74. Madama Butterfly

Puccini saw David Belasco’s one-act play Madame Butterfly on 21 June 1900 at the Duke of York’s Theatre in London, where the play had been performed from 28 April of that year.Note: *See Quaderni 1982, 98. The première of Belasco’s play was on 5 March 1900 in New York. The dates are confirmed in Arthur Hobson Quinn, Representative American Plays, 6th edition (New York and London: D. Appleton-Century Company, 1938), 623-624, and the text of the play is published on pp. 627-636.  The next day, Puccini commented on it in a friendly but reserved manner: “se quel soggetto che vidi: madame butterfly (bettflai, farfalla) non fosse giapponese mi piacerebbe tanto è una cosa bellissima ma non per l’Italia [if that subject I saw: Madame Butterfly, were not Japanese, I would like it so much; it is very beautiful, but not for Italy].”Note: *Quoted in Quaderni 1982, p. 98, without indication of the source and the addressee. Some weeks later, however, he had Ricordi inquire in New York about “quel tal soggetto americano [that American subject]”, which certainly meant Madame Butterfly.Note: *See Carteggi 1958 No. 236 = Epistolario 1982 No. 68; see also MRC 1900/01 - 5/194 of 21 September 1900, quoted in Girardi 1995, p. 202, footnote 8.David Belasco (1853-1931) was a very influential playwright and producer of American theatre around the turn of the 20th century. He wrote a large number of successful plays in various genres for Broadway, and he made his mark by re-defining the role of the producer by meticulously staging his own plays in innovative ways. His work in the theatre as a director was exceptionally demanding and precise, compared to the rather lax standards of his time. His co-author for Madame Butterfly was John Luther Long (1861–1927) upon whose story of the same name (first published in 1898) the play was based. Long’s story was influenced by themes from the novel Madame Chrysanthème (1887) by Pierre Loti (1850–1923), from which André Messager had already made an opera (Madame Chrysanthème, first performed in Paris in 1893). However, Long also used his direct personal knowledge of real events in the semi-colonial Japan of the late 19th century.Note: *See Groos 1991 and van Rij 2001. Moreover, there is a typewritten draft of unknown origin for Belasco’s play in US-NYp (see Kaye 1996, V-VII and 3), with substantial differences from the definitive version of the play. The parallels with the opera found in this draft make it seem very possible that it was made available to Puccini and his librettists — if not specifically prepared for them and for this purpose.While Puccini continued his search for a suitable subject for an opera, examining many different possibilities, he was constantly thinking about Madame Butterfly, the operatic rights to which had not even been acquired by the end of 1900.Note: *See Epistolario 1982 No. 69 of 20 November 1900. But from the beginning of March 1901, Puccini and Illica were already corresponding about a first draft of their treatment of the play as an opera.Note: *See Carteggi 1958 Nos. 243-245, 247, 251, and 253-255. Soon, Giuseppe Giacosa was also involved in the project.Note: *See Nardi 1949, 837: Ricordi to Giacosa, 25 April 1901. Even before a copy of the Belasco play arrived,Note: *See Groos 1996, p. 57. Illica drafted a scenario consisting of a prologue (“prologo”) with the marriage ceremony and the first night of love, followed by the major part of the action, consisting of: “1) La casetta di Butterfly — 2) La villa del Console [later also: Il Consolato S.U.A. a Nagasaki] — 3) La casetta di Butterfly”.Note: *“1) Butterfly’s little house — 2) The consul’s house [later also: The American Consulate in Nagasaki] — 3) Butter-
fly’s little house”; see Carteggi 1958 No. 249. In May, the draft of Illica’s “prologo” was ready,Note: *See Puccini’s letter of 20 May 1901 in Nardi 1949, pp. 837-838. and by that time the contract with the American copyright owners was finally completed.Note: *See Marchetti 1973, No. 247. The details pertaining to this early phase of the libretto were described by Arthur
Groos in Biagi Ravenni and Gianturco 1997, pp. 147-158, in Companion 1994, 174-186, and in Groos 2000, 92-104, in which (107-204) there is also a reconstruction of the complete first version of the libretto. Various libretto drafts are preserved in I-Mr (PP 649-1-1-d) and in the Archivio Giacosa in Colleretto Giacosa (Province of Turin). In I-Baf, there is also an early draft of the beginning and the nearly complete Act 1 (see 74.A.1). The official Inventario degli oggetti e dei cimeli pucciniani conservati nella Villa Puccini di Torre del Lago of May 1952 mentions in No. 71 “un libretto dell’opera ‘Madama Butterfly’ (atto ?)” with “appunti autografi.” However, it no longer appears in the Inventario Archivio Puccini in Torre del Lago of 1980. Giacosa’s work on Act 1 (the former “prologo”) dragged on throughout the summer and autumn of 1901; Puccini probably only received it by the end of November.Note: *See Puccini’s letter of 14 November 1901 in Nardi 1949, p. 840. In the meantime, Illica was writing the three scenes (“quadri”) of Act 2,Note: *See Carteggi 1958 No. 261 of 1 October 1901. for which Puccini already had rather exact ideas — including the music he would write for it. He already mentioned the mysterious humming chorus (“voci misteriose a bocca chiusa”) for the intermezzo before the last quadro.Note: *See Carteggi 1958 No. 263 of 5 December 1901. While the libretto was still under way, Puccini began composing (see 74.A.1, 74.A.2, and 74.A.I.5.a). In 8 December, he informed Illica that he had already accomplished a large amount of work on Act 1.Note: *See Puccini’s unpublished letter in I-PCc, Fondo Illica No. 170, probably dated 22 December 1901. He was also avidly looking for examples of authentic traditional Japanese music. He found some at that time, but termed it “povera e poca cosa [poor and insignificant].”Note: *See Carteggi 1958, No. 266, of 10 January 1902.
