Picture of BachAn Apology and Rationale for Completing J.S. Bach's Fantasia and Fugue, BWV 906

by T.L. Hubeart Jr.
Text and fugue completion © 1996 by T.L. Hubeart Jr.

Go to the Midi Files

Go to the Supplement to this Essay

One of the most celebrated of Johann Sebastian Bach's keyboard works is BWV 906, the Fantasia and Fugue in c minor. Or perhaps one should more precisely say that half of BWV 906, the Fantasia, is celebrated, since the Fugue unfortunately is incomplete. In this paper I endeavor to supply a rationale for attempting to remedy this deficiency in a very modest fashion.

The Evidence

First, a review of the evidence may be in order. The New Grove Bach Family, in its list of J.S. Bach's works, assigns this work a date of "Leipzig, c. 1738," but makes no other mention of it. {1} However, the fantasia has been often recorded and admired, with its homage to Scarlatti duly noted. {2}

That the fantasia was admired as far back as 1802 appears from Johann Nicolaus Forkel's biography of Bach published in that year. He provides an incipit of the fantasia and says,

". . . It is not of the same kind as the preceding, but, like the allegro in a sonata, divided into two parts, and must be performed throughout in the same movement and time. Otherwise it is excellent. In old copies we find a fugue annexed, which, however, cannot belong to it, and is not quite finished (Fig. 4 [gives incipit of BWV 906's fugue]). But it cannot be doubted that at least the first 30 measures are by Seb. Bach, for they contain an extremely bold attempt to make use of diminished and augmented intervals and their inversions in a three-part harmony. None but Bach ever hazarded an attempt like this. What follows the first 30 bars seems to have been added by another hand, for it bears no trace of Sebastian's manner." {3}

The editors of The Bach Reader, who reprint Forkel in full, give a footnote with the following necessary correction to his comments:

"Forkel's doubts are ungrounded--the entire fragment is an authentic composition by Bach and was meant as a sequel to the Fantasy." {4}

This information may be indebted to Philipp Spitta, whose magnificent biography of Bach dates the work at "about the year 1738" and points out that the composer

"has only left us the first forty-seven bars of the fugue. We may assume that he finished it, for the autograph is not a sketch, but a fair copy. He must have been prevented by some mischance from writing it out in its entirety. This is doubly to be lamented, since, to judge from the fragment that remains, the fugue must have been a work of especial boldness, and designed on a grand scale. The irregular proportions which are apparent between the fugue and the fantasia, which in many cases has the character of a prelude, are found in many preludes and fugues of this period. . . . " {5}

In a footnote, Spitta remarks on the rediscovery of the autograph of this work in 1876, and says that "Forkel and Griepenkerl [an early editor of Bach] have been unfortunate in judging of the work, for they consider that the fugue did not originally belong to the fantasia and that only the first twenty-nine or thirty bars are authentic. Both these opinions are proved false by the autograph." {6}

The Bach-Gesellschaft included both Fantasia and Fugue in its monumental edition of J.S. Bach's complete works published from 1851 to 1899. BWV 906 (not numbered as such, of course, since Schmeider's catalog of Bach was a 20th century invention) appeared in 1890, in the fourth volume of the edition's series Clavierwerke, and was edited by Ernst Naumann. Strikingly, the fugue appears in much smaller type than the fantasia, and separated from it, which apparently is meant to denote the expectation that only the latter would be performed, while the unfinished fugue was presumably included only for the sake of rendering the edition complete. {7}

The Daunting Task

I am somewhat astonished that no one, to the best of my knowledge, has attempted a completion of the unfinished fugue. Naturally, in a case like this, no one can hope to fill the shoes of a great composer. But most performers, except in special cases, are not in the habit of playing incomplete works; they know that audiences are unlikely to be pleased by the feeling of being left "hanging"! Thus there is a justification for getting an incomplete piece safely home, in order that the portion that the master finished can be appreciated by more than musicians and scholars. There is a long tradition of works which have undergone this procedure, from Mozart's Requiem to Puccini's Turandot to entire Schubert symphonies completed from sketches. Even some very mediocre works by great masters (e.g., some of the several Mozart keyboard fugues which that composer never bothered to complete) have been so served. {8}

But with Bach the case has been different, and one obvious reason has been this composer's surpassing greatness at fugue. It is one thing to put one's hand to an unfinished fugue by Mozart, whose strengths were elsewhere than this form; one is not so glaringly reminded of one's inadequacies as by the comparison with Bach. Still, this has not deterred the occasional intrepid Bach lover from completing the far more imposing final Contrapunctus of The Art of Fugue (for example, Sir Neville Marriner in his recording). I think we must look for another factor in the long neglect of BWV 906's fugue.

