The Proses lyriques seem to mark a point for Debussy when different aspects of his existence come into focus. At some time during the summer of 1892 he had at last moved permanently out of his parents’ apartment and rented a small furnished flat with Gaby Dupont in the rue de Londres. Vital Hocquet saw them often. Debussy, he recalled in an interview, ‘couldn’t afford to eat or clothe himself. Lunch consisted mostly of a small bar of chocolate, such as schoolboys eat, and what was, in those days, the classic petit pain costing a sou.’

Debussy adored watching games of billiards and going to the circus; he was particularly fond, too, of Guignol. We used to spend hours at the Guignol on the Champs-Élysées or at the Folies-Bergère, where there were often billiard matches. The room on the rue de Londres was a sort of panelled garret, untidily filled with a rickety table, three cane chairs, a sort of bed and a splendid Pleyel [piano], on loan naturally … In this room, where everything had to be done, Achille wrote masterpieces.1

In the year that followed he not only composed the Proses but also finally worked up his sketches for the Mallarmé project into a single orchestral movement, which he called, simply, Prélude à l’après-midi d’un faune. He was probably still at work on this piece when, in May 1893, he attended the world premiere of Maeterlinck’s Pelléas et Mélisande at the Bouffes-Parisiens. Lugné-Poe’s dimly lit, gauze-hung production met with a frosty reception in the press, but drew a very different response from some of Maeterlinck’s fellow writers and artists who attended the one and only performance. Mallarmé wrote up his impressions, in his elliptical, idiosyncratic way:

The painter Henry Lerolle’s response was more succinct: ‘Some very nice things, not very well played, not enough or too much décor … I prefer Ibsen.’3

Debussy, who may have bought and read the play the previous year, reacted to its performance, as we saw, by impulsively setting its most dramatic scene, the lovers’ final meeting by the fountain and Pelléas’s murder by Golaud.4 He could hardly be expected to agree with Mallarmé on the possible contribution of music to Maeterlinck’s mysterious but in fact distinctly unpoetic prose drama. It was Mallarmé, after all, who later that same year responded to a play-through by Debussy of the Prélude à l’après-midi d’un faune by admitting, ‘I didn’t expect anything like that! This music prolongs the emotion of my poem and sets its decor more passionately than its colour.’5 But Debussy still needed to get old Klingsor out of his system. Only a few days before the Maeterlinck he had been to hear the French premiere of Die Walküre at the Opéra, and – what was worse – he had been playing Ring excerpts on the piano for a series of Wagner lectures by his Rodrigue nemesis, Catulle Mendès. He had been revolted by the lecturer’s demonstrative manner. Mendès had talked about Walküre, Debussy told Chausson, ‘in such terms that mothers who had trustingly come with their daughters were obliged to flee the fevered words of this bad priest … There are aesthetic simpletons who see in this work a renewal of music and the death of the jaded old formulae. That’s not my opinion, but it hardly matters.’6 His opinion, of course, was that Wagner was a great composer but a dangerous model. ‘The time is near’, he had told Chausson only a couple of weeks earlier, ‘when this man will take a sweet revenge on the Parisians, and we will suffer as much, because he will be one of those fortresses the public likes to erect against every new aesthetic. And since, in all sincerity, we won’t be able to call it bad, we’ll just have to keep quiet.’7

His own escape from Wagner had already been partly made good in the Proses lyriques, and it was pursued further, not yet in Pelléas, but in the Mallarmé Prélude. It was one thing to have isolated Wagnerian chords and phrases in songs, where the intense verbal imagery invited a disintegration of the harmonic line, quite another thing to compose a ten-or twelve-minute orchestral work, supposedly based on a narrative poem, without falling back on the old harmonic grammar that had kept music going for the past three hundred years. Admittedly Mallarmé’s poem, though narrative in form, is so wrapped up in complex metaphor and arcane syntax that any attempt to turn it into programme music in the old-fashioned sense would surely have got lost in the undergrowth. Debussy later outlined his own very different aim in a letter to the music critic Henri Gauthier-Villars (Willy). The Prélude, he said,

is perhaps such dream as is left at the far end of the faun’s flute? More precisely, it’s the poem’s general impression, for if I’d followed it any closer the music would have run out of steam like a carriage-horse competing for the Grand Prix with a thoroughbred.8

