WTC II/22 in Bb minor - Prelude

II/22.1.1 The prelude-type

The Bb minor prelude is written in polyphonic three-part texture. Two motives pervade the piece - one substantial, appearing much like a "theme", the other short and versa- tile, including inversion and multiple sequential patterns. One can therefore describe this prelude as composed along the lines of a three-part invention.


II/22.1.2 The overall design of the prelude

There are several distinct cadential patterns which help in understanding the structure of this piece. The following six sections emerge:

I bars 1-16d tonic - dominant Bb minor to F minor
II bars 16-24d dominant - tonic F minor to Bb minor
III bars 24-42d tonic - dominant relative Bb minor to Ab major
IV bars 42-55d domin. relative - subdominant Ab major to Eb minor
V bars 55-70d subdominant - tonic Eb minor to Bb minor
VI bars 70-83 tonic confirmed Bb minor


Two substantial structural analogies determine the layout of this composition:

bars 1-16d recur in bars 55-70d
transposed a fifth down, upper and middle voices inverted;
(some variations)
(almost unvaried)
(variations only in bars 55/56)
bars 36-40m recur in bars 77-81m, each voice transposed and slightly varied



II/22.1.3 Practical considerations for performers

This prelude is conceived in rather lively basic character. Evidence for this interpretation can be found in the simple rhythmic pattern which consists predominantly of eighth-notes and quarter-notes, and in the broken-chord pattern which characterizes the main motive (see e.g. M: bars 1m-3m). The alla breve indication of the time signature further supports this decision. The tempo may thus be very fluent; the half-note beats should be perceived as swinging but unhurried.

Articulation contains a light legato touch in the eighth-notes and gentle non legato in the quarter-notes. Exceptions occur only in the keynote / leading-note / keynote formulas which must be legato; see L: bars 1-3, M: 8-10, U: 14-16 etc..

No ornament is indicated in the printed score. Nevertheless, it is a good idea to add one embellishment which would have been understood as implied by any performer of Bach's time. The dotted note on the middle beat of bar 82 represents, together with the anticipation of the resolving keynote (see upper voice: A-Bb-Bb) a typical closing formula which was conventionally decorated with a mordent. This mordent commences with the upper neighboring note Bb and contains two shakes, stopping short just before the seventh eighth-note of the bar.


II/22.1.4 What is happening in this prelude

a) The material


Introduced in the middle voice, the main motive extends over four and a half bars and contains two subphrases.


The first subphrase (M1a) commences with a gentle curve in eighth-notes and continues in quarter-note jumps and broken chords which represent a complete harmonic progression, ending on the downbeat of bar 3. The overall descent of this line is dynamically best represented in a continuous diminuendo.

The second subphrase (M1b) begins with an octave jump followed by what could be described as an ornamentation of the simple group Bb-C-Bb; this segment again is harmonized as a complete progression. The tension grows here until the appoggiatura on the last downbeat (bar 4d) and resolves only through the final half bar.

(ex. 42)

There are altogether eight complete statements of M1. In a few cases (marked below with an asterisk), the second subphrase seems extended. Harmonically, however, these extensions do not form part of the statements, and they are therefore not included in the table below. In two cases, the two subphrases of the motive are shared between different voices. (This division confirms, for performers who might have had doubts, the interpretation of the second quarter-note of the motive's third bar as the beginning of the second subphrase.)


bars 1- 5m M 5. bars 42-44d/44-46m L/M*


bars 8-12m U 6. bars 48-50d/50-52m M/U


bars 25-29m L* 7. bars 55-59m U


bars 31-35m U* 8. bars 62-66m M

In looking for recurring contrapuntal figures one finds that

M1a is accompanied by a keynote / leading-note / keynote formula in the first two and last two entries of the motive;
M1b is twice surrounded by two shortened, transposed and varied versions of M1a (see the variation of M: bars 1-3d in L: bars 3-5d and U: bars 3m-4m; compare also M+L: bars 57-59d with bars 55-57d).


differs considerably from the main motive of this piece. It is much shorter and first heard (see M: bars 7-8d) as a link which prepares the second entry of M1. Consisting of eight eighth-notes which move in scalar motion with a final "hook", it could be described as an extended upbeat leading to a short downbeat. Both metrically and harmonically, this motive clearly represents a gradual increase of tension.
M2 is almost ubiquitous in this prelude. It fills bars where M1 is absent, but also accompanies M1 in many of its entries. Furthermore, M2 is heard in inversion (see e.g. U: bars 24/25), with variations of either its initial (see e.g. M: bars 11/12) or its final intervals (see e.g. L: bars 16/17, 20/21). And while it does not occur in stretto or actual parallel, juxtapositions in contrary motion are used in several cases (see bars 24/25, 41/42, 54/55, 77/78).

