“New Perspectives on Meter in Webern’s Opp. 5, 11, and 29”

Author: William van Geest

Meter in Webern’s music has been a subject of some discussion in recent decades. The topic is intriguing for several reasons. If, as Allen Forte asserts, “it seems reasonable to assume that . . . innovative pitch structures go hand in hand with innovative rhythmic structures,” 2 then we might expect much of rhythmic interest in Webern’s music. Indeed, Wallace Berry considers Webern an innovator with respect to meter: “It is very possible that Schoenberg’s vast influence with respect to the organization of pitch content in post-Romantic music, an influence reaching dominantly across the century, is equalled by Webern’s influence . . . in the direction of increasing metric flexibility.” 3 Pierre Boulez affirms this innovational role, comparing Webern to Alban Berg and their teacher, Arnold Schoenberg: “Only Webern—for all his attachment to rhythmic tradition—succeeded in breaking down the regularity of the bar by his extraordinary use of cross-rhythm, syncopation, accents on weak beats, counter-accents on strong beats, and other such devices designed to make us forget the regularity of meter.” 4 Meter in Webern’s music would later attract a certain amount more of analytical attention, perhaps most notably in a series of articles surrounding the op. 27 Variations. 5

Despite the attention Webern’s use of meter has received, much remains unexplored. Most previous investigations have been devoted to answering the question, “Is it metrical?” While this line of inquiry can produce deep—at times creative—insights into the phenomenon of meter, as well as impressive analytical machinery, the results frequently remain inconclusive or otherwise unsatisfactory. Indeed, such inquiries often amount to a test of fit between predetermined notions of meter and Webern’s music.

My aim in this investigation is to discover Webern’s operative notions of meter, and to situate these within his larger aesthetic goals. I begin from the conviction that meter as notated is meaningful for Webern, and that whatever meaning it does have is related in some manner and to some extent to conventional notions of meter. 6 Beyond this point, however, I allow the problem of metrical projection and discrepancies between notated meter and the musical surface to define the character and shape of his metrical practice.

My investigation builds upon several others, from which it is also distinguished. I will briefly discuss three of these. Carl Dahlhaus explores Webern’s metrical practice in two articles, one on op. 6 7 and another on op. 30, 8 capturing several features important to my approach. For one thing, he takes for granted that notated meter is meaningful throughout the pieces he examines. For another, he recognizes the importance of metrical features in interpreting Webern’s music and understanding his thinking, and he proposes a nuanced and highly sensitive interpretation of the works along these lines. Nevertheless, many of his assertions lack support, apart from an implicit appeal to a refined musical sensibility. Moreover, Dahlhaus largely limits the scope of his observations to the pieces he sets out to examine, failing to consider analytical evidence from, and implications for, other works. By applying analytical evidence both from within a given piece and from other of Webern’s works, I seek to make a stronger case for the types of claims made by Dahlhaus.

A second analyst, George Rochberg, has discussed the development of new approaches to meter and temporality in twentieth-century music, and Webern figures prominently in his accounts. I will invoke several of his observations over the course of this paper, including the concept of spatialization in music, the suppression of pulse, and the production of a sense of immobility through meter. I hope to offer analytical support to these claims where Rochberg often neglects it. Moreover, while agreeing with Rochberg regarding the presence of apparently ametrical or anti-metrical features in Webern’s music, I will counter his notion that, for Webern, notated meter is no more than a referential temporal frame. And while it seems that Rochberg principally had Webern’s later works in mind, I will show that features he identifies are distinctly present in many of Webern’s earlier works, thereby demonstrating continuity in Webern’s metrical practice.

Finally, one of the most important studies for the current investigation is Kathryn Bailey’s “Rhythm and Metre in Webern’s Late Works,” 9 in which she describes the composer’s metrical thinking as evinced in his sketches. Bailey’s work is invaluable for establishing Webern’s preoccupation with metrical matters, and for identifying several main lines along which the composer struggled. My approach differs from Bailey’s in several key ways, however. First, because of the difficulty in accessing Webern’s sketches and the adequacy of the published scores for establishing certain metrical traits, I limit my work to the latter. Second, while I explore one “late” work, the other two I examine are earlier. As we shall see, there is a great deal of metrical interest in these. Third, I seek to situate Webern’s metrical practice within a more general aesthetic and, in so doing, to explore his conceptions of meter more broadly and creatively.

There are several other considerations worthy of mention. First, two unique aspects of my approach are the prevalence of “birds-eye-view” observations and the lack of “blow-by-blow” presentation typical of metrical analysis. This condition is, in part, forced by the nature of Webern’s music: despite the appearance of metricity implied by his employment of conventional metrical notation and particularly by his use a single meter throughout a piece, more often than not his textures are not regular enough to project meter in a conventional sense. Second, I invoke the theoretical writings of Schoenberg. While Webern, unlike Schoenberg, left little in the way of theoretical writings 10 and even less treating metrical matters 11, Schoenberg’s influence on Webern is undeniable. Despite the writings I invoke being written over two decades after Webern had ceased study with him, there is nothing to say that Schoenberg’s views on rhythm had changed in the interim. Moreover, as we will see, their aptness to the phenomena I discuss is striking. Finally, I make occasional recourse to the distinction between the meter as notated and the audible musical surface. 12 As problematic as the terms of this distinction may be—not to mention my pitting them against each other as a dichotomy—employed provisionally, they have utility. Indeed, as we will see, Webern’s music seems to force this distinction.


Five Movements for String Quartet (op. 5): no. 4—Sehr langsam

I will begin with the fourth of Webern’s Five Movements for String Quartet, op. 5 (see Example 1). Hindrances to the projection of meter in this piece are immediately evident: the musical surface is sparse and fragmented and unfolds slowly, and gestures are short and do not align neatly with the notated meter. Indeed, the notated meter, a consistent 3/4 throughout, seems like a frame upon which Webern has disposed the musical surface, what Christopher Hasty calls a “temporal grid for the timing of musical events.” 13 One obvious exception to this lack of regularity is the viola’s ppp pizzicato ostinato beginning in bar 7. But even this runs counter to other elements in the texture, presenting as it does triplet eighth notes against the other instruments’ duple eighth notes, and it begins to decay after only two bars. Despite the poor conditions for the projection of a robust metrical hierarchy, a closer consideration of several aspects of this musical surface and its relation to the notated meter reveals Webern’s careful attention to metrical aspects of the piece, as well as peculiar, even idiomatic, conceptions of meter.

