Chapter Seven—
Pelleas und Melisande , op. 5 (1902–1903)

With Schoenberg's move to Berlin in December 1901 and his assumption of a position at the Überbrettl cabaret, the relatively intense flow of composition represented by the Dehmel works of 1899 and the Gurrelieder of 1900–1 slowed up. Schoenberg's stay in the German capital, which lasted until the summer of 1903, is dominated by a single work, Pelleas und Melisande, which was probably begun in the summer of 1902 (one sketch leaf is dated 4 July 1902) and completed in February 1903 (the full score is dated 28 February 1903 at the end).

With the creation of a large instrumental work, Schoenberg returned to many of the structural issues he had faced in Verklärte Nacht. He now brought to bear the experience in counterpoint, harmony, and thematic structure—and, of course, orchestration—gained in the Gurrelieder.^[1]Pelleas is almost twice the length of Verklärte Nacht. The nature of the programmatic source is very different: a large play, as compared with a short lyrical poem. What Schoenberg achieved in the sextet, capturing in relatively compact form the entire spiritual-dramatic content of a poem, was not possible—or was at any rate not his goal—in the case of Pelleas.

From Schoenberg's own remarks (all printed in Bailey 1984) and from Berg's well-known analysis of Pelleas (Berg 1920), it can be determined that Schoenberg actually selected about eight of the fifteen scenes in Maurice Maeterlinck's drama: Melisande wandering in the forest (Maeterlinck act I, scene 2); the episode at the fountain in the park (II, 1); the tower scene where Melisande combs her hair (III,

[1] This chapter will not deal specifically with Schoenberg's orchestration, which is of course also an essential element of his early compositions. For an extended and sensitive treatment of this aspect of Gurrelieder, Pelleas, and the orchestral songs, op. 8, see Schubert 1975.

---

2); the vaults of the castle (III, 3); in the castle, when Golaud seizes Melisande by the hair (IV, 2); the love scene in the park (IV, 4); the entrance of the women servants in the castle (V, 1); the final scene of Melisande's death (V, 2). Schoenberg himself claimed that in Pelleas "I tried to mirror every detail of it [the play], with only a few omissions and slight changes in the order of the scenes" (cited in Bailey 1984, 61). As has long been realized, however, this process did not involve a mere translation of these scenes into music. Schoenberg used them to frame a genuinely symphonic work, based on a handful of themes that are continuously reshaped.

The large-scale form of Pelleas, like that of Verklärte Nacht, has stimulated different analytical interpretations. The one that has had the most authority is that of Berg, who proposes a four-part division according to the traditional symphonic model:

In the four principal sections of this symphonic poem we can even identify clearly the four movements of a symphony. Specifically, a large opening movement in sonata form; a second movement consisting of three shorter episodes, thus a three-part form (of which at least one scene suggests a scherzo-like character); a broadly spun-out Adagio; and lastly a finale constructed as a reprise.

BERG 1920, 3

Berg's detailed analysis also suggests how Schoenberg distributes elements of the first-movement form over the work as a whole. Although he is not always precise about the formal boundaries. Berg's scheme can be represented as in table 4 (a similar diagram is given in Bailey 1984, 70–71).

Schoenberg's strategy seems to be to introduce in part I a core of themes that are deployed almost continuously throughout the rest of the work. Like leitmotives, they are associated (as Schoenberg himself pointed out) with certain characters or more abstract concepts, and they undergo development that reflects the psychological or dramatic course of the play. For a work of such broad dimensions, there are actually relatively few recurring themes. These are shown in ex. 7.1, where the verbal labels, used mainly for ease of reference, correspond more or less to those provided in the commentaries of Berg and Schoenberg. In order to mark off the larger boundaries of the work—and to provide some needed contrast—Schoenberg presents new themes at or near the beginnings of parts II, III, and IV. Soon after, or even together with, their presentation, these themes are combined with the earlier themes of part I. As with Verklärte Nacht, a consideration of thematic relationships and thematic style is essential to an un-

---

derstanding of Pelleas. It is to this aspect we should turn before considering further the matter of sonata form and the large-scale structure.

