Doing Opera

1.

One of the more amusing hazards of producing opera in Great Britain is the occasional backstage visit of a royal patron. As she makes her way down the awkwardly curtseying line of principals, often more lavishly dressed than the titled visitor, aides and equerries who bring up the rear politely quiz the producer about his job. “Presumably you have to be here every night.” “No, not exactly.” “Oh really! I thought you had to stand in the wings and tell the singers where to go.” “Well, the thing is they know that by the time we open.” “I see. Then what is it you producer chaps actually do?” Well, yes, that is the question.

The difficulty is that unless you attend rehearsals day by day, not to mention the many discussions during which the design is worked out, it’s quite hard to identify those aspects of a performance for which the producer is responsible. The conductor is visible throughout the show and the audience can see, or more often than not, thinks it can see his contribution to the evening’s events. The same goes for the singers. And although he’s not present in person, the work of the designer is there to be seen and often applauded before there are any signs of what you might call production. In fact unless the settings and costumes conspicuously depart from tradition, in which case the producer is usually blamed for encouraging such distracting anomalies, it’s widely assumed that the design is immaculately conceived before he arrives and that otherwise his work is confined to telling the singers “where to go.”

As it is, there was a time, little more than a hundred years ago, when operas, like plays, got themselves on without the help of a producer and there was, as yet, no distinction between the work and how it was put on. The reason is that throughout the eighteenth and much of the nineteenth century a large proportion of the repertoire consisted of works appearing for the first time, and since their staging was unconditionally determined by the theatrical conventions which the composer and librettist would have had in mind when they wrote the work, production as we now think of it wasn’t an issue.

The emergence of mise en scène as something requiring a credit in the program is associated with the development of a repertoire in which operas with an intermittent history of previous productions began to outnumber new ones, a tendency which has now reached the point where so-called classics fill the season and the presentation of original works is a relatively rare event. I’m not suggesting that this was enough to account for the appearance of an unmistakably auteur style. On the contrary, the once-unimaginable policy of reviving operas from the past was well established by the time audiences had occasion to applaud or deplore something which had been “done” to an old favorite. Nevertheless, until operas were given the opportunity of becoming “old favorites,” that is to say, until they began to have a recurrent life in the theater, so that there was now a prospect of their being revived in a cultural environment that was recognizably different from the one in which they were first performed, it’s difficult to imagine how regie, or production, would have got going, let alone assume the controversial importance it now has. Because it’s only by outlasting the period for which it was created that a play or an opera runs the risk of being perceived under a different theatrical description.

Depending on the length of time between its initial run and what one could reasonably describe as its first revival, an opera will eventually disclose, or seem to disclose, aspects which would have been invisible and perhaps unintelligible at the time of its première. And if, as in many cases, the original production has already been dismantled and forgotten, there will be an understandable temptation to start from scratch and develop previously unimaginable stage versions.

The traditionalists would argue that this temptation should be resisted at all costs and that the producer should assume the role of a self-effacing restorer, bending his ingenuity, such as it is, to the faithful reproduction of the staging which realized the composer’s original intentions. “You wouldn’t repaint Piero’s Miracle of the True Cross, so what gives you the right to vandalize Rigoletto by setting it in anything other than Gonzaga Mantua?”

On the face of it this sounds like a reasonable argument and yet the examples are not strictly comparable. In the case of a painting, the artwork is nothing more and nothing less than the unique object bequeathed to posterity by its maker. Additional marks made by anyone other than the artist automatically compromise its autographic identity. But when it comes to plays, operas, or symphonies, where it’s impossible to identify the work with any particular artifact, it’s difficult to say what, if anything, would count as an act of vandalism. Since it would undoubtedly be an act of vandalism to destroy a painting, and thereby deprive posterity of its continued existence, might one also say that it would be vandalism to conclude the first run of an opera without the guarantee of future performances? After all there is a sense in which an opera ceases to exist after what might be its last performance ever. But not really, because in contrast to a painting, which is irretrievably annihilated when the artwork is destroyed, the score of an opera outlives what might be its last performance, so that there is always the possibility of its being revived at a later date.

