You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Skip Navigation

Scandal and Scholarship

Who decides how an opera should be performed?

In July 1958, La Scala, an Italian periodical devoted to news of Italian opera and musical life, published a polemical article by a young Australian musician named Denis Vaughan. Having become an assistant conductor to Sir Thomas Beecham four years earlier, Vaughan had grown fascinated by the asymmetries of phrasing, the subtle gradations of color and dynamics, the non-uniform use of staccato articulation that he felt characterized Beecham's interpretations and gave the music that passed under his baton an inner life of great power and variety. Largely ignorant of the social history of Italian opera and its implications for the editorial processes through which nineteenth-century works were distributed in print, Vaughan was astonished to discover that the autograph manuscripts of certain works by Verdi and Puccini were significantly different from printed editions in circulation during the 1950s. Whereas printed editions offered relatively homogenous dynamics, articulation, and phrasing, the autograph manuscripts--read literally--showed marked asymmetries in phrasing, diverse gradations of dynamics, a selective use of accents, and so on.

For an admirer of Beecham's art, it seemed nothing short of a revelation. Convinced that his discoveries would prompt a reinterpretation of the art of these composers, Vaughan prepared an article called "Discrepancies Between the Verdi Autographs and Their Printed Editions," devoted to the Messa da Requiem and Falstaff, both of whose autograph manuscripts were available in excellent facsimiles. In his introduction Vaughan declared war on the Verdi interpretative tradition, and also on Verdi's editors at the publishing house Casa Ricordi:

The purpose of this study, rigorously critical, conducted on some original autograph manuscripts of Verdi and on recent printed editions of these scores, is to underline the great importance of the musical signs written by Verdi himself, and therefore clearly felt and wanted by him, for all that concerns melody, harmony, tempos, dynamics, phrasing, accents, and articulation. Signs that, although perfectly evident in the original scores, strangely have not been reproduced in the printed editions of the operas in question.
I give below some examples which will serve to demonstrate the enormous value that a critical edition of the works of Verdi might have, by following and reproducing with scrupulous observance the indications, quite precise, left by him. Let it be enough to point out that in the Messa da Requiem alone one can recognize as many as about 8,000 differences between the original and the print, while in Falstaff these differences mount up to 27,000.

Twenty-seven thousand differences! Vaughan's statements no sooner hit the press than an international furor erupted in the musical world.

For many years a picture of Verdi graced the equivalent of the one-dollar bill in Italy, where he has served for more than a century as a national icon. His most famous melodies are still in the air, hummed and whistled by members of every social sphere. Even if mass access to Verdi's music today is largely through television commercials, the symbolic meaning of that music remains strong, and the hold over the popular imagination of a composition such as "Va, pensiero, sull'ali dorati," the chorus of the Hebrew slaves in Nabucco, extends well beyond its musical beauties.

Verdi's carefully self-constructed public image cast him forward as a leading figure in the movement for national independence, and his operas from the 1840s are filled with moments whose potential relevance to contemporaneous political situations was not lost on his compatriots. During the revolutionary uprisings of 1848, the composer went so far as to set a libretto of explicitly patriotic sentiments, La battaglia di Legnano, whose final act is titled "Morire per la patria." And the chorus in Nabucco continued to resonate in the minds and the spirits of the Italian people. When the Teatro alla Scala was reconstructed after the bombings of World War II, the first music that resounded through its halls, under the baton of Toscanini, was "Va, pensiero."

Thus it is not difficult to imagine the public reaction to the idea that Verdi's scores had been so severely misrepresented that there were twenty-seven thousand discrepancies between the composer's autograph of Falstaff and the opera's printed edition. In journalistic circles, "discrepancies" quickly became "errors," and heated letters and denunciations circulated throughout the European press. Vaughan produced letters of support from many musicians and conductors, while others--led by Gianandrea Gavazzeni--ridiculed his examples. At a special concert in Milan, the public was asked to compare passages performed according to the "traditional" versions of the printed scores with the same passages as sanctioned by Vaughan. Parliamentary debates were held in Rome, leading to the foundation of the Istituto di Studi Verdiani of Parma, one of the primary tasks of which was supposed to be the preparation of a critical edition of the composer's works.

And yet Vaughan's campaign for scholarly authenticity fizzled out. There were several reasons, some tactical, some substantive. By mounting a publicity barrage about his "discoveries" and indulging in astronomical numbers of "discrepancies," Vaughan offended Italian national honor. Worse, he emphasized time and again the brilliance of Sir Thomas Beecham and the parallels between Beecham's approach and what Vaughan thought he saw in the Verdi autographs, while explicitly criticizing the scores in use in Italy and implicitly claiming that the interpretations of Italian conductors lacked the inner life that only the great English conductor had been able to achieve. In response, Gavazzeni wrote:

After the cataloguing of all that has allegedly been neglected in Verdi and altered in Puccini to the detriment of the composers' original inspiration and its expression in manuscript, what is Vaughan and with him the school for textual criticism and orchestral conducting of Sir Thomas Beecham trying to prove? Obviously that Toscanini and the other Italians after him who devoted their interpretive powers to the study of Verdi and Puccini mutilated the scores and betrayed the composers in their performances.

This is strong stuff, and Vaughan's credibility was inevitably damaged.

But three substantive problems were more significant: the terrain on which Vaughan joined battle, the logic of his argument, and his reading of the sources. By examining works for which Verdi is known to have participated in the editorial process (in however desultory a manner), Vaughan chose distinctly inhospitable ground. In the 1950s, nobody could accurately assess the extent of Verdi's participation in the editorial process for his late music, and the matter remains unresolved today. It is true that Puccini, another subject of Vaughan's polemic, certainly played a significant role in the printing of his own operas. Perhaps Puccini should never have abandoned the original, two-act version of Madama Butterfly; and perhaps he should not have depended on Toscanini to edit the dynamics and articulation in the printed full score of Manon Lescaut; and perhaps he should have reinstated the canzone dei fiori in Suor Angelica. Yet there is no evidence that this composer, who followed all stages in the dissemination of his works and often revised them for later productions, allowed fundamentally flawed printed editions to circulate. Thus Puccini's autograph manuscripts, however important they are for an understanding of his operas, cannot be considered a court of last resort when editing his music.

