This is an excellent book. Written with clarity and verve, it treats the ontology of music as an issue in mainstream analytic metaphysics: it forces philosophers of art to face up to the commitments of their views, and encourages those working in general metaphysics to test their ontologies on the special case of musical works. I suspect that most people will, like me, reject the book's main conclusions, but this makes it no less worth reading. On the contrary: the book provides such an eloquent and robust defence of its own unorthodox position that it is especially important for those of us who disagree to come to terms with it. Its uncompromising insistence on laying assumptions bare makes reading it an ideal way of coming to understand what is at stake in the issues it is concerned with.

Julian Dodd is concerned with all kinds of works of music which can be performed or played on several occasions. Pure improvisations are not included, but instrumental music of all other kinds is. Works of music, in this sense, are ontologically puzzling for two reasons. First, they have no obvious spatial location. (There is surely something wrong with Heidegger's claim that ‘Beethoven's quartets lie in the storerooms of the publishing house like potatoes in a cellar’ (Heidegger Basic Writings (D. Krell (ed.), Harper Collins, 1994), 145).) And, secondly, they can be performed or played on several occasions. What kind of thing could a work of music be, then, if it has these two puzzling features?

Dodd takes a full answer to this question to contain two components. First, we need to say what ontological category musical works fall into, and secondly, we need to sketch the identity conditions of musical works within that category. For the first part of the answer, Dodd defends a particularly clear-cut version of the type-token theory: musical works are types of sound-sequence-event (performances are their tokens). And for the second part, Dodd defends a form of sonicism: the identity of musical works depends on nothing but their sound (so there could not be two different works which sounded the same). These two parts together constitute what Dodd calls the simple view, clearly favouring the honest virtues of simplicity over the tricksy sophistication of the obvious rivals.

Dodd is not the first to defend a form of the type-token theory. This has often seemed a natural way of explaining the second key feature of musical works: the fact that the same work can be performed on several occasions – which Dodd calls their repeatability. (The view can be found in Richard Wollheim (Art and its objects, (Cambridge University Press, 1980), 74–84)); the most prominent recent advocate is Jerrold Levinsion (Music, Art and Metaphysics (Ithaca: Cornell University Press, 1990).) What is distinctive is Dodd's unflinching characterization of types: they are, he argues, abstract, unstructured, unchanging, modally inflexible, and eternally (not timelessly) existing. Dodd is a Platonist about types, and consequently a Platonist about works of music. One particular consequence of this view is that musical works cannot be created. As for those who have wanted to hold that musical works are types, while still holding that they come into existence when they are composed – this includes both Wollheim and Levinson, among many others – these people, Dodd thinks, fail to face up to the commitments of their view. But Dodd does not just leave us with his Platonism baldly stated: he gives an account of composition as discovery, which is consistent with musical works' existing eternally; and he offers an account of the interpretation of musical works which does not conflict with their being types.

The categorial debate is not settled just by fending off type-token theorists who fail to grasp the nettle of Platonism, of course. There are also those who think that musical works are entities of a different kind altogether. Dodd considers two such alternatives: the idea that musical works might be continuants – entities of the same general kind as ‘people, planets, trees, and stones’ – and the view that musical works might be compositional actions. Against the continuant view (defended, for example, by Guy Rohrbaugh (‘Artworks as Historical Individuals’ European Journal of Philosophy 11 (2003), 177–205)), Dodd argues that it cannot properly explain one of the key features of musical works from which the enquiry began – the feature he calls their repeatability. The continuant view has to understand the relation between a musical work and performances of it as one of embodiment: the work is embodied in the performances, which (on Dodd's view) are occurrences of the work. The basic problem is that something can depend for its existence on other entities, in the manner in which a continuant depends upon what constitutes its embodiment, without those other entities counting as occurrences of the original thing. But that means that the continuant view has nothing to offer in explanation of the distinctive repeatability (as Dodd understands it) of musical works. Consequently, it fails even to address one of the principal sources of our puzzlement about the nature of musical works, and so is in a markedly worse position than the type-token theory.

Dodd considers two versions of the view that musical works are compositional actions: the view that they are action-types (defended by Gregory Currie (An Ontology of Art (New York: St Martin's Press, 1989)), and the view that they are particular actions (advocated by David Davies (Art as Performance (Oxford: Blackwell, 2004)). Dodd argues that both versions have difficulty making sense of the everyday thought that when we listen to a piece of music right through we hear the whole work. Nor can either theory explain how anyone can perform a whole work (except perhaps, in the case of the action-type theory, by composing it again). This last is a particularly serious problem, since it puts compositional-action theories in the same awkward position as continuant theories: they cannot explain one of the principal things that an account of the nature of musical works needs to be able to explain. And Davies' theory faces an additional problem, according to Dodd: it cannot make sense of the natural thought that the same work could have been composed at a different time, without resorting to what Dodd calls ‘ontological obscurantism’ (194).