Is there any way in which one could vandalize the score then, thereby compromising the identity of its subsequent performances? Someone could deface or otherwise modify the composer’s autograph manuscript and that, as J.L. Austin might have said, would be a crying shame, but since there are probably authenticated copies of the original, nothing would be lost apart from the admittedly priceless example of the composer’s hand. Because scores and scripts are nothing more than instructions, and as long as a copy legibly reproduces what the original manuscript specified, the fact that it is printed rather than handwritten is neither here nor there.

In that respect the situation is comparable to the reproduction of biological organisms, in which the development of each short-lived organism represents the “performance” of an inherited script. With the death of the organism, a particular performance vanishes forever, but since the genetic instructions are copied and handed on, the possibility of future “performances” is guaranteed. This doesn’t mean that each individual is an exact copy of its predecessor and that what August Weissmann called the continuity of the germ plasm implies invariability from one generation to the next. On the contrary, even if one ignores the heritable variations introduced by sexual reproduction, not to mention unsolicited mutations—for neither of which there is any equivalent in the case of scripts or scores—the way in which the inherited instructions are expressed or realized is strongly influenced by the environment.

That is why biologists recognize the distinction between the genotype and the phenotype. The genotype, which is represented in and by the biochemically coded chromosomes, dictates what type of individual will develop, i.e., what species it belongs to. However, the physical circumstances in which development proceeds exerts a significant effect upon the individual expression of these instructions. For example, in certain aquatic plants, the leaves which develop beneath the surface of the water are finely dissected or feathery, whereas the leaves which develop above the surface of the water have a more rounded outline. And yet both sets of leaves inherit the same packet of genetic instructions.

Although this is a helpful analogy, it can’t be taken too far because the expression of genetic instructions is a mindless process, whereas the performance of a score or of a script is cognitively mediated. That is to say, it’s the result of conscious interpretation on the part of someone for whom the instructions mean something. In which case the traditionalist would insist that the composer’s meaning should take precedence and that even if the circumstances are different, the producer has an inescapable duty to honor the original intention. The bother is that it’s not all that easy to see how this self-denying principle could be realized.

One method might be as follows. Since the composer was presumably present throughout the rehearsals of his work and was therefore in a position to advise and object, it seems reasonable to assume that the inaugural production captured and expressed most of what he meant, so that the decent thing would be to reproduce this approved prototype as closely as possible. In other words we’re talking about reviving an opera by restoring a particular production of it. This is something with which modern opera houses are all too familiar, although not for the reasons I’ve just described. Let me explain. Apart from mounting brand new productions of operas which have already had previous but now defunct revivals, managements fill out their seasons with frequent revivals of old but still-extant productions. Thus, in any given year, we might have the tenth revival of so-and-so’s twenty-year-old production of Cav and Pag, say, or the third revival of some other producer’s version of La Bohème, and so on.

Now, with their sets and costumes still in existence, not to mention videotapes of the performances, you’d think it was easy to resuscitate any one of these productions. Well, up to a point. The success of the enterprise depends, to some extent at least, on how much time has passed since the last revival of the production in question. As long as they have been carefully maintained, the set and costumes will look just as they once did, but recovering the performance is another matter altogether. In all probability the cast will have changed in the interim and it would be insulting to ask the new singers to watch the videotape and imitate what their predecessors did. Even if you could persuade them to try, copying what someone else is doing is more complicated than reproducing their movements.

The movements have to be recognized as meaningful, and although this may be self-evident you can never be sure. It needs someone who was there at the time to explain what was meant and what was actually going on. More often than not the original producer who might have explained is otherwise engaged, and the assistant who deputizes for him is not necessarily competent to put the case in his own words. And what’s worse, if the assistant is new to the production, he or she is back to square one when it comes to interpreting the helpful recording. The result is that with each subsequent revival the staging drifts further and further away from the prototype.