Thematic Relationships, Style, and Structure

The process of thematic transformation in Pelleas und Melisande is richer and more elaborate than in any of Schoenberg's, earlier works. Some of the most obvious, audible relationships (many noted in Berg's thematic table of 1920) can be summarized as follows (see exx. 7.1 and 7.2):

1. The first theme of the work, MELISANDE 1, is based on a three-note motive, x, which is common to several important themes. The first three notes of MELISANDE 2 are clearly heard as an inversion of x, and the rising

might be heard (although it is embedded in a longer rising line) as x in its original form. The motive appears prominently as the first chromatic element in the broadly diatonic F-major melody based on the GOLAUD motive (ex. 7.2a). It also forms a distinctive part of PELLEAS 1 and appears inverted across the first and second measures of PELLEAS

---

Example 7.1
Pelleas und Melisande,  principal themes.

2; in the latter spot, the inversion occupies the same metrical position with respect to the bar line as in MELISANDE 2. The programmatic import of these recurrences is clear: Melisande's chromatic motive x infiltrates the themes of the two men with whom she becomes involved, Golaud and Pelleas.

2. Another significant transformation on a larger scale involves the reworking of PELLEAS 2 into MELISANDE 4 (aligned in ex. 7.2b). Berg suggests further that MELISANDE 4 is a transformation of MELISANDE 2 (Berg 1920, "Thementafel," ex. 9).

3. Several elements of the themes of part I of Pelleas are reworked in the main LOVE theme of part III (ex 7.2c). Berg suggests that the first measure is a transformation of MELISANDE 1, although he is not specific

---

Example 7.2
Pelleas und Melisande,  transformations of themes.

about the transformation. His example (Berg 1920, "Thementafel" ex. 16) seems to imply that the rhythmic pattern of x and the rising stepwise contour are taken over as

in LOVE. This derivation seems less persuasive than others pointed out by Berg. More convincing is his suggestion that mm. 6–7 of LOVE represent a further transformation of PELLEAS 2 VIA MELISANDE 4 (shown in ex. 7.2c). In addition, the figure of m. 7 of LOVE derives directly from the similar figure in MELISANDE 3.

---

Example 7.3
Pelleas und Melisande,  MELISANDE 2 theme.

Though important, these (and similar) transformations form only a small part of what is most characteristic about the individual themes in Pelleas. A proper understanding of the Pelleas style must take account of the harmonic, formal, and rhythmic contexts. In his often-cited autobiographical essay, "My Evolution," Schoenberg suggested that in Pelleas "many of the melodies contain extratonal intervals that demand extravagant movement of the harmony" (Schoenberg 1975, 82). This statement (for "extratonal" we can probably read "non-diatonic") seems to imply that at least on the local level, harmonic motion is determined or motivated by thematic forces. In fact, the relationship between theme and harmony in Pelleas is really one of mutual interdependence. Not only do the chromatic melodies require unusual harmonic successions; the harmony and the very careful accompanimental voice-leading also tend to give definition to themes that taken by themselves would be almost unintelligible as tonal entities.^[2] What is striking in Pelleas is how this interdependence of theme, harmony, and voice-leading generates thematic structures that are highly chromatic on the detailed level, but are governed by conventional progressions or cadential structures on the higher level. This process can be seen by examining three themes of increasing length and complexity.

The two-measure theme of MELISANDE 2 (ex. 7.3) is based on an open-ended progression of dominant-seventh chords, moving from the E⁷ on the upbeat (spelled with

instead of because of the downward resolution of the third), through the A⁷ on the subsequent downbeat, to the D⁷ (in second inversion) on the downbeat of m. 2. Although this kind of harmonic motion is not unusual in music after Wagner, or even by Wagner (for example, the chains of dominant sevenths at "O süsseste Wonne" in the Siegmund-Sieglinde duet in act I of Die Walküre), Schoenberg's voice-

[2] A sensitive, if brief, discussion of the interaction between the vertical and horizontal dimensions of Pelleas is Harvey 1975, 375–79, who treats the "forest" music at rehearsal no. 2.

---

leading and the intermediate harmonies it produces are distinctive. The vagrant sonorities actually sounded on the beats of the first full measure—an augmented

triad, a half-diminished seventh chord on C, and a C minor-seventh chord in third inversion—make no sense as a "progression"; they are really produced by the motion of the individual parts. After the initial E—A resolution, the bass line progresses by step around A: . This line makes extraordinary but logical counterpoint with the melody. The large downward melodic leap of an augmented fifth, , which alone would sound highly dissonant, is in fact prepared or rationalized by the present in the bass underneath the . Or it might be said that the melodic augmented fifth presents successively in the horizontal dimension what is also occurring simultaneously in the vertical dimension. This reciprocity between bass and melody continues in their exchange of C and on beats three and four.

The theme-complex PELLEAS (ex. 7.4) shows on a larger scale how an extremely chromatic melody and bass line can generate or fill out a harmonic skeleton and a phrase structure that are relatively conventional. The nine-measure theme consists of two phrases, divided 5+4. The first phrase moves from I (E major, in mm. 1 and 3) to V (m. 5); the second returns to the tonic (m. 9). The asymmetry, characteristic of Schoenberg (and of the Brahms themes he admired), is caused by the extension of the first phrase, and the concomitant delay of the arrival on the dominant, from m. 4, where we would normally expect it, to m. 5. The theme begins "on the run," with a remarkable superimposition of subdominant and tonic triads. The first three beats of m. 1 project a kind of rapid IV—V—I cadence; but, as in MELISANDE 2, the chords are produced or carried along by the interaction of a primarily stepwise bass line (and inner parts) with a highly mobile melody. In mm. 1–3 the bass traces a descent from A to the tonic E, then leaps down a major third to C, which moves down to the dominant, B. In the second half of the phrase there is a complementary stepwise ascent back to the tonic, whose final arrival is delayed by the upper chromatic neighbor,

. This unusual final approach to E in mm. 8–9, from a half-step above, creates an important (and undoubtedly intentional) parallel to mm. 4–5, where the dominant is likewise reached from a half-step above and from a note sustained for an entire measure. The   resolution of mm. 8–9 has also been directly anticipated in the bass motion across the bar line of mm. 2–3: indeed, in both cases the tonic triad is preceded by the identical half-diminished seventh ().

Schoenberg fills out the initial I–V progression of mm. 1–5 by means of harmonic motion toward, respectively, chords a major third above and below the tonic. The first gesture is toward iii or

, suggested by the V⁶ of iii in m. 2. Although the melodic "resolves" by fifth to the on the last beat of m. 2, the harmony and the bass line, which continues to descend, fail to support that resolution with a iii triad. Schoenberg returns instead to the tonic, on beat 2 of m. 3. The next move is to the lower major third, C major, on beats 3 and 4 of the measure. The   harmony of m. 4 serves as mediator between C and the true dominant, B, reached in

---

Example 7.4
Pelleas und Melisande,  PELLEAS theme.

m. 5. The chord might initially be heard as a German sixth within C; but the outer voices resolve outward by step to the dominant seventh of E. A deceptive resolution of the dominant leads to an A-minor or iv triad on the downbeat of m. 6, then to a series of vagrant harmonies connected by stepwise voice-leading in the accompanimental parts, and finally back to the tonic in m. 9.

The filling out of diatonic Stufen and a normal phrase structure with chromatic harmonies and voice-leading is carried still further in the main LOVE theme (ex. 7.5). Although this theme is twice as long as the PELLEAS complex just examined, it is likewise clearly rooted in E major, and the principal secondary key is likewise the dominant, B, which is reached at about the halfway point, the end of m. 8, and at the conclusion, in mm. 16–17. On the way to the first dominant is a clear subdominant (m. 6), which is itself prepared by a dominant seventh in mm. 5–6. The mid-

---

Example 7.5
Pelleas und Melisande,  LOVE theme.

