Vasari, Schapiro, Schaafsma: Three Points of Departure for a Discussion of Style in Rock Art

: This article examines the concept of style by combining three approaches: that of Giorgio Vasari, whose work is a classic of Western art history, that of Meyer Schapiro, which mediates between art–historical and archaeological / anthropological disciplines, and that of Polly Schaafsma, an example of what stylistic analysis may achieve in rock art studies. We foreground rock art by reason of its ubiquity and time-depth, at the same time placing it in the context of any kind of depiction. In the course of the argument, we comment on a variety of relevant issues, such as those relating to progress in art; to realism; to the relation of style and history, that is, cultural context; and to quantitative as well as qualitative analytical methodologies.


Introduction
Anyone writing on style will find it necessary to simplify and limit a huge and difficult subject. We limit our article to three "points of departure", aiming to examine theories of style, but with an eye to practice and in relation to diverse scholars, three in particular: Giorgio Vasari, author of a classic art history text, Meyer Schapiro, author of an influential modern article on style, and Polly Schaafsma, a contemporary rock art expert. We stress the argument is less about these three, however relevant, exemplars than about style itself, aiming to put forward eight basic propositions which might initiate discussion of the phenomenon of style. The argument applies to any form of depiction, but focusing on rock art, largely glossed over by art history researchers, has many advantages, not the least of which relates to the ubiquity and time-depth of rock art. (Rock art is found almost everywhere and may date from the near present to great antiquity.) The better our understanding of markings on rocks, the greater our grasp of the issues raised by any kind of art. It goes without saying that throughout this article the term "art" must be taken in quotation marks. But then most of the intentional objects we term "art" were almost certainly not read by their makers in the way contemporary Western observers read them. At the same time, in saying this, we by no means rule out a sense of the aesthetic on the part of the makers of the rock art. While our taste is of its place and time, application of aesthetic judgement is probably universal.

Vasari
Vasari, who published his definitive Lives of the Most Excellent Painters, Sculptors and Architects in 1568, is arguably the chief stylistic theorist in the last five hundred years of European art history ( Figure 1). His views may seem foreign to an account of art in 2019 and even more so to an account of rock art. However, this is not the case. Vasari sets out to tell the story of Italian art in (for him) the last three hundred years on the basis of the new style, that of the Renaissance, which he terms la bella maniera It goes without saying that taste in art has changed since Vasari's day. Few today would subscribe to the view that chaotically profuse Gothic cathedrals might be greatly improved by the application of Renaissance rules of ordered proportion. Gothic and Renaissance architecture are simply different. Likewise, many are likely not to denigrate but to admire the statuesque stillness of Byzantine mosaics precisely because they express qualities and values removed from the everyday. The point we wish to make is a different one, however, namely that a surprising number of Vasari's key ideas retain currency, at least in some quarters and in contemporary form.
In art history, the most notable later-twentieth-century theorist to promote the notion of art as a progression towards realism was Gombrich (1960). In his most famous book, Art and Illusion, he It goes without saying that taste in art has changed since Vasari's day. Few today would subscribe to the view that chaotically profuse Gothic cathedrals might be greatly improved by the application of Renaissance rules of ordered proportion. Gothic and Renaissance architecture are simply different. Likewise, many are likely not to denigrate but to admire the statuesque stillness of Byzantine mosaics precisely because they express qualities and values removed from the everyday. The point we wish to make is a different one, however, namely that a surprising number of Vasari's key ideas retain currency, at least in some quarters and in contemporary form.
In art history, the most notable later-twentieth-century theorist to promote the notion of art as a progression towards realism was Gombrich (1960). In his most famous book, Art and Illusion, he speaks of the classical Greek "conquest of naturalism" (Gombrich 1960, p. 100), perfectly aware that the concept of mimesis, Vasari's imitation of nature, needs to be recast in modern terms. However sophisticated, though, his argument returns us to Giotto's fly: the best art is one that might be confused with reality. In his The Story of Art (Gombrich 1972), chapter titles make his position clear: Classical Greek art comes under the rubric of "the great awakening" and art of the Italian Renaissance under that of "the conquest of reality". Such attitudes meant that Gombrich could only view twentieth-century art as a decline, and the debate with those who thought otherwise because they saw no absolute value in "realism" or progress in art was neatly illustrated in the rivalry with Arnheim (1974). At present, the progressivist mimesis thesis has no followers in the discipline of art history. If we accept Vasari, we have to conclude that Picasso is inferior art. We dismiss most Western art since at least the early twentieth century and much pre-European American, as well as African art and the art of Oceania. We have an insurmountable problem with most rock art. Thus it is fair to say a progressivist mimesis thesis in art has been abandoned, notwithstanding Postmodern quotations of realism, Hyperrealism, and the like. This is not to say various theories based on perceptual illusionism are not going strong, say in disciplines like cognitive psychology. However, we are talking not about perceptual illusionism but the specific idea of mimetic realism as the highest value in art. Perhaps the text which puts the best case for a multitude of equally valid depictions of the real in art is Hagen's Varieties of Realism (Hagen 1986).
Summarizing the above: Eurocentrism in art, based on Graeco-Roman and Italian-Renaissance models, is no longer tenable. We understand that what looks real in one culture need not look so in another. Specifically, we no longer see a figure by Giotto, or by Piero, or by Michelangelo (Vasari's friend and touchstone for mimetic realism) as on the point of stepping out into reality, after the manner of Pygmalion's statue-though we may still casually refer to nudes by Lucian Freud as "naturalistic". In short, we realize that Vasari's "realism" amounted to a particular style, not a realism to end all stylistic representation. We can accept that artistic progress may well obtain within a particular stylistic paradigm. After all, the Italians, from the fifteenth century, did come to terms with the mathematics of a particular form of perspectival projection, namely, the recession of planes as viewed from a single given point. But most cultures, operating within quite different paradigms, might be expected not to conform to this model and to have their own notions of good, better, best in art.
Moreover, in the light of art historical revisions, it might be assumed that today, few rock art scholars would hold to the belief that images on rocks improve insofar as they create an illusion of the object they represent (as if a Coso sheep might leap out of the rock or a Sydney sandstone macropod bound across the researcher's visual field). And yet, almost everyone is in awe at the Chauvet lions, and some certainly attribute this effect to their looking "lifelike". So the Vasari model still has traction, perhaps more in Europe than elsewhere. Breuil (1952) was obsessed with foreshortening and could only describe the canonical form of a bovine-which is best captured in profile and with both horns visible-as a "twisted", that is, "incorrect" perspective. Leroi-Gourhan (1968, who traced stylistic progress in Franco-Cantabrian art, outlined phases of this progress bearing an uncanny resemblance to Vasari's. The terms are mastery of visual or optical "accuracy" (Leroi-Gourhan 1982, p. 17), evolution to depicted movement-that is, "animation" (Leroi-Gourhan 1982, p. 39), in both French and English and in each case pure Vasari-as well as perspective. The aim is expressed concisely (Leroi-Gourhan 1982, p. 31) as "the vision of the subject starting from a unique viewpoint" (see our formulation above: "recession . . . viewed from . . . single . . . viewpoint"). Photography might be the model but, revealingly, Leroi-Gourhan (1982, p. 18) takes it back to evolution of Greek styles from Archaic (stiff, hieratic) to Classical (animated) to Hellenistic ("realistic"). Even as he explicitly denies value judgement, he nonetheless implicitly reintroduces the entire narrative of European artistic developments as generally held until the advent of modernist art in the twentieth century.
While such views are much more difficult to hold outside Europe, say in Australia or the Americas, the notion of depicting things "as they really are" has tenacious appeal in rock art studies. Of course, such depiction would logically bypass any characteristic of "style". It would not have a style but would be an unmediated image of reality. A moment's reflection suggests the difficulties of this idea: there can be no representation unmediated by culture. Can a photograph be any less colored by its cultural context than Giotto's fresco or a Byzantine mosaic from Ravenna? Is the photograph somehow closer to the real? Vasari might well have said this about the Giotto picture. Today we are sufficiently exposed to a plurality of cultures to know that different cultures have different ideas about the real, and diverse styles to represent them. However, the old notion persists, if not in art history at least in rock art studies. Clegg (2012) put forward the view that one of the Chauvet lions was "badly" drawn, to which we would reply, stressing the inappropriateness of his criteria: who could say, outside any cultural context, what constitutes "good" drawing? To give Clegg his due, he had the intellectual curiosity to sense the (for his ideas) immensely problematical implications of an unmediated realism (Clegg 1993).
