WHAT DO WE REALLY KNOW ABOUT BACH’S CELLO SUITES?

Bach’s sublime Six Suites for solo cello are possibly the most frequently published works in western music history, yet their source editions are shrouded in mystery. Cellist and writer Jeffrey Solow puts forward an intriguing new theory as to their origins

When I was a young cellist I learnt the traditional story of Bach’s Cello Suites BWV1007–12 – one offered in programme notes, liner notes and introductions and commentaries to published editions, and taught to most students. We were told that Bach wrote them c.1720 in Anhalt-Cöthen for the court cellist Christian Bernhard Linike (1673–1751) or perhaps for Christian Ferdinand Abel (1682–1761), court gamba player (and maybe cellist). Anna Magdalena Bach’s manuscript had been copied from her husband’s lost fair copy, and the copy made by the organist, composer, and disseminator of Bach’s works Johann Peter Kellner (1705–72) was presumed to have been copied from Bach’s lost working manuscript. Two anonymous later 18th-century copies generated controversy: were they, as Anner Bylsma called them on occasion, the ‘earliest editions’ of the suites, or did they reflect a revised version made by Bach himself, as several musicologists suggested? The Sixth Suite was written for the viola pomposa, a now obsolete, five-string, horizontally played instrument invented by Bach, and the scordatura tuning of Bach’s Fifth Suite – the A string being lowered to G – imitated the tuning of its earlier original version for the lute.

In truth, none of these ‘facts’ can be proved, and at most, only a couple of them are likely to be correct. And is not easy to keep current vis-à-vis the latest theories to their origins. Anything committed to print may become obsolete if new information comes to light or as new and persuasive theories are advanced – sometimes during the short interval between being written and being published!

J.S. Bach by Elias Gottlob Hausmann

Bach’s Cello Suites are probably the most frequently published major works in music history – around the time of the millennium, cellist Dimitry Markevitch counted 97 editions in his library, and dozens more have appeared since then. Given that fact, it is perhaps surprising that we are certain about almost nothing regarding these pieces. We do not know why Bach wrote them; for whom he wrote them – assuming they were written for a specific individual; when he wrote them; if he wrote all six at once or composed them separately over several years; or when they were completed. Were they written before, after or contemporaneously with the Sonatas and Partitas for solo violin? Still being debated is exactly for which instrument he intended either all or some of them and what sort of bow hold would have been used.

The only thing that we can assert with full confidence is that all six suites were complete by the time Kellner copied them in Frankenhain in or about 1726, a date determined by the manuscript paper, his handwriting style and similarities to his copy of the Sonatas and Partitas, dated 3 July 1726. Our lack of certitude, however, does not preclude us from being reasonably sure about many aspects of the text or from making intelligent conjectures about other uncertainties. Historical evidence, logic and tracing the transmission of mistakes, ornaments and other content in the manuscripts can inform us about relationships among the surviving sources as well as illuminate some of the mysteries surrounding the suites.

To organise our thinking, I have identified three main questions that continue to torment, confuse and rile today’s interpreters of the suites:

1. What is the correct text?

2. What did they sound like in Bach’s time?

3. How should they be played now?

Question no.2 is unanswerable. No matter what may have been written during the 18th century or how many scholarly studies and opinions about performance style have been presented since, words are not sounds, and without recordings, or unless someone invents a time machine, we will never know what the suites sounded like when they were played in Bach’s time. Question no.3 has as many answers as there are cellists, Bach lovers and cello aficionados, and this holds true whether one plays the suites on a modern or a Baroque cello or with a Baroque or a modern bow. Consequently, I will confine my discussion to question no.1, which concerns the text (the notes, rhythms and articulations) – the starting point for all interpretations and performances of the suites.

THE SOURCES

Crucial to our understanding of what is most likely to be the correct text are the relationships among the five known manuscript sources A, B, C, D and H (see Sources for the Suites panel, right): what was copied from what? (The currently accepted letter designations for the sources come from Hans Eppstein’s 1990 critical commentary for his edition of the suites in the Neue Bach-Ausgabe.) Other sources must have existed which are no longer extant, at least two of which, [F] and [G] (see panel), are logically undeniable (Eppstein gives lost sources in brackets). Other copies were surely made, but only ones that were part of the chain of transmission can be logically postulated.

