Debussy had lost friends over Thérèse Roger, and he lost more over Lilly. It was not so much that old friends broke with him explicitly or lectured him on his behaviour. Rather, they simply faded out. With some, it was a question of choice between the divorcees. Raymond Bonheur and René Peter stayed friends with Lilly and no longer communicated with Debussy, and the same was true, at least for a time, of Robert Godet, although in point of fact he and Debussy had not corresponded for two years at the time of the split. The Fontaines, whom he had seen less since his marriage to Lilly, vanished finally from his radar. With a few, the break was temporary. With Pierre Louÿs, though, it was decisive. Their respective marriages in 1899 had, it’s true, weakened the intensity of their relationship, but Debussy’s second marriage, to the Jewish Emma, was too much for the anti-Semitic Louÿs. Ironically, what was apparently Louÿs’s final letter to Debussy, in June 1904, seems to have regretted the cooling of their friendship; yet Debussy’s reply was their last communication.

The vacuum created by the loss of old friends was gradually filled, was already being filled, by new ones. Most important, as well as perhaps most surprising, was his friendship with the music journalist Louis Laloy, who had written a sympathetic article on Pelléas et Mélisande in the Revue musicale of November 1902 that attracted Debussy’s attention and inspired him to invite Laloy to call on him. A few weeks later Laloy toiled up to the composer’s fifth-floor apartment in the Rue Cardinet ‘like an explorer in an unknown country’. But Debussy was not the wild beast he seems to have expected. He put Laloy at his ease at once:

Laloy was himself an orientalist, interested especially in the music of the Far East. But he also shared Debussy’s love of the eighteenth-century French clavecinistes, especially Rameau and Couperin. He had studied under Vincent d’Indy at the Schola Cantorum, an institution founded in 1896 as a repository for the teaching of music on the basis of Gregorian chant and fifteenth- and sixteenth-century polyphony. To cap all this erudition of the exotic and the antique, Laloy had written his doctoral thesis on the Greek musical theorist Aristoxenus. But his polymathy did not stop at music. He was also a graduate of the Paris École Normale Supérieure, one of the four French ENSs that creamed off the best university undergraduates to become top scientists, administrators and thinkers. ‘He was’, Lesure remarks drily, ‘the only academic we shall ever encounter in the entourage of the composer, who had a certain mistrust of people who were “too well informed”.’2

Laloy’s first service to Debussy, early in 1903, had been as intermediary in the award to him of the Légion d’honneur, engineered by the proprietor of the Revue musicale, Jules Combarieu, an admirer of Pelléas and, no doubt more conveniently, younger brother of President Émile Loubet’s cabinet secretary. Debussy accepted this unlooked-for accolade with an attitude that he described, somewhat obscurely, as ‘well designed to discourage the best goodwill’.3 But he was more openly grateful for Laloy’s behaviour towards him after the break with Lilly. Laloy had written to him in April 1905 with what he called ‘a clear-sighted sympathy that forbade itself to trample on life without a care’. The composer replied:

Laloy had been out of town when the news broke and became common gossip. But ‘having no direct information’, he later wrote, ‘I refused to pronounce.’

It’s so agreeable to judge one’s neighbour severely that few deny themselves the pleasure. They have fun pitying the victim, blaming the betrayer, exaggerating the sorrows of the one and the crimes of the other, and throwing oil on the fire. As it seemed that Debussy must have been rescued by this event from material cares, they crudely accused him of having acted in a calculating manner, and of having sold himself.5

Obviously Laloy was more interested in Debussy’s music than in his behaviour, a position he shared with the composer himself, who was still finding it hard to understand what people thought he had done wrong. (‘It seems I’m not allowed to divorce like everyone else,’ he had complained to Durand, another friend with a professional loyalty to Debussy’s music.6) Nevertheless the friendship with Laloy was genuine and became close. They would often dine together and even – an arresting image – play bridge (the recently invented auction variety, one hopes).

Laloy had left the Revue musicale early in 1905 and started a new journal, the Mercure musical, to which he was naturally hoping to get Debussy to contribute. But the composer wriggled. At first it was his divorce; then it was the title, Mercure, which he decided would become the butt of vulgar jokes about remedies for syphilis. In due course he offered a new series of ‘Entretiens avec M. Croche’, ‘a man I spent a lot of time with in the past, let’s hope I can find him again’. But M. Croche did not reappear, and a few months later Debussy finally wrote him and his interlocutor out of the Mercure script, with complimentary remarks about Laloy’s own study on Pelléas, and rude remarks about the other contributors, whom he found ‘sinister [and] above all terribly well informed; I really don’t see what this poor M. Croche could manage among so many bold specialists’. Instead he wrote his obituary:

And Debussy never did contribute to the Mercure musical.

