Abstract
This paper proposes the use of blockmodelling, a recent development in the field of network analysis, as a complement to traditional methodologies in studying organization structure. Blockmodelling is considered promising because it studies directly the multiple kinds of interactions which constitute organization structure rather than relying on abstractions of global properties (centralization, formalization, span of control etc.). Tests of some of the propositions of James Thompson and Joan Woodward are proposed as examples of how blockmodelling might be used in the study of organization structure.