The logic of Vaughan's argument was equally problematic. From the generally true proposition that we should give great weight to signs actually written by Verdi in his manuscripts, Vaughan assumed that the absence of signs meant that the composer did not want them. This does not follow logically, nor does it reflect what we know about the preparation of autograph manuscripts by nineteenth-century Italian composers. And Vaughan proved to be a most inaccurate literalist in his reading of Verdi's autographs. He wrote, for example: "On this first page [of Falstaff] there are 125 discrepancies. On the first chord the ff is only for the oboes, bassoons, trumpets, and timpani, the first and second violins and the violas. The others are f." To which Gavazzeni responded: "Nothing and nobody will ever convince me that Verdi intended in the first bar to differentiate ff from f instrumental sections and instruments belonging to the same section." Quite apart from whether Gavazzeni could ever be convinced (and it is difficult not to agree with his musical instincts), Vaughan's list does not accurately reproduce the readings of Verdi's autograph, which has no sign for bassoons and trumpets but does have an unmistakable ff for cellos. The other instruments have no signs, and since when does the absence of a sign signify f?

The argument between Vaughan and Gavazzeni on the subject of slurs for one of the most beautiful phrases in Verdi's Requiem, the Lacrimosa, dies illa, is of breathtaking silliness. According to Vaughan: "Double phrasing; the melody is slurred in one part while the notes are separately articulated in the other. While the first bassoon, the solo tenor, the tenors of the chorus and the cellos, together with the third horn, have a singing legato, the third bassoon, the solo bass, and the basses of the chorus articulate the phrase. Verdi frequently uses this procedure, which was then copied also by Puccini. Thus, it is not an oversight of Verdi's." To which Gavazzeni retorted: "Just try humming the Lacrimosa staccato without slurring and then sing praises to the fetish of `double phrasing.'"

In fact, neither Vaughan nor Gavazzeni looked carefully at either the musical situation or the musical sources. Vaughan's description of Verdi's crowded autograph is an idealization: apart from his frequent misreadings, "slurs" are often groups of slur fragments, and a change of manuscript page in the middle of the melody confuses the issue further. His explanation, too, is filled with absurdities: although Vaughan treats the first bassoon as an "upper" instrument and the third bassoon as a "lower" one, these identical instruments are playing in unison. The presence or the absence of slurs in Verdi's autograph is actually a function of available physical space: there was ample space for a full slur for the choral tenors, and so Verdi wrote it; there was no room for a slur for the choral basses, and so he omitted it. Above the staff for the first bassoon Verdi easily wrote a slur (actually two slur fragments, meeting in the middle of a held note); above the staff for the third bassoon there was no room.

Gavazzeni's self-satisfied sneer was no more justified. What neither realized was that the slurring of the Lacrimosa, dies illa melody in all Ricordi editions of the Requiem through the mid-1980s had been dead wrong. It is slurred in multiple ways in its various appearances, but the diversities are determined by page turns in the autograph manuscript and the division of the music into pages and systems in printed editions. No conductor with any musical sense ever paid any attention to those printed signs, and performers instinctively treated the melody as a legato phrase. Blinded by his theories, Vaughan failed to understand the nature of the problem; and Gavazzeni, accustomed to hearing the melody as a legato phrase, failed to realize that the printed edition was faulty.

Twenty-seven thousand errors like this? Not only did Italian musicians turn their collective back on Vaughan's claims, they conveniently identified "musical philology" and the campaign for "critical editions" with his ideas: if this is what scholars mean when they demand "critical editions," they said, let us return to our vaunted tradition. A chorus line of relieved conductors could sing in unison, "If it was good enough for Toscanini, it's good enough for me." Vaughan's challenge had been met, and one could believe again that commercially available scores of nineteenth-century Italian operas were trustworthy.

There matters rested during the first part of the 1960s. Casa Ricordi, honestly believing that Vaughan's claims were without merit, found no pressing commercial reason to replace its editions, although the publisher did employ a local musician to correct obvious errors in the more popular operas. But meetings to establish a national edition of the works of Verdi led nowhere, because nobody knew where to begin. There were valuable biographical and critical studies pertaining to the composer, but no one had looked carefully at the manuscript sources or the printed editions, no one had analyzed Verdi's compositional process or his involvement in the performance history of his works, and no one had investigated his collaborations with his librettists. With the exception of the autograph manuscripts owned by Casa Ricordi, no one even knew which sources survived or where they were located.

Many people in the music world claim that during this period Casa Ricordi, anxious to avoid further scandals, grew selective about who was permitted to examine its archives. The opinion is so widespread that it is unlikely to be a complete fiction, but my own experience was different. When I arrived in Milan in the fall of 1966, a fresh-faced graduate student working on a doctoral dissertation, I was given access to the collection, and the personnel of Casa Ricordi never ceased to be helpful and interested far beyond what courtesy would require. And I was not alone. Intrigued by the Vaughan affair and nourished by a love for Italian opera, a group of young musicologists were patiently examining the sources of nineteenth-century Italian opera. They came from Italy, the United States, England, Germany, and New Zealand. They knew the impressive textual work being done on the new complete editions of the works of Bach and Mozart; they observed the Berlioz research in England, under the direction of Hugh Macdonald. Without crying scandal, these scholars began to turn their musicological training to Italian opera.

In light of the transformation in our knowledge over the past thirty years, it is difficult to imagine the spirit with which we began our work. I remember sitting in the reading room of the music department at the Bibliotheque Nationale in Paris in autumn of 1965, studying for the first time a complete score of Rossini's Maometto II. What an extraordinary work, I thought--and what a shame that I will never hear it, let alone see it on stage. I can imagine a Donizetti scholar, such as William Ashbrook, having a similar experience. And there were young Verdians, such as David Lawton and David Rosen, who became aware that Verdi left far more music for his operas (suppressed scenes, revisions, alternative arias) than printed editions contained. This scholarly work was beginning to constitute a foundation over which the question of critical editions of Italian opera could again be addressed when the time grew ripe.