The reality was that he was struggling with the composition of a new set of Images for two pianos. The solo piano set had run into difficulties at Eastbourne in August. Now, in September, he was turning to the next component of his original contract with Durand. At that time, in July 1903, he had listed the duo pieces as Ibéria, Gigue triste and Rondes; but by May 1905, this had changed, Rondes had disappeared and been replaced, rather hesitantly, by Valse.8 As it turned out this was the sole mention of any such piece, and by the time Debussy next named the final part of his original trilogy, it was once again Rondes, though whether out of uncertainty about its content or merely about the aptness of this or that title we shall probably never know. In Rondes de printemps, to give the piece its eventual name, there is some triple-time music, but it is not very waltz-like. Most likely, it was the dance idea that lodged itself in Debussy’s mind from the start, while the exact type of the dance and character of the music took time to take shape.

No less mysterious is the fact that he initially thought of these as two-piano pieces (though the possibility of an orchestral version was already mooted in the 1903 contract). The Image concept had been born at the piano, and, to judge from the first solo book, was tied up with the exploration of particular harmonic and textural colourings peculiar to that instrument, with its special resonances and integrated sonorities. Yet from the start the two-piano Images (of which there were to have been six in all, though only three were named) seem to have taken on a more complex type of discourse, more symphonic at least in the sense of a varied procession of thematic materials, if not in the sense of what Debussy called ‘a robust symphony that marches on all fours’.9 It’s hard to imagine the music of Ibéria, especially, as ever having been thought of purely in keyboard terms. Nor indeed is there any indication in the contract of the three-movement form in which that particular Image ended up. Its closest ancestor in Debussy’s work is, it’s true, a keyboard piece, also about Spain, La Soirée dans Grenade. But for Ibéria that piece was no more than the starting-point for an almost cinematic stretch of musical footage, a twenty-minute travelogue that depends for its effect on the multiple layering of different colours and the rapid intercutting of quasi-visual images and slices of musical life. The other two Images will make somewhat different, less specifically visual use of comparable techniques, with the help of fragmented quotations from popular tunes.

Debussy was surely thinking of these orchestral Images when he wrote to Gabriel Mourey early in 1909 about ‘la maladie du retard … et ce curieux besoin de ne jamais finir’ (‘the sickness of delay … and this curious need to never finish’).10 ‘I’m now going to finish the two-piano Images as quickly as possible,’ he assured Durand at the end of September 1905. Early the following July it was ‘if the ironic hazard of things doesn’t come and mess up my papers, I think I shall finish Ibéria next week, and the other two later this month’. But a month later, he still had ‘three different ways of finishing Ibéria; should I toss a coin or look for a fourth?’11 And so it went on. After several more such half-promises, Ibéria was eventually completed on Christmas Day 1908, and Rondes de printemps, in short score at least, a few days later, though Debussy went on tinkering with it for several months thereafter. As for Gigues, now plural but no longer triste except about not being written, it hung fire for almost a further four years, promised and repromised, while Debussy became caught up with other things, most of which remained forever tristes for much the same reason.

As usual the distractions were largely due to Debussy’s unrequited obsession with the theatre. At first it was the King Lear music, which lingered in his mind long after the Antoine production of December 1904 for which it had originally been intended, but was never got down on paper beyond two exiguous episodes, a tiny fanfare, and a slightly more substantial movement called ‘Le Sommeil de Lear’, which Orledge locates to the scene in Act 4 of the play where Lear sleeps in the French camp.12

Then in April 1906 he received a visit from a young writer and amateur musician by the name of Victor Segalen, who was hoping to interest him in an opera on the life of the Buddha. Segalen was a marine doctor by profession who had travelled widely in Oceania, but had also recently spent several weeks in the Sri Lankan capital, Colombo. In Tahiti he had made a study of Polynesian music, and their conversations on this topic seem to have reawakened Debussy’s dormant enthusiasm for this or that music of the Far East, and were probably what led him at least to affect an interest in the Buddha subject, Siddhartha, though he soon realised, as he told Segalen frankly, that ‘in its current form, I don’t know of any music capable of reaching into this abyss! It could scarcely do more than underline certain gestures or make certain settings more precise: in the end, an illustration, much more than a perfect union with the text and the alarming immobility of the central character.’13

Instead he suggested that Segalen might consider writing a libretto on the Orpheus myth, which had formed the basis of his recent novel, Dans un monde sonore. Segalen duly obliged. But Debussy, when confronted with a detailed draft, reacted in much the same way as before. The text was ‘more literary than lyrical’, like ‘whole pages of Chateaubriand, V. Hugo, Flaubert that are said to be bursting with lyricism, but which, in my opinion, contain no music whatever’. ‘No doubt,’ he added soothingly, ‘a few hours together will sort it all out.’14 But they never did; and Laloy considered that Debussy always knew that the Segalen projects were abortive but liked Segalen too much to say so, ‘and wasted a lot of time discussing an Orpheus text, endlessly corrected and redone’.15

Meanwhile Debussy had been placed in an embarrassing position vis-à-vis Segalen by an announcement of 15 June 1907 in Le Courrier musical, to the effect that ‘M. Gabriel Mourey is finishing the libretto of a lyric drama, L’Histoire de Tristan, for which M. Claude Debussy has agreed to write the music’.16 The report, it seemed, was perfectly true. Mourey – last heard of in connection with L’Embarquement pour ailleurs in 1891 – had presented him with an outline for a libretto based on Joseph Bédier’s Roman de Tristan, and Debussy, who had been enthralled by Bédier’s deeply researched novel when it first came out in 1900, and had thought then of turning it into an opera, had promptly accepted. Alas, the Mourey project foundered on the same sort of reef as La Princesse Maleine had done. Bédier had already licensed the work to his cousin Louis Artus, a boulevardier playwright with whom Debussy had no wish to collaborate, but who was unwilling to give up his priority. So a Debussy Tristan went the way of all but one of his other theatre projects to date, to be swiftly followed by a series of other Mourey proposals, all but one of which he rejected more or less out of hand.