If this can happen with a relatively recent production, that is to say, one that has been dormant for little more than five years, imagine the difficulty of reinstating the now defunct but supposedly authentic inaugural production. After a hundred years or more, the set and costumes will have vanished without a trace, there never was a videotape, and in all probability there’s not even a prompt copy which might have said where the long-dead singers “went” and why they did so.

But even if there were some unachievably magical technique for overcoming these difficulties, so that the inaugural production could be restored in all its pristine authenticity, the chances are that the intended meanings it so eloquently expressed at the time would no longer communicate themselves to a modern audience. Theatrically speaking, the production would seem quaint and antiquated. Oddly enough this doesn’t apply to musical restoration. On the contrary, recent efforts to bring back so-called original orchestrations have proved remarkably successful, especially when it comes to the baroque repertoire. After nearly three hundred years of almost silent hibernation the musical performance of Monteverdi’s operas has been lovingly restored, and far from sounding quaint, the scores played on original instruments strike an altogether refreshing note. And yet, even for audiences who pay lip service to the notion of conservation, the occasional attempts to reinstate a correspondingly authentic staging of L’Incoronazione di Poppea, say, have been much less successful.

However, I suspect that this has something to do with the unfamiliar artificiality of baroque stagecraft, and that when audiences insist upon authenticity, what they are actually expressing is a preference for the picturesque realism exemplified by the traditional though not necessarily authentic staging of nineteenth-century historical operas. Even so, if you are prepared to put up with the initial fuss, it’s surprisingly easy to persuade an apparently conservative audience that there’s a legitimate alternative to the stagings that they would regard as canonical.

2.

When I joined the English National Opera twenty years ago, the idea of doing Rigoletto in anything other than doublet and hose would have been inconceivable, and when I managed to persuade Lord Harewood that a Godfather version set in Little Italy would capture Verdi’s meanings just as well, there was general consternation among the subscribers. In the event, though, the gamble paid off, and before long this glaringly inauthentic production settled down to become a steady favorite and after at least ten revivals it still plays to packed houses.

A few years later I pitted my efforts against the notorious conservatism of an Italian audience, this time with a production of Tosca. As soon as the news leaked out that an upstart English producer was about to transpose the opera into the world of Rossellini’s Open City, there was an indignant outcry from the local public. We were threatened with mass picketing and the Christian Democrats in the Florence Commune discussed the withdrawal of the municipal subsidy. On opening night, the tension backstage was something I’ll always remember. And yet when the curtain came down three hours later, the applause was deafening and the production was revived in the following year without a murmur.

I boastfully include these examples to suggest that the proof of the pudding is in the eating and that the notion of “limits,” so often cited by the traditionalists, is more or less meaningless. At the same time, I have to admit that there are productions which undeniably “mess up” the work, not by going too far, but by the clumsy application of what is tendentiously claimed to be a concept. In Germany for example, where I was once warned in all seriousness that without a “concept” I would have “great problematics with my praxis,” productions are often disfigured by half-baked political ideas, such as the one I was offered as an explanation for a recent version of Figaro. When I asked why the characters wore their eighteenth-century clothes inside out, I was pityingly informed that it symbolized the corruption of pre-revolutionary society and that it was after all the English who had coined the phrase “the seamy side of life.”

Although the example I’ve just cited is self-evidently absurd, it is symptomatic of a misconceived urge to exploit “theory” in the name of relevance. The fact is that one way or another, I have always had “problematics with my praxis,” not, as predicted, because I’m reluctant to exploit concepts, but because unless you’re a naive traditionalist, there is something inescapably problematic about reviving operas from the distant past. But it’s a question of thoughtfulness rather than theory.

Figaro for example is too delicate to bear the weight of a “concept,” especially if it encourages the producer to illustrate the corruption of the period or to represent the hero as a sans-culotte manqué who knows that his master’s days are numbered. As with The Cherry Orchard, the characters’ blissful ignorance of the forthcoming revolution lends the opera an irresistible autumnal melancholy, which the audience supplies without having to be didactically nudged. And as for the supposedly invidious social relationships, they speak for themselves, as long as the producer has taken the trouble to represent the now well-documented details of domestic life in an aristocratic household.

The questions that interest me may seem trivial, but when you add them all together you have a reasonable chance of creating an intelligible social life, from which the audience is free to infer some political significance. What services does Susanna provide for her mistress? Where would she fetch the water for the Countess’s morning toilet? Chardin provides a useful picture of a maid stooping to fill a jug from a large copper cistern. And then there’s the room which the Count has so helpfully provided for the newly engaged couple. How would it be furnished apart from the bed mentioned at the beginning? The prompt reply to each of the two bells suggests that Figaro and Susanna will live and sleep where they work, and that in turn means that ironing, needlework, and tailor’s dummies will be in evidence, not to mention wig stands and neatly piled changes of bed linen. In other words it’s a storeroom and a workshop with precious little space for the letto matrimoniale.

And what happens if you introduce the Countess’s children? Admittedly they’re not explicitly mentioned in the libretto, but at the same time there’s nothing to suggest that they don’t exist, and since the Countess conceives a child by Cherubino in the third play of Beaumarchais’s trilogy, it would be rather odd if she’d failed to produce legitimate offspring. In any case it’s a reasonable possibility, and as soon as you allow these unmentioned children, a girl of six, say, and perhaps a baby of six months, something intriguing starts to happen, especially if the two children are ushered in and hustled out during the long introduction to the Countess’s second-act aria. We know, for example, that few if any upper-class women suckled their own babies, and since toddlers were taken back to the nursery as soon as they’d paid their respects in the morning, the entrance and exit of these unexpected infants stresses the solitude which gives rise to the memorable lament which follows. It’s hardly surprising that, neglected by her husband and denied the comforting intimacy of her own children, the Countess is so friendly with her maid. Who else can she talk to? At the same time, decorum requires Susanna to unquestioningly obey her mistress’s peremptory demand for a bandage and a nightcap. Between Figaro and the Count there is a comparably subtle interplay of deference and defiance, all of which demands minute attention to the Goffmanesque rituals of their daily relationship.

Taken one by one, none of these carefully encouraged details would be conspicuous by their absence, but when they are all included, along with many other social observations, the cumulative effect is quite remarkable. Almost effortlessly, the audience gets the uncanny impression of spending one long summer’s day in the otherwise unvisitable past.

Although this reticent strategy works quite nicely with an opera set in a world with which the composer and librettist would have been personally acquainted, a work such as The Magic Flute requires a more managerial attitude on the part of the producer. For all its magnificent music, the Flute can be a tedious theatrical experience, particularly when the Egyptian setting is taken literally. With its deadening solemnities, punctuated by magical high jinks and routine comic shtick, the opera can come across as a priestly prank with a message. You can get some idea of what the traditionalists expect from the question that the management invariably put to the producer who’s been invited to stage The Magic Flute. “How,” they ask, “are you going to bring on the Queen of the Night?” The suggestion that she might walk on is usually met with shocked disbelief. “How can that be? You can’t ask someone like that to walk on!” And indeed in traditional productions she enters on something that looks like a Mardi Gras float, as if she’s topping the bill in Aztec Night at the Copacabana—a bizarre hybrid of Yma Sumac and Carmen Miranda. That, apparently, is how “someone like that” is expected to get onto the stage.

Now although it would be an exaggeration to suggest that the entry of the Queen of the Night is the key to an acceptable production, it’s unarguably true that the way in which she enters depends on the sort of character she’s supposed to be, which means, in turn, that the producer has to identify or perhaps stipulate the possible world in which she occurs. The possibilities are somewhat limited by the fact that, as its title implies, the opera requires an unavoidable element of magic. Even so, if one makes a serious effort to visualize what Mozart and Schikaneder might have meant when they went to such lengths to include the Freemasons, the chances are that more significant aspects of the eighteenth-century mind would become visible. At the risk of recommending the usefulness of a “concept,” I would suggest that The Magic Flute is an opera for which indirect references to the French Revolution might be helpful and indeed legitimate.

