Although the literature on the logic of historical enquiry is already vast and still growing, it continues to polarise overwhelmingly around a single disputed point—whether historical explanations have their own logic, or whether every successful explanation must conform to the same deductive model. Recent discussion, moreover, has shown an increasing element of agreement—there has been a marked trend away from accepting any strictly positivist view of the matter. It will be argued here that both the traditional polarity and the recent trend in this debate have tended to be misleading. The positiviste (it will be conceded) have been damagingly criticised. But their opponents (it will be suggested) have produced no satisfying alternative. They have tended instead to accept as proper historical explanations whatever has been offered by the historians themselves in the course of trying to explain the past. But a further type of analysis must be required (and will be attempted here) if some account is to be given of the status, and not merely the function, of the language in which these explanations are offered. Such an analysis, moreover (it will finally be suggested) has implications of some importance in considering the appropriate strategy for historical enquiries.

My aim in this article is threefold: First, I wish to challenge the view put forward by A. Flew and H. L. A. Hart which has won widespread recognition, namely, that the concept of punishment entails a legal or quasi-legal system of punishment. In my view all that is necessary for a case of punishment to occur is that a person (or body of persons) inflicts deliberately and not primarily for the sake of any beneficial consequences which may flow from his action, anything likely to cause discomfort to a victim who is capable of experiencing it, provided that the person inflicting the discomfort would claim, if asked to give a reason for his action, that he is inflicting it because of something (such as an act, event, state of affairs, or disposition) for which he holds the victim accountable. If I am right it follows that punishment is not necessarily retributive, and that utilitarian ‘punishment’ like reformative and compensatory ‘punishment’ is not true punishment, since the discomfort inflicted in these cases is merely ancillary to another end. Secondly, I hope to show that Hart does not succeed in discrediting the retributive theory, and that his treatment of punishment, though functional in aim, perpetuates the confusion between the meaning of punishment and its justification. Thirdly, I shall contend that retribution is not a proper aim of a legal system, whatever place it may actually have or ought to have in our moral code. Looking to the future, I shall venture to predict that the legal institution of punishment will sooner or later give way to a system designed to reduce crime and to rehabilitate offenders. Such a system will be adjusted to the needs of the community and of individual offenders, and will not involve the infliction of discomfort as an end. Hence, it would be wrong to call it a system of punishment.

In Part I of this paper I attempt to present, in more or less summary fashion, some well-known difficulties in the concept of deontic morality (i.e. the moralty of duty, obligation, what is morally wrong, and what one morally ought to do), as shown by certain features of deontic moral discourse. I make no great claims for originality here, although perhaps there may be some virtue in the presentation and ordering. In any case, Part I is a necessary preliminary to Part II, where I attempt to defend the rationality of and the necessity for deontic language against some recent (direct and indirect) attacks.

A. Pascal's statement of his wager argument is couched in terms of the theory of probability and the theory of games, and the exposition is unclear and unnecessarily complicated. The following is a ‘creative’ reformulation of the argument designed to avoid some of the objections which have been or might be raised against the original.