I/8.1.1 The prelude-type

The E^b minor prelude is mainly determined by melodic processes which unfold around an essentially homophonic accompaniment. The two predominant melodic ideas are introduced in the treble line of bars 1-4 and 20/21 respectively (the former begins in the second half of bar 1 and ends on the E^b in bar 4; the latter stretches from the high C^b in bar 20 to the downbeat of bar 21). While subsequent developments of these melodic ideas show a large degree of variation in the treatment of pitch progressions, rhythmic patterns remain very consistent.

I/8.1.2 The design of the prelude

The first harmonic progression concludes in bar 4 (at the end of beat 1); it is a simple cadence with the subdominant in bar 2 and vii⁷ (replacing V) in bar 3. As this cadential close coincides with the end of the melodic entity, it is certainly of structural importance insofar as it emphasizes phrasing. However, as the development which follows is composed as an immediate continuation, bringing forth its own structure as well as appearing in obvious contrast to the second motive, this phrase ending only marks a smaller breathing within a larger unit.

The subsequent harmonic development embarks on a progression of secondary dominants resolving each into their related tonics. Thus, the listener repeatedly experiences resolutions without truly arriving at any new key.

bar	5	6	7	8	9	10	11	12
	V7/vi --	vi	V7/iv --	iv	V9 -----	i	V9/v ---	v

The following twofold cadence in B^b minor (see bars 13-14m and 14-16) then confirms a modulation to the minor dominant; yet again, this new key is not established with a truly convincing resolution. In bar 14 the new tonic is reached in metrically unaccented position on the second half-note, and in bar 16 the melodic ascent after the downbeat also creates a sense of openness. The subsequent bars then modulate to the subdominant (A^b minor) which is established with a iv-V-i progression in bars 19/20. The point of completion of this cadence marks the end of the first larger section of this prelude.

There are altogether four structural sections in the E^b minor prelude:

The fifth section shows three subdivisions – not only on the level of melodic development but also on that of harmonic progressions – insofar as there are two deceptive cadences prior to the final perfect ending:

-

In bars 25-29, the Neapolitan sixth chord (bar 26) and the vii7 (bar 27) seem to prepare a definite ending. This expectation is deceived on the downbeat of bar 29 by a chord which combines features of the traditional "interrupted-cadence" chord VI (the notes of which are Cb Eb Gb) with the Ab of step iv (reached by way of an appoggiatura Bb ); the i6 of Eb major which follows in bar 30 is reached plagally from iv6 and thus does not conclude this cadential progression either.

-

In a similar process in bars 31-37, the subdominant chord (bar 31d) is redefined on beat 2 as another Neapolitan sixth chord, the vii7 is extended through four bars (bars 32-35), and the preparation of the ending is almost exactly the same as before (compare second half bar 36 with second half bar 28). Again the expectation is deceived, this time because the resolution chord Eb appears topped by its minor seventh – which, as a V7/iv, initiates another iv-vii7-i round which, above a four-bar tonic pedal, finally leads to the resolution into the Picardy-third tonic.

While melodic processes in this prelude reveal several immediate sequences, there are no obvious correspondences on the structural level.

I/8.1.3 Practical considerations for performers

The appropriate tempo in this prelude is determined by rhythm and meter. On the one hand, the sixteenth-notes have to be calm enough to retain their melodic expressivity and at the same time to allow for thirty-second-notes in the trills. On the other hand, the alla breve time signature demands a tempo fast enough to convey a pulse in half-notes. This is very important, particularly as the somewhat complex rhythmic structure in the piece may tempt performers into a transitional stage of counting in quarter-notes. While this may be essential during the process of preparation, it is equally vital that this stage be finally overcome in favor of the larger swinging pulse with only three beats to a bar.

The articulation of this prelude requires legato for all notes pertaining to the melodic processes; this can be achieved, as in other Baroque keyboard pieces, through the fingers alone and thus poses no problems. The accompanying chords also sound best if smoothly linked. Yet connection between chords – particularly if they involve so many repetitions – is not possible on the piano without the help of the sustaining pedal. (The use of the pedal is a very delicate matter and may appear somewhat complicated where sixteenth-note motion in one of the lines requires a clean and unblurred rendition. The only solution is thus to pedal on or slightly after the last rhythmic value in a bar. As every performer will aim at evenness in tone color throughout each bar and each phrase, this means that one can always pedal only on or slightly after the last sixteenth-note fraction of each half-note. The only stretches which might do without any pedal are the beginning of the second half (bars 20/21) and the cadenza-like bar 35.)

