9. Q & A, but few answers

Two weeks ago, I gave a solo recital—something very unusual for me, as I’m really a musicologist rather than a musician.  Although I am regularly involved in various performance activities—most usually as an accompanist or church musician—I cannot recall any solo recital I have done since my senior recital in college.  Indeed, that event is the impetus for this one, as this recital fell intentionally on the twentieth anniversary of that senior recital, in the same hall, on the same instrument.  I decided it would be a good chance to revisit some of the same repertoire; and as I did I was reminded of how I got bitten by the musical text criticism bug to begin with.  (My hitherto unstated goal in this blog has been “writing about very technical points in a non-techical way,” and I fear the technical details of this post will mean that it necessarily falls short of that goal.  I don’t blame anyone for not bearing with me to the end of this post; nevertheless, it will be for me a stroll down memory lane.)

The central item on both recitals was the same:  J. S. Bach’s Canonic Variations on ‘Vom Himmel hoch’ BWV 769—or rather BWV769a, and that brings up the textual point for this post.  The Canonic Variations is a late work, composed in 1746-47 as far as can be determined.  It is transmitted in a number of sources, including two authentic sources—the first edition print (usually allocated the siglum Q, as it will be here) and an autograph manuscript generally described as a fair copy (ditto A).  Both are now freely available on the web.  I have taken the images below from the scans available via the Berlin Staatsbibliothek website (print and autograph), but similar reproductions can be found on Bach Digital and the IMSLP, and other places too.

The relationship between Q and A is not as obvious as one might expect.  It would be reasonable to assume that the fair-copy score came first, and the printed edition later—and, indeed, that the fair-copy score might even have been prepared for the engraver of the print.  In fact these two authentic sources present very different versions of the piece, so that in Wolfgang Schmieder’s Bach-Werke-Verzeichnis we find the version of Q designated as number 769 while the version of A is 769a.  The issue of how Schmieder designated different versions of a work (and the consequences of such designations) deserves another post; some day I’ll come back to it.

Both versions contain five movements, but the sequence and presentation of those movements varies significantly.  (For convenience I will use the abbreviations devised by Walter Emery in his study of the work.)  In A the movements are:

• I.  C8    2-voice canon at the octave in the manuals, chorale cantus firmus in pedal
• II.  C5  2-voice canon at the fifth in the manauls, cantus firmus in pedal
• III.  CF  the cantus firmus in 2-voice inversion canons at the sixth, third, second and ninth, followed by the entire chorale in a stretto coda; the movement starts with three voices, adds an additional voice at the midpoint, and the coda adds two more voices.
• IV.  C7  2-voice canon at the seventh (pedal and left hand), cantus firmus + free cantabile line in the right hand
• V.  CA  2-voice augmentation canon at the octave in the manuals (+ free left hand line), chorale cantus firmus in pedal

A is notated throughout on three staves (conventional for organ music now, but not a default for Bach and his contemporaries).  In Q the movements appear in a different sequence, and are laid out differently:

• I.  C8  on two-stave puzzle notation (giving only the first few notes of the trailing canonic voice)
• II.  C5  ditto
• III.  C7 ditto.  These first three variations fit together on one “opening”—that is two printed pages, the verso of the cover page on the left, the recto of the next sheet on the right.
• IV. CA  in open score (4 staves), requiring two pages—the next “opening” of the print.
• V.  CF on three staves, requiring two pages, the final “opening.”

The BG edition gives only the version of Q (BWV 769), leaving discussion of BWV 769a to the critical commentary.  Largely due to the work of Friedrich Smend, in the first half of the twentieth century the scholarly consensus shifted, and A came to be regarded as the later source—and thus the Fassung letzter Hand [definitive version].  That it has never become the preferred version among performers is probably due to the apparent anticlimax:  the showy stretto which concludes BWV 769 comes at the end of only the third movement of BWV 769a.  (Not that either version is performed all that much!)  Gregory Butler, who has done extensive studies of Bach’s original prints, demonstrates that the first three variations of Q were engraved first (and very likely the only parts of Bach’s original conception of the piece), and that when CF was composed it could not be inserted between them without a considerable waste of labor.  As Bach’s conception of the work changed, he made a virtue of necessity (eventually adding CA), thus producing two very different versions of the same basic material.  By default Q is the more “public” version—that’s what published means, after all.  And it is no surprise that the version of Q was the more widely disseminated version.  We don’t know how many copies of Q were printed, but at least twenty survive now, and the Stemma devised by Hans Klotz for his edition in the NBA indicates many manuscript copies derived from the print.  See the tree on the left side—although, as my pencil scrawl indicates, “O” on the left is a very unfortunate typo—it surely must be Q; the “O” listed on the lower right is an entirely different source, transmitting BWV 769a.  (There are other problems with this Stemma not worth going into here.)