Quite different examples of what some might regard as style-less representation are provided by trance "entoptics" or the more fundamental concept of "phosphenes". Whitley (2000Whitley ( , 2011 appears to think that trance images, especially geometrics spontaneously generated in the first of the three trance stages hypothesized by Lewis-Williams and Dowson (1988), must be independent of a particular culture or style. Watson (2008), who takes up the Bednarik (1984) and Hodgson (2000) view of the phosphene forms as automatic responses to neural resonance, tentatively suggests a comparable conclusion: in his case, that doodles are universals without particular cultural coloring. But even if we accept either the trance or the neural resonance arguments, it cannot be that depicted entoptic or phosphene forms (as distinct from experienced ones) somehow escape their cultural context. Varied entoptics resulting from a migraine presumably look much the same across the globe. However, in a particular cultural context, their depiction will involve a particular design, will be subject to choices regarding placement, association with other images, dimension, medium, etc. Moreover, every hand that makes a picture is idiosyncratic, as in the case of handwriting (see Freedberg and Gallese 2007 on gestural traces). We understand why this point may be missed by scholars focusing on universals (in which group we include ourselves). But it would be a mistake to take the argument so far. We return below to our view that there is no contradiction between arguments based on biological universals and those based on cultural mediation. However universal (and the present authors have focused their work on perceptual universals), depicted forms are bound to have an overlay of a particular cultural style.
We would like to take note of one other form taken by the progressivist thesis in tandem with the appeal to realism (understood as picturing things "as they really are"). This is the distinction, common in many disciplines, between "simple" and "complex". Maynard (1979)-with McCarthy (1967 and Layton (1992), one of a handful of rock art scholars ambitious enough to attempt a stylistic and chronological typology covering the entire Australian continent-divided her material into ("abstract") Panaramitee (loosely equivalent to Archaic Great Basin styles in the United States), Simple Figurative, and Complex Figurative. Naturally, a sense of complexity-as-progress lurks beneath the surface of the argument. At the same time, "complexity" calls up the specter of realism. It means depiction in greater detail, and detail is assumed to characterize things as they really are, not schematized in the process of representation. At this point, we side not against but with Gombrich: all representation is inescapably schematic, what Gombrich (1960, p. 184) calls the "etc. principle", and it is a fallacy to identify realism with greater complexity, that is, with more lines in the picture. Maynard, whose boldness deserves respect, stumbles in another way: she identifies complexity with depicted movement. But in doing so, she alters her definition of complexity from a quantitative to a qualitative one. Stick-figure action may be depicted with (say) ten lines; a static stick-figure may be made with more than ten lines. So how is motion more complex in this case? Clearly, the answer is that Maynard judges a figure pictured in motion as more real than one standing still. We are brought back (inexorably?) to Leroi-Gourhan's "animation" and, over and above that, Vasari's Pygmalion thesis of the statue that comes to life. We would like to anticipate later portions of this article by giving Schaafsma the last word in this first phase of our argument: "there is no correlation between the degree of sophistication in the art and the level of cultural development" (Schaafsma 1980, p. 3). Put another way, the simple-to-complex model does not apply to art. Rock art studies in particular have no need for any thesis of progress. The same applies to the Eurocentric idea of mimetic realism.
Having laid the groundwork for a discussion of style with the propositions that- (1) all depictions have a style (2) historically-recent European realism is itself a style (3) such realism is not a privileged style, there being no progress in art -we now turn to the second part of our exposition.

Schapiro
As an art historian addressing the subject of style with an eye to archaeologists and anthropologists (who have frequently relied on his article), Meyer Schapiro wants to cover the ground as thoroughly as possible. The result is eminently usable. The article (Schapiro [1953] 1973) consists of four sections which (1) offer a general definition, (2) analyze aspects of the definition, (3) comment on the impact of the naturalistic paradigm, and (4) consider the key question of the relation of parts and wholes in style. It points out that for the archaeologist, style is a "motive" or "pattern" which will help to "localize and date" the object of investigation. For the art historian, it has the added value of interest "for its own sake" (Schapiro [1953(Schapiro [ ] 1973. A critical comment of Schapiro's is that style is "directly grasped" (Schapiro [1953(Schapiro [ ] 1973. This would seem to be equivalent to the "distinctiveness" foregrounded in the Rock Art Glossary (Bednarik et al. 2010, p. 16) definition.
Direct perceptual access to attributes of style, taken for granted by Vasari and advanced as the key to stylistic analysis by Schapiro, has begun to be investigated in the present century by researchers into biological aspects of the aesthetic experience who draw insights from neuroscience (Freedberg and Gallese 2007;Gallese 2017;Piechowski-Jozwiak et al. 2017;P. Dobrez 2013P. Dobrez , 2018. It is not, in our judgement, relevant to go into this in a discussion of Schapiro. Nonetheless, we wish to alert the reader to the possibility of identifying biological substrates for the experience of grasping a stylistic feature, in Schapiro's terminology, "directly". Here we would put the matter as follows: that style must be recognizable, however complicatedly and, on occasion, ambiguously. If it is not, then we cannot talk about it. There is also the further point of Schapiro's definition of it as "constant form" (Schapiro [1953(Schapiro [ ] 1973, with the understanding that constancy may obtain for a long time, as with European Palaeolithic art, or a short time, as with Happenings or Performance Art. Contra Gombrich, Schapiro accepts the dominant modernist view that special status can no longer be accorded to "realistic" or "naturalistic" representation. There is no privileged criterion of correct or accurate representation. He allows a definition of style as including technique, subject matter, and material but prioritizes elements of form in relationships which generate specific qualities, and so "expression" (Schapiro [1953(Schapiro [ ] 1973. In general, his approach is holistic, though the nineteenth-century paradigm of organic unity, the idea of the whole contained in its parts (Ast [1808(Ast [ ] 1990, is presented with provisos. Thus there may be a deal of variation in the perceived unity of style, indeed as much heterogeneity as homogeneity in it. Clearly this line can be taken so far as to empty the concept of a given style of any significance, as will be seen below, but Schapiro's examples of heterogeneity are no more than common sense: a work may have more than one style; the individual artist may change style; more than one style may exist within a single culture, etc. In line with general linguistic usage, Schapiro of course understands style as attached to any number of entities, from the individual to the group, possibly very large, or to that open-ended entity known as culture. So while he is fully aware of problematical aspects of the idea of style, he does not problematize to the extent apparent in the last forty-odd years of theorizing. In line with his stress on expressive qualities, he affirms holistic thinking in his comment (to which we shall return) that when we deal with issues of style, qualitative language is more precise than quantitative. Likewise, that there are connections between given works and given cultures. Thus both archaeologist and art historian may take style as cultural evidence, that is, evidence of a "manner" (in Vasari's sense), identifiable, with qualifications, as to time and place. We return to this point below.
It so happens that the latter part of the twentieth century was a period of systematic problematization in all Humanities and Social Sciences disciplines and to an extent beyond, all the way to Law and the Sciences. Since then, the debates of those years have themselves become dated. For present purposes, however, it is relevant to recall some of the issues. The catchword was "heterogeneity" or "difference" (Derrida [1968(Derrida [ ] 1982. Anthropology suffered its crisis of ethnography and turned into Cultural Critique (Marcus and Fischer 1986), archaeology lost its innocence (Conkey 1990, p. 12) and went from the old New Archaeology to the new Postprocessualism-just as the present authors' (then) discipline of Literature went from the old New Criticism to the new Poststructuralism. We will refer to two books which in varying degrees complicate the picture of style as put forward by earlier theorizers like Schapiro, noting from the start, however, that Schapiro's approach is far from simplistic. In 1993 (and, at best, prematurely), Lorblanchet and Bahn declared the "Post-Stylistic Era" in rock art studies. Clottes' (skeptical) contribution to the volume gave two avowed reasons for this, namely, increasing reliance on new dating methods which should make stylistic analysis less necessary, and new generational criticism of past style-oriented chronologies (Lorblanchet and Bahn 1993, p. 19). New and critical readings of such chronologies certainly came into play at the time of the Lorblanchet/Bahn collection. While they could at times amount to the Aunt Sally critique, they were understandable after the excessive claims made for stylistic periods by immensely influential scholars like Leroi-Gourhan. In fact, Lorblanchet (1993) simply wanted to say that pictures that look similar need not belong to the same period. Beyond that, he and Bahn make clear in their Introduction that their manifesto aims at being provocative; style is not dead after all. Indeed, by Clottes' count, some half of the contributors, himself included, stand by the value of stylistic analysis (Clottes 1993). The firmest position against the value of stylistic judgements is held by Bednarik (1993Bednarik ( , 2007, who consistently regards these as subjective and therefore non-scientific. We agree that they are non-scientific but dispute their merely subjective status (L. Dobrez 2011).
A more history-of-theory-oriented attempt at rethinking the idea of style than the Lorblanchet/Bahn one is Conkey and Hastorf (1990), whose work, however, deals with archaeology rather than rock art studies. In her contribution to the volume, Conkey gives a historical account of archaeological theory beginning with culture-history and moving on to Processualism and its Post. Like many others, she accepts modern revisions without wishing to abandon earlier formulations. In particular, she reads Schapiro as an exponent of a culture-history approach, which is perhaps more accurate than not, provided we gloss over his own presentation of complicating factors. In general, though, the Conkey and Hastorf book is part of that problematizing trend evident at the time. (For an account of whose origins and implications see L. Dobrez 2014). Here it suffices to say that the (largely French) ideas which went under the rubric of Poststructuralism or, in a much larger cultural/historical context, Postmodernism, had their "soft" and "hard" versions and that like most-not all-anglophone scholars, Conkey adopts the "soft" version, that is, the "problematization" strategy. She wants to query the too-easy assumption of earlier archaeologists that style acted as a period marker or a marker of culture or group. Of course, the assumption, if not made too facilely, is not wrong: a style must belong to an individual, a group, a culture, a time. But the difficulty is that we cannot identify a source in time and space simply on the basis of a style, as Schapiro seems optimistically to imply. Style is of itself insufficient evidence for time and place. This is the point made repeatedly by Davis (1990, p. 18ff) in his contribution to the Conkey and Hastorf book: there must be, as he puts it, archaeological evidence to support the stylistic. In fact Davis, as an art historian among archaeologists, courteously overstates the point. After all, archaeological evidence may itself be far from secure.
We do not deny that Conkey's argument advances understanding of the complex relationship between culture (read "history") and style in art. It has done no harm to introduce reflexivity and ambiguity into the discussion, to note that neither cultures nor groupings of people need be homogeneous, that culture is, in the jargon, negotiated or contested. But we equally stress the unreality of too-easy dismissals of the historical continuity of theorizing about style. The pattern is that of synthesizing pioneers followed by a generation who can afford to take the large synthesis for granted and who are more concerned with the details-some of which inevitably modify or challenge the ideas of the pioneers. The question may be put this way: at what point does a collection of trees become a "forest"? If you focus entirely on the forest, you miss the diversity of trees; if you focus on individual trees, you miss the larger unifying concept of the forest. Moreover if, in line with "hard" readings of Poststructuralism (Derrida 1978), you take heterogeneity to its logical conclusion, you reach the ground zero of Chaos. The dilemma is not a new one. Before anti-essentialism-in its contemporary form originating in structuralist linguistics and the anthropology of Lévi-Strauss, where everything is defined relationally, nothing as a thing-in-itself-we had what philosophers term Nominalism. An example is found in the Davis piece mentioned above (Conkey and Hastorf 1990, p. 19). Davis references an impressive list of artistic misfits: copies, quotations, forgeries, etc. A 1500 Raphael and its 1989 copy look the same but are of diverse periods. A picture in the style of Velazquez is not the same as one by Velazquez. We may add objets trouvés: is Duchamp's (real) urinal in the style of Duchamp? Perhaps it is. But the most interesting aspect of Davis' article is the idea of style as "polythetic". Diverse art may share a particular and therefore diagnostic X such that all items featuring X will be of the same style. However, Davis adopts a different model of style, one borrowed from Wittgenstein's model of classification (Wittgenstein 1974). It may be illustrated as follows: (1) A (2) BA (3) CB (4) DC

Here, (1) and (2) have A in common, (2) and (3) have B in common, (3) and (4) C in common;
(1) has nothing in common with (3) and (4); (2) has nothing in common with (4). So there is no single common diagnostic factor for all four. At the same time, there is clear continuity from (1) to (4). Wittgenstein called it the "family resemblance" model of classification. And we have alluded to it in a previous article , noting our difficulty with it. The thing to stress is that the model may be termed Nominalist and linked to Conkey's "soft" anti-essentialist approach to style.
We now have three further propositions which contribute to a discussion of style: (4) a given style is necessarily linked to a given time and place, that is, to a cultural context (5) style alone cannot be taken as evidence for a particular cultural/temporal context (6) the relation between style and its historical context may be problematized, but not to the point of denying that there is a relation On the matter of this critical relation of style and history, we add that it is perfectly possible to classify depictions non-historically, that is, non-contextually. We may choose, in short, to bypass issues of style. But this does not mean that style has, as it were, disappeared. Both of the present authors have in the past chosen to investigate universalist aspects of art which may be discussed in, for example, biological terms. But this is not to deny the reality of culture-specific aspects of the same art. As noted above, the two approaches, contextual and universalist, by no means contradict each other. Of course, methodologically, they exclude each other-we cannot choose to study a picture or text historically and non-historically at the same time (or, in the language of Structuralism, "diachronically" and "synchronically" at the same time). In a number of articles, the authors have opted for the latter. But we would not for a moment deny the option of contextual analysis: even if you choose to focus on something other than the cultural/stylistic, culture/style remains. The five horses illustrated below (Figures 2-6) may be analyzed as the universal we called "canonical form" . In terms of stylistic analysis, all that can be said in this case is that we are faced with five distinct styles, each in some way related to a different historical context, respectively Australian, Indian, American, European and, again, Australian (the photograph). Is it necessary to add that the photo also has a particular style?            Three panels from Utah (Figures 7-9) nicely illustrate key points made by Schapiro. Examining these with the aim of registering similarities and differences between them, diverse observers would presumably pinpoint diverse elements. However, some elements would probably emerge as dominant, for example, that two panels feature large tapering frontal figures, but that in one case, the figures are markedly angular; likewise, that large figures do not feature in the third panel, which might be thought to include profile scenes. On the basis of this, the most likely conclusion would be that two of the images differ (Figures 7 and 8), but within a larger pattern of similarity, while the third ( Figure 9) would be judged as entirely different. In all cases, though, we judge similarity and difference immediately, or in a way that is unmediated: we do not need to reflect on the judgement. Moreover, we judge the details that go to make the whole by an immediate synthesizing awareness of the whole. In Schapiro's language, style (in all its complexity) is "directly grasped". It is worth noting that the panel which seems to have least in common with the other two has the advantage of identifiable subject matter: it includes a horse and so is datable to the Historic period. We wish to delay identifying three observably distinct styles, all well known to American researchers, but simply note here that the presence of a depicted horse illustrates the requirement that stylistic analysis have some non-stylistic corroboration. Of course, some rock art recorders not unreasonably consider content (in this case the horse) as part of style. The question then may be one of semantics-though we bear in mind Schapiro's hesitation in giving subject matter, along with technique and material, priority in his definition. Content is certainly relevant to an account of an art panel, but not strictly part of a description of style. We will return to the Thompson Wash example  Three panels from Utah (Figures 7-9) nicely illustrate key points made by Schapiro. Examining these with the aim of registering similarities and differences between them, diverse observers would presumably pinpoint diverse elements. However, some elements would probably emerge as dominant, for example, that two panels feature large tapering frontal figures, but that in one case, the figures are markedly angular; likewise, that large figures do not feature in the third panel, which might be thought to include profile scenes. On the basis of this, the most likely conclusion would be that two of the images differ (Figures 7 and 8), but within a larger pattern of similarity, while the third ( Figure 9) would be judged as entirely different. In all cases, though, we judge similarity and difference immediately, or in a way that is unmediated: we do not need to reflect on the judgement. Moreover, we judge the details that go to make the whole by an immediate synthesizing awareness of the whole. In Schapiro's language, style (in all its complexity) is "directly grasped". It is worth noting that the panel which seems to have least in common with the other two has the advantage of identifiable subject matter: it includes a horse and so is datable to the Historic period. We wish to delay identifying three observably distinct styles, all well known to American researchers, but simply note here that the presence of a depicted horse illustrates the requirement that stylistic analysis have some non-stylistic corroboration. Of course, some rock art recorders not unreasonably consider content (in this case the horse) as part of style. The question then may be one of semantics-though we bear in mind Schapiro's hesitation in giving subject matter, along with technique and material, priority in his definition. Content is certainly relevant to an account of an art panel, but not strictly part of a description of style. We will return to the Thompson Wash example Three panels from Utah (Figures 7-9) nicely illustrate key points made by Schapiro. Examining these with the aim of registering similarities and differences between them, diverse observers would presumably pinpoint diverse elements. However, some elements would probably emerge as dominant, for example, that two panels feature large tapering frontal figures, but that in one case, the figures are markedly angular; likewise, that large figures do not feature in the third panel, which might be thought to include profile scenes. On the basis of this, the most likely conclusion would be that two of the images differ (Figures 7 and 8), but within a larger pattern of similarity, while the third (Figure 9) would be judged as entirely different. In all cases, though, we judge similarity and difference immediately, or in a way that is unmediated: we do not need to reflect on the judgement. Moreover, we judge the details that go to make the whole by an immediate synthesizing awareness of the whole. In Schapiro's language, style (in all its complexity) is "directly grasped". It is worth noting that the panel which seems to have least in common with the other two has the advantage of identifiable subject matter: it includes a horse and so is datable to the Historic period. We wish to delay identifying three observably distinct styles, all well known to American researchers, but simply note here that the presence of a depicted horse illustrates the requirement that stylistic analysis have some non-stylistic corroboration. Of course, some rock art recorders not unreasonably consider content (in this case the horse) as part of style. The question then may be one of semantics-though we bear in mind Schapiro's hesitation in giving subject matter, along with technique and material, priority in his definition. Content is certainly relevant to an account of an art panel, but not strictly part of a description of style. We will return to the Thompson Wash example but at this point wish to direct the discussion specifically to the work of Polly Schaafsma, with reference to the kinds of issues raised so far.

Schaafsma
We have chosen three names for the reason that they suit our argument, and this is no less the case with Polly Schaafsma than with Vasari and Schapiro. We very much began at the beginning (Vasari), before moving on to modern developments (Schapiro). With Schaafsma, we arrive at a contemporary figure. Her writing dates from the late twentieth century to the present, in which she continues to publish major texts, frequently on the contemporary legacy of the rock art record for Pueblo peoples. As we shall see, her training provided a direct connection with classics of art history

Schaafsma
We have chosen three names for the reason that they suit our argument, and this is no less the case with Polly Schaafsma than with Vasari and Schapiro. We very much began at the beginning (Vasari), before moving on to modern developments (Schapiro). With Schaafsma, we arrive at a contemporary figure. Her writing dates from the late twentieth century to the present, in which she continues to publish major texts, frequently on the contemporary legacy of the rock art record for Pueblo peoples. As we shall see, her training provided a direct connection with classics of art history

Schaafsma
We have chosen three names for the reason that they suit our argument, and this is no less the case with Polly Schaafsma than with Vasari and Schapiro. We very much began at the beginning (Vasari), before moving on to modern developments (Schapiro). With Schaafsma, we arrive at a contemporary figure. Her writing dates from the late twentieth century to the present, in which she continues to publish major texts, frequently on the contemporary legacy of the rock art record for Pueblo peoples. As we shall see, her training provided a direct connection with classics of art history

Schaafsma
We have chosen three names for the reason that they suit our argument, and this is no less the case with Polly Schaafsma than with Vasari and Schapiro. We very much began at the beginning (Vasari), before moving on to modern developments (Schapiro). With Schaafsma, we arrive at a contemporary figure. Her writing dates from the late twentieth century to the present, in which she continues to publish major texts, frequently on the contemporary legacy of the rock art record for Pueblo peoples. As we shall see, her training provided a direct connection with classics of art history (Vasari), and there is also a specific link with Schapiro, whose basic ideas she takes up. In this last respect, she is not unique but part of her generation. For us, a significant consideration is that her area of study, the entire southwestern United States, is arguably the richest in the country in terms of rock art. We state this in the knowledge that there are other image-rich parts of the United States, including one with which we are also familiar, the Plains northwest. In addition, Schaafsma has come up with, if not the only impressive taxonomy of a vast area of rock art (see, for example, Keyser and Klassen (2001) and Francis and Loendorf (2002) for the northwest plains), at the very least a taxonomy second to none in scholarship. Finally, she has done this with specific reference to style. As we know, stylistic analysis is part of the practice of many US rock art researchers, including some who do it with distinction. It is probably fair to say, however, that no one has used style as a primary instrument to the extent she has done. In short, what she does, while in line with much rock art practice in America, is nonetheless distinctive. We think she provides illustration of a number of points we wish to make to conclude our discussion of the concept of style. Grant (1983, p. 11) stated firmly that "style is the most important consideration in rock art research", citing a list of exemplars from Steward (1929) to Schaafsma herself. As a preamble to consideration of her approach, we note that rock art classifications originating in Social Sciences disciplines are sometimes less than felicitous. Like attempts to describe religious experience (shamanism included) in objective terms, they may be awkward and/or reductive. In Australia, the pioneering rock art researcher McCarthy (1967) classified first and foremost by distinguishing "engravings" from "paintings", further subdividing each of these into various techniques. Maynard, contra McCarthy and in revisionist mode, based the idea of style on "technique, form, motif, size and character" (Maynard 1979, p. 99). Some of these categories are probably more useful than others, but, like many researchers, Maynard relied on recorded lists of motifs and so gave image content high priority. In the United States, Schaafsma (1985, pp. 250-51), citing varied support, in particular Hedges (1982), critiqued Steward (1929) for his exclusive focus on motif ("element categories"), Grant (1983) for primary reliance on technique, and Heizer and Baumhoff (1962) for reliance on elements and technique. We would make the point that the kinds of classificatory schemes Schaafsma finds wanting have a certain arbitrariness about them. Naturally they bear no relation to categories likely to be applied by the makers of the art. But beyond that, they bear little relation to the phenomenology of the perception of pictures, or to some unified notion of what might constitute rock art-or any art. We might simply take such typologies as constructs, very much abstracted from the art for more or less limited purposes. Schaafsma's criticism is quite specific: while there might just be some use for such classifications, they fail to identify style as it needs to be defined, namely, as "the product of a particular group during a particular time period" (Schaafsma 1985, p. 251). This retains elements of a culture-historical approach, but of a pragmatic rather than ideological kind: someone had to make the art at a given time. So before we concern ourselves with those theoretical issues discussed above, we need to point up the main issue: unlike those classifications Schaafsma has just criticized, hers is anchored in a given context. This is because the idea of style has meaning only insofar as it is situated in time and space. We are not ignoring the complexities of this style/period connection, merely indicating that there has to be a connection, however complex and, from the perspective of the scholar, liable to misinterpretation. This comment addresses our proposition (4) in the previous section of this article. Thus, returning to our five depictions of horses (Figures 2-6), respectively from Australia, India, America, and Europe, and referable to diverse historical periods, these may say something about non-cultural, non-stylistic universalist typologies, but they say nothing about style, other than that they evidence five different styles.
For Schaafsma the historical context, whose reconstruction becomes the aim of rock art research, is given by archaeology. Both in her article (1985) and in those of her books on which we concentrate (Schaafsma [1971(Schaafsma [ ] 1994Schaafsma [1972Schaafsma [ ] 19921980), Schaafsma underlines the Davis point that style of itself cannot be a definitive guide to a particular culture-this being our proposition (5) in the previous section of this article. But a concretely contextual notion of style is only part of the story. Its complement is to grasp the nature of style itself as applied to art. Schaafsma stands out among rock art scholars as taking a particularly integrated approach to art style. Style is not merely an aggregate of "bits" (forms, motifs, techniques, functions, character, etc.) but a unified whole. It is not merely a matter of giving equal weight to various factors, or even more weight to one factor, less weight to another, but of having an eye for the art, for its dominant features and their interactions, for its overall impact-in short, an eye for the aesthetic. It is a matter of (informed) observation: "the aesthetics of any given style . . . can be fully understood only through visual inspection" (Schaafsma 1980, p. xi). The term "aesthetic" regularly recurs (Schaafsma 1985, p. 247;2015b, pp. 252-53). What the eye seeks is "pattern" because, contrary to earlier archaeological belief, rock art marks are not random but structured (Schaafsma 1980, p. 6). The interesting outcome of this is not simply aesthetic appreciation. Rather, it makes sense of the whole business of archaeology utilizing rock art-because stylistic analysis is a "sensitive tool" (Schaafsma 1980, p. 1); indeed, the method which (while ideally incorporating all available data) does most to further the aims of archaeology: "more sensitive categorizations . . . articulate more successfully with temporal, spatial, and cultural considerations" (Schaafsma 1985, p. 252). In short, the better the stylistic analysis, the sounder the style/culture nexus-assuming, of course, that dirt archaeology does its job as well.
The above comments remind us that Schaafsma's earliest training was in art history: "those undergraduate classes when Vasari's philosophy, made explicit or not, ran through all those discussions of European painting" (pers. comm. 19 May 2019). Subsequently, she took up the study of anthropology/archaeology. Concisely put in an interview for the issue of New Mexico Historical Review devoted to her work: Q: Are you an archaeologist with an art historian's bag of tools?
(Lauderdale Graham 2015, p. 131) This is, in fact, if not a unique at least a rare combination in rock art studies. So it comes as no surprise that, like Grant (1983) and Wellmann (1979), Schaafsma, both in her Utah book (1971) and that on the southwest (1980), turns to an art historian for support. For her, as for Schapiro, style is "directly grasped" through observation. Because there is (some degree of) form constancy, and an intrinsic connection with time and place, style points to specificities of culture-with, of course, evidential corroboration from the archaeological record. So "specific styles and traditions" may be identified with "specific cultural relationships and chronological definitions" (Schaafsma 1985, p. 244). If this sounds like cultural history, it is because it is in part that, though in modern form and without the generalities associated with earlier practitioners. As Conkey and Hastorf understand, "we are still working with the results created by this approach" (Conkey and Hastorf 1990, p. 3). Still more to the point: no matter how far it is problematized and, in the process, made more nuanced, the culture-history model continues to serve its original purpose of historical reconstruction. At the same time, it is worth reading Schaafsma's appeal to Schapiro very closely. While accepting Schapiro's style/history nexus, she evidences unease insofar as she puts it in the context of Schapiro's own listing of complicating variables, with stress on the necessity of fixing limits "by convention for simplicity in dealing with historical problems" (Schaafsma 1980, p. 7). We know the complexity of styles that exists under the simple rubric of "the Renaissance", but we continue to speak of a "Renaissance style", say as distinct from "Baroque", tolerably confident that there is at any time some limit to complexity (if not one style, at least a limited number of styles). Looking at all this in the light of archaeological theories in the second half of the twentieth century, we also need to note that in addition to Schapiro, Schaafsma cites Binford (Schaafsma 1980, p. 8)-references to aesthetics notwithstanding. This is understandable given the dominance of Processualist functionalism at the time, though it is also the case that Binfordian dogma was on the way to decline. That Schaafsma is perfectly aware of Postprocessualist developments on the horizon may be read in her highlighting of the problematizing aspect of Schapiro's own position-in which connection, it is also noteworthy that Curtis Schaafsma discussed the new "Post" at the time (Schaafsma 1991). Interestingly, this publication sets out to show that a strictly scientific methodology is quite capable of dealing with Postprocessualist problematization.
At any rate, we are left with a picture of Schaafsma's work, which, after the manner of Conkey's reassessment, seeks to retain the element of historical reconstruction inherited by archaeology from its culture-history days, while keeping a skeptical eye on earlier judgements based on generalized data. The fact that Processualist functionalism intervened between culture-history and the period in which both Conkey and Schaafsma operate explains why it became essential for Shaafsma's scholarly generation to complicate aspects of, on the one hand, older historical approaches and, on the other, some of Schapiro's overly-optimistic judgements about the style/history nexus. Thus Schaafsma's language avoids the presumption that a given identifiable style attaches itself to a particular group, empirical or cultural, at a particular time, or indeed to any entity understood as monolithic. Style is used as a discursive construct. The style/history nexus is real, but as a general rule, not necessarily applicable in any particular case. In this way, Schaafsma conforms to the Conkey post-culture-history formula, looking to empirical detail rather than large hypotheses. For example, in connection with the ongoing American debate about the nature of the impressive ruins at Chaco, she characteristically warns against projecting modern scholars' ideas onto Chaco (Schaafsma 2018). If this is not exactly reflexive Cultural Critique, it decidedly expresses bias in favor of interpretation not exceeding data. The same thing emerges in Schaafsma's skepticism about the identification of the Bluff Sand Island image as a mammoth (Schaafsma 2015a). It might be said in the light of all this that, while her project is a grand synthesis of southwest art, she also belongs to the generation of those who (recalling our metaphor) sought to distinguish the many species of trees that constitute a "forest". In other words, she belongs to the generation of Conkey's problematizers-while managing the grand synthesis of southwest art which constitutes her major achievement.
Following this discussion of Schaafsma's position vis-à-vis theoretical changes occurring towards the end of the twentieth century, it is worth going back to her characterization of stylistic analysis as a "sensitive tool". As we see it, the question implicit in her writings is: how might style back up archaeological research? Or, using Schapiro's characterization of style as "the 'inner form' of collective thinking and feeling" (Schapiro [1953] 1973, p. 271): how might style expose something of the inner form of the thinking and feeling of the makers of the rock art-in a way unavailable from the "outer" forms uncovered by archaeology? The answer is precisely in Schaafsma's insistence on the idea of a sensitive tool. It might be said that there are two fundamental ways of approaching the analysis of images: quantitative and qualitative. Schaafsma's strength lies in her appreciation of the way in which analysis as a sensitive tool requires a qualitative approach.
Schaafsma has always insisted that she works "from an archaeological perspective" (Schaafsma 1985, p. 240). Nonetheless, given stylistic analysis as her point of entry, she also criticizes archaeological focus on the economic, the material (Lauderdale Graham 2015, p. 135), stressing the value of what is not to be found in the archaeological record (Schaafsma 1985, p. 237) but is there in the rock art, equally material but as the "immediate material expression" (Schaafsma 1980, p. 344) of the way people order their world. More simply: "patterns of imagery suggest patterns of thought" (Lauderdale Graham 2015, p. 130). This very specifically recalls Schapiro's definition directed at the "synthesizing historian of culture"-a job description that fits Schaafsma-of style as the "'inner' form of collective thinking and feeling" (Schapiro [1953] 1973, p. 271), a passage Schaafsma cites (Schaafsma 1985, p. 246). We are returned to the entirely observable yet conceptually nuanced and therefore elusive aspects of those stylistic "qualities" alluded to in our mention of the Thompson Wash panels above. What kind of language do we require to describe these (still keeping an eye on archaeology)? A qualitative language is of course readily available in art history. What comes to mind for the present authors is Wölfflin's virtuosic account of the stylistic transition from Renaissance to Baroque in terms of Linear as against Painterly (or color-centered) art (Wölfflin [1915(Wölfflin [ ] 1950. However, this may be balanced with Panofsky's devastating account (Panofsky [1921] 1974) of the entirely mechanical, quantified ratios used by art traditions from the time of the Egyptians-outlined by means of prose no less evocative than Wölfflin's. So even art history does not necessarily offer purely qualitative models of analysis. Such complications aside, the question arises for rock art: in light of the (hoped-for) scientific bias of archaeology, ought there not to be a role for quantitative analysis? To quote a teasingly-serious Maynard speaking at the 2015 Sydney University commemoration for John Clegg: "when in doubt . . . count!" One possible outcome of quantification might be to take rock art, starting with the basic element of the motif, as an aggregate of parts able to be assembled to make, for example, an anthropomorph ( Figure 10). We noted above that the same additive methodology may be applied to the notion of style as an aggregate of characteristics. The Lego principle might get at the structural basics of some stick figures. But there are clear limits to its application. For example, the celebrated "Mountford figures", in particular the panel featuring four running female figures, first recorded by the 1948 American-Australian expedition to Arnhem Land (Mountford 1964, Plate 9), have been repeatedly described as "elegant". Given that they consist of a few simple lines, it might be argued that they are good candidates for quantitative analysis. But such analysis, however revealing, would not address the issue of "elegance". This for the reason that there is no objective, that is, measurable language for a quality like "elegance". We suggest another example, that of "monumentality". A Wandjina (Donaldson 2013, pp. 356-57) recorded by the 1938 Frobenius expedition to the Australian Kimberley is certainly measurable: it is more than four meters long. But is its quality of monumentality, reminiscent of those enormous reclining Buddhas illustrating the passing into Nirvana, simply a matter of size? Vasari's contemporaries Michelangelo, Leonardo, and Raphael all produced images we refer to as "monumental", and indeed, some are simply big. But small ones also exhibit the quality of monumentality. And what about a monumental Buddha carved into a grain of rice? So it would seem to be a complicated matter of relative proportions. By all means, let someone try to quantify this data; we think the outcome would be disappointing. Stylistic analysis, understood as Schaafsma's "sensitive tool", calls for a qualitative approach. role for quantitative analysis? To quote a teasingly-serious Maynard speaking at the 2015 Sydney University commemoration for John Clegg: "when in doubt…count!" One possible outcome of quantification might be to take rock art, starting with the basic element of the motif, as an aggregate of parts able to be assembled to make, for example, an anthropomorph ( Figure 10). We noted above that the same additive methodology may be applied to the notion of style as an aggregate of characteristics. The Lego principle might get at the structural basics of some stick figures. But there are clear limits to its application. For example, the celebrated "Mountford figures", in particular the panel featuring four running female figures, first recorded by the 1948 American-Australian expedition to Arnhem Land (Mountford 1964, Plate 9), have been repeatedly described as "elegant". Given that they consist of a few simple lines, it might be argued that they are good candidates for quantitative analysis. But such analysis, however revealing, would not address the issue of "elegance". This for the reason that there is no objective, that is, measurable language for a quality like "elegance". We suggest another example, that of "monumentality". A Wandjina (Donaldson 2013, pp. 356-57) recorded by the 1938 Frobenius expedition to the Australian Kimberley is certainly measurable: it is more than four meters long. But is its quality of monumentality, reminiscent of those enormous reclining Buddhas illustrating the passing into Nirvana, simply a matter of size? Vasari's contemporaries Michelangelo, Leonardo, and Raphael all produced images we refer to as "monumental", and indeed, some are simply big. But small ones also exhibit the quality of monumentality. And what about a monumental Buddha carved into a grain of rice? So it would seem to be a complicated matter of relative proportions. By all means, let someone try to quantify this data; we think the outcome would be disappointing. Stylistic analysis, understood as Schaafsma's "sensitive tool", calls for a qualitative approach. This is not entirely to rule out the usefulness of quantitative methods. Indeed, there is no need for quantitative methods to contradict qualitative ones. In each case, "getting it right" is understood in quite different, arguably equally valid, ways. The point we make is simply that "sensitive" precision is best suited to analysis of style-and that we need to be thoughtful in our appeal to the quantitative. Of course, from the later part of the last century, quantifiable methodology has utilized computers. These are bound to the rule that "what goes in affects what comes out". In addition to choices made in this