In contrast to Figaro, which was composed in ignorance of these forthcoming events, Flute was written eighteen months after the fall of the Bastille, and since he died before the onset of the Terror, Mozart might have felt, as Wordsworth did, that it was bliss “in that dawn to be alive.” In fact, as Jean Starobinski points out, the opposing themes of darkness and light are too frequently repeated to be an accident.^* With Sarastro’s triumphant declaration that “the rays of the sun have driven away the darkness of night,” it’s difficult to avoid the conclusion that, in some respects at least, The Magic Flute is a millennial work, celebrating the rebirth of humanity under the auspices of Reason and Justice. In which case, the Masonic chorus can shed its implausible Egyptian “drag,” and appear instead as the enlightened eighteenth-century gentlemen Mozart and Schikaneder would have known as fellow members of the Viennese Lodge. No need now for that pompous pantomime which opens the second act. The scene is more convincingly realized by re-creating something like John Trumbull’s tableau of the signers of the Declaration of Independence.

With these adjustments, the rest of the production starts to fall into place. By representing the Masons as they would have been at the end of the eighteenth century, a production can eliminate the Temple and replace it with a Masonic Library, based on the designs of Ledoux or Boullée. During the overture, Tamino begins to drowse over his occult literature and the serpent, like other monsters bred by the sleep of reason, emerges from his dream.

Now, at last we can reinvent the Queen of the Night, so that she can get onto the stage without wheeled transport. If Sarastro is the Master of an enlightened Lodge, it seems reasonable to represent his opponent in comparably naturalistic terms—as a Catholic monarch of the ancien régime, reminiscent of the Empress Maria Theresa, whose raid on the Viennese Lodge may have been the inspiration for the penultimate scene of the opera.

Papageno is another character who would almost certainly benefit from this type of treatment. He is, after all, the epitome of Rousseau’s “natural man,” so that although he catches birds, there’s no conceivable reason why he should look like one, and once he sheds those wretched feathers, his otherwise insufferable cuteness vanishes, and instead of Tweety-Pie, what we see is an amiable eighteenth-century peasant, happy, by his own admission, to live by eating and drinking. All at once he becomes an intelligible and interesting contrast to the princely figure of Tamino, and we can readily sympathize with his common-sense refusal to obey the vows of silence.

Apart from the fact that it would probably disappoint audiences conditioned to visualize it as a fairy tale, the disadvantage of naturalizing The Magic Flute to the extent I just described is that it requires a certain amount of ingenuity to reconcile it with some of the distinctly unnatural events that take place in the opera—the serpent, for example, the musically enchanted animals, and of course the unarguably supernatural ordeals of Fire and Water. These episodes, which would be perfectly acceptable in the dateless elsewhen of “once upon a time,” seem slightly out of order in the world of the eighteenth-century Enlightenment. This difficulty is easily overcome by bracketing the whole production within a dream, in which, as in a fairy story, anything goes.

When it comes to most of the other works in the standard repertoire, the producer is faced with a different type of problem altogether, though judging by the productions to be seen all too often in the more conservative houses, the difficulty I have in mind is scarcely acknowledged, let alone dealt with. With certain conspicuous exceptions—La Traviata, for example—in which the action could have taken place around the time when the opera was written, most of the works in the standard repertoire refer to a previous period, sometimes several centuries earlier, and with the benefit of informed hindsight, or perhaps just the distant vantage point of the early twenty-first century, it becomes increasingly apparent that the events that are represented, and above all the sentiments that are expressed, are completely at odds with what we now know about the bygone world in which they’re supposed to occur.