Apart from a single Tristan theme in a letter to Durand, none of these projects produced a note of music, but merely used up precious time and creative energy. One other operatic scheme that began seriously to obsess him in 1908, without any literary intermediary, was one that he had almost certainly had in mind before but had never seriously addressed, Poe’s tale The Fall of the House of Usher. He had done nothing to his previous Poe opera, The Devil in the Belfry, for the best part of five years, unless one counts his enigmatic remark to Durand in July 1906 that ‘for the Devil I think I’ve found a quite new way of having the voices move around, with the added merit of being simple’.17 Yet in July 1908 he made a formal agreement with Giulio Gatti-Casazza, the newly appointed director of the New York Metropolitan Opera, to give the Met priority in his two Poe operas, as well as, still more speculatively, in subsequent operas including The Legend of Tristan, in return for an immediate advance of 2,000 francs. With the cheque in his pocket, Debussy is supposed then to have told Gatti-Casazza, ‘It’s bad business you’re doing, and I even have some remorse in taking these few dollars, as I think I shall never succeed in finishing any of these works. I write solely for myself, and other people’s impatience doesn’t concern me.’18

Nevertheless, The Fall of the House of Usher did genuinely preoccupy him, on and off, for the rest of his life, and it generated far more material, in however confused a form, than any of his other operatic projects apart from Pelléas and Rodrigue et Chimène. It undermined his concentration on more achievable works. In June 1908, at a time when Ibéria was still incomplete, he admitted to Durand:

These last few days I’ve been working a lot on The Fall of the House of Usher … it’s an excellent way of steadying the nerves against all kinds of terror; all the same, there are moments when I lose all sense of the things around me; and if Roderick Usher’s sister were to walk into the room, I wouldn’t be all that surprised.19

Then, a month later:

And another year on, with Ibéria and Rondes de printemps in Durand’s hands but with Gigues languishing:

Choisnel [Durand’s chief copy editor] conveyed to me your wish to have the next part of the Images … I have to confess that I’ve recently left them a bit to one side, in favour of Edgar Allan Poe. I’ve got so much to do on that that you’ll forgive me, won’t you?21

He had truly become a victim of his own maladie du retard, and was in some danger of losing the impulse of the Images altogether. He did eventually complete Gigues, but it was not without an effort of will that, it might be argued, is not altogether disguised in the finished work.

Not everything in this period was ‘rotting away in the factories of Nothingness’, as Debussy had on one occasion described his state of mind to Durand, quoting Laforgue.22 He had completed the Saxophone Rhapsody and composed a brilliant little suite of piano pieces inspired, initially, by the birth of Chouchou, though well outside even the mental compass of the not quite three-year-old child that she was when he finished it in the summer of 1908. The real birth piece in Children’s Corner, as he called the suite (using the English form for all its titles, perhaps because Chouchou’s governess was English), was Serenade of the Doll, written in March 1906 and imagined, clearly, in the presence of the five-month-old little girl and her actual doll. The other five pieces were added a couple of years later, and the whole suite was dedicated to ‘my dear little Chouchou, with the tender excuses of her Father for what follows’.

Children’s Corner is sometimes very easy to play, sometimes less so. But mostly it has the virtue of sounding harder than it is, which of all instrumental qualities is probably the one that pleases show-off children best. Thus the first piece, Doctor Gradus ad Parnassum, though a gentle parody of Muzio Clementi’s progressive piano tutor Gradus ad Parnassum, lies so well under the fingers that it hardly works as a study at all, despite crossing hands and singing inner voices. When Durand asked how fast it should go, Debussy replied that it was ‘a sort of hygienic and progressive gymnastics; best to play it every morning before breakfast, beginning moderato and ending animé’.23 Serious study, he perhaps wanted it thought, was not quite his cup of tea. In Jimbo’s Lullaby we encounter our old friend from Jardins sous la pluie, the nursery song ‘Dodo, l’enfant do’, as a melody, as a clashing chord of its two main notes, and as a galumphing five-finger exercise in whole tones. The Serenade is correspondingly delicate, with inner and hidden voices like the ones in Doctor Gradus ad Parnassum, which, as there, make for a neat effect that sounds tricky but is not too hard to play, whereas The Snow is Dancing, more Clementi-like in its offset left-hand/right-hand semiquaver figurings, is genuinely awkward if only because of the constant variation in the figures, as the snowflakes swirl around in the wind. At one point the English nursery song, ‘Ding-dong Bell’ materialises unexpectedly out of the mist. The Little Shepherd, simplest and sweetest of the six pieces, seems to derive some of its ideas, especially the plus mouvementé dotted rhythm, from Rimsky-Korsakov’s Sheherazade, which Debussy had liked in the 1890s, then turned against after hearing it in 1906 (‘more bazaar than oriental’, he told Raoul Bardac24). Finally the Golliwog’s Cakewalk, still perhaps Debussy’s most famous piece, with its snappy ragtime rhythms and snide little parody of the opening of Wagner’s Tristan, marked ‘avec une grande émotion’, just in case we miss the point (which no child, and not all music students, would get in any case).

