The Balanchine Dilemma: “So-Called Abstraction” and the Rhetoric of Circumvention in Black-and-White Ballets

Abstract

Choreographer George Balanchine was known for rejecting the premise that his ballets were abstract. Yet, a closer look into his comments on abstraction reveals a greater degree of ambivalence toward the concept than previously noticed. His influential words found response in dance critical writing, where the term “abstract” continued to circulate, but was often applied in vague ways, such as “so-called abstraction.” This and other softened terminological variations formed an ambiguous collection of abstractive terms, like a vague word cloud around the dance concept. This article explores abstraction in Balanchine’s particular ballets, and makes a two-fold argument. On the one hand, by emphasizing the visual aspects of Balanchine’s compositions, we may uncover ways to untangle his dilemma about dance abstraction. Visual theories of “semantic abstraction” by Harold Osborne, and of “the gesture of abstraction” by Blake Stimson, may help us to understand the abstractive modes in several of Balanchine’s black-and-white ballets. On the other hand, whether discussed or not, Balanchine’s abstractive gestures have created powerful representational shifts in some cases. In particular, by examining the interracially cast duet from the ballet Agon (1957) as a visual case study, we may see how Balanchine’s rejections of the concept, amplified by critics’ vague terminological invocations of, or silence about, abstractive choreographic gestures, occluded the work’s participation in the discourse of abstraction. Simultaneously, unnoticed yet potent choreographic gestures of semantic abstraction may have promoted whiteness as a normative structure, one that relies on a hegemonic “bodily integrity” (as discussed by Saidiya Hartman). Such an analysis leads to a recognition that Balanchine’s abstraction could have been a subversive form of dissent similar to Kobena Mercer’s concept of “discrepant abstraction.” However, I posit that, as a result of the Balanchine dilemma and its influence, the interlinked gestures of an abstract nature that have not been recognized as such promoted the self-regulative structure identified by Bojana Cvejić as “white harmony.” Ultimately, a more specific and clear application of the term “abstract” in ballet is needed, as it can help to dismantle or disrupt the system of white supremacy operative in dominant ballet structures.

1. Introduction

I am so often told that my choreographic creations are “abstract.” Does abstract mean that there is no story, no literary image, at best a general idea which remains untranslated in terms of reality? Does it mean the presentation of sound and movement, of unrelated conceptions and symbols in a disembodied state?

(Balanchine [1951] 1992, p. 39)

Choreographer George Balanchine (1904–1983) often questioned whether abstraction was an appropriate aesthetic designation for his ballets. Throughout his career, he mused about the term, and in his recorded comments often seemed to reject it explicitly.1 The questions he posed in the quotation above show an understanding of abstraction in dance as a compounded, unclear term, with connotations of detachment from the body and its reality, as well as the inference of storylessness, even meaninglessness. Through a closer reading of his statements, however, I will propose that Balanchine’s attitude toward ballet abstraction is far more indeterminate than his rhetoric would at first suggest. Balanchine rejected the word “abstract” as a form of disembodiment or bodily masking, as well as non-narrativity. At the same time, he left space for some abstractive possibilities in his choreography. He recognized that abstraction may be part of the audience’s interpretation; furthermore, his choreographic approach involved certain abstractive methods, including the elimination of extraneous details. To denote this ambivalence, I refer to his complex attitude toward abstraction as “the Balanchine dilemma.”

Discussing the Balanchine dilemma today may seem only a matter of interest for the historical record about the processes and artistic intentions that have shaped his works. I propose, however, that this matter is of continued significance for contemporary considerations of specific Balanchine ballets that are still being performed across the international repertoire. The abstraction dilemma is also still relevant because of Balanchine’s towering influence over contemporary ballet training and practice: his seeming rejection of abstraction continues to resonate in dance criticism, scholarship, and ballet profession. For instance, this is how former prima ballerina of the Royal Ballet, Deborah Bull, expressed it:

We’ve all been taught not to say “abstract,” haven’t we, because of the famous Balanchine quote, and I would agree with that actually. I think that the human brain certainly seeks meanings—it can’t help it. So, non-narrative may be more appropriate.

(Bull, interview with the author, 17 May 2011)

Bull belongs to the generation of dancers who performed Balanchine’s works during the late 20th–early 21st century—much later than the time when his words were uttered. Although she did not train in Balanchine’s School of American Ballet (SAB), and was not a member of his New York City Ballet (NYCB), Bull—like many other ballet dancers—nonetheless learnt to refrain from employing the term, plausibly through the oral tradition of ballet teaching and international coaching.2 Bull’s words reveal how the Balanchine dilemma, with its international influence, had a far and lasting reach, leading to terminological changes, including the avoidance of abstraction as a ballet concept.

The hesitance to use the term “abstract” opened ways for alternative terms to emerge in ballet writing: as discussed below, replacements include non-narrativity, plotlessness, Balanchine’s own term “storylessness,” and several others, even non-representationalism.3 Yet, at the same time, Balanchine’s dilemma deterred scholars and practitioners alike from deeper considerations of abstraction in his ballets even if, as was the case all too often, few seem able to fully dispense with the term. Instead of analyzing “abstraction,” however, many qualified the term by putting quotation marks around it, or by questioning it euphemistically with softeners or prefixes.4 The dance critic Edwin Denby’s decision to call Balanchine’s style “so-called abstraction” is one of the representative and influential examples of this widespread, ongoing tendency. This propensity to euphemize abstraction has multiplied in dance writing, leading to a cacophonous effect that I denote as the “word cloud” of “so-called abstraction,” or a choral proliferation of unclear words around Balanchine’s ballets.

In this text, I also argue that this circumvention—the implicit unwillingness to probe abstraction in Balanchine’s specific ballets—is much more problematic than the term abstract itself. I propose that there are works in Balanchine’s rich opus that are unequivocally abstract, but in very specific ways, unrelated to the issue of disembodiment or plotlessness. In the first part of this article, aiming to untangle the Balanchine dilemma, I analyze the mid-20th century use of the term in ballet. Partly supported by the choreographer’s own writing, I aim to chart a mode of abstraction that does not challenge the integrity of the performer’s figure. I suggest that Harold Osborne’s philosophical theory of “semantic abstraction” in the visual arts and pictorial images more broadly (Osborne 1976a, 1979) may help us to understand the way abstract visual modes coexist with figurative representation and realism in Balanchine’s compositions (Osborne 1976a, 1979). If Balanchine’s leotard-clad “black-and-white ballets”5 are considered as dance examples of semantic abstraction, not only do we respond to the choreographer’s critique of abstraction, but are led to pay closer attention to his process of “eliminating” or suppressing connotative details. This analysis further draws on Blake Stimson’s theory of abstraction in photographic reportage, which points out the artistic embodied subjectivity in “the gesture of abstraction” (Stimson 2008). In this case, as it will be seen, woven gestures of abstraction by Balanchine and influential writers as his interlocutors, have promoted an ostensibly detached discourse. Yet, such detachment was still based on selections (of topics for observation and discussion) that were deeply rooted in the authors’ subjective “process of corporeal induction” (Stimson 2008, p. 69).