It’s as if one were to picture the faun lying in the grass with his double flute in the hot Attic sun, amid a general air of languor and sensual promise. In the distance, perhaps, are two naked girls, naiads, bathing. But the faun makes no move towards them. They are only in his mind, a complication of unrealised possibilities, a static image intensified but not extended by thought. They are, so to speak, the final freeze-frame of the poem, alongside what the composer called ‘the decor marvellously described in the text, together with the humanity provided by thirty-two violinists who have got up too early! The ending is the final verse prolonged: “Couple adieu, je vais voir ce que tu devins” (“Farewell, you two, I go to see what became of you”).’9

The famous opening flute solo picks up several key images from the poem: the flute ‘watering the grove with melodies’, ‘malign Syrinx’, and the inertia in which ‘everything burns in the tawny hour’. Musically, too, it sets the scene. The melody swings across the ‘devil’s interval’, the tritone, up and down, a musical symbol of immobility, since by dividing the octave exactly in half (the rest of the octave also forming a tritone) it creates a symmetry that is essentially static. Curiously enough this augmented fourth or diminished fifth – to give it its correct names in tonal theory – is usually thought of as unstable, because it asks a question without answering it. Debussy, however, treats it as stable and inert, in line with his idea of dissonant chords as sonorities in their own right independent of their place in any theoretical grammar.

Meanwhile the melody decorates the interval in arabesque style, creating an ornamental line like a rococo floral moulding. Debussy later referred to this kind of writing as ‘undulating, cradle-rocking [berceuse] music, abounding in curved lines’.10 Arabesque melodies characterise the entire Prélude, movement and flow without direction, like light on rippling water. The idea fascinated Debussy, just as it had fascinated painters and designers, architects and even poets, since the discovery of Japanese prints. It was the core idea of the Art Nouveau, but Debussy had already identified it – in his opinion – in much earlier music, in the polyphony of Palestrina and Victoria. ‘It’s marvellously beautiful,’ he told Poniatowski, after hearing a Palestrina Mass in the church of Saint-Gervais –

As it happens the arabesques in Debussy’s Prélude don’t interweave, but pass from instrument to instrument, while the harmony is provided in the conventional way by the rest of the orchestra. This harmony, however, is for the most part extremely slow moving, sometimes more or less stationary, anchored by held or repeated pedal notes in the bass that often contradict the movement of the inner parts. The elaborate languor of Mallarmé’s faun can almost be smelt in this quietly sumptuous, mobile yet inert orchestral texture. And considering that Debussy had written hardly anything previously for orchestra without voices, and had not heard the one exception (the piano Fantasie), the delicacy and refinement of the scoring in the Prélude, for a reasonably large orchestra but without heavy brass or percussion, are simply astonishing.

The escape from Wagner, though, is by no means complete. Not so very far behind Debussy’s saturated harmonies and flowing arabesques once again lies, at least in general concept, the second act love duet of Tristan; and in particular the start of the Prélude has an obvious kinship with the opening of Wagner’s opera: a solo line ending on the same notorious discord, the so-called ‘Tristan’ chord, which, however, Wagner partially resolves while Debussy, in his Guiraud mood, does not. But the differences are more striking. Though slow moving, Wagner’s harmonies still push relentlessly towards closure, whereas Debussy’s flow gently downstream, sometimes getting caught in side currents or running up against the bank. And Debussy’s music never bullies. The dynamics in the Prélude seldom rise above mezzo-piano, and then only for brief moments of forte or fortissimo which, to tell the truth, are so only in relation to the surrounding music. With Wagner the reverse tends to be the case. With him piano and pianissimo are the relative terms. Loud is his default position; soft is Debussy’s.

*

On 8 April 1893 La Damoiselle élue received its first performance, at an SNM concert in the Salle Érard, conducted by Gabriel Marie. The soloists were Julia Robert and Thérèse Roger, and the programme also included Chausson’s Poème de l’amour et de la mer and works by Dukas and Raymond Bonheur. It was Debussy’s first significant premiere since L’Enfant prodigue nine years earlier, and the very first of a large-scale work of his that was in any sense representative, though in truth this five-year-old score was already past history as far as his evolving style was concerned. It had, nevertheless, a mixed reception. One critic found in it ‘an extremity of chromaticism’, another thought it ‘decadent, even a bit addled’.12 Some found it refined but over-long. On the other hand, it located Debussy firmly among the younger musicians who counted; it cemented his relationships with his fellow practitioners, including those with whom he shared this particular platform, and it brought him to the attention of or into contact with important artists in other disciplines who sensed, what not every musician sensed, a new and sympathetic direction in this work so candidly linked to modern tendencies in painting and poetry.

He had known and been on good terms with Chausson for some years, but the SNM concert brought them closer. ‘I’m furiously bored by your absence,’ Debussy wrote to him a few weeks later.