Debussy was a great admirer of Musorgsky’s song cycle The Nursery, which he will have known in Rimsky-Korsakov’s expanded and edited version of 1882. He adored the clarity and simplicity with which Musorgsky integrated his music and his subject matter; he loved the way the Russian composer handled his materials empirically, without troubling himself with textbook procedures or established forms, everything ‘made up of small successive touches, bound together by a mysterious linkage and by a gift of luminous clairvoyance’. ‘Sometimes too,’ he had added in the same Revue blanche article of April 1901, ‘Musorgsky conveys a sense of shuddering, restless shadow that envelops and grips the heart to the point of anguish.’ Musorgsky had a ‘Doll’s Lullaby’ to compare with Debussy’s Serenade (though the Russian title simply means ‘With Dolly’), which seemed to him to have been ‘divined word by word, thanks to a prodigious power of assimilation, that gift for imagining landscapes of fairy intimacy so special to childish minds’.25 Whether or not Debussy himself quite managed the degree of pure, unselfconscious identification with childhood that he attributed to Musorgsky, he did beautifully map his more sophisticated manner on to the image of a better-behaved childhood such as might have been observed in a drawing room of the Avenue du Bois de Boulogne in 1908.

Like Musorgsky, he often seemed to be feeling his way stylistically, testing every chord and chord sequence, every rhythm, every colour for their precise effect, since he too had abandoned the rule books that, in moments of failing inspiration, guaranteed coherence, if not interest. It was the real cause of his maladie du retard, his slowness to complete and his reluctance to hand over. But was it truly such a maladie? In terms of artistic method, he knew perfectly well, it was his greatest virtue. ‘You have facility,’ he had written to Raoul, ‘and are certainly gifted, but you will never sufficiently distrust the path that your ideas make you take.’

Was he thinking of particular composers? Of Richard Strauss, for instance, whose Salome he had joked about to Durand after studying the score but before hearing it (‘Salome, or the lack of cordiality between all the chords’27), though just before attending a performance in May 1907 he told Gabriel Astruc, with only mild sarcasm, that ‘it seems to me difficult to have anything but enthusiasm for this work, which was “at once” a masterpiece’.28 Or of Ravel, whose latest piano work, the brilliant five-movement Miroirs, Viñes had played to him and Emma three weeks before his letter to Raoul. Or even of Rimsky-Korsakov, or Vincent d’Indy, whose Jour d’été à la montagne he had just heard in a Concert Colonne, and described to Raoul as ‘a bit of d’Indy from the backside of the Cévennes’.

As I’m not exactly well informed about the atmosphere of that region, I can hardly say anything about it. He seemed to me to overuse the bassoon, and one was astonished to hear a piano; I thought pianos were only found on Swiss mountains.29

It hardly needs a depth psychologist to see that these various jests about the star novelties of the day barely conceal a certain unease, even insecurity, in their presence. A year later he would take issue with Laloy over his laudatory review of Ravel’s Histoires naturelles in the S.I.M. (Société Internationale de Musique) journal, ‘astonished’, he said, ‘to read there that a man of your taste deliberately sacrifices the pure and instinctive masterpiece that is [Musorgsky’s] Nursery to the self-conscious Americanism of M. Ravel’s Histoires naturelles. For all its undeniable expertise, it can only be “displaced” music. Leave that to the valet de chambre, [Michel-Dimitri] Calvocoressi.’30 But he was broadly enthusiastic about Serge Diaghilev’s Paris concerts of Russian music that May. ‘They are admirable,’ he told Astruc, who was promoting the series, ‘a shade nepotistic in the programming, and some odd omissions from the ones labelled “historic”. But it’s very good to have played the second act of Boris.’31

It wasn’t, however, only the music of other composers that came between Debussy and his own work. Family life also imposed sacrifices. These included trips to the seaside, no longer to the romantic shore of Jersey 1904, but to the holiday resort of Puy, one of the Dieppe beaches, where they put up in August 1906 in a hotel run by a sinister Italian ‘who does his own shopping, brings back the most doubtful food substances, messes up the fish and the meat, and is patently an assassin’.32 ‘The beach’, he told Laloy, ‘is absurd, the hotel the last word in discomfort, and the English the sort who render any entente cordiale impossible.’ Photographs of Debussy with Chouchou on the sands at Pourville and Houlgate in later years say more than words can about his stoical tolerance of the double-edged seaside ritual: suited, waistcoated and boatered, he stands like a well-dressed attendant, ready to carry his little daughter’s bucket and spade, but not quite equal to the fun and games they offer. For his own work, he goes on, ‘I’ve at last got a 75-centimetre table for writing things that have without fail to revolutionise the world. You’ll tell me that there’s the Sea …! Unfortunately it’s a bit like a “tub” that empties out! The coasts feel too close, and I can’t manage to rediscover my emotion of last year faced with the Ocean.’33 ‘Can you work by the sea?’ he asked Segalen. ‘I can work pretty well anywhere.’ ‘You’re very lucky!’34

