‘Genius gives the rule to Art.’ Thus claimed Immanuel Kant in his Kritik der Urteilskraft (Berlin and Libau: Bey Lagarde und Friederich, 1790), the most comprehensive attempt in the late eighteenth century to account for judgments that we make about our artistic experiences. Genius does not play by the rules, then, but defines through action a way of working, from which rules might then be abstracted. It is as if a genius retains an aspect of childlike behaviour: learning and growing through play, imagination decoupled from intellect. Kant was not writing for or about children, but even so the metaphor of the genius Mozart as the ‘eternal child’ has long been established in our collective consciousness. One reason why this metaphor has proved so attractive is that it short-circuits attempts to rationalize Mozart's compositional processes, thus bypassing the need for musicologists to explain what may have been going on in his creative mind. At the opposite pole from this view we find intensely analytical approaches whereby scholars think they can and must account for everything Mozart wrote down, often using quasi-scientific terms to do so, so that the composer's works can be systematically tested and demonstrated, eventually attaining the status of theory. Both extremes, in fact, offer fascinating possibilities for the interpretation of Mozart's œuvre, although proponents of the one extreme may regard the productions – one hopes not also the proponents – of the other as anathema.

I sketch out this scene not in order to align myself (as reviewer) or Simon Keefe (as author of this splendid new study) with either one of these extremes in the field of Mozart interpretation. Rather, my intention is to help the reader to position Keefe's book as a most welcome and level-headed contribution to Mozart scholarship, which on the one hand makes a claim about Mozart that is quite extraordinary (of which more presently), and on the other provides a justification of that claim with great aplomb. It does the latter not by resorting to a particular theoretical methodology destined to generate more heat than light at Mozart's expense, but by means of a wonderfully common-sensical and empirical approach to Mozart's music that is written in easily comprehensible English rather than in alienating methodological jargon. It is not easy reading: the reader needs to make the effort to follow the prose. But it is worth that effort, especially if, as the reader may find, one is not wholly convinced by some of Keefe's claims or assumptions, but is provoked into thinking carefully about them as a result. That, I believe, is a profound strength of this book: one cannot just accept it; rather, one is moved to test things for oneself. The reader will need Mozart's scores to hand, for they are Keefe's starting-point, and his judgments derive from thoughtful reflection on what he thinks they show us about Mozart's developmental attitude to composition. That turns out to be neither quite ‘Genius giving the rule to Art’ nor (thankfully) something that is explicable only by recourse to theoretical abstruseness. What Keefe shows us is Mozart as the inspired craftsman.

Keefe uncovers a process of continual self-critiquing in Mozart's Viennese instrumental works from the decade 1781–1791 – specifically in the genres of piano concerto, string quartet and symphony, although towards the end of the book he invokes other forms of chamber music too. In this process of self-reflective learning, achievements are assessed and new potential is inferred and enacted in subsequent works. These days we might call this type of activity ‘project management’. Mozart's project involved the development of a musical language in 1780s Vienna. That language is one that was satisfying to the composer; the developmental impact of (or on) the listener is not really the focus of this book, except in so far as Mozart emerges as his own audience. The project was managed by careful attention to progress within the language, which at times led to Mozart's radical re-evaluation of not just how to use that language, but how (and whether) the language itself worked effectively.

Keefe outlines the plan of his argument near the beginning of the book. His central thesis is that Mozart's ‘stylistic re-invention’ took the following twofold form (as he states later):

The same process can apply within and across genres. So, for instance, we are treated to successive chapters establishing Mozart's modus operandi within the Piano Concertos k449, 450–503, 537 and 595, the ‘Haydn’ quartets (k387, 421, 458, 428, 464, 465), the ‘Prussian’ quartets (k575, 589, 590) and the ‘Jupiter’ Symphony, k551 (some of these chapters have appeared previously as articles in periodicals and will already be known to Mozart scholars). A final chapter draws the thoughts together, explaining fascinating cross-generic fertilization, especially between the chamber music with piano and the piano concertos. Underlying his approach is the concept of dialogue as a narrative force in Mozart – a technique that Keefe applied successfully in his previous book on the piano concertos – although that is not the only territory he explores.

Keefe's study sets out with the Piano Concerto in E flat major, k449, a pivotal work in the composer's development, for it bridges the gap between two successive phases of his piano concerto production, having been set aside as ‘work-in-progress’ for some time rather than completed in a sudden burst of activity, as the late Alan Tyson's work on the autographs has revealed (Mozart: Studies of the Autograph Scores (Cambridge, MA, and London: Harvard University Press, 1987)). Looking in particular at the development section of the first movement (or the second solo, to use ritornello terminology), Keefe points to a new conception of the relationship between soloist and orchestra – a kind of dialogic confrontation – which Mozart seems to have regarded thereafter as normative for the concerto. From here Mozart developed this idea in his later concertos, culminating in the Piano Concerto in C minor, k491. Having reached that climactic point of technical control over confrontational dialogue in early 1786, he wrote his remaining concertos in relation to this high benchmark; in other words, the language of these last two concertos had been recast.