One feature which should not be left unmentioned are the pedal notes in the final section. Bar 73 witnesses the beginning of an extended F in the bass which constitutes a four-bar long dominant pedal. A keynote / leading-note / keynote figure in the tenor (see bars 73/74) introduces the dominant six-four chord on the downbeat of bar 74 which then continues throughout bars 75/76, alternating with the dominant-seventh chord on the middle beats. In bar 77, the upper voice takes over the pedal note function with an extended Bb which is retained until bar 80d, before giving way to the melodic line which had been heard similarly at the end of the third harmonic section.

b) The structure

While this prelude's design is thus very clearly outlined, both by the two motives and the cadential closes mentioned earlier, there is another possible interpretation which should be mentioned. The prelude can be read as an - albeit somewhat irregular - three-part fugue, with M1 as the subject. Under this assumption, bars 1-42d would present the exposition, with subject entries in M, U, L and (redundant) U; bars 42-55d then provide the middle section, conveniently characterized by entries which are shared between two voices, and bars 55-83 embody the final section with two more entries in U and M. The striking correspondences between bars 1-16d and 55-70d on the one hand, bars 36-40m and 77-81m on the other support this concept of a ternary layout. The following sketch shows the prelude in Bb minor read as a fugue (ex. 43).



WTC II/22 in Bb minor -- Fugue

II/22.2.1 The subject

Commencing on the downbeat of a three-two bar, the subject of this fugue extends over four bars - including three rests - before closing on the downbeat of bar 5. The pitch pattern displays predominantly small intervals, interspersed only with two high-tension intervals (see the tritones Eb-A in bar 2 and Gb-C in bar 4). The rhythmic pattern contains three different note values - half-notes, quarter-notes and eighth-notes.

Of particular impact are the rests, all three of which are equal in duration (one quarter-note each) and fall on the first half of one of the weaker beats (see bars 1 and 2, beat 3; bar 4, beat 2). These interruptions in the melodic flow pose an important question regarding the structure of the subject: Do all or only some of them mark the end of a subphrase?

In the case of the very first rest, this seems easy to answer. The silence here interrupts an ascending line (see A: bars 1/2: Bb-C-Db-Eb). Complemented by a falling tritone and its resolution which returns to the initial note Bb, the ascent is revealed as the first half of a perfect curve which exerts a strong unifying force. The rest in bar 1 has thus clearly tension-sustaining function, while the one following the return to the keynote in bar 23 represents the "breathing" after the end of the first subphrase.
The next case, however, is ambiguous.
* On the one hand, the figure built by the three notes at the beginning of bar 4 depicts a releasing gesture and thus allows the consideration of another "breath".

On the other hand, bars 3/4 can be identified as a varied sequence of the first subphrase, preceded by one quarter-note upbeat and augmented so as to extend into the quarter-note on the next downbeat:

(ex. 44):


On the basis of this observation, the varied and slightly extended sequence should also be taken as a larger unit, and the rest in bar 4 therefore interpreted as another tension sustaining rest.

The harmonic background to the subject, although somewhat tedious to make out among the highly chromatic lines, supports this concept. As the final entry (see bars 96-100) shows particularly clearly, the two subphrases are harmonized in striking similarity. Of great importance, for the understanding of the rests and of the subject as a whole, is the fact that Bach seems conscientiously to avoid a perfect cadence at the end of the first subphrase but substitutes the expected tonic instead - if only by a major chord (ex. 45):

Combining these findings - on the one hand, the undoubted layout in two corresponding subphrases, and on the other hand, the equally valid observation of a relaxing gesture at the beginning of the fourth bar - into a dynamic interpretation leads to the following concept: Within the first subphrase, the tension increases during the four- note ascent, finds its climax on the downbeat of bar 2 in a vii7 chord, and diminishes during the return to the keynote. As the note before the tension-sustaining rest is simple and straight-forward in its direction towards the subsequent one, no interruption of the crescendo is desirable. Within the second subphrase, the tension rises similarly throughout the (ornamented) ascent until the Gb on the downbeat of bar 4 which is again harmonized by a vii7 chord. The rest, which occurs here not during the ascent but after what is perceived as the climax, is preceded by two ornamental notes. These present a small relaxation - as it were on a secondary level - before the Gb is picked up and a proper decline through falling tritone and (ornamented) resolution is launched.

