Abstract
Four hypotheses are described that explain the structural differences among software development projects in terms of the relationship between team size and productivity and in terms of the differences between new and modification-oriented software development. Methods are proposed for verifying these hypotheses.