Any attempt to write a history of the libretto is fraught with paradox. Almost without exception, a text is the starting point for any opera. Indeed, before Mozart, and often after, the libretto was normally complete before the composer put pen to paper, for all that it might then be revised according to the musical and other demands made upon it. As we shall see, its poetry usually had quite precise musical implications. Moreover, in early opera the poet was normally the prime mover in the operatic enterprise, not just by devising the subject and fleshing it out with appropriate words, but also given his often standard role as director of the production. The libretto was itself the public face of opera in terms of the artefacts that survive to record a given performance: libretti were usually printed for general consumption inside or outside the theatre, whereas musical scores were, on the whole, regarded as more ephemeral performance materials, to be adopted, adapted, and disposed of at will. Poets also acted as the chief ideologues of opera, promoting and defending the genre against its detractors and inserting it into broader literary and cultural debates. In a very real sense, the history of the libretto is the history of opera tout court.¹

Yet as countless librettists have complained, the words rarely come high on any opera audience’s agenda. The music, singers, mise-en-scène, costumes, and choreography vie for the attention of the eye and the ear, while the text, if it is held in any regard at all, is dismissed as a trying necessity or a trifling irrelevance. The beauties of a poet’s verse are as nothing compared with the beauties of a composer’s music, and, in some minds, both pale in comparison to the beauties of a singer’s high C. In this light, the history of the libretto is just one relatively minor branch of opera studies.

The point is confirmed by a platitude: the best poetry can rarely be set to music because it is too self-sufficient, with nothing to be added. While this may or may not be true, the more damaging corollary – that any poetry for music must be second-rate – ignores the fact that libretti should be judged not by the canons of ‘great’ verse (although some are) but, rather, by fitness to purpose. Most theatre poets accepted, gladly or not, that when writing for music, compromises had to be made in terms of plot design and arrangement, and of poetic language, accent, and even vowel sounds (it is hard to sing a melisma on u). The mere fact that it always takes longer to sing something than to say it conditions the nature of poesia per musica, which, in turn, must always expect completion by something beyond itself. Nahum Tate’s verse for his dying Dido (in Purcell’s Dido and Aeneas) is scarcely great, or even good, poetry:²

However, it serves its purpose well. Why that should be the case is something worth exploring.

‘a Poetical Tale or Fiction’

In 1685, the English poet John Dryden published his Albion and Albanius, set to music by Luis Grabu as a full-length opera (the first in English to survive). His preface begins with the nature of operatic subject matter:

This ‘marvellous and surprizing conduct’ extends beyond the implausible plots and dei ex machina so typical of the genre. Still more ‘surprizing’ is opera’s fundamental premise, that drama can be played out in song. The consequent lack of verisimilitude might best be accepted as just a fact of operatic life, but it remained troubling in an age that still paid at least lip service to precepts drawn from Classical poetics, notably the writings of Aristotle and Horace. This explains the subject matter of the earliest operas in the north Italian courts, drawn chiefly from Graeco-Roman myth, where supernatural gods and goddesses could reasonably be expected to differentiate themselves from mere mortals by way of music. It also explains their standard setting in the pastoral utopia of Arcadia, where poetry and therefore music are natural conditions of an idealised life. For Dryden the presence of gods, goddesses, and heroes

The gradual expansion of operatic subject matter throughout the seventeenth century attenuated the pastoral argument and called for further special pleading in printed prefaces, as well as in the other standard forum for operatic apologias, the prologue. Even before opera went ‘public’ in Venice in 1637, its subjects were extending beyond the standard mythological–pastoral fare of earlier court entertainment to embrace epic (Homer’s Iliad and Odyssey, Virgil’s Aeneid, Lodovico Ariosto’s Orlando furioso, Torquato Tasso’s Gerusalemme liberata, Giambattista Marino’s Adone), and even Greek and Roman history. Sacred operas likewise made the transition from the representation of allegorical virtues and vices (Emilio de’ Cavalieri’s Rappresentatione di Anima, et di Corpo of 1600) to quasi-historical accounts of saints’ lives (in the operas staged in Rome under the patronage of the powerful Barberini family from the early 1630s onwards).

