created reached considerable size, these leaders began to spin off junior professors toward other universities. The centers remain major academic institutions in the discipline today.