connection, there are the choices, subsequently less than evident, of the people who made the program we choose to use. Still, with these provisos, who will not opt for computer software that might do the required task? But we need to be clear about hermeneutic premises we might adopt. For example, the Lorblanchet and Bahn collection discussed above includes three examples of quantitative methodologies: Clegg, Smits, and Tratebas. Smits (1993), for example, thinks he can neutralize factors in his sample referable to artist or period (read "stylistic" factors)-in order to arrive at a "functional" conclusion. But this would seem to suggest that style and function mutually exclude each other, which need not be the case. In his 1993 Lorblanchet and Bahn piece, as in an earlier article (Clegg 1977), Clegg also wants to disassociate factors, but, unlike Smits, so as to focus on style. His glory and (we feel) his fatal flaw is that, like Schaafsma, he loved and practiced art. Unlike her, he could not shake off the old Italian demon of mimetic realism. So he took the view, Figure 10. An additive approach to motif. This is not entirely to rule out the usefulness of quantitative methods. Indeed, there is no need for quantitative methods to contradict qualitative ones. In each case, "getting it right" is understood in quite different, arguably equally valid, ways. The point we make is simply that "sensitive" precision is best suited to analysis of style-and that we need to be thoughtful in our appeal to the quantitative. Of course, from the later part of the last century, quantifiable methodology has utilized computers. These are bound to the rule that "what goes in affects what comes out". In addition to choices made in this connection, there are the choices, subsequently less than evident, of the people who made the program we choose to use. Still, with these provisos, who will not opt for computer software that might do the required task? But we need to be clear about hermeneutic premises we might adopt. For example, the Lorblanchet and Bahn collection discussed above includes three examples of quantitative methodologies: Clegg, Smits, and Tratebas. Smits (1993), for example, thinks he can neutralize factors in his sample referable to artist or period (read "stylistic" factors)-in order to arrive at a "functional" conclusion. But this would seem to suggest that style and function mutually exclude each other, which need not be the case. In his 1993 Lorblanchet and Bahn piece, as in an earlier article (Clegg 1977), Clegg also wants to disassociate factors, but, unlike Smits, so as to focus on style. His glory and (we feel) his fatal flaw is that, like Schaafsma, he loved and practiced art. Unlike her, he could not shake off the old Italian demon of mimetic realism. So he took the view, against our first proposition in the "Vasari" section above, that if a picture first and foremost represents the real thing-perhaps because, as with tracks, it is easy to do-then it exhibits, if not no style, at least style at its near-zero minimum. The odd conclusion to this odd assumption (and, to do him justice, Clegg sees some difficulties) is that, by conflating "content" with "naturalism" and eliminating "naturalistic" tracks petroglyphs (read "content"), he thinks that he has isolated the element of "style". All this with a Pearson-r algorithm and on the basis of original use of computerized data by his student McDonald (1982), who sought proof of "naturalism" in tracks at his Sturt's Meadows site. No appeal to Information Technology can override problems at the level of premises. In this context, Tratebas (1993), whose aim is of a different kind from both Smits and Clegg, shows much more caution. Using a Principal Components program reliant, as she says, on a factor analysis text, she believes she has a quantifiable way of identifying stylistic variation at Whoopup Canyon, Wyoming, not evident to the unaided eye. Given that she practices both quantitative and qualitative analysis (pers. comm. 12 April 2019), curiosity prompts our asking whether, after absorbing her results, she returned to test these with observation, and, if so, with what outcome.
In general, we take the view that any approach to rock art, including the quantitative in its many forms, is justified to the extent that it works. At the same time, it appears to be a blunt instrument when dealing with style. Something of the sort would seem to be Schaafsma's view. She accepts that beyond visual inspection, there are "more exact" (Schaafsma 1980, p. 7) analytical methods such as factor and cluster analysis but reaffirms that "although statistical techniques can provide valuable insights, they cannot substitute for a more holistic evaluation based on visual inspection and pattern recognition" (Schaafsma 1985, p. 249). Still more forcefully, she returns to a critique of "materialistic imperialism . . . emphasizing the economic sphere . . . as the fundamental determinant of cultural behaviour . . . more susceptible to measurement and quantification than . . . art" (Schaafsma 1985, p. 267).
Schaafsma's qualitative analysis of rock art is very much her strength (though matched, as noted above, by a capacity for synthesizing her material) and in this connection, we would like to revisit Schapiro's surprising and, to some, doubtless provocative statement that qualitative description of style is more precise than quantitative description (Schapiro [1953(Schapiro [ ] 1973). The reader is asked to bear in mind the two terms, "exact" (Schaafsma referring to the quantitative) and "precise" (Schapiro referring to the qualitative). If, in accordance with general American scholarship, we term the three diverse styles from the Thompson Wash site discussed above (Figures 7-9) "Barrier", "Fremont", and "Historic", we will accept a relation between Barrier figures and the angular Fremont ones and will identify the third Thompson Wash panel as probably Ute and representing a late intrusion of Plains art, quite distinct from the earlier two. If we focus on the first two styles illustrated by the Thompson Wash panels and turn to Schaafsma's stylistic descriptions of Barrier and Fremont examples, we have the following. Figures from the Barrier Canyon type site are described in detail which captures the general outlines of the style. They are "dark, tapering, immobile" anthropomorphs, "frequently ghostly in appearance" (Schaafsma 1980, p. 61); "large staring eyes, bulging heads and the absence or near absence of arms and legs serve to emphasize the spectral aspect" (Schaafsma 1980, p. 64). Schaafsma goes on to comment on the preparation of the rock-support and ways of creating "textural effects": a "spatter technique" contributing to a sense of the "ethereal"; finger application of paint as backdrop for thicker paint, etc. (Schaafsma 1980, p. 66). It may sound like an archaeological reference to "technique". In fact it is equally close to art-historical formal analysis. Likewise, the listing of Fremont features is not so much an archaeology-oriented motif-list as a word-picture equivalent of the representation: "the hallmark of Fremont rock art everywhere is the broad-shouldered human figure in ceremonial regalia. Typically it has a tapering torso and horns or other elaborate headgear. Adornment varies . . . but common embellishments are heavy necklaces represented with dots or as solid yokes . . . " (Schaafsma 1980, p. 166). And so on. In Schapiro's terminology, it is more "precise" than, in Schaafsma's terminology, a more "exact" prosaic description because it does more than supply information, though it does that too. What it conveys is that elusive "quality" that makes the object the way it is, in short, its "style". This is what Schapiro termed the "inner form" of people's thinking and feeling, "inner" because it comes precisely as "form." Of course, this inner/outer dichotomy is no more than a metaphor, but we can see what Shapiro is getting at: he wants to distinguish between awareness of behavior on the one hand and of thought and feeling on the other. Naturally, a picture, insofar as it has identifiable content, will convey information. Additionally, and through its style, it may generate, via the aesthetic response to formal properties, a more direct engagement. This last being not to what the makers thought or felt, but to how they did so, that is, the manner (maniera) or character with which they did so. The "how" cannot be put objectively without losing its quality, whatever it is that makes, for example, the Thompson Wash Barrier panel both similar to and different from the Fremont one. Still, in the form of qualitative description, the immediacy of how the artists in question thought, felt, and overall experienced their lives, may prompt inferences, either more or less speculative, but nonetheless productive of further research-with, as ever, corroboration from archaeology or ethnography. Schaafsma is willing to postulate those "remote, awe-inspiring" Barrier figures as "imbued with spiritual power" (Schaafsma 1980, p. 71), that is, as candidates for a shamanic hypothesis. By contrast, Fremont figures suggest hunting and warfare, perhaps especially in the Classic Vernal variant, so strikingly "forceful and imposing" (Schaafsma 1980, p. 171). In this passage, Schaafsma goes on to art-historical formal analysis of a very evident sort, drawing attention to "clarity of form and a nice use of contrast between line, textural effects and solid areas", as well as "bold use of both angular and curved elements [which] adds further impressiveness . . . " (p. 171). We may sum up the above by saying that Schaafsma perfectly illustrates the point we wish to make about the relevance of a qualitative approach in dealing with style.
However, following our account of Schaafsma's methodological concerns, we need to turn to the larger taxonomic issue foreshadowed at the start of this section. Stylistic analysis, critically applied, has specific use in classifying visual data. In connection with this and above all else, Schaafsma is a scholar with a job to do. The job in question in her three major rock art books (1971, 1972 and 1980), as well as in later texts, is both simple and dauntingly complex: to make order out of the apparent chaos of rock art in her designated, and image-rich, region. In her own words: "the nature of rock art is that one is commonly presented with a bewildering array of imagery that may have been executed over several hundred years by several cultural groups. The sorting of this imagery into style categories is a basic step in the organization and description of the data" (Schaafsma 1985, p. 247). It is a method frequently adopted in the United States as well as elsewhere, including Australia, with Chaloupka (1993) writing on Arnhem Land on the backdrop of radical ecological/cultural change, and Walsh (2000) on the Kimberley plateau-both with a fine eye for the art, if marred in Walsh's case by one misguided premise, Vasari's chestnut of progress and decline. Schaafsma has both the eye and the judgement. Her argument, most definitively put forward in her book on the entire southwest (Schaafsma 1980), operates within a previously-established archaeological framework, something on which she insists. Its four pillars are the horticultural zones of Hohokam, Anasazi, Mogollon, and, last to be recognized, Fremont-in this case characterized as semi-horticultural. But while the archaeology is there for her, no one has incorporated the vast body of rock art in her area into a coherent geographical/historical scheme. She has the Pecos classification as a partial outline. For a start, she needs to bookend her four major presumed stylistic variants with, on the one hand, Archaic art such as Great Basin (Figure 11), investigated by Heizer and Baumhoff (1962), and, on the other, Historic art of the Athabaskan-speaking Navajo/Apache, gradually arriving at separate identities ( Figure 12). Between these, she offers a series of detailed critical analyses which add up to a complete archaeology-based stylistic taxonomy for the southwestern United States. Here we simply comment on two examples of such analysis.  She expands her earlier account of Utah, giving real depth to the idea of Fremont art, with carefully-argued theses concerning the relation of Fremont and Anasazi-and Fremont and Barrier Canyon images, Barrier style clearly separated from Anasazi and given time depth behind both Anasazi and Fremont, that is, most likely located towards the end of the Archaic (Figure 13). This account of Barrier style as representing an early stage (between 1500BP and 2500BP) is an essential move in any attempt to make ordered sense of the Utah material. In fact Shaafsma took a conservative position with respect to Barrier in the 1980 book and continues to do so, stressing the uncertainty of its dating. The case of Glen Canyon Linear, first identified by Schaafsma, would seem to be similar. Finally, there is the large assessment that Fremont artists shared an ideological system with both Anasazi and Barrier "that fostered the representation of heroic and elaborate, possibly shamanic, human figures" (Schaafsma 1980, p. 180). Witness figures from the Great Gallery at Barrier Canyon (Figure 14), frontals above the San Juan at Butler Wash (Figure 15), and Classic Vernal figures at McKee Spring ( Figure 16) and Dry Fork. In addition-but this in the Utah book-Schaafsma makes comparisons between Barrier and Pecos figures illustrated and glossed by Kirkland and Newcomb (1967), suggesting the possibility of a very early representational configuration analogous to the largely (if not solely) non-representational Great Basin tradition ( Figure 17).  She expands her earlier account of Utah, giving real depth to the idea of Fremont art, with carefully-argued theses concerning the relation of Fremont and Anasazi-and Fremont and Barrier Canyon images, Barrier style clearly separated from Anasazi and given time depth behind both Anasazi and Fremont, that is, most likely located towards the end of the Archaic (Figure 13). This account of Barrier style as representing an early stage (between 1500BP and 2500BP) is an essential move in any attempt to make ordered sense of the Utah material. In fact Shaafsma took a conservative position with respect to Barrier in the 1980 book and continues to do so, stressing the uncertainty of its dating. The case of Glen Canyon Linear, first identified by Schaafsma, would seem to be similar. Finally, there is the large assessment that Fremont artists shared an ideological system with both Anasazi and Barrier "that fostered the representation of heroic and elaborate, possibly shamanic, human figures" (Schaafsma 1980, p. 180). Witness figures from the Great Gallery at Barrier Canyon (Figure 14), frontals above the San Juan at Butler Wash (Figure 15), and Classic Vernal figures at McKee Spring ( Figure 16) and Dry Fork. In addition-but this in the Utah book-Schaafsma makes comparisons between Barrier and Pecos figures illustrated and glossed by Kirkland and Newcomb (1967), suggesting the possibility of a very early representational configuration analogous to the largely (if not solely) non-representational Great Basin tradition ( Figure 17). She expands her earlier account of Utah, giving real depth to the idea of Fremont art, with carefully-argued theses concerning the relation of Fremont and Anasazi-and Fremont and Barrier Canyon images, Barrier style clearly separated from Anasazi and given time depth behind both Anasazi and Fremont, that is, most likely located towards the end of the Archaic (Figure 13). This account of Barrier style as representing an early stage (between 1500BP and 2500BP) is an essential move in any attempt to make ordered sense of the Utah material. In fact Shaafsma took a conservative position with respect to Barrier in the 1980 book and continues to do so, stressing the uncertainty of its dating. The case of Glen Canyon Linear, first identified by Schaafsma, would seem to be similar. Finally, there is the large assessment that Fremont artists shared an ideological system with both Anasazi and Barrier "that fostered the representation of heroic and elaborate, possibly shamanic, human figures" (Schaafsma 1980, p. 180). Witness figures from the Great Gallery at Barrier Canyon (Figure 14), frontals above the San Juan at Butler Wash (Figure 15), and Classic Vernal figures at McKee Spring ( Figure 16) and Dry Fork. In addition-but this in the Utah book-Schaafsma makes comparisons between Barrier and Pecos figures illustrated and glossed by Kirkland and Newcomb (1967), suggesting the possibility of a very early representational configuration analogous to the largely (if not solely) non-representational Great Basin tradition ( Figure 17).           The sorting out of these complex mosaics is of course by definition never complete. The present authors are familiar with one illustration of this: the work of Keyser and Poetschat (2017), which extends the range of Classic Vernal figures into southern Wyoming. There is also the "Confluence" panel in southern Wyoming, strikingly reminiscent of the well-known one at Cottonwood Wash, Nine Mile Canyon, which in Shaafsma's scheme comes under the heading of Northern San Rafael-and, as well, there are identifiably Fremont figures in southeastern Utah, at Pipe Spring, a site within the accepted Anasazi area, and with Anasazi material cultural remains (Conti et al. 2018). The classification, in short, continues.
In addition to that of Utah, our other case study must be Schaafsma's cause célèbre: the thesis of a Jornada Mogollon stylistic complex not passed on from the Anasazi, after their migration from the San Juan drainage to the Rio Grande, to their new neighbors, the Mogollon (the "San Juan hypothesis"). Instead, and on the basis of rock art evidence, Schaafsma sees the Anasazi, post-relocation, as coming under the influence of the Mogollon. Thus the change to Anasazi art comes from the south, and the connection with Jornada Mogollon art is not the result of Anasazi influence but rather an influence going from Jornada Mogollon to Anasazi. In the course of this argument, Schaafsma identifies the new Anasazi style, a simplified form of Jornada, calling it the Rio Grande style. In terms of the Pecos classification, it comes under the heading of Pueblo IV. Some of the best examples of Rio Grande-style pictographs come from Abo ( Figure 18) and may be compared with Jornada Mogollon paintings at Hueco Tanks, Texas (Figure 19). Petroglyphs along the length of  The sorting out of these complex mosaics is of course by definition never complete. The present authors are familiar with one illustration of this: the work of Keyser and Poetschat (2017), which extends the range of Classic Vernal figures into southern Wyoming. There is also the "Confluence" panel in southern Wyoming, strikingly reminiscent of the well-known one at Cottonwood Wash, Nine Mile Canyon, which in Shaafsma's scheme comes under the heading of Northern San Rafael-and, as well, there are identifiably Fremont figures in southeastern Utah, at Pipe Spring, a site within the accepted Anasazi area, and with Anasazi material cultural remains (Conti et al. 2018). The classification, in short, continues.
In addition to that of Utah, our other case study must be Schaafsma's cause célèbre: the thesis of a Jornada Mogollon stylistic complex not passed on from the Anasazi, after their migration from the San Juan drainage to the Rio Grande, to their new neighbors, the Mogollon (the "San Juan hypothesis"). Instead, and on the basis of rock art evidence, Schaafsma sees the Anasazi, post-relocation, as coming under the influence of the Mogollon. Thus the change to Anasazi art comes from the south, and the connection with Jornada Mogollon art is not the result of Anasazi influence but rather an influence going from Jornada Mogollon to Anasazi. In the course of this argument, Schaafsma identifies the new Anasazi style, a simplified form of Jornada, calling it the Rio Grande style. In terms of the Pecos classification, it comes under the heading of Pueblo IV. Some of the best examples of Rio Grande-style pictographs come from Abo ( Figure 18) and may be compared with Jornada Mogollon paintings at Hueco Tanks, Texas (Figure 19). Petroglyphs along the length of The sorting out of these complex mosaics is of course by definition never complete. The present authors are familiar with one illustration of this: the work of Keyser and Poetschat (2017), which extends the range of Classic Vernal figures into southern Wyoming. There is also the "Confluence" panel in southern Wyoming, strikingly reminiscent of the well-known one at Cottonwood Wash, Nine Mile Canyon, which in Shaafsma's scheme comes under the heading of Northern San Rafael-and, as well, there are identifiably Fremont figures in southeastern Utah, at Pipe Spring, a site within the accepted Anasazi area, and with Anasazi material cultural remains (Conti et al. 2018). The classification, in short, continues.