In the competition between the two subphrase-climaxes, the second one clearly wins.

Structurally, the Gb is the peak of the slightly longer and more complex subphrase, and lies in an ascending sequence.
Melodically, it represents the sixth degree which, in all minor scales, contains particular tension as a semitone above the fifth.



II/22.2.2 The statements of the subject

The entries of the subject in this fugue follow an extremely regular pattern. There are altogether twenty-four statements; only the two final ones are slightly shortened (see the asterisks).

1. bars 1- 5d A
bars 67-71d Tinv


bars 5- 9d S 14. bars 67-71m Sinv


bars 11-15d B 15. bars 73-77d Ainv


bars 17-21d T 16. bars 73-77m Binv


bars 27-31d T 17. bars 80-84d Sinv
6. bars 27-31m A 18. bars 80-84m T
7. bars 33-37d S 19. bars 89-93d B
8. bars 33-37m B 20. bars 89-93m Ainv
9. bars 42-46d Tinv 21. bars 96-100d S
10. bars 46-50d Ainv 22. bars 96-100d A
11. bars 52-56d Sinv 23. bars 96-99m Tinv*
12. bars 58-62d Binv 24. bars 96-99m


(ex. 46)

The subject's real answer is either identical with the original (see T: bars 17-21) or differs only in its major-mode ending (see e.g. S: bars 5-9). Other modifications concern the end of the phrase which is twice shortened (see the two final entries) and three times extended into a weak-beat ("female") ending (see S: bar 9, B: bar 15, S: bar 56).

Exactly half of the statements are inverted, and the focus on a symmetrical design is even further emphasized with regard to strettos and parallels:

eight statements appear as separate entries
(four of them are in original, four in inversion),

eight statements appear in plain strettos
which involve either only original or only inverted entries
(four of them are in original, four in inversion),

four statements appear in mixed strettos
which involve one original and one inverted entry each
(two of them are in original, two in inversion)

four statements appear in a combined parallel/stretto
(two of them are in original, two in inversion)



II/22.2.3 The counter-subjects

Despite the extraordinary emphasis on the subject itself, Bach invents one regular counter-subject as well as - in lieu of a second contrapuntal complement - a rudimentary figure.


is introduced in A: bars 5-9m. Its characteristic traits are an interrupted but straight chromatic ascent (see bar 5: A-Bb-B-C, bar 6: C-Db-D-Eb-E-F) and a melodic closing figure sequenced after a rest.
The structural correspondences with the subject are subtle but striking in their ingenuity. Like the subject, CS1 consists of two distinct segments each of which involves a rest in its middle. And also like the subject, CS1 features sequences. The differences are that (a) in CS1, the sequences occur within each half while in the subject, they take place between the two subphrases; (b) in CS1, the subphrases are contrasting in material and not separated by a rest but joined by an extended tied note. (Had Bach wanted to separate two subphrases, one can imagine that he could have ended the first subphrase in CS1 after the first half-note in bar 7 - in analogy to the half-note at the end of the initial chromatic segment in bar 5 - and commenced the second one on the tied final quarter-note of bar 7 - in analogy to the tied quarter-note which launches the second melodic closing figure at the end of bar 8.)
The phrase structure and its dynamic equivalent are thus quite different from that depicted in the subject. One finds a long, albeit interrupted, ascent in increasing tension, answered, after a protracted suspension, by a resolving step downwards which is ornamented with a written-out inverted mordent. Then, as if this relaxation did not quite do justice to the extended build-up, the soothing figure is repeated in descending sequence. Two interesting observations meet the eye: (a) this two-fold relaxation contains a chromatic descent (F-E-Eb-D) which partially complements the preceding ascent; (b) the independence of CS1 from the subject is carried so far that the counter-subject regularly closes two or three quarter-notes after the end of the subject (see e.g. bars 9, 15).
CS1 faithfully accompanies each of the single uninverted subject statements. Its inversion, which remains incomplete, breaking off after the long note, supports each of the inverted entries (see A: bars 42-45, T: bars 46-49, A: bars 52-54, S: bars 59-61); it makes a further short appearance in connection with the first mixed stretto (see A: bars 82/83).


is a very unusual companion. Far shorter than the subject which it undertakes to accompany, it only materializes during the build-up to the subject's second (and main) climax. But as if its very restricted scope was not strange enough, its pitch pattern is also entirely vague. The only consistent features are its rhythm and articulation: Three quarter-notes, introduced with wedges and followed each by a quarter-note rest, lead to a final note of varying duration. More often than not, this note group is preceded by an upbeat.
Nondescript as this may seem, CS2 is nevertheless a loyal companion. After its first presentation (see S: bars 12-14), it recurs eight times, accompanying original and inverted entries, single statements and strettos alike.

The following sketch depicts the interplay of phrase structure and dynamic design in the subject and its two companions (ex. 47):



II/22.2.4 The episodes

The fugue encompasses thirteen subject-free passages. Due to the fact that the endings of subject and counter-subject do not coincide and that tight strettos, involving only slightly-shifted entries and beginnings of subject statements, are very frequent, there are several cases where primary and secondary material overlap.*


bars 9m-11d E8 bars 62 -67d
E2 bars 15m-17d E9 bars 71m-73d
E3 bars 21 -27d E10 bars 77m-80d
E4 bars 31 -33d E11 bars 84m-89d
E5 bars 37 -42d E12 bars 93m-96d
E6 bars 50 -52d E13 bars 100/101

bars 56 -58d


* The most consequential cases in which episode material overlaps with subject statements are:

ending on the third quarter-note: E1 bar 11: S  
beginning on the second quarter-note: E5 bar 37: A E9 bar 71: B
  E10 bar 77: S+A E1 bar 84: S

The material employed in the episodes includes partial sequences from subject and CS1 as well as independent episode motives. Apart from E13 at the very end of the composition, no episode is conceived as a cadential close.


describes all motives derived from the subject. In the course of the fugue, these occur in various lengths.

In E1,

the motive derives from the extended ending of the subject (see S: bars 8/9 Db-F, sequenced in free variation of the pitch pattern in E1 bars 9/10 Bb-Eb and bars 10/11 A-Bb).
In E2,

sequences are again free in pitch outline but the subject's extension is not included in the sequence (see B: bars 15/16 Eb-C, 16 Ab-Ab).

In E3,

similar sequences appear in an imitative pattern of the middle voices. Each sequence is extended through a whole bar during which it displays the rhythmic pattern of CS2 (see T: bars 21m-23d and 23m-25d, A: bars 22m-24d and 24m-25 - in this latter case, the CS2-pattern is even anticipated; see A: bars 21/22).

In E4,

both entries involved in the preceding stretto are sequenced freely (see T: bars 31 Bb-F, 31/32 Eb-Eb, 32/33 Db-Db, and A: bars 31/32 Bb-Bb, 32 Ab-Ab).

In E5,

the bass beginning can be read as a sequence of an extended subject ending (see B: bars 36/37 Ab-F sequenced in bars 37/38 Db-Bb).

In E7,

only the final four notes of the extended subject ending are taken as a model for both sequences and imitations (see S: bars 55/56 Bb-Eb sequenced in bars 56/57 Db-Gb, and imitated in B: bar 56 Cb-Fb with sequence in bars 56/57 Eb-Ab).
In E10,

a first and very faithful imitation of the subject ending (see S: bars 77/78 G-C) is followed by a tight pattern of condensed imitations which, had they not occurred right after the subject entry, might be hardly recognizable in their three-note shape (see A+T bars 77/78, S+B bar 78, A+T bar 78, S+B bars 78/79, A+T bar 79, S+B bar 79, T bars 79/80).


presents, after a straight-forward imitation (see B: bars 84/85 Eb-C), sequences and imitations of the subject's ending and its assumed two- note extension (see S: bars 83m-84m G>-Gb sequenced in bars 84-86d Gb-Ab, bars 86/87 Ab-Bb and, partially, in bars 87/88 A-Eb; and imitated in T: bars 85/86 Gb-D, bars 86/87 Ab-C, bars 87-89d Bb-F).


is similarly conceived as a partial sequence of the ending of CS1.

In E1/E2

this occurs quite simply (see A: bars 9/10 Db-C, bars 15/16 Gb-F).

