Mozart's Piano Concerto in C minor, k491, the penultimate in a sequence of twelve written between February 1784 and December 1786, is a momentous work. It is grander yet more intimate than its immediate predecessors, featuring an expansive first movement with both a harsh confrontation between piano and orchestra in the development section and elaborate sequences of dialogue in the exposition and recapitulation, a second movement with some of the richest wind writing in Mozart's orchestral oeuvre, and a terse, intense theme-and-variations finale. Completed on 24 March 1786, less than six weeks before the premiere of Le nozze di Figaro, it may have been played only once publicly by Mozart – at the Burgtheater in Vienna on 7 April 1786.

The autograph score of k491 is also a remarkable document. Sold by Mozart's widow Constanze to the publisher Johann Anton André in 1800, it eventually found its way in 1894 to the Royal College of Music via the Scottish philanthropist Sir George Donaldson, and is now published in the excellent new facsimile under review. Orderly revisions in Mozart's earlier and later Viennese piano concerto autographs usually reveal the composer's thought processes to attentive performers, scholars and editors. But k491 is different, as Robert D. Levin explains, with often ‘disorderly notation’ and a ‘considerable amount of [material] . . . sketched and later overwritten’ (3). Shorthand is another distinctive feature of the manuscript, with only outer pitches written for some solo passagework and left-hand figuration. The sixteen-stave paper is highly unusual (the norm for Mozart's Viennese music is twelve staves), and may have been specially ruled for him (5).

Following Colin Lawson's succinct Preface, documenting k491's journey to the Royal College of Music, Levin provides a brief Introduction. The information will be familiar to Mozart scholars, but helpful to others less experienced with Mozart's autographs, describing (for example) the different phases of composition and notation, and the reordering of chunks of material from the first movement's opening ritornello through a system of signs. More discussion of the second movement would have been welcome. The wind may dominate the strings (3), but parity with the solo piano (not mentioned) is more noteworthy. And exactly how ‘the personalities of the individual [wind] instruments are given music exquisitely suited to their timbre and range’ (3) is left unexplained. It is a pity Levin neither engages with the substantial scholarship on k491 nor provides a select bibliography. Instead of citing the ‘personal psychic balance’ putatively revealed by k491's close compositional proximity to Le nozze di Figaro and the piano concertos k482 in E flat and k488 in A, and a supposed ‘psychic tension that devolves from far-flung modulations’ (2), Levin could have turned to the secondary literature to inform biographical and stylistic perspectives. k491 may indeed constitute ‘one of [Mozart’s] most unsettling works’ (2). But it is unnecessarily tendentious to claim that the manuscript ‘[presents] a visual image as disturbed as the emotional content of the work itself’ (3).

Levin's Critical Notes, published like the Introduction in both English and German, are more detailed, scholarly and incisive. He usefully draws attention to passages where notes have been re-traced, an unusual practice for Mozart, including in the solo exposition of the first movement. Levin also explains how working in different compositional phases may have led Mozart to make errors, forgetting exactly what had been written in an initial continuity draft when eventually orchestrating, such as in bars 39–40 of the middle movement. Descriptions of sketch notation, overwriting and various other modifications are all good. In many cases Mozart's original sketch notation is fleshed out in instrumental staves above the piano in the autograph, sometimes following the plan of the sketch and sometimes not. Elsewhere, though, Mozart's intentions are less clear, including for the piano part in the second variation of the finale. Levin's insightful description of the various layers of revision here identifies a ‘disorder unprecedented in Mozart's autographs’ (13), which at one stage provides ‘no sanctioned version’ (14).

From time to time Levin creates problems for himself by imputing motivation, or calling something an error, with insufficient justification. In bars 175–177 of the first movement, following the secondary theme in the solo exposition, he states: ‘Bassoon 1 drops out; surely it should continue in the next three bars, doubling bassi an octave higher and thereby supporting flute, oboe 1 and clarinet 1 as bassi support violins and viola’ (7). But how can we be sure that the composer inadvertently missed out a bassoon part? Mozart may have intended a change in timbre to accompany the change in piano writing at bar 175, from arpeggiated figures to conjunct semiquavers. Elsewhere Levin is reluctant to identify mistakes on Mozart's part. In bar 357 of the first movement, in the run-up to the recapitulation, ‘Bassoon 1's a-flat¹ exceeds the normal orchestral bassoon range of the time (g¹ ), suggesting that Mozart knew the principal bassoonist at the first performance and that he knew he would be able to play the note’ (9). But, if Levin accepts Mozart errors elsewhere, why could this not have been one? And the cadenza to the first movement (bar 486) draws an equivocal reaction: ‘The lack of the standard seventh chord in the keyboard has prompted cadenzas ending without the trill. Given the sketchiness of the entire solo part, it might be wise not to read too much into the absence of the usual formula’ (10). Judgment calls are probably inevitable in a critical commentary of this kind, but may have more serious consequences if, as Levin implies at the end of his Introduction (4), a new edition were to accommodate (for example) additional material for the bassoon in the exposition of the first movement; stylistic considerations cannot ultimately usurp philological evidence. In many instances, nevertheless, Levin's judgments are sound. Few will question the inclusion of a col basso for the viola in bar 278 in the middle ritornello of the first movement, following two in the previous twelve bars, or the need for two bars of semiquaver scales in the piano part immediately before the cadential trill a few bars earlier.

Minor issues with Levin's Introduction and Critical Notes aside, this beautifully produced facsimile of k491 is a splendid addition to the growing number of recent, high-quality reproductions of Mozart's autographs, including six operas from the Packard Humanities Institute (2006–2009) and the Piano Concerto No. 23 in A major, k488, from Henle (2005). Perusing Mozart's k491, while consulting Levin's commentary, brings into focus the vibrant, colourful nature of composition and performance as Mozart experienced them: he changes a quill; he forgets he is working with a transposing instrument; he neglects to leave a free stave at the top of one of the pages of the second movement; he draws human faces to alert the copyist to insert previously notated ritornello material. In short, Mozart and his compositional process come alive, enriching our appreciation of this magnificent work.