The ornaments in the E^b minor prelude fall into three categories: trills, arpeggios and a grace-note. The grace-note (bar 36) poses the easiest task. According to the Baroque rule of "equal distribution between appoggiatura and resolution in binary note values", the A^b (played together with the notes in the other voices) is held for one quarter-note, leaving the second quarter-note for the resolution note G^b.

Although the trills are indicated by three kinds of symbols – simple mordent (as in bar 3), extended mordent (as in bar 4) and tr (as in bar 15) – their main difference lies not in this indicator but in their context. On the one hand, if the extended mordent in bar 4 is interpreted as a note-filling trill, consistency requires one to play an ornament of similar duration in bars 8, 10 and 12 where only the simple mordent sign appears. On the other hand, the tr in bar 15 ornaments a dotted-note figure and must therefore be executed as a "point d'arrêt trill". This apparent confusion can be solved easily by answering a simple set of questions:

(a)	Does the ornamented note have a tendency to resolve? In this case the trill may fill the entire value with a suffix immediately before the resolution if this resolution should, however, occur on a strong beat.
(b)	Does the resolution occur on an unaccented fraction of a beat, or is it preceded by written-out notes? In this case the ornament has to stop short slightly before the next written-out note.
(c)	Does the ornamented note belong to the same harmony as the subsequent pitch? In this case an ornament does not fill the entire note value, and a trill, even if relatively long, ends on the main note without a suffix.

As we know that Bach often wrote simple mordent symbols for any length of trill including suffix-ending long trills, the above takes care of all ornaments in this prelude. A further detail to be considered is the note with which the ornament begins. As always in ornaments of this era, a beginning from the main note is required only if the ornament is approached in stepwise motion; in all other cases, the ornament commences from the upper auxiliary.

The arpeggios all begin with the lowest note on the beat. They sound best and most even if the rhythmic distance between the lowest and the highest note remains the same (this is better – though perhaps more difficult – than retaining the rhythmic distance between any two consecutive notes); as a result, the speed in the five-note arpeggio in bar 6 should be considerably higher than that in the three-note arpeggios in bar 8.

The more vital question is whether all notes of the broken chords belong to the level of accompaniment, or whether either the top or bottom notes of any arpeggio actually from part of the melodic (horizontal) progression. Whenever the latter is the case, the performer must consider both the touch coloring (melodious vs. neutral) and, in the event of a treble line, the metric placement of the melodic note. The melodic context would sound awkwardly distorted if this note appeared "offbeat" – as it would if played as the end of an arpeggio.

One creative way of finding out which rendition is more appropriate to any given context is to imagine the entire texture of this prelude rewritten for an ensemble of instruments – e.g. the oboe (for melodic lines in the higher register), the bassoon (for melodic lines in the lower register), and the harp (for the accompanying chords). This mind image will, at the same time, greatly improve the sound-color pattern of the keyboard performer.

Questions like the following should then be asked: Who plays

-	the bass-clef Bb on beat 1 of bar 5 as opposed to that on beat 2 of this bar, and the treble-clef Eb on beat 1 of bar 6 as opposed to that on beat 2?
-	the treble-clef Cb - Bb - Ab on beat 1 of bar 8 as opposed to the notes on beat 2 of this bar, and the treble-clef Gb -F- Eb on beat 1 of bar 4 as opposed to the high Gb on beat 2 of this bar?
-	the treble-clef F on beat 2 of bar 12 and the Gb on beat 1 of bar 13; also the treble-clef Gb on beat 1 of bar 25 and the Fb on beat 1 of bar 26?
-	the treble-clef Eb -D in bar 28? (This bar might allow for two equally meaningful solutions.)

A second question worth considering carefully is whether those accompanying chords which follow exactly the same pattern as the others but appear without an arpeggio sign should not also be equally treated. This applies to

-	the left-hand chords in bars 29/30;
-	the two-hand chord in bar 31 beat 1 (with exception of the Cb which continues the "oboe" line);
-	the right-hand chords in bars 32-34; possibly the chords on beats 2 and 3 of bar 36 (here again, a distinction has to be made between melodic top notes which are on the beat, together with the root of the arpeggio);
-	the left-hand double notes in bars 38/39.

I/8.1.4 What is happening in this prelude?

As was already mentioned, the prelude contains two relevant motives, both of which undergo considerable changes in the course of the piece. Furthermore, both motives are determined much more by their rhythmic and metric shapes than by their pitch patterns or the number of notes they encompass.