We honor artists of the past by celebrating the centennials and other anniversaries of their births and deaths. The major celebration related to Italian opera during the 1960s was the hundredth anniversary of the death of Rossini. In 1968, Rossini was a one-opera composer. To be sure, everyone knew that he had written some forty operas, and there had been occasional twentieth-century revivals under Vittorio Gui. The Maggio Musicale Fiorentino in the early 1950s gave hearings to Armida (for Maria Callas), Tancredi, and La donna del lago; and Gavazzeni directed Il Turco in Italia (again with Callas). But in the minds of the public, Rossini was Il barbiere di Siviglia. Figaro, Figaro, Figaro....

When plans were laid for celebrating the centennial of the composer's death, it was to that opera that almost everyone turned. A young Italian conductor named Alberto Zedda was called upon to lead one of those revivals. During an earlier stay in America (as a conductor at the New York City Opera and a faculty member at the Cincinnati Conservatory of Music), he had directed a number of performances of Il barbiere. Several American wind players complained about peculiar readings in their parts, awkward melodic lines, unlikely rhythms. Since the Ricordi archives had no manuscript of the opera, Zedda decided to check these readings directly with Rossini's autograph manuscript, which is preserved at the Bologna Conservatory.

Seeking neither scandal nor publicity (the Vaughan fiasco was still smoldering), Zedda carried his score of the opera from Milan to Bologna, together with instrumental parts rented from Ricordi, on which he planned to enter his emendations. Having never before examined an autograph by Rossini (or any other composer), he had no points of comparison. Unable to identify securely Rossini's hand, he believed the recitativo secco in Il barbiere to be by Rossini, whereas those pages are actually in other hands. He did not know that Rossini had later prepared additional music for his work, which premiered in Rome in 1816; nor was Zedda aware of several Rossini manuscripts containing cadenzas and variations. Faced with serious textual cruxes, he was thrown back upon his own resources, those of an intelligent musician with limited knowledge of Rossini's other works.

But there was so much to see that quibbles over details faded away. The Ricordi edition was fundamentally different from Rossini's manuscript: melodic lines were changed, rhythms modified, harmonies altered, orchestration transformed. Extra brass and percussion had been added. Where Rossini called for a piccolo, the edition substituted a flute. Signs of articulation (slurs, staccati, accents) were unrecognizable. It is not that Rossini's manuscript was structurally different from the printed edition: however much the opera may have been performed with disfiguring cuts, the printed edition was essentially complete. The differences, rather, were in the tissues and the sinews of the opera. Perplexed, disbelieving, Zedda entered into his score and parts as many modifications as possible, performed his corrected version of the opera, and returned the rented materials to the publisher.

Since a publisher must rent the same set of orchestral parts to various conductors and opera houses, rental contracts specify that materials must be returned in good condition. Zedda's parts were so heavily marked up that no other conductor could have used them, and so Ricordi did what any self-respecting publisher would have done: they billed him for the cost of the materials he had rendered useless. Zedda protested: the Ricordi materials were not Rossini's Il barbiere di Siviglia, but a deformation of it. From an Australian conductor concerned with contradictory slurs and seeking publicity, such charges might be rejected; but here was an Italian conductor demonstrating the problems on page after page of the score in the privacy of Ricordi's Milan offices.

Unbeknownst to zedda and Ricordi, many scholars and musicians were fully aware of these problems. In 1864 a Florentine publisher, Giovanni Guidi, had issued a full score of the opera based strictly (even too strictly) on the autograph manuscript, which had been willed to the library of the Bologna Conservatory in 1862 by Rinaldo Bajetti, a Bolognese lawyer and Rossini's friend. Guidi's score was reissued several times, including by a New York publisher, Broude Brothers, which further corrected it on the basis of a good manuscript of the opera in the New York Public Library (believed, erroneously, to be in Rossini's hand). Vittorio Gui, as always a knowledgeable Rossinian, had developed his own more accurate score of the work. Side by side with these scores, however, there circulated the Ricordi edition--the version considered "traditional," the version adopted by most theaters, the version that Zedda had corrected. Whence did this score derive? How could these two versions of Il barbiere be reconciled? Since no one had a response, it was simpler for Ricordi to continue shipping its materials around the world.

Finally convinced that a problem did exist, and confident that Zedda could produce a score that was both faithful to Rossini's autograph and acceptable to performers, Ricordi entrusted him with preparing a critical edition of Rossini's opera. Published at the end of 1969, Ricordi's belated contribution to the Rossini centennial, it was the first critical edition of a nineteenth-century Italian opera ever issued. Whatever its faults, Zedda's Barbiere had the undisputed merit of having tackled with seriousness a difficult, even intractable task. In particular, Zedda was able to show that contemporary manuscripts and printed editions of the opera followed almost without exception the basic outlines of Rossini's original score. There were lacunae, and there were rare substitutions: Bartolo's hilarious aria "Ad un dottor della mia sorte" was sometimes replaced by the simpler (and musically inferior) "Manca un folgio," written in 1816 by Pietro Romani. Also there were small changes in the orchestration: pizzicato cellos accompanied the Count's "Ecco ridente in cielo" in a few manuscripts, reflecting the practice of theaters with no access to a guitar; and Rossini's articulation was copied incompletely and inaccurately, while his rhythms were invariably simplified--eloquent testimony that contemporary copyists made fewer strokes of the pen wherever possible. Still, contemporary manuscripts and printed editions otherwise reflect the text of the opera as the composer notated it in his autograph manuscript.

About the so-called "traditional" version of the opera Zedda was less incisive. He stated that its readings, "even if they may have been produced and taken hold while Rossini was alive, find no confirmation in a written source." In fact, there is no evidence that anything resembling the Ricordi material was in use during Rossini's lifetime. For the "traditional" Barbiere was a deformed version that had been prepared long after Rossini's death, for reasons that may have seemed pressing at the time but have no validity today: filling out Rossini's chamber-like orchestration with heavier sounds; avoiding Rossini's characteristic use of the piccolo; facilitating the process whereby Rosina became a twittering soprano rather than a Rossinian contralto/mezzo-soprano.

