Preferred Citation: Davis, Whitney. Masking the Blow: The Scene of Representation in Late Prehistoric Egyptian Art. Berkeley:  University of California Press,  c1992 1992. http://ark.cdlib.org/ark:/13030/ft7j49p1sp/

Masking the Blow The Scene of Representation in Late Prehistoric Egyptian Art Whitney Davis UNIVERSITY OF CALIFORNIA PRESS Berkeley · Los Angeles · Oxford © 1992 The Regents of the University of California

Preferred Citation: Davis, Whitney. Masking the Blow: The Scene of Representation in Late Prehistoric Egyptian Art. Berkeley:  University of California Press,  c1992 1992. http://ark.cdlib.org/ark:/13030/ft7j49p1sp/

For Alexander Marshack, and his questions

Preface

In this book I interpret a group of late prehistoric Egyptian representations that deserve to be more widely known among art historians, archaeologists, and anthropologists. While writing it I have been thinking of artifacts and images from other cultures—Upper Paleolithic engravings; rock art of the Tassili Mountains, Arnhem Land, or the Brandberg; "animal-style" art from Luristan to Scythia and beyond; the paintings of the Catacombs; late antique ivories; Celtic metalwork; the rock-cut signs of Bronze Age Scandinavia; Viking woodwork; Navaho sand painting; and modern urban graffiti. If art historians convert these artifacts into pure images, archaeologists convert the images into pure artifacts. Both art historians and archaeologists who think anthropologically or historically take the artifact-signs as complete in themselves. Anthropology and history want them to be indexical, iconic, or symbolic wholes and stylistic, functional, or ideological markers—objects in which artifact and representation somehow coalesce without remainder or disruption—rather than made things, elements in a chain of replications, and complex, always incomplete, mediations of intentionality.

I have kept examples other than the late prehistoric Egyptian—they could be unlimited—out of the way and discuss the larger issues briefly. The few objects I have chosen to discuss have undergone exhaustive reinvestigation. For example, to my knowledge no earlier commentator had noted the substantial recutting on the Narmer Palette—not modern alteration but a telling sequence of revisions and slips in the "original" making.

The plan of the book is straightforward. Chapter 1 introduces the problems and possibilities of an interpretive history of late prehistoric Egyptian representation, considering both the evidence and some aspects of historical and critical

method. I note questions of theory as such—for example, the more or less intractable issue of intentionality in the replication of artifacts or images. And I provide information about the date, archaeological context, and function of the objects. Chapters 2–7 interpret a series of late prehistoric images produced about 3300–3000 B.C. , before the emergence of the dynastic state in ancient Egypt with its tradition of official, or canonical, image making. I recommend that readers study the illustrations before reading through my discussions; although I do describe the images, my remarks assume familiarity with what can be seen, and, because the images are complex, some elements of the analysis may be hard to follow without reference to them. Chapters 8 and 9 consider one object, the so-called Narmer Palette, frequently regarded as standing at the beginning of the canonical tradition, about 3000 B.C . Within Egyptology there is a consensus—an important but problematic one—that the Narmer Palette represents a new departure for Egyptian image making. Although I want to be cautious about this point of view, I believe the Narmer Palette requires a detailed, independent assessment; furthermore, in my treatment of it I investigate some questions that I do not address for the images considered earlier in the text.

Finally, in the Appendix, I outline my approach to pictorial narrative, the most immediate background—although not the only background—needed to understand the nuts and bolts of my particular reading of late prehistoric Egyptian images. This discussion considers the ladder used to reach a certain level of substantive analysis—a ladder that can be thrown away once that level is attained. But prehistorians and Egyptologists may well ask about the basis on which I have put forward the substantive account. Since historical or archaeological confirmation for any interpretation of late prehistoric Egyptian art is sketchy at best, an evaluation of my account will depend partly on the theoretical and methodological grounds—on theoretical consistency, for example—discussed in the Appendix. And because my approach to pictorial narrative differs from those of others who have written on prehistoric and Egyptian art, the Appendix also defines my terms and sets out my response to related issues.

One general remark about procedures is in order here. In the broadest sense this book is about the positions people have in, and in relation to, representation. Where the gender of these people is known—for example, the enemies of

the ruler in several of the images are explicitly depicted as male—I specify it. Otherwise, I tend to use the masculine pronoun in description; in fact, it turns out, no women are directly depicted in any of the images. (I use neuter forms for animals to leave the question open; although there are gender distinctions marked by, and therefore possibly wrapped up in, the metaphorical and narrative structures of the animal images, I have been unable to make much sense of them.) But I occasionally break away from this practice when it is worth recalling that the gender of the most important identities in my story—the ruler, the artist, the viewer—is often not known and probably included women. For example, the ruler figured in the images could be—the possibilities are not exhaustive or mutually exclusive—father, mother, hunter, warrior, shaman, matriarch, headman, king, or queen. Viewers of either gender could have occupied, imagined, or identified with these positions, although probably in different ways. Parallel considerations apply to ethnic or racial identity. In some cases the viewer of an image may have had the same ethnic origin as the "enemy" depicted in it. Someone identifying with the position of the "ruler" who did not share the ruler's depicted ethnic identity—in one image he is depicted as an Upper Egyptian—would presumably have had an experience of the image different from that of someone who shared this identity. And so forth; again, I leave the question open where possible.

Curators in Egypt, Europe, and America have kindly allowed me to examine objects in their care, often at very close quarters, and have provided photographs, while several scholars have allowed me to use their drawings of particular images. Individual acknowledgments are rendered in the illustration captions. The artist who prepared the diagrams for this publication, Jandos Rothstein, deserves special credit for his patience and care. Research and travel for the initial draft, completed in 1988 and 1989, were supported by a grant from the Office of Research and Sponsored Projects of Northwestern University. I owe special thanks to Wolfgang Kemp, a faculty member at the Institute for Theory and Interpretation in the Visual Arts sponsored by the National Endowment for the Humanities at the University of Rochester in 1989, for piquing my interest in such black holes of art history as constitutive blanks and imaginary artifacts. Students in two graduate seminars at Northwestern, especially

Laura Weigert, have helped me to develop my analysis. Two anonymous readers for the University of California Press and the art historians Robert S. Bianchi, Celeste Connor, and James Marrow provided detailed comments that have helped to shape the final draft, completed with the support of a Humanities Research Award from Northwestern University for 1989–90. Throughout the process of turning a briefer presentation of some ideas into a complete monograph, it has been a pleasure to work with Deborah Kirshman and her colleagues at the press.

Alexander Marshack has been asking questions that I am not sure any archaeology or art history could answer, and I am not sure he would be satisfied with what I offer here. But I cannot imagine this book without the example of his work on prehistoric marking and symbolic systems. I am delighted to dedicate it to him.

WD, MAY 1992

Chronological Table

1—
History and the Scene of Representation

The register ground line is the most important compositional device of Egyptian canonical representation, the official image-making tradition of the ancient Egyptian state from its establishment about 3000 B.C. until its dissolution in the Hellenistic and Roman periods (Figs. 1, 2). The ground line holds animal or human figures apart. It orders and fixes them, possibly overlapping them slightly, in isocephalous rows. Finally, it orients them in a consistent direction, often toward a figure of authority also depicted in the image—if not always in the same register—and generally the patron of the work, who takes possession of them as his estate.

Register, Composition, Image

Although the register ground line is a compositional device, it does not necessarily delimit the image, understood, as it is throughout this study, as a pictorial statement of an often complex reference using a variety of available "textual" resources like narrative and metaphor. For example, an individual register might frame a group of animal or human figures, but often the action of these figures and the "meaning" of the group cannot be understood without referring to an official, monarch, or divinity depicted not in that band but rather elsewhere in the image. Register bands frame elements of an image—namely, those that are literally depicted on and by the surface of the pictorial medium as it has been cut up and organized. The elements of a literal depiction in any particular passage of an image, however, are not necessarily the same thing as the image itself, the semiotic whole that functions as a narrative or other kind of sign.

Fig. 1.
Principles of compositional organization in canonical Egyptian art, illustrated by unfinished
painted relief, tomb of Perneb, Old Kingdom. After Williams 1932.

Fig. 2.
Tomb relief, exterior wall of tomb of Nofer, Old Kingdom. 
Courtesy Museum of Fine Arts, Boston.

As a compositional device a register might not only frame but also organize an image consisting of several discrete passages of depiction, or "pictorial text." The great painted caves of the Magdalenian period in Upper Paleolithic southwestern Europe, about 15,000 B.C ., provide instructive examples of the relations between composed passages of depiction and the image as a totality (see Leroi-Gourhan 1967, 1986). The caves sometimes contain thousands of figures organized in distinguishable compositions or passages of pictorial text on separate "panels." Nevertheless, each cave might amount to a single complex image, perhaps a narrative or other kind of semiotic structure making use of textual modalities other than depiction (see Vialou 1981, 1982). This image would necessarily be taken in by a viewer in a drawn-out act of viewing that

includes several techniques of "reading." (Conceivably the act of making extended over a considerable time as well; see Marshack 1977.) Even on a tiny object—like the decorated combs and cosmetic palettes we are examining here—the smallest passage of pictorial text, apparently bounded by a framing device that might simply be the edge of the object itself, can be organized internally as an image or images. In sum: composition is the organization of depiction, both within and among its divisions or frames; passages of depiction present—are the "text" of—the image; and the image is the concatenation of passages of depiction functioning as a referential whole for a viewer.

In the Egyptian canonical tradition image makers generally used several registers to organize an image. Composition therefore takes place both within registers—obeying the separateness, the isocephaly, and the constancy of direction of figures, among other criteria (see also Davis 1989: 29–37)—and between registers as they are arrayed on the surface of a wall, within the larger architectural setting, and even within the context of an entire building, building complex, site, or territory (see Tefnin 1979, 1981, and 1983 for fundamental considerations). Figures, registers, and images tend to be organized hierarchically with no extraneous or competing pictorial matter. They can be accompanied by hieroglyphic texts, rebuslike signs, and other symbols that perform complementary, parallel, or identical referential operations (Fischer 1986), with "picture" and "hieroglyph" often working together to constitute the "image." In the canonical tradition an entire decorated object or monument—an item or suite of furniture, a three-dimensional sculpture with applied depictions in relief and incised hieroglyphs, a painted tomb chapel, or a temple—sometimes functions as a single image. Whatever its configuration, this image was taken in by a viewer in a complex, hierarchically organized act of viewing governed by social realities and conventions. For instance, not every viewer could "read" all passages of the pictorial text: some were presented in hieroglyphic writing incomprehensible to many viewers, and some were sacred precincts visited only by the specially qualified. Nonetheless, however complicated it may be as a physical entity, the object or monument presents a reference that was legible in part or as a whole to a few or to many viewers as that reference.

As this description implies, the canonical Egyptian image is an intricate affair. In the real space before the decorated wall, the artist, the official or royal patron, and other viewers face the register band at right angles to the direction of movement of its depicted figures. Human figures within the band are depicted not in absolute profile but in the well-known "frontal/profile" aspect (Davis 1989: 27–29). When reversal of the orientation of figures, detachment of figures from the ground line, and so forth, occur in Egyptian canonical representation, it is because every scene was designed to fit a specific architectural context, and the patron's individual requirements for story and symbol had to be met. But every figure has a "space" where it stands out from the ground of the image in a clear silhouette (see also Groenewegen-Frankfort 1951; Smith 1965; Schäfer 1974; Davis 1989: 7–37, with full references).

Late Prehistoric Images

These features of register composition cannot be assumed to have been present in the same way in Egyptian image making before the consolidation of canonical conventions, a process that began in the early First Dynasty and was apparently not completed until the Third (Davis 1989: 116–19). For example, in prehistoric or early dynastic image making the viewer's line of sight might not have been fixed, by the way figures and composition are constructed, at exact right angles to the line of movement of the figures depicted within the image. If the perspective for the viewer of a precanonical image was different in relation to the figures depicted from that of the viewer of the canonical image, despite continuities between them in the technique of rendering individual figures, the two types of image could have been based on different pictorial principles. It is on the basis of such differences, in fact, that we must distinguish canonical and noncanonical image making in the first place.

The essential point is to avoid assuming that the works under discussion function as images exactly as do canonical images. Hence I avoid labels that would associate them with canonical images, if only as their predecessors. Both types of image have an intricate structure, an often highly ingenious means of setting up figures and compositions in organizing a passage of depictive text.

In investigating the nature of noncanonical image making, I look here at the emergence of the canonical register ground line in six major works of the "earliest"—or what I call late prehistoric —Egyptian art (late predynastic, proto-dynastic, and very early dynastic): the Brooklyn and Carnarvon carved-ivory knife handles and the Oxford, Hunter's, Battlefield, and Narmer carved-schist cosmetic palettes. My reasons for preferring the unfamiliar historical designation "late prehistoric" for these objects will become increasingly obvious as we consider in detail how they should be characterized. In general, the term is useful not least because it is slightly paradoxical. On the one hand, it is illuminating to look at the images from the vantage point of later canonical art, so long as we avoid anachronism. For example, when examining the selective conventionalization of an inherited prehistoric visual culture by artists of the later dynastic or canonical tradition (Davis 1989: 116–91), it is acceptable to label objects as "predynastic" (Williams and Logan 1987), "precanonical" (Davis 1989), or "Preformal" (Kemp 1989) because we are considering the images only as they were viewed by later viewers . But on the other hand, if we are looking at the images from the vantage point of the immediate world in which they were first made to make sense, then teleological and possibly anachronistic labels like "predynastic" are misleading. Within the so-called predynastic period, no artist or viewer could have been self-described as "predynastic," "precanonical," or "Preformal," although he or she certainly did belong to an earlier tradition and may even have had a firm sense of his or her "late" or "latest" position within it. Although no general label suits all purposes, we should seek to use a term representing the real position of the works in the replicatory sequence to which they belonged. A word like "protodynastic" might capture some of the nuances, especially for transitional works like the Narmer Palette (Chapters 8, 9), but it lacks one usual connotation of "prehistoric"—designating society without script—that is important to preserve in relation to the images considered in this book. While I do not focus on the topic as such, I believe late prehistoric images to be intimately implicated in the development of Egyptian hieroglyphic script itself. They are the "writing" that precedes the appearance of a secondary transcription system, the writing of writing, in hieroglyphic script.^[1]

Where appropriate I refer to existing interpretations of particular works by writers like Georges Bénédite (1916, 1918), one of the first to study the images

systematically, or by Helene J. Kantor (1944) and Elise J. Baumgartel (1960), among the first to make use of archaeological evidence in interpreting them. I cite also the iconographical or more broadly iconological speculations of Egyptologists and prehistorians like Henri Asselberghs (1961), Michael Hoffman (1979), Elizabeth Finkenstaedt (1984), Bruce Williams (1988a), Bruce Williams and Thomas Logan (1987), and Barry Kemp (1989). I do not, however, take up every point of agreement or disagreement with all these writers. Except for a pioneering structural analysis of the Hunter's Palette by Roland Tefnin (1979), most available accounts of late prehistoric image making merely list the various figures or motifs, make iconographical comparisons, or offer generalized remarks about the apparent themes of individual passages of depiction. (Williams [1988a] analyzes what he calls the "structural logic" of late prehistoric representation, but he does not employ accepted iconographical, structuralist, or semiological procedures in his descriptive-comparative compilation of similarities among motifs and their supposed "meanings.") By contrast, my principal goal is to account consistently for many features of the images, including some of the most striking, that have remained unexplicated, misunderstood, or even unnoticed.

The task of identifying late prehistoric images as such—getting beyond existing descriptions of motif, composition, and passages of depiction to discover their referential coherence as images—is more difficult than it looks. Although the objects have been the focus of a good deal of writing already (see, for example, Hoffman 1979; Finkenstaedt 1984; Kemp 1989) and have played a major role in the general interpretation of Egyptian art, I believe that their pictorial dynamics have not been completely appreciated.

A part of my purpose is to supplement a brief commentary on the images offered in a study of the canonical tradition in Egyptian art (Davis 1989: 136–71). In that account I was interested in late prehistoric objects for what they can tell us about the origins of later Egyptian artistic conventions. For example, we can understand canonical Egyptian art better if we know that its conventions for rendering the overall aspect of the human figure were deeply rooted in multifarious ancient visual contexts, dating as far back as the fourth and fifth millennia B.C ., while the conventions for the proportions of the figure were of relatively recent vintage and had developed in one particular context at

the end of the fourth millennium B.C . Such information suggests that although the canonical artists and patrons made use of—took account of and made reference to—a widely accepted and historically diffused image of the human body, they also reconfigured that image, grafting onto it new formal features and presumably revising its connotations as they had been inherited. Although throughout my earlier analysis I insisted on the non- or precanonical status of prehistoric and predynastic images in relation to the later canonical tradition—they are good examples of what Barry Kemp (1989) calls the "Preformal" tradition always characteristic of some levels of or moments in Egyptian society—I did not consider late prehistoric images in their own terms.

In this book I reverse the emphasis of my earlier approach. Instead of examining the long-term dynamics of a complex, mostly literate tradition, looking at individual images only insofar as they might exemplify general aspects of that tradition, I consider a few images in detail. Partly because the issue has been treated comprehensively by other writers, I do not concern myself too much about their place, noncanonical and canonical, in the larger prehistoric and dynastic traditions, except in one case, that of the Narmer Palette (Chapters 8, 9), where the issue is especially pressing. By definition such a close focus has disadvantages in not considering the sources for late prehistoric representation. Even this disadvantage, however, may be more apparent than real. As we will see, a rigorous distinction should be drawn between the sources for a motif and its "meaning" as used in a particular pictorial text—for example, as a metaphor or a narrative. That we can identify sources for motifs says nothing about their metaphorical or narrative status in particular contexts. For instance, allegorical uses of a motif in continuous replication can take over what had once been narrative uses. If we studied only the continuous development or "transmission" of the motif in or as a stable formal tradition without investigating the "disjunctiveness" of its meaning (Panofsky 1960) as the very condition sustaining that tradition, we would miss entirely such semantic possibilities as metaphor and probably misconstrue the coherence of the tradition as such. A full consideration of the chain of replications, the "tradition," of late prehistoric image making (Chapters 2–6) will enable us to interpret the iconography in a single image belonging to that tradition and sustained disjunctively by revision into new formats and new contexts (Chapters 7–9).

Sequence, Date, Context

The study of late prehistoric representation in the Nile valley is extremely limited in the archaeological evidence to which it can appeal. Unlike works of later Egyptian art, often closely dated and associated with known historical personalities, events, or institutions—like a particular official's ambitions, the cult of a particular temple, or the building projects of a particular monarch—the late prehistoric works cannot usually be fitted into an informative historical context. Even if they have been excavated, the assemblage or site from which they derive is often puzzling.

Objects that lack archaeological provenance altogether have been acquired on the market in the final years of the last century or have turned up periodically since then. Art historians and archaeologists are often inclined to fit such un-provenanced objects into the overall picture of late prehistoric art in Egypt, for example, by making detailed technical and stylistic comparisons that associate unprovenanced with dated specimens on various grounds of morphological similarity (Fischer 1958; and Davis 1981b, 1989: 155–59 are examples of such efforts). Studies of this kind have limited value, however, especially if they examine objects outside the canonical context, in which the forms and meanings of images are likely to vary more than in the canonical tradition. In principle morphological similarity of mannerisms and motifs among a group of objects says nothing about the way in which a mannerism or a motif was used in a specific context in a particular image. Thus the chronological or cultural contiguity of images can be demonstrated only by showing that they are similar both in their technical and stylistic morphology and in their pictorial mechanics and metaphorics. Of several unprovenanced objects thought to be late prehistoric in date, the Oxford, Hunter's, Battlefield, and (to a lesser extent) Narmer Palettes seem to be similar in these latter respects. Therefore they can be regarded as belonging to the same evolving cultural tradition, a coherent series of making reference or what I call a "chain of replications." Furthermore, other unprovenanced objects fall outside this chain of replications because they violate some or all of what I call its general conditions of intelligibility (Chapter 7). I have almost no doubt that some of these pieces are modern imitations; they were manufactured by forgers who had seen the morphology but did not com-

prehend the pictorial structure of known works of early Egyptian art. To understand the mechanics and metaphorics of such images, as I am attempting here, is to establish a foundation for clarifying the forgery of early Egyptian art—making possible a study that is yet to be written.^[2] Other unprovenanced pieces may not be modern forgeries but probably belong to different, if contemporary, chains of replication in late prehistoric art—for example, a chain of replications influenced if not actually produced by foreign craftsmen that is somewhat independent of the series being considered here.

A few sources of documentation help with the preliminary business of dating. The decorated, frequently large-scale knife handles and palettes that I take to be part of a single chain of replications compare closely in outline shape and technique of manufacture with smaller, more routine examples of these artifacts commonly found in predynastic or early dynastic graves and datable in such contexts (see Baumgartel 1955, 1960). Early in this century, after his excavations of tombs with complements of different grave goods, W. M. Flinders Petrie worked out a fairly convincing morphological history for these routine specimens (see Petrie 1920, 1921, 1953). With subsequent corrections in Petrie's typologies and relative chronologies (Kantor 1944, 1965; Kaiser 1956, 1957), we now possess a general framework for dating the large decorated knife handles and palettes, most of which do not derive from dated graves. On the basis of all such typological analysis, the Oxford, Hunter's, Battlefield, and Narmer Palettes appear to have been manufactured in just this relative sequence—a sequence that is broadly accepted by a variety of scholars working with different evidence and interested in different aspects of the objects (for example, Bénédite 1916; Ranke 1925; Fischer 1958; Needler 1984: 329; Davis, 1989: 136–41; and probably Williams and Logan 1987).

In an absolute chronology (see Chronological Table, p. xvii) the works appear to span the period of the "late predynastic" or Nagada IIc/d—Nagada III (the knife handles and Oxford and Hunter's Palettes) through the early First Dynasty (the Battlefield and Narmer Palettes). In real terms, then, they cover at least eight to ten generations, from about 3300 B.C . (Nagada IIc/d) to the end of the fourth millennium (First Dynasty begins about 3000 B.C .). All of them derive from the period when the consolidated dynastic state and the he-

reditary monarchy emerged in ancient Egypt (see Hoffman 1979; Trigger 1983; and Kemp 1983 for general historical accounts of the period). The most critical state structures do not exist at the beginning of this development or they have not, at least, been fully interrelated one with the next: the period witnesses the emergence, increasing consolidation, and mutual coordination of a hereditary ruling elite headed by a powerful sovereign backed by force; a national bureaucracy of finance and administration and a transregional and perhaps even "imperial" organization of long-distance trade and commercial and military expeditioning; what will become a canonical script, art, and liturgy; and a self-proclaimed consciousness of the wholeness of the state legitimated by an ideology of cosmic order and divine kingship. By the end of this development, however, at the end of the fourth millennium, one of the most important elements of the state, an ideology of rule, will have been constituted within the image-making tradition. And in succeeding generations, in the first two centuries of the third millennium B.C ., the state is consolidated as a set of self-maintaining and interlocked institutions and ideologies (for the structure of the Old Kingdom state, see especially Weber 1978: 231–35, 1044–47; Janssen 1978; and Kemp 1983 and 1989, with Davis 1990c).

My interpretation of images decorating the objects shows this or a similar relative sequence and chronological attribution to be the most likely one. Despite a justifiable concern for exact chronology in the literature, however, little of deep interest depends on the specifics of the chronological sequence I accept. My analysis could be slightly rewritten to interpret a group of variously older, contemporary, or more recent depictions drawing selectively, or "disjunctively," on a common tradition of representation predating the complete emergence of the canonical tradition. Thus, for example, the Hunter's and Battlefield Palettes need not be regarded as respectively earlier and later; the archaeological evidence supports the possibility that they could be contemporary. Similarly, the Battlefield and Narmer Palettes need not be regarded as respectively earlier and later; they could also be contemporary, or even—considering the fluid, emergent nature of state institutions and ideology in the period—respectively later and earlier. The chronological or what George Kubler (1962) terms the "serial" position of an image is independent of its textual position—for example, as a

metaphor, allegory, or narrative—in relation to other existing texts and to what I call the general conditions of intelligibility of the possible varieties and species of textuality as such (Chapter 7). Without absolute dates to assign each object on independent archaeological grounds, we cannot always translate the analysis of textual positions into a specification of real chronological or historical relations.

Despite these subtleties, my analysis depends on showing that image makers knew and took account of one another's work as available or possible varieties and species of image making. These textual relations may go in various, sometimes unexpected directions—the sculptor of the Narmer Palette took account of the metaphorics of the Battlefield Palette, the Battlefield sculptor of the Hunter's Palette, and so on—but they are nonetheless particular, specific, and finite relations. For instance, the sculptor of the Brooklyn and Carnarvon knife handles evidently did not take account of the images produced by the Battlefield and Narmer sculptors. These two works are thus textually and almost certainly also chronologically earlier in the chain of replications.

If we discover that an image maker made use of a motif metaphorically—a possibility that is at the heart of my interpretation of the chain of replications of late prehistoric Egyptian image making—then we assume his awareness of other, alternative uses of the motif, metaphorical or nonmetaphorical. It is implicit in the very notion of metaphor that a standard or at least an accepted use of a linguistic or visual expression precedes or coexists with the secondary or "poetic" metaphorical usage (Black 1962; Goodman 1979; Podro 1991, with Davis 1991). Strictly speaking, however, we should not infer, without further evidence, that the maker of the metaphor depended on chronologically prior expressions. Instead, the totality or some substantial part of a system of expression could have been available to him at a given time, and he chose to use it in a "poetic" way while simultaneously other image makers used it in standard ways. This study concerns such textual relations among the replications of an expression—for example, a motif or style of rendition—and in turn between them and the general conditions of intelligibility that define a chain of replications as a series of possibilities in the first place. But, again, we cannot always qualify in chronological or historical terms the textual relations we might iden-

tify; there could be several plausible historical scenarios accounting for them and little or no independent archaeological means of confirming one or another. Whether it is even desirable to do so is a question I leave open for the moment.

Despite the paucity of independent evidence, it is obviously important to consider, where possible, the social and cultural world in which late prehistoric image making had a place. Unfortunately, although we know a good deal about some aspects of prehistoric society in the Nile valley, our sources fail us when we investigate the functions or meanings of images in particular. Beyond the fact that the objects we are observing date to the period of state formation in the late prehistoric Nile valley, we cannot say precisely what they may have assimilated from or contributed to this political and cultural process—broadly speaking, the replacement of relatively autonomous ranked social structures based on mostly regional economies and political interchange ("chiefdoms") by the transregional economy and politics of the state, certainly as much a structural as a chronological development.^[3] It does appear that the human beings depicted on earlier objects in the group (about 3300 B.C .), like the Ostrich Palette, exhibit a relationship to one another and to the natural world different from those on later objects (about 3000 B.C .), like the Battlefield Palette. Moreover, the later objects conform more closely than the earlier to what would emerge by the end of the sequence and in succeeding generations as the principal standards of canonical representation. The crystallization of the Egyptian state is documented by numerous strands of evidence, from the patterning of settlements and cemeteries (for example, see Kemp 1973, 1982; Kaiser and Dreyer 1982; Davis 1983a; and Bard 1987, 1989) to the appearance of trans-regional and even international technological and perhaps conceptual connections (for example, see Frankfort 1951; Kantor 1965; and Davis 1981a). It is hardly surprising, then, that our objects reflect the real reorganization of the social relations and cultural activities of the neolithic economy according to the new rules of the emerging state.

The historical weight of this statement, however—that images probably reflect real social life and change—is actually rather vague in its implications. Insofar as they are representations, the images must also depict, symbolically and metaphorically, these or other social and historical realities—that is, any

and all states of affairs that can be designated by images within the general conditions of intelligibility of representation. Moreover, the activity of depiction is fully part of, perhaps an active and creative part of, social reality and development: making reference to reality is one of the principal components of reality. But in this regard, when we come to investigate the image makers' knowledge about, attitudes toward, and desires for what they chose to represent, we cannot place the objects specifically beyond recognizing them as instances of the conceptual labor of their time. The images actually constitute our primary evidence for their makers' knowledge, attitudes, and desires and for the nature of their engagement with social reality and development.

Notwithstanding the attempts of some Egyptologists, it would be naïve to take the depictions as straightforward pictorial documents of or for the events and processes of state formation. If we link the representation with the historical reality of what is supposedly documented by the image, we risk mistaking ideology and rhetoric for historical process itself. Whatever was really occurring in the villages, emergent regional polities, and nascent transregional institutions of late prehistoric Egypt, the images necessarily and by definition re-present this reality. They tell a story, perhaps the story of a history—and this, in turn, may be a history that never occurred. Given that information enabling us to back up these observations with a specific historical scenario is almost totally lacking, it makes as much sense to investigate the images as one of the empirical explanations for the emergence of the state in ancient Egypt as to suppose that we could somehow cite that history as an independent empirical explanation for the making of the images.

If in fact the evidence of chronology is any guide, the images may have played a directly formative role in transforming the political-cultural system. By stating symbolic propositions and narrating past and ongoing events, they framed general conditions about the intelligibility of the world and its history on which social decisions, actions, and institutions could be based. Functioning as relays of knowledge, ambition, and desire, they were not, then, just the results but also the instruments of the reconfiguration of the social order. For example, on archaeological and typological grounds the Brooklyn and Carnar-von knife handles, the Davis comb, and the Ostrich, Oxford, and Hunter's

Palettes should all be dated to the Nagada IIc/d or Nagada IIIa period, well before the climactic phase of state formation in the Nagada IIIb period and early First Dynasty. As images they relate a narrative of a human being's mastery of the natural and social worlds quite in keeping with the ideology of rule that characterized the complex dynastic state. But this is not to say that their narrative derives from or depends on the dynastic state. Rather, the images make sense in the context of an agriculturalist society practicing some real or ritual hunting. Thus, it would seem, the ideology of rule characteristic of the dynastic state was overlaid retrospectively by an earlier iconography of hunting for the ways it provided of referring to realities the artist hoped to retain or reconstitute (see also Davis 1989: 232–34).

We do not know whether the masterful human being depicted in the late prehistoric images had a real referent in late prehistoric society. Perhaps this personage was, in the real world, a successful hunter, a family leader, a powerful matriarch, a village headman, a "chieftain" of an internally stratified local polity, a regional warlord, or a conquering invader; but it is also conceivable—the possibilities are not mutually exclusive—that the image represented a partly or wholly imaginary being or state of affairs in a narrative symbol. The images are certainly complex and subtle enough to admit the possibility that this masterful human being was, among other things, an allegory for the artist himself or herself and probably also for his or her patron or the owner of the object. I use the designation ruler partly because other possible terms for a complex identity—mother, father, leader, chieftain, warlord, pharaoh, god, artist, patron—may be too literal, and certainly cannot be confirmed as historical, and partly because some or all of these real identities may be subsumed in the image as a whole. It is admittedly a leap of inference to assign this ruler to the male gender, as I generally do in this study purely for ease and consistency of exposition. Although the later monarchs of Egypt were usually male, we know little about the gender distributions of various forms of rulership in prehistoric society.

Precisely because we are dealing with representations we should credit the possibility that the ruler's "rule" is only within representation itself. The image making could apply as a complex metaphor not just to one literal situation of

someone ruling but also to several domains of life in which "rule" had an actual or conceptual reality—to all kinds of domestic, economic, military, and political contexts involving different people of different genders, which could be categorized and thus compared according to a flexible, general discourse for rule" as such. It seems to me that only this can explain the consistency and continuity of the late prehistoric tradition. Particular social situations of rule come and go, but all are mediated, as situations, by general conditions of intelligibility and a set of symbols, metaphors, and narratives of and for rule. The reality of political-cultural process at its most fundamental level includes the very representations we might otherwise seek to explicate in terms of political-cultural history. Thus any attempt to take the late prehistoric images as documenting this or that particular personage, situation, event, or context of "rule" would be inadequate to the literal reality of the history of representation.

By the same token, the identities I call the ruler's "enemies" within the representations could as well be interpreted as metaphorical. The antagonists presented in the several late prehistoric images I consider here include hunted animals and great carnivores as well as human beings. Not necessarily limited to the "real" enemies of the Egyptian state, these last could include anyone metaphorically "subject to rule," including the viewer of the image.

If we had reliable information about who made the images, when and where, and to what specific ends, we could then begin to disentangle some of these matters. We might, for example, be able to determine whether the images presented versions of well-known themes to a knowledgeable audience, and therefore could be understood to conform to a well-rooted, currently accepted tradition, or whether they projected the more unfamiliar representations of a controversial ideology. Some sense of these possibilities derives simply from evaluating the images in relation to earlier, contemporary, and even later images. For instance, the Battlefield and Narmer Palettes seem to revise substantially the pictorial text used on the earlier Oxford and Hunter's Palettes in a fashion closely tied to the images made on contemporary decorated knife handles and therefore apparently a well-established, widely distributed system of expression. Moreover, the Narmer Palette betrays some inconsistency in its own procedures and thus can be seen as attempting something genuinely novel.

Uses, Audiences, and the Representational Function

We know the owner of the Brooklyn knife handle by the other contents of his burial assemblage, Grave 32 at Abu Zeidan (Needler 1980, 1984). He was a relatively well-to-do townsman, with standard tastes in grave furnishings, who had the good fortune—possibly from family status, economic success, or political connections—to own one work of superb craftsmanship, luxurious, if somewhat conservative, in its design.^[4] We know virtually nothing specific about the owners of the other objects. Like many examples of late prehistoric art from the Nile valley, the Carnarvon knife handle and the Ostrich, Hunter's, and Battlefield Palettes have no recorded provenance.

By contrast, the Oxford and Narmer Palettes come from the so-called Main Deposit at Hierakonpolis (Quibell and Green 1900–1901). But since it was a heterogeneous deposit of various objects, deposited at least several generations after the manufacture of the earliest works it contained, the context shows only that later owners or viewers somehow valued the earlier decorated objects and determined to preserve them. The motivation for their action—and whether they were interested in the images and could understand them—remains obscure.

Before a recent reanalysis and republication of the field books for the turn-of-the-century excavations at Hierakonpolis in which the objects were unearthed (Adams 1974, 1975), it was not fully clear to Egyptologists that the Main Deposit was in fact a heterogeneous collection. Some writers reasonably interpreted it as a dedication at an early temple of Horus or Hathor at Hiera-konpolis. The objects found within it would then have been offerings by royal patrons, such as the "Narmer" named on the Narmer Palette, to divinities (see Baumgartel 1955, 1960; Asselberghs 1961: 259). Although I do not assume a connection between the palettes from the Main Deposit and temple-cult or temple-furniture dedications at Hierakonpolis, my readings will not be incompatible with such a connection if it is confirmed by archaeological evidence yet to be discovered. Early First Dynasty shrines at Koptos (plausibly reconstructed by Williams [1988b]) included colossal statutes bearing motifs used elsewhere in late prehistoric image making but not necessarily with the same meaning;

these shrines are only suggestive evidence, then, that the decorated palettes were votive in function. The proposal that other palettes come from another early temple, in this case at Abydos (for example, Needler. 1984: 328), is unsupported by independent evidence.

Other accounts of the function of the objects (see Hoffman 1979) suggest that valuable and striking knife handles, combs, and palettes might have been given as gifts by senior members of a group to one another or to specially favored junior members. Likewise they might have been commissioned to commemorate important events in an individual's life. These hypotheses are somewhat independent of the question of their actual use after being given or commissioned. While the largest of the decorated palettes would probably have been impractical for continuous daily use as an ordinary cosmetic palette, we must admit, lacking evidence for a specific function, that we can hardly judge whether the palette was "practical" or not. Perhaps the larger palettes were set up for display in a residence or temple or were stored for use on special ritual occasions. But although some such ceremonial status for the objects is often asserted (see Petrie 1953), there is not a shred of independent archaeological evidence in the matter. In fact, prehistorians call an item "ceremonial" when they have no idea how it was used.

Even if correct, the hypothesis of ceremonial function is not helpful when we do not know the nature of the purported ceremony. Some Egyptologists propose that certain examples of late prehistoric or early dynastic Egyptian art document ceremonies known from contemporary textual sources or inferred from anachronistic parallels with later periods. For example, the Narmer Palette has been said to commemorate a celebration of victory over defeated "northern" enemies or a jubilee known as the Sed -festival (Schott 1950; Vandler 1952); and the Narmer mace head, an object apparently contemporary with and closely related to the palette, has been interpreted (Millet 1990) as documenting a festival of the "Appearance of the King of Lower Egypt" mentioned on the Palermo Stone (bearing part of a Fifth Dynasty text describing earlier "history") as having occurred sometime in the First Dynasty. There are several similar proposals in the literature. But even if the images do depict or document particular ceremonies, actually staged in the early dynastic period or merely imagined as ideal stagings by late prehistoric and early dynastic image makers, it is

not thereby shown that the objects or images were also used in and made for these ceremonies. Certainly nothing is said about the way the images, as depictions or documents, represent the ceremonies.

In implicit contradiction with the "ceremonial" hypothesis, the decorated objects are sometimes supposed to have had a magical function. For instance, they could have promoted success in the hunt or in battle or secured the favor of a totem or divinity; they could have been rendered magically efficacious in these or similar functions by their depictions, by a ritual use of the implement, or by both together in the context of a larger performance. The diversity and vagueness of all such accounts (compare Capart 1905; Baumgartel 1955, 1960; and Asselberghs 1961) confuse the issue. They seem to have been influenced as much by outmoded views of prehistoric or "primitive" art in general as by any reliable documentation. (Some Egyptologists have evidently endowed late prehistoric images with some of the "magical" properties of later dynastic funerary statues or liturgical implements, but we cannot assume functions and meanings closely associated with the developed theology and canonical iconography of the hierocratic state to have been present in the same way in the late prehistoric context.) Decorated objects found in tombs, like the Brooklyn knife handle, may be presumed to have served a general "magical" purpose for the life of the deceased beyond the grave. But despite numerous assertions in the literature, there is no archaeological evidence that the great carved palettes or the images carved on them had a magical function.

My reading suggests that construing the works as ceremonial or magical, possibly correct as far as it goes, is at best only a partial interpretation. As we will see, the images afford their viewers a particular representation of the natural and social worlds. They were probably intended less to alter those realities through ceremonial or magical means than to change people's minds through direct representation. We do not need the hypothesis of "magical" function, or any other functional hypothesis, to examine the discursive structure of the representations. In fact, some recent writers have assumed the highly rhetorical nature of late prehistoric image making, describing some of the images as "symbols of power generated as political propaganda" (Hoffman 1979: 299; see also the remarks of Finkenstaedt 1984; Hassan 1988; and Kemp 1989). To be sure, the pictorial mechanics of late prehistoric image making could support this

interpretation. The functional specification itself, however, must still be secured on archaeological grounds—for example, if we could determine that the decorated objects were used in the public display of elite authority—independent of an interpretation of' the representational character of the images as such.

Some aspects of the material could be said to speak against this particular "functional" hypothesis. Propaganda is often addressed by an elite circle to an outside, popular or mass audience that is otherwise not necessarily disposed to accept an interest group's view of the world. But we do not find the palettes in wide circulation, mass production, or public display, in contexts of nonelite or popular consumption, as we might expect if they functioned literally as propa-ganda. Perhaps the archaeological evidence is simply inadequate, as it is for the ceremonial" and "magical" hypotheses; or perhaps the objects had other functions altogether. Whether or not the objects were literally "political propaganda" may be independent of ways in which people might have attached sense to them in light of the pictorial mechanics and metaphorics I am suggesting. My interpretation is not incompatible with the notion that the representations sometimes functioned as propaganda or perhaps as other species of social communication and legitimation, but I regard all such possibilities as matters that might be investigated for historical purposes following a detailed examination of the chain of replications of late prehistoric image making.

Although the literal function of the objects remains obscure, we can make elementary inferences about their makers and about the audience they expected to address. Here again historical confirmations are few and far between. At least two people must have examined the objects and images: the maker and the owner, who probably stood in relation to one another as producer to consumer, retainer to master, or artist to patron. Early dynastic painters, sculptors, and woodworkers sometimes served as specialists in a ruler's immediate entourage, apparently resident at his principal court, for their graves have been found in the architectural complex of early dynastic kings' burials along with those of high officials (see Petrie 1927; Emery 1954: 29, 31, 143–44, 153–54), but many artisans may not have lived this way. While we cannot say definitively whether the objects considered in this study are examples of elite consumption, commission, or patronage, there are reasons to regard them as such: stylistic

similarities exist between them and objects known to have been made at the centers of residence or interment of the elite, like Nagada and Hierakonpolis (Petrie and Quibell 1896; Quibell and Green 1900–1901), or at centers of early cults, like Gebelein and Koptos (Davis 1981b; Adams 1984; Williams 1988b).

It would not greatly matter, however, if a lucky archaeological association showed that the palettes were made in a village for a local head of household, lineage leader, or "shaman," as might have been the case—we have no independent evidence—for objects like the Ostrich and Oxford Palettes. Perhaps they were owned and used simply by the people who made them, for the phenomenon of craftsmen employed full time for an elite patron—one of the achievements of canonical representation (see also Davis 1989: 217–21)—may not have been present at all levels of late prehistoric Egyptian society. It is entirely possible that the sculptors were not in close connection with one another, as they would be if they worked in the same workshop or for the same patron. Local residence of the artisans and regional diversity among them could be one explanation for the distinct stylistic differences among certain groups of carved palettes (see, for example, Ranke 1925; Fischer 1958; and Davis 1989: 155–58).

Whether the objects were displayed to or viewed by a wider public circle beyond maker and immediate consumer, master, or patron cannot be readily determined. Despite their sometimes unwieldly bulkiness, even the largest palettes were made—common sense tells us—to be handled and inspected at close quarters, however few the viewers or how restricted their access. The images seem to be organized as narratives to be "read" in a specific way, if not by actually handling the object at least by moving around it to view its several, separate passages of depiction and the composition of the image as an emerging but open-ended totality. The actual situation of this activity of viewing, however, is unknown. Did it transpire in a ritual setting? Was the object set up for display or could a person actually take it in his or her hands? The various objects in the group, different in size one from the next, may have been handled or viewed in different ways depending on factors we cannot now reconstruct. Finally, the images must assume an audience and a particular mode of viewing at a notional level, but we cannot say whether the assumptions were literally to

be fulfilled. Like the makers of stained-glass windows high in the walls of a medieval cathedral, the late prehistoric Egyptian image makers may have produced objects and images simply according to their proper rules, expecting no one to make out their intricate details.

All these questions about functions and uses are somewhat beside the point I make in this study, not because they are irrelevant but because of what I take to be a prior and overarching consideration. Irrespective of the possible documentary, votive, commemorative, ritual, ceremonial, magical, exchange, propagandistic, or other functions of the decorated objects, they work as pictorial narratives. While their representational function may appear so obvious as to be uninteresting, it is both broader and more penetrating than the other specifications I have reviewed—for all of them must be seen as aspects of a particular function or functions of what is already functioning as representation. In fact, paradoxically, most of the specifications I mention are too general. Other activities or artifacts apart from images, as well as other sorts of images, could have served successfully in ritual, magical, ceremonial, or propagandistic ways; and therefore we still have to develop a specific analysis of these objects as fulfilling such functions by way of the images depicted on them. A scholar outside the disciplines of prehistory and Egyptology—for example, in literary criticism or art history—might be surprised by attempts to identify possible functions in the absence of a rigorous formal identification and textual analysis of the images themselves. How could we determine the function, or social use, of the image even before defining what we are looking at?

The Question of "Sources"

Historical information about the production and use of the objects is so hard to come by that we must simply make the best of an unfortunate situation by exploring the material we do have—namely, the images themselves. Because the problems of evidence are so acute, however, two other possible sources of information have been mined heavily by a number of writers—first, other prehistoric representation, and second, its canonical descendants and analogs.

Several writers have investigated the possible stylistic and iconographic sources of and parallels for decorated ivory knife handles and combs and schist palettes in earlier and contemporary media like rock art, pottery design, wall painting, and textile decoration (see Vandier 1952; Williams and Logan 1987; Williams 1988a; further references and comments are found in Davis 1989: 116–34). The prehistoric traditions of representation in the Nile valley emerged as early as 8000 B.C . in "abstract" engravings produced on rock faces near the present-day towns of Abka and Wadi Halfa at the second cataract of the Nile (Myers 1958, 1960; Hellström 1970; Davis 1989: fig. 6.1). Although production in other media is less well preserved and consequently less well known, there were in fact rich and changing traditions, particularly in rock art and pottery decoration. It is not surprising therefore to discover in them apparent antecedents or analogs for many elements of the late prehistoric images on knife handles and palettes, the latter forming just part of a larger, longer history of production.

It may be that this tradition derived some stylistic mannerisms and iconographic motifs from older or contemporary Near Eastern sources, a possibility urged long ago by Henri Frankfort (1925: I, 93–142; 1941; 1951: 121–37). More recent studies would date this relationship to the relatively well-defined period of the Nagada IIc/d (Boehmer 1974a), suggesting that at least one medium of the diffusion of motifs may have been small, portable roll seals, although the seals actually found to date in Egyptian predynastic contexts do not themselves document transmission of the particular features Frankfort cited (Boehmer 1974b; see also Kantor 1942, 1952, 1965; Trigger 1983: 36–38; Mode 1984; Teissier 1987). Foreign influence is in question precisely because we do not yet have a clear idea of the general conditions of intelligibility of late prehistoric image making and because it is still judged in light of later canonical standards. Thus it is difficult to decide whether a mannerism or motif, by virtue of appearing untypical or "un-Egyptian," is therefore to be ascribed to external influences; and some writers are rightly skeptical of the entire account of diffusion (Kelley 1974). Merely for the purposes of argument, we can grant that motifs like the human being mastering opposed animals, the serpent-necked felines (or "serpopards"), and the winged griffin—all appearing on late

prehistoric Egyptian ivory knife handles or palettes—are virtually certain examples of motifs with an ultimate origin in Mesopotamia, perhaps mediated by small Mesopotamian seals on which the motifs can be identified substantially earlier than the Egyptian uses (Porada 1990).

It is not clear, however, exactly how any of this fine-tuned research pertains to art-historical phenomena. A stylistic or iconographic source can always be cited for an element of a passage of depiction, if not of the image as such; all cultural practices necessarily have cultural heredity, just as biological individuals have biological heredity. But to identify a source, presuming it is a real one, is not thereby to understand the role of the element in any one of its immediate semiological contexts. An artist might appropriate a motif from another artist's work, or even from another culture, and put it to quite independent use, altering its established meaning by inserting it into a new compositional setting. For example, some commentators have suggested that the intertwined serpopards on the Narmer Palette symbolize the unification of Lower and Upper Egypt (Gilbert 1949; see also Finkenstaedt 1984), a meaning that presumably had nothing to do with the "original" Mesopotamian prototype of serpent-necked beasts. I myself contend that the intertwined serpopards on the Narmer Palette play a highly specific role in the narrative mechanics of the image—a role distinct from their appearance, for instance, on the Oxford Palette. Source hunting alone will not expose these relations. Indeed, source hunting based on a ramified description and comparison of many disparate images taken out of different contexts, unqualified by attention to the coherence of each image, distracts from the questions of narrative or symbolic meaning. The possibility of a Mesopotamian or other Near Eastern origin for some late prehistoric Egyptian motifs tells us nothing about the symbolic function—for example, the metaphorical or narrative status—of these motifs as they were replicated by late prehistoric Egyptian image makers. It would be interesting to discover that the late prehistoric image maker expected viewers to see or to sense the Mesopotamian origin as an aspect of the motif chosen and thus a reference or connotation of the image as a whole; but we could establish this textual relation only by scrutinizing the mechanics of the image—not simply by pointing to a Mesopotamian analog or even to a possible medium of diffusion of that analog.

There is a strong chance that in a detailed examination the apparent analogs for late prehistoric Egyptian images will turn out to be paralogs, mere unrelated similarities, or even pseudologs, false similarities. The institutions of canonical image making that enforced conformity to an established set of rules had not yet fully emerged in the late prehistoric period. We cannot assume, therefore, that each use of a motif in the precanonical context meant applying the same rule or rules, as it would in the fully canonical context. There is no guarantee that a source is really a source or a parallel really a parallel; a sign made according to the criteria of one symbol system differs from that made according to other criteria, even if the morphology of the two signs is indistinguishable (see Goodman 1972; Danto 1986).

At the time he laid the theoretical foundations for iconographic histories, Erwin Panofsky (1939, 1960) was aiming for what he called an "iconology"—namely, an explication of styles and motifs within the historical horizons of their unique cultural uses (see also Kubler 1985). Thus, although the Good Shepherd figure of early Christian iconography might derive iconographically from earlier Classical depictions of the beardless, athletic young hero, we would not think to explicate the Shepherd merely as a version of a Classical hero or Apollo type. Rather, in a process Panofsky (1960) termed "iconographic disjunction," the Shepherd figure capitalizes on its contemporary viewers' knowledge of the Classical type in order to make both traditional and novel, metaphorical statements about the identity of Christ, a reference to an object—or, let us say, a new sense for the available reference to a hero—that could not have been known to Classical audiences. At some point the relevance of the Classical reference simply fades away and ceases to matter in identifying and understanding what the latest makers and viewers experienced in the image.

It is virtually guaranteed, I think, that a similar situation obtained in prehistoric, late prehistoric, and early dynastic Egyptian art. For example, an artist like the sculptor of the Battlefield Palette knew the production of a relatively immediate predecessor like the sculptor of the Hunter's Palette. But their relationship was an intricate and subtle one: the Battlefield sculptor took a motif from the Hunter's Palette and by inserting it into a novel narrative context charged it with an entirely different sense. Although almost all mannerisms and

motifs must have sources, their referential dimensions can be seen to have evolved considerably over the span of the many decades or even centuries of the late prehistoric period; only in this way can we make sense of the distinct qualities of several late prehistoric representations.

Such iconographic disjunction, I maintain, characterizes not only the relations among the individual decorated knife handles and palettes but also the relations among these objects and the other media in contemporary or earlier Egypt that are sometimes cited as the iconographic origin or "conceptual basis" (Williams and Logan 1987) for elements of the knife handles and palettes. Making such an assumption means that I spend much less time citing these relations than has been usual in Egyptology. (Some "iconographies" or iconographic histories of predynastic and early dynastic motifs are available [for example, Williams 1988a]; all, however, are subject to the reservations I have mentioned.) In contrast to vulgar iconography's usual source hunting and parallel quoting—many images are reviewed to identify the so-called development of a motif, plucked from its semiological contexts—each image, presuming we can identify it, is taken here as a semlotic whole. It must be interpreted in depth before any iconographic relations among images, from quotation and influence through metaphor and parody to genuine dissimilarity and disjunction, can reasonably be inferred. To follow these relations into other, earlier sources or across media would not only greatly lengthen my account; it would also deflect attention from the properties of pictorial disjunction as the very structure of an image-making tradition. As a practical matter I do not consider all known decorated knife handles or carved palettes of the Egyptian late prehistoric period. Although motifs in all of them can be described and compared, some are simply too fragmentary to support a close and comprehensive interpretation of the organization of the image.

When properly combined with other inquiries, the comparative approach in motif typology—hunting for "sources" and quoting "parallels"—has a limited role to play in historical interpretation. For example, in an investigation of pictorial meaning that recognizes formally similar motifs as having many disjunctive symbolic connotations in a group of images, motif typology might show that one image must be later (or earlier) than another. In some cases motif

typology can be transformed into chronological analysis or an interpretation of historical interrelationships among image makers and viewers. But it cannot establish such conclusions on its own. The objects considered here have a relative chronology based not so much on the morphology of particular motifs as on the morphology of the outline shapes and related attributes of the artifacts themselves (Petrie 1920, 1921; Kaiser 1956, 1957). At any rate, since the motif-typological approach has generally dominated many branches of prehistoric art studies (for instance, see Davis 1990b: 272–74), another approach is required to give proper attention to the coherence of particular images, whether or not we wish to compile larger comparative typologies of stylistic mannerisms, motifs, or other features of passages of depiction.

Canonical Standards and Late Prehistoric Images

Similar considerations apply when we measure a second type of possible evidence for explicating late prehistoric images. Just as they must derive from an earlier history of making images, so too do they form the historical context for subsequent production—namely, for canonical Egyptian representation. The late prehistoric image maker could not have been fully aware at the time of the "serial position" he occupies in a continuous art history (Kubler 1962). He could hardly have planned for all the selections canonical image makers would make from particular aspects of his production—for example, they adapted what I call the several "core motifs" of canonical iconography from the repertory of prehistoric art (Davis 1989: 59–63)—or for all the new references they would assign to the images he left them, for example, in the very act of depositing the Oxford Palette, a late prehistoric object of Nagada IIc/d date, in the Main Deposit at Hierakonpolis, an early dynastic or even later assemblage. Why, then, should the canonical image maker's selections and references be considered relevant in interpreting the late prehistoric image maker's work?

Despite the obvious anachronism many writers on late predynastic and early dynastic art in Egypt have taken what we know of canonical art, hieroglyphic script, and Egyptian ritual, mythology, and cult—a literate tradition based on stable rules associated with the specific social formation and ideology of the

national, authoritarian, theocratic state—and applied it back to images produced in a less complex, almost wholly preliterate society. However ingrained in the Egyptological imagination, the procedure is completely unsound methodologically.^[5] As far as possible I avoid applying decidedly canonical regularities to a precanonical or noncanonical context.

At one point this general rule—in that it is a general rule—will need to be relaxed slightly to afford us a useful line of interpretation in a special, limited case. Sited midway between the Oxford Palette at one end of our chronological continuum (about 3300 B.C .) and the register compositions of Old Kingdom public and mortuary art at the other (from about 2700 B.C . and on), the Narmer Palette (about 3000 B.C .) can be considered at one and the same time as the last exemplar of late prehistoric, noncanonical representation in Egypt and as the first canonical work. We are thus inclined to see it in retrospect as a "mixed," transitional, or transformational production, even though the original maker of the palette and his audience could not have seen the image in this way. But rather than merely worrying about characterizing the Narmer Palette in developmental terms, we should also question our rationale for framing it chronologically or historically as we do. For this and other reasons, it is a matter for delicate judgment in interpreting it to apply canonical rules or noncanonical rules or no rules at all. Reasonable interpreters can reasonably disagree about how the balance should be struck. But this is so for every representation; every representation is a revision within a chain of replications.

By virtue of its application of hieroglyphs, use of a proportional scheme, and tendency to magnify the scale of important figures and scenes, the Narmer Palette is often placed at the beginning of canonical Egyptian art (for instance, Aldred 1980). Partly in order to balance this point of view with its inherent risk of strong anachronism, I suggest that in fact the Narmer Palette is distinctly "prehistoric" in its underlying pictorial mechanics. It is most properly legible according to a narrative symbolic system that seems to have no direct canonical descendants, for canonical artists and patrons in the First through the Third Dynasties elected to replicate the symbolic rather than the narrative dimension of that inherited system.