And likewise for the string quartet, which ultimately allows Keefe to offer a rehabilitation of the last three ‘Prussian’ quartets – seen in relation to the previous ‘Haydn’ quartets – as representative of a systematic revision of the musical language of the quartet, rather than a tailing-off of Mozart's creative powers, as is sometimes claimed of these three late works. Some readers may disagree with the idea that the extraordinary slow introduction of the ‘Dissonance’ Quartet, k465, is explicable within the earlier numbers in that set of six ‘Haydn’ quartets (k387, 421, 458, 428, 464): while Keefe acknowledges that the introduction to k465 is novel in its fantasia-like topic, he believes ‘its most striking harmonic procedures … are drawn to a remarkable degree from Mozart's earlier works in the set’ (97). Perhaps that is stretching a point, but we should remember that harmonic procedure and presentational aspects of harmonic procedure are not the same thing. Just because the slow introduction to the ‘Dissonance’ Quartet does not sound remotely like anything we have heard up to this point in the ‘Haydn’ quartets does not mean that its content is necessarily formed from different material (the fantasia, after all, tended to centre phenomena that were typically peripheral, such as chromatic harmony); perhaps it uses the same material, but differently. Stranger claims by other scholars, including myself, have been made for the sources of Mozart's musical language, after all. If I have a slight reservation about this particular claim, it is merely that it stems from an implication that the ‘Haydn’ quartets are approachable as a unified set of six, although they too were fragmented in their genesis across several years, to judge from the state of the autographs: they were reordered in the process of publication and only really reached objectified form as a set in the Artaria print of late 1785. But none of this means that Mozart's assimilation of harmonic practices could not have proceeded as Keefe implies, and that this slow introductory fantasia could not have been a climactic point, a benchmark for his later configuration of chromaticism as an expressive force while simultaneously a structural marker. (This point is especially pertinent to the case of the solo piano music: try listening to the central Andante of the Piano Sonata in F major k533 in relation to what Keefe shows us about the ‘Dissonance’ Quartet.)

So much for Mozart's music as scores. What about its sound? At times, I wish that Keefe had taken a step back from the scores as visual objectifications of Mozart's thought and engaged with the sound of the music. Take, for instance, the interesting point he makes on page 174 in reference to ‘confrontation’ between the piano and the strings in the first-movement development retransition of the Piano Quartet in G minor, k478, introducing us to ‘an appreciation of interactional confrontation and dramatic intensity and contrast … draw[ing] on the style of dialogic confrontation exploited in the corresponding section of the piano concertos (from K.449 onwards)’. It is the idea of confrontation that I wish to tackle here. Confrontation seems a concerto-like quality, and this passage from k478 certainly looks, on the page, like confrontation. Played on modern instruments, it generally comes across as a confrontation too. But hear it played by Malcolm Bilson (fortepiano), Elizabeth Wilcock (violin), Jan Schlapp (viola) and Timothy Mason (cello) on period instruments (DG, Archiv Produktion 477 6732, 1988/2007) and you begin to wonder if confrontation is the only issue at stake here. In terms of the material being used, there is certainly opposition between a rising minim-plus-semiquaver scale pattern (in the piano part) and references to the movement's opening motive (in the strings' responses). But the term ‘confrontation’ may be putting it too strongly. It certainly doesn't feel like that when you are playing this passage on period instruments (even on a copy of an Anton Walter fortepiano, which can take considerably more stick than Mozart gives it here).

Period-instrument performances of Mozart concertos (from which I suspect Keefe's notion of confrontation springs generically) have begun to show us that what we formerly believed to be ‘the antithesis of the individual and the crowd’ (as Donald Tovey put it so memorably) is more a sonic illusion deriving from traditions of performance practice that were in operation more than a century after Mozart's death, rather than from the internal rationale of Mozart's concerto language. His instruments do not speak in the same way as modern instruments, and that consequence has implications for the narrative continuity of his music. Dialogue there is aplenty, and sometimes confrontation too (most especially in the Piano Concerto in C minor, k491, in which a fortepiano is really pitted against the biggest concerto band Mozart ever conceived, and for which he needed manuscript paper with sixteen staves). But, in the main, historically informed performance practice makes us think more about dialogic cooperation than confrontation.

That realization might just nuance Keefe's claims a little. Those claims are, to my mind, extremely welcome and convincing when applied to ways in which Mozart reviewed his handling of musical materials, at least as reflected on paper. They afford us, I believe, the best understanding we currently have of how Mozart's stylistic development during the Vienna years may have progressed on the micro level. But when, instead, we conceive of those materials as sounds that were encoded in notation (in other words, when we approach Keefe's insights in so far as they affect us as players of this music), we need to translate the principles from an abstract into a narrative form, guiding us as we craft each phrase. For instance, reflecting on Keefe's commentary quoted above, it now seems to me when playing this development retransition from k478 that the ‘confrontation’ is a rather measured one between whole-bar units, which I play on the fortepiano, and answering half-bar pairs from my string-playing colleagues (a dialogue that might be represented as ∪; – –). There is a subtle metrical impulse of hesitation underlying this reading, and it offers us a way of moving through the musical sound-space at this point. In grappling with what ‘confrontation’ might mean (or not mean) in the context of Mozart's sound-world, I have come to a more nuanced understanding of how to perform this phrase. Similarly, I shall be tempted to look out for more development retransitions and test the semantic boundaries of ‘confrontation’ in performance – as in that Andante of k533, for instance, for at just this retransition point there is arguably a confrontation between the languages of fantasia (in the harmonic utterance) and sonata (in the tonal process), demanding a response from the player in handling the narrative in these crucial bars. I had certainly not thought of this as a celebration of what one can do with the materials of the classical musical language, nor specifically in relation to a benchmark of language development attained at the start of the ‘Dissonance’ Quartet, but now – thanks to Simon Keefe's study – I understand k533 differently. Anyone wanting to understand Mozart's mature instrumental music differently, too, could do a lot worse than to let this rewarding and thought-provoking book be one's guide.