The appeal of epic is easily explained. Tasso’s Rinaldo and Armida first appeared in a set of intermedi by Ottavio Vernizzi (Bologna, 1623), followed by Benedetto Ferrari’s Armida (Venice, 1639), Lully’s tragédie en musique, Armide (Paris, 1686), and John Eccles’s Rinaldo and Armida (London, 1698). Characters from Ariosto’s Orlando furioso appear in Marco da Gagliano and Jacopo Peri’s Lo sposalizio di Medoro et Angelica (Florence, 1619), Francesca Caccini’s La liberazione di Ruggiero dall’isola d’Alcina (Florence, 1625), Luigi Rossi’s Il palazzo incantato (Rome, 1642), Lully’s Roland (Versailles, 1685), and Agostino Steffani’s Orlando generoso (Hanover, 1691), to name only a few. The trend increased in the eighteenth century, from Handel’s Rinaldo (1711) and Alcina (1735) through Gluck’s Armide (1777) and beyond. The line between myth and epic was thin, and both Alcina and Armida owe a clear debt to the classic femme fatale of Homer’s Odyssey and its mythological forbears Circe, save that in the later case – and no doubt to the gratification of seventeenth- and eighteenth-century censors – these pagan sorceresses could ultimately be redeemed by the love of a Christian hero. The chief difficulty in both cases, however, was how to turn epic narration into dramatic representation, which usually involved the introduction of extraneous characters (divine or mortal) to explain the plot.

Any subject drawn from history might seem to cause greater problems, although these may be more apparent than real. Here, the question of verisimilitude comes most to the fore. As the librettist Francesco Sbarra admitted in the preface to his Alessandro vincitor di se stesso (1651, set by Antonio Cesti), dealing with Alexander the Great:

Some ‘historical’ characters inhabit a hinterland between fact and fiction: the heroes of the Trojan Wars (Achilles, Aeneas, Ulysses) may actually have existed, but they also have strong mythical properties, and they occupy a world shaped by divine intervention. Both Ariosto and Tasso drew inspiration from history (respectively, the time of Charlemagne and the First Crusade), and yet they subjected their heroes to trials and tribulations inspired by Classical mythology and by medieval romance. The kings and queens of ancient Mesopotamia and the Middle East that also started to populate operas were probably not significant historical presences. Even Roman dictators and emperors (Julius Caesar, Claudius Nero) were not necessarily to be viewed in the same light as characters in, say, a Shakespeare ‘history play’. Indeed, in all these cases, it is often the exotic otherness of their stories that makes them appropriate for operas, which, in turn, were not to be construed, at least directly, as some kind of lesson in the facts of history, even if they raised important questions about how history might usefully be read.

Much has been made of what has often been called the ‘first’ historical opera, Monteverdi’s L’incoronazione di Poppea (Venice, 1643; libretto by Giovanni Francesco Busenello), which is based on the erotic antics of Emperor Nero and his mistress, Poppaea Sabina, and the consequent downfalls of Empress Octavia (sent to exile) and the philosopher Seneca (condemned to suicide). In a preface to his own edition of the libretto, Busenello acknowledged the outline of the events treated in the opera as described by Tacitus. ‘But here’, he states, ‘we represent these actions differently.’⁵ Busenello further tempers any claim for historical veracity by typically relying on the intervention of the god of Love (‘without somewhat of which Passion, no Opera can possibly subsist’, says Dryden). Such treatment might or might not be a case of legitimate poetic license, but it does question the extent to which modern critics should judge this famously immoral plot on the grounds of their own historical knowledge (e.g., that both Poppaea and Nero eventually came to sticky ends).