M1

rests on a figure which could be described as a "question + answer" phrase with two similar halves, each consisting of an upbeat to beat 3, a dotted-note group, and a downbeat. The upbeat can vary in length, the dotted-note group can be ornamented, and the downbeat may have a "female" tail: (ex. 16)   M1 is introduced at the beginning of the prelude, in an extended version which stretches altogether over four bars, from the eighth-note Bb in the middle of bar 1 to the unaccented sixteenth-note Eb in bar 4. In this version with three subphrases, several details which are to characterize the motive later appear slightly obscured. Both the structural design of "question + answer" with two corresponding halves and the texture which enhances this design appear more clearly in bars 4-6d, and the fact that the Bb in bar 1d is stemmed upward and thus seems to belong to the motive also obscures the shape of the melodic unit. Later statements leave no doubt that the motive regularly begins as described above. Dynamically, the two answering subphrases of M1 describe a tension-curve with the main climax on the downbeat which marks the end of the first subphrase, and preferably no heavy accent within the second subphrase. The character of this motive is graceful and mild. This can be heard particularly well after the completion of the first perfect cadence (see bars 4-6, 6-8, 8-10 and 10-12): each time, the "question" sounds in the lower melodic voice, leading harmonically to a dominant-seventh chord, while the "answer", given in the upper melodic voice, brings the respective resolution. Bars 12-14 continue with a free development in the upper voice, followed by a varied version of M1 in what is written as a middle voice (bars 14-16). A last statement, presented in the lower voice bar 16, is extended and merges into the cadential-bass pattern which ends this section of the prelude. In bars 22-25, another variation of M1 appears, with two "questions" in the lower voice merging directly into an upper voice "answer" which shows signs of being infected by the rhythm of M2. Finally, towards the end of the prelude, M1 recurs once more. Stripped of its "question + answer" design it appears here in chain and under different harmonic circumstances (see U: bars 30/31, U+M in parallels bars 31/32, L: bars 32-35 on an extended vii7 chord, U: bars 37-39, varied).

M2

is introduced in bars 20/21. Only one bar long, it shares with M1 the rhythmic feature of the dotted-quarter-note group, the beginning with a eighth-note-upbeat inside the bar, and the end on a downbeat. Its prevalent rhythmic patterns are (ex. 17): M2 is first heard exactly in the middle of the prelude. Its initial statement in the upper voice (bars 20/21) is followed after one beat in stretto imitation by the lower voice. The same imitative pattern is then sequenced (see bars 21/22). These two bars in the middle of the prelude thus stand out for various reasons: apart from marking the beginning of the second half (after the definite cadence in bars 19/20) and introducing the second motive, they are also the only bars in the entire piece to come without any accompanying notes or chords. The tension expressed in M2 is different mainly insofar as the entire figure rests on only one harmonic step. The climax therefore falls on the beginning, i.e. on beat 2 in the leading voice, on beat 3 in the imitation. The character of this motive might be described as stately. This second motive is developed once, in bars 26-28, where it appears in a single-voice version without imitation. Because of the harmonic change at the end of the first of these statements (see downbeat bar 27) and because of the inverted pitch pattern in the second, the dynamic curve would be most convincing if it were also inverted, with each time a crescendo to the final note of the motive. A final statement, only recognizable through its basic rhythmic structure, could perhaps be detected in the middle voice of bars 39/40.

As for the dynamic design in the entire prelude, an overall development of tension with large-scale buildups does not seem the aim in this very meditative piece.

I/8.2.1 The subject

With a little more than two and a half bars, the phrase length of this subject is not metrically oriented. The keynote D^# which falls on the middle beat of bar 3 does not only act as a harmonic resolution of the dominant-seventh chord represented by E^# and G^# at the beginning of bar 3, but also marks the melodic return to the pitch from which the phrase departed on the downbeat of the initial bar. While intuition tells us that this subject is made up of two subphrases, the exact point of phrasing is not quite so certain. Looking at the original statement of the subject alone, there could be two equally valid solutions:

(a)	One option is to regard the jump D#-G# (bar 2m) as a varied sequence of the initial D#-A# at the subject's beginning. This makes sense rhythmically and metrically: both D#s are quarter-notes falling on strong beats, and the target notes of both jumps are syncopations of three-eighth-note length.
(b)	Another option is to view the complete curve from the initial D# to the D# on the middle beat of bar 2 as an entity and thus as the "main thought", with the scale descent from G# onwards as an "afterthought".