Pursued too single-mindedly the question becomes utterly empty, a case of the sterile typology chopping that so debilitates prehistoric archaeology and of the style fetishism that so debilitates art history. General categories—like prehistoric and historic or canonical and noncanonical—break down when the particular semiological dynamics of a specific image are investigated in any depth. It hardly matters whether the Narmer Palette is regarded as canonical or non-canonical or some mixture of both, if we can make sense of the way it has been put together as an image. Unsettled and uncertain, disjunctive and decomposing, the Narmer Palette is an image that literally has two sides.

Most writing on late prehistory in the Nile valley has been dominated by anachronistic decipherments and by a determination to identify what is "Egyptian"—that is, interesting from the point of view of a discipline devoted to the study of dynastic Egypt—in the cultural history of the Nile valley and northeastern Africa (see also O'Connor 1990; Davis 1990b: 274–79). As with iconographic "sources," analogies for late prehistoric practice in later cultural life may be highly suggestive if there is independent evidence that the earlier and the later images did in fact represent their objects, as images, in the same or closely similar ways. It turns out, however, that little of real interest hinges on information gleaned from canonical sources. Here or there a motif might be more precisely identified, a gesture understood, or a glyph deciphered by calling on our knowledge of canonical representation. But by and large the late prehistoric image maker's practices were intelligible to him and to his audience in their own terms, at best only vaguely including the future possibility of a canonical, academic art for a dynastic, patrimonial state.

If depending on earlier and later analogies for late prehistoric practice is a questionable method, the use of general ethnographic or historical analogy outside the northeastern African and Egyptian setting is likely to be even more so. What we know of prehistoric arts, such as Upper Paleolithic cave painting or neolithic pottery decoration, and of historical styles like Roman narrative relief or medieval architectural sculpture might suggest means of understanding late prehistoric images in Egypt. All such suggestions, however, have the same status as the promptings of our intuition or visual scrutiny of the objects; they

are no more and no less reliable. In fact, an ethnographic or historical analogy can be misleading inasmuch as it may distract from the fundamentally disjunctive structure of all chains of replication and from the necessarily specific semiological context in which any given replication has been used. While I have learned a good deal, throughout this study, from what has been written about arts other than the ancient Egyptian, there is no need to introduce my experience in detailed comparisons. An interpretation cannot be strengthened by the existence of the analogies, from whatever visual, ethnographic, or historical context, used in working it out.^[6]

The Temporality of "Meaning"

So far I have been making uncontroversial points. Historical interpretation of late prehistoric image making must avoid, as scrupulously as possible, teleological and anachronistic formulations and misleading visual or historical analogies. It is necessary to insist on these matters only because—in their source hunting and wild analogizing—too many analysts in prehistoric archaeology and Egyptology violate the rules of historical method. Most of the substantive problems for historical interpretation lie well within the purview defined by these somewhat banal methodological injunctions.

For example, although we may wish to avoid teleology in historical analysis, we still need to recognize the goal-driven, the forward-directed quality of an image maker's intentions. An image maker will anticipate a certain response to his work, plan for a certain use, or expect a certain interpretation to be applied to it. He works with extensive present beliefs about future audiences, including a sense of the ways that audience might repeat, revise, or refute the work and therefore, working from the future back into the present, of ways the work might be rendered more legible, more convincing, or more immune to alteration.

Similarly, although we prefer to avoid anachronism in historical analysis created by reading back from later meanings into contexts in which they had not yet been constituted, we must nonetheless acknowledge the memory-filled, backward-looking quality of a maker's intentions. An image maker may fully

incorporate even as he revises an existing proposition or mode of expression remembered or inherited from the past. Taking up materials from the past of image making as an element of his present intentions, the image maker may represent the past in his work as anything from a mere material precondition, the bare necessity of existing materials, to a virtually absolute cognitive determination, a full intentional embrace of the past.

In either case, future aims and remembered resolutions are equally subject to the image maker's "present" intentions or, more broadly, "intentionality" (following Searle 1979)—that is, to his way of both consciously and unconsciously directing his thought toward the world and its objects, including the alms and memories in his thought, in the ongoing present activities of believing, supposing, desiring, imagining, and so forth. (This general concept of intentionality includes but is not limited to what is ordinarily meant by an "intentional" thought or act, that is, one performed by design or according to deliberated purpose; intentionality is any form of the direction of thought toward objects, and may, for example, be completely unconscious or undeliberated.) Thus, the image maker maintains certain beliefs about what is remembered or inherited from the past, such as a belief in the validity of the repertory of styles and motifs he has learned, and also sustains certain desires for what is expected or supposed about the future, such as an ambition to predispose an audience to his work—perhaps precisely for his belief in the validity of the past to be ratified and thereby maintained into the future. In all of this the apparent paradox is that present intentionality, the present structure of beliefs and desires about or for aims and memories, must be partially a function of the aims and memories, the "past" intentionalities or structures of beliefs and desires, in which all "present" beliefs and desires are themselves constituted.

Thus, to continue the example, the image maker's belief in the validity of past styles and motifs, an element of his present intention to replicate them, was constituted in a past intention, present to him at that time, which included beliefs and desires about yet other aims and memories—for example, about his activity, in the past, of making the styles and motifs he will continue to use in the future. And the image maker's ambition to have an audience ratify his belief, an element of his present intention in replicating past styles and motifs,

was also constituted in a past intention, present to him at that time, which included beliefs and desires about yet other aims and memories—for example, about his desire, in the past, to make the styles and motifs for the audience he would encounter in the future.

It is exceedingly difficult to find a language subtle enough to describe the temporal structure or cognitive status of human intentions or intentionality in any context. Although an account of the intention or intentionality of making artifacts and images is surely the prime desideratum of archaeology and art history, a fully satisfying exposition does not exist (although see especially Wollheim 1987; and Walton 1990). In general each image is subject to prospective and retrospective repetition, revision, and refusal. Indeed, as the previous paragraph suggests, it is constituted in such prospective and retrospective replication. There is no single, unitary, centered place or time in which an intention—a structure of intentionality as a set of beliefs and desires about the world, including present aims and memories—is fully present or available for archaeological observation. (This stricture applies to the observation of the one who has the intentions as well as to outside observers; once he or she directs thought toward his or her intentions as intentions in the present, he or she is already outside or beyond that present and his or her thought about intention is directed by his or her further intentionality.) The alternative to identifying a single place or time at which intention can be fully identified—for example, we say that a specifiable intention is "in" this or that particular passage of depiction, composition, or image—must be to recognize that the intentionality of representation is spread out as a recursively nested and emergent temporal structuring of repetitions, revisions, and refusals of representations.

Sigmund Freud (for instance, 1918) coined the term Nachträglichkeit, the property of being "subject to delayed activation," to describe the representations he was studying—namely, the fantasy images of past events in an individual's life shaped in turn by his or her later emerging sexuality in and by which those events are recalled (see also Derrida 1976). The fantasy image cannot be specified as having this or that particular, stable meaning or content at any given point in time. Meaning and content are built up in the ongoing replication of representations of the event—itself an irretrievable and somehow

primal origin of the series—as a sequence of prospective and retrospective directions of belief and desire making reference to and taking account of that event. For Freud the replication of the event in a prospective and retrospective structure of beliefs and desires extends into the most fully present time of observing the event itself—that is, into the moment when the event and its replication become the object of a historical investigation within psychoanalysis. Thus in the transference and the countertransference both patient and analyst continue to make the past the object of present intentionalities—present proposals, hopes, commands, promises, lies, and so on. To the degree that these present intentionalities are determined by the past they replicate, the past is made visible within the psychoanalysis. But to the degree that we can identify no time or place definitively outside or beyond the reach and grip of the spread-out temporal structuring of intentionality, the psychoanalysis is the past rather than its objective representation by a detached, Olympian historian-observer.

Although we need not employ Freud's own preferred language of description and explanation, it is for me crucial to entertain his and other ways of understanding intention in the replication of artifacts and images as ongoing repetition, revision, and refusal temporally structured in prospective and retrospective reference. For my purposes here, then, "teleology" and "anachronism" are within intentionality as its temporal structure. In turn, the point for historical interpretation is to ensure that this structure is not reduced, misrepresented, or ignored by turning it into a historical entity, a time or a place, which is then explained, by the historian-observer—in teleological or anachronistic fashion—as the immanent goal or the immanent origin of other, earlier, or later temporal structures of intentionality, complex and interesting as they may be in their own right. Thus the temporal structure of late prehistoric image making should not be examined solely for the purpose of showing how a later intentionality, the system of canonical image making, originated in, was determined by, and explains its inherent meaning—that is, if the objective is to understand late prehistoric image making. If the objective is to understand canonical image making, then prehistoric image making is somehow present in the temporal structuring of the canonical artist's beliefs and desires, aims and memories, prospective and retrospective references. This is another historical

problem altogether, the interpretive history of the intentionality of canonical image making in which prehistory sometimes appears. Here we may find that the canonical image maker suppresses, or "forgets," as much prehistory as he repeats, or "remembers," and that if prehistoric art is the conceptual basis (Williams and Logan 1987) for canonical image making, it is only as the partial target of later repetition, revision, and refusal.

These observations about the temporal structuring of intentionality should be joined with a recognition of its spatial distribution. An image maker necessarily employs a mode of representation or participates in a convention adopted in his social world among several differently situated agents mutually coordinating their actions and representations (Lewis 1967). To the degree that he does so, his intentions and intentionality are not, strictly speaking, just "his own." Rather, a cultural order, the set of modes of representation and conventions, speaks through and constitutes the mental equipment of the particular maker who replicates representations. Intentionality, then, is distributed in a maker's mind not only temporally as a series of forward-and backward-looking thoughts, each partially targeted and resolved in a "later" fulfillment; it is also distributed spatially among image makers as their very means of apprehending and representing the world. "Ideology" is one usual, useful term for this spatial distribution of a mode of representation coordinated under a social convention common to two or more agents, up to and including the society as a whole. As with the temporal structuring of intentionality, its spatial structuring—for example, as a social ideology—cannot be located in a single, unitary, centered time or place. By definition a convention is the mutual coordination of several agents directing themselves and their thoughts at objects in the world. Therefore the convention is to be found, or observed archaeologically, not only in the content and action of a single agent but also in and among the group of agents as an interacting whole.

These general matters are the province of phenomenology, philosophy, and critical social and cultural theory. They have implications, however, for any project of historical interpretation; although I reject teleological and anachronistic reasoning as means of identifying the "meaning" of late prehistoric Egyptian images, meaning itself is constituted as an indefinitely ramifying and prolifer-

ating temporal and spatial structure of pro- and retrospective reference, repetition, revision, and refusal. To capture this fact, rather than trying to specify "the meaning" or "a meaning" of images, I investigate the apparent pattern of revision in a series of representations evidently linked in or as a continuous chain of replications (although we certainly lack some variants in the chain). No one image can be said to exemplify all the properties of the chain; no single image maker has access to it all at once. This temporally and spatially distributed structure constitutes the intentionality of the image makers—whatever can be identified as their belief and desire, their aim and memory, their consciousness and ideology.

No doubt a chain of replications has a history. For descriptive and analytic purposes we must endow it with a beginning, an ending, and a set of instances, stages, or phases. Furthermore, it must display a degree of irreversibility. Later images are necessarily later because they assume or even refer directly to earlier ones. The structure and properties of the chain as a chain, however, are vastly more important than its individual local expressions, presuming we even have archaeological evidence about these occurrences. If intentionality is temporally and spatially distributed, it cannot be found in or as any particular replication. Any particular archaeological observation, necessarily focused on a replication produced at or deposited in a given time at a given place, will expose it only partially.

I have not gone so far as to adopt the structuralist position that each representation is only an instance of a more general "myth" (see especially Levi-Strauss 1971) or, to use Leroi-Gourhan's (1967, 1986) term in his analysis of paleolithic representation, "mythogram." The semiological uniqueness of each image must be recognized even as its properties of repeating, revising, and refusing others should be identified. My view of intentionality per se requires that the textual organization of any repetition, revision, or refusal is necessarily individual and complex. What might seem to be the most exact repetition of an earlier image could have a place in the chain of replications—in the temporal and spatial distribution of the image makers' intentionality—precisely as a repetition; the past image is repeated only because, for instance, a present belief or desire is directed at that image, becoming the sense or meaning of that past

image, brought into the present, for any future repetition. In other words a thought entertained again is not the same thought. Classic structuralist reasoning seems unable to handle adequately the fact that synonymy of meaning cannot be attained (Goodman, 1949).

I treat each image, therefore, as an autonomous textual organization, if not exactly as a "whole," even as I attempt to identify its status as a replication. The two accounts are not mutually exclusive. We are always looking backward or forward for the meaning of an image—or, in general, somewhere else in the history of its use. In a sense, then, reconstructing a chain of replications means inevitably that we attempt an internal teleological and anachronistic analysis. To be able to interpret an image as a particular instance of repetition, revision, and refusal, we must be able to look backward and forward within the chain of replications. But the point, again, is not to use a tracking of forward-directed and backward-looking replications to identify the meaning of any one replication on the basis of the evidence of the others. In doing so we would have to accept teleology and anachronism as immanent and transcendent within each one of the replications separately understood—precisely as we should not—as single, unitary, centered times and places of meaning. Rather, we must hold close to the fact that intentionality does not stand in a one-to-one correspondence with a particular replication as a stable determination for the form already in the head of the maker. Spread out in and as replication per se, intentionality is material, not ideal, in its temporal and spatial structure.

In what follows I focus on the structure and history of one chain of replications in late prehistoric Egyptian art as it survives partially in a group of typologically, technically, and stylistically similar artifacts. I defer translating what therefore might be called a set of semiological conclusions into, strictly speaking, an iconology—a study of the works of art as necessary symptoms of cognitive, social, and cultural processes. I believe my caution is justified. In the absence of detailed archaeological information, we must look carefully at the works as a possible chain of replications before leaping to "historical" conclusions about the ends they were made to satisfy, the meanings they might have had for individual viewers, the functions they could have served within the social world, and so forth. In principle, if we can discover something of the temporal and spatial structure of intentionality in image making, we will then

have discovered cognitive, social, and cultural processes. We do not have to regard the images as re-presentations, expressions, or trans-scriptions of such processes, for such processes are precisely the timing and spacing of intentionality in representation.

I focus on the sequence and even the slips of the image maker's conception and labor, on the pictorial dynamics of the ensuing images, and on the organization of a viewer's interpretation of them. These features of a chain of replications cannot be deduced a priori or on a mere visual inspection of the artifacts; they must make sense in the total context of what is known of late prehistoric Egyptian image making or to be expected in that context. The constraint is a fairly loose one, however, and not the same as saying that results must entirely accommodate what is known. What is known about late prehistoric Egyptian image making at the moment is so sketchy that any detailed interpretation of a set of images is likely to outrun what we could support archaeologically. A useful interpretation will contribute to, not merely derive from, archaeological understanding of late prehistoric Egypt, for it might enable us to say something about that world which would otherwise remain unseen.

Masking and the Scene of Representation

If we are to place the decorated knife handles and palettes in a view of late prehistoric Egyptian representation, it will have to be on the basis of what sense can be made of the works themselves. Unfortunately, despite attention to sources and to parallels in canonical Egyptian art, late prehistoric Egyptian images have rarely been interpreted systematically as representations. With the exception of Roland Tefnins (1979) structuralist analysis of the Hunter's Palette and Henri Asselberghs's (1961) provocative Geistesgeschichte, no developed semiological or iconological treatments of late prehistoric Egyptian image making exist.

Perhaps the most dramatic evidence of the poverty of our current understanding is the way late prehistoric Egyptian images are reproduced incorrectly in standard textbooks, and even in scholarly monographs, because their pictorial mechanics have not been analyzed thoroughly. For example, the Hunter's

Palette (Fig. 28) is sometimes reproduced sideways, with its long axis horizontal, as if it were meant to be viewed this way. This presentation amounts to starting a viewer halfway through the narrative of the image rather than at its proper beginning. Two-sided images like the Oxford and Narmer Palettes (Figs. 26, 38) are sometimes reproduced in fragments with no indication given that a particular detail of a complex, unitary composition—like only one side of the image—is being shown. In these cases the writer of the analysis presumably imagines that the two sides of the palette, or the various passages or zones of depiction on each side, are independent images rather than elements of a whole. Sometimes the two sides are shown in incorrect sequence, giving the viewer of the reproduction a vague or inadequate understanding of its message.

The pictorial mechanics of late prehistoric Egyptian image making have been profoundly misunderstood at the most general level. For example, H. A. Groenewegen-Frankfort (1970: 113), one of the ablest commentators on Egyptian art, claimed categorically that "no attempt at pictorial narrative was made" on the Narmer Palette—a view shared by many Egyptologists (although see the subtle analysis of Schäfer 1957) and taken up by many nonspecialists. To the contrary, the Narmer Palette exhibits distinct narrative, as well as non-narrative, elements, and in fact I make the case that all or most of the late prehistoric palettes should be regarded as narrative. (Since the assumption is so important for my historical interpretation, a framework for understanding pictorial narrative is presented in the Appendix.)

It remains unclear how the late prehistoric images refer to what they do depict and especially how they substitute for what they cannot depict or disguise, avoid, and suppress—how they "mask"—what they will not depict. The images thwart interpretive and critical unmasking of their mask of representation when even modern viewers continue to take them literally as being what they put forward as depiction—that is, a representation of what is referred to, a substitute for what cannot be referred to, and a mask for what will not be referred to. In such cases the depiction is not recognized as depiction; it is taken as naturally transparent to some real world that becomes the entire focus of the viewer's interest. For example, many writers have handled designs presenting rows of animals, like the Brooklyn knife handle and the Libyan or Booty Palette

(Figs. 8, 53), by trying to identify the actual species depicted and by taking the pictures as documents of the actual hunting, capture, or other literal, real-world encounter with these species. Similarly designs including human beings engaged in a variety of activities, as on the Hunter's, Battlefield, and Narmer Palettes (Figs. 28, 33, 38), have often been understood mostly in terms of the ostensibly documentary status and value of the hunts, wars, or rituals they apparently depict. Elaborate "histories" of late prehistoric Egypt have been built on the basis of the supposed evidence provided by the depictions on the palettes. In these cases modern interpreters have assumed that the mask—the representation, substitution, and suppression performed by the depiction—literally mirrors the world; thus they do not confront the way it acts to imitate the world only partially, precisely to appear to render it but actually to recompose or invent it altogether. (On mimesis and masking as the processes of disguise, substitution, and avoidance in representation, see Girard 1978; Prendergast 1986; and especially Marin 1988, with Douehi 1983.)

A decided advance over literalizing interpretations are relatively recent systematic investigations into the relationship between the formal or compositional organization of late prehistoric images and the social world from which they derive or the social message they put forward (Michael Hoffman [1979: 299] called the decorated palettes "symbols of power generated as political propaganda"; see also Finkenstaedt 1984 and Kemp 1989). Thus, for example, the appearance of register compositions could be thought to reflect the reorganization of Egyptian society in the formation of the dynastic state at the end of the fourth millennium B.C . More emphatically, register composition could be seen as a means of claiming the "registered" organization of the world—well bounded, neatly packaged or classified, and mastered by a human owner or ruler—as a truth or as an ideal by representing it as a fact of experience (see also Davis 1989: chaps. 6, 7). Even here, however, we run the risk of regarding the late prehistoric image merely as a document of some preexisting historical events and processes that it mirrors—in this case, the formation and ideology of the emerging dynastic state. The images are not, of course, depictions of ideology, although what they depict may be ideological; and thus to understand them as ideological it is necessary to consider not only what they literally

depict, what they refer to, but also what they do not depict through substitutions and suppressions—what they mask—as an indissolubly linked aspect of the process of depiction. For example, although the representations render a particular narrative of the ruler's complete mastery over the natural and social worlds, they do so not just through direct reference but also through reshuffling and substituting elements of the narrative, which suppresses or drops some of them into ellipsis altogether—finally, masking the blow of the ruler.

Despite a century of intensive investigation of many individual motifs, a number of immediately pressing iconographic and more broadly iconological puzzles in late prehistoric representation have not been addressed. For example, why are the human beings in several images (Figs. 25, 26, 28, 38) apparently depicted wearing animal masks, costumes, or accoutrements? We could suppose that late prehistoric people literally dressed in this garb, perhaps in the course of carrying out a special kind of hunt or in performing a ritual. We could accept the objective reference of what is depicted even if the exact circumstances of the event remain mysterious. But if people did indeed dress as they are depicted, it was clearly to designate something in turn—namely, whatever is represented by the animal mask or costume. That representation is included in any depiction of them, and we must interpret it in terms of the apparent themes and metaphorical dynamics of the system of late prehistoric representation. The literal depiction is the representation of the representation that is its objective reference. The particular iconographical puzzle—what is the symbolic, metaphorical, narrative, or other significance of the animal mask or costume?—remains within the scene of representation, the chain of representations in which any individual replication takes its place, the setting, a stage or theater of representation with general framing conditions of intelligibility, in which particular representations make sense.

We may well ask, why is the defeat of an enemy rendered as his loss of eyes or as his decapitation (Figs. 33, 38)? Here a narrative, symbolic, or metaphorical point must be deliberately intended; after all, in the ordinary course of a battle, people must have been killed in numerous ways apart from these. Perhaps, then, as is often said, we are being shown a ceremonial or ritual killing and display, a ceremony of decapitation or display of the conquered. But with-

out further qualification this explication does not interpret the specifically representational properties of the ceremony or ritual itself. Why is the enemy ceremonially decapitated rather than something else? Whether or not actual or ritual killings took place by decapitation, the image—either documentary or fictive in status—presents decapitation as an element of the depicted world. But for what reasons or in relation to what pattern of metaphorical association?

Examples of these puzzles could be multiplied indefinitely. For instance, recalling that figures in earlier or contemporary rock art and pottery decoration (Figs. 3, 4, 6, 7) look predominantly in the direction their bodies face, why is there so much over-the-shoulder looking or looking backward in late prehistoric representations (Figs. 23, 26, 28, 33, 38)? Considering the later unidirectionality, isocephaly, and fixity of register compositions (Figs. 1, 2), why do late prehistoric works exhibit what is apparently a great deal of formal and compositional ambiguity, often requiring a viewer to tip, rotate, or flip the decorated surface of the object to understand the depiction spread across it, with its figures oriented, glancing, or moving in different directions? Why is there so much, even, of what seem to be slips of the artist's full composure in advancing a depiction onto the surface?

I take these and many similar questions quite literally. They can be explicated by examining the metaphorics and narrativity of the numerous disjunctively related images making up the chains of replication of late prehistoric Egyptian representation. We cannot wave away a variety of iconological puzzles by viewing late prehistoric artists as merely "primitive." Early but still influential students of the objects like Jean Capart (1905) and Flinders Petrie (1920) often did so, and it is fair to find the legacy of their judgment in later commentators' assessments of predynastic compositions as disorderly or disorganized (Groenewegen-Frankfort 1951; Fischer 1958). The later canonical standards by which the late prehistoric artist might appear to be unsophisticated or disorganized, however, cannot be applied to him retrodictively without gross anachronism. For similar reasons we cannot view the artists' preoccupations as solely "decorative"—as some interpretations within archaeology tend to do in emphasizing the late prehistoric production of luxury objects, defined, in part, by the presence of nonfunctional features like decoration (Hoffman 1979; Hassan

1988). To the extent that the decoration on the luxury craft depended on depiction, some account of its significance must be offered. Circling around the presence of representation as such merely encysts it, unexamined, within superficial descriptions.

Interpretation of late prehistoric representation, then, must focus not only on the passages of depiction per se but also on the scene, never literally shown in its entirety, in which the depictions are organized to constitute an image by referring to some objects and masking others—the scene of representing and specifically of narrating, a setting of and framing for the replication of representation in which any particular representation makes sense. Both the reference and the mask produced by any particular representation can be assumed to have made sense in a tradition of disjunctively related images; thus we must try to solve our iconographical and iconological puzzles in the positive terms of this scene of representation. The visual form must be investigated if not always in its literal denotation and compositional or structural connotation—these elements of meaning may be irretrievable—at least in its status or activity as depiction.

In summary, it is within the scene of representation that the literal references and particular masks of depiction assume symbolic or narrative ends. Therein details that would otherwise be reduced to mere depictive description or ignored as "primitive" or "decorative" can be explicated consistently and with "respect for the mechanisms of the text" (Genette 1988: 8). For instance, apparently minor details like the exact angle an arrow is depicted as entering a body, the direction of the glance of a subsidiary figure, or the place in the scene where the artist's virtuosity, or his misstep, can be felt all have a place in both the mechanisms and the metaphorics of the image and also within the chain of replications of late prehistoric image making.

2—
Outside the Scene

At an elementary formal and historical level, the origins of canonical register composition are not hard to find. On expansive plane surfaces, like a rock, a swathe of cloth, or an interior plastered wall of a tomb, the prehistoric artist was free to dispose his figures quite freely—and so he did, scattering them evenly across the plane (Figs. 3, 4, 5).^[1] When designs were made on surfaces more restricted or contracted than the wide planes of a rock face or a plastered wall, there was an impetus toward a repetitive, rectilinear, sequential formal order. Thus on the decorated vessels of the Nagada II/III period (so-called Gerzean Decorated Ware), the motifs or compositions are ordered so that a viewer must turn the vessel to unfold the image or images wrapped around it (see Fig. 6 for views of a painted vessel and compare Fig. 7 for examples of modern, possibly misleading "unrollings" of the designs). Although I do not consider painted pottery further in this study (see Raphael 1949; Williams 1988a; Davis 1989: 120–24), it forms an important historical and visual context for the images I examine.

On the even more restricted surfaces available for decoration on small objects like combs, knife handles, scepters, incense burners, spoons, and cosmetic palettes—generally intended to be lifted and grasped with one hand—artists came increasingly to carve figures in deeper relief in neat rows,^[2] conforming naturally to the shape and edges of the support and its available space, usually no greater than the palm of a human hand. There may well be a conceptual relation between this space of and for representation and what would become the fundamental module of proportional design in canonical art, the "fist" (Iversen 1975; Davis 1989: 20–27; see also Chapter 9, note 1).

Fig. 3.
Rock drawing, southern Upper Egypt, probably fifth- fourth millennium B.C .
Courtesy Winkler Archive (formerly in Egypt Exploration Society,
London), Frobenius Institute, Frankfurt-am-Main.

Fig.5.
Painted wall from Decorated Tomb (Tomb 100) at Hierakonpolis, late predynastic (Nagada IIc). 
Copy by J. E. Quibell and F. W. Green, Petrie Museum, University College, London.

Fig. 4.
Fragment of painted linen from Gebelein, Upper Egypt, late predynastic (Nagada IIc/d).
 Courtesy Egyptian Museum, Turin.

Fig. 6.
Nagada II (Gerzean) Decorated Ware vessel,
probably Nagada IIc/d.
 Courtesy Metropolitan Museum of Art, New York.

Fig. 7.
Designs on three Nagada II (Gerzean) Decorated Ware vessels.
After Smith 1949.

The small design field was handled by the late prehistoric artist with tremendous virtuosity. For example, on the Brooklyn knife handle, 9.5 cm. long by 5.6 cm. wide (Fig. 8), the sculptor carved ten rows of animal figures on both sides, each presenting ten or more creatures, usually all of the same species, walking in the same direction (storks, lion, cattle, and so on)—nearly three hundred tiny figures in all. Bearing fewer figures, the Pitt-Rivers knife handle, 10-5 cm. long (Fig. 9), the Davis comb, 4 cm. across (Fig. 10), the Gebel el-Tarif knife handle, 9 cm. long (Fig. 20), the Carnarvon knife handle, 5 cm. across (Fig. 23), and several other late prehistoric decorated objects exhibit essentially the same formal organization. At least in the Nagada IIc/d-IIIa period—including the Brooklyn, Pitt-Rivers, Carnarvon, Gebel elTarif, University College (Fig. 21), and Berlin knife handles (Fig. 22) and the Davis comb—no ground lines were used to anchor the rows of figures. The Metropolitan (Fig. 11) and Gebel el-Arak knife handles, dated to the Nagada IIIb—that is, "to the end of the tradition of carved ivory knife handles" (Williams and Logan 1987: 250–51)—also lack ground lines. These latter objects, however, are contemporary with the Narmer Palette (Fig. 38), the Scorpion mace head (Fig. 52), and some of the earlier Hierakonpolis ivories, which do have ground lines.

Despite a number of attempts, no commentator has plausibly explained the late prehistoric artists' selection—an act taking place within what I call the scene of representation—of particular species for inclusion in these images or image-making systems. Georges Bénédite (1918) may have been on the right track in seeing the various animal species as emblems of geographic or social territories in late prehistoric Egypt. Unfortunately his hypothesis is too ad hoc in the details to be fully convincing; although the animals may be territorial emblems, some have no documented associations with localities, shrines, towns, or regions (at least, Bénédite does not provide them), and many of the particular associations that he does suggest rest on anachronisms. Furthermore, even if we could show that every animal species depicted was the emblem of some geographic territory or social group, we have understood only one element of the scene of representation—namely, that representation of geographic or social territories was carried out by way of animal emblems. But the images on the

Fig. 8.
Brooklyn handle: carved ivory knife handle from Abu Zeidan, late predynastic (Nagada IIc/d). 
Brooklyn Museum. From Needler 1984.

Fig. 9.
Pitt-Rivers handle: carved ivory knife handle, late predynastic (Nagada IIc/d).
Pitt-Rivers Museum, Farnham, Dorset. From Asselberghs 1961.

Fig. 10.
Davis comb: carved ivory comb, late predynastic (Nagada IIc/d). 
Metropolitan Museum of Art, New York (ex-collection Theodore M. Davis). 
From Hayes 1953.

Fig. 11.
Metropolitan handle: carved ivory knife handle, late predynastic (Nagada III?). 
Courtesy Metropolitan Museum of Art, New York.

objects make reference to or depict this representation. Why, for example, is there more than one member—often ten or more—of the species, or more than one example of the supposed emblem, depicted in the image? And why are they arranged in long rows filling up all the available space on the field?

If we canvas the surviving artifacts, it seems that all the various domains of the Egyptian natural world make an appearance in the images. We find, for example, fish, stork, and heron associated with the Nile River; cattle associated with the tilled fields and pasturage; badgers and porcupine associated with heavier scrub brush or low forest; dogs associated with the margins of the village; and wild oxen (Bos ), antelope and gazelle species, giraffes, wild dogs (Cape hunting dog) or hyenas(?), jackals, leopards(?), and lions associated with the extant savannah and with the desert steppe.^[3] It has often been remarked, however, that some species we might expect to see are not depicted, such as the crocodile or the falcon, both central symbols in later Egyptian mythology and iconography (see Bénédite 1918: 226). As yet, then, it remains unclear what "world" is depicted in the animal-row designs. But the profusion or plurality of individual figures from different species initially suggests an all-encompassing plenitude of nature—in Bénédite's (1918: 2, 234, 236) perceptive words, an excessive multiplicity" or "entirely unabridged representation"—with the quality of a pictorial encyclopedia.

A closer inspection of the designs reveals that the figures are ordered selectively and tightly according to specific "rules" (Bénédite 1918: 235) or to a "very definite conception" (Vandier 1952: 545), perhaps even "a [mental] model the elements of which are scattered" throughout the images made on the entire series of objects (Bénédite 1918: 228). Although a few of Bénédite's (1918) identifications of depicted species are erroneous, and a proper specification of their ordering must sometimes differ slightly from his, we owe to him the most complete description of these rules. Bénédite showed that all figures belonging to one species are confined to the same single row throughout the entire series, except for the cattle (Bos taurus ), which on two examples extend through two rows (Brooklyn and Pitt-Rivers knife handles). A pair of herons always closes the row of storks (Brooklyn, Pitt-Rivers, Carnarvon knife handles; Davis comb); a five-rayed star twice closes the row of badgers (Brooklyn knife handle, flat

side, row 9, left; Davis comb, obverse, row 5, left); rows of carnivores and noncarnivores tend to alternate.^[4] No two rows in sequence are carnivores. Lions appear in the third row (Brooklyn and Pitt-Rivers knife handles; Davis comb) and jackals in a lower one.

With somewhat more room available for the design than on the knife handles and small comb, the more complex images on the carved cosmetic palettes bring the carnivores and ruminants directly into relation with one another in scenes of animals being hunted by their predators. The sculptor of the Brooklyn knife handle (Fig. 8) imagined this possibility without fully developing it; small details of his image may be both the conceptual and the historical context for the magnified depictions of the chase, attack, and kill on the palettes. The Brooklyn sculptor closed rows of oryx, cattle, and sheep with an attacking (or herding?) dog (Canis familiaris; flat side, rows 6, 8, left; boss side, row 4, right). Likewise the Carnarvon knife handle (Fig. 23), now greatly abraded, includes a lion pursuing(?) bull oxen (flat side, bottom row, left). The badly preserved boss side of this handle should not be regarded as a row arrangement at all but is more like a hunt scene (it has an anomalous order to be considered shortly). As a third example of a similar detail, there is on the Davis comb (Fig. 10), a dog pursuing(?) the lions (obverse, row 3, left), as on the boss side of the Carnarvon handle (top left corner?). (Apparently because he wants to interpret the dog as "herding," Churcher [1984] incorrectly identifies these clearly leonine creatures as "bovids.")

In addition to its orderly rows with a few anomalous interventions, the "formula" for these images includes three types of groupings—that is, sets of different creatures that invariably accompany one another as a single unit. The first is an elephant trampling on a pair of serpents (pythons?) acting as its ground line (Brooklyn handle, flat side, row 1; Pitt-Rivers handle, flat side, row 1; Carnarvon handle, flat side, row 2, right; Davis comb, both sides, row 1 [the serpent is single]; Seyala scepter [Fig. 24]). The second is a trio consisting of a snake (sand boa), a stork, and a giraffe (Brooklyn handle, flat side, row 2, right; Davis comb, both sides, row 2, left and right; Carnarvon handle, flat side, row 1, right [snake missing]; Seyala scepter [snake missing]; the Pitt-Rivers handle edge, where we would expect to find this motif, is missing). The

third is a vulture plucking at prey (a snake?) grasped in its talons (Brooklyn handle, boss side, row 1; Pitt-Rivers handle, boss side, row 1).^[5]

The iconographic import of these groups of linked animals is unknown (see Churcher 1984: 164, 166). The way they appear in the representation, however—exhibiting what Bénédite (1918: 240) called the "jeu d'esprit, " or even the "satiric humor" of the artist—introduces a crucial discontinuity into the uniformity and continuousness of the many rows of identical figures. Driving a wedge, as it were, into the scene, the artist forced apart the neat, "natural" rows of identical animals to insert transnatural representations. These groupings figure and perhaps even denote the artist's power to represent nature in a scene offered as a legible transcription of plenitude merely responding to the constraints of the support. The artist's power is present throughout these images, often unnoticed at first glance. For example, where space to complete a row begins to run out, an anomalous figure is inserted to fill up the available area. Thus the artist maintained the un anomalous order of the overall design, felt first and foremost as an uninterrupted whole plenitude: the viewer does not immediately encounter an interruption or gap in the row because it is subtly, almost surreptitiously revised so as to avoid appearing incomplete or, precisely, as a blank—framing, interrupting, or encroaching on the row of animals as such. On the Brooklyn handle (Fig. 8) two long, thin fish appear where space is tightest (flat side, row 10, left; boss side, row 1, right). Several parallel efforts can be found on the Pitt-Rivers handle (Fig. 9; boss side, row 1, right [fish?]; row 4, right [another species of fish?]; row 3, right [ram?]; flat side, row 1, left [ram?]; bottom row, left [bird]).

Where space on the surface exceeds what is smoothly filled by the rows of figures with their heads, legs, tails, and so on, oriented in particular ways, the artist actively spread the representation into it by interrupting and revising the standard rendition of the animals' bodies. Here again, however, the unanomalous or uninterrupted quality of the whole—presumably, then, as a depiction of a complete plenitude—is preserved by making revisions or inserting anomalies, which have the effect of preventing the encroachment of spatial blanks or gaps on the plenitude. On the Pitt-Rivers handle a ram is inserted at the end of a row of ungulates (boss side, row 2, right) so that its horns can fill up the curve

of the edge. On the Brooklyn handle one lioness's tail points up, rather than down as usual, for the same reason (boss side, row 3, right), a device adopted also on the Carnarvon handle (Fig. 23; boss side, bottom row, right).

These interventions are evidently determined by considerably more than a mundane horror vacui, for at least one anomalous array cannot be explained even in part by such a lowly decorative motivation to fill up the available space. On the Pitt-Rivers handle a gazelle (G. dorcas ?) appears playfully in the middle of the row of oryx (boss side, row 2, fourth from right), where no spatial pressures exist. (Churcher [1984: 167–68] identifies deer, hartebeests, oryx, and Barbary sheep in this row, although it would be more in keeping with the organization of the images to identify only one anomalous creature in the row; the other variations among the figures probably result from chance in drawing and are not significant.)

In sum, the entire design in these works, with its orderly rows, fulfills an ambition to fill the available space for depiction with the represented plenitude—to the extent that representation itself appears absent. The intervention of anomalous and revisionary details in the whole therefore reintroduces the representation that the whole image seems to exclude from the plenitude of nature.

Roughly two hundred years later the canonical artist used a deceptively simple device to organize these neat rows of animals. From the early First Dynasty onward, all figures in the frieze are linked in a continuous band with an underlying ground line (Figs. 12–19).^[6] In an intervention already imagined in the anomalous details we have noted in animal-row designs, the canonical artist established his representation as grounded in nature, for the ground line seems to follow along behind or underneath, literally to underline, the preexisting species and numbers, the connections and intervals, of what is naturally there to be represented.

But despite the way it seemed to be advanced as an underlining—a fillip in grounding a legible depiction—like the anomalous details we have noted, the ground line completing a row of animals masked the composing blow creating the very order to be represented. For the ground line of canonical art (Figs. 1, 2) was the skeleton and armature of the entire composition; drawn

Fig. 12.
Carved ivory spoon from Tarkhan, early First Dynasty. Petrie Museum,
University College, London. After Petrie 1913.

Fig. 13.
Carved ivory plaquette from Hierakonpolis, early First Dynasty(?).
Petrie Museum, University College, London. 
From Quibell and Green 1900–1901.

Fig. 14.
Carved ivory rod from Hierakonpolis, early First Dynasty(?). 
Petrie Museum, University College, London. 
From Quibell and Green 1900–1901.

Fig. 15.
Carved ivory tusk from Hierakonpolis, early First Dynasty(?). 
Petrie Museum, University College, London. 
From Quibell and Green, 1900–1901.

Fig. 16.
Carved ivory plaquette from Hierakonpolis, 
early First Dynasty(?). 
Petrie Museum, University College, London. 
From Adams 1974 by permission of the author.

Fig. 17.
Carved ivory tube, early First Dynasty(?). 
Egyptian Museum, Berlin. From Scharff 1929.

Fig. 18.
Carved steatite scepter head from Hierakonpolis, early First Dynasty. 
Petrie Museum, University College, London. 
From Quibell and Green 1900–1901.

Fig. 19.
Carved shell plaquette, early First Dynasty(?).
Egyptian Museum, Berlin. 
From Smith 1949.

first, not last, it was not an underlining at all but an initial and wholly determining ruling .

In other words, the late prehistoric and the canonical scenes of representation-at the one end, the small decorated knife handles and combs of Nagada IIc/d (Figs. 8–10) and, at the other, the compositions of Old Kingdom art (Figs. I, 2)—are divided by a complex revision of image making. As we will see, the human presence will be introduced obliquely into the depicted world of nature, entering from the edge or side. In the scene of representation this human presence is both within depicted nature, setting out to master it, and outside the depiction altogether, setting out to interpret it. From the point of view of represented nature itself, the oblique or curving progress of the human presence is unnatural for it. When the side entrance is provided in the scene of representation with a straight ground line, however, its reality as oblique is masked and its unnaturalness disguised. Although nature is mastered in and by human image making, the image itself produces a fiction of the absence of a masterful interpretation of the world, a fiction of its own grounding in nature rather than the work of representation.

3—
Circling the Scene

The carnivores and noncarnivores—potentially antagonistic species—are depicted in alternating rows on the knife handles and combs being examined here. They confront one another directly only in anomalous passages; for example, the artist of the Brooklyn handle (Fig. 8) closed a row of cattle with an attacking dog. What appears as an anomaly on the Brooklyn handle, however—a subsidiary intervention, literally on the edge of representation—can become the entire representation in full-scale depictions of the hunt. Presented on three decorated ivory knife handles (the Gebel el-Tarif, University College, and Berlin handles; Figs. 20–22), the carnivores and their prey are not associated with any elements except a plant-leaf rosette and, on the other side of the handle, intertwined serpents. With each beast and its prey alternating along a single baseline,^[1] the plaiting together of the separate rows of the formula is analogous to the two intertwined serpents appearing on the side opposite the hunt scene, filling a long row along one edge of the handle; the intertwined serpents could even be said to indicate how the composition on the other side should be interpreted (see also the serpents below the elephant in the animal-row designs of Figs. 8, 9, 23, and 24). The most extensive of these three images, the Gebel el-Tarif knife handle (Fig. 20), depicts a lioness or spotted leopard attacking an antelope or deer, a lion attacking an oryx, a wild dog(?) attacking a wild boar(?), and a winged griffin attacking an ibex or goat(?) (Vandier 1952: 548–49) in alternating left-facing and right-facing bands.

The Carnarvon knife handle (Fig. 23), more sophisticated and also more innovative than these works, follows the same process of arriving at the order they exhibit but simultaneously revises it. In this respect it is the corollary—at the "beginning" of the chain of replications of late prehistoric representa-

tion—of the Narmer Palette (Fig. 38) at the "end." Both the Carnarvon handle and the Narmer Palette refashion a symbolic image by substantially revising, combining, and suppressing elements of an earlier metaphorics—in the case of the Carnarvon handle, the animal-row and the animal-hunt designs on works like the Brooklyn and Gebel el-Tarif handles, and in the case of the Narmer Palette, the very narrative image the Carnarvon handle begins to develop. Both the Carnarvon handle and the Narmer Palette exhibit the discomposure, the greater or lesser illegibility in terms of established pictorial textuality, that attends radical revision.

On the Carnarvon handle it is not clear whether the animal-rows pattern (from the Brooklyn and Pitt-Rivers or similar knife handles) derives from the carnivores-and-prey motif (from the Gebel el-Tarif or similar knife handles) or vice versa, or whether, more likely, both formulas subsist together as related, dependent terms embedded in a larger scene.^[2] We can easily follow the maker plotting the image. He first laid out the row of storks at the top of the flat side, beginning with the well-established giraffe-and-stork motif from the animal-rows formula. That side of the image finishes, in the bottom row, with the lion following a group of bull oxen. Since the lion is not as bulky as the bull oxen, and since its tail follows the curving edge of the handle, it fits into the row as an anomaly, precisely to preserve the uninterrupted continuousness of the depiction of plenitude on this side of the image. The lion is dropped down a row from its "natural" place in the row of lions, where a lion is "pushed out" of its expected row by the elephant-and-snakes motif used to open it.

Next the maker flipped the handle over to lay out the top row on the boss side, beginning with the lion(?) (Churcher [1984: 168] incorrectly identifies a bovid) pursued by a dog that is overlapped slightly by a following hartebeest. This array violates both the identical-animal-rows formula and the carnivore-following-prey formula as well as the previously allowable revisions and supplementations to these formulas, the insertion of a single anomaly at an edge or side to preserve the uninterrupted continuousness of the formulas. Thus instead of continuing and extending the formulas and their revisions, the boss side of the Carnarvon handle escapes them altogether. More exactly, the top row on this side is a complete synthesis and revision of the entire flat side.

Fig. 20.
Gebel el-Tarif handle: ivory knife handle (covered in embossed gold leaf) and flint blade,
said to be from Gebel el-Tarif (el-Amra), late predynastic (probably Nagada IIc/d).
Cairo Museum. From Quibell 1905.

Fig. 21.
University College handle: carved ivory knife handle and flint blade, late predynastic (probably Nagada III).
Courtesy Petrie Museum, University College, London.

Fig. 22.
Berlin handle: carved ivory knife handle, late predynastic 
(probably Nagada III). 
Egyptian Museum, Berlin. 
From Asselberghs 1961.

Fig. 23.
Carnarvon handle: carved ivory knife handle, late predynastic (Nagada III). 
Metropolitan Museum of Art, New York (ex-collection Lord Carnarvon). 
From Bénédite 1918.

It is not surprising, then, that the remainder of the boss side seems to "degenerate" (Bénédite 1918: 235) from presenting rows of animals to spreading out a complex composition legible only by taking the relative orientations of individual figures to give a "plan" or bird's-eye view of the action. According to this perspective, the image shows a dog pursuing a lion(?); a dog, at the edge, pursuing a hartebeest; and what may be two wild deer (center) and an oryx (bottom left) pursued by felines (leopards?). A rosette adorns the boss itself. (Churcher [1984: 168] identifies the two deer, has the oryx followed by a cow[?], and interprets the degraded bottom figures as antelopes[?], an even more anomalous grouping than the one I propose here.) The last figures to be worked into the image were evidently the leopard(?) at the bottom right edge, shown turning its head back toward the right-hand deer, and the dog, filling up the space on the outside curve of the edge and positioned at a right angle to the neat rows on the flat side of the handle. At least one pursuer, the right-hand leopard(?), does not face or follow but rather turns back to look at its prey, a device replicated and greatly elaborated in later works in the chain of replications; and at least one prey, the left-hand deer, seems not to be pursued at all.^[3] Although the scene may depict beasts of prey and their victims, they are not literally shown as hunting. More exactly, the representation of hunting depicts not the attack or blow itself but rather the entire scene of sighting the enemy and masking the agent of the blow, the preconditions of the blow itself.

On the Carnarvon handle the edge is a place of "wildness," where representation enters the scene of nature obliquely and where the artist and the viewer are most directly represented in it. Here the wild dog who disturbed the serene plenitude of the Brooklyn knife handle is the only figure who can be fully seen in the image when the hunter who owns the knife actually picks it up and wraps his hand around the handle. Thus the wild dog stands for the human hunter in relation to the scene of nature to which the hunter is oriented at right angles. When the knife is used to wield the blow, the dog plunges forward just ahead of the hunter's real force. (If we were to accept these objects as having a "magical" status, then the dog could be said to be a magical guide or reinforcement of the hunter's action.) The dog is therefore the mask of the hunter's blow; the hunter stands and his blow takes "outside" the depiction per se. This and

similar if more complex devices in other works constitute the fundamental pictorial mechanics and metaphorics of the late prehistoric Egyptian scene of representation.

With different compositions on its two sides, the Carnarvon handle suggests that the animal-row images and the animal-hunt images on works like the Brooklyn and Gebel el-Tarif handles were apparently viewed as distinguishable from one another but nonetheless to be juxtaposed—although perhaps only in a revisionary repetition—as related terms. The maker of the Carnarvon handle evidently set out to bring animal-rows and carnivores—pursuing-prey motifs into formal alignment and therefore, presumably, into some kind of metaphorical association. A decorated scepter or mace handle from Seyala in upper Nubia (Fig. 24), probably roughly contemporary with the Carnarvon handle (Kantor 1944: 129–30; Williams 1988a: 4–5), exhibits an identical structure worked out in a somewhat different physical format. Although not presenting two sides of an artifact to be turned over in the hand or flipped, the pictorial text is clearly divided from top to bottom on an object that can be rotated around itself. At the top of the scepter two groups of animals replicate—derive from and present compactly—the animal-rows formula. One element of the formula, the rows of various animals, is the stork and antelope; the other, the specific "groupings" of animals, is the elephant-on-snakes motif plus giraffe. At the bottom of the scepter three groups of animals replicate the carnivores-and-prey or animal-hunt formula, a lioness(?) pursuing an oryx, a lion pursuing a stag, and a leopard(?) pursuing a hyena(?). But as on the Carnarvon handle, the distinction between the two formulas, in its very repetition and reduction to a synthesis, is to some degree broken down. The five groups of animals cannot be divided into two equivalent classes, one for the animal-rows formula and the other for the carnivores-pursuing-prey formula.

In light of what we discover about the metaphorics of the lioness and/or lion in images like the Hunter's and Battlefield Palettes (Figs. 28, 33), it is significant that the figures of this creature, in the middle row and the following fourth row of the Seyala scepter and on the Carnarvon handle at the "end" of the flat side and the "beginning" of the boss side, carry or provoke the elision and revision in the presentation of the two formulas. Later replications of the

Fig. 24.
Embossed gold mace handle from Seyala, Nubia, late predynastic (Nagada III?). 
From Kantor 1944 by permission of the author.

lion figure may refer retrospectively to these prospective revisions in which the power and danger of the beast has been increasingly represented in its central, albeit discomposing, place within the scene of nature's plenitude. It will not be surprising if the figure of the lions comes ultimately to represent the person of the human ruler. The dog has a similar general status as a wild figure, an anomaly within the formal order of the formulas that allows them both to be sustained and to be revised and reduced. Later revisions of the representation of the wild dog—as they return retrospectively to these images—will give it an increasingly literal position as the "frame" of the depicted scene, as on the Oxford Palette, where two wild dogs are placed like heraldic emblems at the top and surrounding part of the image (Fig. 26).

The significance of' lion and wild dog in the images on the Brooklyn or Carnarvon handles and the Seyala scepter cannot be the same as in later images, where we discover that the beasts function somewhat differently in each succeeding replication. But it is fair to say that on the Carnarvon handle both animals are nodes on which the pictorial mechanics literally turn as well as sites of strong metaphorical activity. Although such symbolic or metaphorical meanings are probably irretrievable, the beasts of prey might stand for the human hunter, who actually uses the knife in killing, carving, skinning, or sacrificing the hunted animals. The human hunter is not literally in the depicted scene, but he figures at its edges and surrounds or encloses it on all sides and outside the frame. It is he who sees the entire scene, turns the object over on itself in "following" the lion through the image, and then wraps his hand around this folded-over image to see only the wild dog just ahead of him plunging at its prey. The prey hunted by the dog within the depicted scene is not necessarily the same as the prey hunted by the hunter outside the depicted scene, which that depiction represents. As will become clear, the viewer can be the prey. In the same way the hunter is masked from the identities depicted within the scene by the dog who represents him, the entire image masks the hunter from the viewer, for if the viewer is viewing the scene, the knife is not being gripped—that is, the hunter is not hunting. At most the viewer sees the depicted scene only in the moments immediately before or immediately after the hunter strikes his decisive blow. The blow itself is always outside the depiction, if nonetheless the condition of and for the scene of representation.

4—
Entering the Scene

Decorated ivory artifacts, like the Brooklyn and Carnarvon knife handles (Figs. 8, 23), and other small-scale decorated objects, like the Davis comb and the mace handle from Seyala (Figs. 10, 24), presented the maker with restricted surfaces for the preparation of an image. Many cosmetic slate or schist palettes are also relatively small in scale, usually 10 to 20 cm. in height, and likewise offered a restricted polygonal surface—indeed, two possible surfaces on either side of the palette—bounded by the edges of the object. Incised decoration was a feature of these palettes almost from the beginning of their production in the Nagada I period (see, for example, Capart 1905: fig. 39; Baumgartel 1955: fig. 2). Moreover, again beginning early in the manufacture of the artifact type, many palettes took the form of an animal in their outline, occasionally with incisions and insets to indicate details like muzzle or eyes (for typology, see Petrie 1921). One typical theriomorphic palette had a roughly tapering shape and surface, often unadorned, crowned with two bird heads facing away from each other (Petrie 1921: Types 76–80). "Ceremonial" palettes decorated with often complex images, such as those discussed in this chapter, are believed to derive from such works; although some have the more rectangular form typical of later funerary palettes, Nagada III or early dynastic in date, their scheme of design is consistent with the earlier formats. A reliable archaeological date for what may be the beginning of the group of ornamented double-bird-head palettes—and thus for the chain of replications considered here—is provided by a palette, now in the British Museum, ornamented in very low relief with what has generally been interpreted as the emblem of the god Min (Asselberghs 1961: fig. 117; Davis 1989: fig. 6.8a), a divinity represented also in sculptural colossi known to be late prehistoric in date (Williams 1988b; Davis 1989: fig. 6.21). The palette was discovered in Grave B62 of the late prehistoric cemetery

at el-Amrah, assigned to Petrie's "Sequence Dates" 56–64, probably SD 58 or Nagada IIb/c (Petrie 1920: pl. 51 [B62]; Baumgartel 1960: 90).

Side Entrance

The Ostrich Palette (Fig. 25), at least two palm widths across (16 cm.) and at least two hand lengths long or high (41 cm.), seems to have been made at the same time as the carved-ivory knife handles and comb and certainly at the time of the Carnarvon handle; it has no archaeological provenance (Crompton 1918). On typological grounds its shape—especially the two out-facing bird heads with five pointed radii, each partly drilled through, in between—can be no earlier and is probably not much later than Petrie's Sequence Dates 57–58, that is, Nagada IIb/c (Davis 1989: 139–40). As for its place in the chain of replications of late prehistoric Egyptian art, however, we must consider it to be a stage in the replication of the scene of the order of nature (and of the human presence in relation to it) that is later than the knife handles and comb.

The image, carved in low relief, presents a bird-headed personage with human limbs wearing a belt with penis sheath; although Vandier (1952: 573) asserted that he carries a baton in his left hand (now damaged), there is none to be seen. The protagonist follows and seems to complete a group of three ostriches, somewhat like an anomalous creature inserted at the end of a row in the animal-rows formula and equally like the beast of prey pursuing its victim in the carnivores-and-prey formula. His head shaped to resemble those of the ostriches, he is, then, represented as disguised as an ostrich with distinctly human torso, limbs, and dress. The ostriches certainly cannot see that he is not one of them, for they are depicted as not seeing him at all. Unlike the presence figured at the edges of and outside the scene on the smaller knife handles and comb we have examined, this hunter, masked by an animal, is now introduced into the depicted scene.

The few scholarly commentaries describing this work have dismissed it as merely clumsy, a minor work with awkward figures positioned at peculiar tilts. In fact the Ostrich Palette presents a highly calculated image. It does not set out to replicate the neat order and straight baselines of either the animal-rows

Fig. 25.
Ostrich Palette: carved schist cosmetic palette, late predynastic (Nagada IIc/d, Sequence Date 57–58).
Courtesy Manchester Museum.

formula or the carnivores-and-prey formula. Instead, manifesting the composure evident on the boss side of the Carnarvon knife handle (fig. 23) and on several other carved palettes to be considered later, the image embeds the animal-rows and carnivores-and-prey scenes within a more complex scene of representation. More exactly, it advances the metaphorics of those images into the depicted scene itself.

According to its "bird's-eye view" of the relative orientations of figures, the Ostrich Palette depicts the masked hunter coming up obliquely behind the right side of the third ostrich, becoming a pair with it, following the two leading ostriches in their forward progress. His legs, with knees bent forward, imitate the ostriches' own bobbing gait, their knee joints bent backward; while they are leaving the scene, he is entering. Unlike them, he has two free arms, upraised, and is ready to throw himself on his prey and wrap himself around its body, just as the real hunter (outside the scene) holds the bird-headed palette in his hands to view this depiction of his mastery. (Perhaps it is just because he cannot close one hand over it—as over a knife handle [see Fig. 23]—that the human hunter literally enters the depicted scene.)

The masked hunter enters the scene from its sideline, or, more exactly, from below it. The baseline of the entire group of figures is the forward, masked edge of the space, more than two thirds of the visual field available for depiction (but utterly lacking it), where the human owner of the palette—and the hunter, real or metaphorical, depicted on it—grasps the tapering palette and mixes cosmetic eye paint to decorate and mask his face, or sharpen the sight of his eyes, before entering the scene of wielding his blow, real or metaphorical. The band of depiction on the palette, then, faces and perhaps serves as the metaphor for the band of paint the human owner and "hunter" paints across his own face.

These are, of course, relationships established by and in the scene of representation. We do not know how the cosmetic ground on this or other decorated palettes was used. It could have been applied to a cult image or divine statue as well as to the face of the owner of a palette, perhaps the same person as the "hunter" or "ruler" who is depicted in the images.^[1] In fact we do not know whether pigments were actually ground on the palettes; no trace of cosmetic

material has survived on the surfaces or in the saucers of any palette in the series examined in this study. (That the images evince a literal concern with the "sharp sight" of the hunter or ruler is suggestive, however, as Asselberghs [1961: 266] seems to have been the first to appreciate.) As we will see, the saucer for containing the cosmetic—whether it was really used or not—plays an increasingly prominent role in the structure of the images on the various palettes. It does not appear at all on the Ostrich Palette. On the Oxford Palette (Fig. 26), a "later" replication to be examined momentarily, the saucer is framed decoratively by intertwined animal figures. On the Hunter's and Battlefield Palettes (Figs. 28, 33) it divides the image into separate units that have distinct temporal and spatial positions in the narrative. Finally, on the Narmer Palette (Fig. 38), it has an even more peculiar, apparently symbolic status.

It could be, then, that the meaning of cosmetic, and consequently the function of the palettes and the images they carry, evolved substantially from the earlier to the later phases of the chain of replications—as we would fully expect if applied cosmetic had a representational status in itself and therefore its own textual history of repetition, revision, and refusal. Although we cannot confirm it through archaeological evidence, this is one possible "external" explanation for the "internal" textual development of the images.

On the Ostrich Palette the scene depicted is the representation of the hunters' mask and its effects . It shows as belonging to the scene of nature, to the plenitude of things transcribed by representation, an order of nature the real owner of the palette or the real person of the "hunter" must actually set out to create. To depict the mask, then, is to continue to mask the blow; the blow is struck from behind the mask.

About-Faces

The Oxford Palette (Fig. 26) replicates—repeats and revises—these metaphorics in a much more complicated image. Roughly the same height as (42 cm.) and slightly wider than (22 cm.) the Ostrich Palette, it can be shown on typological grounds to be probably somewhat later in date, certainly Nagada IIIa

Fig. 26.
Oxford Palette: carved schist cosmetic palette, late predynastic (Nagada IIc/d or IIIa),
from Main Deposit at Hierakonpolis. 
Courtesy Ashmolean Museum, Oxford.

(see Davis 1989: 139–40 for full discussion). The masked hunters on these two palettes are the only such instances identifiable in late prehistoric and early canonical Egyptian art. The hunter appears on the Oxford Palette with human dress (belt and pouch or bag) and human limbs, hands, and feet but with a jackal's tail and head or mask, piping a flute at the edge of a group of beasts hunting their victims (reverse bottom left). This puzzling figure has been identified as a "monkey with long ears" (Legge 1900: 132) and as a "fox" (Vandier 1952: 582). He is certainly a human hunter, but it cannot be maintained, as some have said, that he is grasping a stick (Williams and Logan 1987); he holds the implement or instrument he carries up to his mouth. The uncertainty of commentators about the identity and properties of this figure testifies to the general difficulty of determining how the overall image works.

The Oxford Palette echoes the content of the animal-rows and carnivores-and-prey formulas ranged along straight baselines in the images found on the knife handles and comb. And in all likelihood it develops from these images while strongly reconfiguring the formulas. The palette exemplifies the compositional mode in late prehistoric Egyptian art that commentators have sometimes called "chaotic" or "disorderly" or at least "less formal" than canonical art (see, for instance, Groenewegen-Frankfort 1951: 18; Fischer 1958: 65, note 4). Putting the matter in positive terms, the disposition of the figures in the "space" of the scene is indicated by the relative orientation of their individual baselines observed in "plan" or bird's-eye views (Fig. 27); the viewer must look "down" on the scene in handling the image. For example, on the obverse (or saucer) side of the palette, in the first row of animals below the serpent-necked panthers or "serpopards,"^[2] the three horned victims of the attacking dog are meant to be running in the same direction and roughly in a straight line. The right-most creature is shown to be veering slightly to one side, namely, to the left of the general line of flight. The pursuing dog is shown to be attacking the hindmost victim by running at it from behind and one side—that is, again, from the left of the line of movement of the victims.

Once properly understood, this compositional technique (see also Davis 1976) is both informative and flexible. Nevertheless the viewer must shuttle through, twist around, and rise above the scene to grasp it as a scene rather

Fig. 27.
Structure of the Oxford Palette, obverse and reverse.

than as a merely decorative (space-filling or rhythmic) disposition of elements on a field with figures spread at right angles to the line of sight. Indeed, like the masked hunter within the depiction, the viewer enters the scene and becomes part of it in the movement of viewing.

The most hieratic passage of the image—explicitly echoing the animal-row and animal-pair formulas—appears in the top two-thirds of the obverse side of the palette, around the cosmetic saucer. As the evidence of the chain of replications strongly suggests, late prehistoric compositions on decorated palettes are meant to be viewed beginning with this obverse (saucer) side and with the saucer area playing a key role. Thus in this area of the image on the Oxford Palette the viewer is introduced to it through a symmetrically balanced, almost heraldic expression whose narrative complexity and ellipsis unfolds only later in the time of viewing as the object is handled and the image spread on both sides is scanned from top to bottom and side to side. In order to make the point, my discussion will necessarily follow the interpretation of the image in the time of the viewing—the network of internal associations being forged and further questions being raised as the viewer moves through the image and tries to understand its coherence. Here it is crucial to remember that the threads of the image cannot be completely tied up together all at once; rather, they are raveled, and to some extent unraveled, in a complex unfolding as the object is handled and scanned. For example, the viewer encounters ellipses or gaps only to the extent that expectations being set up in the viewing are being frustrated; the viewer feels the image to be a revision of elements of the existing chain of replications only to the extent that the chain seems to be both repeated and modified as the viewer progresses through its latest version. Such facts imply that a verbal description of the viewing experience is difficult to produce and to follow. But I will come back to a summary of the "whole" structure of the image, as finally deciphered.

At the top of the palette and extending along the edges, the saucer is surrounded by two wild dogs (Cape hunting dog) clasping one another's paws. Their heads (one missing) project beyond the main surface. In this respect the Oxford Palette resembles the more routine funerary palettes (Petrie 1920, 1921) with projecting double bird heads or other animal outlines, but it is unusual in

having the two animals facing each other with the entire length of their bodies stretched along the sides. The wild dogs have usually been interpreted as emblematic creatures similar to those in Minoan-Mycenaean painting, where "the intention is not so much a naturalistic presentation as the transfer of power or protection to the object they adorn or confront" (Immerwahr 1990: 136). Thus we might see the wild dogs on the Oxford Palette as "frames" in a sense: they are "outside" the image but provide for its integrity.

The dogs' heraldic, framing function might, however, bear an analogical or metaphorical relation to the narrative presentation proper. Wild dogs also appear within the image on both sides of the palette: on the obverse, there are three running along the lines established by the snakelike necks of the serpopards (the Hunter's Palette [Fig. 28] brings this formal conceit into the narrative itself) and on the reverse one wild dog moves, with its head turned back, toward the left edge. We observed on the Carnarvon handle (Fig. 23), an "earlier" work in the chain of replications, that the wild hunting dog may have been an analog for the person of the human owner, perhaps the "hunter" himself. It is possible that the framing dogs on the Oxford Palette do so as well, inasmuch as the palette is grasped by its human owner precisely with thumbs placed over the wild dogs' bodies at the edge of the object, their heads projecting in front as an extension of the owner's hands.

There is an important revision in this representation, even if it has indeed been maintained from the earlier to the later replication. The wild dogs that surmount the image as a whole do not frame everything the pictorial text relates. Outside the area bounded by the bodies of the wild dogs along the edges of the palette, the bottom third of the image depicts not wild dogs—the Cape hunting dog has the status of a natural carnivore like a jackal or lion—but domesticated hunting mastiffs, differentiated by floppy ears and heavy studded collars, who more directly figure the human hunter's presence in the scene. The surmounting wild dogs provide one overall frame or title, "announcing" the theme of the representation of the human hunter within the scene of nature. Once some metaphorical and narrative expectations are set up, getting the viewer started, the further development of the image supplements this announcement in several ways.

At the top of the framed area, below the clasped paws of the wild dogs and between the heads of the two serpopards, is depicted a single ostrich—a creature that is clearly near the beginning of the pictorial text in the formal structure of the image. If we regard the surmounting dogs as a framing announcement providing a preliminary specification of the image, the ostrich initiates its development proper in and through the act of viewing.

Within the chain of replications of late prehistoric representation, the derivation of the ostrich almost certainly lies in earlier works such as the Ostrich Palette (Fig. 25). Its replication in later contexts substantially revises and even refuses the metaphorical or narrative message of the earlier, much simpler one. As we saw in the first part of this chapter, the ostrich on the Ostrich Palette was about to be attacked and captured by a masked human hunter with upraised arms about to throw himself on his prey. Here, at the top of the Oxford Palette, the ostrich lifts its wings as if to take flight, escaping the human or any other hunter. Although the masked human hunter does appear on the Oxford Palette, it is not as a predator following and pursuing his victim, as on the Ostrich Palette's revision of the animal-rows and carnivores-and-prey formulas. Instead, on the Oxford Palette the masked human hunter is shown on the other side of the image (bottom reverse) as if luring and setting up an attack to be carried out by the hunter's dogs. Thus the depiction of the masked hunter on the Ostrich Palette has been wedged apart.

The ostrich opening the image on the Oxford Palette establishes not only the repetition of the theme of the hunt but also the very fact of its revision precisely as the wedging apart and supplementation of earlier formulas. At this level the ostrich probably alludes to the disturbance of the rows of birds in the animal-rows formula. What had intervened in those rows—the giraffe appearing in the grouping of snake and stork and giraffe—appears now in a crucial position on the Oxford Palette at the farthest available distance from the ostrich at the top of the obverse: namely, at the bottom of the reverse. In other words, the two elements of the formulaic grouping (snake and stork = serpopards and ostrich) are separated from the third (giraffe) in the creation of an extensive image spread on both sides of the palette. Thus the image reinstates the animal-rows and carnivores-and-prey configurations, supplementing them, however,

with a complex metaphorical and narrative depiction of the presence of the human hunter.

Presuming that this complex image is not viewable all at once, like the earlier animal-rows or carnivores-and-prey formulas or the hunting scene on the Ostrich Palette, how does the viewer know how to proceed through it? Although the question is difficult to answer definitively, it is likely that the figure of the ostrich beginning the pictorial text provides the necessary instructions for viewing. A ground-running bird depicted as about to take its short and awkward flight, the ostrich plausibly represents a key to its cipher, the mechanism of viewing the viewer must adopt to interpret the image: the viewer follows a straight line and then takes a "jump" or "leap" in a flip from the obverse to the reverse of the palette.

In this way the viewer activates chronological and causal transitions from one state of affairs to another in the underlying narrative. (A pictorial narrative must somehow generate such mobility in what would otherwise be depictions that do not represent ongoing processes; see the Appendix.) Thus he or she produces the narrative in a structured interaction with a text that provides the appropriate conditions for its own intelligibility—for example, by framing, "sign posting," and other devices organizing passages so that the viewer goes through them in the relevant order. In late prehistoric Egyptian art a cipher key of the kind I identify here seems to appear in all the complex images it is possible to reconstruct more or less completely as narratives. The artist produces the act of viewing the narrative—produces the viewer's production of narrative—by a textual specification that is, as such, non-narrative, although the same element, in this case the ostrich, may also have a thematic status in the image as a whole. While the framing wild dogs announce the general theme of the image, to be supplemented and modified as the viewer proceeds, the ostrich provides the appropriate "guide for the reader."

Within the area of the image surmounted by the wild dogs and below the ostrich, two serpopards are shown licking the back of a stumbling gazelle. The motif is often interpreted as depicting a nonviolent if incipiently aggressive action; its specific designation and connotations are not known. A related passage appears in the same position on the reverse side of the palette: two

rampant" lions are depicted as biting two gazelles belonging to the same species (with short up-curved horns and short tail) as the gazelle on the obverse. As replications of the carnivores-and-prey formula, these particular obverse and reverse motifs are unusual; more standard versions of the formula appear in lower portions of the image on both sides. Rather than chasing their prey, the carnivores in the top passages approach frontally; and the action in both cases is ambiguous—on the obverse "licking" rather than attacking, and on the reverse "biting" in a fashion that almost resembles kissing. The face-to-face encounters involving serpopards and lions with gazelles—seemingly free of conflict—must be metaphorical rather than literal. The wild dogs facing one another and clasping paws, surmounting the whole and framing the top two-thirds of the image on both sides of the palette, thus provide a symbolic announcement of this theme.

To make sense of this portion of the image, the viewer will consider it in the context of the whole: he or she proceeds from the obverse, where the single gazelle is licked by two serpopards, to the reverse—encountered after a flip of the palette—where two gazelles are confronted by two lions. Because they have the same formal arrangement—both motifs appear directly below and are framed by the wild dogs clasping paws—the viewer regards each in terms of the other. Consequently a question arises in viewing the image about the passage of depiction at the top of the obverse. Because there are two gazelles on the reverse, should there not be another gazelle on the obverse, the "twin" of the gazelle at the top? The viewer discovers this "absent" gazelle at the bottom of the obverse, outside the area framed by the wild dogs' bodies and below the serpopards. Here, in a standard carnivores-and-prey formula, the gazelle is pursued by one of the hunter's dogs, which attacks from behind and pounces on the back of its victim. Having encountered the motifs of "licking" and "biting" or "kissing," the viewer now construes the standard formula as an explicit alternative to or revision of the passages of depiction at the top of the image on both sides of the palette.

The discovery of the "second" or "twin" gazelle and the interpretation of the overall image it helps to sustain is likely to be made in the middle of the time of viewing the scene as a whole. The viewer understands the gazelle at

the top of the obverse and the gazelle in the complex group at the bottom of the obverse as necessarily related only after he or she has turned to the other side of the palette, at the top of which the two gazelles are brought close together in their relation to the lions. This search for and understanding of the "twin" gazelle is structured by the other creatures appearing at the top of the obverse side of the image along with the "first" gazelle and the serpopards—namely, the ostrich between the heads of the serpopards and the three wild dogs running along their sinuous necks. Just as there are three animals in the scene of nature hunting (the gazelle and two serpopards), so must there be three wild dogs accompanying the scene—for the wild dogs, as the surmounting dogs attest, represent the overall theme of the hunt. Two of these internal wild dogs are in symmetrical placement, facing each other on the left and right edges. The third, just below the cosmetic saucer, is linked symmetrically not with a twin of itself but rather with the gazelle and ostrich above the cosmetic saucer. The gazelle runs and the ostrich "jumps" and "flies" from left to right (that is, from obverse over to reverse) while the third wild dog runs right to left (that is, back again). Jointly, then, they indicate, first, the flipping of the palette—in this move, the viewer discovers the two gazelles and two lions in exactly the same position on the reverse as the single gazelle and serpopards on the obverse and thus raises the question about the "twin" gazelle on the obverse—and, second, the return to the obverse, in which the viewer discovers the metaphorical place of the entire top of the image on both sides in relation to the bottom outside the wild dogs. The ostrich is the node of the activating mechanism, under the wild dogs, between the heads of the serpopards, and directly above the stumbling gazelle; the two (left and right) wild dogs run "up" toward it; and it is opposite, although reversed in orientation to, the third wild dog below the saucer.

If the viewer discovers two gazelles on the reverse and therefore asks about two gazelles on the obverse, so, then, does a similar question arise about the wild dogs. There are three on the obverse. Should there not be a fourth, completing the parallel with two gazelles and two attackers, on the reverse? The construction of the parallel takes place in and through the time of viewing as the viewer moves back and forth between the two sides of the palette. The

fourth wild dog appears more or less where the viewer comes to expect to find it—namely, in the complex scene of nature hunting presented on the top of the reverse opposite the three dogs on the obverse. Here, next to the hind foot of the framing wild canine on the left edge of the palette, the dog, running right to left (or reverse to obverse) seems to halt and turn back on itself. It has sighted an antelope running toward it, but this creature, although running right to left, also turns its head and thus does not see its danger.

As the viewer is led from the top side of the obverse—with ostrich, gazelle, serpopards, and wild dogs—to consider the top of the reverse, he or she encounters there, just below the two lion-and-gazelle pairs, another face-to-face pair of attacker and supposed victim: a serpopard bites the leg of an oryx. Despite the precedent of the topmost passage of depiction on the other side, containing the two serpopards, and the passage of depiction above it on its own side, containing the pairs of lions and gazelles, this serpopard does not have a twin. Instead it is paired with its prey—who does not, in the end, turn out necessarily to be a victim. (The encounter may not be fatal, since the oryx is bitten on the leg, rather than face or throat, and actually reappears on the obverse, outside the framed portion of the top of the image, where its "twin" is pursued by, but may still escape, the hunter's dogs.) This portion of the image—the entire passage on the top two-thirds of the reverse, framed by the wild dogs at the sides—takes the viewer, from top to bottom, through a complex sequence depicting the relations of hunter and hunted within the scene of nature hunting.

All in all, six pairs of hunter and victim are depicted in the framed portion of the image on the reverse side of the palette. Like those just described—two lions attacking two gazelles and the serpopard biting an oryx—the next two creatures see one another directly. But, for the first time in the sequence, they are presented almost as a standard carnivores-and-prey pair: the leopard follows the hartebeest, who turns around to look at its attacker.

This arrangement is reversed immediately below, where an antelope evades the preceding conflict—in bird's-eye view, it is running "out" of the melee—but looks back over itself at the hartebeest, which confronts own death.

It is the antelope's possible attacker, the wild dog to the left, who turns back over itself toward its possible victim. But whether the wild dog will necessarily attack and kill the antelope is not clear, for the wild dog's "twin" in the same compositional position on the obverse is running right to left, or "away from" the melee. As the viewer comes to understand it, then, it is as if the wild dog on the reverse pauses—poised between flight and attack—and then continues to run away. Should the unseeing antelope therefore escape death at this moment, in direct contrast with the hartebeest aware of (but unable to prevent) its fate immediately above, the viewer must ask about the conclusion. If the wild dog continues to run away, then what happens to the antelope? It would be an ellipsis in the pictorial text of the reverse top portion, as well as an anomaly in the entire image, for the matter of the antelope's fate to be left hanging. Like the oryx escaping the serpopard, the antelope is depicted as attacked in the obverse bottom portion of the image—but by a different enemy altogether. Here it turns its head to look in the direction from which the danger had been coming on the reverse—that is, to the "left" and therefore in the direction of the place where the wild dog had been standing. But its action cannot now avail it, for the wild dog is replaced by the hunter's dogs, which attack from both sides.

The six-part sequence of scenes of nature hunting on the reverse top portion of the image is completed, finally, by the pair, at the bottom of the framed portion of the image on the reverse, depicting a winged griffin attacking a fleeing bull ox.^[3] This motif is the only absolutely standard presentation of the carnivores-and-prey formula (compare the attacking griffin on the Gebel el-Tarif knife handle [Fig. 20]), with an attacker placed behind and seeing a quarry that is unaware of danger. The viewer encounters this pair after the entire sequence, from top to bottom, has presented other possibilities: a face-to-face encounter (lions and gazelles), a face-to-face encounter with attacking beast of prey but a possible escape and postponed death for the victim (serpopard and oryx), an attack from behind with the victim's death despite seeing its danger (leopard and hartebeest), and an attack from the front unseen by the victim

whose death is nonetheless deferred (wild dog and antelope). We can put these results in tabular fashion as follows:

The summary shows it to be a fact of the world of nature hunting—as presented in the top portion of the image spread across obverse and reverse sides—that the victim may or may not be killed if approached from in front, when it must see what it confronts, and may or may not be killed if attacked from behind when it can still see its attacker, but is always killed if attacked from behind when failing to see the attacker. The presentation begins with the most ambiguous aspect of this "rule," the serpopards and gazelle on the top of the obverse, and ends with the most unambiguous, the griffin attacking a bull ox, toward the bottom of the reverse.

Juxtaposed with the entire statement in the portion of the image framed by the wild dogs, enclosing the scene of nature hunting, a revised and reversed statement appears in the portion of the image outside the frame in the bottom one-third of the palette on both sides. Here the victim, the antelope at the bottom of the obverse, is killed even when it can see its danger—precisely because, when hunted by man, it can never see the whole of that danger; the threat, like its very depiction on the palette itself, is always partly on the other side of what is seen .

Although the top portion of the image establishes a definitive statement—it relates a narrative of nature hunting—questions about or ellipses within it are progressively encountered by the viewer. For example, the viewer must wonder about the twin of the top gazelle on the obverse or about the fate of the oryx and the antelope in the middle part of the sequence on the reverse. These gaps are completed in the bottom portion of the image as spread across the obverse and reverse sides. Here the "twin" gazelle, the oryx, and the antelope reappear in a new context: the masked human hunter and his surrogates are introduced into the scene of nature hunting in a way that explicitly recounts the difference between nature hunting per se and nature hunted by man, a difference not specifically observed on the Ostrich Palette (Fig. 25). The full image, then, is an intertwined juxtaposition of top (framed) portion and bottom (unframed) portion, each completing itself on the obverse and reverse sides of the palette while relating to the other above or below it.

Unlike the top portion of the image, which works from obverse to reverse, the bottom implies a chronology that works from reverse to obverse. The hunter's dogs, on the obverse, must have been released by the hunter, on the reverse, even if the viewer's experience—technically speaking, the story of the fabula in the text of the image (see Appendix)—may go initially in the other direction. At the chronological and causal beginning of the narrative, the bottom zone of the reverse of the palette and perhaps the last zone to be viewed, a giraffe stands between the masked human hunter and a bearded ibex running toward them. According to the bird's-eye view of the composition, the ibex runs "into" the giraffe by moving "down" from the melee of the scene of nature hunting presented in the framed portion of the image above it. The neck, face, and ears of the giraffe are treated so as to be formally similar to the flute, face, and ears of the human hunter; it stands between the hunter, placed at the edge of the scene and apparently advancing into it, and the scene of nature hunting itself. The giraffe, then, is in effect the mask of the masked hunter.

Running toward the lure of the hunter's music—if this is indeed the sense of the group depicted here—the ibex sees the giraffe rather than its real enemy. The danger of the hunter for the ibex and the other animals lies outside the scene altogether, as is made clear on the other side of the palette. Here, at

exactly this level in the composition of the image (the top of the bottom, unframed zone), one of the hunter's mastiffs, the beast that will actually perform the kill, pursues the group of three animals (gazelle, ibex, and oryx) carried around the palette by the fourth (the ibex's twin) coming up to face the masked hunter standing in the very position of his dog on the other side.

From the vantage point of the ibex at the bottom of the reverse side, the danger in front of it is masked in two senses. First, the hunter is masked from the ibex's direct view by the giraffe. Second, even if the ibex were to see him as he really appears, it would only, still, encounter a mask—namely, the jackal-masked hunter piping a flute rather than preparing to kill. The ibex's real danger, the hunter's dogs, remains unseen and behind it on the "other side" of both hunter and ibex. Thus, in this compositional position on the obverse, the ibex and the oryx are depicted, from the vantage point of the reverse, as turned around and running in the other direction, away from the hunter's dogs—a decision fraught with danger, for, in this direction of viewing the image, the oryx will encounter the serpopard.

On the bottom of the obverse, the gazelle to the left and behind the ibex is being killed by the hunting dog; the oryx to the right and in front of it will be attacked by the serpopard in the immediate "future" on the reverse of the palette. The bearded ibex itself is not shown being directly attacked and killed. As the viewer inspects the two sides, the image therefore suggests the hopeless position of the ibex; avoiding a literal depiction of the ibex's destruction, it leaves the obvious implication for the viewer to construct. But this condition of nature when hunted by the human hunter is represented metaphorically at the bottom of the unframed zone on the obverse. Here a hapless gazelle, beset by two of the hunter's dogs, twin of the gazelle with which the scene begins at the top, is the only creature in the entire image spread across the two sides of the palette to be attacked from both sides. Wholly unlike the two gazelles attacked by two lions at the top of the reverse, the gazelle at the bottom of the obverse is depicted as being unable to see either of its two enemies, however it tries. Its head is turned back from the dog coming up toward its "right" front, but it is also turned to its own "right" back rather than to the "left" back where the other dog appears. To represent the confusion in its hopeless situation—

figuring the entire scene of representation, including the viewer's interpretations in producing it—the gazelle's horns are depicted in a surreal overlapping twist, perhaps the most important key to the cipher of the image.

This metaphor of the hunter's dog attacking the twisted gazelle synthesizes for the viewer the image as a referential whole. In particular, it explains the forward mask of the masked hunter in the same place on the other side—namely, the peaceful giraffe that the ibex does see depicted as disguising a destructive blow it cannot see. But the human hunter himself, entering the scene with his dogs, remains masked. He is not directly shown to wield the blow—in this case, to set his dogs upon the animals—as the master might be shown in canonical Egyptian art. Instead his decisive act must take place outside or, as it were, between the sides of the image. Like the "licking" serpopards and the "kissing" lions in the top portion of the image, his depicted relation with his prey is decidedly ambiguous: he plays a flute, which could be both entertainment and lure. The truth of his action is unfolded only progressively in the relating of the full narrative. Unlike the masked hunter depicted on the Ostrich Palette (Fig. 25), coming up directly behind his prey as if he were actually an animal, the human hunter on the Oxford Palette is behind his dogs. As in all instances of nature hunting, it is the dogs—or other carnivores—who actually kill the prey; thus the human hunter pursuing the animal hunters hunting their prey, like the jackal whose mask he wears, is represented as if he were a scavenger who only discovers the corpses of the animals. But of course the viewer understands that the human hunter is not actually a scavenger, for he owns and trains the dogs and sets them on their prey; his jackal form is the mask of his real identity as deadly hunter in front of his prey.

As we will see in Chapter 5, the image on the Oxford Palette will be replicated and revised. For example, the Hunter's Palette (Fig. 28) will show the nonmasked human hunters pitted directly against their prey. Here the image is on one side of the palette—and the hunter surrounding the victim will come into the literal depiction in another fashion. On the Battlefield Palette (Fig. 33) the scavenger will have a distinctively different role in the pictorial metaphorics and narrative, but, as on the Oxford Palette, a specific relationship between the enemy's sight and the ruler's representations will be sustained.^[4]

In my description of the Oxford Palette, I attempt to follow the structure and sequence of the narrative—its symbolic content or meaning is still obscure—as it is distributed across the surface of the object, in active manipulation, and through the time of its viewing. Like the two sides of the Carnarvon handle (Fig. 23), the two sides of the Oxford Palette are related closely as passages of depiction in a single complex image (Fig. 27). In effect, the top of the obverse masks the blow at the top of the reverse, whereas the bottom of the reverse masks the blow at the bottom of the obverse; but even the blows depicted are not presented fully or literally. Figured by the jumping, flying ostrich and the gazelle's plaited horns, the palette constructs a sophisticated double twist, tied together from top to bottom and from side to side. Compared with the decorated knife handles and comb and the Ostrich Palette, the place from which the human hunter advances is shown on the Oxford Palette in some of its power and danger: it is advanced further into the depicted scene. Indeed, compared with the Brooklyn and Carnarvon knife handles or the Ostrich Palette, it is only superficially that the masked hunter seems to be at the edge of the scene. In the double twist, wrapping around, and folding over of the image as it is spread across the palette in the time of the viewing, the masked hunter becomes its crucial central node—even as his blow remains literally undepicted on the other side of his mask.

5—
Failing to See on Contested Ground

The Hunter's Palette (Fig. 28) was one of the first carved palettes to be discovered. Since it looked substantially different from canonical Egyptian images, it was thought initially to have a Mesopotamian origin. On typological grounds we can reasonably assign it to the Nagada IIc/d or IIIa period, roughly contemporary with the Oxford Palette (Fig. 26). In terms of the history of the chain of replications, however, it makes sense to take it as a later work: its metaphorics seem to require the preexistence of works like the Ostrich and Oxford Palettes.

Two Grounds

The image depicts a group of human hunters, not literally masked but costumed in garb that includes adornments like wild animals' tails hanging from their belts and ostrich feathers in their hair, setting out to hunt desert animals—a deer, an antelope, an ostrich, and so forth (see generally Tefnin 1979 for a penetrating analysis). Several of these creatures appear in ways that might have reminded a viewer of works like the Ostrich and Oxford Palettes (Figs. 25, 26)—for example, an ostrich runs along the ground, lifting its wings before launching into its peculiar hopping flight. The animals all have a distinctive place in the novel metaphorics and narrative of this ingenious image, however, and the hunters themselves are literal representations of the masked human hunters who appear completely outside or only at the edges of the natural world earlier in the chain of replications.

Evidently by stumbling across the den where its young cub resides, the hunters provoke and do battle with a desert lion. On the Oxford Palette two

Fig. 28.
Hunter's Palette: carved schist cosmetic palette, late predynastic (Nagada III). 
Photograph courtesy British Museum, London; drawing from Smith 1949.

lions at the top of the reverse attack two gazelles in a scene of nature hunting; the animal-masked hunter observes from the sidelines at the bottom of the same side of the image. On the Hunter's Palette the hunters march from the direction of the bottom of the palette, where they first enter the scene, to the top, where the lion turns away from its expected victims—such as an antelope positioned just below, at right angles to it—to face these human antagonists.

Not masked but rather bedecked in animal finery, the hunters march in two files along the two long, left and right sides of the palette. This arrangement can be seen both as a prototype of the later bands of figures in canonical representation (Figs. 1, 2) and as an echo of the rows of animals on earlier or contemporary decorated knife handles and combs. But there are two important revisions in the new image on the Hunter's Palette. First, whereas the rows of animals depicted in earlier images had been interrupted by anomalous beasts and shadowed by the human presence outside the scene, here the beast, the fierce lion, interrupts rows of human beings depicted within the image. Second, the two long rows of hunters replace the wild dogs announcing the theme and framing the narrative on the Oxford Palette and related works (for example, Fischer 1958; Asselberghs 1961: figs. 129–30, 133–34, 138, 142[?], 170); the wild dogs of the earlier works had masked the hunters who appear here, without masks, in the wild dogs' place. In both of these replicatory revisions, the hunter's mask has been advanced somewhat further into the scene.

The two files of hunters are separated from one another by various other figures. Some of the right-hand hunters carry small round shields, whereas none of the left-hand hunters is so equipped (see Tefnin 1979: 224–25). Both right and left files of hunters carry the same standards, however, which may imply that the groups share a common totem, moiety, or territorial affiliation. Since the minor differences between the two files may be fortuitous, I do not assume that they represent two social groups, however defined, although the finding would not be incompatible with my interpretation, only slightly rewritten. As we will see, the two files do play different parts in the narrative.

Since the hunters surround the lion on all sides, the hunter's place on the encircling outsides or sidelines of nature in the animal-row images conforming to the formulas and on the Ostrich and Oxford Palettes (Figs. 25, 26) has been advanced on the Hunter's Palette literally to enclose nature. The hunters frame

the scene; and in the context of replications like the Oxford Palette, the topmost group of hunter and lion, to be considered in more detail momentarily, can be understood to announce its central theme. Like hunting scenes produced in the later canonical art of the Old Kingdom, several centuries later, where animals are often depicted as driven into fenced enclosures to be hunted for sport by a courtier and his dogs (for example, Vandier 1964: 787–815; Davis 1989: figs. 4.6a, 4.9a, and compare figs. 4.15a–b, 4.16, 4–17a), the files of hunters on the Hunter's Palette could be driving game. No fences are actually depicted in the scene, however, and the figure of a sporting master is not singled out as he would be in canonical hunt scenes; furthermore, the dogs in the central band of the design on the Hunter's Palette are wild rather than domesticated. (Inasmuch as the Hunter's Palette replaces the Oxford Palette hunter, masked by dogs who actually perform the kill, with huntsmen who carry out their own killing with spears, throwsticks, maces, and bows-and-arrows, the hunting mastiffs need no longer be included in the scene of the hunt.) On balance, then, the maker of the Hunter's Palette probably meant to depict a lion and cub surprised in the wild open landscape—indicated by the row of fleeing wild animals pursued by a wild dog—in the course of a hunting expedition rather than the quasi-ritual killing of a previously captured lion and cub in ceremonial combat or sport. Instead of framing an enclosed stretch of ground, the hunters must be advancing through the brush in two files or—depending on how we decipher the composition—trapping animals between two files of men coming together from two directions.

The round-topped building (a "shrine"?) and "double-bull" signs (Fig. 30) at the top right of the image suggest—if we are to take them, like later hieroglyphs, as informational labels or captions—that the entire episode bears some relation to a particular, perhaps sacred building or locality (Vandier 1952: 577–78). The implication is not necessarily that the hunt takes place within this building or locality; instead, the hunters may be affiliated with it, or, among other alternatives, might have dedicated their activities or trophies to it. Another reading could conclude that the signs label the temple to which a palette of this kind was supposedly dedicated. It hardly matters for our general analysis of the narrative structure exactly how we decide these questions. What-

ever their other representational functions might have been, the signs have a role to play in the internal metaphorics and narrative of the image itself.

With the human hunters at the edges of the palette—like the anomalous animals at the ends of rows in the animal-rows formula—the hunt that takes place within nature occupies the central band of space remaining on the palette. In this central passage the scene of nature hunting is presented as less visible than the details around the periphery; it is squeezed between the human hunters and disposed on either side of the cosmetic saucer. Here three jackals are pressed farther into the center of the image from the former place of such scavenging creatures (for example, the wild dogs on the Oxford Palette; see Chapter 4) by the arrival of the human hunters from their former place at the edges of and outside the depicted scene altogether. In keeping with the core formula in which they are often presented, the jackals attack a group of non-carnivores—a wild deer or stag, two antelopes, a gazelle, an ostrich, and a hare—that flee from and either see or fail to see their pursuers. Like the top and bottom and side-by-side zones of the Oxford Palette, the central zone on the Hunter's Palette, containing the scene of nature hunting, is specifically related to the side zones, folding around the top of the image, containing the scene of confrontation between nature and human hunters.

The literal mimesis of nature by the hunters on the Ostrich and Oxford Palettes retreats literally to the sidelines on the Hunter's Palette—to a marginal place in the space or to a former place in the time of the depicted scene; the episode depicted on the Ostrich Palette, the stalking and capture of an ostrich, presumably to acquire its plumes, has clearly already occurred for the hunters on the Hunter's Palette, and the outcome of that episode now appears in their hair. In the central passage of the image, however, the depiction of three jackals and an ostrich taking flight recalls that hunters and animals are intertwined in a continuous cycle of contests, with the hunter's blow (never quite depicted) always decisive.

The animal masks depicted as actually used by the hunters on the Ostrich and Oxford Palettes are removed from the faces of the hunters on the Hunter's Palette. The real site and force of the hunter's blow is advanced into the depicted scene: in their outstretched arms the hunters carry bows, spears, single- or

double-headed maces, and curved throwsticks, and some wear shields or game bags slung over their shoulders. Whatever the symbolic meaning of the insignia (Vandier [1952: 578] and Baumgartel [1960: 98] wish to take them literally, if anachronistically, as territorial emblems of early provinces of Egypt), the standards carried by several of the hunters seem to identify them as heavily armed members of a ruler's retinue. They have evidently marched from their village, as one or more groups, into the wilderness, entering nature not—as the image directly reminds us—for the first time.

With three exceptions, each figure of a hunter is constructed in the manner that will later become canonical in Egyptian art (see Fig. 2): the head and legs of the human body are displayed in profile, while the torso is displayed frontally. But we should not look at this technique of rendition only in the terms of later standards of canonical representation; instead we should observe that the twisting movement on the Oxford Palette in and through which the presence of the hunter appears there has been brought, on the Hunter's Palette, into the depicted aspect of the hunters' figures. Moreover, the viewer does not flip the palette around itself, as with the Oxford Palette, to understand the rendition of figures and the composition of the scene—but rather turns it on a central pivot.

The canonical construction of the aspect of the human figure, in which component parts of the body are pivoted around a central vertical median (Iversen 1975: 33–37; Davis 1989: 28), can thus be seen as a replication of a certain technique of rendition that serves in late prehistoric image making as a metaphor for the way the hunter's identity is present in or introduced into the natural and the human worlds. Although I do not explore the topic further in this book, the entire pictorial mechanism of late prehistoric representation is advanced into the depicted body of an individual human figure drawn according to the canonical rules of components, proportions, and aspects (Davis 1989: 10–29). In canonical representation, then, the human body becomes the site of late prehistoric metaphors for and narratives of the masking of the ruler's blow. It is for this reason, I suspect, that canonical Egyptian portrayals do not depict the particular physiognomic, ethnic, and sexual characteristics of the individual body, but instead develop stereotypes or symbolic markers for them.

The body of canonical image making is not the physical body, the body with passions, appetites, and desires, but the corporate or social body, the body politic, the body under rule, the body that traces or indexes the order of the state legitimated in its transcendental theodicies—embodying the metaphor for and narrative of the institutions of the hunter's or ruler's mastery of nature (see also Davis 1989: 36–37, 221–24).^[1]

Back and Forth, up and down

Although the action of the hunt depicted on the Hunter's Palette (Fig. 29) is somewhat obscure, it appears to show the two files of hunters as hunting the noncarnivores being attacked by the three jackals—creatures that disguise the approaching hunters from the other animals, and serve literally, then, as the hunters' forward masks (recalling the jackal-masked human hunter on the Oxford Palette [Fig. 26]). Only one animal, the antelope directly below the cosmetic saucer, sees the human hunters.

During this action one unfortunate hunter, carrying a bow and a mace, stumbles across the lion cub's hideaway. As is indicated by the position of the cub at the extreme right edge of the scene, the den was apparently across the hunter's line of march rather than in the area, depicted in the central zone, to the hunter's "left." The cub's sire leaps ferociously to defend his offspring and surprises the hunter (see Fig. 29, Group 1). Since there are ten hunters in the left row, a "full complement," and only seven in the right (restoring two in the missing fragment), it would seem that the fallen hunter comes from the shorter, right-hand row. Therefore the lion is breaking into that row from its right—that is, altogether from outside the scene. He is constituted, then, like an anomalous beast entering a row of identical animals or like a carnivore attacking its prey in the basic formulaic images earlier in the chain of replications. From the outside he pounces into the center of the scene. But the hunters, although they are like a row of identical animals in the earlier formula, are not the same as animals; and the lion's fate will be, precisely, that of an anomaly to be suppressed by the order in which the hunters are arrayed and for which they stand. The narrative on the palette unfolds this moral.

Fig. 29.
Principal groups of figures on the Hunter's Palette. After Smith 1949.

The doomed bowman is depicted as completely detached from the two regular files of his companion hunters. It is implied in the orientation of his baseline in relation to the other figures that he lies spread-eagled beneath the lion, whose decisive blow has just taken place. Like the dying lion at the bottom of the palette (for the viewer initially holding the palette upright, with long axis vertical), the doomed hunter's grounding is utterly torn from him: neither on his companions' ground nor in the lions', he is inserted "upside down" in the right angle formed by the intersection of the two groups.

The fallen bowman managed to shoot the arrow from his bow at the lion as it approached and pounced. One of his companions, the third hunter back in the right-hand file, apparently did the same; although his figure is now partly missing, he appears to run forward—toward the scene of confrontation—holding up his bow. At any rate the lion is depicted as having been hit by two arrows before mauling the surprised hunter.

At the top left curve of the palette, another hunter attempts to rescue or avenge his fallen companion. Leaning forward to fit an arrow to his bow, the approaching hunter reveals his right shoulder in profile and the quiver on his back, details not rendered for any other hunters in this image. Consistent with his status as the hunter who successfully kills the lion, he exposes more of the real site and force of his blow than the others—a force precisely to be found behind his back, on the other side of the usual side of representation. Furthermore, in confronting the lion the approaching hunter must strike out on his own ground, "twisting away" from his companions: on first glance he appears to turn sharply right out of the left-hand file of hunters. In fact he is now on the lions' ground; his baseline runs parallel with lion and cub. Evidently, then, the temporal and spatial unity formed by the four figures at the top of the palette—cub, lion, surprised hunter, and approaching hunter—is expressed in their disposition along the same baseline.

Supplanting the lion's natural victims presented in the carnivores-and-prey formula of other images (such as gazelles and lions on the Oxford Palette [Fig. 26]), the approaching hunter is the lion's antagonist in a confrontation (in time) following upon and (in space) replacing the first antagonists confronting each other—namely, the lion cub and the surprised hunter. This narrative sequence

breaks apart and interleaves the separate, symmetrical pairs of carnivores and prey of earlier images (like the two pairs of lions and gazelles on the Oxford Palette; Zone 1, Fig. 27). Viewing from right to left, the earlier and later confrontations appear as a continuity from before to after—namely (1), the lion cub (2) is defended by the lion (3), who attacks a hunter (4) rescued by the approaching bowman. But viewing from left to right, the story is not so much reviewed as understood to move on to the next stage of the action: after entering the lions' ground from his erstwhile place among the hunters (4), the rescuer shoots his arrow (3) over the body of the fallen hunter (2) into the head of the lion (1) attempting to defend its cub. Placed immediately above the four figures, the shrine with closed door, "to be opened and entered," and the joined foreparts of two wild bull oxen or buffalo "looking both ways" (Fig. 30) seem to indicate how this two-paired group is to be viewed—namely, from right to left (one "door") and from left to right (the other "door"), back and forth, producing the characteristic "mobility of internal elements" that establishes a pictorial narrative's "summation in time" (Lotman 1975; for discussion of this understanding of pictorial narrative, see the Appendix).^[2]

Although stooping visibly, the hunter approaching to rescue his fallen companion is the tallest figure in the composition, like the ruler in the Decorated Tomb painting (Fig. 5) or like Narmer on the Narmer Palette (Fig. 38). He is also the topmost and the only "upright" human figure. It is he who is depicted about to shoot his arrow at the lion, to strike the decisive blow. Not counting the fallen hunter, the two huntsmen closest to him carry Falcon and Feather standards (royal standards of the West, if interpreted anachronistically, but otherwise simply emblems of the group's social affiliation).^[3] On all these grounds, the topmost hunter is almost certainly the leader, the "ruler" of the group. "The rather clumsy way in which the bowman has been represented has often been pointed out" (Baumgartel 1960: 97); like the bobbing, masked hunters on the Ostrich and Oxford Palettes (Figs. 25, 26), he ducks, twists, and insinuates himself into the proper position rather than marching straight into things. His stooping profile aspect, ignored in all existing commentaries, seems a literal depiction of his sideways entry into the lions' ground.

By contrast, with two exceptions all the hunters unengaged with the lion march fully upright on a more or less straight baseline along the side edges of

Fig. 30
The "shrine" and "double bull" on the Hunter's Palette. After Smith 1949.

the palette. They fail to see the lion cub and the danger posed by the lion, for they are oriented at right angles to the scene of confrontation. They are also at right angles to the viewer's initial view of the image (holding the palette vertical). To bring them upright the viewer does not just look right or left but must literally turn the palette downward to the right or the left; in doing so the viewer will, like the hunters, fail to see the decisive blows of the principal antagonists, the lion (who has already struck) and the ruler (who is about to strike). Turned to the right or to the left with its long axis horizontal, the image depicts what is before, what is after, or what is simultaneous with the scene of confrontation—and moving it to the upright will be to see what the hunters on the edges, moving into the scene of nature, could not themselves see awaiting them there. The viewer's position, then, shifts depending on how the palette is held and the image spread across it is viewed: the viewer oscillates from being like the ruler to being like the other hunters and vice versa.

Behind the Back

So far we have dealt only with the first group in the image (Fig. 29, Group 1) and its relation to the two files of hunters along the edges. At first glance the ruler appears to turn sharply "right" out of the left-hand file of hunters to confront the lion directly, face-to-face. The apparently straight lines of his straight arrows do not fly straight ahead, however, as they would if hunter stood directly before lion. Rather, they twist or angle in a visibly downward, rightward direction into the lion's ground. Their trajectory figures, or prefigures, the viewer's own next move—namely, to turn the palette down from the upright (long axis vertical) to the right (bringing the long axis to the horizontal). Despite appearances, then, the ruler—perhaps because he is not masked by the features of an animal—is on the lion's ground only in an oblique position. In fact, although the ruler appears to confront the lion face-to-face, the fallen hunter masks him from the lion, who is actually killed from behind. Facing to the "right" at the topmost position in the composition, the ruler is not the eleventh person in the left-hand row of hunters but rather the eighth and foremost in the right-hand row, almost matching the full complement of ten in the other file.^[4]

The scene of representation on the Hunter's Palette incorporates on its one and only side the twisting flip required to interpret the two-sided image on the Oxford Palette (Fig. 27). The lion breaks into the right-hand file of hunters from its "right"—that is, from outside the scene altogether—and the ruler, in turn, backs out of the right-hand file so as not to turn and face the lion but to go around behind it. The ruler's arrow flies, as it were, around the palette, taking an elliptical path through a "wild" space unrepresented in the image before hitting the target decisively; in passing out of and back into representation, the arrow masks the depiction of the blow literally falling (see Fig. 32). The ruler's motion and the act of killing the lion are unfolded by moving the palette: when the viewer twists the upright palette (long axis vertical) down to the right, tracking the downward trajectory of the ruler's arrow and moving into the "next" unit of the narrative, the ruler appears to move downward to be brought up, as it were, behind the lion at the "bottom" (now left) of the palette. Here the beast is shown "upside down," placed on the same long horizontal axis as the files of hunters but facing away from them—that is, not seeing

the ruler "behind" him (see Fig. 29, Group 2). The two arrows fired earlier at the lion's head, presumably one by the fallen hunter and one by his running companion in the right-hand file (third hunter back), are now reinforced by three more arrows shot at and into the back of the lion's head. It is probably not the ruler who fires these three arrows; the sixth hunter back in the left-hand file—who could also be "behind" the lion—is depicted as entering the scene carrying precisely three arrows. But making six shots in all, a fourth and final arrow fired during the "rescue," the only fully straight shot along the line of the ground and evidently the ruler's decisive shot, plunges into the lion's anal-genital region.^[5]

There could hardly be a more explicit depiction of the lion's unfortunate fate. His natural ability to strike his own blow—in the top scene (Fig. 29, Group 1) his powerful legs and claws and open jaws appear to be a match for the hunters' weapons—was, it turns out as the viewer unfolds the narrative, pointed in utterly the wrong direction. As on the Ostrich and Oxford Palettes (Figs. 25, 26), the hunter's mastery of nature takes place where nature cannot see it, moving up from the sidelines, "behind its back," to close a deadly circle. When the hunter enters the scene, even the clearest sightlines and straightest shots take a deadly turn.

A Natural Interlude

So far we have seen the story of the lion's death in two episodes—first, the lion's attack on the surprised hunter to defend its cub and, second, the ruler's corresponding attack on the lion to rescue or avenge his companion. Although the story follows the "natural" chronological order of events, the pictorial text presents them elliptically. The ruler's blow is not directly depicted; instead, in the first unit of story-text (Fig. 29, Group 1), the ruler prepares to strike his blow (fits an arrow to his bow) and in the second unit of story-text (at the bottom of the palette), the blow has fallen (the lion has been killed). As in other images in the chain of' replications, the spatiotemporal and causal relations of the narrative, its logic of before and after, can be rearranged in the sequence of presentation into alternative, complex combinations at the very same time as the pictorial text offers its oblique representation of the episodes in the sequence.

On the Hunter's Palette the relation between the actual chronological and causal logic of events and their presentation in the composition of the image is not entirely clear. The other zones of the image (Groups 2, 3, and 4) do not seem to depict additional episodes in the death of the lion (Groups 1 and 2), although they might be regarded as variously before or after that main event. Rather, they repeat or rehearse its general theme in a context that depicts different actors and events, like a second narrative running alongside the main story. Whether or not they are events literally continuous with or before or after the lion's death, they have been folded into the pictorial text of the story of the lion's death and must be viewed along with it as a kind of metaphorical shadow.

It is likely that the Hunter's Palette relates a single large-scale narrative of various events of the hunting expedition as a whole, broken into two "chapters" in the presentation. These chapters relate either sequential or simultaneous and—a more interesting point—metaphorically associated episodes (the lion's death in Groups 1 and 2, and the wild animals' hunt in Groups 2, 3, and 4), each chapter containing two scenes (therefore four scenes in all) related to the story in such a way that one of the most crucial events in the full narrative, although presented in its preconditions and results, drops into ellipsis. This structure seems to be a revising replication of the structure to be observed on the Oxford Palette (Fig. 27), where each of the two sides includes two zones (top and bottom), or four textual zones in all. Turning now to the other zones of the image on the Hunter's Palette, we find (in Group 2) an antelope in the row of animals on the same baseline as the defeated lion turning its head to see a confrontation—a jackal pouncing on its prey—it successfully manages to evade. The jackal and the fleeing deer or stag (which does not see its enemy) form a typical carnivore-and-prey pair, framed on the "left" by the doomed lion, defeated by his own flawed sight, and on the right by the escaping antelope, saved by proper sight.

For an image maker or viewer who had seen the Oxford Palette (Fig. 26) or similar images, the jackal could presumably be a deliberate metaphor for the human hunter who wore its mask in the earlier replication. But even if this equivalence was not intended, a direct association between the story of the lion's death (Groups 1 and 2) and the wild animals' hunt (Groups 2, 3, and 4) is

clearly made, for Group 2 also presents the jackal's attack, repeated, in turn, in other ways in Groups 3 and 4. Thus the viewer can unfold the entire image as a set of possible metaphorical relations that function independently as a continuous narrative.

The center of the image, around the cosmetic saucer, includes two extraordinary groups (Groups 3 and 4) tying together the principal top/bottom and left/right axes of the composition. In the pictorial metaphorics they juxtapose the hunters' mastered ground and the ground of nature, containing its own contests but contested by the hunters.

Immediately below the cosmetic saucer (Group 3), a gazelle—on the Oxford Palette the prey of the lions—runs in the same direction as or even "toward" the left-hand file of hunters; the gazelle looks back over its shoulder at a hunter there about to throw a lasso at the melee of animals. Running past the gazelle in the other direction, a jackal directs a backward glance at it. Like the bowmen and lions at the top of the palette (Group 1), this group (Group 3) depicts the moment just before the blow.

Immediately above the saucer and below the bowmen and lions (Group 4) is the moment just after the blow: the scene below the saucer (Group 3) is literally flipped over from bottom to top and left to right and advanced a moment in time beyond the decisive strike, like the depiction of the defeated lion at the bottom of the palette (Group 2) in relation to the figure of the ruler at the top (Group 1). The antelope is caught by the horns with a lasso thrown by a hunter in the right-hand file who now leans forward to pull in his victim. Unlike the fleeing gazelle, the antelope above the saucer turns its head around to look back only to mistake its real pursuer, for it is depicted as turning back to see the jackal instead; the jackal is now flipped over and reversed in direction running toward and along the "right" side of the antelope about to pounce on it. Precisely as it did on the Oxford Palette, the jackal running along the ground line of the lasso masks the deadly power of the antelope's real enemy—the hunter—who, wearing the tail of the carnivore, comes up behind the antelope as it looks at the jackal.

Groups 2, 3, and 4, then, present a continuous story ranged along the baseline established by turning the palette down to the right and bringing its

long axis to the horizontal. First (Group 2) the jackal chases the stag while the antelope, turning its head, flees ahead of it. Next (Group 3) a gazelle in the melee of beasts—fleeing their pursuers but at the same time encountering the files of human hunters moving through the brush—sees the hunter in the left file preparing to throw his lasso at the animals. Running past the gazelle in the other direction, "toward" the antelope in Group 2, a "twin" of the jackal—that is, the "same" jackal—from Group 1 in turn sights the gazelle. Finally (Group 4) a "twin" of the antelope—that is, the "same" antelope—from Group 2 turns its head around to the other side to see the jackal coming toward it, as depicted in Group 3, but in so doing fails to see the hunter, in the right file, throwing the lasso that catches it (Group 4). The gazelle presumably escapes—after all, the gazelle's antagonist in the earlier replication, the lion of the Oxford Palette, is otherwise occupied—for in the terms of the narrative image, the action from Groups 2 through 3 and 4 passes it by.

Group 3 is separated from Group 2 by the figure of an ostrich running along the ground and flapping its wings; Group 4 is separated from Group 3 by the figure of a bounding hare. In the literal depiction these creatures are presumably part of the melee of wild animals being pursued by the jackal(s) and human hunters. But like the shrine and double bull at the top of the image, they also indicate, for these portions of the image, how the narrative image should be read—that is, at least for this central passage (as it is brought into view by having turned the palette down to the right), not by going back and forth or right and left but instead by "flying" or "hopping" to the next, "later" piece of ground, from Group 2 to Group 3 to Group 4 as the stages of a continuous action. Certainly the ostrich in Group 3, if the image maker or viewer had seen the Oxford Palette (Fig. 26), would have had an established textual function of this kind, and the hare, placed in the same position in the next group (Group 4), could be construed likewise. Once the viewer comes to the end of the central passage of depiction along the long horizontal axis, the top scene (Group 1) is reencountered; the first instructions of the "double door" and bull—go back and forth, right and now, presumably, left—apply again.

In the entire central passage of the image (Groups 2, 3, and 4), the hunter's blow is not depicted. It falls between Groups 3 and 4, just as occurs between

Groups 1 and 2, where the cosmetic saucer intervenes in the composition of the image as a whole—that is, in a wild, unrepresented place where the human owner of the palette reaches down to mix eye paint and mask his or her own face before entering the scene. It does not matter here whether the palette was used to prepare for an actual or a ritual, ceremonial hunt, to recall or commemorate one, or to depict something else altogether for which "hunting" and "masking" are the appropriate metaphors. Its pictorial mechanics and internal metaphorics dictate in the central passage of depiction the metaphor of masking the blow: the huntsman's lasso takes the same angling turn, coming around and out from behind what masks it, as the ruler's deadly arrow fired at the lion.