---

Example 7.6
Pelleas und Melisande,  rewriting of mm. 5–7 of LOVE theme.

point harmonic articulation is to some extent obscured by the thematic structure: four phrases, of which the third (mm. 9-11) is an almost exact sequence (up a whole step) of the second (mm. 5-8) and the fourth is an extended variation of the two preceding. The form might be schematized as:

A (mm. 1-4)

B (5-8)

B sequence (9-11)

B varied (12-17)

This is a theme that seems literally to get carried away with itself. Although the harmonic design is essentially symmetrical or balanced, the thematic content is progressive, abandoning A after four measures for sequential and variation treatment of B. The phrase structure, too, begins to unravel: after an initially well-balanced 4+4 measures, we get 3+6.

From the very opening, the theme also projects tension or asymmetry in an apparent conflict between the notated

meter, supported by the bass and inner parts, and the tendency of the melody in mm. 1-3 toward , or a duple division of the measure. The last measure of the first phrase, m. 4, seems to establish the more firmly, but the metrical grid begins to dissolve again at the beginning of the B phrase, where Schoenberg extends the initial E of the neighbor motion. By analogy to the rhythm of mm. 1 and 2, we might expect this figure to be contained within m. 5, leading to the high on the downbeat of m. 6, as in ex. 7.6a. But the arrives only on the second beat of m. 6. Schoenberg has

---

thus "stretched" the notated

to an implied , which is continued across mm. 6-7 (ex. 7.6b). This "melodic" meter by no means displaces the notated meter; it is in fact contradicted by the bass, which remains tied to the bar line and arrives on the dominant B on the downbeat of m. 8.

As suggested above, mm. 9-11 repeat the pattern of mm. 5-8 up a whole step, without the initial augmentation. The sequence seems about to continue with another leg in m. 12, but is broken by the appearance of the climactic half-diminished chord, B-D-F-A, which is sustained for two and a half beats, like the E of m. 5. This extension may be said to balance, or cancel out, the earlier one. It now brings the

of m. 14 onto the notated downbeat (cf. the of m. 6 and of m. 9). Bass and melody are at last fully aligned and move together toward the half-cadence.

As in the PELLEAS theme-complex, Schoenberg expands or fills out the basic diatonic framework of the LOVE theme by means of harmonic substitutions and extensions. The first of these comes in m. 3, with the abrupt move to the C⁷ chord. The relationship between this chord and the tonic, which, as we have seen, was central to Verklärte Nacht, is one Schoenberg exploits frequently in Pelleas und Melisande, where German sixths come to function almost as substitute dominants. Here, however, the chord moves neither to the real tonic nor to its own"tonic," F; it resolves to a half-diminished seventh on D, thence to a C-minor triad in first inversion. The result is that the first phrase, rather than concluding on the tonic, dominant, or another diatonic Stufe, ends on

vi.

The harmonic motion to this remote area has its corollary in the melodic avoidance of E in m. 3. Instead of ascending by step from

to to E, the melody leaps up to . This note then descends chromatically across mm. 3-4, through    and The E that was avoided earlier appears at the start of the B phrase in m. 5, where the harmony also returns to the tonic region: the E is supported by an easily "readable" of IV.

In one sense, the phrase's ultimate goal in the bass, the dominant B, is reached in m. 8 and is then embellished and extended by what follows. The sequence of mm. 9-12, up a whole step from mm. 6-8, can be heard to move in the bass from the B up to

to , then back down stepwise (in m. 12) to the original point of departure, B. The B of m. 12 is harmonized not with a dominant seventh, but with the half-diminished seventh. The bass now continues to climb across mm. 13-14 up to the tonic E, which forms part of a final I-vi-V/V-V progression that concludes on the dominant seventh.