Later Zedda suggested that this version dates from approximately 1885, when Ricordi, instead of seeking out Rossini's original, printed a full score that reproduced one being used for performances at the Teatro alla Scala of Milan. Rossini had nothing to do with Milan after 1840, nor did he pay any attention to La Scala and its practices. Thus the pre-1968 Ricordi materials, rather than reflecting a long-standing performance tradition, merely reflect a lamentable editorial decision at a particular historical moment to print an easily available score of Il barbiere di Siviglia rather than search out Rossini's manuscript. Even Zedda's critical edition has not completely erased the devastating effect of that decision on the performance history of Rossini's opera.

Critical editions of musical works are fundamentally different from those of literary works. While the critical edition of a poem or a novel can be read with pleasure, its details dissected by the devoted scholar, its obscurities and its curiosities enjoyed by the informed amateur, a critical edition of a musical work is not intended for the library or the study alone. It is intended to be used as the basis for performances. When the work is as much a part of the popular imagination as Il barbiere di Siviglia, the first performance based on a new edition can become a cultural event. In 1968, the Teatro alla Scala was the site, an extraordinary cast was assembled (featuring Teresa Berganza as Rosina), Jean-Pierre Ponnelle was the stage director, and the man entrusted with bringing the new Barbiere to light was the rising star of Italian conductors, Claudio Abbado.

Abbado at that moment was a figure of controversy. Having become interested in Bellini's setting of the Romeo and Juliet story, I Capuleti e i Montecchi, he prepared a new performing edition of the opera. He altered the arrangement of the voices by substituting a tenor as Romeo for the mezzo-soprano required by Bellini, who had followed a tradition to which many Rossinian opere serie (Tancredi, La donna del lago, Semiramide) belong. Abbado's aim was comprehensible: he wanted to revive Bellini's opera in a way that the public might more easily accept, since few works being performed during the 1960s featured female heroes en travesti. (I will never forget the conversation of two elderly women during a performance of Rossini's Bianca e Falliero at Greater Miami Opera in 1988, before supertitles had transformed audience understanding. "Do you see what I see?" one whispered to the other. "Two women ... making love....")

But Bellini's score was not so easily manipulated. He had planned the music with certain effects of sonority, and a tenor Romeo was decidedly not what he had in mind. Perhaps the most impressive moment in the opera occurs at the end of the first act's finale, when, from opposite sides of the stage, the forcibly separated Giulietta and Romeo sing in unison an extraordinary Bellinian melody, "Se ogni speme e a noi rapita," over a sotto voce staccato accompaniment from male soloists and the all-male chorus. The significance and the beauty of the passage lies in those two female voices, lost in each other, soaring over the male ensemble, intoning a melodic idea that goes on and on in ever inventive and rhythmically subtle detail, thirty-one measures of continuous melody. ("Lungo, lungo, lungo," as Verdi described Bellini's lengthy melodies.) Substitute a tenor for the second woman's voice and the magic is lost.

The implications of the revised vocal scoring were broader still. In an opera as dependent upon ensembles as I Capuleti e i Montecchi, Bellini's delicate web of vocal parts unravels when Romeo is recast as a tenor. In their duet, Romeo and Tebaldo (the latter played by the young Luciano Pavarotti in Abbado's production) frequently sing in parallel sixths, with Romeo above Tebaldo. One cannot transpose Romeo down an octave and hope for acceptable results: the music is not conceived for two male voices in thirds.

Nor did abbado's interventions stop with the vocal parts. Despite Bellini's considerable talents, he was far less expert in handling orchestral sonority than Rossini or Donizetti. The autograph manuscripts are replete with alterations that suggest insecurity, not an idealized search for perfection. The resulting sound is often heavy, for Bellini began with a larger orchestra than Rossini's and kept most instruments playing too much of the time. Similar problems are not unknown to the German symphonic tradition. Should orchestras today play Schumann's symphonies or a revision of them by another composer or conductor? Schumann was an active participant in the first performances of these works, working directly with his conductor (Felix Mendelssohn) and making alterations where he felt artistic results obtained during rehearsals and performances were unsatisfactory.

Bellini had similar responsibilities: he was contractually required to rehearse his new operas and to participate in their first three performances. The sound was verified directly in the theater by the composer. If we think that Bellini's operas are worth performing, then they are arguably worth performing as they were conceived, with problems of balance resolved through careful control of orchestral size and seating, the use of dynamic gradations, and so on. (Performances with "early" instruments may be fascinating, but they do not make these problems go away.) But Abbado chose instead to "revise" Bellini's orchestration, and his interventions were present on every page of the score.

Abbado's score of Bellini's work might have circulated in this form had not the dean of Italian music critics, Fedele d'Amico, blown the whistle. In a sharply worded article he deplored the operation, lamenting that Bellini's beautiful opera lay fallow while this pointless "revision" was allowed to circulate. Abbado apparently was stung by the criticism. His edition of I Capuleti e i Montecchi was withdrawn from circulation shortly thereafter; and when the first performance of a nineteenth-century Italian opera based on a critical edition was planned for La Scala, it was Abbado who took command.

His barbiere was a revolutionary reading of the opera. Not only did Abbado employ the critical edition, but he adhered to the text with almost fanatical strictness. No significant cuts were sanctioned, and few ornaments were permitted. This was a performance with a message: Rossini was to be presented at a level of seriousness and precision usually reserved for the German masters. The slapstick antics that the work had endured for decades were replaced by a clean, precise, and eminently funny staging by Jean-Pierre Ponnelle, in which physical action emerged from the music. The singing was elegant, while the precision of the orchestral execution brought out every detail of the Rossinian palette.

Yet not everyone approved. Isolated voices, paying no heed to the evidence of the sources and insisting on a "tradition" invented at the end of the nineteenth century, preferred the old version. (As late as 1990 Giuseppe Patane made it a point of honor that his recording of the opera did not use the critical edition.) And there was the more complicated objection of those who wondered whether this was how a Rossini opera should be performed. Did purging the opera of the vocal fireworks that coloratura sopranos had appended to melodic lines never written for them in the first place also mean that a mezzo-soprano Rosina or a contralto Rosina was compelled to sing only the notes printed in the score? Did eliminating traditional licenses (speeding up, slowing down, introducing pauses for stage business) also mean that the music had to be performed with quasi-metronomic regularity?