The problem is not restricted to opera. Contemporary spoken drama runs through a similar gamut of genres, styles, and subjects, and seems equally fluid in terms of potential interpretations. So, too, do commedia dell’arte scenarios – which range far more widely than just the stereotypical characters and plots often viewed as standard in the genre – and likewise a relatively unexplored source of material for Baroque opera, contemporary novellas and related ‘popular’ literature. All these establish a number of plot-types that, in turn, prompt variations on a set of standard themes. For example, two young lovers, one or both in an unhappily arranged engagement or marriage, will staunchly resist social and other pressures applied by a pedantic tutor or a busybody nurse, surmounting all obstacles to find true happiness. The fact that this is the foundation of L’incoronazione di Poppea as much as it is of, say, Rossini’s Il barbiere di Siviglia need not cause too much discomfort. But it suggests that originality of invention is not what matters most. In part, this is blatant commercialism: audience familiarity leads to ‘brand loyalty’ and hence increasing consumption. Furthermore, it permits efficient short-cuts in the re-telling of well-known stories. Finally, it creates a strongly intertextual world where operas are to be compared less with ‘real life’ than with other similar works.

While earlier librettists such as Ottavio Rinuccini, Busenello, and Giovanni Faustini had tended to write one-off libretti for a small circle of composers – a practice that, of course, remained in use – some later libretti seem to have become reified as ‘works’ of and for themselves that could therefore gain wider distribution. Giacinto Andrea Cicognini’s text for Orontea had settings by Francesco Lucio (Venice, 1649), Francesco Cirillo (Naples, 1654), Cesti (Innsbruck, 1656), and Filippo Vismarri (Vienna, 1660). Their complex genealogy has yet to be fully disentangled, and, in general, the mechanisms of libretto transmission have not yet been properly studied: presumably they involved complex networks of personal contacts (among impresarios, poets, composers, and singers) and also, and increasingly, of printed editions whether of single works or of collected opera omnia. One result, however, was that some literate audiences might well have started to identify particular works as belonging to their librettists independent of the different musical clothing offered by a succession of composers and singers. The same is true of, say, the libretti of Pietro Metastasio in the eighteenth century, which had a strong literary presence quite apart from their repeated operatic settings.

The latest catalogue of printed opera (etc.) libretti contains some 5,800 entries covering the years 1600–1699, and 24,000 for 1700–1799.⁶ It is hard nowadays to conceive the sheer scale of the operatic enterprise in the seventeenth and eighteenth centuries – especially given the highly limited repertory of most modern opera houses – as the burgeoning opera ‘industry’ created complex infrastructures of supply and demand. But if the various tendencies towards standardisation identified above are certainly to be viewed in this light, they also reflect a codification of genres for further academic and related reasons. Here the need was to resist, rather than promote, certain trends that were coming to be viewed as deleterious to the notion of opera as some kind of drama; it also played into broader debates, such as the Querelle des Anciens et des Modernes that animated French (and thence European) cultural discourse from the second half of the seventeenth century into the eighteenth.

For example, the ‘Arcadian Academy’ was founded in Rome in 1690 for the reform and ‘purification’ of Italian poetry, in particular the opera libretto. It emerged like many such Roman gatherings from the circles of specific patrons, in this case Cardinal Pietro Ottoboni, although its influence spread widely through Italy and abroad. Librettists associated with the Arcadians included Ottoboni himself, Apostolo Zeno, Gian Vincenzo Gravina, Silvio Stampiglia, and Metastasio. Their spokesmen, including Giovanni Maria Crescimbeni (La bellezza della volgar poesia, Rome, 1700) and Ludovico Muratori, ranged widely in their attacks on the abuses of contemporary poetry, advocating a return to Classical simplicity, in part via French models drawn from Corneille and Racine. Giacinto Andrea Cicognini’s libretto for Cavalli’s Giasone (Venice, 1649) came under particularly harsh critique by Crescimbeni:

Thus the Arcadians sought to restore order by regularising opera’s structures, themes, and affective content. But their appeal for a more ‘moral’ form of art went back to Horace’s dictum that art should not just entertain but also educate. Lip service to the ideal was conventionally paid in operatic prologues that sought to justify, or at least explain away, the action that followed. According to its prologue, L’incoronazione di Poppea is a demonstration of the power of Love over Fortune and Virtue, which, if not ‘moral’ enough in itself, might at least prompt a satirical reading of the work by negative example. Other morals were even clearer in those operas where ancient heroes exhibited the qualities of bravery, virtue, honour, wisdom, and clemency that, in turn, could stand as allegories for modern princely patrons; this provides a basis for reading most of Lully’s tragédies en musique as some form of propaganda for Louis XIV. Yet allegory is always a slippery tool. Stories from the appalling life of Nero might well serve pro-Venetian republican propaganda, so one reading of L’incoronazione di Poppea goes.⁸ But in Antonio Giannettini’s L’ingresso alla gioventù di Claudio Nerone (Modena, 1692; libretto by Giovanni Battista Neri), the same ‘historical’ character serves to celebrate the wedding of Francesco II d’Este, Duke of Modena. Either contemporary audiences were able to make subtle and sensitive value judgements about the subjects placed before them, or they did not care much about these subjects at all.

Of course, different members of different audiences would no doubt read different works in different ways. Indeed, such polyvalence was surely an essential condition for opera taking Europe by storm. But if one can perhaps find common ground in what opera ‘taught’ its consumers, it was probably not so much at the level of grand historical, political, or ethical sermons as it was in more immediate modes of human emotional behaviour. Those who cried or laughed at the characters and actions represented on the stage received a sentimental education in the nature of human feeling through which to construct their lives. Some might view this as social engineering; others might claim it as what is most uniquely liberating about the operatic experience.

‘softness and variety of Numbers’

Dryden also discusses the requirements of poetry for music:

Despite the preposterous notion ‘That Rhyme, on any consideration shou’d take place of Reason’, Dryden says, one can only follow the models established by the masters, in this case, the Italians.

Dryden’s ‘softness and variety of Numbers’ refers to the nature of Italian poetry, defined by the number of syllables in a given line and the position of the final accent.⁹ Poetic lines can be from three to eleven syllables in length (thus ternario, quaternario, quinario, senario, settenario, ottonario, novenario, decasillabo, and endecasillabo): the endecasillabo is the ‘classic’ norm, with its chief component, the settenario, in second place. The verso piano has the final accent on the penultimate syllable. An accent on the final syllable produces a verso tronco, and one on the antepenultimate syllable a verso sdrucciolo. Versi tronchi and sdruccioli are counted as modified versi piani: so, a settenario tronco or sdrucciolo will have, respectively, six and eight actual syllables. Syllable counts are also affected by various treatments of synaloepha and diphthongs.

The verso piano is the standard form, while versi sdruccioli and tronchi are used in special circumstances. For example, in libretti versi sdruccioli often invoke pastoral resonances (on the precedent of Jacopo Sannazaro’s Arcadia of 1504); they also have a long history of association (usually, in quinari) with infernal, demonic, or magic scenes, as in the response of the woodland gods to the summons of the wicked witch Artusia in Benedetto Ferrari and Francesco Manelli’s second Venetian opera, La maga fulminata of 1638 (Act III scene 3: ‘Insana femina’) or Medea’s ‘L’armi apprestatemi’ in Cavalli’s Giasone (Act III scene 9; 1649). The verso tronco can be comic – and it is sometimes associated with nonsense syllables – but, on a structural level, it becomes most significant to articulate closure: the result has strong musical implications, given the greater suitability for musically perfect cadences of versi tronchi (with a masculine ending, weak–strong) over versi piani (with a feminine ending, strong–weak).