At a fundamental level, the three themes from Pelleas that we have examined in some detail all show traditional diatonic harmonies or progressions, as well as strong traces of conventional phrase structure. The basic skeleton supports a highly mobile melodic style and flexible, largely stepwise voice-leading. In Schoenberg's ability to expand or flesh out the skeleton by means of chromatic harmony and voice-leading,

---

the thematic idiom or style of Pelleas goes far beyond that of Verklärte Nacht. The contrast can be seen most directly by comparing the LOVE theme of the symphonic poem with the somewhat similar theme 5 from the sextet (ex. 7.7). Both are slow, broad themes in E major that contrast with the more agitated passages preceding them. The Verklärte Nacht theme (incorporating 5a and 5b) is ten measures long, probably the longest individual theme in a work in which, as we have seen, themes tend to be quite brief. The first four measures of both themes show something of the "sentence" proportions and structure: 1+1+2. Both themes move, on the broadest span, between basic diatonic degrees, the tonic and dominant. But in the theme from Pelleas und Melisande the upper voice and the bass range much farther afield, and the span is filled in with denser chromatic harmonic motion and voice-leading.

Both themes show a certain metrical-rhythmic flexibility that overrides the notated bar lines. In Verklärte Nacht that flexibility is most apparent in 5b, where the basic unit is actually two beats long (a beat is a dotted quarter note) and begins not on the downbeat of m. 111, but on beat 2 (with a preceding upbeat). The metrical extension and displacement thus generated continue through m. 114 and are dispelled only with the return of 5a in m. 115. In the LOVE theme the metrical ambiguities begin right away and affect the entire theme up until the final cadence in mm. 16-17.

Tonality and Form

Having examined the thematic style of Pelleas und Melisande in some detail, we can now look again at the larger formal structure in which the themes are placed. Although there is some merit to Berg's analysis of the work as a sonata-symphony Mischform (see table 4), this approach becomes less persuasive the more specific it gets. Schoenberg himself, although he seemed on the whole to approve of Berg's analysis (Berg 1987, 293) and on two occasions referred to the first fountain scene as a "scherzo" (Bailey 1984, 61, 66), nowhere elaborated any sonata-like view of Pelleas (unlike in the case of the First Quartet, op. 7).

Berg's analysis is, as its title implies, almost purely thematic; he gives no tonal references for any of his sections. The sonata-form analogy becomes less plausible when the harmonic dimension is considered, as has been observed by both Philip Friedheim (1963, 207) and Walter Bailey (1984, 72). To be sure, Pelleas begins and ends in D minor, and the opening material is recapitulated in the final section (part IV). From the sonata-form point of view, however, the tonal relationships are odd: Berg's "introduction" is in D minor and the first theme (GOLAUD) in F major; in the "recapitulation" the introduction reappears (which makes one suspicious of calling it an introduction at all) and is not in the tonic, but in

minor; the "first theme" is recapitulated in D minor. The sonata analogy is even hazy in Berg's analysis of part I. Although he locates the beginning of the "kurze

---

Example 7.7
Verklärte Nacht,  theme 5.

---

Reprise" at the reappearance of GOLAUD in F major at rehearsal number 14, he might more logically begin it seventeen measures earlier with what he calls the Schlußbsatz, the return of the "forest" themes of Melisande (MELISANDE 1 and 2; at 4 after 12 ), which leads into GOLAUD, as at the opening of the work.

If we avoid the urge to cast part I in sonata form and look instead at the tonal and thematic processes with a fresh eye—and an awareness of the programmatic source—a different kind of plan emerges (table 5). By stressing the status of F major, commentators on Pelleas und Melisande have ignored the large role played by A major in part I. The F major in which the GOLAUD theme appears at 5 is only a temporary stopping place on the way to A, whose key signature replaces that of D minor at 6. The GOLAUD theme is presented as a point of tonal repose in A major at 4 after 7, is interrupted by the FATE theme at 8, and is then restored in A at 6 after 8. Particularly in this latter passage, A has the feel of a fully established key area. Indeed, if there is a sonata-like "secondary key" in part I of Pelleas und Melisande, it would have to be A major, rather than F.

