La Valse.

Unlike a great many opuses in Ravel's catalogue, the editorial history of La Valse is extensively documented — with remarkable precision regarding the chronology of the various manuscripts and the first Durand engravings — from the earliest pencil strokes on the sketches in 1919 to the first printing of the orchestral parts in 1921. A rare circumstance for source studies: all the manuscripts referenced by Ravel in his correspondence are accessible in two research libraries — the Morgan Library in New York and the Bibliothèque nationale de France¹ in Paris.

The urgency of the composition — completed in just four months — of Wien² / La Valse, a long-standing project for a "Choreographic poem"³ as the composer described it (conceived from the outset for the theatre and not the concert hall, originally intended for performances by the Ballets russes in June 1920 at the Théâtre national de l'Opéra de Paris), forced the composer — who had taken refuge far from Parisian bustle in a hamlet in the Ardèche — to finish his manuscripts on the tightest of schedules. A private audition of the new score, under Diaghilev's supervision, was to be scheduled for the first days of spring.

Ravel then speaks of his "turbin" [grind], his long days of "boulot" [toil], rising "well before the sun," ironically noting "lie-ins until 7.30 am." Throughout the surviving correspondence of the Ardèche winter of 1919-1920 — swinging between bursts of enthusiasm and depressive episodes — the following chronological elements emerge⁴:

On examining the various surviving musical manuscripts, the following dates appear on the final page, near the composer's signature and the location "Lapras": the solo piano score bears the period December 1919 to February 1920; on the ink manuscript of the orchestration fair copy, we read: "December 1919 – March 1920."

Curiously, the two manuscripts of the two-piano Réduction — "transcription" in Ravel's own terminology — held at the Morgan Library carry no start or completion dates. These parallel undertakings — the black-and-white versions required for the deadline audition for the Ballets russes as well as for choreographic preparation, completed by the colouring of the orchestration — were thus conducted simultaneously. On the piano parts appear several mentions of orchestral desks, illuminating this or that phrase or flourish within the two-stave writing so often too narrow to contain the richness of the text.

Although the episode of Diaghilev's rejection of the score remains little documented, we know for certain that a piano audition took place in May 1920, and that the Russian impresario turned down the new work⁵. One can easily picture the composer's frustration — but also imagine his publisher's relief at the postponement of a punishing schedule, with the simultaneous engraving of La Valse in three forms.

Through the correspondence of 1920, we learn that Ravel returned to his two-piano "Réduction" in June — having ultimately regarded his first draft of winter 1919-1920 as a preparatory sketch containing "dubious passages" — and completed this final version in July, adding the dedication to Misia Sert as well as the "Viennese" ballet synopsis, which appeared here for the first time. In September⁶, we discover that the composer envisaged a premiere of La Valse at the Opéra de Paris while Lucien Garban worked on the orchestral material at Durand⁷.

However, this material — used for the premiere at the Salle Gaveau with the Orchestre Lamoureux, conducted by Camille Chevillard on 12 December 1920⁸ — could only have been provisional and almost certainly manuscript or in the form of a mere proof, with no trace surviving of these scores from the very first performance. On 17 December, Ravel wrote to Lucien Garban: "La Valse must not be engraved before I revise the score." And the Durand engraving-register⁹ clearly indicates that the engraving of the orchestral score¹⁰ and the separate parts was completed in March 1921 (deposited at the Dépôt légal on 19 and 30 March of that same year)¹¹.

We do not, unfortunately, know from which score Chevillard conducted the first performance. Certainly not from the complete ink manuscript that has come down to us — this is entirely free of blue or red pencil markings for performance, or even of any preparatory work for engraving. The most plausible hypothesis is that another "reference" orchestral score existed — most probably autograph or in a copyist's hand — used by Chevillard at the desk and by Garban for the engraving, and which very likely also contained the composer's post-premiere revisions and corrections. An essential document now lost — or destroyed to avoid confusion at the publisher when handling the various proofs, and never retained by Ravel¹².

Contemporary witness to this premiere, the archives of the Orchestre Lamoureux still preserve a complete Durand set — from the first print run — which bears on the cover pages the performance dates added by the musicians at their desks¹³. An essential support for considering the corrections, repentirs, the string forces then in use¹⁴, or the famous modification of the opening double-bass divisi that was also adopted in the performances conducted by Manuel Rosenthal.

Through the comparative study of the musical text across the various available manuscripts, we have emphasised the analysis of a document that, in our view, had not yet found its true value: that of the orchestral sketches formerly in the possession of Roger Désormière¹⁵, whose study brings fascinating clarification of the composer's initial intentions. We know Ravel's reluctance and lack of scribal appetite when it came to recopying his scores in haste, coupled with his tendency to omit whole bars or numerous orchestral details without ever returning to them¹⁶. Elements that we were able to observe — once again — in reading this very advanced preparatory work, with instrumental variants that we chose to present in this revised 2022 edition.

As an indispensable complement, we carefully considered the notes, corrections and queries of two Ravel contemporaries whose scores have come down to us: that of Charles Münch¹⁷ — who wonderfully introduced the "rhythmic suppleness" so desired by the composer¹⁸ right through to the final apocalypse and its brutal collapse — and that of Roger Désormière¹⁹, whom we deeply regret was never able to record the work. Additionally, the annotations of Pierre Boulez²⁰ on his pocket score confirm his predecessors' corrections, chiefly on the transposing instruments.

It will fall to the performers of 2022 to rediscover the "chic" that Ravel intended in the interpretation of this homage to the Vienna of the imperial balls — which heard, from the hall or at the desk, the strings scattering the melodic lines with glissandi and other portamenti, according to the taste and performance conventions of the early twentieth century. Take for example Philippe Gaubert's 78-rpm recording at the head of the Orchestre de la Société des Concerts du Conservatoire in 1927: one of the very first recordings of the score²¹.