Having settled the categorial question in favour of a Platonist type-token theory, Dodd then turns to the question of the individuation of musical works. Here he defends what he calls timbral sonicism: the view that in the case of musical works identity is acoustic identity. Dodd motivates timbral sonicism on the basis of a certain ‘moderate empiricism’ about music, according to which the aesthetic properties of a work of music supervene upon its acoustic properties (roughly: no aesthetic difference without an acoustic difference). And he takes this moderate empiricism as the default position, the view that strikes as us intuitively natural. But it is plausible that the ultimate motivation for Dodd's timbral sonicism is that it is the natural partner of the Platonist type-token theory which he embraces in response to the categorial question. This becomes apparent when we consider the two sources of objection to timbral sonicism which Dodd addresses: the views he calls instrumentalism (according to which its instrumentation is essential to a work's identity) and contextualism (according to which a work's location in its musico-historical context is essential to it). It is hard to make sense of instrumentalism without some broadly teleological conception of works (being composed for the piano, for example), and it is hard to make sense of any such teleological conception within Dodd's Platonist type-token theory. And similarly, it is hard to see how historical location can be essential to something which exists eternally, as the Platonist type-token theory holds that musical works do.

When this is appreciated, it is clear that the view of musical works which Dodd calls the simple view – the view comprising a Platonist type-token theory together with timbral sonicism – is not just a splicing-together of two unrelated parts. The view is almost an organic whole, and, as such, presents itself, as a whole, as the clearest and most obvious position to be considered by anyone engaged in musical ontology. Dodd concludes, with characteristic flourish: ‘When it comes to the ontology of music, the simple view is not merely a serious player: it is the best game in town’ (276).

When we consider what other games are to be found in this particular frontier zone, I think Dodd is right, and has, indeed, shown that he is right. I might quibble over the odd argument, and there is, perhaps, an over-reliance at times on appeals to untutored intuition, as well as more weight being sometimes placed on theoretical neatness than it can easily bear. But I am convinced that Dodd's Platonism is the only properly intelligible version of the type-token theory, and that the type-token theory is preferable both to the continuant theory and to the different versions of the compositional-action view, for reasons which at least include those which Dodd adduces. Dodd's view has an integrated coherence which none of these alternatives can match.

And yet. And yet the view strikes me as mad. Let us begin from the back, with a question about Dodd's rejection of contextualism. It is often important to works of music that they refer to, cite, quote, or parody other pieces of music. Thus it is crucial to Schumann's C major Fantasy, op 17, that it works with the theme of the last of Beethoven's ‘An die ferne Geliebte’ songs. Dodd may be able to dismiss the fact that Schumann used this as a kind of love letter to his own distant beloved as of no more than art-historical interest. But the same cannot easily be said of the fact that Schumann uses this theme as a form of nostalgic tribute to Beethoven, the nostalgia being intensified by his apparently deliberate decision not to jolly himself out of the theme's wistful mood in the way that Beethoven himself did. This kind of reference and quotation is ubiquitous in music. Whenever it occurs, it seems crucial to the meaning of the later work that the earlier piece is already there to refer to, but this is something that does not supervene on the acoustic properties of either work.

Dodd's defence of timbral sonicism against instrumentalism may seem all very well – until we consider the position of the performer, rather than the listener. An important part of the point of Chopin's Black Keys Etude, op 10, no 5, is that the right-hand part is to be played almost exclusively on the black keys of a piano. And it has been said that Schubert's sexuality can be determined from what is involved in playing his piano four-hands music, once it is also known whom he expected to play it with. In fact, it is unclear what sense the simple view can make of performance in general. It seems central to performance that the work itself is the object of the performer's attention: it is the work which the performer aims to understand and interpret. This means that the ‘of’ in the phrase ‘performance of a work’ is an intentional ‘of’, which is precisely not the case for the ‘of’ in the phrase ‘token of a type’. Since the point of the type-token view is to understand the ‘of’ in ‘performance of a work’ in terms of the ‘of’ in ‘token of a type’, that must put all versions of the type-token view in question.

For all that, the simple view really is ‘the best game in town’. What I think this means is that we need quite a major revision of the way we set about ontology. What Dodd does is approach the issue of the nature of musical works with the general shape of his ontology already in place. It is an ontology which is meant to be adequate to characterizing the world, as the world has been encountered before art is considered. He then asks, in effect: to which of these categories of entity, which we have reason to accept before we consider art at all, do musical works belong? And the answer ‘To none’ is not even contemplated. Dodd is not unusual in this: all those whom he discusses here do the same. Nor is the ontology of art an unusual place to apply this kind of approach: it is the dominant approach to ontological questions in all fields. But it seems to me that if this approach serves up the simple view, quite rightly, as ‘the best game in town’, it must be wrong.

Here is one way of trying to find an alternative, which is not just a matter of patching together an implausible combination of existing views. We might ask: what is the point of the notion of a musical work? (We might ask a similar question in all ontological enquiries.) One answer in the musical case might be: the point of the notion of a musical work is to single out an object of attention and understanding, on the part of both performer and audience, which is distinct from the object of any merely psychological or historical interest. If that answer were right, it might explain why the simple view, as well as compositional-action theories, seems so unsatisfactory.

Dodd's book is a pleasure to read: clear, thoroughly argued, uncompromising, and witty. It is the book to read on its subject-matter. It shows the simple view which it advocates to be a rounded and integrated position, and it makes a compelling case for the claim that any other way of fitting musical works into antecedent ontological categories will be, at best, a fudge. Any serious challenge to it will have to be a challenge to the general practice of ontological enquiry.