The following summer it was Pourville again, and this time for two whole months, while a faulty boiler in the Paris house was being repaired. So for once Debussy was forced to work by the sea, despite another horrible hotel, inedible food, stomach upsets, and the usual army of uncordial Englishmen. It was probably during these months that he somehow got Ibéria and, less certainly, Rondes de printemps into something close to their final form. But he must also have worked on a very different but similarly titled work, his second book of Images for solo piano, which he was able to present to Durand in its finished state at the end of September 1907. The perfection of this work, achieved under frequently hostile conditions and without his usual equivocations and prevarications, is in startling contrast with the erratic progress and in some ways uneven outcome of the orchestral set. Of course, the orchestral composition is in many ways the more complex. The concentration and unified character of the piano pieces was more in the direct line of Debussy’s recent work. But if one wants to consider what he meant about revolutionising the world at a 75-centimetre table, it’s necessary to take account of both these brilliant sets of contrasted pieces.

The three pieces in the second book of Images all have titles referring to things seen or visible, but in each case the reference is somewhat oblique. Cloches à travers les feuilles (‘Bells through the leaves’) relates a visual impression, the leaves, to an aural one, since one doesn’t watch bells through foliage, but might hear them through it. Even so, the suggestion is slightly odd. To what extent would foliage affect the sound of the bells? The question is naive, and misses Debussy’s point, which is surely a transference between the different senses: bell sounds converted into the rustle of leaves, leaves thought of as an image of the intricate polyphony of pealing bells. Laloy is supposed to have inspired this idea with a description of the tolling of bells on All Souls’ Day in the Jura, ‘passing from village to village, through the yellowing forests in the silence of the evening’.35 But Debussy might equally have stored up his own memories of English bell peals, a very different phenomenon from anything to be heard, then or now, in France, and one that sets up its own complex associations with nature and the outdoors.

English change-ringing, with peals of six, eight or up to twelve bells, is a unique kind of music that is at once static and highly mobile: static, because the content of each round never changes, each bell being rung exactly once; mobile, because the order of the bells changes all the time, creating a complex network of interior patterns within the unvarying harmony. Nobody – and certainly no musician – who has ever listened to change-ringing of this type will be unaware of the contrast between the fixed reverberation – Tennyson’s continuous ‘mellow lin-lan-lone of evening bells’ – and the subtle rhythm of the interior changes, mathematically determined, but deeply affecting aesthetically. Debussy might well have sensed a parallel with the gamelan music that he still revered: there was the same kind of interior polyphony (genuine with the gamelan, merely implied in the case of the bells) within the same kind of static harmony. There was the same tintinnabulation, the same ‘molten golden-notes’, that he must have known from his beloved Edgar Allan Poe’s poem ‘The Bells’.

Cloches à travers les feuilles is, consciously or unconsciously, his version of this intriguing and complex idea. Take the opening page (plus one bar: so, the first eight bars). With the exception of two C naturals in bar six, the entire passage is derived from a single whole-tone scale, a completely static harmony but with an intricate filigree of layered interior motion, laid out on three staves for clarity, mainly scales (as in the early and final stages of ringing a peal), similar and contrary motion, but with much rhythmic variety and extremely refined articulation, which Debussy, when he sent the music to Durand, made a point of asking the engraver to respect. At bar 9 the harmonic colour changes abruptly, and it does so again at bar 13, switching to a new colour altogether, in a soft left-hand ostinato that Debussy wants ‘very even, like an iridescent vapour’. These changes are not unlike the changes of angle or focus with fixed cameras in a film about nocturnal animals. Debussy seems to be peering into the innards of his harmonies and sonorities, trying not to disturb them, then after a few minutes moving quietly on to the next. He doesn’t ignore the needs of music as a temporal structure. ‘But I’m more and more convinced’, he told Durand, ‘that Music is not, in its essence, something that can flow along in rigorous traditional forms. It’s a matter of colours and rhythmicised time.’36 So he makes a textural climax, ‘a little animated and brighter’, with sharper dynamic and rhythmic contrasts and more brilliant figuration, but still unchanging harmonically for five full bars (24–8), then another four (33–6), before the piece dies away triple piano in a variant of the opening material.

For the pianist, the difficulty of this incomparably subtle music lies, first, in observing the complexities of Debussy’s instructions on dynamics and articulation, and, secondly, in pedalling, without over-pedalling, music so dependent on resonance, for which Debussy gives no instructions at all. Paul Roberts has explained the importance of pedalling in the first two bars, a ‘simple’ five-finger exercise, down and up, with the top note held, the bottom note accented (slightly lengthened), and the intervening notes semi-staccato.37 Should all five notes resonate, with the sustaining pedal held down; or is it a question of half-pedalling – with the dampers almost touching the strings – giving a delicate blurring of each note but avoiding the Turneresque fog that passes for Impressionism with certain pianists who may remain nameless? Too much pedal obliterates the all-important sense of touch; too little leaves a dryness that cannot have been Debussy’s intention. This problem is if anything still more acute in the next Image.