In E3,

the motive emerges as an imitation of CS1's final inverted-turn figure, extended with a tie-prolongation (see S: bars 21-23d and, in a varied sequence, bars 23-25d).


stands for episode material derived from the chromatic segment of CS1.


presents a version in which the CS1 ending with syncopation and inverted-mordent figure is preceded by two chromatic steps (see A: bars 37-38d, S: bars 38-39d, T: bars 39-40d, A: bars 40-41d, S: bars 41-42d). Interestingly, this motive anticipates the first occurrence of the CS1 inversion in bars 42-45.

In E9,

the chromatic ascent alone is recalled (see B: bars 71-72d and, with a varied beginning, bars 71-73d).

Finally, there are four motives which confess no immediate connection (neither sequential nor imitative) to a preceding segment of primary material. Each is limited to a single episode. Two of them are hybrid motives which combine traits from different primary sources.

Mcs2 in E3 (see B: bars 21/22 G-Bb, bars 22/23, 23/24 and, in a non-chromatic version which retains only the rhythm, in bars 24/25) portrays a blend of the two counter-subjects; CS2 provides the rhythmic feature of quarter-notes interspersed by quarter-note rests, while CS1 contributes the chromatic ascent.


in E12

recalls a longer segment, the two-fold syncopation (although not interrupted here for a chromatic step) with two inverted-mordent figures from the CS1 ending. Preceded by an upbeat, this version appears in an imitative pattern (see S/A: bars 93-95).


in E6 is introduced in the alto (see A: bars 50/51 Ab-Db) and imitated in the tenor (see T: bars 51/52 C-F).


in E8 is the distinguishing feature of this episode. Although it contains a syncopation with subsequent inverted-mordent figure which is clearly reminiscent of CS1, it is otherwise fairly independent, both in its melodic line and insofar as it is the only episode component which involves a contrapuntal feature. An imitative pattern between alto and soprano (see A: bars 62/63 Eb-F, S: bars 63/64 Bb-Bb) is accompanied by a two-bar bass line (M2a see B: bars 62-64d). The complete pattern is then sequencedone step higher (see bars 64-66d) and rounded off by an additional quotation of M2 in the soprano (see bars 66-67d).

As it turns out, the final portion of E3 (see bars 25-27d) is the only episode segment that is not distinguished by very particular motivic features.


II/22.2.5 Character, tempo, articulation, ornament realization

This fugue represents basically calm character. Although the time signature indicates that each half-note equals an undivided beat, these beats should be stately and composed. The relative tempo of the prelude to the fugue is a complex one; it can be attained in either of two different mental processes:


two bars
correspond with
one bar
in the prelude

in the fugue


an assumed triplet quarter-note
an eighth-note

in the prelude

in the fugue

(Approximate metronome settings: half-notes in the prelude = 88, half-notes in the fugue = 66.)

The basic articulation in this fugue is legato. Several important exceptions do, however, occur:

In the subject's first subphrase, Bach has marked the unornamented ascent with wedges indicating a purposeful abbreviation of each note. Although these markings are not consistently repeated throughout the composition, the same approach must be retained in all further entries of the subject.
CS2 equally carries wedges on three of its few notes. These seem particularly telling since they precede rests which would have interrupted the sound flow in any case. This shows that it is not so much the interruption but the forceful attack of the notes which is requested. As in the subject, this articulation of CS2 must be regarded as an integral characteristic and transferred to all further statements. Two questions, however, remain at the individual interpreter's discretion: should the wedged approach also be undertaken for the occasional parallel in equal rhythm (see A: bar 19, B: bar 44, T: bar 60), and should the same articulation be retained in the CS2-derived episode motive? (The latter is more unlikely since there are several examples in other fugues where Bach repeats an indication if he desires it in episode material.)
Non legato is also appropriate for consecutive jumps (as e.g. in S+T bars 25/26, B: bars 39-41, 62-67) and for unexpected extensions to the subject (see S: bars 9, 56 and 84; B: bar 15) and their sequences or imitations.

All long notes in cadential-bass patterns - whether occurring in an episode or accompanying a subject entry - must be played detached (see e.g. bars 45/46, 52/53).

The score gives only one ornament symbol for this fugue, the cadential mordent in bar 100. Approached stepwise, it commences on the main note, shakes twice (in four sixteenth-notes) and ends on a eighth-note A natural which precedes the eighth-note Bb.