Abbado's performance had been technically perfect, but still there were complaints that it lacked the wit and the vivacity that characterized Rossini's art. And in a transference that has become standard, uncertainties about the performance grew into doubts about the edition. Is that what it meant to use a "critical edition"? Did the new edition encourage, or even require, this kind of performance? Was spirit the price of scholarship?

One year later a different Rossinian production graced the same theater. The work was L'assedio di Corinto, and it was given under the direction of Thomas Schippers. It also marked the debut at La Scala of two of the greatest American singers of our time, Beverly Sills and Marilyn Horne. With several changes in the score and cast (and, sadly, without the presence of Horne), this production was imported to New York five years later for the arrival of Sills in the promised land of the Metropolitan Opera, after her two decades wandering in the desert of the New York City Opera. Schippers and his colleagues subjected Rossini's score to alterations far worse than those imposed by Abbado on I Capuleti e i Montecchi. Although they did not reorchestrate the music, they cut and rearranged so much of it that large parts of the opera were unrecognizable. From a serious work of music drama they concocted a showpiece for two prima donnas.

L'assedio di Corinto, or The Siege of Corinth, has a complicated history. Originally written as Maometto II for the Teatro San Carlo of Naples in 1820, Rossini revised it for Venice in 1822, and in 1826 he used it as the source for Le Siege de Corinthe, his first opera in French. This version was then translated back into Italian as L'assedio di Corinto. But the versions performed by Schippers at La Scala and the Met were not merely this re-translation into Italian of Rossini's French opera, but a confused blend of the various versions, to which was added an extraneous piece (from a revision of the opera by another composer) to favor even more the part of the soprano.

The original Maometto II and Le Siege de Corinthe are both coherent works of art, but they are very different. The unusual dramaturgical and musical design of Maometto II, Rossini's most innovative Italian serious opera, must have seemed bewildering to his contemporaries, even among the relatively sophisticated public of Naples. What Rossini calls a terzettone in his autograph manuscript--a big, fat trio--is a continuous musical composition that occupies almost a third of the first act. Anna's heroic scene at the end of the second act opens with some of the most difficult and expressive florid music that Rossini ever wrote, but instead of concluding with an elaborate rondo, the opera's final moments witness the arrival of the Turkish forces, Anna's abrupt suicide, and the shocked reaction of Maometto and the rest of the cast.

In the best Neoclassical tradition, Maometto II is a tragedy of love and honor, focused on four principal characters: Paolo Erisso, a tenor, the leader of the Venetian colony at Negroponte; Anna, a soprano, his daughter; Calbo, a contralto en travesti, a Venetian warrior in love with Anna; and Maometto II, a bass, the leader of the besieging Turkish forces. Anna and Maometto had fallen in love at an earlier time and a different place, with Maometto in disguise; but now they must play out their personal story in hopeless circumstances. She betrays her beloved in order to save her father and her people, weds Calbo (whom she respects but does not love), and ultimately kills herself before her mother's tomb.

In Le Siege de Corinthe, Rossini transformed his Neapolitan masterpiece into a French grand opera, heir to Spontini, precursor to Meyerbeer. The protagonists become Greeks and Turks instead of Venetians and Turks, to reflect the political events of the 1820s; but this was the least significant alteration. The vocal lines of the Italian original were greatly simplified, following the more declamatory style in use at the Opera in Paris. In further homage to French traditions, Rossini expanded the spectacular elements of the score. Choruses, dances, and pantomimes often overwhelm those elements of the tragedy that remain from Maometto II. A scene of prophecy for a new character, Hieros, and a group of Greek soldiers, invoking Marathon, exalting martyrdom and promising a glorious future for Greece, foretell the conclusion of Le Siege, a mass suicide rather than the individual suicide of Anna. Almost every detail of this scene was imitated by Verdi in the well-known passage that concludes the third act of Nabucco.

These alterations in dramaturgy were accompanied by changes in the solo roles. Heroic parts en travesti were not acceptable in French opera; hence the contralto role of Calbo was transformed into the tenor Neocles. Since Calbo's original aria ("Non temer: d'un basso affetto," with its cabaletta "E d'un trono alla speranza") is a quintessential solo for coloratura contralto, and absolutely inappropriate for a tenor, Rossini replaced it with a new aria for Neocles. But rather than lose the Calbo aria altogether, the composer modified its cabaletta for the soprano (now called Pamyra) and allowed it to conclude her major aria (derived from the incomplete aria that Anna had sung at the end of Maometto II), which in Le Siege opens the second of the opera's three acts. Then, in order to provide solo music for Pamyra near the end, before the mass suicide, Rossini inserted the prayer for Anna that had been included within the terzettone from Maometto II.

While most scholars and performers are convinced that Maometto II is musically and dramatically more powerful than Le Siege de Corinthe, there are legitimate reasons to favor one opera or the other. But Schippers tried to merge the two. His fundamental error was to imagine that Le Siege could be performed with a contralto (Horne) as Neocles and a high soprano (Sills) as Pamyra. Each decision had unhappy ramifications. By reintroducing the hero en travesti into a French opera written for tenor, Schippers found himself forced to turn to Maometto II to find appropriate music for Horne's aria. The resulting piece was monstrously large, and it was constructed on the assumption that more is better (the kitchen sink principle, we might call it), drawing freely on music written for both Calbo and Neocles.

The presence of a coloratura contralto in Le Siege compromised the work's overall shift to a simpler vocal style. With Neocles now reassigned Calbo's original cabaletta, the Pamyra aria at the beginning of the second act suddenly found itself without a concluding movement. What should one do? Choose a cabaletta, of course, any cabaletta, and stick it in. Sills sang one from Ciro in Babilonia, an opera written by Rossini early in 1812, with a ludicrous result. From the elaborate orchestral web of Rossini at his most mature, the sound suddenly dissolves into the Cimarosan ideal of his youth. And as if that were not sufficient, Sills vocalized along with the rambunctious orchestral theme between the two statements of the cabaletta theme, strewing high notes hither and yon. Bubbles needed to have something to sing, after all, since her effective tessitura was so much higher than what Rossini wrote for either Anna or Pamyra.