In the spring of 1902, Puccini tried for a long time to meet the Japanese actress Sadayacco who was touring in Italy with her kabuki theatre company. Those efforts were in vain, but Puccini certainly noticed the detailed reports of her appearances, which obviously influenced his ideas for the opera. Ultimately, he was able to obtain other samples of published versions of Japanese music.Note: *See Groos 1995, p. 75; see also Carteggi 1958 No. 277. He later called on the services of the Belgian musicologist, Gaston Knosp, who was involved in the study of the music of Japan,Note: *See Puccini’s letter of 5 June 1902 in Knosp 1937, p. 105. and the wife of the Japanese ambassador to Italy.Note: *See Epistolario 1982, No. 74, of 18 September 1902. Concerning Puccini’s specific use of Japanese music, see the particularly detailed findings in Powils-Okano 1986, which have been revised and corrected by more precise information in Groos 1995.  By the end of June 1902, the libretto was ready in a provisional form, and Puccini was satisified with it.Note: *See Puccini’s letter of 26 June 1902 in Nardi 1949, 844; see also Ricordi’s confirmation on the same day: MRC 1901/02 - 22/325. Progress with the composition of the music was slow during the typical and frequent revisions of the libretto, but Puccini finished the composition draft of Act 1 on 7 September 1902 (see 74.A.I.136.a).
While working on Act 2, Puccini became convinced that Illica’s scene set in the consulate was unnecessary.Note: *See Carteggi 1958, No. 287, and Epistolario 1982 No. 77, both of 16 November 1902. As with Manon Lescaut (64) and La Bohème (67), this is another example of a large passage being omitted early on in the composition. By the end of November 1902, Puccini and Illica established the new version of the libretto necessitated by the cut.Note: *See Carteggi 1958 Nos. 289 and 290. According to 74.B.1, by 28 November 1902 Puccini already had been working on the instrumentation of Act 1, which occupied him until 15 September 1903. Of course, for many months of that time Puccini was prevented from working at all because of his serious automobile accident in February of that year; but already in June 1903 Ricordi’s periodical Musica e musicisti (p. 475) reproduced a page from the autograph full score of Act 1 (fig. 40-3/+1), which meant that Puccini had already delivered the completed part of the manuscript to Ricordi (probably everything up to the middle of the love duet). It seems, however, that Puccini used an earlier orchestral version of parts of Act 1 when he finished his instrumentation (see 74.B.1).
Simultaneously, from the beginning of June 1903, when he was able to resume working, Puccini continued with the composition of the second act.Note: *See Carteggi 1958 No. 316. He finished the intermezzo by the end of August.Note: *See Epistolario 1982 No. 81 of 29 August and Puccini’s unpublished poem to Illica, I-PCc, Fondo Illica, of 27 August 1903. At the beginning of September, he worked on the trio for Pinkerton, Suzuki, and Sharpless, but by then he had already sketched parts of the final scene of the opera as well.Note: *See Carteggi 1958 No. 322 of 3 September 1903. Puccini wanted to orchestrate the very extensive second act in Paris,Note: *See Epistolario No. 81 of 29 August 1903. where he stayed in September and October for more than a month during the preparations for the first French performance of Tosca (69). He had already started writing the full score of Act 2 on 12 September while he was also finishing the orchestration of Act 1 (see 74.B.1). The different paginations in 74.B.1 indicate that in Paris Puccini orchestrated large parts of Act 2 out of sequence.