Towards the end of May, he spent a few days with Chausson, Henry Lerolle, and Raymond Bonheur at Luzancy, on the River Marne, where Chausson had rented a small château for the summer. During the day they worked on their own projects; there were river excursions, and in the evening they played and discussed music. ‘The new Musorgskys will perhaps be there,’ Chausson had written, urging Debussy to ‘borrow whatever Russian music you can’.14 The new Musorgskys certainly included a vocal score of Boris Godunov, and perhaps some songs, music that was beginning to make an impact on these French composers. ‘How boring of you not to be here any more,’ Chausson wrote after Debussy had returned to Paris. ‘No more Russian music, no more boat trips, no more billiards.’15 For Debussy, it was a new scenario in every respect.

Ernest Chausson, though a pupil of Massenet and Franck, was a composer who belonged neither to the automatic Conservatoire set that Debussy made a point of detesting, nor to the bohemian café circuit he frequented. His world was that of the well-to-do, cultivated Parisian bourgeoisie. His father had made a fortune as a building contractor in Haussmann’s redevelopment of Paris, and he himself lived well, if without ostentation, in an elegant hôtel particulier built by Chausson père on the Boulevard de Courcelles, where he and his wife Jeanne, née Escudier, entertained a circle of mainly artistic and art-loving friends, including Chausson’s brothers-in-law, the painter Henry Lerolle and Arthur Fontaine, a mining engineer and civil servant increasingly concerned with employment law and social welfare. These friendships gave Debussy entrée into a social world of refined and discriminating artistic taste and musical and intellectual sophistication beyond his experience. Chausson’s walls and ceilings were decorated with the work of Lerolle himself and the young Maurice Denis, and there were ‘Delacroix drawings, a veritable gallery of Degas pastels and drawings, canvases by Manet, Corot, Gauguin, Signac, the Orphée of Puvis de Chavannes, an important collection of Japanese prints and several albums of lithographs by Odilon Redon’.16 As an artist, Lerolle kept out of the limelight, but he was on dining terms with many leading painters, including Degas, Eugène Carrière (who painted him and his family), Renoir, Denis, as well as writers already known to Debussy: Mallarmé, Régnier, Valéry, Gide, and Pierre Louÿs, regulars at the Mallarmé Tuesdays. How Debussy fitted into this milieu in the question of manners can only be imagined, but his musical genius and his shared enthusiasm for Symbolist poetry and painting and oriental art in general opened many doors. Photographs of him at Luzancy show a shirt-sleeved, mildly tousled young man at the piano surrounded by admiring or at least attentive Chaussons and Lerolles. Posed as they are, these photos exude the cultured but unpretentious air of the sub-aristocratic artistic life that came, in France and elsewhere, from the new money associated with business success in the latter part of the nineteenth century.

Tousled or not, Debussy stood out in this environment by virtue of his extreme poverty. For the best part of a year after the Damoiselle performance Chausson helped him discreetly with gifts of money, and with an unsuccessful attempt to get him a job as assistant conductor at the casino in Royan, in return for which Debussy advised him on his opera Le Roi Arthus, which he had been struggling with (he called it ‘my battle’) for the past seven years. Chausson’s mother-in-law, Mme Escudier, arranged for him a series of Saturday afternoon sessions at her house, in which he played and sang Wagner to an audience of paying guests. Alas, not just the gifts and the Wagner sessions, but even the friendship, came to an abrupt end in the early spring of 1894, as a result of a marital indiscretion on Debussy’s part. Though apparently settled in his life with Gaby Dupont, now in an apartment in the rue Gustave Doret, for which Chausson had provided funds, he suddenly, apparently out of the blue, announced his engagement to Thérèse Roger, his nice but not startlingly attractive Damoiselle soprano, with whom he had just performed two of the Proses lyriques at an SNM concert on 17 February. He explained the situation in a letter to Lerolle:

Lerolle promptly wrote to Chausson:

You know Debussy’s getting married … He’s in the seventh heaven. Thérèse is ravishing, young and all the rest. I had dinner with him last night after he’d played the first act of Tristan on the piano. And I went with him to arrange for him to play more Wagner, at 250 francs a month. He’s very happy about it. He has to earn money and not live on Thérèse’s.18

Chausson, though ‘stupefied’ by the news, was also delighted, feeling that the marriage would settle Debussy and put his life on a more regular footing than with Gaby, who in the eyes of the Boulevard was no better than a courtesan, and was not received there.