Most people have immediate and fairly strong feelings as to which version they prefer. However, as it turns out in the course of the fugue, the choice is not really ours for there are two statements which reveal what phrasing Bach himself had in mind. The middle-voice entry in bars 19-22 and the upper-voice entry in bars 20-22 both feature a rest within the phrase. This rest occurs after the end of the curve; it cuts the assumed sequential jump in two and thus gives undeniable evidence for the phrasing into "main thought" and "afterthought", the second option described above.

The pitch pattern in the subject contains two fifth intervals, at both ends of the main subphrase; all else is stepwise motion. (The succession D^#-G^# is not counted as an interval since the two notes do not belong to the same phrase.) Rhythmic features in the subject include eighth-notes, quarter-notes and syncopated dotted (or tied) quarter-notes.

The harmonic background to this subject is that of a simple cadence with an extended first chord. The complete curve of the first subphrase is firmly rooted in the tonic, but interspersed with dominant chords which quickly resolve back into D^# minor. The syncopation at the beginning of the second subphrase represents the subdominant, followed by a gradual return to the home chord.

(ex. 18)

 

The shaping of tension within the subject leaves no room for doubt in the second subphrase where a simple decrease after the syncopation is certainly the only logical answer. In the first subphrase, however, it allows for two slightly different interpretations, depending on the chosen tempo and on individual preference for rhythmic vs. melodic processes.

-	Performers who feel the rhythmic feature stronger than the melodic one, will place the main climax on the dotted A#.
-	Performers who feel the melodic feature strongly enough, may wish to play a tension-rise past this A# to the B – which is the (high-tension) minor sixth degree – thus giving this subject a very special emotional quality.

I/8.2.2 The statements of the subject

The D^# minor fugue contains thirty-five subject entries: in this fugue, the subject seems all-important. There are no rivaling counter-subjects or even any transitorily prominent motives. Instead, subject entries encompass an unusually large portion of the composition, and the subject appears in so many different shapes, distributed with such obvious determination and purpose, that nothing else seems to matter.

(The table above lists the subject statements with a short comment regarding the specifications under which they appear: inverted statements, augmentations and rhythmically varied versions are marked (inv), (augm) or (rh) respectively.)

 (ex. 19)

Apart from the usual adjustments in the tonal answer (see e.g. bars 3-6), the most frequent guise of the subject is the inversion in which all intervals appear upside down. Augmentations, i.e. statements in which each note value is twice as long, and rhythmic variations each occur three times. In these variations, the dotted-quarter-note value of the second subject note also "affects" the fourth and sixth notes, thus stretching the first subphrase in such a way that the second subphrase has to be shortened. Furthermore, two-part and three-part strettos occur frequently.

Beginnings and endings of subject statements suffer only small variations. The first note is shortened in bars 12 and 61, lengthened by anticipation in bar 26 and ornamented in bar 39. The final note is omitted in bar 50, delayed in bars 14 and 26; it is reached after an ornamenting escape note in bars 29 and 79, or after a chromatic passing note in bars 63/64. Finally, in the very dense three-part strettos of bars 52/53 and 54/55, only the first subphrase of the subject can be tracked, but even this contains irregularities at its end.

I/8.2.3 The counter-subject

This fugue does not contain a single counter-subject. Consequently, the sketch which performers draw to visualize the phrase structure and dynamic design as created in the combinations of primary material, would not rely on the blend of subject and counter-subject. Quite on the contrary, the intriguing juxtapositions are those three-part strettos in which one of the voices appears in augmentation. The sketch in ex. 20 shows bars 77-83, with

(ex. 20)

 

I/8.2.4 The episodes

There are ten subject-free passages. As can be seen from a cursory glance at the music, episodes occur more often in those sections where the subject appears in separate statements, but they are very scarce – and short – within the context of the strettos.

None of the episodes is related to the subject. The only subject-related component between entries occurs in bars 22/23 where both the upper and the middle voices extend their respective subject statements by sequencing the second subphrase. These sequences, however, appear inseparably linked to the preceding entries. As they do not give the impression of a contrast between statements, these bars are counted here as a subject extension rather than as an episode.

None of the episodes establishes any independent motive. There are, though, note groups which reappear once or twice occasionally. They should be mentioned not because of their thematic importance but because a performer would wish to pay attention to shaping them consistently.