In addition to that of Utah, our other case study must be Schaafsma's cause célèbre: the thesis of a Jornada Mogollon stylistic complex not passed on from the Anasazi, after their migration from the San Juan drainage to the Rio Grande, to their new neighbors, the Mogollon (the "San Juan hypothesis"). Instead, and on the basis of rock art evidence, Schaafsma sees the Anasazi, post-relocation, as coming under the influence of the Mogollon. Thus the change to Anasazi art comes from the south, and the connection with Jornada Mogollon art is not the result of Anasazi influence but rather an influence going from Jornada Mogollon to Anasazi. In the course of this argument, Schaafsma identifies the new Anasazi style, a simplified form of Jornada, calling it the Rio Grande style. In terms of the Pecos classification, it comes under the heading of Pueblo IV. Some of the best examples of Rio Grande-style pictographs come from Abo ( Figure 18) and may be compared with Jornada Mogollon paintings at Hueco Tanks, Texas ( Figure 19). Petroglyphs along the length of the Rio Grande, especially accessible at the large complex known as West Mesa across the river from Albuquerque (Figure 20), echo such Jornada complexes as Three Rivers (Figure 21).              Stylistic analysis of the relation of Jornada and Rio Grande styles leads Schaafsma to her more specific thesis of a Mesoamerican influence on Pueblo IV (i.e., later Anasazi) art as a result of increased contact with the Jornada Mogollon. This would seem better to explain change in Anasazi art than did the San Juan hypothesis. Indeed if we insist on a north-to-south influence, changes from Pueblo III to IV are very hard to understand. Schaafsma sees instead a new ideology, expressed in new imagery, which is relatable to the myths and imagery of central Mexico. She argues that in the process of the shift of emphasis to the Rio Grande settlements, the Anasazi radically alter their iconography, along with, presumably, their own myths and rituals-now associated with, among others, the southern triad of rain-deities Quetzalcoatl ( Figure 22) and Tlaloc (Figure 23), as well as the war-deity, the planet Venus ( Figure 24). Likewise, she suggests that the immense preponderance of the mask (Figures 25 and 26) motif is itself a legacy of Mesoamerica. Much of this imposing iconographic complex and the stories it carries metamorphosing over time into the beliefs and practices of the descendants of the Anasazi and Mogollon, the modern Pueblo. The critical point to stress, however, is that the Mesoamerica thesis is essentially built on the evidence of the vast data of New Mexico rock art, in this case independently of archaeology-though without contradicting archaeology. It is, in other words, an example of what can be done with painstaking and judicious style analysis, and by a broad comparative knowledge of iconography extending from the southwestern USA to central and southern Mexico.  Stylistic analysis of the relation of Jornada and Rio Grande styles leads Schaafsma to her more specific thesis of a Mesoamerican influence on Pueblo IV (i.e., later Anasazi) art as a result of increased contact with the Jornada Mogollon. This would seem better to explain change in Anasazi art than did the San Juan hypothesis. Indeed if we insist on a north-to-south influence, changes from Pueblo III to IV are very hard to understand. Schaafsma sees instead a new ideology, expressed in new imagery, which is relatable to the myths and imagery of central Mexico. She argues that in the process of the shift of emphasis to the Rio Grande settlements, the Anasazi radically alter their iconography, along with, presumably, their own myths and rituals-now associated with, among others, the southern triad of rain-deities Quetzalcoatl ( Figure 22) and Tlaloc (Figure 23), as well as the war-deity, the planet Venus ( Figure 24). Likewise, she suggests that the immense preponderance of the mask (Figures 25 and 26) motif is itself a legacy of Mesoamerica. Much of this imposing iconographic complex and the stories it carries metamorphosing over time into the beliefs and practices of the descendants of the Anasazi and Mogollon, the modern Pueblo. The critical point to stress, however, is that the Mesoamerica thesis is essentially built on the evidence of the vast data of New Mexico rock art, in this case independently of archaeology-though without contradicting archaeology. It is, in other words, an example of what can be done with painstaking and judicious style analysis, and by a broad comparative knowledge of iconography extending from the southwestern USA to central and southern Mexico. Stylistic analysis of the relation of Jornada and Rio Grande styles leads Schaafsma to her more specific thesis of a Mesoamerican influence on Pueblo IV (i.e., later Anasazi) art as a result of increased contact with the Jornada Mogollon. This would seem better to explain change in Anasazi art than did the San Juan hypothesis. Indeed if we insist on a north-to-south influence, changes from Pueblo III to IV are very hard to understand. Schaafsma sees instead a new ideology, expressed in new imagery, which is relatable to the myths and imagery of central Mexico. She argues that in the process of the shift of emphasis to the Rio Grande settlements, the Anasazi radically alter their iconography, along with, presumably, their own myths and rituals-now associated with, among others, the southern triad of rain-deities Quetzalcoatl ( Figure 22) and Tlaloc (Figure 23), as well as the war-deity, the planet Venus ( Figure 24). Likewise, she suggests that the immense preponderance of the mask (Figures 25 and 26) motif is itself a legacy of Mesoamerica. Much of this imposing iconographic complex and the stories it carries metamorphosing over time into the beliefs and practices of the descendants of the Anasazi and Mogollon, the modern Pueblo. The critical point to stress, however, is that the Mesoamerica thesis is essentially built on the evidence of the vast data of New Mexico rock art, in this case independently of archaeology-though without contradicting archaeology. It is, in other words, an example of what can be done with painstaking and judicious style analysis, and by a broad comparative knowledge of iconography extending from the southwestern USA to central and southern Mexico.           We will end by going back to the larger question of style in art with which we have been concerned in this article.
First, we add two propositions to the six given in the first and second sections above: (7) "exact" quantitative analysis of art is applicable wherever measurement is possible (8) "precise" qualitative analysis is applicable where measurement is impossible or simply unsuited to the aim of (historically contextualized) analysis of style The reason for our choice of Schaafsma's work on rock art as an example of the procedures and advantages of a qualitative approach to iconography, as well as of the uses of style analysis for visual data classification, should now be clear. We illustrate not least Schaafsma's own point that archaeology badly needs to find ways of incorporating art analysis into its methodology. As we have seen, Schaafsma's analysis remains anchored in extra-stylistic evidence but in some instances comes up with results which cannot be got by archaeological means alone. Looking at the wider picture indicated by our title "points of departure for a discussion of style in rock art", Schaafsma illustrates what analysis of style has to offer to studies of rock art. Beyond that, and given her art history qualifications, it suggests avenues for art history itself which, while it long ago refined its own methods of stylistic analysis, has not yet found adequate ways of incorporating the vast body of rock art into its discipline.
Style in art is a very subtle thing-and, with our (1) to (8) summarizing points, we have barely begun treatment of the issues it raises, at best laying the groundwork for an understanding of these. We stress that, while the eight points represent key steps in our argument, they should not be taken as conclusions. This is why we characterize our procedure as clearing the ground. They are basic propositions intended to initiate discussion of style-all of which returns us to our title, with its emphasis on "points of departure". Style is an extremely complex phenomenon. Against this, and at the same time-as Schapiro tells us and Schaafsma shows-it offers a simple, immediate and direct engagement with the past-in the field of rock art with artists who may never be otherwise known to us.  We will end by going back to the larger question of style in art with which we have been concerned in this article.
First, we add two propositions to the six given in the first and second sections above: (7) "exact" quantitative analysis of art is applicable wherever measurement is possible (8) "precise" qualitative analysis is applicable where measurement is impossible or simply unsuited to the aim of (historically contextualized) analysis of style The reason for our choice of Schaafsma's work on rock art as an example of the procedures and advantages of a qualitative approach to iconography, as well as of the uses of style analysis for visual data classification, should now be clear. We illustrate not least Schaafsma's own point that archaeology badly needs to find ways of incorporating art analysis into its methodology. As we have seen, Schaafsma's analysis remains anchored in extra-stylistic evidence but in some instances comes up with results which cannot be got by archaeological means alone. Looking at the wider picture indicated by our title "points of departure for a discussion of style in rock art", Schaafsma illustrates what analysis of style has to offer to studies of rock art. Beyond that, and given her art history qualifications, it suggests avenues for art history itself which, while it long ago refined its own methods of stylistic analysis, has not yet found adequate ways of incorporating the vast body of rock art into its discipline.
Style in art is a very subtle thing-and, with our (1) to (8) summarizing points, we have barely begun treatment of the issues it raises, at best laying the groundwork for an understanding of these. We stress that, while the eight points represent key steps in our argument, they should not be taken as conclusions. This is why we characterize our procedure as clearing the ground. They are basic propositions intended to initiate discussion of style-all of which returns us to our title, with its emphasis on "points of departure". Style is an extremely complex phenomenon. Against this, and at the same time-as Schapiro tells us and Schaafsma shows-it offers a simple, immediate and direct engagement with the past-in the field of rock art with artists who may never be otherwise known to us.
Author Contributions: L.D. and P.D. researched and photographed the rock art sites illustrated here, read the literature, conceived and designed the manuscript, and together revised the final version. With critical interventions from P.D., L.D. wrote the paper. Both authors contributed substantially to this work.
Funding: This research received no external funding.