Example 1: Five Movements, Op. 5: no. 4—Sehr langsam


The first metrical feature to note is the unusual breadth of the bars. Webern provides a metronome marking of ♪ = ca. 58 to accompany the indication Sehr langsam. These do not match, of course, if metrical conventions apply: normally metronome indications are given with reference to the tactus, which would be the quarter note in this meter. This decision may have been simply pragmatic: if Webern kept the same tempo and adjusted his metronome indication according to the quarter note, the indication would be 29, a setting that does not exist on most metronomes. But even if so, this pulse might be too slow to be helpful to the performer; as Justin London discusses, the lower threshold for the speed of metrical pulses is approximately two seconds. 14 Likewise, in combination with this meter, the tempo would put the duration of a bar at just over six seconds, which is also beyond conventional accounts of the lower limit for metrical entrainment. 15 While Webern likely did not have much, if any, musical perception–cognition research at his disposal, he must have known by convention that this was an unusually long bar, so we might wonder why he notated the piece thus. Moreover, this is not an isolated instance of this notational practice; we find it in several of Webern’s early works—and among these, this is a modest example. 16

However, discerning Webern’s precise motivation here is perhaps inconsequential; the point is that the length of these bars is unusual and pushes the limits of metrical conventions. Indeed, the metrical background that Webern establishes is by no means arbitrary. With long, slow bars, it constitutes the metrical expression of a more general aesthetic approach in the piece, one in which the various elements are put under a metaphorical magnifying glass. For example, the most intense dynamic level the composer indicates is pp, and he provides indications such as äußerst ruhig and so zart als möglich at several points. Moreover, the texture is sparse, motion is minimal, and the ideas are very short. By Webern’s understatement of these elements, the significance of the slightest change or activity increases markedly. Indeed, one frequently encounters such an aesthetic of extreme subtlety in Webern’s oeuvre. 17 In the realm of meter, the remarkably slow tempo of this movement permits contemplation of the various elements of metrical experience, principally the various metrical positions: downbeat, second beat, and so on. Interestingly, because of the slow tempo, these are extracted from the natural projection of meter, causing listeners to wonder whether the various metrical elements retain their identity when so isolated: Can a downbeat still sound and feel like a downbeat when the tempo is so slow?

Advancing deeper into the piece, another prominent feature is Webern’s exploration of different metrical dispositions for gestures. For example, in the second section, 18 a figure of four descending eighth notes appears in three different metrical dispositions (bars 3–5), 19 and overlapping with this, the cello imitates a gesture in the first violin, but it is displaced by an eighth note (bars 4–5). In a similar spirit, Webern presents the ascending figure that closes three formal units in a different metrical position and in a slightly different rhythmic guise each time (bars 6, 10, and 13). Additionally, the ostinato in bars 7–10 decays in a related manner. 20 This practice is, of course, by no means novel. Harald Krebs has explored such phenomena—what he calls “metrical dissonance”—in a variety of musical styles, most from the nineteenth century, with this particular type constituting “grouping dissonance.” 21 Such deployment of a gesture in several different metrical dispositions is in fact characteristic of Webern’s music of this period. 22 At the very least, this suggests that metrical matters play a significant role in the music and, thus, that the other metrical phenomena observed here are likely not incidental.

Webern’s handling of downbeats also bears consideration. To begin with, we find a suggestive correlation between downbeat-activation and form. 23 I propose four formal units (bars 1–2, 2–6, 7–10, and 11–12) plus a coda (12–13). 24 The boundaries of each of these are defined with a rit . . . tempo. 25 Interestingly, Webern leaves the initial beat of each formal unit unactivated in every case, and often conspicuously so. For example, we find a constant stream of sixteenth notes leading up the downbeat of bar 7 and a constant stream of triplet eighths following, but on the downbeat itself, we find a rest. 26 In fact, non-activations on the eighth-note level are generally rare in this piece (Example 2). Excluding the rests at the beginning of the first bar, at the end of the last, and those blurred over by the ostinato, there are only eleven non-activated eighth-note positions, four of which are the beginning of formal units. What explains this practice? As we will see, the non-activation of the beginning of formal units is not an uncommon feature of Webern’s music; it is found other works and other guises. I will suggest that this practice relates to a view of strong metrical positions as inherently accented. If such a belief is operational here, it is easy to see why Webern leaves these positions unactivated: beginning a formal unit with an accent would shatter the ruhig effect he clearly seeks, and it would highlight formal boundaries in wholly inappropriate fashion.

Example 2: Five Movements, Op. 5: no. 4—Sehr langsam


In conjunction with such relations between formal units and meter, divisions between each section, with the exception of the coda, correspond to bar lines, in spite of the general lack of clear correspondence between gestures and notated meter noted earlier. While this may suggest a slippage between meter and form, one not unheard of in rhythmic theory, 27 this close correspondence more specifically suggests a treatment of visual metrical cues (i.e., the bar line) as boundaries, for there is no a priori reason that a phrase or formal unit may not, for example, include a pick-up or end on a downbeat, thus spilling over a bar line. For a composer like Webern who is sensitive to the visual representations of meter in musical scores, it is not difficult to understand how such a slippage may come about. Indeed, this alignment of meter and form, found elsewhere in Webern’s works, 28 suggests that he is repurposing metrical cues for imaginative ends.

We should also note Webern’s distribution of musical energy relative to these formal units. In each formal unit, sounded musical material continues until the final eighth-note position, and in two formal units, at the end of the first unit (bar 2) and immediately preceding the coda (bar 12), this final eighth-note position features a punctuation of sorts. 29 On one hand, this practice might be considered a formal marker of a unit’s end, since, for example, two of the four op. 5 pieces (the first and the third) end with such a punctuation. On the other, this practice seems to ignore or even subvert the typical internal dynamics of a bar, whereby energy either moves toward or emanates from the downbeat.

What notion of meter is in operation if such internal dynamics are ignored? Hasty describes a prevalent—and in his view inadequate, misleading, and unmusical—notion of meter as a container that may be filled with musical contents, or a frame with specific positions upon which to affix material. 30 Rochberg describes the notated meter in Webern’s Variations, op. 27, as “a frame of reference only.” 31 Webern’s shunning of conventions pertaining to the internal dynamics of bars and the liberty he takes in distributing onsets irrespective of, or even in direct opposition to, these dynamics is congruent with such a view. He seems unconcerned with establishing a meter; either he takes its power for granted, or he is deliberately trying to subvert it.