Like the novels of Walter Scott, from which so many of their plots are derived, these operas represent the past in a characteristically nineteenth-century Romantic fashion, so that when they’re staged conventionally, the productions tend to resemble the floridly picturesque tableaus of the French salon, projecting the pompier style of painters such as Hayez and Delaroche. The difficulty is that if, in the understandable effort to avoid such kitsch, the producer commissions more accurate decor, that is to say decor which is historically authentic rather than operatically authentic, the result is not much better, since the appearance of the stage is now conspicuously inconsistent with what sounds like a nineteenth-century melodrama.

One way of dodging this difficulty is to do the work in concert or, as they say, “semi-staged.” An opera such as Il Trovatore, say, or La Forza del Destino, whose claim to the title “historical” is paradoxically compromised by what is fondly supposed to be a realistic production, can often achieve an unexpectedly truthful effect when it’s performed in front of the orchestra without costumes or decor, but with the singers allowed a modicum of acting. By the same token, although the paying public never sees it, the final run-through in the rehearsal room is often much more convincing than the lavishly scenic production which appears in front of an audience a few days later on the stage. With the entrances and exits marked out with tape on the floor, and perhaps a few token walls knocked together out of plywood, the action has a forceful intensity soon to be subverted by the spectacular “period” decor which audiences love to applaud as the curtain goes up.

I’m not recommending these as substitutes for fully staged productions, but something valuable is to be learned from such makeshift simplicity. Apart from the fact that it avoids some of the awkward aesthetic contradictions I’ve already mentioned, the relatively unfurnished mise en scène allows the action to sing for itself, and as Peter Brook’s productions at the Bouffes du Nord in Paris show again and again, the “empty space” dignifies the performer and restores some sort of mystery to the peculiar art of pretending to be someone else.

Nevertheless, I admit that audiences who pay what they maddeningly refer to as “good money” for their operatic entertainment would probably feel cheated by such austerity, and many of the sponsors, if they hadn’t vetoed the production to begin with, would go ape on the first night. In which case there has to be some other way of breathing life into the increasingly effete genre of nineteenth-century historical opera. I’m not seriously suggesting the Marx brothers, though as we know from A Night at the Opera they certainly helped to give Trovatore a lift. No, I’m referring to the widely exploited and deeply resented technique of “updating.” This can take one of two forms, the most controversial of which is straightforward modernization. An opera which now seems quaint and somewhat awkward when staged in the period in which it’s supposed to occur can often achieve a remarkable vigor when the action is brought forward to the present day or at least sometime within living memory.

Unlike some of my colleagues, for whom the destination of such impudent time travel is right now or not at all, I prefer to stop several decades short of today, if only to preserve a sense of historical distance or “otherness” comparable to the one unconsciously intended by the composer and his librettist when they backdated the action. So, as in the examples I cited earlier, I shifted Rigoletto no further than the 1950s and my production of Tosca was set ten years earlier. My production of La Bohème came to what I regarded as an interesting halt in 1930, carefully based on what now look like “period” photographs of Brassaï, Doisneau, and Kertész. And in order to avoid the touristic Spain of some of the more traditional productions, I packed off Carmen for a relatively short journey into the Seville of Cartier-Bresson. The Mikado, whose inaugural production is the subject of Mike Leigh’s current film Topsy-Turvy, survived its loss of Japanese decor and flourished when I transposed it into the Fredonia of Duck Soup. I can still recall the incredulous laughter when Eric Idle, playing KoKo, opened the letter from the Mikado and said indignantly, “I can’t read this, it’s in Japanese!”

There are, of course, many operas whose plots are such that they obstinately resist transposition, especially when they happen to include well-known historical characters such as Anne Boleyn or Mary Stuart. Neither of these Tudor queens “travels” well. Apart from the fact that they are celebrities famously stuck in their own time, there is no way in which they or anyone else could plausibly lose their heads on arrival in the twentieth century.

If these and other anachronisms compromise the credibility of the piece, there is no point in bringing the opera up to date, since one of the purposes of doing so is to reduce the dramatic inconsistencies associated with the traditional setting. In which case, the most prudent policy is to concede the period as indicated, but without representing it in embarrassing detail. In other words, as long as it suggests a world in which monarchs can believably order the execution of their unfaithful wives, the stage picture itself can be quite lean. However, this is a scenic idiom which has to be learned and understood, and as I’ve already indicated, audiences in some of the more conventional houses consistently misinterpret such reticence and deplore it as something done on the cheap. Witness the objections to Robert Carsen’s elegantly spare production of Eugene Onegin some years ago at the Met.