He’s deeply in love [Chausson replied to Lerolle], and I find lovers so utterly captivating, not to say a rarity in our refined intellectual circle. This announcement is certainly going to set tongues wagging interminably … Personally, I’m confident about the outcome; I think it’ll be a very happy marriage, precisely because it’s not one that the ultimate in common sense would approve.19

It soon, however, became apparent that Gaby had not gone off at all, but was still installed in the Gustave Doret flat. As late as 8 March Debussy was still parroting to Chausson the prospective joys of wedlock.

I could stray into louche places! but I preserve a horror of them that will forever protect me, and I’m still young enough to be able to say that I am bringing a fully renewed soul to a new life, and I have feelings in me that have never been able to be formulated, with good reason, and which happily I have kept intact, hoping always for the moment when I shall have the intimate joy of seeing them flower.20

Whether or not he believed these platitudes at the time, within just over a week the engagement had been broken off. Tongues had indeed been wagging, but about Gaby, the rumours had reached Chausson, and he had written to Debussy on the 15th demanding an explanation. To call Debussy’s reply of the 16th devious would be like calling Iago a naughty boy. Passing over his fiancée with an assurance that he intended to be open with her about Chausson’s financial support, he insisted that ‘it would be impossible for me to act any differently, and I thirst for a life that is very transparent and without mysterious undercurrents’, then proceeded to dun his friend for a further loan.21 Yet he had probably already ended the engagement, which was certainly dead by the 17th. Chausson reacted in a letter to Lerolle:

Truly, the more I hear, the less I understand. At a pinch I can understand the lies, the palliatives, the subterfuges, stupid and pointless as they always are, but to lie in one’s teeth, with protestation and indignation and on matters of such gravity, that defeats me.22

Debussy’s behaviour certainly was as peculiar as it was reprehensible. Lesure suggests that he was partly motivated by the recognition that his liaison with Gaby was holding him back socially, a feeling he will have owed to his new Boulevard friendships, whether or not anything specific had been said. There was certainly outside influence of a kind. Lesure points the finger at Marguerite de Saint-Marceaux, a well-to-do sculptor’s wife who had a salon into which Debussy had been introduced that winter. She was, he alleges, a matchmaker, a Mme Verdurin (who, in Proust’s À la Recherche du temps perdu, liked to pair off the members of her ‘little clan’). It may be that Debussy had begun to tire of Gaby, thought she had left him, and felt genuinely disconcerted, not to say trapped, by her return. But after all they remained together for another four years after this blip. The real explanation probably lies in certain unreconciled conflicts in Debussy’s own character: a strong libido, which he was inclined to mistake for lifelong devotion; a genuine desire for stability and respectability, and a degree of ruthlessness in the interests of his creative work. Genius may not openly claim special moral privileges in this sublunary world, but it will often act as if it assumed them. Behind Debussy’s unsatisfactory treatment of the women in his life lay the instinctive feeling – which ordinary men usually manage to suppress – that emotional ties are a nuisance unless kept firmly in the drawer marked ‘when I need them’.

*

After ripping up his first, too Wagnerian, attempt at the death of Pelléas, he lost no time in composing a replacement, which was effectively the scene as we know it. By mid-October he had finished this new version, and soon afterwards he set to work a little more systematically, starting with the first act, then leapfrogging to Act 3, which was done by August 1894, and finally mopping up Acts 2, the rest of Act 4 and all of Act 5, completing the whole long opera apart from the orchestration – more than three hours of music – in August 1895. All the time he was working with Maeterlinck’s text in front of him, setting it as it stood, without the intervention of a librettist. A feature of the play was its rat-tat-tat of short scenes, nineteen of them in all. Debussy cut out four and a number of lines from the scenes he retained. But in essence he was simply composing the play, responding to its texture and pacing rather than trying to concoct an operatic equivalent, whatever that might have been. Though he might not have known it, he was working in a specifically Russian tradition, the tradition of Dargomïzhsky’s setting of Pushkin’s Stone Guest, and Musorgsky’s of Gogol’s Marriage and Pushkin’s Boris Godunov (but the first version of that opera, rather than the very different revised score they had been studying at Luzancy). Like Marriage, but unlike most of Boris, Pelléas et Mélisande is a prose play, a fact that distinguishes it also from Wagner’s libretti and imposes particular constraints on the musical setting. To be more exact, it allows freedoms that are really tyrannies of a different sort and which forced Debussy into a style of vocal writing that, as it happened, answered a question of Guiraud’s four years before about word-setting: ‘But then where is your poet?’ and explained Debussy’s answer: ‘One of things half-said … Music rules overbearingly in the lyric theatre. They sing too much …’ Perhaps this, as much as the atmosphere of the plays themselves, is why he was so drawn to Maeterlinck, even though his plots are by no means without their Wagnerian aspects, however much frailer and less muscular, less decisive, his characters.