We observed Forkel's comments earlier, that the final bars of the fragment were "added by another hand" and carry "no trace of Sebastian's manner." He states that only the fugue's "first 30 measures are by Seb[astian] Bach," but Spitta's paraphrase of Forkel and Griepenkerl (cited above) saying that "only the first twenty-nine or thirty bars are authentic" leads to an important point. It would seem that the stylistic discontinuity alleged by Forkel actually begins at bar 30 rather than after it--or more precisely, between bars 29 and 30. For the continuous motion of sixteenth notes from bars 25 to 29 is stopped dead with a half-diminished seventh chord, announced bluntly by a solitary quarter-note f-sharp in the bass (first quarter), and confirmed by the upper parts' c and e-flat (second quarter). (See bars 28-30 of the fugue, below.)

Fugue, bars 28-30

I would encourage the reader to consult the entire fugue in score to get a precise picture of the situation here. However, enough has been said to call attention to the abrupt textural change which misled Forkel and Griepenkerl. There is a clear break at this point which should be taken into account. We get very far afield from the "diminished and augmented intervals and their inversions" which were employed by Bach at the outset. And it is perhaps the uncertainty as to what Bach would have done next (unlike the final Art of Fugue contrapunctus, where we at least know that he intended to combine four subjects) that has contributed to the relative neglect of this fugue.

A Possible Solution

However, although Bach's autograph breaks off before the point at which he would have recapitulated his opening material, I feel that it is virtually certain that he would have done so in a very conservative manner (i.e., in something very close to the initial statement) to provide coherence to the whole fugue. After the rather sharp stylistic break at bar 30, the opening needs to be recapped, with little change, in order to remind the listener of what the fugue is about.

Given the probable linkage in the 18th century mind between the fugue and the study of rhetoric {9}, it would not be going too far to draw an analogy between the present fugue and a formal debate. In a fugue, the "proposition," much like that in a debate, is stated formally, then refuted, then reasserted and confirmed--sometimes several times. (Of course, the fugue's opening material is always "vindicated" at the end, which is not always the case with a debate's proposition!) In Bach's BWV 906 fugue, as I read it, we seem to have the formal statement (bars 1-8), followed by some further "shoring up" of the "proposition" (bars 9-24), leading to an episode or "refutation" (25-29), after which the opening "thesis" seems smashed beyond recognition (30-31). Then, after some transitional material (32-33) and a seemingly new idea (34-35), the original motif is stealthily reintroduced and elaborated (36-48). One could, of course, follow this fragment with further refutations and reassertions of the theme; we have already noted Spitta's impression that the fugue must have been "designed on a grand scale." But since only someone of Bach's caliber could accomplish this adequately, and since he is not presently accessible, it seems best to end the fugue as expeditiously as possible. We must, then, proceed to the restatement of the opening material.

This is perhaps an appropriate point at which to summarize the fifteen bars of conclusion devised by me. I have supplied the following:

Bars 48-49: transition back to the tonic;

Bars 50-57: reuse of Bach's bars 1-8, substantially unaltered;

Bars 58-62: coda--theme taken to higher octave, full cadence, and Picardy third.

The success or failure of the attempt will be evident to the listener, and probably there are many who could complete the work in a more satisfactory manner than I have done. However, I would venture to say that any acceptable completion by anyone other than Bach himself would be virtually compelled to follow an outline similar to mine. Just as one would not want to subjoin further episodes to the fragment, which would tend to increase the importance of one's own contribution to the piece at the expense of Bach's, one should not end too perfunctorily--for example, by following the fragment with just enough material to get to a full cadence. The latter course would call attention to the completing hand's inability and incompetence, which, again, would detract from the genuine portion by Bach.

Regarding my reuse of bars 1-8 without alteration: I felt that, although it would have been nice to introduce some slight variation into the repetition, this desideratum was outweighed by the benefit of the reused music being from Bach's own hand. As Suessmayr's completion of Mozart's Requiem (admittedly a more substantial piece than this one) demonstrates, employing an earlier section to help conclude an unfinished piece is a respectable and effective procedure.

I hope the completion I have essayed, whether satisfactory or a spur to some abler hand to finish the fugue in a worthy manner, allows this piece to finally be heard in something closer to the form Bach intended. The fact that he followed his fantasia with this fugue shows that he intended the former to serve as a prelude to something, rather than being the free-floating work that its performance history has made it. And the boldness of the fugue, noted as early as Forkel, should earn it a reprieve from the obscurity of the complete works editions of Bach. If my completion brings these facts to light, I will consider it successful.

June 29, 1996
(Illustration added and footnotes 3 & 4 updated 6 Aug 2001.)