Transposition doesn’t necessarily mean modernization. There are certain operas in which the drama and music are so insistently reminiscent of the period in which they are composed that when they are reset accordingly, the theatrical effect is equivalent to a homecoming. Der Rosenkavalier is just one example. It’s not that the eighteenth century is conspicuously misrepresented. On the contrary, Hofmannsthal has conjured up a surprisingly plausible theatrical fiction which compares quite favorably with some of Verdi’s sixteenth-century romances. Still, at a distance of almost one hundred years, the Theresian setting seems comically inconsistent with Strauss’s waltzing music, and through the increasingly diaphanous veil of its eighteenth-century decor, the world of Musil’s Kakania becomes almost distractingly visible. As soon as it’s reset in the year that it was written, that is to say just before the onset of the First World War, the opera acquires an ominous wistfulness, so that it’s difficult not to hear the Marschallin’s first-act aria as a prophetic lament for what one of Strauss’s contemporaries suspected were the last days of mankind.

Pelléas and Mélisande is yet another work which benefits from being restored to the period in which it was composed. In contrast to Der Rosenkavalier, which seems to flourish quite comfortably in its traditional setting, Debussy’s opera is stiffened and disabled when the action is dutifully set in the Middle Ages, as the text indicates. The music has such a striking affinity with the appearance of some of Monet’s later paintings, especially the Bassins de Nymphéas, that it’s irresistibly tempting to try to find a literary counterpart to both. And what could be better than Proust? Apart from the fact that the reminiscent reflections and refractions of À la recherche du temps perdu bear a striking formal resemblance both to the music of Pelléas and to the paintwork of Monet’s lily ponds, Proust’s recursive allusions to the ancient nobility of the neighboring Guermantes allow the producer to reconcile the notional Middle Ages of Maeterlinck’s play with the fading world of late-nineteenth-century French aristocracy.

In any case, there are dramatic themes in Debussy’s opera which have almost exact equivalents in Proust’s novel. For example, the disconcerting scene in which Golaud forces his young son, Yniold, to climb on his shoulders and spy on what he suspects to be the adulterous lovemaking of his wife and half-brother bears a striking resemblance to Swann’s jealous lurking beneath Odette’s lighted window (see illustration on page 15). If the opera is rescued from the reproduction tapestry world of Maeterlinck’s Middle Ages, so that the action unfolds in a more recognizable social context, the work becomes much more energetic and intelligible. What’s more, it sets up a number of interesting expectations about the psychological consequences of casting a child in such a conflicted role—a question which would scarcely arise in a world in which children were not yet credited with inner lives.

How might this affect the mise en scène? Bearing in mind the way in which Proust recalled his own curiosity about what was going on downstairs when he impatiently awaited his mother’s good-night kiss, it’s not unreasonable to assume—or let’s be honest and say stipulate—that having had his prurient curiosity aroused in one scene, Yniold becomes an autonomous voyeur, so that in the production that Irecently revived at the Metropolitan Opera, I allow the child to lurk almost but not quite invisibly in a distant corridor so that he inadvertently witnesses the scene in which his father brutally abuses Mélisande for her infidelity.

This in turn allows a much more intelligible staging of the scene that follows immediately—that is to say, the episode in which Yniold tries to rescue a flock of sheep from their forthcoming slaughter. Here is where the device of the dream came to my rescue. Instead of having to do literally what the stage directions seem to suggest, the scene can be much more plausibly represented by having the child sing in his sleep, as he tosses and turns in the grip of a nightmare provoked by the violence he has just witnessed. And I can conclude the scene with a Proustian touch by having Mélisande arrive just in time to comfort her stepchild with a consoling good-night kiss.