Two Turns, Looking One Way and the Other

The figures of the hunter lassoing the antelope and the figures of the lion mauling the fallen hunter are clearly meant to be viewed in relation to each other. They may be either successive or simultaneous episodes in the narrative chronology, perhaps arranged as above and below although the pictorial text—the site of metaphorical linkages—requires that they be viewed together. Both groups appear in the portion of the image above the cosmetic saucer, the zone the viewer is first likely to inspect and ordinarily where late prehistoric images of this sort begin to unfold. Moreover, it seems to be no accident that the viewer cannot quite distinguish the fallen hunter from the doomed antelope, their arms and legs tangled together and both oriented "upside down" in the scene to which they belong: both have been destroyed by what they did not see. In fact the fallen hunter in Group 1, the viewer might now conclude after having surveyed the central passage of depiction (Groups 2, 3, and 4), stumbled across the lion cub and provoked its sire precisely because he was watching or helping his companion lasso the antelope that failed to see its danger; there is, certainly, a narrative (re-)connection between Group 4 and Group 1, completing the movement or circle of the entire viewing. We can also see now that the viewer is constituted as akin to fallen hunter and doomed antelope. Lured by the apparent order of the image, the viewer falls to see the decisive blows occurring

outside, around, or deep within it—placed, in the image, where the depiction goes blank at the point of its interruption by the cosmetic saucer.

The top area of the palette above the cosmetic saucer is both the "beginning" of the image and a passage of depiction (re-)encountered after working through all four groups—for example, when the palette is brought upright again after the central passage, with long axis horizontal, has been surveyed. Thus, as a passage of metaphorical and narrative relations between what the viewer unfolds as Groups 1–2 and Groups 2–4, the top area of the palette measures human success, the hunter's victory over the antelope, against human defeat, his companion's destruction by the lion. But the forces of man and nature are not equally matched; there is much more to the story. Whereas man can imitate nature, nature cannot imitate man. Whereas nature, in the pair of a carnivore and its prey, must literally confront itself in its contests, therefore always having opportunity to see its danger (the narrative message of the Oxford Palette), man stands back and strikes from the sidelines. He fills the gap between the unseeing target at a distance from him and the real ground of his deadly force, an advancing edge coming into the scene, with all manner of costumes, lures, traps, projectiles, and masks—in general, with representations.

Like the decorated knife handles (Figs. 8, 9, 11, 20–23) and the Oxford Palette (Figs. 26, 27), the image on the Hunter's Palette has a tight structure (Fig. 31). "We are in the time and space of the sign, a sign that brings into play a representation of each of the orders of life—carnivore, herbivore, and human—and establishes among them a relation of combat" (Tefnin 1979: 225). Although our interpretation cannot be confirmed by independent external evidence, it seems that no figure is left out of the representational mechanism pivoting around the wild space of the cosmetic saucer—its central node—upright, down to the right, back upright, down to the left, and repeated, just as many individual figures throughout the image pivot on their own axes.

As with the ostrich launching its flight and the gazelle with twisted horns on the Oxford Palette (Fig. 27), a series of cipher keys, four in all and corresponding to the four groups of the image, guide—by representing—the motions of and for the viewing of the Hunter's Palette. We can choose to understand them as symbols activating or "narrativizing" the otherwise static image

(for this quality of pictorial narratives, see the Appendix). The "shrine" and the joined foreparts of two wild bull oxen or buffalo at the top right curve of the palette (Fig. 30) should probably be read to say "go through the door on the left and look back and forth, left and right, up and down," while the flying ostrich and leaping hare should be read to say "jump to the next stage." In their preliminary position within the pictorial text, like the wild dogs on the Oxford Palette, the topmost signs probably also announce the theme of the image—to be provided with metaphorical and narrative substance in the very viewing they prescribe for the viewer. The open door of the "shrine" is on the left, while its right side is a closed, blank facade, corresponding to the positions, the actions, and the fortunes of the two chief actors in the tale, ruler on the left and lion on the right. The bulls look in both directions at once, but as individuals the bull on the left cannot see what is to the right and the bull on the right cannot see what is to the left—a condition corresponding generally to all the identities in the image, except for that of the ruler, who is precisely in the place of or able to go through the door and become the "double bull," that is, able to see in both directions at once.^[6]

Having surveyed the various elements of the image, we can now review the "looking one way and the other" constructed by the pictorial mechanism (Fig. 31) by going through the ordered temporal activity of viewing.^[7] Initially holding the palette upright (long axis vertical), the viewer sees, on the short horizontal baseline at the top of the image, the lion just after striking down the doomed hunter but also just before the ruler strikes his decisive blow against the lion. That blow is not literally depicted in the scene; at the bottom of the palette the lion is shown as defeated, the decisive blow having fallen. Here, like the hunter he had destroyed, the lion has been pivoted from his earlier horizontal axis and turned "upside down" to represent his own death.

In examining this scene of confrontation and following the instructions of the "double bull," the viewer traces the implied trajectory of the ruler's arrow and turns the palette down to the right, moving the ruler "down," "around," and "behind" the lion. The palette is now reoriented in the viewer's hands with its long axis horizontal to the viewer's line of sight. With the bottom lion now appearing on the left side of but facing away from the files of hunters, the viewer

Fig. 31.
Viewing the Hunter's Palette. After Smith 1949.

begins to see properly how the beast was slain. (In other words, from Group 1 to Group 2 the pictorial text shows the viewer how the lion was killed after showing it about to be killed; it moves from a "present" to a "past" tense.) In particular, the viewer notices the arrows penetrating the lion from behind, three arrows in the back of its head and one arrow in its anal-genital region.

Scanning from the doomed lion on the left, from left to right along the long horizontal back into the scene, the figures turning up next in the time of the viewing—the jackal, the unseeing stag, and the escaping antelope (Group 2)—represent the victim's alternatives: namely, seeing the force coming up behind (and escaping) or failing to see (and being destroyed). Next on the horizontal as viewed from left to right, the following figures (Group 3) depict the moment just before the lassoing, while the final figures (Group 4), on the other, right side of the saucer, depict the moment just after the lasso has coiled over the antelope's horns. Both groups around the saucer represent the attack from behind and mistaking the mask for the blow—that is, they metaphorically, re-present the death of the lion.

For the viewer holding the palette with right side down, the last figure on the long horizontal, the doomed antelope, is now "upside down." That is, the antelope is in the same position of "death" as was the doomed lion when the palette was upright. Once the viewer brings the palette back upright to its original position, with its long axis in vertical and its short axis in horizontal orientation, Group 1—ruler, fallen hunter, lion, and cub—comes into view again so that it can be more fully understood. The text returns—now in a "past" tense—to the episode it had already partially related; the viewer now sees the lion's danger clearly and knows in advance what its consequences will be, because the text has enabled (in Group 2) a "look ahead" at the lion killed from behind.

But the action of viewing does not end here. To bring the fallen antelope upright, the viewer must continue to turn the palette on its pivot downward to the left. Holding the palette in this orientation, with the left side down and the top of the palette to the viewer's left, the real truths of the scene are made fully visible. The text goes through the narrative chronology of events one more time with complete retrospective clarity. The doomed antelope and lassoing

hunter now become upright for the viewer. At the same time the jackal running "upside down" beside the antelope reveals itself to be intervening between antelope and hunter; it becomes a forward mask for the antelope's true danger, what the antelope manages to see when to save itself it should see its human enemy. In parallel the doomed hunter mauled by the lion is equally revealed as what masks the ruler, coming up behind the lion, from the lion's sight; while the lion faces the fallen hunter, the ruler edges into position behind the beast to shoot the fatal arrow. The remainder of nature, for the viewer now on the right side of the saucer along the long horizontal axis, is entirely "upside down"; that is to say, it is revealed as target and victim of the hunters' blows—as what has died or will do so—like the antelope and the lion itself. At the beginning of the story the hunters march into a nature that does not observe them. In the viewer's first viewing orientation the animals engage only themselves, upright and at right angles to the hunters. But now, at the end of the story in the viewer's final viewing orientation, the hunters march out of nature, having mastered it. The "upside-down" lion at the far right end of the image, facing out of the scene altogether, is returned, destroyed, to the place outside the files of hunters whence it first advanced.

Finally the true directions and targets of the hunter's lasso and the ruler's arrows become visible. The lasso, snaking out from the file of hunters with the jackal running along it (a visual conceit employed on the Oxford Palette, Fig. 26), figures the double twist required to interpret the image (Fig. 32); holding and moving the palette upright, down to the right, upright, and down to the left unfolds the narrative. The hunter employs the lasso to wrest the antelope onto his own ground; the throwing of the lasso is one of the main events of the narrative. In the pictorial text—in the formal functioning of the image itself—it is a literal ground line along which the human hunters' forward masks are advanced; it takes possession of the represented world, contested unsuccessfully by the lion. Natural master in nature, and even threatening the regularity of human order as he tears the doomed hunter from his baseline, the lion still lacks the force of representation. Unlike the human ruler, the lion cannot assume a distance from his target and advance a representation against it, and consequently he is mastered by the ruler's arrow fired from behind.

Fig. 32.
The ruler's arrow and the hunter's lasso on the Hunter's Palette. After Smith 1949.

The spatiotemporal logic of the events related in the image requires that the ruler's arrow fly straight from his bow to strike its target; but in the story as it is depicted, we are given only the preconditions (Group 1) and the consequences (Group 2) of this action. In the pictorial text, the literal depiction within the image, the arrow must in fact fly "outside," "over," and "around" the scene to get from its initial to its target position. In the first viewing orientation (long axis vertical), the arrow launched by the ruler (Group 1) must fly "around and behind" the lion and "down" the palette itself (to Group 2); and in the final viewing orientation (long axis horizontal, left side down), the arrow must fly "over" the scene of the hunters' mastery of nature (reading from left to right, "over" Groups 1, 4, 3, 2) to hit the lion. The lasso, by contrast, is thrown in the other direction; it moves up from below (in the first viewing orientation), from left to right (in the next orientation, with long axis vertical and right side down, it is thrown by the hunter in the left-hand file), and from right to left (in the final orientation, with long axis vertical and left side down, it is thrown by the hunter in the right-hand file). Since the flight of the lasso is directly depicted while the flight of the ruler's arrow is not—in the final viewing orientation the arrow crosses "over" the lasso going in the other direction—the lasso, like the jackal running along it, is in itself a metaphor for the ruler's blow.

The flight of the ruler's arrow links the top and bottom of the image just as the lasso ties together the left and right. Indeed, to link the first (top) and second (bottom) stages of the scene of the lion's destruction—to connect before and after—the path of the arrow about to come around behind the lion, at the top, and then plunged into him, at the bottom, is to take the very twist the image literally provides for the end of the lasso representing the ruler's decisive force within the depicted scene. In fact, like the end of the lasso, the arrow must make a complete circle, equivalent to turning the palette all the way around its pivot. But this circle, the flight path of the ruler's arrow, could never be literally depicted. In the first viewing orientation (long axis vertical), the arrow can fly "down" the right side, but cannot hit the lion at the bottom, who is "upside down" in relation to the orientation of the ruler at the top; and in later viewing orientations (short axis vertical), the arrow can plunge into the

lion at the bottom, who is hit from behind, but cannot at the same time be seen as coming from the ruler, never depicted as standing "behind" the lion on the long horizontal axis but always necessarily at right angles to it, "upside down," from the vantage point of a lion who is not able to see him. Thus the arrow's flight, the deadly circle enclosing and destroying the lion, hovers somewhere above and outside the depiction. In this wild space the oblique force of the ruler's blow, advancing behind representations, escapes representation. It is here too that the viewer twists the palette back and forth, pivoting it around itself, in mastering the depicted scene of mastery.

6—
In the Morgue

The somewhat misleadingly named Battlefield Palette (Fig. 33) was found with the Hunter's Palette toward the end of the last century (Baumgartel 1960: 96-97), neither having an archaeological provenance (for the fragment fitting into the upper right corner, which was discovered later, see Mueller 1959 and Harris 1960). Compared with the Hunter's Palette, the Battlefield is wider and has a blunter, squared-off bottom edge similar to the shapes seen in later types of routine funerary palettes; it should probably be assigned to the Nagada IIIa/b or Horizon A /B period. In narrative structure and metaphorics it is probably later than—because it seems to assume a viewer's full familiarity with—the Hunter's Palette (Fig. 28) but earlier than the Narmer Palette (Fig. 38), which it probably resembled in having a canopylike decorated top edge making a gentle double curve. Because of its fragmentary state the image is difficult to understand.

Parts and Sides

The missing top edge should perhaps be reconstructed like the top of the Narmer Palette to display "heraldic" animal heads, a "royal"-palace-facade emblem (serekh ), and hieroglyphs giving the ruler's name. But excluding this symbolic border—on the Narmer Palette it is related to, not directly part of, the narrative image—the Battlefield, like the Hunter's Palette, presents four principal groups of figures, each including four agents or actors (Fig. 34).

On the Battlefield Palette, however, the textual positions of the groups are reversed. Rather than appearing at the top and bottom of the image, the two episodes closest to the moment of the blow are brought together into the center

Fig. 33.
Battlefield Palette: carved schist cosmetic palette, late predynastic (Nagada III) or early First
Dynasty. Ashmolean Museum, Oxford (top fragment, photo), and British Museum, London
(bottom fragment, photo). 
Photograph courtesy of British Museum; line-drawing restoration of the palette includes 
fragment from Kofler-Truniger Collection, Lucerne, Switzerland (top right edge).

Fig. 34.
Groups of figures on the Battlefield Palette, obverse.

of the composition. The other elements of the story are pushed to the outside, to the top and bottom of the image. In general, then, in comparison with the Oxford and Hunter's Palettes (Figs. 26, 28) the central area on the Battlefield Palette, including the cosmetic saucer, is given over more directly to the depiction of the ruler's power and danger. This rearrangement can be regarded as a continuation of the process we have been tracing through the entire chain of replications in which the ruler's presence—beginning outside, circling, and at the sides or edges of the depicted scene—is advanced into it.

If the Battlefield Palette presents an encounter with the ruler's blow at the visual and temporal center of the viewing of the image, the blow itself, as on the earlier works, is not depicted. Required in the narrative chronology of events and implied in the arrangement of those events, it is deflected from view in the pictorial text. Although this general device is the fundamental text of late prehistoric representation, its replication on the Battlefield Palette, like every other one examined here, has its own dynamics.

Each of the four groups of figures in the image forms a rough register band, especially by comparison with the groups of figures on the Oxford and, to a lesser extent, the Hunter's Palettes (Figs. 26–29), which occupy irregular fields in the compositions, arrayed in relation to one another more like pieces of a jigsaw puzzle than as a tier of friezes. Despite its adherence to an overall compositional device, however, no register ground lines are actually drawn in; the Battlefield Palette is certainly not a canonical image in this regard.

Whereas the Oxford Palette had its four principal groupings spread on both sides, two on either side, the Battlefield Palette, like the Hunter's, presents all four groupings on only one side. The reverse bears a single, mysterious image that might be no more than a metaphor for the narrative image on the obverse. Moreover, it does not seem to be a separate narrative image in its own right. Thus, whereas the Oxford Palette consists of a single four-part narrative image on two sides and the Hunter's of a single four-part narrative image on one side, the Battlefield may consist of two images, one on each side, one a four-part narrative image and the other perhaps not a narrative at all. In its turn the Narmer Palette (Fig. 38) will replicate and revise this arrangement. It spreads a six-part narrative image on both sides of a two-sided palette, but one zone is

particularly magnified (like the image occupying the entire reverse side of the Battlefield Palette, it is placed in the center of its reverse), as if in addition to being an element of a narrative image, it can be seen as having a specific symbolic status in relation to the whole.

With Arms Bound

The topmost band on the obverse of the Battlefield Palette depicts the legs of two slain enemies of the ruler. While the palette is badly damaged here at the top left, the two figures should probably be restored like the enemy farthest opposite them in the composition—that is, in the bottom right angle of the palette. They have been vanquished, their limbs splaying lifelessly; we are meant to understand them as sprawled on the ground.^[1] Next to the sprawling enemies and directly above the cosmetic saucer, the top band might have presented a central motif such as hieroglyphs giving the name of the ruler.

The right side of the topmost band of the image depicts a kneeling enemy prisoner with arms bound. He is mastered by a wild dog or a jackal, creatures familiar to us from the Oxford and Hunter's Palettes and other images and with somewhat similar roles to play in the pictorial metaphorics of those works. Since the top of the creature's head has been lost, we cannot determine what kind of ears it has (rounded for a wild dog, pointed for a jackal); the bushy tall with band markings is consistent with earlier representations of either species. The poses of the two figures suggest that the beast is pulling back the enemy's head, perhaps in preparation for the human ruler to throttle or decapitate him. Alternatively, like the falcon on the reverse of the Narmer Palette (Fig. 38), the beast may be inserting a hook or cord into the enemy's nose.

Behind these figures, over the shoulder of the beast, there remains the bottom portion of what could have been a hieroglyph (Harris 1960: 104). It may represent the base of a stand of papyrus, like the one placed in a similar position on the Narmer Palette (reverse, top right). If so, following a common but somewhat disputable reading of the Narmer glyphs, the sign may have served to label the number of conquered enemies, metaphorically depicted by the bound prisoner being mastered, or to name the locality in which the battle

took place. In making an indication of this kind, however, the image on the Battlefield Palette would be going well beyond notations used to this date in late prehistoric representation—namely, what I call the "cipher keys," or symbolic motifs that represent the act of viewing rather than the content of the scene to be viewed, employed to activate the image. Instead it would be functioning as a hieroglyph proper—a "caption" of an episode or a statement of a more general message. Whatever the case, whether a cipher key, a hieroglyph proper, or something else altogether, the artist was not quite comfortable about including it in the image. Evidently used to working with uninterrupted pictorial narrative text, he squeezed the right side of the composition awkwardly downward, about half the height of a single band, to accommodate it. Many commentators on the Battlefield Palette, not noticing the anomaly introduced by the sign at the edge of the image, have therefore failed to see that it presents four bands of four figures each. They describe a composition more like that of the Oxford and Hunter's Palettes, even though the maker of the latter (Fig. 26), facing a similar formal challenge, had managed to include the "shrine" and "double bull" above the figures of a lion and cub without difficulty. (The same problem is solved differently in a later example: the right-hand group in the top band of the image, presented here as three separate forms [bound prisoner, mastering dog or jackal, and cipher key or hieroglyph] is brought together on the Narmer Palette as a single compact form with a tripartite structure: the rebus" [Fig. 38, reverse, top right] comprises the conquered enemy and the ruler represented in animal form [wild dog/jackal or falcon] plus a notation or symbol for place and/or number.)

Moving down from the top of the palette, three forms representing victorious captors in the second band of the image, apparently figuring the ruler or the ruler's victorious army, enter from the sidelines. The Ibis and Falcon standards, on the left side of the cosmetic saucer, and a robed personage, on the right, each follow three defeated, bound enemies wearing penis sheaths but otherwise naked. Although the robed personage has been called a goddess or a Libyan (see Vandier 1952: 586), the entire narrative and metaphorical structure of the image—in the overall context of the chain of replications—suggests identification as one of the ruler's retainers; each zone of the image on the

Battlefield Palette consistently presents three enemies and three "representatives" of the ruler, while the ruler's presence as such remains outside depiction.^[2] With arms bound behind their backs, the enemies are being marched by the victorious captors toward the center of the compositional field—that is, toward the cosmetic saucer or the wild space outside representation—where presumably the ruler is placed.

An indecipherable motif, now mostly missing, appears in front of the enemy on the right, mastered by the robed personage. It might be a weight hanging around the prisoner's neck, or perhaps an independent glyph—for example, the bottom part of a rebus like the one on the reverse of the Narmer Palette (although its ovoid outline does not have the shape of Narmer's stand of papyrus stalks). The fate of the three bound prisoners in this band of the image is not directly depicted; they may be about to be judged and then strangled or decapitated. (Perhaps they are to be choked by a heavy rope noose or decapitated by a blow of the mace like the ten bound prisoners depicted after the decapitating blow on the Narmer Palette [Fig. 38, obverse, top right.] Their death must precede having their broken bodies thrown to scavengers, depicted in the bottom band of the image and possibly in the top band as well; and their death must follow the moment of judgment, in which—as the viewer will discover in unfolding the narrative image—there apparently remains the possibility of being spared by acknowledging the ruler.

Williams and Logan (1987: 253, 271) have interpreted the "return and sacrifice of prisoners"—what seems to be depicted in this band of the image and perhaps an aspect of its theme as a whole—as part of a larger "pharaonic" iconography that they hope to reconstruct in late prehistoric image making: a cycle" depicting the "expected liturgical activities of the ruler"—namely, smiting his enemy (or its representation in ceremonial actions), "possibly hunting," traveling by river in a sacred bark to a paneled palace facade, and officiating at the Sed -festival or jubilee ceremonials. There are numerous later parallels for the capture, display, and execution of prisoners depicted on the Battlefield Palette—including the Narmer Palette and the roughly contemporary rock relief from Gebel Sheikh Suleiman (Fig. 35) showing bound prisoners and slain enemies (Murnane 1987), a door socket from Hierakonpolis (Quibell

Fig. 35
Relief of bound and fallen prisoners, etc., from Gebel Sheikh Suleiman, Upper Egypt, early First
Dynasty (time of Narmer?). 
From Murnane 1987 by permission of the author.

Fig. 36
Incised reliefs of slain enemies from base of limestone statue of Khasekhemuwy 
(see Fig. 51), late Second Dynasty. 
From Quibell and Green 1900–1901

and Green 1900–1901: I, pl. 3; Davis 1989: fig. 6.26) representing a prisoner or slain enemy with arms bound behind his back, and the incised reliefs of slain enemies on the statue bases of King Khasekhemuwy from the end of the Second Dynasty (Figs. 36, 51), as well as other minor examples, especially among the group of early dynastic carved ivories from Hierakonpolis (see the full list of parallels cited by Williams and Logan 1987: 247, 256, 267–71). Deriving from historical contexts somewhat or considerably later than the Battlefield Palette, all these examples are "later" replications, likely to be revisionary in the same way the Battlefield Palette revises the Hunter's and the Hunter's revises the Oxford. It is not feasible here to explore the disjunctive relations between the Battlefield and Narmer Palettes and later images. But at least it is clear that the motif of the bound and slain prisoner was taken up by early dynastic state artists for their own purposes, as a selection from a retrospectively constituted group of possibilities, regardless of its "earlier" significations. In canonical representation it was developed as an individual image of established pharaonic rule worthy of replication in its own right; expressively magnified, it was often combined with other motifs in a complex iconography (see Davis 1989: 64–82; Hall 1989).

Since the Battlefield Palette is one of the earliest unequivocal versions of the theme of sacrificed prisoners bound by ropes, however, it would be anachronistic to read back from the later, canonical images to specify its meaning. Rather, the meaning of later and canonical images must be seen as repeating and revising whatever the Battlefield Palette represents. The act of smiting was depicted in the Decorated Tomb painting from Hierakonpolis (Fig. 5), probably to be dated to the Nagada IIc/d period and thus well before the Battlefield Palette. Immediate preparations for an act of smiting seem to be depicted on an incense burner from Grave L24 at Qustul, Nubia, dated to the Nagada II/III transition (Williams 1986: 158, fig. 56). The Narmer Palette—which I would date after the Battlefield Palette on both typological and structural grounds—has the scene of smiting occupying the central place on the reverse side of the image. Therefore we can assume that the motif of prisoners bound and slain on the Battlefield Palette was associated with an established—but, if replicable, therefore revisable—story about smiting the enemy; we have no independent, nonanachronistic evidence, however, about what that story "meant" for its late prehistoric viewers.

Although in either case the narrative tells a story about a "ruler" and an enemy," it cannot have signified exactly the same thing for the resident of a small-scale ranked polity led by a local ruler, however elaborate his court, as for the subject of the national, authoritarian state—if such is the measure of the social difference between a viewer in the age of the Battlefield Palette and a viewer in the reign of Khasekhemuwy (Figs. 36, 51). For example, the enemy of a late prehistoric polity might have been the leader of a neighboring natural irrigation basin fifteen or twenty kilometers away (see Butzer 1976) or the population of another area of Egypt; by contrast, the enemy of the dynastic state was usually a foreigner, or sometimes a criminal, social revolutionary, or religious reformer (for social and religious controversy at the end of the Second Dynasty, see Edwards 1972: 30–35 and Kemp 1983: 54). The motif of prisoners bound and slain could be replicated to have a meaning in either of these contexts.^[3]

We must, then, look carefully at the way the image on the Battlefield Palette actually presents the general narrative—an available if possibly evolving

one—of capturing and killing an enemy. It is not surprising that the text should be related to the existing pictorial texts of animals in rows, carnivore hunting prey, and human beings hunting and otherwise doing battle with wild animals, as on the Ostrich, Oxford, and Hunter's Palettes (Figs. 25–29). Given the formal and structural similarities between these earlier images and the Battlefield and Narmer Palettes, the narrative of capturing and killing human enemies probably depended metaphorically on such earlier narratives and was initially produced as a replicatory revision of them.

In the band below the cosmetic saucer (the third band from the top) on the right, one of the enemies, his arms still bound but his penis sheath absent or removed, lies on the ground. If it is a battlefield, the battle has already taken place and the victors have left the scene; a scavenging raven plucks out this enemy's eyes. Like the "battlefield" scenes above the saucer, which also present an "aftermath," this depiction would be preceded by the forced march of the prisoners toward the saucer (in the second band). Evidently the saucer is the wild unrepresented place containing the ruler's decisive blow—here advanced more literally into the scene than on the Hunter's Palette (Fig. 28)-where the prisoners are slain before their corpses are tossed out on the field.

Indeed the owner actually holding the palette must grasp it at just the left and right edges of the second band where the prisoners are being thrust into the center of the composition. The viewer no longer literally twists the palette around to view the image—as on the Hunter's Palette (Figs. 31, 32), where the narrative is pictured in a way that requires physical reorientation of the image from upright to horizontal and back. Here the same ellipsis does not call for such handling in the process of unfolding the narrative: the image includes in the pictorial text what was earlier physically performed by the viewer on the pictorial text.

The bottom (fourth) band of the image presents a final depiction of the battlefield or, more exactly, of the morgue—the place where the enemies, captured, judged, and executed by the ruler and the ruler's representatives, are thrown to be torn apart by scavengers. Here three enemies, "twins" of the three enemies in the second band about to enter the wild space and encounter the ruler, lie slain on the ground. Unlike their living "twins" above, however, their

arms are no longer bound, and they have been stripped of their penis sheaths; since they are dead, they no longer need to be confined, and in fact they are utterly exposed to the elements.

All three enemies are depicted with their heads completely twisted around—an especially vivid detail that can be fully understood only by reconstructing the narrative events as the image requires that they be constituted. Before the battle the enemies presumably failed to see and acknowledge the power and danger of the ruler; after the battle, as the literal depiction shows, they are rounded up by the ruler's representatives. And after their capture—here the pictorial text drops into ellipsis, in the wild space in the middle of the second band—the enemies must actually see the ruler who comes up directly before them and condemns them to die; the ruler or his representatives then kill them by breaking their necks, and their bodies are tossed onto the battlefield or into the wilderness. The pictorial text takes up again after this ellipsis: in the bottom band is a group of three vultures and three ravens—one of each species for each of the dead enemies—in the act of alighting on the battlefield. The scavenger birds come from the same direction as the ruler's standards and officials—the ruler's "hands"—and the place of the real owner holding the palette itself. Two of the ravens peck at the hands and feet of the left-hand enemy. Moving to attack the right-hand enemy, the third raven, noticing the scene immediately "above" it where a lion mauls an enemy, apparently anticipates being able to make off with the victim's entrails. While two of the vultures have not yet landed, the third, right-band vulture approaches the head of the middle enemy to pluck out his eyes. In sum, having first failed to see the power and danger of the ruler, the prisoners must die with their heads twisted around, as if to be able to see what had come up "behind" them, but with their sight obliterated any such knowledge is gained too late to spare them the ruler's judgment.

Since the viewer literally occupies the ruler's place above the saucer, holding the palette and looking down into the depicted scene, it is the viewer who in constructing the narrative performs the "twisting around" that the image figuratively builds into its composition and narrates as the condition of the enemies' bodies. It is the viewer as well as the ruler who thrusts the captured prisoners

into the morgue. Unseen like the ruler, whose fierce action always remains outside view, the viewer masks the ruler. He or she is like one of the ruler's officials holding a prisoner with arms bound: he or she holds the palette, which is no longer to be moved physically in the hands but rather is held upright, its image folded around itself.

The Ruler's Mask and the Ruler's Enemy

Just below the cosmetic saucer, the third band in the composition depicts a moment immediately after the bound prisoners emerge from the wild place they are shown as entering in the band above. (The blow remains unrepresented in that wild place.) The metaphorical and narrative relation between the third band and the rest of the image is complex and, in view of the missing information once contained in the lost portions of the palette, somewhat uncertain. Presented at noticeably larger scale than any others in the image, implying their preeminent place in the narrative, are two figures flanking a lion disemboweling a fallen enemy. The latter could be a chieftain or, at least, could symbolize an army, territory, or population inimical to the ruler. His arms are unbound, and he has been stripped of his penis sheath; in the literal terms that are operative here, he has been—like his "twins" above and below—captured, condemned, and killed, his broken body thrown onto the battlefield or into the wilderness.

The identity of the lion is more mysterious. It is conceivable that the image should be taken literally at this point—in the same way the Hunter's Palette (Fig. 28) might be interpreted—as depicting a lion actually mangling the abandoned body of one of the executed enemies he encounters in scavenging. But unlike the vultures and ravens, the lion is not naturally a scavenger; rather, he attacks and kills his own prey. Moreover, the structure of the overall image—with two bands already presented, viewing from top to bottom, prior to this group—suggests that the lion has metaphorical connotations. Each of the four bands in the image pairs a metaphorical representation of the ruler, his power, and its preconditions and consequences—the wild dog or jackal in the top band, the standards and official in the second band, and the vultures and ravens in the bottom band—with a group consisting of three enemy figures.

According to this scheme the lion in the third band, paired with a group of three slain enemies, stands in for the ruler. (For a discussion of the lion on the Battlefield Palette as the king, see, for example, Spencer 1980: no. 576.)

An association between the lion and the "ruler" —whether this real identity is a matriarch, a village headman, or a hereditary national monarch—seems to have been current from the late Nagada II/III period (for instance, on the incense burner from Grave L24, Qustul, Nubia [Williams 1986]). In later Egyptian state religion and canonical iconography, the king's identity—like that of some divinities—was continuous with or immanent in other creatures; for example, early dynastic lion sculptures, probably set up in temple grounds, may have had associations with the monarch (Davis 1981b; Adams 1984). On the Narmer Palette (Fig. 38) the figure of King Narmer, depicted in human form in his identity as king of Upper Egypt, is singled out in scale and compositional placement in the same way as the lion on the Battlefield Palette.

We cannot be certain, however, that the later beliefs about the king's nature were held also in late prehistoric society, despite the possibility of some available associations between lions and "rulers." And other representations must have been current as well: here the evidence reviewed in this study is quite direct. As the Oxford and Hunter's Palettes imply, associations between animals and human beings involve something other than direct symbolic equation; rather, they are complex metaphorical patterns of relationship, poorly understood even for later canonical Egyptian art (see Davis 1989: 77–82). For instance, on the Hunter's Palette, ruler and lion are not identical but stand as victor to vanquished, unseen to seen, representer to represented, and so forth.

Whether or not late prehistoric rulers or Egyptian pharaohs (or both) were imagined to have a nature including an animal aspect or appearances, the ruler's human aspect should still be distinguished from his animal, which—unless we happen to believe Egyptian iconography—must be understood as symbolic and so interpreted in the scene of representation. Given that the human person of the ruler wielding the blow against his enemy does not literally appear in the image on the Battlefield Palette, the lion that does appear—either as separate from or as a symbolic aspect of the ruler—masks the ruler's presence.

The lion probably "is" the ruler in a general, extended sense; but this statement is shorthand for a particular relationship of representation we should

not elide if we hope to understand the images under consideration. For example, it is not clear why the lion's tail is curving up between his legs rather than floating over his back, as might be more natural (and as was certainly more usual in late prehistoric and early dynastic depictions of lions in general [see Baumgartel 1960: 100; Davis 1981b; Adams 1984]). Nevertheless, like many apparently insignificant or fortuitous details in these images, this one makes sense in the context of the chain of replications. If the lion on the Battlefield Palette does represent the ruler, the image maker may be specifically concerned to differentiate his lion-ruler, not exposed to an attack from behind, from the earlier lion-enemies —for on the Hunter's Palette, an upraised tail had exposed the lion's anal-genital region to the lead hunter's arrow. Indeed the end of the Battlefield lion's tail stands in quite vividly for the beast's phallus—the bulbous tip emerges like the glans of the penis—and therefore, in the metaphorics of the image as a whole, for the phallus of the ruler.^[4] On the Hunter's Palette the anal-genital region is the site of the lion-enemy's defeat, his powerlessness before his human antagonist standing at a distance from him and coming up behind; thus, on the Battlefield Palette, the anal-genital region of the lion-ruler is securely defended and aggrandized. It is almost certain that the maker of the Battlefield Palette had seen the Hunter's Palette, for the two lions have much in common stylistically, except, precisely, for the way each has been inserted into the narrative metaphorics. Clearly the possibility of attack from behind concerns the maker of the Battlefield Palette in that the enemies killed by the ruler or his minions have their heads twisted around. What is behind them, what they see or do not see there, continues to be a subject of representation. In fact the third band of the image is quite straightforward about these relations: it relates or narrates them as its subject. Whereas the left-hand, escaping, fleeing, or freed enemy looks back over his shoulder to see the lion-ruler—protected from behind by his tail—facing away from him and mauling the fallen enemy, the right-hand dead enemy cannot see the lion-ruler; he faces in the other direction, and his eyes are plucked out by a scavenging raven.

An image does not always produce or enable a viewer to render a successful presentation of the narrative it ostensibly intends; in fact, for a variety of reasons, depiction always exceeds narrative statement as such (see the Appendix).

Prior to exploring some striking examples of these difficulties in the Narmer Palette, the central passage of depiction in the third band of the image on the Battlefield Palette—the representaton of the lion-ruler and his enemy—serves as a preliminary case in point.

Because the image maker was attempting to pull together the three axes of the scene—left and right, up and down or above and below, and in and out or back and forward—he had considerable trouble rendering the figure of the fallen enemy. Depicted in profile and with what seems to be a broken back, the enemy looks sightlessly up at the lion who masks the ruler placed "above" the scene (or, from the point of view of the fallen enemy, "behind" his forward representation in the figure of the lion). But we should look carefully at the way the enemy's thighs, arms, and legs appear in relation to the forequarters of the lion. The lion's right forward leg, in front of the other leg, is placed on the enemy's left leg, thrown to the right over his abdomen (the stretched muscle around the iliac crest is carefully rendered); but the lion's left forward leg, "behind" the other, is placed on what seems to be the enemy's right arm, thrown to the left over his midriff, for the enemy's head is actually—just as we would expect—twisted around. The image maker depicted the enemy as totally twisted around himself and bent around the forequarters of the lion in a half-circle on the ground.

All the obliquity and torque in the larger scene of the enemy's defeat—what served on the Oxford Palette (Fig. 26) as the key to the cipher of the image and on the Hunter's Palette (Fig. 28) as the entire mechanism of representation—is advanced into and made literal in the enemy's very body in the scene on the Battlefield Palette. Just as the viewer "twists" and breaks apart the image on the palette to unfold its narrative of the defeat of an enemy, so does the narrative text twist and break apart the image maker's production of the depicted figure of the enemy. As we will see with the Narmer Palette, the blow that falls within the literal depiction also hits the image maker producing that depiction in the scene of representation as a whole.

In the pictorial text the enemy's head, although twisted around, appears to face upward. If the enemy is bent in a half-circle around the lion's forequarters "back" to "front," then the viewer—looking "down" at the image and inter-

preting this figure—reconstructs the enemy as gazing sightlessly not just at the lion "above" him but also out at the viewer . The pictorial text finds a literal means to relate how the viewer figuratively stands behind and looks over the shoulder of the lion beneath whom the enemy gazes back upward in his moment of death. This is to say, in turn, that the human person of the ruler must stand beside and behind the viewer on this side of the depiction. The "twisting around," the mechanism of the scene of representation on the Hunter's Palette, is now embedded as the central node of the intersection of the three axes of literal depiction on the Battlefield Palette—left and right, across the bands of the composition; up and down, between the tiers; and back and forth, in the fictive space that the pictorial text attempts to render. Given that the construction is ambiguous and the image maker was undeniably disconcerted in his efforts at rendition, it is not surprising to find the "twisting around" elsewhere in the scene of representation as well: the image maker, as viewer, goes through contortions in attempting to render the body of the enemy twisted around by the ruler; the enemy's body is deformed in a wild space where the ruler, as we have seen, stands not only before his enemy but also behind the viewer. The image maker's own narrative—or, more exactly, his "voice" as the image's narrator—thus expresses the positions of both ruler and enemy, positions that can never be reconciled.

Losing Sight, Being Upright

To reiterate, on the right side of the central pair of lion-ruler and fallen enemy a prisoner lies on the ground, stripped of his penis sheath, his arms bound and his eyes being plucked out by a scavenging raven. This victim is matched on the left side of the central pair by the figure of another enemy—one who is able to see—standing upright, with arms unbound. The only enemy figure not directly confined, attacked, or approached by one of the ruler's representatives, he apparently escapes death, perhaps freed by the ruler. As he seems to edge or "tiptoe" to the sidelines or outside the scene—toward the mastered ground, beyond the literal depiction, whence the representatives of the ruler had ad-

vanced and where the ruler himself is placed—the upright enemy looks over his shoulder to witness the lion's victory over the fallen enemy. The lion does not see him, for he faces in the other direction, but nonetheless the escaping enemy acknowledges what he sees—namely, the defeat of his comrade—and lifts his hand in recognition.^[5]

In sum the third band of the image—escaping enemy, lion-ruler and fallen enemy, dead enemy—depicts the ruler's position and the victim's alternatives in relation to it, like differently constructed groups on the Hunter's Palette (Figs. 28–30). The victim may be allowed the three axes of "life" —left/right, up/down, back/forward—or may lose them, precisely according to what the ruler gives or withholds in a decisive blow establishing the order of the depicted scene.

Complicated to describe in prose, the ingenious narrative format of the image, once identified, is quite straightforward. In the context of existing images like the ones we have already considered, it must have been readily legible to contemporary viewers. Although we have examined how the viewer gradually unfolds the narrative in the process of viewing, working back and forth among many passages of pictorial text, it is useful to organize our results in a brief review of the entire narrative in uninterrupted sequence. (Notice, however, that this review is not a summary of the narrative experience; a viewer does not obtain the entire narrative all at once or uninterruptedly but rather generates it through repetitions and recollections of pictorial text and interpretations of ellipsis and ambiguity. Rather, the summary states what a viewer, in looking back retrospectively over his or her viewing of the Battlefield Palette, presumably knows he or she has obtained precisely as the narrative coherence of that viewing. It is also a partial statement of what a viewer, having constructed the narrative of the Battlefield Palette, might then take to the viewing of other, "later" images in the chain of replications, such as the Narmer Palette [Fig. 38]; once encountered, these later images require the viewer's retrospective experience with the earlier one.)

The image is divided into four zones, each depicting an episode in a narrative of the ruler's victory in capturing, judging, killing, and disposing of the bodies of three enemies. The story begins on the left side of the top band with

a vivid depiction of the battlefield with strewn enemy corpses. While the missing material precludes a definitive interpretation, the scene probably depicts the immediate aftermath of the battle. If this is so the top left group of enemies represents those killed in battle, and the following bands represent the capture of surviving enemies who will be judged (top right band and second band), then killed by the ruler or his representatives (third band) and tossed onto the field (fourth band). Whereas the top band would thus be the logical beginning of the narrative, it could also depict enemies to be killed after being judged by the ruler or his representatives; the wild dog or jackal is apparently binding a surviving enemy, a group that resembles the enemies with standards on the left and the enemy and robed personage on the right in the band below. This scene—perhaps it is confined to the right side of the top band—would be a later episode in the entire narrative of the ruler's contest against his enemies. The two possibilities are not mutually exclusive and seem to be separated as the left and right portions, the "first" and the "next," of the "beginning" of the narrative. Especially if the top band included a cipher key or "hieroglyphic" symbol in the upper right corner, then it would be an announcement of the general theme as well as an instruction for viewing. Certainly as narrative material the top band is a preliminary run-through of the principal events of battle and capture, setting up a frame of reference for the more detailed representation of the theme in succeeding bands. In other words it has the narrative status both of beginning the story and of providing a textual image of the entire sequence.

The story proceeds in the second band to present three captured enemies being prepared for actual or ritual slaughter by the ruler's representatives. It implies that the captives are being brought before the ruler for judgment—they are being marched "toward" some goal or fate—although his person does not directly appear. The ellipsis is filled in by the following, bottom two bands, which show the dead enemies (one with arms still bound) as well as one living, freed or fleeing enemy. Because the third band contains the figure of the lion best construed as representing the ruler and what can then be interpreted as the victim's alternatives, the viewer is able retrospectively to construe the center of

the second band, the cosmetic saucer, as the wild place where the ruler actually confronts his enemies, condemns them to death, and carries out the execution.

The third band dramatizes the ruler's decision and explicitly states the fortunes of, and the alternatives facing, the enemies—with special focus on a principal enemy, presumably the enemy leader, represented as the special target of the ruler's leonine power, his ferocity and implacability "as a lion," inviolable "from behind" whereas the enemies are in danger there. Although one enemy on the left side of the band manages to escape or has been freed—he seems to acknowledge or accept the ruler—another enemy on the right side and the enemy leader himself have been executed and left to the scavengers on the field.

As a kind of "fade-out," and echoing the first band, the bottom band concludes the story with another view of the battlefield, a more literal presentation of material covered indirectly in the zone above it. The splayed corpses of the enemies who did not recognize the ruler are left to rot as scavenger birds pick apart their eyes and limbs.

This review indicates that it is possible to follow the story—as a sequence of four bands viewed top to bottom, with passages of depiction arranged from left to right—as obeying chronological and causal logic, with some foreshadowing and flashback introduced to construct the prinicpal ellipsis, the presence and place of the ruler's blow. But another construction is equally possible, shifting the specific metaphorical and narrative valence of individual motifs. As episodes of the story the four bands can also be understood to thread together two strands running parallel simultaneously—namely, the scenes of death and destruction on the battlefield (top left, third, and fourth bands) and the scenes of capture, judgment, and killing by the ruler (top right, second, and third bands). In an image that relates the ruler capturing, judging, and killing his enemies in the process of doing battle with them, it is easier to construe the third band as part of the story of death on the battlefield if the lion in the third band is taken literally as a scavenging animal rather than metaphorically as a representation of the ruler. But these two strands of story material cannot be separated absolutely, and either construction of a story line with its appropriate pictorial text is possible. Indeed, it is significant that the image maker appears

not to have cared which particular interpretation of the image a viewer adopted. As well as producing a tightly constructed ellipsis, the text permits narrative ambiguity, multiple unfoldings of the narrative logic in different elliptical sequences realized by the pictorial text, and hence a substantial metaphorical play with what would otherwise be the discrete, specific "meaning" of motifs.

The Symbolic Side of the Narrative

While it is possible to understand most elements of the image on the obverse of the Battlefield Palette, despite its missing segments, the relation of the other side to the narrative remains obscure. The reverse presents a single group of four figures—equivalent, it would seem, to one band on the obverse. It shows two gerenuk or giraffes—from their hooves to the tips of their ears, spanning the full height of the four bands on the obverse—searching for dates in the fronds of a palm tree, the central vertical axis of the scene. From the surviving fragment we can reconstruct the tree as having ten fronds on each side and two bunches of dates above. The bottom edge of the object serves as the baseline for the animals; it is straight except for slightly curving tips at the bottom angles of the palette. Two plump birds of unidentified species, sometimes said to be guinea fowl, flank the gerenuk on right and left. Apparently the top of this side of the palette was decorated with a form, now unreadable, echoing the shape of the bottom edge. It is possible that it should be reconstructed as a canopy, plain on the bottom but ornamented—as on the Narmer Palette (Fig. 38)—with bucrania (perhaps heads of the cow goddess), royal palace facade (serekh ), and ruler's name. (I cannot follow Williams and Logan [1987: 264] in taking the form as the king's sacred bark, for this object—for example, on the obverse top right of the Narmer Palette—has a distinctively different outline.)

It may be that whereas the obverse of the Battlefield Palette depicts a state of war (Asselberghs 1961: 285—or, as I describe it, a complex narrative of the ruler's blow), the reverse depicts peace, stability, or mutuality. Expressing the metaphor of animals quietly browsing for food, it could come before or after (or both) the image on the obverse, as the order disturbed by and returning after war. In this case, as on the Oxford Palette (Figs. 26, 27), the viewer can

take the scene as a chapter of the narrative belonging somewhere in the story related by the two sides of the palette. That the reverse of the Battlefield Palette presents only a single motif, however, rather than the band or zone format of the obverse and all other narrative images we have considered, suggests a difference from the other scenes in the story and hints at additional functions.

Perhaps, then, we should see the reverse as the key to the cipher of the overall narrative, like the running ostrich and twisted gazelle on the Oxford Palette or the "shrine" and "double bull" on the Hunter's Palette (Fig. 28). The gerenuk on the right is depicted lifting up a frond to peer within, searching for dates (the missing gerenuk on the left is probably doing likewise) somewhat as the viewer moves down through the bands of depiction on the obverse. After scanning from top to bottom and interpreting the narrative of enemies' capture and death, the viewer could flip to the reverse for both a completion of the story—filling in the initial and the final states of the narrative in a metaphorical presentation of peace—and a confirmation of the way to produce the narrative. While it is puzzling to find the principal key to the cipher of the narrative on the reverse of the image, at the "end" of the text, rather than on the obverse top at its "beginning," as had been done on the Oxford and Hunter's Palettes, keys to the cipher of late prehistoric narrative images are inserted where they are required; the twisted gazelle appears at the bottom of the obverse of the Oxford Palette, its narrative status unfolding "late" in the viewing, while on the Hunter's Palette flying ostrich and leaping hare function when the central passage of depiction is viewed with the long axis of the object in horizontal orientation. Furthermore, as is clear from the same contexts, the function of being a passage of depiction set within the narrative as a story episode and that of being a symbol guiding narrativization itself, a literal or metaphorical representation of the act of viewing, are not mutually exclusive. If the reverse of the Battlefield Palette begins or ends the narrative with a depiction of the peace the motif is believed to connote, then it can simultaneously represent the interpretation of that narrative in its specific details.

Yet another possibility is that the image on the reverse of the palette should not be integrated into the narrative on the obverse in any sense. It could be an independent, even non-narrative image somehow related to the narrative as a

metaphor for its themes or perhaps as an entirely unrelated symbol for something else. For example, if the decorated palette had a "magical" function (with the magical efficacy deriving in turn from the power of the representation), then the reverse of the palette, physically "behind" and protecting the "rear" of the cosmetic saucer and the image around it, could have been meant (as a magical prophylactic or apotropaic sign) to protect the object and the potency of its image by warding off harmful influences. The protection of the lion-ruler from behind—in the context of the chain of replications—is thus observed not only within the depicted scene but also outside it, on the other side of the image. There are possible parallels for this setup: for example, the "smiting" image on the reverse of the Narmer Palette (Fig. 38) is placed "behind," both masking and protecting, the cosmetic saucer that, as on the Battlefield Palette, is the place of the ruler's blow. (On the Narmer Palette, however, whatever their further functional implications, the passages of depiction on both sides of the object are connected as units of a single narrative image.) It remains to be seen, however, what defines the narrative and/or metaphorical relation between the depictions on the two sides of the Battlefield Palette: why should gerenuk grazing in palm trees "magically" protect the palette?

Natural and Unnatural

A certain proportion of the metaphorics of the Oxford and Hunter's Palettes is retained on the obverse of the Battlefield Palette; the general structure of the narrative image replicates as it revises the earlier images. There are, however, some striking disjunctions between them. In the earlier works the lion was a natural force quite different from man (Oxford Palette) or even an antagonistic force to be mastered by man (Hunter's Palette). By contrast, on the Battlefield Palette the lion's natural properties—for example, his ferocity, size, and strength—are assimilated to the ruler's identity, previously rendered as the unnatural force of the human hunter masked as, and by, nature. On the Battlefield Palette, then, the ruler has gained the ground approached from the sidelines on the Oxford Palette and vigorously contested on the Hunter's Palette. It is now his ground; its natural inhabitants are aspects of his being; no longer outside

him and his natural antagonist in nature, the lion represents an aspect of his own being, for the ruler is now nature. Whereas the lion-enemy of the Hunter's Palette was exposed to a fatal attack from behind, the lion-ruler of the Battlefield Palette is fully protected: his tail is tucked up over and around his analgenital region; the enemy in back of him flees instead of attacking him there. In general, rather than masked hunter representing nature in order to master nature, nature now represents the ruler in order to master the human world. As a consequence, on the Battlefield Palette the ruler as a natural, leonine force defeats his human enemies, who are, so the pictorial metaphorics has it, unnatural among men—that is, failing, like the nature mastered on the Oxford and Hunter's Palettes, to see the power and danger of the hunter. From a central, well-grounded place within or as the heart of nature, the ruler wrenches his unseeing enemies from their baselines, topples them from their axes, obliterates their sight, and casts them out into a nature that consumes them. The vultures and ravens picking apart the limbs and gouging out the eyes of the fallen enemies are the latest replications of the carnivores-and-prey formula; they are the natural antagonists for the ruler's enemies, human beings unnatural among men.

Represented by his leonine double and identified with the properties attributed to it by the pictorial text, like carnivores and prey in the established formula the lion-ruler must still confront directly those human agents who would resist his dominion. Thus the image maker showed the lion's claw biting deeply into the flesh of the fallen enemy, just as on the Hunter's Palette the image maker had taken special interest in the beast's ravening jaws. (On another palette fragment [Fig. 37], the king-as-bull tramples an enemy as his standards lasso another enemy figure. Too little of this striking work, the Bull Palette, has been preserved to enable us to construct the narrative image; it may relate closely to the Battlefield Palette.) No longer advancing from the outside of society to the inside of nature, as on the Oxford Palette, but coming from the inside of the natural order of society to the outside, to what is unnatural to it, the ones who represent the natural and the unnatural in society meet in the face-to-face confrontation of the lion-ruler and his enemy. But no gap has really been closed. What is truly unnatural—the human person of the ruler himself,

Fig. 37.
Bull Palette: carved schist cosmetic palette, early First Dynasty. 
Courtesy Musée du Louvre, Paris.

merely represented by his natural doubles—does not appear in the scene. For the identities within the depicted scene, the ruler is outside it, still fully in the wild. No one in the scene of representation, including the viewer beside or behind whom he stands, sees him directly. His presence is immanent in the twisting around, the oblique progress, that insinuates his power into the scene, bringing the scene within his striking range.

7—
Looking Back

Common to the Brooklyn and Carnarvon knife handles (Figs. 8, 23) and the Ostrich, Oxford, Hunter's, and Battlefield Palettes (Figs. 25, 26, 28, 33) is that each advances the masterful human being's presence—the hunter's or ruler's as well as the image maker's or viewer's—in a distinctive way into the depicted scene of his or her mastery. For example, the human figure on the Ostrich and Oxford Palettes, appearing as a lone masked hunter on the edge of a group of wild animals, is figured on the Battlefield Palette as a leader of men, accompanied by retainers (or standards) and also represented by his animal aspects, vanquishing his human enemies. At a very general level the ruler's mastery of the natural and social world will become the central theme of canonical representation as well (Davis 1989: chaps. 4, 7).

Without Interruption

Despite the variety of pictorial metaphorics in the late prehistoric images examined here, they display a common figurative preoccupation. However each image may relate the narrative of human mastery, and whatever it meant to contemporary viewers—a meaning that must have been evolving continually—every image works both in the literal depiction and in the scene of representation in which that depiction is produced to mask the blow of the hunter or ruler. To use the terms of narratology (see the Appendix), the ruler's blow is not presented in the pictorial text for the story related, the sequence of episodes that makes up the image (although it is logically required in the fabula, the chronological and causal logic of the underlying narrative material). Rather, the viewer sees the preconditions of the blow, like a hunting expedition

(Figs. 25, 26, 28), or its consequences, like a battlefield strewn with corpses (Fig. 33).

For example, on the Hunter's Palette (Fig. 28) at the top and "beginning" of the image the viewer encounters a hunter, possibly the leader of the hunting expedition, fitting an arrow to his bow, evidently just about to shoot a lion attempting to protect its cub. The viewer does not literally see the flight of the arrow; that fact is logically implied in the concluding sequence of the narrative—namely, the death of the lion after being struck by six arrows, presented at the bottom of the image. This ellipsis in the pictorial text of the image is in fact even greater than might first appear. At the top of the image the hunter seems to stand in front of the lion to shoot directly at it; their confrontation is face-to-face. At the bottom of the image, however, we discover that the lion has been struck by one arrow in the anal-genital region. In synthesizing the narrative from the pictorial text provided by the image, the viewer must reconstitute the moment of the kill as having involved the hunter's sneak attack from behind. Retrieving an earlier passage of depiction (image of the lion about to be killed) to reinterpret the later passage (image of the dead lion), the viewer literally experiences the "filling in" or "filling out" of the narrative as a forward-and backward-looking temporal movement. For all of these works, the ellipsis is a clear example of the general narrative metaphorics—namely, masking the blow.

It would be fundamentally misleading to isolate individual figures, groups of figures, MOTIFS, or signs in an image and try to identify their meaning or meanings independent of their metaphorical and narrative context. To reiterate a definition with which I started, an image is a concatenation of passages of depiction functioning for a viewer as a referential whole. No iconographic detail in or compositional zone of the pictorial text can be ignored, for almost all details and zones are related to the others as structures of metaphor and narrative. As the example of the lion hunt on the Hunter's Palette shows, if a viewer ignores the sixth arrow depicted as plunging into the lion's anal-genital region, it is then impossible to reconstruct the preceding developments in the narrative (for example, a sneak attack) that, once the viewer realizes their necessity, alter any further understanding of the story and the way it has been realized picto-

rially. Only after such reconstruction do we see that an insignificant, perhaps "descriptive" or even fortuitous detail, the angle of the top hunter's arrow set into his bow, takes on a narrative status with ramifying consequences for the entire viewing experience.