One of the many intriguing questions about Et la lune descend sur le temple qui fut (‘And the moon goes down over the temple of old’) is whether Debussy had any music in his head when he thought of the title. For a long time it was accepted that the title, with its oriental flavour, came from the orientalist Laloy, the dedicatee of the piece. But it already figures on the contract Debussy signed with Durand in 1903, before he knew Laloy well. More to the point, as the eventual music hardly seems to represent anything specific about the title beyond its generally mysterious atmosphere, one might suppose that Debussy came up with one he liked the sound of, then wrote the music he wanted to write, possibly changed the title, then reinstated the original, as is proved by an otherwise incomprehensible remark to Durand in the letter accompanying the Images. Was it just the sound, the metre of the title, ‘a faultless alexandrine’, as he joked to Durand?38 Or was it the imagined scene, a Japanese print to be set to music, in the way that the third Image, Poissons d’or, is a lacquered bowl set to music?

Whatever it represents, Et la lune descend sur le temple qui fut is one of the most perfectly imagined and executed of all Debussy’s piano pieces. In a sense it establishes its own sound world with a single chord, an open fifth (E–B) with a hanging dissonant whole tone (A) attached to the upper note like a limpet, repeated more softly, in the manner of an echo, then with the same chord attached to each note of the ensuing melody as a particular colour: a colour, not a harmony. In fact the colour varies slightly. At one point, for the sake of fluency, the whole tone becomes a semitone, and on the last two chords of the opening phrase the open fifth contracts to an augmented fourth, as if questioning the previous chords. Debussy’s judgement here is incredibly refined. Play this phrase with the last two chords the same as the others and it will sound flat and closed, leading nowhere. Play it, as Cécile Ousset does in her recording, with the final chord contracted to a perfect fourth, and it sounds banal. In the second phrase the chords gradually evaporate, leaving the unison B, with or without whatever resonance the player chooses to allow to survive through the sustaining pedal.

To describe this music and what follows – the parallel triads, the two-part counterpoints of vaguely oriental melody, the different variants of the opening chord sequence – is easy enough. To explain its unique, haunting beauty and its formal perfection is another matter. The sequence of apparently unconnected events is like an irregular series of lines of poetry in a half-understood language. But just as verse is bound together, however informally, by metre and rhyme schemes, so Debussy’s piece is linked by its procession of balanced phrases, its alternating reprises, and crucially by the note B, which is the top note of the first and last chords, begins or ends most of the intervening phrases (either as a single pitch or as a chord), and is left hanging on its own at the very end, probably without the reverberation of the previous bar, though as usual Debussy is silent on the issue of pedalling. His own description of Musorgsky’s Nursery as ‘made up of small successive touches, bound together by a mysterious linkage and by a gift of luminous clairvoyance’ could just as well have been prompted by Et la lune descend. His other remark, about the ‘sense of shuddering, restless shadow that envelops and grips the heart to the point of anguish’, might overplay the element of emotion in this restrained, enigmatic Image. But context is all-important, and it’s precisely the restraint that empowers Debussy’s discreet expressif markings, and gives his single très expressif (bar 32) the force of a minor explosion.

The final Image, Poissons d’or (‘Goldfish’ or perhaps, in view of the supposed inspiration, ‘Golden Fish’), is altogether more specific in its imagery, almost a return to normality after the mysterious excursion to the temple of the setting moon. Debussy owned an exquisite Japanese lacquered panel depicting two goldfish – one light, one dark – swimming in a stylised river overhung by a golden-leafed willow, against a black sky. (The panel can now be seen in the Musée Debussy at his first home in Saint-Germain-en-Laye.) This panel may or may not have been the direct inspiration for Poissons d’or, but the music suggests it was. Somehow, partly through the angle of the fish to each other, partly through the slope and implied flow of the water, the anonymous artist has achieved an extraordinary sense of swirling motion, an icon of realism in an essentially stylised context. Debussy converts all this, con bravura, into a musical image of flowing water and pirouetting fish. The smooth, rippling waters of the opening are soon invaded by all kinds of fishy activity: darting, plunging, ‘capricious and supple’, almost as if the composer had sat down at the piano and improvised a variety of musical representations of the panel before his eyes. One or two details might indeed appear capricious on the part of the composer; at one point, the fish seem to pause while the pianist throws in two or three fragments of blues (‘expressive and without strictness’) a year or so before Debussy could possibly have known about that genre; then they laugh and go back to their pirouetting.