Up to the appearance of the PELLEAS theme at 9, then, the tonal scheme of the work is based on a chain of ascending thirds, D-F-A, which outline the tonic triad. The close harmonic association of the keys D, F, and A is programmatically and psychologically appropriate: Golaud has found the wandering Melisande and has "captured" and married her. The appearance of PELLEAS in E major literally breaks apart this scheme; it is intended to be tonally distinct (but not remote: just one notch on the circle of the fifths past A). During the initial presentation of the PELLEAS complex, up to 7 after 11, none of the previous themes is heard (I am ignoring in this context the relationship of the PELLEAS theme to the other themes).

At 4 after 12, MELISANDE 1 and 2 from the opening reappear, followed by PELLEAS and MELISANDE 3 together. Despite the additions, we have a clear sense of some kind of return; and although D minor is even less in evidence than at the actual beginning

---

of the piece, the distinct A⁷ harmony at 4-5 after 12 could be said to stand in for the key. As before, this section is followed by GOLAUD in F major (at 14 ), which is then followed by more MELISANDE and PELLEAS material, a climax, a ritardando, and the beginning of the "scherzo" at 16. The scherzo begins in A major and thus replicates the ascent from F to A at 6. Indeed, the A major of the scherzo could be said to fulfill or elaborate the earlier approach to that tonality.

The thematic-harmonic design outlined here suggests at the highest level an ABA ' form (as shown in table 5) whose basic tonal structure tends to override the actual formal boundaries of parts I and II. Even though many aspects of the scherzo-its mood, meter, and primary thematic material—are new, its key provides continuity with (or completes) what has preceded. It bears stressing, too, that the ABA ' structure is no conventional, tidy ternary form, but rather a skeleton, which Schoenberg fleshes out with an intricate layering of thematic and harmonic processes.

One of the harmonic relationships Schoenberg exploits most systematically throughout Pelleas und Melisande is that between the German sixth and the tonic. This tonal nexus, already prominent in Verklärte Nacht and Gurrelieder, can be said to govern many of the major articulation points of part I of the symphonic poem:

These cadential patterns, in each case circumventing the traditional dominanttonic resolution, create a distinctive harmonic syntax within part I of Pelleas. The question remains, however, just how these German sixths are actually preceived in context: do we come to hear the German sixth-tonic resolution as normative? Or, for instance at 7, do we hear the F⁷ clearly as pushing toward

, for which the A major acts as a deceptive resolution? Such questions are not easily answered. Indeed, Schoenberg seems consciously to be exploiting this very ambiguity of the German sixth, which is a dominant in its own right. It is clear too

---

that Schoenberg is working with conventional dominant expectations. The F⁷ that is (from one point of view) thwarted by the resolution to A at 6, at 4 after 7, and at 8 (here without the seventh), at last resolves normally at 15, where MELISANDE 4 enters in the key of

. This resolution is short-lived, however, as the F⁷ returns within a few measures and leads to the scherzo theme in A major (at 16 ).

As was suggested by the scenario outlined in table 4, part II of Pelleas is more episodic, less formally and tonally focused than part I. Despite the scherzoarting opening, it certainly does not conform to a scherzo-trio-scherzo structure. Schoenberg may have intended the somewhat looser construction of part II to give freer rein to his powers of thematic transformation (and to his orchestral prowess). One is struck especially by the reworking of PELLEAS 2 and MELISANDE 4 into the theme that both Schoenberg and Berg refer to as "Golaud's jealousy" (first heard at 23 ) and by the further transformation of MELISANDE 2 into the "flowing hair" figure in the tower scene at 25.

We can pass on to part III, the love scene. The action of part III, in which the lovers meet, attempt to consummate their passion, and are interrupted by Golaud, is very close to that of act II of Wagner's Tristan, where the tryst is similarly broken off by King Mark. One commentator has suggested (but not demonstrated) that Schoenberg's movement is actually modeled on the love duet of Tristan (Nitsche 1974, 15). On the broadest level there may be some truth to this assertion. Schoenberg's love scene, like Wagner's, is based on increasing sexual passion, and the prominent half-diminished seventh chord in m. 12 of the theme (see ex. 7.5) seems an almost direct reference to the "Tristan" chord. But formally Schoenberg's scene has its own distinctive shape that is rondo-like and is also infused, like part I of Pelleas, with sonata-like elements (table 6).