Schoenberg reflects another relatable concept in drafts for his treatise, entitled Coherence, Counterpoint, Instrumentation, Instruction in Form. In considering compositional approaches to meter, Schoenberg claims that constant changes of meter threaten “metrical unity.” 32 His language implies that an unchanging meter supplies unity, a view that makes the notated meter an active agent in a musical surface. Schoenberg goes on to describe his own practice: “I, however, avoid changes of meter if possible and instead use shifts of accent; it has this advantage: that the [metric] unit is not continually altered … yet every kind of rhythmic subtlety can still be presented.” 33 This statement seems to take for granted that the musical surface will involve “shifts of accent,” perhaps as an aesthetic desideratum, and similarly to express Schoenberg’s desire for “every kind of rhythmic subtlety.” It is easy to sense resonances between these views and Webern’s employment of an unchanging meter signature combined with the (occasionally strong) activation of weak metrical positions in the present piece. Both approaches—Schoenberg’s in theory and Webern’s in practice—suggest a view of the written meter as exerting a power on the musical surface that puts the two stances in dialogue. This, in turn, suggests that strong metrical positions are somehow inherently accented and need no further encouragement. While this notion may appear fanciful, it exhibits close parallels with one reflected in contemporary theoretical literature: what Lerdahl and Jackendoff call “metrical accent.” 34According to this, strong metrical positions do not merely prescribe an accent or stress, they somehow produce accent. 35 We will return to this notion below.

Two other matters of metrical import warrant exploration. First, Webern activates the downbeat in the middle of formal sections just as consistently as he leaves the downbeat of the beginning of formal units unactivated. 36 Admittedly, in the first two formal units, this activation can hardly be called considered monumental; the first (bar 2) is constituted of by the beginning of a ppp tremolo and the second (bar 5) by the terminus of a descending line and a change in texture. The third and fourth, however, are different. The downbeat in the middle of the fourth formal unit (bar 12) is rather obviously activated quite clearly a climax, as evidenced by the hairpins surrounding this climactic moment and the single-mindedness of the homophonic texture. The middle downbeat of the third formal unit (bar 9) is more subtle: the downbeat immediately preceding (bar 8) is rather conspicuously unactivated except for the somewhat impervious ostinato, as is the one following (bar 10). Moreover, in the violin line leading up to this activation, we find not only one of the longest hairpin pairs in the piece, but also a systematic activation of off-beat eighth-note positions. 37 This systematic off-beat activation may be found in other anacrustic situations in Webern’s music, 38 indeed including the lead-up to the climax in the fourth formal unit. It is difficult to identify the conception that underlies this practice. Given his particular handling of downbeat positions, some aspects of which will be discussed below, it is conceivable that Webern understands such activation of weak positions as productive of anacrustic momentum in an abstract sense, but he may also be imitating a practice common enough in earlier music. 39

One other observation should be added to this discussion. In his essay, “Today’s Manner of Performing Classical Music,” Schoenberg describes the over-accentuation of strong beats as exhibiting “poor musicianship,” and he asserts that “[bringing] out the ‘centre of gravity’ of the phrase is indispensible to an intelligent and intelligible presentation of its contents.” 40 Webern’s combination of unactivated downbeats and activated mid-phrase downbeats could readily be seen as a manifestation of this approach. An argument made along these lines would need to account for two matters: first, how and whether the strategic activation of downbeats may constitute bringing out the “centre of gravity” of the phrase; and second, whether these formal units can be considered phrases, as form and phrase structure seem to require conceptual reconsideration in light of Webern’s particular aesthetic and the extreme density of his music. While the second point lies beyond the scope of this investigation, I will pursue the first below.

Despite its brevity, Webern’s op. 5, no. 4 features much in the way of metrical interest—enough at least to demonstrate Webern’s preoccupation with metrical features. Indeed, this preoccupation is clear in other of the four pieces of the op. 5 set, in some of the ways explored here and others not explored. 41 Webern’s handling of metrical features is also particular; while he seems to accept some conventional notions of meter, his acceptance takes the form of subversion, problematizing, or repurposing. Ultimately, he takes such an approach not for its own sake, but rather to serve his expressive ends.


Three Little Pieces, op. 11: no. 1, Mäßige ♪

The first of Webern’s Three Little Pieces for cello and piano, op. 11 (see Example 3) presents a metrical background in some respects similar to the piece just discussed. The notated meter is consistent throughout while the musical surface does not project it. If anything, in contrast to the nearly constant stream of activations in op. 5, no. 4, the musical surface here is more disjointed and features more rests, which together should make conventional metrical projection even more unlikely. Moreover, as with op. 5, no. 4, this piece exhibits a similar overall ruhig affect with corresponding  slow tempo, and thus we may also speak here of a magnification of various musical elements, including those related to meter. 42 Interestingly, the bars are the same absolute duration as those in the op. 5 piece, though the internal organization is different, at least by conventional notions of meter: we find two groups of three eighth notes instead of three groups of two. This is a significant difference, for if the meter is merely a grid upon which a musical surface is imposed, there should be no difference between the handling of 3/4 and 6/8; but as we will see, there are aspects of Webern’s practice here that are proper to 6/8. This piece is also shorter than the previous by several bars, which accords, perhaps, with the even greater density of its gestures.

Example 3: Three little pieces, Op. 11: no. 1—Mäßige ♪


At least one of the specific metrical practices found in op. 5, no. 4 is also found here. I propose four formal units (bars 1–3, 3–5, 5–6, and 7–9), with divisions determined again by rit . . . tempo pairs. As in the earlier piece, Webern leaves the first beat of each formal unit unactivated. While, statistically speaking, this feature should be less surprising, given that the musical surface features more rests than the op. 5 piece, several of these non-activations are again rather conspicuous, suggesting that Webern has taken special care in this matter. For example, the cello line that ends the first section in bar 3 continues to the very end of the formal unit, and the piano only comes in an eighth rest after the tempo marking, our indicator of the beginning of a new formal unit. The following formal unit likewise begins with an eighth rest followed by two successive entries of sustained notes in the piano (bar 5), conspicuously surrounding this rest by sustained notes as well. The same could also be observed for the rest at the beginning of the final section (bar 7), but here, the rest is even shorter.

Webern’s practice of leaving initial beats unactivated applies here to individual bars, a level lower than formal units. The overall effect on both levels is a blurring of the musical surface, a rounding of sharp metrical edges. Formal units seem to taper off at the end; the exact moment of their inception is unclear, both because the moment of return to tempo is musically unmarked, and because a strong metrical position is passed over unactivated. With regard to bars, the omission of downbeat activation similarly results in a lack of metrical orientation that such activation would provide. In a comment he made to Schoenberg following a performance he conducted of the latter’s Drei Volksliedsätze, Webern indicated that such an optimally-smooth texture satisfies an aesthetic desideratum. Webern reports, “We sang the song Herzlieblich Lieb without bar lines. I did not give a definite beat at all.” 43 Interestingly, the relation between form and notated meter may even participate in this smoothing: unlike the op. 5, no. 4, two out of the three formal divisions do not align with bar lines.

Webern’s non-activation of downbeats is again combined with another metrical characteristic: he activates the middle position 44 of bars with a surprising frequency. In contrast to the overall lack of consistency in the activation of other metrical positions, this middle position is activated in five of the piece’s nine bars. 45 Furthermore, this activation is often remarkably strong; in all cases but one it consists of a dense chord, made all the more striking by an otherwise sparse texture. Moreover, with the small playbook of metrical practices already developed, we may propose specific reasons for why Webern may have left some middle positions unactivated. In two of these, bars 3 and 5, the position constitutes a formal division, and we have already observed Webern’s tendency to leave such positions unactivated, sometimes conspicuously so. In bar 9, we can consider the piece’s sounding material finished by this point. This leaves only the middle position in bar 8 unexplained, the gestures of which could arguably constitute the activation of weak metrical positions for anacrustic effect discussed earlier, where the climax of this figure is the downbeat of bar 9 (we will return to this matter below). Thus the activation of middle positions may be considered the rule, to which exceptions in nearly every case may be justified by some other means.

This combination of unactivated initial beats and regularly activated middle positions suggests several points. First, it should put to rest any question of whether the notated meter in this piece is significant, and this with respect to the specific internal dynamics that a meter signature conventionally suggests. The question, if any, is what notions of meter guide Webern in this practice, and what effect he intends this practice to have. It would be tempting to imagine that Webern seeks to establish a meter, one out of phase with the notated meter by half a bar. 46 This, however, seems wholly inconsistent with his metrical practice in general, for he rarely, if ever, seems concerned with establishing a meter conventionally, that is, by emphasizing strong metrical positions. Perhaps, then, he seeks to draw energy away from the downbeat and toward the metrical position farthest from it. If this were the case, Webern’s practice here would truly constitute an application at the level of bars of that observed in the op. 5 piece on the level of formal units. This, in turn, would accord with the general intensification of features we have observed from that piece to this, specifically the shorter length and denser gestures.

There is one more metrical feature in this piece that bears discussion: while Webern often leaves downbeats unactivated, there are two bars in which the downbeat is activated, namely bars 5 and 9. 47 It is suggestive that these bars are the middle and final bars of the piece. Bar 5 seems quite clearly to constitute a climax, given the accelerando leading to it and rit departing from it. This moment also constitutes a dynamic climax. 48 I propose that, in line with the conception of downbeats as inherently accented, such an accent is desired here as marking—or, rather, contributing to—a dramatic climax. What stands as remarkable is Webern’s treatment of strong metrical positions as a resource to be called upon at opportune times and left otherwise untouched, like certain registers of an instrument or intense dynamics. Where by convention meter is both projected by the music and a condition by which the music is understood, 49 Webern here uses meter as an expressive device. 50 With regard to the activated downbeat in the final bar, I would suggest that it is activated similarly to take advantage of the inherent accent, but as a formal, rather than dramatic, indicator. It signals formal closure.

This differs, of course, from the use of off-beat activations to mark (or perhaps effect) formal closure, as discussed above. That Webern employs at least two different practices of activation as formal closure should not be worrisome. On the contrary, too much consistency would be predictable in a way antithetical to Webern’s broader aesthetics. Moreover, we find strong-position activations at critical formal moments in other works by the composer. For example, at the beginning of op. 31, no. 5 (“Freundselig ist das Wort”), an activated strong metrical position closes the choir’s first phrase and signals the beginning of the piece proper (see Example 4). In other words, the activation of a strong metrical position in a context where such positions are infrequently activated may constitute a point of formal articulation much as, for example, the arrival of tonic signals formal inception in a sonata introduction or a final cadence indicates the end of a piece. As William Caplin points out, in both cases musical material may precede or follow, but formally speaking, the piece has begun or ended. 51 This interpretation of Webern’s handling of downbeat activation is admittedly speculative and depends on a very particular view of meter. Indeed, it could easily be dismissed as analytical fancy were not a similar structure also found in the third piece of the op. 11 set, where, in addition to such activations in the middle and end, we find one marking the beginning as well.

Example 4: Second Cantata, Op. 31: V. “Freundselig ist das Wort” (bars 1–3)


In summary, op. 11, no. 1 shares some metrical characteristics with the op. 5, no. 4, including Webern’s avoidance of activating the initial beats of units and reserving downbeat activation for strategic moments, especially the middle of units. In several ways it constitutes an extension of the procedures found in op. 5, no. 4, including the application of downbeat non-activation and center-of-gravity emphasis to the level of the bar and reserving activations for dramatic climaxes and formal indications.


First Cantata, op. 29: I. “Zündender Lichtblitz,” bars 1–13

The final work I will discuss is the introduction (bars 1–13) to the first movement of Webern’s First Cantata, op. 29 (see Example 5). 52 The challenge this piece poses to the projection of meter is significant, most obviously owing to the frequent change of meter. With a new meter signature in every bar but one, this piece represents a significant departure from the previous two, which maintain the same time signature throughout. But the metrical differences go deeper. In comparison to the dense, somewhat erratic gestures of the previous two pieces, this excerpt features a significantly attenuated durational vocabulary. Nevertheless, I hope to show that, as with the previous two pieces, the musical surface relates closely to its notated meter, albeit in different ways. Despite these markedly different metrical styles, I will also argue that these pieces do share some important metrical features.

Example 5: First Cantata, Op. 29: I. “Zündender Lichtblitz” (bars 1–15)



Let us begin by examining the changes in meter signature. We find multiple orders of change. In most cases the numerator changes from one bar to the next, but the denominator also toggles between 2 and 4. This complicates the situation significantly, since each change entails substantial differences that do not necessarily operate on the same plane; changes of numerator conventionally pertain to the internal organization and length of bars, whereas changes of denominator pertain to which note value corresponds to the main motion, the tactus. Nevertheless, we may still make some sense of these changes. With respect to the changes of denominator, we do find corresponding conventional differences in tactus. It is tempting to draw conclusions about form from these differences; for example, the four bars with a denominator of 2 might suggest the division of the introduction into four units, each of which begins with a slowed bar. But the way Webern employs such bars in the remainder of the movement readily contradicts this hypothesis, at least if the phrase structure of the music is assumed to align with that of the text. It is probably best, then, to rely on the indicators of formal units employed above, namely moments of rit . . . tempo, with two such units found in this excerpt. This disjunction is noteworthy because it implies that the tactus may change over the course of a given phrase, a feature that further mitigates the projection of meter.

Pursuant of meter signatures with denominator 2, the basic unit of motion in these bars would seem to be the whole note and not the half-note that the meter signature would suggest, its duration measured off by accompanimental half-notes. Presumably, Webern employs meter signatures with a denominator of 2, despite their inaccuracy with regard to conventions of metrical notation, 53 to avoid more accurate but less conventional signatures with denominator 1. 54 An obvious factor in this matter is Webern’s employment of relatively large note values, more simply notating the piece in values one-quarter the length of those used, for example. Why does he use such large values? It is easy to imagine that, given his fascination with Renaissance music, 55 Webern sought to capture an antiquated practice in which the durational palette involves larger values than those typically used in the nineteenth or twentieth century. While this may be the case, it does not greatly advance our understanding of Webern’s metrical practice. The Cantata’s second movement may be of significant metrical interest, since there Webern employs mostly small values—the denominators of its meter signatures toggle between 8 and 16—and frequently he has beams crossing bar lines. Thus by metrical notation alone, he is apparently seeking to establish contrast between the two works, but to what effect, and on what grounds? Adequately accounting for Webern’s thinking here is at best difficult. Perhaps further investigation will shed light on this question, but at the very least, we may note the striking visual representation of meter for Webern.

With respect to changes of numerator, we may wonder to what extent the numerators Webern employs reflect the internal organization of bars according to metrical conventions, and if any such relationship is more than haphazard, that is, whether he affirms conventional internal organization, subverts it, or outright ignores it. This question is difficult to answer because of a combination of two factors. First, the numerators Webern employs are low numbers; generally, meters with a higher numerator—particularly those of compound meter—reflect a specific subdivision, while those with a lower numerator do not. Second, Webern indicates a rather brisk tactus, meaning that the notated meter will not suggest a metrical structuring of the musical surface, suggested on lower metrical levels by beaming and other notational conventions. This reflects a practice found in several of Webern’s works that I have elsewhere called “non-committal meter,” an effect owing to these two factors whereby the notated meter prescribes no subdivision of a bar’s contents. 56 Consequently, there are only three bars whose numerator could meaningfully reflect a conventional internal organization, namely the two 4/4 bars (4 and 8) and the 6/2 bar (6). The denominators of the other bars are too low to prescribe a metrical organization. 57 This, however, points up a significant difference between the first two pieces examined and the present one. Owing again to the choice of tactus, one bar of 5/4, the longest bar of denominator-4 meter here, would pass by more quickly than would a quarter note in the previous two pieces. Thus, it is clear that the magnification of elements observed in the earlier pieces does not obtain here, which constitutes further evidence of a different metrical practice.

It is worth exploring the relation of the notated meter to the musical surface. The constant changes in numerator certainly suggest that Webern conformed the notated meter to the musical surface, lending the musical surface a certain priority not evident in the previous pieces, in which the musical surface seems to wend around and otherwise respond to the notated meter. In conjunction with this idea, Webern never sustains a tone over a bar line in this piece. 58 In light of the foregoing analyses, we may make some inferences regarding this intriguing feature. First, it confirms the close relationship between musical surface and meter and, with it, the meaningfulness of the notated meter, even if it is wholly adapted to the surface. This is another example of Webern’s striking sensitivity to metrical elements. But along with it is a misunderstanding that leads to a reimagining of these elements: vertical lines should be used to represent divisions of bars is of course arbitrary, but if taken as non-arbitrary, they may easily be understood as barriers, which is how he treats them here. Thus, this may represent another substantiation of Webern’s treatment of visual metrical cues as barriers for elements of the musical surface. 59 What motivated Webern here, however, remains obscure.

There are several additional metrical features of note in this excerpt. First, as mentioned above, Webern employs only three durations, namely the quarter note, half note, and whole note. 60 Second, we should note the peculiar distribution of musical energy in this piece, particularly the near-constant stream of activations at the quarter-note level in lebhaft sections (see Example 6). While we observed a similar frequency of activations in op. 5, no. 4, it also had a greater variety of activations, especially on lower metrical levels. This regularity of musical surface coupled with a limited durational vocabulary is in fact characteristic of several of Webern’s later pieces. 61 Such austerity of duration seems related to Webern’s adoption of new principles of pitch organization beginning with op. 15 and consummated perhaps in op. 21, for it is in the latter piece that we first find similar patterns of metrical austerity. 62 To these might be added the irregular accentuation lent to the texture by varying articulation, frequent changes in register and timbre, and prescribed accents. On one hand, then, the regularity of activations creates a strong metrical potential, but on the other, the lack of consistency in the grouping of these activations lends the texture a certain metrical anonymity, even timelessness. 63 In this sense, the fluctuation of meter signature, a feature that generates variety, Webern takes so far that it contributes to the nondescript texture: without metrical consistency, beats do not take on strong metrical identity. We may relate this to op. 5, no. 4, where we observed the unusual spreading of musical energy by the non-activation of strong positions and the continuation of musical material to the end of bars. Both, it would seem, deliberately mitigate the projection of meter, but they do so in different ways. The most metrically distinct bars are those with a denominator of 2, with the duple division of whole notes, but given the variability of their numerators, such moments are fleeting.

Example 6: First Cantata, Op. 29: I. “Zündender Lichtblitz” (bars 1–15) activations



One final metrical feature relating this passage to the earlier two pieces is the downbeat non-activation in its opening bar, which bears the unusual meter signature of 7/2. In addition to the relative rarity of such a signature, 64 its relation to the bar’s contents is striking. The gesture it encloses is the same as that in bar 6, which is notated in 6/2, except that it begins with a half rest. This impels us to consider the initial rest seriously, since Webern could easily have omitted it. What it suggests is not wholly clear. This may be simply another example of Webern’s desire to avoid beginning units, metrical or formal, with a downbeat activation, a habit consistent with the general ruhig character of this gesture, if the conclusions we drew regarding the earlier two pieces are reliable. Or, the rest’s position at the beginning of the bar may force the gesture to begin on a weak metrical position, which, depending on how one interprets the metrical subdivision of the meter, would put at least some of the activations on weak metrical positions. Interestingly, Webern also leaves the downbeat of the following bar (the first lebhaft bar of the movement) unactivated, yet again blurring the transition to this new metrical character similar to op. 5, no.4 and op. 11, no. 1. While the non-activation of downbeats does seem to feature in this movement, the patterns they form, if any, are difficult to decipher. 65

This piece thus poses a set of challenges to metrical projection rather different from those identified in the other pieces I have examined here, and thus represents a different metrical practice, indeed, one found in numerous later works. Still, several practices shared with the earlier pieces also point to a certain continuity in Webern’s metrical practice.



In this investigation, I hope to have shown first and foremost the value of metrical analysis for illuminating Webern’s music, even as the analytical means I have employed here are decidedly simple. I have examined Webern’s handling of downbeats and initial metrical positions of formal units, relations between metrical and formal units, the distribution of activations across bars and formal units, and the role of metrical features in Webern’s broader aesthetic. This examination has brought to light patterns that suggest some peculiar underlying notions of meter, including meter as a latent force subverted or tapped into by the musical surface, strong metrical positions as inherently accented, and downbeat activation as formal indicator. Though some of the notions I have discussed are admittedly imaginative, analytical evidence exists to support them, and as I have shown, they are not so different from ideas that have been advanced in existing theoretical literature. I also hope that this investigation may aid in deciphering Webern’s other works, for in an oeuvre of such intricate construction, there must remain much worthy of analytical attention.

William van Geest is a Ph.D student in Music Theory at the University of Michigan. He specializes in the history of music theory, rhythm and meter, and the medieval grammar tradition. His dissertation explores the intellectual context surrounding the emergence of mensural rhythm in thirteenth-century France. William has presented papers at several national and international conferences, including those of the Society for Music Theory, the Society for Music Analysis UK, the Canadian University Music Society, and the International Conference of Students of Systematic Musicology. William holds a Master of Arts (Music Theory) degree from McGill University, where he wrote a thesis entitled Metre in the Music of Anton Webern (supervised by Christoph Neidhöfer). He also holds a Master of Music degree (Piano Performance) from the University of Ottawa, and he completed his undergraduate work at Calvin College with majors in philosophy, music history, and piano performance.

1. I wish to thank the Brandeis Music Graduate Student Society for selecting this paper for presentation at the 2015 Brandeis Musicology Conference, and particularly Charles Stratford and Jacques Dupuis for their assistance with the details of my participation in this conference. I also wish to thank Stephen Lett, Kája Lill, Charles Stratford, and my anonymous readers for their helpful comments on earlier drafts of this paper.

2. Allen Forte, “Foreground Rhythm in Early Twentieth-Century Music,” Music Analysis 2, no. 3 (1983): 240.

3. Wallace Berry, Structural Functions in Music (Englewood Cliffs, NJ: Prentice-Hall, 1976), 397.

4. Pierre Boulez, “Proposals,” in Stocktakings from an Apprenticeship, trans. Stephen Walsh (Oxford: Clarendon Press, 1991), 49.

5. The small wave of publications on this piece was apparently spurred by Edward T. Cone’s article “Analysis Today,” in “Problems of Modern Music. The Princeton Seminar in Advanced Musical Studies,” special issue, Musical Quarterly 46, no. 2 (April 1960): 172–188. For a brief discussion of these articles, see my “Metre in the Music of Anton Webern” (master’s thesis, McGill University, 2014), 15–18.

6. I leave this obviously fraught term undefined, since defining it with any precision would be impossible. I take Fred Lerdahl and Ray Jackendoff’s account of meter in A Generative Theory of Tonal Music (Cambridge, MA: MIT Press, 1983) as a good account of this and Danuta Mirka’s situation of this theory in a broader picture of metrical experience in conjunction with Christopher Hasty’s theory as described in Meter as Rhythm (New York: Oxford University Press, 1997) as a salutary re-orientation of this theory: see Metric Manipulations in Haydn and Mozart: Chamber Music for Strings, 1787–1791 (Oxford: Oxford University Press, 2009), in particular 13–30.

7. Carl Dahlhaus, “Rhythmic structures in Webern’s Orchestral Pieces, Op. 6,” in Schoenberg and the New Music: Essays by Carl Dahlhaus, trans. Derrick Puffett and Alfred Clayton (Cambridge: Cambridge University Press, 1987), 174–80.

8. Carl Dahlhaus, “Problems of Rhythm in the New Music,” in Schoenberg and the New Music: Essays by Carl Dahlhaus, trans. Derrick Puffett and Alfred Clayton (Cambridge: Cambridge University Press, 1987), 45–61.

9. Kathryn Bailey, “Rhythm and Metre in Webern’s Late Works,” Journal of the Royal Musical Association 120, no. 2 (1995): 251–80. The “late works” in question basically consist of those after op. 21, though the author considers opp. 17 to 20 in statistics she presents early in the study.

10. One obvious exception is Anton Webern, The Path to the New Music, ed. Willi Reich, trans. Leo Black (Bryn Mawr, PA: T. Presser Co., 1963).

11. One of the few discussions of meter in Webern’s writings is an analysis of his op. 28 String Quartet; see Appendix 2 in Hans Moldenhauer and Rosaleen Moldenhauer, Anton von Webern: A Chronicle of His Life and Work (London: Victor Gollancz, 1978), 751–56. I explore Webern’s notions of meter as revealed by this analysis in a forthcoming paper.

12. By “meter as notated” I mean not only the meter signature but also the bar lines, the metrical implications of the notation (i.e., as reflected in beaming conventions), and so on. By “musical surface” I mean the sounds of a musical composition independent of any metrical interpretation—obviously a theoretical construct, but helpful here.

