For dancers, choreographers, and audiences, the preservation and reperformance of dance works is a difficult and exigent practical concern. Dance is perhaps the last of the art forms to rely heavily on kinesthetic and visual memory for repeated performances of works. These are unstable means and subject to question. But another source of the problem is general unclarity about what it is essential to preserve. Specifically, what serves as an acceptable criterion of identity on the basis of which a performance can be said to be a performance of a given work? Although we do not intend to provide a definition of dance, we will pick out and elucidate some of its most important properties and their relation to the dance work. The examples we will use are generally recognized by dancers, critics, and audiences. They provide a reliable database, though they do not exhaust the possibilities by introducing those cases on the extreme experimental fringe of the art form.

     One popular answer to the identity question is that the work is determined relative to a stable choreography or plan of movement, construed as the specification of a sequence of movements. The performance is the work made manifest. As we will argue, neither this approach nor the more sophisticated version of it offered by Nelson Goodman (1968: 211-18) is an adequate criterion of identity for the dance work of art. Choreography, properly construed, is important to determining the identity of a dance work, but these theories have an inadequate notion of choreography.¹ We will argue, first, that choreography construed as the plan of a work often includes as integral, elements which on the proposed criterion would be ruled incidental. In addition, the proposed criterion fails to take into account the very important role played by style in determining the identity of the work. The proposed criterion is too weak, as it fails to include alleged "incidentals" and rests on a thoroughly inadequate notion of dance movement. Finally, it can be shown that, both with regard to works which are viewed as traditional and to works produced by non-standard choreographic processes, the proposed criterion is too strong. In the final section of the paper, we propose to explain some of the ways in which choreography properly construed and dance scores actually work.

I. The Dance Process: The "Incidentals"

Traditional aesthetics has tended to isolate the art object from the creative process. This separation of the work from the activity which produces it appeases our desire for a rational, simply constructed theory about art. It offers a theory which in fact deals only with the termination of the artistic process in a product which is the object of the beholder's contemplation. It is usual to claim in addition that the art object of contemplation may be manipulated by the members of society, but that the additional function does not change the basic status of the work of art as object of contemplation.

     Dance does not lend itself easily to this model. A good deal of dance has been and still is bound to special settings and use-contexts. More importantly, dance is a process-art. At least part of the creative process is crucial to the identification of the work. As a result, with dance it is difficult to determine the exact extent to which the creative process may be taken as irrelevant. In this sense, dance is like the other performing arts. But in dance, the problem is complicated by the simple fact that some of the elements commonly considered incidental in the identification of a work of music or theater are or can be integral in the identification of the dance work.

     It is quite true that we ordinarily talk as if there were some central notion of the dance work—the movement—with respect to which such factors as stage effects, music, and specific performers are incidental. Closer inspection shows that this view is not so plausible as it seems. The main focus of choreography is movement, but in a given work, the quality of the movement may be drastically changed by variation of one of the alleged "incidentals," and this is often not a matter of indifference to the choreographer.

     We are now past the age in which "stepping to the music" or "interpretation of the music" was a primary determinant of the dance work. Musical constraints are the first to be done away with when dancers experiment with movement, and works are often done in silence. Some works, e.g., Twyla Tharp's "The Bix Pieces"² have a variable musical element. She created the movement for the piece to the music of Haydn. The final work was performed to the music of Bix Biederbecke, though during the performance an actress explains the motivating role of Haydn's music.³ Still, such examples show only that music is sometimes a variable element and that a work in progress may search around for the right music, not that music is always incidental in the final product. Certain traditional and modern pieces would lose their identity if the music were changed, even if another accompaniment could be found which did not do violence to the planned movement. This is perhaps most obvious from the side of the dancer, for whom the music is an important motivator. But even for the audience, "Romeo and Juliet" needs its music to be the dance it is, and Eric Hawkins's "Cantilever" would cease to be "Cantilever" if performed to a Bach fugue. Denotational meaning is often carried in the music or the music-dance combination.

     Costuming is often as important as movement in creating the properties of a dance work. The creation of movement qualities by costuming seems obvious in explicitly multimedia works, such as those of Alwin Nikolais;⁴ but their importance to classical works often comes as a surprise. Balanchine's "Chopiniana," a formalist "version" of "Les Sylphides," performed in leotards instead of the classic long tutus, demonstrated that the properties created by the movement-costume interaction were necessary for establishing the work's identity. It is fortunate that Balanchine titled his work "Chopiniana," not simply "Les Sylphides" (Levin 1973: 29-48).

     Lighting may also prove important in creating the properties of the work, not only in "setting the mood," but in defining what is seen. The particular lighting scheme is incidental to many works. But a work such as "Winterbranch" by Cunningham is planned to have lighting set by chance, so that each performance provides new variations. Moves often begin in light and dissolve into darkness, or the reverse. It is a source of constant surprise to the dancers and audience which parts of the movement are revealed. Performed on a fully lit stage, "Winterbranch" would loose the ominous quality that characterizes the work. The dance would not be "Winterbranch" at all.

     Finally, individual performers are of more than incidental importance. It is plausible to claim that we have "Frontier" by Martha Graham only if it is performed by Graham or by some surrogate explicitly designated by her. There is some question of whether anyone but Pavlova can be considered to have performed "The Dying Swan." Other examples tend to show that it is not so much the specific identity of the performers as an important consequence of what dancers are chosen that makes the difference—the style of the performance.

     Dance style is a double-aspect, two-level phenomenon. Both levels of style involve a distinctive choice in "spatial imagination," the ability to conceive of movement to fill a given time-space interval. A general style choice, once made, produces what we will call a "spatial vocabulary" and a distinctive pattern of kinesthetic motivation. A spatial vocabulary is an inventory of acceptable positions and position sequences; and kinesthetic motivation is best described from the dancer's side as a sense of the pattern of the movement flow, the originating impulse, the stresses, and transitions. From the audience's point of view, kinesthetic motivation is the quality and direction of movement thus produced. Many modern dance groups use roughly the same inventory of movements or spatial vocabulary. The vast differences in movement quality which make these groups different from each other stylistically stem from their very different ways of motivating the movement. A cohesive vocabulary and motivation choice produce what we call style₁ constraints. Contrasted with these are the style₂ characteristics produced by the dancer's individual manner of execution at the performance level. Style₂ is both an articulation and a further development of style₁. Every dancer has a personal style₂, an individual internalization of general style₁ constraints; but we tend to notice its presence only when personal style₂ is exceptional. Both levels of style have a performer and an audience aspect. The performer experiences the style₁ constraints and his style₂ implementation of them as kinesthetic phenomena. The audience apprehends both style₁ and style₂—as the dancer does not—as qualitative external results of the dancer's activity.

     In the case of "Frontier," a performer's style₂ choices have become integral to the work's identity. Ordinarily, though, it is the style₁ constraints of the work which are crucial to its identity. A 1975 performance of José Limón's "The Moor's Pavane" by Nureyev, Fonteyn, Kain, and Bortoluzzi is a case in which style₁ constraints were violated. This work was not officially "wedded" to particular performers. As it happens, certain modern dance steps were changed to ballet steps. (Some modern dance turns were changed to pirouettes, and there were no pirouettes in the original 1949 performance.) But even had the steps been kept intact, it was felt that the work had changed sufficiently in other respects to eliminate it as a performance of the work:

The larger issue is whether the distinction between modern dance and classical ballet is worth preserving. The more immediate question is whether ballet stars with no previous training in any modern dance technique and no apparent conviction in the ideology that sparked the revolt of modern dance against the ballet should transform (perhaps destroy) such works as "The Moor's Pavane." (Kisselgoff 1975)

The real question is not how the dancers were trained or what they believe, but how they performed and whether they should not perhaps have retitled the piece, given that they destroyed its distinctive style₁.

     This is exactly what Tharp did when she adapted "Deuce Coupe," which was originally choreographed for her company and the Joffrey ballet, for the Joffrey alone. She retitled the piece "Deuce Coupe II." This work is related to but distinct from the original:

So in 1975 Tharp produced "Deuce Coupe II," "customized" just for the Joffrey. It's a far more masterfully constructed ballet than its predecessor, but some of us feel that what we have here is the refined, slightly frosty elder sister of a bright, impulsive untidy child we loved (Jowitt 1976: VI, 11.)

Tharp anticipated the "Moor's Pavane" problem and solved it in advance. She eliminated the style₁ element specific to the training of her dancers and restricted the piece to the style₁ of ballet, a vocabulary within which the Joffrey members could motivate and execute the movement.

     These examples are not meant to be exhaustive. They are meant to show both that something more than the object-model is needed for dealing with the dance work of art and to suggest that a simple appeal to "the movement" is likely to prove inadequate for purposes of establishing a work's identity. The crucial role played by such elements as we list cannot simply be treated as an extreme example of the importance of display conditions. A painting in darkness is still the painting it was; a dance performance with a change in one of the alleged "incidentals" may not be a performance of the work at all.

II. The Work and Its Identity

Criticism of traditional models is doubtless the first step in the creation of a process-work model. It has been suggested (Price 1970) that we subsume both the performing and the non-performing arts under a new category of activities, some of which admit of performance. By claiming that all art has an activity component, we are able to understand the relation of that process to the final work. This kind of process-model creates special theoretical problems in the case of dance, however. Style, stage elements, and performers are sometimes allowed to vary widely from performance to performance. By admitting that the elements involved in producing a performance and the activity of the performance itself are invariably relevant to determining what work we are seeing, we should place ourselves in grave danger of losing the notion of "the work" as some central core which persists despite differences due to the increments and variations incidental to a particular performance. This notion of "the work" is useful and close to common sense; no one wants to lose it entirely. To save it, we need to know how much difference makes a difference.

     It is in the search for clear and invariant criteria of identification of the dance art work that Nelson Goodman suggests that we begin by separating the arts into those which are "autographic" and those which are "allographic."

     Let us speak of a work of art as autographic if and only if the distinction between the original and a forgery of it is significant; or better, if and only if even the most exact duplication of it does not thereby count as genuine. If a work of art is autographic, we may also call that art autographic. Thus, painting is autographic, music non-autographic, or allographic (Goodman 1968: 113):

In sum, an established art becomes allographic only when the classification of objects or events into works is legitimately projected from an antecedent classification and is fully determined independently of history of production, in terms of a notational system. (Ibid., 198)

The relevant criterion of identity for the dance work would thus be the choreography as a plan of movement recorded in an adequate notation system. The criterion would then allow us to pick out the "compliants," or performance events, which constitute the work. Choreography so construed and adequately notated will, on Goodman's view, allow us to identify the work "independent of history of production." Strictly speaking, Goodman does not require that the notated score suffice for "retrieval" or production of a performance of the work. But there is every reason to believe that certain kinds of difficulties in retrieval point to parallel difficulties in providing a record adequate for identification.

     Goodman's proposal has the obvious drawback that it offers no ready way of dealing with the integral "incidentals" we described in section I. And, as Goodman should be the first to admit, if there is no significant and

generally applicable consensus about what is and is not incidental, then an art form just is not amenable to notation, which presupposes such a distinction. Notation clarifies and reinforces antecedent distinctions, but it does not create them. (Goodman 1968: 121)

Let us assume, for the moment, though, that we could secure sufficient agreement on the importance of "incidentals." It may yet be the case that some important element of the art form itself causes it to resist notation, as, on Goodman's account, painting does.

     On this view, the dance score is seen in obvious analogy to the score of music, usually produced by the composer, which is then in turn used to produce or identify the musical events which are performances of the work. The score itself can in theory be uniquely determined from a single accurate performance. It is in fact the uniquely recordative function of the score which makes it an adequate criterion of identity for the work. The problem is that there is in most cases no analogous production score for dance, even in the traditional cases for which Goodman's suggestion seems most apposite. Sometimes the choreographer does a great deal more than draw up a definitive plan of movement—and sometimes a good deal less. If we could simply dispense with the choreographer's activity for theoretical purposes of identification and just use the notator's score, all would be well. But choreography, correctly understood, is one of the main criteria for identifying a performance as a member of the class of allographic compliants which counts as a performance of the work. The problem is that choreography, as it is in fact done and as it determines the identity of a work, will not produce the kind of compliants that Goodman (and the notator) want. The second problem is that there are difficulties, even when the choreographer restricts himself to specifying precisely defined movement, in producing a notational system for scores which can then serve the dual purpose of recording and retrieval—and that there is considerable doubt about whether they are adequate, even for identification. The difficulties are endemic, not merely practical.

     The problem initially seems to be that the common-sense basic unit of movement construction, the step, is ill defined. Choreographers deal in spatial vocabularies and kinesthetic motivational systems. Some choreographers, like Balanchine, graft into established systems. Others arrive at a distinctive system by experimentation on and with dancers. In some cases, such as that of the autonomous choreographer, who creates and dances alone the works he designs, the dancer completely shapes the vocabulary. The more ordinary situation is that described by Paul Taylor in "Down With Choreography":

More often than not, the kind of dance we work on together turns out to be dependent on these different dancers as individuals. Sometimes their limitations are as interesting as their strong points. These eight dancers are not exactly like tubes of paint with which to cover the canvas of space. Not exactly. They have character and personality which they assert. They have individual traits, and just when you think you know how to handle them, they change. (Taylor 1965: 91-92)

For the choreographer, the natural proclivities of his dancer are much in evidence. They constitute and shape his work as it meets the beholder, and they influence his choice and implementation of spatial vocabulary. Even choreographers in such stable dance forms as ballet have consistently allowed their dancers to shape their vocabularies.

     The items of spatial vocabularies are the real basic units in terms of which choreographers compose and in terms of which their works are interpreted for performance. There is no one-to-one correlation between spatial vocabularies and sets of steps. The unwieldy nature of the connection is illustrated by the phenomenon of stylistic borrowing. Heuristically, we often say that it is steps which are borrowed. Borrowing produces a number of effects ranging from novelty to the choreographic in-joke. But we can say that the choreographer borrows steps only if we are prepared to complicate that notion. The choreographer who borrows a step actually has only a rough-and-ready criterion of how much to take: enough to get across the "feel" of ballet and not so much as to get it confused with his own basic vocabulary inventory. The borrowing is successful when it accomplishes this.

     Style, which incorporates spatial vocabulary and kinesthetic motivation, is in fact the ultimate criterion by which we identify and judge steps. Working backwards from a score, knowledge of the choreographer's style is crucial to the interpretation of what steps he is specifying. This means that a score which is adequate for producing a work in a known style would not ordinarily be adequate if the style, its vocabulary, and its characteristic kinesthetic motivation and ideals were not antecedently known. The common-sense unit of choreographic composition is, thus, not interpretable independent of its history of production. This accounts for the fact that it is extremely important in dance to know both who did the scoring and whose work was being scored.

     At this point, though, it may be objected that the way people do score and the instability of the usual basic unit does not preclude the possibility of a serviceable universal notation. The step may be relative to a style, but the notion of a sequence of finely differentiated body positions is not. Such a notation would function, much as the scores of music do, independently of knowledge of the choreographer's style, since the elements in which he deals could be built up of units of a much more finely articulated system. The problem with this suggestion is that sequences in dance have an element which seems likely to resist notation. With great care and extreme disregard for economy, most spatial vocabularies could be captured by the kind of system described. But choreographers also deal in kinesthetic motivation.

     A sequence of positions may be "letter perfect." Still, if there is not in addition the correct kinesthetic motivation, the sequence is quite literally wrong. Kinesthetic motivation is not at all a matter of making dancers feel or want to move in a certain way. It has to do instead with the way sequences of movement are organized. It may be taught by example, explained in terms of where the impulse starts and how it develops, or elucidated by images, e.g., "hinge the arms out like a fan." Slight differences in kinesthetic motivation make a great difference in the quality of movement produced. Movement wrongly motivated will be and look wrong, i.e., inaccurate. Even rightly motivated movement may fall short of the ideals of the dance style or form. This is not because the proper motivation may be known in the abstract without being translated into movement, but because kinesthetic knowledge, like any other kind, may fail to guarantee performance if ability and training are inadequate. Obviously, a complete understanding of kinesthetic motivation presupposes the ability to carry it out, since the knowledge is, in the main, knowledge of what it feels like to do the movement correctly. But a dancer may very well know, for example, that a successful outward spiral requires that the head follow the body axis and yet be unable to keep balance while doing this. Wrongly motivated movement is considered inaccurate, not just inadequate, as "The Moor's Pavane" example shows. There is no point in asking dancers to pry themselves loose from an insane standard of compliance, and admit that movement motivation, like the expressiveness of a pianist, is something which may vary widely without affecting the integrity of a work. It is simply too obviously a fact of the dance world that a performance which is stylistically deviant because wrongly motivated (though all the positions are correct) is not just suboptimal, but fails to be a performance of the work.

     Given the nature of choreography as we have sketched it, it is not surprising that theorists and artists view the notion of an adequate notation for dance with suspicion. Under the best of circumstances, there is a good deal of information to be encoded, information which is usually carried in visual or kinesthetic memory precisely because it has to do with vocabulary and motivation.

     Notation, as notators see it, has the role of providing a code for recording the information necessary for retrieval and production of a performance which is a member of the compliance class of dance events which, in Goodman's terms, comprise the work.⁵ It has proved inadequate for these purposes. Some choreographers object in principle to the very notion of a stable choreographic process with its record in precise notation as a means of producing a performance:

Notation—all of these systems based as they are on symbols which are translated by the dancer, are out of whack. The element of them that has always troubled me was the translating act. The notator looks at a step, translates it into a symbol, writes it down, then sometime later, the dancer looks at the symbol, translates it back into a step, and then does it. But this is not the way a dancer acts. In his class and in rehearsing, he looks directly at the step, or someone doing a movement, and reorganizes that immediately into his own body. It is more direct than the symbol syndrome. . . . I say the symbol is an unnecessary hang-up with the past. (Cunningham 1968).

It is a fact that dancers and choreographers are unwilling to rely on notated scores to produce performances of established works. This leads inevitably to the suggestion that the technique might be appropriate for identifying performances of a work, though inadequate for producing a performance of it. Goodman allows for this possibility. But there are difficulties with this suggestion too.

     There is, to begin with, the technical problem posed by orderings and deletions. It seems that the notator would have to claim that performances of Cunningham's "Suite by Chance" (1953) were not performances of the same work, since individually stable segments may turn up in nearly any order. The sequence is determined by chance devices. This seems an incorrect classification. Part of the point of doing or seeing that work is to see what results a new ordering will produce. There is no point in having a theory which cannot cope with such works, which are no longer borderline cases. In any event, a retreat to "stable classics" will turn up analogous difficulties. Some traditional ballets have sections which are not often or not always performed. And a cursory inspection of films of "classic" performances will usually show that either they or current performances are deviant. This particular difficulty is perhaps not insurmountable. The notator could, in the first case, record the segments and allow any ordering of them to count as a performance of the work, so long as all are performed and done accurately. In the second case, the sections in question could be considered official options.

     A much more serious problem is that, in the case of the autonomous choreographer or the highly trained ensemble, it is not clear that the information which is vital can be encoded or that all that is encoded is vital, if we construe notational systems as means of recording movement in minute detail. Autonomous choreographers, who plan and execute their own works can make non-trivial changes in a work from performance to performance without placing the identity of the work in doubt. The same is true of established choreographers, such as Graham. The outside boundaries are clear enough. Alterations in style₁ constraints are inadmissible. But in such a case, a notation precise enough to be useful will also be precise enough to force rectification in cases where alteration is not drastic enough to warrant it.

     The most serious problem of all is the fact that a notated score adequate for identification of works in a way that conforms to general practice must capture the crucial element of kinesthetic motivation. The Effort-Shape system does try to account for some such factors. But it is an analog system; it sacrifices precision for completeness. Labanotation is precise, but becomes descriptive, and even pictographic when it attempts to notate movement ideals and motivation. The Eshkol-Wachman system is more subtle in its distinctions, and is still little known. It is digital and, hence, less susceptible to blurry interpretation. In any event, none of these systems [seems] to capture movement motivation,⁶ except by recourse to verbal and pictographic devices. The phrase markings and verbal supplements to musical notation play much the same role as the corresponding devices in dance notation. The difference between the two situations is that the supplements to musical notation do not, while the corresponding elements in dance notation do, specify elements which are integral to the identity of the work. It is perhaps true that, strictly speaking, a musical performance which violates mood, tempo, and general stylistic convention is deviant in a nonessential way. This is just not the case in dance, where at least some of these considerations are important to the identity of the work at least some of the time. The fact that musical notation, strictly construed, does not include these determinants suggests that they are very difficult to include in notation.

     In very ordinary cases, then, notated choreography seems to provide a criterion of identity which is too loose to incorporate fully style₁ and its style₂ manifestations. The criterion is too weak. The case of the autonomous choreographer suggests that it is also too strict to accord with dancers' and audiences' abilities to recognize compliants. The problem is that producing movement isomorphs which are exact is not the operative concern in preserving the identity of works in dance, though in a weakened form it is an important consideration.

III. Dance Scores

In fact, there are operationally adequate scores in dance, though they are not notational in Goodman's sense. Many modern choreographers do not see what they call the "score" as a plan of movement drafted or derived from one performance and used to produce and identify other members of the compliance class of events which comprise the work. In some cases, a single score is used to produce several dance events which are not properly considered performances of the same work. In other cases, the dance events produced are properly considered performances of the same work, though the traditional notator would have to disallow them. The kind of symbol scheme used to put together a score depends upon what the score is supposed to do. In almost every case, the scheme is very different from the one which is useful in recording or providing an analysis of movement. Since, as Cunningham points out, the creators of such dance works have no intention of using their scores in this way, it is of no concern to them that their notation would be drastically inadequate for those purposes. In general, these "scores" are used to generate the work, and can be construed on a "rule-instantiation" model (Tormey 1974). Lawrence Halprin defines the score as "symbolization of processes which extend over time," but goes on to add that they "describe the process leading to the performance" (Halprin 1969: 1-2). Scores of this kind are manifestly not independent of the history of production.

     In the case of dance works designed by "chance," like the works of Merce Cunningham, the number of score symbols is finite, and their references are discrete; the ordering is variable, and in many cases arbitrary. The sequence of steps, accompaniment, and lighting vary from performance to performance according to "chance" selective devices, such as the throw of the dice. The duration of intervals determined by chance is regulated by a stopwatch. The dancers do consider the performances produced in this way to be performances of the same work, since there are performance "rules" and the overall style is relatively homogenous. A good choreographer working in this way with his own trained dancers produces spontaneity, not chaos. The "rules of the work" and the qualitative style similarity are the criteria dancers go by in deciding which work is being performed. Both elements are important. If the rules are changed, though the dancers remain the same, the result is a different work. If the rules are followed, but by dancers who are not Cunningham trained or taught that training, the result is likely to be a stylistically deviant and inaccurate work. Compliance with these two criteria may be more apparent to the dancers than to the audience. Cunningham and his company have recently given "collage" performances, which are made up of sequences of past works. The audience is left to guess the origin of the sequences, and often guesses wrong. (The dancers presumably know precisely what work they are performing.) It is on the whole easier to know the rules one is playing by than to recognize what rules someone else is playing by.

     Works such as those produced by Anna Halprin and the San Francisco Dancers' Workshop present a different kind of problem. In this case, an initial workshop setting determines the score, which employs an idiosyncratic choice of symbol scheme. The score precedes the performance, though it does not specify either style₁ or style₂ constraints. Each of several groups collectively makes a drawing or "score" which is a broad, diagrammatical outline representing the action of the piece and including parts for the members of the other groups. Since the symbol scheme is idiosyncratic, each "score" has to be explained by some member of the relevant nuclear group when it is implemented. The entire group then participates in the performance of each of the several scores without prior rehearsal. Each original score is so broad that there is no predicting what the performances will look like, though the score is in each case indeed the score of the work and is being used to generate the performance. The scores could be re-utilized, and the production of performances from such scores is surely not arbitrary. But despite the constancy of the rules instantiated, subsequent productions are not likely to produce performances of the same work, for they may be radically different stylistically. There is no antecedently developed spatial vocabulary or motivational system. The style is produced on the spot. The training of such dancers aims at producing the ability to develop stylistic cohesiveness quickly and the physical ability to execute the movements required by a spontaneously developed style, not at producing a clear and recognizable movement idiom.

     Finally, there are scored works which are similar to the cases previously mentioned, in which the score determines a work to be performed by large groups of people who have had little previous contact. These scores can be verbal or diagrammatic and the instructions can vary in specificity. Works of this type are based on a concept of communal experience and creativity and are sometimes termed "environmental pieces," because the features of the setting often become an integral part of the performance. The necessity of putting together a program for advertising purposes might force the choreographer to label the dance events produced from the score "performances of the same work," but dancers are not likely to agree with the implication of the labeling. "Happenings," "rituals," or "environmental pieces" are considered one-shot works. There is a simple reason for this. Works of this type have preconditions defined by the rule of the score, and they may have a built-in propensity for style₁ constraints. But the distinctive spatial vocabulary and movement motivation [are] dependent mainly upon what happens during the performance. Style develops out of group cohesiveness, and the precise quality of the cohesiveness will be determined by the performers and the setting.