The rondo aspect consists in the twofold return of the LOVE theme and its alternation with sections based on MELISANDE 1. The sonata-like features include the strong dominant reached and sustained before 43, the development-like quality of what follows, and the return (not in E major, however) of LOVE at 44. But both formal models, rondo and sonata, are overridden to a large extent by an ongoing developmental process that builds toward successive climaxes, the most powerful at 46-47. These climaxes are based on three appearances of the thematic-harmonic component introduced in mm. 12-14 of LOVE (see ex. 7.5). The high A in m. 12, supported by the half-diminished chord B-D-F-A, serves to break the sequential pattern of mm. 6 and 9. Underneath the melody, the half-diminished chord now moves in m. 13 through stepwise voice-leading to a diminished seventh on

which leads to a first inversion A-major triad.

The model of mm. 12-14 first recurs at 40, where, through overlapping imitation, it now unfolds over five measures. The second recurrence is still more prolonged: at 7 after 46 Golaud appears, but Pelleas and Melisande continue their lovemaking a half-step higher (the half-diminished chord is now on C). The third and final pre-

---

Example 7.8
Pelleas und Melisande,  climax of LOVE theme.

sentation, at 2 before 48, deviates strikingly from the earlier ones. The initial harmony is no longer a half-diminished, but a dominant seventh (with ninth and thirteenth) of

(ex. 7.8). Here we might expect a functional resolution. In fact, the bass does resolve by fifth to (in the low brasses), but instead of a triad, this pitch supports a diminished seventh,  This is the same harmony (transposed) that occupies the second position in the original model in m. 13, and as in the model it leads stepwise to a triad in first inversion, here . The modification, just before 48, of the model of mm. 12-14 of LOVE is a signal that the scene is drawing to a close.

Berg suggests that part IV of Pelleas und Melisande serves the dual function of finale

---

Example 7.9
Pelleas und Melisande,  half-step approach to D.

to a four-movement work and "free reprise" of a single large sonata form (Berg 1920, 10). According to this scheme, the reprise begins with the "introduction" (comprising MELISANDE 1), followed by the full GOLAUD melody (Berg's Hauptsatz) and LOVE. The recapitulation is then interrupted at the death of Melisande, which leads to an "epilogue" in ABA' form. A is the original Hauptsatz; B is based on earlier themes, including MELISANDE 1, 2, 3, PELLEAS 1, and LOVE.

As Bailey has rightly pointed out (1984, 73), the large-scale tonal scheme of this "finale" shows Schoenberg recreating almost exactly (although in reverse order) the procedures used in part II of Verklärte Nacht: the tonic D minor is approached first from a half-step below,

, then from a half-step above, . The "introduction" or forest music is recapitulated in minor (beginning at 50 ), a key that becomes explicit with a long sustained pedal at 53. The pedal returns just before 55 and shortly thereafter moves to D (ex. 7.9). The thematic parallelism (bracketed in the example) makes this half-step juxtaposition as clear as possible: the motive (from GOLAUD) in the cellos and basses is imitated directly by the in the horn.

The scene of Melisande's death unfolds in

minor, which is likewise sustained by a long pedal. Almost exactly as in Verklärte Nacht, the pedal moves, not directly to the D, but at the last moment to its own dominant, , which resolves down by half-step to A, the dominant of D (ex. 7.10a and b). In both passages the resolution is marked "Breit."

Just as the basic tonal strategy in part IV of Pelleas is similar to that of the sextet, so too is Schoenberg's apparent desire to combine or associate the "twin" approaches to D with the phenomenon of recapitulation. And it is here that he still shows himself less than secure, even three and a half years after Verklärte Nacht. It will be recalled from chapter 5 that in Verklärte Nacht, the recapitulatory part II makes frequent approaches to D and actually culminates twice in the same passage of music. In part IV of Pelleas, where the harmonic language is more complex, the tonic is not implied as often or as directly, but it could be said that there is too much thematic recapitulation, that part IV is overlong.