II/22.2.6 The design of the fugue

The design of this fugue is so ingeniously laid out that it requires no sophisticated analysis. There are five sections, each encompassing four subject statements (one in each voice) as well as one or more intermittent episodes and a closing one, and a coda with another set of four subject statements and the final episode.

Section I contains four single entries in original shape (A S B T). The two episodes which link the subsequent entries (E1, E2) are both short and entirely based on sequential prolongation. The closing episode E3, by contrast, presents in its first four bars slightly more developed material which is then topped off with two non-motivic bars.
Harmonically, the four entries are in the home key (alternating tonic and dominant), and even the final episode does not leave Bb minor.

Section II

begins in three-part texture and only regains the full ensemble after the intermittent episode. It contains two strettos; these combine first the inner voices (T+A), then the outer ones (S+B). Corresponding with the design of section I, the linking episode (E4) is short and made up of sequencing material while the closing passage is five bars long and presents more complex motives.
This section begins in Bb minor, but the second stretto and most of the closing episode (until bar 41d) are in the relative major key (Db major). The final bar brings the return to the home key.

Section III

introduces the inversion of the subject with four single statements. As in section I, the first two subject entries follow one another without interruption, while the other two are each preceded by two-bar long episodes (E6, E7). With regard to material, however, these bridging episodes appear much more independent than their earlier counter- parts. The closing episode (E8) is again five bars long.
Harmonically, these four statements modulate from the tonic to the subdominant (Eb minor). Due to the inversion of the subject and its main counter-part, entries here do not even necessarily close in the key in which they had set off (compare e.g. bar 52 with bar 56 and bar 58 with bar 62). Yet, as in the preceding section, the closing episode returns once more to Bb minor.

Section IV

corresponds with section II insofar as it consists of two strettos, a short bridge in-between and a closing episode; it follows up on section III in that it uses exclusively the inversion of the subject. Harmonically, the two strettos refer to the tonic and dominant key respectively.

Section V

combines in each of its two strettos one straight and one inverted subject statement. Despite the unexpected length, the inner episode develops as a sequential prolongation from the preceding entry, while the closing episode (this time shorter) displays more independent material.
Harmonically, the second of the two strettos marks the final return to the tonic. The closing episode with its descending soprano line, its repeatedly sustained dominant note F in the lower voices and its final dominant-seventh chord, clearly prepares a confirming cadence in the home key. The expected resolution on the downbeat of bar 96, however, turns this close into a deceptive (interrupted) cadence. This provides a truly convincing transition to the short but extremely powerful coda of the piece.

The coda

surpasses everything that has been heard so far by combining the mixed strettos of section V into a single four-part block of two parallel original entries and two parallel inverted statements. The latter are slightly shortened, giving way to a final cadential formula which closes this extraordinary fugue.

There is a stunning analogy between this layout and that of the D# minor fugue from the first volume of the Well-Tempered Clavier (compare chapter 8 of this book).

For a sketch showing the design of the fugue in Bb minor, see ex. 48.


II/22.2.7 The overall dynamic outline of the fugue

The development of tension in this fugue presents a long, extremely powerful increase launched in two consecutive thrusts. Section II surpasses section I just like section IV exceeds section III. Within the first pair of sections, all episodes are decreasing. With- in the second pair, by contrast, the intermittent episodes express no clear dynamic intention (E6, E7, E9) while the closing passages either embark on a relaxation which is later converted into a last-minute increase (E8 ), or produce a straight-forward crescendo.

It is this difference in the dynamic impact of the episodes which guarantees a progressive motion in sections III/IV despite the structural correspondence with sections I/II. Section V therefore begins on a dynamic level which is higher than that of the preceding sections. Its intermittent episode follows up on this development by presenting two bars of ascending sequences which also profess an increasing gesture. Only then does the tension abate somewhat towards the second mixed stretto; the fact that this marks the return to the home key enhances the effect of relief that is felt on this very high level of tension.

After a harmonic as well as dynamic release, it would be natural to expect a further completion of the curve in a resolving cadence. (Interpreters should uphold this "promise" as long as possible, and not give away the pending interrupted cadence too early - this would considerably impoverish the glorious conclusion of this fugue.) The double stretto of the coda adds a structural surprise to the harmonic one, by topping a process which seemed already to have passed its peak with a final climax.