Subversion in this study was defined as two or more prisoners getting together and deliberately acting to undermine good order. Four experimental groups of inmates (total sample size of 179) at Parkhurst Dispersal Prison participated in the study: nonsubversives, subversives, possible subversives, and those of uncertain status. The researchers hypothesized that the distribution of "subversive scores" for the nonsubversive group would be skewed to the lower end of the scale. The findings to date highlight four significant issues: that the youthfulness of a dispersal population is not necessarily a negative in terms of subversiveness, that institutional unrest is predictable from the concentration of subversives in one area, that prison riots are indeed often confined to one wing or a small number of wings, and that architectural or regime differences may account for differences between wings in one institution and between wings in different prisons. 2 figures