Pelléas might certainly be described as Tristan und Isolde without the Will and without the attendant mythology. The inhabitants of Maeterlinck’s Allemonde live on a historical island if not a physical one. They have neither past nor future but merely exist in a misty dynastic but apolitical present with nominal attributes such as kingship, old age, half-brotherhood, a castle, and a famine in the land that has no known cause and no consequences. The characters seem not even to know their own environment. Golaud gets lost in his own forest; Pelléas has only once, a long time ago, entered the vaults of his own castle. Mélisande (like Parsifal) seems to know virtually nothing at all, though at least she knows her own name. In Tristan, King Mark seeks a political marriage with Isolde, unaware that she is already in love with the knight he has sent to bring her back. In Pelléas Golaud simply finds Mélisande and marries her (though King Arkel, his grandfather, had a political marriage in mind for him, but lets it go). Pelléas falls in love with Mélisande without knowing it and without, it seems, any moral scruples about the outcome.

In one sense it’s a Symbolist drama characteristic of its time, full of unrelated incidents heavy with secondary meaning. For instance, Mélisande’s hair tumbling from her window in the tower obviously represents her sexuality, just as the doves that fly away at the climactic moment of Pelléas’s love-making with the hair clearly stand for orgasm and the loss of virginity. Yet in another sense it’s nothing but an everyday story of country folk. The orgasm idea makes no sense in view of the scene of the lovers’ final meeting, which is evidently their first physical encounter. So the doves were doves and the hair hair, and Golaud’s worries about his wife’s daughter’s paternity are needless. The sick father is just a redundant character, like Pelléas’s dying friend, Marcellus, who amounts to a sort of Bunbury figure concocted as an excuse for Pelléas to be always about to leave (though he never does, except in a coffin). Mélisande’s death is a bourgeois misfortune to set beside the Liebestod, sadder perhaps, but distinctly less seismic. The play’s banality of event is easy to mock, and may be why, though beautiful in its writing and penetrating in its psychology, it has not held the stage except in the form of Debussy’s masterpiece.

From the start of the opera the music emphasises the strangeness and remoteness of the setting and the curious impenetrability of the characters, and glides over the banalities, either by omitting them or by investing them with gnomic significance. When, in the play, Arkel remarks, ‘If I were God, I would have pity on the hearts of men’, it sounds like the platitude it is. When sung, at the end of the violent scene of Golaud swinging Mélisande by the hair (another possibly, if more obscurely, sexual moment), it confers a spiritual exorcism on a moment of insanity, much as Boris Godunov’s ‘Oh God! You do not wish the death of a sinner, have mercy on the soul of the criminal Tsar Boris!’ exorcises his hysteria at the end of Musorgsky’s Kremlin scene, not because of the text, but because of the music.

In general Debussy’s music both intensifies and elevates Maeterlinck’s play, but not always exactly in the manner he had described to Guiraud: the things half said, the monochrome and grisaille, and the ‘music where the word finishes’, music that emerges from the shadows. The score of Pelléas is initially soft and discreet, but it rides what is in due course a violent and passionate drama that eventually generates an emotional electricity at least as great as anything in Puccini or Richard Strauss. What is new is not its discretion (Debussy’s term), but the means by which it achieves something a great deal more powerful, specifically through the way it sets the French text. When Musorgsky composed Marriage, line by line and with attention to the spoken contours of the language, he soon got bored with the flatness of the result and gave up after one act. But Gogol’s play is a comedy, dispassionate and cynical. Maeterlinck’s, by contrast, is tragic and emotionally violent, for all its prosaic language, which often seems designed to conceal as much as it reveals. Debussy showed that music can unlock these inhibitions without altering what is being said, and in doing this he evolved the technique for which, perhaps more than anything, his opera is famous.