Italian opera libretti draw only rarely upon the standard poetic forms of Renaissance Italian (Tuscan) hendecasyllabic verse: for example, Dante’s terza rima (rhyming aba bcb cdc…), Petrarch’s fourteen-line sonnet (with two quatrains – abab abab – and two tercets, e.g., cdc dcd) and the ottava rima stanzas of Ariosto and Tasso (abababcc). When they do, it is often for special (archaic, moralising, etc.) effect, as with Orpheus’s ‘Possente spirto, e formidabil nume’ in terza rima in Act III of Monteverdi’s Orfeo. Instead, the basis of early libretti – as of the late Renaissance pastoral plays that provided their most immediate precedent – was a mixture of free-rhyming endecasillabi and settenari, producing versi sciolti (‘loose’ or ‘free’ verse): more regular rhymes and/or metrical consistency could define structural units within this flow, and lines could be divided between characters to enhance the effect of dialogue. However, one major, and crucial, exception is the strophic canzone/canzonetta, generally in other than seven‑ or eleven-syllable lines, that appears even in the first operas. Rinuccini, for example, introduced strophic groupings mixing ottonari and quaternari in Dafne (1598) and Euridice (1600) specifically for the end-of-‘scene’ choruses. His model was the anacreontic verse introduced (in part, on French precedent) by the poet Gabriello Chiabrera (1552–1638), who said that he was catering specifically for composers of the ‘new music’ and their ‘arias’. In Orfeo, the librettist Alessandro Striggio brought such structures into the acts themselves to produce formal songs – often distinguished as such, whether dramatically (e.g., diegetically) or structurally – that stand apart from the prevailing versi sciolti for the recitative. The opening of Act II, for example, has a series of four-line stanzas for Orfeo and his companions (one in ottonari, six in settenari) separated by instrumental ritornelli, culminating in four quatrains in ottonari for Orfeo, producing an aria both in the technical sense (a strophic setting of strophic verse) and in the musical one:

Eleven- and seven-syllable versi sciolti remained standard for recitative (and variants thereof) through the nineteenth century and beyond: their fluidity and flexibility were well suited to its dramatic function and musical style. However, the style of arias (choruses, ensembles, etc.) favoured shorter lines in clear-cut patterns with regular metre and rhyme. From early opera onwards, such dramatic and structural shifts were essentially cued by the librettist, whom the composer could ignore only with potential prejudice to the musical outcome. Thus reading a libretto allows one to predict with a fair degree of certainty what the music is meant to do with it, and therefore also encourages us to be surprised when something different occurs.

Such matters remained fluid throughout most of the seventeenth century as the canons of opera were forged by developing social, political, and even literary contexts. Lodovico Zuccolo (Discorso delle ragioni del numero del verso italiano, Venice, 1623) showed clear contempt for the new canzonetta, claiming it to be a mere sop to musicians. More sympathetic theorists of the second quarter of the century, such as the anonymous author of Il corago (c. 1630), still felt ambivalent about shifts away from versi sciolti: they approved the variety thereby achieved but warned against anti‑Classical improprieties. But as opera entered the public domain, the rising fortunes of the aria could scarce be resisted. Thus at the beginning of Act I scene 2 of Cavalli’s Giasone (1649), Cicognini gives the lovesick hero two stanzas of senari (with a refrain and cadential versi tronchi) before Ercole interrupts his amorous babble. The strong amphibrachs (weak–strong–weak, two per line) almost force a setting in triple time.

The structures adopted for ‘aria’ poetry through the seventeenth century are strikingly varied, at least until the rationalisations prompted by the Arcadians and followed by Metastasio. Here the pattern becomes relatively standard: two isometric stanzas, usually of four or three lines each (settenari – sometimes with concluding quinari – or ottonari tend to be preferred), with regular (and generally parallel) rhyme-schemes, and each often ending with a verso tronco rhyming with its counterpart. This then meshed with the musical structure that was emerging to dominate opera in the early eighteenth century, the da capo aria, with one stanza each for the A and B sections and a subsequent return to A, therefore ending with the first part of the text.

A representative example of what one might call an intermediate stage in this long process of development is provided by a scene from Antonio Sartorio’s Giulio Cesare in Egitto, to a libretto by Francesco Bussani, first performed in Venice in the 1676–1677 season (although the main score that survives comes from a performance in Naples in 1680).¹⁰ Cesare (Julius Caesar) has arrived triumphant in Egypt and is pursued by Cleopatra, who has disguised herself as the maid, Lidia; her servant, Nireno, is also in on the game. In Act I scene 4, Cesare is alone on stage musing on his amorous state – observed by Nireno in hiding – but then is suddenly surprised to hear a voice singing in the distance:

Cesare begins with two stanzas in quinari – which Sartorio sets in ABA′ form, repeating the first stanza after the second – and then moves into versi sciolti (recitative). Cleopatra’s offstage song (‘V’adoro pupille’) is a single four-line stanza (senari) that is interrupted (after line 2) by a brief question in recitative from Cesare (‘What voice do I hear?’); Nireno then takes over this poetic line (indicated by the arrow), identifying the owner of the voice for the benefit of the audience (‘This is Cleopatra’) and then commenting on feminine wiles. The song resumes, and, at its end, Nireno reveals himself (recitative), allowing Cesare to wax rhapsodical over the ‘angelic voice’ he has just heard.

There are several typical games in play here. Cleopatra’s aria is diegetic (performed as a ‘real’ song meant to be heard as such by the other characters onstage), whereas Cesare’s is not (he is just in love); we also have the typically meta-operatic strategy of commenting on the act of singing and on the ‘angelic’ qualities of the singer. It is a scene that invites a mixture of arousal and wry humour; it also merits comparison with Handel’s handling of this same moment in his own Giulio Cesare in Egitto (London, 1724).¹¹

‘they labour at Impossibilities’

Dryden greatly admired the Italian language, ‘the softest, the sweetest, the most harmonious, not only of any modern Tongue, but even beyond any of the Learned’, one that ‘seems indeed to have been invented for the sake of Poetry and Musick’, rich in vowels and with a pronunciation that is ‘manly’ and ‘sonorous’. Other nations, however, can only envy the Italians:

Dryden’s reference to French’s ‘natural harshness’ presumably refers to its consonants, and the ‘perpetual ill Accent’ to the final es left mute in speech but articulated in song. Of course, a seventeenth-century French académicien would not have agreed.

The French equivalent of the Italian endecasillabo is the alexandrine, with twelve or thirteen syllables, depending on whether the rhyme is masculine (accent on the final syllable) or feminine (accent on the penultimate syllable, followed by a mute e). Alexandrines may be subdivided into six-syllable hemistichs by medial caesuras (thus offering the possibility of two parallel musical phrases for a single line of verse). Line-endings tend to alternate between masculine and feminine, often in rimes croisées (abab), although one can also find rimes plates (aabb) and rimes embrassées (abba). But the alexandrine is not always amenable to fluid musical setting given the internal caesura and the strong end-accents; it also seems too long, save where sententiousness or grandeur is required. Accordingly, the French librettist Philippe Quinault tended to opt for freer vers libre, mixing lines of different length. The argument between the flirtatious Céphise and her suitor Straton in Act I scene 4 of Lully’s Alceste (1674) is typical enough (syllable counts ignore feminine endings):¹²

Straton is given alexandrines when it comes to voicing his complaint and when he jumps to the (wrong) conclusion that Lychas has lied to him. The opening couplets are nicely balanced, and Céphise’s final comment wittily plays off two hemistichs to deflate her importunate lover. Her four seven-syllable lines (‘Un ton grondeur et sévère … d’un amant’) also mark a generalised moral that Lully sets apart from the prevailing recitative in the manner of a duple-time aria. However, and in general, French verse has fewer clear structural implications for music than Italian, often making it harder to predict from the text what the composer would do with it. This might have been seen as an advantage, given the French preference for more ‘natural’ and fluid forms of declamation drawing upon (as Lully himself is reported to have done) the rhetorical strategies of spoken drama.

Dryden was equally doubtful about his native tongue:

The suggestion that ‘female Rhymes’ (i.e., based on words with strong–weak endings) are important for song is interesting, while Dryden’s sense of the ‘Effeminacy of our pronunciation’ perhaps relates to the lack in English of strong stresses, equally necessary for good melodic writing.

The standard form of ‘noble’ English verse, the (usually iambic) pentameter, had similar problems to the alexandrine in terms of its length and its tendency to fall into repetitive patterns.¹³ As a result, English librettists of the mid-seventeenth century such as Richard Fleckno (Ariadne Deserted by Theseus and Found and Courted by Bacchus, 1654) and William Davenant (The Siege of Rhodes, 1656) tended to adopt an equivalent of vers libre with two-, three-, four- or five-foot lines: Davenant claimed that such variety was ‘necessary to recitative music’, although the tendency towards rhyming couplets produces a certain sameness. The techniques remained similar in later works. Thus Tate’s libretto for Purcell’s Dido and Aeneas is predominantly in rhyming iambic (weak–strong) and trochaic (strong–weak) tetrameters that become rather plodding (this is typical of German verse, too). Take, for example, Dido’s first speech in Act I as presented in Purcell’s score (the 1689 libretto has some differences):

Purcell’s best option is often to treat this almost as prose, with word repetitions and also a tendency to favour enjambments (thus weakening the rhyme) even at the expense of sense: ‘Ah, ah, ah, Belinda, I am prest with torment, / Ah, ah, ah, Belinda, I am prest with torment not to be confest.’ In more dance-like sections, however, Purcell seems to enjoy piquant mismatches between textual and musical metre and stress (as in the duet ‘Fear no danger to ensue / The hero loves as well as you.’)

Given the regularity of much of Tate’s libretto, his ending (given towards the beginning of this chapter) is rather strange. Like any good librettist, he provides an appropriate number of key words to prompt the composer (‘darkness’, ‘Death’, ‘remember me’, ‘ah!’). But the metre is odd. ‘Thy hand, Belinda’ is in tetrameters, although the feminine line-endings (‘… shades me’, ‘… invades me’) maintain a flow. However, after a reasonably regular quatrain, the metre shifts to a pentameter (‘When I am laid in earth, my wrongs create’), a trimeter, and a final pentameter. Tate seems to want to set apart this portion of Dido’s final speech, as indeed does Purcell: ‘When I am laid in earth’ is, of course, Dido’s lament, over a ground bass typically based on a descending chromatic tetrachord. However, Purcell does also offer one further intervention that must be his. Tate’s syntax is somewhat elliptical, and Purcell seems (consciously or not) to have wanted to clarify the subjunctive, leading to the a-metrical ‘When I am laid in earth, may my wrongs create’.

* * *

‘Prima la musica, e poi le parole’ (‘First the music, and then the words’) was a catchphrase enshrined in the title of a satirical divertimento teatrale by Antonio Salieri (1786; libretto by Giambattista Casti). For all that it is a procedural illogicality, it reflects a common aesthetic presumption about the nature of opera. My aim here, however, has been to demonstrate the benefits of taking libretti seriously in terms of their poetic structures, and also, one might add (though I have not covered it here), for what they tell us about staging. Nor are these benefits limited just to early opera. Poetry was the standard format of opera libretti at least until the late nineteenth century and the rise both of Literaturoper and of a preference for more ‘naturalistic’ dramatic and musical prose. Thus the principles established here operated through the Classical and Romantic periods and had no less impact on, say, a Mozart or a Verdi. Ottavio Rinuccini’s legacy was powerful indeed.