There are only a few possible exceptions to the general finding that the text on these late prehistoric decorated objects is wholly coherent as a pictorial narrative image. First, on the Ostrich and Oxford Palettes (Figs. 25, 26), the "heraldic" or "emblematic" animals forming the top outline of the palette—in one case, the out-facing bird heads, and in the other, the in-facing wild dogs—may be a frame for the narrative image enclosed within it rather than actors in any particular episode of the story. And on the Oxford Palette, the wild dogs depicted clasping paws seem to serve an apotropaic function, apart from their possible role in a narrative relating the peaceful and violent relations among wild animals.

Second, on the Oxford Palette keys to the cipher of the narrative seem to be provided within the image—specifically in the figures of the ostrich flapping its wings and the peculiar, almost surreal, gazelle with double-twisted horns (Fig. 27). These creatures seem to function as symbols denoting the viewing of the narrative image, even though, as pictorial text, they can also be interpreted as episodes in the story of the wild animals' hunt. They must be seen, therefore, as having a double pictorial function—as both pictorial narrative and pictorial non-narrative text. Similar keys to the narrative cipher appear in the image on the Hunter's Palette (Fig. 28), where the ostrich flapping its wings and the leaping hare probably indicate how the viewer must move from one passage of depiction to another in a particular portion of the composition (the central long horizontal zone). Again, the motifs have a double pictorial function: they depict both actors in the story and the act of viewing that story. (Elsewhere on the Hunter's Palette these two functions may be divided between two different passages of pictorial text.) Abstracted from the pictorial elaboration of the story and placed "before" it—namely, at the top of the palette or "beginning" of the text—is the key to the narrative cipher: "shrine" and "double bull" (Fig. 30). These signs depict the action of viewing that allows a viewer to unfold the narrative coming afterward. They may have no other status in the text, although

according to another, not incompatible, interpretation, if these signs designate the locality of an episode within the story, then they may function as pictures depicting that place or as "hieroglyphs" denoting it.

Finally, on the Battlefield Palette (Fig. 33), the image on the reverse, presenting gerenuk browsing in palm trees, while it could be a key to the narrative cipher, may be independent of the narrative image on the obverse. (Or the two images may stand in a symbolic or metaphorical relation to one another.)

These differences among the palettes suggest that the late prehistoric image makers constructed a complex combination of pictorial narrative images with other forms of text, pictorial but non-narrative as well as sometimes altogether nonpictorial. (It is not feasible here to relate these findings to the conceptual and historical origins of Egyptian hieroglyphic writing; a full-scale exploration of the issue would need to refer not only to the late prehistoric images considered here but also to certain parallel problems of textuality and notation in prehistoric pottery decoration and prehistoric rock art.) With the possible exception of the reverse side of the Battlefield Palette, the texts seem to have been used always with a narrative image dominating the visual field and establishing the formal or compositional axes of the whole. And with the possible exception of the Battlefield Palette, no object bears two different texts. In all cases the non-narrative or nonpictorial text used along with the pictorial narrative text is distinguished from it compositionally; it abuts, edges, or frames either passages of pictorial narrative depiction or the whole of the narrative image as such. Notwithstanding the tendency of vulgar iconographers to pluck individual motifs from their referential context, late prehistoric pictorial narrative, then, constructs an uninterrupted text or viewing experience—even though, and perhaps especially because, it unfolds temporally through the very movement of the object itself.

Chasing the View

The visual text of late prehistoric Egyptian representation is principally iconic, a pictorial statement or set of statements of what is going on in the narrative. Although some "hypo-iconic," quasi-symbolic supplements were used by the

image makers, such as "pictures" of actors in the story that also denote the required actions of viewing, the pictorial text, as such, cannot directly recount the spatial and temporal transformations required in the narrative. Strictly speaking, the picture is an image of "what is" in the fictional world but not necessarily of "what happens" in that world (see the Appendix). Therefore a viewer cannot produce a narrative without clear markers for steering through the image and for interrelating its separate elements precisely as connected by chronological or causal transformations. For example, the viewer needs to know that the orientations of individual figures relative to one another are, in fact, iconic—that they provide information about the way figures "move"—rather than the result of other considerations (such as the simple horror vacui often invoked to explain the organization of the compositions). Such knowledge could not be taken for granted in interpreting the image or constructing it as a narrative. All the works depend therefore on what I call general minimum conditions for the intelligibility of the image. These conditions were not revised throughout the chain of replications because altering them would have been so radical, by definition, as to risk rendering the images unintelligible or unrep-licable. But within this purview—the "style," or horizon of unrevisability, defining the replicatory sequence as such—each replication arrives at an expressive text of its own precisely because the conditions of intelligibility are general and minimum ones.

The general conditions of intelligibility are not, in themselves, immutable or outside convention and a history of representation. All are necessarily constituted in a process of replicating—repeating, revising, and refusing—a chain of earlier replications with its own conditions of intelligibility. I examine aspects of this process (see Chapters 2–4) in considering the revision of the animal-rows and carnivores-and-prey formulas on the decorated ivory knife handles and comb (Figs. 8–10, 20–23) and the earliest of the palettes (Figs. 25, 26). At the "beginning" of our chain of replications, as the radical revision of an earlier chain of replications, the Carnarvon knife handle (Fig. 23) exemplifies the risk of confusion and discomposure that we find also at its "end," in the Narmer Palette (Fig. 38). Confusion and discomposure result when the general minimum conditions of intelligibility of the chain of replications as a whole undergo

substantial replicatory revision and refusal; an image made in this context risks being unintelligible. Thus the Carnarvon knife handle can be described as "degenerating" (Bénédite 1918: 235), and it is only because late prehistoric representation has been misunderstood, and later canonical standards taken as the measure of earlier image making, that the Narmer Palette has not been so described as well (see Chapter 8). By contrast, when the general minimum conditions of intelligibility are substantially repeated and sustained in most of their significance, the image maker and the viewer know what is "correct" and any possible mistakes can be seen readily as such.

Just as later canonical image making had its own extraordinarily comprehensive and flexible standards of intelligibility (Davis 1982; 1989), so too did late prehistoric representation. First is the condition of the referential wholeness of the image—that is, of the pictorial and nonpictorial, narrative and non-narrative text spread across the surfaces of these objects. A viewer will get nowhere trying to decipher each and every figure or motif by recalling and quoting their many formal and iconographic parallels, whether earlier, contemporary, or later. Using such a "dictionary," presuming it even existed, would be both unreliable (disregarding the possibility of disjunction) and misleading (treating each motif as a discrete, well-defined "entry" with an autonomous meaning rather than as a term defined only by its interrelations with other terms). On both counts the procedure would ignore the pictorial quality of the image as such. Unlike the elements of natural language, pictorial motifs have no independent definitions listed somewhere that, toted up, give the meaning of the entire image; rather, the viewer takes passages of depiction as the constitutive fabric of a complex reference. If there is a dictionary for looking up the sense of this entity, then it must be a handbook for defining an image as a whole, less a thesaurus than a treatise on allegory.^[1] For our purposes the chain of replications and what I later consider as the "Internal stratigraphy" of the image are its dictionary: they are what the viewer must know something about—a matter of the archaeology of his or her understanding—and must work through to construct the reference.

The viewer's first objective is to identify the image. Here, of course, there is no guarantee of success. The viewer simply may not have access to all the

properties of the image; it may be fragmentary or unavailable in a thousand different ways. Furthermore, the image could fail to fulfill the available, ostensible criteria for its wholeness, the criteria from which it seems to have been projected in the first place. Its intentionality may be substantially distributed and divided; the making may have gone beyond the "Intention" to establish the wholeness known and desired in advance—for example, when in carving the Battlefield Palette, the maker dealt awkwardly with the new notational elements of his text and produced a revisionary image of the human body.

In fact, as my initial brief consideration of intentionality in image making suggests (see Chapter 1), wholeness cannot be a condition of any individual replication. Rather, the meaning of the image lies partly in its backward-looking repetitiveness and revisionism (its refusals) and partly in its forward-looking intention to repeat, revise, and refuse. The wholeness of the image lies distributed in the chain of replications, not in the links.

It is axiomatic that the chain of replications can never be brought into view all at once, or retrieved as an established, available, whole context of meaning. By the same stricture that applies to the meaning of any particular motif or image, the meaning contained in the chain of replicating images can be "reviewed" only by going back through or over it, as well as moving away from it, and only in the terms of the latest repeating, revising, and refusing replication from which the reviewing begins. Even if the reviewing gives a view of the whole chain of replications, in looking at this view in its turn the viewer cannot also look at the reviewing. The wholeness of the image in its chain of replications, then, is entirely notional, an intentional property of the image in the strict sense—something the image maker or viewer thinks about the image—determined, in its turn, by a complex structure of pro- and retrospective beliefs and desires. Image and meaning always come up around each other; and whatever wholeness of image there may be obtains, unviewably, in and as this encirclement.

These very facts—the apparent circularity, ungraspable as such, in which the meaning of an image is constituted—suggest how an image becomes a referential whole for a particular viewer at a particular time and place, always subject to later, ongoing replication. In the temporally and spatially extended

activity of chasing the meaning of the image through the viewing of it, the viewer necessarily stitches together many motifs and figures, interrelating and interanimating them, taking account of them or keeping them in mind. Reviewing the image and its chain of replications is directed—looking forward and looking back—at securing a whole, for only the possibility of attaining it sustains the work of memory and projection, the keeping-in-mind and the taking-account-of, that constitutes the activity of viewing. We could say that the image is a whole without edges, a knot. A knot is tied when it holds itself together: the image is a whole when viewing stitches up, for a viewer, just what can be taken as an image for that viewer—that is, as a view that can be viewed. This meaning-producing event is historical in the pragmatic rather than the syntactic and semantic dimensions of image making; it defines the viewer's actual relationship with the pictorial text, which he or she investigates in a spatial interaction and a temporal duration. To return to an earlier phrasing (see Chapter 1), wholeness is not immanent and transcendent; it is a material condition of the image, a necessary feature of its temporal and spatial structure for the intentionality of a viewer.

Such considerations, in practice, imply that the wholeness of the image can only be a function of some real activity of handling and viewing the object itself. (Despite its importance, this matter is seldom discussed in art histories or archaeologies that analyze images and the historical construction of their meaning.) Of primary concern in interpreting a given chain of replications is that the viewer chases the wholeness of the images made within it by assuming more particular conditions of their intelligibility than their wholeness as such. After investigating these conditions in the construction of late prehistoric narrative images (Chapters 4–6), it is now possible to state them in summary.

Each late prehistoric image requires the viewer to begin with the top of the obverse side of the palette. Although we have no independent archaeological evidence about the actual handling, use, or display of the palettes, it is reasonable to suppose that the "obverse"—or "first"—side of the image is specified for a viewer as the side of the object carrying the cosmetic grounds, which would spill if and when the palette is turned over. The only one-sided image in

the group, the Hunter's Palette (Fig. 28), has the saucer on its decorated side; and in the whole corpus of predynastic palettes there are no examples with saucer on one side and images on the other. The "top" (the "beginning" in the top-to-bottom axis of the story) was probably specified for contemporary viewers by the fact that the routine funerary palettes, from which the decorated palettes derived, oppose an emphasized, three-dimensionally sculpted edge—often ornamented with double bird heads (Petrie 1920; see Fig. 25)—to a plain, tapering edge. The cosmetic saucer is usually closer to the sculpted edge, which in predynastic graves generally faces in the same direction as the head of the dead body (see Petrie and Quibell 1896). Despite these conventions, however, the most literal "top" and "bottom" of a palette or other object is always relative to a viewer's particular orientation, a relation that need not be constant throughout manipulation or viewing; since the palette can be turned in the hands, "top" can become "bottom" and vice versa. A narrative can unfold through this reorientation. Narrative experience, in fact, would not be totally compromised—although it might be rendered more difficult or ambiguous—if a viewer were to begin with the "reverse" and the "bottom" rather than with the "obverse" and the "top." The underlying narrative logic can be reconstructed from the sequence of passages of depiction however they are ordered in the viewer's experience of them. (Distinct from the reconstruction of the underlying narrative, the viewer's experience of the pictorial text as such must vary with his or her point of entry, for example, at the "top obverse" and "beginning" or the "top reverse" and "half-way through." Pictorial narrative can play on this fact. On the Narmer Palette [Fig. 38], it is only by "reading through" all of the pictorial material on both sides that the viewer will discover how the "beginning" of the story on the "obverse top" is actually the logical end of a chronology or causal chain.)

On either side of each palette the image presents either three or four distinct groupings of figures arranged in visually discrete zones or bands of depiction. These zones are linked with one another as the episodes in the story of the narrative or the poles of a metaphor in top-to-bottom, left-to-right, and side-to-side or obverse-and-reverse relations. Typically each zone contains one or

more pairs of actors—the several animal or human figures in the image—engaged in some mutual action or interaction. These zones tend to have somewhat different textual qualities one from the next (for example, in compositional arrangement, pictorial detailing, or use of complementary signs and "hieroglyphs"), although late prehistoric representation is characterized by considerable regularity among zones at the level of the narrative material each presents (for example, all zones usually contain four actors in two pairs).

Within each band of depiction, the nature of the mutual relation between actors is established in three ways simultaneously: first by their placement—at the sides or in the center (or bottom or top, right or left) of the zone; second by their bodily orientation—moving toward, moving beside, or moving away from one another, from or toward the side or edge of the zone, from or toward the center of the palette, and so on; and third by the direction of their glances—looking at one another or looking elsewhere.^[2] These relations may be repeated or revised (for example, reversed or inverted) in other zones; that is, the placement, orientation, and directionality of zones create mutual relations among them. Such interrelations generally establish the linkage of two zones as a continuous segment of the story, whatever its chronological arrangement (for example, as before to after), or as a completed metaphor in the text.

To identify all these relations within and between zones, the viewer is generally required to twist, rotate, or flip the object to produce a series of views of its surface. For example, the viewer can twist the object from "left" side up to right" side up or turn it from obverse to reverse. Views of the passages of depiction thus produced could be generated also by the viewer moving around the object displayed or held in a fixed position. In fact such movement could even be carried out in a viewer's imagination, as in viewing three-dimensional sculpture by conjecturing its appearance on the planes invisible from a given vantage point. If an image can be apprehended in such a notional way, "twisting" or "flipping" an object can likewise be a conceptual rather than an actual physical component of the viewing activity. In either case, we would want to understand the determination of such conditions for viewing, whether real or notional, within the context of the scene of representation.

Depending on where the viewer's hands grasp it, on what can be seen as upright in holding it from that angle, on what disappears from view, and so on, the individual figures within the zones and the relations both within and between zones have different, changing aspects. To find the whole narrative image the viewer literally moves and scans the depictions in a particular order: "An internal transformation and subsequent combination . . . become[s] a summation in time . . . linked to the mobility of its internal elements. . . . The narrative is constructed as the combination of an initial stable state with a subsequent movement" (Lotman 1975; see also the Appendix).

The Real Viewing

Since archaeology has produced no independent evidence about the way carved cosmetic palettes were manipulated and displayed, questions remain about their function and use (see Chapter 1). If, as many prehistorians and Egyptologists suppose, they were votive or ceremonial objects, they may have been handled infrequently; they may have been displayed or stored, and in either case visual and physical access to them might have been restricted for practical or social and ideological reasons. If exhibited ceremonially, they could have been carried or displayed by one, two, or more handlers in such a way that viewers could see both sides of the image. It is also conceivable that only one side of the object was visible at certain times—for example, if eye paint was actually being mixed in the cosmetic saucer on the obverse side—or that it was perpetually hidden from view. In the latter instance the viewer's exposure to a whole image requiring close scanning, rotation, turning over, and so on, was only notional. Like the topmost registers in a Roman triumphal column or the highest stained-glass windows in a medieval cathedral, the image would have existed to satisfy conceptual conditions that could not involve physical practice (see also Brilliant 1984).

Smaller palettes—like the Ostrich (Fig. 25) at 16.5 cm. high and the Oxford (Fig. 26) at 42 cm. high (fully complete, the Battlefield Palette [Fig. 33] would be approximately 42 cm. high)—could be lifted easily and turned

in the hands. The surviving larger palettes, however—the Hunter's (Fig. 28) at 64 cm. high and the Narmer (Fig. 38) at 63.5 cm. high—are too heavy and bulky to be handled in this way. Perhaps, then, although conforming to the conceptual requirements of an image that could also be moved in a viewer's hands, the largest (probably later) replications ceased to have this literal function. Does this suggest that viewers' encounters with such works were more distant or more diluted than with the small replications, that they could not be viewed as intimately and comprehensively? Perhaps—but as noted above, there are ways the image could be viewed without actually rotating or flipping it, for example, by moving around it or even by imagining the necessary movements. For simplicity's sake I regard all the decorated objects in the chain of replications as capable of being handled directly by one person examining the several zones of the depiction in a sequence of twists, rotations, and flips. Of course, at least one person did indeed handle every one of the objects this way—the image maker.

These questions are somewhat beside the main point given that all the objects except the Hunter's Palette present passages of depiction on both sides. Any interpretation of the image must surely accommodate this obvious property. Indeed, we may discover it to have been a constitutive one: it is not incidental but intrinsic to the significance of the works that they require the literal or physical "transposition of internal elements" identified as the essential property of narratives in nondiscrete texts such as pictures (Lotman 1975; see also the Appendix).

It might be said that if the palettes were used to grind eye paint, they could not have been rotated or turned over without risk of spilling the grounds. If we conclude, therefore, that rotation and flipping were actually to be carried out on the objects, it seems we must also conclude that the palettes were nonfunctional. At the very least they could not have been used for their ostensible practical function of grinding and carrying cosmetic. This conclusion, in turn, seems to reinforce the common assertion that the great carved palettes—certainly the largest, heaviest ones—had to have had a ceremonial function; they were otherwise impractical.

The string of inferences thus drawn has no credibility, however. For one thing it is risky to proceed as if we know intuitively what late prehistoric Egyptian society would have regarded as practical or impractical, given its own understanding of the functions its images were intended to have. As stated earlier, an archaeologist's designation of an object as ceremonial frequently serves only as a euphemism, for no one can point to independent evidence that shows how the object was actually used. Since the term ceremonial is virtually empty, we should work toward a specification of function by analyzing the representational scheme or pictorial mechanics.

In fact, in both logical and practical terms the pictorial mechanics I have identified are compatible with the use of the objects, at times, to prepare eye paint—in the process of which, presumably, the palette could not be turned over without spillage. Consider the parallel of a decorated chair. When a person is sitting on the chair, using it for its literal or practical function, any decoration applied to the back of the chair cannot be seen without great difficulty, the chair cannot be tipped over, and the sitter cannot simultaneously move around it. But none of this implies per se that the decorations on the chair are not systematically structured as narratives or other complex images. Certainly the images can still be related to the literal, practical function of the chair or of the sitting carried out in it—for example, as symbols of the person displayed on the chair, metaphors of "sitting" as being settled, being sited, being in the seat, and so forth. So it is too with the decorated late prehistoric Egyptian palettes. It is perfectly permissible to suppose that the general representational or rhetorical function of the objects encompassed both the occasional grinding of eye paint—the ostensible use of the object—and the organization of the images, also "used" in the viewing of them. There is no need even to decide that one of these uses is primary and the other secondary, one literal and the other notional, or one functional and the other nonfunctional or ceremonial. Both are aspects, probably interconnected, of the use of a complex artifact that also happens to be an image—an artifact picturing something—and a complex image that also happens to be an artifact, or an image to be handled.

Open Handling

The particular order of the handling of an object and of viewing the image that accompanies it is determined partly by the physical properties of the object—for example, as in flipping from the obverse to the reverse of a flat palette. Despite such fixed conditions, we cannot dictate a self-evident or natural order for scanning the depictions, some fully obvious route for chasing the image that any viewer, even one uneducated in the conditions of intelligibility of the image, could presume. Consider a viewer looking at the right edge of the composition on one side of the palette about to make a flip to the other side, the right edge coming over the left. Should this viewer follow the gaze and look again at the right edge of the composition on the other side or follow the edge and look first at the left edge on the other side (see Fig. 27)? The viewer's preference will have major consequences for the narrative if, for example, right and left portions of the composition on one or both sides are organized as presenting episodes respectively before and after one another in the story. To take a second instance, if the viewer twists the image in one direction to bring certain figures or motifs into upright orientation, then to unfold the continuing narrative, the image could either be twisted further in the same direction, to the extent of being spun all the way around, or be inverted to bring it back to the original position, in which, for example, other figures or motifs might become upright (see Figs. 28, 31).

The viewer always has options, despite being guided by keys to the cipher setting up more or less constraining specifications for those crucial routes through the narrative without which there would be no narrative. For example, on the Oxford Palette (Fig. 27) the flying ostrich on the obverse top, as a key to the overall cipher of the narrative, seems to instruct a viewer to flip the palette from obverse to reverse, left edge over right edge: the ostrich "flies" from left to right in making its "jump." But this key to the cipher, although establishing a crucial, if general, route for the narrative, says nothing about the place or moment in which the flip from obverse to reverse should be performed. Thus the viewer could begin with the obverse top zone, flip the palette to inspect the reverse top zone, drop down to the reverse bottom zone, and finally flip back to the obverse bottom zone. If both flips are conducted according to

the key, then the whole flipping takes the viewer "there and back again" rather than "all the way around," a fact that can probably be integrated into our understanding of the metaphorics and narrative image as such. But in principle, even following the key to the cipher, the viewer could begin with the obverse top, drop to the obverse bottom, flip to the reverse top, and drop to the reverse bottom. The entire flipping here is an "all of this, all of that." This organization, too, could be understood as having metaphorical or narrative standing. In fact, without any extrinsic, independent evidence or any internal, structural evidence, the Oxford Palette is generally assumed in standard commentaries to work in the latter fashion (for example, Vandier 1952; Kemp 1989)—mostly, I think, because we are heavily influenced by the modern Western habit of reading entire pages from top to bottom before turning the page (compare Winter 1985: 20). (I prefer the first, somewhat more complex, construction. "There and back again," as a way of establishing metaphorical and narrative relationships, is more flexible and comprehensive than the simple structuralist opposition "this, that," which "there and back" can subsume within itself. But I do not think it is necessary to make an absolute choice.) To reiterate, in all these situations, the viewer's decision is guided, if not decided, by two considerations—first, the way figures in one zone direct the gaze by their bodily placement, orientation, and glance into one place rather than another as the "next" zone (for example, the arrow on the Hunter's Palette [Fig. 28] moves us down and to the right), and second, the operating instructions provided in the text by individual motifs functioning dually as actors in the story and keys to the narrative cipher.

No one element in the pictorial text can always be assumed to provide helpful information or any information at all. Moreover, the use of an element one way in one replication does not guarantee its use in the same way thereafter. The general minimum conditions of intelligibility, then, guide viewing sufficiently for the viewer to chase the image; whether it can be captured is another matter altogether. As I have suggested, the movement of chasing will itself stitch up the elements of a possible image—and the narrative is in the motion.

In late prehistoric Egyptian representation, it seems a viewer is quite free to scan the zones in various possible sequences depending upon the interpreta-

tion and weight given to this or that placement, orientation, glance, actor, or cipher key. Especially because late prehistoric representation does not draw hard and fast lines within and between passages of pictorial text, it allows different and to some extent diverging possible sequences and cross-viewings. The viewer is not forced to construct one passage of text as overshadowing the rest or one episode of the story as summarizing everything else. Even the Battlefield Palette (Fig. 33), which noticeably magnifies the scale and compositional centrality of a principal group, conforms to such order; if more of this palette were preserved, the dominant motif would probably seem less dramatic. Different possible sequences followed out produce different possible story arrangements and textual renditions of the underlying narrative, for example, by putting this or that zone "before" or "after" another. As a correlative, metaphorical relationships in the image are construed in different ways depending on how the metaphor is distributed textually in the possible sequences—for example, one pole of the metaphor, like the "human" or "animal," being "before" or "after" or "on top of" or "below" the other. The apparently deliberate openness of the late prehistoric image has been called "chaotic" and "disorganized" (Groenewegen-Frankfort 1951; Fischer 1958), but this judgment derives from an experience with later Egyptian art—in which boundary, weighted and unipolar hierarchy, unidirectional narration, and closure have become canonical. As an intelligible artifact the Narmer Palette (Fig. 38) stands as the watershed image—although not necessarily fully successful as such—between the late prehistoric and the canonical scenes of representation.

8—
About-Face

The Narmer Palette (Fig. 38) refuses and replaces as much as it repeats and revises the chain of replications of late prehistoric image making. On the one hand it replicates the narrative images on the Oxford, Hunter's, and Battlefield Palettes (Figs. 26, 28, 33); it has thoroughly assimilated late prehistoric image making. But on the other hand it revises some of the general conditions of intelligibility of late prehistoric images, conditions that had established them as disjunctive replications of a common pictorial metaphorics and narrativity.

A Link in the Chains

The ambiguity in the Narmer Palette—in itself appearing to be an effort to control the ambiguity of late prehistoric image making—makes the image on the palette difficult to understand. In principle it could be interpreted as a replication within two different chains—namely, that of late prehistoric image making and that of the subsequent tradition—the palette being an early, if not the earliest, replication in what would emerge in the next two hundred or more years as the canonical tradition (for First and Second Dynasty work, see Davis 1989: 159–71, 179–89). Neither placement of the image in its "serial position" in an art history (Kubler 1962) would be incorrect, but neither is complete or sufficient. While it does revise the late prehistoric chain of replications, it would have to be replicated in and for itself—a chain of replications occurring only after the Narmer Palette had been made and necessarily assuming it—for this revision to achieve the status of an alternative tradition. The replication of the revision transforms it into the conditions of intelligibility on which a coherent, if always necessarily disjunctive, iconography can be founded; thus we

Fig. 38.
Narmer palette: carved schist cosmetic palette from Main 
Deposit at Hierakonpolis, early First Dynasty. 
Courtesy Cairo Museum.

can say not that the Narmer Palette performs this transformation for itself but that later images do so retroactively by replicating it. Our difficulty in understanding the image, however, is attributable not only to its peculiar place in the chains of late prehistoric and canonical image making; it is also possible that the narrative on the palette did not fully amount to an image, for its composition as an image in both the late prehistoric and the canonical chains of replication is seriously disturbed.

On typological grounds the Narmer Palette is usually assigned by Egyptologists to the Horizon A /B or early First Dynasty—that is, to the shadowy era of the synthesis of late prehistoric polities in a dynastic state ultimately ruled by national monarchs (see Trigger 1983; Kemp 1989). Narmer himself seems to have been one of these kings; his name appears in various other contexts (see Edwards 1972: 1–15; the stela from Tomb BIO at Abydos, attributed by Petrie [1900–1901: 1, pl. 13, no. 158] to Narmer, is now thought to belong to King Hor-Aha at the beginning of the historical First Dynasty [Kaiser and Dreyer 1982: 213–191]). It is unlikely, however, that the historical Narmer was solely responsible for the so-called unification of all Egypt, and if the palette must be regarded as a representation of real historical events—at best this would be a partial interpretation of its metaphorical and narrative structure—it could recount a local conflict, for example, between "Narmer" and neighbors to the immediate north or south. The simplest historical interpretation puts the Narmer Palette later in an absolute chronology than the other palettes examined here and defines it as the type of representation appropriate to the interests of the emerging state.

Like the canonical tradition of image making, the national dynastic state cannot have emerged all at once. Pre- or nonstate image making would have continued to exist, at the same time state-affiliated images were being produced, in different centers of social life—for example, in small villages, in regional polities not yet absorbed into and possibly even resisting nationalization, or among competing elites. Thus the Battlefield Palette (Fig. 33) and the Narmer Palette could be regarded as contemporary—on typological and stylistic grounds, this is perfectly possible—although they were conceived in different ways. For expository convenience we can regard the Narmer Palette as later

than the other images and as belonging to a different social and cultural formation, but we do not yet know how one merged into another. The Narmer Palette is itself part of the evidence for this process.

From Before to After:
Replicating Late Prehistoric Representation

The top edges of both sides of the Narmer Palette are decorated with frontal heads of bovids, two on each side, possibly depicting one of the cow goddesses, Hathor or Bat, of later mythology (Vandier 1952: 596; see also Baumgartel 1960:91). There is nothing to justify the anachronistic parallel, or indeed the supposition that the bovids are female (although according to the anachronistic interpretation Hathor, as female, should be represented without a beard). It is more consistent with the general narrative metaphorics of the entire palette to take the heads as human-faced bulls. Like the ruler's force "as a lion" on the Battlefield Palette, the ruler's power "as a bull" is figured within the pictorial text: on the obverse bottom, a great bull attacks an enemy's citadel, and on the reverse middle the ruler, smiting his enemy, is depicted wearing a bull's tail in a handle ranging from his belt, decorated also with four bulls' heads—possibly to match the four at the top of the palette—affixed to what appear to be strips of beaded cloth. Although subtle differences among the four heads on the top of the palette have often been remarked, it is unclear how these might be reconstructed in the image as a whole; perhaps they are merely variations in textual morphology that do not bear on the metaphorics. The heads flank a palace facade (serekh ) containing the nar -fish (a species of catfish) and mer -chisel, signs placed also before the largest figure in the obverse top zone, apparently naming the ruler himself. The entire decorated top edge of the palette is sometimes interpreted as depicting a roof or canopy for a facade with bucrania; although not necessary for our purposes, this reading accords with other features of the image.

Below the decorated top edge, set off by a thin register line, the two sides of the palette are divided into three horizontal zones. Except for the "rebus"

(on the reverse top right), each zone on the Narmer Palette containing a representation of the ruler has a register ground line—namely, the obverse top, with the ruler and his retainers inspecting the corpses of ten defeated enemies; the obverse middle, with the ruler's two retainers mastering two serpopards; the obverse bottom, with the ruler-as-bull vanquishing an enemy; and the reverse middle, with the ruler smiting his enemy. By contrast, ground lines do not underscore enemy figures at the obverse top right, the obverse bottom, and the reverse bottom. In each of these zones the enemies are depicted as defeated, twisted on and from their baselines and upright axes, an image presented in another way on the Battlefield Palette. The special place of the enemy on the ruler's ground line in the reverse middle zone will concern us later. The registers therefore mark the different status of figures within the image as being either associated or not associated with the ruler. Thus the Narmer Palette proffers the latest version of the animals' and the hunters' differing grounds on the Hunter's Palette; moreover, as on the Battlefield Palette, the animal enemies of the hunters have become the human enemies of the ruler. In addition to marking and maintaining distinctions indicated in earlier images in the chain of replications, however, the registers on the Narmer Palette divide each zone from the others on that side of the Palette. As elements of composition, passages of visual text, they separate the zones—and thus presumably the elements of the story—in a way that is not presented on the late prehistoric images we have examined so far.

A viewer whose visual culture included the chain of replications of late prehistoric representation would presumably seek to interpret the Narmer Palette in the same way the Oxford, Hunter's, and Battlefield Palettes (Figs. 26, 28, 33) would have been viewed—that is, as a narrative image. Accordingly, on the Narmer Palette (Fig. 39) each side presents a zone depicting what comes after (at the top) and what precedes (at the bottom) the ruler's decisive blow of victory, with the obverse depicting its aftermath and the reverse its preconditions. The story is therefore arranged in the four top and bottom zones of the palette, top-to-bottom/obverse-to-reverse! as consistently after-to-before.^[1] The pictorial text, however, also includes elements relating material in the other temporal direction, from before to after (strictly speaking, a counternarrative),

Fig. 39.
The structure of the narrative image on the Narmer Palette.

as well as elements resisting or breaking out of the narrative sequencing altogether. In view of this complexity it is useful to go through the zones one by one before considering their metaphorical and narrative interrelations.

Beginning with the obverse top, the victorious ruler, wearing the Red Crown (it would later signify Lower Egypt, but we cannot necessarily assume this denotation here) and labeled with the two hieroglyphs of his name, is preceded by four standard bearers. He is flanked on the left by his sandal (and seal?) bearer and on the right by a figure most commentators have identified as a "priest" carrying scribal equipment. Carrying the seal's case(?) around his neck and the sandals of the ruler, presumably strapped to his wrist, the sandal bearer is labeled with a rosette and pendant plantlike form. The image on the ruler's seal is apparently depicted above the sandal bearer to the left in the upper left

corner of the zone. Although its status as a seal imprint has not been recognized, the central form in this sign has been tentatively identified as the "reed float" used by Old Kingdom sportsmen to hunt hippopotami (see Baumgartel 1960: 92–93) or the "bilobate Khons sign" (Williams and Logan 1987: 248, note 13); more likely it is a schematic depiction of the ruler's sandals with the strap the sandal bearer uses to carry them around his wrist (see Vandier 1952: 596). The "priest" is labeled with a rope noose or tether and a loaf. There are reasons to suppose that these two personages have special narrative significance; it is even possible to interpret them as members of the ruler's family. The entire group of the ruler and his followers inspect two rows of five decapitated enemies spread on the ground (without ground lines) at right angles to the progress of the victors. The several signs or hieroglyphs placed above this group may or may not be read like later hieroglyphs. One depicts a door on its pivot; it could be taken as the hieroglyph "great door" (= wr -swallow + door; the bird is to the right of the door)—for example, the door of the temple where the palette may have been dedicated or the door of the palace before which the enemies are displayed. Another depicts a falcon, possibly denoting the ruler, carrying or surmounting a harpoon, probably denoting the enemy shown on the reverse of the palette, where he is labeled (with this sign and a sign depicting a pool of water) as "harpoon" or "coming from harpoon territory," presumably a swampy or marshy region. (In both contexts in the pictorial text, the harpoonlike sign is difficult to decipher; the object may be one of the smaller implements—such as bone and ivory pins or hooks—commonly found in predynastic funerary assemblages. For convenience, the label "harpoon" is used here.) A third depicts the "sacred bark" of the ruler. (For possible readings of these signs, see Kaplony 1966; Williams and Logan 1987.) Considering that the obverse top of such images has so far always been constituted as the beginning of the narrative text, the first pictorial text encountered on the Narmer Palette, then, concerns the achievement of the king, the celebration of his victory.

What precedes this episode is related in the top zone on the reverse side of the palette. Here a rebus depicts the enemy or enemies brought to the ruler by a falcon—an aspect, double, or representation of the ruler or perhaps his divine protector. The falcon inserts a hooked cord into an enemy's nose, possibly to

prepare him for the beheading whose aftermath is related in the obverse top zone, where the king inspects decapitated corpses. It is sometimes said that the six stalks of papyrus growing from the enemy's body—treated abstractly and looking like the later hieroglyph for swampy or watered land—signify "six thousand [enemies]," on the basis of a presumed parallel between the papyrus and the lotus hieroglyph (= numeral 1,000), but there seems to be little justification for the equation. Instead, the papyrus may denote the enemy's home territory, Papyrus Land (Vandier 1952: 596). However it should be interpreted, the rebus indicates in a general sense that the ruler—in his aspect as (or with the protection of) Falcon—has defeated his enemies and prepares them for their judgment or destruction.

Continuing the progression backward through the narrative chronology, the scene preceding this one appears on the obverse bottom, where a bull uses its horns to break down the walls of a fortress evidently inhabited by the enemies, one of whom is thrown to the ground, facing away from the bull, and trampled. Like the falcon on the reverse and like the lion on the Battlefield Palette (compare also the bulls on the Louvre Bull Palette, Fig. 37), the bull can be taken as an aspect, double, or representation of the ruler, although he should not be identified with the ruler's human person as such. Within the encircling outer walls of the fortress is a smaller structure with two towerlike sides; it might be equivalent to the sign used to label the right enemy in the reverse bottom zone, which probably depicts a cattle pen or storehouse, but because of its size might indicate the enemy leader's residence. Three small squares in front of the fallen enemy's face and below Bull's horns are difficult to interpret. They may represent fragments of the walls of the fortress broken down by Bull; on the Libyan or Booty Palette (Fig. 53), probably produced as a replication of the Narmer Palette or similar images, they resemble small rectangular houses within the walled towns. This zone seems to depict the moment in the battle when the enemies' defeat was secured. In his aspect as (or with the protection of) Bull, the ruler enters their citadel.

Finally, what comes before this episode is related on the reverse bottom; in the Narmer Palette's sequencing of the episodes of its narrative, the chronological and causal beginning "concludes" the text. Here two enemies are depicted

as fleeing toward the right, that is (returning through the text to the preceding stage, on the obverse bottom), toward the citadel in which they will take refuge to no avail. They look back over their shoulders as if being pursued by the ruler. Although the ruler does not appear in the pictorial text of this zone of the image, the larger text is clear about his presence; the viewer already knows (having seen the obverse bottom) that Bull will break down the enemies' citadel, a building the text had depicted as a broken version of the sign here used beside the left enemy figure to label him. Furthermore, the enemies' glances over their shoulders are directed upward to the figure of the ruler smiting his enemy presented immediately above this zone. Although coming last in the image as it would be viewed in the context of existing late prehistoric representation, at the very bottom of the reverse side of the palette, the narrative "begins," then, with the ruler pursuing his enemy.

Almost all commentators agree that the sign for the left enemy on the reverse bottom depicts or denotes a fortress with bastions. Werner Kaiser (1964: 90) took it specifically to name the northern town of Memphis, presumably, at this period in protodynastic history, an enemy of the Upper Egyptian polity in or for which the Narmer Palette is usually thought to have been made. (This latter assumption is based on the fact that the principal figure of the ruler in the image, in the reverse middle zone, wears the White Crown, in later iconography the symbol of Upper Egypt; that the palette was found at Hierakonpolis in Upper Egypt might be taken as supporting evidence, although the context speaks only to the final deposit of the palette and not to its manufacture and use.) The sign for the right enemy was interpreted by Yigael Yadin (1955) as depicting a "desert kite," a fortified enclosure used by much later desert herders in Transjordan to protect their flocks. (While kites might have existed in prehistoric times, there is no archaeological evidence for them [see Ward 1969: 209]; more likely the structure had a more all-purpose status as a farm building or storage house.) Yadin's reading led him to see the enemies on the Narmer Palette as Western Asiatics and the entire palette as documenting an Egyptian invasion of southern Canaan-an interpretation that could be sustained even if we reject his reading of the sign. (Alternatively, Kaiser read the sign for the right enemy tentatively as denoting the delta town of Saïs, presumably a twin

of "Fortress," or Memphis, in its struggle with Upper Egypt; other readings have also been proposed [see Kaplony 1966: 159, note 190]. It could be a depiction of the knotting of the loincloth worn by the enemy in the reverse middle zone.) If all three signs labeling the enemies on the Narmer Palette, in the obverse and reverse bottom zones, refer to their actual hometowns, then, in Kaiser's (1964) view, the palette could document a contest between Narmer of Upper Egypt and an eastern delta coalition in which the principal enemy, Buto, or the "harpoon" enemy on the reverse middle (Newberry 1908; Vandier 1952: 596; Kaiser 1964: 90), was assisted by Memphis and Saïs.

But I question the wisdom of forcing the Narmer Palette into serving as a document for actual historical events, and of "reading" its images and signs chiefly on the basis of anachronistic parallels with the meanings of particular hieroglyphs in Middle Egyptian script, in order to reconstruct it as an annalistic or commemorative "statement," somewhat like later kings' chronicles, of Narmer's victory. (Some of the signs on the palette have parallels in Archaic Egyptian script, but the denotations of many of these signs are not definitively known and they largely postdate the Narmer Palette. "Reading" the Narmer Palette's signs as hieroglyphs could be greatly elaborated; only a small selection of the possibilities have been noted here.) In some cases the signs certainly have replications in hieroglyphic script, which could be seen precisely as evolving from late prehistoric pictorial narrativity as it is replicated and revised on the Narmer Palette; the palette is complex precisely because it can sustain interpretation in terms of both late prehistoric image making and later canonical representation. But historical method dictates that we interpret it according to the former, the conditions of legibility apparently assumed or accepted by its original maker and viewers, rather than the latter, a mode of intelligibility not completely constructed until the Third Dynasty at the earliest, a full two hundred years later. Moreover, neither a "decipherment" of some of the images and signs on the Narmer Palette as (later) hieroglyphs nor an interpretation of the image as documenting historical events can, in itself, allow us to grasp the scene of representation in which this ostensibly annalistic and documentary statement was produced.^[2] For example, even if specific references for the image were found in actual events, they alone could not explain their representation

metaphorically or narratively in the image as such; the scene contains much more than the literal denotation—whether depictive or hieroglyphic—of individual figures, motifs, and signs. For example, to represent or recount the contest and defeat of the enemies, the maker of the Narmer Palette replicated a motif, the ruler "smiting," that had been produced in other historical contexts of replication presumably different from the context of the "wars of unification" and well before even the most preliminary formation of script—namely, the conflict scene depicted in the Decorated Tomb painting at Hierakonpolis (Fig. 5) some 250 years earlier than the palette. Again, in deciding to use Bull to stand for the ruler, the maker was influenced in the chain of replications by earlier or other representations of the ruler in the aspect of a great animal. His choices in replicating the existing metaphorics of "looking backward" date to the Oxford Palette at least. In all these cases the meaning of figures, motifs, and signs on the Narmer Palette must have derived, in context, from this earlier history of signification; an appeal to the later meaning of the images—in hieroglyphic script or canonical art—besides being unnecessary, is likely to ignore the inherent structure of disjunction in the replication of symbol systems.

For the moment, then, it is best to leave open the question of the specific denotation of the signs labeling the enemies; we can take them as generic labels for a "harpoon" enemy and two others, "fortress" enemy and "desert-kite" (or "knot") enemy, all of which could have had a symbolic or metaphorical as well as a literal significance for contemporary viewers. Whether the ruler, in his battle with "Harpoon," "Fortress," and "Desert Kite," is conquering new territory, suppressing a revolt, or carrying out some other action is difficult to determine (although many commentators would take the image to denote the territorial expansion of the Upper Egyptian polity, whether to the north or to the south, in the so-called unification of Egypt). The pictorial mechanics of the image—independent of any "meaning," probably irretrievable, it might have had for certain viewers—probably require only that the ruler's activity be construed as a victory with certain particular symbolic features. In fact, it was precisely because a metaphorical and narrative system of representation had evolved in the chain of replications of late prehistoric image making that any

particular activity of any individual ruler could be rendered intelligible to viewers. Whether or not episodes of actual conflict that had taken place or were taking place at the time the image was made sustained interest in the replication, the visual text could be replicated in later image making with or without reference to the "real" denotanda of any particular, individual replicatory version. (For the continued replication of the "smiting" motif in canonical image making, see Davis 1989: 64–68; Hall 1989.)

Being throughout

If we invert the order of their arrangement in the image, the four episodes presented in the four top and bottom zones of the Narmer Palette can be reconstituted as a narrative chronology with an implied causal armature (Fig. 40). I label them as Episodes One through Four, going from the implied causal beginning to the implied ending:

Although the direct blow of the ruler is not literally presented in any of the relevant passages of depiction, it links each of the episodes in the story chronologically and causally to the next. Reading from the "beginning" of the narrative (presented at the "end" of the pictorial text), the ruler must—between the first two episodes of the story—successfully track the fleeing enemies, locate their citadel, and scale or undermine its walls; we are shown the enemies fleeing (Episode One) and the citadel, already taken, with the enemies being trampled (Episode Two). Between the second and third episodes of the story, the ruler or his army must round up the conquered inhabitants and prepare them for judgment and imprisonment; we are shown the town being overrun just after the walls have been demolished (Episode Two) and Falcon mastering the captives taken (Episode Three). Between the third and fourth episodes of the story, the

Fig. 40.
Four episodes in the fabula of the narrative on the Narmer Palette.

ruler or his executioners must lop off the heads of the enemies condemned to die or to be sacrificed in the victory ceremonials; we are shown Falcon lifting an enemy's head (Episode Three) and the ruler inspecting the decapitated corpses (Episode Four).

The ruler's decisive blow must occur, then, several times in the narrative. Indeed, it animates the narrative chronology and causality (in narratological jargon, the fabula), for it is the transition, the event or process, that takes the story inexorably from one state of affairs to another. (Readers unfamiliar with the narratological concepts fabula, story, and text may wish to consult the Appendix before reading further.) When the viewer flips the palette from obverse top, where the text "begins," to reverse top (that is, goes from Episode Four to Episode Three), despite the logical progression to the rebus in the reverse top zone, the most prominent feature occupies the reverse middle zone. Here, accompanied by his sandal bearer and wielding a mace, the human person of the ruler—wearing the White Crown and other regalia—is about to strike his enemy, labeled "Harpoon." The figure of Narmer is presented at larger scale than any other figure in the image, larger, in fact, than any figure in the entire known chain of replications of the image of the ruler's victory. The rebus in the reverse top zone appears almost squeezed in. In the next flip from reverse top to obverse bottom (from Episode Three to Episode Two), the picture of the ruler smiting is replaced by what seems to be its metaphorical equivalent—namely, the picture, in the obverse middle, of the ruler's two retainers mastering two serpopards. (The retainers are distinguished by hairstyle and dress both from the ruler's other followers, such as the standard bearers in the obverse top zone, and from the ruler's enemies, such as the fleeing or crushed figures in the obverse and reverse bottom zones, but it is not clear what these differences are supposed to indicate. The men look somewhat like the curly-haired, bearded enemies of the lion-ruler on the Battlefield Palette [Fig. 33], although they clearly do not have the same status in the narrative.) If the ruler is Falcon and Bull, the serpopards are possibly "Fortress" and "Desert Kite," or, in another popular (but overliteral and anachronistic) interpretation, the two territories of the Nile valley, Lower Egypt (Red Crown) and Upper Egypt (White Crown), united by Narmer (see Gilbert 1949). Finally, in the flip from obverse bottom

to reverse bottom (from Episode Two to Episode One), the picture of Narmer smiting his enemy, removed from view after the first flip, reappears. Thus, no matter how the viewer flips the palette to determine the relation between the zones of depiction spread across it, Narmer smiting his enemy continually appears in the text.

Assuming the general condition of intelligibility of such images to be that the viewer must begin with the top of the obverse, the most logical sequence—Episode Four (obverse top) to Episode One (reverse bottom)—is an inversion of the "natural," forward chronological and causal direction of the fabula. For this reason the viewer is encouraged to retrieve other sequences as well. In investigating these alternatives, the one completely non inverted sequence, which would present the fabula in the order of its forward logic as Episodes One-Two-Three-Four, would be the most difficult for a contemporary viewer to attain. It would require the strongest violation of the established conditions of the intelligibility of images, for it would mean beginning with the reverse bottom of the palette (Episode One) rather than the obverse top (Episode Four) and scanning bottom to top rather than top to bottom.

Other possible sequences and segmentations—other possible "stories" working with the underlying narrative material—would arrange differently the chronological and causal armature of the fabula as presented in the "natural" sequence of Episodes One-Two-Three-Four. All of these fall somewhere in between the two possibilities respectively closest to and furthest from the established conditions of intelligibility for viewing images in the chain of replications (the order Episodes Four-Three-Two-One and Episodes One-Two-Three-Four, respectively). And all of them seem to be more or less equally available options for a viewer handling and scanning the six separate zones of depiction on the two sides of the Narmer Palette. For example, because on the Narmer Palette the fabula is consistently storied after-to-before in two dimensions of the viewing of the text, top to bottom and obverse to reverse (Fig. 39), it can be coherently interpreted—depending on which dimension a viewer selects to pursue first—as breaking into parallel sets of episodes in turn possibly to be regarded as simultaneous, successive, or metaphorical in relation to each other. Thus the obverse can be viewed from top to bottom as a first two-

episode segment of story (Episodes Two-Four: ruler as Bull captures enemy town and decapitation of enemies). This segment would be succeeded, after flipping the palette, by a second two-episode segment of story from top to bottom (Episodes One-Three: enemies fleeing and Falcon's presentation of the enemies of the ruler). The transition between the two episodes in each set and between the two sets is appropriately depicted as the ruler's blow, in the pictures of the retainers mastering serpopards (obverse middle) and the ruler smiting his enemy (reverse middle). This particular sequence takes the form of Episodes Four-(blow)-Two-(blow)-Three-(blow)-One.

As an alternative arrangement, these two sets—obverse top to bottom and reverse top to bottom—need not be regarded as causally or chronologically successive episodes, or "chapters," of the story. The obverse and reverse may simply be parallel recitations of the same story: Episodes Four-Two (obverse top/obverse bottom) = Episodes Three-One (reverse top/reverse bottom). Here the same story has a different, distinctive visual text in the case of each set of two zones, which thus stand in a metaphorical relation. For example, in the obverse story (Four-Two) the ruler appears as Red Crown, Flail, and Bull, while in the reverse story (Three-One) the ruler appears as White Crown, Mace, and Falcon. Again the two parallel stories each depict the ruler's blow.

At first glance this general finding—that several possible story sequences can be constructed by a viewer chasing the narrative image—seems to be consistent with what we have already seen of the open texture of preceding late prehistoric narrative images in the chain of replications. In one sense the presence of the ruler's decisive blow is masked by the ambiguity of the text; in another sense the ambiguity of the text is progressively resolved as centered in the ruler's blow.

Compared with earlier replications, however, the Narmer Palette exhibits many more possibilities for constructing the narrative image. For one thing the composition of the image presents more clearly separate zones of depiction than had been produced earlier, multiplying the possible choices for segmentation, grouping, or sequencing. Moreover, in its placement, scale, and register framing, the image repeatedly brings the viewer back to the picture of the ruler wielding his blow. Whereas in earlier images the text elided or masked the

Fig. 41.
Detail of Narmer Palette, obverse: ruler's retainers mastering 
two serpopards. 
From Wildung 1981.

picture of the ruler wielding his blow, the Narmer Palette, on first glance, is more explicit: the human person of Narmer is about to smite his enemy. It is as if the very elaboration of the viewing activity requires a stronger, more direct and mimetic presentation of exactly what is being or is to be viewed—a closure of the openness in the image. Because the viewer chases the view, the view must now chase the viewer. The Narmer Palette both exemplifies and figures this encirclement, this tightening of the narrative noose—and to that extent masks it. But before we investigate this matter, there are other elementary structural features of the narrative organization of the image to be considered.

With Two Hands

The text provides the viewer with a key to the cipher of the narrative image at exactly the place where the palette would be both grasped and flipped—namely, the depiction of the ruler's two retainers mastering the two serpopards (Fig. 41). They represent the ruler's as well as the viewer's own blow in viewing the image: the two serpopards are the two sides of the palette to be twisted around one another in the viewer's hands. Whatever their role as metaphors for the ruler's victory, the figures of the retainers, textually associated with the

viewer's own hands, are depicted as moving in a clockwise circle (from obverse flipping right hand over left hand toward reverse), gradually twisting together the long necks of the beasts from their heads down to their bodies. The palette must be viewed by starting with the obverse top scene and always flipping the palette right hand over left in the same direction and always scanning top to bottom.

If the viewer adopts and, without interruption, continues to follow this means of chasing the view (Fig. 42), then a simple continuous twist will be generated. It can be summarized as follows:

The viewer's own interruptions of this twist, and the order of the story episodes that it unfolds, might include retrieving the sequence as close to the logic of the fabula as the text provides if the general conditions of intelligibility are still to be observed (that is, Episodes Four-Three-Two-One in the story for One-Two-Three-Four in the fabula). Alternatively, violating the general conditions, the viewer might retrieve sequences that follow the logic of the fabula even more closely still (for example, One-Two-Three-Four in the story for One-Two-Three-Four in the fabula).

In contrast to these possible, and partial, sequencings, the full twist (Fig. 42) both obeys all the general conditions of intelligibility of prehistoric image making and, given those conditions, takes the viewer through all the

Fig. 42.
Sequence of flips in reading the Narmer Palette, obverse and reverse.

available sequences of all zones of the depiction. On completing this twist at the obverse bottom, the viewer—presuming a continuing observance of the general conditions of intelligibility—would repeat the same full twist over again, beginning another series of flips. This simple, continuous twist allows the viewer to construct what I call the "canonical" form of the narrative image available in the Narmer Palette. Obeying all the general conditions of intelligibility of this mode of image making, it incorporates all possible forms of sequencing available within that context. Returning to the obverse, whence the construction of the narrative began, the viewer progresses through this canonical pictorial text making ten complete flips of the two-sided palette. An appropriate metaphor for this activity, or a key indication for adopting it, might be the two rows of five decapitated enemies, ten in all, depicted in the obverse top zone where the text "begins."

Emphasis and Ekphrasis

We have examined so far the arrangement of the fabula in the story presented by the image—especially the basic, "canonical" presentation (Fig. 42)—simply in terms of the connectedness of the four episodes. But the properties of the pictorial text invest the canonical sequence (and, in different ways, other possible sequences) with cumulative properties—what might be called a narrative "tone." The most important element of the pictorial text is the continual reappearance, the underlining, of the ruler standing firm on his ground line and stabilizing the "center" of the image on a horizontal axis—the pictures of the ruler's blow—as the palette continually rotates around its vertical axis. Moreover, although the top and bottom zones on the two sides are all episodes in the story, they are not all textually alike.

For example, consider the textual qualities of and relation between the first two logically closest episodes in the palette's inverted storying of the fabula (that is, the properties and relations of Episode Four, obverse top, and Episode Three, reverse top [Fig. 39]). In the canonical form of the image (Fig. 42) and in the inverted storying of the fabula the viewer retrieves in accepting the canonical form, Episode Four precedes Episode Three. In fact, in the canonical

from and any other sequencing that still obeys the general rule of beginning with the obverse of the palette, Episode Four, the obverse top zone, "begins" the text, while Episode Three, the reverse top, turns up later and necessarily after a flip of the palette. Already, then, the text of Episode Four, the depiction on the obverse top zone, must prepare the viewer for the text of Episode Three, the rebus on the reverse top. In the pictorial text of Episode Four the ten decapitated enemies, depicted frontally, and the ruler's retinue, depicted in profile, literally ground the viewer's point of view: the viewer is placed above and looking down on the enemies while being on the same level as and at the side of the ruler's group. The picture is a graphic depiction of the human person of the ruler with crown, costume, and regalia, his chief retainers, and his army as well as the specific fate of the enemies, with many attributes of costume and individual gesture carefully rendered. Some of this ekphrastic detail has a narrative status. For example, the decapitated heads of the enemies must be connected with the depiction in the rebus—"later" in the viewing of the narrative, if "earlier" in the fabula—of Falcon lifting an enemy's head for him to be executed. Some of this description, however, according to the logic of pictorial textuality, is superfluous for the purposes of the narrative (see the Appendix). Compare the textualization of similar narrative material in the rebus, pictorially the least explicit of all the zones of the image, where ekphrasis is at a minimum. To represent the ruler's many enemies the maker used a single enemy head, without details of body, and plants possibly symbolizing the enemy territory of Papyrus Land. In fact, as a rebus, this passage of the text on the Narmer Palette modulates pictorial textuality into another species of textualitY altogether, quite close to if not identical with ideographic script.