{1} Christoph Wolff, et al., The New Grove Bach Family, NY: Norton, 1983, p 208.

{2} E.g., in Philipp Spitta, Johann Sebastian Bach, trans. Clara Bell & J.A. Fuller-Maitland, Vol. 3, NY: Dover, 1951, p. 182.

{3} Johann Nicolaus Forkel, "On Johann Sebastian Bach's Life, Genius, and Works," trans. credited to A.C.F. Kollmann, in Hans T. David & Arthur Mendel, The Bach Reader, rev. ed., NY: Norton, 1966, pp. 342-3. [In The New Bach Reader, a revision of the David-Mendel volume by Christoph Wolff (Norton, 1998), this passage appears on p. 468.]

{4} Ibid., p. 343, note 49. [New Bach Reader, p. 468, note 37.]

{5} Spitta, op. cit., Vol. 3, pp. 182-3.

{6} Ibid, note 340.

{7} Bach-Gesellschaft vol. xxxvi, 145, 238. Dover Publications provides a helpful reprint of this work (and other selected portions of the Bach-Gesellschaft's edition) in J.S. Bach, Italian Concerto, Chromatic Fantasia and Fugue, and Other Works for Keyboard, NY: Dover, 1987 (BWV 906's fantasy and fugue appear, joined together, on pp. 72-6). A library search for a corresponding volume of the Neue Bach Ausgabe was unsuccessful; I can only assume that this is one of the volumes still in preparation in this new critical edition of Bach's works.

{8} Two of these, K. 153 and 154, are even condemned as "worthless except as documents" by Arthur Hutchings in The Mozart Companion (ed. H.C. Robbins Landon and Donald Mitchell, NY: Norton, 1956, p. 59). These were both completed, Hutchings informs us, by Viennese theorist Simon Sechter "long after Mozart's death," and are printed with Sechter's completions in the old edition of complete works of Mozart done by Breitkopf and Hartel in the last quarter of the 19th century. Some of the completion work is rather questionable: over half of K. 153 is not even by Mozart, but by Sechter (the piece has 66 bars, but Mozart stopped at bar 27)!

{9} See "Fugue" in Don Michael Randel, ed., New Harvard Dictionary of Music, Cambridge: Belknap Press of Harvard UP, 1986, pp. 327-9.


© 2002, 2008, 2011 by T.L. Hubeart Jr.

Given the additional evidence I have obtained in the years since I wrote the above essay, I decided that it was better to leave that piece to speak for itself, rather than dismembering it and attempting a revision. Instead, I will summarize the additional material I have since encountered here in order to provide an update.

I wrote in 1996 of being "somewhat astonished that no one, to the best of my knowledge, has attempted a completion of the unfinished fugue." My knowledge has now been broadened considerably, thanks in good measure to some helpful friends, and I now can identify no fewer than three such attempts:

1.      First, apparently, to the field was the Italian pianist and composer Ferruccio Busoni (1866-1924), who inserted an arrangement of Bach's Adagio in G (BWV 968) between an adaptation of the Fantasy and Busoni's own idiosyncratic completion of the fugue, entitling the result "Fantasie, Adagio, und Fuge." I don't personally find the completion very Bachian, though it is indeed interesting and impressive in its own way. Those wishing to hear it can find it in a fine performance by the pianist Andrea Padova on an album entitled "Busoni nach Bach, Piano Works 1908-1921" on the Stradivarius label, catalog number STR 33657.

2.      Jim Michmerhuizen called my attention to another completion of this fugue by Igor Kipnis, on a recording released on the Arabesque label, catalog number Z6577. Though this was unavailable for purchase at the time he alerted me, through the kindness of Bill Palmer, son of the late Dr. Willard A. Palmer, I was able to borrow a copy of this 1987 Kipnis recording that had belonged to Dr. Palmer. (It has since become available again through retailers such as Amazon.com.)  On track 6 of the album, Igor Kipnis plays both the Fantasia and Fugue that comprise BWV 906; to the incomplete fugue's break-off point, at bar 48, he joins a repeat of bars 3 through 32, concluding with one measure of his own to bring the piece to a close. I was pleased to find that the Kipnis completion anticipated mine in seeing the need to repeat the opening measures, although I am left uneasy by what I feel is an awkward join between the break-off point and the recapitulated material. This is a matter of taste, of course, and although it is no longer possible to ask the deceased harpsichordist his rationale (the CD's liner notes are by Dr. Klaus Hofmann, not by Kipnis, and simply give the reason as "[i]n order to provide a more effective conclusion..."), my impression is that Kipnis chose to err on the side of caution in this matter, rather than to supply additional bars of his own composition which he may have felt were unworthy of Bach.

