The rapidly increasing amount of information retrieval experimentation necessitates that considerable attention be paid to the problems of experimental methodology. The authors feel that the generalized and well-known experience in the physical, life and behavioral sciences concerning the role of methodology in scientific research must be applied to information retrieval experimentation. The scope of this paper is limited to experimentation involving retrieval systems and sub-systems. After defining the terms used in the paper, the authors discuss the structure of information retrieval systems, variables within retrieval system components, and experimental controls. It is concluded that the need for controls in experiments is clearly evident. Some unresolved questions which warrant further research are listed. (Author/CC)