According to MRL, Ricordi began engraving the piano-vocal score (74.E.1) on 6 November, even though — as usual — Puccini had neither finished the composition nor the orchestration. A full month after his return from Paris, he was still occupied with the intermezzo between the two parts of Act 2.Note: *See Puccini’s unpublished letter to Illica of 28 November 1903, I-PCc, Fondo Illica; note also the date of 30 November 1903 in 74.B.1.  At the last moment, he even made additions in the flower duet at the end of Act 2, Part I, and composed some new music for the final scenes of the opera.Note: *See Ross and Berra 1979, 853 and 857, with letters of 5 and 7 December 1903. At the beginning of Butterfly’s death scene in 74.B.1, there is the autograph date of 11 December 1903. Late in the evening of 27 December, Puccini finally finished the full score.Note: *See 74.B.1, confirmed by Carteggi 1958 No. 335. In a note at the end of the manuscript, he told Ricordi to be sure that the libretto was printed with the changes in the stage directions resulting from his last modifications of the score (see 74.B.1).
By this time, the rehearsals for the first performance at La Scala probably had already begun, even though the integral piano-vocal score was not yet available. There were single leaves for the singers, which had to be exchanged from time to time to reflect Puccini’s revisions and to correct mistakes.Note: *See Carteggi 1958, No. 340, of 27 January 1904. Fortunately enough, the orchestra parts were prepared by the end of January, so the orchestra rehearsals could begin on 30 January and be finished on 6 February.Note: *See Puccini’s unpublished letter to Illica of 27 January 1904, I-PCc, Fondo Illica, as well as Carteggi 1958, No. 342.
Puccini decided very late to dedicate the opera to the Italian queen Elena.Note: *See Puccini’s unpublished letter to Illica of 4 February 1904, I-PCc, Fondo Illica, and Carteggi 1958, No. 345, of 8 February 1904. Since the earliest copies of the first edition of the piano-vocal score (74.E.1) dating from January 1904 did not contain the dedication, the score may only have been ready in the last days of that month. More importantly, it is very probable that the majority of the changes in 74.B.3, obviously made during rehearsals, were already realized in the first performance.
That first performance on 17 February 1904 became the most spectacular failure in the history of Puccini premières. Not all of the reasons for that debacle are quite clear. Even though the opera was criticized for being ponderous with the very long second act, political motives and the rivalry between the publishing houses of Ricordi and Sonzogno were also probably important factors.Note: *See Girardi 1997. In any case, Puccini and his librettists withdrew the opera the same night in a joint statement to the management of La Scala,Note: *See the facsimile in Marchetti 1973, p. 297. and Puccini also cancelled the production scheduled immediately thereafter in Rome, so he could take his time to make some “piccoli lavorucci [little reworkings] all’opera”.Note: *See Carteggi 1958, No. 352, of 21 February 1904. He also stated that he wanted to make some cuts in the score and divide the long second act into two separate acts.Note: *See Marchetti 1973, No. 293, of 22 February 1904.
He started working immediately.Note: *See Carteggi 1958, No. 354, of 23 February 1904. At first Act 1 was slightly shortened by removing approximately 130 bars, amounting to less than one tenth of the act. This was finished on 29 February (see Carteggi 1958 No. 361). Essentially, only one small change at the end of Part I of Act 2 was necessary for Puccini to manage the break before Part II (later also designated as Act 3). However, the revisions in Part II of Act 2 were more complicated.Note: *See Marchetti 1973, Nos. 364-367, of 7-10 March 1904. Here, among other things, a passage in the dialogue between Pinkerton and Sharpless was newly composed, and Pinkerton’s aria “Addio fiorito asil” was added to the score by 16 March (the date stamped in 74.B.1). On 24 March Puccini informed Illica that the revisions were finished.Note: *See Puccini’s unpublished letter I-PCc, Fondo Illica No. 274. In April, Ricordi implemented the necessary changes in the second edition of the piano-vocal score (74.E.2) for the production of the revised version of the opera in Brescia.
Puccini was not completely satisfied with those revisions. Even though the successful performance in Brescia would launch the opera’s worldwide success, during rehearsals Puccini informed Giulio Ricordi that there would be a cut in the intermezzo between the two parts of Act 2 (at fig. 6 – fig. 13; see Epistolario 1982, No. 87). So the performances in Brescia were obviously not strictly according to the published version of 74.E.2.
Soon after the Brescia performances Puccini made further changes, particularly involving cuts in the score. Each of the following performances probably took place in a different version, whenever Puccini was present at the rehearsals. The enormous extent of these cuts (nearly 500 bars less than 74.E.2) is documented in a copy of the piano-vocal score (in I-Lgp) used in rehearsals for performances at the Teatro Dal Verme in Milan (October 1905) and in Palermo (April 1906). Some of those changes were probably entered in that rehearsal score in Puccini’s handwriting.Note: *The changes Puccini made after the failed first performance at La Scala and after the success of the second performance at Brescia have mainly been presented inaccurately or are imprecisely described in the majority of the literature pertaining to this opera (for details see Schickling 1998b); Kaye 1996 is an exception, for he exactly reproduced a comparative text of the published (and some unpublished) versions, with commentary on the musical changes as well.  The two English editions of the piano-vocal score (74.E.3 and 74.E.4), probably published between May and July 1906, preserve the actual state of the revisions at that time. Most of them were derived from the Milan/Palermo rehearsal score, but they also introduced some new modifications.