-	In bars 14-16, the middle voice (C#-F#-E#-D#-C#) is imitated an octave higher (see upper voice, bars 16/17 from C#); it might also be recognized, with a variation in the first interval and a cut before the last, in bars 18/19 (see M: C#-B#-A#-Gx).
-	Also in bars 17/18, the upper voice (Cx-D#-E#-F#-E#-D#-C#) is followed by a varied sequence (see U: bars 18/19 from D#). The same curve recurs in inversion twice in bars 33/34 (see U: from E#; M: from G# - varied), and six more times, in several slightly different variations, in bars 41-44 (see U: bar 41 from G#, bar 42 from F#, bar 43 from C#; M: bar 42 from C#, bar 43 from F#, bar 44 from F#). In bars 82-85, overlapping with the end of the preceding subject entry, the same curve recurs three times in inversion (see M: bar 82 from A#, bar 83 from A#, bar 84 from F#).
-	Also in bars 82-85, the syncopated final note of the subject entry, together with the subsequent falling broken chord, set the model for a little curve which recurs twice (see U: bar 83 from A#, bar 84 from F#).
-	Finally, all three voices are involved in stating a four-note upbeat figure in bars 85-87.

Comparing the above-mentioned bars with the list of episodes drawn up above, one finds that the only episodes to develop any melodically lasting figures are E3, E5 and E10; these are not only the longest episodes in this fugue, but, as will emerge later, indicators of important structural caesuras. (Beyond this correspondence on a higher level, there is no further remarkable relationship between the episodes.)

All episodes without exception serve as periods of relaxation between the measures of tension built up by the subject statements. Only a few have any definite dynamic tendency:

-	In E3, a decrease can be felt from bar 17 towards the cadence in bar 19;
-	in E5, the descending sequences also express diminishing tension;
-	in E6, the tension seems to rise, both because of the unrelentingly driving eighth-notes in all voices and because of the harmonic intensification towards the A# major chord in bar 52;
-	E8 presents descending lines in all three voices;
-	E10 begins with falling lines and relaxing tension but then approaches the last bar of the composition in a powerful tension-rise.

 

I/8.2.5 Character, tempo, articulation, ornament realization

Both the variety of note values (half-notes, quarter-notes, eighth-notes, sixteenth-notes and various syncopations) and the predominantly stepwise motion in the fugue indicate a rather calm basic character. The range of an appropriate tempo is suggested by two inherent characteristics of the fugue:

-	The calm character requires a tempo slow enough to allow for each eighth-note to be felt with full melodic impact;
-	the augmented subject statements require a tempo fluent enough to allow for each subphrase to be felt in one breath.

(For performers: it seems a very good idea to establish a desirable tempo – and take note of the metronome mark – before starting to practice the details.)

The relative tempo of the prelude to the fugue can be simple:

The articulation which best expresses the character of this fugue is an overall legato, only interrupted by phrasing. These interruptions, which occur frequently, should be brought out very clearly.

The fugue contains only one ornament, a complex trill in the lower voice of bar 74. This ornament seems somewhat arbitrary: proving neither structurally important (like the trills in cadential closes) nor thematically consistent (like the typical subject-embellishing trills in many fugues), it appears redundant if not a bit confusing. (If it were played, it should consist of five notes: B^#-C^#-B^#-A^#-B^#.)

I/8.2.6 The design of the fugue

In this fugue, the design is obvious from the particular way the material is grouped:

I	The first section contains four separate subject statements: M U L / L. It is interesting to see that the regular lower-voice entry (bars 8-10) remains in the pitch range of a tenor while the fourth (redundant) entry sounds like a bass – thus giving the unsuspecting listener the illusion of a four-part fugue.
II	Bar 19 introduces the first of three strettos, all of them using the uninverted shape of the subject: MU UM UM. * In the first of these strettos, the imitation is at the octave at two quarter-notes' distance, with both voices equally faithful to the original. (The middle voice is in the lead merely because it enters first.) * In the second stretto, the subsequent voice comes in after only one quarter-note but sounds in a rhythmic variation and with a shortened second subphrase. * In the third stretto, the upper voice enters first but shows modifications at both the beginning and the end, while the middle voice gives a faithful rendition of the subject and also relates more immediately to the target key of F# major. This is why the middle voice should be regarded as the leading voice in this stretto.
III	The third section begins in bar 30 with three separate statements of the inverted subject: U(inv) M(inv) L(inv). As has been shown above, the closing episode of this section bears resemblance to that episode which ended the first section (compare bars 41-44 with bars 17-19), thus confirming the structural analogy of these two sections.
IV	Bar 44 introduces two strettos with subject inversions only: LU MU. * In the first of these strettos (just as in the first of the earlier strettos) the imitation is at the octave at two quarter-notes' distance, with both voices equally faithful to the original. (The lower voice is leading merely because it enters first.) * In the second stretto (just as in the second of the earlier strettos) the subsequent voice joins in after only one quarter-note but sounds in a rhythmic variation and with a shortened second subphrase.
V	A new grouping is introduced with the three-part strettos in bars 52-55, together with the return, after a long stretch of inverted statements, to the original shape of the subject. These two strettos seem both to sum up and further expand the earlier developments in the fugue. They sum them up in that they bring, in rapid succession, the original subject and its inversion; the expansion is caused by the increase in number of entries involved in a stretto from two to three. Both strettos show the lower voice in the lead, followed by the middle voice and complemented by the upper voice. The imitations are at the octave and separated by one quarter-note each. All statements appear in a shortened version. After this quite dramatic buildup, a single entry in the original shape (bars 57-60) presents no new pattern but serves to release tension.
VI	The final section of the fugue is distinguished from the previous ones by an entirely new feature: strettos containing an augmented subject entry. From bar 61 onward there are three such strettos: M LUinv L MU L,M UM. * In the first two of these strettos, the augmented entry imitates an unaugmented one at a distance of two quarter-notes; another statement in original values follows once the first is completed, i.e. halfway through the augmented statement. All voices remain very close to the original shape. (The augmented voice is clearly in the lead because of its much greater impact.) * The second group is slightly extended in that the voice which carried the augmented entry follows with a "repeated statement" – a separate entry reminiscent of the similar one after the second stretto in the fifth section. As before, this entry may serve to release the tension before the forthcoming greater climax. * The last stretto is the densest of all; it brings a heightening combination of features from earlier processes: (1) the augmentation of one of its entries links it to the two preceding strettos; (2) the fact that it commences in three-part stretto with octave imitation recalls the strettos in the fifth section; (3) the combination of the first two unaugmented entries recalls the second pair in the earlier two-part stretto sections: the subsequent voice enters after only a quarter-note and sounds in a rhythmic variation; (4) as in the previous stretto there are four entries, yet the fourth one appears here integrated as the expected final subject statement which comes in after the end of the rhythmically varied entry. This stretto thus outranks everything the listener has heard so far in this fugue.

There are few explicit cadential closes in this fugue, above all a twofold harmonic development, with a return to the tonic and a new progression occurring roughly in the middle of the fugue. The first four separate statements are in D^# minor, followed by a modulation to the dominant (A^# minor; relevant cadence in bar 19). The first stretto then modulates to F^# major which is reached in closing-formulas on the downbeat of bar 30. After this, the three separate inverted statements return to the home key of D^# minor (from bar 39 onwards). The second stretto group sets out from the tonic and leads to the dominant on the downbeat of bar 52, thus repeating the harmonic process of the first section. Subsequently, the fifth section begins and ends in A^#, with a short detour to F^# in the second three-part stretto. Similarly, the augmented-entry strettos begin and end in A^# while the middle group touches several other transitory keys. It is left to the final episode to modulate back to the home key. Owing to this harmonic device, this impressive fugue ends without a coda. (For a sketch showing the design of this fugue, see ex. 21.)

I/8.2.7 The development of tension

Within the first section, there is a very gradual tension-rise which is supported by the growing ensemble. At the same time, this growth in tension is counteracted on various levels: by the strong tension-release in the subject itself, by the lack of contrapuntal tension and by the neutral character of the episodes. In the second, third and fourth sections, the dynamic development between one entry (or stretto) and the next is negligible; emphasis thus remains only on the contrast between subject-carrying bars and episode bars. The fifth and sixth sections allow to distinguish three levels: a first level of extremely high tension in the strettos, a second level of drastically reduced, medium tension in the sudden separate statements (bars 57 and 72), and a third level with a strong tendency towards relaxation in the subject-free passages.

The overall development in this fugue shows the outlines of continuous tension-growth, with two analogous departures and two small internal relaxations. The steep drop or rise in tension between adjacent sections is so pronounced that in comparison each section seems built on a solid plateau. The only remarkable increase within a section occurs in the sixth section; here the final augmentation-stretto is evidently conceived as a climax which outranks any preceding ones. The following simplified diagram shows the structural processes in relation to the dynamic levels: