For Andrew R., the relevant graphs from Adams et al. 1996.
The blocks are periods of forty seconds. I would say that the laboratory setting was not particularly inhibiting to subject reaction. It must have been the comfortable reclining chair.
Adams did not eliminate the alternate causal explanation of circumference increasing due to anxiety -- Intelligent Design, my ass -- so perhaps future experiments will include several minutes of anxiogenic video footage, say the last scene from Pink Flamingos. (No link. I have standards. Somewhere.) The resulting graph might need a negative y-axis.