Understanding this problem requires knowing something about the sources that transmit Bach’s work. We know the St. John Passion, first of all, from a large stack of vocal and instrumental parts Bach used in his various performances. It turns out that there are four layers, each representing a performance different from the others: Versions I (1724, Bach’s first passion season in Leipzig), II (1725), III (c. 1732), and IV (c.1749, near the end of Bach’s life).

The layers of parts and the performances they represent suggest a useful working definition of a “version” of the St. John Passion: a form of the work as it was performed under Bach’s direction and as documented in a set of parts. In practice this is not so simple because what survives is not four complete sets of parts. Rather, we have the set-aside remnants of the parts for Version I (the rest are lost); the bulk of the surviving material is from version II, which is represented by an essentially complete set. Further, Bach did not make new parts for versions III or IV but instead marked up the parts from Version II. A version, then, does not necessarily correspond to a set of parts, but to the state of a set of parts at a certain moment.

Bach’s composing score is lost, but we do have a copy of it. It is partly in Bach’s hand, having been started in the late 1730s, but never completed, leaving that task to a copyist who finished it years later. The assistant made a faithful copy of his portion, but Bach, apparently not content simply to copy music he had composed almost fifteen years earlier, revised the piece as he wrote, making changes to the first ten movements. This score, then, represents a revision of the St. John Passion by the composer and is arguably yet another version of the piece. But these revisions were never heard in Bach’s time because the new readings never found their way into any of Bach’s performing parts, even those of version IV, which took place after the revisions were made.

Whether or not we regard the music in the later score as a true version we have a wealth of choices in performing the St. John Passion. The version that most modern listeners know today resembles Version I (1724). It opens with the chorus (really a choral aria) “Herr, unser Herrscher”, whose text is a poetic paraphrase of a psalm, and ends with the choral aria “Ruht wohl” and a simple chorale setting, “Ach Herr, laß dein lieb Engelein.” It includes a number of accompanied recitatives and arias among the commentary movements distributed throughout the story.

When Bach performed Version II in 1725 he replaced the poetic opening chorus with an elaborate chorale setting, “O Mensch, bewein dein Sünde groß,” the same movement that in 1736 would close the first part of the St. Matthew Passion. The closing chorale was replaced by a different one, “Christe, du Lamm Gottes,” in a setting borrowed from a cantata Bach had performed at his Leipzig audition in 1723. Bach added or replaced some of the work’s solo arias as well; altogether there are three new arias. Their texts particularly emphasize apocalyptic images and offer different perspectives of the meaning of the passion. They make it essentially a different piece, at least from a theological point of view.

In Version III, from c.1732, Bach restored the opening chorus from version I and some solo numbers, but replaced one aria with a substitute and another with a (now-lost) instrumental sinfonia. Version IV, undertaken many years later, was essentially a return to Version I, though we know that Bach added a contrabassoon to the orchestra and used harpsichord rather than organ for the continuo. He also provided revised texts for some of the poetic movements, texts that commentators have described as more rationalistic than the original versions.

Most recordings of the St. John Passion, following an influential edition, present a mixed version using the revised readings of the first portion (the part Bach copied in his new score) filled in with readings from Version IV—but without that version’s revised texts—and some reworkings of arias from Versions II and III. From the strictest point of view, of course, this is no version at all, but rather a modern pastiche.

More recently, recordings have begun to present specific versions of the St. John Passion. Version II (1725) is available on one of the recordings conducted by Philippe Herreweghe (HMC 901748.49). One conducted by Masaaki Suzuki (BIS CD-921/922) offers Version IV (1749, including the use of harpsichord continuo), and adds the three arias from Version II as an appendix. The recording conducted by Kenneth Slowik (Smithsonian ND 0381, with particularly good liner notes) presents a mixed version but also includes the alternative movements from Version II in such a way that a CD player can be programmed to present that version’s pieces (less the original readings of Nos. 1-10) in order. Listeners have a wealth of St. John Passions from which to choose, in every sense.

Bach’s vocal and instrumental forces

Few issues in eighteenth-century music have attracted as much attention as the size and composition of Bach’s vocal forces. Today we typically hear forces that are not only larger than Bach’s, but that are also organized and deployed very differently. Our performing tradition has inherited the use of large choirs (and orchestras to match) and also a sharp distinction between the roles of vocal soloists and choir members.

We typically divide singers into two categories: chorus members and soloists. In Bach’s time there were also two kinds of singers, but they were distinguished differently. The principal kind of eighteenth-century singer, called a “concertist,” was essential to the performance of a vocal work. There was one concertist in each range in a typical performance. For example, one soprano concertist was responsible for the entire soprano line, singing all the recitatives and arias in that range. Similarly, alto, tenor and bass concertists were responsible for the music in their ranges. But each of these singers was also responsible for his line in ensemble pieces that called for soprano, alto, tenor and bass singing at the same time. This kind of piece, which typically involved most or all of the instruments as well, was most often found at the beginning and end of a church cantata or similar work, and the eighteenth-century name for a movement like this was a “chorus.” In this sense a chorus is a kind of movement calling for all the voices together, usually with all the instruments as well. Such a movement is a chorus even if sung just by these four singers—it does not require a big ensemble of the kind we often associate with the word. A piece like a passion could be (and, it seems, sometimes was) sung just by these four principal singers singing solo numbers on their own and functioning as a group in choruses.

But the director of a performance could choose to add more singers known as “ripienists,” the other category of eighteenth-century vocalist. These singers had no musical numbers of their own but joined the concertists as reinforcements in appropriate numbers. In a passion or church cantata by Bach, for example, they might typically sing the choruses and chorales—and only those movements—leaving arias and recitatives to the concertists. It is essential to understand that the choral passages are sung both by the concertists and the ripienists—the concertists keep singing when joined by ripieno singers. Concertists sang everything and ripienists, who had no music of their own, simply doubled the concertists when told to do so. We know that this was Bach’s practice in his church music because we have so many of his original performing parts, and their design strongly suggests that they were meant to be used by a relatively small number of singers who fell into these categories.

We have an especially rich collection of parts for Bach’s surviving passion settings, and they suggest a clear picture of his own performances. For the St. John Passion, for example, Bach prepared four principal vocal parts that contain essentially all the music in each vocal range: a soprano part containing the soprano arias and the soprano lines of each chorus and chorale; an alto part with equivalent material in its range (arias, choruses and chorales); a tenor part (labeled “Tenore Evangelista”) that includes all the tenor range music including the tenor lines of the choruses and chorales, the tenor arias, and the tenor recitatives conveying the Evangelist’s words, and a bass part (“Basso Jesus”) including almost all the bass-range music, including Jesus’ words.

These are the parts for four vocal concertists who performed essentially everything. A few small parts provide the small character roles, but they are marked with the instruction to remain silent in other movements. In addition, Bach prepared four additional parts containing the lines for choruses and chorales. These are ripieno parts, meant to reinforce the sound in certain movements. There is internal evidence suggesting that each of these parts was designed for use by one singer, meaning that Bach’s performances of the St. John Passion used about 10 singers: four principal singers, four additional voices for choruses and chorales, and a couple of others for certain small roles.

Bach’s St. Matthew Passion, famous for its employment of double chorus and orchestra, is actually set up very similarly. For this work Bach also prepared four principal vocal parts, including one for a tenor who sings the Evangelist’s words and one for a bass who sings the role of Jesus. As in the earlier work, these four parts contain all the music in their ranges: choruses, chorales, recitatives and arias. There are also four additional parts for four more singers in the St. Matthew Passion. In many movements the singers who used them were essentially ripieno singers, but in this work they also formed Chorus 2 (the principal singers are Chorus 1), and were also assigned arias.

That is, in the St. Matthew Passion Bach asked his additional singers to take on some independent music, not just to double the principal singers. This distribution is often not reflected in modern performances and recordings, which frequently employ only one aria singer in each range. Altogether, Bach’s performances of the St. Matthew Passion called for eight singers (who make up Choruses 1 and 2 and who sing solo music), plus a few others for small roles and for chorale melodies in the opening and closing movements of the first part of the passion.

In both this work and the St. John Passion, this ensemble is not only substantially smaller than that typically heard in modern choral performances but also conceptually different as well. For one thing, there was no distinction between “soloists” and “chorus members.” In this way of thinking a “chorus” is not a distinct ensemble; rather, it is a kind of piece sung by all the concertists together, joined (where present) by ripieno singers. Another difference is that the principal singers served several different functions in the work, presenting poetic and hymnic commentary (arias and chorales), narrative in recitative (especially the tenor who sang the Evangelist’s music and the bass who sang Jesus) and the portions of the narrative sung by the vocal ensemble together (choruses of groups in the passion narrative).

A recording of the St. John Passion that uses forces in the manner discussed here is conducted by Andrew Parrott (Virgin Classics 62068, together with the Easter Oratorio, BWV 249 and Mass in B minor, BWV 232). The recording of the St. Matthew Passion conducted by Paul McCreesh (Archiv 474 200-2) also deploys forces almost exactly as documented in Bach’s performing materials for that work. These recordings give the modern listener a chance to hear these works from a very different perspective.

A lost Bach passion, and one that never existed

A careful observer will notice that the Bach passions make a numerical jump from the St. Matthew Passion, BWV 244 and St. John Passion, BWV 245 directly to the St. Mark Passion, BWV 247, skipping over BWV 246. Readers of the New Testament will note the lack of a passion based on Luke’s Gospel in Bach’s output. In fact, there is a piece that fills both of these gaps: a St. Luke Passion, BWV 246 that sets Luke’s words as a biblical oratorio with interpolated poetic arias and chorales, just like Bach’s other passions, and that was once attributed to him.

The modern history of the St. Luke Passion begins with the acquisition of a manuscript score in the early 19th century by a collector who recognized it as being in the hand of J.S. Bach. The collector’s devotion to the work eventually developed into a conviction that Bach had composed the work, not just copied it. The passion found a further champion in 1880 when Philipp Spitta published the second volume of his magisterial study of Bach’s life and music. Spitta argued strenuously for Bach’s authorship of the St. Luke Passion, basing his judgment largely on comparisons with other early Bach vocal works. (Unfortunately, some of the “Bach” pieces to which he compared it have since turned out not to be by Bach at all, raising serious questions.)

The St. Luke Passion was published in a performing edition in 1887 and again in 1898 in the prestigious complete edition of Bach’s music, events that led to many performances under Bach’s name—at least until a torrent of critical writings and some manuscript studies eventually debunked the Bach attribution. For the last 100 years or so, the work has been regarded merely as part of Bach’s working repertoire, with many authors claiming he performed it in 1730 and again in the later 1740s. These supposed performances turn out to be poorly documented, but at least their adherents no longer claim Bach was the work’s author.

The true identity of the composer has not been uncovered, but the St. Luke Passion’s Bach connection has helped keep this otherwise obscure piece—a work definitely worth hearing—on the radar screen. There is an excellent recent recording conducted by Wolfgang Helbich (cpo 999 293-2). Its title (“Johann Sebastian Bach: Apocryphal St. Luke Passion”) shows how tightly Bach’s name has stuck to this work that is actually not his, and gives listeners the chance to hear a nearly-contemporary passion setting that Bach himself knew.

We have the opposite problem, in a way, with Bach’s St. Mark Passion, BWV 247. We know that Bach presented this work on Good Friday 1731, but his musical materials do not survive, depriving us of the score of this companion work to the St. John and St. Matthew Passions. A lost Bach passion is a tantalizing challenge, but the only way to perform the St. Mark Passion would be to reconstruct it. In fact, any number of musicians and scholars have undertaken this task over the years. A close look at the results forces us to ask whether Bach’s name can legitimately be applied to the results.

We can contemplate a reconstruction of the St. Mark Passion because its text survives in a collection of poetry published in 1732 by its librettist Picander, who wrote the eight poems and probably chose the sixteen hymn stanzas distributed as commentary throughout the narrative. We can recover many of the chorales in the St. Mark Passion because they are probably represented (though without labels that make their exact identity certain) in collections of Bach’s chorale settings prepared in connection with his teaching.

A different approach can be taken to the arias and choruses. It was noted in the nineteenth century that some of the poetic texts in Picander’s libretto correspond to verses Bach had set in an earlier piece, the 1727 Trauer Ode, BWV 198 composed for a University memorial service for the late Electress of Saxony. In fact, we know that Bach often re-used vocal music with new texts for new purposes, a process known in modern terminology as “parody.” Some of Bach’s most important vocal/instrumental works (including the Christmas Oratorio and Mass in B minor) were assembled this way. His contemporaries used parody, too: Handel arranged Italian chamber duets of his own composition to produce “For unto us a child is born” and other choruses in Messiah.

Five choruses and arias from the St. Mark Passion look to be poetic parodies of movements from the Trauer Ode, and in principle we should be able to restore them using the old music and new text. But we do not know whether Bach followed through on the planned parodies, or how extensively he might have reworked the borrowed music, or the keys and instrumentation of the new versions. Neither do we know what music he used (or composed) for the other arias in the work. Even worse, Bach’s setting of the core of the passion setting—the Gospel narrative—is lost. It has been claimed that a few passages can be recovered from their later re-use in works that do survive, but this is doubtful and would account for only a small part of the music in any event.

Over the years, hopeful reconstructors of the St. Mark Passion have dealt with the loss of the Gospel portion in different ways. They have borrowed music from other composers’ settings, embedding Bach’s arias in them in a hybrid work; they have written new recitatives and choruses themselves; they have used music from Bach’s St. Matthew Passion; they have presented the narrative purely as spoken dialogue and they have eliminated the Gospel text altogether, offering a purely reflective passion cantata consisting only of poetic arias and choruses.

The interested listener can hear the work (or perhaps wishful thinking) of various reconstructors on recordings. Among the attempts are one conducted by Peter Schreier (Philips 456 424-2), with spoken narration; one conducted by Roy Goodman (Brilliant Classics 99049) using the gospel narrative from a St. Mark Passion attributed to Reinhard Keiser; one conducted by Geoffrey Webber (Gaudeamus 237-2) using the same narrative; one conducted by Hans Gebhard (Eres 24) presenting the work as a series of reflections using no narrative at all; and (most recently) one led by Ton Koopman (Erato 8673-80221-2) using newly composed recitatives and other movements freely adapted from various works by J.S. Bach, but systematically avoiding the pieces from the Trauer Ode.