Bach must have made a composing or working manuscript, now lost (it has been called Ω or [X] but is now commonly referred to as [F]). Eppstein argued that the existence of a single source for C and D, which he called [G], is supported by the layout of those manuscripts’ pages and the similarity of their notes, newly introduced articulations and the many added ornaments. In the commentary for his 2016 edition, Andrew Talle convincingly proposes that [G]’s scribe may have been Bach’s student and son-in-law Johann Christoph Altnickol (1719–59), who was a singer, violinist, organist and cellist. He was also a trusted copyist, and Bach tasked him with making two copies of his Well-Tempered Clavier, Book 2. If [G] was made by Altnickol or another Bach student, it may contain ideas from Bach. Indeed, Bach’s student Heinrich Nikolaus Gerber (1702–75) copied the Inventions and Sinfonias in 1725, and Bach scholar Yo Tomita notes: ‘In it we can confirm the traces of Bach’s teaching in the form of added ornaments, some of which being identified as in [Bach’s] own handwriting […] Naturally one would speculate that they were added extemporarily during lessons to demonstrate how to execute them effectively on the keyboard.’

The earliest printed editions (see panel, page 39) reflect which sources were known at the time they were published. Even though many of today’s scholars, editors and cellists give precedence to A (which was recognised as an important source only decades after it was acquired by the Berlin State Library in 1841), it is clear that the two earliest editions (E and Probst) were directly or indirectly based on [G]. For proof, one has only to look at the notes of the third and fourth beats of bar 27 in the G major Prelude, which follow C and D (which are themselves directly based on [G]) and are quite different from A and B (see figures 1 and 2). This bar and others that are correct only in B (for example, the penultimate semiquaver (s) in bar 95 of the D major Prelude) show that cellist and editor Friedrich Dotzauer somehow had access to that manuscript (or a related one) when he prepared his edition and recognised the superiority of some of its readings over those of E or the Probst edition (which are basically identical). The fact that Dotzauer corrected slurs in the Fifth Suite Sarabande, a movement that Kellner did not copy, points to his having had access to a related manuscript, rather than B itself.

FIGURE 1 Bach Cello Suite no.1 in G major – Prelude, second half of bar 27: in C, D, edition E (1824) and Probst edition (1825)

FIGURE 2 Bach Cello Suite no.1 in G major – Prelude, second half of bar 27: in A, B and Dotzauer edition (1826)

Anna Magdalena Bach (1701–60)

RELATIONSHIPS AMONG THE SOURCES

In the recent publishing history of the suites, scholarly editors generally fall into two groups (see Traditional Stemma, opposite): the Anna Magdalena Bach (AMB) camp, and the C and D camp. The former posits that her copy, source A, was made directly from Bach’s fair copy and is the closest we can get to Bach – as Ulrich Leisinger stated in his urtext edition (2000), ‘There is no reason to doubt that Anna Magdalena had access to a manuscript of her husband’. The C and D camp holds that Bach revised the suites and his later thoughts were reflected in [G] (the lost source for C and D) – ‘It is unlikely that new findings in the future might show that the version of sources C to E does not represent JSB’s intentions,’ or, as Markevitch puts it in his 1964 edition, ‘We can get an almost perfect rendering of what Bach intended in regard to notes, bowing and ornaments.’ Hardcore AMB followers believe that source A’s inconsistent slurs truly reflect Bach’s intentions, while some who have given their allegiance to [G] feel that, ‘It is not impossible that the manuscript in Carl Philipp Emanuel Bach’s possession was a second autograph by Johann Sebastian Bach’ (Leisinger).

Kellner’s manuscript, source B, has generally been viewed as an outlier in that he was thought to have worked from Bach’s composing manuscript (not a fair copy), thus reflecting the composer’s earlier, unrefined thoughts about the suites, and also because being an organist whose copy was not intended for cellists, Kellner did not pay careful attention to Bach’s slurs. As a presumed pre-revision text, B was deemed to be related to A – Eppstein even prepared two urtext versions for the Neue Bach-Ausgabe, one based on A and B, the other on C and D. Hence was formed the traditional view of the relationships among the sources.

HARDCORE ANNA MAGDALENA FOLLOWERS BELIEVE THE INCONSISTENT SLURS TRULY REFLECT BACH’S INTENTIONS

ANNA MAGDALENA IMAGE BACCHAUS EISENACH

[X] (working manuscript, now called [F]) → Source B (Kellner)

[X] → fair copy (once called [F]) → Source A (Anna Magdalena)

[X] → fair copy → Source [G] → Sources C, D, E

[X] or fair copy → Source H (Bach’s lute arrangement)

As noted above, the traditional story of the relationships among the sources assumes a fair copy in the hand of Bach. If such a fair copy ever existed, however, no trace of it has yet been found and it did not serve as a source for any surviving copies. Recently, Talle has conclusively demonstrated that the numerous errors present in all four MS cello sources, A, B, C and D, preclude the existence of a fair copy or a revised version in the chain of transmission. ‘Plentiful examples of shared notational conventions and common errors reveal that all extant sources – including Source B – trace their ancestry to a single manuscript; Source [F],’ he says. ‘If Bach made two autographs of the Cello Suites it is inconceivable that he would have made so many of the same errors twice.’ This makes [F] (previously known as [X]) the archetype: the latest source from which all other sources descended (see Revised Stemma, page 43).

How, then, to explain the divergent pitches, ornaments and slurs in [G] without invoking a second, revised manuscript? In his commentary, Talle hypothesises that [F] existed in two states, an early ‘clean’ version that served as the source for A and (indirectly) for B, and a later ‘revised’ version with written emendations by Bach. Talle offers a scenario in which Bach, after making his lute arrangement as well as having given lessons to cellists who had made their own copies of the suites, rethought various details and ‘revised Source [F] […] chiefly through the addition of ornaments, dynamics, and articulation marks’. He explains various similarities between the earlier source B and the later additions and modifications to [F] that found their way into [G] as being ‘attributable to the fact that they all came from one mind’.

IS THERE A PLACE FOR A NEW THEORY?

Not only did this scenario feel contrived to me, but as Sherlock Holmes famously declared, ‘when you have eliminated the impossible, whatever remains, however improbable, must be the truth’. And upon closer examination, Talle’s hypothesis cannot be correct.

A differing theory is advanced and discussed by James Grier in his 1996 book The Critical Editing of Music. Grier stresses that ‘errors, not good readings, are the only evidence admissible in stemmatic discussions’, and he references two types of significative errors in descendent sources (which he terms witnesses): conjunctive and separative. ‘A conjunctive error is one in which two (or more) witnesses agree, and it constitutes evidence of the parallel descent of those witnesses from a single common ancestor in which the error was originally committed,’ he writes. Thus, conjunctive errors were most likely copied from a common ancestor; it is less probable that different scribes ‘made the same error independently’. This supports Talle’s conclusion that [F] is the single source for A, B, C and D. However, an error is separative when a ‘witness that transmits the correct reading could not have been copied from [an] erroneous witness, or [a] common ancestor’.

Bach’s autograph manuscript for source H (his lute version of Suite no.5) serves a critical function here. When Bach prepared H he corrected the conjunctive errors in A, B, C and D for that suite. For example, on the third note in the first bar of the Prelude, the natural sign (½) that must have been missing before the A in [F] is added in H; it is A flat in all the cello sources (see figure 3).

FIGURE 3 Bach Cello Suite no.5 in C minor – Prelude, bar 1: as in A, B, C, D, H and edition E (1824)

FIGURE 4 Bach Cello Suite no.5 in C minor – Allemande, bar 25: (a) in B and H; (b) in A, C, D and edition E (1824)

FIGURE 5 Bach Cello Suite no.5 in C minor – Courante, bar 11: (a) in B and H; (b) in A, C, D and edition E (1824)

However, other errors-in-common in A, C and D are separative; B’s readings are corroborated by H. For example, on the second beat in bar 25 of the Allemande, B and H have two semiquavers followed by a quaver (e) (figure 4a), whereas A, C and D all have a quaver followed by two semiquavers (figure 4b). Another instance occurs in the middle of bar 11 of the Courante, where B and H have a quaver figure in the melody line (figure 5a), whereas A, C and D all have a dotted quaver–semiquaver figure (figure 5b).

Since Grier stresses that only errors, not good readings, are admissible as stemmatic evidence, the many common errors in A, C and D in the other suites that are generally accepted as being correct in B are not conclusively separative – although if I were a trial lawyer I would still present them to the jury to support my argument.

WERE THERE ANOTHER TWO SOURCES?

Title page of a 1736 songbook, thought to depict J.S. and Anna Magdalena Bach

The question, then, is to explain how some errors-in-common among sources A–D are conjunctive, having all descended from [F], yet other errors are separative between A, C and D, and B. In his discussion of the suites, Grier offers three possibilities. Either Kellner amended his copy through conjecture (which he deems highly unlikely), or he based it on a copy corrected by Bach (which I will discuss below), or most intriguingly: ‘the only other conceivable explanation is that A was not copied from the autograph but from a copy in which the errors common to A, C and D were committed.’ Thus, Grier argues that a hyperarchetype (a source from which only some surviving witnesses descend, and which I will call [X]), must have existed between [F] and A, C and D.