The significance of these unexamined modes of visual abstraction is explored in more detail through a socio-historical reading of Balanchine’s famous ballet Agon (1957), and particularly its originally interracially cast duet. It is well known that Balanchine’s predominant interest was in choreo-musical explorations of dance, particularly when working with Stravinsky’s scores (Jordan 2007; Joseph 2002). Yet, by looking at Agon’s visual aesthetic through the lens of abstraction, we may observe that more is at stake. The power of the compounded gestures of abstraction may be seen as abstracting away from the lived reality of the dancer’s body. This is to say, Balanchine’s choreographic and casting choices, and writers’ unchallenging responses, projected uncomfortable—racially charged—meanings onto dancers’ bodies.6 Whilst seeming to promote equality between Black and white dancers, Balanchine’s unnoticed gesture of semantic abstraction, that erased racial signifiers from the ballet in several later iterations, inadvertently promoted “white harmony”—a socially structured self-regulatory undercurrent that smooths over the potent frictions (Cvejić 2018).

As one of the key examples of semantically abstract ballets in Balanchine’s oeuvre, Agon initially strongly animated questions of racial equality, and used abstractive strategies for their subversive, dissenting potential (much like Mercer’s (2006) mode of “discrepant abstraction”). Yet, through the interlinked gestures of abstraction, Agon—perhaps unintentionally, but nevertheless effectively—avoided rattling ballet’s historically sedimented social structures. Supported by writers who euphemistically treated the ballet as a “so-called abstraction,” attention was deflected from Agon’s revelation of the unequal terms of participation, or the fact that its casting changes promoted the single uncontested and historically grounded “bodily integrity,” normative and white (Hartman 1997). Thus, it is still crucial today to dedicate attention to Balanchine’s abstraction dilemma, as it leads us to ask what is extraneous and withdrawn, and what is the “essential” content retained in the work.

2. Untangling the Balanchine Dilemma

Whilst such debates are not exclusive to dance,7 it is important to note that a dilemma about abstraction was present already in ballet before Balanchine’s career.8 Mark Franko has pointed out that one of the principal ways in which abstraction was invoked in 20th century Euro-American theatre, including ballet, had to do with the ways in which dance historicized its own past. One such manifestation was a questioning of mimesis, and shifting the terms of representation by breaking down narrative structures (Dickerman and Franko 2013, p. 38).9 Such a meaning may be distinguished from the related positions in the early modern dance traditions of the US and Europe, particularly choreographies that abstracted the dancer’s body by masking, visually deconstructing it, or challenging its three-dimensional representationality (Salazar-Sutil 2014; Andrew 2020). Rather, the mode of abstraction that resisted narrative unity and characterization was often paired with naturalistic presentation of bodies, highlighting the movement of “the self-connoting human form” (Dickerman and Franko 2013, p. 36). This strand of abstraction broadly related to non-mimetic experiments in modernist theatre and other time-based arts, in which “abstractionism” stood for “a decidedly episodic structure in contrast to the causally stringent development of events” (Meyer-Dinkgraefe 2005, p. 58).10 This meaning of the adjective “abstract” arrived through the tradition of balletic divertissements,11 but the direct term was popularized in writings of the 1920s by the writer André Levinson, proponent of “pure dance.”12 In his critical essays, inspired by Symbolist poetry, Levinson argued that abstraction could liberate ballet classicism from its baggage of mimetic meaning-making, whilst still allowing the form to retain its intellectual sublimity.13 For Levinson, then, abstraction was not particularly firmly related to the dancing figure, but was also an outcome of compositional principles that involve choreographic and musical organization of dancers’ interactions on the stage.14

Ballet abstraction in relation to the performer’s body—in dances that employ an unaltered human form as the carrier of movement—was considered by the writer and critic Rayner Heppenstall in the 1930s. Heppenstall observed that corporeal artistic forms, including dance and singing, could resolve the seemingly paradoxical relation of abstraction to materiality. For Heppenstall, the non-mimetic dancing of a naturalistic body, set free from literary libretti, principally referred to embodiment (of song and movement). The “human organism” uplifted as a sole artistic material and medium then produced a form of art that is “Abstract by the way of Concrete” (Heppenstall 1936, p. 129).15 Heppenstall’s contemplation of the dancer’s unmasked body as a dance material and a medium capable of polysemic meaning-making may be relevant for understanding of the Balanchine dilemma, but not quite sufficient to address the reservations concerning the appropriateness of the term “abstract.”

A related direct link between abstraction and Balanchine’s ballets was forged a few years later, in texts of the prolific dance writer Edwin Denby. In his early writings (late 1930s–early 1940s), Denby, a close ally of New York artists from the Abstract Expressionist scene, liberally applied the adjective to Balanchine’s choreography.16 Later, he shifted his position. As Andrea Harris demonstrates in detail, Denby ultimately tended to omit the word abstraction, speaking of Balanchine’s “dance works” that highlight the “direct enjoyment of dancing as an activity” (Harris 2017, p. 128). Harris emphasizes that Denby positioned Balanchine’s style as “pure” dance neoclassicism (Harris 2017, p. 128).17 I add to Harris’s analysis, with a further look into Denby’s writings, which show that when using the term abstraction, he often softened it in various ways, including the “so-called” syntagm.18 From the late 1940s onward, there is a silent enunciation: when writing about Balanchine’s ballets, Denby qualified “abstract” with quotation marks.19 It is not clear whether this is incidental to Balanchine’s own comments or informed by them, but by the 1950s–1960s, Denby rarely used the term in his essays. We may observe also that Denby’s explicit invocations of the word were carefully qualified, as seen in the celebrated essay “Three Sides of Agon” (1959). Here, Denby conceived of abstraction as an interpretive prerogative on the spectatorial side of the proscenium: “While the ballet happens, the continuity one is delighted by is the free-association kind. The audience sees the sequence of action as screwball or abstract, and so do I” (Denby and Cornfield [1959] 1998, p. 265).

Following later writings about Balanchine during the mid- and late-20th century, we may observe a similar trend. Abstraction is often mentioned but rarely analyzed, and made tentative through euphemisms. Particularly, the influential critical writing that had a broad reach and impact on the audience’s understanding and reception of Balanchine’s choreography often added quick-fixes such as quotation marks or prefixes.20 Such articulations allowed the idea of abstraction to remain in the discourse, whilst absolving the writers of using a term that the choreographer disliked. In The New York Times, for instance, Roger Copeland wrote about Balanchine as “the first major creator of truly ‘abstract’ ballets” (Copeland 1978); in The New Yorker, Arlene Croce positioned Balanchine’s work in the “so-called abstract ballet” as a “genre” vaguely related to “pure” dance.21 Such fuzzy, uncommitted uses of the term enveloped Balanchine’s opus with a vague word cloud that I call the discourse of “so-called abstraction.”

Considering the nebulous terminology, the choreographer’s doubts seem legitimate:

I said on another occasion22 that no piece of music, no dance can in itself be abstract. You hear a physical sound, humanly organized, performed by people, or you see moving before you dancers of flesh and blood in a living relation to each other. What you hear and see is completely real.

(Balanchine [1951] 1992, p. 39)

The connotation of “abstract” as dissociated from the performers’ reality seemed particularly illogical and uncomfortable for an embodied art. Such objections are not exclusive to Balanchine; many artists were apprehensive of terms imposed from the “outside” by spectators and writers.23 Indeed, we may agree that even the most radical examples of non-representational performing arts, such as Kandinsky’s theatrical experiments at the Bauhaus, in which the dancer’s body was occluded so that it became an abstract stage presence, still required living, breathing humans to carry out the event, whether they were visible or not.24 In Balanchine’s works, and particularly in his visually stripped-down black-and-white ballets (named for the practice-costume tones), dancers and their embodiment of movement arguably was highlighted to an even greater extent.25 The human figure delineated through tight-fitting body suits was further framed by the background-lit color cyclorama, often the only décor.26

Although he was not the only ballet choreographer to experiment with such a pared-down aesthetic, Balanchine’s monochrome designs are emblematic of his opus.27 Particularly in the period of 1950s–1970s, he proliferated a number of structurally focused explorations that shifted away “from lavish and referential costuming” toward “simple body coverings reminiscent of classwork [attire] which forefronted what the body was doing” (Preston-Dunlop and Sanchez-Colberg 2002, p. 22). Those works accentuated the dancer’s performing personality, rather than treating them as characters in a story. Led by musical scores, the choreographer deployed a full range of motion afforded by the medium so as to arrive at precise choreo-musical responses. He developed a style of ballet training that famously extended dancers’ range of musical eloquence28—a spectrum of soft and rounded, as well as angular, three-dimensional postural and gestural actions, expressed a dancing response to the unaltered tempi.29 His movement style involved speed, variability, and expansive qualities. Dancers famously travel across the stage as if to devour the space through movement, often with swift shifts of bodily orientation. Balanchine’s choreographic legibility thus deeply highlighted dancers’ subtle and surprising physical explorations and spectators delighted, as Denby suggested above, in a range of actions which originate deeply from the body.

Such a movement style, with its emphasis on intricate embodiment (expansiveness, variability, dancers’ musicality), makes the Balanchine dilemma very compelling. How could dances that highlight the reality of the performer’s neurologically and socially constructed body be abstract? Such reasoning may be behind Balanchine’s strong rejection of “abstract” as a “dangerous” term:

Music is often adjectived as being too abstract. This is a vague and dangerous use of words and as unclear to me as when my ballets are described that way. Neither a symphony nor a fugue nor a sonata ever strikes me as being abstract. It is very real to me, very concrete, though ‘storyless.’ But storyless is not abstract. Two dancers on the stage are enough material for a story; for me, they are already a story in themselves.

(Balanchine [1951] 1992, p. 40)

Yet, upon a closer reading of the whole essay from which this quotation was extracted, we can observe how Balanchine later obliquely acknowledged at least two possible manifestations of dance abstraction. The first is based on the idea that abstraction may emerge as an effect in the spectator’s imagination. This is similar to Denby’s previously cited point about Agon: as Balanchine put it, “the after-image that remains with the observer may have for him the quality of an abstraction” (Balanchine [1951] 1992, p. 39). The choreographer’s example was his own experience of listening to Stravinsky’s score Apollon musagète (1927): “I myself think of Stravinsky’s Apollon, for instance, as white music, in places as white-on-white” (Balanchine [1951] 1992, p. 39).30 Balanchine thus condoned abstraction, if it was outside of the work itself.

The second manifestation in Balanchine’s text, although more covert, in fact is even more revealing. In explaining his choreographic credo later in the essay, Balanchine revealed an affinity toward restraint and elimination as core aspects of his choreographic process:

My first real collaboration with Stravinsky began in 1928 when I worked on Apollon. I consider this the turning point of my life. This score, with its discipline and restraint, with its sustained oneness of tone and feeling, was a great revelation to me. It was then that I began to realise that to create means, first of all, to eliminate … I began to see how I could clarify by limiting and by reducing what seemed previously to have multiple possibilities.

(Balanchine [1951] 1992, p. 42)

Although Balanchine did not expand upon his terms “to eliminate,” “reduce,” and “limit,” it appears that he was interested in “restraint” as a way of retaining in the work that which is essential, and eliminating what he thought of as extraneous. Nevertheless, it is difficult not to draw the links between the notion of subtraction/reduction and the definition of the verb “to abstract” as the “action of taking something away; the action or process of withdrawing or removing something from” (OED, Abstraction).31 This definition is also linked etymologically with Balanchine’s rejected adjective, through the Latin, abstrahere.32 Then, Balanchine’s strong rhetoric notwithstanding, his own words invite further questions. For instance: in which ways do his acts of “eliminating” or reducing affect the choreographic work itself? This is where his argument—that his ballets cannot be abstract—begins to break down.

That “eliminating,” as Balanchine’s medium-specific methodology, produced visible or visual effects within the work can be seen in several ballets—including Balanchine’s own example, Apollo, and its evolution.33 Here I focus my argument only on the aspects of the visual aesthetic, rather than on choreo-musical or movement-compositional properties of the work. As dance and art historian Juliet Bellow explains, in response to this score Balanchine choreographed for Diaghilev’s Ballets Russes troupe the original Apollon musagète (1928), “a pure-dance ballet blanc,” originally performed in white “knee-length tutus worn by the three Muses” (Bellow 2007, pp. 241, 243). The original short-lived decor and costumes by André Bauchant involved well-selected referential elements that illuminated the Greek mythical story.34 Subsequently, Balanchine deployed a process of further reduction as the ballet underwent a series of revisions.35 The stylistic subtraction of narrative and decorative details can be traced through photographs: we may follow Balanchine’s gradual stripping down of the work to a minimalist mise-en-scène with black-and-white costuming (Figure 1). Ultimately, Balanchine arrived at an all-white, pared-down version, perhaps as the closest correlate to his “white-on-white” after-image (Figure 2). Recognizing that the lack of “literary stimulation” represented “a great challenge to the designer’s imagination,” Balanchine claimed that “there are the costumes and sets which can underline and help—with their composition of colour and form—to make the visualisation of music plastic and dramatic” (Balanchine [1951] 1992, p. 41).

According to Balanchine’s vision, then, elements seen as essential were carefully considered, and those remain in the ballet as performed today. Through subtraction, the choreographer arrived at a spare design that accented the choreo-musical legibility. Balanchine even left certain, carefully selected, referential elements that infuse the ballet with story-like properties: the props, including Apollo’s lute, and the Muses’ signifiers of lyre, mask, and scroll. The ballet itself is then a very good illustration of the Balanchine dilemma about abstraction. Balanchine’s verbal rejection of abstraction notwithstanding, the outcome of the process of abstracting has produced visible effects within the work and aligned them with definitions of abstraction. Apollo particularly may closely exemplify Susan Manning’s continuum of “mythic abstraction” whereby “choreographers represented dancers as inhabitants of abstract worlds” (Manning 2004, p. 118)—even with the aforementioned referential elements, which may prompt us to question its non-narrativity. But, if we continue to inquire into the duality of representation and abstraction, coupled with Balanchine’s propensity toward eliminating, we may be encountering a visual aesthetic of “semantic abstraction.”

3. From “Semantic Abstraction” to “Gestures of Abstraction”

“Semantic abstraction” is part of a broader taxonomy of abstraction in visual objects and pictorial depictions, developed by British art historian Osborne (1976a, 1979). Osborne’s theory could be understood to unite the perspectives of aesthetic philosophy and art history, offering a range of pictorial examples predominantly from European art history, design, and illustration. He used concepts from information theory to show how the process of pictorial abstracting retains and uplifts a certain referential content. In addition to semantic abstraction, Osborne discussed another broad, fully non-figurative, or nonrepresentational mode, which he called “syntactic/non-iconic.”36 Both modes individually, or at times blended, may project certain abstractive “expressive information,” in his parlance (Osborne 1976a, p. 242).37

As a discrete mode with several sub-categories, semantic abstraction is found in “the work of figurative or representational art, i.e., one which […] transmits information about some segment of the visible world other than itself” (Osborne 1976a, p. 243). Osborne introduces several sub-types of semantic abstraction (decorative, aesthetic, voluntary, selective) and illuminates the distinctions between them through a myriad of art historical and contemporary examples that today we would classify as the domain of visual culture, but that he broadly deemed as pictorial objects or “pictures.”38 The abstractive mode in images is an outcome of certain informational content deliberately left incomplete, or fully omitted: “when we say, for example that a certain […] scene is abstract rather than naturalistic,” semantic abstraction is gradational, and “equivalent to incomplete specification” about the object, subject, or event presented (Osborne 1976a, p. 243). This definition may remind us of the progressive reduction of details in Balanchine’s Apollo, which ultimately articulated a pared-down design that provided incomplete, yet sufficient, specification about the event depicted.

The concept of semantic abstraction can also help us to understand abstractness as a way of stripping down redundant detail, with Balanchine’s careful selection of several props as referential symbols left in the work. Osborne particularly addressed semantic abstraction’s ability to retain the duality of depiction and restriction of information. When artists aim to present “a new way of seeing the world,” they may provide “only a small proportion of the information.” This “selective [semantic] abstraction” promotes “a selective focusing of attention upon some items” in an aim to establish informational clarity by forming selected “associations of relevance” (Osborne 1976a, p. 248). This does not exclude representational artworks: Osborne discusses landscapes by Constable, and Post-Impressionist paintings by Cézanne and Monet.39

Osborne’s pictorial theory has its limitations as an explanatory tool for visual art and for the moving, embodied art of dance. However, his semantic modes offer us useful points of entry into the Balanchine dilemma, especially the choreographer’s comments about abstraction. As the reviewer of Osborne’s Abstraction and Artifice in Twentieth-Century Art (1979), Ralph A. Smith noticed, “Osborne forsakes historical and sociological criticism” of abstraction “in favour of an approach in which he examines the problems the artists themselves say they faced” (Smith 1981, p. 210). Particularly given Balanchine’s propensity for the gradual reduction of visual elements, semantic abstraction as a gradational mode is helpful in discussing the black-and-white leotard ballets. Although limitation of individuating details certainly seems integral to any art work as it is not “a replica of a real life,” Osborne proposed that we tend to denote a given work as abstract only when the “information is notably less complete than the medium would allow.” Such reduction could have been avoided by the artist “even though he may not have been consciously planning to abstract in this way” (Osborne 1976a, pp. 245–246).40

Thus, in semantic abstraction, the reduction is “voluntary” or “deliberate”—even if unintended, abstraction is there, and some information is highlighted as more pertinent by virtue of the process of elimination of other elements (Osborne 1976a, p. 245). Extending Osborne’s analysis to dance, in Apollo, we saw that Balanchine’s progressive reduction of referential details in decor and costuming (Figure 1 and Figure 2) was intentional, and greater than what was afforded by the theatrical medium.41 Furthermore, Osborne’s concepts might help to untangle Balanchine’s word “storyless” as a potential kin to abstraction, but also an oxymoron, because of human bodies’ polysemic connotations.42 Semantic abstraction speaks about such a duality: the abstractive restraint of informational details may accentuate particular expressive information. In political posters or caricatures, for instance, certain referential-contextual details may be omitted in order to emphasize an ideological message, such as to spin a news story (Osborne 1976a, p. 250).

Osborne’s discussion also, then, invites us to inquire into the function of “voluntary” abstraction. The function of suppressing informational detail might be didactic. Some individuating elements are stripped in devising a generic type of a species in botanical illustrations or other similar imagery. Likewise, details may be omitted in schematic and instructional images, engineers’ drawings, architects’ plans, or graphic design symbols that urge particular action. The human figure, for instance, is particularly stripped down in traffic or road signs (Osborne 1976a, p. 246). Yet, semantic abstraction’s aim equally may be ideological, as we saw in political caricature, or artistic, as in Fauvist line drawings by Matisse. The “selection has been done for us” in order to perceive the artist’s vision of the world, or “we are predisposed to find interesting the artist’s own manner of perceiving and presenting the things he depicts” (Osborne 1976a, p. 249).

On one level, then, we can also understand Balanchine’s aesthetic as a didactically-oriented form of semantic abstraction. Jordan (2007), Joseph (2002), and others have demonstrated how Balanchine’s choreographic responses to music guide the spectator to listen to the less noticeable nuances in the orchestration.43 In a stripped-down work, such as Apollo, if we follow Balanchine’s cues we may be more focused on music, and could even be predisposed to hear what Balanchine may have heard. However, we should not miss potential ideological aspects either. Balanchine’s words in the same essay cited above indicated a tension between his attention to the “reality” and the aim of his art:

My imagination is guided by the human material, by the dancers’ personalities. I see the basic elements of the dance in its aesthetic manifestations, that is, in the beauty of movement, in the unfolding of rhythmical patterns, and not in their possible meaning or interpretation; I am less interested in the portrait of any real character than in the choreographic idea behind the dance action. Thus the importance of the story itself becomes reduced to being the frame for the picture I want to paint.

(Balanchine [1951] 1992, p. 41)

Although he was inspired by his dancers as human individuals, for Balanchine the “real” world was just a pretext to explore an artistic affinity and to promote the correspondences between music and dance.44

Stimson’s (2008) theory of “the gesture of abstraction” may help us to understand the deeper ideological implications of Balanchine’s self-professed approach. To be inspired by human reality, but simultaneously to shift away from it, is to perform what Stimson describes as the “gesture of abstraction.”45 Writing in the context of aestheticized photographic reportage, Stimson points to the photographer’s click of the shutter. The moment involves two close, deeply embodied acts: the instance of the maker’s looking out into the world to seek (and be inspired by) an image, in a moment of selection. This is closely followed by the instance of “turning away” from the external event being captured, toward the medium’s apparatus (a look into the eye of the camera). The “turning inward of the photographer’s attention from the world outside to his own affective response” is a two-fold “gesture of abstraction” seen even in “realistic” images (a reportage) that purport to represent more objective, even universal experiences (Stimson 2008, p. 78).46 However, an artistic point made about the reality of the world (“[t]he cool, distant, and objective ‘over there’”) rather deeply encapsulates the specifically embodied moment of authorial selection (“the wooly, intimate, and subjective ‘in here’”) (Stimson 2008, p. 70). “[T]he process of abstraction itself” inscribes “the removal of understanding outward from any particular experience to a general, all-purpose explanation or figure or type”; at the same time, such a shift highlights the author’s subjective and sometimes parallactic “locus of affective or embodied engagement” in their concern with the medium (Stimson 2008, p. 69). In other words, the purportedly removed representations and generalizing abstractions obscure power structures and “the terms of participation” upon which they are based (Stimson 2008, p. 80).

How Stimson’s “gesture of abstraction” may relate to dance is demonstrated by Kélina Gotman, who drew upon the theory to analyze historical writings about the dance phenomenon of choreomania in the 16th–19th centuries (Gotman 2018). Gotman extends Stimson’s analysis, including the point that an author-observer (a photographer or an ethnographic writer, in her cases) may use its device “not as a vehicle for connection, but as a device for distancing othering, abstracting” (Stimson 2008, p. 80). Gotman observed ways in which chroniclers of choreomania projected ideological meanings (including pathologization) on events and participants they observed. In “describing, categorizing, and ultimately displacing vital events onto a discursive plane,” these writers strongly and invisibly inscribed “the paradox”—the gesture of abstraction-as-generalization alienates, “shifts and transforms the event itself just as it attempts to capture it” (Gotman 2018, p. 14). Gotman’s detailed study shows how gestures of abstraction in the dance context may produce strong ideological meanings, including generalizations and negative stereotyping—an issue to which I will return shortly.

By applying Stimson’s concept of gestures of abstraction to dance, then, we can understand also how both Balanchine’s choreography, and critics’ vague discussions of it, were based on selection and elimination rooted in their own subjective corporealities. Yet, such abstractive interventions may have ideologically infused or shifted the “reality” of embodied events and thus projected further meanings upon Balanchine’s semantically abstract ballets. The way in which Balanchine described his own approach aligned with the way Stimson theorized the “gesture of abstraction”: whilst capturing the reality of human dancers, he ultimately prioritized the concerns of his medium. Thus, when writer Roger Copeland observed that Balanchine’s “essence of the dance is […] the human body in motion; and according to the modernist mandate, anything extraneous—decor, characterization, costumes, story—should be progressively eliminated” (Copeland 1978), we need to ask further questions. If we accept that specific works by Balanchine are semantically abstract, we have to pay attention to the acts of subtraction. What is selected as the key content? What is minimized, or even taken away as “extraneous” within choreographic and critical gestures of abstraction? How do the terminological obfuscations and Balanchine’s “dilemma” play a part in this process? In the following section, I apply these concepts to the famous pas de deux from Agon (1957), where we can observe how Balanchine’s gestures, compounded and supported by the writers’ own abstractive acts, may have produced complicated, perhaps even unwanted, ideological messages.

4. The Agon Duet and the Implications of Interlaced Gestures of Abstraction

Am I a subject or a vessel, an agent or a channel?

Gutierrez (2018)

Much has been written on the subject of Balanchine’s much-loved modernist black-and-white ballet Agon, ever since the early reviews of the December 1957 NYCB premiere (Reynolds 1977, pp. 182–86). Much of that conversation has focused on the relationship between music and choreography. Nancy Reynolds succinctly described the ballet “in twelve movements for twelve dancers” as a choreographic response to Stravinsky’s “dodecaphonic score” (Reynolds 1977, p. 182).47 Agon as a choreographer’s “total collaboration with the composer” has stood across time as a “ground-shifting masterpiece” that “occupies a central position in Balanchine’s oeuvre” (Dunning 2007).48 The musicologist Robynn Stilwell analyzed the cellular compositional structure of the “compact, ostensibly non-narrative work lasting approximately twenty three minutes”; Agon’s ensemble dances, trios, solos, and a central duet, she argues, represent a unique melding of the serial score and “Balanchinian neoclassical technique modified by modern dance influences” (Stilwell 1994, p. 2). Following Stravinsky’s music, the ballet also hints at certain historical codes: European social dances of the 17th century (sarabande, galliard) are framed through the idea of “contest” (agon) in a title word of Greek origin.49 These references are not all overt. As we already saw, spectators such as Denby suggested that Agon’s playfulness motivates the viewers’ delight in surprises and interpretations of the “free-association kind,” including abstraction.

Writers often emphasized that Agon’s stripped-down monochrome design aided Balanchine’s choreo-musical visualization.50 Balanchine’s key conceptual collaborator Lincoln Kirstein described it thus:

Costumes are black-and-white practice uniforms, near nudity. ‘Production’ consists of execution alone, determined by the austere complexity of music in the serial method, patterns of cellular structure involving a continuum of variations on a twelve-tone set. […] The choreography projects a steel skeleton clad in tightly knit action, lacing a membrane of movement over a transparent net of contrapuntal design, which in shifting concentration on spare dancers gives an impression of intense drama.

(Kirstein, cited in Reynolds 1977, p. 185)

Although Franko and Harris detailed Kirstein’s complicated relationship with, and eventual rejection of abstraction,51 here we may observe how the writer developed an abstractive tone, even without introducing aesthetic labels. Even with the “drama,” Kirstein’s imagery of cellular, tightly knit steel skeletal action-structure with cellular membranes may remind us of Balanchinian abstract spectatorial “after-images,” as well as of the discourse of the “so-called abstraction.”

Other influential writers sometimes deemed the work as abstract, and many equally resorted to the actual phrase “so-called abstraction.”52 David Michael Levin’s text “Balanchine’s Formalism” (Levin [1973] 1976) observed a more particular abstractive elimination: the choreographic process “requires not only that he [Balanchine] suppress the theatricality of stage and costume, but also that he purge [emphasis added] the classical dance syntax of its theatrical allusiveness.” The “structure and content” thereby “become identical to the degree that each submits to the process of abstraction” (Levin [1973] 1976, p. 224).53 Writers’ statements complemented the artists’ own vivid abstract images: for Stravinsky, the “subject was figures and the relation of figures” and for Balanchine, the subject was “dancing, just numbers,” as Reynolds reveals:

‘Construction,’ incidentally, would seem to be the appropriate word for this ballet: said Stravinsky, ‘The mechanics (of contrapuntal music, the predominant texture of Agon) must be adjusted like that of a modern car’; while Balanchine described the ‘IBM ballet’ as ‘more tight and precise than usual, as if it were controlled by an electronic brain’.

(Reynolds 1977, p. 182)

From the sample of influential writings above, we see that statements by the artists and writers enhanced each other, promoting the elusive “so-called abstraction” discourse. This is to say, abstraction did not need to be always named for it to radiate through a dreamy collection of imagery of mathematical permutations, generalized organic forms, modernist machinery, and artificial intelligence.

Such statements strongly affiliate Agon with Franko’s axis of dance abstraction as “the reduction of art to the essence of its formal means,” which in turn promotes a “depersonalized (‘universal’) embodiment of subjectivity” (Franko 1995, pp. x–xi). Thus, through the so-called abstraction discourse, Balanchine’s works upheld what Gillian Lipton called the “rhetoric of detachment from the mundane” (Lipton 2020, p. 81). Yet, as performance and dance scholar Arabella Stanger warns us, it is important to pay attention to those very dances that are considered as “eminently formal” because they may reveal “social situations” that surround them. As she writes, the “spatial ideas concerning harmonic, liberatory, democratic, authentic, individuated, or anarchic forms of social relation can come to conceal” the inequality and violent dispossession on which they are systemically grounded (Stanger, forthcoming).

It is important, then, to take a closer note of the ways in which Balanchine’s depersonalizing imagery also projected content and ideological meanings onto the bodies of dancers. A closer look into archival videos and photographs shows how Agon’s minimalist design, in tandem with novel movement combinations, highlighted dancers’ bodies, and promoted visual effects. These are particularly striking when dancers move together, whether as a full group (in opening and closing sections), or in tighter ludic partnerships. The female dancers’ black bodices and male dancers’ white tops contrasted with the alternating tones of black tights for the men next to the women’s light tights. In movement, which included some classically-poised virtuosic interplays, the contrasting costume tones projected an additional layer in response to Stravinsky’s score; the checkered visuals could be understood as embodied analogues of transcoded black and white piano keys in a musician’s interplay.54 In one sense, the tiled patterns may again remind us of Osborne’s mode of semantic abstraction where pared-down visual information didactically draws the spectator’s attention to essential considerations of the artist (musical interpretation).

Particularly striking entanglements of black and white tones could be seen in the images from the pas de deux (Figure 3), an important section that we know Balanchine choreographed first. The famous photographs by Martha Swope show the ballerina in a deep arabesque penchée over her kneeling partner’s shoulder. Their leotard tops, in converse tones, are very close to one another, with the reversing contrast of dressed legs, more and less extending in opposite directions. Whilst the tessellated effects may have prompted “after-images” of an achromatic tone-blocked “construction,” as Reynolds said, this interplay equally highlighted the closeness of bodies and skin (as Kirstein suggested, the effect was “near nudity”). Thus, the haptic interplay is part of the duet’s expressive content.

This is particularly important given that the original casting was interracial: a white ballerina, Diana Adams, danced with a Black dancer, Arthur Mitchell. Before I take this analysis further, I want to remark on a problem that my line of inquiry presents. It is potentially dehumanizing to the dancers to equate their skin tones with non-human visual aesthetic elements, such as costumes and scenery. Yet, I follow Mitchell, who collaborated with Balanchine in the creation of Agon, and his discussion of the interracial casting in these very terms: he notes that the choreographer intended to extend the structural play of the black and white aesthetic to the dancers’ bodies:

There was a definite use of the skin tones in terms of Diana being so pale and me being so dark, so that even the placing of the hands or the arms provided a color structure integrated into the choreographic one.

(Mitchell in Mason 1991, p. 395)

Mitchell repeated this idea several times across the years.55 Speaking purely in visual terms, the effect described by Mitchell may be noticed in many other archival documents (Figures 3 and 5, but not as much in Figure 4, for example), so that we may observe how dancing in the inter-racial casting promoted intricate contrasts of skin tones framed by the dancers’ monochromatic attire.56

Balanchine never publicly discussed the casting or its aesthetic effects. Many critics described the duet in formal terms, but, importantly, did not address the visuality of the contrasts of skin that Mitchell described. Kirstein, for example, promoted the duo’s “impersonalization” of movement shapes, “of arms and legs into geometrical arrows” (Kirstein 1970, p. 242). Levin emphasized the disembodied “purely pictorial structural line,” with optical effects that accentuate weightlessness and immateriality: “it is as if their [duet’s] substance has magically evaporated!” (Levin [1973] 1976, pp. 232–33). Such descriptive choices by dance writers cited above, because of their emphasis on the geometrical, “purely pictorial,” and “impersonalized” structures of the “evaporated” body substance, recall the gestures of abstraction. Those are similar to what Gotman (2018, p. 14) observes: the writerly abstractive pivots shift the meaning of the dancing event whilst describing it. One of the more probing interlocutors at the time was Denby, who noted the potential significance of the interracial encounter, but ultimately did not inquire into it further:

Turning pas de deux conventions upside down, the boy with a bold grace supports the girl and pivots her on pointe, lying on his back on the floor. At one moment classic movements turned inside out become intimate gestures. At another a pose forced way beyond its classic ending reveals a novel harmony [emphasis added] …The fact that Miss Adams is white and Mr. Mitchell Negro is neither stressed nor hidden; it adds to the interest”.

(Denby and Cornfield [1959] 1998, p. 268)

Denby’s gendered language, in particular his reference to the dancers as a “girl” and a “boy,” also deserves further consideration.57 Here, though, I will focus on the racialized aspects of the visual aesthetic.

Agon, we should recall, was created at a moment of heightened attention to the interrelationships of Black and white bodies in the United States. Reynolds notes that “In 1957, the struggle for civil rights was very much in the news, and Balanchine (and Mitchell of course) were well aware of this” (Reynolds 2018). Melissa Hayden, one of the original Agon soloists, noted that “The first time you saw Diana Adams and Arthur Mitchell doing the pas de deux it was really awesome to see a black hand touch a white skin. That’s where we were coming from in the fifties” (Hayden in Mason 1991, p. 359). From his experience, Mitchell observed that Balanchine’s “mind was always involved with” the socio-political events, “but he was never overt about it” (George Balanchine Foundation 2002a).58 Whilst the choreographer never liked to discuss what his works may have expressed ideologically, we can still further inquire into the significance of racial representation as the duet’s integral aspect. Brenda Dixon Gottschild suggests that “it has become a tradition in revivals of this ballet that the role Mitchell originated is danced by a black male partnering a white female” (Dixon Gottschild 1996, p. 200). Yet, a closer look into Agon’s casting history does not quite confirm this hope—if the blackness of Mitchell’s role was considered crucial, it was not as long lived as the ballet itself. Mitchell exclusively danced the role until he left the company (still appearing a bit later, in the early 1970s in the NYCB production).59

From around 1970 onward, Mitchell’s role in the company was reprised by several white dancers: first Frank Ohman (Figure 4), then Peter Martins, Jean-Pierre Bonnefoux, and Bart Cook.60 The whiteness of the role persisted at least until the early 1980s, when Mel Tomlinson’s casting restored the interracial encounter which thus reemerged, and disappeared again from time to time, throughout Agon’s later NYCB history.61 Some writers at the time may have commented on an all-white cast, but many did not; Levin, for example, wrote about the ballet in the mid-1970s, and made no remark upon the casting of this role. Nevertheless, we have to observe how even the inattention to the racial dimensions of casting added further gestures of abstraction to the ballet. Here, for instance, is how the influential dance reviewer for The New York Times, Clive Barnes, wrote about the new cast in 1970:

This performance had some casting new to me. I have never previously seen Frank Ohman in the Pas de Deux (Arthur Mitchell’s usual part) where he partnered, efficiently yet characterlessly, Allegra Kent. […] Yes, I have seen the company dance Agon better—I really do miss the subtle force of Arthur Mitchell—but it stands up magnificently”.

(Barnes 1970)

Ballets inevitably change in various ways, including casting, particularly in long-lived productions such as Agon. Barnes’s ambiguous commentary may have been led by that custom. Yet, here is how Mitchell later reflected on the shift toward mono-racial re-castings of the duet:

Mitchell thinks that Balanchine was consciously making a political statement in the pas de deux by pairing him with Diana Adams. ‘Mr. Balanchine was politically aware of what was going on racially in America … I think one of the major things that’s missing now is the use of the skin tones as part of the choreography. My being black and Diana being very pale meant the color of the skin tones was incorporated into the choreography’.

(Mitchell 2001)

Over time, Mitchell maintained his point about the play upon contrasting “skin tones” as the integral choreographic aspect of this section of the work. This is, then, a moment to remember that semantic abstraction invites questions about acts of selection: what is kept as essential, and what is subtracted in artistic interventions? As Osborne said, in semantic abstraction the “[i]rrelevant detail is eliminated with the object of throwing into prominence a message which it is the purpose of the work to communicate” (Osborne 1976a, p. 246).

We already have seen how Balanchine, following his strategy of elimination, visually stripped the ballet Apollo gradually to arrive at its indispensable elements. As with Apollo, in wondering what was essential—or not essential—to Agon, we can follow its visual history to observe which elements have persisted. In this case, the contrasting black-and-white costuming never changed. We may therefore conclude that the retention of contrasting tones of black and white is an essential aspect of this ballet. This is quite unlike Apollo, in which the men’s black tights (Figure 1) were replaced to achieve an all-white idea (Figure 2). Following Mitchell’s point that the skin contrasts were originally used as an additional and integral visual-choreographic device, we come to ask: if the contrasts of black and white are important in Agon, how is it that the extended effect of dancers’ skin tones did not remain across all the versions?

If the skin interplay was later viewed as a redundant aspect of Agon, we already face an uncomfortable question about the discursive content projected onto the dancers’ bodies. Certainly, the question—whether the blackness of the male role was, as Copeland said, one of the “extraneous” elements—may seem simplistic or unfair. It may be that this was a pragmatic decision. Even if a Black dancer would have been preferable for Balanchine’s vision, were there suited dancers available to embody the role at that time in NYCB after Mitchell left?62 Yet, Agon was performed by Mitchell’s Dance Theatre of Harlem (DTH) from 1971, and it is plausible that guest appearances of Black dancers with the NYCB were possible.63 There may have been other reasons for the color-blind casting too. Perhaps Balanchine hoped to promote Mitchell’s body as unmarked in Agon.64 Even then, at least we have to ask: were those issues brought to the surface by the Black dancer’s casting with the white ballerina, including the issue of African-American civil rights, no longer relevant in later decades? If social change had made the casting of a Black dancer in that role irrelevant, then how is it that other roles in the NYCB productions of Agon have retained their whiteness?65

5. The Issue of Bodily Integrity in Ballet

In asking such questions, it is not my intention to reduce Mitchell’s contribution to Agon to his skin color. It is possible that this is exactly what those (white) dance critics in the 1970s were trying to avoid by skirting the issue of race. We know that Balanchine deeply valued the creativity and intelligence that Mitchell brought to his performances.66 For his part, Mitchell routinely praised Balanchine and Kirstein for uplifting his position in the company.67 However, we also know that he was the Black performer who “broke the color wall” in the ballet world (Kourlas 2018).68 Thus, not to observe the major effect of racial representation, and to minimize the relevance of critical silence around the casting changes is problematic. Although Agon in general, and Mitchell’s casting in particular, were a great hit with the liberal New York audience at the time,69 we also know that this approach was not made easy for Balanchine:

‘Can you imagine the audacity to take an African-American and Diana Adams, the essence and purity of Caucasian dance, and to put them together on the stage?’ Mr. Mitchell said. ‘Everybody was against him. He knew what he was going against, and he said, ‘You know my dear, this has got to be perfect’.

(Mitchell cited in Kourlas 2018)

If we consider Mitchell’s words, then Balanchine’s pressure (“this has got to be perfect”) during Agon’s creation may reveal the choreographer’s awareness of ways in which Black performance was “inevitably burdened by political circumstances,” as DeFrantz (1996) notes.

As we have seen, some key New York critics did not focus on identity politics in their descriptions of Agon; instead, they employed a humanistic, yet detached tone that promoted the equalizing connotations of the duet. However, critical responses outside the NYCB audience base exposed a racialized discourse, exoticization, even fully racist reactions.70 Claire Croft’s analysis of the company’s 1962 Soviet Union tour reveals that Mitchell’s neo-classically precise performance in the duet (which, if it explicitly promoted any heritage, this was French 17th-century court dance) was interpreted by the local writers as Balanchine’s shifting of movement style towards Africanist influences (Croft 2015, pp. 75–83). Although Dixon Gottschild has famously critiqued Balanchine’s appropriation of Africanist movement in various ballets,71 the Soviet reviewer’s point made here was specifically about Mitchell’s performance. Thus, no matter how balletically “perfect,” responsive, and talented a classicist, to a segment of the white audience, Mitchell’s body was marked by blackness and thus “carr[ied] only the information of its surface” (DeFrantz 1996).

Across the US, too, the pairing of a Black dancer with a white ballerina was not always viewed in formalist terms, seemingly free of racialized associations. As Lipton found, for certain US viewers “the entangled partnering between a black man and a white woman” was a “metaphor for the simmering racial and sexual tensions of the time” (Lipton 2020, p. 80). We know that in duets in particular, as dance scholars Preston-Dunlop and Sanchez-Colberg (2002, p. 77) explained, the “[p]roxemic behaviour, articulated in the psychology and sociology of interpersonal signs” promotes “spatial orientation of people to each other [which] carries meaning within a culture”. We have to remember, then, that Agon arrived on the tails of a long US history of, as Saidiya Hartman explains, white fantasies and anxieties about the “proximity and intimacy of black and white bodies.” After the abolition of slavery, many white Americans in both the North and South feared an “amalgamation of the races,” that is, “[e]quality was blamed for the increased likelihood of miscegenation” (Hartman 1997, p. 184). This fear certainly was not resolved by the time of Agon’s premiere, which took place ten years prior to the Supreme Court’s “Loving v. Virginia” decision that legalized interracial marriage throughout the US.72 It is possible, then, that the pas de deux, as a semantically abstracted event, promoted an undisclosed, and potentially provocative, discursive ideological narrative: a general story of a man and a woman that strongly animated certain white Americans’ “fixation on imagined sexual trespasses” (Hartman 1997, p. 185).73 The polysemic meaning in Agon, thus, also may represent a move toward normalization of the close proximity of black and white skin (Figure 5), as the above-cited quotation from Hayden suggested (see Section 4 in this text). Casting a Black dancer in a “noble” principal role74 was progressive in 1950s ballet, or as Dixon Gottschild (1996, p. 64) suggested, “near revolutionary.”75

These are ideological meanings and “stories” that Agon initially plausibly promoted through semantic abstraction, with Balanchine’s visually pared-down staging highlighting the particular materiality and reality of the dancing bodies. In light of that, Balanchine’s casting of Mitchell initially promoted what Stimson called a moment of collectivization, yet one that is in fact “withdrawn from the social field and attached to a putative theatrical element” (Stimson 2008, p. 75). This is because Agon, with its original casting, could be seen as an extension of Balanchine’s and Kirstein’s vision of institutional equality articulated many years earlier. Writing in 1934 for Edna Ocko’s leftist publication), Kirstein described “the destruction of the proscenium arch” where the use of Black dancers “in conjunction with white dancers” would lead to “the replacement of an audience of snobs by a wide popular support”—this was “all part of Balanchine’s articulate program” of ballet modernity (Kirstein in Garafola 2002, p. 58). Thus, even if we consider Mitchell’s role as tokenistic, since he was the only Black dancer in the production, we can recognize seeds of the vision of equity in ballet in Agon.

In Balanchine’s original casting, haptic elements, such as the dancers’ close proximity and the emphasis on their touch, played an important aesthetic and socio-cultural role—arguably, they were part of Agon’s “drama,” tensions, risks, and anxieties that writers such as Kirstein hinted at when discussing the impersonalization and witticism of choreographic cellular structures (cited in Section 4 of this article). Then the abstractive discourse of “just numbers” and IBM machinery also might have been a veil under which to smuggle the subversive vision. In this respect, Agon held a potential of “discrepant abstraction”—a dissenting mode that has “disruptive agency” and power to decenter institutional imbalances, as the art historian Kobena Mercer articulates (Mercer 2006, p. 17). If the interracial casting was retained consistently, as a truly essential aspect of the ballet, then Agon’s abstraction would plausibly illuminate an “inherently multi-directional” and far reaching mode, as defined by Mercer: ”[w]hen viewed in the round,” he writes, abstraction holds a “defining quality” of “openness” that can amplify “the dynamics of inter-cultural exchange” (Mercer 2006, pp. 7, 9). Balanchine’s ballet also would have materialized DeFrantz’s observation about ballet’s unique power to reverse stereotypes about Black performers: “Ballet locates its aesthetic power in the refinement of gesture away from everyday bodies and politics,” and it therefore has significance for the normalization of “the black male body to the degree that the idiom unmarked the lingering minstrel persona” (DeFrantz 1996, p. 11).

