In some recent discussions of roll calls in Congress a model of interacting blocs has often been adopted and to a considerable degree verified. This model assumes the existence of several fairly cohesive blocs along with, perhaps, some unattached members. Furthermore, it is assumed that some of these blocs are fairly consistently opposed on roll calls, while others ally now with one side, now with the other. This model is attractive, not only because it accords with the usage of journalists, but also because it seems to provide a rational explanation of what sometimes appears to be the almost random confusion of Congressional voting behavior. As the evidence here presented suggests, however, this model is somewhat too neat and requires modification to account for shifting alliances over (often relatively short periods of) time. In a trial, reasoning from the assumptions of this model, we attempted to pick out those blocs and members who shifted from side to side. We were, however, unable to do so except in a few instances, largely, we believe, because the model as heretofore developed is static.