With his elegant and stimulating book Mozart's Così fan tutte: A Compositional History, Ian Woodfield has built on and drawn together two kinds of English-language Mozart scholarship that have thrived, simultaneously but for the most part separately, during the last quarter of a century: the study of the manuscripts that record Mozart's music, and the study of his operas.

Alan Tyson's Mozart: Studies of the Autograph Scores (Cambridge, MA, and London: Harvard University Press, 1987) comprises an important series of his own articles from the 1970s and 1980s on Mozart's autographs and the paper on which they are written. Tyson's catalogue of the watermarks in the paper that Mozart used (Wasserzeichen-Katalog, 2 volumes (Kassel and London: Bärenreiter, 1992)) constituted another monumental contribution to our knowledge of the autograph scores. Dexter Edge's brilliant doctoral dissertation ‘Mozart's Viennese Copyists’ (University of Southern California, 2001) did for manuscript copies (that is, the work of professional copyists) what Tyson had done for the autographs. It made available vast amounts of new information and important methodological insights whose influence will undoubtedly be felt for a long time – and not only by Mozart scholars.

While Tyson and Edge subjected Mozart's manuscripts to meticulous examination, other scholars participated in an equally fruitful study of his operas. Wye Allanbrook, Bruce Alan Brown, David Buch, Sergio Durante, Edmund Goehring, Daniel Heartz, Mary Hunter, Dorothea Link, John Platoff, Michael Robinson, Julian Rushton, Andrew Steptoe, Jessica Waldoff and James Webster are among many writers who have contributed to an astonishing efflorescence of scholarship on Mozart's operas and their place in the theatrical culture of their time, in the form of monographs, articles and collections of articles and conference papers.

Taking full advantage of the new material that these two schools of Mozart research have made available, and energized by the intellectual excitement stirred up by them, Woodfield has done something entirely new. He has written a book about one of the most fascinating and enigmatic of Mozart's operas, based on an exhaustive and scrupulous study of the autograph score and the manuscript copies in which it circulated during the first two decades after its premiere.

Woodfield's book consists of three parts. Part 1 revisits, in much greater detail, Tyson's attempt (in ‘On the Composition of Mozart's Così fan tutte’, Chapter 13 of Mozart: Studies of the Autograph Scores) to learn from the autograph about the circumstances of the opera's composition: the paper, the colours of ink, Mozart's corrections and second thoughts. Part 2 consists of a series of daring and imaginative speculations about changes involving the opera's cast and its plot that might have occurred during the process of composition. Part 3 is a study of early manuscript copies, and what these copies tell us about the changes that the opera underwent during and after the first production.

I find much of Part 1 thoroughly illuminating and persuasive – a model of lucid exposition. The recent publication of a facsimile of the autograph score of Così fan tutte (Los Altos: Packard Humanities Institute, 2007) allows the reader to follow Woodfield's argument closely, and I for one rarely disagree with him. A rare case of dissension (or at least of my not being able to follow his argument) is in his discussion of the duet ‘Fra gli amplessi in pochi istanti’ (72). There he says that the beginning of the Larghetto (Ferrando's plea ‘Volgi a me pietoso il ciglio’) ‘is the only example in this opera of a substantial change to melodic identity after the particella stage. … Mozart did not make the changes until he had added in the first violin line which doubles the singer's melody’. Woodfield seems to mean here that Mozart changed Ferrando's tune after he had written out a first violin part that doubles the original tenor melody; the change in the vocal line thus required him to change the first violin part as well. Yet in the autograph score (to judge from the facsimile (page 532)) there is no evidence of any changes to the first violin part, which doubles the rewritten tenor line. This suggests that Mozart rewrote Ferrando's melody before he wrote down the violin part.

Although Woodfield's discussion of the autograph should be of interest to all Mozartians, some of his generalizations may not apply to all of Mozart's autographs, or even to all the autographs of the later operas. For example, his statement that ‘as a rule Mozart copied his opera scores in several stages’, beginning most commonly with a particella that contains an aria's ‘vocal line, some or all of the bass line, and occasional hints as to the rest’ (16), is more applicable to Così fan tutte than to Don Giovanni or La clemenza di Tito. One important use of particelle was to allow vocal parts to be prepared for the singers before the opera was fully scored, so that Mozart or his assistant could help the singers learn their roles. For an opera like Così fan tutte, which was to be performed in Vienna by singers resident in Vienna, it made sense for Mozart to write down the music in a way that would allow him to work with those singers at leisure. But in the case of Don Giovanni and La clemenza di Tito, operas whose premieres took place in Prague, Mozart wrote down most of the music while still in Vienna, far from the singers. There was no need for vocal parts to be prepared in advance, and consequently no need for Mozart to write particelle first. Thus it is not surprising that in the autograph scores of Don Giovanni and La clemenza di Tito we find much less evidence of particelle than in the autograph of Così fan tutte. Nor is there evidence in these autographs to support Woodfield's claim on page 43 that ‘it was Mozart's usual practice not to complete an aria until he had checked it with the singer’ (though this certainly may have been his practice when composing Così fan tutte).

Although Woodfield's writing is almost always very clear, he occasionally seems to have been in doubt about the level of technical detail appropriate for his ideal reader. On the one hand, he introduces the word ‘rastra’ (11) without definition, as if he expected his readers to know what it means. Yet when discussing the two librettos printed in connection with the first production of Così fan tutte, he refers to them quite casually: ‘A draft libretto was produced, quickly followed by a final version. The changes made are often interesting’ (40–41). Woodfield suddenly – and uncharacteristically – seems to be writing for a layman who might be intimidated by bibliographical references.

Part 2 constitutes the heart of the book and, I guess, its principal raison d'être. In it Woodfield seeks to explain certain idiosyncrasies in the autograph score with reference to the opera's reception history – the distaste that some critics (starting with Mozart's early biographer Franz Xaver Niemetschek) expressed in reaction to the opera's cynical view of romantic love and its disconcerting use of musical beauty in the service of deception. Woodfield's most surprising and intriguing idea is that Mozart or Da Ponte (or both of them) felt uneasy about the idea of the two young men seducing each other's fiancées and tried an alternative plot development in which each man, disguised, seduces his own girlfriend. He cites a series of alterations and inconsistencies in the autograph as evidence of a change of plot attempted but eventually abandoned. For example, at the beginning of Guglielmo's aria ‘Rivolgete a lui lo sguardo’ (which was itself removed from the opera shortly before the premiere) Mozart seems to have been unable to decide whether the pronoun should be ‘me’ or ‘lui’: and his stage directions originally had Guglielmo address the opening lines to Dorabella (whose name was later crossed out and replaced with that of Fiordiligi). Another instance of a change of plot having possibly left a trace is in the several ensembles in Act 1, where Fiordiligi's part was originally assigned to Dorabella, and vice versa.

Woodfield's speculation, always closely tied to the autograph, is ingenious and thought-provoking. But sometimes I get the impression that he is piling one supposition on another, as when, on a single page, we read: ‘It is not inconceivable that …’; ‘On balance, it seems likely that …’; ‘It is not difficult to imagine …’ (95). Such turns of phrase, when they become too frequent, undermine the persuasiveness of Woodfield's arguments.

An eighteenth-century opera did not stop evolving after rehearsals started or after the first performance; but in the case of Mozart's operas at least, subsequent changes were recorded for the most part in copies rather than the autograph score. In Part 3 of his book, Woodfield examines those changes, and the way they coalesced into several versions documented in different manuscript traditions. Much of the argument here depends on the existence of an early Viennese manuscript that Woodfield refers to as V2, which is now lost. I wish Woodfield had explained more fully his reasons for positing the existence of V2. He states: ‘Once an opera had been accepted for performance in Vienna, at least two copies were produced almost simultaneously. I have labelled them V1 (the reference copy) and V2 (the conductor's copy)’ (152). I know of no evidence for the existence of this pair of copies in the case of Così fan tutte or other Viennese operas. From Mozart's letter of 25 September 1782 we know that the court theatres had only a single copy of the score of Die Entführung aus dem Serail. It is perhaps the extra burden of justifying the existence of a lost score that causes Woodfield to write with something less than his customary clarity, as when he has recourse to a double negative in stating of V2: ‘This amounts to a pedigree of some weight, yet it is equally undeniable that the reference copy V1, responsible for the dissemination of the opera throughout Europe, did not incorporate the changes needed for this version’ (193).

This is a handsome book, printed on good paper and solidly bound. The illustrations are clear, and reproduced on a scale that allows us to see what needs to be seen. I only wish the many notes had been printed at the bottom of the page rather than at the back of the book. I understand that some publishers fear that footnotes will scare away less scholarly readers. But this is an unashamedly erudite book, and the Mozart specialists who are bound to read it would probably do so with a little more ease and pleasure if they were to be spared the task of repeatedly turning back and forth from the text to the endnotes. That is only a minor inconvenience, however, considering the rewards that await readers throughout this exceptionally rich and adventurous exploration of Mozart's compositional process.