Despite this travesty of the music of Rossini, the ladies sang their hearts out, and L'assedio di Corinto was a triumph for the prima donnas. In the case of Marilyn Horne, it was also a revelation of what Rossini's vocal lines could be when sung by an artist with the requisite technical skills. Still, just as voices were raised against Abbado's Barbiere, many protested this operation. Few knew the opera well, of course, and almost no one had even a passing acquaintance with Maometto II; but a vocal score of L'assedio was available from Ricordi, and some had even heard the reprise of the opera early in the 1950s, with a young Renata Tebaldi as Pamyra. I threw in my two bits at a well-attended public lecture at the New York Public Library at Lincoln Center the night before the opening, to which many music critics flocked. For my pains I earned this barb from Sills: "Some so-called musicologists remind me of middle-aged men who talk about sex."

The contrast between Abbado's Barbiere and Schippers's Assedio embodies two extreme approaches to the performance of Italian opera. For the one, the text of an edition, especially a "critical edition," is sacrosanct and must be respected in every detail; for the other, an opera is an entertainment and can be freely manipulated as long as the result is a good show. Since the music of Rossini was largely unknown and unknowable during the late 1960s, the second approach seemed feasible. But the uproar that greeted the La Scala revival of L'assedio di Corinto had one important result: it helped give impetus to the formation of the Edizione critica delle opere di Gioachino Rossini.

It was a stroke of good fortune that, after the Vaughan skirmishes, the real battle for critical editions of nineteenth-century Italian opera and a revitalized performance practice to go with them was engaged first on Rossinian soil. After all, neither audiences nor musicians had much knowledge of Tancredi, La donna del lago, Il Turco in Italia, or Il viaggio a Reims. The appearance of new editions became a cause for rejoicing: singers and conductors accepted them readily, and audiences were delighted to hear what amounted to new works. An occasional crotchety critic (particularly one who disliked the bel canto repertory anyway) might snort something about "scholarship," but know-nothingism has never seemed a worthy platform.

When the terrain shifted back again to the operas of Giuseppe Verdi, however, howls of anguish and even anger arose from some quarters, and studied indifference from others. The more familiar the work, the stronger the reaction. A belief in the rectitude, the sanctity, of what is perceived to be a modern performance tradition spills over into a belief that the printed text lays equal claim to validity. This confusion is at the heart of many contemporary controversies about editing and performing the music of Verdi. And the name around which these controversies have swarmed for almost three decades is that of Riccardo Muti.

With a rigorous approach to the scores that overshadows even the legendary Toscanini, Muti's Verdi is inextricably allied with the search for a re-conceptualized performance practice of Italian opera. More than any other Italian conductor, he has been closely tied to a strict reading of the printed score. Yet this "cause of fidelity," as Muti has defined it, "must not be understood as the cold reproduction of a text, but as an intuitive interpretation through the written signs of a whole spiritual world that exists beyond the written signs." This spiritual position is achieved by insisting on integral performances and by refusing to allow singers "traditional" liberties and interpolations.

When the works involved were tangential to the repertory, Muti's approach won a chorus of approval. His Guglielmo Tell at the Maggio Musicale in 1972, although sung in Italian and based on a problematic edition, was a revelation. Yes, the work lasted more than five hours, and one had to be in optimal physical and mental condition simply to attend. Yes, it was performed in an execrable nineteenth-century translation, when it should have been done in the original French. Yes, the tenor Nicolai Gedda, as appropriate a choice for the role of Arnold as was possible at that time, bowed out before the end of the run. All this is true, but perfection is something that we seek, not something that we attain. Muti's rendering of Rossini's William Tell for the first time clarified for a modern listener the fascination that Rossini's last opera had for nineteenth-century musicians.

With the works of Verdi, matters were very different. Whatever else anyone may or may not have heard in Florence when Muti directed Il trovatore in December 1977 at the Teatro Comunale, all that anyone talked about was the end of "Di quella pira," the aria that concludes the third act. The reason was that Muti's Manrico did not sing the high C traditionally interpolated to bring down the curtain amid a storm of applause. Indeed, the production has gone down in public consciousness and critical lore as Muti's defiant challenge to the contemporary performing tradition, the "Trovatore without a high C." (A similar reaction emerged when he conducted the same opera to open the 2000-2001 season at La Scala.)

It was, in fact, much more. The American Verdi scholar David Lawton had taken the initial steps toward preparing the critical edition of the opera (which was ultimately published in 1993), carefully controlling the printed text of the Ricordi edition then in commerce with Verdi's manuscript. A control of this kind is a crucial part of the process of preparing a critical edition, identifying errors and misreadings and clarifying the meaning of a composer's notation. But the efforts of Lawton, Muti, the singers, the orchestra, and the production team were overshadowed: everything dissolved into the hullabaloo over the high C.

What makes the controversy particularly absurd is that the high C in question never existed in any printed edition of the opera. The great moment is in fact an interpolation. In this respect, Lawton's edition of the opera was unchanged with respect to the older printed edition. Muti's insistence that the interpolated note not be sung, however, marked not only his performance, but also Lawton's edition, which became "the edition of Il trovatore without a high C." But why should anyone care so much? Does it really make such a difference if the tenor ascends to that note rather than remaining on the G, as in Verdi's written text?

Looking exclusively at the end of the aria, the interpolated note is nothing but harmless pyrotechnics. As assertive a cabaletta as Verdi ever wrote, the piece closes the third act, brings down the curtain, and precipitously moves the opera to the final catastrophe. Verdi's conclusion demonstrates his wish to preserve an unusual level of tension. Manrico would ordinarily have ended the aria by descending from the G of "All'armi" to the lower C. Instead Verdi holds the voice on the G so that Manrico concludes on the fifth of the C major chord, while the first tenors take the E below it and the second tenors and the basses sing middle C. The result is a full tonic triad, with Manrico alone on the highest note.