Other zones in the depiction stand between these two relatively extreme possibilities of greater or lesser descriptive richness in the pictorial notation. Three quasi-hieroglyphic zones of pictorial textuality (obverse middle, obverse bottom, and reverse top) can be distiguished from three descriptive zones (obverse top, reverse middle, and reverse bottom). In the descriptive zones pictorial text is supplemented with symbolic labels (the signs for Nar-Mer, Harpoon, Fortress, Desert Kite, and so forth)—that is, with nonpictorial text—but such symbols are not to be expected in the "hieroglyphic" zones (cipher key, Bull

and citadel, and rebus), which are nearly or completely modulated into nonpictorial text anyway.

These and other textual differences among passages of depiction necessarily play out in the way the narrative image is interpreted by the viewer. For example, in the canonical construction of the image the viewer encounters Episode Two (Bull attacks citadel), in the obverse bottom, after the descriptive Episode Four (ruler inspects corpses), the obverse top, and before the hieroglyphic Episode Three (Falcon presents enemy), the reverse top (Figs. 41, 42). In this midway position Episode Two seems pictorially more descriptive than a group of hieroglyphs or a rebus: its figures of Bull and enemy are visually separated one from the other, and the hieroglyph labeling the hometown of the enemies on the reverse bottom is treated here as an actual picture of their citadel. But the same Episode Two on the obverse bottom is less a pictorial description than is the picture on the obverse top, where the ruler inspects his enemies. Therefore, in the first segment of the canonical sequence (obverse top-obverse middle-obverse bottom-reverse top = Episode Four-[cipher key/picture of blow]-Episode Two-Episode Three), the narratiVe image begins with a graphic, vivid look at the "end" of the fabula with the defeated, decapitated enemies (obverse top); next, gives the viewer the key required for unfolding the story leading up to it, the retainers strangling the serpopards (obverse middle); next, goes back in sketchy detail to a much earlier episode, the taking of the town (obverse bottom); next, telegraphs the connection between the taking of the town and execution of prisoners, presented in the two episodes already viewed, should the viewer fail to infer it, and simultaneously anchors further material it introduces (the rebus in the reverse top); and so forth. As in experiencing modern novels and films, the viewer-reader progresses through paragraphs of dense description for and brief summaries of events nested in an armature of foreshadowings and recapitulations—the montage editing of long or short takes in close-up, deep-focus, and zoom perspective, with greater or less clarity, color, and so on.

It is not feasible to continue the description begun in the previous paragraph of the Narmer Palette's narrative image, enumerating all the textual properties of every possible segmentation and sequencing. We should, however,

at least notice how the four episodes of story at the top and bottom of the two sides of the palette are related to the two central passages of text, those with the narrative function of seeming to picture the ruler's blow that the fabula requires as taking place throughout the story—from pursuing the enemies to taking their town, to rounding up the prisoners, to executing them. One passage, with retainers and serpopards, is more hieroglyphic and the other, with the ruler smiting his enemy, has a more descriptive text. Both zones have singular emphasis—they appear at larger scale and in the center of the composition on either side—precisely because of their relatively greater narrative significance: that is, the fact that the incident related in them occurs repeatedly. Consistent with the overall organization of the image—obverse-to-reverse/ top-to-bottom—as after-to-before (Fig. 39), the obverse middle depicts the moment immediately after the decisive blow has fallen: here the serpopards have just been caught and their heads begun to be twisted together. The reverse middle depicts the moment immediately before the blow: here Narmer throws back his right arm, carrying his mace, just about to smite his enemy. The blow as it strikes its target, after the preliminaries to the blow and before the aftermath of the blow, must be in the center of the image—namely, in the wild place, the cosmetic saucer, where the owner of the palette or even the viewer reaches down to mix eye paint.

Escaping the Noose

Whereas in preceding late prehistoric images (Figs. 26, 28, 33) the absence of the ruler's decisive blow appeared as an ellipsis in the text while the viewer chased the narrative image, the pursuit of the ellipsis on the Narmer Palette is advanced into the text itself. As they twist together the necks of the serpopards (Fig. 41), the ruler's retainers are also depicted as tightening a noose around the wild space in the center of the image, the cosmetic saucer, where the ruler's blow at the instant it strikes its target must fall. In continually flipping the palette to retrieve the image caught and recaught in its twist (Fig. 42), the viewer thus seems to come closer and closer to seeing the falling of the blow, to pulling the noose tight around it with his or her own hands.

Fig. 43.
Canonical form of the narrative image on the Narmer Palette.

But in fact, although the picture of Narmer raising his mace appears again and again in the canonical form of the image, the blow escapes depiction and the image continues to mask it. While there is no need to describe all of its textual properties and relations (we have looked briefly at the first segment), the full viewing sequence through ten flips gives the narrative image on the Narmer Palette a complete, canonical text of eighteen zones—that is, the three sets of possible ways of scanning the six zones on the two sides according to the general conditions of intelligibility of late prehistoric image making, these sets sequenced in turn to accord with the same conditions. This complete text can be mapped as follows (Fig. 43):

1. OBV top (Episode Four)

2. OBV middle (cipher/after the blow)

3. OBV bottom (Episode Two)—flip

4. REV top (Episode Three)

5. REV middle (before the blow)

6. REV bottom (Episode One) [completes top-to-bottom set]

7. REV top (Episode Three)—flip

8. OBV top (Episode Four)

9. OBV middle (cipher/after the blow)—flip

10. REV middle (before the blow)

11. REV bottom (Episode One)—flip

12. OBV bottom (Episode Two) [completes side-to-side set]—flip

13. REV top (Episode Three)—flip

14. OBV middle (cipher/after the blow)—flip

15. REV bottom (Episode One)—flip

16. OBV top (Episode Four)—flip

17. REV middle (before the blow)—flip

18. OBV bottom (Episode Two) [completes double-twist set]

Several features stand out in this text. First, the rate of flipping speeds up as the text progresses—it literally becomes three times faster at the end than at the beginning. By the end of the text casual viewers may long since have abandoned it; but viewers accepting the general conditions of intelligibility and following the instructions of the cipher key will experience its accumulating excitement as a physical sensation—like a roller-coaster ride that begins slowly and then rockets through the final stretch.

Second, in the first set of flips (top to bottom, zones 1–6), the enemies' execution (Episode Four) and the enemies' flight (Episode One) begin and end the segment or "chapter" (as zones 1, 6). In the fabula these episodes are separated by the greatest temporal span. In the next and second set of flips (side to side, zones 7–12), the enemies' capture (Episode Three) and the taking of the town (Episode Two) begin and end the segment (as zones 7, 12). In the fabula these episodes are separated by a lesser temporal span. In the center of this segment, and of the text as a whole (zones 9, 10), the viewer seems as close temporally to the blow as it is possible to get insofar as the blow is not directly depicted (9, 10 = "after the blow" followed by "before the blow," the least temporal span of all). But in the third and final set of flips (double twist, zones 13–18), that temporal closeness is forced open again. The enemies' capture

(Episode Three) and the taking of the town (Episode Two), as they had earlier (zones 13, 18 = zones 7, 12), enclose the aftermath of the blow and what precedes the blow (zones 14, 17 = zones 9, 10); but they, in turn, enclose the greatest temporal span in the fabula (zones 15, 1 6)—namely, the enemies' flight (Episode One) and the enemies' execution (Episode Four). In sum, despite the increasing pace of the viewing, or precisely because of it, the image relates the greatest possible temporal span of the fabula as entirely caught up within the king's blow, the central, focal moment. The viewer is not led closer and closer to the blow, for the blow is outside and around whatever the viewer's viewing unfolds. In the way it relates details and passages of depiction, the storying works to establish a principal metaphor of the image—an expression of the continuousness of the ruler's blow.

Third, it is not just that the viewer fails to see the ruler's blow; it comes at the viewer from precisely where the viewer is not looking. While in the ten obverse-to-reverse flips of the palette the ruler's retainers twist the serpopards together clockwise (Figs. 41, 42), the ruler with his mace raised to strike (Fig. 44)—poised precisely on the other side of the place where the viewer begins—is always about to strike counterclockwise in the reverse-to-obverse direction. Therefore—like the lion on the Hunter's Palette (Fig. 28)—the viewer fails to see the ruler coming up around and "behind" him. The noose will twist off the head of the viewer before it is tight enough to capture the ruler's blow; in the viewer's cumulative and dizzying viewing of the image, the ruler pulls the noose tight as his blow, and the viewer becomes its victim. At the beginning of the text, then, like the ruler viewing his executed enemies, the viewer is invited by the image to enter the scene (obverse top, zone 1): the viewer marches beside the ruler and looks down on the decapitated corpses. But at the end of the text (by zone 18), the viewer understands the real import of the scene initially viewed—it was, after all, the "last" scene (Episode Four): the viewer leaves the text as yet another victim of the ruler's blow.

Considering what we have noted so far, it is not surprising that although in viewing the image the viewer may seem to unmask the ruler's appearances and come ever closer to seeing the blow taking place throughout, ten flips of the palette serve finally to remask the ruler. The ruler is always escaping the

Fig. 44.
Direction of retainers' twist and viewer's flips and direction of ruler's blow on the Narmer Palette, obverse and reverse.

depiction of the identity the viewer believes to be wielding the actual blow. According to the scheme whereby the six zones of the two sides of the palette (eighteen in all in the canonical form of the image [Fig. 43]) are presented in two visibly different modes of pictorial textuality—a descriptive or ekphrastic mode (D ) and a hieroglyphic (H ) mode—the sequence of these modes can be mapped as follows:

D H H H D D (top-to-bottom set, zones 1–6);

H D H D D H (side-to-side set, zones 7–12);

H H D D D H (double-twist set, zones 13–18).

In the first set of flips (top to bottom), after a graphic description of the final episode, Episode Four, the enemies' execution (zone 1), the preceding episodes are presented in less descriptive, hieroglyphic pictorial text (zones 2–4), then reviewed in more descriptive terms (zones 5, 6). For the viewer setting out to view the image, it would seem that the "masked" depictions of the ruler—as Bull and Falcon—are to be unmasked and his human person smiting his enemy revealed.

The lure of unmasking the ruler, and of finding the real climax of the narrative, sets the viewer off on a further series of flips, resulting in entrapment in the ruler's noose (zones 7–18). A parallel device can be found in other late prehistoric images—for example, on the Oxford Palette a victim moves toward the lure of the hunter's presence only to be attacked from behind (Fig. 27)—but the Narmer Palette handles this arrangement as a metaphor in its turn: the viewer is now a "victim." After a dizzying exchange of maskings and unmasings (zones 7–12), in which the viewer can never be certain where the ruler really is in the narrative (indeed he appears to be throughout it), the viewing ends with the thRee moSt descriptive depictions of the ruler (zones 15–17) enclosed by the three most hieroglyphic (zones 13, 14, 18). Since the arrangement of episodes in the story embeds the gReatest temporal span within the least temporal span, the pictorial text for this story can be said to embed descriptive specificity within hieroglyphic generality. It turns what would ordinarily be a simple inversion of spatiotemporal and causal logic, for the purposes of arranging a story, into a metaphor for the narrative as a whole—an expression of the universality of the blow paralleling its continuousness. Visible every time

the viewer flips to the reverse side of the palette (flips 1, 3, 5, 7, and 9), the picture of Narmer smiting his enemy is textually present in a full half of the narrative.

For these reasons the picture of Narmer smiting his enemy becomes not only an episode of or story element in the narrative in which it is embedded but also a symbol of that narrative. In fact it advances into a single passage of text (reverse middle zone) the pictorial dynamics of the entire narrative image as it is spread over a series of passages segmented and sequenced in various ways. In other words the reverse middle zone with the ruler smiting his enemy presents the metaphor that is not directly depicted in any one of the zones but rather is constructed by the complete concatenation of all the zones: it is the pictorial symbol of the metaphorical content of the narrative.

Extracting a Symbol from the Narrative

In what has been called the main "symbolic image" of the Narmer Palette (Schäfer 1957), the rebus in the reverse top right zone (Fig. 45) is not divided from the picture of the ruler smiting his enemy by a register line like those separating the pictorial zones elsewhere on the palette. It would seem, therefore, that whereas the late prehistoric narrative image calls for considering the reverse top and the reverse middle zones as distinct passages of pictorial narrative text, the pictorial symbol of the narrative also contained on the palette dictates that they should not be so distinguished.

In the symbolic image, in which reverse top and reverse middle zones are not separated, the depiction explicitly shows that Falcon lifts the enemy's face to gaze directly at the face of the ruler Narmer. Unlike the fleeing enemy on the Battlefield Palette (Fig. 33, obverse, third zone), who looks behind him to see and acknowledge the victorious lion-ruler facing away from him, here Narmer's enemy, subdued by Falcon, faces the master who takes possession of him. Just below, however, is a less compliant enemy (labeled as "harpoon" or coming from harpoon territory"), grasped by Narmer himself; although his body faces away from Narmer, the ruler twists his head completely around. The ruler must be grasping the enemy's topknot, for his clenched fist reaches into

Fig. 45.
Symbolic image on the Narmer Palette, reverse. 
From Wildung 1981.

the enemy's hair. A small rectangular form protrudes slightly over or "behind" Narmer's fist, but as it is not placed immediately "behind" Narmer's thumb resting over his curled fingers it cannot actually represent the enemy's hair. It appears to be an object Narmer is wearing on the back of his hand or wrist, perhaps the cylindrical roll seal belonging to the case (?, possibly a square box with a lid) suspended by a strap around the neck of his sandal bearer, standing behind him to the left. (The sign or image on Narmer's seal is apparently depicted in the upper left corner in the obverse top zone of the palette, at the very beginning of the pictorial text. The scene of smiting an enemy is replicated, in later images like the Libyan or Booty Palette [Fig. 53] and many canonical representations of figures adopting the "smiting" pose [for example, Smith 1949: 169–71, 245–48; see Davis 1989: 80–82], as a metaphor for taking possession of, surveying, or enumerating properties—an act literally accomplished in ancient Egyptian and ancient Near Eastern societies by impressing a seal.)

Narmer stands fixed to a solid ground line drawn below the group from side to side of the palette. Since the mace points up, rather than down, and considering the way Narmer's arm, wrist, and hand are rendered, the blow will swing in the moment of smiting forcefully to the side of the enemy's head, caving it in or knocking it off; presumably the ten decapitated enemies in the obverse top zone are the aftermath of such a blow. Narmer's decisive blow, then, must arc through a wild, unrepresented space—an ellipsis in the image's specification of "place"—from the point of view of figures ranged along the depicted ground line. As with the preceding late prehistoric images, the ruler's deadly force enters the world from outside, where it cannot be seen. In this case his force comes directly from the place—like the cosmetic saucer on the obverse—where the real viewers of the palette must be standing, invited to view the narrative image as followers of the ruler, and where the ruler himself must be standing behind them, for his blow will, in the full unfolding of the narrative, take the viewers too. In the depiction the doomed enemy seems about to slump on the ground line. When the blow falls his broken body will collapse forward along the ground line or at right angles to it, in any case "falling off " the ground line and "out of" the picture into an unrepresented place. But, again,

in the full unfolding of the narrative image, all the places "in back of," "beside," and "in front of" the palette have been mastered by the ruler in the counterclockwise direction of his blow.

In addition to the rebus and the figure of Narmer smiting the enemy, the symbolic image on the reverse contains a third element. Behind the ruler and placed on his own small ground line, Narmer's personal servant waits on the sidelines to perform his services. The viewer has already met him in the introduction to the narrative image (obverse top zone), where he follows "behind" Narmer. In both passages he carries the ruler's sandals (apparently bound to his wrist), a water jug or oil container, a small wooden case for holding the ruler's seal(?) strapped around his neck, and slung around his waist a bowl with two long hanging flaps of cloth, possibly for washing.^[3] Above his head two "hiero-glyphs"—a "royal" rosette, a flower from a lileacious plant, and an object that could contain sandals (Vandier 1952: 597) or might be the bulb of the plant—apparently give his title (see also Schott 1950: 22).

While the sandal bearer in the obverse top zone has a relatively straightforward place in the highly detailed description of the enemies' execution, matters are by no means as clear on the reverse of the palette. The sandal bearer appears to be part of the reverse middle passage of depiction; he seems to accompany the ruler just before the moment of his decisive blow. And yet, since he has his own ground line placed in the left portion of the composition, he could also belong to the reverse top zone, facing and complementing the rebus. According to this view the passage of pictorial text above the principal ground line on the reverse (middle and top zones) presents a "close-up" but also more "hieroglyphic" view of the scene depicted on the obverse top—that is, the ruler, accompanied by his personal servant and other companions (represented by Falcon), executing his enemies.

Why, then, should the sandal bearer be the only one of the ruler's retainers, of the several depicted in the introduction to the narrative image, to stand near the ruler just before the decisive blow? Although on the obverse top he has an obvious, if limited, place in the narrative image by filling out the ekphrasis at the beginning of the sTory, why should he be included in the text that symbolizes the narrative as a whole? Why, in the symbolic image, does he occupy the

place walking beside or behind the ruler—the place the viewer occupies in the beginning of the narrative and by its end discovers as being struck by the ruler's blow? Finally, if we treat the entire reverse side of the palette as part of the symbolic image, can it be significant that the fleeing enemies below the ground line look over their shoulders not only at the ruler but also at the sandal bearer (no other such glances across register lines appearing on the Narmer Palette, though common between zones in earlier replications)? Are we looking simply at the compositional and iconographic ambiguity necessarily inherent in a passage of text that functions simultaneously as part of a narrative image and as a symbol of the narrative whole?—or are we looking at the maker's, or the viewer's, difficulty in extracting a symbolic whole out of a narrative image?

If the symbolic image on the reverse of the palette extracts from the entire narrative the same metaphor for the sandal bearer as for the group of ruler and enemy, then the sandal bearer's position as "coming up behind" spatially might be interpreted as his continuous, universal temporal place. Like the ruler smiting his enemy, he too must be before, after, and throughout the story of the blow—with the difference that throughout he is masked from the sight of the enemy, for Narmer stands between the sandal bearer and the falling enemy.

The pictorial text places the sandal bearer directly below the ruler's raised arm clenching its mace. After the blow falls, then, he could move "forward" to assume the ruler's place. Thus he can be regarded as the ruler's heir-apparent.^[4] With the heir-apparent on the left ("future" or "after") side of the blow, the two enemies depicted below Narmer's ground line (reverse bottom zone) are correlatively placed to the ruler's right (the "past" side of the blow), looking back in the very direction the blow will be coming toward them (Fig. 44). Like the ruler's heir-apparent, they could be said to have a place in the symbol in both the "before" and the "after" stages of the blow. Before the blow—in the forward direction of the narrative image—they flee the ruler's deadly wrath, are pursued, and are destroyed (Figs. 40, 43). After the blow—and in the message of the narrative image extracted by its symbol—they acknowledge Narmer's successor. Their apparent change of heart is indicated in the pictorial text by a gesture somewhat like the raised hand of the fleeing enemy on the Battlefield Palette earlier in the chain of replications (Fig. 33): while the left enemy raises

his left hand and looks back over his right shoulder, the right enemy reverses direction, raising his right hand and looking back over his left shoulder. In either case, the pictorial text depicts them as able to see fully only what is behind them—the power and danger of the ruler, whom they can observe by twisting their heads back even further to their left (resulting in death by decapitation) or completely around to their right (that is, changing their point of view).

Chasing the Viewer

Considering its place in the composition of the narrative text and its status as an independent symbolic text, the major image on the reverse side of the palette (Fig. 45) presents the continuity of the blow, plaited through and tightening a noose around history and therefore an order both outside and a precondition of that history as a continuous cycle of after-to-before and before-to-after. In all of history the ruler—and his dynasty—is always coming around in the other direction from his enemy.

In short, considering both the narrative image and the symbolic image metaphorically related to it, the pictorial text on the Narmer Palette—the viewer's experience of the representation of the ruler's blow—has become the homolog of the message of the fabula. Viewing the Narmer Palette is, itself, an example of the ruler's victory.

Henceforth this mechanism and ones like it will become the canonical narrativity of the art of the dynastic Egyptian state, which literally subordinates the viewer of its representations to and by those representations (Davis 1989: 192–224). But it is also true that in the Narmer Palette the open texture of late prehistoric representation becomes closed and another mode of representation is instituted in its place. The open activity of the viewer chasing an image masking the blow of the ruler now becomes an image masking the blow of the ruler chasing the viewer—and closing in (Figs. 40–45). Notwithstanding the two sides and several zones of depiction, and the possibility of different ways of scanning, segmenting, and sequencing them, the insistence of the image on the Narmer Palette cannot be ignored: in the text it regulates the story sequence

by the use of register lines and repeatedly introduces or inserts and magnifies the scale of the ruler; in metaphor it expresses the continuousness and universality of his blow; and in symbol it represents and reduces the narrative. It is not possible to escape when the ruler is always there, everywhere.

Some viewers of the Narmer Palette may not, in fact, have pursued the narrative image all the way through its complex canonical form (Fig. 43), despite it's having derived from and replicated late prehistoric narrativity. For them the symbolic image on the reverse side of the palette—the choice of a single striking icon to stand for the whole narrative image—was the dominant aspect of the pictorial text (Fig. 45). Singular in its compositional scale and framed by register ground lines, the symbolic image of Narmer about to strike his enemy pushes the other zones of depiction to the side. As the representation of the narrative in which it functions and which it replaces in the very operation of so doing, the pictorial symbol on the reverse side of the Narmer Palette becomes interpretable independent of the narrative image. It can be treated as an entirely discrete text, a second image on the Narmer Palette existing alongside and representing the narrative image. As has been written of the frescoes in the synagogue at Dura Europos, "The two modes, the 'abbreviated' and not always fittingly called the 'symbolic' on the one hand, and the 'narrative' on the other, were not always sharply separated from each other but sometimes used side by side within the same work of art, and they may penetrate each other and become fused" (Weitzmann and Kessler 1990: 149; see also Winter 1985: 20–21, 27–28).

In the constitution of the canonical tradition in the early dynastic period, later artists selected the symbolic image for replication as such (Figs. 48–50). They took it as an independently meaningful visual text that could be applied in new scenes of representation the image maker of the Narmer Palette could never have anticipated. Moreover, as far as we can tell, these image makers also suppressed the replication of the late prehistoric narrative image from which this symbol derived and in which it was initially embedded (see Davis 1989: 64–82, 159–71, 189–91; Hall 1989). The complex disjunctive metaphorics by which the ruler edges into and turns about in the ground of nature—the metaphorics of the Ostrich, Oxford, and Hunter's Palettes (Figs. 25, 26,

28)—no longer needed to be reproduced when the metaphor had fully gained its ground and become standard usage, a literal reality that could be designated in symbol; in the emerging canonical tradition representation now did an about-face to face the ruler.

The chain of replications, then, is radically revised at the time of the Narmer Palette. A break occurs in the continuity of the metaphorics of the chain; it shifts from the continuous replication of a metaphor of the ruler's victory to a replication of an independent symbol of the metaphor. Rather than symbol appearing as a reductive selection from the complexity of a narrative, as it must have seemed to the immediately contemporary viewers of the Narmer Palette familiar with late prehistoric image making, narrative now appears for succeeding viewers as an elaboration of an established symbol. On the Narmer Palette late prehistoric narrativity thus comes to an end. Precisely because the late prehistoric narrative precipitates a non-narrative symbol that claims to state the full generality of the narrative, any and all possible versions of the narrative become metaphorical instances of the symbol. The previous metaphorics recede from view and the new symbol seems to refer, as the earlier metaphorics had done, directly to experience in the world.

This is not to say that the newly established symbol of the ruler's victory would not be narratively constructed in itself in canonical representation, and in all kinds of disjunctively related ways—just as, for example, the Carnarvon knife handle (Fig. 23) and succeeding late prehistoric images (Figs. 25, 26, 28, 33) had narrativized the symbols on the animal-rows and carnivores-and-prey designs (Figs. 8–10, 20–22). As we would expect, canonical narrative metaphorics are subjected equally, in turn, to the extraction of particular symbols—for instance, a remarkably stable image of the happy and successful courtier (Fig. 2)—and the suppression of sometimes alternative, but frequently unacceptable, metaphorical associations within a general horizon of unrevisability. (For instance, while sexuality is not absent from Egyptian image making and perhaps had a remote metaphorical root in the phallic prowess of great animals and successful hunters or rulers connoted in certain late prehistoric images, the official image of the courtier did not usually include explicit representation of his sexual pleasures, as it did in Greek Archaic and Classical vase

painting or in certain Japanese elite arts.) Canonical image makers took up several "core motifs" with diverse historical origins. In fact they balanced dynastic symbols, like the ruler smiting (Hall 1989), with equally or even more archaic symbols perhaps more familiar to the ordinary experience of the Egyptian population, like scenes of gaming, fishing, or farming (useful selections in Klebs 1915, 1922, 1932; Vandier 1964). They could establish further metaphorical relations among these symbols, such as an image of the hunter, fisher, or farmer as the ruler smiting or vice versa (see Davis 1989: 80–81). But all this complex activity in the scene of representation is a matter for the intricate history of the canonical tradition after the revision of the chain of replications of late prehistoric image making being examined here.

9—
In the Wild

If the Narmer Palette (Fig. 38) is any guide, the radical revision of late prehistoric image making was not accomplished easily. We cannot suppose the maker already had in mind a full conception of a completely new symbolic image that would derive from but thoroughly revise and replace late prehistoric narrative images. Since the image maker could not yet know the later canonical tradition, he would have had to select from his own past tradition of late prehistoric image making, replicating his revisions in such a way that they ceased to be revisions and became the accepted way of making images. Moreover, we cannot suppose that his contemporary patron or larger audience had a new image in mind, demanding the new symbolic image on the Narmer Palette to match the reality of some new social and cultural formation that they had already determined to recognize and represent.

To the degree that representation constitutes the reality of the social and cultural formation, such apparently historical explanations for the creation of the Narmer Palette would be empty. They would imply that the new image was somehow literally produced elsewhere—for example, "in the heads" or "by the design" of fully foresighted image makers or contemporary patrons—and then, somehow, merely translated on the Narmer Palette. We cannot explain, or explain away, a new mode of representation or a new social and cultural formation apart from the image constructed on the object itself. The Narmer Palette is the very site of the appearance of the new image; it is the very theater in which the new scene of representation is being staged.

How, then, can we interpret the disjunctive properties of the Narmer Palette as replicating the existing tradition of late prehistoric representation?

Radical Positions in Continuous Replication

We should not overdramatize the historical problem posed by the Narmer Palette. It is only in double retrospect that the palette could have been regarded, and can now be regarded, as especially radical in its revision. Its decisive and influential qualities are constituted retrospectively both from its own vantage point, looking back over the history of late prehistoric representation before it, and from the vantage point of later image making, looking back over the history of representation including it.

For the first, the replicatory distance between Ostrich and Oxford Palettes (Figs. 25, 26), and between Oxford and Hunter's Palettes (Figs. 26, 28), Hunter's and Battlefield Palettes (Figs. 28, 33), and Battlefield and Narmer Palettes (Figs. 33, 38) is substantial in all cases. All these images are disjunctively related one to the next; in no case does the disjunction amount to a trivial or completely predictable difference, the ordinary by-product of repeating a motif. Rather, the disjunction is the symptom of metaphorical activity: in the chain of metaphors the ruler circles, enters, contests, and finally gains all the ground occupied, initially, by other identities as well—animals, human enemies, companions, followers, and even the viewer. No single advance is possible without the preceding ones, and none is necessarily greater than any other. In fact the "first" step, the Carnarvon handle (Fig. 23) or the Ostrich Palette, is perhaps more decisive than the "last," the Narmer Palette.

At any rate, we cannot find in a chain of replications the wholeness of any one replication, its significance or finality, apart from the chain as a whole. The seeming finality of Narmer's advance depends in part on cumulative advances made by the ruler through earlier images; whatever is radical in Narmer's image must be attributed in part to the narrative structure of the existing chain of replications. From this point of view the Narmer Palette is less a new hypothesis than an obvious and well-supported deduction, "radical" because, by the sheer seriality of the case, it happens to occupy the position of the Q.E.D. rather than the premise. To be more exact, neither premise nor conclusion is, in itself, radical or not radical; the decisiveness of the image in the continuous chain of

replications—the only way it can catch the viewer up—lies in the argumentation as a whole.

The status of the Narmer Palette as the climactic and radical revision of the chain of replications of late prehistoric image making is explained in part by the accident of its survival. There do exist fragments of what must have been other complex palettes datable by their sculptural technique and style to the same period as the Battlefield and Narmer Palettes—that is, to the Horizon A /B or early First Dynasty period. Some of these intriguing objects were quite as well made as the Narmer Palette. Among those that almost certainly presented sophisticated narrative and/or symbolic images are the Louvre Bull Palette (Fig. 37), the Cairo bird-and-boat palette (Petrie 1953: pl. B 4; Asselberghs 1961: fig. 159), the Cairo-Brooklyn Palette (Asselberghs 1961: fig. 181; Bothmer 1969–70; Needler 1984: no. 266; Davis 1989: fig. 6.13b), and the Libyan or Booty Palette (Fig. 53). Moreover, there are three early First Dynasty carved mace heads (Asselberghs 1961: figs. 172–80) that are in my opinion narrative/ symbolic images constructed according to many of the principles considered here, with some differences attributable to the different shape of the artifact itself, the roughly spherical head of a mace. One of them, the mace head of the so-called King Scorpion (Fig. 52), replicates—repeats, but also revises—the metaphorics of the Narmer Palette.

From the vantage point of later image making, in the same way the Narmer Palette became decisive because of the accumulating power, the compounding interest banking up in the chain of replications to which it belongs, its influence also derived from decisions of the later image makers who invested in it. After all, it was these later artisans, like the maker of the mace head of King Scorpion, who seem to have ceased making decorated palettes as such. This decision, in retrospect, gives the Narmer Palette a sense of finality it may not have had for its immediately contemporary maker and audience. Furthermore, later image makers determined to replicate some, but not all, of its narrative and symbolic elements, and, equally important, determined not to replicate other, "earlier" forms of late prehistoric narrativity hitherto available. These two decisions again conferred on the Narmer Palette qualities of preferability or

superiority it may not have had for its immediately contemporary maker and audience.

As all these arguments should imply, it is not self-evident that the Narmer Palette was, in fact, regarded as exclusively preferable, a decisive revision, despite the replication of its symbolic image within the canonical tradition. After all, later users also determined to preserve the Oxford Palette (Fig. 26) in the same archaeological deposit as the Narmer Palette, the Main Deposit at Hiera-konpolis, along with various other materials (Quibell and Green 1900–1901; Adams 1974). The survival of other major carved palettes, the Ostrich, Hunter's, and Battlefield Palettes (Figs. 25, 28, 33), was conceivably attributable to scrupulous preservation by contemporary and later owners, although in these instances we do not have specific archaeological provenances. (The Cairo-Brooklyn Palette [Needler 1984: no. 266; Davis 1989: fig. 6.13a] survived—whether through deliberate preservation or sheer luck we do not know—into the Eighteenth Dynasty and was recut by a sculptor in the reign of Amenhotep III.) It may be, then, that it is from our vantage point alone that the Narmer Palette seems radical—the "end" of late prehistoric representation, the "beginning" of canonical representation—because the full history of the modulation of replication, the pro-and retrospective construction of meaning for contemporary makers and audiences, is so out of good archaeological focus for us.

At the same time we recognize the Narmer Palette as completing a narrative consisting of many narrative images in a way that was selected for continued replication only by later viewers, we must still ask about the status of this already-radical, yet-to-be-radicalized revision that is the Narmer Palette itself. Why was this image produced, and under what conditions? After all, although its apparently radical qualities may result from pro-and retrospective con-ferrals of decisiveness and influence, those conferrals could not have been made if there was nothing they could "nominate for continued regard" (for this latter vocabulary, I am indebted to Levinson 1979, 1989). The text was produced—in all its proficiency or lack of proficiency and in all its backward- and forward-looking hesitancy and ambition—chasing or (as may be) eluding the meaning coming out of the past and out of the future to meet it.

Symptoms of Discomposure

Because we have no independent archaeological evidence for the conditions under which the Narmer Palette was produced, we will never achieve anything like the fine resolution we would need to answer all our questions satisfactorily. Moreover, as we have seen, an "archaeological" observation of the conditions of the production of a given individual replication would not tell us about the determination of that replication. Rather, the replication is produced as a spatial and temporal distribution of meaning, or nonmeaning, that eludes archaeological observation of the conditions at this or that particular place or time.

Nevertheless, in principle, the very fact that representation is produced in the pro- and retrospective construction and distribution of meaning—a metaphor, a narrative, a symbol, an image, a chain of replications—means that the image does contain its own archaeology. The image itself is a spatial and temporal structure with a history of construction. It looks back over itself; and others look it over. We can excavate this folding-over by recognizing that the work of representation necessarily exhibits a material stratigraphy. It displays a recoverable sequence and an apparent direction, an intentionality of making that cannot but reveal something of its determining conditions. The representation is simultaneously the disease and the symptom—although disease and symptom must be differently "placed" and "timed" in the making, in the artifact itself.

As we would expect if the conditions of its creation had a decisively radical effect on the chain of replications, the image on the Narmer Palette—an image spatially and temporally constructed in and distributed on an artifact—is shot through and through with symptomatic material. What we have already seen of the viewer's place in the scene of representation will demonstrate the point. The viewer twists the Narmer Palette in his or her hands to view the narrative of the ruler's blow (Figs. 42, 43), but the blow comes at the viewer in the other direction (Fig. 44); the viewer is caught up in it like one of the ruler's enemies. Upon "beginning" the narrative, however, the viewer had been invited to enter the scene beside the ruler (obverse top zone) or, more generally, to be like the

ruler: the viewer had been solicited by the organization of depiction, its placing of the viewer's "point of view" on the scene, to progress through and master the scene. The "ruler" of the scene continues to stand beside the viewer outside the depicted scene throughout the narrative; his presence outside the scene does not cease to be felt, and in fact the viewer outside the literal depiction, like the identities within it, comes progressively to see its power and danger.

To be specific, the viewer in the scene of representation is placed to the right. Initially (obverse top zone), the viewer looks "down" on the ten decapitated enemies and walks "beside" the ruler's retinue (on the ruler's right hand); and later (reverse bottom zone), the viewer—now constituted as and represented by the fleeing enemies—"looks over the (right) shoulder" at the ruler coming up behind. Indeed the viewer moves right hand over left in chasing the image and constructing the narrative (according to the cipher key presented in the obverse middle zone; Fig. 41). By contrast, the ruler of the scene of representation is placed, with the depicted ruler, to the left (obverse top and reverse middle zones) and moves left hand over right—that is, in a direction away from the viewer's twist (Fig. 44). Like the viewer initially well grounded on the obverse top zone, looking "down" on the ten decapitated enemies, the ruler of the scene is well grounded on the reverse of the palette. Here, in the reverse middle left zone, the sandal bearer carries his two sandals, facing toward the outside of the depiction and at the exact angle at which the ruler of the scene must be standing there, "beside" the viewer on his right-hand side.

If the viewer's action is to view the image, to chase the unfolding narrative by twisting the palette, then the action of the "ruler" outside the depiction is to make the image. He is—in part—the maker himself. Whereas the viewer's first action is to scan the completed depiction from top to bottom, the maker's first act is to draw the ground line from left to right on the blank surface, establishing the armature to guarantee the order of that depiction. Furthermore, whereas the viewer begins the progress through the narrative with the obverse of the palette, various internal clues—what I call the stratigraphy of the image—suggest that the maker's progress began with the reverse . In so doing the maker inverted the direction for viewing the image's story but accepted the primacy of the symbolic image to be precipitated from it. We can

describe the direction of the maker's intentionality as through the symbolic image toward the narrative—a reconstruction, then, of all late prehistoric narrativity by way of the symbol already cast up as its organizing metaphor (the smiting motif) and by way of the compositional format (the register band) long ago adopted, pried apart, and made the ground of the ruler. On the Narmer Palette the register line carrying the motif of the ruler smiting was drawn first, not last.

But in fact this was the first time the image maker had attempted such a thing and thus his work was uncertain, almost botched. In cutting the register line from left to right across the reverse side, the maker had not yet discovered that straight lines drawn across a slightly convex surface and therefore inserted in the edges in nonperpendicular fashion will not appear straight to the eye. The ground line on the reverse side looks lopsided because the maker failed to account for the curvature of the object. He corrected his mistake in the next register ground line drawn, below the top scene on the obverse. Establishing the ground of the ruler as the underlining and armature of the composition was not, in fact, a step the image maker could fully control or understand in all its implications: the ruler, as we will see in more detail, was only just beginning to enter the maker's ground in the scene of representation.^[1]

With the armature of the ground lines in place, the maker then sketched the figures and began carving them in light relief, removing and smoothing the negative matter as he worked. Again we can see clearly that on each side he worked from left to right and top to bottom for each figure or group, his left hand holding the palette steady, turning it to the proper angle, and brushing away chips and dust as his right hand carried out the cutting. (For a right-handed engraver, this is, of course, the most natural and efficient procedure.) But working from top to bottom and left to right, he ran the risk of running out of space on the bottom right. Sure enough, at the right bottom corner of the reverse side, he failed to leave quite enough space for the left arm of the right-hand enemy figure, the last element to be produced (Fig. 46). He compensated by cutting back the figure's abdomen slightly and squeezing the arm partly behind the left leg. To make sure the arm would not appear to be lacking a hand, he then drew four fingers of the hand poking out on the other side of

Fig. 46.
Detail of Narmer Palette, reverse: fleeing enemies.
From Wildung 1981.

the leg. Noticing, in turn, that these might look like the genitals of the figure, he carefully added true genitals above them. He completely forgot, however, to return to the first enemy, already drawn on the left side, and add the corresponding genitals there. In revising his figures, the maker was clearly under pressure; and it led to an egregious error precisely in the place where the enemies look over their shoulders to see and acknowledge the ruler's blow coming at them (Fig. 44), or where, in general, the maker reversed the direction of his work—and perhaps its place in the entire chain of replications.

The passages of depiction above the ground line on the reverse side of the palette are, if not incorrect, at least ambiguous; it is not obvious how, as late prehistoric images, they should be divided into zones (left plus top right and middle, versus left plus middle and top right) or whether they should be divided into zones at all—that is, considered entirely within the conditions of intelligibility of the narrative image. Here too the ruler's blow breaks the composition apart. The decision to magnify Narmer's figure in relation to the other figures, and at a scale beyond anything attempted in late prehistoric representation so far, disturbs and arrests late prehistoric narrativity. From the vantage point of a viewer familiar with the general conditions of intelligibility of late prehistoric narrative on the Oxford, Hunter's, and Battlefield Palettes (Figs. 26, 28, 33), the static composition of the reverse side of the Narmer Palette (Fig. 45), with its heavy, lopsided ground line below and floating ground line coming partway in from the left edge, may have seemed clunky, even disorderly,

creating ugly blank spaces and alignments that do not quite seem to flow with the integration of the Oxford or Hunter compositions.

In sum, if we eschew looking at it through the lens of canonical art, the Narmer Palette is no more organized or composed than earlier images, with their clear differentiation and a mutual relatedness of pictorial zones we could pass smoothly between, like the intricate interlocking of pieces in a jigsaw puzzle. The revision of late prehistoric narrativity is not altogether confident and complete, not a correction so thorough that mistakes will cease to be visible and awkwardness avoided. Rather, the revision is, in advance, threatened and to an extent fragmented by the ruler's blow. The ruler's blow against the maker is already present in and as his revision of the image.

The Circumstance of Revision:
Looking at Himself Being Looked at Looking at His Work

Although viewer and maker are separate identities imagined as placed and divided from one another outside the depicted scene, they are the same actual person with two sides or faces—namely, the artist, whose positions as maker and viewer of the image are twisted together as a single identity in the continuous sequence of his work. To make the image he must view it, and to view the image he must make it. He is constantly exchanging one position for the other—coming around his hands with his gaze, coming around his gaze with his hands. And at the same time, he is looking over his shoulder—for another is gazing back at him and at the twists he accomplishes in and as his work. In looking at him looking at his work, this other—the ruler still outside the scene—causes him to twist himself to attend to the other's gaze. In all of this, then, revision becomes re-vision; the artist looks again at what he is doing, and, in looking again, he may not be looking with entirely the same eyes or even with his own eyes at all, for meanwhile he has been looking back to see how the other is looking at him and his work.

Like the figure of the right-hand enemy at the reverse bottom, other telling slips in the progress of his work are symptoms of the artist's double "two-sidedness," the fact that he must come around himself to make and view the work in front of him at the same time he must turn away from it to attend to

Fig. 47.
Detail of Narmer Palette, obverse: (left to right) sandal bearer, Narmer, scribe 
priest, standard bearers, and ten decapitated enemies. 
Photo courtesy of Werner Forman.

what is behind him. For example, to draw the ten decapitated enemies on the obverse top (Fig. 47), he needed to turn the palette (left hand coming around right) at right angles to the vertical orientation of the register ground lines. He placed the first small enemy figure in the top row against the side edge, the ground line. With his hands coming around his gaze and perhaps thinking of the enemies on the reverse bottom, he drew the enemy's feet both facing right. Doing exactly the same for the next enemy (that is, for the figure second from the left in the top row of enemies), with his gaze in turn coming around his hands, he recognized the compositional difficulty he was creating for himself. With the third enemy in from the left, he turned the two feet to face inward. In this passage his power as artist—his ability to take possession of making and viewing as an uninterrupted, undisturbed condition of replicating and

revising—seems to be confirmed even as the trace of the danger of discomposure persists. As artist, he seems to have escaped what he represented to be the case for his depicted subjects, for whereas the depicted enemies' arms are bound and their heads lopped off, the hand that drew them moved freely with the gaze, always taking control as the work threatened to go out of control. Here we can accept his "mistake" as a successful revision, suppressing a possible difficulty and even adding greater descriptive detail to this passage of depiction.

But when the maker must depict himself in the scene, the two-sidedness of his identity as an agent in control of his work and the fact that he is followed, or shadowed, by another catch him up in a twist. At the introduction to the narrative on the obverse top, the scribe-priest marches in the ruler's retinue in front of the ruler himself—or the ruler follows behind him—to inspect the ten decapitated enemies, the ruler's handiwork and, in the canonical form of the image, a metaphor for the palette itself. Slung over his left shoulder the scribe carries his two paint pots with the rope hanging down in front of his torso. One pot would contain red ink for a preliminary sketch of a text or pictorial composition or for making rubrics and the other black ink for its completion. Whether or not this figure depicts the maker of the Narmer Palette—we cannot confirm it literally—he represents scribes and artists, makers of pictures and texts.^[2]

The real artist's interest in him is figured in the complex sequence of slips and revisions in the cutting. For one thing, the artist could not quite manage the two paint pots. They seem to hang impossibly in front of rather than beside or behind the figure, although the rope hangs down his torso and is clutched by his left arm in such a way that the pots could be carried only by being slung over the left shoulder. Here we might recover the mistake as an element of the image. For example, perhaps the descriptive function of this passage of depiction required the artist to inform his viewers of the person's occupation—and somehow he had to depict the paint pots otherwise out of view, like the lead bowman's quiver of arrows, slung over the shoulder, on the Hunter's Palette (Fig. 28).

Why not then do what the maker of the Hunter's Palette had done, and hang the paint pots properly over the right shoulder (on the viewer's left) and "behind" the figure? The maker's problem here is that working from left to

right he had obviously produced the figure's right arm and right shoulder already —with or without rope and pots; and working from left to right, he was forced into an awkward arrangement for the left shoulder: he had no choice but to place the paint pots "in front of" it. In fact the right arm, despite its failure, is a revision of an earlier attempt, cut over something else after the torso and left shoulder had been completed. The new right arm cuts off part of the old left hand. It seems likely, indeed, that the old right arm would have carried the rope and paint pots in proper orientation hanging behind the back. Why then revise the work by moving the rope and paint pots to the other side, the figure's left, and recutting his right arm to hang down rather than clutch the rope, as it probably did in the initial version?

The answer lies behind the figure even farther—that is, where the figure of the ruler is placed. Working from left to right, the artist had placed the scribe's back, right foot close to the forward, left foot of the ruler in the same way the ruler had been placed close to his sandal bearer following behind him at the left edge of the composition. But then it turned out that the figure carrying rope and paint pots over the right shoulder would be too close to the ruler's mace to be fitted in comfortably. Therefore, in what must have been the first revision in this sequence of recuttings, the scribe's back, right leg (on the viewer's left) had to be stretched out at just enough of an awkward angle to put him out of the ruler's way; the right leg angles up from the ground line with an extreme tilt. And then, in relation to the figure's excessively long right leg (on the viewer's left), his forward, left leg coming down to the ground line (on the viewer's right) would appear stumpy, like a peg leg protruding from his skirt—so the skirt is cut at an upward slant toward the front to provide room for a longer, "taller" forward leg matching the back one. Despite this effort the figure would still appear too stooped or lame, dragged backward by the paint pots; so the paint pots come around to the front, the left arm clutching the rope is produced, and the back, right arm, formerly clutching the rope holding the pots, is cut over again. Note, then, that at one point in the sequence of recuttings, the figure would be carrying two sets of paint pots, one over each shoulder.

By this point in his work of revision, the artist was unable to handle the right arm. He did not know which part of it to show himself or the

viewer—back, left side, right side, or front—and awkwardly revealed a little of each: precisely because the ruler comes up forcibly behind him, more of the maker now comes "into view" in the scene of representation.^[3] The new right arm and hand—used by a right-handed artist, like the maker of the Narmer Palette, to wield the brush or hold the chisel—hang loose and useless; and its hand finally comes out as a peculiar upside-down version of the hands of enemies acknowledging the ruler (as on the reverse bottom zone of the palette and on the upright fleeing enemy on the Battlefield Palette [Fig. 331]). The left shoulder carrying the paint pots clutched by the left arm is completely twisted around itself. Everything becomes contorted or useless precisely because the artist attempted to revise the figure away from the ruler coming up close behind. There is no confidence in this effort; no full composure can result.

As a replication revising late prehistoric images, the Narmer Palette seems, on the evidence of its internal stratigraphy, to have been made under the very conditions it represents in narrative and symbol. It was made with the ruler just about to wield his blow standing beside and behind the maker, perhaps not literally over his shoulder but certainly close enough—sometimes too close for comfort. The ruler was close enough for the maker to turn his head away from the replication he was making, an activity determined by the history and the general conditions of intelligibility of late prehistoric representation, and toward the person of the ruler coming up behind him. He had to make room for the ruler, both in the image itself and in the making of it. Despite what the narrative and the symbolic images seem to say in their deliberated, literal presentations, the maker seems to have been caught partially unaware. He did not quite see the power and danger of the ruler, or right away; his corrections—of the entire tradition of replication, and of his own image as a deliberated correction of that tradition and of itself—revise him away from his failure to see and toward a greater recognition of the ruler.

The artist's knowledge must have come both pro-and retrospectively. Late prehistoric representation had implied that the ruler is decisively powerful and dangerous (Figs. 28, 33), but the maker could not know how much ground the ruler had gained until he looked over his own shoulder and found him standing there. The work, in other words, had to be hit by the very blow it narrates and symbolizes. The narrated or symbolic blows are only in representation, but the

Fig. 48.
Carved ivory label from Abydos, showing Den smiting
his enemy, First Dynasty. After Spencer 1980.

blow that hits them is real. It hit the maker himself as he composed the image, inevitably losing full control over the replication. The measure of proficiency, subtlety, and sophistication he and earlier makers had achieved in the ongoing elaboration of late prehistoric representation is noticeably disrupted, as if he had been almost pushed aside by the identity looking over his shoulder to inspect the representation of itself.

As a symptom of what it represents, the revision precipitated by the maker of the Narmer Palette could be selected for continued replication as a coherent symbol when other, later image makers, following his example, turned to face the ruler, now fully on this side, the viewer's and maker's side, of the text rather than on its reverse (Figs. 48–50). His revision is no longer uncomfortable, no longer the cause of discomposure or confusion, when later image makers no longer look over their shoulders—when they turn to work with the ruler directly in their field of view and as the frontal field of view. Discomposure will be reserved now entirely for the enemies of the ruler who do not see and acknowledge him (Figs. 51, 52). And with the ruler no longer beside or behind

Fig. 49.
Cliff relief from Wadi Maghara (Sinai), showing Sekhemkhet smiting his enemy
and standing in majesty, Third Dynasty. After Petrie 1896.

Fig. 50.
Cliff relief from Wadi Maghara (Sinai), showing Sanakht smiting his enemy, 
Third Dynasty. After Gardiner and Cerny^[*]  1952–55.

Fig. 51.
Limestone seated figure of Khasekhemuwy in majesty, with figures of his enemies incised
on the base (see Fig. 36), late Second Dynasty. 
Courtesy Ashmolean Museum, Oxford.

them, but rather in front, later image makers ceased to narrate the story of his sideways, oblique, and contested advance from the wild outside representation—off to the side, from the back—into the world they could represent. The representable world, directly before them, becomes the ruler. More exactly, it does not matter whether they look straight ahead or over their shoulders when in any case they must face the ruler who can still take them from behind.

Positions

The creation of the Narmer Palette as a replication of late prehistoric images could be described much more literally. I do not believe it is necessary, however,

or even fully desirable, to do so. There are some advantages, but many drawbacks, in trying to say how the ruler actually stood behind the artist or how the artist literally found himself looking over his shoulder. This holds good for the entire chain of replications; for earlier representations we could ask how, in a literal sense, the hunter edged into the natural world or how its ground was contested.

For example, we could say that on the evidence of its internal stratigraphy the Narmer Palette must have been made under the "direct, daily, personal" supervision of the ruler and his immediate entourage (Bourdieu 1977: 190), probably in the context of a political and cultural struggle for national domination (for various accounts, see Edwards 1972; Trigger 1983; and Kemp 1989: pt. 1). In this struggle the labor of artists, scribes, and others was highly valued, their intellectual loyalty sought, and their economic subordination desired (see also Davis 1989: 215–21); the objects or texts they made may have functioned literally as media of state propaganda (Hoffman 1979: 299) or other elite self-representation. According to an account of this kind, the "earlier" images in the chain of replications, by contrast, could have been made by independent craftsmen or craftsmen affiliated with local elites, either historically prior to or contemporary with the emergent state, whose attachment to traditional forms of late prehistoric representation continued to be strong. The upward social and economic mobility of certain artisans (Davis 1983b) and the dynastic ambitions of some elite patrons (Kaiser and Dreyer 1982) had a noticeable if somewhat confusing impact on contemporary image making whether or not makers or patrons intended it, introducing a series of changes into the chain of replications that, whether or not it was initially noticed and selected for further use, constitutes the measure of the revisionary, the "radical," in the Narmer Palette. Some such scenario, to the extent that it might be consistent with other evidence, could be erected on the basis of our findings about late prehistoric image making and its revision.

A literal account such as this might, in principle, give rise to an even finer-tuned account. For instance, the image maker's professional anxiety could become the subject of our historical analysis. On the one hand, the maker of the Narmer Palette seemed to want to replicate late prehistoric representation—that is, to produce a well-formed image according to established standards of

pictorial narrativity. On the other hand, he seemed to want to revise the replication to produce an image more directly attentive to the presence of his ruler, a symbol of the ruler's real substance. Desires like this make sense in a time of upward social mobility for some, but not all, craftsmen and artisans (Davis 1983b). A craftsman on the way up had to detach himself from primary agricultural production and most likely had to become a full-time specialist in a craft workshop. There were significant risks, including possibilities that the workshop would fall, the patron cease to provide, or the agricultural year offer too small a surplus to permit full-time specialization. Inevitably the craftsman's identification with the ruler's ambitions was an ambivalent one.

We might even see the image maker's own craft, the business of painting or sculpting pictures, undergoing swift change, challenge, and doubt in the late prehistoric period—whether it was practiced in a village or at a court, in small, part-time fashion or as a large-scale, full-time operation. The decades reviewed in this book, from about 3300 B.C . to about 3000 B.C ., see the emergence of hieroglyphic script (Schott 1950; Kaplony 1972; de Cénival 1982; Ray 1986; Baines 1989; Millet 1990), based, at least in part, on traditional practices of image making in rock art, pottery painting, and other media (see Arnett 1982). Although I do not focus systematically on this process and it has yet to be investigated in detail as a theoretical problem, we have encountered some of what I believe to be its most important aspects—namely, the use of "keys to the cipher" of iconic signs and the precipitation of standard replicable symbols from narrative pictorial images.^[4] Did the image maker learn to read, affecting his replication and revision of images?—or was he one of the first writers of hieroglyphs to begin with? Was his status as maker of images and transmitter of knowledge threatened, or shored up, by the invention of script and the gradual creation of a semiliterate elite? What kind of stories could he continue to tell when a highly efficient mode of notation came into being? Did the very invention of a standard notation, readable under widely varying conditions and transmitted across the length and breadth of the state and as part of the growth of the state sector, make it more difficult to depend on the conditions of intelligibility of late prehistoric narrativity? To be able to answer such questions would be to provide a literal scenario of the image maker

looking over his shoulder at the ruler standing behind him—in the context of producing a representation masking the ruler's blow.

Although they need not be archaeologically implausible, if difficult to piece together, the methodological problems with such literal accounts should be obvious. A literal account might be consistent with archaeological evidence, but such consistency does not constitute explanation. For one thing, in its seemingly straightforward historical statements—based, in turn, on scrutiny of settlement patterns, burial data, and so forth—the literal account begs the hard question of cause and effect. It fails to say in what sense the images were the conditions for, or the results of, one or another element of the full scenario. For instance, in the example above, were some artisans upwardly mobile because they produced images satisfying the preexisting demands of an elite? Or was the success of the elite the result of their ability to persuade artists and other specialists to work for them? Even with the most finely resolved chronological knowledge of the sequence of events, these and other patterns of cause and effect will not be given in the historical evidence cited by any literal account; they will have to be inferred by the historian. The material evidence—which is simply a series of archaeological observations—needs to be narrated; and the narrating is the work of a historian observing patterns of relationship as well as occurrences of fact.

Despite its apparent seamlessness, its narrative drive, the literal story is full of doubts and gaps. Its status as the material account it purports to be is no more than possible at best. For instance, at least for the images under consideration here, archaeological evidence about their actual manufacture by independent regional versus court-affiliated national artists is completely lacking. A distinction of this kind must rely on "stylistic" differences among the works, in turn linked to a few debatable archaeological provenances (for example, see Davis 1989: 155–59 for "Delta" and "Upper Egyptian" styles in late prehistoric and early dynastic relief). But stylistic statements are only descriptions of sets of similarities among artifacts that are explained, by hypothesis, as having been caused by common descent from the same artifact "production system" (Davis 1990a). Thus a stylistic analysis does not allow us to make historical inferences but rather smuggles in a fundamental historical inference from the

beginning. The literal account might rely also on the internal stratigraphy of the images, rather than an independent archaeological record, to suggest possibilities of explanation. But, at least in the way I present it, that internal stratigraphy—although depending on numerous and minutely observed details—is not put forward as a literal history of anything but the material making of the object itself. As evidence for the determination of that material making in some actual context of "masking the blow," it serves as a general narrative of possible coherence and incoherence that must in turn be open to literalization when, or if, independent evidence becomes available.

Furthermore, whatever its measure of historical plausibility, the literal account is strictly limited by its own literalness, which might even be self-defeating. For example, it is always possible that we could secure some positive evidence about the social status of late prehistoric image makers and about particular patrons and viewers for whom their work was made. Taken as such, this evidence would not enable us to determine whether the internal stratigraphy of the images is symptomatic of the history of their production. Following the example of a literal scenario, an "independent" image maker—an artisan not affiliated with any court under the supervision of a powerful patron—could nevertheless be well aware of the possibility of a ruler "standing over his shoulder," and consequently he, like a court artist, would look back. Awareness of this kind might well be one element of his independence as defined in economic, social, or professional terms: to be independent is to be aware of what stands before or behind one. How does the literal account observe here what is, by definition, the imaginary or the symbolic in the real?

Taken too far—as it always must be if it aims to specify the "real" conditions of the production of images and the "actual" dimensions of the scene of representation—a literal account will confuse the cognitive "position" taken up, in the scene of representation, by a human subject in relation to the world and in relation to other representations of the world with real position—economic, social, professional, generational, family, ethnic, sexual, or other. If an artist represents himself in, or is represented as having, a certain economic, social, or other position, then, the account implies, he must really be in that position. By the logic of representation, however, the object depicted and the representing subject are divided from one another. Even in self-representation,

an artist looking at himself in order to represent himself is in a "position" that differs from the side or aspect of himself that has become his subject. To be able to represent being in a position or being represented as having one is already to show that the "real" material position from which the representations are made is partly outside those representations—that is, fully in the wild from the vantage point of identities within, and viewers of, the literal depictions as such. But that is not to say that they do not belong to the scene of representation.

Most worrying of all, in its sheer willingness to tell a story about what really occurred the literal account tends to exclude other possible literal accounts by other storytellers, then and now. Any and all of the positions someone has, including even the most transient "positions" taken up in irreducibly individual, nonconventional aspects of someone's form of life, are possibly relevant conditions of making representations. None can be ignored. The cognitive reality of any one of them in relation to its apparent presence in representation is a matter for investigation following upon, and in the context of, the archaeology of representation in replication.

For example, although certainly not usual in Egyptology or prehistoric archaeology, a literal tale of the maker's affiliation with the burgeoning state sector (as in Trigger 1983; compare Winter 1987) could be entirely rewritten in terms of some other "real" position, such as sexuality or ethnicity (for a provocative attempt, see Bersani and Dutoit 1986, with comments by Davis 1988). On the evidence of the temporal and spatial construction and stratigraphy of the images considered here, sexual, ethnic, community, and generational position certainly seem to be present in the late prehistoric Egyptian scene of representation. Thus the metaphorics that I explore in this book—the narratives of coming up behind, of seeing and being seen, of not-seeing and being destroyed—could be translated or rendered in literal scenarios as resistance and penetration, dominance and submission, ignorance and recognition, being inside and being outside, apartness and belonging, powerlessness and authority, and so forth, in every case assuming that the simple binary oppositions are metaphorically shaded and transformed in complex networks.

Some of these literal translations may be fully compatible with a historical account of the rise of state institutions and state ideology, perhaps even a necessary element in it. Some, however, might resist any such assimilation. And

why should we be forced to choose among them at all, settling, somehow, on the "real" or most basic scenario, the story of what is most compelling or determining in the last instance? For example, dominance of one actor's position over another's entails something specific (and possibly incompatible) in the two contexts, respectively, of real ethnic or racial diversity and real if sometimes unconscious sexual desire in a population. And these contexts cannot be separated: needless to say, ethnic or racial diversity and sexual desire can be the most literal or direct historical or social functions of one another. Thus we can only conclude that in fact the more general the account of "being dominant or being dominated"—or resisting and penetrating, ignoring and recognizing, and so forth—the less it will violate the realities of the world that a more literal, less inclusive account might reduce and obscure.

My account of the archaeology of replications "masking the blow" is, I hope, as open as possible to various, equally necessary literal accounts to be produced by historians with differing skills, knowledge, and interests—to the projections of all interpreters wanting to tell a story about the pastness of representation. To some extent this procedure is simply making a virtue of necessity, since we have no independent evidence for the real conditions of the production of late prehistoric Egyptian images. But I can see nothing intrinsically misleading or mistaken in that. Moreover, by no means does the general, open account have an easy pluralism, a free heterodoxy in its statements. For example, it has been quite tightly constrained by theoretical understandings of, and predictions about, intentionality, depiction, and narrative and their interrelations. Here it is not a virtue, if sometimes a necessity, to state matters in a broad way, for generality does not mean looseness; this framework establishes definite expectations for and limits on both the confirmability and coherence of historical interpretations.

Some historians would suggest that in the circumstances—we cannot know definitively what someone means or meant because we lack full evidence for it—we should say nothing. We would be instructed simply to give up making statements about late prehistoric Egyptian image making. Despite its appeal for those wishing to reserve their energies for what is supposedly secure and definite, I must reject this line of reasoning not only as the usual attempt to

hide "embarrassing questions" about "the place of history and the ultimate ground of narrative and textual production" (Jameson 1981: 32) but also as an abdication of the most literal-minded historical responsibility. In fact, it is an avoidance of the historical project as such. We can never know definitively what anyone means or meant because, by definition, meaning is not open to archaeological or historical observation at this or that time or place.

The hope for a fully general and open interpretation is an analytic ideal that cannot be completely realized in practice. My account of late prehistoric representation and of the Narmer Palette—which some readers must consider overly metaphorical—necessarily literalizes in all kinds of ways. For example, I want to relate the visual (or optical) and the physical (or manual) conditions of figures within the image and of the identities making and viewing it—the place and position of eyes and bodies. But in describing these eyes and bodies as "positions," it is difficult not to fall back into literalization, into a story about "real eyes," which cannot possibly look in both directions at once, and "real bodies," which cannot possibly twist around themselves. It is sometimes necessary, therefore, to maintain descriptions of seeing, viewing, holding, or moving the image or the object between quotation marks. That statements must be thus rhetorical and that the object of study is, itself, a metaphor whose translation is unknown may not be directly relevant in producing a general, nonliteral account—for the general account is about this figurative condition, or the conditions of figuration, in the first place.

The literal existence of some real condition of seeing, viewing, holding, or moving the image or the object is not in great doubt for any human being who could look at images or handle objects. Thus it can be a basic fact entered in any literal chronicle about the scene of representation. A "flipping" of the palette may well be a metaphor for a material scenario about someone's real eyes and bodies still or perhaps never to be specified. But for the purposes of analysis it serves us well enough as one material or metaphorical, particular or general ground, the ground of the eyes and the body, on which further complex metaphors could be propped—extending the optical and the physical into other spheres, such as the economic, social, professional, ethnic, generational, or sexual. As a point of interpretive theory, we may call the position of eyes and body the metaphor,

the "construction," of "real" positions—economic, social, or other. As a point of analytic method, however, we have to say just the opposite. Because the literal reality of some condition of the eyes and the body is not in doubt, it can become the grounds for a general account open to other literalizations in which the material connection between representation and the form of life is, indeed, in considerable doubt.

Always Outside

If the identity outside the literal depiction on the Narmer Palette is simultaneously the maker and the viewer—that is, the artist—then these are positions not only of the actual artist of the work, whoever he or she might have been, but also of the real ruler. Looking over his shoulder to take note of the ruler standing behind him as he produced the representation of the ruler coming up behind his enemy, the artist—along with the figures he drew—is the ruler's representative in the scene of representation. Other works of the early First Dynasty, while fragmentary, seem to advance a literal depiction of the ruler as maker into the depicted scene; his identity as the one who sees had already been well established by the chain of replications (Figs. 25, 26, 28, 33, 38).

In the image on the carved mace head of the so-called King Scorpion (Fig. 52), probably to be dated roughly to the time of Narmer, the ruler makes the fields of Egypt and, extending the depiction into the metaphorical domain, presumably Egypt's abundance and prosperity as well. The figure of Scorpion is singled out, first, by his size—he is the tallest in the composition; second, by the "royal" rosette and scorpion sign in front of his face, apparently the only such label in the composition; and third, by a narrative device we cannot be surprised to discover, considering how other images in the chain of replications have been put together. The wide band of depiction running around the middle portion of the mace head is divided internally into two (and at points three) stacked registers. The only exception is the figure of Scorpion, facing to the right in the surviving piece of the mace head, who spans the entire band from top to bottom. In the top register of the remaining piece of the central band, figures are oriented as moving away from Scorpion, and in the bottom, as

moving toward him. The switch in direction—where figures moving away must be coming together and figures moving toward must be proceeding outward—is undoubtedly made on exactly the other side of the mace head, in a passage of depiction that can be viewed only when the viewer is not viewing the figure of the ruler opposite it.

But what should appear in this other, exactly opposite position—predictably behind or on the other side—but another figure of the ruler? Although the mace head is almost completely missing for this segment of the image, a surviving fragment of a rosette emblem (not the same as the one used to label the ruler in the main surviving piece of the object) indicates that Scorpion was probably depicted here as well. Presumably facing to the left, he must have been engaged in an activity involving the women of the household, being carried toward him (one carrier and two women have survived), as well as rejoicing women (four surviving figures are visible) and advancing retainers with standards (three surviving figures are visible).

If this interpretation is correct, the portion of the image that survives on the mace head is the "other side" of the "opening" of the visual text—or, more accurately, the depictions of the ruler on both sides of the mace head are the "opening" of a pictorial narrative text, viewable in both directions, of which the ruler opposite is on the other side. In viewing this image by turning the mace head, the two depictions of the ruler are also the "middle" point reached on the two strips of the central band wrapped around the object. Starting with one ruler, we pass by the other ruler in the process of turning the mace head all the way around to the beginning again. In the top band of the image, above the figure of each ruler, the standards with hanging elements—lapwings (possibly the hieroglyph for the people of Egypt) and bows (perhaps the "nine bows" that later symbolize the foreign enemies of Egypt)—face in the direction the ruler is facing. Thus the lapwings hanging from standards by ropes strung around their necks make a switch in direction, with the birds replaced by hanging bows (and vice versa), precisely—in the most plausible restoration I offer here—between the points where the two rulers are placed.

In the portion of the image that survives, Scorpion wields a hoe, the same implement used elsewhere in early dynastic image making to depict the breach-

Fig. 52.
Mace head of King Scorpion: carved limestone mace head from Main Deposit at Hierakonpolis, early First Dynasty. Ashmolean Museum, Oxford. Detail of principal
figure of King Scorpion (photo courtesy Ashmolean Museum, Oxford); line drawing of surviving surfaces (corrected after Smith 1965) and suggested structure of the image.

ing of enemy fortresses (see Fig. 53, right). Here he is cutting an irrigation canal, indicated along the bottom register of the central band by a channel of water that forms the ground line for his figure and those of several others. Two bearers approach "from the right" to stand before him, preparing to sweep up and carry off the dirt.^[5] Two fan bearers come up behind him "from the left" (possibly emerging from the building depicted behind them, according to a recent restoration; Millet 1990: 59) to shade him from the sun. The ruler has the White Crown of Upper Egypt, the tunic, and the tail he wears on the Narmer Palette (Fig. 38); the locality seems to be specified as Papyrus Land.

In the register below the ruler's ground line and to the right (that is, coming into view from the right if the mace head is turned around left behind right), the ruler's sacred bark—only its prow visible—sails on the river, coming from the direction of the scenes now completely lost. Thus as the mace head is turned around to the left, moving "to the right" away from the figure of the ruler, the boat "sails toward" the viewer. (Since the boat presumably bears the ruler, as one figure of the ruler disappears from view, it is as if the other figure—on the other side of the palette—is being transported into view; as always, the narrative is in the motion.) In its progress the boat will pass by or between two cultivated plots. The right-hand plot in the preserved portion of the image is inhabited by a hoeing farmer, below, and another man, above, who seems to thrust his hands into the river. The left-hand plot, now badly damaged, probably contained the same two men; the top figure, thrusting his hands into the water, is still preserved. All these characters may be defeated, bearded enemies of the ruler put to work or settled on the land. Although it is not possible to understand the arrangement of the lower register, it seems that it was divided by the waterlines, and possibly other devices, into at least two and probably more episodes concerning the life of the countryside. The figures seem to be repeated from one zone to the next, as is a round-topped building. The register was probably arranged to establish a series of "befores" and "afters," or of preconditions and consequences, of the ruler's actions as they appear in the central band above it.

In sum, the image as a whole—top band, internally divided central band, and bottom band—divides into two principal movements, separated in the

central band by the register line that runs around the middle of the mace head. The two movements always go in opposite directions as far as we can tell from any single view of the image stretched around the mace head. When the top portion of the image is moving away from Scorpion "to the right," the bottom portion is moving toward him "to the left," and when the top portion is moving away from him "to the left," the bottom portion is moving toward him "to the right." The two figures of Scorpion, opposite one another in the image and each oriented in reverse direction from the other, span these two movements, whose "switch" is not visible when either figure of the ruler is being viewed. Thus, as in the other images in the chain of replications, the ruler is always outside, coming around and behind that which faces toward him and that which faces away from him. This much of the general structure of the image is clear. Unfortunately its metaphorical and narrative content remains obscure; because only about one-third of the whole has survived, we cannot say how the several episodes fitted together.^[6]

As the awkward drawing of Scorpion's hands and forearms suggests (there may be substantial recutting here), the artist struggled to position the main handle of the ruler's hoe almost exactly parallel with the ground line. This formal relation in the text of the image is crucial in the general metaphorics. It is the ground line itself that is wielded by the king, produced by him from the land of Egypt and—in continuing or "shooting out" from the hoe and around the mace as the viewer turns it—supporting the register of his rejoicing retainers and standards. The ruler makes the very order that canonical representation finds in the world.

This order is presented also in both early dynastic and canonical image making, as that which the ruler possesses as his estate (Figs. 1, 2). The Libyan or Booty Palette fragment (Fig. 53) preserves roughly the bottom third or half of an image perhaps derived from late prehistoric images as an independent symbol of their metaphorics of the ruler's victory, and now probably of state rule. A missing top register on one side (top )—we cannot tell obverse from reverse because the cosmetic saucer is absent—might be restored as prisoners being marched to the right by the ruler's retainers (for example, like the scene on the Beirut Palette fragment [Asselberghs 1961: fig. 183; Davis 1989:

Fig. 53.
Libyan or Booty Palette: carved schist cosmetic palette,
early First Dynasty. Courtesy Cairo Museum.

fig. 6. 12]); remains of human feet are visible at the edge of the break. Although a figure of the ruler might be restored in this register, he would presumably have to be placed in the center or on the left side so as not to violate the general organization of late prehistoric images, in which the ruler and the ruler's representatives never face "out" of the scene but are always coming into it. All the figures on the palette, however, should probably be "read towards the left in accordance with the orientation of Egyptian writing" (Terrace and Fischer 1970: 21). As we would expect, then, considering the replicatory revisions of the Narmer Palette, this palette exemplifies an emerging canonical revision of the order of late prehistoric image making.

Below this passage of depiction, whatever it presented, fortified cities—inside each, town-name hieroglyphs accompany small rectangles apparently indicating buildings—are breached by representatives of the ruler, including a falcon (top row) and a lion, a scorpion, and falcons on standards (bottom row). In its metaphorics the scene is related, on the one hand, to the image of ruler-as-bull breaking down the enemy citadel on the Narmer Palette (Fig. 38) and, on the other, to the ruler-as-wielder-of-the-hoe, as "maker" of Egypt, on the mace head of King Scorpion (Fig. 52).

On the other side of the palette (bottom ) three registers of animals survive—from top to bottom, bull oxen, donkeys, and rams, all males of their species. The image on this side might originally have included another register or two in the missing piece. There are four beasts in each surviving row of animals, with the exception of the last. The creatures are usually interpreted as booty taken by the ruler from plundered towns in the "olive-tree land" symbolized by the olive trees and "Tjehenu" sign (= western delta, Libya) in the bottom zone on this side of the palette (Vandier 1952: 592). Like the animal-rows motifs produced much earlier in the chain of replications, but separated from them by the complex set of disjunctions we have surveyed, the animals march along ground lines drawn from one edge of the scene to the other, eliminating, as they go, the possibility of a sideline or another ground for humankind and nature. This image has "tamed the unruly mob of the early palettes (although it seems to have robbed every creature of some of its mysterious potency)" (Groenewegen-Frankfort 1951: 20).

Only the young ram at the end of the line in the bottom row—observed last if the image is viewed like hieroglyphic writing from right to left—turns back its head to look outside the depicted scene it inhabits as the king's estate. This figure reminds us (would any contemporary viewer have remembered?) of one of the commonest devices of late prehistoric image making—namely, the creature pursued by what it does or does not see about to catch up with it. But what could be behind this ram but a register line, the continuous, endless ground of the ruler? Or does the register line wrap around the palette—where in fact it disappears in this position—and the ram gaze back into a wild place where the register line cannot be found? The ambiguity is surely inherent and unresolved. The maker of the palette, standing in line facing the ruler like everything and everyone else, did not absolutely intend or fully anticipate how his ram would function. He simply ran out of space toward the bottom of the tapering palette and was forced to squeeze one figure in by drastically scaling down and shortening its body and turning its head back. And yet this slip, or resistance, of the maker plotting the ground figures some possibility of wildness within the unalterable armature of canonical representation. It constitutes as possible the continued presence, for canonical Egyptian art, of the late prehistoric scene of representation that is now literally behind and completely outside it.

The possibility of wildness is only that, for the actual person of the ruler of Egypt is still outside the scene of representation and in the canonical tradition will never be within it. His "real" appearance outside—as much as his represented appearance within—the literal depiction masks the way he continues to stand beside and behind. He is merely represented by that apparently real body with its composing blow of canonical figuration—that is, by the person, action, and gaze of the artist. The ruler himself, as ruler of the entire scene of representation, remains outside. He continues to designate how he will be seen—namely, precisely as the source of the representation he appears to inhabit from any angle or from all directions and about whom such a representation is naturally made. The artist who, as the representative of the ruler, enters a representation that depicts the king entering an unseeing nature can be nothing more than another nature entered by the ruler. The artist is "nature" carried to

the second power, or to as many powers as might be nested within one another; the artist is part of the scene of representation as it is always framed by a final, absolute power outside it. That power is absolute in the world it attempts to guarantee by putting forward this very representation of it. The ruler remains beyond representation—masks his own blow—until his wildness can be entered as a nature by one who has his or her own place outside the ruler's representations.

Disjunctive as it is, a movement has been completed—a series that has taken us through images none of which can stand alone (or as a whole, a beginning or an ending), each subsisting only in terms of what is behind and outside it. From the Oxford Palette (Fig. 26) to the Narmer Palette (Fig. 38)—a temporal and spatial slice of the chain of replications—the deadly forces of humanity and nature have completely traded places, shifted ground, twisted about, absorbed into each other, and advanced from outside to inside and inside to outside. Throughout, the ruler—father, hunter, matriarch, warrior, king—is forever in advantage by the very fact of being behind, the only one to approach always obliquely behind a mask with the other always failing to see him there. Unseeing nature, which cannot represent, does not design a place for the ruler; the ruler designs a place for it—that is what defines him as a ruler, absolute within the temporary reign of uncontested representation.

On the Oxford Palette the wild dogs face one another and clasp paws, while below them the masked hunter stands near the sidelines playing his flute. On the Narmer Palette the heraldic cows turn outward to face the one who looks at the entire scene: the ruler himself, never literally depicted in the scene of his mastery but only figured there as the condition of its composition. And within the depicted scene, at right angles to the person viewing him as his or her self, his represented natural double—the figure of the ruler smiting his enemy—has exchanged his mask and flute for crown and mace.

Appendix:
Pictorial Narrative

My interpretation of late prehistoric Egyptian image making assumes that many of the images are pictorial narratives. In the absence of independent evidence about the production or significance of the images, this assumption can be evaluated only on its own merits; it has to prove itself in practice. We must ask, first, whether an account based on it respects general logical considerations about narrative and about pictorial representation, and second, whether it generates an interpretation of the images that is faithful to the evidence, internally consistent, and more comprehensive and penetrating than existing interpretations. I believe the second question can be answered favorably, but I leave readers to judge for themselves. In this appendix I review the first question, beginning with the narratological concepts central to my account: fabula, story, and text. The distinctions among them, specifically as they relate to pictorial representation, provide much of the background for my analysis of late prehistoric Egyptian images. I forgo reviewing approaches to pictorial narrative that do not employ these distinctions—approaches quite different from my own (for instance, Weitzmann 1970; Brilliant 1984; Winter 1985). Thus it may be helpful to explicate my point of view, partly to set my interpretation in the context of an approach that still has little currency in prehistoric archaeology or Egyptology and even many branches of art history. This appendix, then, offers a brief introduction to one narratological approach to pictures for readers who want to evaluate the consistency and logic of my interpretation or pursue other issues, problems, or examples. Since this is not a book on pictorial narrative as such, my remarks are not systematic. Rather, I highlight a few topics I find especially significant, intriguing, or troublesome in a historical and critical examination of late prehistoric Egyptian image making as narrative. Although

pictorial narrative is not my only concern in this book, it is the most immediate concern in deciphering the visual evidence.

Narratives:
Fabula, Story, Text

Narrative should be conceived as the verbal designation or the graphic, sculptural, choreographed, or other depiction —broadly, the discursive "relating"—of a transition from one state of affairs to another. The transition entails and requires change in some, but not necessarily all, of the properties that events, actors, times, and places in the story are initially represented as having. This requirement, however, is not necessarily carried over fully into the narrative text itself, which implies but need not relate everything that logically changes in the "world" represented as so changing.^[1]

My general and technical definition of narrative differs from the colloquial sense of the term. As understood here a narrative does not necessarily require, although it can certainly include, the relating of separate events or actions linked together causally and arranged in a temporal order. A single event or action can itself be narrative or at least "narratable"; it can be related as a transition from one state of affairs to another. "As soon as there is an action or an event, even a single one, there is a story because there is a transformation, a transition from an earlier state to a later and resultant state. 'I walk' implies (and is contrasted to) a state of departure and a state of arrival. That is a whole story, and perhaps for [Samuel] Beckett it would already be too much to narrate or put on stage" (Genette 1988: 19). An image necessarily depicts "me walking" all at once. It may not simultaneously present pictures of earlier and later moments of my walking. Yet this constraint in itself does not prevent the image from being narrative.

Readers or viewers interpret the narrative's presentation on several levels. They proceed at the level of logic—that is, "what must be so in" or "what is true of" the world of the story if it can logically have the states, properties, and transitions it is related as having. The logic of fictional, mythical, or fantastical worlds and the truth value of statements about them present many puzzles (see

Goodman 1978; Wolterstorff 1981; Riffaterre 1990; Walton 1990). Readers or viewers generally must be able to assume that at least some "laws" of spatiotemporal order and cause and effect as they apply in the "real" world must apply in the fictive world of the narrative, if it is to be understood and discriminated from the "real" world as a representation requiring an effort of understanding. But as my quotation marks imply, the "real world"—the world of the readers or viewers—might be produced partly by fiction itself, by narrative and the intentions that animate it.

The narrative text might or might not present all the elements that readers or viewers require and assume in interpreting the story as presented. For example, an event that transforms one state of affairs into another—like the introduction of information impelling a character to change a plan—might not be directly related in the text, although it is logically required for the story to go forward; it will probably be "filled in" by readers' or viewers' "imaginations" in the "act of reading" (Sartre 1949; Iser 1978). Similarly, two states of affairs linked as cause and effect by a process detailed in the narrative text may themselves not be presented. Readers or viewers must infer them as causes and effects or as preconditions and consequences. Any explication of the narrative must address the logic and interpretation of the narrative "in the reader's head" and outside the text as well as what the text literally relates.

This preliminary distinction between what the literal text presents and what the narrator of and "in" the text implies and readers or viewers "outside" the text infer is uncomfortably artificial. We surely want to say that a novel includes what is not stated by the narrator but must be so for that narrator's direct statements to make sense. But here, too, well-known puzzles arise. The projected fictional world could be indefinitely and infinitely complex. To make sense of what it seems to contain, we must project further facts, and yet further facts, endlessly. At a certain point our projection of the world of the novel leaves behind what the novelist's projection manifestly requires.

Contemporary theorists have been more willing than earlier schools of thought to suppose that readers or viewers each construct, or project, their own narrative text (the central work remains Iser 1978; and see the interesting quasi-empirical investigation in Holland 1975). According to this view the text exists

in a complex, mutual, temporally and spatially structured interaction between a spoken, written, or depicted image, produced jointly by a narrator, an author, and a reader/viewer. Despite the plausibility of this analysis, we must also recognize that not all readings or viewings are as possible or as practicable as others. The text seems to be built so as to encourage one response or a limited range of possible readings or viewings that make sense of the writing. We cannot retrieve the individual acts of imagination on the part of individual earlier readers and viewers. But it is possible to investigate what the text, to the extent that it survives, might have offered them.

A single perspicuous formula (Klaus 1982) that captures these and other complexities of a narrative relating—a narrative statement or depiction—takes the following form:

SxPy ®SxP -y .

S is a (grammatical) subject, P its (grammatical) predicate(s), x and y the stated or depicted properties thereof, the bar (as in -y ) some transformation of properties (addition, reversal, or negation), and the ® a causal relation.^[2] E. M. Forster (1927: 31, 82) provided the classic example of a "simple" narrative:

The king died, then the queen died of grief.

The formula could be used to develop the following analysis of this narrative (as we will see, other analyses could be accommodated as well):

Although as a single English sentence of two clauses the statement appears simple, its narrative logic requires scrutiny, as the breakdown according to the formula suggests.

Following Aristotle, Forster distinguished between two dimensions of the statement, the story (the chronology in which the king dies and then the queen dies) and the "plot" (the queen dies "of grief" because the king died). At least three levels of a narrative relating are commonly distinguished in contemporary narrative theory. Narratologists label them according to their own methods and theoretical interests, but for my purposes here the concepts can be phrased in terms of the analysis of narrative relatings provided by the general formula.

First, the formula analyzes the intrinsic logic of the case, of the fluid state of affairs the statement relates. At one point (labeled State of Affairs 1) the king is dead but the queen is alive. At a later, causally related, point (labeled State of Affairs 2), the king is (still) dead, and the grieving queen dies (or the queen dies of grief). A more finely tuned application of the formula might distinguish more than two states of affairs—including, for example, the death of the king (State of Affairs 1), the grieving of the queen (State of Affairs 2), and the resulting death of the queen "from grief" (State of Affairs 3). According to this understanding, the sentence links State of Affairs 3 to State of Affairs 2, for the queen's death is caused by her grief, but says nothing about a causal link between State of Affairs 2 and State of Affairs 1, for it does not state that the queen's grief, and consequently her death "of grief," was caused by the king's death, only that it succeeds the king's death. Many readers will infer this causal link, especially if no other possibility is intimated; the "correct" reading will depend on the context of both the sentence and the narrating as a whole as well as on readers' understandings of words (for some, grief might connote bereavement; for others, it might have different connotations). However the statement is analyzed, the structure of the formula remains constant. At the first level, then, we look at the chronological and causal logic of narrated situations and events, "what is so in" as well as "what must be true of" the situations and events the narrative relating requires—sometimes called the "story material" but which I call the fabula .

Second, the formula allows us to analyze the statement "The king died, then the queen died of grief" as the sentence under consideration presents it. For example, the sentence happens to relate the king's death before it relates the queens death, whereas it could invert the sequence of presentation: "The queen died of grief after the king died." This statement preserves the content of the fabula, its information—direct or implicit—about chronological successions and causal links, including any ambiguities about these (as between State of Affairs 1, king's death, and State of Affairs 2, queens grieving, compared with the link between State of Affairs 2, queen's grieving, and State of Affairs 3, queen's death). We can recover the underlying structural logic of the narrative, the fabula, by analyzing the second, inverted, sentence according to the formula just as we analyzed the first, uninverted, version. But we might be interested in more than simply recovering this "story material." We might be interested in the second level itself, what I call the story, the particular arrangement of chronological successions and causal links in the narrative or, more simply, the fabula presented in a particular fashion .

Third, as a piece of writing, the sentence relates the properties of the agents and events in the fabula and story in a particular way. Preserving the underlying structural logic and the sequential presentation of that logic, the sentence could certainly have offered other words in another grammatical arrangement. For example, another narrative relating it might have read, "The king passed away, and his queen followed him to the grave in her despair." This sentence preserves the chronology and causal relations of the fabula as well as the manner of presentation of one story version, the "uninverted" version, but it puts the information in words quite different from Forster's "The king died, then the queen died of grief." This level of particularity is the level of the narrative text, the particular manner of relating the story as its particular manner of relating the fabula.^[3]

The words, phrases, clauses, sentences, and so on, relating a story of the fabula can be, but are not necessarily, identical with those making up what is often colloquially understood as a " text"—namely, the total concatenation of everything brought together in a single piece of writing like a "short story" or

a "novel." The narrative text can be combined with other forms of text—such as passages describing characters, settings, and so on—that do not affect the fabula and story themselves. These passages of text, however, may strongly affect the reader's experience of the narrative text. For example, they may slow down the reader's reading of an episode where word choice, syntax, and so on, obscure the succession of the story or make it difficult for the reader to recover the chronology or the relation of cause and effect in the fabula. I often consider text" in general—that is, late prehistoric Egyptian images as referential wholes—differentiating, however, between narrative text and, for example, the text of descriptions. But it can be difficult (and sometimes, for theoretical reasons, pointless) to distinguish in the textual presentation what is specific to the fabula and story and what is not.

The greatest disadvantage of distinguishing analytically between fabula, story, and text is that the triad, presented in this order, "corresponds to no real or fictive genesis" (Genette 1988: 14). This observation applies also to distinctions between story and plot (Forster 1927) or histoire and discours (story and discourse; Chatman 1978) and to relations between narrative text and the whole concatenated text, including passages of (non-narrative) description or punctuations and intercalations of various kinds. The method of working (for expository convenience) from fabula to story to narrative text, or back from narrative text to story to fabula, encourages the belief that the narrative text is but the representation of a story and fabula preexisting somewhere "in the writer's head" (or readers' heads) or "out there" in the collective consciousness or myth system of a culture. This attitude has debilitated the art-historical study of pictorial narratives. It reinforces a tendency to explain a narrative image in relation to preexisting stories and fabulas—to be found in the texts of scripture, mythology, annals or chronicles of the deeds of sovereigns, and so forth—treating the pictorial text as merely a re-production or illustration of that other text. In such cases the specific textual organization of the narrative image may be ignored, and certainly that of the preexisting text—which must have its own structure of fabula, story, and text different from that of the image—will be left wholly out of account.

Thus we must stress that the analytic distinctions are precisely analytic, valid (and, in fact, desirable) for precisely that reason. The distinctions are logical rather than psychological or experiential, either conscious or unconscious; and they are not, in and of themselves, genetic analyses. Because in psychological experience the fabula and story can be known by readers or viewers only "in" or "by" the text, the textual dimension of the narrative relating is experientially primary. And as a genetic matter, it may be only by progressively clarifying a text that readers construct the story and fabula that historically become the cognitive basis for narrative relatings. Out of the text of the late prehistoric Egyptian image can be constructed both a narrative (a story and fabula) and a symbol with metaphorically related content but decidedly different cognitive and historical status (see Chapters 6, 8). "The prime reality is that of the text, which does not reproduce a model that might be fictive or real, but which, more precisely, produces a narrative. The narrative itself does not organize a story [and fabula], which would be prior to what it says, but on the contrary, it allows itself to be reduced to the story [and fabula]" (Cros 1988: 95).

The temptation—extrinsic to the distinctions themselves—to extract fabula and story from the text as its prior "meaning" and cognitive determination must be countered by insistence on the literality of the fabula and story, on their textual production. We are not, as it might appear, caught in tautologies. The construction and reconstruction of fabula, story, and narrative text in and as the writing-out of an entire text are distributed temporally and spatially along the chain of replications of representation, like all other phenomena of intentionality. This activity has a complex and recursive cognitive structure (see Chapter 1). Thus, for example, a particular text produces a narrative for readers or viewers, who progressively reconstruct fabula and story from it in the reading or viewing. These then become the cognitive and historical bases, in ongoing replication, for producing another text that repeats but also revises and refuses those constructions. Perhaps the fabula is indeed repeated but "restoried" as a very consequence of productions arising within the new context, like metaphorical associations or even slips of rendition. Such dynamics have been the

matter of my investigation and interpretation of late prehistoric Egyptian images. At no single time or place along the chain of replications can we specify that here, now, is the fabula—a pure, unmediated thought of only what is and what must be so in the chronological and causal structure of some projected world—followed, here and now but somewhere else, by the production of the story to mirror that thought, followed, here and now but somewhere else yet again, by the text to reproduce that story. Strictly speaking (I have not relied on the terms because they are cumbersome), we should conceive fabulation, storying, and textualization as the temporally and spatially distributed, backward-and forward-directed intentionality of narrative, a narrating .

Gérard Genette (980, 1988: 13–15) has proposed the triad of terms story, narrative, and narrating, which correspond roughly to terms used by Mieke Bal (1985): Bal's fabula = Genette's story, while Bal's story and text = Genette's narrative . Genette's third term, narrating, is the act or situation of uttering the narrative, what linguists would designate the "pragmatic" aspect. It is implicit in Bal's understanding of fabula, story, and text. Because I want to maintain the distinctions among fabula, story, and text for the formal structure of the narrating, I prefer to say that the narrative consists of fabula, story, and text, whose production—fabulation, storying, and textualization—is the "narrating." The narrating transpires in the scene of representation, the time and place of producing particular representations that make sense in a series of pro- and retrospective references, repetitions, revisions, and refusals.

The analysis of and distinctions among fabula, story, and narrative text and between narrative text and text in general are fundamental to narrative theory. No study of verbal and pictorial denotation can attend to their "narrative" dimensions without making these distinctions in some way.

Pictorial Text:
Nondiscreteness, Density, Iconicity

Because they fail to employ the distinctions presented in the preceding section, many studies of "pictorial narratives" seem narratological neither in their premises nor in their results. All too frequently they do not even investigate pictorial narratives, or they completely miss those that should be identified, instead

concerning themselves with matters like the representation of motion, of elapsed time, or of "historical" or "actual" events; the sequencing of pictures of events; degrees of realism; and so forth. Although some of these topics may be relevant to pictorial narrative, many of them are red herrings.

Lacking minimally necessary narratological distinctions, most self-described descriptions of "narrative" in Egyptian canonical art (for example, Kantor 1957; Gaballa 1977) cannot serve as studies of Egyptian pictorial narrative. One example—by an art historian rather than an Egyptologist—is especially instructive. H. A. Groenewegen-Frankfort, who offered valuable insights into Egyptian representations of "space and time" in a superb book, Arrest and Movement (1951), worked with an overnarrow conception in a later discussion of Egyptian "narratives" (1970). Like many Egyptologists, she regarded Egyptian images (including most temple and tomb paintings and reliefs) as "static." Occasionally, however, Egyptian images strove for "dramatic actuality," as she put it—for instance, images produced by artists working for King Akhenaten at Tel el-Amarna in the Eighteenth Dynasty or by Ramesside artists in the Nineteenth (compare Smith 1965). Although her observations on particular images are highly suggestive, the problem of narrative as such cuts across her interest in the static versus "dramatic" or "dynamic" properties of images. Thus a pictorial narrative might include several "static" images to depict the individual states of affairs of the fabula as it is related in the story. In fact, it is impossible to reconcile Groenewegen-Frankfort's comments about the narrative possibilities of "dramatic" images, which presumably include single pictures like the well-known relief of Akhenaten, his queen Nefertiti, and their children (Davis 1978, 1989: fig. 2.11), with her statement—incorrect, as it happens—that the "juxtaposition" of separate pictures is the "only" means of creating a pictorial narrative (1970: 113). Groenewegen-Frankfort's two criteria for identifying narratives are "drama" and "actuality." Because she found "dramatic actuality" in Ramesside art, she also (without further analysis) found narrative. (Indeed, for some Egyptologists [for example, Gaballa 1977] the "actuality," or documentary value, of an image is the sole test of its narrativity.) But the "actuality" of an image is an independent matter with no immediate bearing on pictorial narrative as such. Simply put, narratives may relate actual

or non-actual, fictive, events; to look for narratives only in depictions of actual or historical occurrences is to rule narrative out of religious or mythological images.^[4]

Part of the problem is inherent in the material of images. In contrast to spoken or written texts ("natural—language text," or NLT ), all drawn, sculpted, or mimed texts ("depiction" or "depictive text") have a special, perhaps defining, property. Whereas NLT employs discrete elements, such as the phonemes of a natural language or the characters in a script, the elements of depictive text are nondiscrete or "continuous," both syntactically and semantically (Bach 1970; Polanyi 1970; and Goodman 1972 are especially clear statements).

Various ways of phrasing this general distinction bring out some of its implications. First, structural semiologists like Roland Barthes (1977) say that whereas in the discrete text separate signs (usually words) can be identified, individual signs cannot be singled out of pictures or films as materially separate or separable from one another. For example, in a painting the figures do not literally detach from their environment. In fact, the very same brushstroke can establish both a figure and elements of the background. Nonetheless, the figure "stands out" for a viewer familiar with the relevant conventions of the system, such as the general rule of Western illusionistic painting that a dark smudge at an outline usually stands for shadow around rather than dirt on the figure. For depictive text the whole composition—the framed picture or movie frame (or, for some writers, movie shot)—must be treated as a single sign. Because no principles for demarcating discrete signs within nondiscrete text can be developed, attempts to identify the "language" or "grammar" of pictures—if we take those terms literally—must fail. In nondiscrete text there are no well-defined elements of the "language" that could be "grammatically" related in the first place.

From a second point of view, theorists of the logical characteristics of symbol systems like Nelson Goodman (1972) and Catherine Elgin (1983) say that depictive text differs from NLT —a sketch differs from a score—in its relatively greater syntactic and semantic "density." Roughly speaking, density measures how much the perceived morphology of the sign functions in the syntax or semantics of that sign. It is a measure of the "signliness" of sign morphology.

In my view this analysis successfully avoids the obscurities of elementary pictorial semiotics, especially its difficulty in saying exactly what the elements of depictive text might be.

In principle, in a sketch (or other depiction) any change in the morphology of the pictorial medium, such as an alteration in the direction of the line or in the pressure of the pen, affects the interpretation of what is depicted and how. This is so because the morphology has a constitutive role in the form ("syntax") and reference ("semantics") of the sign: change the morphology, and the sign itself is changed. In Goodman's (1972) example, changing one tiny squiggle in Hokusai's view of Mount Fuji makes the sketch no longer Hokusai's view of Mount Fuji, whatever else it might (now) be.

By contrast, in diagrams, graphs, or scripts—less "dense" types of symbolic notation—such constraint is far less extensive. For example, a graph may show the points representing the highest temperature levels achieved in a scientific experiment on each of the successive days in which it was conducted. The continuous line connecting the points representing each temperature may make the graph easier to interpret but is entirely irrelevant to the information it carries. We can write the letter A as "A ," "a," "a, " and so forth, with a fairly wide tolerance of variation. So long as we can determine that we are reading the letter A , and not B or R or F , these minor or major variations in morphology have neither syntactic nor semantic consequences.

When analyzing "density" we should not fall into the trap of supposing that pictures always function in a more-"dense" symbol system and script in a less-"dense" symbol system. We can in fact make a written element work either pictorially or according to the logic of a symbol system other than script itself. When we examine someone's handwriting to detect possible forgery, the shape of the letters matters down to their finest modulations. But in all such cases, in examining the sign morphology we are simply shifting out of NLT and into another logic of notation; the same morphology can function under the rules of any number of different symbol systems. Conversely, we can treat the pictorial element "linguistically," as in Egyptian hieroglyphic script, where the picture of an owl has the consonantal value "m." In this case a morphology devised in the context of depictive text (the hieroglyphs probably originated in rock art,

pottery decoration, and related enterprises of marking or notation) is inserted into a context of NLT .

Social and cultural prescriptions for relative "density" in both NLT and depictive text have varied widely. We may find in a society like that of ancient Egypt, which apparently did not value expressive individuality in representational practice, a social scrutiny of pictorial "density" less exacting than that in what Goodman calls the autographic Romantic or modernist depictive practices of the West (see also Davis 1989: 113–15). Similarly, some societies treat script according to a "dense" symbol system—as, for example, in calligraphy-whereas industrial Western society has little such interest.

The archaeology of chains of replication is one of our best sources of evidence in this investigation. If we do not know what symbol system was being used, we would expect the producer of a text who was making NLT to Care little about fine details of the morphology of characters. In replications over time the details of sign morphology that are not regulated by the relevant symbol system would be subject to continual (and to a great extent random) revision, replacement, error, and erasure. By contrast, we would expect the producer of depictive text to be attentive to pictorially meaningful morphology and to preserve it from one replication to the next (see also Davis 1987 and n.d.; the matter is considerably more complicated than I indicate here). Thus in a chain of replications observed over time we should, in principle, be able to distinguish between the production and revision of both depictive text and NLT even if their morphologies appear similar or identical—as in the case of hieroglyphs.

Yet a third point of view on the distinction between discrete and nondiscrete text concerns the way a linguistic sign refers to its object arbitrarily, whereas a pictorial sign refers to an object by resembling it (even though resemblance is not sufficient for depiction and might itself be regulated by convention; Burks 1949). In Peircean sign theory, or "semelotic" (Peirce 1932), a picture displays "iconicity," as the sign resembles an object; "indexicality," as the sign is literally a symptom of the processes that caused it; and "symbolicity," as the sign stands for things. These functions penetrate and animate one another in all signs or sign systems. Thus in Peirce's schema a picture is "hypoiconic" among other things because a symbol (or the symbolicity of the picture) establishes that some possible resemblances between sign morphology and po-

tential. objects should be taken iconically (for example, a line depicting a horizon) rather than indexically (for example, that line as tracing the artist's gesture in making it). Other signs, like words or maps, are hypo-indexical, hyposymbolic, and hypo-iconic in other ways. Peirce outlined the possibilities in his full sixty-six-fold classification of signs.

According to the general understanding advanced by "semeiotic," depictive text, then, is never wholly iconic. Iconicity, like indexicality and symbolicity, is an abstract logical dimension of a sign. Nevertheless, on this account pictures are substantively and definably different from diagrams, charts, graphs, words, and scores, in each of which iconicity, indexicality, and symbolicity are represented in different measure and relationship.

We expect to find social and cultural diversity among sign-making practices. It has been said repeatedly of Egyptian representation that it emphasizes a picture-sign's symbolic dimension, even its capacity literally to stand in for things, as much as the iconic dimension more familiar to modern Western viewers. But this description, correct as far as it goes, must be substantially qualified. For it is equally possible to say that the Egyptian canonical image is highly iconic by its own standards, resembling as well as symbolizing objects such as the human body (Schäfer 1974; Iversen 1975). The sign morphology "resembles," however, not the objects found in the world of ordinary human experience, like those depicted by Chardin or Courbet, but rather those learned only by and through representation itself; they have a transcendent nature that is immanent in human history (see Davis 1989: 204–10). Despite what Egyptologists sometimes assert about the "hieroglyphic" quality of Egyptian image making, the image is, after all, both a hieroglyph, an icon that works symbolically in a specific way, and a depiction, a symbol tied iconically to the object of its reference.

In addition to the mutual relationship of iconicity and symbolicity in the picture-sign (contrasted, perhaps, with the hieroglyph-sign), we must consider its indexical dimension. Here, unlike Western modernist arts, Egyptian image making apparently places little or no weight on the indexical aspect of the picture-sign; but for this a more subtle judgment is in order. To suppress evidence of the artist's own handiwork, authority, and individuality (see also Davis 1989: 108–15) is indexical in itself. It traces the symbolic fact that both the

image and object it renders iconically are the work not of the maker of the sign morphology but of the "tradition." The canonical image may not be the indexical sign of the artist's hand (where traces of the artist's hand can be found, they are not usually significant); nonetheless it is the indexical sign of what is "spoken by the ancestors" and constituted by the gods "in the first time," as Egyptian ideology puts it (Davis 1989: 208).

To the "semeiotic" way of phrasing the general distinction between discrete and nondiscrete text we can, finally, relate an observation frequently made about the difference between NLT and depiction—implicitly in the tradition of ut pictura poesis (Lee 1940) and also to an extent in Lessing's Laocoön and its descendants (see generally Mitchell 1986: chaps. 1, 4). Whereas a writer need not designate the physical properties of an object referred to in his or her natural-language text, like the color of an animal or the shape of its body, a picture, logically, represents some properties of objects (see Fodor 1973: 177–95). No picture can avoid "descriptive" or ekphrastic specifications altogether. "Unlike the director of a movie, the novelist is not compelled to put his camera somewhere; he has no camera" (Genette 1988: 73); conversely, the picture maker is compelled to put the "camera" somewhere, to capture the properties of things instead of simply labeling them. As is apparent in Chapters 6 and 8, we must decide whether the properties of objects depicted in pictorial narratives have a narrative status, a non-narrative status, or possibly both.

Pictorial Narrative

The formulations distinguishing NLT from depiction help us to specify the difficulties that narrative depiction might encounter. First, if narrative relates a transition from one state of affairs to another, then it is easy to see how the discrete spoken or written text makes possible a narrating or narrative text. In NLT the separate states of affairs as well as the transitions can be designated by discrete "words" (sentences, paragraphs, chapters) in a linear sequence. As we have seen, however, this is not possible in depictive text, where there are no discrete elements that could relate a state of affairs.

Second, in NLT both spatiotemporal and causal relations among disjunct agents and events can be specified, whereas the depictive text has no straightforward equivalents for words or phrases like "and then," "as a result," "next," or "long ago and far away." Fabulation, storying, and textualization of chronological and causal relations must transpire in some other fashion.

Third, in NLT "descriptive" or even "ornamental" detail—all non-narrative material—can usually be discriminated easily from the narrating itself: for example, descriptive adjectives stand apart from substantives functioning as subjects in the subject-predicate structure of narrative statements. By contrast, in depictive text the ekphrasis is indissolubly bound to narrative function. The painter depicting the narrative event (such as the death of an animal) must simultaneously depict some properties of that event (such as the animal's shape, size, and color) not necessarily relevant to the fabula or story. The picture cannot directly demarcate any property of actors, events, times, and places as significant in or as the narrative. Given, moreover, that pictorial morphology is indefinitely "dense" or hypo-iconic, it is impossible to depict just the properties that also happen to have a narrative function. Thus in depiction every instance of narrating is also a describing, and some proportion of that description inevitably exceeds what the narrative requires. The converse, however, does not hold. Depictive descriptions produced with no narrative structure as fabula, story, and text can nonetheless provide the basis for a subsequent or ongoing narrative construction. Therefore in depictive narrating, narrative and descriptive functions are necessarily but only partially implicated one in the other.

Fourth, NLT can (though it need not) specify various durations in the fabula, such as the span covered by a flashback or -forward as well as an ellipsis in the story. By contrast, depictive text must always find a means of textual designation other than direct discrete discourse for all the temporal durations that a fabula logically contains and the jumps, anachronies, diachronies, and so on, that the story may present in relating them. This other means necessarily employs passages of nondiscrete, dense, hypo-iconic depictive text. For example, the temporal duration of an event could be indicated by its place and scale in the composition and consequently by the "time" it takes (or seems to take) to

view it; by lighting differences within a single composition that imply, for the reconstruction of a fabula, the passage of time in the day or through the seasons; and so forth. Two visually distinct compositions within the narrative image could indicate a temporal jump, as could a passage of special depictive text, such as the "whoosh" lines of modern comic strips that signify both the passing of time and the motion of things that usually accompanies it.

Because of these and other peculiarities of depiction, some writers have supposed that narratives, a strictly "verbal transmission" (Genette 1988: 16), cannot be produced in pictures. Noting that a picture cannot literally depict temporal and causal relations between states of affairs, Seymour Chatman (1975: 309) described a picture as a "stasis statement." Supposedly it represents what is" in some fictional world but not, strictly speaking, "what happens" in that world; for him, a "text which consist[s] entirely of [such] stasis statements, that is, [one that] stated only the existence of a set of things, could only imply a narrative." No doubt, as Chatman noted, this picture could somehow be "activated" as a narrative. For example, the picture's "static" presentation of a "given topic prompts the [viewer's] recall of a specific 'narratable' event" (Van Dijk 1975: 288). Even more directly, the picture maker could append a label in NLT that translates the "stasis statement" of the image into a "narrative" statement, as on the Narmer Palette (Fig. 38), whose rebus in the upper right zone of the reverse might be read, "Horus brings the defeated enemy to the king of Upper Egypt," implying that the enemy has earlier been defeated. According to this view the image may be narratable in NLT but is not narrative in itself; although the picture may be story material, no story is told without NLT . Apparently for reasons such as these, some of the most systematic and nuanced narratologies are limited explicitly to natural-language narrative text.^[5]

For some time, however, art historians have referred comfortably to Roman historical narrative relief or to nineteenth-century narrative history painting. Moreover, theories about narrative NLT suggest that in the presentation of "an action or an event, even a single one," as in the supposed "stasis statements" of depiction, "there is a story because there is a transformation, a transition [implied] from an earlier state to a later and resultant state" (Genette 1988: 19). But especially in light of the four peculiarities of depiction reviewed at the beginning of this section, the habitual practice of art historians and the logical

possibility of a narrative "stasis statement" in NLT do not, in themselves, show precisely how images "manifest narrative" (Chatman 1975: 317).

In fact many art historians seem to accept tacitly the suggestion that a single picture cannot be narrative by locating pictorial narrative in sequences of pictures instead. Examples of such sequences include large-scale pictorial cycles like Neo-Assyrian or Roman historical relief and Renaissance wall fresco as well as modern comic strips and the cinema, all exhibiting arrays of separate pictures. In comic strips the individual pictures are separated by distinct blank spaces, lines, bars, or registers; in "continuous narrative," however, such divisions are not used, and the picture seems to unfold as a single enormous strip. The best-known ancient and medieval examples are Trajan's Column (see Brilliant 1984) and the Bayeux Tapestry (see Bernstein 1987; McNulty 1989). Although each picture in either the divided or the continuous strip format might well be a "stasis statement"—the depiction of a state of affairs with no implied transition into or away from it—narrative supposedly transpires because the picture has been sequenced with other pictures presenting "earlier" and "later" states of affairs. The sequencing, then, has been "superadded" to the individual pictures precisely "to denote successivity" and accomplish the narrating (Chatman 1975: 317).

Many art historians take the narrative properties of the strip format for granted. But for all we have seen so far, narrative in the strip format materializes extrapictorially. If no single picture can depict the temporal and causal relations in the narrative story and fabula, there is no reason to believe that simply stringing together a set of single pictures will succeed in doing so. Just as mere sequence is potentially "narratable" in the sense noted already, it is also potentially not narratable if the states of affairs that are juxtaposed have no plotted relationships with one another. Alternatively, the array might present versions or "views" of the same single state of affairs rather than a single continuous narrative of changing states of affairs. This possibility is explored brilliantly in Akira Kurosawa's film Rashomon: the "strip" of images at a certain point repeats the representation of the same events from another character's point of view. Whether the narratings are chronologically and causally related and whether an overall continuous narrative image works in and through the array of separate point-of-view narratings as such are important questions in the

filmic text as a whole. At any rate, as the example indicates, a sequence of separate pictures is not intrinsically narrative. Comic-strip textuality and pictorial narrativity are potentially crosscutting, even if they sometimes overlap in particular cases.

If we would violate narratological theory in supposing that pictures become narrative merely by being arrayed, so too would we go astray in supposing that arrayed pictures, even if they are narratives, are the only species of narrative depictive text. Art historians do this when they contrast pictorial arrays with single compositions—the "iconic," "emblematic," "allegorical," or "symbolic" images produced in one or another supposedly non-narrative genre: "still life," "landscape," "portraiture, and so forth.^[6]

Yet clearly in a broader sense art historians want to recognize many single, nonarrayed, compositions—like some Renaissance religious and most modern history painting—as narrative depictive text. Even if such unitary compositions are neither divided internally by "narrativizing" devices like vertical register lines nor strung together in supracompositional arrays, they can nevertheless narrate in the strongest sense. We find that a single, unitary, nonarrayed image can present "a network of overlapping descriptions [of objects and actors] bound up in a present that incorporates past and future, experienced together" (Brilliant 1984: 18). Thus many Archaic and Classical Greek pictorial compositions combine episodes from different temporal points in the mythic story, present them all at once as an image, a referential whole, and thereby—but exactly how?—were and are understood to narrate or "emplot" it.^[7] Ostensibly "iconic," emblematic," "allegorical," or "symbolic" images outside our familiar pictorial systems—such as images from many prehistoric, ancient, medieval, or non-Western traditions—could fall under this rubric as well. To say this is not to abandon the constraints I have already noted. Whatever their mechanics turn out to be, such pictorial narrative compositions must exemplify the general properties of depictive text—nondiscreteness, density, and iconicity. For this reason, if no other, they are unlikely to be well analyzed from the point of view of NLT .

The steps to be taken in examining the particular structures of pictorial narrative—specifying the general features of any and all narratives as discursive relatings, distinguishing depiction from natural-language text, and recognizing

the probable diversity of pictorial narrative beyond the forms familiar to us—remain to be clarified. I conduct the clarification in this book in an extended historical and critical example rather than a pursuit of systematic typology.

I assume that the images found on late prehistoric Egyptian knife handles, combs, and palettes are not simply "decorative" or wholly "allegorical" or "symbolic," as they have generally been understood if their representational status has been considered at all (for example, Kemp 1989). Although they can be characterized as "writing" in the extended contemporary sense of the term,^[8] they are not hieroglyphs—even if they are sometimes associated with hieroglyphs (compare Millet 1990). ("Decoration," allegory, and hieroglyphs are not, of course, incompatible with pictorial narrative, for a narrative can also be an allegory or a decoration; conversely, allegories and hieroglyphs can be incorporated in an image that is also narrative. The categories are crosscutting. In a chain of replications, some images may be non-narrative and others narrative, one type even deriving from and revising or replacing the other.) Rather the late prehistoric Egyptian images are narrative images. They tell a story—properly, textualize the story of a fabula—whose "meaning" may be archaeologically irretrievable for both practical and theoretical reasons (see Chapter 1) but whose presence in a chain of replications can be traced.

While late prehistoric Egyptian images may be narrative, they are not likely to exhibit the same features as pictorial narratives in other historical contexts either in Egypt or in other societies. Thus they certainly do not exhibit the Greek vase painters' attempt to summarize, in a single carefully focused picture (Hanfmann 1957), the "unity of action" of a myth, even though an ambition to isolate the "pregnant moment" can be identified in some of them (see Chapter 9). They do include some devices later taken up in canonical Egyptian representation (see Chapters 3, 4). Pictorial narrative in the canonical tradition, however, has many different properties as well, including the constant and highly rule-governed use of the register ground line and a tighter association between NLT and depictive text (Davis 1989: chap. 3). The late prehistoric Egyptian images might be compared also with the comic-strip formats of pictorial narrative cycles like Neo-Assyrian or Roman historical relief (Winter 1981; Brilliant 1984). In late prehistoric Egyptian images, however, the long linear arrays we find in Ashurbanipal's palace or on Trajan's Column could

not be produced simply because of the size and shape of the supporting medium; the late prehistoric images wrap around relatively small and portable objects. This formatting of passages of depiction, composition, and image may have been in large part intrinsic to the surface to be "decorated." Nevertheless, as we have seen, the format seems to have a specifically narrative or metaphorical significance in itself. Since not all cosmetic palettes, for example, were decorated, and the decorated palettes are sometimes quite large and possibly not designed for actual use, the artifact type might have been selected for "decoration" because of an intention to make an image with a certain format rather than the other way around. Many other specific properties of late prehistoric Egyptian images have become clear in the course of my analysis (for a summary see Chapter 7).

Once we determine that an investigation of the images as narrative is appropriate, we are pretty much on our own. The broadest narratological propositions, such as the distinctions among fabula, story, and text, probably hold good across the board. As my comments in this appendix imply, however, many details of pictorial narratology remain to be developed, especially in light of a detailed understanding of particular historical cases. Following Juri Lotman (1975: 333–38), I favor a general analysis of nondiscrete or pictorial text that identifies an organized way of timing and spacing the transformation of the internal elements of the image as its "narrative" dimension (I make a few glosses to Lotman's statement):

Under what conditions . . . can such an image [a nondiscrete text] take on the [signifying] function of a text, become the conveyor of a [narrative] message? . . . [In the case of an image] there are, essentially, no [discrete] signs: the message is communicated by the text in its entirety. If we do treat it as discrete, and single out signlike structural elements, this is because of our habit of seeing verbal intercourse as the fundamental, even the sole form of communicative contact, and a result of making the pictorial text seem like a verbal [text]. . . . For the internally nondiscrete text-message of the iconic type . . . narration is a transformation, an internal transposition of elements [my italics]. A child's kaleidoscope may serve as a model . . . : the asemantic quality [of the kaleidoscopic pattern], in this case, helps only to reveal the mechanism of the narration, at the basis of which lies not the syntagmatic combination of the elements in space [in a sequence], which necessarily entails an increase [by linear expansion] in the size of the text [until it becomes

a string or array of many individual pictures], but an internal transformation and subsequent combination in [the ] time [of viewing ] (my italics). One figure is transformed into another. Each of them constitutes a certain synchronically organized segment [that is, it is viewed as a whole at a given temporal point in the extended time of viewing]. These segments . . . are transformed one into another, becoming a summation in time. . . . The verbal description is constructed as a narrative which is formed by the addition of new pieces of text, whereas the [pictorial narrative] may be seen as the transformation of one picture [in the time of viewing]. . . . The capacity of an iconic text to transform itself into a narrative is thus linked to the mobility of its internal elements [my italics]. . . . In the iconic text, which cannot be divided into discrete units [like the verbal/written text or even arrayed pictures], the narrative is constructed as the combination of an initial stable state with a subsequent movement [of the pictorial text in the time of the viewing].

In presenting the narrative dimension of late prehistoric Egyptian image making in this light, I prefer to offer a positive characterization of the pictorial narratives (Chapters 2–6) before stressing the instability or erasure of the narrative of the image, whether attributable to inherent cognitive and textual factors or to "extrinsic" social or cultural processes (Chapters 7–9). I also insist throughout on the place of individual images in the chain of replications as the siting of a pro- and retrospective distribution of intentionality; no single image embodies the narrative, which is disseminated throughout the entire chain of replications.

Notes

1— History and the Scene of Representation

2— Outside the Scene

3— Circling the Scene

4— Entering the Scene

5— Failing to See on Contested Ground

6— In the Morgue

7— Looking Back

8— About-Face

9— In the Wild

Appendix: Pictorial Narrative

Bibliographic Note

The literature on the prehistoric Nile valley and on predynastic and early dynastic Egypt is complicated and scattered; dates, relative chronology, and nomenclature have been constantly revised; and there are few synthetic monographs. These comments are intended to provide starting points for nonspecialist readers, citing publications mainly in English with many containing full bibliographies.

For introductions to the archaeology of early Egypt, see Edwards 1972, Butzer 1976, Hoffman 1979, Trigger 1983 (all with full bibliographies), Hassan 1988, and Kemp 1989. Edwards emphasizes the political history of the early dynasties, reconstructed from textual sources; Butzer examines the relation between the environment and early settlement; Hoffman offers a vigorous anthropological perspective on the epipaleolithic and neolithic Nile valley, as does Trigger; Hassan offers up-to-date chronostratigraphic perspectives; Kemp stresses what Egyptology has learned from the archaeology of predynastic and early dynastic material culture.

There can be no substitute for examining the original excavation reports (see the bibliographies in Hoffman and Trigger). For the most important archaeological context of objects examined in this book, the Main Deposit at Hierakonpolis, see Quibell and Green 1900–1901 and Adams 1974 and 1975, with further research at the site reported by Hoffman 1982. The excavations at Abydos provide important points of reference; the fundamental source is still Petrie 1900–1901. The standard archaeological typologies for the predynastic period also provide access to the excavated data; basic are Petrie 1920 and 1921, and see also Petrie 1913. Nonspecialists should note that Petrie's "Sequence Date" system has been revised in its details; both Hoffman and Trigger discuss the seriations and typologies. Baumgartel's 1960 monograph on late predynastic material culture, although detailed and essential for specialists, must be used with extreme caution on interpretive matters.

For late prehistoric and early dynastic Egyptian art, two compilations are invaluable: Vandier 1952 provides careful descriptions of the objects, mostly eschewing speculation, and Asselberghs 1961 publishes excellent photographs and includes a lengthy

English summary of his provocative analysis (in Dutch). Selected objects are especially well described and well illustrated by Scharff 1929 (publishing the important collection in Berlin), Kantor 1974 (including the most well-known objects), and Spencer 1980 (for the extensive British Museum collections). The treatments in standard surveys of Egyptian art are sketchy, with the exception of Smith 1949, which places the late prehistoric materials in the general context of the evolution of early dynastic and Old Kingdom style. In the text of this book I note specific publications on individual objects; see also the discussions and references in Davis 1989 (chap. 6), considering the origin of canonical conventions in prehistoric and predynastic art, a perspective somewhat different from the one developed here. The art and architecture of the first two dynasties is treated in Emery 1961; Emery 1949–58 is a systematic publication of early material from Sakkara; and Smith 1949 discusses the major reliefs and paintings of the early dynastic period.

In the earlier literature brief discussions of many objects can be found in Capart 1905 (significant as the first systematic overview of late prehistoric Egyptian culture and remarkably influential in its classification and interpretation of the material) and Petrie 1953 (with several important, although tendentious, iconographic identifications). Petrie's speculations about the functions of many types of objects (in Petrie 1920 and elsewhere) have been surprisingly popular; his stress on "magical" artifacts, taken up by Baumgartel and others, should be approached with caution. The earlier literature is more or less completely recorded in Porter and Moss 1950: 104–5 (major palettes), 106 (Gebel el-Tarif knife handle), and 193 (palettes and mace heads from the Main Deposit at Hierakonpolis); however, many of the citations are to minor discussions or to reproductions. The most detailed studies of the individual knife handles and palettes are those of Baumgartel 1960 (chaps. 4 [palettes] and 5 [mace heads], bringing a close familiarity with the objects, contexts, and historical parallels to occasionally idiosyncratic interpretations of the images) and Asselberghs 1961 (which attempts a synthetic interpretation emphasizing the aesthetic and spiritual dimensions of the works). Other specific interpretations are cited in my footnotes.

For late prehistoric and early dynastic iconography in particular, there are a number of important monographic studies: Bénédite 1905, 1916, and 1918; Kantor 1944; Vandier 1952; Baumgartel 1960; Asselberghs 1961; Williams and Logan 1987; and Williams 1988a. I do not register my agreement or disagreement with these authorities on each point of description and interpretation.

It is difficult not to look at late prehistoric image making in Egypt through the lens of later canonical image making. Smith 1949 still provides the most comprehensive descriptive introduction to Old Kingdom painting and sculpture; the chapters on early

dynastic and Old Kingdom art in Smith 1981 bring his account up to date. The most substantial technical analyses of canonical image making are in Schäfer 1974 and Iversen 1975, while Davis 1989 provides further references and discussion in the context of a broader study. Groenewegen-Frankfort 1951 offers a brilliant analysis, strongly influenced by Ernst Cassirer's philosophy of symbolic forms and Panofskyan iconology; Tefnin 1979, 1981, and 1983 introduces semlotic and semiological approaches to the study of Egyptian art.

It is essential to the approach of this book, however, that canonical image making not dominate our view of late prehistoric representation. Unfortunately, epipaleolithic and neolithic ("predynastic") image making in the Nile valley is still poorly understood. For rock art in the Nile valley, see Winkler 1937–38, Dunbar 1941, Hellström 1970, and Davis 1990b (other monographs and articles are listed in Davis 1979). For prehistoric material culture Petrie's many excavation reports and typological monographs are still essential, especially Petrie and Quibell 1896 and Petrie 1920; Baumgartel 1960 is a useful guide; and see also de Morgan 1896–97, Massoulard 1949, Brunton and Caton-Thompson 1928, Brunton 1937 and 1948, and Kantor 1944 and 1965. Nagada I and II (Amratian and Gerzean) painted pottery (White Cross-Lined Ware and Decorated Ware) have been widely published in scattered museum handbooks, exhibition catalogues, and other sources; George 1975 is an especially detailed presentation. Raphael 1949, a stimulating work by a sophisticated nonspecialist, is well worth studying. A systematic modern interpretation of the social and symbolic properties of neolithic painted pottery in Egypt is still needed.

Bibliography

Adams, Barbara. 1974. Ancient Hierakonpolis, Warminster: Aris and Phillips .

———. 1975. Ancient Hierakonpolis Supplement . Warminster: Aris and Phillips.

———. 1984, The Koptos Lions . Contributions in Anthropology and History 3. Milwaukee: Milwaukee Public Museum.

Aldred, Cyril. 1980. Egyptian Art in the Days of the Pharaohs, 3100–320B.C . London: Thames and Hudson.

Alpers, Svetlana. 1988. Rembrandt's Enterprise . Chicago: University of Chicago Press.

Arnett, William S. 1982. The Predynastic Origin of Egyptian Hieroglyphs . Washington, D.C.: University Press of America.

Asselberghs, Henri. 1961. Chaos en Beheersing: Documenten ult aeneolithisch Egypte . Documenta et Monumenta Orientis Antiqui 8. Leiden: E. J. Brill.

Bach, Kent. 1970. "Part of What a Picture Is." BJA 10:119–37.

Baines, John. 1989. "Communication and Display: The Integration of Early Egyptian Art and Writing." Antiquity 63:471–82.

Bal, Mieke. 1983. "The Narrating and the Focalizing: A Theory of Agents in Narrative." Style 17:234–69.

———. 1985. Narratology: Introduction to the Theory of Narrative . Trans. Christine van Boheemen. Toronto: University of Toronto Press.

Bard, Katherine. 1987. "The Geography of Predynastic Excavated Sites and the Rise of Complex Society." JARCE 24:81–93.

———. 1989. "The Evolution of Social Complexity in Predynastic Egypt: An Analysis of the Naqada Cemeteries. " Journal of Mediterranean Archaeology 2:36–49.

Barthes, Roland. 1977. Image-Music-Text . Trans. R. Howard. London: Hill and Wang.

Baumgartel, Elise J. 1955. The Cultures of Prehistoric Egypt . Vol. 1. 2d ed. Oxford: Oxford University Press/Griffith Institute.

———. 1960. The Cultures of Prehistoric Egypt . Vol. 2. Oxford: Oxford University Press/Griffith Institute.

Bénédite, Georges. 1905. "Le Stèle dite du roi serpent." Fondation Eugène Piot, Monuments et Mémoires 12: 5–17.

———. 1916. "Le Couteau de Gebel el-'Arak: Etude sur un nouvel objet préhistorique acquis par le Musée du Louvre." Fondation Eugène Piot, Monuments et Mémoires 22, fasc. 1.

———. 1918. "The Carnarvon Ivory." JEA 5:1–15, 225–41.

Bernstein, David J. 1987. The Mystery of the Bayeux Tapestry . Chicago: University of Chicago Press.

Bersani, Leo, and Ulysse Dutoit. 1986. The Forms of Violence: Narrative in Assyrian Art and Modern Culture . New York: Schocken.

Black, Max. 1962. "Metaphor." In Models and Metaphors, pp. 25–47. Ithaca: Cornell University Press.

Boehmer, R. M. 1974a. "Orientalische Einflüsse auf vierzierten Messergriffen aus dem prädynastischen Ägypten." Archäologische Mitteilungen aus Iran (DAI Teheran) NF 7: 15–50.

———. 1974b. "Das Rollsiegel im prädynastischen Ägypten." Archäologischer Anzeiger 4:495–514.

Bothmer, Bernard. 1969–70. "A New Fragment of an Old Palette." JARCE 8:5–8.

Bourdieu, Pierre. 1977. Outline of a Theory of Practice . Trans. R. Nice. Cambridge: Cambridge University Press.

Brémond, Claude. 1973. Logique du récit . Paris: Seuil.

Brilliant, Richard. 1984. Visual Narratives: Story-Telling in Etruscan and Roman Art . New York: Columbia University Press.

Brunton, Guy. 1937. Mostagedda and the Tasian Culture . London: Egyptian Research Account.

———. 1948. Matmar . London: Egyptian Research Account.

Brunton, Guy, and Gertrude Caton-Thompson. 1928. The Badarian Civilization. London: Oxford University Press.

Bryson, Norman. 1979. Vision and Painting: The Logic of the Gaze . Cambridge: Cambridge University Press.

———. 1990. Looking at the Overlooked . Cambridge: Harvard University Press.

Burks, Arthur. 1949. "Icon, Index, and Symbol." Journal of Philosophy and Phenomenological Research 9:673–89.

Butzer, Karl W. 1976. Early Hydraulic Civilization in Egypt . Chicago: University of Chicago Press.

Capart, Jean. 1905. Primitive Art in Egypt . Trans. A. S. Griffiths. London: Grevel.

Case, H., and J. C. Payne. 1962. "Tomb 100: The Decorated Tomb at Hierakonpolls." JEA 48:5–18.

Cénival, Jean-Louis de. 1982. "La Naissance de l'écriture [en Egypte]." In Naissance de l'écriture, pp. 61–62. Paris: Galeries nationales du Grand Palais.

Chatman, Seymour. 1975. "Towards a Theory of Narrative." NLH 6:295–318.

———. 1978. Story and Discourse: Narrative Structure in Fiction and Film . Ithaca: Cornell University Press.

Churcher, C. S. 1984. "Zoological Study of the Ivory Knife Handle from Abu Zaidan." In Needler 1984, pp. 152–68.

Clark, Timothy J. 1985. The Painting of Modern Life: Paris in the Art of Manet and His Followers . New York: Alfred A. Knopf.

Coste, Didier. 1989. Narrative as Communication . Introd. by Wlad Godzich. Minneapolis: University of Minnesota Press.

Crompton, Winifred M. 1918. "A Carved Slate Palette in the Manchester Museum. JEA 5:57–60.

Cros, Edmond. 1988. Theory and Practice of Sociocriticism . Trans. Jerome Schwartz, introd. by J. Link and U. Link-Heer. Minneapolis: University of Minnesota Press.

Culler, Jonathan. 1980. "Fabula and Sjuzhet in the Analysis of Narrative." Poetics Today 1:27–37.

Danto, Arthur C. 1986. The Transfiguration of the Commonplace: A Philosophy of Art . Cambridge: Harvard University Press.

Davis, Whitney. 1976. "The Origins of Register Composition in Predynastic Egyptian Art." Journal of the American Oriental Society 96:404–18.

———. 1978. "Two Compositional Tendencies in Amarna Relief." AJA 82:287–94.

———. 1979. "Sources for the Study of Rock Art in the Nile Valley." GM 32:59–74.

———. 1981a. "The Foreign Relations of Predynastic Egypt, I: Egypt and Palestine in the Predynastic Period." Journal of the Society for the Study of Egyptian Antiquities 11:21–27.

———. 1981b. "An Early Dynastic Lion in the Museum of Fine Arts." In Ancient Egypt, the Aegean, and the Sudan: Essays in Honor of Dows Dunham, ed. William Kelly Simpson and Whitney Davis, pp. 34–42. Boston: Museum of Fine Arts.

———. 1982. "Canonical Representation in Ancient Egyptian Art." Res 4:20–45.

———. 1983a. "Cemetery T at Nagada." MDAIK 39:17–28.

———. 1983b. "Artists and Patrons in Predynastic and Early Dynastic Egypt." SAK 10:119–39.

———. 1987. "Replication and Depiction in Paleolithic Art." Representations 19: 111–47.

———. 1988. "Pleasure and Its Contents." [Review of Bersani and Dutoit 1986.] Art History 11:445–56.

———. 1989. The Canonical Tradition in Ancient Egyptian Art . Cambridge New Art History and Criticism. New York: Cambridge University Press.

———. 1990a. "Style and History in Art History." In The Uses of Style in Archaeology, ed. Margaret W. Conkey and Christine Hastorf, pp. 18–34. New Directions in Archaeology. Cambridge: Cambridge University Press.

———. 1990b. "The Study of Rock Art in Africa." In A History of African Archaeology, ed. Peter W. Robertshaw, pp. 271–95. London: John Currey.

———. 1990c. Review of Kemp 1989. AJA 94:686–90.

———. 1992. "On Depiction and Metaphor: Comment on Michael Podro [see Podro 1991]." In Visual Theory, ed. Norman Bryson, Michael Ann Holly, and Keith Moxey (forthcoming).

———. 1992. "The Deconstruction of Intentionality in Archaeology." Antiquity 66 (in press).

———. n.d. Seeing Through Culture: The Possibility of the History of Art (unpublished manuscript).

Derrida, Jacques. 1973. Speech and Phenomena . Trans. D. B. Allison. Evanston: Northwestern University Press.

———. 1976. Writing and Difference . Trans. Alan Bass. Chicago: University of Chicago Press.

———. 1981. Dissemination . Trans. Barbara Johnson. London: Athlone Press.

———. 1987. The Truth in Painting . Trans. Geoff Bennington and Ian McLeod. Chicago: University of Chicago Press.

Douehi, Milad. 1983. "Traps of Representation." [Review of Marin 1988.] Diacritics 14, no. 1:66–77.

Dunbar, J. H. 1941. The Rock Pictures of Lower Nubia . Cairo: Government Printing Office.

Earle, Timothy K. 1978. Economic and Social Organization of a Complex Chiefdom: The Halelea District, Kaua'i, Hawaii . University of Michigan Publications in Anthropology 63. Ann Arbor: University of Michigan Press.

Edwards, I. E. S. 1972. "The Early Dynastic Period in Egypt." In The Camhridge Ancient History, ed. I. E. S. Edwards, C. J. Gadd, and N. G. L. Hammond, vol. 1, pt. 2, pp. 1–70. Cambridge: Cambridge University Press.

Elgin, Catherine Z. 1983. With Reference to Reference . Indianapolis: Bobbs-Merrill.

Emery, Walter B. 1949–58. Great Tombs of the First Dynasty . 3 vols. London: Egypt Exploration Society.

———. 1954. Great Tombs of the First Dynasty . Vol. 2. London: Egypt Exploration Society.

———. 1961. Archaic Egypt . Harmondsworth: Penguin Books.

Erlich, Victor. 1965. Russian Formalism: History, Doctrine . The Hague: Mouton.

Faulkner, R. O. 1969. The Ancient Egyptian Pyramid Texts . 2 vols. Oxford: Oxford University Press.

Finkenstaedt, Elizabeth. 1984. "Violence and Kingship: The Evidence of the Palettes." ZAS 111:107–10.

Fischer, Henry G. 1958. "A Fragment of Late Predynastic Egyptian Relief from the Eastern Delta." Artihus Asiae 21:64–88.

———. 1986. L'Ecriture et l'art de l'Egypte ancienne . Paris: Collège de France.

Fodor, Jerry A. 1973. The Language of Thought . Cambridge: Harvard University Press.

Forster, E. M. 1927. Aspects of the Novel . New York: Harcourt Brace.

Frankfort, Henri. 1925. Studies in Early Pottery of the Near East . 2 vols. Royal Anthropological Institute Occasional Papers 6. London: Royal Anthropological Institute.

———. 1941. "The Origin of Monumental Architecture in Egypt." American Journal of Semitic Languages and Literatures 58:329–58.

———. 1951. The Birth of Civilization in the Near East . Bloomington: Indiana University Press.

Freud, Sigmund. 1918. "From the Case History of an Infantile Neurosis." Trans. James Strachey. In Standard Edition of the Complete Psychological Works of Sigmund Freud, ed. James Strachey et al., vol. 17 (1952), pp. 7–122. London: Hogarth Press/Institute of Psycho-Analysis.

Fried, Michael. 1981. "Representing Representation: On the Central Group in Courbet's Studio ." In Allegory and Representation, ed. Stephen Greenblatt, pp. 94–127. Baltimore: Johns Hopkins University Press.

———. 1987. Realism, Writing, Disfiguration: On Thomas Eakins and Stephen Crane . Chicago: University of Chicago Press.

Fried, Morton. 1967. The Evolution of Political Society . New York: Random House.

Gaballa, G. A. 1977. Narrative in Egyptian Art . Mainz-am-Rhein: Phillip von Zabern.

Gardiner, Alan H. 1938. "The House of Life." JEA 24:157–79.

Gardiner, Alan H., and Jaroslav Cerny^[*] . 1952–55. The Rock Inscriptions of Sinai . 2 vols. Oxford: Oxford University Press.

Gasché, Rodolphe. 1986. The Tain of the Mirror: Derrida and the Philosophy of Reflection . Cambridge: Harvard University Press.

Genette, Gérard. 1980. Narrative Discourse . Trans. Jane E. Lewin. Ithaca: Cornell University Press.

———. 1988. Narrative Discourse Revisited . Trans. Jane E. Lewin. Ithaca: Cornell University Press.

George, Beate. 1975. "Frühe Keramik aus Ägypten: Die dekorierte Negade II-Keramik im Medelhavsmuseet." Medelhavsmuseet Bulletin 10.

Gilbert, Pierre. 1949. "Fauves au long cou communs à l'art égyptien et à l'art sumérien." Chronique d'Egypte 22: 38–41.

Girard, René. 1978. "To Double Business Bound": Essays on Literature, Mimesis, and Anthropology . Baltimore: Johns Hopkins University Press.

Godelier, Maurice. 1986. The Mental and the Material . London: Verso.

Goodman, Nelson. 1949. "On Likeness of Meaning." Analysis 1:1–7.

———. 1972. Languages of Art: An Approach to the Theory of Symbols . Indianapolis: Bobbs-Merrill.

———. 1978. Ways of Worldmaking . Indianapolis: Bobbs-Merrill.

———. 1979. "Metaphor as Moonlighting." Critical Inquiry 6:125–30.

Greimas, A. J. 1966. Sémantique structurale . Paris: Larousse.

———. 1971. "Narrative Grammar: Units and Levels." Modern Language Notes 86: 793–806.

Groenewegen-Frankfort, Henrietta A. 1951. Arrest and Movement: An Essay on Space and Time in the Representational Art of the Ancient Near East . London: Faber and Faber.

———. 1970. " 'Narratives' in Ancient Egypt." Art News Annual 36:112–21.

Haas, Jonathan. 1982. The Prehistory of the State . New York: Columbia University Press.

Hall, Emma Swann. 1989. The Pharaoh Smites His Enemies . Münchner Ägyptologische Studien 46. Munich: Münchner Ägyptologische Studien.

Hanfmann, George M. A. 1957. "Narration in Greek Art." AJA 61:71–78.

Harris, J. R. 1960. "A New Fragment of the Battlefield Palette." JEA 46:104–5.

Hassan, Fekri A. 1988. "The Predynastic of Egypt." Journal of World Prehistory 2: 135–85.

Hayes, William C. 1953. The Scepter of Egypt . Vol. 1. From the Earliest Times to the End of the Middle Kingdom . Cambridge: Harvard University Press.

Hellström, Pontus. 1970. The Rock. Drawings . 2 vols. Scandinavian Joint Expedition to Sudanese Nubia 1. Odense: Scandinavian University Books.

Hoffman, Michael. 1979. Egypt before the Pharaohs . New York: Alfred A. Knopf.

———. 1982. The Predynastic of Hierakonpolis: An Interim Report . Egyptian Studies Association Publication 1. Cairo: American University in Cairo.

Holland, Norman. 1975. Five Readers Reading . New Haven: Yale University Press.

Husserl, Edmund. 1970. Logical Investigations . Trans. J. N. Findlay. New York: Humanities Press.

Immerwahr, Sara. 1990. Aegean Painting in the Bronze Age . University Park: Pennsylvania State University Press.

Iser, Wolfgang. 1978. The Act of Reading . Baltimore: Johns Hopkins University Press.

Iversen, Erik. 1975. Canon and Proportions in Egyptian Art . 2d ed. with Y. Shibata. Warminster: Aris and Phillips.

Jameson, Fredric. 1981. The Political Unconscious: Narrative as a Socially Symbolic Act . Ithaca: Cornell University Press.

Janssen, J. J. 1978. "The Early State in Egypt." In The Early State, ed. Henri J. Claessen, pp. 213–34. The Hague: Mouton.

Jonnes, Denis. 1990. The Matrix of Narrative: Family Systems and the Semiotics of Story . Approaches to Semiotics 91. Berlin and New York: Mouton de Gruyter.

Kaiser, Werner. 1956. "Stand und Problem der ägyptischen Vorgeschichtsforschung." ZAS 81:87–109.

———. 1957. "Zur Inneren Chronologie der Naqadakultur." Archaeologia Geographica 6:69–77.

———. 1964. "Einige Bemerkungen zur ägyptischen Frühzeit, III: Die Reichseinigung." ZAS 91:86–125.

Kaiser, Werner, and Gunther Dreyer. 1982. "Umm el-Qaab: Nachuntersuchungen im frühzeitlichen Königsfriedhof." MDAIK 38:211–69.

Kantor, Helene J. 1942. "Early Mesopotamian Relations with Egypt." JNES 1:174–213.

———. 1944. "The Final Phase of Predynastic Culture: Gerzean or Semainean JNES 1:110–36.

———. 1952. "Further Evidence for Early Mesopotamian Relations with Egypt." JNES 11:239–50.

———. 1957. "Narration in Egyptian Art." AJA 61:44–54.

———. 1965. "The Relative Chronology of Egypt and Its Foreign Correlations before the Late Bronze Age." In Chronologies in Old World Archaeology, ed. R. W. Ehrich, pp. 1–46. Chicago: University of Chicago Press.

———. 1974. "Ägypten." In Frühe Stufen der Kunst, ed. Jan Filip and Machteld Mellink, pp. 67–72. Propylaen Kunstgeschichte 13. Berlin: Propylaen Verlag.

Kaplony, Peter. 1966. Kleine Beiträge zu den Inschriften der Ägyptischen Frühzeit . Wiesbaden: Harrassowitz.

———. 1972. "Die ältesten Texte." In Texts et langages de l'Egypte pharaonique, vol. 2, pp. 3–13. Cairo: Institut français d'archéologie orientale.

Kelley, Allyn L. 1974. "The Evidence for Mesopotamian Influence in Predynastic Egypt." Newsletter of the Society for the Study of Egyptian Antiquities . 4:2–13.

Kemp, Barry J. 1973. "Photographs of the Decorated Tomb at Hierakonpolis." JEA 59:36–43.

———. 1982. "Automatic Analysis of Predynastic Cemeteries: A New Method for an Old Problem." JEA 68:5–15.

———. 1983. "Old Kingdom, Middle Kingdom and Second Intermediate Period c. 2686–1552 B.C. " In Bruce G. Trigger et al., Ancient Egypt: A Social History, pp. 71–182, 352–56. Cambridge: Cambridge University Press.

———. 1989. Ancient Egypt: Anatomy of a Civillsation . London: Routledge and Kegan Paul.

Klaus, Peter. 1982. "Description and Event in Narrative." Orbis Litterarum 37:201–16.

Klebs, Luise. 1915. Die Reliefs des Alten Reiches . Heidelberg: Winter.

———. 1922. Die Reliefs und Malereien des Mittleren Reiches . Heidelberg: Winter.

———. 1934. Die Reliefs und Malereien des Neuen Reiches . Heidelberg: Winter.

Kubler, George. 1962. The Shape of Time: Remarks on the History of Things . New Haven: Yale University Press.

———. 1985. Studies in Ancient American Art: Collected Essays of George Kubler . Ed. Thomas F. Reese. New Haven: Yale University Press.

Lee, Rensselaer W. 1940. " 'Ut Pictura Poesis': The Humanistic Theory of Painting." Art Bulletin 22:197–206.

Legge, F. 1900. "The Carved Slates from Hieraconpolis and Elsewhere." Proceedings of the Society for Biblical Archaeology 22:125–39.

Leroi-Gourhan, André. 1967. Treasures of Prehistoric Art . Trans. Norbert Guterman. New York: Harry N. Abrams.

———. 1986. The Dawn of European Art . Cambridge: Cambridge University Press.

Levinson, Jerrold. 1979. "Defining Art Historically." BJA 29:232–50.

———. 1989. "Refining Art Historically." Journal of Aesthetics and Art Criticism 47:21–33.

Lévi-Strauss, Claude. 1971. L'Homme nu . Paris: Plon.

Lewis, David. 1967. Convention: A Philosophical Study . Cambridge: Harvard University Press.

Lewis-Williams, J. D. 1984. "Ideological Continuities in Prehistoric Southern Africa: The Evidence of Rock Art." In Past and Present in Hunter-Gatherer Studies, ed. Carmel Schrire, pp. 225–52. Orlando: Academic Press.

Lewis-Williams, J. D., and T. A. Dowson. 1988. "The Signs of All Times: Entoptic Phenomena in Upper Palaeolithic Art." Current Anthropology 29:201–45.

Lewis-Williams, J. D., and J. H. N. Loubser. 1986. "Deceptive Appearances: A Critique of Southern African Rock Art Studies." Advances in World Archaeology 5: 253–89.

Lingis, Alphonso. 1986. Phenomenological Explanations . Dordrecht and Boston: Kluwer.

Lotman, Juri. 1975. "The Discrete Text and the Iconic Text." NLH 6:333–38.

Marin, Louis. 1988. Portrait of the King . Trans. Martha M. Houle (French ed., 1981), introd. by Tom Conley. Minneapolis: University of Minnesota Press.

Marshack, Alexander. 1977. "The Meander as a System: The Analysis and Recognition of Iconographic Units in Upper Paleolithic Compositions." In Form in Indigenous Art: Schematisation in the Art of Aboriginal Australia and Prehistoric Europe, ed. Peter J. Ucko, pp. 286–317. Canberra: Australian Institute of Aboriginal Studies.

Martin, Wallace. 1986. Recent Theories of Narrative . Ithaca: Cornell University Press.

Massoulard, Emile. 1949. La Préhistoire et protohistoire de l'Egypte . Travaux et mémoires de I'Institut d'ethnologie 53. Paris: Institut d'ethnologie.

Mathieu, Michel. 1977. "Analyse du récit." Poétique 30:226–59.

McNulty, J. Bard. 1989. The Narrative Art of the Bayeux Tapestry Master . New York: AMS Press.

Meillassoux, Claude. 1978. "The 'Economy' in Agricultural Self-Sustaining Societies." In Relations of Production, ed. D. Seddon, trans. Helen Lackner, pp. 127–58. London: Cass.

Meyer, Elizabeth. 1985. "Component Design as a Narrative Device in Amarna Tomb Art." In Pictorial Narrative in Antiquity and the Middle Ages, ed. Herbert L. Kessler and Marianna Shreve Simpson, pp. 35–51. Studies in the History of Art 16. Washington, D.C.: National Gallery of Art.

Meyer, Klaus-Heinrich. 1974. "Kanon, Komposition, und 'Metrik' der Narmerpalette." SAK 1: 247–65.

Millet, Nicholas B. 1990. "The Narmer Macehead and Related Objects." JARCE 27:53–59.

Mitchell, W. J. T. 1986. Iconology: Image, Text, Ideology . Chicago: University of Chicago Press.

Mode, Markus. 1984. "Frühes Vorderasien und frühes Ägypten: Motivgeschichtliche Berührungspunkte in der Kunst." Halle'sche Beiträge zur Orientwissenschaft 6: 11–35.

Morgan, Henri de. 1909. "L'Egypte primitiv (Suite)." Revue de l'Ecole d'anthropologie de Paris 19:263–81.

Morgan, Jacques de. 1896–97. Récherches sur les origines de l'Egypte . 2 vols. Paris: E. Leroux.

Mueller, Hans-Wolfgang, 1959. "Ein neues Fragment einer reliefgeschmückten Schminkpalette aus Abydos." ZAS 84:68–70.

Murnane, William J, 1987. "The Gebel Sheikh Suleiman Monument: Epigraphic Remarks." JNES 46:282–85.

Murray, Margaret A. 1905. Saqqara Mastabas I . London: B. Quaritch.

Myers, Oliver H. 1958. "Abka Re-Excavated." Kush 6:131–41.

———. 1960. "Abka Again." Kush 8:174–81.

Needler, Winifred. 1980. "Two Important Predynastic Graves from Henri De Morgan's Excavations." Bullétin de l'Association international pour l'étude de la préhistoire égyptienne 1:1–15.

———. 1984. Predynastic and Archaic Egypt in the Brooklyn Museum . Brooklyn: Brooklyn Museum.

Newberry, Percy E. 1908. "The Petty-Kingdom of the Harpoon and Egypt's Earliest Port." Liverpool Annals of Archaeology and Anthropology 1:17–22.

O'Connor, David. 1990. "Egyptology and African Archaeology." In A History of African Archaeology, ed. Peter J. Robertshaw, pp. 236–51. London: John Currey.

Pager, Harald. 1989. The Rock Paintings of the Upper Brandberg . Pt. 1, Amis Gorge . 2 vols. Ed. Rudolph Kuper and Tilman Lenssen-Erz. Africa Praehistorica (Monographs on African Archaeology and Environment) 1. Cologne: Heinrich-Barth-Institut.

Panofsky, Erwin. 1939. Studies in Iconology . Oxford: Oxford University Press.

———. 1960. Renaissance and 'Renascences' in Western Art . Copenhagen: Russak.

Peirce, Charles Sanders. 1932. Elements of Logic . Ed. C. Hartshorne and Paul Weiss. Collected Papers of Charles Sanders Peirce 2. Cambridge: Harvard University Press.

Petrie, W. M. F. 1900–1901. Royal Tombs of the First Dynasty . 2 vols. London: Egypt Exploration Fund.

———. 1906. Researches in Sinai . London: J. Murray.

———. 1913. Tarkhan I and Memphis V . London: Egypt Exploration Fund.

———. 1920. Prehistoric Egypt . London: Egyptian Research Account.

———. 1921. Prehistoric Egypt Corpus . London: Egyptian Research Account.

———. 1927. Tombs of the Courtiers . London: Egyptian Research Account.

———. 1953. Ceremonial Slate Palettes and Corpus of Protodynastic Pottery . Ed. H. F. Petrie and M. A. Murray. London: Egyptian Research Account.

Petrie, W. M. F., and J. E. Quibell. 1896. Naqada and Ballas . London: Egypt Exploration Fund.

Podro, Michael. 1992. "On Depiction and Metaphor." In Visual Theory, ed. Norman Bryson, Michael Ann Holly, and Keith Moxey (forthcoming).

Polanyi, Michael. 1970. "What Is a Painting?" BJA 10:225–36.

Pollini, John. 1992. "The Gemma Augustea: Ideology, Rhetorical Imagery, and the Creation of a Dynastic Narrative." In Narrative and Event in Ancient Art, ed. Peter Holliday. New York: Cambridge University Press (in press).

Porada, Edith. 1990. "The Earliest Seals and the Rise of Civilization." In Edith Porada, "Near Eastern Art, Mostly in Seals." Una's Lectures in the Humanities, University of California at Berkeley (unpublished lecture, February 27, 1990).

Porter, Bertha, and R. L. B. Moss. 1950. Topographical Bibliography of Ancient Egyptian Hieroglyphic Texts, Reliefs, and Paintings . Vol. 5, Upper Egypt . Oxford: Oxford University Press/Griffith Institute.

Prendergast, Christopher. 1986. The Order of Mimesis: Balzac, Stendhal, Nerval, Flaubert . Cambridge: Cambridge University Press.

Prince, Gerald. 1987. A Dictionary of Narratology . Lincoln: University of Nebraska Press.

Quibell, J. E. 1905. Archaic Objects . Catalogue Générale des antiquités égyptiennes du Musée du Caire 23–24. Cairo: Institut francals d'archéologie orientale/Service des antiquités de l'Egypte.

Quibell, J. E., and F. W. Green. 1900–1901. Hierakonpolis . 2 vols. London: Egypt Exploration Fund.

Ranke, Hermann. 1925. Alter und Herkunft der Löwenjagd Palette . Heidelberg: Winter.

Raphael, Max. 1949. Prehistoric Pottery and Civilization in Egypt . Bollingen Series. Princeton: Princeton University Press.

Ray, John D. 1986. "The Emergence of Writing in Egypt." World Archaeology 17:307–16.

Ricoeur, Paul. 1981. "Narrative Time." In On Narrative, ed. W. J. T. Mitchell, pp. 165–86. Chicago: University of Chicago Press.

———. 1984. Time and Narrative . Trans. Kathleen McLaughlin and David Pellauer. 2 vols. Chicago: University of Chicago Press.

Riffaterre, Michael. 1990. Fictional Truths . Baltimore: Johns Hopkins University Press.

Sagan, Eli. 1985. At the Dawn of Tyranny: The Origins of Individualism, Political Oppression, and the State . New York: Alfred A. Knopf.

Saleh, J. M. 1987. "Rémarques sur les représentations de la peinture d'Hierakonpolis (tombe no. 100)." JEA 73:51–58.

Sartre, Jean-Paul. 1949. The Psychology of the Imagination . Trans. Hazel E. Barnes. New York: Free Press.

Scamuzzi, E. 1965. Egyptian Art in the Egyptian Museum of Turin . New York: Harry N. Abrams.

Schäfer, Heinrich. 1957. "Das Niederschlagen der Feinde." Wiener Zeitschrift für die Kunde des Morgenlandes 54:168–76.

———. 1974. Principles of Egyptian Art . 4th ed. Ed. Emma Brunner-Traut, trans. J. R. Baines. Oxford: Oxford University Press.

Scharff, Alexander. 1929. Die Altertümer der Vor- und Frühzelt Ägyptens . Vol. 2. Mitteilungen aus der Ägyptischen Sammlung Staatliche Museen zu Berlin 5. Berlin: Staatliche Museen zu Berlin.

Schott, Siegfried. 1950. Hieroglyphen: Untersuchungen zum Ursprung der Schrift . Abhand-lungen der Geistes- und Sozialwissenschaftlichen Klasse 24. Wiesbaden: Akademie der Wissenschaften und der Literatur (Mainz).

Searle, John R. 1979. "What Is an Intentional State?" Mind 98:74–92.

Service, Elman R. 1975. The Origins of the State and Civilization . New York: Norton.

Smith, William Stevenson. 1949. A History of Egyptian Sculpture and Painting in the Old Kingdom . 2d ed. Cambridge: Harvard University Press.

———. 1965. Interconnections in the Ancient Near East . New Haven: Yale University Press.

———. 1981. The Art and Architecture of Ancient Egypt . Rev. ed. William Kelly Simpson. The Pelican History of Art. Harmondsworth: Penguin.

Spencer, A. J. 1980. Early Dynastic Objects . Catalogue of Egyptian Antiquities in the British Museum 5. London: British Museum Publications.

Stewart, Andrew F. 1983. "Stesichoros and the François Vase." In Ancient Greek Art and Iconography, ed. Warren G. Moon, pp. 53–74. Madison: University of Wisconsin Press.

———. 1987. "Narrative, Genre, and Realism in the Work of the Amasis Painter." In Papers on the Amasis Painter and His World, pp. 29–42. Malibu: J. Paul Getty Museum.

Tefnin, Roland. 1979. "Image et histoire: Refléxions sur l'usage documentaire de l'image égyptienne." Chronique d'Egypte 54:218–44.

———. 1981. "Image, écriture, récit: A propos des représentations de la bataille de Qadesh." GM 47:55–76.

———. 1983. "Discours et iconicité dans l'art égyptien." Annales d'histoire de l'art et d'archéologie 5:5–17.

Teissier, Beatrice. 1987. "Glyptic Evidence for a Connection between Iran, Syro-Palestine, and Egypt in the Fourth and Third Millennia." Iran 25:27–53.

Terrace, E. L. B., and Henry G. Fischer. 1970. Treasures from the Cairo Museum . London: Thames and Hudson.

Trigger, Bruce G. 1983. "The Rise of Egyptian Civilisation." In Bruce G. Trigger et al., Ancient Egypt: A Social History, pp. 1–70, 349–52, 365–71. Cambridge: Cambridge University Press.

Vandier, Jacques. 1952. Manuel d'archéologie égyptienne . Vol. 1, Les Epoques de formation . Paris: Picard.

———. 1964. Manuel d'archéologie égyptienne . Vol. 4, Bas-reliefs et peintures: Scènes de la vie quotidienne . Paris: Picard.

Van Dijk, Teun A. 1975. "Action, Action Description, and Narrative." NLH 6:273–97.

Vialou, Denis. 1981. L'Art pariétal en Ariège magdalénienne . 3 vols. Mémoires de la Laboratoire de paléontologie humaine et de préhistoire 13. Paris: Institut de paléontologie humaine.

———. 1982. "Niaux, une construction symbolique magdalénienne éxemplaire." Ars Praebistorica 1:19–45.

Vitoux, Pierre. 1982. "Le jeu de la focalisation. "Poétique 51:359–68.

Walton, Kendall L. 1990. Mimesis as Make-Believe: On the Foundations of the Representational Arts . Cambridge: Harvard University Press.

Ward, William A. 1969. "The Supposed Asiatic Campaign of Narmer." Mélanges de l'Université Saint-Joseph (Beirut) 45:205–21.

Weber, Max. 1978. Economy and Society . Ed. Guenther Roth and Claus Wittich. 2 vols. Berkeley: University of California Press.

Weitzmann, Kurt. 1970. Illustrations in Roll and Codex: A Study of the Origin and Method of Text Illustration . 2d ed. Princeton: Princeton University Press.

Weitzmann, Kurt, and Herbert L. Kessler. 1990. The Frescoes of the Dura Synagogue and Christian Art . Dumbarton Oaks Studies 28. Washington, D.C.: Dumbarton Oaks.

Wildung, Dietrich. 1981. Ägypten vor den Pyramiden: Münchner Ausgrahungen in Ägypten . Mainz-am-Rhein: Phillip von Zabern.

Williams, Bruce. 1986. Excavations between Abu Simhel and the Sudan Frontier . Pt. 1, The A-Group Royal Cemetery at Qustul: Cemetery L . Oriental Institute Nubian Expedition 3. Chicago: University of Chicago Press.

———. 1988a. Decorated Pottery and the Art of Naqada III . Mürichner Ägyptologische Studien 45. Munich: Münchner Ägyptologische Studien.

———. 1988b. "Narmer and the Coptos Colossi. " JARCE 25:35–59.

Williams, Bruce, and Thomas J. Logan. 1987. "The Metropolitan Museum Knife Handle and Aspects of Pharaonic Imagery before Narmer." JNES 46:245–84.

Williams, Caroline Ransom. 1932. The Decoration of the Tomb of Per-Neb: The Technique and the Color Conventions . New York: Metropolitan Museum of Art.

Winkler, Hans A. 1937–38. The Rock Drawings of Southern Upper Egypts . 2 vols. London: Egypt Exploration Society.

Winter, Irene J. 1981. "Royal Rhetoric and the Development of Historical Narrative in Neo-Assyrian Reliefs." Studies in Visual Communication 7:2–38.

———. 1985. "After the Battle Is Over: The 'Stele of the Vultures' and the Beginning of Historical Narrative in the Art of the Ancient Near East." In Pictorial Narrative in Antiquity and the Middle Ages, ed. Herbert L. Kessler and Marianna Shreve Simpson, pp. 11–32. Studies in the History of Art 16. Washington, D. C.: National Gallery of Art.

———. 1987. "Legitimation of Authority through Image and Legend: Seals Belonging to Officials in the Administrative Bureaucracy of the Ur III State." In The Organization of Power: Aspects of Bureaucracy in the Ancient Near East, ed. McGuire Gibson and Robert D. Biggs, pp. 69–106. Studies in Ancient Oriental Civilization 46. Chicago: University of Chicago Press.

Wollheim, Richard. 1987. Painting as an Art . Princeton: Princeton University Press.

Wolterstorff, Nicholas. 1981. Works and Worlds of Art . Oxford: Clarendon.

Wood, David. 1989. The Deconstruction of Time . Atlantic Highlands, N.J.: Prentice Hall.

Yadin, Yigael. 1955. "The Earliest Record of Egypt's Military Penetration into Asia? (Some Aspects of the Narmer Palette, the 'Desert Kites,' and Mesopotamian Seal Cylinders)." Israel Exploration Journal 5:1–16.

Index

A

Abka, 23

Abu Zeidan, 17 , 257 n. 4

Abydos, 17 , 164 , 279

Akhenaten, 243

Amarna, 243 , 274 n. 4

Amenhotep III, 204

Assyrian art, 221 , 251 , 253

B

Battlefield Palette, 6 , 9 , 10 , 11 , 12 , 13 , 16 , 17 , 25 , 68 , 75 , 91 , 119 –43, 145 –60, 161 , 164 , 165 , 166 , 169 , 175 , 192 , 196 , 202 , 203 , 204 , 208 , 213 , 261 n. 4, 269 n. 2

Brooklyn knife handle, 6 , 12 , 14 , 17 , 19 , 38 , 39 , 48 –55, 62 –63, 67 –70, 71 , 92 , 145 , 259 n. 4

Bull Palette, 143 , 169 , 203 , 263 n. 1

Buto, 171 , 271 n. 6

C

Cairo-Brooklyn Palette, 203 –4, 261 n. 4, 264 n. 2, 266 n. 3

Canonical art, 1 –5, 7 , 27 –30, 33 –34, 39 , 41 , 43 , 52 , 55 –61, 98 –99, 128 , 133 , 145 , 150 , 161 , 171 –72, 173 , 194 , 197 –200, 214 –16, 217 , 219 , 224 , 232 , 243 –44, 257 n. 5, 260 n. 1, 262 n, 1 , 263 n. 1, 266 n. 4, 267 n. 1, 274 n. 2, 280 –81. See also Composition; Hieroglyphs

Carnarvon knife handle, 6 , 12 , 14 , 17 , 48 –55, 62 –70, 71 , 74 , 81 , 92 , 145 , 149 –50, 199

Chain of replications, 9 , 12 , 30 –37, 148 –60, 161 –65, 198 –200, 202 , 204 , 210 , 213 , 241 –42, 246 , 259 n. 4. See also Iconography; Intentionality

Combs, 18 , 43 , 48 –61, 77 , 253 . See also Davis comb

Composition, 1 –5, 38 –39, 41 , 43 , 74 , 77 –80, 166 . See also Canonical art; Prehistoric art

D

Davis comb, 14 , 48 –55, 71

Decorated Tomb, 43 , 102 , 129 , 172 , 258 n. 1, 263 n. 2, 264 n. 5, 269 n. 2. See also Hierakonpolis

Decorated Ware. See Pottery, decorated

Dura Europos, 198

E

Early dynastic art. See Canonical art; Khasekhemuwy, art of

Egyptian art. See Canonical art; Prehistoric art

G

Gebel el-Tarif knife handle, 48 , 62 –63, 68 , 87 , 266 n. 4

Gebel Sheikh Suleiman. See Rock art

Gebelein, 21 , 258 n. 1, 269 n. 2

Greek art, 199 , 252 , 253 , 278 n. 7

H

Hathor, 165 . See also Narmer Palette

Hierakonpolis, 17 , 21 , 27 , 48 , 126 –28, 204 , 259 n. 6, 260 n. 1, 279 . See also Decorated Tomb; Narmer Palette; Oxford Palette

Hieroglyphs, 4 , 6 , 11 , 27 , 28 , 96 –97, 124 –25, 126 , 138 , 147 –48, 165 , 168 , 171 –72, 218 , 231 , 245 –46, 253 , 256 n. 1, 262 n. 2, 266 n. 2, 269 n. 2, 270 n. 4. See also Narrative

Hor-Aha, 164 . See also Abydos

Hunter's Palette, 6 , 9 , 10 , 11 , 12 , 14 , 16 , 17 , 25 , 37 –38, 39 , 68 , 75 , 81 , 91 , 93 –118, 119 , 123 , 124 , 125 , 128 , 130 , 133 , 134 , 136 , 137 , 141 , 142 , 143 , 145 –60, 161 , 166 , 189 , 198 , 202 , 204 , 208 –9, 211 , 261 n. 4, 264 n. 5, 274 n. 4

I

Iconography, 7 –8, 24 –27, 29 , 41 , 54 , 127 –29, 133 –34, 146 , 150 , 277 n. 6, 280 . See also Chain of replications; Scene of representation

Incense burners, 43 , 129 , 133 , 260 n. 3

Intentionality, 30 –37, 150 –52, 201 –5, 222 , 271n. 1. See also Iconography; Narrative; Scene of representation

J

Japanese art, 200

K

Khasekhemuwy, art of, 128 –29, 216 , 263 n. 1

Knife handles, 16 , 18 , 24 , 26 , 43 , 48 –63, 110 , 253 . See also Brooklyn knife handle; Carnarvon knife handle; Gebel el-Tarif knife handle; Metropolitan knife handle; Pitt-Rivers knife handle

Koptos, 17 , 21

L

Late prehistoric art. See Prehistoric art

Libyan Booty Palette, 38 , 169 , 194 , 229 –32

M

Mace heads, 203 , 266 n. 3. See also Narmer mace head; Scorpion mace head

Main Deposit. See Hierakonpolis

Masking. See Scene of representation

Meaning. See Intentionality

Medieval art, 22 , 29 , 155 , 241 –42, 251

Memphis, 170 –71

Mesopotamia. See Near East

Metropolitan knife handle, 48 , 262 n. 2

Min Palette, 71 –72

Minoan-Mycenaean art, 81

N

Nagada, 21

Narmer mace head, 18 , 2 62n. 2 , 264 n. 2, 265 n. 2, 266 n. 4, 269 n. 2, 271 n. 6

Narmer Palette, 6 , 8 , 9 , 10 , 11 , 12 , 16 , 17 , 18 , 24 , 28 , 38 , 39 , 48 , 63 , 75 , 102 , 119 , 123 , 124 , 125 , 126 , 128 , 129 , 130 , 133 , 135 , 137 , 140 , 142 , 149 –50, 156 , 160 , 161 –233, 250 , 258 n. 5, 259 n. 4, 260 n. 2, 261 n. 4, 263 nn. 1-2, 264 n. 5

Narrative, 21 , 38 , 41 , 80 –81, 83 , 89 , 102 , 129 , 135 , 145 –60, 175 , 177 –78, 182 –85, 197 –99, 208 , 217 –18, 222 , 234 –55. See also Composition; Scene of representation

Near East, 23 –24, 93 , 194

O

Old Kingdom art, 11 , 28 , 96 , 168 , 214 –16, 224 , 280 –81. See also Canonical art

Ostrich Palette, 13 , 14 , 16 , 17 , 21 , 72 –75, 82 , 83 , 91 –92, 93 , 95 , 97 , 102 , 105 , 130 , 145 , 147 , 155 , 198 , 202

Oxford Palette, 6 , 9 , 10 , 11 , 14 , 17 , 21 , 24 , 27 , 28 , 38 , 70 , 75 –92, 93 , 95 , 96 , 97 , 98 , 99 , 101 , 102 , 104 , 105 , 106 , 107 , 108 , 110 , 111 , 115 , 123 , 124 , 125 , 128 , 130 , 133 , 135 , 140 , 141 , 142 , 143 , 145 –60, 161 , 166 , 172 , 191 , 198 , 202 , 204 , 208 –9, 233 , 260 n. 4

P

Paleolithic art, 3 , 29

Palermo Stone, 18 , 266 n. 2

Palettes, 18 , 24 , 26 , 43 , 71 –72, 74 , 152 –57, 203 , 229 , 253 , 258 n. 2, 260 nn. 1-2, 261 n. 4, 264 n. 2. See also Battlefield Palette; Bull Palette; Cairo-Brooklyn Palette; Hunter's Palette; Libyan Booty Palette; Min Palette; Narmer Palette; Ostrich Palette; Oxford Palette

Pitt-Rivers knife handle, 48 –55, 63 , 77

Pottery, decorated, 23 , 29 , 41 , 43 , 245 , 259 n. 5, 281

Predynastic art. See Prehistoric art

Prehistoric art, 5 –8, 21 , 33 –34, 36 , 55 –61, 129 , 133 , 158 –60, 199 , 219 , 240 , 242 , 253 –55, 257 n. 5, 262 n. 1, 265 n. 1, 279 –81. See also Combs; Incense burners; Knife handles; Palettes; Pottery, decorated; Rock art; Scepters

Q

Qustul. See Incense burners

R

Register composition. See Composition

Representation. See Scene of representation

Rock art, 23 , 41 , 43 , 126 , 245 , 258 n. 1, 281

Roman art, 29 , 155 , 251 , 253 , 277 n. 6

S

Sais, 170

Scene of representation, 14 , 19 , 22 , 32 , 37 –42, 61 , 74 , 93 , 104 , 109 , 136 , 171 –72, 205 –24, 269 n. 2, 272 n. 1. See also Narrative

Scepters, 53 , 68 –70, 71

Scorpion mace head, 48 , 203 , 224 –29, 231 , 265 n. 2, 266 n. 4

Seals, decorated, 23

Seyala. See Scepters

W

Wadi Halfa, 23

Writing. See Hieroglyphs

Preferred Citation: Davis, Whitney. Masking the Blow: The Scene of Representation in Late Prehistoric Egyptian Art. Berkeley:  University of California Press,  c1992 1992. http://ark.cdlib.org/ark:/13030/ft7j49p1sp/