The purpose of this paper is to consider a recent account of the nature of retributive claims in the theory of punishment. This account, which appears in T. Honderich's book Punishment, The Supposed Justofcation, 1 will be criticized as a misrepresentation of classical retributive thinking. Before considering this account, it will be useful to first suggest the nature of a distinction the neglect of which seems responsible for Honderich's view: this is a distinction between critical and conventional roles of moral principles. The sense in which I wish to use these terms is a simple one. A principle fulfills a critical role when it is appealed to as a test of proposed solutions to moral questions. In its conventional role it is merely a part of the given phenomena which help define the problem and its context. The clearest illustration of these two roles, because it concerns a critical principle universally endorsed, is that of fairness. This may be appealed to as a justification of some social arrangement, in which case it plays a critical role; on the other hand, the fact of a morality of fairness in the community may be taken as a matter relevant among others when proposed social arrangements are submitted to the scrutiny of critical principles-including that of fairness. There is a certain trickiness in maintaining this distinction in some ethical discussions, because of the constant temptation to slide from one role into the other. One has to adopt an anthropological view of moral standards while at the same time being aware of a commitment to them as the proper rules for solving moral problems, and this raises the seemingly odd specter of being in a position to morally criticize our critical principles. But anyone who thus claimed to take an objective view of critical principles would not have grasped the point of the distinction. Whatever is the subject of critical appraisal is at that time fulfilling the conventional role; its role changes when it becomes the principle to which a critical appeal is made.2