One might reasonably ask why Anna Magdalena did not directly copy from Bach’s autograph. Indeed, as a distinguished colleague put it: ‘It is very far-fetched to think that AMB would have copied from anything else other than her living husband’s own hand.’ However, if there were enough people who needed to make copies at the same time, it would have been practical to have more than one exemplar for them to work from. Perhaps Bach required that each cello student in the Thomasschule make his own copy of at least some of the suites. There is no doubt that he required keyboard students to copy and study his keyboard works, so that would make sense. It should be noted that in a footnote to a 2020 article in the Journal of the Riemenschneider Bach Institute, Talle commented that he had become convinced, largely through conversations with me, that Grier was correct in arguing that Anna Magdalena’s source could not have been a Johann Sebastian autograph. He states: ‘There is no reason to believe that Anna Magdalena’s copy is closer to the composer’s lost autograph than are any of the other surviving manuscripts.’

As to the possibility that Source B was copied from a Bach-corrected intervening manuscript, not only is that idea torpedoed by the demonstration of there being only a single autograph for the suites, but if a corrected manuscript existed, would Bach have sent it to a 21-year-old organist that he (at the time) had never met, in a city 128 miles away? Not likely. But this also raises the question: what exactly did Kellner copy?

IF A CORRECTED MANUSCRIPT EXISTED, WOULD BACH HAVE SENT IT TO A 21-YEAR-OLD ORGANIST HE HAD NEVER MET?

ALONE A MONG THE COPYISTS, ANNA MAGDALENA WAS TRYING TO MAKE A CELLO-FRIENDLY COPY

Here Talle’s very reasonable conjecture that cellists would have heard about the existence of the suites and visited Bach in order to copy them and take some lessons makes perfect sense. (Kirsten Biesswenger also suggests that B’s source was a cellist’s copy.) In fact, this theory is supported by textual evidence (figure 6). On the first beat of bar 8 in the D minor Sarabande, Source A has a trill on a single note E, Sources C and D have the trill on the upper note of a double-stop 3rd, which is awkward to play, while Source B has the trill on the E above the open C and G strings, a much more cellistic solution. Since Kellner was an organist, it is highly unlikely that he would have initiated this change himself; it was almost certainly present in his source (which I will call [Y]) and that probably came briefly into his hands when it or its owner was in Frankenhain.

Hence, I offer two additional lost sources and a revised system of relationships. The only logical explanation for the numerous errors in Sources A, C and D that are not present in Source B is that they all descended from [X]. And the most logical explanation for the changes and additions to Sources C and D is that, while some may have been suggested by Bach, they all first appeared in the anonymously copied [G], which was indeed, as Bylsma described it, the earliest edition of the suites.

LIKELY LOST SOURCES

Source [X] → source for A and [G] → hence C, D and E

Source [Y] → intermediate source for B

So where does that leave us regarding the question of the text, to which I said I would confine my discussion? Let us first consider the notes and rhythms. Although most discrepancies among the sources can be resolved through harmonic considerations, musical logic (i.e. parallel passages) and common sense (clear copying errors), some plausible alternatives remain. However, if Sources A and B are only one step removed from Bach, while C and D are twice removed, to me that argues for their primacy in spots such as in figure 1 (bar 27 in the G major Prelude). And the fact that many of Kellner’s unique readings are universally accepted as correct, such as bar 30, beat 2 in the C major Prelude, and bar 95, beat 4 in the D major Prelude, inclines me to trust other standalone readings from Source B, e.g. E flats in bar 3 and 7 of Menuet 2 in the First Suite, and bar 18, beat 2 in the D major Gigue. Following this logic resolves most of the important note and rhythm questions.

Articulations and ornaments are another story – here deciding among the sources is pretty much a crapshoot. The fact that Kellner was not copying with an eye to cello playing weighs against his slurring, although not against his ornaments. C and D’s twice-removed descent and the lack of provenance for [G] makes me suspicious of many (but not all) of the added ornaments, and the slurs, while clearly written, do not always make musical sense. Alone among the copyists Anna Magdalena was trying to make a cello-friendly copy without editorial interventions. Unfortunately, she was clearly hurried and probably harried – coping with familial distractions by day or copying at night by expensive candlelight – resulting in her famously unclear and error-strewn slurs. Personally, I accept or reject, and pick and choose among the sources as my musicality and a lifetime of study and experience guides me.

Luckily, any interpretation of the suites will not succeed or fail because of textual details, ornaments, or bowings – the music is the important thing. Character, tempo, rhythm, timing, phrasing, energy, and flow are the crucial elements of a performance, and these things, as always, remain in the hands of the player.

FIGURE 6 Bach Cello Suite no.2 in D minor – Sarabande, bar 8 in A, C and B