THE PRINCIPAL EVIDENCE IN SUPPORT OF THIS HYPOTHESIS IS THE STABILITY OF THE IMPRISONMENT RATES IN THE UNITED STATES FROM 1930-1970, NORWAY FROM 1880-1964, AND CANADA FROM 1880-1959. THIS PAPER EXTENDS THE THEORETICAL BASIS OF AN EARLIER WORK AND SUGGESTS SOME PROCESSES THAT MIGHT GENERATE THE STABLE LEVEL OF PUNISHMENT. THE TIME SERIES OF THE IMPRISONMENT DATA FOR THE THREE COUNTRIES ARE ANALYZED TO PROVIDE AN EMPIRICAL DESCRIPTION OF THE STRUCTURE OF THE DATA, WHICH INDICATES STRONG SIMILARITIES IN THE THREE COUNTRIES STUDIED. DIFFERENT MODELS OF THE CRIME AND IMPRISONMENT PROCESS ARE THEN EXPLORED IN AN EFFORT TO CHARACTERIZE AN UNDERLYING PROCESS THAT WOULD GENERATE THE KINDS OF TIME SERIES OBSERVED. A SENSITIVITY ANALYSIS IS THEN PERFORMED TO IDENTIFY HOW THE DIFFERENT PARAMETERS OF ONE SUCH MODEL CONTRIBUTES TO NATIONAL DIFFERENCES IN OBSERVED LEVELS OF PUNISHMENT. (AUTHOR ABSTRACT MODIFIED)