SOURCE:  detail of scan of Kritischer Bericht for NBA Ser. IV Bd. 2, ed. Hans Klotz (1957), p. 88.

It may seem odd to copy out printed works by hand, but in the centuries before photocopiers it was extremely common.  In this work it was particularly necessary, as the puzzle notation of the first three variations of Q would make them virtually unplayable without realizing them in notation.  Some of the manuscripts descending from Q (Klotz’s J4, E1, and B2, for example) transmit only movements that needed to be realized.  Thus rather than being bootleg manuscript copies of an out-of-print or otherwise inaccessible piece, these might have been used along side a copy of Q to play the complete work.

There are a plentitude of textual differences between Q and A, in and his comprehensive studies of this work Butler also convincingly argues that the very concept of a “definitive” version is meaningless in this piece.  Butler’s chronology is essentially this:

1. Initial conception:  C8, C5, and C7 composed and subsequently engraved.  Indeed, these seem at first to have been engraved without even indicating (beyond the signum S) the incipits of the canonic entries:

SOURCE:  cropped screenshots of Q as available at http://digital.staatsbibliothek-berlin.de/werkansicht?PPN=PPN614768373, pp. 4 and 5.

Especially in variations 2 and 3 the incipits have been crowded in with the utmost awkwardness.  Butler suggests that these might have intended to look more like the puzzle notation of Bach's other published canons, lacking any incipits at all, but that these were added as the concept of the work shifted.

2.  A new phase:  possibly inspired by his work on BWV 1087 (the fourteen extra canons based on the first eight notes of the bassline of the Goldberg Variations), Bach employed the chorale tune itself canonically, thus producing the series of 4 canons + stretto that make up CF.  At this point Bach started writing out a clean copy of the work—A—placing this new variation before the cantabile C7.  It is possible, indeed, that he considered the work complete after writing out C7, as he drew some final flourishes after the double bar at that point:

SOURCE:  cropped screenshot of A as available at http://digital.staatsbibliothek-berlin.de/werkansicht?PPN=PPN812577051, p. 106.  The three flourishes which I have encircled in red) appear only at the end of the fourth variation (C7).

If one were pursuing “definitive” readings, this would suggests that variants readings contained in A would supercede those in Q for C8, C5, C7, and CF, as A is the later source for these movements.  Certainly Bach was making revisions even as he was preparing A, a source which Butler notes combines the characteristics of a composing score, a clean copy, and a revised copy.  For example, in C7 bb. 6-7, Butler posits a revised reading that initially appeared in A, subsequently modified, but faintly visible now:

SOURCES:  composite cropped screenshots of Q and A (as above) with my transcription from Butler B-Jb 2000, p. 18.

3.  The final phase:  despite the apparent final flourishes that conclude C7 in A, Bach revisited the work to produce CA, the most complex of the canons.  While this could be put nowhere but the end in A, in Q it could be placed either at the end or in between C7 and CF.  Butler has demonstrated that CA was certainly the last to be engraved, and analysis of the variant readings supports the argument that Q is the later source for this movement.  Walter Emery had raised this possibility some fifty years ago (without the benefit of Butler’s meticulous study of the printing of Q):

As a performer, I had to commit myself to one version or the other.  (As Susan Hellauer once memorably expressed it, “You can't sing a footnote.”)  I opted in this recital for the version of A basically using the text as printed in the NBA, but with a few alterations imported from the critical report (and a fistfull of wrong notes scattered here and there, too).  Scholars don’t have to commit, though, and as Werner Breig puts it in his new edition of the work for Breitkopf,

What is an appropriate type of close?  Whether it is a contrapuntal concentration such as the six-part stretto or a canon in augmentation that points beyond itself, so to speak, through the unfinished canonic imitation in one part—Bach would certainly not have wanted to commit himself to any particular solution. (p. 20)

Moreover, as Bach’s habit of not commiting himself was so pervasive throughout his career—and Bach editions in the last three centuries have had to deal with some perplexing variants—with this post I am launching a

For the first post of each month from now through November 2017 (approximating the liturgical year that generated Bach’s own Jahrgänge of cantatas in his first years in Leipzig) I will deal with some textual issue relating to the works of J. S. Bach.  While the quantity of text-critical work that has been done on Bach cannot rival that done on Shakespeare (still less the Bible), it is a massive body of literature and editions.  Needless to say there is no shortage of things to talk about—and I hope that these posts won’t be as tedious as this one might have been.