This study of Aristotle's central philosophical doctrines is animated by the idea that the notion of necessity is key to understanding the deep unity of Aristotle's philosophy and especially its cogent rationality. An important methodological assumption is that Aristotle's work forms a system, and can be investigated as such, because it displays ‘dynamic coherence’. Although Aristotle developed his philosophy over a lifetime, leaving many unresolved inconsistencies, he organised his writings according to a coherent didactic project. This is testified by the incessant signposting and cross-referencing the reader encounters in the treatises. Such indications can offer crucial clues for reconstructing the framework of the system and helping to smooth inconsistencies.

The material covered in the book includes the Organon (with De interpretatione getting the lion's share and the Categories deliberately marginalised), the general physical works, the Metaphysics and the Ethics. The analysis is mainly philosophical with textual problems occasionally dealt with in the footnotes. Aristotle's texts are accurately paraphrased and expounded, and most of the traditional interpretative problems are addressed with subtlety and competence. Moreover, the specialist focus on necessity has not prevented D. from offering a comprehensive and informative exposition of the doctrines involved. Even in the discussion of minute details, an extremely lucid presentation keeps the reader engaged and aware of each part's contribution to the big picture.

While admitting that Aristotle's necessity is characteristically ‘said in many ways’, the book's main ambition is to show that the system exhibits a unitary concept of necessity. D. sees no discontinuity between the logical necessity of the syllogism and the necessity that reigns in the different domains of Aristotle's cosmos. A distinctive claim of the book is that the necessity of syllogisms and the necessity of definitions rest on the Principle of Non-Contradiction (PNC). Hence the unity of the whole system and the interconnectedness of its parts ultimately derive from this principle. Necessity is not just found in the world, but imposed on it by the philosopher structuring experience through language, and language in turn, being a flexible tool, can be shaped by the philosopher to make it totally informed by the PNC. Aristotle is then a sort of idealist: experience becomes reality when, and in so far as, it is structured by a language sharpened up by the philosopher.

What about the first principles of an Aristotelian science? The most important are definitions of the form ‘X is YZ’. They enjoy absolute necessity, which is transferred in turn to the conclusions by syllogistic deduction. D. contends that this absolute necessity too is underwritten by the PNC. However, to say that the necessity of ‘X is YZ’ is the necessity of the PNC only means that for X it is impossible, i.e. self-contradictory, not to be YZ. This does not imply that the existence of X is absolutely necessary. So it would seem that we must recognise an altogether different concept of necessity, one that is beyond the limits of the legislation of the PNC. D. would not be fazed by this objection. The notion of being he attributes to Aristotle would not leave room for this separate claim of existence, since he argues that the copula, the verb ‘to be’ used to connect the subject and the predicate, and the attribution of absolute being – when we just say ‘X is’ as opposed to ‘X is Y’ – are all of a piece and mean determination. In predications of a term the predicate determines the subject, while ascriptions of absolute being mean that the subject is completely determinate in itself. Affirmative sentences have no existential import (the author sees no peculiarity in Aristotle's example with an empty subject ‘Homer is a poet’), and Aristotle's science is not so much interested in establishing the existence of its objects, but their full determination, which is expressed by the definitions. Their existence is already grasped by experience, albeit in a confused way, to be refined and polished by dialectic and diairesis, until it can be made fully determinate by scientific definition. The only absolute being in the sense of determination is substance, especially the primary substance or eidos of the Metaphysics, and its determination increases with its simplicity and uniformity. So if absolute necessity is absolute determination, one can see how it is ultimately secured by the PNC as well.

This doctrine of being as determination is interesting, but I fear it is not served particularly well by Metaphysics Z1, 1028a20–1, invoked by D. According to the account of predication proposed, the nominal part of the predicate always determines the subject, but in Z1 Aristotle claims that the inflected forms of non-substantial categories (e.g. the walking thing) are more of a being than the corresponding infinitival forms (to be walking) because their subject, a substance, is something determinate. Supposing with D. that this passage claims that being is determination, it seems to me that, if anything, it is the subject that lends determination to the predicate, not vice versa.

Somewhat surprisingly in a book on necessity, and in a book of such ambition, the least developed part is the section dealing with modal syllogistic, the logic of necessity and possibility. What I found wanting in its brief treatment is a philosophical assessment of the intense work that has recently been carried out on this subject. To mention only the most prominent example, M. Malink's interpretation of Aristotle's modal logic (Aristotle's Modal Syllogistic [2013]) is mentioned only in passing, in a footnote (p. 106 n. 1), and dismissed all too quickly. Since D. rejects the traditional charge that Aristotle made systematic confusion between de dicto and de re readings of modal propositions, it would have been helpful to do justice to Malink's much admired attempt to clear Aristotle of that blame. Instead, D. insists that necessity in Aristotle's account of modal propositions attaches to the copula (so the proposition is interpreted de dicto), and even when it seems to be part of the predicate (de re), it does not apply to the res predicated, but only to the predicative function of the predicate. Here D. should have paused to explain how the controversial syllogistic mood Barbara NXN, held by Aristotle to be valid, can be explained on the assumption that, however expressed, the modality always attaches to the copula.

Many original interpretations set forth in the book deserve discussion. A few of them are startling and unpalatable despite an able defence (e.g. the idea that even the primary substance of the Categories is a ‘function of predication’, pp. 60–1, contrast Cat. 5, 3a36), but in reviewing such a magnificent book it is only fair to end on a highlight.

This is the interpretation of Aristotle's discussion of future contingents and logical determinism in the much debated Chapter 9 of De interpretatione. D. argues that Aristotle is not attacking the Principle of Bivalence (‘every sentence is true or false’), but only a strong interpretation of this principle according to which, in a contradictory pair, one sentence is true and the other false. On this interpretation, the principle harmlessly holds for declarative sentences about the past and the present, but implies determinism when extended to sentences about the future. From the point of view of standard modern logic, this is the only viable interpretation of the Principle of Bivalence; so the only way to avoid determinism would be to restrict this principle and accept that sentences about future contingents are simply neither true nor false. Several interpreters think that this is also Aristotle's conclusion. By contrast, D. argues that Aristotle's strategy is not to show that the scope of Bivalence must be narrowed, but rather to define this principle as the claim that every sentence is true-or-false, where the hyphens indicate that the alternative must be read sensu composito. In the case of sentences about a contingent future, this weaker formulation only requires that they are bound to get their truth value when the time comes, regardless of the fact that now they are still neither true nor false. In this way Aristotle can concede that at the time of the utterance sentences about the future lack truth value, and claim at the same time that Bivalence holds for all declarative sentences, whatever their tense. I think this suggestion deserves serious attention. There is a chance that some progress has been made on one of the most discussed texts in Aristotle.