He may have found some kind of model in those parts of Boris Godunov that survived from the original version in the edition they had on the piano at Luzancy. Musorgsky had devised a kind of flexible recitative that could move easily into arioso (a more regular, lyrical recitative) and back again, without ever settling into formal aria. There are good examples in the Chudov Monastery scene (Pimen and Grigory) and the scene of Boris’s death, both in this respect essentially unaltered in the revision. The difference for Debussy was that the Russian language has a strong tonic accent, like English, whereas French is a lightly stressed language capable of inventing accents – as we saw in connection with the songs – on unimportant words and syllables. But this proved an advantage, since it permitted a much freer and above all swifter delivery of what is, after all, a longish play set verbatim with relatively few excisions, and this allowed drastic variations in the pacing of the text and corresponding variations in the density of different emotions: the reflective, the anxious, the fearful, the reproachful, the passionate, etc. There are graphic illustrations of this in the score. In the garden scene, Act 1 scene 3, where Mélisande, Geneviève and Pelléas talk calmly about the light and the sea, the tone is conversational with only minor changes of pace, whereas in Act 4 scene 4 – the lovers’ final meeting and Golaud’s murder of Pelléas – there are indicated changes of pace every two or three bars: taking a page at random we find librement, en retenant, plus lent, serrez, très retenu, modéré, all within the space of fourteen bars. These kinds of change are matched by variations in the density of the orchestral accompaniment, from silent through every gradation of volume and weight and a huge variety of motion. The writing here is far from Debussy’s idea of the discreet or half said, and verges almost on the Wagnerian, though with a mercurial, needlepoint quality that is decidedly French rather than German.

Both in Debussy’s time and since, Pelléas et Mélisande has often been accused of lacking melody, to which his understandably irritated reply was that ‘Pelléas is nothing but melody’, a remark that might sound dangerously close to the Wagnerian mantra of ‘endless melody’ (unendliche Melodie).23 What the accusation means, of course, is that there are no ‘good (vocal) tunes’, in the manner of Gounod or Bizet. The orchestral part is in fact rich in what one might call melodic gestures, brief lyrical phrases that seem to break out in response to a situation or something said and that owe no specific allegiance to the vocal line. There is no trace here of Wagner’s idea of motives that originate in the ‘musical-poetic-line’, that is in the setting of the text (an idea that Wagner himself adheres to only spasmodically in any case). Debussy does use leitmotifs, but rarely in any symphonic way, and never in the vocal line. They pop up in the orchestra in relation to particular characters, more or less like what Debussy himself mocked in Wagner as their ‘calling cards’. The vocal line, however, concerns itself only with the words, with their contour and weight of feeling.

At the back of this technique are two apparently childlike effects: note repetition and silence. Debussy’s discovery of the expressive value of these two devices was of major significance, as he was well aware. After tearing up his Wagnerian false start, he wrote to Chausson:

What he means, perhaps, is that Wagner’s silences and near-silences, of which there are plenty in The Ring and elsewhere, were used mainly as special effects – the tense pauses that prepare the Todesverkündigung in Act 2 of Die Walküre, the expectant thinning of the texture at Parsifal’s final-act appearance as the Black Knight – whereas for Debussy silence and repeated notes are simply the first two components of an expressive scale that shares, in somewhat enhanced form, the properties of spoken language. No doubt he slightly exaggerates this distinction. The many silences in his first scene are partly at least Mélisande’s personal silences: in a sense she is silence. And sometimes these gaps suggest the awkwardness of talking to someone who won’t speak or who answers the wrong questions. Surely that makes them dramatic effects, or at least effects of character. But if you relate these silences to the next stage in Debussy’s scale, the low density of short phrases made up largely or entirely of repeated notes in more or less even values, then to the many phrases that move largely or entirely by step, you can begin to appreciate Debussy’s skill in moulding the simple elements of speech into a sung language of extraordinary expressive beauty and flexibility. Even at the opera’s most violent or overwrought moments – Golaud swinging Mélisande by the hair; the lovers’ feverish embrace before Pelléas’s murder – the vocal lines remain rooted in note repetition and stepwise motion, with isolated thrusts of upward or (less often) downward energy – just as in daily life, when we are angry or upset, we still mostly keep to the normal contours of speech but with explosions on individual words or phrases. In general, only the clinically insane talk in the jagged lines of Parsifal and Kundry in Wagner’s second act or of the patently disturbed heroines of Strauss’s Salome or Schoenberg’s Erwartung. For all its strangeness of context, no one in Pelléas is in need of treatment. With odd exceptions, like the search for Golaud’s ring in a cave nowhere near where it was lost, they all behave quite rationally, and if Golaud’s killing of his brother falls outside that rubric, one has to admit he is provoked. A French court in 1895 would have acquitted Golaud of murder on grounds of crime passionnel. It might also have noted that on a previous occasion when he seemed to be contemplating murder but in a calmer spirit (after taking Pelléas down into the vaults) he had drawn back from the act. In Allemonde, in any case, there are no law courts, and Golaud is free to regret his actions by his wife’s deathbed.

The other crucial aspect of Pelléas et Mélisande – the aspect by which most of us recognise Debussy’s style even if we could never identify the reason – is its harmonic language. We have seen this developing through his songs, and in a single short orchestral masterpiece, over the previous ten years or so. But a three-hour opera obviously demanded a range of idiom and an overarching coherence not called for even in longer songs such as ‘Le Balcon’ or ‘De rêve’. Debussy had admittedly had some practice in Rodrigue et Chimène, and it may well have been his difficult and sometimes reluctant experience with that work that helped him with the discipline needed, and achieved, in Pelléas. Fear of the problems involved might even have influenced him in his choice of a play in short scenes with an average length of less than ten minutes, though in the event he then made life harder for himself by the decision to link the scenes with orchestral interludes rather than simply lowering and raising the curtain, which is presumably what happened in Lugné-Poe’s production of the play. The original interludes were admittedly much shorter than the extended versions he had to compose at the last minute to cover the scene changes in the eventual first production of 1902. But the continuity was there from the start, and continuity – in a style that breaks almost every known conservatory rule of harmony without ever casting the listener adrift in a sea of modernistic complexity – is perhaps the work’s greatest achievement of all.

The elements of this language are the ones we have encountered before: for the most part more or less conventional chords of tonal harmony, but treated as isolated events or colours without past or future, much like Maeterlinck’s dramatis personae. These rich chords of the seventh, ninth, eleventh and thirteenth, with or without added sharps or flats, constantly suggest the tonal grammar they are culled from, but hardly ever acknowledge their parentage. When Debussy does, rarely, settle on a simple tonic chord, or even, still more rarely, an ordinary perfect cadence such as concludes almost every work in the classical repertoire, it is enough of a surprise to count as a particular gesture. For instance, Golaud’s assurance to Mélisande at the start of Act 2 scene 2 that he is unhurt by his fall from his horse, or his enquiry later in the scene whether her unease was caused by Pelléas, both moments supported by unadorned perfect cadences, seems to imply some lack of subtlety in his personality, a sort of plain-man, that’s-that quality remote from the complications that we the audience know lie behind both the statement and the question. He later admits, ‘I’m like a blind man looking for treasure on the ocean bed! … like a new-born baby lost in the forest’, at which point the music tries, but fails, to settle on a common chord of G major. All the time one feels that home keys are never far away, even when Debussy whizzes through sequences of parallel ninth chords, in honour of his remark to Guiraud that ‘il faut noyer le ton’ (‘you have to drown the tonality’). It may be an open question which tonality. What we feel is a certain general presence, something one specifically does not feel, or should not feel, in atonal music, where the tonality is not drowned but strangled at birth.

The ‘field’ approach, which carries the music along with surprising fluency, is especially noticeable in two particular kinds of harmony, both well established here and there in Debussy’s songs, but now given specific meanings relevant to this particular story. One kind is the whole-tone scale that we came across in ‘Recueillement’, where its lack of direction created a moment of rather sinister stillness appropriate to the ‘obscure atmosphere enveloping the town, bringing peace to some, anxiety to others’. A simple illustration of the effect of these harmonies is at the end of the vault scene, when Pelléas emerges on to the sunlit terrace and breathes in the smell of the sea and the scent of the just-watered roses, and the harmony changes from dark whole-tone to bright arpeggios of sevenths and ninths: dissonances, at the Conservatoire, that need resolving, but here enough of a resolution in themselves. The other kind is the modal harmony with which the opera begins, and that creates the once-upon-a-time flavour that goes with the story’s sense of timeless antiquity. Golaud’s motive, in the fifth bar, is whole-tone on this first appearance, so there’s an early opportunity to hear how Debussy controls the atmosphere by fluctuations of harmony, a process that, in one way or another, runs through the entire opera. Above all, this montage of harmonies is managed by Debussy with such brilliance that the listener hardly has time to wonder how one thing follows from another or whether it will be apparent when it’s all over.

Pelléas et Mélisande is a masterpiece of many ambiguities. On the face of it, it is a Symbolist work set in a dream-like environment; yet it is a coherent love story with a beginning, a middle and an end played out by believable, rounded characters with comprehensible motives. It is a fairly derivative score. It plainly owes a lot to Wagner in its broad discourse, in some of its materials and procedures, and even in some aspects of its subject matter. There are indirect allusions to Musorgsky’s Boris Godunov, not only in the style of word-setting, but in various orchestral details – for instance, the oscillating quaver and semiquaver figures that abound in the accompaniment (as in Musorgsky’s Chudov and Kremlin scenes) – and in particular gestures, such as the flute solo that opens Debussy’s second act, an apparent reminiscence of the start of the Kremlin scene. These are background presences, but they seldom impinge on the originality and individuality of Debussy’s opera, which have evolved through his own earlier work and in response to ideas of his own that may indeed have been helped along by reactions against Wagner and by sketchy encounters with assorted Russians, just as Bach learned from (and stole from) every foreign genius he could lay his hands on, and became the great Bach. Even if his songs and Mallarmé Prélude had not already done so, Pelléas gave notice of the great Debussy, a score of astonishing beauty and theatrical impact in itself, and a confirmation of the validity and workability of ideas that might previously have seemed not much more than the gurglings of a baby pulling the plug out of its own bath.