3.      A third attempt at completing the fugue by the harpsichordist, conductor, and early music specialist Richard Egarr appeared on the 2003 Harmonia Mundi France release J.S. Bach Per cembalo solo… (catalog number HMU 907329). Mr. Egarr writes in his liner notes that with regard to completing Bach’s fugue, it seems to him “an impossibility to second-guess, recreate or pastiche such a master as Bach,” but “The material of this Fugue…seemed so pregnant and fascinating that I felt an unusual desire to respond to it creatively. I offer my completion, or (perhaps better) ‘reworking’ not as Bach, but as my own ‘Tombeau’ for this genius of the Fugue.” Egarr’s version freely rearranges and supplements Bach’s incomplete fugue, but although he explicitly does not present it as a completion, I thought it deserved to be mentioned here as he is one of the few aside from Busoni and Kipnis who have attempted to provide a musical whole out of Bach’s fragment. (By the way, Mr. Egarr’s album precedes his “reworking” with BWV 906 as we have it from Bach, including the incomplete fugue.)

Another resource graciously supplied to me by Bill Palmer was the Alfred Masterwork Edition of BWV 906's Fantasia, edited by Dr. Willard A. Palmer and published in 1971. Here, Dr. Palmer provides an impressive reproduction of his source for the edition, an autograph in the possession of the Bach Choir of Bethlehem, Pennsylvania bearing the title "Fantasia per il Cembalo di G.S. Bach." Apparently it caused Dr. Palmer a good deal of difficulty to locate this source material, for he said in a 1973 interview:

. . . Well, I couldn't find it [the autograph] where the Schmeider Bach-Werke-Verzeichnis said it was located. I looked for years, and when I finally found it, much to my amazement it was on the wall of the Bethlehem Bach Choir in Bethlehem, Pennsylvania. I wrote to them and asked if I could use a facsimile. They gave me permission and sent me a beautiful photograph of the autograph. I published the entire manuscript in my edition of the C Minor Fantasia. . . . [--As qtd. in "Conversation with Music Editors Dr. Willard A. Palmer and Mrs. Margery Halford," The Harpsichord, Vol. VI, No. 2 (May-July 1973), p. 7.]

However, nowhere in the edition published by Dr. Palmer is there any mention of the incomplete fugue. Since he stated that he published the "entire manuscript" to which the Bethlehem Bach Choir had given him access, the obvious conclusion is that the Choir's two-page manuscript did not include anything but the Fantasia. This conclusion is validated in the liner notes to the Kipnis CD, where Dr. Klaus Hofmann states that there are two autographs: the Choir's autograph of the Fantasia only, dating "from around 1729," and the autograph of the Fantasia and (incomplete) Fugue in the Sächsische Landesbibliothek in Dresden, dating from about 1736-7. That latter manuscript can now, through the kindness of the Sächsische Landesbibliothek, be viewed online. Plainly the existence of two autographs was not known to Schmeider, the compiler of the standard catalog of Bach's works. So when Dr. Palmer, following Schmeider's information, located an autograph for this work on the wall of the Bethlehem Bach Choir, he naturally deduced that it was the one Schmeider meant and used it as the source on which to base his edition.

Although we now know it is not the sole autograph, the Choir's manuscript does provoke an intriguing reassessment of what I said in my 1996 essay regarding BWV 906: that only "half of BWV 906, the Fantasia, is celebrated, since the Fugue unfortunately is incomplete." It was my opinion at the time that the segregation of the performed Fantasia and the unperformed Fugue by modern keyboardists postdated Bach, and was simply a consequence of the Fugue's incompleteness. However, the fact that the Choir's autograph contains only the Fantasia suggests that J.S. Bach himself considered performances of the Fantasia alone to be acceptable. Indeed, given the gap of about a decade between the two autographs, the Fantasia may have been conceived c. 1729 as a stand-alone piece, with the Fugue being started in the late 1730s as an afterthought. (You can make up your own mind as to whether you believe Bach never finished the fugue, or he finished it and--as Spitta maintained in the quote given in my essay--the completion was lost.)

Whatever the truth of the matter, the Fugue deserves to be heard and appreciated, and I hope that others will be encouraged to discover it.

(last revised 6 Feb 2011)

Hear the Results!

If you have a computer capable of playing General MIDI files (Type 1), you can access Bach's Fantasia and my completion of his Fugue by clicking open the following two files.

(Files are © 1996 by T. L. Hubeart Jr.; all rights reserved.)

Download Bach's Fantasy in c minor (MIDI format, 23kb).

Download Bach's Fugue as completed by me (MIDI format, 10kb).


Return Home