     Anna Halprin's "Trance Dances" are of this kind. The style₁ is created out of the interaction and accommodation of individual style₂'s. In each dance, the score prescribes the presence of live music, often drumming; there is a beginning moment when everyone sits in a circle and finds his particular breathing rhythm. Out of this rhythm develops a song or chant and an ending determined by Halprin's or the group's need for closure. The ending involves a verbal sharing of the experience. Style₁ features are created in the process of the enactment of the score. Dancers who have not previously worked together in ensemble and who have different training face the artistic challenge of finding a common style₁ by merging and standardizing different style₂'s. The piece is successful only if this merging takes place.

     Thus, choreography, even in its expanded definition, adequately notated or not, is a most unreliable identity criterion for the dance work. Even if in some cases, the "score" does produce multiple performances of the same work and could doubtless be used to identify them on occasion, the performances need not resemble each other to the extent demanded by the scores of traditional notation. A notator's score is not a substitute for the underlying structure or constitutive rules of a given work. And sometimes, as in Halprin's works, the score is not a criterion of identity, though it is necessary to produce the piece. In fact, in this case, the score is designed to produce multiple works. The fact that scores often do precede performance and are sometimes used to preserve the intricacies of movement for future dancers is the source of the mischief. It leads to the infelicitous tendency to overstate the parallel between the scores of music or drama and those of dance.

     In dance, it is the performance which is the primary work. Grouping performances together into compliance classes which are then labeled "the work" is a harmless abstraction—so long as one is not overly strict about the nature and the extent of compliance required and does not take common choreography as an absolute determinant of class membership. If compliance is defined too strictly, we shall be forced to disallow the integrity of works such as those of Cunningham, as well as many works of autonomous choreographers. That this ought not to be done is not a "soft" intuition. It is hard data. If, on the other hand, choreography more reasonably interpreted is taken as the sole determinant of identity, we should be forced to group together works more properly considered distinct. This latter problem is less intractable, since it is probably not very clear to artists and theoreticians just how we ought to handle the problem of multiple works from a single score. The disinclination to accept the performances of Halprin's "Trance Dances" as performances of the same work is, of course, an extension from the mainstream of dance theory, where distinctive style is an important determinant of identity. Improvisational works in the performing arts have always caused precisely this sort of problem, and it is possible that the question of how to deal with them ought to be kept separate from the question of how to deal with works in an art form which are not of this kind.

     It remains clear that style, construed as spatial vocabulary and characteristic kinesthetic motivation, is an important determinant of identity in the dance work of art. It is style which allows us to reject certain performances, while others, though bad, are allowed to count. The average civic ballet's performance of "Swan Lake" is stylistically deficient. The Fonteyn-Nureyev presentation of "The Moor's Pavane" was stylistically deviant. The problem was that the style₁ of the original piece was replaced (perhaps inadvertently) by a well-defined and alien style.

     We have argued that style as we construe it is difficult, if not impossible to record in a notational scheme. This does not mean that traditional notation systems cannot serve as mnemonic or analytical devices or that choreography is not an important element in determining what work a performance is a performance of. Indeed it is precisely the importance of choreography as a link to the specific choreographer which causes difficulties for "common-denominator" notation schemes which aim at bypassing this link. But if style is important and not amenable to notation, then identity criteria based solely on notational scores are inadequate. The broader theory based on the "game model" (which may in some cases include a score) with its "rules of the work" is not adequate either. But it provides a model which allows the introduction of the two levels of style as further determinants.

     Our model elucidates the important aspects of the dance process conceived in general terms. It also provides a convenient way to characterize the relation between the contribution of the individual performer and the identification of the work. A model which can account for the generation of the properties the work has can begin to deal with the important problem of what constitutes a violation of the stylistic constraints of a work—without, we feel, undue emphasis on factors more properly related to a work's history than to its identity.

     Goodman's model is too tight to accord generally with incidental, non-incidental distinctions as they are now made. Dance clearly tends toward being an allographic art, but it is not yet clearly notatable. The fact that a need for notation is present and that notation is partially successful for known works in stable styles⁷ indicates that dance is much more likely to be an art form in transition to being allographic than an art form inherently autographic. It will become allographic only when either notation succeeds in capturing style or general practice decides that style is incidental. Either might occur, but neither has to date. It is unlikely that style ever will be considered incidental, given the tight connection between stylistic compliance and authenticity. Probably no system of notation which satisfies Goodman's requirements even will succeed in capturing style. But it is likely that the increased demand for re-performance of dance works and the increasing homogenization of dancer training already underway will gradually free dance works from the idiosyncratic control of their creators and increase the number of persons who can adequately interpret inadequate scores.