The question of symbolism has been well worked over in the Debussy literature. Maeterlinck’s play is presumably Symbolist in the sense that it tells a story set in a strange, nowhere land about people who hardly belong to the real world. But apart from a few unexplained circumstances and impedimenta – Mélisande’s personal history, the crown in the fountain, the old men asleep in the cave – it lacks the allusiveness, solipsism, and counter-intuitiveness we normally associate with true Symbolist art from Huysmans’s À rebours to the poetry of Eliot and the paintings of Chagall and Magritte. Debussy adds the symbolic power of music, but that is a generic property shared with all music, perhaps all art. In his influential book Debussy: Impressionism and Symbolism, Stefan Jarocinski attempts to draw a distinction between those two troublesomeisms and to argue for the replacement of the former label with the latter in Debussy’s case. There is no denying that, at the time of Pelléas, Debussy was much in the company of poets and writers whose work employed obscure syntax and hidden meanings to a far greater extent than Maeterlinck, and of painters such as Maurice Denis who used visual images as insignia of properties not portrayed. But this is a poor reason for a tendentious labelling of Debussy’s own work. Whether La Damoiselle élue, the score of which was published in 1892 with a cover by Denis, is any more ‘Symbolist’ than, say, Berlioz’s Nuits d’été is a topic for discussion, to the point at which one might weary of the whole issue of labelling. One might simply point out the categorical error of comparing Impressionism and Symbolism at all. Symbolism is a relation of subject to meaning, while Impressionism is a relation of subject to technique. Debussy might be neither, or both at once, as Jarocinski seems to admit when he suggests that ‘in [Monet’s] latest works (e.g. Nymphéas) Impressionism rejoins Symbolism’.25

Debussy was himself well aware of the special properties of his work, without attributing to them smart pigeon-holing labels, which, like most artists, he detested. He thought of his characters as real people, and lived intensely in their company while writing. His account to Lerolle of the violent scene between Golaud and Mélisande is revealing in this respect.

‘I live’, he had told Pierre Louÿs a year previously, ‘solely in the company of Pelléas and Mélisande, who are always very accomplished young people.’27 A month later he had finished the vault scene, ‘full of sinister terror’, he informed Lerolle, ‘and mysterious enough to give even the most temperate souls vertigo’; and the following scene, emerging from the vault, ‘full of sunshine, but sunshine bathed by our mother the sea’. But the lovers themselves had

begun by sulking and no longer wanting to come down from their tapestry, so I was obliged to play with other ideas, at which point they came and leant over me, and Mélisande, with that sweet morbid voice of hers that you know, said to me: ‘Drop these silly little ideas so loved by cosmopolitan audiences and stick to your dreams of my hair, you know very well that there’s no love like ours.’28

Eventually he admitted to Louÿs that ‘Pelléas and Mélisande are my only friends at this minute; moreover we’re perhaps getting to know one another too well, and we tell each other nothing but stories whose endings we know perfectly well; but then, finishing a work is a bit like the death of someone you love, is it not? …’29

Pelléas et Mélisande was complete, apart from the orchestration, by August 1895, but it lay on the composer’s desk for a good many years before reaching the stage of the Opéra-Comique in April 1902 and being published in vocal score that same year by Fromont. One problem, as Debussy foresaw, was the unconventional nature of the characters and their vocal castings. Mélisande, the only substantial female character, is hardly a typical operatic soprano, while the ‘tenor’ part, Pelléas, is equally suitable for a high, light baritone, what the French call a ‘baryton Martin’, and has been sung as often by baritones as by tenors. ‘In France,’ he wrote to Lerolle after finishing the score,

But then he dreaded the whole business of production and performance and the associated hullabaloo. ‘I hate crowds, universal suffrage and tricolore phrases!’ he told Lerolle in the same letter. He had played the work, in whole or part, a number of times in private before it was accepted in principle for staging at the Opéra-Comique. Even then it was four years before the production actually happened, during which time Debussy made substantial revisions, particularly to the fourth act. The orchestration was written down – though presumably largely present in Debussy’s mind – only after the work’s formal acceptance in 1901.