One can talk about some Debussy in this programmatic way. But, as ever, in turning such images into music, he is concerned exclusively with musical values. Though formally free (after early hints at a rondo), Poissons d’or is immaculately designed and balanced, with just enough of a climax generated, so to speak, by the incessant motion of the figuration, and a firm tonal framework (F sharp major) within which the chromatic fish can cavort without muddying the waters too much. As usual with Debussy, the default dynamic is piano dying on the final pages to pianissimo then triple piano at the end. Refinement and clarity are required of the player throughout; clumsy virtuosity has no place in this music. Whether or not Debussy specifically intended these three pieces to be played as a set, they make perfect sense that way, and were in fact done complete by Ricardo Viñes when he premiered them in February 1908: the bells, the leaves, the moonlit temple, the river, the fish – a series of panels from an imaginary gallery of symbolic images, with every degree of subtle colouring, texture and motion, concluding with the most brilliant and the most delicate.

Debussy might have boasted to Durand that the second set of piano Images was written ‘with the skill born of habit’. In fact he made this remark about the orchestral Images, and not as a boast but as a confession. Writing from Pourville that August (1907) a slightly more guarded version of his usual promise to finish the work soon, he told his publisher that there had remained ‘a lot of places that bothered me … it was well written, but with that skill born of habit that one has so much difficulty conquering and which is so tedious’.39 What these bothersome places were we have no means of knowing, but one might guess that they had to do with linkage, continuity and how to end. The question of montage arises also in the keyboard pieces, because of Debussy’s technique of the static moment connected to the next by an instantaneous switch of angle or texture, with or without the shared notes that provided the logic for composers such as Schubert or Bruckner who cultivated modulation to remote keys. In the orchestral pieces, however, the issue of changing colour is more acute, because of the many different timbres involved, added to which Ibéria, especially, has a specifically cinematic kind of form with graphic effects of montage.

The other problem may have been the orchestration of what, as we saw, had started life as music for two pianos. Rondes de printemps, he explained to Durand, was ‘immatérielle’ (probably best translated as ‘insubstantial’) and not to be handled like an ordinary symphony. As for Ibéria, here an old rivalry would soon rear its head. Debussy certainly knew Ravel’s Alborada del gracioso, as yet a piano work only, and he may or may not have known that his younger colleague had been working on an opera about Spain, L’Heure espagnole. There had been a rather too obvious similarity between La Soirée dans Grenade and Ravel’s Habanera, and there would soon be another near-collision, with Ravel’s Rapsodie espagnole, not composed – initially as a two-piano work – until October 1907, but then rapidly orchestrated, and premiered at a Colonne concert in March 1908, several months before Debussy completed Ibéria. Both composers’ Spanish music depends to a significant extent on style clichés, particular dance rhythms, flamenco idioms of melody and ornamentation, and also images of rural Spain and its people. Avoiding each other’s take on these matters was plainly no trivial matter in the small world of Parisian music.

As it happens, the first movement of Ibéria is closer to a less threatening model, the Fandango finale of Rimsky-Korsakov’s Capriccio espagnol, which Debussy had first heard at the Trocadéro concerts in 1889, conducted by Rimsky himself. Debussy’s own fandango first movement is more complex, with subtleties of rhythm and metre presumably intended as a response to the displaced accents of the Andalusian dance, with its on-and offbeat stamping and clapping. One wonders, equally, whether he had yet encountered those parts of Albéniz’s Iberia piano collection that were being published in Paris in 1906 and 1907.40 But his own approach, notwithstanding the music’s supposed origin as a work for two pianos, is so quintessentially, meticulously orchestral in its detailing as to distance itself, not only from other people’s piano music, but from his own. Instead of the self-contained, whole-cloth quality of the keyboard Images, Ibéria feels composed in a linear, almost anecdotal fashion, its music passed from hand to hand like the baton in a relay race, Par les rues et par les chemins (‘Through the streets and byways’), as Debussy in fact called the piece. Halfway through, horns and trumpets announce what amounts to a rival dance in the rhythm of a tarantella, with brilliant fanfares, and a brief expressive climax (violas and cellos) where the composer seems to step aside from the physical dance and reflect on its emotional undercurrents, after which the music reverts to the original fandango.

In the same way the second movement, Parfums de la nuit (‘Perfumes of the night’), mingles observation and passion, but inextricably, so that the nocturnal atmosphere, its rustlings and chirrupings and distant fragments of melody – including the daytime melody of the first movement – become images of the emotions they arouse, those intangible feelings of love, sensuality and regret that seem inseparable from hot southern nights. We are back in the world of La Soirée dans Grenade, with the same habanera rhythm, but less in the thick of things, less actively engaged. We stand in the gardens and simply listen to – and smell – the intensity of the darkness, just as we previously listened to the bells through the leaves, the one sense – or two – turned into music, and back again.

From this experience we wake up to the morning of a feast day (Le Matin d’un jour de fête), by way of a transition that brings into play Debussy’s montage technique, the dying night intercut with the dawning day. But the day itself is also a montage, of singers and guitarists, a solitary violinist, the laughter of schoolboys and the braying of donkeys. The whole piece is spectacularly scored and makes a brilliant finish to Ibéria as a three-movement suite in its own right. Musically, though, it feels riskily lightweight, with a flavour of the travel poster or the tour guide, and a disconcertingly abrupt ending that might suggest that Debussy was having difficulty extending his snapshot material into a fully fledged narrative. As a whole Ibéria lives by the beauty of its colourings and the vividness of its atmosphere, rather than by the accumulated structural power of La Mer or the intense focus of the piano Images. But at least it adds up to a substantial concert work with a beginning, a middle and an end, all beguilingly imagined and heard. The exact role of the other two Images was to prove much harder to define.