Why did Verdi not intensify this effect further, by giving C, E, and G to the male chorus, with Manrico free to ascend to high C? Musical analysis, which can be invoked to support all sides of an argument, is painfully unsuited to this kind of question. But here is a relatively simple explanation--a historical explanation, not a musical one. Verdi wrote the role of Manrico for Carlo Baucarde, a tenor whose range it presumably reflects. The part has a high tessitura, sitting for long stretches in the sixth between middle C and high A, but A (which recurs frequently) is the highest note that Verdi expected Manrico easily to sing. Only at a single point in the opera does the composer notate a high B-flat for Manrico, in the context of the stretta that concludes the first-act trio; and to avoid interrupting the octaves being sung by Manrico and the prima donna, Verdi gave the tenor a choice: either B-flat (the only time in the opera) or the lower D-flat, should the singer be unable to handle the higher pitch. Baucarde, in short, had no high C, no high B, and only a most uncertain high B-flat.

It is quite irrelevant whether or not Verdi actually told the singer Enrico Tamberlick, who had boasted of the success of his high C with the public, "Far be it from me to deny the public what it wants. Put in the high C if you like, provided it is a good one." The real disaster of the interpolated high C is its effect on the choice of an appropriate tenor to sing the role of Manrico. For the sine qua non for an opera house today, as it casts the part, has become the ability of a tenor to let loose a stentorian high C at the end of "Di quella pira." The interpolated note has come to dominate the conception of the part, and everything else is planned around an effect that Verdi never intended. To produce the high C, moreover, singers generally cut the cabaletta by half and omit the notes that they should be singing with the chorus, so as to preserve breath and energy for the holy final pitch.

At the time of Muti's Trovatore in Florence, one Italian critic commented that the high C, even if not written by Verdi, was a gift that the people had given to Verdi. This bit of sentimentality hides the basic issue: whether Manrico does or does not bellow his high C is of little artistic importance. What is artistically devastating is that the perceived need to hit the stratospheric note has transformed our conception of the role. Give me a tenor who can sing Manrico as Verdi conceived the part and chooses to add a ringing high C, and I will join the loggione in applauding him. Failing that, let Manrico, in Rossini's famous words to the same Tamberlick, leave his high C on the hat rack, to be picked up on his way out of the theater.

The opening of the operatic season at La Scala on December 8, the day honoring the city's patron saint, Sant'Ambrogio, is Italy's most keenly awaited musical event each year. "Scala supplements" are published in the major newspapers; what the Italians call the cronaca pages report gleefully on those members of the government and society likely to be in attendance, and what they will be wearing; protest movements (involving party politics or animal rights) know that their activities will be publicized. Through all the social trappings, some attention is even paid to the music. The performances of Verdi's Ernani at La Scala for Sant'Ambrogio in 1982 were particularly in the public eye: it was the first time that this prestigious event was to be entrusted to Riccardo Muti.

For weeks before the premiere, the word filologia ("philology") was tossed around by critics and the public. For this Muti was more than a little responsible. Just before the performance he told Corriere della Sera: "This will be the Ernani of Giuseppe Verdi, that is, an edition extremely faithful to the manuscript, without cuts; the opera will be performed exactly as it was conceived." This was a peculiar statement. First of all, Muti used the manuscript of Ernani largely to investigate only a single issue. The score that he employed was the standard edition, prepared by Ricordi at the end of the nineteenth century, with more than its share of mistakes and misreadings. Muti could have justifiably talked about a performance "extremely faithful to the Ricordi edition," but hardly about fidelity to Verdi's autograph.

But there was another, graver problem. What does it mean to perform an opera "exactly as it was conceived"? How do we know how a work was conceived? We cannot ask Verdi how he conceived Ernani. Perhaps we must settle for something far more banal: choose an edition derived as closely as possible from authentic sources, and then use that edition as the basis for a performance, bringing to bear what the composer wrote down, and other historical information, and our knowledge about performances that Verdi himself directed, and our awareness of modern performing traditions, and--most of all--our own musical and dramaturgical intuitions. In any event, the claim to an "Ernani at La Scala the way Verdi wanted it" was imprudent under the best of circumstances.

Just as a polemic about a single note characterized Muti's Trovatore in Florence, a single piece served as the lightning rod for his Ernani in Milan: the cabaletta for Silva, "Infin che un brando vindice," just before the first-act finale. The piece was not in the original version of the work, which was performed at the Teatro La Fenice in Venice on March 9, 1844. Although this movement is included in most editions of the opera, Muti decided to omit it. He set forth his reasons in the same Corriere della Sera interview:

Just recently I happened upon a vocal score from the second half of the nineteenth century, in which the cabaletta for the bass, Infin che un brando vindice, is missing. My curiosity aroused, to keep faith with my rigor and my philological scruples I decided to examine Verdi's manuscript: from Casa Ricordi I obtained a copy of the original and I had confirmation that the cabaletta is not there. Driven, as always, by a need to return to the sources and to arrive at the truth of the written sign, I sought documentation by calling on the musicologist Francesco Degrada and consulting the book of Julian Budden. These explain clearly that almost certainly the cabaletta is not by Verdi and was inserted for the first time by the bass Ignazio Marini, who sang it in the autumn of 1844 in Milan.

At the time the critic of a Milanese paper reacted negatively, going so far as to protest strongly this abusively inserted cabaletta and to hold Marini (then a well-known bass) responsible for having introduced it into the opera. Having verified this, I also found the answer to a problem that had been disturbing me: why was it that the [arias for the] soprano, tenor, and bass had both cantabiles and cabalettas, while the baritone alone had no cabaletta, leaving the relationship among the four unbalanced? There was the answer in Verdi's original manuscript: the soprano and tenor had both a cantabile and cabaletta, but neither the bass nor the baritone had a cabaletta. I thus decided to follow the precise indications of the composer and to omit a passage almost certainly not by Verdi or, at least, of which there is no trace in the original manuscript.

Notice the uneasy slippage here between analytical explanation and philology. For the former, do we really know how many characters in a "typical" Italian opera of 1845 have multi-partite arias and how many have arias in a single tempo? For the latter, Degrada and Budden are much more cautious than Muti: they describe factors that might suggest the piece is by Verdi and others that suggest it is not, leaving the matter open.