From the summer of 1906, Puccini worked to prepare what he probably hoped would be the definitive version for the first French performance of Madame Butterfly at the Opéra-Comique. It contained most of the previously tested cuts, but not all of them. Before rehearsals began for this performance, Ricordi published the first French edition of the piano-vocal score (74.E.5), which was soon followed by a second French edition (74.E.6) with additional changes implemented during the actual rehearsals in Paris. Carignani went to Paris specially for the purpose of producing the new version of that piano-vocal score.Note: *See Puccini’s unpublished letter to Sybil Seligman of 6 December 1906 in MEL. Even so, probably neither of those editions present the complete extent of the cuts made in the Paris performance on 28 December 1906.
Parallel to the Opéra-Comique production, Ricordi’s intention to print a full score, which originated in the summer of 1906, was being realized. Previously, only copyists’ full scores were rented for performances. However, owing to the revisions made for Paris, the completion of this full score (74.E.7b) was delayed until June 1907.
In principle, the newly corrected French version was adopted for a new Italian edition of the piano-vocal score (74.E.7) representing, as it were, the current version of the opera as authorized by Puccini and published in the original language. Although, according to the earliest blind stamp, 74.E.7 was only published in April 1907, the performance on 11 February of that year at New York’s Metropolitan Opera may have been based on it, since Puccini certainly would not have accepted an outdated version for a performance he personally supervised.
After that, only 74.E.7e (the full score published in 1920 for sale to the public) contained slight revisions of the earlier editions (even though it was printed with what might be some publisher’s errors). However, it cannot be determined from the extant sources precisely when Puccini carried out the revisions for the engraving of 74.E.7e. There is no corresponding piano-vocal score.
However, the existence of Puccini’s corrections (“accomodi”) contained in 74.B.5 raises doubts as to whether the full score 74.E.7e did indeed represent Puccini’s final intentions.Note: *The modifications for the Teatro Carcano are imprecisely described in Smith 1980, 236-237. These corrections were intended for a performance at Milan’s Teatro Carcano apparently scheduled for the spring season 1920,Note: *According to the Annali del Teatro italiano, vol. 1: 1901-1920 (Milan: L’Eclettica, 1921). but actually given on 9 December 1920.Note: *The hint to the source for this information in the Annali del Teatro italiano is to be found in Fairtile 1996, 34; however, the correct date is documented in the daily newspaper La Perseveranza of 10 December 1920. Strangely enough, this performance is not even mentioned in Musica d’oggi, but the performances of the opera at the Teatro Carcano given two years later, in January 1922, are reported in Musica d’oggi of February 1922, 58, as well as four further different performances in the same theatre from 1922–1924 (see Musica d’oggi of October 1922, 284; July 1923, 227; January 1924, 25; and October 1924, 319). It remains unclear whether Puccini’s revisions for the Teatro Carcano were retained in these subsequent performances. For this production, Puccini wanted to reinstate three passages in Act 1, which were either cut at the performances at the Teatro Dal Verme, or in Palermo, or in Paris, and in all subsequent editions of the score. This involves the reinsertion of no fewer than 140 bars of music. But since the full score (74.E.7e, which originated around the same time and was at least accepted by Puccini) does not include these reinsertions, it must remain open whether Puccini really wanted these partial restorations from the original version or only tested them at the Teatro Carcano, as he did on previous occasions with other revisions in different theatres.Note: *That Puccini may indeed have wanted the material permanently reinstated can be concluded from an unpublished letter by Toscanini to Guido Zuccoli of 31 October 1925 (sold some time ago by the music antiquarian J. Voerster, Stuttgart); the conductor wanted to adopt those reinsertions for his performance of the opera on the occasion of the first anniversary of Puccini’s death (on 29 November 1925 at La Scala).There have been some attempts to stage philologically questionable conflations of the various versions of the opera by combining or deleting elements of 74.E.1 and 74.E.2.Note: *The first of these was probably in 1978 in Joachim Herz’s production at the Komische Oper, Berlin; see Werk und Wiedergabe, edited by Sigrid Wiesmann (Bayreuth: Mühl’scher Universitätsverlag 1980).  They may have been theatrically acceptable but were by no means authentic. Puccini’s original version is not completely preserved in orchestral score, due to his own distortion of the autograph manuscript (74.B.1). Moreover, there were certainly some important changes made prior to the first publication, in rehearsals, and at the first performance (see 74.B.3). The existence of so many authentic and diverse sources proves that there is no definitive version of this opera. On the contrary, for almost two decades, Puccini permitted different versions of the score, all of which should be documented in a critical edition.