Debussy’s description of Rondes de printemps as ‘immatérielle’, feather-light in texture and content, located it as far as possible from the symphonic and operatic traditions of the day as represented by, say, Mahler or Strauss, or in France by Franck or, recently, Dukas, whose Ariane et Barbe-bleue Debussy had found beautiful but somewhat implacable when he attended its dress rehearsal in May 1907. If anything, Rondes de printemps has the opposite fault of being insubstantial in the negative sense, another possible but unintended meaning of immatérielle. Ideas flow past, exquisitely but a touch shortwindedly, a quality highlighted by the tendency to play each phrase twice; and the now familiar children’s songs – ‘Dodo, l’enfant do’ and ‘Nous n’irons plus aux bois’ – seem to come out of the blue, like old standbys in adversity. Admittedly they fit the image of the piece. ‘Vive le Mai,’ runs the epigraph at the head of the score, invoking the Tuscan spring festival of the Maggiolata, ‘bienvenu soit le Mai, avec son gonfalon sauvage’ (‘Long live May, welcome, May, with its wild streaming pennant’).41 Debussy’s spring rounds are another of his 1908 tributes to childhood. Ideally they might provide the scherzo for a larger but more conventionally formed work. But their position in the whole set of orchestral Images has always raised questions.

‘These poor Images’, as Debussy called them after handing over Ibéria and Rondes de printemps to Durand, were never published as a set, even when eventually completed with Gigues, only as three separate works. When Debussy conducted the complete premiere, in January 1913, he played them in the order Gigues – Ibéria – Rondes de printemps, and this order has tended to stick, on the comparatively rare occasions when all five movements have been played, even though the elusive Rondes de printemps makes a less than ideal finale, not least because its own actual ending is somewhat contrived. André Caplet, who worked with Debussy on the orchestration of Gigues, conducted the Images after the composer’s death in the order Rondes de printemps – Gigues – Ibéria, which has the virtue of ending with the strongest music. But the problem is ultimately intractable, and lies with the fact that the concept changed from a book of pieces to a multi-movement concert work, without sufficient corresponding change to the contents.

Debussy began work on Gigues a few days after handing the other two over, but it would be almost four years before he finished the seven-minute work. At the start he was suffering rectal pains, apparently caused by haemorrhoids and severe enough to be treated with ‘morphine, cocaine, and other nice drugs’, as he grumbled to Durand a month later.42 But during that same period, as we shall see, he was writing works that belong with his most perfect, so the problems with Gigues can’t reliably be blamed on illness. They were inherent in the idea of adding an English image to the Spanish and French ones already completed, or rather in the character of that addition. According to Lockspeiser this derived from a song by Charles Bordes, a setting of Verlaine’s poem ‘Streets’ called ‘Dansons la gigue’, based on the tune of the Northumbrian folk song ‘The Keel Row’.43 Since Debussy certainly knew Bordes, a former fellow pupil of Franck’s and one of the co-founders of the Schola cantorum, Lockspeiser’s is as good an explanation as any of Debussy’s appropriation of that tune, though there is no particular evidence that he knew Bordes’s piece (composed in 1890), and ‘The Keel Row’ was a pretty well-known song. In any case, Debussy was taken with it and used it to the point of saturation in Gigues.

The result is intriguing: sometimes successful, sometimes less so. For the most part, Debussy uses only the first line of the tune (‘As I came thro’ Sandgate’), sharpening the fourth and flattening the seventh, as in L’Isle joyeuse, so that the effect is curiously melancholy, especially in conjunction with an achingly beautiful oboe solo that hints at the spirit, if not the music, of Wagner’s cor anglais in the final act of Tristan: a gigue triste, indeed. But when the music speeds up, the incessant, bouncy dotted rhythms of the folk tune begin to sound slightly crude. Lockspeiser, who greatly admired Gigues, found these effects ‘subtly ironic’, but it’s hard to believe that Debussy was setting out to make fun of his material, any more than one feels that way about his treatment of ‘Nous n’irons plus aux bois’, where the issue is more a mildly sentimental version of nostalgie de la boue. The truth is, surely, that Debussy does many lovely and some clever things with this actually very fine folk melody, but in the end expects too much of it, both melodically and rhythmically. It’s striking that he uses the answering title phrase, ‘O weel may the keel row’, only on a single page late in the piece, almost as an afterthought.

I’ve expressed reservations about these Images. Yet all the time when one listens to them one is listening to authentic Debussy, everything heard and balanced to a degree of refinement beyond the reach of almost any other composer, and with breathtaking moments that remind one that a genius is a genius even in his difficult times. A page of Gigues is worth a hundred pages of – well, I leave the name to be filled in by the reader, in the spirit of the late lamented M. Croche.