But a performance does not have the luxury of being indecisive. Silva cannot proceed to the footlights and explain the situation to the audience: "Ladies and gentlemen, we are not sure whether the cabaletta that follows in some sources is really by Verdi, so I'm going to (a) sing it or (b) leave it out." And the glamour of opening night at La Scala blends poorly with uncertainty. Muti took a legitimate doubt and made an entirely plausible artistic decision: after all, composers of Italian opera in the nineteenth century frequently were called upon to make just such decisions for particular occasions on practical grounds, to meet the needs of theaters and singers. And then the conductor buttressed his decision with claims that it was philologically motivated and analytically just. Rumors circulated widely in Milan that the real motivation for omitting the cabaletta was the inability of an aging Nicolai Ghiaurov to sing it well.

By the time of the old Spanish grandee's entrance toward the end of the first act, the opening-night audience at La Scala was already hostile to the staging by Luca Ronconi and the sets of Ezio Frigerio, whose price was the subject of vociferous debate and whose awkward multiple levels subjected the singers to unaccustomed gymnastics. The singers in turn reacted with a tentative performance that did not sit well with melomanes in the audience. Finally Silva appeared and sang his cantabile, "Infelice, e tu credevi." As he attacked the finale, omitting "Infin che un brando vindice," a cry was heard from the recesses of the theater: "La cabaletta, filologo!" The last word was spat out with palpable scorn. Muti dug in his heels and carried on.

The incident was worthy of the theater of the absurd. Ostensibly, Muti acted with philological responsibility by omitting the cabaletta, while the gallery patrons demonstrated their attachment to the "traditional" way of singing Ernani. But in truth all of them got it wrong. What the loggionisti did not understand was that in the performing tradition of Ernani, the Silva cabaletta was usually omitted. Although it is present in Ricordi materials (which, as so often, take their cue from a Milanese production), in most extant copies the music is firmly crossed out. And Muti erred in not understanding that the music he actually performed was a philologist's nightmare, an erroneous conflation of separate versions.

Whoever introduced the cabaletta into Verdi's score also felt it necessary to modify the preceding two bars of music. When Ricordi first printed string parts of the opera soon after its premiere, they followed Verdi's autograph, without the Silva cabaletta. When the string parts were reprinted to include the cabaletta, Ricordi modified the two previous bars. Parts for winds, brass, and percussion for Ernani were first printed in the 1880s, and they contain only the modified bars introducing the cabaletta. Each version has its own musical and harmonic integrity. What Verdi never intended, however, was the version played at La Scala.

Muti performed the bars that were rewritten in order to introduce the cabaletta, and then he cut the cabaletta. Philology indeed! But he was hardly alone in having made this mistake. Ever since Ricordi re-did the parts, there was no way to perform Verdi's own music, and everyone who omitted the Silva cabaletta fell into the same trap. Only when the critical edition of Ernani became available was it possible to perform the music either as Verdi had originally written it or with the added cabaletta. Performances, by their very nature, cannot pretend to be philological: that is the purpose of editions.

Subsequent research has clarified much of the history of "Infin che un brando vindice." The cabaletta is indeed by Verdi--but he did not write it for Ernani. It was composed at the request of Marini for insertion into a revival in Barcelona in 1842 of Verdi's first opera, Oberto, Conte di San Bonifacio. Roger Parker first identified the Spanish libretto, which includes the relevant text; and Claudio Gallico, in the preface to the critical edition of Ernani, transcribed the letter from Verdi to Marini that accompanied the manuscript on its voyage to Barcelona. It should come as no surprise, then, that it was the same Marini who introduced "Infin che un brando vindice" into Ernani. What we still do not know is whether Verdi gave his blessing to the operation or merely tolerated it.

By the time of the Ernani follies at La Scala, there was good reason to hope that similar problems might be easier to avoid in the future. The University of Chicago Press and Casa Ricordi had announced their intention to publish The Works of Giuseppe Verdi in new critical editions, and the inaugural volume in the series, Rigoletto, was about to appear. The conductor who first used the new Rigoletto in the theater was the conductor who had demonstrated how deeply he cared about the issues that it sought to address: Riccardo Muti.

The words "critical edition" strike a mixture of scorn and terror into the hearts of many conductors, singers, and music administrators. Their reaction is comprehensible: musicians have enough to do preparing a performance of a work without being told that the printed scores that they are using are inaccurate or incomplete. The less they need to think about such problems, the happier they are. Moreover, most of them can recite an endless number of anecdotes about new editions that are worse than the old, and musicologists who cannot interpret properly the notation of transposing instruments, and pages strewn with extraneous symbols, and so on. I have seen critical editions embraced by conductors with a rigor contrary to their true meaning, and ignored by nervous singers unwilling to change a note or a word of an interpretation that they believe to be authorized by some mystical tradition, and denounced by pecuniary administrators as the brainchild of rapacious music publishers.

Critical editions of nineteenth-century Italian operas make available the best texts that modern scholarship, musicianship, and editorial technique can produce. Fully cognizant of modern textual theory, they do not return blindly to a single "original" source, although the composer's autograph manuscript is often our best single guide. Instead they reconstruct the circumstances under which a work was written, the interaction of composer and librettist, the effect of imposed censorship, the elements that entered into the performance, the steps that led to publication, and the role the composer played in the subsequent history of the work. They interpret the notation, often incomplete or contradictory, so that musicians have available not only a text prepared with an ear toward eventual performance, but also one that permits them to distinguish signs that stem directly from the composer from signs that derive from secondary sources or are provided by the editor. For works that exist in multiple versions, they choose a basic performing text (not always a simple matter), with appendices that make possible the practical realization of any version prepared under the direction of the composer.

Musical editions exist on the page, not in the theater; but all theatrical presentations begin with a printed score and parts. Critical editions of this repertory engage with the history of each work, and struggle with contradictions and uncertainties, and seek feedback from performers, and proofread over and over in order to eliminate inadvertent error (not even the best edition can hope to be free of error). But this scholarship does not mean to be imperious. These critical editions continue to recognize the composer as the central figure in the Italian operatic landscape during the nineteenth century, and they seek where possible to reproduce his voice fully and accurately. What really matters, of course, is what happens after the curtain goes up.

This article originally ran in the July 2nd, 2001, issue of the magazine.