13. Hasty, Meter as Rhythm, 4. Here, Hasty describes a view of meter against which he eventually seeks to elaborate his own. His description aptly summarizes the appearance of many of Webern’s scores, and indeed, analysts have described the latter’s metrical practice, particularly in his late works, along these lines. See, for example, Kathryn Bailey’s reluctant conclusion in “Rhythm and Metre,” 280. Rochberg describes this approach similarly: “The beat may remain as a referential point but has no other function in many traditionally notated scores.” See Rochberg, “The New Image of Music,” in The Aesthetics of Survival (Ann Arbor, MI: University of Michigan Press, 2004), 20.

14. Justin London, Hearing in Time: Psychological Aspects of Musical Meter, 2nd ed. (Oxford: Oxford University Press, 2012), 27–30. London reports several ways in which this threshold has been determined, but he is deliberately vague about where exactly this threshold lies for a number of reasons, including the nature of the experiments and the importance of the particular musical context in question.

15. Ibid.

16. The first song in Webern’s op. 4, “Eingang,” is an extreme example of this practice. According to the provided tempo and bar length, the bars in this piece should last over fifteen seconds. Webern does, however, provide dotted bar lines to divide up these bars.

17. Of the numerous possible examples, some such pieces composed around the same period are op. 4, nos. 1 and 3; op. 5, nos. 2 and 5; and op. 6, nos. 5 and 6.

18. I indicate sections in Example 1 by brackets above the score. I discuss the grounds for my formal divisions below.

19. Following Harald Krebs’s practice for grouping dissonance, discussed below, I indicate this figure with a “4” at the beginning of each of its manifestations. With the exception of the ostinato, the following gestures are not similarly marked since they lack clear cardinality.

20. Arguably the tremolo in the first section (bars 1–2) exhibits this practice as well, appearing as it does in two different metrical dispositions. Indeed, one might even be permitted to speak of a hemiola at play here. But in both cases, and as is characteristic for Webern, there is so little musical activity that mounting such a case is difficult.

21. Harald Krebs, Fantasy Pieces: Metrical Dissonance in the Music of Robert Schumann (New York: Oxford University Press, 1999), 31–33. Krebs points out that he borrows the term “grouping dissonance” from Peter Kaminsky. Krebs has also used music by Webern, including another movement in the op. 5 set, to illustrate his theoretical points. See “Some Extensions of the Concepts of Metrical Consonance and Dissonance,” Journal of Music Theory 31, no. 1 (1987): 110–11. Note that since no audible meter has been established, this would be an example of what Krebs calls “subliminal dissonance.” See Fantasy Pieces, 46–52.

22. This is particularly true for Opp. 5 and 7. I discuss op. 5, no. 1 and op. 7, no. 3 in my “Meter in Webern,” 44–59.

23. By “activation” I refer to the onset or “attack” of a sound. For good or for ill, this term reflects a notion of an underlying meter waiting for the composer to realize it in sound. Strange as such a notion may be, however, it will prove not wholly out of place here. The negation of this term plays an important role in this paper, but unfortunately no adequate term exists to differentiate between a simple absence of activation, agnostic as to the composer’s agency, and a deliberate omission of an onset, which invokes the composer’s agency. While the difference between these is significant, I believe that my intention will be clear according to the context of a given claim.

24. I have indicated these in Example 1 by means of brackets above the top staff. It has been suggested to me that the indication zögernd at the end of bar 2, which for the purposes of my formal divisions I am treating as synonymous with rit., in fact belongs with the im tempo found at the beginning of bar 3, and thus this longer, single indication refers not to a slowing of the tempo followed by its resumption but instead to a hesitation that nonetheless remains in tempo. This latter interpretation is occasioned by the 1949 edition of the piece, in which the only obvious evidence for the separation of zögernd and im tempo is an unusually large space between these two parts. A 1922 edition of the work, however, has zögernd and im tempo offset more obviously, and indeed in his orchestral version of this piece, Webern has rit . . . in place of zögernd. Moreover, the composer does not seem to use the indication zögernd im tempo in at least opp. 1–10, if at all in his published output; indeed, he uses im tempo only very rarely. The composer’s autograph of the work, which may be viewed at http://www.themorgan.org/music/manuscript/115911, confirms the above (intriguingly, Webern here has zögernd also in bar 1). I thank Professor Áine Heneghan for a stimulating exchange on this topic.

25. Besides the intuitive prudence of this approach, particularly in the absence of other salient formal markers, it should be noted that this means of differentiating formal units is confirmed across Webern’s oeuvre. The analysis of the last division as a coda results from an ambiguity between, on one hand, other markers of formal division and, on the other hand, the brevity of this unit relative to the others, as well as its misalignment with a bar line, as discussed in what follows. If anything, however, the tension this ambiguity presents accords well with this formal category.

26. I indicate non-activation in Example 1 with arrows, following Wallace Berry’s means of indicating thesis. See Berry, “Rhythm and Meter,” in Structural Functions in Music (Englewood Cliffs, NJ: Prentice-Hall, 1976), 301–424.

27. Notorious in this regard, for example, is Grosvenor Cooper and Leonard Meyer’s assignment of metrical feet to increasingly larger chunks of a piece up to an entire movement, as in their analysis of the first movement of Beethoven’s Symphony No. 8. See Grosvenor Cooper and Leonard B. Meyer, The Rhythmic Structure of Music (Chicago: University of Chicago Press, 1960), 203. Lerdahl and Jackendoff criticize this approach in Generative Theory, 25–29.

28. See, for example, the fifth piece of the op. 5 set.

29. These are indicated with downward arrows. Admittedly, two such punctuations seems like scant evidence on which to build a case, but in fact this remarkable feature is typical of Webern’s works in this period. I discuss off-beat punctuations in my “Meter in Webern,” 45–55.

30. Hasty, Meter as Rhythm, 7, 13, and 34.

31. Rochberg, “New Image,” 22.

32. Arnold Schoenberg, Coherence, Counterpoint, Instrumentation, Instruction in Form = Zusammenhang, Kontrapunkt, Instrumentation, Formenlehre, ed. Severine Neff, trans. Charlotte M. Cross and Severine Neff (Lincoln, NE: University of Nebraska Press, 1994).

33. Ibid., 215. Brackets are the editor’s.

34. Lerdahl and Jackendoff, Generative Theory, 17–18.

35. Mirka provides this distinction in Metric Manipulations, 23.

36. I have indicated these in Example 1 with thick downward arrows. Incidentally, all of these formal sections, with the exception of the coda, are two or four bars. Symmetries—in so far as a notion so natively visually may be applied to music—may also be found in pitch structures, for example in the violins in bars 1–2 and in the four-note descending line in bars 3–4. Webern’s use of such palindromes is well known and occurs with particular frequency in his later works; see, for example, op. 21 and op. 27. See also Kathryn Bailey, “Symmetry as Nemesis: Webern and the First Movement of the Concerto, Opus 24,” Journal of Music Theory 40, no. 2 (Autumn 1996): 245–310. Rochberg also discusses such symmetries in Webern’s music within the context of what he calls the “spatialization of music” (“New Image,” 17, 21–22), although he has in mind Webern’s later, “mature” works.

37. I have indicated these with “2”s, following Krebs’s practice for displacement dissonances: see Fantasy Pieces, 33–35. Admittedly, there is one on-beat eighth-note position activated in this group, that immediately preceding the climactic downbeat activation (indicated with an arrow in parentheses).

38. I discuss several other such situations in my “Metre in Webern,” 85–88.

39. David Lewin notes a similar procedure at the beginning of the second movement of Webern’s Variations, op. 27, linking it to metrical play in Brahms: see “A Metrical Problem in Webern’s Op. 27,” Music Analysis 12, no. 3 (October 1993): 349.

40. Arnold Schoenberg, “Today’s Manner of Performing Classical Music,” in Style and Idea, ed. Leonard Stein, trans. Leo Black (Berkeley, CA: University of California Press, 2010), 321. Presumably, the excesses in accentuation to which Schoenberg refers are of frequency and not stress of individual downbeat positions, but admittedly, both interpretations are plausible.

41. I discuss the first piece of the set in my “Meter in Webern,” 45–56.

42. Interestingly, while they abound in the op. 5, no. 4, Webern almost completely eschews expressive indications like ruhig and zart in this piece. What this indicates about Webern’s approach or interests here is unclear.

43. Moldenhauer and Moldenhauer, Anton von Webern, 335.

44. By “middle position,” I refer to the fourth eighth-note position of the bar, misleading though such language is in this context. The term is only accurate on a conceptual level.

45. I have indicated these activations on the annotated example again with thin downward arrows.

46. Here I employ terminology from Lerdahl and Jackendoff, Generative Theory, 30.

47. I have indicated these in Example 3 with thick arrows.

48. It is curious that, if this moment is the piece’s climax, the most intense dynamic level occurs before the downbeat, which is also the peak of the accelerando. Following Schoenberg, as discussed above, this may be a case of reluctance to “overaccentuate” strong positions, even at this moment of climax.

49. On this, see Mirka, Metric Manipulations, 22.

50. Carl Dahlhaus also observes the notion of metrical position as expressive—in the prescriptive sense—in the context of the op. 6 pieces. See Dahlhaus, “Rhythmic Structures,” 177.

51. William E. Caplin, Classical Form: A Theory of Formal Functions for the Instrumental Music of Haydn, Mozart, and Beethoven (New York: Oxford University Press, 1998), 15. Caplin calls this “before-the-beginning” or “after-the-end.”

52. While this passage is not representative of the texture of the rest of the piece in that it omits vocalists, it is nevertheless representative in other important respects. Moreover, the presence of a sung text sways metrical interpretation in ways that may not be helpful here, particularly when comparing this texture to the previous pieces.

53. This inaccuracy is most pronounced with the 6/2 signature of bar 6, whose internal division should, by convention, be two groups of three half notes rather than the three whole notes by which it is actually divided.

54. Interestingly, Webern doesn’t so limit himself in other works; see, for example, op. 23, no. 2; op. 22, no. 2; and op. 31, no. 1. It is difficult to understand the reason for this.

55. This possibility is more convincing in the context of op. 31, where Webern gives individual voices their own barring in the final movement. In most cases, these do not correspond with the other voices.

56. See my “Meter in Webern,” 74, 77, and 85. In the examples I discuss there, however, the meter remains the same for long sections, if not the entire piece.

57. One exception to this is the 5/4 bar (2), but since there is no convention by which 5/4 breaks down to either of its possible subdivisions, namely 3+2 or 2+3, this sheds no light on the matter.

58. This is not to say that gestures do not cross bar lines, as evidenced by the many slurs that do so (e.g., clarinet, bars 2–3) or by places where dynamic markings for a given instrument begin on the upbeat to a bar (e.g., bass clarinet, bars 2–3).

59. One also finds examples of the opposite in Webern’s works: for example, in several works he has beamings that cross bar-lines; see, for example, the Quartet, op. 22. This practice can, however, be found in earlier works (see Brahms’s Intermezzo, op. 76, no. 7, bars 22–23, and even Beethoven’s Piano Sonata in D, op. 10, No. 3, fourth movement).

60. Webern does employ both dotted notes and eighth notes later in the movement—indeed, in the bar that the choir enters. While the implication of the introductory attenuation of durational vocabulary is unclear, his reliance on a similar approach in the first movement of the Symphony, op. 21, may shed light on this.

61. See, for example, op. 28, no. 2; op. 21; and op. 24, no. 2.

62. See the second movement in particular.

63. Along similar lines, Rochberg writes of “the impression of immobility” in the context of the op. 27 Variations; see “New Image,” 22.

64. Webern does use a meter signature with numerator 7 as early as op. 4, no. 1 (“Eingang”). If anything, the use of such a signature so early in his writing suggests once again the longevity of his interest in pushing metrical conventions. The use of this meter was evidently increasingly common in the first decades of the twentieth century. In the 1911 first edition of his Theory of Harmony, Schoenberg states, “Only recently, 5 was adopted, and 7 is still completely strange to us,” while in the 1922 version, he revises this to, “it was only recently that 5 and 7 began to appear [as numerators].” Arnold Schoenberg, Theory of Harmony, trans. Roy E. Carter (Berkeley, CA: University of California Press, 1978), 204.

65.Later in the movement, Webern leaves both downbeats and upbeats of bars conspicuously unactivated, in part as a consequence of a sort of suggested but unfulfilled rhythmic symmetry. See, for example, in the choir’s first line (bars 14–22), where the downbeats of bars 16 and 18 and the upbeats of bars 17 and 19 are unactivated. I have yet to understand Webern’s intentions with this metrical configuration. It should be noted that the practice of leaving downbeats unactivated features significantly in the fifth movement of the Second Cantata, op. 31, evidence of the continuation of this practice late in his oeuvre.