Yet, even if unintentionally, Balanchine carried out a different gesture of abstraction. In reassigning Mitchell’s role to white dancers, he stripped Agon of such a potential to disrupt and subvert dominant structures. As in Stimson’s gesture of abstraction, in turning away from the “real” world of his inspiring dancers as socially located people, Balanchine shifted his focus inward, towards his compositional medium. Balanchine’s selection of white dancers to replace Mitchell implicitly disclosed that the questions provoked by the blackness of the male key role can be withdrawn, and that space was opened for normative (white) dancing bodies to occupy it. This decision was strongly protected by writers who exercised their own gestures of abstraction, in essence immunizing Balanchine’s ballet from further probing into its racial implications. Here, for example, is how Charles Joseph defended Balanchine’s prerogative to deal only with aesthetic “visual impacts”:

The choreographer’s provocative coupling of Mitchell and Adams, however, should not be misconstrued: Balanchine had no desire to use dance as a public soapbox for defiant political statements. Rather, he had always delighted in creating jarring visual impacts. More than anyone, he realized that such a stark contrast would viscerally intensify the visual polarity that Agon often sought. It was just one further aspect of the “contest” that went to Agon’s core. Whatever racial implications others would read into the pairing of Mitchell and Adams did not seem to concern him one way or another.

(Joseph 2002, p. 258)

The white re-casting was, as Joseph describes it, a non-event that could be carried out freely because references to particular human stories were just a non-essential “frame.”

Joseph’s statement has to make us pause and think about the question asked by the contemporary American choreographer Gutierrez (2018), in his article “Does Abstraction Belong to White People?”: “Who has the right not to explain themselves?” The choreographic choices and withdrawals in Agon also position this work in line with Stanger’s argument that the “liberation of ballet from its old-world stability” in “Balanchine’s case depended on both the availability and the absence of Black dancing bodies as instruments for white (self-)invention” (Stanger, forthcoming).76 Furthermore, Balanchine’s short-term attention to the political implications of his choreography might remind us of the known habit of whiteness, denoted by the choreographer Adrienne Ming as the “white antiracist tourism”: interested allies emerge at a politically charged moment, only to revert “back to normal” when the events settle (Ming 2020). Was Balanchine’s recasting a way of reverting to the habit of casting white dancers—a shift back to “normal," a phenomenon also observed by Rebecca Chaleff in US postmodern dances? Chaleff suggests that dance abstractions of the Judson Church choreographers depended upon the unmarkedness of white dancers and their “presumptive ontological facticity” of being “unseen, unremarkable, and, above all, ordinary” (Chaleff 2018, p. 72).77 Black dancers, however, fought the historically tangled legacy of African-American performers’ struggle to shed away, as Hartman demonstrated, the “fungibility” in the representation of their bodies.78 Similarly, Agon then, through artistic acts of Balanchine and those who wrote about him, was positioned within a theatre dance lineage of abstraction that acts as a technology of whiteness, in Gutierrez’s sense, in that it reveals “[t]he ways that power, access, possibility, and the ability to dream into the future were constricted by the happenstance of economic or racial contingencies that defy reduction into shape and color” (Gutierrez 2018).79

A new covert ideological message thus emerged and solidified through interlinked gestures of abstraction that cleansed the discourse, moving it away from the controversial issues of racial tensions in the society into a more ‘neutral’ territory of formalist concerns. Here, we may remember Stimson’s analysis of interlaced abstractive gestures, and wonder about further ways in which they also encapsulate power relations, given that they are “attained by a particular relay or switch that defines the terms of participation” (Stimson 2008, p. 70).80 As Hartman showed, even innocuous and “gentler forms” of entertainment were rooted in unequal terms of participation, disguised through a “veil of enchanted relationships” (Hartman 1997, p. 42). In fact, “a neutral or egalitarian guise” (and we may think here of Balanchine’s close collaboration with Mitchell) all too often involved the embodied viewpoint of the white maker as the only one with the normative properties and an implicitly assumed, unchallenged “bodily integrity” (Hartman 1997, pp. 25–26). The gestures of abstraction also protected a normative reset: the unquestioned body-switch and the restoration of the hegemonic white bodily base was not questioned by the writers as influential spectators.

The reset to the ballet default-setting in Balanchine’s Agon thus strongly signals ballet’s capacity to preserve its hegemonies by self-regulating. This is also known as the concept of the “white harmony,” articulated more broadly in contemporary dance context by the performance philosopher Bojana Cvejić. Cvejić characterizes harmony as a process of purification that is carried out under the flag of social cooperation, often unnoticeably integrated within corporeal/somatic and structural and formal organizational aspects of choreography (Cvejić 2018). Such self-regulating harmony “is white, because it white-washes its materials”:

White harmony is over-determined. First of all, it is a gesture of inclusion that remains in denial about violence and exploitation, as if conflict was just a matter of difference. But, second, whiteness or blankness is a result of specifically modernist procedures of harmony such as purification, immunisation, and neutralisation. Taking the stance of not knowing and starting from scratch, enables sometimes the choreographer to white-out nuances and whitewash narrative contexts.

(Cvejić 2018)

I borrow Cvejić’s concept, and apply it to Agon because we could follow the language of white harmony in writers’ images of alike mind-bodies meshing (Reynolds); in Joseph’s withdrawal of attention to the issues of race; or in the “novel harmony”, to borrow Denby’s term (see Section 4 in this article). The accumulated discourse of the vague “so-called abstraction” thus immunized Agon, pivoting away from the frictional and tensile representations of the real world.

Although in this case unnoticed and undiscussed, or unchecked abstraction operated as obfuscation, dance abstraction is not a problem in itself. We know of strong examples where abstraction does not belong to white people (to borrow Gutierrez’s title), and rather helps choreography to “creak with polyvocal alternatives” (Mercer 2006, p. 7). The decentering force of dance abstraction has been observed in choreographies by Akram Khan, Bill T.Jones, Reggie Wilson, and Tao Ye, for instance (Mitra 2015; Bill T. Jones in Brody 2008; Wilson 2019; Yapp 2020, pp. 489–504). As we have seen, even whilst he doubted the abstract adjective, Balanchine’s gestures of semantic abstraction in Agon (as seen in Figure 3 and Figure 5) held a similarly strong potential, as they initially articulated a challenge to ballet’s institutional inequalities. The Balanchine dilemma, however, stifled the discussion of abstraction, allowing the abstractive gestures to operate below the surface, on an implicit assumption that the dancer’s stripped-down image can be generalized and unpolluted by the historicity of their body. Rather than simply accepting Balanchine’s terms and stipulations, critics could have probed into semantic mechanisms that involve the selection and elimination of informational detail—in this case, the elimination of important racial and social questions that resulted from the color-blind casting changes.

As Hartman suggested, only by looking at the less extreme relationships and “disassembling the ‘benign’ scene can we confront the everyday practice of domination, the nonevent, as it were” (Hartman 1997, p. 43). Along those lines, by thinking in terms of the visual concept of semantic abstraction and its attention to what is taken away and what is retained, are we led to read the stage, or as Gutierrez says, to look at “who is not there” (Gutierrez 2018). Avoiding the vague, euphemistic discourse of “so-called abstraction” we may also avoid the pendular force of balletic re-centering that erases diverse experiences, or refuses to confront the politics of exclusion operative in dance companies and institutions. This is why, upon closer reflection, it seems much more “dangerous” (to borrow Balanchine’s word) to brush over abstraction as a dance word. Our avoidances may support a long-term immunity of the discriminatory dance (ballet) institutional apparatus that perpetuates hierarchical modes of embodied representation.