In this respect it is significant that, just as Bruno Walter picked up on the potentially redundant climax of Verklärte Nacht and requested a cut, so did a conductor of Pelleas. In 1918 Zemlinsky, who was to conduct the work in Prague, apparently

---

Example 7.10
Verklärte Nacht  and  Pelleas und Melisande,  approaches to D from

wrote to Schoenberg requesting permission to cut from 50 up to 59— that is, from the beginning of the return of the introduction music in

up to Melisande's death. As he was to do with Walter many years later, Schoenberg refused,^[3] arguing that the passage in question is in fact "the best in the whole work" (Schoenberg 1964, 54), pointing out that a new theme is introduced at 50, and attempting to demonstrate subtleties of phrase structure and motivic variation.

From the viewpoint of tonal balance or symmetry (something Schoenberg does not raise in his letter), the composer was right to veto the cut: it would mean the loss of the approach to D from the half-step below. But despite Schoenberg's clever analytical pleading, it must be admitted that—as in the case of Verklärte Nacht— the conductor was on to something. It does seem redundant to have the GOLAUD/Hauptsatz theme recapitulated four different times in the tonic: at 55, and then three more times after Melisande's death, at 62, at 4 before 67, and at 69. In the first two instances the theme leads similarly to the LOVE theme, at 56 and at 5 after65, respectively. (In the latter case, PELLEAS I intervenes.) The thematic recurrences are not, of course, identical: each time the Hauptsatz theme is given a new bass line/ counterpoint, and the LOVE theme is presented differently. Yet in a work that despite its length has been, up to this point, as taut as Pelleas, the repetitions seem excessive. One could wish, with Zemlinsky, for a swifter denouement.

As in Verklärte Nacht, then, we see Schoenberg grappling with certain fundamental principles of large-scale form. Beneath all the brilliant thematic transformations and development in Pelleas und Melisande lies a basic uncertainty (not confessed by

[3] See Schoenberg 1964, 54-56. Bailey 1984, 67-70, cites and discusses the letter, as well as including portions that were cut from Schoenberg 1964.

---

Schoenberg, of course) about how much tonic and how much thematic return are enough. It was an issue that had occupied Schoenberg as early as the songs of the mid 1890s and that continued to occupy him as he tried to reconcile the apparent demands of traditional tonal forms with newer impulses coming from within.

Although there are many wonderful moments in Pelleas und Melisande, it is probably the least successful of the large-scale instrumental works of Schoenberg's early period. That it falls chronologically between Verklärte Nacht and the First Quartet seems to have had technical and expressive ramifications. We sense Schoenberg struggling to reconcile programmatic and thematic-formal demands. In the relatively compact dimensions of Verklärte Nacht, problems of this kind tended to be swept away by the bold strokes of inspiration. Despite compositional awkwardnesses, the sextet easily convinces us of its status as a masterpiece. Pelleas und Melisande fails to do so; it seems bloated, its shortcomings (or long-comings) more exposed.

Pelleas und Melisande also shows obvious affinities with Gurrelieder, but here too it suffers by comparison. Gurrelieder represents the very best in Schoenberg's extravagant late romantic style; under its decorated, overripe surface lies a taut synthesis of harmonic, thematic, and formal processes. In Pelleas und Melisande, the lush orchestration and rich harmony, outwardly similar to those of Gurrelieder, are put to different use: they too often become a cloak for intricate contrapuntal experimentation, as in the whole-tone canon based on MELISANDE 2 that begins at 2 after 2. Schoenberg soon came to realize that the cloak fitted such a technique poorly. In the songs of 1903-4, and then in the First Quartet, the contrapuntal preoccupations are laid bare in leaner textures that allowed (or inspired) Schoenberg to recapture something of the vitality of Verklärte Nacht and Gurrelieder.

---