Search

(P1) is what Evans (1982) calls the Generality Constraint. It holds that mental states with conceptual content must be closed under all meaningful recombinations of the constituents of the sentences that best express them. For example, anyone who can conceptually think that Albert is friendly and that Bob is gracious must also be able to conceptually think that Albert is gracious and that Bob is friendly. The Generality Constraint does not deny that some mental states might fail to be systematic. It just maintains that non-systematic mental states will lack conceptual content. Gillett thus seems to misunderstand my argument when he claims that I ‘argue that the [generality] constraint is not definitive of conceptual states in the way that Evans claimed’ (Gillet 2014, p. 1147). On the contrary, I assume that the Generality Constraint is a necessary truth and use it as a premiss in my argument for (C). At no point do I deny that the Generality Constraint is definitive of conceptual content.

Among Kant's innovations in the understanding of logic (‘general logic’) were his claims that logic had no content of its own, but was the form of the thought of any possible content, and that the unit of meaning, the truth-bearer, judgement, was essentially apperceptive. Judging was implicitly the consciousness of judging. This was for Kant a logical truth. This article traces the influence of the latter claim on Fichte, and, for most of the discussion, on Hegel. The aim is to understand the relations among self-consciousness, reason and freedom in the idealist tradition.

I. IntroductionIt has often been seen as a crucial feature of modernity that the old chains of necessity-religious, political, social, or otherwise-have been shaken off and human beings have been released into a radical form of freedom. This kind of freedom has both a positive and a negative side: on the one hand, it may seem to allow us to make a kind of individual decision and to act according to our own will. On the other hand, it also brings with it the dire difficulty of dealing with what we can call the problem of contingency: how to decide what to do if it is equally open to us to perform a certain action as well as its opposite, if there is no necessity, no sufficient reason that tells us to do the one but not the other? In this paper, I argue that this problem of contingency has a prominent place in Robert Musil's The Man Without Qualities. I do not claim that Musil's novel suggests any definite answers, but rather that it presents and explores different strategies for dealing with contingency.Given the limits of this paper, I will not be able to give an exhaustive account of this problem in Musil's novel; a lot of the important ethical implications will have to be left out, as, for example, the relation of this problem to any form of violence. The paper will start by clarifying the notion of contingency as it is taken up in the novel and its connection to what Musil calls the sense of possibility. Subsequently, I will show how Kakania, the "servants" and citizens of this state, deal with contingency. The Parallel Campaign will be argued to be one big attempt to ground this state in some kind of necessity. We will then move on to Ulrich's way of dealing with contingency: with Ulrich, Musil introduces the thought experiment of assuming a person who is fully aware of the challenges that contingency raises for human agency and who attempts to face up to them. Finally, I will contrast Ulrich's way of dealing with contingency with the way of some of the other characters in the novel.II. CONCEPTUAL CLARIFICATIONS: CONTINGENCY AND THE SENSE OF POSSIBILITYLet me first clarify the notion of contingency at work in this paper somewhat further, which will help us in our analysis of the different contingent phenomena sketched in the novel. I want to understand contingency here as the opposite of necessity1-so as a first pass we can understand it positively with the help of notions like chance, indeterminacy, or possibility.In The Man Without Qualities, necessity can be found in three different spheres:2 there is (1) logical necessity-Ulrich's ideas for new ways of living our lives are based on his attempts to follow only logical necessity; (2) causal necessity-referred to in the discussions of the laws of nature;3 and, most importantly for the great majority of citizens, (3) necessity within the sphere of politics or society. The rules and regularities of society are explicitly understood as a form of necessity by certain figures of this novel; for example, Bonadea considers the changes of fashion as following a form of necessity,4 and initially the regulations of the military have the same status for Stumm von Bordwehr.This last form of necessity-political or social-is put into doubt by what Musil calls the "sense of possibility," which takes this alleged necessity as just one possibility among others. We get a famous account of this sense right at the start of the book. Chapter 4, entitled "If there is a sense of reality, there must also be a sense of possibility," introduces the sense of possibility in the second paragraph:Whoever has it does not say, for instance: Here this or that has happened, will happen, must happen; but he invents: Here this or that might, could, or ought to happen. If he is told that something is the way it is, he will think: Well, it could probably just as well be otherwise. So the sense of possibility could be defined outright as the ability to conceive of everything there might be just as well, and to attach no more importance to what is than to what is not. …

The overgeneration argument attempts to show that accepting second-order validity as a sound formal counterpart of logical truth has the unacceptable consequence that the Continuum Hypothesis is either a logical truth or a logical falsehood. The argument was presented and vigorously defended in John Etchemendy’s The Concept of Logical Consequence and it has many proponents to this day. Yet it is nothing but a seductive fallacy. I demonstrate this by considering five versions of the argument; as I show, each is either unsound or lacks a troubling conclusion.

The Tarski T-Schema has a propositional version. If we use ϕ as a metavariable for formulas and use terms of the form that-ϕ to denote propositions, then the propositional version of the T-Schema is: that-ϕ is true if and only if ϕ. For example, that Cameron is Prime Minister is true if and only if Cameron is Prime Minister. If that-ϕ is represented formally as [λ ϕ], then the T-Schema can be represented as the 0-place case of λ-Conversion. If we interpret [λ…] as a truth-functional context, then using traditional logical techniques, one can prove that the propositional version of the T-Schema is a tautology, literally. Given how well-accepted these logical techniques are, we conclude that the T-Schema, in at least one of its forms, is a not just a logical truth but a tautology at that.

Deflationists about truth seek to undermine debates about the nature of truth by arguing that the truth predicate is merely a device that allows us to express a certain kind of generality. I argue that a parallel approach is available in the case of logical consequence. Just as deflationism about truth offers an alternative to accounts of truth's nature in terms of correspondence or justification, deflationism about consequence promises an alternative to model-theoretic or proof-theoretic accounts of consequence's nature. I then argue, against considerations put forward by Field and Beall, that Curry's paradox no more rules out deflationism about consequence than the liar paradox rules out deflationism about truth.

A necessarily true sentence is ‘brute’ if it does not rigidly refer to anything and if it cannot be reduced to a logical truth. The question of whether there are brute necessities is an extremely natural one. Cian Dorr has recently argued for far-reaching metaphysical claims on the basis of the principle that there are no brute necessities: he initially argued that there are no non-symmetric relations, and later that there are no abstract objects at all. I argue that there are nominalistically acceptable brute necessities, and that Dorr's arguments thus fail. My argument is an application of Godel's first incompleteness theorem.

Truthmaker theorists claim that for every truth, there is something in virtue of which it is true—or, more cautiously, that for every truth in some specified class of truths, there is something in virtue of which it is true. I argue that it is hard to see how the thought that truth is grounded in reality lends any support to truthmaker theory.

I define ‘skim semantics’ to be a Davidson-style truth-conditional semantics combined with a variety of deflationism about truth. The expressive role of truth in truth-conditional semantics precludes at least some kinds of skim semantics; thus I reject the idea that the challenge to skim semantics derives solely from Davidson's explanatory ambitions, and in particular from the ‘truth doctrine’, the view that the concept of truth plays a central explanatory role in Davidsonian theories of meaning for a language. The fate of skim semantics is not determined by the fate of the truth doctrine, so rejecting the truth doctrine does not in itself open the way to skim semantics. I establish my thesis by showing that some recently proposed versions of skim semantics fail because of truth's expressive role. I also discuss the conditions that might permit skim semantics.

In response to Hoeltje I concede the main point of his first section: for each logical truth S of the object language, it is a logical consequence of the Davidsonian theory of meaning I offered in my paper that S is logically true, contrary to what I asserted in the paper. However, I now argue that a Davidsonian theory of meaning may be formulated equally well in such a way that it not a logical consequence of the theory that S is a logical truth. Nonetheless, the revised theory of meaning will still ‘entail’ in a wider sense that S is a logical truth, for we can prove by induction on the consequence class of the revised theory that S is a logical truth. So far, my disagreement with Hoeltje is over the more charitable interpretation of a passage from Davidson. I close by arguing that Davidson was mistaken on one point, a theory of meaning will entail a threefold distinction among the sentences of the object language, not a twofold distinction as he claimed.

Donald Davidson has claimed that for every logical truth S of a language L, a theory of meaning for L will entail that S is a logical truth of L. Jim Edwards has argued (2002) that this claim is false if we take ‘entails’ to mean ‘has as a logical consequence’. In this paper, I first show that, pace Edwards, Davidson's claim is correct even under this strong reading. I then discuss the argument given by Edwards and offer a diagnosis of where he went wrong.

D.M. Armstrong has objected that the dispositionalist theory of laws and properties is modally inverted, for it entails that properties are constituted by relations to non-actual possibilia. I contend that if this objection succeeds against dispositionalism, then Armstrong's nomic necessitation relation is also modally inverted. This shows that at least one of Armstrong's reasons for preferring a nomic necessitation theory is specious.

The goal of this paper is to make sense of relativism about truth. There are two key ideas. (1) To be a relativist about truth is to allow that a sentence or proposition might be assessment-sensitive: that is, its truth value might vary with the context of assessment as well as the context of use. (2) Making sense of relativism is a matter of understanding what it would be to commit oneself to the truth of an assessment-sensitive sentence or proposition.

We introduce an abstract algebraic structure — a lattice defi ned on a generalized truth value space of constructive logic. For background one can refer to the idea of ‘under-determined’ and ‘over-determined’ valuations (Dunn), a ‘useful four-valued logic’ (Belnap), and the notion of a bilattice (Ginsberg). We consider within one general framework the notions of constructive truth and constructive falsity, as well as the notions of non-constructive truth and non-constructive falsity. All possible combinations of th e basic truth values give rise to an interesting ‘16-valued logic’. It appears that these 16 truth values constitute what we call a trilattice — a natural mathematical structure with three partial orderings that represent respectively a n increase in information, truth and constructivity. The presentation of the paper is essentially conceptual: the stress is laid on introducing new concepts and structures as well as on their general interpretation.

Mixed inferences are a problem for truth pluralism, a doctrine which aims at combining truth assessability and anti-realism with respect to allegedly non-descriptive sentences, such as moral sentences. It seems that truth pluralists have to give up the classical account of validity. Against this,JC Beall suggests that truth pluralists can adopt the account of validity usedin many-valued logics: validity can be defined as the conservation of designatedvalues. The problem, I argue, is that there is ground to believe that on this account sentences which are true in one or other way also fall under a generic truth predicate. I also argue that mixed conjunctions make for a further problem for truth pluralism, and more particularly for Beall's version of it. Finally, I consider the deeper worry that the distinction between truth which does and truth which does not entail realism is inferentially irrelevant.

Hale (1999) presents an argument from McFetridge (1990) for his conclusion, and states (1999, p. 32) that there is an obvious objection. His paper aims to supplement McFetridge's argument to meet that objection. He concludes that this establishes the necessity of necessity. Below are two independent complaints against Hale. The first is directed against McFetridge's argument. It is independent of Hale's "obvious objection". The second is directed against Hale's reply to that obvious objection.

Dominic Hyde (1997) has presented a new logical system for vague words that employs paraconsistent logic. He called it subvaluationism, in analogy with supervaluationism Fine (1975) and others have made popular. Hyde maintained that subvaluationism is substantially different from supervaluationism, and is at least as good for a logic of vagueness. In this note, I shall argue for the following: First, we may reasonably take Hyde's so-called subtruth not as truth simpliciter, as Hyde does, but just as possible truth in a certain sense of "possible". Second, we also may take supertruth in supervaluationism as the dual of subtruth, a kind of necessary truth; we do not have to take it as truth simpliciter, either. We can regard superand subvaluationism essentially as duals. Finally, if we interpret superand subtruth this way, we can make inference rules for vague words much simpler and in compliance with classical logic. The logic of superand subvaluationism can be regarded not as an alternative to classical logic but as an extension of it. We do not need non-classical logic to understand vagueness. Superand subvaluationism make use of the notion of (admissible) precisification. An admissible precisification of a vague predicate (or singular term) is an admissible interpretation under which a particular precise extension (or denotation) is assigned to the word. Even a vague sentence is either true or false on an admissible precisification. (In what follows I shall drop the adjective "admissible" for the sake of simplicity and just say "precisification" when I mean "admissible precisification". What is admissible and what not has been a matter of controversy, but I set aside that problem.) According to supervaluationism, a sentence is supertrue if and only if it is true on all precisifications, superfalse if and only if it is false on all precisifications, and neither supertrue nor superfalse if and only if it is true on some precisifications and false on the others. According to subvaluationism, a sentence is subtrue if and only if it is true on some precisifications, and subfalse if and only if it is false on some precisifications. If a sentence is true on some precisifications and false on the others, it is considered both subtrue and subfalse. In both theories, all logical truths are superor subtrue because they are true on all (thus, some)

Must we believe in logical necessity? I examine an argument for an affirmative answer given by Ian McFetridge in his posthumously published paper "Logical Necessity: Some Issues", and explain why it fails, as it stands, to establish his conclusion. I contend, however, that McFetridge's argument can be effectively buttressed by drawing upon another argument aimed at establishing that we ought to believe that some propositions are logically necessary, given by Crispin Wright in his paper "Inventing Logical Necessity". My contention is that Wright's argument, whilst it likewise fails, as it stands, to establish the necessity of necessity, establishes enough to close off what appears to me to be the only effective-looking sceptical response to McFetridge's original argument. My paper falls into four principal parts. In the first, I expound McFetridge's argument and draw attention to the possibility of a certain type of sceptical counter to it. In the second, I begin a response to this sceptical move, taking it as far as I can without reliance upon argument of the kind given by Wright. Turning, then, to Wright's argument, I explain to what extent I think it is successful and seek to rebut some objections to the argument which, were they well-taken, would show that the argument cannot enjoy even the partial success I wish to claim for it. Finally, I return to my main theme and try to show, with the assistance of what I take to be solidly established by Wright's argument, that the sceptical response collapses.

For a large class of cases-though not for all-in which we employ the word “meaning” it can be defined thus: the meaning of a word is its use in the language . Wittgenstein (1953, §43) The purpose of this paper is to defend Wittgenstein's idea – his so-called “usetheory” of meaning – against what is perhaps the most influential of the many arguments that have been levelled against it. I'm thinking of Kripke's critique of “dispositionalism”, which is a central component of his celebrated essay, Wittgenstein on Rules and Private Language. Kripke argues that meaning a certain thing by a word is not a matter of being disposed to use it in a certain way. And his argument has been well-received. Most commentators, whatever they say about Kripke's overall line of thought (leading up to his “sceptical conclusion” about meaning), tend to agree at least that the use-theory has been elegantly demolished. My main objective is to combat this impression. Just what Wittgenstein himself had in mind is not entirely clear; but that's not my topic. Rather, what I want to do here is to explore and support a certain version of the use-conception of meaning – one which seems to me to have some attractive features (and which I believe can be pinned on Wittgenstein).

Non-cognitivism in ethics holds that ethical sentences are not in the business of being either true or false-for short, they are not truth apt. No-truth theories of indicative conditionals (on one labelling of the relevant class of conditionals) hold that indicative conditionals have assertability or acceptability conditions, but not truth conditions; they are not truth apt. The arguments for these views are typically local to ethics and conditionals, respectively. They are not usually set within a specific theory of truth, and the question of how they connect to the various theories of truth is typically left unaddressed. This is surprising. ' An obvious question to ask about non-cognitivism and no-truth theories of conditionals is how they fare in the light of various views about truth. This paper is concerned with an increasingly popular position on the obvious question. According to this position, a certain view about truth, often called "minimalism about truth", leads pretty well immediately, and in any case without recourse to the considerations distinctive of the debates concerning ethics and conditionals, to the conclusion that ethical sentences and indicative conditionals are truth apt. A variant on this position distinguishes kinds of truth, and holds that thin truth, or disquotational truth, or at any rate some core, non-robust notion of truth, cannot sensibly be denied to ethical sentences and conditionals, and that, accordingly, the live issue should be thought of as whether, on some thick or robust notion of truth-perhaps tied to correspondence with reality in some metaphysically heavyweight sense, or to mind independence, or to evidence transcendence, or ...-ethical sentences and conditionals are truth apt.2 In this paper, we argue that this is a mistake. The distinctive considerations cannot be sidestepped. The paper divides into six main sections. In ? 1 we characterise minimalism about truth and argue that the usual arguments from minimalism about truth to

obtain at a world w, if and only if w is a member of (the set) P. PWA provides intuitively compelling accounts of necessity, contingency, and impossibility. Conjunction and disjunction are representable as intersection and union. We can say that P entails Q just when P c Q, thereby explaining why the objects of thought stand in entailment relations; and it turns out that P entails Q just when the material conditional (P :D Q) is a necessary truth. Kripke's [11] account of names as rigid designators fits well with PWA. And following Stalnaker [18], one can embed PWA in a 'pragmatic' account of intentional states: desiring P is being disposed to perform actions that would tend to make P obtain in a world in which one's beliefs, whatever they are, were true; believing P is being disposed to perform actions that would tend to satisfy one's desires, whatever they are, in a world in which P and one's other beliefs were true.' This account has

It is often held that laws of nature bear a special relation to counterfactual conditionals. In particular, it is suggested that one should retain, under a counterfactual antecedent, all of those claims that one believes to be law-statements, if the antecedent is consistent with all of those claims. I'll call this principle 'nomic preservation': If p q (read 'Had p obtained, then q would have obtained') and p is consistent with L, the conjunction of all and only the statements of natural law, then pq and if p is consistent with L and it is a logical truth that (pL neither Lewis nor Slote denies that it would. Slote's counterfactual seems strange because our standard reason for contemplating a counterfactual antecedent, such as 'Had I been depressed yesterday', is not to determine whether every past moment would then have been different in some respect or other, but to consider some particular moment and to examine whether it would then have been different in some particular way. Had I been depressed yesterday, would I have cried all day or phoned you? Had I been depressed sometime yesterday, would someone have to have first reminded me of the famine in Somalia, since I was in such a good mood?

In his monograph Parts of Classes (1991), David Lewis argues that mereology, the calculus of whole and part, applies to classes. What are the parts of classes on his view? They are all classes themselves: subclasses of the classes in question. It follows that singletons, that is, unit classes (I will use the two phrases interchangeably), are mereological atoms. This enables Lewis to define class membership in a striking new way. The relation of member to its singleton is taken as primitive. Membership of classes generally is then defined by saying that x is a member of a class y if and only if the singleton of x is a part of y. ' This leads him to focus his attention onto the relation of member to singleton. He finds it a very mysterious one. At the present stage of philosophical enquiry it is a pretty mysterious relation on anybody's view. But it is particularly mysterious for Lewis. He argues thus. If a singleton class is a mereological atom, then it can have no internal structure. Very many members of singletons will have an internal structure, some a very complex internal structure. But in each case the singleton itself will have no internal structure. Using this lack of structure as a premiss, Lewis draws the conclusion that singletons are "wholly distinct" from their members. This conclusion of Lewis's appears to involve taking up one of at least three positions about the location of classes: (i) they are not located, except perhaps at structureless points or point-instants; (ii) they are located where their members are located, but are not located where the proper parts of these members are located; (iii) they are located-wholly-where each part, proper or improper, of their members is located. I am not sure that solutions (ii) and (iii) are available where the member is a relation, at any rate where relations are not reduced in some way (say to classes!). All these views seem rather unattractive. But a more important problem for Lewis is this. Member and singleton are wholly distinct. Yet it appears to be a necessary truth that, given an individual, its singleton exists. And is it not still more obviously a necessary truth that, given a singleton, its member exists? Here we seem to have a symmetrical necessary connection between wholly distinct existences. For myself, I agree with David Hume

Quine's 'Two Dogmas of Empiricism'" proclaimed a more radical empiricism than anything before it. Beginning from the observation that 'our statements about the external world face the tribunal of experience not individually but only as a corporate body', Quine inferred that since our theories of the world are tested only as wholes there will always be a choice as to which part of a theory to revise when experience proves 'recalcitrant'. Moreover, he claimed, no part of the theory is in principle immune from revision. Since, then, logic and mathematics do undoubtedly form part of our physical theories, they too are subjected to the test of experience no less than any other parts, and it is conceivable that they should fail the test no less than other parts. That is, it might turn out that we have good reason to revise them, when seeking for a better theory to replace one that has proved unsatisfactory, and it is a mistake to regard them as protected by their alleged a priori status. I shall argue first that this claim of Quine's should be accepted, and that logic and mathematics are therefore in this sense empirical. But I shall also observe that there appears to be another sense in which they are not empirical, but remain just as problematic as ever. I shall confine the discussion to the simplest part of logic, namely propositional logic. If the case can be made out here, then it will apply a fortiori to more complex areas of logic and mathematics. I remark also that, if we are prepared to accept a Kripkean separation between a priori truth and necessary truth, then of course we may consistently hold that the truths of logic are both necessary and empirical. But I shall not be concerned with the question whether these truths are necessary, only with the point that we think of them as necessary. This, evidently, is the immediate objection to the Quinean thesis. We do not in fact suppose that empirical investigations are in any way needed to establish the correctness of an argument that can be certified as valid simply by the methods of logic, for example by a truth-table calculation. And we do not see how any experiences, however 'recalcitrant', could upset such methods.

The doctrine of middle knowledge was developed by the Spanish Jesuits Luis de Molina and Francisco Suarez in the late sixteenth century, and immediately became the subject of fierce controversy, being bitterly opposed by the Dominicans, especially Domingo Bafiez and Diego Alvarez, and requiring the intervention of the Vatican (in 1607) to cause the bitterness to subside. As recent work has shown, the intellectual interest of the doctrine extends far beyond that of an inhouse debate in the Catholic Church of the Counter-Reformation; for it touches on issues of central importance not only in analytic theology, but also in philosophical logic. According to Molina, there are three moments in God's foreknowledge of the history of the world which He creates.' These moments are not to be thought of as stages in a temporal series, but as ordered by the relation of logical or conceptual dependence of later on earlier.2 In the first moment, God knows by natural knowledge all metaphysically necessary propositions. These include not merely truths of logic and mathematics, and natural laws, but also facts about which contingent states of affairs are possible, since that a state of affairs is possible is itself a necessary truth. God's natural knowledge would suffice to equip Him with complete foreknowledge of the history of His creation, if the world He created were entirely deterministic, and if He undertook to refrain from free intervention in that world. The creation

i. In The Possibility of Altruism, Nagel argues that an awareness of the benefit or harm an action will cause to another can in itself, without our having to assume the agent to possess any specific desires he might conceivably not possess, move an agent to perform or desist from that action. Hence his claim to have demonstrated the possibility of altruism, the possibility of acting simply out of regard for the welfare of another. I want to suggest that it is possible to argue for something a little stronger: not, perhaps, the necessity of altruism but at any rate something more than its bare possibility. It may be that something stronger follows from Nagel's own arguments, but that is not a possibility I shall explore. I want to focus upon the more specific question of our response to the distress of others. I shall argue that it is a necessary truth that to perceive another's distress as distress is to perceive it as prima facie a bad thing, as other things being equal something to be avoided or prevented. This does not mean that to perceive another's distress is necessarily to be moved to prevent it. What it does mean is that where someone is not moved to prevent distress, where they are unmoved or are moved in some other way-amused, excited, pleased or whatever, we must either deny that they are genuinely aware of the other's distress or find, in the special character of their view of that person in particular or the world in general, an explanation of why they should take what is prima facie an evil to be amusing, exciting, gratifying, or whatever.

Divine command theories of morality have not been particularly popular among philosophers for some time. Consequently, such theories have received very little serious attention in recent years. One notable exception to this rule has been Robert M. Adams, who on several occasions has defended what he calls a modified divine command theory of ethical wrongness (1973, 1979a, 1979b). Most recently he has formulated this theory in terms drawn from current work on the theory of meaning. Specifically, he has proposed an account of the meaning of the term "wrong" similar to Hilary Putnam's account of the meaning of natural kind terms such as "water" and "gold." In this paper I will (1) describe the major features of Putnam's account of the meaning of natural kind terms, (2) describe Adams's analogous account of the meaning of the term "wrong," (3) discuss what I take to be some advantages of Adams's theory, (4) discuss two problems with it, and (5) propose an alternative which retains the advantages and avoids the problems discussed here. According to Adams it is a necessary truth (if it is true at all) that any action is ethically wrong if and only if it is contrary to the commands of a loving God. But, and this is where the new theory of meaning comes into play, this is not an analytic proposition. It is not the result of an analysis of a concept or of moral language. Nor is it a proposition that is known a priori. It does not represent what everyone thinks that wrong actions are. Nor does it represent how everyone recognizes wrong actions for what they are. Nor does it represent just the meaning of the concepts, or language, of Judaeo-Christian believers. Rather, it represents the actual nature of wrong actions. As such, the proposition is necessarily true (if it is true at all), although it is a synthetic one and epistemically a posteriori.

Professor George Fletcher's book Rethinking Criminal Law' throws out a challenge to those concerned with criminal jurisprudence to re-examine the critical apparatus with which they analyse and explain the criminal law. He examines most areas of the substantive criminal law with a view to 'reworking the apparatus with which we think about criminal liability' (p xxi) and invites his readers to carry his enterprise forward. Nobody who is interested in the criminal law can treat his challenge lightly. The book is encyclopaedic, spilling over with suggestions, ideas and insight that take arguments about the criminal process onto a different plane. In doing that, Fletcher earns our thanks and our admiration. But when we come to ask how, for example, the central distinction that he wishes to make between definition and justification might be applied to a crucial problem of the criminal law, such as the role of mistake as a defence, the suggested prescriptive pattern provides no helpful guidance. This part, at least, of the new apparatus must be approached with considerable caution. In his quest for unifying themes, Fletcher presses us to a too rigid view of the effect of mistakes. Indeed, I would suggest that his quest for a 'theory of mistake'2 is itself misguided, and that mistakes occur in too many different guises within the criminal law to be susceptible to a system of classification of the sort that is suggested. There is a practical purpose in isolating this aspect of Fletcher's work. The English courts have recently insisted, on a number of occasions, that in order to exculpate, the mistake made by a putative criminal must have been a reasonable one.3 In doing this, the courts appear to be resiling from the position adopted by the House of Lords in Morgan4 where it was held that any mistake defeating the required mental element must exculpate as a matter of logical necessity, whether it was a reasonable one or not. My question is, does Fletcher's rethinking enable us to see perennial questions such as these in a new perspective? Does it confirm and reinforce what we already know, or articulate reasons that justify and explain our present practices? These are, perhaps, difficult questions to be posing of new theories. Yet if a rethinking is to be worthwhile at all, it must perform some of these tasks.